superscripts = EAR.

#### 3.2.2. YCM/PCM Consumers vs. Non-Dairy Consumers

Non-dairy consumers had significantly lower intakes than YCM consumers in most macro- and micronutrients, except for fibre in children aged one to two years (*p* < 0.05). No significant differences were seen for carbohydrates, fibre, and magnesium and selenium between the PCM and non-dairy consumer groups in children aged three to four years. Non-dairy consumers had mean nutrient intakes far below the recommendation. The only nutrients where non-dairy consumers were in line or above the recommended levels were protein, sodium, selenium, and niacin (all ages); magnesium, B6 and B12 (three to four years only).

#### *3.3. Percentage of Children with Inadequate Intakes*

Table 3 shows the proportion of children with nutrient intakes below the EAR or AI across the three consumer groups. Among non-dairy consumers aged one to two years, except for carbohydrates and selenium, more than 50% of children had intakes below the EAR/AI. In particular, almost all children in this group were inadequate in calcium, vitamin D, and iron (99%, 97% and 91%, respectively). The highest level of inadequate intakes among other milk consumers (one to two years) were vitamin D (98%), vitamin E (97%), iron (92%), folate (82%), and vitamin B6 (73%). YCM consumers (1 to 2 years) had the lowest percentage of inadequate intakes when compared to the other groups. The highest levels of inadequacy in this group were vitamin E (68%), phosphorus (55%), vitamin D (52%), and folate (49%).

Among preschool children, the proportions with inadequate intakes were slightly lower than the younger children, except for calcium in PCM (67%) and other milk consumers (65%), and vitamin D (72%) in PCM consumers. Non-dairy consumers still had the highest proportion of children with inadequate intakes across most nutrients.

#### *3.4. Odds of Nutrient Adequacy Per Consumer Group*

Table 4 illustrates the odds ratios of other milk consumers and non-dairy consumers meeting nutrient adequacy when compared to YCM/PCM consumers (reference group), after adjusting for potential confounding factors (wealth quintiles, dwelling location, mother' education level, mother's occupation). Among children aged one to two years, other milk consumers were significantly less likely to meet the AMDR for carbohydrates, as well as the EARs for iron, zinc, thiamin, niacin, vitamin B6, folate, vitamins C, D, and E (all *p* < 0.001), and vitamin B12 (*p* = 0.002), than YCM consumers (reference group). Other milk consumers were more likely to meet the EAR of protein (*p* = 0.003), phosphorus, selenium, and riboflavin (*p* ≤ 0.001) than YCM consumers. Non-dairy consumers had significantly lower odds of meeting the adequacy of all nutrients when compared to YCM consumers, except for carbohydrates and selenium (both non-significant).

In preschool children, other milk consumers were less likely to reach nutrient adequacy in iron, folate, vitamins C, D, E (all *p* < 0.001), zinc, thiamin, and vitamin B6 (all *p* < 0.05), but were more likely to meet adequacy in phosphorus and riboflavin when compared to PCM consumers. Non-dairy consumers had lower likelihood of meeting adequacy in total fat, calcium, iron, zinc, folate, vitamins C, D, E (all *p* < 0.001), and thiamin (*p* = 0.005) and vitamins B6 (*p* = 0.024), but more likely of being within AMDR of carbohydrates (*p* = 0.016) than PCM consumers.


**Table3.**Percentageofchildrenwithinadequateintakesperconsumergroupbyagegroupsonetotwoyears,threetofouryears,and


*Nutrients* **2020**, *12*, 392

**Table 4.** Odds ratios of children meeting the estimated average

requirements/adequate

intake/Acceptable

Macronutrient

 Distribution

 Ranges

(EAR/AI/AMDR)

 of

The results also highlight the sociodemographic factors that were significantly related to increased or reduced odds of meeting nutrient adequacy. In children one to two years, living in urban areas increased the likelihoods of meeting protein, phosphorus, zinc, thiamin, niacin, and vitamin B6 adequacies; being in the richest wealth quintile were associated with adequacy in total fat, calcium, zinc, thiamin, and vitamin B6 (all *p* < 0.05). On the other hand, being in poor or poorest wealth quintiles, and/or mothers with education at or below high school levels reduced the odds of meeting adequacy in most nutrients, including protein, total fat, calcium, selenium, zinc, B vitamins, and vitamin E (all *p* < 0.05). Similar trends were also observed in children aged three to four years. The only nutrient that had an opposite trend than the rest of the nutrients was carbohydrate. In children aged one to two years, being in the richest wealth quintile lowered the odds of meeting the AMDR for carbohydrates, and similarly among three to four years, those who were the richest, living in urban dwelling, and with mother currently working reduced the likelihood of meeting AMDR for carbohydrates, while being in the poor or poorest wealth quintile increased the odds.

None of the sociodemographic factors were associated with iron, vitamins C and D adequacies in children aged one to two years, and vitamin B6, folate, vitamins D and E adequacies in children three to four years.

#### **4. Discussion**

The current study assessed the contribution of different milks to daily nutrient intakes among young and preschool children in the Philippines, by comparing the mean nutrient intakes and percentage of children with inadequate intakes between three consumer groups: YCM/PCM consumers, other milk consumers, and non-dairy consumers. Our analysis has shown that children in the Philippines who consumed dairy products, including YCM/PCM and other milks, had higher intake of most nutrients and lower nutrient inadequacy than non-dairy consumers. Non-dairy consumers, in particular, had intakes of energy, total fat, calcium, phosphorus, iron, potassium, folate, vitamin D, and vitamin E far below recommendations. On the other hand, YCM/PCM consumers had mean intakes of micronutrients that were closer to the recommendations when compared to other consumer groups. While other milk consumers had similar macronutrient intakes to YCM/PCM consumers, a higher percentage of other milk consumers had inadequate micronutrient intakes as compared to YCM/PCM consumers. These findings support the fact that dairy is an important food group that significantly contributes to the intakes of children in these two age groups. Moreover, the choice of dairy products (e.g., YCM/PCM vs. other milks) contributes differently to total daily nutrient intakes, and that YCM/PCM is a good option in terms of reducing micronutrient inadequacy.

#### *4.1. Nutrient Inadequacy in Filipino Children*

The alarming prevalence of inadequate intakes, particularly for energy and micronutrients, among young children in the Philippines deserves further attention. Overall, the energy intakes of the three consumer groups were below the age and gender specific Recommended Energy Intakes in the Philippines. Moreover, 77% of children (one to four years) were inadequate in calcium, 81% inadequate in iron, 74% in folate, 60% in zinc, 62% in vitamin C, and 90% in vitamin D. Previous studies have found that the diets of young Filipino children lacked diversity, which contributed to high levels of nutrient inadequacy [21,22]. While improving diet diversity (increasing number of food groups consumed) could incrementally increase micronutrient adequacy, a more recent study found that, even with high diet diversity, Filipino school children had difficulty in achieving adequacy in calcium, folate, iron, vitamin A, and vitamin C. This is likely due to the low quantities of food consumed, or that the current food supply in the Philippines might not contain enough of these nutrients to fulfil the needs of children [23]. Indeed, our logistic regression analysis provided interesting insights that those living in urban areas and wealthier socio-economic quintiles were more likely to achieve total fat and micronutrient adequacies, while those from poorer socio-economic backgrounds and having mothers with lower education levels had reduced odds of meeting micronutrient adequacies, except

for carbohydrates. It is possible that those from higher socio-economic quintiles ate better diets and that living in urban areas meant better accessibility and availability of more nutritious foods (e.g., meat) and fortified beverages; and, those from lower socio-economic quintiles had a predominately carbohydrate-rich (e.g., rice) diet that lack variety. Our findings support the literature on key strategies to reduce stunting in Southeast Asia [24] and in Filipino children [23]. There is an urgent need for better access to a variety of nutrient-rich foods, particularly in the lower socio-economic groups, increase availability of fortified foods that are tailored for young children, in tandem with increased parental education on dietary intake, to close nutrient gaps among young children.

#### *4.2. Large Proportion of Filipino Young Children Did Not Consume Dairy on a Given Day*

The Daily Nutritional Guide Pyramid for Filipino Children aged one to six years recommends consuming one glass of milk & milk products per day, which includes one glass of whole milk, or 1/2 cup evaporated milk diluted with 1/2 glass water, or four tablespoons of powdered whole milk diluted with one glass of water [18]. However, the current study found half of Filipino children (43% aged one to two years and 58% aged three to four years) were not consuming any milk or milk products on a given day. Non-dairy consumers had mean energy intakes approximately 40% lower than the Recommended Energy Intakes (REI) (one to two years: mean intake of 543 kcal/d vs. REI of 1000 kcal/d for boys and 920 kcal/d for girls; three to four years: mean intake of 828 kcal/d vs. REI of 1350 kcal/d (boys) and 1260 kcal/d (girls)), suggesting these young children were at serious risk of undernutrition.

Indeed, the consumption of milk and animal-sourced foods is known to be limited among low-income countries. It is estimated that animal-source foods, such as milk, provide between 5% to 15% of total daily energy in Asian countries, when compared to over 20% daily energy in western countries, such as the U.S. and Europe [4]. A study on South East Asian countries, including Thailand, Malaysia, Vietnam, and Indonesia also observed sub-optimal milk and dairy intake in children aged one to 12 years. Only around half of Indonesian (52%) and Vietnamese (47%) children consumed dairy products on a daily basis. The study found that children who consumed less than one portion of dairy per day had significantly lower nutrient intakes and higher prevalence of underweight and stunting than children who consumed ≥1 portion of dairy per day [25].

Our analysis also suggests that a high percentage of non-dairy consumers came from poorer wealth quintiles, from rural areas, and with mothers who had lower education levels; while, the opposite sociodemographic characteristics were observed for those who consumed YCM/PCM. Financial constraints, limited product availability, and lack of parental awareness of the importance of nutrition and dairy food could be the reasons for the high proportion of children not consuming YCM/PCM or other milks.

#### *4.3. The Role of YCM*/*PCM in the Diet of Filipino Children*

A previous study has highlighted that both cow's milk and YCM/PCM were two of the top five contributors of energy, carbohydrate, protein, fat, thiamine, riboflavin, vitamins A, C, calcium, iron and zinc for Filipino children aged 12 to 35.9 months [3]. Particularly, YCM was a top contributor of iron and zinc in children aged 12–23.9 months. Other studies in Europe and Oceania found that YCM/PCM consumers had higher intakes of nutrients, such as iron, vitamins C, and D than cow's milk consumers [5–8]. Our study adds to the evidence that YCM/PCM and other milk consumers had nutrient intakes that were more in line with recommendations than non-dairy consumers, and that the dairy food group can contribute hugely to the daily intake of nutrients in young children. Furthermore, the fact that lower percentages of YCM/PCM consumers had inadequate intakes, and that they had higher odds of achieving adequacy in iron, zinc, thiamine, niacin, folate, vitamins B6, B12, C, D when compared to other milk consumers, suggests that YCM/PCM might be a good choice of dairy product for Filipino children.

In contrast to the expert opinions and recommendations on toddler milk and YCM from Europe and the U.S. [13,26–28], the routine use of YCM is not deemed a necessity, as the missing nutrients in the diet of young children can also be provided by other dietary sources in these countries, the findings of the current study did support the role of YCM in the diet of young children in the Philippines. Furthermore, our study also supported that the use of YCM/PCM can increase the intake of iron and vitamin D and decrease the intake of protein when compared to other milk, in accordance to the European Food Safety Authority (EFSA) scientific opinion [26], conclusion from the European Society for Paediatric Gastroenterology Hepatology and Nutrition (ESPGHAN) committee [13], and findings from previous studies [5–11].

However, one caution to highlight is that, while our data supports that, overall, YCM/PCM consumers have lower nutrient inadequacies than other consumer groups, not all YCM/PCMs on the market have favorable nutritional compositions for children. Table S1 highlights the minimum, maximum, and mean nutrient compositions across 18 brands of YCM/PCMs. Parents and health care professionals should consider the nutritional profiles of YCM/PCMs when choosing YCM/PCM for children, as, evidently, YCM/PCMs have highly variable macro- and micronutrient content, and not all are fortified with essential micronutrients to the same extent.

