Importance of nutrition to human health
In 1912, the term ‘vitamine’ was coined leading to the discovery of the first vitamin (vitamin B1 also called thiamine) in 1926(Reference Semba1). Modern nutrition science, thus, first focused on the discovery, description and treatment of diseases and conditions due to single nutrient deficiencies, the definition of recommended daily allowances and food fortification. Over the decades, the focus changed, and nutrition science confronted the challenge of studying the associations of foods and dietary patterns with cardiometabolic diseases and other chronic diseases, using data from prospective population-based cohorts or intervention studies(Reference Mozaffarian, Rosenberg and Uauy2).
The Global Burden of Disease study provides the most comprehensive estimates of diet-related burden worldwide. It is clear from the Global Burden of Disease study that there is enormous potential to improve population health by reducing nutritional risks and seizing nutritional opportunities. According to the Institute for Health Metrics and Evaluation, diet was the third highest risk for the Global Burden of Disease in 2019, following high blood pressure and tobacco. In 2017, 22 % of all deaths were attributable to an unbalanced diet, mainly via the increased risk of CVD(3). Overall, it is estimated that obesity was responsible for 160 million disability-adjusted life years (DALY) and 5 million deaths in 2019, with a high burden across all regions of the world(Reference Chong, Jayabaskaran and Kong4). Obesity-related DALY and mortality are expected to increase by nearly 40 % in the coming decade(Reference Chong, Jayabaskaran and Kong4).
Nutrition is on the top priority list of many supranational organisations such as the WHO, which defined objectives to promote healthy nutrition and decrease the risk of noncommunicable diseases that should take place within the boundaries of sustainable development(5). The 2030 Agenda for Sustainable Development, adopted by all United Nations Member States in 2015, includes several goals related to nutrition such as ‘End hunger, achieve food security and improved nutrition and promote sustainable agriculture’ or ‘Ensure healthy lives and promote well-being for all at all ages’ (United Nations Decade of Action on Nutrition 2016–2025: https://www.un.org/nutrition/; WHO 2030 Agenda for Sustainable Development: https://www.who.int/europe/about-us/our-work/sustainable-development-goals). One-third of all man-made greenhouse gas emissions are a result of food systems(Reference Crippa, Solazzo and Guizzardi6,7) . The EAT Lancet Commission report stated that about half of man-made greenhouse gas emissions could be attributable to food choices by 2050(Reference Willett, Rockström and Loken8). Considering the climate crisis, diets are likely to substantially change in the coming years. This emphasises the importance to continuously monitor what people eat and to continuously assess links between diet and health status.
In Switzerland today, food is always available in great variety and abundance, but many people still consume an unbalanced diet including excessive intake of salt, sugar and fatty foods, which increase the risks of developing noncommunicable diseases such as diabetes, obesity or CVD. In addition to the human suffering they cause, such diseases account for around 80 % of Swiss healthcare costs. While data on nutritional deficiency are lacking on a country level, it has been shown in smaller studies that vitamin D, folic acid and Fe are at-risk nutrients also in the Swiss population, in particularly in more vulnerable populations sub-groups(Reference Christoph, Challande and Raio9–Reference Andersson, Egli and Zimmermann11). Also, iodine, a historical public health focus in Switzerland and a global success story, requires constant monitoring(Reference Fischer, Andersson and Braegger12). To prevent nutrition-related diseases, the Swiss Nutrition Policy 2017–2024 has been developed along with an action plan focusing on four actions areas, that is, information and education; framework conditions; coordination and cooperation and monitoring and research. In addition, a strategy and action plan have been developed to promote the implementation of the UN 2030 Agenda for Sustainable Development. In the field of research, it is important to have a coherent contribution of the key actors in nutrition, especially researchers from main institutions in Switzerland. To reach this goal, an initiative has been developed, the Swiss Research Network-Healthy Nutrition(13).
Current status of epidemiological nutrition research in Switzerland
To date, various population-based studies collecting information on diet or nutritional status have been conducted in Switzerland. Table 1 summarises studies with assessment of diet, whereas Table 2 provides an overview of studies with assessment of specific food items intake or nutrient status conducted in Switzerland in the last decades.
Table 1. Overview of population-based studies conducted in Switzerland including assessment of the diet of participants
NRP1A MONItoring of trends and determinants in CArdiovascular disease.
Table 2. Overview of population-based studies conducted in Switzerland including the assessment of specific food items intake or nutrients status
The MONItoring of trends and determinants in CArdiovascular disease(Reference Wietlisbach, Paccaud and Rickenbach14) and the National Research Project 1A(Reference Gutzwiller, Nater and Martin16) were two population-based studies aiming to investigate cardiovascular and lifestyle risk factors in Switzerland between the 1970s and 1990s. In both studies, a mortality follow-up was established through linkage of census and mortality data, and simplified food checklists with yes/no questions on the consumption of specific foods were used to assess diet(Reference Wietlisbach, Paccaud and Rickenbach14–Reference Gutzwiller, Nater and Martin16). The Swiss Health Survey is a nationally representative cross-sectional survey, conducted every 5 years since 1992, with the aim to collect information on the health status of the Swiss population(17). The dietary assessment is conducted as part of an extensive questionnaire on health behaviours, via short questions on the consumption of selected food groups. The Swiss Food Panel 1·0 and 2·0 are two longitudinal studies focusing on eating behaviours and covering the German- and French-speaking part of the country(Reference van der Horst and Siegrist18,Reference Siegrist and Hartmann19) . The dietary assessment methods used differed slightly among the two studies: the Swiss Food Panel 1·0 used a FFQ specifically developed for the study and considering food groups with unfavourable health effects or established by dietary guidelines, whereas the Swiss Food Panel 2·0 used a semi-quantitative FFQ adapted from the US Nurses’ Health Study and inquiring about the consumption of forty-seven types of food and beverages(Reference van der Horst and Siegrist18,Reference Siegrist and Hartmann19,Reference Hartmann, Dohle and Siegrist37,Reference Hagmann, Siegrist and Hartmann38) .
