*Review* **The Impact of the Quality of Nutrition and Lifestyle in the Reproductive Years of Women with PKU on the Long-Term Health of Their Children**

**Maria Inês Gama 1, Alex Pinto 2, Anne Daly 2, Júlio César Rocha 1,3,4 and Anita MacDonald 2,\***


**Abstract:** A woman's nutritional status before and during pregnancy can affect the health of her progeny. Phenylketonuria (PKU), a rare disorder causing high blood and brain phenylalanine (Phe) concentrations, is associated with neurocognitive disability. Lifelong treatment is mainly dietetic with a Phe-restricted diet, supplemented with a low-Phe protein substitute. Treatment adherence commonly decreases in adolescence, with some adults ceasing dietary treatment. In maternal PKU, elevated blood Phe is harmful to the fetus so a strict Phe-restricted diet must be re-established preconception, and this is particularly difficult to achieve. A woman's reproductive years introduces an opportunity to adopt healthier behaviours to prepare for successful pregnancies and positive health outcomes for both themselves and their children. Several factors can influence the health status of women with PKU. Political, socioeconomic, and individual food and lifestyle choices affect diet quality, metabolic control, and epigenetics, which then pre-condition the overall maternal health and long-term health of the child. Here, we reflect on a comprehensive approach to treatment and introduce practical recommendations to optimize the wellbeing of women with PKU and the resultant health of their children.

**Keywords:** adherence; epigenetics; health; phenylketonuria; preconception; women

#### **1. Introduction**

Phenylketonuria (PKU, OMIM 261600) is an inherited metabolic disorder caused by mutations in the phenylalanine hydroxylase (PAH) enzyme that impairs phenylalanine (Phe) metabolism, leading to high blood and brain Phe concentrations. It is managed with a lifelong Phe-restricted diet and an adjunct pharmacological treatment, such as sapropterin or pegvaliase [1]. In maternal phenylketonuria (MPKU), it is established that Phe crosses the placenta's blood membrane through a concentration gradient [2,3] and elevated blood Phe levels have a well-recognised teratogenic effect on the developing fetus, particularly in the early stages of pregnancy [4]. MPKU syndrome is characterized by foetal intrauterine growth retardation, facial dysmorphism, microcephaly, congenital heart disease, infant low birth weight, developmental delay, and intellectual disabilities [4]. There is also an increased risk of miscarriage, usually associated with poor maternal metabolic control [5]. Although there are several reports of pregnancy in women with PKU, little is known about the conception rates compared with the general population, though one recent UK/PKU study reported that 37% of 300 women aged ≥18 years had one or more children [6]. MPKU

**Citation:** Gama, M.I.; Pinto, A.; Daly, A.; Rocha, J.C.; MacDonald, A. The Impact of the Quality of Nutrition and Lifestyle in the Reproductive Years of Women with PKU on the Long-Term Health of Their Children. *Nutrients* **2022**, *14*, 1021. https:// doi.org/10.3390/nu14051021

Academic Editor: Yugo Shibagaki

Received: 3 February 2022 Accepted: 24 February 2022 Published: 28 February 2022

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

**Copyright:** © 2022 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 (https:// creativecommons.org/licenses/by/ 4.0/).

syndrome is preventable if women achieve rigorous blood Phe control by adhering to a Phe-restricted diet that is commenced preconception and continued throughout pregnancy. A considerable amount of professional health time and support is given to women during this challenging time.

In women with PKU, less consideration is given to the overall quality of nutritional care in the reproductive years (spanning from mid-adolescence until mid-adulthood) and interpregnancy. There is mounting evidence in all women of reproductive age that poor maternal and pregnancy health leads to a higher risk of disease in their children as they age [7]. The nutritional health of many women with PKU at the time of conception is likely to be sub-optimal, particularly if a strict dietary treatment has not been maintained through adult life. Some may have adopted an unhealthy eating pattern even if they are able to maintain optimal metabolic control. Furthermore, unplanned pregnancies at any point in time may increase the risk of nutrient imbalances. In England, 45% of all pregnancies are unplanned [7], and similar figures are observed in women with PKU, despite active health professional education to avoid unplanned pregnancy [5].

Therefore, the lifestyle choices of all women in reproductive years can have an enduring influence on the lifetime health of their children, and a clear focus on interventions before conception is necessary. Cohort studies have shown that improving dietary patterns for up to three years prior to conception can influence pregnancy outcomes, including lowering the risk of preterm birth [8]. Preconception environmental and nutritional factors that may affect the foetal outcome in women with PKU are presented in Figure 1. This review aims to highlight the importance of optimal nutrition, lifestyle, and environment in women with PKU in their reproductive years and offers proposals for pragmatic interventions that may improve the outcome of their children.

**Figure 1.** Preconception environmental and nutritional factors that may affect the foetal outcome in women with PKU. ↓-lower; ↓-higher; -lower/higher; broken line-arrows-potentially lower/higher.

#### **2. Nutritional Vulnerability of Women with PKU in Their Reproductive Years**

*2.1. Distal, Social, and Economic Causes of Nutritional Vulnerability in Adult Women with PKU*

There are many economic and political factors that may lead to suboptimal nutritional outcomes associated with the availability of treatment for women with PKU. Health provision varies around the world, and some women with PKU have limited access to 'free' health care from public funding, while few hospitals provide PKU health care teams that provide expertise in the management of adult patients. Low Phe protein substitutes and special low protein foods (SLPFs) are an essential part of treatment but are expensive and may be unaffordable unless provided by insurance or state health care systems. Pharmaceutical treatments may be unavailable or even ineffectual (e.g., sapropterin) for adult patients with classical PKU without residual enzyme activity [4]. Many adult women may be unemployed, receive low earnings due to part-time work, or have minimal earning capacity due to impaired cognitive functioning, affecting their economic security, life quality, and ability to afford their dietary treatment. Political legislation that aims to improve the health of the entire population, e.g., food labelling laws and sugar taxes, may indirectly create additional treatment challenges because of further unintentional dietary restrictions for people with PKU.

Women who do adhere to dietary treatment are dependent on a Phe restricted diet and, if they have classical PKU, usually tolerate <500 mg/day Phe (equivalent to 10 g/natural protein) supplemented with protein substitutes. The protein substitutes are mainly comprised of Phe-free L-amino acids (AA) or low-Phe glycomacropeptide (GMP) and may potentially supply up to 80% of protein intake. Although they usually contain added tyrosine, micronutrients including vitamins, minerals, and long-chain fatty acids, such as docosahexaenoic acid (DHA), the lifetime outcome of habitually taking an artificial protein source is unknown. Amino acid supplements, compared with natural protein, are associated with less efficient utilization and early oxidation, and they may alter insulin release, glycaemic control, and endocrine regulation [9]. The impact on gut microbiota and long-term renal health is undetermined. SPLFs are high in carbohydrates [1,10,11] and contain isolated starches that are more refined or have a higher glycaemic index than equivalent foods made from wheat flour [12,13].

#### *2.2. Proximal Causes Directly Related to Nutritional Vulnerability in Adult Women with PKU*

Dietary adherence becomes increasingly challenging with age and metabolic control commonly deteriorates from adolescence [14–18]; it is estimated that 25% to 40% of adults who remain in clinical follow up discontinue treatment [19]. Most adults have difficultly reestablishing dietary control after a period 'off diet' or dietary relaxation [20]. Although more natural protein is consumed than prescribed, clinical practice suggests that the quality of foods eaten is poor, potentially leading to nutritional inadequacy [21,22]. Women may have a low IQ (associated with poor blood Phe control during childhood) and poor executive functioning and possibly have left home and lost the practical support of their parents. This affects their ability to self-manage a Phe restricted diet owing to the daily organisation and planning required [18,23]. Low mood or denial of the condition may also obstruct the ability of people to comply by reducing self-control or motivation. Poor knowledge of diet and food suitability, limited cooking skills and meal choices, the inability to read and interpret protein amounts on food labels, being unable to estimate protein exchanges, and difficulty accessing supplies of protein substitutes/SPLFs also influence the ability to adhere to the diet [24].

#### *2.3. Health of Women with PKU*

**Obesity**: The prevalence of overweight and obesity in all women of childbearing age is high, and approximately 39% of the world's adult population is overweight, with 13% being obese [25]. Although a recent systematic review and meta-analysis of women with PKU [26] found that the body mass index (BMI) of patients with PKU was similar to their healthy controls, a subgroup of patients with classical PKU had a significantly higher BMI. The authors also noted a trend towards a higher BMI in females with PKU in all studies with male and female datasets. The BMI was also higher in an uncontrolled study in women with PKU, particularly if they had poor blood Phe control [27]. Adolescence is a critical period for the development of overweight and obesity [28], with a recent study illustrating that 28% (*n* = 101) of adolescents with PKU were overweight or obese [29].

**Eating disorders**: There is increasing evidence of eating disorders, food neophobia, and adverse attitudes towards food in adults with PKU [24,30–32]. Disordered eating refers to abnormal behaviours focused on eating or feeding, but it does not fit the pattern of a specific eating disorder [33]. It can manifest in restrictive, emotional, or uncontrolled eating. It is lower in severity and intensity than that of an eating disorder but impacts everyday life.

Fourteen percent of adults (*n*= 40/286) self-reported disordered eating in a survey reported by the UK National Society for PKU, with 4% receiving therapy for eating disorders. Individual patient stories described how they had an unpleasant relationship with food; others described how they used food as a reward [24]. Bilder et al. reported that 3.4% of patients (*n* = 128/3714) with PKU had an eating disorder compared with 0.9% in the general population [31]. Viau et al. discussed that 53% of adults (*n*= 9/18) on pegvaliase therapy had food neophobia with low enjoyment of food which did not appear to improve with a relaxed protein intake [32]. Luu et al. [33] found that in a group of adults with PKU (*n* = 15) aged 12–35 y, patients with poor metabolic control had symptoms of disordered eating at a higher frequency than those with good metabolic control. They were more likely to have been overweight, and there was an association between dieting and dissatisfaction with body image.

Food neophobia in adults with PKU may have its origins in childhood [34–38] and is likely to impede long-term dietary patterns, alter food selection, and lower nutritional quality later in life. Intransient feeding problems are very challenging to change, and diagnosing an eating disorder in a patient with PKU is difficult. Existing validated tools for the assessment of eating disorders may not be appropriate for individuals with PKU on a prescribed dietary treatment [33,39].

