**All You Can Feed: Some Comments on Production of Mouse Diets Used in Biomedical Research with Special Emphasis on Non-Alcoholic Fatty Liver Disease Research**

#### **Sabine Weiskirchen 1, Katharina Weiper 1,2, René H. Tolba <sup>2</sup> and Ralf Weiskirchen 1,\***


Received: 5 December 2019; Accepted: 31 December 2019; Published: 7 January 2020

**Abstract:** The laboratory mouse is the most common used mammalian research model in biomedical research. Usually these animals are maintained in germ-free, gnotobiotic, or specific-pathogen-free facilities. In these facilities, skilled staff takes care of the animals and scientists usually don't pay much attention about the formulation and quality of diets the animals receive during normal breeding and keeping. However, mice have specific nutritional requirements that must be met to guarantee their potential to grow, reproduce and to respond to pathogens or diverse environmental stress situations evoked by handling and experimental interventions. Nowadays, mouse diets for research purposes are commercially manufactured in an industrial process, in which the safety of food products is addressed through the analysis and control of all biological and chemical materials used for the different diet formulations. Similar to human food, mouse diets must be prepared under good sanitary conditions and truthfully labeled to provide information of all ingredients. This is mandatory to guarantee reproducibility of animal studies. In this review, we summarize some information on mice research diets and general aspects of mouse nutrition including nutrient requirements of mice, leading manufacturers of diets, origin of nutrient compounds, and processing of feedstuffs for mice including dietary coloring, autoclaving and irradiation. Furthermore, we provide some critical views on the potential pitfalls that might result from faulty comparisons of grain-based diets with purified diets in the research data production resulting from confounding nutritional factors.

**Keywords:** animal experimentation; diet; nutrition; ingredients; lard; fibers; fructose; diet coloring; autoclaving; irradiation

#### **1. General Aspects of Mice in Biomedical Research**

The laboratory mouse derived from the house mouse (*Mus musculus*) has been first used in biomedical research as a model system since the 17th century [1]. The earliest documentation of the use of mice in scientific research was done in the year 1664 in England, where Robert Hooke in his study used this animal model to study the biological consequences of an increase in air pressure [1]. In the 19th century, mice were used for a couple of breeding experiments, in which coat color or behavioral mutations were studied. Since that, these rodents have been used in many research areas. In 1981, a first genetically engineered transgenic mouse model was introduced that expressed the herpes simplex virus thymidine kinase [2].

Thereafter, both transgenic and knockout mouse models have become essential tools in the field of immunology, oncology, toxicology, genetics, and many more. These models allow the determination of the general consequences of alterations in individual genes and their cooperation with other genes. Particularly, inbred mice, which are "isogenic organisms" (nearly) identical to each other in genotype and phenotype, are frequently used for such studies. In respective experiments, these highly similar "linear test animals" are most suitable to establish reproducible results and conclusions. Moreover, testing in mice is a central part of drug development for humans, in which they are vital as a means for pre-clinical safety and efficacy testing before starting a human trial with a candidate drug. Therefore, it is not surprising that experimental work in mice has developed as an integral part of biomedical research in the building of basic knowledge. Exemplarily, mice experiments have developed as the gold standard to confirm a proposed disease-associated mechanism in hepatology research, in particular, non-alcoholic fatty liver disease (NAFLD) [3]. Specialized protocols have been developed closely mimicking typical clinical situations, including cholestasis, poisoning, metabolic injury, portal hypertension, inflammation, fibrosis, cirrhosis, and hepatocellular carcinoma [3]. During the last decade, the individual steps in such procedures were summarized in highly standard operating protocols (SOPs) to achieve uniformity in performance and outcome [4].

At the end of 2013, the seventh report on the statistics on the number of animals used for experimentation and other scientific purposes in the member states of the European Union (EU) was published [5]. This concise dispatch contains detailed information about the number of laboratory animals used for biomedical research in 2011 in the 27 member states with the exception of France. According to this report, a total of 11.5 million animals were used in biomedical research in 2011. From these, mice are the most commonly used species with 60.9% (~7 million animals) of the total use [5]. Although precise numbers for the worldwide total numbers of mice in biomedical research are not really available, first conservative estimates of annual laboratory animal use suggested at least 115.3 million animals to be sacrificed for scientific purposes in the year 2005 [6]. If globally the frequency of mouse usage in all kinds of animal studies is the same as in the EU (~61%), this means that about 70.2 million mice are annually sacrificed by scientists and clinical researchers. However, the strict implementation of the three Rs (replacement, reduction, and refinement) concept for animal experimentation proposed by Russell and Burch in 1959 [7] that is now mandatory standard in biomedical research in many countries [3], the introduction of relevant replacement methods [8], and finally the increasing political and public concern about animal experimentation influencing people's view toward the use of animals in research [9] have led to a significant reduction in animal number [5].

Nonetheless, all these laboratory mice bred for scientific purposes and kept in laboratory must be supplied with food. When estimating that each of the 70.2 million mice typically eat 3.5–3.75 g of food per day (10–15% of their body weight) [10] and assuming that the life span of a mouse in a typical setting of an Institute for Laboratory Animals Science is about six months (Tolba and Weiskirchen, unpublished), it can be calculated that about 44.23–47.39 million kg dietary food products are necessary to feed these mice. Surely, this value is a high underestimation because potential losses due to throwing away, passing of expiration dates, unwanted food spoilage, and many other circumstances have not been taken into account in this simplified calculation. Moreover, based on special requirements necessary in the individual mice experiments, a large variety of companies have developed dedicated to developing and providing products that meet the unique challenges for all kinds of experiments.

Such "special needs" are considered in products manufactured and marketed as "custom diets". In comparison to "standards diets" or "chows", these products are more expensive in formulation because they require costly dietary manipulation such as the addition of vitamins, minerals, special fats, cholesterol, proteins, dietary fibers, drugs or other compounds. Unfortunately, in the literature the terminology of diets is used somewhat inconsistently. Different reports use terms such as "mouse diet", "rodent chow", "custom diet", "defined diet", "purified chow", "special diet", "purified ingredient diet", "grain-based diet", "standard chow", and many others. However, there are basically only

two types of diets, namely the "grain-based diets" and "purified ingredient diets". While "grain-based diets" are made out of grain, cereal ingredients, and animal by-products, "purified ingredients diets" are composed of highly refined ingredients [11].

In addition, rodent diets may require sterilization techniques when animals sensitive to normal or opportunistic microbes, such as immune compromised or germ free mice, are investigated. Food sterilization or decontamination is possible by exposure to γ-irradiation or by high-vacuum autoclaving. Highly sensitive diets can also be vacuum-, gas-, or modified atmosphere-packed (i.e., nitrogen-purged), which minimizes the risk of spoilage by oxidation.

Therefore, all these circumstances illustrate that the production of standard and customized diets intended to be used as mice feed is a complex business and a science in itself. We here will summarize some important issues on leading manufacturer, the production and diversity of mice research diets, their ingredients, and treatments during production.

#### **2. Producers of Mouse Diets**

Usually most laboratory mice are kept in centralized, well-designed, managed animal facilities, which allow efficient, economical and safe animal experiments. Depending on the design and size of the animal facility, the mice are either kept in high barrier, specified pathogen free (SPF) areas with restricted access to animal facility staff only or in low barrier (conventional) areas with additional access for licensed scientists. Usually, trained and skilled staff takes care of the animals and scientists usually don't care about the formulation and quality of diets the animals receive during normal breeding and keeping. However, scientists investigating certain immunodeficient strains, analyzing diet-induced impairments, or conducting experiments in which nutritional factors interfere with the outcome of their experiments are more interested in the products feed to their mice.

Grain-based diets are commercially manufactured in an industrial process and the safety of products should be addressed through the analysis and control of all biological and chemical materials used in the production process. Companies with an international reputation for quality often are certified by quality assurance systems and work in strict accordance with the guidelines provided by either local (e.g., England: Food Standards Agency, FSA; France: French Agency for Food, Environmental and Occupational Health & Safety, ANSES) and/or international institutions such as the International Organization for Standardization (ISO), the European Commission (e.g., https://ec.europa.eu/food/safety/animal-feed\_en) or the good manufacturing practices (GMP) of the World Health Organization (WHO). If necessary, these specifications must then be adapted locally by the responsible animal welfare authorities. These guidelines or directives ensure that manufacturing, testing processes, and labeling and batch processing record are clearly defined, validated, reviewed, and documented, providing the basis to conduct good laboratory practice (GLP) studies. Furthermore, these regulations guarantee that mouse diets are prepared under good sanitary conditions and truthfully labeled to provide information of all ingredients. As such, they are rather similar to the guideline used for human foods.

#### **3. The Production Process**

#### *3.1. Grain-Based Diets*

Grain-based diets are made in "closed formulas" containing natural ingredients such as soybean meal, ground corn, fish meal, animal byproducts, and very high levels of both soluble and insoluble fibers [12]. In addition, such chows frequently contain non-nutritive but biologically active compounds such as phytoestrogens and toxic heavy metals. Grain-based diets for biomedical research purposes are made in accredited facilities using SOPs that guide all facets of diet production. Each diet formula is manufactured by a fixed formula or are produced by supplementation of a "constant nutrition" designed and supervised by a nutritionist. Depending on the ingredients, fixed formulas can reduce variation of nutrients from batch-to batch. However, grain-based diets may still be subject to variation

due to the complex nature of the ingredients in these diets, which contain multiple nutrients and non-nutrients known to be subject to variation. Different work steps are integrated into a linear sequence within the production process (Figure 1).

**Figure 1.** Schematic overview about the production process of mouse diets. Pellet and extruded diets are produced from the same raw materials. In a first step, the different ingredients are put together in the intended proportion, mixed, grinded to the desired density, and moistened to a desired moisture content (pellet diets: ~14–16%; extruded diet: >20%). Pellet diet is then processed in a pellet mill and dependent on the water content included either directly cooled or dried to a moisture content lower 12.5%. Thereafter, this diet is packed and sterilized for example by irradiation. In contrast, during the production of an extruded diet, the conditioned materials are then forced through an opening of a perforated plate or die to create a product in desired shape and size. It is then dried to moisture content of 8–12%, cooled, and packed. These diets are more or less germ-free because of the high temperature in the extruder barrel and drier. If necessary, these diets can be further sterilized by autoclaving before use. However, the high temperature during the extruding process already warrants low concentrations of microorganisms.

Key steps in the production process of grain-based diets in large quantities are the choice and delivery of raw materials (Figure 2A–C), quality control of these materials (Figure 2D–E), compilation and assignment of lot numbers (Figure 2F), and the computer-controlled mixing of individual compounds (Figure 2G–I) in suitable mixers. When producing an "extruded diet", the mixed substances are then mixed with steam and hot water and fed into the extruder barrel and forced through the die opening to form a product in desired shape and size (Figure 2J–L). During this process the quality and moisture content is continually monitored and/or adjusted. The final product is then packed in multilayer paper bags or sacks (Figure 2M) and screened for unwanted stray metal particles by passing through an in-line metal detection capability (Figure 2N). The packed diets are then transported and stocked in suitable facilities with controlled temperature and humidity conditions to avoid food spoilage (Figure 2O). From there, the diets are quickly retrievable on demand.

**Figure 2.** Overview of some work steps in the production of an extrusion diet in large scale. (**A**,**B**) The first step in production of mouse diets is the selection, procurement and approval of high quality bulk ingredients. In this regard, each company might have a different source of raw materials. (**C**–**E**) From each bulk, samples are taken and screened by a well-trained technician for mycotoxins (e.g., aflatoxin, vomitoxin, fumonisin), protein, fat, fiber, and moisture using near infrared spectroscopy and approved/certified test kits. (**F**) When quality is assured, each lot of raw material receives a lot number that is used to track each ingredient through the entire production process. (**G**–**I**) The different ingredients are mixed together in a certified mixer, in which flow from ingredient bins, scales and processing is critically monitored. (**J**–**L**) To produce an extruded diet form, the mixed ingredients are sent first to a post grind hammer mill and then to an extruder, in which the mixture is forced through a die. In this device, the product is expanded by a stream that is injected under pressure. In a next step, the product is passed through a dryer and several screeners to ensure that no metallic traces or other unwanted compounds are passed through. The moisture and bulk density of the product is evaluated and recorded. (**M**,**N**) Finally, the diet is packed in packing lines and once validated for unwanted stray metal particles by an in-line metal detection capability. (**O**) All diets are stocked in suitable facilities from which they are quickly retrievable on demand. All images were kindly provided by Dr. Jörg Lesting (Envigo Teklad Diets, Madison, WI, USA). A well-arranged movie showing the complete manufacturing process is viewable online at: https://www.envigo.com/p/teklad/.

#### *3.2. Purified Diets*

Purified diets are composed of refined ingredients that have undergone further processing and the composition is open to the researcher [12]. They usually contain a standardized and balanced quantity of proteins, carbohydrates, fats and fibers provided as a mixture of casein, corn starch, soybean oil and cellulose. Therefore, these diets should be always constant in composition from batch to batch. Importantly, they can be individually tailored with all kinds of compounds and further changed in regard to their relative amount of standard ingredients.

More specialized diets produced on request of the customer are usually prepared in much lower quantities. The production process is more laborious and the equipment used such as the mixers are much smaller (Figure 3A). Furthermore, most of these diets are irradiated and are delivered in smaller heat-sealed packages. To avoid mix-ups of these customized diets with other diets, these products are frequently colored with artificial food colors. Therefore, the appearance of these diets can be highly diverse (Figure 3B). Of course, such products are more expensive because of the labor-intensive work process and the often expensive ingredients requested by the customer.

**Figure 3.** Production of customized purified mice research diets. (**A**) The manufacturing of a mouse diet used for basic and clinical research is a complex and highly-controlled process resembling that of producing bakery products for humans. Shown is a mixer in which the ingredients are blended. After that a known amount of water (based on the diet formula) is added and the product is properly shaped. The wet powdered diet is forced through a spinning die and without added heat, the pellets from their cylindrical shape, are cut to approximately the same length and then dried under low humidity conditions to remove the water added for pelleting. Dependent on the production process, different amount of diet ranging from several kilograms to several tons can be produced in a single batch. (**B**) Color coding by the addition of non-toxic dyes allow discriminating the different food products in animal facilities, in which animals are kept requiring different nutrients. The images depicted were kindly provided by Dr. Matthew Ricci (Research Diets, Inc., New Brunswick, NJ, USA).

High-fat diets (HFD) are widely used in studies of diet-induced obesity (DIO) and metabolic injury [13]. Diets enriched with different dietary fibers such as barley beta-glycan, apple pectin, inulin, inulin acetate ester, inulin propionate ester, inulin butyrate ester or combinations thereof have recently been shown to induce specific differences in cecal bacteria composition [14]. The beneficial effects of inulin-enriched diets and their modulatory role in microbiota composition were also sufficient to reduce inflammatory gene expression in hippocampus of APOE4 transgenic mice that develop systemic metabolic dysfunction and symptoms of Alzheimer's disease [15]. In line, the supplementation with galactooligosaccharide improved the intestinal barrier in hyperlipidemic mice lacking the low-density lipoprotein receptor (LDLR) [16]. In mice, dietary fructose impairs mitochondrial size, function, and protein acetylation, thereby decreasing fatty acid oxidation and development of metabolic dysregulation [17]. Strikingly, the exposure of mice in utero and in the pre-weaning period to a methyl donor-supplemented diet provoking DNA methylation resulted in significant attenuation of repetitive motor behavior development that persisted through early adulthood [18].

All these examples show that the diet has tremendous impact on mouse health and disease and that profound changes in the composition of diets may affect the reproducibility or outcome of mice experimentations. The knowledge of the composition of a diet during an experiment is highly crucial to maximize experimental reproducibility requiring highly standardized conditions.

To reduce experimental variation among laboratories, the American Institute of Nutrition (AIN) established a committee in 1973 with the aim to establish general guidelines in preparing dietary standards for nutritional studies with laboratory rodents [19]. These recommendations should help scientists with limited experience in experimental nutrition and facilitate interpretation of results among experiments and laboratories [19,20]. The first diets introduced by the respective committee, i.e., AIN-76 and AIN-93, contained fixed formula supporting growth, reproduction and lactation [20]. Although subsequent changes in respective diets were introduced later, these nutritional guidelines are still applicable today and provide a global standard for purified mouse diets [19].

In line with these efforts, small size feed suppliers or larger companies accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) or other councils produce and/or market standardized mouse diets with fixed formulation, which allow reducing the fluctuation in study results (Table 1).


#### **4. Pasteurization**

Pasteurization is a term named for the French scientist Louis Pasteur for a mild heat-treatment process used to destroy pathogenic microorganisms and preventing of spoilage of foods and beverages. For pasteurization of milk, for instance, a low-temperature, long-time process (LTLT) for 30 min heating at 63 ◦C or a 15 sec high-temperature short-time process (HTST) is used to inactivate large (but not all) spoilage-causing vegetative forms of microorganisms [21]. Beside LTLT and HTST, very short heating to 138 ◦C or above for at least 2 sec (ultra-pasteurization) is used for specialized applications [21].

Autoclaving referring to a process of sterilization under pressure is believed to be one of the most efficient methods of sterilization and common practice in many research institutions [22]. This method is inexpensive, convenient and guarantees the destruction of all microorganisms including spores and viruses. Therefore, pasteurization is commonly used for "sterilization" of mouse diets (Figure 4).

Usually, animal diets are autoclaved at 121 ◦C for 20 min, which can result in pellet hardness resulting in a significant reduction in wastage and in apparent and true consumption of the pelleted diet [23]. In addition, losses of pantothenate, vitamin A and vitamin D were found during autoclaving, while thiamine, riboflavin and pyridoxine were less affected [24]. In particular, autoclaving at higher temperatures for shorter period were more detrimental than autoclaving for longer time intervals at

lower temperatures [24]. Therefore, feed fortification with vitamins is potentially necessary, especially after autoclaving at high temperatures to ensure that the maintained mice receive an adequate and balanced supply with these essential nutrients.

**Figure 4.** Sterilization of mouse diets by autoclaving. (**A**) Industrial autoclaves that can be used for sterilization of large batches of rodent diets must have large capacities. In principal, these devices are large pressure chambers, in which the goods are sterilized by subjecting them to a pressurized saturated steam at 121 ◦C (249 ◦F) maintained for about 20 min in a locked chamber. Different electronically-controlled valves and lines regulate steam flow and temperature in the steam chamber. (**B**,**C**) A typical industrial computer-controlled autoclave is depicted. (**D**–**F**) Autoclaves used for food production can be highly variable in size. Images showing different autoclaves from Steriflow (Roanne, France) were kindly provided by Kai Bergner from Vos Schott GmbH (Butzbach, Germany). A vivid 3D animation video of a typical water cascading process for sterilization by autoclaving can be found at: https://youtu.be/bWD87VVtzKU.

In addition, it is known that this procedure exerts undesirable effects on feed quality due to production of toxic compounds (e.g., acrylamide) and reduction of the overall nutritional value [22]. The autoclaving of a standard rodent diet resulted in a 14-fold increase in acrylamide, while the content of endogenous acrylamide in diets subjected to irradiation was reduced [25]. The forming of acrylamide is strongly correlated to the temperature used for sterilization [22,25]. In addition, autoclaved food products with high quantities of acrylamide produce elevated concentrations of epoxides, which are highly reactive chemicals, acting as mutagens [22]. Therefore, investigators and institutions should consider the detrimental and toxic effects that autoclaving might provoke in mouse diets.