Four studies used a validated semi-quantitative FFQ to assess diet. The Bus Santé study is a community-based long-term survey, designed to assess cardiovascular risk factors of the population of Geneva,(Reference Morabia, Bernstein and Héritier20) CoLaus(Reference Firmann, Mayor and Vidal22) and PsyCoLaus(Reference Preisig, Waeber and Vollenweider23) represent two components of a single-centre cohort aiming at assessing risk factors for CVD and psychiatric disorders, respectively, in the population of Lausanne aged 35–75 years at baseline(Reference de Abreu, Guessous and Vaucher39). Sapaldia 3 is the third follow-up of the nationally representative Sapaldia cohort study with associated biobank designed to explore broad health and ageing effects of air pollution and the exposome more broadly(Reference Steinemann, Grize and Pons25,Reference Vlaanderen, de Hoogh and Hoek40) .
Finally, the first National Nutrition Survey of the Swiss adult population was conducted in 2014/2015 (menuCH)(Reference Chatelan, Beer-Borst and Randriamiharisoa27), and an analogous study in children is conducted in 2023/2024 (menuCH-Kids)(28). Both studies are cross-sectional and assess diet using two non-consecutive 24-h dietary recalls, providing crucial insights into the food consumption and dietary habits of the Swiss adult and children populations.
With respect to the assessment of specific food items or nutrients, only Se, iodine and Zn status have been measured in the Swiss population on a national level(Reference Burri, Haldimann and Dudler33,35) . Additionally, a survey on the intake of dietary supplements was recently conducted by the Federal Food Safety and Veterinary Office(Reference Gut and Fröhli29), and a population-based survey investigating the consumption of salt in the Swiss population was conducted twice(Reference Glatz, Chappuis and Conen30).
Gaps in Swiss nutrition research
Large-scale Swiss longitudinal data
Only few studies described above provide comprehensive dietary data collected using validated dietary assessment methods and were conducted nationally and longitudinally. As shown in Tables 1 and 2, Switzerland has conducted several studies that assessed diet. They do, however, differ strongly with respect to how diet was assessed and, thus, the quality of dietary information. A major disadvantage of most of these studies is that they were not intended to be cohort studies (with the exceptions of SAPALDIA, CoLaus-PsyCoLaus, Swiss Food Panels). Existing cohorts only cover parts of the Swiss population and are limited in size. However, Switzerland has three main language regions with cultural habits mirroring those of the large neighbouring countries, which influences the dietary habits of the population(Reference Pestoni, Krieger and Sych41). It is therefore important to conduct geographically comprehensive health and nutrition studies.
Some of these studies used a rather crude dietary assessment method, for example, checklists with ‘yes’ or ‘no’ questions on the intake of specific foods(Reference Wietlisbach, Paccaud and Rickenbach14,Reference Gutzwiller, Nater and Martin16) , short questions on the consumption frequency and occasionally quantity of broadly defined food groups(42), or focusing on specific foods only(Reference van der Horst and Siegrist18).
Although menuCH(Reference Chatelan, Beer-Borst and Randriamiharisoa27) and menuCH-Kids(28) provide essential detailed and nationally representative data on the food consumption and dietary habits of the adult and children population of Switzerland, they are cross-sectional(Reference Chatelan, Beer-Borst and Randriamiharisoa27,28) and, in the case of menuCH, do not include the collection of biological samples(Reference Chatelan, Beer-Borst and Randriamiharisoa27).
Most studies that have assessed dietary information in Switzerland were mainly conducted among adults. menuCH Kids(28) will at least partly fill this gap. However, studies with a life-course approach are missing in Switzerland. Nutritional trends such as an increasing prevalence of plant-based diets (vegans, vegetarians, flexitarians), the consumption of plant alternatives for meat and dairy, but also cooking skills and nutrition literacy are likely to differ between age groups and generations(Reference Brombach, Haefeli and Bartsch43,Reference Brombach, Bartsch and Gertrud Winkler44) . This underlines the importance for cohorts that do not only include middle-aged populations that are vulnerable for chronic diseases in the near future, but also younger and older populations. Besides missing information on some age groups, studies in pregnant and lactating women, people with handicaps, or people with migration backgrounds are missing.
Misreporting of nutritional intake is a major limitation of current nutritional studies. For example, Subar and colleagues(Reference Subar, Kipnis and Troiano45) reported that energy intake is considerably underreported on 24h recalls (12–14 % for men; 16–20 % for women) and even more on FFQ (31–36 % for men, 34–38 % for women). The direct assessment valid biomarkers of nutrients and food intake(Reference Subar, Kipnis and Troiano45,Reference Ulaszewska, Weinert and Trimigno46) (see sections Biomarkers and Reference methods) are thus key to more objectively measure dietary intake. In the future, the possibility of incorporating machine-learning based methods aimed reducing various sources of misreporting(Reference Popoola, Frediani and Hartman47) is worthy of investigation.
In addition to the above gaps, Swiss studies that assess dietary information hardly ever allow linking dietary data with health outcomes. For example, menuCH cannot be linked to health outcomes on an individual level. Even if this limitation was overcome by using geographical linking methods(Reference Suter, Pestoni and Sych48,Reference Suter, Karavasiloglou and Braun49) , individual level data would allow for increased precision and the possibility to better adjust for potential confounders. Also, although the cohorts Colaus-PsyColaus and Sapaldia have been linked to health outcomes, they are restricted to specific regions and only few clinical conditions can be studied. Larger surveys, including the Swiss Health Surveys, can be linked to mortality and cancer incidence using established linking methods, but are limited by crude dietary assessments(Reference Krieger, Pestoni and Frehner50–Reference Lohse, Faeh and Bopp53).
The existing Swiss cohorts have the potential to provide important information in the future. In particular, several of these cohorts already integrate repeated measures in their design (see Table 1); resurvey of existing study participants could help to address many issues around changing diets across differences ages or time periods. These possibilities are however limited as none of the studies above is currently able to address questions of nutritional transition with respect to, for example, (i) the health impacts of poorly (ultra) processed foods, (ii) the persistence of nutritionally caused cardiometabolic conditions and (iii) climate change and the insurgence of plant-based diets and the potential lower nutritional value of plant-based meat and milk alternatives, and (iv) the overall consequences and extent of a broader uptake of vegan and vegetarian diets.