**Dietary pattern quality**: There are many concerns about the quality of diets consumed by women who have stopped dietary treatment, potentially causing nutritional fragility in reproductive years. Some patients remain on a self-imposed low-protein diet, avoiding protein-rich foods such as meat, fish, and milk for many years. If they eat higher protein foods, it is commonly only intermittently as many report guilt and having less food enjoyment if they eat foods contraindicated in their dietary treatment [24]. The discontinuation of a protein substitute, supplemented with vitamins and minerals, intensifies the risk of micronutrient deficiencies [18,22]. Women may have unpleasant memories of the taste, smell, and texture of protein substitute from childhood, or they may associate it with causing gastrointestinal symptoms such as reflux and constipation [24]. The absence of protein substitute intake may lead to the thinning of hair and poor skin condition associated with inadequate nutritional status [32]. There are reports of reduced or low normal serum urea levels [40]. In patients on a partial or minimal dietary treatment, a protein [41] and amino acid deficiency, particularly tyrosine [42] with low normal free carnitine values [43], are described.

Overall, there is little qualitative data discussing the dietary patterns of adults with PKU, and it is undetermined if they consume an adequate intake of fruit and vegetables. The habitual intake of meat, fish, dairy products, wholegrain cereals, and nuts and seeds is unknown but thought to be minimal. It is established that teenagers commonly eat high amounts of carbohydrates with a limited intake of fruit and vegetables [44], despite extensive dietary education.

**Nutrient deficiency**: Women may be at particular risk of iron deficiency due to menstruation and the low intake of Phe-free/low-Phe protein substitutes. In a group of nonadherent UK adult patients with PKU (*n*= 14) who did not take protein substitute as prescribed, dietary intakes of iron, zinc, vitamin D3, magnesium, calcium, selenium, iodine, vitamin C, vitamin A, and copper were significantly lower than adherent patients (*n* = 16) and were below the UK Reference Nutrient Intakes [21]. Rohde et al. demonstrated that in 67 patients with PKU who consumed a ≤0.5 g/kg protein equivalent from a protein

substitute that calcium and vitamin D intake was low, and the majority had low plasma 25-OH- vitamin D levels [22]. Vitamin B12 [41,45], zinc [21,46], and selenium [21,41,47] inadequacies are also reported in adult patients. Lower dietary adherence is associated with mild iodine deficiency and lower urinary selenium levels [48]. Pregnancy also increases the requirements for several macro- and micro-nutrients, compounding the risk of nutritional imbalance in women. The influence on maternal and foetal outcome of genetics, foetal programing, dietary management, and lifestyle of women with PKU are presented in Figure 2.

**Figure 2.** Foetal metabolic programming in women with PKU: influence of genetics, dietary management, and lifestyle on maternal and foetal outcome. ↓-lower; ↓-higher; -lower/higher; broken line-arrows-potentially lower/higher.

#### *2.4. Nutrition, Foetal Metabolic Programming, and Epigenetics*

The foetal programming concept suggests that maternal nutritional imbalance may have a persistent effect on the health of their children. It may pre-condition for metabolic syndrome and lead to long-term, irreversible changes in the organs and metabolism [49]. Poor maternal nutrition has been linked with early embryogenesis and foetal growth abnormalities, cardiovascular disease risk, and metabolic and renal dysfunction [50,51]. The Dutch famine studies clearly demonstrated how poor nutritional intake affects foetal outcomes. Children from pregnancies influenced by famine in early gestation had increased disease and metabolic risk in adulthood [52]. Even second-generation children of women who experienced famine in pregnancy were at increased metabolic risk, creating a transgenerational effect. Foetal epigenetic programming could play a key role in foetal metabolic programming [53,54].

Epigenetics is defined as changes that modify gene expression and cellular function; they do not change the DNA nucleotide sequence. Unlike genetic changes, these are reversible [53,55]. Epigenetic changes occur when environmental conditions, such as malnutrition or stress during critical periods in early life, modify metabolic and developmental pathways, in turn leading to alterations in their function [55–57] and the predisposition of individuals to disease in later adulthood [58]. Barker [59] first suggested that environmental events occurring during pregnancy could have consequences in adult life, leading to cardiometabolic disease. Thus, the quality of nutrition and nutritional imbalances, dietary

restriction, eating behaviors, lifestyle, and nutritional supplementation may affect nutritional programming before, during, and between maternal PKU pregnancies [49,57,60,61].

Micronutrients, including iron, zinc, folic acid, and other vitamins, contribute to epigenetic modifications during organogenesis in early pregnancy [58,62]. Methyl-donor groups, such as folate and vitamin B12, are vital for embryo and early foetal development [62]. Preconception zinc deficiency compromises foetal and placental growth and neural tube closure [63]. Folate, vitamin B12, methionine, choline, and betaine can affect DNA methylation and histone methylation. Folic acid, vitamin B12, and zinc participate in brain DNA and RNA synthesis, which begins early in gestation. Decreased vitamin B12 in the first trimester, associated with raised levels of folate, predicts increased central obesity and insulin resistance in the offspring [62]. Vitamin B12 has also been shown to affect myelination, which begins during gestation, and may affect cognitive functioning.

Folic acid and vitamin B12 participate in the folate–methionine cycle [64]. They are essential in the remethylation of homocysteine into methionine, which, consequently, generates S-adenosylmethionine, a methyl-donor molecule and folic acid essential in the prevention of neural tube defects (NTDs) [65]. There is evidence of inadequate intakes of folate and vitamin B12 in adult patients with PKU [41,66–68]. Many countries have a folic acid food fortification policy to decrease the incidence of NTDs or recommend folic acid supplementation during preconception and early pregnancy. However, regular foods fortified with folic acid (e.g., bread, pasta, and flour) are unsuitable for people with PKU. Protein substitutes are supplemented with folic acid, but reports of inadequate folic acid intake are described in non-adherent adults. In women with PKU, 400 μg/day of folic acid supplementation is recommended during preconception and the first 12 weeks of gestation [4]. Vitamin B12 is obtained from animal foods, which are excluded in a Phe-restricted diet, and acceptable intake is usually only associated with adherence to a nutritionally fortified protein substitute.

There is also evidence from animal and clinical studies that maternal overnutrition can lead to epigenetically mediated alterations in different physiological homeostatic regulatory systems and is associated with increases in the cardiometabolic risk in infants [56]. Observational evidence suggests that metabolic changes due to parental overweight/obesity affect epigenetic markers in oocytes and sperm alike and may influence epigenetic programming and reprogramming processes during embryogenesis [69]. However, mechanisms underlying overweight development and foetal adipogenic programming through influences of early-life stages are still poorly understood.

#### *2.5. Role of Key Micronutrients in Reproductive Nutrition*

**Iron**: A major public health problem that affects all women of reproductive age is anaemia, and in 2019 the global prevalence of anaemia in women of reproductive age (15–49 years) was 29.9% [70]. Anaemia has been associated with an increased risk of poor birth outcomes (low birth weight, preterm births, being small for gestational age, stillbirth, and perinatal and neonatal mortality) and adverse maternal outcomes (maternal mortality, postpartum haemorrhaging, and preeclampsia [71,72]. Perinatal iron deficiency is associated with long-term cognitive abnormalities as iron plays an important role in normal neurodevelopment through enzymes controlling neurotransmitter synthesis, cell division, neuronal energy metabolism, and myelination [73].

Preconception iron status is critical [65], and in women with PKU, the main sources are protein substitutes; women are particularly at risk of deficiency if adherence to this nutrition source is low. Several studies have reported an inadequate micronutrient status, including iron, particularly in non-adherent patients [21,22,74]. Green et al. identified that off-diet individuals with PKU with a blood Phe ≥600 μmol/L had iron intakes below the country-specific recommendations [74]. In a further two studies, patients with PKU who had stopped dietary treatment had significantly lower iron intake compared to adherent patients [21,22].

**Iodine**: Iodine is important in early foetal development and is associated with its involvement in thyroid function and foetal brain development [65]. Due to an increase in the iodine requirement for brain development in early pregnancy, iodine deficiency in the preconception period increases the risk of developmental delay in a child [65]. A meta-analysis by Levie et al. showed that a lower urinary iodine-to-creatine ratio during pregnancy was associated with a lower verbal IQ [75]. In women with PKU, iodine status is strongly influenced by a dietary adherence to protein substitutes supplemented with micronutrients, the main dietary source of iodine [21,22,48,74,76].

**Zinc:** In an in vivo model, acute dietary zinc deficiency before conception compromised oocyte epigenetic programming and disrupted embryonic development [77]. It is also important for immune function, foetal growth and neurological development, and potentially lowers the risk of preterm birth [65]. Low zinc intakes are commonly observed in women with PKU [21,74].

**Long-chain polyunsaturated fatty acids (LC-PUFAs)**: These play an important role in the inflammatory response as eicosanoid precursors, as well as an important role in foetal–infant brain development in the later stage of pregnancy and early infancy. It is crucial that adequate maternal LC-PUFAs reserves are maintained early in pregnancy and for foetal use in later stages of development [78]. The placenta relies on fatty acids as a major energy source and disturbances in nutritional status could cause placental dysfunction, such as angiogenesis occurring in the first trimester and, consequently, compromise of foetal development [78].

The placental transport of LC-PUFAs is altered in maternal obesity and diabetes, which consequently has implications for foetal metabolic status [78]. Low DHA concentrations are reported in patients with PKU and during pregnancy [79–82] if women do not receive a supply from a protein substitute supplemented with DHA. Pregnant women should be supplemented with an additional supply of ≥200 mg DHA/day, over and above the intake recommended for an adult's general health, and usually achieves a total intake of ≥300 mg DHA/day [83]. This should be given to all women with PKU considering pregnancy and throughout pregnancy [4,83].

**Over-nutrition:** Obesity is associated with an increased risk of most major adverse maternal and perinatal outcomes, including infertility, miscarriages, complications during pregnancy (pre-eclampsia and gestational diabetes) and delivery (macrosomia), congenital anomalies, stillbirth, unsuccessful breastfeeding, and even maternal death [65,84–88]. A higher BMI before pregnancy is associated with a more significant fat mass gain during pregnancy and is correlated with fat retention postpartum. It is also a strong predictor for increased birth weight, as well as for childhood overweight and obesity [69].

Obesity in pregnancy has been shown to significantly alter glucose metabolism leading to impaired fasting glucose reduction in early pregnancy and a considerable increase of peripheral and hepatic insulin resistance [56]. Any obesity-related, pre-pregnancy insulin resistance is associated with an increase of gestational diabetes and, consequently, a higher risk of foetal glucose metabolism impairment, hyperinsulinemia, and type 2 diabetes.