#### **5. Irradiation**

In some cases, the diets are irradiated with γ-rays to eliminate remaining microorganisms residing in the feed. The microbial reduction strategies are often used to sterilize diets used for animals kept under SPF conditions [26]. In a typical sterilization process, the required dose depends on the "initial bioburden" and irradiation doses of between 20 and 30 kGrays (kGy) are used most frequently to treat diets intended for SPF animals, while larger doses (40–50 kGy) are recommended for diets intended for gnotobiotic or germ-free animals [26]. The physical unit Gy is defined as the absorption of one J of radiation energy per kg of irradiated material. For irradiation of large quantities of mouse diets, the pallets of product are loaded onto conveyors moving around the γ-ray source. In most of the commercial facilities for food irradiation, cobalt-60 (60Co) is the most common source of γ-rays. The respective facilities contain a number of safety systems, which are designed to avoid exposure of personnel to radiation. Furthermore, such irradiation devices are girded by a thick shield that hampers the penetration of γ-rays to the outside (Figure 5).

**Figure 5.** Irradiation of large batches of mouse diets. To minimize the risk of diet spoilage by pathogenic organisms, diets can be exposed to ionizing radiation. For irradiation in large scale, the packed diets are most commonly irradiated with γ-rays from a cobalt-60 (60Co) source that has high penetration depth and dose uniformity and is able to penetrate relatively dense products. When not in use, the radiation source is stored in a water-filled storage pool, which absorbs the radiation energy. This Cherenkov radiation results in a blue appearance of the water bath, which is commonly known as "blue glow". For radiation, the 60Co rods are lifted out from this pool and the emitted energy is directed to the goods to be irradiated or the goods are moved around the γ-ray source. The facility is surrounded by a thick concrete wall to avoid radiation leakage into the environment. The photos of the irradiation facility and the blue glow were kindly provided BGS Beta-Gamma-Service GmbH & Co. KG (Wiehl, Germany, ©BGS/M. Steur).

Although it is often argued that the doses used to destroy microorganisms are rather low, caution is advised because γ-rays at these doses have profound effects on some the integrity of the individual ingredients of the diet. This was impressively shown in a systematic study, in which the amounts of fat, protein, carbohydrate, and vitamins was investigated, showing that γ-rays at a dose of about 30 Gy have profound and selective effects on the stability of vitamin A and peroxide content of dry animal diets [26]. In line, a more previous study summarizing the main finding of published literature showed profound losses of the vitamins C, B1, E, and A in food after its irradiation [27]. Moreover, destruction of highly polyunsaturated fatty acids up to 98% and destruction of fatty acids with two double bonds

up to 46% with accompanying lipid peroxide formation after irradiation with doses of 2–10 kGy [28], in which the effective dose of γ-rays refers to the amount of radiation that penetrate in the middle of the product to be irradiated.

#### **6. General Remarks on the Origin of Nutrients in Purified Diets**

In nature, mice have an extremely diverse diet, consuming practically any food source to which they have access. In order to be fed up, wild mice spend many of their active hours (~20 h) searching for these food products. Contrarily, the amount of time taken for a laboratory mouse to gnaw and eat well-balanced food directly from the cage hopper is considerably less [29]. Ideally, these preformed diet products are nutritionally either complete for various life stages from breeding through long-term maintenance, or adapted to special needs occurring during husbandry and housing. Global sold mouse diets contain a well-balanced mixture of proteins, carbohydrates, fats, vitamins, minerals, and potential additives that may be modified for the special needs of the biomedical research undertaken. In these formulations, substances that have been reported to have adverse confounding effects on experimental results or are toxic to the animals should as far as possible be omitted and should conform to the nutrient requirements of mice established by the National Research Council (see below). When properly stored at room temperature or cooler, depending on the composition of the diet, with ideally lower 50% relative humidity, these diets are usually stable for 6–12 months. To prevent continuous exposure to light and air, the storage in the original packaging or closed containers is recommended. Special diets enriched with temperature-sensitive additives may require the storage at lower temperature (4 ◦C or −20 ◦C). To avoid contamination, the products should be stocked in a proper environment.

The individual substance classes of which a diet is composed may originate from different sources (Table 2). The respective source may have an impact on energy intake, feed efficiency, apparent nitrogen and fat digestibility, composition of gut microbiota, and of course, of body weight development in mouse [30].



#### *6.1. Proteins*

Casein, soy protein isolate, egg white proteins often serve as sources for proteins in respective chows. Caseins are the most frequent protein constituent in animal milk from cow, sheep and buffalo containing with an intrinsically disordered structure forming large colloidal particles with calcium phosphate to form casein micelles [31]. It is enriched in proline, which distorts protein folding into α-helices and β-sheets preventing the formation of higher proportions of secondary and tertiary protein structures. However, casein proteins are important nutritionally because of their high phosphate content due to which they bind significant quantities of calcium ions [31].

Compared to casein, soy protein isolate has a hypocholesterolemic effect [32] due to a lower intestinal absorption of cholesterol, increased steroid excretion, and a greater biological activity in decreasing hepatic lipogenic enzymes [32]. As a consequence, mice fed soy protein isolate or soy protein isolate hydrolysate diets have lower body weight, lower plasma cholesterol and glucose levels compared to animals that are fed with a casein diet [32].

Egg white contains proteins high in amino acid balance, only low quantities of carbohydrate, and is almost free of fat [33]. Compared to other diets, the consumption of egg white in mice has no significant impact on total cholesterol, high density lipoprotein (HDL), low density lipoprotein (LDL), or triglyceride levels, and suppresses food intake, dietary fat absorption, and fat accumulation, thereby preventing the formation of glucose tolerance [34,35]. However, white egg supplementation is supposed to induce oral desensitization and immune tolerance in mice [36]. Nowadays, egg white as a protein source is not often used. Nevertheless, this protein source can be helpful in studies analyzing effects of zinc-deficient diets. This is due to the fact that the zinc-binding capacity of casein is about 8.4 μg–30 μg/mg casein [37,38], while the maximal amount of bound zinc in egg white products is estimated to be only 1.3–1.6 ng/mg [39]. Moreover, egg white contains large quantities of the anti-nutrient avidin having strong affinity for biotin, preventing its absorption across the gastrointestinal tract [40]. Therefore, supplementation with biotin is sometimes necessary when using egg white as the major protein source.

Chemically-defined diets containing crystalline amino acids as the sole source of nitrogen as an alternative to complete proteins have also been successfully used for mouse maintenance [41]. It has been shown already decades ago that amino acid rations, when properly compounded, will provide a rate of growth in mice closely approaching that obtained with casein [42].

#### *6.2. Carbohydrates*

It is well-known that mice become obese when offered free access to sugars, but it is not established whether specific sugars are more likely to cause DIO [43]. Most common in purified diets as sugar sources are sucrose, fructose, and corn starch. The disaccharide sucrose is a disaccharide composed of glucose and fructose produced naturally in plants and after oral uptake efficiently hydrolyzed by sucrose in the intestinal mucosa to its constituent monosaccharides [44]. Free glucose elicits a glycemic and insulinemic response that stimulate the uptake of this sugar into cells, while fructose is mainly metabolized in hepatocytes via insulin-independent mechanisms not regulated by energy supply [44]. Sucrose has an energy of 16.8 kJ (4 kcal) per gram. Interestingly, sucrose stimulates higher daily intakes than isocaloric fructose solution in mice [43]. In animals, the fruit monosaccharide fructose produces profound metabolic disturbances, including insulin resistance, impaired glucose tolerance, high insulin and triglyceride levels, hypertension, dyslipidemia, and microvascular hepatic steatosis [45]. However, the susceptibility to sugar-induced obesity varies with strain [45]. It has an energy density of 15.75 kJ/g (3.75 kcal/g) and feeding mice with a high fructose diet induces hepatic lipid accumulation by activating lipogenic gene expression and de novo lipogenesis [46]. Therefore, this sugar is supposed to be one of the key dietary catalysts in the development of non-alcoholic fatty liver disease [47]. Interestingly, elevated uptake of fructose in mice can result in dysbiosis, increased hepatic lymphocyte infiltration, and further inflammation of gut, liver and fat tissue [47].

Corn starch also known as "cornflour" is a glucose polymer, highly branched carbohydrate (e.g., the starch) derived from the endosperm of the kernel of corn (maize) grain. In its pure form it is a tasteless, odorless, and cold water insoluble powder. In the body, starch is hydrolyzed by amylases into its constituent sugars. Its energy content is about 15.95 kJ/g (3.8 kcal/g). Depending on its formulation, certain glucose polymers may resist digestion in the small intestine in mammals and arrive in the colon where they will be fermented by the gut microbiota resulting in a large variety of products including short chain fatty acids (acetate, propionate, butyrate) that provide as a prebiotics a range of physiological benefits [48]. This should be critically kept in mind when performing experimental studies analyzing the impact and composition of gut microbiota on energy homeostasis, development

of obesity and its metabolic consequences [49]. Similarly, the feeding of C57BL/6J mice with HFD supplemented with resistant starch derived from maize resulted in an altered gut bacteria composition and corroborated with a significant shift in the liver metabolome [50].

#### *6.3. Fats*

A normal rodent diet contains about 10 kcal% fat, while diets enriched with 30–60 kcal% are defined as HFD, provoking significant weight gain and insulin resistance [51]. Typical fat constituents in mouse diets are lard, corn oil, safflower oil, or Menhaden oil.

#### 6.3.1. Lard

Lard is a semi-soft fat derived from adipose tissue of the pig and contains a high content in saturated fatty acid (~30%) and <1% trans-unsaturated fatty acids (i.e., trans fats). In some cases, lard is hydrogenated or treated with bleaching and deodorizing agents, emulsifiers, and antioxidants to improve its stability. The energy content of lard is about 37.6 kJ/g (9 kcal/g). Interestingly, in rodents lard-based HFD accentuated the increase in weight gain and the development of obesity and insulin resistance more than a diet that was based on hydrogenated vegetable-shortening diets, suggesting that the outcome of consuming HFD is strongly dependent on the used fat constituent [52]. Moreover, lard-based diets were significantly more inferior than soybean oil in protecting mice after application of the powerful hepatotoxin carbon tetrachloride twice a week for three weeks, which is a model to generate liver necrosis and steatosis, potentially indicating its less antioxidant activity [53].

#### 6.3.2. Corn Oil

Refined corn oil is derived from the germ of maize and typically contains 99% triacylglycerols with 59% polyunsaturated fatty acid (e.g., linoleic acid), 24% monounsaturated fatty acid (e.g., oleic acid), and 13% saturated fatty acid (e.g., palmitic acid, stearic acid, arachidic acid) [54,55]. It is categorized as one of the richest sources of health-promoting phytosterols and tocopherols protecting against DNA damage, hypertension, platelet aggregation, hypercholesterinemia, and diabetes [54]. In line with these beneficial effects, high corn oil dietary intake was shown to improve health and longevity of aging mice when fed at normal energy balance [56]. In addition, disease development and progression as well as deposition of extracellular matrix within the liver in a mouse model of non-alcoholic steatohepatitis (NASH) was significantly reduced when the HFD was composed of corn oil instead of non-trans fats [57].

As mentioned, the dietary intake of corn oil is known to improve health and longevity of mice, which corroborates with reversing aging-increased blood lipids and decreasing serum pro-inflammatory markers [56]. In addition, the olfactory cues and the oily texture of corn oil are important orosensory factors provoking a strong appetite in mice [57]. However, other reports showed that the excess dietary intake of polyunsaturated fatty acids is associated with loss of spontaneous physical activity and development of insulin resistance [58]. In addition, polyunsaturated fatty acids are subject to oxidation. Therefore, the AIN recommended the supplementation of antioxidants in formulations containing large quantities of corn oil [19].

#### 6.3.3. Safflower Oil

The safflower (*Carthamus tinctorius*) or safflor is a thistle-like annual plant of the *Asteracea* family from which vegetable oil can be extracted from its seeds. Safflower seed oil is flavorless and colorless and in its composition similar to oil from sunflowers, olives, and peanuts, typically containing high content of linoleic acid (63–72%), oleic acid (16–25%) and linolenic acid (1–6%) [59]. In particular, the high content of linoleic acid was shown to have highly beneficial health-promoting effects by reducing the expression of lipogenic enzymes and increasing the activity of hepatic fatty acid oxidation enzymes [60].

#### 6.3.4. Menhaden Oil

The forage fish menhaden (*Brevoortia tyrannus*) belongs to the herring family and forms large flocks occurring on the North American Atlantic coast from Nova Scotia to Florida and related forms are also found up to the coasts of Argentina. The oil derived of these animals is rich in omega-3 polyunsaturated fatty acids such as eicosapentaenoic acid (EPAc) and docosyhexaenoic acid (DHAc), both supposed to have anti-inflammatory activities [61]. Recently, it was demonstrated that EPAc and DHAc supplementation in the context of HFD partially mitigated reductions in insulin sensitivity and maintaining cell function [62]. Moreover, the polyunsaturated fatty acids in menhaden oil prevented high-fat diet-induced fatty liver disease in mice [63].

#### *6.4. Vitamins*

Mice, like humans, require some essential micronutrients in small quantities that cannot be synthesized by their own. These vitamins are organic molecules and must be obtained through the diet, or alternatively synthesized by microorganisms in the gut flora. They have diverse biochemical functions and are commonly sub-classified as either water-soluble (vitamin C, vitamin B1, vitamin B2, vitamin B3, vitamin B5, vitamin B6, vitamin B7, vitamin B9, vitamin B12) or fat-soluble factors (vitamin A, vitamin D, vitamin E, vitamin K). In comparison to fat-soluble vitamins that can accumulate in the body, water-soluble vitamins are readily excreted from the body. Deficient intake (primary deficiency), malfunction during absorption or use of a vitamin (secondary deficiency), or increased consumption results in hypovitaminosis. In contrast, excess intake results in hypervitaminosis occurring mainly only with fat-soluble vitamins (e.g., vitamin A and D). The different vitamins are involved in many biochemical processes (Table 3). Therefore, any shortage might result in complex illnesses potentially affecting different organs. General guidelines defining the nutrient requirements of the mouse are available (see below) [10].




**Table 3.** *Cont*.

Based on the heterogeneous character of the different vitamins, their stability is highly variable. Quantitatively deterioration in content over time of vitamins can be affected by many factors, including temperature, moisture, oxygen, light, pH, oxidizing and reducing agents, catalytic activity of metals, mutual damage by other vitamins, detrimental compounds (e.g., sulphur dioxide), or combination of these factors [79]. For example, vitamin B12 is decomposed by light, alkali, acids, and oxidizing or reducing agents, while on the contrary vitamin B2 (riboflavin) and vitamin B3 (niacin) are rather stable [79].

In addition, during production and handling of mouse diets, there are several factors affecting the stability of vitamins during extrusion. These occur for example during handling of raw material, mixing, conditioning, processes, changes in moisture, heat or pressure treatments during extrusion and expansion [80]. Aspects of stability of vitamins and reduced levels of vitamins during processing of fish feed were concisely discussed by Riaz and coworkers [80]. Since the production of grain-based diets is rather similar, the reported values should be comparable with these values. For the different vitamins, the factors affecting vitamin destruction during processing and storage are different (Table 4). Sufficient supply with vitamins in mouse diets can be guaranteed by food fortification.


**Table 4.** Vitamin losses during pelleting, extrusion and storage of feeds and factors affecting vitamin deterioration.

Information of this table was taken in simplified and modified form from [80] and complemented with data from [79,81].

#### *6.5. Minerals and Trace Elements*

Minerals, also known as macrominerals or micronutrients, are inorganic elements which generally occur in large quantities, while trace elements or microminerals normally are present only in small amounts in organisms. Calcium, phosphorus, chloride, magnesium, phosphorus, potassium, sodium, and chloride are the elements playing vital roles in the body [82]. They regulate the proper composition and function of the body fluids, tissue, bone, teeth, muscles and nerves. Some of them also have function as a coenzyme in metabolic reactions and guarantee biochemical functions in body homeostasis, including energy production, growth, wound healing and proper utilization of vitamins and other nutrients. Essential trace elements required in smaller amounts by animals and plants are iron, zinc, copper, nickel, molybdenum, manganese, selenium, iodine, and others. These elements are involved in vital enzymatic reactions by acting as cofactors or by stabilizing cellular structures [82]. Based on their involvement in hundreds of biological processes, inadequate mineral and trace element intake can result in severe health conditions that can affect nearly all organs and tissues. These can be highly variable and become most evident after chronic shortage (see below).

#### *6.6. Fibers*

In its simplest definition, fibers are non-starch polysaccharides composed of a large number of monosaccharides that are linked through covalent bonds [83]. The term "dietary fibers" is often used to designate the sum of non-starch polysaccharides with its complex fibrous, tasteless organic polymers (i.e., the lignin) forming key structural materials in the supportive tissue of plants of which it is derived of [83]. Based on its composition, these dietary fibers have different physiochemical properties regarding size, hydration, viscosity, fermentability, and impact on satiety. In addition, the proportion of the cell wall components varies from plant to plant and is further dependent on the age and type of plant tissue.

Dietary fibers are roughly grouped into soluble and insoluble fibers. Soluble fibers (non-cellulosic polysaccharides, arabinoxylans, β-glucans, some hemicelluloses, pectins, gums, mucilages, inulin) dissolves in water and are broken down in the gut some of which form a thick, spread-out gel, while insoluble fibers (cellulose, some hemicelluloses, lignin, resistant starch) are left intact as food moves through the gastrointestinal tract [84]. Some soluble fibers block the uptake of fats and are used as a fermentable energy source for gut bacteria. On the contrary, insoluble fibers are indigestible and speed up the elimination of toxic waste in the digestive tract through promoting bowel movement in the colon, thereby preventing constipation.

Fibers can be further sub-classified as neutral detergent fiber (NDF) and acid detergent fiber (ADF), in which NDF is the complete fraction of insoluble residue following neutral detergent digestion and ADF is the harder to digest part of the fiber. In other words, ADF is the sum of cellulose and lignin and NDF is the sum of ADF and hemicellulose. ADF is the fraction of fibers that contain virtually no fermentable ability and reduces overall digestible energy from the diet and NDF is a measure of most of the fiber in the diet (except for soluble fiber, which is not part of this fraction). Therefore, a high content of ADF in a diet will provide lower amounts of energy than a diet with lower ADF amount [85].

There are a large number of fiber sources used to dilute the nutrient and energy density of the diets (cf. Table 2). When fibers are included in rodent chows, the weight of the cecum and colon may increase and microbial fermentation results in short-chain fatty acid (SCFA) production such as acetate, propionate and butyrate having beneficial effects on mice health [86,87]. Moreover, addition of fibers that dilute the nutrient density of the diet will have effects on food intake and body weight and further impact the fecal and urinary nitrogen excretion as a result of microbial fermentation [10]. In humans, diets with a high content of fibers are suggested to have beneficial effects, including increasing the volume of fecal bulk, decreasing the overall time used for intestinal transit, promoting the elimination of toxic waste, stimulating the intestinal flora, and finally reducing the onset risk of metabolic syndromes (Figure 6) [88]. Comparable to humans, a low-fiber diet was shown to promote expansion and activity of colonic mucus-degrading bacteria, suggesting respective diets are ideally

suited as nutritional models for analyzing aspects of colonic mucus layer dysfunction and altered pathogen susceptibility [89].

**Figure 6.** Beneficial effects of dietary fibers in humans. The ingestible parts of plants help to speed up the elimination of toxic waste in the digestive tract through promoting bowel movement in the colon and reacting with bacteria in the lower colon, thereby producing short chain fatty acids (acetate, propionate, butyrate) causing cancer cells to self-destruct. Inadequate fiber intake during malnutrition results in distortion of the mucosa, reduced intestinal barrier function, and inflammation. Similarly, fiber-free diets provoke degradation of mucus layer and barrier dysfunction in mouse.