Quality of the Swiss food composition database
High-quality food composition databases are crucial in nutrition research to ensure accurate and reliable results, regardless of the dietary assessment methods used to record food consumption(54). Food composition databases should be updated frequently to ensure a good coverage of new products entering the market. In addition, they should ideally include a variety of different foods and a wide range of nutrients, vitamins, and minerals, as well as other food components linked to health outcomes. The Swiss Food Composition Database (https://naehrwertdaten.ch/en/) currently includes around 1’100 different generic foods and provides data for 40 nutrients. These numbers are rather low when compared to food composition databases from the large neighbouring countries Germany (approximately 15’000 foods and approximately 140 nutrients), France (approximately 3’200 foods and approximately 70 nutrients) and Italy (approximately 1’000 foods and approximately 120 nutrients). Compared to these food composition databases, the Swiss database lacks specific data for fatty acids (e.g. n-3, n-6, EPA and DHA), sugars (e.g. fructose, lactose and glucose), proteins (e.g. amino acids), trace elements (e.g. manganese and copper), and vitamins (e.g. vitamin K, including vitamin K2 and vitamin B7). Furthermore, anti-nutritive factors such as polyphenols, lectins, saponins or phytic acid, which are essential to calculate the nutrient bioaccessibility and bioavailability of key nutrients are not present in the database. The Swiss Food Composition Database should therefore be further developed and regularly updated, and resources invested in its maintenance.
Understanding consumer choices
The Swiss Food Panels by the ETH Zürich, among others, aims at connecting dietary habits and food consumption with aspects of food literacy(Reference Hartmann, Dohle and Siegrist37), drivers of food consumption(Reference Brunner, van der Horst and Siegrist55), and predictors of diet quality(Reference Sob, Siegrist and Hagmann56). Besides the Food Panels, no studies targeted psychological aspects of diet and food consumption to address questions such as ‘why do consumers choose foods’, ‘what drives eating patterns’ etc.
Even though Switzerland is a country with high mean household income and general mandatory health insurance, there are disparities in health and access to health care(Reference Long, Mackenbach and Klokgieters57–Reference Baggio, Abarca and Bodenmann59). Dietary habits are known to differ by region and by socioeconomic status, but what is missing is whether these differences are due to lack of knowledge (‘nutrition literacy’) or differences in accessibility of food (‘food deserts’) or lack of infrastructure in rural areas, or a mix of these factors. There is also no data available on overall cooking skills and ability as well as barriers to implement healthy nutrition in a private household. It will thus be important to conduct longitudinal research studies on dietary habits to understand how Swiss consumers actually handle food. This becomes even more important with respect to challenges due to climate change and the need for a transformation of the worldwide food system(Reference Willett, Rockström and Loken8).
Filling the gap with a Swiss nutrition cohort
State-of-the art assessments of diet and food environments are crucial in any large-scale population-based cohort like that one that is foreseen in Switzerland(Reference Probst-Hensch, Bochud and Chiolero60). A regular assessment of diet will help to overcome the limitations of existing studies as mentioned above. We argue that an investment in setting up a nutrition cohort in Switzerland is crucial. Ideally, such a cohort is implemented into the planned large-scale population-based Swiss Cohort & Biobank.
The model proposed in the Swiss Cohort & Biobank White Paper(Reference Probst-Hensch, Bochud and Chiolero60) is one of an internationally harmonised, large-scale (i.e. over 100’000 participants) long-term prospective population-based cohort covering all age groups. Such a cohort would provide the necessary data to support evidence-based policies, conduct population-based surveillance, and advance public health knowledge within the Swiss context. The large cohort is complemented by selected sub-cohorts targeting specific populations of interest (e.g. pregnant women, patients, vegetarians/vegans, etc). Such a cohort would allow for covering topics related to prevention, including risk and disease screening, and health promotion aiming at producing the evidence to implement health-in-all policies in Switzerland. The planned collection of medical imaging and biological samples would allow for producing population-based reference data, including in the field of nutrition. We here present and discuss in detail some of the methods that we consider most relevant for such a national project.
Laboratory tools for a nutritional cohort
Dietary intake assessment tools
Dietary intakes vary daily, across seasons and ages. Regional eating habits, increasing heterogeneity as in Switzerland, may further complicate the measurement process. In addition, individuals consume multiple foods and beverages with varying nutrient profiles and may consume dietary supplements.
To collect dietary intake data, researchers use self-report tools, which are affected by different types and degrees of measurement error(Reference Kirkpatrick, Reedy and Butler61). This leads to the suggestion of complementing them with objective measures (see Section Analytical targets of personalised nutrition). Nonetheless, validated biomarkers that reflect true intake are known for only few dietary components and mostly reflect recent food intake, and objective measures alone do not provide insights into what people actually consume and the related contextual factors. Thus, there is a clear value in the continued use of dietary intake self-report tools, acknowledging their strengths and limitations(Reference Subar, Freedman and Tooze62).
Established methods of dietary intake assessment in research mainly include FFQ, 24-h dietary recalls, (weighted) food records, and screeners(Reference Kirkpatrick, Vanderlee and Raffoul63). Besides their risk of recall bias, all these methods have specific advantages and drawbacks. For surveys, the European Food Safety Authority recommends to keep the burden for participants at a minimum by using two non-consecutive 24-hour dietary recalls in adults, and the 24-hour dietary recall method followed by a computer-assisted personal or telephone interview in infants and children(64). A short food propensity questionnaire is also recommended to collect information on the consumption of some less frequently eaten foods and food supplements. The combination of both allows for computing a participant’s habitual diet using methods such as the Multiple Source Method(Reference Harttig, Haubrock and Knüppel65). Various new technologies have emerged to collect dietary intake, including web-based dietary assessments with self-administered record such as myfood24 or the Automated Self-Administered 24-Hour Dietary Recall(Reference Carter, Albar and Morris66,Reference Subar, Kirkpatrick and Mittl67) .
