Inflammation and response to nutrition interventions

Abstract

The complex interplay between nutrition and inflammation has become a major focus of research in recent years across different clinical settings and patient populations. Inflammation has been identified as a key driver for disease-related malnutrition promoting anorexia, reduced food intake, muscle loss, and on a cellular level, insulin resistance, which together stimulate catabolism. However, these effects may well be bidirectional, and there is strong evidence showing that nutrition influences inflammation. Several single nutrients and dietary patterns with either proinflammatory or anti-inflammatory properties have been studied, such as the long-chain ω-3 fatty acids eicosapentaenoic acid or docosahexaenoic acid. The Mediterranean diet combines several such nutrients and has been shown to improve medical outcomes in the outpatient setting. In addition, there is increasing evidence suggesting that inflammation affects the metabolism and modulates the response to nutrition support interventions. In fact, recent studies from the medical inpatient setting suggest that inflammation, mirrored by high levels of C-reactive protein, diminishes the positive effects of nutrition support. This may explain the lack of positive effects of some nutrition trials in severely ill patients, whereas similar approaches to nutritional support have shown positive results in less severely ill patients. The use of biomarkers, such as C-reactive protein, may help to identify patients with a lower response to nutrition, in whom other treatment options need to be used. There is need for additional research to understand how to best address the malnourished patient with inflammation by specifically lowering inflammation through anti-inflammatory medical treatments and/or nutrition interventions.

INTERPLAY BETWEEN NUTRITION AND INFLAMMATION

Inflammation has been classified into acute or chronic inflammation, which differ from duration of the process as well as various immune factors.¹ Although chronic inflammation has been studied extensively and has been shown to be associated with cardiovascular disease and other illnesses,^2-4 acute inflammation is predominant in acute illness, lasting normally from minutes to a few days. Acute inflammation has a complex regulation and is mediated by innate immunity and the increase of several acute-period proteins, including C-reactive protein (CRP) (Figure 1).¹ In inflammatory response, leukocytes and mast cells are present, which lead to a “respiratory burst” as a result of an enhanced uptake of oxygen and, therefore, enhance the production and release of reactive oxygen species (ROS). Free radicals can also further induce inflammation^{5, 6} if not resolved within antioxidative capacities. In healthy body condition, there is a balance between ROS/free radical and endogenous antioxidant defense mechanisms. A disturbed balance, however, can lead to oxidative stress and associated damage. This oxidative stress state can cause injury to all vital cellular components, such as DNA, proteins, and membrane lipids, and it may lead to cell death (Figure 2). Additionally, it plays a major role in the development of illnesses, including cardiovascular diseases, diabetes, degenerative diseases, cancer, anemia, and ischemia.^{1, 7}

Figure 1

Different types of inflammation: the cascade of acute inflammation and selected associations of chronic inflammation with diseases. CRP, C-reactive protein; PAI-1, plasminogen-activator-inhibitor type 1; PCT, procalcitonin.

Figure 2

Oxidative stress, inflammation, and the influence of immunomodulatory nutrients. ROS, reactive oxygen species.

In addition to anti-inflammatory strategies, such as the use of glucocorticoids, there is important evidence suggesting that the inflammatory response can be modulated by nutrition. In fact, specific ingredients have been shown to have proinflammatory or anti-inflammatory properties. The most studied nutrition components in this regard with strong anti-inflammatory properties are specific polyunsaturated fatty acids (PUFAs), for instance the long-chain ω-3 FAs eicosapentaenoic acid, docosapentaenoic acid, or docosahexaenoic acid. On the other hand, certain long-chain ω-6 FAs like the arachidonic acid (AA) have shown proinflammatory effects.⁸ These nutrients are substrates for immunomodulatory mediators, such as prostaglandins, thromboxanes, and leukotrienes, which mediate either an anti-inflammatory or a proinflammatory effect depending on the starting substrate. Moreover, because of the use of the same enzymes in both pathways, ω-3 FAs have the potential to competitively reduce the metabolism of ω-6 FAs to proinflammatory mediators.⁹ Yet the effect of ω-3 FA supplementation on important clinical outcomes, such as mortality, has not been conclusively demonstrated in randomized controlled clinical trials.^{10, 11} Next to specific long-chain FAs, there are various bioactive substances, such as polyphenols, that influence inflammation in different ways, for example, by interfering with immune cell regulation, proinflammatory cytokines’ synthesis, and gene expression. They can also inactivate NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells) and modulate mitogen-activated protein kinase and AA pathways and influence inflammation this way.¹²