**Maternal gut microbiome**: Maternal health and diet play a critical role in the foundation of a child's gut microbiome with long-lasting health implications. The rise in oestrogen and progesterone during pregnancy alters the gut function and microbiome composition, increasing vulnerability to pathogens. Throughout pregnancy, the gut microbiota progressively changes, with the greatest change occurring in the ratio of specific key bacteria (e.g., Firmicutes/Bacteroidetes ratio) mimicking the higher levels of Firmicutes seen in obesity [89]. Gut microbiota [90] can interact and be modulated by dietary factors. Prebiotics, such as fructooligosaccharides and galactooligosaccharides, have a positive influence on the gut microbiota composition. Little is known about the carbohydrate intake of adults with PKU. In a Phe-restricted diet, many of the carbohydrate sources allowed are based on simple sugars, e.g., sucrose and fructose, and this may cause rapid deregulation in the composition of the gut microbiota and, hence, metabolic dysfunction in the host [91]. Although some SPLFS contain added fibre, it is usually in the form of hydrocolloids to

help their structure rather than provide nutritional benefits [11,44]. There is evidence that patients with PKU may have dysbiosis with less variety of bacteria, which may interfere with an optimal metabolism [92]. As well as the quality of carbohydrate intake, the high consumption of snacks, late-night eating, and skipping breakfast can also affect the gut microbiota composition [91].

**Sleep hygiene**: Sleep patterns may be disturbed in adult patients with PKU [93]. Quantity and quality of sleep play important roles in metabolic regulation and homeostasis [57,94]. A good night's sleep is associated with improved glucose, lipid, and energy metabolism, cardiovascular risk, inflammatory response, neurocognitive function, and mental health status [94,95].

#### *2.6. Interventions to Improve Nutritional Health in the Reproductive Years of Women with PKU*

Preconception care has been defined as "any intervention provided to women of childbearing age, regardless of pregnancy status or desire, before pregnancy, to improve health outcomes for women, newborns and children" [96]. In MPKU, it is important to identify any opportunities for improving nutrition prior to pregnancy using evidence informed interventions. It should be accepted that improving women's nutritional status may take several years and may be particularly challenging to maintain due to the high levels of food neophobia, maladaptive feeding behaviours, and limited food choices. In addition, individual motivations to engage with improving preconception nutrition will differ according to age, mental health, cognitive ability, and executive function. Understanding and harnessing these motivations will be key to successful intervention. Interventions to improve the nutritional status of PKU patients during their reproductive years are presented in Table 1.

**Table 1.** Interventions to improve nutritional health in women with PKU in their reproductive years.





Abbreviations: PKU, Phenylketonuria; MPKU, Maternal Phenylketonuria; BMI, Body mass index; HbA1c, hemoglobin A1c; EFSA, European Food Safety Authority; Phe, Phenylalanine.

#### **3. Conclusions**

The health of a mother and her children cannot be completely separated, and a heightened awareness of the importance of preconception health, particularly diet and nutrition, is essential in women with PKU. Birth outcomes are influenced by the long-term interaction of a woman's biology, behaviour, social and environmental factors, and quality of diet. Therefore, the optimal health status of women with PKU before and inter-conception is essential. It is important that there is attention to dietary adequacy, healthy weight, and lifestyle. Women should be encouraged to maintain dietary and pharmaceutical treatments for PKU for optimal neuropsychological functioning and the provision of self-care during their reproductive years. In addition, the attainment of optimal nutrition should be the goal of health professionals. Any approach that improves the long-term nutritional health of women with PKU will help enhance the well-being of their future children.

**Author Contributions:** Conceptualization, A.M., M.I.G., A.P. and J.C.R.; methodology, M.I.G., A.P., J.C.R. and A.M.; writing—original draft preparation, M.I.G. and A.M.; writing, review, and editing— M.I.G., A.M., A.P., A.D. and J.C.R.; supervision, A.M. and J.C.R. All authors have read and agreed to the published version of the manuscript.

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

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** A.P. has received an educational grant from Cambrooke Therapeutics and grants from Vitaflo, Nutricia, MerckSerono, Biomarin, and Mevalia to attend scientific meetings. A.D. has received research funding and financial support from Nutricia and Vitaflo to attend study days and conferences; Vitaflo has funded a PhD advisory member for APR., and J.C.R. is a member of the European Nutritionist Expert Panel (Biomarin), the Advisory Board for Applied Pharma Research and Nutricia, and received honoraria as a speaker from APR, MerckSerono, Biomarin, Nutricia, Vitaflo, Cambrooke, PIAM, and Lifediet. A.M. has received research funding and honoraria from Nutricia, Vitaflo International, Metax, Applied Pharma Research, and Biomarin. She is a member of the advisory board entitled ELEMENT (Danone-Nutricia) and Applied Pharma Research.

#### **References**


## *Article* **The Challenges and Dilemmas of Interpreting Protein Labelling of Prepackaged Foods Encountered by the PKU Community**

**Imogen Hall 1, Alex Pinto 2, Sharon Evans 2, Anne Daly 2, Catherine Ashmore 2, Suzanne Ford 3,4, Sharon Buckley <sup>5</sup> and Anita MacDonald 2,\***

	- <sup>4</sup> North Bristol NHS Trust, Southmead Road, Bristol BS10 5NB, UK

**Abstract:** Phenylketonuria (PKU) can lead to severe intellectual impairment unless a phenylalaninerestricted diet starts early in life. It requires expert user knowledge about the protein content of foods. The ability of adults or caregivers of children with PKU to calculate protein exchanges from food labels on manufactured foods and any difficulties they encounter in interpreting food labels has not been studied systematically. Individuals with PKU or their caregivers residing in the UK were invited to complete a cross-sectional online survey that collected both qualitative and quantitative data about their experience when calculating protein exchanges from the food labelling on prepackaged foods. Data was available from 246 questionnaire respondents (152 caregivers of patients with PKU aged <18 years, 57 patients with PKU aged ≥18 years or their caregivers (*n* = 28), and 9 teenagers with PKU). Thirty-one per cent (*n* = 76/246) found it difficult to interpret food protein exchanges from food labels. The respondents listed that the main issues with protein labelling were the non-specification of whether the protein content was for the cooked or uncooked weight (64%, *n* = 158/246); labels stating foods contained 0 g protein but then included protein sources in the list of ingredients (56%, *n* = 137/246); the protein content being given after a product was prepared with regular milk rather than the dry weight of the product (55%, *n* = 135/246); and the non-clarity of whether the protein content was for the weight of prepared or unprepared food (in addition to non-specification of cooked or uncooked weights on food labelling) (54%, *n* = 133/246). Over 90% (*n* = 222/246) of respondents had experienced problems with food labelling in the previous six months. Misleading or confusing protein labelling of manufactured foods was common. The food industry and legislators have a duty to provide accurate and clear protein food labelling to protect populations requiring low protein diets.

**Keywords:** phenylketonuria; food labelling; protein

#### **1. Introduction**

Phenylketonuria (PKU) is a genetic condition in which there is an inability to metabolise the amino acid phenylalanine into tyrosine. The treatment strategy for this condition is a lifelong phenylalanine-restricted diet to prevent adverse neurocognitive and psychological outcomes. This maintains blood phenylalanine levels within a narrow target therapeutic range but still delivers enough phenylalanine to support physiological protein synthesis, growth, and development. Patients with classical phenotypes usually have a natural protein tolerance that limits amounts to only 20% or less of what is expected in a regular

**Citation:** Hall, I.; Pinto, A.; Evans, S.; Daly, A.; Ashmore, C.; Ford, S.; Buckley, S.; MacDonald, A. The Challenges and Dilemmas of Interpreting Protein Labelling of Prepackaged Foods Encountered by the PKU Community. *Nutrients* **2022**, *14*, 1355. https://doi.org/10.3390/ nu14071355

Academic Editor: Bahram H. Arjmandi

Received: 23 February 2022 Accepted: 22 March 2022 Published: 24 March 2022

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

**Copyright:** © 2022 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 (https:// creativecommons.org/licenses/by/ 4.0/).

diet [1]. High-protein foods such as meat, fish, eggs, cheese, seeds, and nuts are avoided with controlled and measured intakes of cereals, potato, breakfast cereals, and some vegetables allocated in a 1 g protein exchange system (1 exchange is equivalent to ~50 mg phenylalanine) [1]. The amount of natural protein tolerated is individual and influenced by the patient's phenotype, use of adjunct therapy (such as sapropterin), growth rate, and dosage of protein substitute intake.

Since the 1960s, the UK has adopted a straightforward approach to dietary management, allocating foods such as fruit and vegetables containing phenylalanine up to 75 mg/100 g weight without measurement. Although there is a long history of including manufactured foods in a protein-restricted diet, the range of prepackaged foods available has exponentially increased, and food choice is now almost indefinable. Every major British supermarket stocks 30,000 to 40,000 consumable items, including a diverse range of prepackaged foods. The breadth of food additives is continually expanding, and many prepackaged foods contain a multitude of ingredients with some contributing extra protein or phenylalanine, such as artificial sweeteners, spirulina extract as a colour additive; cereal; gelatine thickeners and taste enhancers, e.g., yeast extracts. In particular, aspartame, an artificial sweetener, is a peptide rich in phenylalanine. In the EU and UK, prepackaged foods should list the protein content as one of six mandatory nutrients and state the amount of protein per 100 g or per 100 millilitres [2]. However, it is not mandatory to issue food label warnings if the food product recipe changes and alters the nutritional content. Navigating food labels and understanding the suitability of individual manufactured foods has intensified the complexity of dietary management.

In 2020, the British Inherited Metabolic Disease Dietitians Group (BIMDG-DG) published consensus statements about the suitability of foods in a phenylalanine-restricted diet for PKU to help standardise interpretation, particularly of prepackaged foods [3]. Statements divided food and drink into categories based on defined protein content. It included foods allowed without restriction, which contain protein ≤0.5 g/100 g, and foods that should be calculated/weighed as an exchange food if they contain protein exchange ingredients (categorised into foods with a protein content of: >0.1 g/100 g (milk/plant milks only), >0.5 g/100 g (bread/pasta/cereal/flours), >1 g/100 g (cook-in/tabletop sauces/dressings), and >1.5 g/100 g (soya sauces) [3]. The practical statements were endorsed and translated into practical dietary advice for patients and caregivers by the National Society for PKU (NSPKU).