In Switzerland, some studies used an electronic FFQ. Automated Self-Administered 24-Hour Dietary Recall is tested in different subgroups of the populations including children, adolescents, adults, and elderly(68) and accuracy of the automated dietary app MyFoodRepo is evaluated against controlled reference values from weighted food diaries(Reference Zuppinger, Taffé and Burger69). A multilingual (German, French) web-based FFQ for adults that captures food consumption of the past four weeks has been developed(Reference Pannen, Gassmann and Vorburger70) and its validation is underway.
Biosampling
Biological sample collection is a critical factor in study design. Various factors such as nutritional status, physical activity, and circadian rhythm, and the exposome more broadly can significantly influence metabolite levels and leave molecular fingerprints in human tissues and fluids(Reference González-Domínguez, González-Domínguez and Sayago71). However, the process of collection also alters the sample due to necessary or incidental additives in collection containers and sampling devices.
Blood is considered rich in information for clinical chemistry-based research and provides a temporal snapshot of an individual’s physical condition.(Reference Williamson, Munro and Pickler72–Reference González-Gross, Breidenassel and Gómez-Martínez74). Drawing large volumes of blood should be avoided especially in vulnerable subjects or where venous access is difficult as in children or the critically ill(Reference Williamson, Munro and Pickler72). To minimise the blood sample volume, capillary blood sampling and blood spot cards have been introduced for blood collection(Reference Locatelli, Tartaglia and D’Ambrosio75,Reference Hoffman, McKeage and Xu76) . These techniques have been applied in point-of-care assessments, drug development, medical monitoring, and nutritional studies but has been linked to analyte loss(Reference Hoffman, McKeage and Xu76,Reference Zimmer, Christianson and Johnson77) . Because of the relatively large volume of blood required for studies conducting a broad range of analytical tests, for example, multi-omics studies, traditional blood sampling remains the method of choice. However, the sensitivity of analytical technologies constantly improves as illustrated by the development of nanomaterials-assisted proteomics and metabolomics(Reference Wang, Li and Shu78) as well as single-cell analysis of omics datasets(Reference De Biasi, Gigan and Borella79).
Urine has been regarded as a reservoir for numerous metabolites originating from exogenous nutrients and drugs or endogenous substances(Reference Garde, Hansen and Kristiansen80). While spot urine (urine taken at a specified time of the day) collection is common, a 24-hour urine collection (pool of all voids within a 24-hour period), is considered the ‘gold standard’(Reference Witte, Lambers Heerspink and de Zeeuw81–Reference Fernández-Peralbo and Luque de Castro83). However, a complete urine collection in a 24-hour time interval may be difficult to obtain, especially when proper sample storage and transportation are considered to maintain sample integrity(Reference Bi, Guo and Jia73,Reference De Wardener84,Reference Harris, Purdham and Corey85) .
Saliva offers a less invasive yet powerful alternative to blood for clinical applications(Reference Drummer86,Reference Bellagambi, Lomonaco and Salvo87) . The amount and composition of proteins in saliva vary according to circadian rhythm, diet, age, sex, and physiology(Reference Battino, Ferreiro and Gallardo88). Molecules are generally found in nano- or picograms which makes the reliable detection of biologically active molecules difficult(Reference Lusa Cadore, Lhullier and Arias Brentano89,Reference Palanisamy, Sharma and Deshpande90) . The standardisation of saliva collection protocols reduces the high variation of saliva parameters and facilitates downstream analysis(Reference Jacobs, Nicolson and Derom91).
Less commonly investigated human tissues offer interesting complementary sources of biomarkers whose potential should be further investigated. For example, the carbon and nitrogen stable isotope ratios 13C/12C and 15N/14N in hairs can be used as dietary marker(Reference Votruba, Shaw and Oh92); nails emerge as an adequate matrix to evaluate the nutritional status of zinc(Reference Wessells, Brown and Arnold93); also the nutritional status of carotenoids can be measured by evaluating this compound in skin(Reference Ahn, Hwang and Kim94).
A wide range of metabolic products are produced by the microorganisms within the gut and are important to human health(Reference den Besten, van Eunen and Groen95). While faeces are readily accessible, the recruitment of individuals willing to participate in a study may be difficult due to due to various barriers. For metabolome studies, the most common practice is to freeze samples at –80°C, –40°C, or –20°C, sometimes aided by flash freezing in liquid nitrogen as it is unclear if stabilizing solutions adversely affect metabolite profiles(Reference Deda, Gika and Wilson96).
Biomarkers
The classical measurement of food and nutritional intake are self-reported food intake measurements(Reference Picó, Serra and Rodríguez97). While these have inherent limitations, the use of biomarkers enables the objective measurement of nutrient intake. Biomarkers are indicators that can be measured to inform about the normality or the disfunction of specific biological processes in response to multiple environmental and/or genetic factors such as gene polymorphisms, diet (e.g. nutrient intake and levels in the body), physiological status (e.g. pregnancy, lactation, ovarian cycle and menopause, physical exercise), physical and chemical exposures (e.g. environmental pollutants), lifestyle (e.g. stress levels) and various pathogenic processes and diseases. In nutrition, biomarkers can be either direct (e.g. nutrient itself) or indirect (e.g. nutrient-associated endpoints or functional biomarkers) measurements of the nutrient(s) of interest. Nowadays, a series of biomarkers are available in routine for both nutrition clinical practices and research. Such conventional biomarkers that are based on single nutrients show limitations and encourage the developments of a new generation of biomarkers that better reflect the metabolic processes in relation to diets. Metabolomic approaches enable to simultaneously quantify multiple metabolites representative of the systemic metabolic regulatory processes (see section Metabotypes). This makes metabolomics a suitable approach to discover new nutrient-associated metabolic patterns and thus additional functional biomarkers for nutrition. Whatever their direct or functional nature, biomarkers must fulfil the following specifications:
-
correlation with the rate of nutrient intake, at least within the nutritionally significant range, and respond to deprivation of the nutrient;
-
acceptable specificity and selectivity for the nutrient(s) of interest;
-
relation to a meaningful period of time;
-
indication of normal physiological function;
-
measurable in an accessible biological sample (e.g. typically blood and urine);
-
validated analytical method (linearity, accuracy, reproducibility) deployable in routine and at affordable cost;
-
availability of established normative data.