#### *4.4. Strengths and Limitations*

The large sample of children (*n* = 2992) from a nationally representative survey of the Philippines covering the young toddlers one to four years is a strength of this study. The nutrient intakes of this age group have not been extensively studied in this country or in the Southeast Asia region, and this study has added to the existing literature that dairy is a key food group to the diet of young Filipino children beyond infancy. This study, to our knowledge, is the first to identify the high percentage of young children who are not consuming any dairy in the Philippines. This highlights the opportunity for the public health authority in the Philippines to improve dairy consumption in children.

One limitation of the current study is the method for estimating total quantity of powdered milk and YCM/PCM consumed. Powdered milk and YCM/PCM consumption were recorded as gram weight in powder form in the survey and a conversion factor of 6.7 was applied to estimate fluid weight (g), as consumed. While we had checked several milk brands' on-pack preparation instructions, and the conversion the factor used was deemed appropriate, it is possible that the total quantity of other milk and YCM/PCM as consumed was overestimated. Indeed, the estimated total YCM consumed by children aged one to two years (641 g/d) was higher than anticipated. A report of an expert panel recommended two to three servings, with a total of 400–600 mL of YCM/PCM per day are appropriate for children age one to six years [29]. The nutrient composition of the milk or YCM/PCM did not change and would not affect the calculation of mean nutrient intakes or percentage of children with inadequate intakes, despite the conversion to weight as consumed.

Another limitation is that the high proportion of missing data for some of the nutrient variables (vitamin A, saturated fat, monounsaturated, and polyunsaturated fats) meant that they were excluded for analyses. Given that vitamin A inadequacy is known to be high in the Philippines from the literature and that the fatty acids also are of public health importance, it was unfortunate that they were not investigated in the current analysis.

#### **5. Conclusions**

This study highlights the high prevalence of inadequate intakes, particularly for energy and micronutrients, among young children in the Philippines, regardless of milk consumption type. Secondly, the current study also provides novel insights on the importance of dairy food group including YCM/PCM and other milk to energy and macronutrient intakes, and meeting micronutrient adequacy in the diet of Filipino children. YCM/PCM have demonstrated to be a more superior option than other milks to achieve adequacy in key nutrients, such as iron, zinc, vitamins C, D, E, and some B vitamins in this population. Furthermore, expert opinions and recommendations on the role of YCM/PCM from Europe or the U.S may not apply to countries, such as the Philippines, where nutrient inadequacies and deficiencies are much more common. Finally, strategies targeting specific

socio-demographic segments (e.g., by wealth status, dwelling location, and mother's education) should be adopted to increase the intake of YCM/PCM or other fortified nutrient-dense foods and beverages among young children in the Philippines to improve nutrient intakes and reduce nutrient inadequacy.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/2/392/s1, Table S1: Nutritional compositions of YCMBs, Milk (fluid), Milk (powdered) considered in the current study.

**Author Contributions:** T.-N.M. conceptualized and designed the study, interpreted the data, and drafted the manuscript. M.T. conducted the data analysis, interpreted the data, and drafted the manuscript. I.A.-A. and M.V.C. guided the data analysis and critically reviewed the draft of the manuscript. E.A.O. conceptualized the study, oversaw the project and critically reviewed the manuscript. All authors proofread and approved the final manuscript.

**Funding:** This research and the APC were funded by Nestlé Research.

**Acknowledgments:** We would like to thank Emma F. Jacquier, Lolita Bazarova, Delphine Egli, Hervé Vancheri and Ryan Carvalho for their guidance of the manuscript. We would like to thank FNRI for granting access to the data.

**Conflicts of Interest:** The authors declare no conflicts of interest. T.N.M., M.T., E.O. are employees of Nestlé Research, Vers-chez-les-Blanc, Lausanne, Switzerland. This research study is made possible to Nestlé Research via virtual access of the requested data of the 2013 National Nutrition Survey, funded by Nestlé Research, Lausanne, Switzerland.

#### **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* **Milk Polar Lipids: Underappreciated Lipids with Emerging Health Benefits**

#### **Liya Anto 1, Sarah Wen Warykas 1, Moises Torres-Gonzalez <sup>2</sup> and Christopher N. Blesso 1,\***


Received: 6 March 2020; Accepted: 2 April 2020; Published: 4 April 2020

**Abstract:** Milk fat is encased in a polar lipid-containing tri-layer milk fat globule membrane (MFGM), composed of phospholipids (PLs) and sphingolipids (SLs). Milk PLs and SLs comprise about 1% of total milk lipids. The surfactant properties of PLs are important for dairy products; however, dairy products vary considerably in their polar lipid to total lipid content due to the existence of dairy foods with different fat content. Recent basic science and clinical research examining food sources and health effects of milk polar lipids suggest they may beneficially influence dysfunctional lipid metabolism, gut dysbiosis, inflammation, cardiovascular disease, gut health, and neurodevelopment. However, more research is warranted in clinical studies to confirm these effects in humans. Overall, there are a number of potential effects of consuming milk polar lipids, and they should be considered as food matrix factors that may directly confer health benefits and/or impact effects of other dietary lipids, with implications for full-fat vs. reduced-fat dairy.

**Keywords:** polar lipids; dairy; sphingomyelin; heart disease; gut health; cancer; inflammation

#### **1. Introduction**

Polar lipids are essential components of all biological membranes and found in the human diet as phospholipids (PLs) and sphingolipids (SLs). Consumption of dietary polar lipids is relatively common in the Western dietary pattern and estimated to be in the range of 2–8 g/day for PLs (~1–10% of daily fat intake) [1] and 50–400 mg/day for SLs [2,3]. In milk, polar lipids are primarily located within the milk fat globule membrane (MFGM), which is a trilayered biological membrane that surrounds the fat globule. MFGM are typically derived from cell membranes of lactating cells and the endoplasmic reticulum membranes [4]. Animal cell membranes, in general, have PLs as the major structural lipids and SLs, such as sphingomyelin (SM), are typically components of lipid rafts in association with cholesterol [5]. Of the total milk lipids, polar lipids account for approximately 1% of total lipids in milk. SM content encompasses approximately 25% of total milk polar lipids, and SM is found at ~3:1 ratio to cholesterol by mass [6]. While health effects of PLs from eggs [7] and SLs [8] have been reviewed previously, the health effects of consuming milk polar lipids have not been reviewed extensively. Since the SLs found within MFGM are known to impact various aspects of lipid metabolism [9,10], gut microbiota [10], and inflammation [9,11], milk polar lipids may be considered as food matrix factors that may confer health benefits and/or impact effects of other dietary lipids, with implications for full-fat vs. lower-fat dairy varieties. This review summarizes the recent basic science and clinical research examining food sources and health effects of milk polar lipids, as well as to identify gaps in the scientific literature related to milk polar lipids research.

#### **2. Polar Lipids**

#### *2.1. Classes of Polar Lipids*

Phospholipids and sphingolipids are the two major classes of polar lipids in milk. PLs in milk includes SM (considered both a PL and a SL) and glycerophospholipids. The glycerophospholipids consist of phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylcholine (PC), and phosphatidylinositol (PI). Sphingomyelin is a phosphosphingolipid and is the major SL found in milk. Other SLs in milk include glucosylceramide (GluCer) and lactosylceramide (LacCer). Classification of milk polar lipids is given in Figure 1.

**Figure 1.** Classification of milk polar lipids. Abbreviations: GluCer, glucosylceramide; LacCer, lactosylceramide; PC, phosphatidylcholine; PE, phospatidylethanolamine; PI, phosphatidylinositol; and PS, phosphatidylserine.

Glycerophospholipids are amphipathic molecules comprised of a glycerol backbone, two esterlinked fatty acids (FAs) which form the hydrophobic tail (at sn-1 and sn-2 positions of glycerol), and a polar head group linked by a phosphate residue (at sn-3 position). The polar head group can be ethanolamine, serine, choline, or inositol, which makes it PE, PS, PC, or PI, respectively [12]. Fatty acids seen in glycerophospholipids are usually unsaturated and long chain. However, PC is reported to have more saturated FAs when compared with other PLs seen in milk [6]. Esterification of very-long-chain fatty acids in glycerophospholipids is also reported, although to a lesser extent than long-chain FAs [13].

Sphingolipids are polar molecules that differ from glycerophospholipids in that instead of glycerol, they contain a long-chain amino alcohol, known as a sphingoid base, as their backbone [14]. The predominant mammalian sphingoid base is the 18-carbon sphingosine. Ceramide is formed when the amino group of sphingosine is linked to a FA via an amide bond. Sphingomyelin (or ceramide phosphocholine) is an amphipathic sphingolipid in which a phosphorylcholine head group is linked to ceramide [15]. In glucosylceramide (GluCer), the 1-position of ceramide is linked to a glucose residue, whereas in lactosylceramide (LacCer), it is linked to lactose [8,16].

Previous literature has cited the presence of other minor PLs in milk, including lysophosphatidylethanolamine (LPE) and lysophosphatidylcholine (LPC) [17–19]. However, the origin of these lipids in milk is unclear, as they are the hydrolytic products of PE and PC and can be formed by hydrolysis during milk handling [12]. Another minor class of PLs seen in milk are the plasmalogens, which are structurally characterized as a glycerophospholipid having a vinyl-ether linkage with a fatty alcohol at the sn-1 position of the glycerol backbone [20,21].

#### *2.2. Biological Functions*

Phospholipids are present in all biological membranes, where they display amphipathic properties. The amphipathic nature of PLs is characterized by two hydrophobic tails and a hydrophilic, polar head group. This feature helps them to aggregate spontaneously in the aqueous phase by exposing their hydrophilic head and hiding their hydrophobic tails away from the aqueous phase, thereby attaining an energetically favorable conformation [22]. Phospholipids aggregate in two different ways; they can

either bury their hydrophobic tails in the interior forming a spherical structure, as seen in micelles, or they can sandwich their tails between the hydrophilic head groups forming sheets or bilayers, as seen in cell membranes [23].

Almost 50% of the total mass of the animal cell membrane is made up of lipids, with PLs being the most abundant. The major PLs in cell membranes include PC, PE, PS, and SM. They serve a wide variety of structural and biological functions in the cell membrane [23]. Phospholipid bilayers in cell membranes help in maintaining a semi-permeable barrier between the cells and the organelles [24,25]. Due to their amphipathic nature and cylindrical structure, the PLs in cell membrane can also demonstrate a self-healing property. A tear in the membrane can produce an energetically unfavorable form, which compels the PLs to rearrange spontaneously to heal/close the tear [23]. This specific property of PLs is fundamental to the existence of a cell. Within the lipid bilayers, individual lipids can move freely, which provides fluidity to the membrane [26]. The presence of PS in the cytosolic surface of the bilayer is important for the enzyme activity of cytosolic protein kinase C. Phospholipids are important intracellular mediators. Extracellular signals can activate various membrane phospholipases that can cleave PLs and the fragments produced can act as intracellular mediators [24,27,28]. Thus, PLs are integral parts of cellular membranes and are also critically involved in vital functions of the cell.

During lipid digestion in the small intestine, mixed micelles are formed, which also contain PLs on their surface. As previously mentioned, the spherical structure of the micelle is due to the amphipathic property of PLs. Phospholipids are arranged in micelle in such a way that the hydrophilic head group face to the exterior, while the hydrophobic tails are arranged on the inside. This particular arrangement of PLs, along with bile salts, helps in the packing of highly hydrophobic triglyceride (TG) and cholesteryl esters in the core of mixed micelle, so that they can be transported through the aqueous environment in the intestine where they can be digested and absorbed. Likewise, the emulsifying function of PL monolayers in lipoproteins derived from the intestine and liver is similar to what is observed in micelles.

