Inflammation and Nutritional Science for Programs/Policies and Interpretation of Research Evidence (INSPIRE)^1,2,3,4,5

a. NF-κB signaling and induction and maintenance of T-regulatory cells.

NF-κB function and mechanism of activation are described in Text Box 8.

b. Induction and maintenance of T-regulatory cells.

T-regulatory (Treg) cells play an integral role in self-regulatory processes through their involvement in both shutting down immune responses after successful elimination of pathogens and in preventing autoimmunity (130, 131). Members of the Treg cell family are numerous, but the best-characterized are those that express CD4, CD25, and forkhead box P3 (FoxP3). Again, specific nutrients have been implicated in various aspects of Treg activity.

c. Nutrients and the acquired immune system.

As outlined in Table 3 and documented in Supplemental Tables 2–8, studies in animal models have revealed the following:

• ● PEM and deficiencies of vitamin A, zinc, or iron have a robust, albeit variable effect in reducing antigen-presenting cell (APC) function (150–160).

• ● PEM and vitamin A, zinc, and iron deficiencies exert their greatest effect on reducing CD4⁺ T cell function (79, 161, 162). Proliferation is often reduced, and polarization to a particular antigen is reduced/altered (163–165).

• ● VAD causes reduced trafficking of T and B cells to tissues (158, 166, 167).

• ● CD8⁺ T cell function is the least studied of the major components of acquired immunity.

• ● VDD has a modest effect on cell-mediated and humoral immunity with the greatest effect seen on APC and T cell function (85, 168–173). In particular, NK T cell development in mice is critically dependent on vitamin D (173).

The data from human studies, presented in Table 4, reveal the following:

• ● Only PEM has evidence attesting to an effect in reducing human APC function (174).

• ● PEM and vitamin A and zinc deficiencies have their largest effect in reducing CD4⁺ T cell function (98). There are suggestions of reduced proliferation for all 3 and modest evidence of alteration in polarization for VAD and supplementation (106).

• ● VDD has a relatively minor effect on cell-mediated and humoral immunity in humans, with the greatest effect seen on APC and T cell function (175, 176).

• ● ID primarily affects CD4⁺ T cell function, with a lesser effect on B cell function and no conclusive evidence with regard to its impact on APCs (177–181).

• ● Megaloblastic anemia due to folic acid deficiency has been associated with depressed cell-mediated immunity (182).

Parenthetically, the implications of the effect of PEM, vitamin A, zinc, and iron on CD4⁺ T cell counts are important in the context of interpreting the impact of nutritional interventions in conditions such as HIV in which CD4⁺ T cell counts are used as a primary outcome. Because we appreciate that nutrient deficiency in itself affects CD4⁺ T cell count, we must design studies that can distinguish between a disease (e.g., HIV pathogenesis or treatment) effect and the impacts of malnutrition or its amelioration. For example, a study that evaluates the impact of a single or multiple micronutrient intervention on HIV might include an HIV-negative control to be able to distinguish between an HIV-specific effect and the impact of ameliorating nutritional insufficiency. This distinction has obvious biological and programmatic/public health implications.

Among the limitations of the extant evidence is that most of the research to date has focused on the effect of single-nutrient deficiencies on immune response. Few studies examined the simultaneous association of multiple nutrients, or their status, with immune function (183, 184). The design of such studies in humans is challenging for several reasons, including

• ● our relatively limited understanding of the nature of micronutrient interactions within biological systems,

• ● the complexities of the health context in low- to middle-income settings where multiple-micronutrient insufficiency may be common but often coexists with multiple other conditions contributing to both acute and chronic inflammation, and

• ● low prevalence of multiple-micronutrient deficiencies in otherwise healthy participants in developed countries; when they do exist in certain subgroups such as adolescents, certain racial groups, or the elderly, confounders such as age, disease, etc., further complicate experimental design and data interpretation.

d. Potential points of interaction between micronutrients involved in dependent pathways.