In recognition of the complex interactions of different nutrients within a particular diet, the focus has shifted towards research on the effects of dietary patterns instead of single nutrients.¹³ Two of the most extensively investigated dietary pattern in terms of its anti-inflammatory effects in the outpatient setting is the Mediterranean diet and a fiber-rich diet. Next to high intake of plant-based nutrients and nonprocessed food, both are often associated with a high intake of polyphenols and complex carbohydrates, which may have anti-inflammatory effects, including their turnover of nondigestible carbohydrates into immuno-regulators substances like short-chain FAs by the gut microbiota.^{14, 15} Importantly, most of the studies focused on long-term nutrition interventions and outpatients with chronic inflammation and did not investigate acute inflammation in acute medicine patients.¹⁶ There are, however, several medical options to reduce the systemic inflammatory response, including corticosteroids and anti–interleukin (IL) 6 inhibitors, which have been used successful in hyperinflammation during COVID-19.¹⁷ Nevertheless, within this review, we will focus mainly on anti-inflammatory effects of nutrition. Although there are convincing data showing that nutrition has a strong effect on inflammation, we are now starting to understand how inflammation affects the way the body responds to diet, which will be the focus of this review.

DISEASE-RELATED MALNUTRITION

Disease-related malnutrition (DRM) is a multifactorial complex syndrome that is negatively affected by disease-related systemic inflammatory response,¹⁸ contributes to high morbidity and mortality,^{19, 20} and seriously interferes with recovery from acute illness.²¹ In industrialized countries, malnutrition is mainly disease-related and its relevance is reflected by its prevalence. Data from Europe and the United States demonstrate that about 30% of patients have malnutrition or are at risk at hospital admission.^{18, 22, 23}

The European Society of Clinical Nutrition and Metabolism (ESPEN) proposes three etiological groups of malnutrition: DRM with inflammation, DRM without inflammation, and malnutrition/undernutrition without disease.²⁴ To diagnose malnutrition, the Global Leadership Initiative on Malnutrition (GLIM) proposes a 2-step approach consisting of a nutrition risk screening followed by a more detailed assessment. If a nutrition risk is detected, a nutrition assessment for diagnosis should be performed, including etiological (reduced food intake or assimilation, and disease burden/inflammation) and phenotypic (nonvolitional weight loss, low body mass index [BMI], and reduced muscle mass) criteria. According to GLIM, the diagnosis of malnutrition is made if one etiological and one phenotypic criterion apply for the patient.²⁵ The ESPEN classification and also the GLIM criteria include inflammation as an etiological factor, revealing its relevance in DRM. They also suggest that DRM with or without inflammation are different malnutrition phenotypes. However, the question whether these phenotypes also need a different therapeutic approach has not been conclusively answered today.

WHY IS OUR PATIENT BECOMING MALNOURISHED?

The pathogenesis of DRM is complex and often results from different factors, including starvation, acute or chronic disease (eg, polypharmacy, disease-related inflammatory mechanisms, and compromised intake or assimilation of nutrients), immobility-associated muscle wasting, and older age or social isolation. Focusing on inflammation, which has been suggested to be the main driver of DRM, there are several effects on the metabolism. There are measurable elevated levels of proinflammatory cytokines like tumor necrosis factor α (TNF-α), CRP, IL-1β, and IL-6. The release of those is mainly caused by activation of the hypothalamic–pituitary axis, a systemic response to a stressor like disease or other noxious stimuli. Affecting brain circuits, proinflammatory cytokines control food intake, delay gastric emptying, and increase skeletal muscle catabolism.^{18, 26} Further modulation of the hypothalamic-pituitary-adrenal axis provokes the release of stress hormones, including cortisol and catecholamines, and suppresses other hormones regulating sex, thyroid, and other peripheral functions.²⁷ In malnutrition, the deiodination of thyroxine (T₄) to triiodothyronine (T₃) was shown to be down-regulated—a process called “low T₃ syndrome,” which is an adaptive metabolic mechanism to reduce energy expenditure and prevent catabolism.²⁸ Catecholamines and cortisol increase glycogenolysis and hepatic gluconeogenesis excessively while simultaneously inducing peripheral insulin resistance and inhibiting glucose from entering peripheral tissues, such as the skeletal muscles.²⁷

HAVE WE MADE PROGRESS IN TREATING MALNUTRITION?