#### *2.3. Milk Polar Lipids: Classes and Quantity*

Not all dairy products are created equal when it comes to PL and SL contents. There is a considerable amount of variability when comparing polar lipids to total lipid content in dairy products. For example, some fat-rich products like butter and cream have relatively low levels compared to other products. For the content of SLs, which is mostly SM, dairy products vary from trace levels of SM in anhydrous milk fat to higher levels in buttermilk (~12–21.5 g/100 g of PL) and butter serum (also called beta serum) (23.8–28.92 g/100 g of PL) [29]. The major PL and SM content of raw milk and other dairy products is summarized in Table 1. Whole milk contains about twice the PLs of skim milk, while some dairy co-products can become relatively enriched in PLs, including buttermilk and butter serum [29]. Phospholipid content in whole milk can vary between 0.7 to 2.3 g/100 g of fat [30–34]. This variation in the level of reported PL content in milk can be attributed to the different analytical techniques used, as well as the diet [35], season of the year [36], age, and stage of lactation of the animal [37,38]. The most abundant PL in milk is reported to be PE, followed by PC and SM [20,29,31,33,34,37,39–41]. Fat-rich dairy products like butter and cream are found to be less enriched in PL content as a proportion of total fat, when compared with their aqueous co-products. Cream is rich in fat (~32%) and protein (~2%) [31] and has 0.3–5.65 g of PL/100 g of fat. Most authors reported PE being the most abundant PL found in cream, followed by PC and SM [30,32,42,43]. Butter (~78% fat) [32] has a PL content ranging from 0.195–5.31 g/100 g fat. The order of abundance of PL in butter is the same as cream, with PE being the richest followed by PC and SM [30–32,44]. Buttermilk (4.485–35.32 g/100 g of fat) and butter serum (46.69–48.39 g/100 g of fat), which are the co-products of butter and anhydrous milk fat production, respectively, are rich sources of polar lipids, yet are low in fat [32]. Cheese whey is also a good source of milk PL (5.3–23.66 g/100 g fat).


**Table 1.** Milk PL and SM content of raw milk and dairy products.

Abbreviations: DM, dry matter; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PI, phosphatidylinositol; PL, phospholipid; PS, phosphatidylserine; and SM, sphingomyelin.

It has been reported that various dairy production methods (homogenization, heating, etc.) can disrupt the MFGM [30]. They can also affect the MFGM protein content in the final product and will enhance the milk PL content in the serum phase [51–53]. In addition, several reports have shown that milk products which are rich in PL are also rich in MFGM proteins [6,30,54–56]. As a consequence, it is suggested that, due to the close association of milk PL with MFGM proteins, these fractions may be migrating together during dairy processing [6]. Churning, centrifuging, homogenization, and spray-drying are some of the dairy processing techniques which are known to affect the composition of milk PLs and SLs [30]. This could be one probable reason behind this unexpected inverse relationship between fat and milk PL content in dairy ingredients and co-products. Due to the high PL content on a dry matter and fat base, buttermilk, butter serum, and cheese whey are the most suitable sources of milk PL purification. However, these ratios are much higher for butter serum when compared with buttermilk and cheese whey, which makes it the most attractive source for milk PL purification [49]. However, researchers have also successfully used buttermilk [57,58] and cheese whey [56] for the extraction of milk PLs.

#### **3. Health E**ff**ects of Milk Polar Lipids**

#### *3.1. E*ff*ects on Intestinal Lipid Absorption*

Dietary PLs can inhibit intestinal lipid absorption when added to the diet in significant amounts by interfering with lipid mobilization from mixed micelles (as reviewed by Cohn J et al. [1]). Dietary SM, in particular, is known to dose-dependently reduce the intestinal absorption of cholesterol, TG, and fatty acids in rodents [10]. Products of SM digestion, such as ceramides and sphingosine, also inhibit cholesterol and fatty acid absorption [59–61]. Compared to SM derived from egg yolk, SM from milk has been shown to have stronger effects on inhibiting the intestinal absorption of fat and cholesterol in rats, potentially due to stronger hydrophobic interactions between milk SM and other lipids [62]. Evidence of dietary SM and its hydrolytic products inhibiting intestinal lipid absorption through cell-dependent and cell-independent mechanisms is provided in previous preclinical studies [63]. Enrichment of the dietary PL pool by milk SM has shown to decrease intestinal cholesterol absorption by decreasing the active concentration of cholesterol monomers in mixed micelles [64]. In vitro studies suggest that SM in mixed micelles can reduce TG hydrolysis by inhibiting human pancreatic lipase-colipase

activity [65,66]. Long-chain bases of SLs and long-chain fatty acids also have been reported to compete with each other for cellular uptake, since they both utilize acyl-CoA synthetases [61].

Other milk polar lipids such as PC, PE, and gangliosides are also known to reduce the intestinal absorption of dietary lipids. Phosphatidylcholine, which when present in the bile can facilitate intestinal absorption of dietary lipids, can inhibit lipid absorption when present in large amounts in the diet. The presence of PC in taurocholate containing mixed micelles reduced the uptake of cholesterol by Caco-2 cells [67]. Incubation of Caco-2 cells with micelles containing 200 μM of PC reduced cholesterol absorption, accompanied by reduced cellular esterification and secretion of cholesteryl esters. In contrast, the presence of lysophosphatidylcholine showed only a minor effect. Additionally, previous research also revealed the capacity of intact PC in mixed micelles to inhibit the absorption of cholesterol and FAs using in vivo (isolated jejunal segment technique) and in vitro (everted sac experimental model from rat jejuna) studies [68]. Similar effects of PC are also reported by Rampone and Long (1977) using the same in vitro model [69]. In human trials, the administration of PC via intraduodenal infusion attenuated cholesterol absorption when compared to the placebo group, which received the same amount of safflower oil with similar FA composition [70]. Later studies showed that phospholipase A2 hydrolysis of surface PL in lipid emulsions prior to pancreatic lipase/co-lipase-mediated TG hydrolysis is necessary for the cellular uptake of cholesterol and FAs. The cellular uptake of cholesterol from lipid emulsions with high PL/TG molar ratio (>0.3) was significantly lower [71]. However, the exact mechanism by which high PC concentration reduces intestinal cholesterol absorption remains unclear. Although, a plausible mechanism can be attributed to an increased solubility of cholesterol in the micelle and, thus, shifting the partition coefficient away from the cell membrane [72]. Additionally, a higher concentration of PC in lipid emulsions may also lead to a decreased availability of cholesterol for absorption, due to the increased packing density of the micellar surface [71].

Phosphatidylethanolamine is shown to have a hypocholesterolemic effect in animal studies [73,74]. Supplementation with 2% PE for two weeks in rats fed with a 1% cholesterol-containing diet reduced serum cholesteryl ester compared to control animals, with the cholesteryl ester concentration being inversely related to the level of hepatic PE [74]. A similar effect of PE in rats was also reported by Imaizumi and co-workers [73]. It is also known that mono- and di-unsaturated PE can exhibit a similar affinity to cholesterol as that of SM and can influence cholesterol absorption [75,76]. Therefore, PE found in milk may also have the capability to reduce intestinal lipid absorption, due to its affinity for cholesterol; however, further studies are needed to confirm this.

#### *3.2. Anti-Inflammatory E*ff*ects*

Dietary SM and its hydrolytic products (i.e., sphingosine and ceramide) have shown promising anti-inflammatory action in preclinical studies. Mazzei et al. [77] showed that sphingosine can activate a peroxisome proliferator-activated receptor-gamma (PPARγ) reporter in macrophages. PPARγ is a nuclear receptor that represses the transcriptional activation of inflammatory response genes in mouse macrophages [78]. Similarly, consumption of milk SM effectively decreased disease activity and colonic inflammatory lesions in mice with chemically induced colitis, partly through PPARγ [77]. Norris et al. [11] found that milk SM was not cytotoxic to RAW264.7 macrophages at physiological dosages tested but strongly decreased LPS-stimulated pro-inflammatory gene expression. The major bioactive component in these experiments was sphingosine, as only sphingosine and sphingosine-containing SLs recapitulated the anti-inflammatory effects of milk SM. Consistent with this finding, sphingosine reduced TNF-α production from macrophages stimulated with LPS [79]. The choline moiety of milk PLs may also contribute some protection against macrophage inflammation. Choline dose-dependently reduced TNF-αrelease from macrophages stimulated with endotoxin by inhibiting NF-κB activation [80]. In mice, intraperitoneal injection of choline (50 mg/kg) prior to the endotoxin treatment significantly improved the survival rate and decreased plasma TNF-α level [80]. This anti-inflammatory effect of choline was found to be nicotinic acetylcholine receptor subunit α7 (α7nAChR)-dependent, which is

an essential component of the cholinergic anti-inflammatory pathway [80]. Collectively, these data demonstrate that milk SM has potential to be anti-inflammatory in macrophages.

Milk polar lipids and their metabolic products also appear effective against endotoxemia. Endotoxemia is a persistent sub-clinical, low-grade inflammation due to circulating endotoxins, primarily LPS, which may be absorbed into circulation due to defects in the gut barrier. Milard et al. [81] observed gut barrier effects when Caco-2/TC7 cells were treated with milk SM incorporated into mixed micelles. They observed significant inductions in the gene expression of tight junction proteins, which was related to a specific induction of interleukin-8 (IL-8) by milk SM. While IL-8 has been reported to have pro-inflammatory effects in some conditions, this study showed recombinant IL-8 specifically increased tight junction protein expression. More research is warranted to investigate the role of milk SM on IL-8 and gut barrier function. While these effects were observed with milk SM, some studies have been conducted with MFGM as a source of polar lipids. Snow et al. [82] studied the effect of MFGM on gut-barrier and systemic inflammation in LPS-challenged mice by feeding them with 10% MFGM containing diet (6 g of milk polar lipids/kg of the diet). The MFGM supplemented diet significantly attenuated LPS inducted systemic inflammation, partly by improving gut barrier integrity. Moreover, a four-week parallel intervention study in healthy adults showed that consumption of 10 g of a commercially-available MFGM-rich milk protein concentrate (Lacprodan PL-20, containing 16% PL by weight) twice daily can provide in vivo resistance to food-borne infections [83]. Norris et al. [10] was the first to report that circulating LPS activity was reduced in high-fat diet (HFD)-fed C57BL/6J mice supplemented with milk SM (0.25% *w*/*w*) for 4 weeks. Additionally, supplementation of milk SM (0.1% *w*/*w*) to an obesogenic diet dramatically reduced serum inflammatory cytokines/chemokines and mRNA expression of inflammation markers in adipose tissue in C57BL/6J mice [9,11]. Accordingly, Li et al. [84] reported that supplementation of MFGM (200 mg/kg body weight) for eight weeks attenuated HFD-induced intestinal inflammation, improved gut barrier tight junction protein expression, and reduced LPS activity and inflammation biomarkers in the circulation of C57BL/6 mice.

#### *3.3. Modulation of Gut Microbiota*

An emerging area of importance is the potential modulation of gut microbiome by dietary polar lipids [63]. Humans harbor trillions of bacteria in their gastrointestinal (GI) tract as part of the natural gut microbiota [85]. The residency of these microbes in the GI tract is closely linked with human physiology and plays a vital role in the function of the gut. The three most abundant phyla in human gut microbiome are Bacteroidetes, Firmicutes, and Actinobacteria [86]. Both Firmicutes and Bacteroidetes make up about 90% of phyla in the human gut microbiota [87]. The significance of the Firmicutes/Bacteroidetes ratio to human health has yet to be fully elucidated, yet it appears that lean individuals have a greater proportion of gut microbiota as Bacteroidetes compared to obese individuals, whereas the opposite is seen for Firmicutes [88]. Thus, the Firmicutes/Bacteroidetes ratio is often used as a marker of gut dysbiosis related to obesity and HFD consumption. Bifidobacterium belongs to the phylum Actinobacteria, and it is the most abundant bacteria seen in the gut of breastfed infants [89,90]. Lower levels of gut bifidobacteria are commonly associated with many disease conditions, including hepatitis B [91], cystic fibrosis [92], Type 1 and Type 2 diabetes [91,93], and obesity [94,95].