In order for patients/caregivers to fully adhere to dietary management, they are expected to acquire expert knowledge about the protein content of foods. It is the role of dietitians specialising in inherited metabolic disorders to teach parents and patients about the application of the complex set of BIMDG dietary rules. This enables patients/caregivers to understand and interpret food label ingredient lists and explain how to calculate 1 g protein exchanges directly from protein labelling. Patients and caregivers are given a range of dietary resources, including 'pocket' protein exchange calculators, dietary information books, detailed food lists, and a collection of suitable manufactured food picture books.

In practice, reading and interpreting food labels adds an additional task to a dietary regimen already associated with a heavy time burden [4]. The ability of adults with PKU/caregivers to calculate protein exchanges and any difficulties they encounter in interpreting food labels and calculating protein exchanges have not been studied systematically. This project aimed to explore the perception and opinion of patients with PKU and their caregivers about their experiences when calculating protein exchanges from the food labelling of prepackaged foods.

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

#### *2.1. Methodology*

This was a cross-sectional study using an online survey collecting qualitative and quantitative data from caregivers of children with PKU and adult patients. Respondents were excluded if they did not reside in the UK.

The questionnaire was built in an Online Surveys platform (https://www.onlinesurveys. ac.uk, accessed on 17 July 2020). This was shared on the UK National Society for Phenylketonuria (NSPKU) website, with additional promotion on the NSPKU Twitter, Instagram, and Facebook sites. The questionnaire was open from the 18 July 2020 until the 1 February 2021.

#### *2.2. Questionnaire*

The non-validated questionnaire contained 24 questions. There were fourteen multiplechoice, four multiple-responses, and six open-ended questions. Five questions consisted of more than one part (2–7 parts). Four other questions invited additional comments. There were 4 questions about alcohol labelling that were targeted at adults aged ≥18 years; these data will be included in a separate publication.

The questionnaire was developed by dietitians with expert practical and scientific knowledge of PKU (AP, SE, CA, AD, AM), a colleague from the NSPKU (SF), and a student dietitian from Birmingham City University (IH). It was reviewed by colleagues and lay people to ensure its readability and then amended according to feedback.

#### *2.3. Data Collected*

The questionnaire was divided into 3 sections. Section 1 collected information on patient age, sex, type of supermarket they commonly shopped at, and ease of calculating protein exchanges from food analysis labels for known problems previously identified [5]. These included 4 groups of manufactured foods:


Section 2 contained information about interpreting the protein content of alcohol that was only collected from adults.

Section 3 contained information collected about the problems with food labelling, examples of issues experienced in the previous 6 months, the respondents' approach to dealing with food labelling issues, emotions when identifying misleading labelling, and changes that should be made to food labelling legislation. All data collected were based on the patient's/caregiver's knowledge of their own experiences when interpreting the suitability of foods and calculating protein exchanges from food labelling.

#### *2.4. Statistics*

Questions were analysed with descriptive statistics only.

Qualitative data analyses of open-ended responses were carried out in NVIVO v 12 PRO (QSR International Pty Ltd., Australia, New Zealand and Oceania Level 5, Suite 5.11 737 Burwood Road Hawthorn East, Vic 3123). The whole survey dataset was imported into NVIVO so that coding of open-ended responses could be broken down by survey questions. All open-ended question responses were analysed thematically.

#### *2.5. Ethics*

Ethical approval was obtained from the Birmingham City University Ethics Committee prior to commencement of the study (Hall/7499/R(B)/2020/Jul/HELS FAEC–MSc Healthcare Project: What are the current issues with protein labelling for PKU patients?). At the beginning of the online questionnaire, respondents gave consent, and it was emphasised that questionnaire completion was voluntary. Potential respondents were advised that data from the survey would be published in an anonymised form. Names or hospitals mentioned in verbatim abstracts were removed from results presented in this manuscript.

#### **3. Results**

#### *3.1. Demographics*

Two hundred and forty-six respondents from the UK answered the questionnaire. Twenty-three per cent (*n* = 57/246) were adults with PKU (aged >18 years), 11% (*n* = 28/246) were parents/caregivers of adults with PKU, 62% (*n* = 152/246) were parents of children with PKU, and 4% (*n* = 9/246) were children/teenagers with PKU. Forty-eight per cent (*n* = 117/246) of the respondents or respondent's children with PKU were male, 50% (*n* = 124/246) female, and 2% (*n* = 4/246) non-binary, and one respondent (0.4%) preferred not to answer. The four main regular supermarkets used by respondents were: Tesco (62%, *n* = 153/246), Asda (54%, *n* = 132/246), Aldi (39%, *n* = 97/246), and Sainsbury's (39%, *n* = 95/246).

#### *3.2. Rating of Food Labelling in General*

This received a mixed response from respondents, with 2% (*n* = 5/246) describing it as very good, 41% (*n* = 101/246) as fairly good, 30% (*n* = 74/246) as neither good nor bad; 19% (*n* = 47/246) as fairly bad, 7% (*n* = 17/246) as poor, and 1% (*n* = 2/246) did not know.

#### *3.3. Ease of Calculating Protein Exchanges from Food Labels*

There was difficulty in calculating protein exchanges from food labels for food and drinks for at least one-third of the respondents (Table 1). For some individually manufactured foods, increased problems were described, including dried powdered products such as sauces, soups, dessert powders, dried custard powders, drinking chocolates, pot noodles, and noodles. In an open comment question, 398 verbatim comments were received about food labelling. The mixed responses were thematically analysed into the following categories: (1) finding food labelling easy to understand, (2) difficulty with interpreting food content, (3) difficulty with understanding how to calculate protein exchanges, and (4) did not use protein labelling. Examples of responses are given in Table 2.

Many respondents commented that protein labelling was unclear when the protein analysis was given after theoretical preparation, particularly when the manufacturers had assumed a product was prepared with cow's milk or egg. Ice cream was complicated as protein analysis was commonly given by volume as mL rather than weight as g. Some commented that it was difficult when food products such as jelly or yoghurt had to be checked for both protein content and the presence of aspartame. It was also remarked that due to deficits with cognitive functioning, particularly mathematical and reading skills, some respondents were unable to calculate protein exchanges. Some respondents with sight difficulties were unable to read the small font of some food analysis labelling, and some did not calculate protein intake but preferred to use food picture books showing suitable manufactured foods provided to them by their hospitals and NSPKU, as they had confidence that these were likely to be correct. Others did not deviate from the foods they knew were safe and did not try new manufactured foods.

#### *3.4. Main Issues with Protein Labelling*

The respondents listed that the main issues with protein content on food labels were (Table 3): not specifying if the protein content is for the cooked or uncooked weight; a manufactured food stating that it contains 0 g protein but the ingredients list contains a source of protein such as milk or gelatine; protein amount given only after a product has been prepared with regular milk; and non-clarity if the protein content was for prepared or unprepared food weight (in addition to cooked or uncooked weight).

#### *3.5. Issues with Protein Food Labelling in the Previous 6 Months*

Over 90% (*n* = 222/246) had experienced problems with food labelling in the previous 6 months. In fact, *n* = 97/246 (39%) identified having problems at least 10 times in the 6-month period, with *n* = 68/246 (28%) describing weekly issues with food protein labelling. One hundred and sixteen respondents listed examples of problematic food labelling, and

these were thematically analysed into nine categories: (1) inadequate aspartame warning (*n* = 27); (2) dried products that are made up/served with milk (*n* = 16); (3) no differentiation of dried, unprepared, or uncooked weight vs. cooked/prepared weight (*n* = 16), (4) unclear protein labelling in general (*n* = 13); (5) suspect/doubtful protein content (*n* = 11); (6) foods purchased in multi-packs with unclear protein labelling (*n* = 9); (7) recipe change of a food item without warning (*n* = 9); (8) unclear protein content of imported foods (*n* = 8); and (9) analysis of protein content by volume rather than weight (*n* = 7). Examples of verbatim comments by the respondents are given in Table 4.

**Table 1.** Rating of interpretation of protein exchanges from food labels by respondents (*n* = 246) for food and drinks.



manufactured

 food labels.

**Table 2.** Verbatim comments about ease of calculating protein exchanges from protein content on


*preparation'*

*how to work out protein content'*




**Table 4.**

Verbatim comments of respondents

 in 9

thematically

 analysed categories explaining their practical problems with protein food

If respondents were unsure about the interpretation of food labelling, the majority said they would not use the food products (57%, *n* = 140/246), 47% (*n* = 115/246) would ask their dietitian or other health professional for help, 30% (*n* = 73/246) would ask others on social media, and 14% (*n* = 35/246) would guess the protein content and use it. Eight per cent (*n* = 20/246) said they would either try looking at other sources of information on websites, ask their relatives, or try and calculate it themselves.

#### *3.6. Respondent Emotions Associated with Food Labelling*

Respondents reported that misleading or inadequate information on protein food labelling made them feel frustrated (67%, *n* = 165/246), anxious (33%, *n* = 82/246), angry (33%, *n* = 81/246), upset (28%, *n* = 70/246), unhappy (28%, *n* = 68/246), and excluded (27%, *n* = 67/246).

#### *3.7. Suggested Changes to Food Labelling by Adults with PKU/Caregivers*

Suggested changes to protein labelling are presented in Table 5.

**Table 5.** Suggested changes to protein labelling as requested by questionnaire respondents (respondents could choose more than one response), *n* = 246.


• None (1%, *n* = 2/246).

There were 33 other suggested changes, including that manufacturers should not assume that products are prepared with cow's milk and give the protein analysis only after theoretical preparation; aspartame should always be in bold; all protein analysis should be made available on every supermarket website; and products should state accurate protein analysis and not use protein <0.5 g/100 g, which is unhelpful for low-protein diets. Some suggested that the protein content should always be in a uniform position on the food analysis list. It was also suggested that nutrient analysis should be in a larger font, and the protein content should be included on the front of the packet alongside the energy content.

#### **4. Discussion**

This paper highlights the considerable problems faced by both adult PKU patients and caregivers of children with PKU when trying to calculate exchanges from the protein analysis provided on food labels of prepackaged foods. Although there was a consensus that overall food labelling was satisfactory, the findings indicate that many patients/caregivers find protein calculations a complex process and identified several difficulties when interpreting protein labelling.

It was disconcerting that over 90% of respondents described specific issues with food labelling in the previous 6 months. Several respondents were frustrated that some potentially suitable instant dessert mixes and dried cereals had a protein content given on the food analysis after manufacturers had assumed they would be reconstituted/prepared with added cow's milk or egg, rendering the products unsuitable for people with PKU; no data were provided about the protein content of the dry products as purchased. There were many examples of ice creams that gave protein content for volume (in millilitres) rather than weight, and prepackaged foods that only gave a protein content of <0.5 g/100 g. Some commented that it would make a 'massive difference' if food labelling was clearer as there would be more foods that could be consumed, that the 'confusing protein labelling made it very hard when choosing suitable foods in the supermarket', and 'the problems of interpreting protein labelling will not help my son become independent.' These issues were also identified by Kravela et al. 2020, who examined the accuracy of protein analysis from supermarket websites [5].