Several points of interaction among the various micronutrients seem likely and are outlined in the accompanying Text Box 9. For the purpose of this overview, the general molecular actions of 4 micronutrients (vitamins A and D, zinc, folic acid) will be the primary focus, including their specific points of convergence on the regulation of 2 major regulators of inflammation: NF-κB activity and the induction, and maintenance, of Treg cells.

a. Mechanisms of vitamin A effect on the immune system.

Most of the effects of vitamin A on immune or inflammatory responses can be explained via binding of the vitamin A metabolite, ATRA, to 1 of 3 zinc-finger proteins containing members of the nuclear receptor superfamily, RAR-α, RAR-β, and RAR-γ (185). These function as ligand-dependent transcriptional regulators by binding, usually as heterodimers to RXR-α, -β, and -γ and to RARE in target genes. Apo (unliganded) RAR can also occupy regulatory elements at their target genes and repress their expression. Ligand-dependent repression of gene expression by RAR can also occur. Recent estimates, in various human cell lines, indicate that RARs are constitutively bound to ∼500 genomic sites, and ATRA treatment induces RAR binding to 500–600 DNA sites (186).

The number of genes actually regulated in this manner is unknown. It has been >10 y since the last large-scale survey was done. At that time, the estimated number of genes directly regulated by retinoic acid was very large (187), whereas the number of direct targets was fewer. Although new data are needed to further elucidate the magnitude of these relations, it is clear that vitamin A has multiple genetically mediated effects on the immune response via the activities of ATRA.

b. Mechanisms of vitamin D effect on the immune system.

Although some effects of vitamin D are mediated independently of VDR, the large majority of the activity of vitamin D can be explained via binding of the vitamin D metabolite [1,25(OH)₂D] to a single, zinc-finger containing, ligand-activated transcription factor, VDR. The heterodimeric partner, RXR, is required for transcriptional activity. The VDR/RXR complex binds to specific sites within the promoter region of genes known as VDRE and activates transcription (188).

The presumed number of genes containing a VDRE and/or regulated by VDR/RXR is very large. Recent human data suggest that VDR can bind to >2700 sites in the genome and directly regulates >200 genes (189). Like vitamin A, vitamin D can affect immune function through this genetic interaction. Because vitamins A and D share a nuclear receptor (RXR), compete for coactivators and corepressors, and reciprocally regulate the expression of their receptors, a wide variety of combined effects on gene regulation likely occurs (190–192).

c. Mechanisms of the zinc effect on the immune system.

The molecular actions of zinc appear to be more complex than those of vitamins A and D. In humans, the zinc-containing proteome contains <3000 members (193). More than 10% of this proteome encodes enzymes, including alcohol and aldehyde dehydrogenases, involved in vitamin A metabolism (193). Zinc is also necessary to maintain the structural integrity of proteins such as zinc-finger containing transcription factors (194). Approximately 50% of all transcription factors, including the vast majority of transcription factors involved in leukocyte lineage commitment (T cell vs. B cell commitment, CD4 vs. CD8 lineage commitment, Treg commitment) and effector function, contain zinc (195). Furthermore, 477 members of the poly-zinc-finger (poly-ZF) family of putative transcriptional repressors in humans (330 in mice) involved in epigenetic gene silencing contain zinc (195).

The ability of free zinc to act as an intracellular signaling molecule, similar to calcium ion (Ca²⁺), was recently described (196). An appreciation of the molecular complexities of zinc is just emerging. It will be essential to determine what the hierarchy of functional pathway maintenance is during ZnD.

d. Mechanisms of folic acid effects on the immune system.

The molecular actions of folic acid are also extraordinarily complex. Folic acid, along with other nutrients involved in one-carbon metabolism (e.g., methylcobalamin, choline, betaine, and methionine), is required for the synthesis of DNA and S-adenosylmethionine (SAM)–dependent methylation of DNA and histones. The number of specific sites [specifically sites on DNA where cytosine and guanine are separated by a phosphate; referred to as cytosine-phosphate-guanine (CpG) sites] in the genome that are potential targets for methylation is staggering. The recent Encyclopedia of DNA Elements (ENCODE) project has identified 1.2 million putative CpG methylation sites in the human genome (8.6% of nonrepetitive genomic CpGs) (197). The methylation pattern is unique to each cell type. The number of genes regulated by histone methylation is similar to that regulated by DNA methylation, and the pattern of histone methylation is unique to each cell type (198). The roles of folate in methylation-dependent DNA modifications will be covered in greater detail in the following section on epigenetics.

a. Epigenetics.