#### **7. Nutrient Requirements of the Mouse**

Mice as humans need a balanced, fresh and healthy diet that meets their nutritional needs. Nutrients designed for rats, guinea pigs, hamsters or other herbivores are not necessarily suitable for mice, because they need sufficient quantities of essential amino acids, fatty acids, vitamins, and minerals that might vary in content to other animals.

When housed under a standard 12-h light/12-h dark cycle, mice typically consume the majority of their food during the dark period, with short bouts of feeding during the light period [90]. There are a number of factors impacting food uptake, including strain differences, genetic background in transgenic and knockout mice, age, stress, habituation, forced movement, and discomfort resulting from drug treatments or surgeries. Moreover, the energy balance in female mice is strongly affected by hormonal variation associated with the estrous cycle [90].

Detailed guidelines for the nutrient requirements of laboratory animals were first published in 1962 and updated several times [10]. In the most recent edition, published in 1995, the composition of an adequate nutrition of mice maintained in conventional animal facilities is in detailed listed (Table 5). It should be noted that mice kept in a germ-free SPF facility or subjected to experimental-induced stress have altered nutrient requirements that should be adapted accordingly.


**Table 5.** Nutrient requirements of mice maintained in conventional animal facilities \*.

\* The information depicted for individual nutrients was taken from the National Research Council (NRC) guidelines [10]. According to these guidelines, the nutrient requirements are expressed on an as-fed basis for diets containing 10% moisture and 16–17 kJ metabolizable energy per g and should be adjusted for diets of differing moisture and energy concentration. \*\* This calculation of this parameter assumes that the average nitrogen (N) content of proteins is about 16 percent, which led to the use of the calculation N × 6.25 (1/0.16 = 6.25) to convert nitrogen content into protein content. The amount of protein is given for animals maintained under regular growth conditions.

It was demonstrated some years ago that the composition of the commensal gut microbiota in humans correlates with diet and health in the elderly [91]. Moreover, in aged mice some of the alterations associated with aging can be rescued by fecal transfer [92]. In this context, it should be noted that the eating of fresh feces, which is a natural behavior of mice, is possibly not only helpful to better absorb nutrients/minerals they need to stay healthy, but is further a requirement to slow down the aging processes caused by nutritional deficiencies. The consequences of a chronic shortage in a specific nutritional compound have been best documented in mouse studies in which diets were fed lacking individual substances. Such studies have shown that the permanent lack in a specific component evokes severe consequences (Table 6).


**Table 6.** Consequences of insufficient supply with selected nutritional components.


**Table 6.** *Cont*.

The resulting phenotypes resulting from chronic shortage in specific elements or compounds can vary dramatically. The nutritional status impacts growth, reproduction, longevity and determines the response to pathogens, environmental stress, and organ function. Therefore, the avoidance of inadequate intake is one important factor in guaranteeing the welfare of the mouse kept in an animal facility.

#### **8. Representative Examples of Diet-Induced Obesity and Fatty Liver Disease**

The composition of a diet has strong impact on the health of an organism. It influences the composition of the gut microbiota and overfeeding or fasting can cause disease. Therefore, scientists frequently use such model to analyze aspects of diet-related diseases. On the contrary, lifelong caloric restriction is an effective experimental tool to reprogram hepatic fat metabolism and to extend life span in diverse species [126].

In biomedical research, mice are the most widely used animals and the nutrient requirements might depend on development state, reproductive activity, age, and stress factors induced by the experimental conditions. Moreover, there is a great danger that mice, when housed in standard laboratory under ad libitum feeding conditions having continuous access to food but virtually no environmental stimulation become overfed and sedentary and are potentially not suitable as proper controls in animal experiments [127]. Similarly, unwanted contaminants such as pesticides, mycotoxins, heavy metals, nitrosamines, nitrates, nitrites, phytoestrogens with estrogenic activity, and polychlorinated biphenyls in dietary products may affect the outcome of animal studies when present at a sufficient high concentration [128]. Maximum allowable concentrations of these undesirable substances in mouse diets are also specified by the guidelines provided by organizations such as the US Environmental Protection Agency (EPA), the Food and Drug Administration (FDA), the British Association for Research Quality Assurance (BARQA), or the Society of Laboratory Animal Science, GV-SOLAS [128].

Therefore, depending on the research question, the nutritional requirements must be carefully considered. Contrarily, the usage of specific diets in mice is widely applied to induce diseases mimicking

human pathologies including liver disease, metabolic dysfunctions (insulin resistance, diabetes type 2), heart failures, immune system alterations, neurological disorders, or even cancer. These diet-induced models are enriched in specific fats, sugars, toxins, metals, or alternatively lack essential nutrients that are indispensable for the proper synthesis of essential nutrients. In particular, studies analyzing aspects of the immune system require special needs in regard to sterility and compounds that might interfere with the composition of the gut microbiome or function.

In most countries, the feeding of diets provoking the formation of disease or animal suffering require permission of responsible animal welfare authorities and should be carried out in an ethical framework that minimize fear, pain, stress and suffering of animals. This is best done, when respective experiments are carried out following established SOPs providing details about the scientific background, its implementation, experimental details (handling, concentrations, duration of procedure, biometric aspects, readout systems), and about the animal burden associated with this procedure [4]. The diversity of mouse diets is extremely versatile. The diets might vary in size, form, color and of course in nutritional composition (Figure 7).

**Figure 7.** Diversity of mouse diets. (**A**) Diets can be produced in different appearance. (**B**) Diets can be produced in variable shape and color. (**C**) Uncolored diet, (**D**) yellow-stained diet, (**E**) blue stained high-fat diet, and (**F**) doxycycline hylate added diet. These figures were kindly provided by Dr. Dr. habil. Annette Schuhmacher (Ssniff Spezialdiäten GmbH, Soest, Germany).

Nowadays, a large number of diet-induced disease models are established in biomedical research. Most common are models to induce atherosclerosis, obesity, diabetes type 2, or liver damage. In the following, we will discuss some aspects of representative diets used in hepatology research, in particular, NAFLD. These are a typical control diet and four diets that are used to induce fatty liver injury, namely a diet to induce DIO, a typical Western diet (WD), a diet rich in fat, fructose and cholesterol (FFC), and the so called methionine-choline deficient (MCD) diet. Representative compositions of such diets are given in Table 7.

These four diets are some examples of diets commonly used in experimental hepatology research as nutritional models to induce a spectrum of disorders associated with accumulation of excess fat in the liver. The most common form is NAFLD and a more serious condition named non-alcoholic steatohepatitis (NASH). NAFLD and NASH and are the most prevalent liver diseases in Western society and the third leading cause for liver transplantation in the US [129]. Furthermore, there is evidence that NAFLD precedes and is associated with the metabolic syndrome characterized by obesity, diabetes, insulin resistance, and hypertension [130]. Phenotypically, patients with NASH/NAFLD are characterized by liver cell injury and damage, inflammation, and an increased risk for liver fibrosis and carcinogenesis [129]. Based on the eminent importance of NAFLD, several experimental dietary mouse models were developed to mimic the pathogenesis of human NASH and NAFLD.




**Table 7.** *Cont*.

\* The concentration of the individual components were taken from Ssniff diets with order numbers E15712 (control), E15742 (DIO), E15721 (Western diet), E15766-3402 (NASH diet), and E15653 (MCD diet), respectively. \*\* Abbreviations used are: DIO, diet-induced obesity; FFC, fat-, fructose- and cholesterol-rich diet; HF, high fructose; MCD, methionine-choline deficient; MNB, menadione nicotinamide bisulfite; WD, Western diet. \*\*\* 1 mg Vitamin A corresponds to 3333 IU. \*\*\*\* 1 μg Vitamin D3 corresponds to 40 IU. Percentages [%] are given in relation to the whole weight of the diet. Underlined values mark special features in the respective diet.

When comparing a diet used for DIO with a control diet, the most striking difference is the high-fat content of the DIO diet (cf. Table 7). Typically, mice fed a DIO diet containing 40–60% of calories from fat for 7–30 weeks increases their body weight and propensity to develop pre-diabetic symptoms and metabolic syndrome. This type of diet is, therefore, often used in studies investigating aspects of food intake, energy expenditure, glucose tolerance, insulin resistance, and elevated blood pressure [130,131]. When using this model, a slight increase in body weight can be noticed already after 2–4 weeks, while the body weight gradually increases thereafter and is 20–30% higher in mice after 16–20 weeks compared to chow-fed mouse [132]. However, the outcome of the DIO model is influenced by many factors, including genetic background, gender, age, and environmental factors such as cage placement, mice density, and mice handling [132].

While in a typical control diet, the fat and sugar content is not higher than 10%, a WD is characterized by a high content of fat combined with a high amount of a sugar as sucrose or fructose. In some cases, these diets are enriched with trace of SCFA such as C4:0 (butyric acid) and medium-chain fatty acids (MCFA) such as C6:0 (caproic acid), C8:0 (caprylic acid), and C10:0 (capric acid) are added to these diets. The rationale of this supplementation is the notion that ghrelin activation requires acetylation of its third residue, serine, with caprylic acid by ghrelin O-acyltransferase [133].

After prolonged feeding of a DIO diet or WD to mice for 30–50 weeks, the animals become severely obese, fat deposition occurs, ectopic fat accumulates in the body and the liver size significantly increases (Figure 8).

**Figure 8.** Diet-induced obesity and high-fat diets. (**A**,**B**) Comparison of mice receiving either a grain-based diet (**A**) or a diet enriched in fat (**B**) for prolonged times. While the body weight of the mice receiving a control diet was 25 g, the mouse fed a diet enriched for the same time was 52 g. (**C**–**F**) In obese animals, excess fat deposition and ectopic fat accumulation in the body occurs (**C**,**E**). In addition, the liver size is much higher in animals that received a diet rich in fat compared to animals at same age fed a grain-based diet (**D**,**F**). Depicted figures in (**A**,**B**) and (**C**–**F**) were kindly provided by Dr. Angela Schippers (Department of Pediatrics, UKA, Aachen, Germany) and Anastasia Asimakopoulou (IFMPEGKC, UKA, Aachen, Germany), respectively.

In cardiovascular research, WDs enriched in cholesterol, cholate, sucrose, and/or saturated fatty acids have also atherogenic effects and are frequently used to induce or accelerate atherosclerosis in mice [134,135].

Diets highly enriched in fructose rapidly induce an early diabetic state in mice [136,137]. In the liver fructose can be converted in several steps to glycerol-3-phosphate and metabolized by de novo lipogenesis to fatty acids, which can then be esterified to triglycerides (Figure 9A). Therefore, the chronic intake of excess dietary fructose leads to increased formation of triglycerides that accumulate in the liver (Figure 9B), insulin resistance and formation of very low density lipoprotein, attributes that are hallmarks in NAFLD [46].

**Figure 9.** Fructose metabolism and consequences of increased fructose uptake. (**A**) The metabolism of fructose is initiated by phosphorylation of fructose to fructose-1-phosphate, which is subsequently hydrolyzed to form dihydroxyacetone phosphate and glyceraldehyde. Glyceraldehyde can also be converted to dihydroxyacetone phosphate or metabolized to glycerol 3-phosphate. Dihydroxyacetone phosphate can be isomerized via glycerol to glyceraldehyde 3-phosphate. Dihydroxyacetone phosphate can be further reduced to glycerol-3-phosphate or converted into glyceraldehyde 3-phosphate, and subsequently sequentially to phosphoenolpyruvate, pyruvate, and lactate. Pyruvate is central in feeding the citric acid cycle by transferring acetyl groups to coenzyme A, which is essential for the generation of fatty acids. Fatty acids can be esterified to glycerol-3-phosphate to generate triglycerides. Compound images were prepared with the Jmol program (www.jmol.org), version 14.2.15 using the compound identification (CID) nos. 5984, 65246, 751, 753, 1005, 754, 668, 754, 107735, 444493, 91435, 985, and 11147, respectively, (**B**) feeding of a fructose-enriched diet for 4–8 weeks results in progressive accumulation of hepatic fat in mice, which become evident in Oil Red O stain. Interestingly, the fat deposition is higher in female mice than in male littermates. More details about the biological effects and pathomechanism of fructose-induced fatty liver disease can be found elsewhere [46,47].

Interestingly, fructose-induced steatosis and damage induced by feeding a diet enriched in 60% fructose for four to eight weeks was more severe in female than in male mice, suggesting that respective diets provoke gender-specific differences during progression of disease [47]. Feeding of fructose in combination with fat and cholesterol for four days was already sufficient to induce hepatic triglyceride accumulation demonstrating that individual "unhealthy" compounds within a diet can be additive or synergistic [138]. Moreover, the feeding of fructose (60%) for four to eight weeks provoked impairment of olfactory epithelium, resulting in reduced olfactory behavioral capacities [139].

Other diets are characterized by the lack of essential components. In the MCD, sulfur-containing supplements are missing that cannot be synthesized de novo. When missing the essential amino acid methionine, S-adenosylmethionine (SAM or AdoMet) representing a common co-substrate involved in transmethylation, transsufuration, aminopropylation that further blunts inflammatory reactions, cannot be synthesized [140]. The lack of this compound results in lower quantities of cysteine, lecithin, phosphatidylcholine and many other macromolecules (Figure 10A) provoking significant fat accumulation and fibrosis progression in liver (Figure 10B).

Chronic shortage in methionine is, therefore, associated with a progressive physiopathology characterized by increased oxidative stress, hepatic upregulation of pro-inflammatory and pro-fibrogenic genes, liver damage as indicated by increased levels of aminotransferases, and manifestation of other NASH-associated symptoms [13]. Similarly, a shortage in choline, which is an integral part of phosphatidylcholine, sphingomyelin, and acetylcholine, results in significant intrahepatic lipid accumulation through a decreased production of very low density lipoproteins (VLDL), down-regulation of key enzymes involved in triglyceride synthesis, and impaired *de novo* lipogenesis [13]. As a consequence, harmful reactive oxygen species (ROS) are generated and the inefficient β-oxidation causes ballooning of hepatocytes, diffuse necrosis, and hepatic fibrogenesis, and on long-term liver cancer [13,141].

Cholesterol-enriched diets are widely used in studies investigating aspects of the metabolic syndrome. When mice were fed with a high (1%) cholesterol diet for 12 weeks, animals developed hyperlipidemia, hyperinsulinemia, and showed hepatocyte hypertrophy with extensive intracellular accumulation of lipid vacuoles and droplets [142]. It is suggested that in atherogenic diets, which are enriched for example in cholesterol and cholic acid, cholesterol is the key component driving oxidative stress resulting in steatohepatitis and insulin resistance [143]. In addition, these diets induced immune-related responses that may be related to liver damage in 12 inbred mouse strains tested [144].

In sum, these examples demonstrate that "unhealthy" diets enriched in or lacking of ingredients usually part of a balanced diet are suitable to provoke hepatic damage. Therefore, these diets are most popular in biomedical research to investigate mechanisms of initiation and progression of liver disease. However, many of these studies draw conclusions by comparing health aspects of animals fed a grain-based diet with a purified diet such as HFD. However, the effects of the dietary fat will be confounded with the effects of other components that differ between the diets. This fact has been already critically highlighted twelve years ago in a thought-provoking commentary in which 35 studies published in five prestigious high-impact journals were critically evaluated in regard to their performance [145] and this trend has continued as demonstrated by a more recent survey of a larger sampling of the same journals [11]. This exemplarily illustrates the fact that it is critical to draw conclusions when comparing dietary effects obtained in animals receiving either "grain-based diets" or "purified diets". Although diets are normally produced in fixed formulation, minor differences might also result when comparing findings obtained with diets produced by different companies. However, these variations should be relatively negligible.

**Figure 10.** Choline and methionine are essential dietary supplements. (**A**) Methionine is an essential sulfur-containing amino acid that is part S-adenosylmethionine (SAM), which is indispensable as a methyl group donor in pathways driving synthesis of nucleic acids, proteins, lipids, and secondary metabolites. Choline is an integral part of phosphatidylcholines, sphingomyelins and necessary precursor for the synthesis of the neurotransmitter acetylcholine. Choline is necessary for production of very low density lipoproteins (VLDL), down-regulation of key enzymes involved in triglyceride synthesis, and proper function of de novo lipogenesis. Compound images were prepared with the Jmol program using the compound identification (CID) nos. 305, 6137, and 34755. (**B**) Mice fed a methionine-choline deficient (MCD) diet for four weeks develop severe hepatic liver damage, steatosis, ballooning, lobular inflammation, and fibrosis. In hematoxylin eosin (H & E) stain, the architectural changes are visible. In Oil Red O stain, the increased fat accumulation during the diet is assessable, while the Sirius Red stain is suitable to demonstrate increased deposition of collagens. Space bars correspond to 100 μm (H & E) or 50 μm (Oil Red O, Sirius Red). More details about the biological effects and pathomechanism of MCD diet-induced fatty liver disease can be found elsewhere [13].

#### **9. Special Ingredients**

For some studies, mouse diets are fortified with special ingredients (Figure 11). Since the mid-1990s many genetically modified mice were developed, in which the transgene is directed under the control of a tetracycline (Tet)-dependent regulatory system [146]. In these "Tet-on" or "Tet-off" systems, doxycycline is preferable as an inducer in these systems due to its high biological potency,

excellent tissue penetration, and its widespread availability [146]. This compound is rather stable in food products and its concentration is not significantly influenced by storage at room temperature or by exposure to light [146].

**Figure 11.** Some special ingredients in mouse diets used in biomedical research. Doxycycline, tamoxifen, genistein, daidzein, cholesterol, myo-inositol are compounds that are added to specific diets. Compound images were prepared with the Jmol program using the compound identification (CID) nos. 54671203, 2733526, 5280961, 5281708, 5997, and 892, respectively.

In other genetically-modified mouse systems, proteins are expressed as fusions with a modified estrogen receptor ligand binding domain. In these systems, the binding of this moiety to tamoxifen results in a conformational change that allows the fusion protein to translocate to the nucleus. The nuclear translocation of a dominant active transcription factor induces transcriptional activation of susceptible genes, while the expression of a dominant negative receptor fusion might provoke silencing of respective genes. Using this concept, also several inducible tamoxifen-dependent Cre recombinases (from English causes recombination) such as the estrogen receptor-dependent recombinase (CreER recombinases) were cloned that are widely used in biomedical research. They can direct the excision of LoxP-flanked DNA to generate genome modifications in mice [147]. However, using these models and respective diets enriched in tamoxifen, it should be noticed that this drug could influence locomotor activity, social interaction and anxiety in mice requiring critical planning of experimental design [148].

The isoflavone genistein is a phytoestrogen with antioxidant activity targeting numerous intracellular targets leading to retardation of atherogenic activity, possessing suppressive effects on both the cell-mediated and humoral components of the adaptive immune system, and inhibiting cancer progression by inducing apoptosis or inhibiting proliferation [149]. On the molecular level, it was shown that this compound inhibits a large number of enzymes, including adenosine triphosphate (ATP)-utilizing enzymes such as tyrosine-specific protein kinases, topoisomerase II and enzymes involved in phosphatidylinositol turnover [150]. Moreover, this substance has anti-angiogenic effects, modulates estrogen activity, and impacts DNA methylation and/or chromatin modification [151]. Based on this complex repertoire of activities, many mice studies have been performed with genistein-enriched diets. In one study, in which aspects of energy expenditure were analyzed in obese mice, 600 mg genistein/kg diet fed for a period of four weeks resulted in significantly increased food consumption without affecting body weight [152]. In a WD, the supplementation of 1.5 g genistein/kg diet decreased mouse food intake, body weight, and improved glucose metabolism [153]. In a study analyzing the impact of genistein on DNA methylation a concentration of 300 mg genistein/kg diet for four weeks was applied [154]. In the respective investigation, it was shown that male mice fed

a casein-based diet containing genistein showed significantly higher DNA methylation in prostate than control male mice [154].