The SL fraction of milk polar lipids are reported to possess antibacterial effects. Sprong et al. [96] were the first to report the antibacterial effect of milk SLs. In their study, they showed the antibacterial effects of galactosylsphingosine and lysoSLs against Gram-positive and Gram-negative bacteria. In addition, Fischer et al. [97] showed that sphingoid bases (e.g., sphingosine, phytosphingosine, and dihydrosphingosine) also have broad-spectrum antimicrobial activity, which was supported by the induction of ultra-structural damage subsequent to pathogen uptake [98]. Nejrup et al. [99] conducted a 24 h in vitro fermentation study in the fecal sample of healthy infants to determine the effect of digestive products of milk lipids on modulating gut bacteria. They found that long-chain non-esterified fatty acids (LC-NEFA) with 10% sphingosine can increase bifidobacteria relative abundance in fecal content, whereas the LC-NEFA alone did not have an influence on bifidobacteria populations [99]. These in vitro findings suggest that milk SLs and their metabolic products may potentially exert changes in the gut microbiota when consumed regularly. Milk SM is digested and absorbed in the middle and distal part of the small intestine in rats and, presumably, in humans [100]. However, a large fraction of dietary SM and its digestive products reach the colon [101], where they may exert their bactericidal and gut-modulating effects. The effects of milk PLs on gut microbiota composition in both pre-clinical and clinical studies are summarized in Table 2.

Norris et al. [10] first reported the gut microbiota modulating effects of purified milk SM. Supplementation of a HFD (45% kcal as fat) with 0.25% (*w*/*w*) milk SM for four weeks increased the relative abundances of fecal Bifidobacterium, Firmicutes, and Actinobacteria, as well as reduced Bacteroidetes in C57BL/6J mice [10]. Supplementation of 0.25% (*w*/*w*) milk SM to a semi-purified low-fat diet also increased fecal Bifidobacterium in C57BL/6J mice [63]. However, a longer supplementation (10 weeks) of a lower dose of SM (0.1% milk SM) in mice fed an obesogenic diet (60% kcal as fat) had weaker effects on gut microbiota, with little change except for an increase in Acetatifactor relative abundance [11].

With SM being the most bioactive component in milk PLs, effects of the total milk polar lipids fraction on gut microbiota have shown similar results to the animal studies using purified SM. We have recently observed that feeding 2% milk PLs (containing 0.4% milk SM) to HFD-fed LDL-receptor knockout mice resulted in a similar increase in fecal Bifidobacterium relative abundance, as previously observed by Norris et al. [102]. Interestingly, there was also an increase in the Bacteroidetes phylum, which significantly reduced the Firmicutes to Bacteroidetes ratio in the 2% milk PL-supplemented group [102]. Milard et al. [103] found that supplementing an HFD with 1.6% milk PLs (0.38% SM) induced a reduction in fecal Lactobacillus in C57BL/6 mice, while 1.1% milk PLs (0.25% SM) induced an increase in Bifidobacterium compared with the HFD-fed controls. Akkermansia muciniphila, which is classified under the Verrucomicrobia phylum, was also significantly higher in milk PL-fed mice. Akkermansia muciniphila is noted for its positive metabolic effects, which includes improving insulin sensitivity and protecting against metabolic endotoxemia-induced inflammation [104–106]. In addition, there was a significant positive correlation between Bifidobacterium animalis and Akkermansia muciniphila for the milk PL-supplemented mice [103]. However, a study conducted by Reis et al. [107] did not find any effect on the gut microbiota composition by supplementing total polar lipids, PLs, or SLs in HFD-fed C57BL/6J mice. This particular experiment by Reis et al. [107] differed from other experimental designs, as mice were first fed with HFD for five weeks and then later supplemented milk polar lipids along with the HFD for an additional five weeks. As the HFD consumption likely resulted in altered gut microbiota after five weeks, the delay in polar lipid treatment may have contributed to the insignificant effects in this case. Modulation of the gut microbiota by MFGM, i.e., the milk PL-rich fraction of milk, has also been tested in rodents. The pup-in-a-cup model was used in five-day-old rats to examine the feeding of formula with fat from vegetable sources only, formula with supplemented MFGM, or mother's milk on intestinal development and the gut microbiota [108]. After 10 days, the pups fed a formula supplemented with MFGM had a more similar intestinal development and gut microbiota to those fed mother's milk when compared to those fed formula only. In another study, Li et al. [84] reported that supplementation of MFGM (200 mg/kg body weight) to C57BL/6 mice for eight weeks attenuated gut dysbiosis that occurs with HFD, including increasing the Bacteroidetes/Firmicutes ratio. Recently, modulation of the gut microbiome by ethanolamine, which is the base constituent of PE, was tested by supplementing 0, 250, 500, and 1000 μM ethanolamine to the drinking water of rats [109]. Ethanolamine supplementation at 500 and 1000 μM significantly increased Bacteroidetes and decreased Proteobacteria, Elusimicrobia, and Tenericutes. In addition, a reduction in Spirochetes was also noticed in the mice provided with 500 μM ethanolamine [109]. Overall, most pre-clinical studies in mice have noted some impact of milk polar lipids on gut microbiota composition; however, there are differences in microbiota profiles across studies. The observed differences between studies may be related to varying dosages, forms of milk polar lipids, base diet composition, or the use of prevention or treatment models.




**Table 2.** *Cont*.

 diversity Abbreviations: ACE, abundance based coverage estimator; BW, body weight; HFD, high-fat diet; LC-NEFA, long-chain non-esterified fatty acids; LDLr, low-density lipoprotein receptor; MM, mother's milk; MFGM, milk fat globular membrane; NEFA, non-esterified fatty acids; PL, phospholipid; SM, sphingomyelin, SPL, sphingolipids; and TPL, total polar lipids.

1%

MPL:

Increased

Shannon

While the pre-clinical findings reporting modulation of the gut microbiome by milk polar lipids are promising, only one human clinical trial has been conducted investigating this area. In a recent study by Vors et al. [110], post-menopausal women who supplemented their diets for four weeks with either 3 g or 5 g of milk polar lipids daily through butter serum showed no significant changes in major phylogenetic groups or bacterial species of gut microbiota when compared with the control group fed only butter oil. This may be attributed to the much lower dosage used in this study when compared to animal studies. A simple allometric approach considering the body surface area can be used to convert mice dose in mg/kg to human equivalent dose (mg/kg) by multiplying by 0.081 [111]. For example, the animal dose of milk PLs in Millar et al. [102] shown to modulate gut microbiota was ~1.25–2.5 g PLs/kg of body weight, which is equivalent to ~0.1–0.2 mg/kg of bodyweight in humans. This would equate to a dose of 7–14 g of milk PLs in a 70 kg human. However, Vors et al. [110] reported there was a significantly greater amount of fecal coprostanol, as well as a higher coprostanol/cholesterol ratio in those supplemented with milk polar lipids compared to control. While not examined further, these effects suggest there were changes in gut microbiome metabolism specific to increased coprostanol conversion in the gut with milk polar lipid supplementation. Some gut microbes are known to have the ability to convert cholesterol to coprostanol [112,113]. Research has shown an inverse relationship between blood cholesterol concentrations and the coprostanol/cholesterol ratio in human feces, suggesting the ability of coprostanol to modulate cholesteremia [114]. Thus, metagenomic effects of milk polar lipids and their influence on coprostanoligenic bacteria warrant more investigation.

Although comparisons of studies investigating gut microbiota composition are often challenging, five out of eight studies described above, including in vitro studies, showed an increase in fecal bifidobacteria by supplementing milk polar lipids at different concentrations. Strains of Bifidobacterium are commonly used as probiotic agents [115–118] and have been shown to have preventive and therapeutic effects in infant gut diseases [119] and in respiratory and gastrointestinal disorders in adults [120]. If supported in human studies, milk PLs may have potential as prophylactic or therapeutic agents against these diseases. It is quite interesting to note the differences in the changes in Firmicutes/Bacteroidetes ratio by the supplementation of milk PLs and milk SM. Norris et al. [10] reported a decrease in Bacteroidetes and increase in Firmicutes (hence an increased Firmicutes/Bacteroidetes ratio) by supplementing 0.25% milk SM for four weeks to HFD-fed mice. It is noteworthy that SLs can be produced in the gut by a small fraction of bacteria belonging to the Bacteroidetes phylum. A plausible justification may be that chronic exogenous SM supplementation may be triggering a feedback signaling pathway that is lethal to Bacteroidetes [63]. Another possibility may be related to greater amount of lipids getting to the colon of animals supplemented with high amounts of purified milk SM due to its noted inhibitory effects on lipid absorption. However, supplementing the same or higher amount of milk SM in the presence of other milk PLs did not change the Firmicutes/Bacteroidetes ratio [103] or decreased it [102]. The varying responses of the gut microbiota across studies may be due to differences in animal models or diets used, as well as the duration of milk polar lipid supplementation. While effects on gut microbiota composition observed at the phylum level are intriguing, more research is warranted in this area to investigate genusand species-level compositional changes, as well as the metagenomic effects.

Recent research also suggests that choline-containing PLs can be metabolized by gut bacteria to generate trimethylamine (TMA) and, subsequently, oxidized to trimethylamine-N-oxide (TMAO) in the liver after absorption. Many observational and metabolomic studies have reported TMAO as a predictive risk factor for cardiovascular disease (CVD) [121,122] and colorectal cancer [123]. Choline, as one of the precursors for TMA, is found in many foods as free choline or as part of phosphatidylcholine, phosphocholine, or SM. Since PC is one of the most abundant polar lipids found in milk and milk products, researchers have analyzed the association between milk consumption and TMAO production. A cross-sectional study conducted in a German adult population reported a positive association between elevated plasma TMAO levels and milk consumption [124]. In contrast, another study conducted on healthy adults (KarMeN study) found no association between milk consumption and plasma TMAO [125], while intervention studies by Zheng et al. reported lower

urinary TMAO levels in overweight women on high dairy diets [126] and adult men on both high milk and high cheese diets [127].

Current evidence appears insufficient to associate milk polar lipids and milk to high plasma or urinary TMAO levels. The direct effects of elevated plasma TMAO levels in promoting CVD risk is also controversial. Fish and seafoods are the rich sources of TMAO and TMA [128]. Many reports associate high fish intake to elevated plasma TMAO levels [128,129]. If Plasma TMAO can increase the risk of CVD, it should be speculated that high fish intake can increase the risk of CVD. However, epidemiological and observational studies report a protective effect of fish consumption on CVD risk [130–133]. On the other hand, fish oil has reported to ameliorate the adverse effects caused by TMAO in HFD fed mice [134]. A recent Mendelian randomization trial suggests that type 2 diabetes and kidney disease can increase circulating TMAO and evidence for the association between TMAO and CVD in observational studies may be due to reverse causality or confounding [135]. Overall, current evidence suggest choline-containing milk PLs may be used as substrates for TMA generation by gut microbiota, but the effect on disease risk is unclear. Fortunately, a number of studies have investigated the effects of milk PLs on other CVD risk factors.

#### *3.4. Cardiovascular Disease*

Cardiovascular disease (CVD) remains the most prominent contributor to mortality in the United States [136]. Atherosclerosis, which is characterized by the deposition of fatty plaque in the inner walls of the arteries, is a key player in the development of CVD. Dietary modification is recommended as a primary prevention strategy for managing blood lipid levels to reduce the risk for CVD [137]. The health effects of milk PLs on serum lipid levels are summarized in Tables 3 and 4. Genetically-obese KK-Ay mice displayed significant reductions in plasma LDL-cholesterol when fed a diet supplemented with SL-concentrated butter serum (0.35% SLs in diet by weight) or milk-derived ceramides (0.35% *w*/*w*) [138]. The feeding of milk SM (~0.25% *w*/*w* of diet) has been shown to significantly reduce serum cholesterol by ~15–25% when fed to C57BL/6 mice consuming both milk fat-enriched [10,139] and low-fat diets [63]. In humans, milk SM also shows potential to improve serum lipids. In a single-blind, randomized, controlled isocaloric parallel study, Rosqvist et al. [140] observed that an eight-week consumption of 40 g milk fat/day as whipping cream (rich in MFGM) in overweight adults resulted in lower plasma LDL-C, non-HDL-C, and apoB:apoA-I ratio compared to the same amount of milk fat as butter oil (free of MFGM). Conway et al. [141] reported that ingestion of 45 g/day of buttermilk for four weeks resulted in reductions in plasma cholesterol and TG compared to placebo, in a double-blind randomized study of healthy adults. The lower plasma LDL-C concentrations observed with buttermilk were associated with changes in plasma β-sitosterol, a marker of cholesterol absorption. In a clinical trial in postmenopausal women by Vors et al. [110], a dose of 5 g/day of milk PLs (1.3 g/day milk SM) via a butter serum concentrate lowered total and LDL-C, as well as decreased total/HDL-cholesterol and decreased apoB:apoA-I ratio, compared to a control cream cheese devoid of milk PL. Overall, it appears that the hypolipidemic effects of milk PLs observed in animal studies were also observed in several human studies, even when using lower dosages than in animal studies.