Understanding the role that epigenetics plays in determining the plasticity and heterogeneity of human immune responses is essential for planning and interpreting data generated by nutrition intervention trials. Basic principles regarding epigenetics and immunity/inflammation are highlighted in the accompanying Text Box 10. In the following section, we will consider the role of nutritionally regulated epigenetic control of genes involved in immunity and/or inflammation.

Aside from the role of miRNA in immune development and function, and despite the surging interest in epigenetics in the past decade, surprising little is known about how immunity or inflammation is modulated via epigenetic mechanisms. The majority of epigenetic studies were conducted with the use of T cells and macrophages. Factors that control T cell development such as IL-2 are regulated via promoter demethylation in T cells (202). Naive CD4⁺ T cells differentiate into specific and stable T-helper (Th) cell phenotypes (including, but not limited to, Th1, Th2, Th17, and Tregs). Chromatin remodeling allows increased binding of specific factors to the regulatory regions of IFNG in Th1 cells or IL4 (203) in Th2 cells.

Few studies have addressed how nutrients play a regulatory role in altering immune function or inflammation via epigenetic pathways. A small but rapidly growing literature suggests that non-nutrient, bioactive food components may regulate epigenetic pathways. Claycombe et al. discuss these in a companion article to this supplement “Epigenetics of inflammation and immune dysfunction: the role of nutrition” (36). The following is a summary of the current knowledge about nutrient control of epigenetic events in general with a focus on immune or inflammatory responses.

b. Role of specific nutrients in genetics/epigenetics.

a. Nutritional/metabolic implications of the APR.

The APR and accompanying changes in APP concentrations can result in profound metabolic consequences with significant nutritional implications (Text Box 15).

Although evidence exists for extrahepatic synthesis of APPs, the majority are derived from hepatocytes in higher order vertebrates. Table 5 shows a survey of the most commonly encountered human APPs along with their putative functions. Recent evidence from proteomic studies in stress-challenged mice and humans suggests the existence of several hundred more proteins that might meet this definition (249, 250).

b. Roles and impact of APPs.

Approximately 50% of the major APPs are involved in nutrient transport or regulation of nutrient status. This challenges the ability to distinguish between changes in nutrient status consequent to exposure vs. the response of a nutrient-related APP to an APR. It might also compromise efforts to implement and monitor an intervention effectively in a patient experiencing an APR, because the metabolic responses to the APR may, by their nature (i.e., a physiologic override of normal metabolic/nutritional processes), be intransigent to usual approaches to mitigation (e.g., use of a dietary supplement to increase concentrations).

Conversely, it is also the case that certain aspects of the APR are blunted in severely malnourished children (251–253). Another important consideration in this context is the impact of chronic stress on the APR and consequent metabolic problems. The latter scenario is exemplified by what we have learned about the impact of HIV infection on the inflammatory response and subsequent changes in metabolism that might account for some of the chronic untoward complications of that disease and its treatment (254–256). These types of disease-mediated chronic relations will be discussed in further detail by WG 3.

In addition to the broad impact of the APR cascade, APPs also play specific roles within the human body and are outlined in Text Box 16.

In many ways, the response to stresses such as infections may be seen as paradoxical (i.e., having a negative impact on both host and pathogen). This view is particularly evidenced by the nutritional responses to infection (e.g., nutrient sequestration, anorexia) (257). To understand how this paradox is manifested and how it is intended to defend the host, it is useful to look at what is known about specific nutrients in this context.