It was worrying that some respondents identified that manufacturers changed the recipes of some of their products, affecting the protein content, without any 'front of package' warnings, possibly causing dietary error. This commonly occurred in foods such as breakfast cereals following the Public Health England voluntary sugar-reduction programme (2017), which requested that manufacturers lower the sugar content of foods by 20% [6]. Some manufacturers replaced sugar with other ingredients containing protein. If people with PKU or their caregivers do not detect changes in protein labelling immediately, it may potentially lead to a long-term miscalculation of protein intake. It is well established that some patients with PKU struggle with maintaining satisfactory blood phenylalanine control [7–9]. This is often attributed to poor dietary adherence, but inadequate standards of food protein labelling could contribute to this. Misinterpretation of protein food labelling may cause some of the day to day blood phenylalanine variation that is observed in PKU, although this remains an area not considered by researchers.

Respondents also described an unfortunate trend for average protein labelling on multi-packs of different individually wrapped foods (e.g., small boxes of breakfast cereals, mixed flavoured bags of popcorn and crisps, and sweets and chocolates) with each individual item in the multi-pack having a different protein content per 100 g. For many multi-packs, respondents described how the protein content was given as an average on the outer packs, with no protein content stated on individual packs. For one product, a different protein content was stated on the outer compared with the inner packaging, which suggested careless protein labelling practice by the manufacturer. There appears to be no mandatory law to inform manufacturers that this practice is misleading and unsafe for people with PKU as well as other patient groups following protein-restricted diets. It is extraordinary that the UK Food Standards Agency has allowed this practice to occur.

There was respondent mistrust around the accuracy of protein labelling, with examples given of discrepancies of protein analysis between websites and actual food product labelling. Some food products declared high-protein-containing ingredients in the first two or three items listed on their labels, yet the protein analysis was 0 g/100 g. One product contained less protein per 100 g than was given for an 80 g portion size. There were examples of decimal place typing errors that had clearly not been detected by the proofreaders of the manufacturer's labels; this could have serious consequences for patients with PKU. There were descriptions of protein analyses being hidden/lost in packaging 'folds,' or the protein analysis being written in a linear format with other nutrients listed on the same line, making it difficult to distinguish protein from other nutrients. There were also important concerns about the protein labelling of imported foods. Food labels from the USA state protein content in portion sizes only. Imported foods from the USA only acknowledge the presence of protein on food labels if a prepackaged product contains more than 1 g of protein/portion; otherwise, they inaccurately state that the product contains 0 g of protein/portion. Some imported foods were reported to not include any Englishlanguage food analysis on the labels, although all labels need to comply with the UK food labelling laws, and this is mandatory.

Over one-third of respondents found drinks labelling a particular issue. Any alcoholic drink with a volume content above 1.2% does not legally require protein content to be declared, although appropriate allergen information should be given [10]. Importantly, aspartame content is exempted from inclusion in the labelling of alcoholic drinks. [11]. Several examples were given of inconsistent aspartame identification on the labels of fruit squashes or drinks bought from shop vendors. Detailed information about the perceptions

of aspartame and food labelling of patients or caregivers of patients with PKU has been reported [11].

Except for the mandatory guidelines that manufacturers should state the product protein analysis per 100 g or 100 mL, there are few legal requirements about protein labelling [12]. The legislation allows manufacturers to use different methods to calculate the protein content of foods. It does not necessarily require laboratory analysis, and it may be possible for a food business operator themselves to perform a calculation from the known, or actual, average values of the ingredients used or to utilize established and accepted data [13]. Food regulations consider that a protein amount of ≤0.5 g per 100 g or 100 mL to be negligible, and so neglects the needs of people with PKU. Manufacturers may give the protein content per portion and/or per consumption unit, but this is not mandatory [2]. There is much that is lacking in protein legislation. Legislators must be aware that an inattentive approach to protein food labelling is a source of increased stress and burden for people with PKU and their caregivers. It limits their food choices, may induce unhealthy/repetitive food patterns, reduces variety in the diet, and may contribute to food neophobia [14].

This study has some limitations. Recruitment of participants for the online survey was performed via the NSPKU website and promoted on PKU social media sites, so respondents were limited to individuals with access to the internet using appropriate technology. Hence, it is likely that respondents were people who accessed social media sites frequently, and their views may not fully represent those of the broader population of PKU patients or their caregivers. However, problems deciphering food labels may be just as frequent in non-social media users, and this could be further investigated. Although there was a large response from caregivers (*n* = 180), there was a low response from adults with PKU (*n* = 57). It is known that in England alone, there are around 1100 adults on diet therapy with PKU. It is unclear whether this was due to a low interest in this area; unchanging dietary habits; limited reading of food labels; or low usage of websites, or PKU sites in particular, by affected adults [15]. The questionnaire was not validated prior to use, and the respondents' levels of education were unknown. We did not examine the amount of teaching they had received about a phenylalanine-restricted diet, which may have affected their answers, and the data from adult patients were not compared with those of caregivers.

#### **5. Conclusions**

Calculating PKU protein exchanges whilst considering portion sizes and checking for ingredients such as aspartame is a complex process with significant health implications. It is crucial that the quantity and presentation of protein and additive information on food labels enable patients with PKU or their caregivers to interpret this correctly. The range and extent of the issues identified around food labelling and interpretation suggest that the food and drinks industry is not currently providing clear and accurate information.

There appears to be no monitoring system examining the reliability of protein analyses on product labelling. Food manufacturers and legislators have a duty to provide a safe environment by ensuring accurate and clear protein labelling for populations requiring therapeutic low-protein diets.

**Author Contributions:** Conceptualisation, A.M., A.P. and I.H.; methodology, A.M., A.P., S.E., S.F., S.B. and I.H.; formal analysis, A.M. and S.E.; writing original draft preparation, A.M.; writing, review and editing, I.H., A.P., S.E., A.D., C.A., S.F., S.B. and A.M.; supervision, A.M. and A.P. All authors have read and agreed to the published version of the manuscript.

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

**Institutional Review Board Statement:** This study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Birmingham City University Ethics Committee prior to the commencement of the study (Hall/7499/R(B)/2020/Jul/HELS FAEC–MSc Healthcare Project: What are the current issues with protein labelling for PKU patients?).

**Informed Consent Statement:** Informed consent was given by all subjects when filling in the questionnaire.

**Data Availability Statement:** The data will be made available from the authors upon reasonable request.

**Acknowledgments:** We would like to acknowledge and thank all the patients and families that took the time to complete this survey.

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

#### **References**


## *Article* **Validation of a Low-protein Semi-Quantitative Food Frequency Questionnaire**

**Sharon Evans 1,\*, Catherine Ashmore 1, Anne Daly 1, Richard Jackson 2, Alex Pinto <sup>1</sup> and Anita MacDonald <sup>1</sup>**


**Abstract:** Analysis of dietary patterns and their role in long-term health is limited in phenylketonuria (PKU). Food frequency questionnaires (FFQ) are commonly used to assess habitual intake. A semiquantitative 89-item FFQ with a portion size photographic booklet was developed for children with PKU as a tool for collecting data on habitual intake of foods, food groups, energy and macronutrient intake. Twenty children with PKU aged 11–16 years, 30 parents of children with PKU aged 4–10 years, and 50 age/gender-matched control children were recruited. To test reproducibility, FFQs were completed twice with a mean interval of 5 weeks (range: 4–10). In order to test validity, FFQs were compared with five 24-h dietary recalls with a mean interval of 10 days (range: 6–18). Energy and macronutrient intake and quantity/week of individual food items were calculated and compared. There was good reproducibility for the FFQ with macronutrient correlations *r* > 0.6 and good validity data with most correlations *r* > 0.5. Bland–Altman plots for reproducibility and validity showed mean levels close to 0 and usually within 2 standard deviations. FFQ comparisons of PKU and control groups identified expected differences in % energy from macronutrients (PKU vs. control: carbohydrate 59% vs. 51%, fat 26% vs. 33%, protein 15% vs. 16%). This FFQ for PKU produced comparable data to repeated dietary recalls and is a valid tool for collecting data on habitual food and nutrient intake. It will be useful in assessing changes in dietary phenylalanine tolerance of new pharmacological treatments for PKU.

**Keywords:** phenylketonuria (PKU); dietary patterns; food frequency questionnaire; validation; reproducibility

#### **1. Introduction**

Phenylketonuria (PKU) is a rare genetic condition, resulting in the failure to metabolise the amino acid phenylalanine, resulting in severe neurocognitive disability if untreated. It is managed with a low phenylalanine diet supplemented with a protein substitute (either phenylalanine-free L-amino acids or glycomacropeptide (GMP), typically with additional micronutrients), and special low-protein foods (SLPFs). The remaining diet consists of food starches, sugars, fruit and low-protein vegetables.

Dietary pattern analysis is increasingly used to examine food intake and the synergistic effect of food and nutrients [1,2], but this is unreported in PKU. Conceptually, dietary patterns provide a broad picture of food and nutrient consumption and may be more predictive of disease risk than individual foods or nutrients. In dietary pattern analysis, food consumption patterns are characterised by habitual intake [3]. With PKU, although much is known about the dietary prescription, little is known about what is consumed, including food preferences, range of meal choices and food patterns. Whilst it is assumed that patients eat plentiful amounts of low-phenylalanine fruit and vegetables, evidence suggests the converse position [4–6]. Furthermore, food neophobia appears to be more

**Citation:** Evans, S.; Ashmore, C.; Daly, A.; Jackson, R.; Pinto, A.; MacDonald, A. Validation of a Low-protein Semi-Quantitative Food Frequency Questionnaire. *Nutrients* **2022**, *14*, 1595. https://doi.org/ 10.3390/nu14081595

Academic Editor: Shanon L. Casperson

Received: 7 March 2022 Accepted: 9 April 2022 Published: 12 April 2022

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

**Copyright:** © 2022 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 (https:// creativecommons.org/licenses/by/ 4.0/).

prevalent in children with PKU [5,7,8], and they appear reluctant to eat a wide range of fruits and vegetables [6].