Classical biomarker measurement
Biomarkers of nutrient status measure the level of biological adequacy of nutrients in the organisms, for example, vitamin status biomarkers. Selected status biomarkers for micronutrients are reported in Table 3. Although classical biomarkers have advantages to be widely deployed in routine analysis for both general population and patient groups(Reference Berger, Shenkin and Schweinlin98) it is worth mentioning that basically all biomarkers have limitations and special attention needs to be paid to their interpretation.
Table 3. Examples of direct and functional biomarkers of micronutrient status
Biomarkers of nutrient exposure are used to quantify the recent levels of consumed foods or nutrients in biological fluids. Such biomarkers can help stratifying individuals according to their consumption patterns such as whole grain(Reference Ross, Bourgeois and Macharia99), fruit and vegetable, or meat and fish intake(Reference Dragsted100). However, many of the biomarkers of exposure still need validation as a high variation between individuals is often observed(Reference Dragsted100).
Show cases for classical nutrients in Switzerland: folic acid, vitamin D and iron
Although the Swiss population is generally considered ‘well nourished’ and mineral and vitamin deficiencies are not considered a major public health problem, no representative data exist on the prevalence and temporal development of nutritional deficiencies. Most of the data relies on small studies conducted in subgroups of the population.
Subgroups of the Swiss population may be at risk of folic acid deficiency, as a recent non representative survey identified 58 % of the study population with low plasma folate (14 nmol/l)(Reference Schüpbach, Wegmüller and Berguerand10). In addition, severe vitamin D deficiency defined as 25-dihydroxycholecalciferol concentrations below 25 nmol/l was found in 34,2 % of 1’382 pregnant women attending prenatal care between 2012 and 2015, while low status 25-dihydroxycholecalciferol concentrations < 50 nmol/l was identified in 73 % of the sampled women(Reference Christoph, Challande and Raio9). While prevalence of anaemia is low, a screening study including 672 young women of reproductive age recruited from high schools, the University of Zürich and ETH Zürich resulted in an estimated prevalence of iron deficiency of 22·7 % (serum ferritin <15 µg/l)(Reference Andersson, Egli and Zimmermann11). These studies are limited in scope, are not representative, and could be affected by sampling bias. Additionally, some of these analyses were conducted with convenience samples that were originally collected with different aims, not with the primary aim to assess nutritional status and its determinants, which can affect the outcome. Furthermore, the lack of a longitudinal component hampers in-depth analyses of predictors and associated factors. This is crucial for the identification of etiological patterns of nutrient deficiencies to design cost-effective and efficacious interventions.
Nutritional phenotyping
In order to expand abilities to capture nutrient-nutrient interdependencies and potentially to discover new biomarkers of nutrient status, the approach of nutritional phenotyping was introduced(Reference Rezzi, Collino and Goulet101). Nutritional phenotyping relates to the analytical possibilities to quantify a broad profile of nutrients and their related metabolites in biological fluids. This can be achieved thanks to the parallel use of complementary analytics including high pressure liquid-mass spectrometry, gas chromatography, inductively coupled mass spectrometry, and clinical chemistry. Within nutritional phenotyping, inductively coupled mass spectrometry is used to provide a quantitative profiling of elements enabling the so-called domain of ionomics. By analogy with metabolomics, ionomics aims at measuring the entire elemental composition of a living organism and its dynamics relative to genetic, physiological and metabolic variability(Reference Salt, Baxter and Lahner102). Although less known and applied than metabolomics, ionomics has proven efficient in the study of element metabolism in isolated cells and in biological fluids(Reference Konz, Monnard and Restrepo103,Reference Konz, Santoro and Goulet104) . Combined with recorded dietary information, nutritional phenotyping has the potential to study molecular interactions between the different nutrient families (amino acids, fatty acids, vitamins and minerals) while delivering information on classical biomarkers. This approach opens possibilities to identify nutrient patterns associated with various genetic, environmental, or phenotypic determinants that may help to identify novel nutrient status biomarkers.
A diverse range of pre-existing technologies, such as photography, microfluidics, wireless sensors and artificial intelligence may be combined and applied to nutrition research. These applications include the use of mobile phones to record and subsequently analyse dietary intake(Reference Mohanty, Singhal and Scuccimarra105), glucose sensors(Reference Boscari, Vettoretti and Amato106) or microfluidic-based skin sensors measuring nutrients(Reference Kim, Wu and Luan107).
Precision nutrition
Defining precision/personalised nutrition
The concept of personalised nutrition was put forward two decades ago in relation with the nutrigenomics approach promising to deliver personalised, health-directed, dietary guidelines, based on knowledge of the interactions between genes and diets(Reference Kaput, Ordovas and Ferguson108). This concept has evolved to integrate a more systematic approach of the interaction of diets with the human organism. It investigates gene–diet interaction, but also integrates different intrinsic datasets (epigenomics, transcriptomics, proteomics, metabolomics) at different structural levels of the human organisms (cells, organs, gut microbiota) under dynamic conditions, and considers environmental, extrinsic, factors such as physical activity(Reference van Ommen, van den Broek and de Hoogh109). The German Nutrition Society also proposes a model for personalised nutrition that goes beyond genetics to integrate phenotypic traits and the consumer(110–Reference Renner, Buyken and Gedrich113). Of note, a more detailed evaluation of the chemical composition of foods, for example, foodomics, is increasingly being recognised as an important component of nutrition research(Reference Bordoni and Capozzi114). During the last years, nutrition researchers have integrated new technological tools such as wearable technologies(Reference Shi, Li and Shuai115), Apps(Reference DiFilippo, Huang and Andrade116) and photographic evaluation of food labels and dietary intake(Reference Lazzari, Jaquet and Kebaili117,Reference Salathé, Bengtsson and Bodnar118) . Cutting edge bioinformatics and biostatistical approaches consequently became essential tools for an efficient extraction of the information derived from modern nutritional studies(Reference Dang and Vialaneix119). Increasing the technology around nutrition research has led to the more recent concept of precision nutrition, although definitions for differentiating these two terms (i.e. personalised nutrition and precision nutrition) are still unclear(Reference Ferguson, De Caterina and Görman120) and often used interchangeably(Reference Blaak, Roche and Afman121). The abbreviation PN (referring indistinctly to both personalised and precision nutrition) will consequently be used below to encompass this point. All in all, PN aims at using state-of-the-art, validated, analytical approaches to investigate the impact of nutrition, that is, nutrients, foods, and diets, on specific subgroups, even individual consumers. Given the level of detail accessible in modern nutritional studies, paradigms relying on discrete, homogeneous groups, may be less central and new data analysis tools such as principal component analysis (PCA) may be more promising, as the one-size-fits-all concept is this no longer valid. Some researchers push the boundaries of nutritional studies to the extreme case of single individuals in the case of n-of-1 studies(Reference Kaput122).