Similar to the biological effects of genistein on energy expenditure, diets enriched with a related isoflavone daidzein-rich isoflavone aglycone extract at 0.6% of the diet was able to reduce HFD induced body weight gain via reduced hepatic production of triglycerides and subsequent reduction of adipose tissue mass [155]. It was, therefore, suggested that the beneficial effects of daidzein and genistein on food intake and body weight gain in mice are mediated by alterations within the liver X receptor (LXR) signaling pathway [153].

The sugar myo-inositol representing one of nine distinct stereoisomers of inositol is a structural component of many second messengers and lipids such as phosphatidylinositol and its derivatives. Interestingly, when chronically given, this substance has insulin-sensitizing potential in mice provoking a significant decrease in white adipose tissue [156]. Diets enriched in myo-inositol at 2.64 g myo-inositol/kg diet also showed potent reduction in the number, size, and stage of lesions in cancer-prone transgenic mice through triggering alterations in macrophage recruitment and phenotype switching [157].

These examples show that special diets have become an essential research tool in biomedical research. The companies specialized in the production of rodent diets offer limitless custom formulated diets for virtually each application. The fields of applications are numerous, including addition of special fibers, fat, or sugars to modulate the murine microbiome, incorporation of metals or environmental toxins such as microplastic to test their toxicity, feeding of "drug diets" to test the safety of compounds, and many others. The mentioned representative examples of special ingredients demonstrate the high diversity that is possible to modulate diets for a specific purpose. Custom research diets spiked with substances or lacking essential compounds are one critical puzzle piece of biomedical research that help to unravel individual risk factors contributing to disease formation. Usually these diets are produced in small quantities are formulated after consultation with the manufacturers.

#### **10. Diet Coloring**

Food colorants can be divided into three groups: (i) Naturally-derived colors used for food coloring may originate from crushed insects (e.g., carmine), saffron, turmeric, carrot, beet or their color-making ingredients, such as riboflavin and β-carotene. These can be extracted with or without intermediate of final change of identity from biological sources. The addition of these colors to foodstuffs only needs to be approved in the country in which the product is sold or manufactured. (ii) In addition, several mineral or synthetic inorganic colors such as iron oxide, titanium dioxide, chromium oxides are certified as natural food colorings, or for use in drugs, cosmetics, medical devices, or animal food. In particular iron oxides black, red and yellow are intended to be used as colorings and restore color to animal feeding stuffs at a recommended concentration between 500 and 1200 mg/kg without posing a risk to the environment [158]. Since these dyes are highly stable and excreted essentially unchanged in the faces of the animal, they are also considered as safe additives. In many cases, they can be added without requiring to be certified by regulatory bodies when applied in amounts not exceeding preset maximum concentrations. (iii) On the contrary, artificial food colors also categorized as synthetic food dyes or certified color additives are dyes produced by chemical synthesis. In the US, these synthetic compounds must be approved for their usage by the FDA. Once approved, these food dyes are typically named by Federal Food, Drug and Cosmetic Act (FD&C) numbers, while in the European Union and Switzerland certified color additives are classified with European (E) numbers. In some cases, these synthetic dyes are classified as "coal-tar colors" because they were originally produced from petroleum or coal. Actually, the usage of seven colorings and their lakes are permitted in food products in the US (Table 8, Figure 12).



\* Acceptable daily intake (ADI) values are given for humans and were taken from the Joint FAO/WHO Expert Committee on Food Additives (JECFA, https://www.who.int/foodsafety/areas\_work/chemical-risks/jecfa/en/) or from the Internationally Peer Reviewed Chemical Safety Information (http://www.inchem.org). Abbreviations used are: bw, body weight; \*\* Dye allowed by the FDA for limited applications. \*\*\* The usage of this dye in food products is forbidden in the EU.

**Figure 12.** Artificial coloring of mouse diets. The artificial dyes Brilliant Blue, Indigotine, Fast Green, Erythrosine, Allura Red, Tatrazine, Citrus Red 2, Sunset Yellow, and Orange B or using their lakes has been permitted by the US Food and Drug Administration to color food products. These dyes are also approved for mouse diets. Compound images were prepared with the Jmol program using the compound identification (CID) nos. 19700, 2723854, 16887, 12961638, 33258, 164825, 22830, 17730, and 11685735, respectively.

Usually the dyes used in rodent diets are applied as ionic salts rendered partially insoluble by interaction with a metal such as calcium or aluminum. The FDA defines these water-soluble dyes as "lakes", in which the proportions of dye to metal are more or less fixed and given in percentages (e.g., FD&C Red No. 40, aluminum lake, 36–42%).

Commonly, the dyes for coloring of food are used to modify the appeal of food for humans [159]. However, in comparison to humans, mice do not have the visual ability to distinguish the abundance

of colors [160]. Instead, the uptake of food by a mouse is strongly dependent on smell perception and olfactory system that are extremely important for controlling energy homeostasis [161]. The coloring of mouse diets is, therefore, in principle of no importance for the animals. However, the coloring has some useful properties. They allow the researcher to distinguish one diet from another and they ensure that contamination and transmissions to other diet products during the production pipelines can be recognized. In the following, we will give some short information about the artificial colors used for food coloring.

Brilliant blue (FD&C Blue No. 1, E133) is a reddish-blue triarylmethane water-soluble dye used as a blue colorant. It has reasonable stability when exposed to light, heat and acidic conditions, but it has overall low oxidative stability [162]. This dye is considered harmless and most of the dye is excreted undigested [163]. There is no evidence that this dye in rats or mice is carcinogenic, or neurotoxic [164]. However, this dye can act as purinergic inhibitor without pharmacological selectivity, thereby modulating some organ and tissue functions [162]. In addition, a recent report has demonstrated that this dye showed significant greater absorption in septic patients with reduced intestinal barrier function [165]. The daily maximum FDA-approved uptake of Brilliant blue for humans is 12.5 mg/kg body weight/day, while the EU scientific committee suggested an acceptable daily intake (ADI) of 10 mg/kg body weight/day [162].

Indigotine (FD&C Blue No. 2, E132) or indigo is a dark blue water-insoluble anionic pyrrole-based dye originally isolated from the leaves of certain tropical plants. Nowadays this dye is one of the most used coloring agents in the textile industry and synthesized by various methods [166]. In toxicity studies, indigo carmine, representing an organic water-soluble salt derived from indigo by sulflonation, showed no genotoxicity, developmental toxicity or modification of hematological parameters. An ADI up to 5 mg/kg body weight is presently considered as harmless [167]. Moreover, in traditional Chinese medicine, indigo as "Qing-Dai" alone or in combination with other compounds is used as an overall safe and effective drug for treatment of sun stroke, convulsions associated with epilepsy, cough, chest pain, hemoptysis, and phlegm and childrens convulsions [167]. In rodents, the majority of this dye is not absorbed, but readily broken down in the gastrointestinal tract to 5-sulfoanthranilic acid that is absorbed and excreted mostly in the urine [164].

Fast Green FCF (FD&C Green No. 3, E143) is a FDA-approved triphenylmethane dye, while its usage as a food dye is prohibited in the EU [168]. When administered orally 200 mg of this dye to rats, the dye was excreted unchanged in the faeces and no dye was found in the urine [169]. Mice fed diets containing up to 2% Fast Green FCF for 78 weeks, showed no lesions attributed to feeding of the color [170]. The estimate of temporary ADI for man is set up to 12.5 mg/kg body weight [171].

Erythrosine (FD&C Red No. 3, E127) is a cherry-pink poly-iodinated xanthene used as artificial red colorant in foods, drugs and cosmetics [172]. In the past, this dye was commonly used in many countries but is less commonly used in the US, where it is most often replaced by Allura Red AC (FD&C Red No. 40). Chronic toxicity and carcinogenicity studies performed in rats and mice revealed an increased incidence of thyroid follicular cell hyperplasia and adenomas in animals that received 4% erythrosine in the diet for 30 months following in utnero exposure [173]. However, this dye is non-mutagenic and thus the observed tumorigenic activity is most likely not the result of genotoxic initiation [174]. However, the dye has negative effects on thyroid function and therefore the temporary ADI is only in the range of 0–0.05 mg/kg body weight [175].

Allura Red AC (FD&C Red No. 40, E129) is a highly popular red azo dye that may cause allergic reaction such as urticaria or asthma, especially when administered together with orther synthetic color additives [167]. However, in general this dye at 0–7 mg/kg of body weight per day is considered as safe [167]. In the US population, Allura Red AC belongs to the three highest cumulative eaters-only exposures of FD&C color additives in food products [176,177]. Although the European food safety authority expressed concerns about the usage of Allura Red AC as a food color additive, the dye has no genotoxic activity in different test systems [178].

Tartrazine (FD&C Yellow No. 5, E102) is a water-soluble yellow monoazo dye used all over the world for food coloring. In a community-based, double-blinded, placebo-controlled food study, this dye in a mix with other artificial color additives provoked increased hyperactivity in young children [179]. However, the compound has an overall low toxicity with and LD50 value of greater than 2 g/kg body weight and an ADI for humans of 0–7.5 mg/kg body weight was established [177]. Similarly, in mice high-dose level in excess of this ADI were shown to produce only a few adverse effects in neurobehavioural parameters during the lactation period that were however unlikely produce any adverse effects in humans [180].

Sunset Yellow FCF (FD&C Yellow No. 6, E110) is an orange azo dye supposed to have no carcinogenicity, genotoxicity, or developmental toxicity in mice [181]. According to the WHO/FAO guidelines, the ADI was increased in year 2014 from 0–1 mg/kg body weight per day to 0–4 mg/kg body weight [182]. When high content of Sunset Yellow FCF (up to 5% for 23 months) were fed to mice, the mortality rate was not significantly different than in mice receiving no dye and the histopathological changes in organ and tissue observed were considered unrelated to the dietary administration of Sunset Yellow FCF [182].

Citrus Red (Citrus Red 2, E121) is an yellow to orange dye. Testings in mice showed that the feeding of diets containing 3% Citurs Red 2 caused increased morbidity and mortality in both sexes [183]. Based on a number of similar reports suggesting that Citrus Red 2 has carcinogenic effects, the FDA approved this dye only for limited applications such as coloring the peel of oranges, while in the EU it is not permitted at all [168]. However, there it was recommended that this dye should not be used as a food additive [184]. Therefore, this dye should not be incorporated into rodent diets.

Similarly, Acid Orange 137 (Orange B) is approved in the US for use in small traces (150 ppm) only in Frankfurter and sausage casings, while it is forbidden as a food additive in the EU [164,184]. Structurally, it is a pyrazolone dye that is reduced in the gut to form naphthionic acid [184]. In rodents this dye induces lymphoid atrophy of the spleen, bile-duct proliferation, and moderate chronic nephritis when applied for long-term [184]. Therefore, the usage of this dye as a food additive is forbidden in the EU [168].

In sum, the FDA has approved seven synthetic color additives for usage in food products. Three of them (Fast Green FCF, Citrus Red No. 2, and Orange B) are not permitted in the EU as food additives. Although the impact of artificial dyes on mice has not been intensively investigated, there are some studies showing that individual artificial dyes or combinations thereof might be neurotoxic when applied in high concentration [185]. However, typically the concentration necessary to induce adverse effects in male and female mice are extremely high. Brilliant blue FCF for examples showed no adverse effects even at high dietary concentration (7354 mg/kg/day and 8966 mg/kg/day) in male and female mice for 104 weeks [186]. These concentrations are far beyond the concentrations that are used for diet coloring of mice research diets.

#### **11. Diversity of Diet Ingredients may Confound Data Interpretation**

As discussed above, many papers using nutritional models in mice draw conclusions about dietary effects from comparison of grain-based diets with purified diets [145]. However, such mismatched diets potentially hamper the investigator's ability to draw useful conclusions from otherwise well-designed studies [11]. The reproducibility of research findings is adversely affected by the use of improper control diets in metabolic disease research and the lack of adequate diet descriptions in resulting publications [11]. In many publications, grain-based diets referred vaguely as "chow diet", "normal diet" or "control diet" are compared with purified ingredient diets also named as purified diets or semi-purified diets. Grain-based diets are made with grain, cereal ingredients, and animal by-products that may be somewhat variable from formulation to formulation, while purified diets are composed of highly refined ingredients [11]. Given these inherent differences between these diets, data produced from them should not be compared to each other or matched to one factor specifically different (e.g., fat or sugar composition). In particular, soluble fibers fermented

by bacteria in the gut to SCFAs can for example change the gut pH, absorption of bile acids, and chelation of minerals [11]. Exemplarily, Chassaing and coworkers have demonstrated impressively that mismatched diets can result to erroneous conclusions [187]. In their study, the authors investigated the extent to which HFD-induced adiposity is driven by fat content vs. other factors that differentiate purified HFD, grain-based diet, and compositionally-defined diets (i.e., purified diets). Interestingly, the study revealed that high-fat content and lack of soluble fiber are both acting as obesogenic factors promoting rapid and marked loss of cecal and colonic mass and increased adiposity [187]. Therefore, the diet with its ingredients has to be considered as a key environmental factor that critically affects the outcome of a specific experiment. Properly matched "control diets" are therefore an indispensable prerequisite to draw conclusions regarding diet-driven phenotypic differences in respective studies.

#### **12. Conclusions**

Nutritional factors are crucial in laboratory animal science. To guarantee reproducibility of mouse experiments, it is necessary that they receive reliable food with constant composition. There are many providers that have concentrated on the production of grain-based and purified diets. Some are certified and produce their products according to national and international guidelines. The ingredients used are analyzed extensively and the production process guarantees nutrient stability and purity. Besides modification of the different ingredients of a diet, diets can be produced in varying shape, grains and colors using approved non-toxic food dyes. γ-rays and pasteurization are frequently used to sterilize diets fed in SPF facilities. However, these treatments might result in vitamin loss and formation of toxic substances, including acrylamide and peroxide radicals. An important but largely underestimated problem in conducting animal experimentation relying on nutritional models is the impact of confounding factors when choosing unsuitable control diets. These factors might impact the outcome of a specific experiment and incorrect conclusions. Confounding factors in this context are all ingredients differing between the control diet and the intervention diet. Likewise, the coloring with dyes such as Erythrosine and Tartrazine that have already shown to have biological effects in mice or humans should be ommitted for diet coloring. If an investigator has special requirements or wishes for his dietary interventions, it is urgently advisable to contact the manufacturer of the diet product before starting an animal experiment. In most cases, the manufacturers of diets offer consultations with expert nutritionists to assist the scientists in the selection of the right diet for the planned study requirement.

**Funding:** This research was funded by the German Research Foundation (SFB/TRR57) and the Interdisciplinary Centre for Clinical Research within the Faculty of Medicine at the RWTH Aachen University (IZKF Aachen, Project O3-1). None of the funding sources exerted influence on the content or decision to submit this review for publication.

**Acknowledgments:** The authors are grateful to Jörg Lesting (Envigo RMS GmbH, Rossdorf, Germany), Matthew Ricci (Research Diets, Inc., New Brunswick, NJ), habil. Annette Schuhmacher (Ssniff Spezialdiäten GmbH, Soest, Germany), Kai Bergner (VOS Schott GmbH, Butzbach, Germany), and the BGS Beta-Gamma-Service GmbH & Co. KG (Wiehl, Germany) for providing photos and information. The staff and representatives of the mentioned companies were extremely friendly, helpful and answered almost every question arising during writing of this review. In addition, we cordially thank our colleagues from the UKA, Angela Schippers and Anastasia Asimakopoulou, for providing images depicted in Figure 8.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **Abbreviations**



#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **Dietary Strategies for Metabolic Syndrome: A Comprehensive Review**

**Sara Castro-Barquero 1,2,3, Ana María Ruiz-León 2,3, Maria Sierra-Pérez 1, Ramon Estruch 1,2,3 and Rosa Casas 1,2,3,\***


Received: 7 September 2020; Accepted: 27 September 2020; Published: 29 September 2020

**Abstract:** Metabolic syndrome is a cluster of metabolic risk factors, characterized by abdominal obesity, dyslipidemia, low levels of high-density lipoprotein cholesterol (HDL-c), hypertension, and insulin resistance. Lifestyle modifications, especially dietary habits, are the main therapeutic strategy for the treatment and management of metabolic syndrome, but the most effective dietary pattern for its management has not been established. Specific dietary modifications, such as improving the quality of the foods or changing macronutrient distribution, showed beneficial effects on metabolic syndrome conditions and individual parameters. On comparing low-fat and restricted diets, the scientific evidence supports the use of the Mediterranean Dietary Approaches to Stop Hypertension (DASH) diet intervention as the new paradigm for metabolic syndrome prevention and treatment. The nutritional distribution and quality of these healthy diets allows health professionals to provide easy-to-follow dietary advice without the need for restricted diets. Nonetheless, energy-restricted dietary patterns and improvements in physical activity are crucial to improve the metabolic disturbances observed in metabolic syndrome patients.

**Keywords:** metabolic syndrome; dietary pattern; Mediterranean diet; plant-based diet; DASH diet; low-carbohydrate diet; high-protein diet; low-fat diet

#### **1. Introduction**

Following unhealthy dietary patterns and sedentary lifestyles has led to a notable increase in the prevalence of overweight and obesity worldwide. Non-communicable chronic diseases (NCDs) related to unhealthy dietary patterns and weight gain have expanded in parallel, being the major cause of morbidity and mortality both in developed and underdeveloped countries [1]. Among NCDs, cardiovascular diseases (CVD) and type 2 diabetes mellitus (T2DM) are public health priorities, not only for their high prevalence and outcomes but also for the huge economic burden imposed on the health system [2,3].

Metabolic syndrome (MetS) is a clinical condition characterized by a clustering of metabolic risk factors, which is defined by the simultaneous occurrence of at least three of the following components: central obesity, dyslipidemia, impaired glucose metabolism, elevated blood pressure (BP), and low levels of high-density lipoprotein cholesterol (HDL-c), according to the consensual definition of the International Diabetes Federation, the American Heart Association, and the National Heart, Lung and Blood Institute [4]. In developed countries, the prevalence of MetS has risen up to 20–25% in the adult

population, and its incidence continues to increase over time [5–8]. In Spain, the prevalence of MetS is currently reaching epidemic proportions, affecting approximately 22.7% of the population, taking into account that its incidence increases with age [5]. In addition, MetS increases the risk of T2DM onset and major cardiovascular events by two-fold and five-fold, respectively, and other chronic disease such as cancer, neurodegenerative diseases, non-alcoholic fatty liver disease, the risk of reproductive, lipid and circulatory disorders, atherosclerosis, and all cause-mortality are also increased [8–13].

Recent evidence has demonstrated the association between the incidence and prevention of MetS and modifiable lifestyle factors, especially dietary habits. Steckhan et al. analyzed the positive effects of different dietary approaches on MetS inflammatory markers [14]. Regarding the prevention of MetS, Godos et al. also conducted a meta-analysis to demonstrate the preventive role of the promotion of healthy dietary patterns to reduce the prevalence of MetS [15]. Furthermore, some sub-studies from the PREDIMED-Plus cohort showed associations between some dietary components of the traditional Mediterranean Diet (MedDiet) and improvement in MetS components [16–20]. The aim of the present review was to analyze the potential benefits of different dietary approaches on MetS status and their use as efficient strategies to prevent and treat MetS and its comorbidities.

#### *Dietary Patterns*

A single-nutrient dietary intervention has several limitations, and dietary advice must be focused on the overall dietary pattern as part of MetS treatment. Recent evidence supports the implementation of healthy food-based dietary interventions instead of calorie or isolated nutrient restriction [1,21] diets. The health benefits, regarding MetS, of dietary macronutrient patterns and different dietary approaches are summarized in Table 1.