Sequestration of nutrients to limit the growth and function of pathogens has been proposed as one reason for the redistribution of some nutrients during the APR. This is most evident for iron (259) but may also occur with zinc (260). Another intriguing possibility is that nutrient deprivation increases the level of stress on all host cells, causing those that are subsequently stressed by intracellular pathogens or trauma to die (257). Similarly, to limit host tissue damage, sequestration of pro-oxidant nutrients and a change in the antioxidant balance in favor of the host in response to reactive oxygen species (ROS) produced during the immune response have been proposed (18). Last, compartmentalization of nutrients during inflammation may create a permissive environment for inflammation to occur via temporary removal of nutrients (e.g., vitamin A, zinc) that negatively regulate NF-κB activity (118, 126).

c. Case study: vitamin A and the APR.

Rather than a full discussion of each of the putative APP mechanisms, a case study exploration of the various proposed scenarios using vitamin A as an example will be illuminating. Although vitamin A is absolutely required for mounting an effective immune response, it is tempting to speculate that temporarily inhibiting plasma delivery of vitamin A to tissues is beneficial to the host by creating a favorable environment for inflammation. The accompanying Text Box 17 outlines potential mechanisms that might describe how this limitation might occur and evidence for how this may manifest in the context of infection.

Several strategies have been designed to assess nutrient status in the face of the APR, including the use of a single APR to identify subjects with and without an active APR, reporting nutrient concentrations and prevalence rates of deficiency in those with and without elevated APPs, or using multiple APPs to adjust nutrient values for phase or severity of infection (271, 272). Each of these strategies may be useful, but they are limited, and identification of biomarkers of nutrient status that are not sensitive or responsive to inflammation remains an important avenue of research. This will be addressed extensively by WG 4. The application of genomic metabolomic and/or proteomic approaches to this problem may be useful (127, 273, 274).

a. Epidemiology of VAD.

The challenge of obtaining vitamin A, a fat-soluble vitamin, is great in many settings because preformed vitamin A is found primarily in animal-source foods including liver, eggs, and whole milk. Although it can also be obtained through precursors (β-carotene and other provitamin A carotenoids), which are found in orange-yellow vegetables and fruit and green leafy vegetables, the conversion of carotenoids to vitaminA is variable. Text Box 20 provides a summary of causes contributing to VAD.

On the basis of serum retinol concentrations <0.70 μmol/L (or 20 μg/dL), 122 countries are characterized as having moderate to severe VAD problems in preschool-aged children, whereas 88 countries fulfill this criterion with regard to pregnant women; it is estimated that a total of 190 million children and 19 million women are affected (5). The regions that are most affected by VAD are Africa and Southeast Asia (325). Preschool-aged children and pregnant women face the highest risk, reflecting high nutritional demands and high infectious disease burden (325). The principal clinical manifestation of VAD is xerophthalmia in various levels of severity, ranging from night blindness and Bitot’s spots to corneal scarring and blindness. Other consequences of VAD are anemia and a higher risk of child mortality from infectious diseases (325).

b. Effects of VAD/vitamin A supplementation on immune function and disease risk.

As discussed by WG 1, VAD and vitamin A supplementation affect a range of innate and adaptive immune functions. Overall, VAD has been associated with decreased barrier function and skewing toward a Th1 response. Several human studies documented that VAD is associated with increased Th1-driven delayed-type hypersensitivity (DTH) responses and reduced Th2-driven antibody responses to vaccines; vitamin A supplementation reverses these responses (104, 105, 326). An animal study indicated that a high vitamin A intake may be linked to asthma (164), an association that remains to be studied in humans. Vitamin A supplementation has mostly been reported to have reciprocal effects of VAD (Table 6).

The important effects of vitamin A on immune function are reflected in the many community and clinical studies that showed that vitamin A affects mortality and morbidity. Large randomized trials led to meta-analyses concluding that prophylactic vitamin A supplementation of children between 6 mo and 5 y of age reduces overall mortality by 23–30% (290, 327–328) (Table 7). Trials in infants <6 mo of age produced conflicting results. No study found a beneficial effect between 1 and 5 mo of age, and the effect in neonates is still under debate; whereas some studies suggested a beneficial effect in South Asia and in boys, the results have been heterogeneous (332). One possible explanation for the apparent lack of benefit of vitamin A supplementation below 6 mo of age is that infants in such studies are protected from infection by maternally derived immune factors (e.g., serum IgG, factors in breast milk) to a greater degree than are older infants. Thus, the benefits of correcting VAD on survival are less pronounced.