There has been intensive research on how to detect and diagnose malnutrition.²⁵ These findings have led to an improvement in early detection and thus the start of therapy. Although the possible benefits of nutrition therapy in malnourished patients have long been unclear,^{29, 30} recently there has been growing evidence in favor of nutrition therapy in the medical inpatient setting, also due to large high-quality randomized controlled trials (RCTs).^{19, 20} In a recent meta-analysis, Gomes et al found that nutrition therapy significantly reduces the risk for mortality by 25%. Also, nutrition interventions during the hospital stay reduced the risk of complications and hospital readmission, improved functional outcome, and prevented body weight loss in medical inpatients. Compared with studies published before 2014, which were predominantly heterogeneous and of small sample size, studies published after 2014 had an even more pronounced beneficial effect on mortality (odds ratio [OR]: 0.47, 95% CI: 0.28–0.79), which indicates an improvement in nutrition treatment.³¹ Further stratifications demonstrated that trials using high-protein nutrition interventions and whose intervention lasted longer than 60 days were the most effective. Among the trials, nutrition strategies used were dietary counselling, food modifications and optimization, supplement prescriptions, oral nutritional supplements, individualized nutrition plans, or nutrition care from healthcare assistants. There were no studies included using enteral or parental feeding strategies.³² These investigations provided important evidence that nutrition support is a highly effective therapy in medical inpatients with improvements in clinical outcomes.

Interestingly, these positive results were not consistent across different patient populations. A Cochrane review from 2017 also including critically ill and surgical patients (and higher-risk parenteral and enteral nutrition trials) has only found few significant beneficial effects on clinical outcome and some controversial effects.²⁹ Herein, a highly vulnerable population are patients treated in intensive care units (ICUs). These patients are severely ill and often are unable to feed volitionally by mouth for periods up to weeks.²¹ The catabolic response is much more pronounced than that evoked by fasting in healthy persons, since the energy deficit in acutely ill patients is often superimposed on immobilization and pronounced inflammatory and endocrine stress response.³³ Even after ICU discharge, the majority of patients remain nutritionally compromised and oral intake of food continues to be hampered by various barriers like low appetite, taste changes, impaired swallow function, gastrointestinal disturbances, or psychosocial factors.³⁴ A comparison of five large ICU trials (eg, EPaNIC trial with 4640 patients³⁵ comparing early or full nutrition support to standard care and more [Table 1]) showed that every nutrition intervention improved energy and protein intake. Nutrition interventions included a combination of oral, enteral, and parenteral nutrition. Interestingly, reaching higher energy and protein goals did not translate into improved clinical outcome and mortality, length of ICU stay, or need for mechanical ventilation in these trials. The findings of these recent adequately powered RCTs limit the number of nutrition interventions that can be confidently recommended for critical care patients. There is need for research on the mechanisms underlying benefit or harm from nutrition support and additionally to identify biomarkers of the anabolic recovery period to guide the initiation of more intense feeding. A validated scoring system to predict treatment response would potentially further help selecting patients and improving nutrition treatment.

Table 1. Comparison of mortality and length of ICU stay in five randomized controlled trials during critical illness.

	EDEN trial	Early PN trial	EPaNIC trial	SPN trial	TICACOS
Full name	Trophic vs full-energy enteral nutrition in mechanically ventilated patients with acute lung injury	Early parenteral nutrition in critically ill patients with short-term relative contraindications to early enteral nutrition	Early parenteral nutrition completing enteral nutrition in adult critically ill patients	Impact of supplemental PN on infection rate, duration of mechanical ventilation, and rehabilitation in ICU patients	Tight calorie control study
Number of patients	1000	1372	4640	305	130
Length of stay in ICU	Unaffected	Unaffected	Longer with early PN	Unaffected	Longer with REE
Mortality in ICU	Unaffected	Unaffected	Unaffected	Unaffected	Unaffected

• Abbreviations: ICU, intensive care unit; PN, parenteral nutrition; REE, resting energy expenditure.

WHY DO WE SEE DIFFERENCES IN TREATMENT RESPONSE?