**1.**DietarystrategiesandpotentialhealthbenefitsforMetabolicSyndrome



Approaches to Stop lipoprotein cholesterol, HDL-c, glycated hemoglobin, HbA1c;

Hypertension,

 DASH; unsaturated fatty acids, UFAs; body mass index, BMI; diastolic blood pressure, DBP; low-density lipoprotein cholesterol, LDL-c; high-density

monounsaturated

 fatty acids, MUFA.

#### **2. Mediterranean Diet**

The MedDiet refers to the dietary pattern, culture and culinary techniques adhering to countries and populations living in the Mediterranean Sea basin [64]. This dietary pattern has stimulated a great deal of scientific evidence, demonstrating the potential health benefits associated with adherence, and the primary and secondary prevention of many health outcomes, such as CVD, T2DM, and MetS [10,22]. Recent scientific evidence concluded that the MedDiet not only has beneficial effects on health but also has beneficial effects on sustainability and culture [22,65]. Additionally, the MedDiet has been recognized by UNESCO as an Intangible Cultural Heritage of Humanity [66] and the 2015–2020 American Dietary guidelines referred to the MedDiet as an example of a healthy dietary pattern [21]. The MedDiet is a plant-based diet characterized by a high intake of vegetables including leafy green vegetables, fruits, whole-grain cereals, pulses, legumes, nuts, and extra virgin (cold pressed) olive oil (EVOO) as the main source of fat. Moreover, classical recipes are seasoned with sauces such as *sofrito*, whose main ingredients are olive oil, tomato, garlic, onion or leek, rich in phenolic compounds and carotenoids, such as naringenin, hydroxy-tyrosyl, lycopene and β-carotene [67]. Moderate alcohol intake of fermented alcoholic beverages such as red wine, mainly during meals, is also characteristic of the MedDiet, which also comprises a low to moderate intake of fish and poultry, and low consumption of red meat, butter, sweets, pastries and soft drinks [12,23,68]

The traditional MedDiet is a high fat and low-carbohydrate (CH) dietary pattern, which provides a 35–45% of total daily energy intake from fat, about 15% from protein, and 40–45% energy from CH [12,68]. However, the profile of this fat is mainly one of monounsaturated (MU) and polyunsaturated (PU) fatty acids (FA) and the main food sources of total fat intake are EVOO and nuts. EVOO is one of the key foods of the MedDiet and is the main contributor of monounsaturated fatty acids (MUFAs) in MedDiet countries. Oleic acid is the major component of EVOO and many studies have linked MUFA intake to improvements in insulin resistance, one of the main risk factors for MetS, and in blood lipid profile, and a reduction in both systolic and diastolic BP levels [12,24,69]. EVOO is also rich in polyphenols, which present anti-inflammatory and antioxidant effects and contribute to improving the lipid profile and endothelial function [70] Besides the beneficial effects of unsaturated fats, the whole dietary pattern characterized by the high intake of fruits and vegetables together with moderate red wine consumption provides wide nutritional components, such as antioxidant vitamins (vitamin C, E and β-carotene), phytochemicals (such as polyphenols), folates and minerals, which may exert beneficial effects [31,70].

Considering the effects of the MedDiet on MetS, Di Daniele et al. conducted a review addressing the impact of MedDiet adherence on MetS criteria, obesity and adipose tissue dysfunction [10]. The authors reported that prescription of the MedDiet can be used as a possible therapy for MetS, as it prevents the excess of adiposity and obesity-related inflammatory response. Franquesa et al. concluded that there is a strong evidence for the effect of the MedDiet on obesity and on MetS prevention in healthy or high-CVD risk individuals, as well as on the risk of mortality in overweight or obese individuals [22]. As previously cited, a meta-analysis of 12 cross-sectional and prospective cohorts showed that higher adherence to the MedDiet was associated with a 19% lower risk of developing MetS (relative risk (RR): 0.81 (95% confidence interval (CI) 0.71 to 0.92)), and individual components, such as waist circumference and BP, were also improved (RR: 0.82 (95% CI 0.70 to 0.96); RR: 0.87 (95% CI 0.77 to 0.97), respectively) [15]. Several prospective studies observed the same protective effects in Mediterranean and non-Mediterranean countries [25–27]. The CARDIA (Coronary Artery Risk Development in Young Adults) study is a prospective study including 4713 individuals which evaluated the evolution of CVD risk factors in black and white populations in the United States [28]. They observed a lower incidence of MetS in individuals with a higher adherence to the MedDiet (Hazard ratio (HR): 0.67 (95% CI 0.49 to 0.90)) compared to those with lower adherence, showing a linear trend according to the five score categories (*p* for trend = 0.005) [25]. Kesse-Guyot et al. conducted a prospective 6-year follow-up with 3232 subjects in the SU.VI.MAX study to evaluate the association between different MedDiet adherence scores and the incidence of MetS. They found that participants with higher adherence had a 53% lower

risk compared to the lowest tertile of the MedDiet score (odds ratio (OR): 0.47 (95% CI 0.32 to 0.69) and 0.50 (95% CI 0.32 to 0.77 for each MedDiet score) [26]. In addition, MedDiet adherence scores were associated with improvements in some individual criteria for MetS, such as waist-circumference, BP, triglycerides and HDL-c levels [26]. Moreover, lower MetS prevalence was observed in Korean adults with medium to high MedDiet adherence (OR: 0.73 (95% CI 0.56 to 0.96) and 0.64 (95% CI 0.46 to 0.89), respectively) [30].

MedDiet adherence has been inversely associated with the incidence of CVD and mortality, as well as cancer and degenerative diseases [23,71]. In the case of CVD, the MedDiet is associated with clinically meaningful reductions in the risk of developing the main CVD outcomes, including coronary heart disease and stroke [72]. In a prospective cohort study with 25,994 healthy women from the US Women's Health Study, Ahmad et al. observed an inverse association between the highest MedDiet adherence score and the incidence of CVD compared to the lowest score (HR: 0.72 (95% CI 0.61 to 0.86), *p* for trend < 0.001) [73]. Among the health effects observed, MedDiet interventions have shown improvements in body composition by reducing total and segmental fat, which might have an effect on metabolic profile [10]. Furthermore, the MedDiet has contributed to a decrease in the incidence of T2DM and CVD, while lessening severity and associated complications in individuals who have already been diagnosed [12,22,23,29,31]. Due to the health benefits associated with this easy-to follow dietary pattern, the MedDiet should be considered as one of the first treatment strategies for the prevention and management of MetS.

#### **3. DASH Diet**

In 1997, the Dietary Approaches to Stop Hypertension (DASH) diet became a promising strategy for the treatment of high BP [74], and subsequent randomized clinical trials (RCTs) have supported this evidence [32]. This eating pattern promotes vegetables, fruits, whole grains, low- or free-fat dairy products, legumes and nuts intake, while restricting the intake of red and processed meat and sugar-sweetened beverages [74,75]. The DASH diet is characterized by a low-fat content (27% of daily calorie intake from fat), especially saturated fats (6% of energy) and dietary cholesterol (150 mg/d approximately), and reduced sodium content (from 1500 to 2300 mg/day), but it is rich in fiber (>30 g/day), potassium, magnesium and calcium compared to other dietary patterns [55,76]. The DASH diet has proven to be a useful strategy for the treatment of hypertension [32,33,55,77], and several epidemiological studies have associated higher adherence to the DASH diet with a better cardiometabolic profile [34,36–39,78–80]. In a meta-analysis of several cohort studies, Schwingshackl et al. reported that higher adherence to the DASH diet was associated with a significant reduction in the risk of all-cause mortality (RR: 0.78 (95% CI 0.77 to 0.80), the incidence of or mortality by CVD and cancer (RR: 0.78 (95% CI 0.76 to 0.80); RR: 0.84 (95% CI 0.82 to 0.87), respectively) and the incidence of T2DM (RR: 0.82 (95% CI 0.78 to 0.85)) [40].

Regarding the use of DASH diet as an approach for the treatment of hypertension, a recent meta-analysis of 30 RCT with 5545 hypertensive and non-hypertensive participants concluded that the DASH diet together with lifestyle interventions significantly decreased systolic and diastolic BP measurements compared with a control diet (mean differences: −3.2 mm Hg (95% CI −4.2 to −2.3) and −2.5 mm Hg (95% CI −3.5 to −1.5), respectively) [32]. This effect was more pronounced when sodium intake was lower than 2400 mg/d, in subjects under the age of 50, and in participants with hypertension but without antihypertensive medication [32]. Moreover, on comparing the antihypertensive effects of the DASH diet with 13 other eating patterns (including low-fat diet, Nordic diet, MedDiet, Paleolithic diet and low-sodium diet), the DASH diet was the most effective, especially in comparison with low-fat diets [33]. In contrast, Ge et al. identified the Paleolithic and Atkins diets as the most effective dietary patterns for both systolic and diastolic BP management after six months of intervention compared to usual dietary advice, although this effect was not observed after one year of intervention [55].

The DASH diet intervention has also shown potential effects against excess body weight and abdominal obesity [35]. Middle-term dietary interventions have shown a significant reduction in body mass index (BMI) (weighted mean difference: −0.42 kg/m2 (95% CI −0.64 to −0.20)) and waist circumference (−1.05 cm (95% CI −1.61 to −0.49)) [35]. Nevertheless, in overweight or obese individuals, DASH dietary approaches showed significant weight loss compared to other dietary patterns (−3.63 kg (95% Credible Interval −2.52 to −4.76)) whereas this weight loss was lower after one year of intervention (−3.08 kg (95% Credible Interval −0.48 to −5.66)) [55].

The results are not consistent in the case of blood lipoproteins [55,77]. Ge et al. did not observe significant differences in HDL-c or low-density lipoprotein-cholesterol (LDL-c) levels after a DASH dietary intervention versus usual diet [55], whereas in a meta-analysis of 1917 participants with some CVD risk factors, Siervo et al. observed a reduction in total cholesterol and LDL-c levels after the DASH intervention (mean differences: −0.20 mmol/L (95% CI −0.31 to −0.10) and −0.10 mmol/L (95% CI −0.20 to 0.01), respectively), but reported no significant differences in HDL-c and triglyceride levels [77]. Similar results were obtained in a recently published controlled trial in 80 T2DM patients after 12 weeks following the DASH diet compared to an antidiabetic diet based on American Diabetes Association guidelines [81]. Both dietary interventions significantly reduced triglycerides, total cholesterol and very-low-density lipoproteins.

Epidemiological evidence suggests an association between higher adherence to the DASH diet and a better cardiometabolic profile and lower risk of CVD [36–39]. A cross-sectional study of 1493 adults showed that higher adherence to the DASH diet was associated with 48% less risk of developing MetS, whereas BMI, waist circumference, pro-inflammatory markers and adiposity measures were significantly lower compared to individuals with lower adherence [34]. Interestingly, Ashari et al. observed that higher adherence to the DASH diet was associated with a 64% lower risk of MetS in 425 healthy children and adolescents from 6–18 years of age [82]. In addition, the authors also observed inverse associations among adherence to the DASH diet and BP, fasting plasma glucose levels and abdominal obesity [82]. In this sense, adaptation of the DASH diet to type 1 diabetes glucose requirements (a reduction in CH of around 10% and 15% increase in fat content) resulted in better glucose control and improved the quality of the whole diet, showing a higher intake of fruits, vegetables, fiber and protein compared to the usual intake [83].

The health benefits associated with the DASH diet are probably due to its nutritional quality and distribution. The DASH diet is rich in vegetables and fruits, which translate into high potassium, magnesium and fiber intake, and these nutrients have shown to have a role in BP control, glucose metabolism and insulin response [84]. Furthermore, vegetables and fruits are the main food source of antioxidants and polyphenols, which have been linked to better glucose and insulin blood levels [84]. Moreover, it is limited in sodium and fat, mainly saturated fatty acids (SFA), which are closely related to CVD [84]. Nonetheless, Pickering et al. suggested that the potential health effects of the DASH diet are dependent on eating pattern adherence, with subjects with lower adherence to the DASH diet showing greater benefit from DASH dietary interventions in BP control than those with higher adherence before the dietary intervention [85]. Nonetheless, the commitment and implication of the patient are critical in all life-style interventions based on dietary modifications [86,87].

#### **4. Plant-Based Diets**

Plant-based diets include a wide variety of dietary patterns, which are characterized by a reduction or restriction in animal-derived food intake and the promotion of plant-source food intake, such as fruits, vegetables, nuts, legumes, and grains. Among plant-based diets, strict vegetarian diets, also known as vegan diets, are defined by the exclusion of all animal-derived products, including dairy products, eggs and honey; lacto-vegetarian diets restrict animal food intake except for dairy products; lacto-ovo-vegetarian diets exclude meat, seafood and poultry but include eggs and dairy products; and pesco-vegetarians or pescatarians are similar to lacto-ovo-vegetarian but include fish [88]. Despite the fact that plant-based diets are defined by the exclusion of some or all animal products, recent evidence defines plant-based diets as dietary patterns that promote a reduction in animal-source food intake along with an increase in plant-based food intake, such as the MedDiet [41,88–90].

Plant-based diets have consistently been associated with beneficial cardiometabolic effects, specifically with a lower risk of developing MetS and all of its components [91]. Moreover, these dietary patterns are associated with decreased all-cause mortality and a decreased risk of obesity, T2DM and CVD [43,47,48]. Some studies have found a lower risk of mortality from ischemic heart disease in vegetarians compared with non-vegetarians [43]. Additionally, recent systematic reviews and meta-analyses found significant associations between adherence to the MedDiet and DASH diets and a 38% and 20% lower risk of CVD, respectively, while a 28% reduction in the risk of coronary heart disease was observed following a vegetarian diet [46].

Regarding BP, a meta-analysis of seven RCTs reported a mean reduction of 4.8 mmHg in systolic BP (95% CI −3.1 to −6.6; *p* < 0.001) and a 2.2 mmHg reduction in diastolic BP (95% CI −1.0 to −3.5; *p* < 0.001) in participants following a vegetarian diet compared to an omnivorous diet [42]. These results were confirmed by the same authors in a meta-analysis of 32 observational studies including 604 participants, in which an association was observed between vegetarian diets and reductions in systolic and diastolic BP (−6.9 mmHg (95% CI −9.1 to −4.7; *p* < 0.01) and −4.7 mmHg (95% CI −6.3 to −3.1; *p* < 0.01), respectively) [41,42].

The effects of plant-based diets on blood lipid concentrations are controversial. Wang et al. conducted a meta-analysis of 11 RCTs to evaluate the effects of vegetarian diet on triglycerides, LDL-c, HDL-c and non-HDL-c levels [92]. Total cholesterol levels, LDL-c and HDL-c, were significantly reduced after following a vegetarian diet compared to an omnivorous control diet (0.36 mmol/L (95% CI 0.55 to 0.17; *p* < 0.001), 0.34 mmol/L (95% CI 0.57 to 0.11; *p* < 0.001) and 0.10 mmol/L (95% CI 0.14 to 0.06; *p* < 0.001), respectively). No significant effects were observed for triglyceride levels. This study also described a significant weight-loss in participants who followed the vegetarian compared to the omnivorous diet (−2.88 kg (95% CI −3.56 to −2.20; *p* < 0.001)). Similar results were observed in another meta-analysis of 12 RCTs involving 1151 individuals, in which subjects randomized to the vegetarian diet intervention group showed significant weight loss compared to the non-vegetarian group (mean difference −2.02 kg (95% CI −2.80 to −1.23; *p* < 0.001)) [44]. Other studies assessing plant-based dietary patterns, such as the MedDiet, have also described positive effects on body weight and waist circumference [45].

The health benefits observed are mainly explained by the nutritional quality of plant-based diets as they promote the intake of a wide variety of plant-based foods while cutting down the intake of animal-derived products, such as red and processed meat, which have been associated with a higher risk of developing T2DM, CVD and certain types of cancer [93]. However, it is important to consider that the term "plant-based" does not necessary mean "healthy", as there is evidence supporting adverse health effects of the excessive intake of some plant-derived foods, such as refined grains, snacks, pastries or sugar-sweetened beverages [41,88,94]. A healthy plant-based diet promotes the intake of whole grains, fruits, vegetables, legumes, and non-hydrogenated vegetable oils, such as EVOO. Thus, plant-based diets have low-energy density and high fiber content, which may contribute to CVD prevention, weight loss and long-term body weight maintenance [41,44,88]. Moreover, the profile of fat is mainly MUFA and polyunsaturated fatty acids (PUFA), while SFA intake is lower compared to other dietary patterns. Replacing SFA by MUFA and PUFA has been linked with anti-inflammatory effects and improvements in insulin sensitivity [88]. Among plant-derived foods, the antioxidant effect exerted by several nutrients and bioactive compounds such as vitamin C and E, β-carotenes and polyphenols has been linked to the prevention of CVD and MetS [41,88,95]. Finally, the replacement of some animal-derived foods implies intake restriction of the harmful components mainly present in red and processed meat, such as excessive sodium, heme iron, nitrates and nitrites, which have been linked to CVD outcomes [41,88,96].

In conclusion, recent evidence has demonstrated the protective effect of plant-based diets against MetS, CVD and their individual risk factors. However, healthy plant-derived food choices are crucial to ensure these beneficial effects. Thus, dietary guidelines should consider healthy plant-based dietary patterns as a potential dietary strategy for the prevention and treatment of MetS.

#### **5. Low-Carbohydrate Diet**

Low-CH dietary patterns are characterized by a reduction of total CH intake (<50% of daily calorie intake from CH). This type of diet implies a restriction in the intake of several ultra-processed foods, refined grains, starches and foods rich in simple or added sugars [1]. The association between CH intake and the prevalence and management of the MetS is discrepant [97]. In a meta-analysis of 18 studies with 69,554 MetS patients, Lui et al. concluded that the risk of developing MetS was increased in individuals with higher CH intake (2.5% increase in the risk of MetS per 5% energy from CH intake (95% CI 0.4 to 4.8)) [97]. Moreover, some effects on lipid profile were observed in individuals with high CH intake, such as elevated BP, triglycerides and LDL-c and reduced HDL-c levels [98,99]. The mechanisms underlying the health benefits observed in low-CH diets are the avoidance of the rapid absorption associated with some types of CH, such as glucose and refined grains, which leads to an increase in insulin resistance and insulin demand [53,54]. Therefore, in the case of T2DM, recent clinical guidelines do not recommend a specific CH distribution or restriction, and dietary individualization must be prioritized in the treatment and management of this condition [49]. Bazzano et al. conducted a RCT to analyze the effect of a low-CH diet (<40% of total energy intake from CH) compared with a low-fat diet (<30% of total energy from fat, <7% SFA) without energy restriction or physical activity advice in obese adults (BMI 30 to 45 kg/m2) [50]. After 1 year of intervention, subjects on the low-CH diet without energy restriction showed greater weight loss (−3.5 kg (95% CI −5.6 to −1.4 kg)), specifically in fat mass (−1.5% (CI −2.6% to −0.4%)). Moreover, some cardiovascular risk factors were improved in the low-CH group, such as triglycerides, HDL-c and total cholesterol to HDL-c ratio [50]. Regarding the management of T2DM, low-CH compared to low-fat dietary interventions (<30% of total energy from fat) showed higher reductions of body weight, glycosylated hemoglobin (HbA1c), triglycerides and BP levels and increased HDL-c concentrations and, consequently, a modification in glucose-lowering medications was observed [49,51].