Disease-specific studies have likewise not shown uniformly beneficial effects. Although vitamin A may have a preventive effect on the incidence or severity of diarrhea and measles in community trials, it has had little or the opposite effect on respiratory infections (105). The same is seen in studies of vitamin A as a therapeutic agent (Table 7).

More in-depth mechanistic studies have provided clues to these conflicting results. For instance, neonatal vitamin A supplementation had a different effect on in vitro whole-blood cytokine responses depending on the sex and the vaccination status of the children (301). Vitamin A supplementation of 6- to 15-mo-old children differentially modified the association between gut chemokine and cytokine responses and duration of infections, depending on the infectious pathogen (351). Also, although it is clear that inflammation is associated with decreased serum retinol and RBP concentrations, the role of vitamin A supplementation on inflammation and its markers is less clear. Vitamin A supplementation of children <5 y of age increased their APP, CRP, and AGP values in response to vomiting and diarrhea but not in response to cough and fever (297) (Table 8).

Clearly, VAD is associated with critical impairment of immune function. However, the effects of vitamin A supplementation on the immune system may lead to variable clinical outcomes, depending on the nature of the immune response needed to combat the infecting pathogens. For this reason, vitamin A supplementation may be a double-edged sword. For some—and on the basis of the clinical trials, presumably most—infectious diseases, vitamin A supplementation may be beneficial; in other situations it may be harmful (369, 370). Important effect modifiers may be population vitamin A status, age, sex, vaccination status, season, pathogen prevalence, disease type, and disease burden. The latter is also reflected in differences in effect over region and season (332, 370).

a. Epidemiology of ZnD.

ZnD is a global problem that contributes significantly to disease morbidity and mortality, particularly in low-income countries (372, 378). Although severe ZnD is rare, mild to moderate deficiency is common throughout the world (372) and is a predisposing factor for a range of infections, particularly diarrhea (379). ZnD is estimated to account for 450,000 deaths annually, >4% of deaths among young children, and 1% of all deaths in children globally (378).

The estimation of zinc availability in the diet in combination with childhood stunting rates to generate a likelihood of deficiency remains the accepted method for assessing population-based ZnD. The estimation of zinc availability in the diet in combination with childhood stunting rates to generate a likelihood of deficiency has been used by the International Zinc Consultative Group (380) and remains the accepted method for assessing population-based ZnD. On the basis of estimates of dietary zinc availability and the prevalence of childhood stunting, it is estimated that ∼15–30% or 1–2 billion of the world’s population is zinc deficient (378, 381). The prevalence of ZnD is greatest in South and Southeast Asia, sub-Saharan Africa, and Central America (372, 382). Mild to moderate ZnD is thought to be common in low-income populations, especially in populations with a low consumption of zinc-rich animal-source foods and high intakes of foods rich in phytates, which inhibit zinc absorption (383). Children <5 y of age, pregnant women, and the elderly comprise the groups at high risk of deficiency. Within those groups, premature and small-for-gestational-age infants and preschool-aged children, predominantly those 6–23 mo of age are at greatest risk (372, 379). Infants are particularly vulnerable because the zinc content of human milk declines sharply in the postpartum period regardless of maternal zinc status (372).

b. Effects of ZnD or supplementation on immune function and disease risk.

As discussed by WG 1, zinc is known to play a critical role in the integrity, function, and maintenance of host defense systems (384, 352, 353) (summarized in Text Box 22).

Furthermore, interactions of ZnD and environmental enteropathy (EE) that plague children in developing areas may well be synergistically interactive, each worsening the other and their potential impact on impaired child development (386). Human and animal studies showed that mild ZnD can impair multiple mediators of host immunity, ranging from physical barriers of the skin to innate and acquired immunity, increasing the risk of severe infection (387, 354). Globally, these effects translate to 10% of diarrheal diseases, 16% of acute lower respiratory infections, and 18% of malaria episodes being attributed to ZnD (378). Of the 450,000 deaths attributed to ZnD, >50% result from diarrhea (378). In addition, poor zinc status has been implicated in ∼7% and 10% of pneumonia and malaria deaths, respectively (378).