An important research question in nutrition research is why there are such differences in treatment response among different patient populations. Although recent research has focused more on the detection and diagnosis of malnutrition, it will be important to now focus on understanding whether different malnutrition phenotypes exist that respond differently to different types of nutrition interventions. Herein, the severity of inflammation may be an important consideration. A hypermetabolic, burned, and highly inflamed patient unable to eat volitionally, who is treated in the receiving ICU, is obviously a different patient with different nutrition needs compared with an older geriatric patient with low appetite, who is not eating enough. Although there is already a consensus that nutrition protocols should be individualized regarding nutrition targets for inpatients (mainly based on their BMI and severity of illness) in critical care and hospital ward settings, current research suggests that tailoring nutrition to the specific medical and metabolic condition could further improve the effectiveness of nutrition intervention. Herein, a better understanding of the underlying pathophysiology is key to developing effective new interventions.¹⁸

PERSONALIZED NUTRITION AND THE USE OF BIOMARKERS

The concept of personalized nutrition is based on the observation that not all patients show the same response to nutrition interventions.¹⁸ Even in a homogenous cohort, clinically relevant differences can be seen regarding treatment response.¹⁹ The EFFORT trial (Effect of early nutritional support on Frailty, Functional Outcomes and Recovery of malnourished medical inpatients Trial) investigated the effect of individualized nutrition support to reach energy and protein targets in polymorbid medical inpatients at nutrition risk. This large non-ICU nutrition trial with >2000 patients was conducted in eight Swiss hospitals. Although the overall effect of nutrition support on mortality (adjusted OR: 0.65, 95% CI: 0.47–0.91; P = 0.011) and adverse events (adjusted OR: 0.79, 95% CI: 0.64–0.97; P = 0.023) was positive, there were some noteworthy differences within certain subgroups.

Focusing on different illness-related factors, Bargetzi et al investigated the effects of nutrition support on mortality in subgroups of patients stratified according to kidney function at the time of hospital admission (estimated glomerular filtration rates [eGFRs] <15, 15–29, 30–59, 60–89, and ≥90 ml/min/1.73 m²) in the EFFORT cohort.³⁶ Patients with a chronic kidney disease admitted to the hospital are often at risk for malnutrition or already malnourished and their status can further deteriorate during the hospital stay.³⁷ Bargetzi et al demonstrated that, within this population, patients with a lowered kidney function (eGFR 15–59  ml/min/1.73 m²) had a more pronounced survival benefit through nutrition support compared with patients with an eGRF >60 ml/min/1.73 m².

Another important analysis focused on inflammation within the EFFORT cohort. The most used inflammatory marker to assess inflammation in the acute phase of illness is CRP. When stratifying the EFFORT cohort according to CRP groups, nutrition support did not reduce mortality in patients with high CRP levels >100 mg/L as compared with patients with CRP concentrations ≤100 mg/L (Figure 3).³⁸ These effects were independent of infection and severity of disease and suggested that nutrition support is less effective in individuals with severe inflammation. Interestingly, a further analysis within EFFORT showed that the subgroup of patients with different types of cancer had a benefit of nutrition support, which again was not found in cancer patients with high inflammation.^{39, 40} This finding may explain the heterogeneity of results of nutrition trials, with some trials in the critically ill not showing beneficial results. Yet how to best approach the patient with severe inflammation is not yet clear, but it could include nutrients with anti-inflammatory properties.⁴¹ Further, it needs to be determined if CRP is a good marker to assess inflammation and possible treatment response in the population of ICU survivors, as a recent secondary analysis of the EPaNIC trial shows that a rise of CRP was not related to cytokine responses and thus did not reflect higher systemic inflammation.⁴² Furthermore, CRP as an acute-phase protein cannot represent the entire systemic inflammatory response, is influenced by noninfectious factors (eg, cardiovascular diseases, cirrhosis, cancer or systemic lupus erythematosus), and has a high interpersonal variability, and values can fluctuate from day to day.⁴³ Furthermore, it seems complicated to assess whether high CRP values reflect acute or chronic inflammation. Additional biomarkers, including proinflammatory cytokines (eg, interleukin, TNF), biomarkers of activated neutrophils and monocytes (eg, cluster of differentiation), infectious organisms, and related protein or receptors (eg, toll-like receptors, TNF receptors), would provide a more sophisticated assessment for inflammation.⁴⁴ To date, to our knowledge, no study has conclusively examined inflammatory status and response to nutrition therapy with a combination of the above inflammatory biomarkers.