Initiatives on precision nutrition
Several countries have recognised the strategic and public relevance of moving nutrition research to a PN. Among these, in the USA, the NIH has recently started, within its 2020–2030 Strategic Plan for NIH Nutrition research, a program awarding $170 million over 5 years for PN(Reference Kaiser123,Reference Lee, Ordovás and Parks124) . The awarded program will investigate 10’000 participants who are part of the NIH’s USA cohort with 1’000’000 participants. The NIH consortium includes several centres integrating clinical evaluations, dietary assessments, metabolomics analyses and clinical assay, gut microbiome analyses, as well as data modelling and bioinformatics. These tools will be used by the NIH to integrate lifestyle, biological, environmental, and social factors to develop eating recommendations for individual that improve overall health(125).
At the European level, the Food4Me project was a precursor in 2012–2014 in developing and evaluating the personalised approach in nutrition research(Reference Stewart-Knox, Rankin and Kuznesof126), in particular by comparing three levels of personalisation based on diet, phenotype and genotype. This study concluded that PN-based advice achieves greater impact on dietary management by the participants(Reference Livingstone, Celis-Morales and Navas-Carretero127). The European Innovation Council has launched a call for a Pathfinder Challenge on precision nutrition. One of the objectives is to investigate causal relationships among diet, microbiome and glycans, with potential impact on personalizing human diet(128).
Analytical targets of personalised nutrition
Omics sciences at each level of the molecular flow of information in human cell are now integral parts of nutrition research, including nutritional cohorts. The analytical targets of PN not only investigate these molecular levels individually but also integrate them, and even goes beyond them, as presented in the following sections.
Genomes
A wide range of gene–diet interactions has been reported in the literature that impact on an equally broad range of phenotypic traits associated with metabolism(Reference Floris, Cano and Porru129) and metabolic diseases(Reference Barrea, Annunziata and Bordoni130). Besides several examples of monogenetic nutrigenetic tests targeting clinical endpoints such blood pressure(Reference Poch, González and Giner131), liver fibrosis(Reference Vilar-Gomez, Pirola and Sookoian132), myocardial infarction(Reference Cornelis, El-Sohemy and Kabagambe133), or obesity(Reference Corella, Peloso and Arnett134), the combination of a range of polymorphisms involved in a particular phenotypic trait was superior to monogenetic tests in the context of obesity management(Reference Kalantari, Doaei and Keshavarz-Mohammadi135,Reference Cha, Kang and Lee136) . Also, a genetic risk score based on SNPs associated with blood pressure may identify persons responsive to salt reduction(Reference Nierenberg, Li and He137).
Genetic testing in nutritional counselling has benefits and limitations, highlighting the need for reproducing the reported study and, more importantly, identifying clinically useful gene–diet interactions(Reference van der Horst, von Meyenn and Rezzi138). Therefore, reaching clinical usefulness in PN certainly requests that PN goes beyond genetic tests and polygenetic scores, and integrates post-genetic molecular factors in its evaluation.
Epigenomes
The environment, including the diet(Reference Delage and Dashwood139,Reference Alegría-Torres, Baccarelli and Bollati140) , is a key source of epigenetic modifications on DNA and histones. Nutriepigenetics is emerging as a key tool in PN. Dietary nutrients directly modulate DNA and even histones through the one-carbon pathway delivering methyl groups for epigenetic modifications. Work by Pembrey and colleagues suggested that access to calories of grandparents during puberty influence mortality of the grand-children though epigenetic modifications maintained across generations(Reference Pembrey, Bygren and Kaati141). Although highly controversial due to the difficulty in going beyond mere statistical associations, this work found echoes in other studies investigating the association between access to food in 1945 during the Dutch Hunger winter of 1944–1945, epigenetic modifications and a range of cardiometabolic endpoints in the subsequent generations, including birth weight(Reference Lumey and Stein142). As such, nutrition is a key source of interindividual variability in human biology and epigenetic marks may contribute to the existence of metabotypes(Reference Hellbach, Baumeister and Wilson143). In particular, interindividual variation in DNA methylation is associated with obesity(Reference Kühnen, Handke and Waterland144) and epigenetic modifications of genes involved in gene–diet interactions were shown, in addition to genetics, to modulate the efficiency of weight loss programs(Reference Sun, Heianza and Li145).
Metabotypes
The development of metabolomics in analytical chemistry has quickly found its application in nutrition research, evidently due to the primarily metabolic nature of human nutrition(Reference Ulaszewska, Weinert and Trimigno46). Metabolomics has become extremely popular in nutrition research thanks to its ability to study the quantitative expression of the real endpoints of the physiological regulatory processes, that is, the metabolites, in relation with health and disease outcomes, including nutrition-related metabolic risk factors for primary prevention(Reference Rezzi, Collino and Goulet101). Metabolomics is nowadays also a well-established approach to identify metabolic signatures associated with specific dietary intake from food groups to very specific foods such as dark chocolate(Reference Guasch-Ferré, Bhupathiraju and Hu146,Reference Rezzi, Ramadan and Fay147) or citrus fruits, proline betaine being one of the best characterised biomarkers of dietary intake(Reference Heinzmann, Brown and Chan148).
Recent research indicates that metabotypes associated with unfavourable metabolic status and incident disease occurrence are also characterised by diets low in vegetables, dairy products, and fibres, and highest intakes of total red and processed meat(Reference Riedl, Hillesheim and Wawro149). These findings open the door to the stratification of dietary guidance for consumers, based on their metabotypes(Reference Hillesheim, Ryan and Gibney150).