Food frequency questionnaires (FFQ) are common tools used to measure dietary patterns. Respondents are given a list of foods and they describe how often each is eaten, e.g., how many times per day/per week/per month [9]. Compared with traditional dietary assessment methods, such as food diaries or 24-h recalls, FFQs require limited health professional time (both in data collection and analysis), low participant commitment, and may be completed by individuals with lower education or motivation [9,10]. They can be completed on paper or electronically in hospital clinics or the home environment. The results obtained by FFQs represent usual intakes over time and are suitable for ranking subjects into low, medium or high intake groups for individual foods or nutrients.

A FFQ should be tailored to each diet therapy. In PKU, portraying the full range of diverse foods permitted in a phenylalanine-restricted diet is challenging. This includes SLPFs and differing food types and quantities based on natural protein tolerance. Patients with classical phenotypes may tolerate only 3 g/day of natural protein (150 mg/day phenylalanine) but mild phenotypes tolerate ≥25 g/day (1250 mg/day phenylalanine), resulting in varying dependencies on SLPFs and protein substitutes. Pharmaceutical treatments, such as sapropterin dihydrochloride (BH4), may increase natural protein intake and the types of foods consumed in a subset of patients with PKU [11].

Ideally, a FFQ should contain no more than 100 commonly eaten foods grouped into sections, as only marginal gain is associated with more detailed questionnaires [12]. All FFQs should be validated to ensure that they measure what is intended and that they yield consistent results from repeated samples over time. This in turn improves the quality of the data collected and enables comparisons between studies using the same tool. There are different types of validity, meaning that a questionnaire is never fully validated but is valid for certain populations under specified conditions [13]. Validation of FFQs can be achieved in various ways and it is suggested that a combination of methods should be used to assess reproducibility and validity [14]. Checking that the questionnaire content is relevant and valid (content validity), that it can differentiate between different subject groups (construct/discriminative validity), that it produces reliable/reproducible results (reproducibility) and compares well with an existing standard (criterion validity), provides more credibility to the resulting data. Similarly, reporting that experts established questionnaire face validity, that the questionnaire was pilot tested on a subset of participants for understanding and relevance, and that appropriate statistical tests were used, also improve the integrity of the data [15].

Any FFQ designed for PKU should be validated by comparing it with a control group population to demonstrate that it is able to distinguish between the variations in macronutrient intake associated with the different food items eaten in a phenylalaninerestricted diet. Food intake will also vary according to the age, ethnic, social, educational, and economic background of the study population. Thus far, only one PKU-specific FFQ has been validated from the USA; 29 adults/adolescents were studied, and they compared the results of a FFQ with a 3-day food diary [10]. Whilst this study found good agreement between the different dietary methods and between repeated measures of the FFQ for protein intake, it was not validated in children, is likely to be specific to the USA population for food types and portion sizes, and it did not report on the validity of energy or other macronutrient intake such as carbohydrates, fat or fibre.

The aim of this study was to develop and validate a semi-quantitative FFQ for use in children with PKU, providing a tool that collects data on habitual intake for foods, food groups, energy and macronutrient intake, which can be utilised for dietary pattern and lifestyle analysis nationally.

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

#### *2.1. Construct Validity (Ability to Differentiate between Different Subjects)—Study Subjects*

Fifty children with PKU and 50 age and gender-matched healthy control children were recruited to test the FFQ for construct validity, which is the ability to differentiate between the dietary patterns for different groups. For children aged 4–10 years, data was completed by a parent/carer with assistance from an inherited metabolic disorder (IMD) dietitian; and for children aged 11–16 years, data was completed by children with assistance from the parent/carer and IMD dietitian. Inclusion criteria for subjects with PKU comprised the following: diagnosed by newborn screening; dietary treatment only (i.e., not prescribed sapropterin), supplemented with a prescribed free/low phenylalanine protein substitute from diagnosis and SLPFs; and no co-existing medical conditions, other special dietary requirements or intercurrent infection. All subjects with PKU were recruited from Birmingham Children's Hospital over a 30-month period (2018–2020). For control subjects, inclusion criteria comprised the following: age (within 6 months) and gender-matched to subjects with PKU; and on a regular diet (special diets, including vegan, vegetarian, and dairy-free were excluded). Control subjects were recruited from siblings of other children with inherited metabolic disorders, friends, or family of Birmingham Children's Hospital staff during the same time period as PKU subjects. The average nutrient and individual food intake from the 2 FFQs for each group were compared to establish construct validity.

#### *2.2. Content Validity (Checked by Experts in the Field)—Food Frequency Questionnaire Development*

Other UK and European PKU centres who were members of the SSIEM (Society for the Study of Inborn Errors of Metabolism) or BIMDG (British Inherited Metabolic Diseases Group) were invited to share their food frequency questionnaires. Five questionnaires were received (2 from England, 2 Scotland and 1 Germany). From these, a draft FFQ was developed for PKU and then adapted for control children (low-protein meal choices were matched with regular foods in the control FFQ, and SPLFs were also added to the PKU FFQ). The FFQs were reviewed by 5 IMD dietitians and piloted on 5 children with PKU and 5 control children to assess content validity. Following minor modifications, an 89-item PKU FFQ (+11 general dietary pattern questions) and a 69-item control FFQ (+5 general dietary pattern questions) were produced, including portion sizes for each item. The difference in the number of food items on the 2 questionnaires was a result of the additional SLPFs on the PKU version. General dietary pattern questions focused on meal frequency, missed meals, addition of salt to food, vitamin and mineral supplements and frequency of eating out (restaurants/cafes). The PKU FFQ also included 6 questions about quantity and frequency of protein substitute dosage and the number and amount of protein exchanges (the weight of food/drink that yields 1 g protein or 50 mg phenylalanine is one exchange).

#### *2.3. Reproducibility—Food Frequency Questionnaire*

The FFQ was completed at recruitment and at the end of the study (1–2 months apart) to test for reproducibility (test/retest reliability). This length of time was chosen in order to minimise changes over time but also to minimise recall of previous answers. The questionnaires were administered by one of four trained IMD dietitians using a standard script (see Supplementary Materials Supplementaries S1–S4). The same dietitian completed both questionnaires with each subject. For each food item on the FFQ, both the number of daily and weekly food portions consumed were recorded. Items consumed less than once a week were omitted.

#### *2.4. Portion Size Booklet*

Photographic portion size booklets were designed to accompany each FFQ, with pictures of the portion sizes specified in the questionnaire. Of 425 food photographs (captured from a range of distances and angles for each food), 65 were selected for the PKU FFQ and 58 for the control FFQ (24 foods were the same; 21 were the same but for the PKU FFQ were SLPFs instead of regular foods, e.g., low-protein pasta, bread, burgers; 8 were the same food but differed in portion sizes to show the amount that was equivalent toa1g protein/50 mg phenylalanine exchange; the remaining were diet specific foods, e.g., SLPFs with no regular comparative, or meat products with no low-protein comparative). Foods were prepared and presented on a plain white plate or bowl or transparent glass, on a neutral background and photographed immediately to maximise the aesthetic appearance of the food. Foods that were pre-packaged in standard portion sizes or were equivalent in size to another food did not have photographic representation. Portion sizes for all foods (including SLPFs) were described by weight (g), volume (mL) (if applicable), and household measurements, e.g., tablespoons, teaspoons, glasses, slices, packets/sachets, whole items, or a combination of these. Portion sizes for protein/phenylalanine containing exchange foods were usually described as the amount yielding 1 g protein or 50 mg/phenylalanine, often described as a weight or volume range, e.g., 50–70 g to allow for small variations in intake. Average portion sizes were generally used, these could then be multiplied up (e.g., double) or down (e.g., halved) if a larger or smaller portion size was consumed. For dietary analysis purposes, the smaller value was used to calculate nutrient intake.

#### *2.5. FFQ Database*

In order to analyse energy and macronutrient intake (protein, fat, carbohydrate and fibre) from the FFQ, food items were assigned a nutrient content based on composition data compiled from *McCance and Widdowson, The Composition of Foods* [16] supermarket nutrient analysis data (Tesco website accessed May 2017) and SLPF nutrient composition data from manufacturers. For each FFQ item, the nutrient analyses were selected from one or more of these sources and the nutrient contents were averaged to obtain a single value for each nutrient. These values were then entered in the *Nutritics* [17] software computer analysis program as 'new foods', including portion sizes, and data from the FFQs were then analysed using the items as entered in *Nutritics*. Data entry was completed by the same dietitian and cross-checked for accuracy by a second dietitian. Nutrient intakes for each of the 2 FFQs for each subject were obtained and converted into average daily intakes for energy (kJ), carbohydrate, protein, fat, dietary fibre, starch and sugars. The 2 FFQs were then compared to establish reproducibility, and the average of the 2 FFQs compared with the average of the 24-h recalls to establish criterion validity.

#### *2.6. Criterion Validity (Comparison with an Existing Standard)—24-h Dietary Recalls*

As a means of comparison and to test criterion validity (comparison with an existing standard), five 24-h dietary recalls were completed with subjects by one of 4 trained IMD dietitians experienced in taking diet diaries and using the same standard script for asking questions. All food and drink consumed the day before were recorded including type and quantity, and time of day consumed. All 5 dietary recalls for each subject were completed by the same dietitian during the same time period as the FFQs (over a 1–2-month period) and with a minimum of 5 days between each one including at least 1 weekend day and no more than 2 of the same weekdays, to ensure a representative intake (Figure 1).

Dietary recalls were all analysed by the same dietitian using the *Nutritics* [17] computer analysis program, and results were averaged to obtain an average daily intake. These were then compared with the average of the 2 FFQs to establish criterion validity.

#### *2.7. Anthropometry*

Body weight, height and BMI (body mass index) were measured and z-scores calculated at recruitment. Weight was measured to the nearest 10 g using Seca electronic scales; length was recorded to the nearest 1 mm using a Seca 213 portable stadiometer (Seca, Hamburg, Germany).

**Figure 1.** Study design. FFQ: Food frequency questionnaires.

#### *2.8. Statistics—Sample Size*

Data from a previous study [5] in a subset of children with PKU suggested that for a high fat, high carbohydrate food such as potato fries, power to detect a clinically relevant difference in mean intake of 1.7 days/week (the difference between a PKU group mean, μ1, of 3.2 days/week and a control group mean, μ2, of 1.5 days/week). Assuming a common standard deviation (SD) of 2.08 and a two-sided significance level of 0.05, a sample size of 33 in each group will have a power of 90%. Sample size calculations were performed using nQuery Advisor.