Figure 3

Kaplan-Meier estimate for time to death within 30 days according to inflammatory status. Source: Adapted from Merker et al.³⁸ CRP, C-reactive protein.

Besides CRP, no other parameter has been considered to select patients for nutrition support. Albumin serves as a negative acute-phase protein and was historically used to assess nutrition status or to evaluate progress of nutrition support.⁴⁵ A recent study investigated the question whether selecting patients for nutrition support by serum albumin level or prealbumin is helpful. Although patients with a low serum albumin level (<30 g/L) had a significantly higher risk to die within 30 days (adjusted hazard ratio [HR]: 1.4, 95% CI: 1.11–1.77; P = 0.005), the effects of nutrition support on 30-day mortality were similar for patients with a low serum albumin level compared with patients with normal serum albumin concentrations, with no evidence for a subgroup effect.⁴⁶ As albumin has a particular long half-life time, prealbumin with a half-life time of approximately 2 days, may be more accurate for characterization of patients in the acute care setting. A post hoc analysis demonstrated a similar prognostic value of prealbumin (adjusted HR: 1.59, 95% CI: 1.11–2.28; P = 0.011) for mortality. However, regarding the predictive value of treatment response, prealbumin also failed to identify patients who benefit from nutrition support.⁴⁷

It is worth noting that there is a wide range of possible biomarkers that have been studied in regard to assessnig malnutrition. Next to common serum visceral proteins like mentioned albumin or prealbumin, also transferrin or retinol-binding protein have been considered. Other laboratory markers that are associated with malnutrition are urinary creatinine, urinary 3-methylhistidine, serum cholesterol, blood lymphocyte count, serum insulin-like growth factor-1 (IGF1), serum leptin and many more. Also, proteonomic and metabolomic markers have been studied.^{48, 49} Because of their association with malnutrition, these different markers may be considered potential candidates for predicting response to nutrition treatment, but it is important to consider individual limitations of each marker.⁵⁰ Besides, hypothesis-generating studies on treatment response are lacking so far.

To summarize, in medical inpatients at nutrition risk, there are some promising biomarkers, which might help to personalize nutrition support in the near future. Clearly, validation studies will be needed, and results are so far rather hypothesis-generating and mostly generated by secondary analyses. A possible potent and well-studied biomarker for treatment response is CRP in non-ICU patients, which represents acute inflammation but has low specificity. Using biomarkers to stratify patients according to their malnutrition phenotype and their treatment response will be key to evolve from traditional treatment to individualized and eventually personalized/precise treatment (Figure 4).

Figure 4

The way to precision nutrition. BMI, body mass index; CRP, C-reactive protein, eGFR, estimated glomerular filtration rate.

EFFECTS OF INFLAMMATION ON METABOLISM

As a key driver of malnutrition, inflammation has several effects on the metabolism, which might explain the less pronounced effects of nutrition support in highly inflamed patients observed in previous studies. Several mechanisms may explain these effects. A well-known effect of acute inflammation in medical inpatients is cytokine-induced anorexia. Inflammation also interferes with gastrointestinal motility, leading to gastroparesis and nausea.⁵¹ Further, neuroendocrine response, together with inflammatory response, leads to mobilization of energy storages, triggering the release of FAs by lipolysis, the release and degradation of glucose from glycolysis, glycogenolysis, and gluconeogenesis in the liver and release of amino acids from muscle proteolysis. These metabolic changes cause hyperglycemia, together with peripheral insulin resistance, hindering glucose entering the cells. All this culminates to uncontrolled catabolism, which leads to an increase in the release of ROS and therefore enhanced inflammation.²⁶

Nutrition support alone appears to be ineffective in reversing the impaired metabolism because insulin resistance and other previously mentioned mechanisms prevent adequate metabolism of the ingredients. Past trials in the ICU or with patients with advanced cancer support this hypothesis.^{33, 52}