We have previously reported that supplementing purified milk SM at both 0.25% and 0.1% (*w*/*w* of diet) attenuated dyslipidemia and inflammation in HFD-fed C57BL/6J mice. Our research group [142] and others [143] have previously reported that dietary egg SM could attenuate atherosclerosis development in apoE−/<sup>−</sup> mice, even in the absence of changes in serum lipids. In a recent study, we report that supplementation of 2% (*w*/*w*) milk PLs to LDLr−/<sup>−</sup> mice fed a milk fat-rich diet strongly reduced atherogenic lipoprotein cholesterol, modulated gut microbiota, modestly lowered inflammatory markers, and markedly attenuated atherosclerosis development [102]. Milk polar lipids were provided by supplementing diets with butter serum, a dairy co-product rich in both PLs and SLs. Thus, due to potential beneficial effects on both serum lipids and inflammation, milk polar lipids may be important to consider when choosing foods for the prevention of CVD.

#### *3.5. Non-Alcoholic Fatty Liver Disease (NAFLD)*

NAFLD is the most common cause of chronic liver disease worldwide [144]. In Western societies, it affects 20%–30% of the general population and over 75% of obese individuals [145]. The "two-hit hypothesis" model of NAFLD states that the disease progresses in a stepwise manner, with a "first hit" from obesity and insulin resistance resulting in hepatic lipid accumulation, i.e., hepatic steatosis. Subsequently, a "second hit" in the form of oxidative stress and inflammation promotes liver injury and fibrosis [146]. In laboratory animals, dietary SM appears useful in preventing hepatic lipid accumulation (as extensively reviewed previously by Norris and Blesso [15]). The health effects of milk PLs on hepatic lipid metabolism are summarized in Table 3. Previous research in our laboratory has reported that HFD-fed mice supplemented with milk SM (0.25% *w*/*w*) had reduced hepatic TG after fouur weeks, compared to HFD-fed control animals [10]. In addition, our experiments investigating 10-week supplementation of a lard-based HFD (31% lard, 0.15% cholesterol by weight) with either 0.1% (*w*/*w*) milk SM or egg SM significantly attenuated the development of hepatic steatosis and adipose tissue inflammation in C57BL/6J mice [9]. Moreover, gene expression analysis revealed lower hepatic mRNA expression of stearoyl-CoA desaturase-1, indicating a reduced capacity for hepatic TG synthesis. Likewise, Cohn and colleagues have also noted that chronic supplementation of Western-type diets (21% butter fat, 0.15% cholesterol by weight) with various milk PL extracts (0.25–0.35% SM *w*/*w* of diet) significantly attenuated hepatic cholesterol and TG accumulation in mice. Recently, we have reported that supplementing milk fat-based diets (21% milk fat by weight) with milk PLs (at 1% and 2% *w*/*w*) resulted in significantly lower hepatic cholesterol concentrations in LDLr−/<sup>−</sup> mice, although no effects were seen in hepatic TG content [102]. For human studies, there is limited data available evaluating the effects of milk PLs on NAFLD-related markers, although one study by Weiland et al. [147] found beneficial effects of 2–3 g/day of milk PLs for 7–8 weeks on serum γ-glutamyl transferase (GGT), a marker of fatty liver disease, with no changes in alanine transaminase and aspartate transaminase (markers of liver injury) in two separate clinical trials of overweight or obese men.






sphingomyelin-rich

 fraction; SPL, soybean polar lipids; TC, total cholesterol; TG, triglyceride; and VLCFA, very long-chain fatty acids.






GS, gangliosides; HDL-C, high-density lipoprotein cholesterol; HOMA-IR, homeostatic model assessment of insulin resistance; hs-CRP, high-sensitivity C-reactive protein; IL-6, interleukin -6; LDL-C, low-density lipoprotein cholesterol; MFGM, milk fat globule membrane; PC, phosphatidylcholine; PL, phospholipids; sICAM, soluble intercellular adhesion molecule; SL, sphingolipids; SM, sphingomyelin; TAG, triacylglycerol; TC, total cholesterol; TG, triglyceride; and VLCFA, very long chain fatty acid.

#### *3.6. Insulin Resistance and Type 2 Diabetes*

Insulin resistance occurs when the body's cells cannot effectively import glucose in response to the release of endogenous or exogenous insulin within the bloodstream [162]. Insulin resistance can cause hyperglycemia, which makes it a risk factor for the development of type 2 diabetes mellitus (T2DM). Diabetes is a major cause of death in the U.S. and, also, contributes to significant comorbidities related to micro- and macrovascular complications, including CVD, NAFLD, kidney disease, and blindness [163]. While studies specifically examining the effects of milk polar lipids in T2DM models are lacking, the health effects of milk polar lipids on insulin resistance and glycemia are summarized in Table 5.

Nagasawa et al. [164] observed the effects of dihydrosphingosine (DHS) on activating GPR120, a receptor expressed by enteroendocrine cells that promote the secretion of the incretin GLP-1. DHS, along with phytosphingosine, was shown to strongly activate GPR120 in vitro, although sphingosine did not. Interestingly, milk SM is known to have more saturated sphingoid backbones than other SM sources [165] and, thus, would provide DHS as a hydrolytic digestive product. GLP-1 is known to have inhibitory effects on insulin resistance and T2DM; thus, more research should investigate the effects of milk polar lipids on regulating incretin production.

In contrast to in vitro effects, results from in vivo studies have been less promising. Yamauchi et al. [153] studied the effects of SM supplementation in male obese/diabetic KK-Ay mice. Mice that were supplemented with 1% (*w*/*w*) milk SM in several low-fat diets (7% by weight as lard, soybean oil, or linseed oil) for four weeks showed no significant effects in body weight, adiposity, or blood glucose compared to control groups. Correspondingly, there were no significant differences in blood glucose concentrations when measured at various time points (days 0, 7, 14, 21, and 28). Similarly, Norris et al. [9] reported that blood glucose and HOMA-IR were not significantly affected by 0.1% (*w*/*w*) milk SM supplementation in C57BL/6J mice fed an obesogenic HFD for 10 weeks, although egg SM was shown to significantly reduce fasting glucose in the same study. Weiland et al. [147] observed the effects of three different milk interventions administered to overweight and obese men. Within this report, there were two double-blind parallel-group trials that occurred involving PL-enriched milk supplementation in overweight/obese men. Trial 1 consisted of administering milk enriched with 2 g milk PL or 2 g milk fat (control) to 62 male participants over an eight-week time period. Trial 2 consisted of administering milk enriched with 3 g milk PL or 2.8 g soy PL to 57 male participants over a seven-week time period. The overall results showed a reduction in waist circumference in participants that received 2 g of milk-PL (intervention) in Trial 1 when compared to those that received the control. However, there were no differences in fasting glucose, insulin, or the insulin sensitivity index in both trials.



#### *3.7. Cognitive Function and Neurodevelopment*

There is a growing interest in the potential health benefits of milk PLs on neurodevelopment. The development of an infant's brain starts at two weeks of conception and stops when the individual has reached about 20 years of age or early adulthood [166]. As galactosylceramide (cerebroside) content of myelin is important for central nervous system function [167], it is hypothesized that milk polar lipids may influence cognition when introduced into the diet at a young age. The health effects of cow's milk polar lipids on cognitive function and neurodevelopment are summarized in Table 6.

Oshida et al. [167] studied the effects of cow's milk SM supplementation on l-cycloserine (LCS) (an inhibitor of myelination) treatment in male Wister rat pups. Rat pups were divided into two treatment groups that were either administered LCS treatment only (control) or LCS + cow's milk SM (0.81% *w*/*w* of diet) for 28 days. This study found an increase in brain weight and CNS myelin dry weight in the LCS + SM group when compared to the LCS only group. Gurnida et al. [168] examined the effects of infant formula supplemented with complex milk lipids and gangliosides on Griffith Scale values and serum ganglioside levels in infants. The treatment group consisted of infants that were given a supplemented infant formula with complex milk lipids and gangliosides (11–12 μg/mL) (derived from cow's milk), while the control group consisted of infants that were given unsupplemented, standard infant formula. The reference group consisted of infants given breastmilk only. The infants received the treatment or control products starting at two to eight weeks of age and consumed them until six months of age. Overall, findings showed an increase in Griffith Scale values within the treatment group when compared to the control group. Correspondingly, the Griffith scale values were comparable to the values of a reference group of breastfed infants. Additionally, an increase in serum GM3, GD3, and total ganglioside levels were observed in the treatment group when compared to the control group.

#### *3.8. Colorectal Cancer and Colitis*

While milk polar lipids, particularly SM and other SLs, have shown promise in controlling inflammation and modulation of gut microbiota, there has also been interest in their effects on chronic diseases of the GI tract, such as colorectal cancer and inflammatory bowel disease (IBD). Colorectal cancer is the second leading cause of cancer-related deaths in the United States [169]. Meanwhile, IBD has rapidly expanded in both Western civilizations and newly industrialized nations in the 21st century [170]. In North America, over 1.5 million people suffer from IBD, which includes ulcerative colitis (UC) and Crohn's disease (CD) [170]. These two diseases of the gut are linked, as patients with IBD have a greater risk of developing colorectal cancer [171]. The health effects of milk PLs on colorectal cancer and colitis are summarized in Table 7.




**7.** Studies examining e ffects of milk polar lipids on colon cancer and colitis.

**Table** 

#### *Nutrients* **2020**, *12*, 1001

Kutchta-Noctor et al. [173] observed the effects of buttermilk, containing SM, lactosylceramide (LacCer), and ceramide, on growth inhibition of SW480 human colon cancer cells and noncancerous fetal human colon (FHC) cells. They reported that buttermilk containing SM and LacCer led to growth inhibition of SW480 cells and was selective towards cancer cells, with no effect on FHC cell growth. Early experiments reported decreased aberrant crypt foci formation in CF1 mice treated with 1,2-dimethylhydrazine (DMH) when supplemented with milk SLs in the diet [174–176]. In terms of the control groups, CF1 mice were fed a semi-purified diet (A1N76A) consisted of 0–0.005% SL content throughout the various studies [174–176]. A 66–70% decrease in aberrant colonic crypt appearance was observed by Schmelz et al. [176] when mice were supplemented with 0.1% SM by weight. Similar findings were observed by Dillehay et al. [175], who reported decreased incidence of DMH-induced colon tumors in mice fed 0.05% milk SM (*w*/*w*) when compared to the control group. Schmelz et al. [174] showed that DMH-injected CF1 mice supplemented with 0.025%–0.1% (*w*/*w*) of glucosylceramide (GluCer), LacCer, or GD3 had a measurable decrease in aberrant crypt foci formation of >40%. When compared to milk SM supplementation, GluCer, LacCer, and GD3 supplementation yielded similar decreases in aberrant crypt foci formation in DMH-injected CF1 animals. Snow et al. [177] reported a significant reduction in aberrant crypt foci when AMF-containing diets were supplemented with MFGM in Fischer-344 rats. Mazzei et al. [77] studied the effects of 0.1% (*w*/*w*) milk SM supplementation on azoxymethane (AOM) and dextran sulfate sodium (DSS)-induced colitis and colon tumor formation in PPARγ+/+ and PPARγ−/<sup>−</sup> C57BL/6 mice. Compared to AOM + DSS control animals, those fed milk SM had reductions in disease activity index and colonic inflammatory lesions, with greater effects in PPARγ-expressing animals. Additional findings show decreased AOM-induced colon tumors only in PPARγ−/<sup>−</sup> mice fed SM. Thus, pre-clinical studies show promising effects of milk SLs on attenuating colitis and colorectal cancer. Clinical studies are necessary to find out if these favorable effects also translate to humans.