Recent evidence has shown an association between dietary CH intake and the risk of mortality. The Prospective Urban Rural Epidemiology (PURE) study is a cohort study of 135,335 individuals aged 35–70 years from 18 countries from five continents [98]. The aim of this study was to assess the association of dietary fat and CH intake and total mortality and CVD, differentiating this intake according to the profile of FA and CH. The findings of this study suggest the need for an update in dietary guidelines, with emphasis on fat restriction to promote low-fat and CH dietary patterns (around 50–55% of daily energy intake from CH) rich in PUFA and whole-grain CH. Other studies have also observed an association between refined CH intake and a higher risk of cardiovascular events, such as stroke or myocardial infarction [100,101]. However, there is insufficient scientific evidence on low-CH diets (<50–55% of total energy intake) and metabolic improvements have not been demonstrated in order to support or recommend very-low CH diets [98]. Seidelmann et al. observed that with high (>70% of total energy intake) and low (<40%) CH diets the total mortality increased, with 50–55% showing the lowest risk of mortality, representing a U-shaped association [102]. The replacement of CH with other nutrients has shown different effects on total mortality, which was increased in low-CH diets rich in animal-derived fat and/or protein. By definition, low-CH diets promote the restriction of foods rich in CH, such as vegetables, fruits, whole-grain cereals, legumes, etc. Consequently, low-CH diets, compared to diets with 50–55% of energy from CH, showed lower amounts of bioactive compounds such as fiber, PUFAs, polyphenols, vitamins and minerals [103]. Therefore, the use of very low-CH diets as a dietary approach for MetS should promote plant-based fat and/or protein food sources [102].

Among low-CH diets, it has been postulated that very low CH ketogenic diets have a therapeutic role in several NCDs, including overweight and obesity, CVD and MetS [104]. Although there is no standardized definition of the ketogenic diet, it is characterized by a reduction in CH to less than 10% of daily energy intake, which means around 30 to 50 g of CH per day, and a relative increase of fat intake (fat to CH and protein intake ratio of 3:1 to 4:1) [105]. This restrictive dietary pattern has shown protective effects for obesity and CVD by reducing body weight and improving the lipid profile [104,106–108]. The meta-analysis of Bueno et al. observed greater weight loss (weighted mean difference −0.91 kg (95% CI −1.65 to −0.17 kg)), and reduced triglyceride (−0.18 mmol/L (95% CI −0.27 to −0.08)) and diastolic BP levels (−1.43 mmHg (CI −2.49 to −0.37)), while HDL-c levels increased (0.09 mmol/L (95% CI 0.06 to 0.12)) after following a ketogenic diet compared to a low-fat diet [52]. The mechanisms of action underlying these protective effects are as follows: the absence of dietary CH intake leads to a decrease in insulin secretion, which is translated into an inhibition of lipogenesis and fat accumulation and an increase in lipolysis; a satiety effect of protein intake and its effect on appetite control hormones, such as leptin and ghrelin; and the modulation of insulin secretion and ketone body production which might lead to metabolic improvements, especially in insulin signaling [104,109]. Moreover, CH restriction and the glycogen depletion characteristic of this type of diet lead to the use of ketone bodies as the main source of energy. Nevertheless, energy restriction is necessary to maintain ketone body production. Thus, recent evidence suggests that body weight and CVD benefits observed with ketogenic dietary interventions are due to energy restriction, in spite of the macronutrient distribution of the diet [104,110]. However, health care professionals should consider the difficulties in following a ketogenic diet and the absence of healthy foods such as vegetables, fruits and whole-grain cereals, the intake of which is associated with a lower risk of developing chronic diseases such as CVD, T2DM and some types of cancer.

#### **6. Low-Fat Diet**

By definition, the fat content of the low-fat diet comprises less than 30% of total energy, of which <10% are SFA, with a moderate PUFA content and limited *trans* FA [111,112]. In proportion, CH intake is higher and protein intake is moderate (around 15–17% of total energy intake). Low-fat diets usually include foods and products with reduced total fat content, such as low-fat dairy products instead of whole-fat products and derivatives. Low fat diets in weight-loss oriented dietary interventions showed a reduction in the risk of premature mortality in obese adults [56]. In this sense, a meta-analysis of 34 RCTs observed an 18% lower risk of all-cause mortality in weight-loss oriented dietary interventions in obese adults (95% CI 0.71 to 0.95), while no significant effects were observed in CVD mortality or incidence [56]. Recently, a network meta-analysis described the effectiveness of the low-fat diet for body weight reduction compared to the usual dietary advice and dietary patterns after short-term intervention (6 months), with this effect being attenuated after one year [55].

Clinical trials evaluating the effect of a low-fat diet on the prevalence of MetS have shown conflicting results [113–116]. Dietary interventions based on low-fat intake (around 20% of total energy intake from fat) slightly reduced MetS components, but no significant effects were observed for CVD or the incidence of coronary heart disease in postmenopausal women compared to the usual diet [113,114]. Nevertheless, following a low-fat diet was not associated with a lower prevalence of MetS in older subjects at high CVD risk [115]. In this sense, Veum et al. compared the effect of a low-fat, high-CH diet (around 30% of total energy intake from fat) vs. a very high-fat, low-CH diet (around 73% total energy intake from fat) on MetS components [117]. No significant differences were observed in MetS components, body weight and body composition in the medium-term [117]. Similar findings were described by Gardner et al. in the DIETFITS trial, in which both low-fat and low-CH dietary approaches showed significant weight loss with no differences between the two interventions [118].

Regarding BP and blood lipoproteins, a low-fat diet showed beneficial effects on systolic and diastolic BP management, and improved HDL-c and LDL-c levels in the short term compared to usual diet, but these effects were reduced in long-term interventions [33,55]. However, in a meta-analysis of RCTs including 17,230 hypertensive and pre-hypertensive participants, Schwingshackl et al. suggested that the MedDiet and the DASH diet are more effective in long-term BP management compared to low-fat diets [33]. Likewise, low-CH diets showed greater effects on the control of glycated hemoglobin and blood lipid levels than low-fat diets in a short to medium term intervention [52,112,119,120]. In the case of glucose metabolism and insulin control, some RCTs have not identified significant effects of low-fat dietary interventions versus other dietary approaches, while higher triglyceride levels were observed, mostly when simple CH proportion is increased [121–125]. In the case of T2DM management, Basterra-Gortari et al. found that a low-fat diet did not have an effect on glucose-lowering medication management while the MedDiet supplemented with EVOO could delay its requirement in older people at high CVD risk [126].

MetS is associated with a pro-inflammatory state, and it has been proposed that a low-fat dietary intervention inducing weight loss slightly reduces inflammatory biomarkers such as high-sensitive c-reactive protein (CRP), interleukin−6 (IL) and tumor necrosis factor alpha (TNF-α) levels [16,127–129]. These results are inconclusive, and the effects observed depend on weight loss and diet composition, particularly dietary fiber, fruits and vegetables [16,128]. Additionally, some studies observed that low-fat dietary interventions could improve the microbiome dysbiosis linked to MetS by increasing α-diversity [130,131]. However, limited results are available and more evidence regarding long-term response is needed [132]. Furthermore, nutrigenetic interactions have been described between dietary fat content and metabolic response [133].

Based on the evidence available, current dietary guidelines, such as the 2015–2020 American and European Dietary Guidelines, should avoid stating upper limits of total fat intake, mainly from healthy unsaturated FA. Moreover, this recommendation should include not exceeding 10% of total energy intake from SFA and the replacement of SFA by MUFA and PUFA [22,134].

#### **7. High-Protein Diet**

Recent evidence suggests that a high-protein dietary pattern leads to greater weight-loss and CVD improvements than standard protein diets (0.8 g protein/kg body weight). High-protein diets are characterized by a 20–30% of daily energy intake from protein, which means around 1.34 to 1.5 g protein/kg body weight [57]. Currently, the use of high-protein dietary interventions has been postulated for the treatment of obesity, MetS and glycemic control [58,93]. The effect of high-protein dietary strategies for weight management is controversial. A meta-analysis of 18 studies on the effect of a high-protein diet in T2DM patients showed that a high-protein diet did not significantly decrease body weight compared to a regular protein diet [135]. Moreover, no significant effects where observed for glycemic control parameters, such as fasting glucose, insulin and HbA1c, blood lipid profile or BP levels. Nevertheless, a significant reduction of triglyceride levels was observed in participants who followed a high-protein diet. In the case of MetS, high-protein diets with CH restriction have shown effective weight-loss in obese adults with MetS [57,58]. Campos-Nonato et al. performed a RCT in 118 adults with MetS to evaluate the effect of a hypocaloric high-protein diet compared to a hypocaloric standard protein diet (500 kcal/day less than the metabolic rate and 1.34 g protein/kg body weight or 0.8 g protein/kg body weight, respectively) [57]. Weight-loss after 6 months of the dietary interventions was significantly higher in participants who followed a high-protein diet (−7.0 kg ± 3.7; *p*-value = 0.046) compared to the standard protein diet (−5.1 kg ± 3.6; *p*-value = 0.157) [57]. MetS criteria, including fasting blood glucose, insulin, homeostatic model assessment for insulin resistance (HOMA-IR) index, and triglyceride, and cholesterol levels, improved in both intervention arms, but non-significant differences were observed in the comparison between groups. The Optimal Macronutrient Intake Trial to Prevent Heart Disease (OmniHeart) study was a randomized, controlled, three-period, crossover nutritional study with 164 participants with overweight or obesity and prehypertension or stage 1 hypertension free of T2DM [136]. This study aimed to evaluate insulin sensitivity with the quantitative insulin sensitivity check index among three dietary interventions: a high-CH diet (58% of daily kcal from CH; 15% from protein and 27% from fat); a protein diet (replacement of 10% of total CH to protein, 25% of daily kcal intake from protein, mainly from plant-based protein sources); and an unsaturated diet (replacement of 10% of total CH to unsaturated fat, 37% of daily kcal intake from fat, mainly from seeds and oils such as olive, canola and safflower oils and nuts). The protein and high-CH dietary patterns did not affect insulin sensitivity, while the unsaturated diet showed improvements in insulin sensitivity, suggesting that the replacement of CH by unsaturated fat, such as in the MedDiet patterns, are alternative dietary approaches to improve insulin sensitivity. The mechanism underlying the potential health benefits of a high-protein diet is that protein induces satiety, which is translated into

reduced energy intake in the next meals [137,138]. Furthermore, high protein intake avoids muscle mass loss during energy-restrictive dietary interventions for weight loss [139].

Among protein food sources, meat and meat derived products have been associated with a higher risk of developing T2DM, CVD and MetS [12,13,140]. Dietary guidelines recommend prioritizing plant-based protein food sources such as soy, legumes, beans, nuts and seeds instead of meat and processed meat [21]. Plant-based protein food sources are rich in fibre, phenolic compounds and PUFA, while cholesterol, trans or SFA are in lower proportions [141]. In a recent meta-analysis of 36 RCTs, red meat consumption had no effect on the blood lipid profile or BP, while after analyses stratified by the type of comparison diet, the substitution of red meat with plant-based protein foods showed a reduction in total cholesterol and LDL-c levels [93]. Thus, strong evidence promotes the intake of plant-based protein food sources, and this should also be recommended to promote environmental sustainability.

#### **8. Other Dietary Patterns and Strategies**

Other dietary alterations have been shown to improve the MetS condition, such as the Nordic Diet, which is characterized by a high content of whole-grain high-fiber products (such as rye, barley, oat, rice, vegetables, fruits and nuts), with rapeseed oil as the main source of dietary fat and a high intake of fish and shellfish [142,143]. Similar to the DASH and the MedDiet, the Nordic diet is considered to be a healthy dietary pattern in that it promotes the intake of vegetables, fruits, fish, poultry, nuts, and is low in sodium, red meat and processed foods. A recent meta-analysis of 5 RCTs including 513 participants demonstrated the effectiveness of the Nordic diet in improving some MetS criteria, mainly systolic and diastolic BP (weighted mean differences −3.97 mmHg (95% CI −6.40 to −1.54; *p* < 0.001); −2.08 mmHg (95% CI −3.43 to −0.72; *p* = 0.003), respectively) [59]. Moreover, improvements in LDL-c (0.30 mmol/l (95% CI −0.54 to −0.06; *p* = 0.013)), but not in HDL-c and TG levels, were observed compared to control diets [59]. Further studies are needed to evaluate the beneficial effects of the Nordic diet on MetS management and prevention.

Among dietary strategies, intermittent fasting has shown benefits for CVD, T2DM, metabolic disturbances and cancer, mainly because of the daily caloric restriction involved [63]. The main cardiometabolic effects observed after an intermittent-fasting intervention are weight loss and improvements in insulin resistance, dyslipidemia, BP levels and inflammation [60–62]. Despite the evidence and potential health benefits of intermittent fasting, the applicability of this dietary strategy is complex and trained health care providers are needed to avoid side effects. Furthermore, De la Iglesia et al. postulated other potential dietary approaches for the prevention and treatment of MetS, such as diets rich in omega−3 FA, low glycemic index, high antioxidant capacity or high meal frequency dietary interventions [144].

Thus, dietary intervention based on energy restriction, independently of the distribution of macronutrients, might influence BP and CVD. Accordingly, most scientific evidence highlights the relevance of dietary quality rather than quantity, especially in the management and prevention of MetS [8,145–147]. Moreover, the effectiveness of every dietary intervention is associated with the previous metabolic state (e.g., presence of insulin resistance, T2DM, altered fasting glucose levels, etc.) [148,149]. While multifaceted lifestyle interventions focus on weight-loss and the promotion of physical activity, adherence is the key factor in achieving the beneficial effects observed in each dietary pattern, with intervention adherence being decisive in the results observed independently of the type of diet [150–152].

#### **9. Conclusions**

The protective effects of healthy dietary patterns on MetS seem to be due to the sum of small dietary changes rather than the restriction of any single nutrient. On comparing low-fat diets and very-restricted diets, the scientific evidence supports the use of the MedDiet intervention as the new paradigm for MetS prevention and treatment. The nutritional distribution and quality of the MedDiet allows health professionals to provide easy-to-follow dietary advice without the need for a restricted

diet. Nonetheless, RCTs on the effects of a low-CH MedDiet style diet, promoting the intake of whole grain and plant-based protein food sources in patients with MetS, are needed to demonstrate the efficacy of this dietary pattern.

**Author Contributions:** Conceptualization, R.C. and S.C.-B.; writing—original draft preparation, S.C.-B.; A.M.R.-L.; M.S.-P.; writing—review and editing, R.E. and R.C.; supervision, R.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** This work has been partially supported by PIE14/00045, PI16/00381 and PI19/01226 from the Instituto de Salud Carlos III, Spain. CIBER OBN is an initiative of the Instituto de Salud Carlos III, Spain. S.C.-B. thanks the Spanish Ministry of Science Innovation and Universities for the Formación de Profesorado Universitario (FPU17/00785) contract.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Review* **The Legacy E**ff**ect in the Prevention of Cardiovascular Disease**

#### **Esther Viñas Esmel 1, José Naval Álvarez <sup>1</sup> and Emilio Sacanella Meseguer 1,2,\***


Received: 26 September 2020; Accepted: 19 October 2020; Published: 22 October 2020

**Abstract:** The "legacy effect" describes the long-term benefits that may persist for many years after the end of an intervention period, involving different biological processes. The legacy effect in cardiovascular disease (CVD) prevention has been evaluated by a limited number of studies, mostly based on pharmacological interventions, while few manuscripts on dietary interventions have been published. Most of these studies are focused on intensive treatment regimens, whose main goal is to achieve tight control of one or more cardiovascular risk factors. This review aims to summarise the legacy effect-related results obtained in those studies and to determine the existence of this effect in CVD prevention. There is sufficient data to suggest the existence of a legacy effect after intensive intervention on cardiovascular risk factors; however, this effect is not equivalent for all risk factors and could be influenced by patient characteristics, disease duration, and the type of intervention performed. Currently, available evidence suggests that the legacy effect is greater in subjects with moderately-high cardiovascular risk but without CVD, especially in those patients with recent-onset diabetes. However, preventive treatment for CVD should not be discontinued in high-risk subjects, as the level of existing evidence on the legacy effect is low to moderate.

**Keywords:** legacy effect; metabolic memory; cardiovascular disease; diet; diabetes; hypertension; dyslipidaemia

#### **1. Introduction**

Cardiovascular disease (CVD) has emerged as a major cause of morbidity and mortality, accounting for 30% of all deaths worldwide. The intensive management of cardiovascular risk factors is required to reduce its incidence, and some authors suggest that achieving tight control at early stages of the disease, before vascular damage has developed, is a determinant of outcomes [1].

The legacy effect concept refers to long-term sustained benefits after a period of intensive treatment intervention, even after cessation of the intervention [2]. Initially described in diabetic patients, it has also been observed in patients with hypertension or hypercholesterolemia [3]. Moreover, the concept of metabolic memory, mostly described in the study of diabetic models, refers to DNA's ability to store information related to prior poor metabolic control; for example, persistent adverse effects of hyperglycaemia may reduce the potential benefit of subsequent improvements in glucose control, and induce the development of vascular complications in target organs [4,5]. Therefore, achieving good glycaemic control in the early stages of diabetes could be critical in preventing late-stage complications [6,7].

Metabolic memory was first described in 1987 when Engerman et al. observed a higher incidence of diabetic retinopathy among dogs in which good glycaemic control was preceded by a longer period of abnormal glucose levels [8]. Likewise, long-term overproduction of fibronectin, both by endothelial cells cultured in a high-glucose medium and by the kidneys of streptozocin-induced diabetic rats after the restoration of near-normoglycemia, was also detected [9]. Further investigation in 1993 showed that transplanting the islets of Langerhans shortly after diabetes mellitus onset in rats could reverse diabetic retinopathy [10].

However, most studies examining the legacy effect on CVD and its related risk factors for vascular complications are based on pharmacological interventions, with low emphasis on dietary interventions; thus, the long-term effects of strict dietary regimens remains unknown. In this review, we focus on the recent evidence of the legacy effect and metabolic memory in CVD development after intensive pharmacological and non-pharmacological interventions addressing cardiovascular risk factors.

#### **2. Methods**

The aim of this narrative review was to assess those scientific studies that evaluated the existence of the legacy effect in the prevention of CVD after a nutritional or pharmacologic intervention. We searched for scientific studies published in the last ten years and written in English in PubMed Medline Database by using specific search terms ("legacy effect", "metabolic memory", "cardiovascular disease", "diet", "diabetes", "hypertension" and "dyslipidemia"). In addition, recent reviews and meta-analysis about the legacy effect, metabolic memory and its pathophysiological mechanisms were also included. To sensitise the search and select the best articles, we used the aforementioned keywords and applied different Boolean operators. Finally, a total of 64 articles were selected for this review. Most of them were clinical trials in humans, and there was also a small proportion performed in animal models. The differences observed among the outcomes of the trials and their post-trial follow-up studies have been described by using the difference in means, risk ratio (RR), odds ratio (OR) and hazard ratio (HR) measures. The publication, attrition and co-intervention bias, and the baseline characteristics of the patients included in the trials should be taken into consideration for the interpretation of the results.

#### **3. Pathophysiological Mechanisms Involved in Metabolic Memory**

Multiple mechanisms have been described as relevant to the development of the legacy effect. Diabetic models have been used to thoroughly study the concept of metabolic memory. Furthermore, studies performed on hypertension models have shown that similar mechanisms are involved in the maintenance of beneficial post-intervention effects.