#### *2.9. Data Analysis*

Analyses were performed to evaluate the differences between the FFQs for PKU and control groups, as well as between the FFQ and dietary recalls data using GraphPad Prism version 6.01 for Windows, GraphPad Software, La Jolla California, USA. Continuous data were summarised as median (IQR) and categorical data were summarised as frequencies of counts with associated percentages. The strengths of association between dietary components were estimated using Spearman's correlation coefficients with Wilcoxon signed rank tests used to evaluate any differences between PKU and control groups. Comparisons of continuous data were performed using Wilcoxon sign rank for paired data and rank sum test for unpaired data. Categorical data were compared between groups using a Fisher test. Bland–Altman methods were used to assess the agreement between the FFQ and the 24-h dietary recall data. A *p*-value of 0.05 was used to determine statistical significance throughout.

#### *2.10. Ethical Approval*

This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and a favourable ethical opinion was obtained from the London—Queens Square National Research Ethics Service (NRES) Committee (REC reference: 15/LO/1463 and IRAS ID: 185896). Written informed consent was obtained from the parent/carer of all subjects, and assent from children was obtained where appropriate, according to their level of understanding.

#### **3. Results**

#### *3.1. Subjects*

#### 3.1.1. PKU Group

Fifty children (24 male) with PKU, (mean age 9.3 years; range: 4–16 years) on a phenylalanine-restricted diet only, were recruited from one specialist PKU centre (Birmingham Children's Hospital, UK). No changes were made to dietary intake during the study period. For 30 of the children aged 4–10 years, their mothers completed the questionnaires

with a dietitian, and for the remaining 20 children, (aged 11–16 years) they self-completed the questionnaires with assistance from parents/carers and a dietitian.

#### 3.1.2. Control Group

Fifty control children were age (within 6 months) and gender matched to the subjects with PKU. Questionnaires were either completed by parents/carers or by teenagers, following the same criteria as in the PKU group.

#### *3.2. Demographics and Anthropometry*

Most children were white UK/European origin (n = 45 subjects with PKU and n = 47 controls), with the remaining being of either Asian (n = 3 PKU, n = 2 control) or mixed-race (n = 2 PKU, n = 1 control) origin. There was no significant difference between mean z-scores for BMI, weight or height between PKU and control groups (see Supplementary Materials Table S1).

#### *3.3. Meal Patterns—FFQ 1 vs. FFQ 2 (Reproducibility) vs. Dietary Recalls (Criterion Validity-Comparison with an Existing Standard)*

The mean time between the two FFQs was 5 weeks (range: 4–10). Recalls were completed at a mean interval of 10 days (range: 6–18).

For meal patterns, there was little difference between the two FFQs, or between the FFQ and dietary recalls (Table 1). For the PKU group, there was a difference in the median number of meals that were consumed for FFQ 2 (4 meals/day) compared with FFQ 1 and the dietary recalls (5 meals/day). The PKU group varied across assessment methods in the percentage consuming mid-morning snacks, and this group was also less likely to consume a mid-morning snack compared with controls (FFQ 2 *p* = 0.0005). However, some children took their protein substitute at this time.

**Table 1.** Meal patterns—percentage of subjects consuming meals and snacks, or missing meals for FFQ 1, FFQ 2 and dietary recalls (PKU and control groups).


\*\* average of 5-day dietary recalls; \* Fisher test; # Wilcoxon signed rank; (n) = number of children; FFQ = Food Frequency Questionnaire; PKU = Phenylketonuria

#### *3.4. Protein Exchanges and Protein Substitute Intake*

Dietary patterns related to intake of protein substitute and natural protein exchanges were comparable between repeated FFQs and between FFQ and dietary recall data (Table 2). The median daily number of 1 g protein exchanges (50 mg phenylalanine) from the 24-h dietary recalls was 0.5 g protein (25 mg phenylalanine) less than prescribed (5.0 g vs. 5.5 g). Dietary recall data showed that 70% (n = 35) of children with PKU were taking their daily number of 1 g protein exchanges to within 0.5 exchanges of prescribed amounts. Of the remaining 30%, 6% (n = 4) were allocated ≥ 10 g/day of protein and 18% (n = 9) were prescribed ≥5 g/day, indicating that these patients were less protein restricted.


**Table 2.** Percentage of subjects consuming natural protein exchanges (1 g protein = 50 mg phenylalanine) and protein substitute at meals and mid meals (PKU group only).

\*\* average of 5 days; \* Fisher test; (n) = number of children; # Wilcoxon signed rank.

*3.5. Macronutrient Intake*

3.5.1. Reproducibility (A Measure of Whether the FFQ Produces the Same Results at Different Times)—FFQ 1 vs. FFQ 2

There was no significant difference between the two FFQs for PKU for any nutrients or between the two control FFQs except for protein and starch (*p* = 0.05) in control children, with FFQ 1 reporting values slightly higher than FFQ 2 (Table 3). Similarly, correlation *r* values all exceeded 0.5 for nutrients in both PKU and control groups, showing good correlation between FFQs taken at different intervals. Bland–Altman plots also demonstrated no clinically significant differences with mean levels close to 0 and homogeneous data mostly within the upper and lower levels of agreement (2 standard deviations—SD) (Figure 2).



**Figure 2.** *Cont*.

**Figure 2.** Bland–Altman plots for PKU and control group macronutrient intake FFQ 1 vs. FFQ 2. bias line (mean); upper and lower levels of agreement 95% confidence (2 SD). CHO = carbohydrate.

#### 3.5.2. Criterion Validity (Comparison with an Existing Standard)—FFQ vs. Dietary recalls

For the PKU group, there was a trend for the FFQ to report higher intakes of all nutrients compared to the dietary recalls (Table 3). In the control group, the same was observed except for energy, starch and fat. Nutrient correlations for the PKU group were close to or above 0.5 (r) except for fat. For the control group, correlations were less strong, ranging from 0.33 to 0.55. Conversely, most Wilcoxon *p* values for the PKU group were significant except for protein and starch, whilst in the control group fewer nutrients showed statistical differences, with only protein, fat and energy not showing a difference. However, from a clinical perspective, differences were not of relevance. For example, the difference in sugar intake between FFQ and dietary recalls for the PKU group was around 25 g or approximately 1 tablespoon per day, whilst the difference in fat intake was around 5 g or 1 teaspoon of fat per day. The Bland–Altman plots show homogeneous data with most values falling within the upper and lower levels of agreement (2 SD) and mean values close to 0 (Figure 3). The exceptions to this were sugar and fibre for the PKU group, and sugar for the control group.

**'LIIHUHQFHYVDYHUDJH%ODQG\$OWPDQRI&21752/N-))4YUHFDOOV \$YHUDJH 'LIIHUHQFH**  

**Figure 3.** *Cont*.

355

**Figure 3.** *Cont*.

**Figure 3.** Bland–Altman plots for PKU and control group macronutrient intake FFQ vs. dietary recalls. Bias line (mean); upper and lower levels of agreement 95% confidence (2 SD); CHO = carbohydrate.

3.5.3. Construct Validity (Ability to Distinguish between Different Groups)—PKU FFQ vs. Control FFQ

As expected, and in agreement with previous research, due to the composition of a phenylalanine restricted diet there were significant differences in macronutrient intake between the PKU and control groups when using the FFQ (Table 3). The PKU group had significantly higher carbohydrate and starch intakes, and a higher percentage of energy from carbohydrate and a lower percentage of energy intake from fat compared to controls.

#### *3.6. FFQ Individual Food Items*

3.6.1. Reproducibility (A Measure of Whether the FFQ Produces the Same Results at Different Times)—FFQ 1 vs. FFQ 2

Most food items for both PKU and controls showed good correlation between FFQ 1 and 2 (*r* > 0.40), demonstrating good reproducibility (see Supplementary Materials Table S2). Foods with a lower correlation coefficient were usually consumed by fewer than 10 subjects. The exceptions for the PKU group were vegetarian gummy sweets, pasta sauce, dried fruit and regular biscuits. For the control group, exceptions were meat pie, meat curry and butter/margarine.

Similarly, for commonly eaten foods (> 10 subjects) there was no significant difference (*p* > 0.05) between FFQ 1 and 2 for most food items in either group. Exceptions in the PKU group included: corn/rice/oat-based breakfast cereal, sweet drinks and vegetables containing phenylalanine < 75 mg /100 g and those with >100 mg/100 g. Exceptions in the control group included the following: dairy desserts, wheat-based breakfast cereals, mayonnaise/dressings, pizza and crackers. No items with a low *r* value (<0.40) were significantly different (*p* < 0.05).

#### 3.6.2. Criterion Validity (Comparison with an Existing Standard)—FFQ vs. Dietary Recalls

There was a trend for the FFQ to report higher intakes compared with the dietary recalls for just over half the items (n = 54/89, 61% PKU; n = 35/69, 51% control) (See Supplementary Materials Table S3). Similarly, for most foods, the FFQ reported more people consuming individual foods than the dietary recalls (FFQ 85%, n = 76/89 foods vs. recalls 15%, n = 13/79 foods for PKU; FFQ 83%, n = 57/69 foods vs. recalls 17%, n = 12/69 foods for controls).

Most food items for both PKU and control groups showed good correlation between the FFQ and the dietary recalls (*r* > 0.40), demonstrating satisfactory criterion validity except for items consumed less often (<10 subjects). The exceptions for the PKU group were as follows: vegetarian gummy sweets, chips, vegetarian burgers and low-protein biscuits. For the control group, exceptions were greater in number and similar to those that varied between the 2 FFQ: meat pie, meat curry and butter/margarine, in addition to chips, processed meats, ice cream, cheese, cake, pizza, pasta and chocolate.

Similarly, for commonly eaten foods (>10 subjects) there was no significant difference between the FFQ and dietary recalls for most food items in either group. Those that did tended to be different than the items that had low *r* values (<0.40). Exceptions in the PKU group included vegetarian gummy sweets, and in the control group cake, gummy sweets and table sauces—which had low *r* values (<0.40) and were significantly different (*p* < 0.05).

There were some commonly eaten foods (>10 subjects consuming) that showed significant differences in the mean g/week consumed between the FFQ and dietary recalls in both control and PKU groups. These included the following: boiled, mashed and jacket potato, table sauce, crisps and vegetables containing phenylalanine >75 mg /100 g. These foods tended to be considerably higher in the FFQ, except for crisps which were lower compared with the dietary recalls.