#### **4. Gaps in Scientific Literature and Future Directions**

Significant research has been conducted to understand the health benefits of phytochemicals, whereas less is known about zoochemicals, such as SM found in milk polar lipids. Our research laboratory as well as others have shown in animal studies that milk polar lipids may impart health benefits through lowering blood cholesterol, inflammation, and altering gut bacteria. However, very little research has been conducted to investigate how milk polar lipids affect lipoprotein profiles, inflammation, and gut health in men and women at risk for CVD. Vors et al. [110] recently reported that postmenopausal women who consumed 5 g/day of milk polar lipids had beneficial effects on plasma lipids and increased the coprostanol/cholesterol ratio in feces. While crucial data were revealed with this important study, there are many questions that remain. For example, (1) what are the impacts of chronic intake of milk polar lipids on systemic and intestinal inflammation? Additionally, (2) how are lipoprotein particle characteristics that are known predictors of CVD risk (e.g., LDL size and HDL particle number) affected by milk polar lipids? While Vors et al. [110] did not observe changes in major bacterial phyla composition after four weeks (i.e., who is there), the increased coprostanol conversion in feces with milk polar lipid consumption suggests significant changes in gut microbiome metabolic capacity; thus, (3) are there effects of milk polar lipids on the metagenome of gut bacteria (i.e., what they are doing)? Finally, while the study by Vors et al. [110] was conducted in moderately hyperlipidemic, overweight post-menopausal women, (4) do milk polar lipid health effects differ between men and women? Further research should address these important unresolved questions.

It is quite challenging to arrive at a comparative assessment and extrapolate in vivo experiments across different mammalian models. Apart from the category of polar lipid used in different studies, some of the confounding factors could be partly due to the differences in diet, dosage, and duration of the studies across different species. A significant number of studies reviewed in this paper used SM, SLs, milk PLs, total polar lipids, or MFGM as the source of milk polar lipids. When interpreting the results of studies that used total polar lipids or milk PL, the likelihood of the synergistic effects of

different polar lipids should be considered. Unlike SM studies, the effect of SLs in animal models could also be attributed to the various fractions involved. Likewise, in studies that used MFGM the effect of membrane proteins, glycoproteins, and gangliosides should be considered as potential bioactive compounds influencing the results.

#### **5. Conclusions**

In this review, we evaluated the potential health effects of milk PLs in humans by examining in vitro and in vivo studies. Milk PLs were shown to favorably influence health in relation to inflammation, CVD, NAFLD, gastrointestinal diseases, and neurodevelopment, with most effects observed in pre-clinical studies (Figure 2). As described above, this can be attributed to the much lower dosages used in human studies when compared to animal studies and more clinical studies with higher doses are needed to confirm these effects in humans. Evidence from such studies may further support the development of "designer" dairy products rich in milk PLs and SLs to enhance value and promote health. Additionally, inexpensive dairy co-products rich in PLs, such as butter serum or buttermilk, could be promoted and utilized as value-added sources of milk PLs. Overall, Milk PLs are emerging as commonly consumed dairy matrix components that may be important to consider when planning diets for the prevention of chronic disease.

**Figure 2.** Functional properties of dietary milk polar lipids on various organs. Dietary milk polar lipids appear to have local effects in the GI tract on gut microbiome, colon health, and lipid absorption. Although the reported effects of milk polar lipids on gut microbiome are quite variable, a consistent finding in most studies is an increase in bifidobacterial population. Broken black lines indicate the hypothetical contribution of gut modulating effect of milk polar lipids on changes seen in other organs. Solid blue lines indicate the known underlying mechanisms by which milk polar lipids exert their systemic effects. Abbreviations: FA, fatty acids; TC, total cholesterol; TG, triglycerides; LDL-C, low-density lipoprotein-cholesterol; non-HDL, non-high-density lipoproteins.

**Author Contributions:** Conceptualization, M.T.-G. and C.N.B.; methodology and literature search, L.A. and S.W.W.; writing—original draft preparation, L.A. and S.W.W.; writing—review and editing, M.T.-G. and C.N.B.; supervision, M.T.-G. and C.N.B.; and project administration, M.T.-G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by a research grant (#3085) to C.N.B. by National Dairy Council.

**Conflicts of Interest:** M.T.-G. is a National Dairy Council employee. C.N.B. has received research funding from National Dairy Council. L.A. and S.W.W. 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/).

## *Protocol* **Galacto-Oligosaccharide RP-G28 Improves Multiple Clinical Outcomes in Lactose-Intolerant Patients**>

**William Chey 1,\*, William Sandborn 2, Andrew J. Ritter 3, Howard Foyt 3, M. Andrea Azcarate-Peril <sup>4</sup> and Dennis A. Savaiano <sup>5</sup>**


Received: 25 February 2020; Accepted: 3 April 2020; Published: 10 April 2020

**Abstract: Background and Aims:** Lactose intolerance (LI) is a global problem affecting more than half of the world's population. An ultra-purified, high-concentration galacto-oligosaccharide, RP-G28, is being developed as a treatment for patients with LI. The efficacy and safety of RP-G28 in reducing symptoms of lactose intolerance were assessed in a blinded, randomized, placebo-controlled trial. **Methods:** In this multiclinical site, double-blinded, placebo-controlled trial, 377 patients with LI were randomized to one of two doses of orally administered RP-G28 or placebo for 30 days. A LI test and symptom assessment were performed at baseline and on day 31. The primary endpoint was a ≥4-point reduction or a score of zero on LI composite score on day 31. Voluntary milk and dairy intake and global outcome measures assessed patients' overall treatment satisfaction and quality of life before therapy and 30 days after therapy. This study received Institutional Review Board (IRB) approval. **Results:** For the primary endpoint, 40% in the RP-G28 groups reported a ≥4-point reduction or no symptoms on LI symptom composite score compared to 26% with placebo (*p* = 0.016). Treatment with RP-G28 also led to significantly higher levels of milk and dairy intake and significant improvements in global assessments compared to placebo. RP-G28 but not placebo led to significant increases in five *Bifidobacterium* taxa. **Conclusions:** RP-G28 for 30 days significantly reduced symptoms and altered the fecal microbiome in patients with LI. Treatment with RP-G28 also improved milk/dairy consumption and quality of life and was safe and well tolerated.

**Keywords:** lactose intolerance; lactase non-persistence; galacto-oligosaccharide; gut microbiome; abdominal pain; bloating; gas; diarrhea

#### **1. Introduction**

Non-persistence of the lactase enzyme in the small intestinal mucosa (LNP) affects 65–70% percent of the population globally, impacting health and potentially causing distressing gastrointestinal (GI) symptoms, which are commonly referred to as lactose intolerance (LI) [1]. LNP results in lactose malabsorption, which allows undigested lactose to move into the colon. Fermentation of lactose in the colon can produce carbon dioxide, hydrogen gas, methane and short-chain fatty acids, leading to a range of abdominal and bowel-related symptoms that can include abdominal pain, cramping, discomfort, bloating, distension, flatulence, increased stool frequency and/or loose or watery stools [2].

Patients with LI often avoid dairy products. Prolonged restriction of dairy can result in insufficient dietary calcium and vitamin D, which can, in turn, lead to osteoporosis, osteomalacia, and hypertension [3,4]. The daily calcium intake in patients with LI, ranging from 320 to 388 mg/day, is significantly less than the recommended amount of 1000–1200 mg/day [5,6].

There is accumulating evidence that treatments which alter the gut microbiome can improve a wide range of diseases, either by reversing disease-associated alterations in the microbiome or "dysbiosis" or by modifying the "normal" gut microbiome. One example of this latter strategy includes galacto-oligosaccharides (GOSs), which pass intact through to the colon, where they stimulate the growth of lactose-metabolizing bacteria. Elevated populations of *Lactobacillus* and *Bifidobacterium* enhance β-galactosidase activity and GOS utilization [7,8], resulting in enhanced fermentation of lactose to glucose, galactose, and short-chain fatty acids as well as reduced lactose-derived gas production, which could be beneficial to the symptoms of LI [9].

RP-G28, an ultra-purified, high-concentration (>95%) galacto-oligosaccharide, has been evaluated in a Phase 2a study in 85 patients [2]. RP-G28 reduced abdominal pain in 50 percent of patients after treatment and 30 days post-treatment (*p* = 0.019). Additionally, RP-G28-treated patients were 6 times less likely than patients given placebo to report LI 30 days post-treatment after reintroduction of dairy foods (*p* = 0.039).

The aim of the current study was to further evaluate the efficacy and safety of two doses of RP-G28 in patients with LI in a larger, multicenter, randomized, double-blind, placebo-controlled, parallel group trial.

#### **2. Materials and Methods**

#### *2.1. Study Overview*

A multicenter, randomized, double-blind, placebo-controlled, parallel group clinical trial was conducted to determine the efficacy, safety, and tolerability of two doses of RP-G28 in subjects with moderate to severe LI. The study took place between March and October 2016 at 15 investigative centers throughout the U.S. and included a 7 day screening, a 30 day treatment, and a 30 day post-treatment "real-world" observation, during which dairy was re-introduced to patients' diets (Figure 1).

**Figure 1.** Protocol Design.

The clinical trial consisted of 3 phases: a screening phase, a treatment phase and a post-treatment phase. Patients were screened for LI for 7 days prior to study. Patients were then stratified into higher and lower-dose RP-G28 treatment or placebo for a 30 day treatment phase, during which patients did not consume lactose. On day 31, post-treatment, LI symptom assessments were made. Following this, "real-world" dairy intake began, and LI symptoms were assessed over a 30 day period.

Study patients, investigators, study site staff, the sponsor, the medical monitor and the study monitors were blinded to the treatment during the trial. All sites obtained Institutional Review Board (IRB) approval. This investigation was carried out in accordance with the Declaration of Helsinki of the World Medical Assembly and its revisions, as well as the rules of Good Clinical Practice (GCP) of the United States FDA (Protection of Human Subjects, 21 CFR 50; IRB, 21 CFR 56; and IND, 21 CFR 312). IRB approval for protocol G28-003 was approved on 26 January 2016, with approval of Amendment #1 on 5 May 2016, and Amendment #2 on 17 June 2016. This clinical trial can be found on the clinical trial registry website (www.clinicaltrials.gov), trial number NCT02673749.

The primary and secondary endpoints provided a comprehensive evaluation of the effect of treatment on the symptoms of LI, milk and dairy consumption, quality of life experiences, and fecal microbiome to assess meaningful treatment benefits.

#### *2.2. Screening Period*

To establish eligibility, a pre-screening questionnaire was administered to confirm patient perception of LI. Eligible patients were required to have a moderate to severe symptom severity score, and a positive Hydrogen Breath Test (HBT) following a standardized in-clinic lactose challenge. The criteria for diagnosis of LI, and the main criteria for inclusion into the study, are provided in Table 1.


**Table 1.** Diagnosis and Main Criteria for Inclusion.

A 5 h HBT at baseline confirmed LNP in patients. Hydrogen, methane, and carbon dioxide concentrations were measured in exhaled breath following a single-blinded lactose challenge three times during the study. Anhydrous food-grade lactose was administered at 0.35/kilogram body weight on day 0 (baseline, in order to confirm that the patient had symptoms of LI for eligibility), on day 31 (end-of-treatment, primary endpoint), and on day 61 (end of post-treatment period). After fasting overnight, patients were assessed for symptoms of LI using a validated symptom questionnaire (LI Symptom Questionnaire) to create a composite symptom score (comprised of abdominal pain, cramping, bloating and gas) at 6 time points starting at 30 minutes and hourly thereafter for 5 h.

Subjects who met the eligibility criteria completed a global assessment questionnaire and a 7 day lactose consumption assessment recall, provided stool samples, and were randomized to 1 of 2 dose regimens (higher or lower) of RP-G28 or placebo in a 1:1:1 ratio via a centralized, randomized Interactive Response (IXR) system using the subject ID number assigned at the beginning of screening. The IXR system was then used to assign drug kit numbers.