#### *3.1. Oxidative Stress*

Chronic hyperglycaemia induces uncoupling of the mitochondrial electron transfer chain in endothelial cells, causing excessive production of superoxide anion and other reactive oxygen species (ROS), thereby increasing cellular oxidative stress and inducing endothelial dysfunction. Moreover, ROS can cross membranes and damage macromolecules (nucleic acids, mitochondrial DNA and proteins), many of which may have longer half-lives, and hence may exert metabolic effects. These oxidative stress markers can persist in endothelial cells despite glucose normalisation after prolonged hyperglycaemia, suggesting a metabolic memory phenomenon (Figure 1) [4,5]. Furthermore, postprandial glucose oscillations in diabetic patients have been associated with increased ROS marker levels and vascular stress; the cells' inability to adapt to this dynamic environment leads to cell damage and elevated collagen and fibronectin levels that remain abnormal for several days after glucose levels have normalised [11–13]. In hypertension models, the overproduction of ROS by NADPH oxidase upregulation due to renin-angiotensin system activation appeared to persist after cessation of the hypertensive period and was related to deleterious effects. Otherwise, the intensive blocking of the renin-angiotensin system showed a long-term reduction in the overproduction of ROS [14].

**Figure 1.** Pathophysiological mechanisms of metabolic memory in chronic hyperglycaemic conditions. Abbreviations: AGE: advanced glycation end-products; DAG-PKC: diacylglycerol-protein kinase C; GAPDH: glyceraldehyde 3-phosphate dehydrogenase; HbA1c: glycated haemoglobin; miRNA: micro RNA; NF-κB: nuclear factor κB; RAGE: AGE receptor; ROS: reactive oxygen species.

#### *3.2. Non-Enzymatic Glycosylation of Proteins and Chronic Inflammation*

Superoxide anion inhibits glyceraldehyde 3-phosphate dehydrogenase (GAPDH) activity, leading to the accumulation of all glycolytic intermediates, which can react non-enzymatically with proteins, lipids, and nucleic acids to form advanced glycation end-products (AGEs). AGEs may accelerate ageing processes, and they cannot be degraded easily by the enzymes involved in normal metabolism. In addition, the most exposed proteins, such as haemoglobin, are also highly glycated [4]. AGEs also promote oxidative stress, vascular hyperpermeability, pathological angiogenesis, and thrombogenic reactions via interaction with their receptors (RAGEs), which promote endothelial dysfunction and atherosclerosis [5,15,16]. Hyperglycaemia also causes the activation of the diacylglycerol (DAG)-protein kinase C (PKC) pathway, fructose production, and increased flux through the hexosamine pathway, which intensifies the expression of various adhesive molecules, proinflammatory cytokines, and growth factors. These mechanisms activate nuclear factor κB (NF-κB), a rapid-response transcription factor involved in inflammatory reactions and proapoptotic programs in diabetes.

Even after glycaemic control has been achieved, AGEs may be used as markers to measure cumulative diabetic exposure, especially in those with vascular complications, whereas intensive treatment is associated with significantly lower AGE levels. These pathways are responsible for metabolic memory and contribute to the development and progression of CVD and diabetic complications. Glycaemic memory can be measured by glycated haemoglobin (HbA1c), although AGEs have also been correlated with retinopathy progression, independent of HbA1c level [16]. Some medications could alter these pathways and, consequently, prevent the development of CVD. Metformin, pioglitazone, glucagon-like peptide-1 (GLP-1) receptor agonists and dipeptidyl peptidase-4 (DPP-4) inhibitors have been shown to prevent AGE formation or decrease inflammation by blocking AGE and NF-κB pathways, reversing the metabolic memory phenomenon [12,16–18]. Also, angiotensin-converting-enzyme inhibitors (ACEI) and angiotensin II receptor type 1 (AT-1) blockers can reduce AGE formation and, subsequently, the inhibition of superoxide generation [17]. It is widely recognised that benefits of renin-angiotensin-aldosterone system (RAAS) inhibition extend beyond blood pressure (BP) reduction and may last even after switching from an intensive BP-lowering strategy to a conventional one; its effectiveness in preventing diabetic complications indirectly suggests that RAAS dysregulation is involved in triggering organ damage in diabetic patients [19].

#### *3.3. Epigenetic Modifications*

Chronic hyperglycaemia can induce epigenetic modifications, which could underlie endothelial dysfunction and the development of metabolic memory in diabetic complications. Recent research suggests that DNA methylation and post-translational histone modifications in a hyperglycaemic environment could lead to enhanced expression of proinflammatory genes, such as the increased expression of the key p65 subunit of NF-κB in various target tissues, which persists long after restoration of normoglycemia. Moreover, in cell and animal models exposed to transient-high and subsequent normal glucose levels, an increase in differentially expressed miRNAs has been observed; these may play a role as regulators of endothelial dysfunction in metabolic memory by increasing the constitutive activation of NF-κB pathway. However, the potential reversal of glucotoxicity and lipotoxicity depends on how long the patient has been exposed to this poor metabolic environment [20,21].

#### **4. Legacy E**ff**ect after Dietary Intervention**

Few studies have assessed the legacy effect after dietary intervention. Most of these have been done in animals. Below, we describe the most relevant findings.

One study showed that mice switched from dietary restriction (DR) to ad libitum (AL) intake had significantly better glucose tolerance at 6–10 months compared to the AL group, suggesting that there was positive glycaemic memory in the DR group [22–24]. Moreover, it was also observed that short-term reversion to a normal diet in rats after initial exposure to a high-fat diet (HFD) could not restore the insulin resistance-induced complications. However, administering metformin to these animals induced remarkable amelioration of anomalies associated with insulin resistance and endothelial dysfunction via lipotoxicity reduction [25,26]. In another study, mice fed a HFD (60% kcal from fat) for 21 weeks were compared to those fed a low-fat diet (LFD, 10% kcal from fat); the HFD animals developed type 2 diabetes mellitus (T2DM) and gained much more weight compared to mice fed an LFD. During the 4-week intervention period following diabetes onset, insulin-treated mice maintained on an HFD exhibited significantly improved glucose tolerance test results compared to sham-treated animals. However, mice remaining on a HFD following cessation of the insulin treatment exhibited no benefit from the early insulin therapy and continued to gain weight, had worse glucose tolerance test results, and displayed significantly higher fasting insulin, C-peptide, and leptin levels compared to sham-untreated controls, and early insulin therapy in mice maintained on the LFD. In fact, early insulin therapy only conferred a beneficial effect in animals switched to an LFD after the insulin treatment. These findings indicated that the legacy effect of early insulin-treatment was only observed

in diabetic mice switched to an LFD after the insulin treatment and emphasised the vital role of diet adherence in diabetes control at any stage of disease progression [27].

The legacy effect was also confirmed in rats subjected to an HFD, when time-restricted feeding was alternated with AL feeding, with no significant differences in body weight observed between the alternated feeding and the control group (AL feeding only) in a short term-study (12 weeks). However, in the long-term study (25 weeks), when mice were maintained on time-restricted feeding for 13 weeks and then switched to AL feeding for 12 weeks, they showed significantly less body weight gain after returning to an HFD compared to the control group, which were maintained on time-restricted feeding throughout the 25 weeks (112% versus 51% body weight increase, respectively) [28]. Further research performed on male rats showed that maintaining an HFD induced persistent changes in sperm cells that could be transmitted to the female descendants, impairing their glucose tolerance and insulin secretion [29].

To our knowledge, there is limited evidence assessing the existence of a legacy effect in humans after dietary intervention. The effects of caloric restriction (approximately 25% of daily energy intake) or intermittent fasting (up to 18 h/day) in humans have been demonstrated mainly in observational studies, and only one clinical trial, the Comprehensive Assessment of the Long-term Effects of Reducing Energy Intake (CALERIE) study, was performed. Both strategies improve general health indicators and slow or reverse ageing and disease processes, such as obesity, insulin resistance, dyslipidaemia, hypertension and inflammation, although intermittent fasting seems to exert a greater effect. Improvements in health indicators typically begin within the first month, after the start of intermittent fasting, and then dissipate over a period of several weeks after resumption of a normal diet [30]. This effect was also evaluated in a small clinical trial of 24 obese older patients (mean age 65–79 years) after caloric restriction (CR) intervention plus resistance training (RT) compared to RT intervention for five months. Despite clinically meaningful weight loss and concurrent favourable shifts in total body and thigh composition in the CR+RT group compared to RT group during the intervention period, these improvements were generally not sustained over the long-term (18 months). At the end of the study, only weight was significantly lower compared to baseline in the CR+RT group. Therefore, this trial showed a temporary legacy effect which lost intensity during the follow-up [31].

The Oslo cardiovascular study was a five year randomised intervention conducted in healthy middle-aged men at high risk of coronary heart disease (CHD) assigned to a dietary advice group (main objectives: to reduce daily intake of saturated fats, sugar and alcohol; to increase daily intake of fish, vegetables and fruit; to reduce weight; to quit smoking) compared to a control group. At 40 years of follow-up, a significant reduction (19%) in the risk of death at first myocardial infarction (MI) was detected in the intervention group (HR 0.71, 95% confidence interval (CI), 0.51–1.00), with no significant difference in total mortality. The legacy effect was achieved through the dietary intervention (as evidenced by lower cholesterol and serum triglyceride levels and weight reduction) because few men quit smoking; therefore, the effect of this measure was small [32].

In the Look Action for Health in Diabetes (Look-AHEAD) trial, overweight/obese patients with T2DM were randomly assigned to an intensive lifestyle intervention to achieve and maintain a weight loss ≥7%, which included: hypocaloric diet (1200–1800 kcal/day; less than 30% of daily energy intake from fat and less than 10% from saturated fat; at least 15% of daily energy intake from protein), physical activity and diabetes education, compared to standard care for one year. Patients included in the intensive lifestyle intervention had greater weight reduction with better control of cardiovascular risk factors (glycaemic control, systolic BP and lipid profile), and more improvement in fitness levels throughout the four year follow up period compared to the control group [33].

#### **5. Legacy E**ff**ect in Diabetic Patients after Intensive Glycaemic Control**

Several clinical trials have assessed legacy effects in different populations of patients with diabetes, the main results of which are listed below. The specific data from each article are shown in Table 1.


**Table 1.** Studies assessing the legacy e ffect after intensive glycaemic control.

mellitus; DCCT: Diabetes Control and Complications Trial; EDIC: Epidemiology of Diabetes Interventions and Complications; UKPDS: United Kingdom Prospective Diabetes Study;ACCORD: Action to Control Cardiovascular Risk in Diabetes; VADT: Veterans Affairs Diabetes Trial; ADDITION: Anglo-Danish-Dutch study of Intensive Treatment in People withScreen-DetectedDiabetesinPrimaryCare.\*Thenumberofrandomisedpatientsenrolledintheinterventionalphaseandthecohortofpost-trialfollow-up,respectively.

#### *Nutrients* **2020**, *12*, 3227

The Diabetes Control and Complications Trial (DCCT) was performed in patients with type 1 diabetes mellitus (T1DM), who were randomly assigned to either intensive or standard glycaemic control regimens. After 6.5 years of intervention, the development of severe proliferative or non-proliferative retinopathy (HR 0.47, 95% CI, 0.14–0.67), clinical neuropathy (HR 0.60, 95% CI, 0.38–0.74), and microalbuminuria (HR 0.39; 95% CI, 21–52) was significantly lower in the intensive treatment group [34]. After the DCCT trial, patients were followed up for 17 years in the Epidemiology of Diabetes Interventions and Complications (EDIC) study. A legacy effect was observed in the former intensive-therapy group during this period, as evidenced by a significant reduction in the prevalence of cardiovascular events (nonfatal MI, stroke, or death from CVD; HR 0.43; 95% CI, 0.12–0.79), and nephropathy (HR 0.54, 95% CI, 0.34–0.84) (Table 1) [35].

The United Kingdom Prospective Diabetes Study (UKPDS) recruited patients with newly diagnosed T2DM (six months from diagnosis); after a three-month diet (main features: low and moderately high daily intake of saturated fat and fibre, respectively; about 50% of daily caloric intake from carbohydrates; hypocaloric diet in overweight patients), they were randomised to intensive glucose control with medication (sulfonylurea, insulin, metformin) or to conventional dietary treatment, and they were followed-up for ten years. Intensive glucose management was associated with a significant reduction for any diabetes-related endpoint (12%) and for any diabetes-related death (10%) compared to the control group. Most of the risk reduction in the "any diabetes-related aggregate" endpoint was due to a 25% risk reduction in microvascular endpoints. However, there was no significant reduction in macrovascular endpoints between both arms of the study. Moreover, patients in the intensive group had more hypoglycaemic episodes than those in the conventional one [36]. Although differences between both groups in HbA1c levels were lost after the first year, a clearly different incidence in clinical endpoints in the post-trial follow-up of the UKPDS was observed. Thus, relative reductions in risk persisted at ten years for any diabetes-related endpoint (RR 0.91, CI 95%, 0.83–0.99), for microvascular disease (RR 0.76, CI 95%, 0.64–0.89), MI (RR 0.85, CI 95%, 0.74–0.97), and for death from any cause (RR 0.87, CI 95%, 0.79–0.96), and emerged over time in the intensive therapy group, as more events occurred. In the metformin group (overweight patients), significant risk reductions persisted for any diabetes-related endpoint (RR 0.79, CI 95%, 0.66–0.95), MI (RR 0.67, CI 95%, 0.51–0.89), and death from any cause (RR 0.73, CI 95%, 0.59–0.89) (Table 1) [37,38].

The Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial included patients with T2DM with a mean duration of ten years and previous cardiovascular events or multiple cardiovascular risk factors. The patients were assigned to receive intensive or standard therapy (targeting HbA1c <6% versus vs. <7.9%, respectively) for 3.7 years, with a combination of different hypoglycaemic drugs including insulin [39]. It was not observed statistically significant benefit during the intervention period nor after ending the trial in the intensive treatment group, except for the rate of nonfatal MI, which was lower than in the standard therapy group (HR 0.76, 95% CI, 0.62–0.92). In fact, a significant increase in cardiovascular mortality (HR 1.35, 95% CI, 1.04–1.76), and death from any cause (HR 1.22, 95% CI, 1.01–1.46) was detected during and after the ACCORD trial in the intensive treatment group, attributed to a higher incidence of severe hypoglycaemia in those patients (Table 1) [39,40].

The Veterans Affairs Diabetes Trial (VADT) was performed in patients with poorly controlled long-standing T2DM (mean duration of 11.5 years), 40% of whom had already suffered a cardiovascular event. They were randomly assigned to intensive or standard therapy for a median of 5.6 years. Other cardiovascular risk factors were treated uniformly. The intensive glucose-lowering treatment led to a between-group difference of 1.5 percentage points in HbA1c levels (6.9% in the intensive-therapy group vs. 8.4% in the standard-therapy group) during the in-trial period. The only benefit observed was a slower progression of microalbuminuria in the intensive treatment group. However, a significantly lower incidence of CVD (HR 0.83, 95% CI, 0.70–0.99) in patients originally assigned to intensive therapy was observed after ten years of follow-up. Nevertheless, over a 15-year follow-up period, the risks of major cardiovascular events (HR 0.91, 95% CI, 0.78–1.06), or death from any cause (HR 1.02, 95% CI, 0.88–1.18) were not lower in the intensive-therapy group than in the standard therapy group,

suggesting a modest and transient long-term cardiovascular benefit of intensive glucose-lowering therapy in patients with more advanced diabetes (Table 1). In fact, the cardiovascular benefit (legacy effect) of intensive therapy lasted when HbA1c curves of both groups were separated (Table 1) [41–43].

The Anglo-Danish-Dutch study of Intensive Treatment in People with Screen-Detected Diabetes in Primary Care (ADDITION) study recruited middle-aged adults with a moderate to high risk of T2DM for diabetes screening, in order to evaluate the risk of CVD and mortality among incident cases of T2DM in a screened group compared to an unscreened population. Participants diagnosed with T2DM were offered intensive multifactorial treatment and compared to those receiving routine care for five years, with a reduction in fatal and nonfatal CVD of 17% shown. After a 10-year follow-up, patients diagnosed after population-based screening had a significant 21% lower risk in all-cause mortality (HR 0.79, 95% CI, 0.74–0.84) and a significant 16% lower risk in cardiovascular events (HR 0.84; 95% CI, 0.80–0.89), but no significant risk reduction on cardiovascular mortality compared to unscreened patients [44]. Moreover, population-based diabetes screening was associated with less need for insulin therapy after ten years and slightly better long-term glycaemic control compared to patients diagnosed during usual care (Table 1) [45].

The Diabetes and Aging Study was a large observational study of newly diagnosed T2DM patients and followed those with long survival post-diagnosis (≥10 years). The study assessed the impact of HbA1c levels ≥6.5% in the first year after T2DM diagnosis on later CVD risk. The risk of microand macrovascular events was higher in those patients with HbA1c levels ≥6.5% compared to those with lower levels, whereas mortality risk was significantly higher in patients with HbA1c levels ≥8%. These results suggest it is necessary to achieve normoglycemia in the first 12 months after T2DM diagnosis to generate a legacy effect (Table 1) [46].

#### **6. Legacy E**ff**ect after Blood Pressure Control**

Several clinical trials have assessed legacy effects in different populations of hypertensive patients, the main results of which are listed below. The specific data from each article are shown in Table 2. A separate UKPDS post-trial follow-up study assessed whether risk reductions for microand macrovascular complications were achieved and maintained in a subgroup of hypertensive subjects with newly diagnosed T2DM with tight versus less-tight BP control with captopril or atenolol. However, differences in BP between groups (tight vs. less tight) disappeared within two years after trial termination, and most cardiovascular benefits were not sustained during the 10-year post-trial follow-up period. Only a reduced risk for peripheral vascular disease associated with tight BP control was significant (RR 0.50, 95% CI, 0.28–0.92). Hence it would be necessary to maintain good BP levels over time to preserve this previous benefit (Table 2) [36,47].

The Systolic Hypertension in the Elderly Program (SHEP) trial was performed in elderly patients with isolated systolic hypertension who were randomised to chlortalidone therapy or placebo (plus atenolol or matching placebo, if BP remained uncontrolled). Over a mean follow-up of 4.5 years therapy, reduction in fatal or nonfatal stroke (RR 0.64, 95% CI, 0.50–0.82), MI (RR 0.67, 95% CI, 0.47–0.96), and heart failure (RR 0.51, 95% CI, 0.37–0.71) in the chlortalidone group compared to placebo was observed. However, there was no significant risk reduction in all-cause and cardiovascular mortality. After a 22-year follow-up, significant, albeit modest life expectancy gains free from CVD-related deaths were observed in the chlortalidone group, corresponding with approximately one day (HR 0.89, 95% CI, 0.80–0.99) gained for each month of treatment. (Table 2) [48].


**Table 2.** Studies assessing the legacy e ffect after intensive BP control.

Program; ALLHAT:

ANBP2: Second Australian National BP study; ASCOT:

and the cohort of post-trial follow-up, respectively.

Antihypertensive

 and Lipid Lowering Treatment to Prevent Heart Attack Trial; ROADMAP: Randomised Olmesartan And Diabetes

Anglo-Scandinavian

Cardiovascular

 Outcomes Trial. \* The number of randomised patients enrolled in the interventional

MicroAlbuminuria

 Prevention;

 phase

In the Antihypertensive and Lipid Lowering Treatment to Prevent Heart Attack Trial (ALLHAT), 32,804 hypertensive patients with at least another cardiovascular risk factor were randomised to receive chlorthalidone, amlodipine, or lisinopril for 4 to 8 years, and thereafter, passive surveillance continued for a total follow-up of 8 to 13 years. During the in-trial period, no statistically significant differences in CHD or nonfatal MI were observed. In the post-trial follow-up, the only significant differences were observed in secondary outcomes, such as heart failure and stroke mortality, which were higher with amlodipine and lisinopril, respectively, compared to chlorthalidone. However, a significant treatment-by-race interaction was detected and, after accounting for multiple comparisons, none of these results were deemed significant (Table 2) [49].