#### 3.6.3. Construct Validity (Ability to Distinguish between Different Groups)—PKU FFQ vs. Control FFQ

There were significant differences between the intake of PKU and control groups using the FFQ, particularly in the foods expected to be different (see Supplementary Materials Table S4). This included higher protein foods that were consumed in greater quantities by controls: milk, cheese, soft cheese, dairy desserts, cream, wheat-based breakfast cereal, sandwich spreads, milk sauces, legumes, vegetables containing phenylalanine >75 mg/100 g, eggs, meat pies, meat curries, sugar-free drinks (usually containing aspartame), hot chocolate powder, nuts/seeds, and regular varieties of bread, bread rolls, pasta, pizza, biscuits, cakes, puddings, jelly, chocolate, gummy sweets and crackers. In addition, the higher carbohydrate/fat foods allowed since they are low protein/aspartame-free, were higher in the PKU group; these included: sweet spreads, mayonnaise/dressings, sweetened drinks (aspartame free), sugar, other sweets and butter, in addition to vegetarian varieties of foods such as burgers, pies and curries. Additionally, some foods commonly used as protein exchange foods were higher in the PKU group: tinned pasta, processed potato and potato or corn-based crisps.

#### **4. Discussion**

This is the first FFQ validated for children with PKU, with data suggesting that it is an effective, accurate and practical tool for estimating energy and macronutrient intake as an alternative method to dietary recalls. It identified dietary patterns, the quality of natural protein consumed, and adherence with protein prescription.

This FFQ demonstrated excellent reproducibility when administered at a mean time interval of 5 weeks. PKU group meal patterns were similar, and all nutrients showed good correlations (*r* > 0.6). The protein amounts in the PKU group had a correlation of 0.91, demonstrating that the FFQ reliably estimated usual intakes with similar accuracy to repeated 24-h dietary recalls. In addition, individual foods generally showed good correlation (*r* > 0.4) if they were commonly consumed items (eaten by >10 individuals). Discrepancies between FFQ 1 and 2 for individual foods may be explained by differences in interpretation between the two questionnaires or in participant memory of the types of food consumed at the various time points. For example, parents were sometimes ambivalent about the sugar content of drinks their children consumed, particularly if drinks were consumed at school/nursery or outside of the home. However, this could equally vary across the dietary recalls. Furthermore, some foods on the FFQ were rarely consumed (by <3 individuals). In order for a food itemisied on a FFQ to contribute to absolute intake or differentiate between individuals, it should be eaten regularly and by a significant number of the study population [9]. Therefore, some foods were removed from the study FFQ following analysis.

A minimum correlation coefficient of 0.3 to 0.4 has been suggested to detect associations when validating FFQs [9]. In this study, all nutrient correlation coefficients were above 0.5 for the comparison of the 2 FFQs in both groups, and above 0.4 for the comparison of FFQ and dietary recalls, except for fat in both groups and energy and protein for the control group only. Similar correlation results were shown in other validation studies [18–23].

Bland–Altman plots were used to display the stability and direction of the bias across levels of intake [19]. Agreement was considered reliable if the difference between the two measures for reproducibility (FFQ 1 vs. FFQ 2) or validity (FFQ vs. recalls) was within 2 standard deviations (SD) of the mean [10]; the mean was close to 0; and demonstrated homogeneous data. Expert consensus suggests a combination of correlation or regression statistical methods together with Bland–Altman analysis should be used to assess reproducibility and validity of a FFQ, rather than any one single method [14].

To be truly valid, reported dietary intake from any assessment method should not be significantly different to actual intake, however there are practical difficulties with measuring 'absolute validity'; thus alternatively, 'comparative validity' (comparing with an alternative or 'reference method') is reported [24]. There is no gold standard method for recording dietary intake, all have limitations: weighed food records require a high level of subject commitment, adherence and understanding that would have excluded some recruits from this study; 3-day food diaries represent the current diet, rather than typical or usual intake over time; doubly labeled water, a more accurate method for comparison of energy intake, is expensive and requires specialised equipment that may be intimidating to children. Repeated dietary recalls have been previously used for validating FFQs [22,23,25]; whilst single day recalls do not account for day-to-day variability in food intake or episodically consumed foods [19], we chose to complete multiple 24-h recalls over a 4–10-week period to capture a more realistic picture of usual intake over time. This approach is supported by a systematic review of the validity of different dietary assessment methods compared with doubly labeled water, suggesting that multiple 24-h dietary recalls conducted over at least 3 days and using parents as proxy reporters was the most accurate method for children aged 4–11 years [25]. Neither the FFQ nor the multiple 24-h dietary recalls are likely to measure actual macronutrient intake with precision, as both are subject to recall bias; however, both methods produced a similar picture of intake.

Our FFQ designed for PKU demonstrated acceptable criterion validity when compared with the chosen reference standard, repeated 24-h dietary recalls. Total natural protein intake only varied by 0.5 g (25 mg phenylalanine) per day between methods, which is a very good correlation. There was some variability between assessment methods for the percentages of children reporting that they consumed food at mid meals; however, this is likely to be something that varies in individuals from day to day. Some individuals with PKU may also choose to consume their protein substitute in place of a snack between meals so as not to reduce appetite at main meals.

In keeping with other validation studies comparing FFQs with other methods [18–20,22,23] there was a tendency for the FFQ to report higher nutrient and individual food intakes than the dietary recalls. FFQs have been reported to overestimate dietary intake in children resulting from the use of adult portion sizes [10]; however, we overcame this by developing pictorial child-size portions. The main difference between the FFQ and dietary recalls was not so much the quantity of a food consumed, but less variation in the types of foods consumed for the dietary recalls. This reflects one of the limitations of dietary recalls in that they only capture recent intake rather than habitual food intake.

Consistent with previous studies looking at the macronutrient content of the PKU diet [26–28] our results demonstrated that the FFQ can differentiate the differences in macronutrient and individual food intake between children with PKU and children in the general population that would be expected. This substantiates good construct validity.

FFQs rely on recall over a longer assessment period than other methods and hence are associated with less accurate quantification. It is suggested that children under the age of 8 years may have difficulty recalling food intake, estimating portion size and conceptualizing frequency of food consumption [24,25]. The ability to cognitively selfreport dietary intake accurately is commonly given as approximately 12 years [24,25]. Previous research has shown that when older children complete a FFQ, they receive less assistance from parents, and this can result in a greater number of inaccuracies [2]. There may also be anomalies in data (from both dietary assessment methods) for adolescents due to inaccurate self-reporting and the highly variable food patterns commonly seen in this age group [19]. In this study, children aged 11–16 years completed the FFQ and

recalls themselves which may have led to misreporting, although parents were able to assist. Furthermore, children with PKU have a more repetitive food pattern, receive dietary education and are accustomed to measuring portion sizes and completing dietary assessments. Correlations for the FFQ compared with the dietary recalls were stronger for the PKU subjects than for controls, suggesting that PKU subjects or their parents may have had better dietary recall than the control group [14,29].

Completion of any dietary assessment method may draw participants' attention to their diets [9], and there is also the risk of subjects responding in a way that demonstrates good adherence only in the presence of a dietitian. FFQs were administered by an IMD dietitian trained and experienced in dietary assessment rather than self-completion due to anticipated initial difficulties of comprehension and interpretation. As four different dietitians were involved in administration, there may have been some degree of interrater reliability. However, a standard script was used to administer questionnaires and food recalls to minimise this. It is anticipated that with repeated use, parents/carers and adolescents (>12 years) would be able to self-administer the FFQ independently. Recent studies have demonstrated that technology-assisted methods, such as an online FFQs, performed equally as well in estimating intakes as doubly labelled water and other methods [23,24]. As such, further analysis of this tool after regular use and with an online version may be warranted.

#### **5. Conclusions**

A FFQ can simplify dietary data collection in PKU, particularly if patients are familiar with the tool and can complete it electronically before clinic appointments. This low-protein FFQ designed for use in patients with PKU yielded comparable data to repeated dietary recalls, and can be validly used to collect data on usual food and nutrient intake in place of other dietary assessment methods. It will also enable assessment of the dietary patterns that may lead to lifestyle diseases, such as obesity in PKU, and in turn will facilitate tailored health messages to the PKU population that will help to reduce the incidence of healthrelated illness. It could also be particularly important in assessing the impact of dietary changes associated with pharmaceutical treatments in PKU. Further testing of an online version of the FFQ is warranted.

**Supplementary Materials:** The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/nu14081595/s1: Table S1: Mean z-scores for BMI, weight and height for PKU and control children; Table S2: FFQ food items mean g/week FFQ 1 vs. FFQ 2 (PKU and Control); Table S3: FFQ items mean g/week FFQ vs. Dietary recalls (PKU and Control); Table S4 Kapp coefficient statistics; Supplementary S1: Script for FFQ (PKU); Supplementary S2: Script for FFQ (Control); Supplementary S3: Script for 24-h dietary recall (PKU); Supplementary S4: Script for 24-h dietary recall (Control).

**Author Contributions:** Conceptualization, S.E. and A.M.; data curation, S.E., R.J. and A.M.; formal analysis, S.E., R.J. and A.M.; funding acquisition, S.E.; investigation, S.E., C.A., A.D., A.P. and A.M.; methodology, S.E., C.A., A.D., A.P. and A.M.; project administration, S.E. and A.M.; supervision, A.M.; validation, S.E., R.J. and A.M.; visualization, S.E. and A.M.; writing—original draft, S.E. and A.M.; writing—review & editing, S.E., C.A., A.D., R.J., A.P. and A.M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was funded by the Birmingham Children's Hospital Charities Research Foundation (BCHRF407).

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki, and approved by the London—Queens Square National Ethics Research Service Committee, UK (protocol version 3, 18 April 2019, approved 16 May 2019; REC reference: 15/LO/1463; IRAS ID: 185896).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We would like to thank all the children and parents/carers who participated in the study.

**Conflicts of Interest:** S.E. has received research funding from Nutricia, financial support from Nutricia and Vitaflo to attend study days and conferences; C.A. received honoraria from Nutricia and Vitaflo International to attend study days and conferences; A.D. research funding from Vitaflo, financial support from Nutricia and Vitaflo to attend study days and conferences; A.P. has received an educational grant from Cambrooke Therapeutics and grants from Vitaflo International, Nutricia, Merck Serono, Biomarin and Mevalia to attend scientific meetings; A.M. has received research funding and honoraria from Nutricia, Vitaflo International, and Biomarin. She is a Member of the Advisory Board Element (Danone-Nutricia). The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Nutrients* Editorial Office E-mail: nutrients@mdpi.com www.mdpi.com/journal/nutrients

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel: +41 61 683 77 34

www.mdpi.com

ISBN 978-3-0365-5320-7