Subjects were asked to refrain from ingesting lactose-containing beverages/foods during the treatment period and record LI symptoms and adverse events (AEs) on a daily basis. The first dose of the study drug was administered on day 1 and the final dose on day 30. Subjects were followed for an additional 30 days after the final dose, during which time subjects were encouraged to ingest lactose-containing foods and record symptoms of LI daily. The demographics and characteristics of the study population are provided in Table 2.

#### *2.3. Treatment Dose and Period*

The lower dose of RP-G28 was 5 grams twice daily on days 1–10 followed by 7.5 grams twice daily on days 11–30. The higher dose of RP-G28 was 7.5 g twice daily for days 1–10, followed by 10 g twice daily on days 11–30. The placebo (powdered corn syrup that matched the consistency, color, sweetness, and taste of the drug) was administered in a blinded matching packet.


**Table 2.** Demographics and Baseline Characteristics (mITT Population).

BMI = body mass index; max = maximum; min = minimum; mITT = modified intent to treat; *N* = number of subjects; SD = standard deviation; kg = kilogram.

#### *2.4. At 30 Days Post-Treatment "Real-World" Observation Period*

The amount of lactose consumed each day for 7 days pre-treatment, and for 30 days post-treatment, was assessed via a food diary. Further, during the 30 day post-treatment period, global patient assessment questionnaires measured feelings, experiences and dietary changes resulting from treatment (Table 3). Global assessments are widely accepted as qualitative tools to evaluate the efficacy of treatment for functional disorders of the gastrointestinal tract [10].

#### *2.5. Gut Microbiome Analysis*

The 16S rRNA amplicon sequencing was performed utilizing patient stool samples collected on days 0, 31 and 61. Amplification using universal primers targeting the V4 region of the bacterial 16S rRNA gene was performed on 12.5 nanograms of total DNA from collected samples. <sup>10</sup> Each 16S rRNA gene amplicon was purified using the AMPure XP reagent (Beckman Coulter, Indianapolis, IN). Next, each sample was amplified using a limited cycle Polymerase Chain Reaction (PCR) program, adding Illumina sequencing adapters and dual-index barcodes (index 1(i7) and index 2(i5)) (Illumina) to the

amplicon target. The final libraries were again purified using the AMPure XP reagent, quantified and normalized prior to pooling. The DNA library pool was then denatured with NaOH, diluted with hybridization buffer and heat denatured before loading on the MiSeq reagent cartridge and on the MiSeq instrument (Illumina). Automated cluster generation and paired-end sequencing with dual reads were performed according to the manufacturer's instructions.



1. NRS = Numerical Rating Scale.

The 1050 DNA samples corresponding to 345 subjects receiving placebo, lower-dose or higher-dose treatments at three time points (days 0, 31, and 61) were analyzed by high-throughput quantitative PCR (qPCR) targeting *Bifidobacteria* and *Lactobacilli* using specific 16S rRNA gene and GroEL probes [11,12]. Microfluidic qPCR was performed using a BioMark HD reader (Fluidigm Corporation, San Francisco, CA) with a Dynamic Array 24.192 chip processed following manufacturer's instructions.

#### *2.6. Statistical Analysis*

#### 2.6.1. Primary Efficacy Endpoint

A LI Symptom Questionnaire was developed, validated and applied during this clinical trial. This questionnaire rated individual symptoms of LI on an 11-point Numerical Response Scale (NRS) as well as a Verbal Rating Scale (VRS).

The primary efficacy endpoint was the proportion of responders on day 31. A responder was defined as a patient with a reduction from baseline in the composite score of 4 points or greater or a composite score of 0 (i.e., symptom resolution) on day 31. The composite score was calculated from the maximum symptom scores for each symptom (abdominal pain, cramping, bloating, and gas movement) after a lactose challenge test. Each symptom was rated on a 10-point Likert-type scale over 5 h after a lactose challenge. Each maximum symptom was then averaged into a composite score ranging from 0 to 10, where 0 indicated no symptoms and 10 indicated symptoms at their worst. A 4-point change was considered a meaningful improvement in symptoms of LI, based on psychometric analysis (blinded review of the clinical data prior to unblinding) and two rounds of cognitive interviews (N = 30 and 23). In the cognitive interviews, a 4-point change was meaningful for 84% of the subject responses and 53% of the subject responses in the first and second round of interviews, respectively. Symptom resolution was also considered meaningful. The 4-point threshold was further supported by empirical cumulative distribution functions using patient global severity anchor and was associated with sensitivity of 71%, specificity of 68%, positive predictive value of 63%, and negative predictive of 75%.

Statistical analysis used two-tailed tests at the α = 0.05 level of significance. The proportions of responders on day 31 were analyzed by a stratified (by quartile of baseline LI symptom composite score) Cochran–Mantel–Haenszel (CMH) test comparing both doses combined versus placebo, higher dose versus placebo, and lower dose versus placebo. Prior to unblinding, the protocol was modified to specify that the primary endpoint was a comparison of the 2 active arms combined versus placebo, using a two-sided test at the α= 0.05 level of significance. Pooling the data provided greater statistical power. The modified intent to treat (mITT) analysis population was all randomized subjects who received at least 1 dose of drug.

The sample size of 372 subjects was designed with the standard deviation of percentage abdominal pain reduction being 50%. It was hypothesized that the mean percentage reduction reported would be 50% for placebo-treated subjects and 70% for actively treated subjects. A simulation using Dunnett comparisons indicated that with 113 evaluable subjects in each arm, there would be 90% power to detect at least 1 of the active arms to be superior to placebo. The primary endpoint was based on a dichotomization of the distribution of the composite symptom score, which was thought to have at least as much power as percentage abdominal pain reduction.

The SAS software version 9.4 was used.

#### 2.6.2. Secondary Efficacy Endpoints

A number of secondary analyses were conducted. The proportion of participants in each group reporting no symptoms by the lactose intolerance composite score as well as the individual symptoms of abdominal pain, cramping, bloating, and gas movement after lactose challenge on day 31 was determined. The amount of milk consumed by participants on day 61 after RP-G28 or placebo was assessed. Global endpoints as defined in Table 3 were also compared on day 61 after RP-G28 or placebo. Analysis was performed by a stratified Cochran–Mantel–Haenszel (CMH) test for 3 comparisons: both doses combined versus placebo, higher dose versus placebo, and lower dose versus placebo. Analyses were conducted using two-sided tests at the α = 0.05 level of significance.

#### *2.7. Gut Microbiome Analysis*

Paired-end fastq files were joined into a single multiplexed, single-end fastq using the software tool fastq-join. Demultiplexing and quality filtering were performed on the joined results. Quality analysis reports were produced using the FastQC software [13]. Bioinformatics analysis of bacterial 16S amplicon sequencing data was conducted using the Quantitative Insights into Microbial Ecology (QIIME) software at a 25,000 reads/sample depth [14]. Operational Taxonomic Unit (OTU) picking was performed on the quality filtered results using pick\_de\_novo\_otus.py. Chimeric sequences were detected and removed using ChimeraSlayer [15]. Summary reports of taxonomic assignment by sample and all categories were produced using QIIME summarize\_taxa\_through\_plots.py and summarize\_otu\_by\_cat.py. The script group\_significance.py was used to compare taxa frequencies in sample groups and to determine whether there were statistically significant differences between abundances in the different groups. The non-parametric Analysis of Variance (ANOVA) test (Kruskal–Wallis) with False Discovery Rate (FDR) correction was used to compare treatments and placebo groups. For high-throughput qPCR data, the relative proportion of *Bifidobacterium* and *Lactobacillus* species was computed based on the Livak method. Quantitation cycle (Cq) values for each sample were normalized against the Cq value for the universal primers. Fold differences were calculated by 2 –ΔΔCt. Paired t-test and ANOVA with Tukey tests were conducted to assess statistically significant differences between groups.

#### *2.8. Safety*

All treated subjects were included in the safety analyses using summary statistics by treatment group of AEs, concomitant medications, vital signs, physical examinations, and clinical laboratory measurements. AEs were reported from initiation of treatment through 30 days of the post-treatment

period. An AE was classified as a Severe Adverse Event if it interfered significantly with the patients' usual functions.

#### *2.9. Irregular Site*

A for-cause audit of one study site (out of 15 sites) was conducted due to significant data irregularities. The audit found significant deviations from the protocol and failure to comply with regulatory requirements for Good Clinical Practice (GCP), and identified protocol deviations including medical history and concomitant medications, inclusion/exclusion entry criteria, and administration of the Hydrogen Breath Test (HBT), lactose challenge, and patient diaries. Compared to the other sites, there were significant differences in multiple baseline symptom scores at this site. Patients reported two times more dairy intake prior to entry into the trial compared to subjects at other centers (*p* = 0.04). In addition, patients reported higher symptom severity scores during the in-clinic lactose challenge compared to the other site's subjects (*p* = 0.035). High milk consumption in addition to high symptom severity scores based on a blinded in-clinic lactose challenge is inconsistent with patients with LI. Further, the screen failure rate was also significantly lower at this center compared to other centers.

Thus, efficacy analyses were conducted for both the mITT population which included all randomized subjects that received at least one dose of the drug and a mITT Efficacy Subset population, which included all randomized subjects that received at least one dose of the drug, excluding those who were enrolled at the site with numerous GCP violations.

#### **3. Results**

A total of 1398 subjects were screened for LI, 377 subjects from 15 study sites were enrolled and randomized (127 lower dose, 123 higher dose and 127 placebo), and 344 (87% placebo, 92% lower dose, 94% higher dose) completed the study. A high screen failure rate (>70%) was expected based on the rigorous inclusion and exclusion criteria required, including testing positive for LNP from a HBT and meeting LI symptom thresholds (63% of screen failures were due to HBT or LI symptoms scores not being met).

#### *3.1. Primary Endpoint Analysis*

#### Symptom Reduction in RP-G28-Treated Patients

In the Efficacy Subset mITT group, significantly more patients in the pooled RP-G28 group responded as compared to patients in the placebo group—40% versus 26% (*p* = 0.016). In total, 41% of subjects treated with lower-dose RP-G28 (*p* = 0.043) and 38% of subjects treated with higher-dose RP-G28 (*p* = 0.029) responded. (Table 4) In the mITT population, the pooled RP-G28 group trended toward significance *p* = 0.062 (40% with treatment versus 31% with placebo).

#### *3.2. Secondary Endpoint Analysis*

#### Individual Assessed Symptom Response to RP-G28 Post-Treatment

In the Efficacy Subset mITT, RP-G28-treated patients were significantly more likely than patients treated with placebo to report complete elimination of LI symptoms. RP-G28 was more likely to lead to complete elimination of the LI symptom composite score (*p* = 0.004) and individual symptoms of abdominal pain (*p* = 0.014), abdominal cramping (*p* = 0.002), abdominal bloating (*p* = 0.015), and gas movement (*p* = 0.001) than placebo (Figure 2A).


**Table 4.** Primary Endpoint <sup>1</sup>

1. Proportion of LI symptom composite score responders post-treatment (day 31). 2. Efficacy Subset (mITT)—mITT data in which observed inconsistent data from one study center was removed from analysis. 3. mITT—modified intent to treat (all patients who received at least one dose of drug). 4. CMH = Cochran–Mantel–Haenszel. *p*-value versus placebo. N = number of subjects.

**Figure 2.** Secondary Endpoints Analysis. (**A**) The proportion of patients reporting complete elimination of lactose intolerance (LI) symptoms (abdominal pain, cramping, bloating, and gas movement) with RP-G28 treatment or placebo. (**B**) The proportion of patients reporting a response (≥4-point improvement from baseline or a score of 0 on day 31) in each key LI symptom (abdominal pain, cramping, bloating, and gas movement) with RP-G28 treatment or placebo. (**C**) RP-G28 led to a significantly greater increase in daily average milk consumption from baseline 30 days after treatment (day 61) compared to placebo.

Patients treated with RP-G28 exhibited a consistent and significantly greater decrease in LI symptoms including cramping (*p* = 0.026) and bloating (*p* = 0.028). Non-significant trends for improvement were seen for abdominal pain (*p* = 0.105) and gas movement (*p* = 0.060) (Figure 2B).