In that regard, defined as an extension of the pharmacokinetic concept in nutrition, nutrikinetics offers unique perspectives to study interindividual metabolic differences related to different food matrices(Reference van Velzen, Westerhuis and van Duynhoven151,Reference van Duynhoven, van Velzen and Westerhuis152) . It can capture interindividual differences in the response to nutrition and particularly to dietary phytochemicals that are metabolised by the gut microbiome, for example, microbiome co-metabolites hippuric acid, 4-hydroxyhippuric acid and 1,3-dihydroxyphenyl-2-O-sulfate as indicators of polyphenol-rich black tea consumption(Reference van Velzen, Westerhuis and van Duynhoven151). Because it enables to differentiate individuals based on their actual capability to process nutrients, also via the gut microbiome, nutrikinetics is foreseen as a powerful approach not only to infer host–microbiome nutrient interactions but also to objectively categorise individuals into fast/slow metabolisers for subsequent nutritional intervention trials.
Microbiome
Following the reduction in the cost of DNA sequencing and progresses in bioinformatics, the last decade has seen an explosion of research activities on the interactions between the gut microbiota and the human organism and, a few years later, on the impact of these interactions on health(Reference Flint, Scott and Louis153). A characteristic of the gut microbiota is its large interindividual variability(Reference Leeming, Johnson and Spector154) and the concept of enterotypes has emerged early in this field(Reference Arumugam, Raes and Pelletier155). Diversity in the gut microbiota is intimately associated with variability in dietary intake. This implies that dietary regimens aimed at improving health, for example, the immune status(Reference Wastyk, Fragiadakis and Perelman156), should be tailored based on the gut microbiome(Reference Johnson, Vangay and Al-Ghalith157).
In essence, the direct interplay between diet and the microbiota, with intestinal microorganisms directly utilizing ingested nutrients, underscores the microbiota’s pivotal role in human nutrition. Therefore, alongside genetics and epigenetics, understanding the composition and functions of the gut microbiota is crucial to understand how diet impacts human health and disease.
Reference methods
Nutrition is characterised by subtle effects, which when added over long term, exert a high impact on health outcomes. Cohort studies offer the opportunity to deploy reference methods longitudinally to measure these subtle effects directly and with high precision and link them to the nutritional and environmental exposome. This allows for the discovery and identification of novel, previously undescribed associations, as well as the improved understanding of the effect of dietary patterns and components on the underlying human physiology.
Such methods currently include the use of stable or long-lived isotopes, such as the long-term monitoring of bone calcium balance via the use of 41-Ca isotopes(Reference Hodges, Maiz and Lachcik158,Reference Denk, Hillegonds and Hurrell159) , stable iron isotopes to label body iron over the long term and to measure iron absorption and losses directly in the study participants(Reference Speich, Mitchikpè and Cercamondi160), or the measurement of energy expenditure with double labelled water(Reference Speakman, de Jong and Sinha161). A further, non-isotopic example for a high precision, established reference method is the utilisation of the CO rebreathing technique to assess blood volume and Hb mass, which improves the precision of the haemoglobin measurement substantially(Reference Schmidt and Prommer162). At term, PN is expected to make used of newly validated nutritional biomarkers to fuel the panel of reference methods available to researchers.
Non-Nutritional factors
The exposome refers to the totality of exposures from internal and external sources during the lifetime, including exposures to pollutants and other chemical and biological agents in addition to dietary compounds. The exposome is also contributed by psychosocial factors such as the socio-economic status(Reference Miller and Jones163). Understanding the impact of nutrition on the health of the Swiss Cohort and Biobank proposed in the white paper of Probst–Hensch and colleagues(Reference Probst-Hensch, Bochud and Chiolero60) thus requires that interactions between nutritional and non-nutritional elements of the exposomes and the human organism be taken into consideration. For illustration, the ability and/or willingness of participants in cohort studies to fully and adequately answer questionnaires, including dietary(Reference Ahn, Paik and Ahn164) or socio-demographic(Reference Andreeva, Galan and Julia165) questionnaires is subject to a large inter-individual variability that is influenced by factors such as older age, lower educational level, poorer health status and unhealthy lifestyle habits(Reference Tsiampalis and Panagiotakos166). These biases need to be considered, for example, by imputing missing data or decreasing their occurrence by improving the response rate based on neuro-psychological tools. In addition, physical activity can modulate the interaction between diet and human metabolism, for example, for the impact of genes on body weight(Reference Muhammad, Sulistyoningrum and Huriyati167). Also, the impact of the food environment on the intake of consumers at their residence, school, or workplace as well as their perception of this environment needs consideration(Reference Myers168,Reference Pineda, Bascunan and Sassi169) .
Towards nutritional systems biology
Can PN deliver on its promises?(Reference Ferrario, Watzl and Møller170) Penetrant phenotypic traits such as phenylketonuria are clear demonstrators of the potential of PN to translate knowledge into public health policies(Reference Lichter-Konecki and Vockley171). However, most chronic diseases are complex and the research field of PN consequently moves towards a combination of factors. The use of polygenic scores for the management of obesity illustrates this research direction although the clinical utility of this score has not been demonstrated yet(Reference Höchsmann, Yang and Ordovás172). Even traits with apparently clear causes cannot be pinpointed to isolated molecular events. For illustration, lactose intolerance not only involves polymorphisms upstream of the gene coding for lactase but is also modulated by epigenetic events as well as by the gut microbiota(Reference Porzi, Burton-Pimentel and Walther173).