Neither ZnD nor zinc supplementation has an appreciable effect on responses to bacille Calmette-Guérin (301), diphtheria and tetanus toxoid (388), rabies (389), influenza (390), Haemophilus influenza type b (321), and pneumococcal vaccines (389, 391). Zinc supplementation, however, may improve antibody responses to cholera vaccination in children (392). It should be noted that the influence of zinc status on vaccination has not been well studied. Most of the evidence is drawn from secondary analyses or small observational studies. Few account for the effects of age, sex, degree of deficiency, supplement dose, and duration of dosing.

Results of preventive trials in children living in areas of endemic ZnD showed that zinc supplementation significantly reduced the incidence of pneumonia and persistent diarrhea (379, 337, 338) (Table 7). Zinc prophylaxis in children is associated with a reduction in diarrhea mortality of 13% and pneumonia mortality of 15%; however, despite the implied association of zinc status with malaria mortality (199), zinc prophylaxis does not appear to affect deaths from malaria (339), raising questions about both the value of current methods of zinc assessment and the specific role of zinc in malaria. Significant reductions in acute and persistent diarrhea morbidity were also observed in zinc-supplemented children in therapeutic trials (340). On the basis of these results, the WHO recommended in 2004 that children be given oral zinc for 10–14 d with oral rehydration salts for the treatment of diarrhea (393).

Evidence of whether or not zinc supplementation provides a similar therapeutic benefit to children with severe pneumonia is conflicting (343–345). However, results from a recent trial in India indicate that zinc adjunctive therapy can reduce treatment failures in infants aged 7–120 d with severe infections of suspected bacterial etiology (342). On the other hand, some data suggest that supplementation during the clinical phase of gram-negative infections may exaggerate APRs and prolong hospitalization (347, 348, 394). The clinical and immunologic effects of therapeutic zinc supplementation on the outcome of infection are likely to depend on age, zinc dosage and frequency, degree of deficiency, region, and infectious agent. Further mechanistic studies are needed to optimize zinc therapy and to define populations most likely to benefit.

a. Epidemiology of ID.

Nutritional ID and ID-anemia are highly prevalent, with the WHO estimating that 16.6% of the world’s population was affected in 2000, with prevalence rates being highest (>20%) in sub-Saharan Africa, India, other countries in Asia, the Pacific Islands, and the Middle East (398). ID typically results from a combination of dietary inadequacy and high physiologic need resulting from growth in infants and children and from the demands of the fetus during pregnancy. Menstrual blood loss is another condition that increases iron requirements and hence the risk of deficiency (397). Thus, infancy and early childhood, pregnancy, and childbearing years are the principal life stages in which ID is a risk (399, 400).

b. Effects of ID on immune function and disease risk.

As outlined by WG 1 and summarized in Text Box 24, ID affects key aspects of innate immunity.

A recent review found that ID in humans does not affect antibody responses to vaccination (including responses to T cell–dependent antigens), although the data are limited and impaired responses were noted in animal studies (103). The contribution of ID to morbidity and mortality from infectious diseases worldwide is less than that attributed to VAD and ZnD (406).

A critical element of the interaction of iron with infection is the restriction of iron availability to invading pathogens (which, like host cells, require iron for growth). This is a major part of the innate defense of mammals against microorganisms that have specific mechanisms for acquiring iron in low-iron environments, such as host tissues (259, 407). In brief, the acute phase peptide hepcidin is synthesized in the liver in response to infection or inflammation (IL-6 plays a specific role in inducing its transcription) and it blocks the normal flow of iron from the gut, senescent erythrocytes, or tissue iron stores via transferrin to the bone marrow or other tissues in need of iron. This activity of hepcidin markedly decreases serum iron concentrations, which restricts iron availability to extracellular pathogens and may also affect iron availability to intracellular pathogens (e.g., malaria parasites) that depend on transferrin-mediated iron acquisition. This host response prompts the inference that iron supplementation during infection may be detrimental, a topic that is covered in greater detail by WG 3. Although results are equivocal for most infections, it is clear that iron supplementation in areas where malaria is endemic can have adverse effects, perhaps by enhancing survival of the parasite but possibly via other mechanisms as well, including promotion of iron-catalyzed oxidative damage or facilitation of concurrent bacterial infections by opportunistic, invasive pathogens such as Salmonella (28, 259, 349, 350).

a. Epidemiology of underweight, stunting, and wasting.