The Randomised Olmesartan And Diabetes MicroAlbuminuria Prevention (ROADMAP) trial included patients with T2DM with at least one additional cardiovascular risk factor and normoalbuminuria. They were randomly assigned to olmesartan or placebo for 3.2 years; the main outcome was significantly delayed microalbuminuria onset in the olmesartan group. After a 6.5-year observational follow-up period, researchers concluded that congestive heart failure (CHF) requiring hospitalisation (OR 0.23, CI 0.06–0.85) and diabetic retinopathy (OR 0.34, CI 0.15–0.78) were significantly lower in the former olmesartan group, indicating a legacy effect in these outcomes (Table 2) [50,51].

The Second Australian National BP study (ANBP2) observed a lower risk of cardiovascular (HR 0.47, 95% CI, 0.27–0.81) and all-cause mortality (HR 0.63, 95% CI, 0.46–0.86) in the "treatment-naive" group (participants without BP-lowering medication at study registration) compared to BP-lowering medication or "previous treatment" group (BP-lowering treatment at registration; median duration of previous therapy was five years) during the in-trial period in an elderly cohort. No differences were found between groups with respect to randomised treatment allocation to either ACEI or diuretic-based regimens; likewise, there were no differences in cardiovascular outcomes or all-cause mortality when the data from the in-trial and ten-year post-trial follow-up were combined (Table 2) [52].

The Anglo-Scandinavian Cardiovascular Outcomes Trial (ASCOT) enrolled patients with hypertension and at least three other cardiovascular risk factors. In the BP-lowering arm, patients were randomised to either an amlodipine-based regimen (adding perindopril as required) or an atenolol-based regimen (adding bendroflumethiazide and potassium as required). The in-trial results demonstrated a significant reduction in cardiovascular events, mortality and all-cause mortality in subjects assigned to the amlodipine-based group, but no overall difference in cardiovascular mortality (HR 0.90, 95% CI, 0.81–1.01) among treatments. These results could suggest that not all hypotensive drugs are equally effective in preventing CVD, although the resulting BPs are similar. After 16 years of total follow-up (ASCOT Legacy Study), there was no overall difference in all-cause mortality between treatments, although significantly fewer deaths from stroke (HR 0.71, 95% CI, 0.53–0.97) occurred in the amlodipine-based treatment group (Table 2) [53].

Finally, a systematic review and meta-analysis of three clinical trials were performed to analyse whether early versus late initiation of antihypertensive treatment was better at reducing cardiovascular morbidity and mortality. The trials involved 4746 mildly hypertensive (systolic BP 140–159 mmHg) middle-aged patients free of CVD at baseline and with low cardiovascular risk. No differences were seen between strategies during the in-trial period (5 years) or during post-trial follow-up (ten years). Therefore, the review showed no clinically adverse legacy effect on mortality or major CVD with delayed pharmacotherapy in middle-aged mildly-hypertensive subjects with low cardiovascular risk [54].

#### **7. Legacy E**ff**ect after Lipid Control**

Several clinical trials have assessed the legacy effects in different populations of patients with dyslipidaemia the main results of which are listed below. The specific data from each study are shown in Table 3.



the

interventional

 phase and the cohort of post-trial follow-up, respectively.

The ALLHAT lipid-lowering trial (LLT) evaluated the impact of large, sustained cholesterol reductions in all-cause mortality in a cohort of hypertensive patients with at least one other cardiovascular risk factor. All-cause mortality did not differ significantly between the pravastatin and usual care treatment groups (RR 0.99, 95% CI, 0.89–1.11) at six years of follow-up. Similarly, no significant reductions in CHD events (RR 0.91, 95% CI, 0.79–1.04) were observed at the end of the follow-up period (Table 3) [55].

The West of Scotland Coronary Prevention Study (WOSCOPS) was a primary prevention trial performed in men (45 to 65 years old) with hypercholesterolemia who were randomised to pravastatin or placebo for five years. The 20-year follow-up study after the initial intervention identified a legacy benefit in the pravastatin group compared to placebo, with a significant reduction in all-cause mortality (HR 0.87, 95% CI, 0.80–0.94), attributable mainly to a 21% decrease in cardiovascular death (HR 0.79, 95% CI, 0.69–0.90), reduction in hospital admissions for any coronary event (18%), for MI (24%), and for heart failure (35%). The authors suggested that these benefits may be the result of reduced infarct size, prevention and regression of atherosclerotic changes, or pleiotropic effects of statins. On the other hand, there was no difference in non-cardiovascular or cancer death rates between groups (Table 3) [56].

The ASCOT lipid-lowering arm (LLA) is a substudy of the ASCOT study previously reported. ASCOT-LLA enrolled ASCOT study participants who had total cholesterol of 6.5 mmol/L or less and no previous lipid-lowering treatment. They were randomised to receive atorvastatin or placebo as primary prevention for CHD; lower nonfatal MI and fatal CHD was demonstrated in the atorvastatin group during the interventional period of the study. The legacy effect was studied for 16 years after initial randomisation and also showed a significant risk reduction in cardiovascular mortality (HR 0.85, 95% CI, 0.72–0.99) in those assigned to atorvastatin group. No differences in all-cause mortality or stroke were seen between the atorvastatin and placebo groups during the post-trial study period (Table 3) [52].

The ACCORD-Lipid study included a subgroup of ACCORD study participants. Briefly, T2DM patients were randomised to receive simvastatin plus fenofibrate vs. simvastatin plus placebo for five years; no evidence of a beneficial effect of statin-fibrate combined treatment compared to statins alone on cardiovascular outcomes and mortality was found. However, a beneficial reduction in major CHD (HR 0.65, 95% CI, 0.48–0.90) was observed in a subgroup of participants with dyslipidaemia. The extended post-trial follow-up study (ACCORDION) showed lower rates of all-cause mortality, cardiovascular mortality, nonfatal MI, CHF and major CHD in the simvastatin-fibrate group in the ten years following randomisation. Moreover, the trial period's combined statin-fibrate treatment arm conferred a beneficial legacy effect, which was observed in the post-trial follow-up on all-cause mortality (HR 0.65, 95% CI, 0.45–0.94). Fibrate therapy, offered as an add-on to statin therapy, may be beneficial for people with diabetes with hypertriglyceridemia and/or reduced HDL-C (Table 3) [57].

A recent meta-analysis, including eight placeboes vs. statin randomised clinical trials for primary and secondary cardiovascular prevention, evaluated the legacy effect during a mean post-trial follow-up ranging from 1.6 to 15.1 years. The results mostly showed a significant reduction in cardiovascular and all-cause mortality within the trial period, and less benefit in the post-trial period. The legacy effect was observed in relation to lower all-cause mortality but, appeared to have no effect on CVD mortality. Additionally, in a subgroup analysis, there appeared to be a greater legacy effect when statins were used for primary prevention compared to secondary prevention in CVD and all-cause mortality (HR 0.87 and 0.90, respectively), suggesting the importance of long-term prevention in these patients [58].

Another meta-analysis carried out by Hirakawa et al. (62) analysed the legacy effect in the post-trial follow up (mean duration six years) after an intervention period with antihypertensive and lipid-lowering treatments. The study demonstrated significant reductions in all-cause and cardiovascular mortality (about 9% and 12%, respectively) after discontinuation of antihypertensive and lipid-lowering treatment during the overall follow-up, although this effect was lower than that observed during the in-trial phase. Furthermore, progressive attenuation of post-trial benefits was noted as the length of the post-trial observation increased. However, no clear differences between the effects of BP or lipid-lowering therapies were detected across trials studying different types of therapies or patient populations. These findings indicate that it is important to continue antihypertensive and lipid-lowering treatment in the long-term to provide optimal cardiovascular protection [59].

#### **8. Legacy E**ff**ect after Multifactorial Intervention**

The Action in Diabetes and Vascular Disease (ADVANCE) trial was performed in 11,140 T2DM patients ≥55 years old, with at least one additional risk factor for CVD. The trial proposed a double strategy to optimise glycaemic control (intensive glucose control or standard control) and BP control (perindopril-indapamide or placebo) for five years. During the intervention phase, there was a reduction in microvascular events (14%), primarily due to reduction in nephropathy incidence in the intensive glucose-control group, as well as a reduction in the relative risk of all-cause mortality (14%) and cardiovascular mortality (18%) in the BP control group. The six-year post-trial follow-up study found a significant reduction in all-cause mortality (HR 0.91, 95% CI, 0.84–0.99), and cardiovascular mortality (HR 0.88, 95% CI 0.77–0.99) in those patients originally treated with perindopril-indapamide compared to placebo. Although the confidence limits were wide, the results suggested that one death from any cause would be prevented for every 79 patients assigned to active therapy for five years. On the other hand, there was no evidence that the effects of the treatment could be inferred from initial BP levels or concomitant use of other treatments at baseline. However, intensive glucose control did not provide any long-term benefits during the intervention period nor after post-trial follow-up (Table 4) [60–62].


**Table 4.** Studies assessing the legacy effect after multifactorial intervention.

Abbreviations: BP: blood pressure; CVD: cardiovascular disease; T2DM: type 2 diabetes mellitus. \* The number of randomised patients enrolled in the interventional phase and the cohort of post-trial follow-up, respectively.

The Steno-2 study recruited T2DM patients (*n* = 160; mean age 55 years) with microalbuminuria who were randomly assigned to standard treatment or intensified multifactorial intervention during a mean follow-up of 7.8 years. This intervention included simultaneous control of glucose levels, BP, lipid profile, and antiplatelet therapy through pharmacological and non-pharmacological intervention. The latter included dietary advice (total daily intake of fat ≤30% of total daily energy intake and less than 10% kcals as saturated fat), light to moderate exercise at least 30 min three to five times

a week, and smoking cessation. During the active intervention period, there was risk reduction for nephropathy progression (OR 0.27, 95% CI, 0.10–0.75), retinopathy progression (OR 0.45; 95% CI, 0.21–0.95), and autonomic neuropathy progression (OR 0.32, 95% CI, 0.12–0.78), as well as for cardiovascular complications and all-cause mortality combined (OR 0.45, 95% CI, 0.22–0.93) in the intensive treatment group. After a 21-year follow-up from randomisation, a significant reduction in all-cause mortality (45%), cardiovascular mortality (62%), macroalbuminuria (48%), retinopathy progression (33%), and autonomic neuropathy (41%) was demonstrated in the intensive treatment arm, suggesting a legacy effect from intensive multifactorial treatment compared to the standard approach. For patients in the intensive treatment group, death and time to first cardiovascular event was delayed by 7.9 and 8 years, respectively, compared to patients in the standard treatment group (Table 4) [63,64].

#### **9. Discussion**

Most studies evaluating the legacy effect on the primary or secondary prevention of CVD are clinical trials promoted and financed by the pharmaceutical industry in which an observational study has continued at the end of the intervention phase. As a consequence, the legacy effect has been assessed mainly after pharmacological intervention. Few studies have been carried out after a nutritional intervention, and most focused on animal and diabetic models.

Different clinical outcomes have been evaluated, including microvascular complications (retinopathy, neuropathy, nephropathy), major cardiovascular events (heart failure, MI, stroke, and ischemic gangrene) and mortality, as well as analytical parameters. The results obtained have been quite heterogeneous, probably due to differences in the baseline characteristics of the patients included, the type and duration of the interventions, and post-trial follow-up. In general terms, the legacy effect has been observed to be less intense than that achieved in the intervention phase and tends to diminish with longer follow-up. All the data suggest that this effect is greater and longer lasting in patients without known CVD undergoing intensive therapy for recent-onset diabetes, hypertension or dyslipidaemia.

A limited number of studies related to the legacy effect after nutritional intervention have been published to date. In studies performed with rats and mice, some authors have demonstrated that after CR (reduction of daily caloric intake by approximately 25%) or intermittent fasting, glycaemic control and insulin resistance could improve for several weeks. The evidence in humans is lacking, as health benefits of CR or intermittent fasting have been evaluated mainly in observational studies and a single clinical trial (CALERIE Study). The effect during the intervention period appears to be favourable but disappears shortly after the intervention ends. Therefore, CR or intermittent fasting has not been shown to generate a legacy effect in humans [30]. In addition, we would like to highlight safety concerns regarding intermittent fasting in patients with diabetes mellitus due to the higher risk of hypoglycemia. Data from the Oslo Cardiovascular study suggest that systematic advice on a healthy diet and smoking cessation for five years could be associated with a legacy effect translated into a reduced risk of cardiovascular mortality in the next 40 years. However, it is difficult to ensure that no confounding factors could have altered these results [32]. Currently, there are no data available on the possible legacy effect in other, more recent nutritional intervention studies, such as the PREDIMED (Prevención con Dieta Mediterránea) Study [65]. The limited scientific evidence published on the possible legacy effect in CVD prevention after a dietary intervention may be due to multiple factors, including: difficulty in maintaining high adherence to the proposed nutritional intervention, absence of specific biomarkers of dietary compliance and difficulty in having a real control group or blinding the interventions. In addition, these studies require a very long follow-up as well as a high economic cost that can be more difficult to finance if there is no financial support from public institutions. Due to all these limitations, there are few nutritional intervention studies that have carried out a long follow-up at the end of the intervention [66,67].

The first scientific evidence of the legacy effect in patients with diabetes comes from the DCCT and the UKPDS studies, in which an intensive glycaemic control compared to standard treatment resulted in a significant reduction in cardiovascular mortality, nonfatal MI, stroke and nephropathy. These beneficial effects persisted for more than a decade after the intervention was finished [35,37]. Furthermore, the VADT trial provided the first evidence of the legacy effect in the reduction of a composite of major cardiovascular events (MI, stroke, CHF or amputation for ischemic gangrene) in patients with poorly controlled and long duration T2DM after ten years of randomisation. Notwithstanding, these significant cardiovascular outcomes disappeared over a longer follow-up period (15 years) [42,43]. On the other hand, the ADVANCE and ACCORD trials, which included long-standing T2DM patients, showed no legacy effect in CVD. Differences observed among these trials may be explained due to the fact that the DCCT and UKPDS studies recruited younger patients with new-onset diabetes without known CVD, in contrast to the characteristics of the patients included in the ADVANCE and ACCORD studies [40,62].

Several clinical trials have analysed the legacy effect after tight BP control with controversial results. On the one hand, the SHEP and the ROADMAP trials showed reductions in cardiovascular mortality, retinopathy and delayed onset of microalbuminuria, whereas the ASCOT study only demonstrated a small reduction in stroke mortality in the amlodipine-treatment group at the end of the entire follow-up [49,50,52]. On the contrary, the HDS and the ALLHAT studies did not provide clear evidence of a legacy effect, suggesting that this beneficial effect in patients with higher cardiovascular risk and probable subclinical CVD is unlikely to be achieved once in the post-trial phase, when BP is less strictly controlled. In addition, the benefits of antihypertensive treatment on major cardiovascular outcomes usually appear shortly after treatment implementation, and are attenuated when BP differences between groups are lost [47,48]. Therefore, based on clinical trials conducted, it appears that the legacy effect does not exist or seems to be mild and transient after tight antihypertensive regimens.

Different lipid-lowering trials, including ASCOT, WOSCOPS and ACCORD lipid studies, have also observed a legacy effect, resulting in a reduction in all-cause mortality and CHD mortality for 10–20 years after ending the interventional phase (statin or fibrates treatment), whereas the ALLHAT-LLT study did not provide evidence of a legacy effect after intensive lipid-lowering treatment [52,55–57]. It should be noted that in those trials in which a legacy effect was observed, the statin, a drug with pleiotropic effects, was compared against placebo. Moreover, the observation that five years of statin therapy led to a lower long-term risk of all-cause and CVD mortality raises the question of whether treatment with statins for 5–10 years would be sufficiently beneficial, while limiting lifetime exposure to the drug. However, there are still concerns about whether this therapeutic strategy could be effective in any patient, or in select populations only.

Until now, the legacy effect after multifactorial intervention has only been observed in the Steno-2 study, which included pharmacological and non-pharmacological measures. Indeed, lower mortality, CVD incidence and microvascular complications were detected in the intensive treatment arm of the trial compared to the standard treatment group. Surprisingly, the results were obtained in a sample of only 160 patients with intermediate cardiovascular risk but without CVD at baseline. It has been suggested that the positive long-term cardiovascular effects must be the result of a synergistic effect of the multifactorial intervention on cardiovascular risk factors [64]. Although these results are impressive, they have not been reproduced in subsequent studies.

The knowledge of pathophysiological mechanisms underlying the legacy effect adds plausibility to its existence. The best-known mechanisms are those related to the deleterious effects of hyperglycaemia through the development of metabolic memory induced during the first years of diabetes onset, which cannot be reversed with better glycaemic control. Thus, early interventions against hyperglycaemia could reduce ROS production and oxidative stress in the mitochondria of endothelial cells, decrease AGE formation and RAGE expression, and therefore, prevent activation of inflammatory processes and epigenetic changes in the arterial walls in the long term [4,5,15–17,20,21]. Likewise, dysregulation of RAAS may be involved in vascular complication development, through increased oxidative stress and AGE formation. However, these mechanisms are not yet fully understood [19].

This review has several strengths and limitations. Firstly, one strength involves the inclusion of well-designed multicentre prospective clinical trials with large participant samples during the interventional phase. Secondly, most trials have a long post-interventional follow-up (6–20 years), which is enough time to detect differences between groups. Likewise, the study of different types of cardiovascular risk factors (diabetes, hypertension or dyslipidaemia), as well as variable evolution time and treatment objectives (primary or secondary prevention of CVD) may allow us to identify in which studies the legacy effect may be more or less relevant. However, some limitations should be taken into account. First, loss of patients during post-interventional follow-up can reach 25–30% of subjects included in the trial (attrition bias), therefore data collection during this period may not have been as exhaustive as in the intervention period. The publication of only those trials with positive results (publication bias) and the absence of blindness in several trials (co-intervention bias) should also be considered. Furthermore, the inclusion of post hoc analysis together with the detection of the legacy effect is influenced by different confounding factors (recommended therapeutic targets, adherence to treatment, global cardiovascular risk), which may lead to a questionable interpretation of the results. Finally, another limitation that must be taken into account is the fact that few non-pharmacological studies (i.e., dietary intervention studies) have been carried out. Although dietary recommendations are included in most studies, they are usually quite generic, and it is also difficult to know the degree of adherence to them.

In conclusion, there is sufficient data to suggest the existence of a legacy effect after intensive intervention on cardiovascular risk factors in subjects with moderate-high vascular risk. However, this effect is not equivalent for all risk factors and could be influenced by patient characteristics, disease duration and the type of intervention performed. Currently, the available evidence suggests that the legacy effect would be greater in subjects with moderate-high cardiovascular risk but without known CVD, especially in patients with recent-onset diabetes. However, we should not withdraw any treatment to prevent CVD in these individuals as the level of available evidence on the legacy effect is low to moderate. Further investigation should be promoted to determine whether there is a legacy effect associated with nutritional interventions.

**Author Contributions:** Conceptualization, E.V.E., J.N.Á. and E.S.M.; methodology, E.V.E., J.N.Á. and E.S.M.; software, E.V.E., J.N.Á. and E.S.M.; validation, E.V.E., J.N.Á. and E.S.M.; formal analysis, E.V.E., J.N.Á. and E.S.M.; investigation, E.V.E., J.N.Á. and E.S.M.; resources, E.V.E., J.N.Á. and E.S.M.; data curation, E.V.E., J.N.Á. and E.S.M.; writing—original draft preparation, E.V.E. and E.S.M.; writing—review and editing, E.V.E., J.N.Á. and E.S.M.; visualization, E.V.E., J.N.Á. and E.S.M.; supervision, E.S.M.; project administration, E.S.M.; funding acquisition, E.S.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Sociedad Española de Medicina Interna (SEMI), Spain, grant number DN40585 (2015).

**Acknowledgments:** We thank Cambridge Proofreading LLC for manuscript proofread and edition.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Review*