The road to translate nutritional data into information that is relevant to the consumer’s health may thus well take the direction of systems biology and artificial intelligence(Reference Ferrario and Gedrich174). Indeed, biomedical research is currently embracing the concept of systems biology, which combines structural, dynamics, modelling, and omics analytical approaches to further foster the translation of research from the laboratory to the bed(Reference Yang175). Nutrition research follows this path by integrating elements such as the concept of the virtual patient(Reference de Graaf, Freidig and De Roos176), whole-body models integrating metabolism, physiology and the gut microbiota(Reference Thiele, Sahoo and Heinken177), phenotypic flexibility allowing for real-time evaluation of metabolism in response to a dietary challenge(Reference van Ommen, van den Broek and de Hoogh109), imaging techniques, such as functional MRI of the brain(Reference Francis and Eldeghaidy178), as well as combinations of multiple omics dataset(Reference Burton-Pimentel, Pimentel and Hughes179,Reference Badimon, Vilahur and Padro180) . For illustration, a retrospective cohort study used digital twin technology to reverse type 2 diabetes though precision nutrition(Reference Shamanna, Joshi and Shah181). This technology platform uses artificial intelligence to build a dynamic digital twin model of the patient with a broad range of data including, among others, clinical chemistry, dietary intake, exercise, and sleep recommendations.
The Swiss cohort as a tool towards precision nutrition
A key point to establish the Swiss cohort will be to characterise the relevant health outcomes in relevant population groups(182). The health outcome should be measured using appropriate clinical endpoints, including clear adjudication processes and validated risk factors to allow for the establishment of a high level of evidence for the investigated risk-outcome relationships. International efforts providing state-of-the-art insight into the assessment of risks, in particular dietary risks, will serve as basis for establishing the analytical strategy of the cohort(3,Reference Willett, Rockström and Loken8,Reference Zheng, Afshin and Biryukov183,Reference Murray, Ezzati and Lopez184) .
Determining the size of a cohort is a strategic issue that must consider, among others, scientific, economical, and logistic factors. For example, compared to other countries, Switzerland possesses a rather homogenous population when measured by socio-economic status; on the other hand, the geography (alpine region, plateau…) and culture (four national languages, high percentage of migrants, etc.) can be considered heterogenous. Estimating the size needed to have a representative cohort based on these factors is thus a complex task. Based on the experience gained internationally from existing large cohorts, Probst-Hensch and colleagues estimate in their White Paper that the Swiss Cohort & Biobank should enrol 100’000+ participants to account for the number and complexity of chronic diseases to be monitored and to allow for the identification of rare diseases(Reference Probst-Hensch, Bochud and Chiolero60). Nutrition research should thus join force with medical research to add its arsenal of research tools to the analysis of the exposome of the Swiss Cohort & Biobank.
Translating nutritional research into information that will impact on the health of Swiss consumers thus requests that established research tools be combined with ground-breaking technologies. In addition to validated risk factors, new technologies and will also lead to the discovery of candidate biomarkers that will fuel the conduct of additional studies to validate them (see section on biomarkers below). The diet of the Swiss Cohort and Biobank will be linked to the phenotypic traits of the participants using state-of-the-art methodologies. The phenotypic traits include the medical history of the participants, their clinical chemistry, focusing on validated risk factors, as well as omics-analyses along the cellular flow of information (DNA, RNA, proteins, metabolites) in the biological samples collected in the cohort. In particular, blood cells will be used for genetic and epigenetic analyses; faecal water, serum/plasma, and urine for metabolomics, and the faeces for genomics analyses of the microbiome. The nutritional biomarkers identified in the cohort will provide information on (i) dietary intake, to complement classical dietary assessment, (ii) effect of dietary intake on the metabolism, to better evaluate the nutritional properties of the nutrients, foods or diets of interest, and finally (iii) susceptibility to dietary intake, to foster personalised nutrition(Reference Gao, Praticò and Scalbert185). The nutritional biomarkers will be validated according to the following criteria: plausibility, dose-response, time-response, robustness, reliability, stability, analytical performance, and inter-laboratory reproducibility(Reference Dragsted, Gao and Scalbert186). Modern bioinformatics tools, including artificial intelligence, will be used to analyse associations in the triad diet–biomarkers–health.
Although longitudinal data in the Swiss cohort will provide some hints at mechanisms at play in the interaction of nutrients, foods and diets with the human organism and strong indication of causality can only be inferred from randomised controlled trial. Information gathered from the analysis of the triad diet-biomarker-health in the cohort will lead to the establishment of new nutritional hypotheses that will need to be tested in randomised controlled trials. These interventions could be conducted in subgroups of the Swiss cohort, using the so-called trials within cohorts design(Reference Gal, Monninkhof and van Gils187) or in independent study groups. An analysis of the dietary behaviour of Swiss consumers has identified a dietary cluster that is specific to Switzerland (‘Swiss traditional’) and close to Western diet(Reference Krieger, Pestoni and Cabaset188). Developing a healthy and sustainable Swiss diet and demonstrating its benefits in an intervention study could serve as a proof a concept for the ability of Swiss nutritional research to translate knowledge into practice. To this end, understanding the Swiss consumer and developing the methodologies to motivate changes in dietary habits will be key.
The integration of high precision measurements in the cohort will allow for both the precise characterisation of selected aspects of nutritional status, the measurement its longitudinal development and the identification of relevant health associations to inform future interventions and to identify and discover novel risk factors and health associations. This will expand the knowledge base for nutritional sciences, promoting discoveries, but also, by employing reference methods, allowing for overcoming long-standing controversies in the nutrition field.
Although public health and PN appears at first to follow two opposite strategies with regards to the number of persons targeted by research, namely entire populations v. the individual, the public health nature of a Swiss cohort can indeed be fostered through PN by targeting large groups of consumers. PN advice can be targeted to consumer clusters with specific dietary patterns with a potential impact on health(Reference San-Cristobal, Navas-Carretero and Celis-Morales189) or to groups of citizens in living specific geographical areas or specific environments(Reference Tsiampalis, Faka and Psaltopoulou190). PN will thus contribute to an increased credibility of the nutritional sciences with the public and to an overall advancement of public health.
Acknowledgements
The co-authors are members of the Swiss Research Network – Healthy Nutrition whose mission is to bundle and strengthen nutritional research in Switzerland. Swiss Research Network-Healthy Nutrition is organised in seven working groups covering all aspects of nutritional research, including food and diet characterisation, mechanisms and physiology, epidemiology, interventions, consumer sciences, life cycle assessment, and translation.
The authors are financed by their own institutions.
All authors wrote and/or critically revised the manuscript.
The authors declare no conflicts of interest.
Open access