"Real-world" Observation Period—Milk and Dairy Intake 30 days Post-Treatment: After 30 days post-treatment (day 61), patients treated with RP-G28 reported drinking significantly more milk, drinking an average of 1.5 cups/day versus 0.2 cups/day prior to treatment. The mean increase in milk intake of 1.3 cups/day (SD 1.479) in treated patients was significantly higher than that of the placebo group, which was 0.7 cups/day (SD 1.591) (*p* = 0.008, Efficacy Subset mITT) (Figure 2C).

In addition, 59% of treatment patients consumed ≥1 cups/d of milk after being treated with RP-G28 in comparison to 42% with placebo (*p* = 0.01, Efficacy Subset mITT)). Patients treated with RP-G28 also consumed more dairy in general, ingesting 5.4 cups/day on day 61 versus only 1 cup prior to treatment, at baseline. In comparison, the placebo group ingested 4.5 cups/day on day 61 versus 1.7 cups/day at baseline. The mean increase in dairy intake of 4.3 cups/day trended toward significance compared to the placebo group (*p* = 0.057, Efficacy Subset mITT).

"Real-world" Observation Period—Global Patient Assessments 30 days Post-Treatment: Significantly more treated patients (82%) reported "no symptoms" or "mild symptoms", respectively) as compared to 64% in the placebo group after treatment (*p* = 0.001, Efficacy Subset mITT). In addition, significantly more treated patients reported satisfaction with the ability of RP-G28 to prevent or treat their LI symptoms, with 66% of treatment patients reporting "very satisfied" or "extremely satisfied" as compared to 52% in the placebo group (*p* = 0.030, Efficacy Subset mITT). Patients' perception of adequate relief from LI symptoms and patients' global impression of change were also improved with treatment (*p* = 0.042 and P = 0.034 respectively, Efficacy Subset mITT) (Figure 3).

**Figure 3.** Global Patient Assessments. Patients' global assessments were evaluated 30 days after treatment (day 61) and were based on a patient questionnaire.

#### *3.3. Gut Microbiome Analysis*

Of 543 bacterial species identified overall, 28 were differentially represented (Kruskal–Wallis, FDR corrected *p* < 0.05). Of those, relative abundances of five *Bifidobacterium* taxa (*Bifidobacterium\_Other*, *Bifidobacterium sp*., *Bifidobacterium adolescentis*, *Bifidobacterium longum* and *Bifidobacterium pseudolongum*) were increased by both RP-G28 treatments. High-throughput qPCR quantitative data confirmed a significant (*p* < 0.05) increase in the abundance of the phylum Actinobacteria, the family *Bifidobacteriaceae*, and the genus *Bifidobacterium* on day 31 in the higher- and lower-dose treatment groups, but not the placebo group (Figure 4). Specifically, treatment with RP-G28 resulted in an elevated abundance of *Bifidobacterium longum*, *Bifidobacterium bifidum*, *Bifidobacterium breve*, *Bifidobacterium catenulatum*, *Bifidobacterium angulatum*, *Bifidobacterium gallicum*. In total, 78% (77/99) of patients treated with RP-G28 had elevated *Bifidobacteria* levels, as compared to 52% (49/94) of patients in the placebo group (*p* < 0.001). Data using Firmicutes, *Lactobacillaceae* and *Lactobacillus* species-specific 16S rRNA gene probes showed a significant increase in the relative abundance of the family *Lactobacillaceae* on day 31 in the lower-dose treatment and a non-significant increase in the higher-dose group, but not in the placebo group.

On day 31 after treatment with RP-G28, qPCR quantitative data revealed an increase in the relative abundance of the genus *Bifidobacterium* in the lower-dose galacto-oligosaccharide RP-G28 and higher-dose RP-G28 treatment groups, an effect not seen in the placebo group.

#### *3.4. Safety*

RP-G28 was well tolerated. In total, 6.9% of patients had treatment-emergent adverse events (TEAEs) with no differences between placebo and RP-G28. None of the serious adverse events (SAEs) were treatment related. No TEAEs resulted in hospitalization or death (Table 5).

**Figure 4.** Gut Microbiome Analysis.



AE = adverse event; N = Number of subjects; SAE = Serious adverse event; TEAE = treatment-emergent adverse event. 1. Treatment-related TEAT was defined as any TEAT that was possibly, probably, or definitely related to study drug. 2. One subject (a randomized placebo) experienced the SAE of spontaneous abortion.

#### **4. Discussion**

Though lactose intolerance is remarkably common, treatment options have not changed for decades [16,17]. The cornerstone of treatment, abstinence from consuming dairy products, is potentially nutritionally detrimental, inconvenient and limiting to patients. This study assessed a novel treatment strategy for lactose intolerance involving supplementation of RP-G28, which promoted the proliferation of lactose-fermenting bacteria, including *Lactobacillus* and *Bifidobacterium* [12]. Increased abundance and a higher diversity of these species is associated with enhanced levels of beta-galactosidase [17], reduced lactose-derived gas production in the colon [9,17], and mitigation of clinical symptoms in patients with LI [12,18].

To assess the efficacy and meaningfulness of treatment, three broad groups of LI patient experiences were evaluated—a LI composite score to measure specific symptoms, milk and dairy intake to evaluate patients' ability to increase dietary lactose with treatment, and global assessment measures to assess patients' overall satisfaction with therapy. This three-pronged assessment provided a comprehensive evaluation of how treatment influenced symptoms of LI and the day to day lives of patients.

A composite symptom score and a stringent threshold of meaningful change was established prior to unblinding the study in order to identify patients who reported a meaningful decrease in LI symptoms. Psychometric analyses indicated that the four symptom severity items were distinct yet related enough to support the creation of a composite score, and this composite score was shown to be reliable, valid, and responsive to change in LI symptoms over time. This tool provides a valuable resource for accurately assessing symptoms from the patient perspective in future LI clinical trials, and for identifying subjects who have experienced a meaningful treatment benefit in terms of patient-reported LI symptom severity. This study is one of the first to define and measure a meaningful treatment benefit for LI patients. The tool was developed adhering to good measurement principles following the FDA PRO requirements [19], and with consultation with the FDA. The instrument content was developed through a literature review, patient surveys, and concept elicitation and was tested through several rounds of cognitive interviews.

In the Efficacy Subset mITT population, the primary endpoint of response on day 31 was achieved in the pooled and individual RP-G28 groups. Individual symptoms were consistently reduced. The most rigorous endpoint for symptom relief is complete elimination. In the Efficacy Subset mITT population, RP-G28 administration resulted in complete symptom relief in a significantly greater proportion of patients than placebo, for the measures of abdominal pain, cramping, bloating, gas movement and overall symptoms (*p* < 0.05).

RP-G28-treated patients drank significantly more milk after treatment compared to the control group, drinking an average of 1.3 cups/day, as compared to an average of 0.7 cups/day more by the placebo group. Milk is the primary source of lactose in the diet, and thus milk consumption is an important indicator of the clinical benefits of supplementation with RP-G28. The ability to drink milk without LI symptoms supports optimal nutrient intake, contributing to the USDA's recommended 2–3 cups of dairy per day [20]. The sustained improvement in dairy intake and symptom relief on day 61 are further validation of the durability of treatment.

Consistent, statistically significant improvement in global assessments support the clinical meaningfulness of RP-G28. When consuming dairy foods for 30 days after treatment, 82% of patients reported "no or mild symptoms", 66% reported being "very or extremely satisfied," 83% reported adequate relief and 40% reported "very much or much improvement".

Overall, RP-G28 was safe and well tolerated. There were no differences in adverse events in those receiving RP-G28 or placebo. There were no serious adverse events in either group. Of note, there were no increases in GI side effects with RP-G28. This is important as RP-G28 is a GOS and, thus, part of the family of fermentable, oligo, di, monosaccharides, and polyols referred to as FODMAPs. FODMAPs have been shown to trigger GI symptoms in patients with irritable bowel syndrome. It is reassuring that at the doses administered, RP-G28 did not induce any significant GI side effects.

Gut microbiome changes and a reduction in net hydrogen gas production support the hypothesized mode of action [18]. In total, 78% of individuals treated with RP-G28 had elevated *Bifidobacteria*, and *Lactobacillaceae*. Patient microbiomes adapted further with the reintroduction of lactose into patient diets during a 30 day period post-treatment, with an increase in lactose-fermenting *Roseburia* species [2,12]. Elevated abundance of *Bifidobacterium* and *Lactobacillus* resulting from RP-G28 treatment, known to enhance lactose fermentation, is likely instrumental in better clinical outcomes. Further, stool from patients adapted to lactose produce less hydrogen, due to an absolute reduction in hydrogen production, rather than an alteration in hydrogen uptake by the microbiome [21]. Additional studies to understand how the changes in gut microbiome induced by RP-G28 lead to clinical improvements are warranted.

#### **5. Limitations**

One of the 15 study sites was excluded from the Efficacy Subset mITT due to significant GCP violations. Nevertheless, this is the largest and most rigorously designed double-blinded randomized trial for the treatment of LI ever conducted. Another limitation is that the construct that identifies LI is subjective, depending on a single lactose dose and a subsequent time course. Real-world LI is likely intermittent, and depends on diet, dose, transit and other environmental and biological factors that

are impossible to fully control within a clinical trial. However, in this study, the primary endpoint is part of a comprehensive assessment, including primary and secondary efficacy endpoints, global patient assessments, lactose consumption assessments, and correlation of treatment to changes in the microbiome—all of which are indicative of a beneficial effect of RP-G28 in improving symptoms of LI.

Another limitation of the study is the duration, which does not allow determination of long-term durability. While significant improvements in diet quality and lactose tolerance were evident on days 31 and 61, we cannot determine whether retreatment will be necessary or effective.

#### **6. Summary**

RP-G28 was safe and effective for reducing or eliminating symptoms of LI. Treatment with RP-G28 led to increased milk and dairy intake and improved quality of life. RP-G28 was safe and well tolerated at the doses administered. RP-G28 led to changes in the microbiome, which may be involved in the clinical benefits observed in patients with LI. These findings are relevant not only to lactose intolerance, but also usher in an era of using prebiotics to manipulate the microbiome to facilitate gut health.

**Author Contributions:** Conceptualization, W.C., W.S., A.J.R., H.F. and D.S.; methodology, W.C., W.S., A.J.R., H.F. and D.A.S.; formal analysis, W.C., W.S., A.R., H.F., M.A.A.-P. and D.A.S.; investigation, W.C., W.S., A.J.R., H.F., M.A.A.-P. and D.A.S.; data curation, W.C., W.S., A.R., H.F., M.A.A.-P. and D.A.S.; writing—original draft preparation, W.C. and D.A.S.; writing—review and editing, W.C., W.S., A.R., M.A.A.-P. and D.A.S.; funding acquisition, A.J.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** Support for this research was provided by Ritter Pharmaceuticals.

**Acknowledgments:** We thank Dana Barberio, of Edge Biocience Communications and Tara Nassari for their assistance in drafting and revising the manuscript.

**Conflicts of Interest:** W. Chey, W. Sandborn and D. Savaiano are on the Medical Advisory Board of Ritter Pharmaceuticals, H. Foyt; A. Azcarate-Peril are consultants of Ritter Pharmaceuticals; A. Ritter is an employee of Ritter Pharmaceuticals.

#### **Abbreviations**

Adverse Events (AE), Analysis of Variance (ANOVA), Code of Federal Regulations (CFR), Cochran–Mantel–Haenszel (CMH), False Discovery Rate (FDR), Food and Drug Administration (FDA), Gastrointestinal (GI), Galacto-Oligosaccharides (GOSs), Good Clinical Practice (GCP), Hydrogen Breath Test (HBT), Institutional Review Board (IRB), Investigational New Drug (IND), Lactase Non-Persistence (LNP), Lactose Intolerance (LI), Modified Intent to Treat (mITT), Numerical Response Scale (NRS), Polymerase Chain Reaction (PCR), Serious Adverse Event (SAE), Treatment-Emergent Adverse Events (TEAEs), and Verbal Rating Scale (VRS).

#### **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/).

*Article*