In addition to elevated ROS levels, other mechanisms may exist. Macroautophagy, also called autophagy, is responsible for clearing potentially toxic protein aggregates and damaged organelles and is the only pathway able to degrade such bulk cytoplasmic material. It has a crucial role in maintaining integrity of cells, organs, or tissues and has a fundamental role in immunity. The cytoplasmic clean-up function of autophagy is by default anti-inflammatory in any type of cells and like this contributes to the inflammatory response to a stressor.⁵³ There is evidence from some studies in critically ill patients suggesting that [over]nutrition interferes with autophagy, thereby lowering detoxication of cells with negative effects on outcomes.⁵⁴ Vanhorebeek et al studied the activation of autophagy in hepatic and skeletal muscle biopsies within a trial focusing on hyperglycemia prevention in prolonged critically ill patients. Morphologically, both liver and skeletal muscle revealed an autophagy-deficiency phenotype. Proteins involved in the key steps of autophagy were induced 1.3- to 6.5-fold by critical illness (P = 0.01), but mature autophagic vacuole formation was impaired by 62% (P = 0.05) and proteins normally degraded by autophagy accumulated up to 97-fold (P = 0.03). Also, specific mitophagy markers were unaltered or down-regulated (P = 0.05). Despite obvious attempts of the ill body to activate the autophagic machinery, this pathway appeared incomplete and insufficient as indicated by a reduced number of mature autophagic vacuoles, accumulation of specific substrates, and damaged organelles normally eliminated by autophagy. Autophagy can be influenced—two of the most powerful suppressors of autophagy are nutrients and insulin through the activation of mTOR pathway.⁵⁵ Considering the previously mentioned impaired autophagy in critical illness and the assumed additional suppressive effect of nutrients on autophagy, the role of artificial (enteral and parenteral) and aggressive feeding during critical illness should be further investigated.⁵⁴ This finding supports the hypothesis that especially overfeeding might be more harmful in this patient population.

Therefore, there is a need to find biomarkers that indicate when the patient is ready to be fed or to receive more intensive nutrition therapy. In a secondary analysis of the EPaNIC trial, Van Dyck et al aimed to find a “ready-to-feed indicator” for critically ill adults. They investigated the predictive potential of the circulating growth differentiation factor 15 (GDF15), a cellular stress marker that abruptly increases during critical illness. Within the EPaNIC cohort GDF15 was elevated throughout ICU stay, but similarly in early parenteral nutrition and late parenteral nutrition patients, and remained high beyond ICU discharge. It was only weakly related to gastrointestinal tolerance and the potential as “ready-to-feed indicator” appears limited.⁵⁶ So far, no other biomarker that indicates the optimal timing for nutrition initiation could have been identified. Thus, the intention to find a suitable marker remains important for the future of personalized nutrition.

COMPLEXITY OF INVESTIGATING INFLAMMATION

Investigating the interplay between nutrition and inflammation is a highly complex task with many confounding factors, and there are many questions to answer. There are several processes within the cellular level and the core that occur beforehand and thus not fully accounted for.⁵⁷ Simultaneous analysis of complement factors, cytokines, chemokines, and other mediators would provide a more integrated picture of inflammation and the disease process⁵⁸ but will be difficult to transfer into clinical practice. In addition, a difficulty in studying the effect of inflammation on treatment response is that many of the used biomarkers are also biomarkers of disease severity. Therefore, there is overlap, and the individual effects cannot always be determined so far.

CONCLUSIONS

Inflammation is a key driver of DRM leading to anorexia and a compromised metabolism of ingested nutrients in medical inpatients. There is current evidence suggesting that stratifying patients according to their inflammatory state (eg, on the basis of CRP levels) might be useful for predicting the benefit of nutrition support. The approach on how to feed highly inflamed patients is unclear today. There are certain nutrients and dietary patterns that have anti-inflammatory properties, but there are no data showing that these nutrients would be of benefit for the patient with high inflammation today. Within this research area, the complexity of inflammatory processes at different levels (nucleus, cell, and systemic level) must be kept in mind. In the future, biomarkers might help to guide initiation of nutrition therapy and to select patients who would benefit from nutrition support.

AUTHOR CONTRIBUTIONS

Carla Wunderle, Franziska Stumpf, and Philipp Schuetz contributed to the conception and design of the manuscript; and Carla Wunderle drafted the manuscript. All authors critically revised the manuscript, agree to be fully accountable for ensuring the integrity and accuracy of the work, and read and approved the final manuscript.

CONFLICT OF INTEREST STATEMENT

Dr P. Schuetz is supported by grants of the Swiss National Science Foundation (SNSF Professorship, PP00P3_150531) and the Research Committee of the Kantonsspital Aarau (1410.000.058 and 1410.000.044). Dr P. Schuetz's institution has previously received unrestricted grant money from Nestlé Health Science and Abbott Nutrition. The remaining authors declare no conflict of interest within the last 36 months.