The prevalence of under-5 underweight, stunting, and wasting due to undernutrition, although diminished due to aggressive public health interventions, remains high (1). The most vulnerable remain in low-income countries. Regional differences are seen, with underweight being most prevalent in Asia and eastern Africa, stunting in eastern and middle Africa as well as in South Central Asia, and wasting in South Central Asia. Severe wasting (<3 SDs) is often used as a trigger for nutritional rehabilitation programs and is most common in South Central Asia and middle Africa. The causes of PEM are diverse, but in these affected regions early childhood infections are typically associated with poor water and sanitation, poverty, food insecurity, and inadequate health care (408). In particular, a low availability of micronutrient-rich, animal-source foods is associated with the development of malnutrition (406). Repeated episodes of infection leading to impaired absorptive function also likely contribute to the development of undernutrition and may compound the problem of inadequate food intake (323).

PEM can also develop in adults and is typically defined as a BMI (in kg/m²) <18.5. Older adults are at particular risk of malnutrition, with the prevalence of low BMI increasing 4-fold above 75 y of age compared with those aged 45–64 y in the United States (409). Older adults also tend to have lower lean body mass and higher adiposity than younger adults due to loss of muscle mass with aging (410). The development of malnutrition in the elderly is complex and may include poverty, decreased ability to prepare food, depression, or the development of illness, but a common feature is a decrease in appetite or ”physiologic anorexia” (322).

b. Effects of PEM on immune function and disease risk.

The relation between PEM and host resistance to infections is bidirectional. This relation is magnified at either end of the life course, but particularly in the first 2 y of life in children living in areas with heavy pathogen exposures. Infections can precipitate poor nutrition via several mechanisms (406, 411), and antibiotic use during refeeding from malnutrition can enhance recovery, suggesting that an untreated (and undiagnosed) bacterial infection was inhibiting recovery (412). In addition, the frequency, severity, and risk of mortality from diarrhea, pneumonia, malaria, and other infections are highest in malnourished children due, at least in part, to impaired host defenses. Moreover, PEM indirectly affects the risk of infection and perhaps even NCDs via its role as the core of several “vicious cycles of poverty” (322, 397).

The effects of PEM occur through both its effects on immune and inflammatory function as well as on host protective barriers and development (2). The putative mechanisms by which PEM affects immune function and inflammation were covered by WG 1.

Acquired or cell-mediated immunity (CMI) is impaired by anthropometrically defined undernutrition. Infants and children with underweight, wasting, and stunting are typically also at risk of multiple-micronutrient malnutrition deficiencies; thus, it can be difficult to specifically associate particular immune deficiencies with these conditions. CMI is also specifically impaired in elderly adults with PEM. Serum antibody responses to vaccination also seem to be impaired in the elderly (256) unlike in younger malnourished subjects. Some additional, relevant aspects of the impact and risks associated with PEM are summarized in Text Box 25.

Although acquired immunity, including CMI, is most consistently affected by PEM, some aspects of innate immunity may also be affected in underweight, wasted, and stunted individuals, with effects being more pronounced in more severely malnourished patients, (e.g., those 3 SD below the expected mean, or with a clinical presentation of Kwashiorkor or Marasmus). Relevant aspects of the impact of PEM on the innate immune systems are highlighted in Text Box 25.

a. Epidemiology of obesity.

According to the WHO (418) and the recent GBD analysis (1), the global prevalence of overweight/obesity (BMI >30) is increasing at alarming rates, even in LMICs. Of particular concern is the dramatic increase in overweight/obesity in children <5 y of age.

Biochemical markers of obesity include high serum leptin, low adiponectin, and elevated inflammatory makers including CRP, TNF, and IL-6 (362–365). Insulin resistance and type 2 diabetes are often comorbidities with obesity. Coincident with the increase in obesity rates has been a concurrent increase in those comorbid NCDs, including hypertension, diabetes, CVD risk, and cancer (1). It has been suggested that the development of insulin resistance and diabetes may be due to the state of chronic low-grade inflammation found in obesity (366, 367).

b. Effects of obesity on immune function and disease risk.

Obesity and its concomitants (e.g., insulin resistance) have been associated with impaired immune/inflammatory responses (419, 420). A summary of the reputed effects of obesity on immune function/inflammation and disease risk are found in Text Box 26.

With respect to vaccinations, there are several studies reporting impaired responses in obese adults and children including the following:

• ● An impaired response was found to hepatitis B vaccination (444, 445).

• ● A decreased anti-tetanus antibody response after tetanus vaccination was observed in obese compared with healthy-weight children (446).

• ● Trivalent influenza vaccination was associated with a lower antibody titer 1 y after vaccination, and impaired CD8 responses were observed against vaccine strains in obese individuals (447).

In support of the human studies, animal models of diet-induced obesity clearly demonstrated immune impairment and increased inflammation. Closely mimicking human obesity, mice with diet-induced obesity developed insulin resistance, elevated leptin, and elevated lipid TGs leading to fatty liver and increased inflammatory markers (448–450). Compared with lean mice, diet-induced obese mice had greater morbidity and mortality from primary and secondary influenza infections (451, 452), impaired resistance to Leishmania major infection (453), and increased mortality from S. aureus–induced sepsis (454).

a. Evidence that chronic stress can be harmful to the host.

The link between immune function/inflammation and the acute stress response has clinical implications that have just begun to be explored. Although the acute stress response effects are temporary, such that catabolism and immunosuppression are initially beneficial, over a prolonged time period these effects can be damaging. As a result, the host could be more vulnerable to infections, neoplastic disease, and metabolic derangements. General causes and implications of chronic inflammation were reviewed by WG 4.

The impact of this type of chronic stress can result in an increase in glucocorticoids, thereby antagonizing growth hormone actions on fat tissue. Chronic episodes of stress may lead to an attenuation of growth hormone actions to induce lipolysis, which could contribute to increased visceral adiposity. In addition, enhanced glucocorticoid secretion can lead to insulin resistance and profound hyperinsulinemia (464).

Experimental basic science showed that in monosodium l-glutamate (MSG)–induced hypothalamus-damaged and hyperadipose rats, exposure to LPS resulted in HPA hyperactivity likely due to leptin resistance, hyperinsulinemia (which may precede insulin resistance), and hypertriglyceridemia (465). In addition, catecholamines can induce production of the proinflammatory cytokine IL-6 locally at the myocardium (466); IL-6 stimulates hepatic production of CRP, which can contribute to progression of atherosclerosis. Persistent elevations in catecholamines can alter hemodynamics and lead to endothelial injury. The injured endothelium releases chemokines that attract monocytes, lymphocytes, and glucocorticoids, which could facilitate the early stages of atherosclerosis through induction of vascular cellular adhesion molecules (467).

The cascade of events associated with HIV infection exemplifies the potential detrimental effects of a chronic acute stress response. HIV-infected monocytes and macrophages are stimulated to produce IL-1, TNF, and IFN-α, which can increase CRH release and thus potentially affect immunomodulation of cytokine production (254). Taken together, the chronic activation of glucocorticoids and catecholamines can cause changes in the inflammatory cytokine milieu to selectively alter the pattern of cytokine secretion from Th1 to Th2. This may be linked in part to the increased predominance of visceral adiposity (255), insulin resistance (256), and atherosclerosis (468) in HIV-infected patients.

a. Exposure to specific and multiple infections through the life course.

It is not uncommon for individuals in low-income countries to have multiple, coexisting infections and NCDs. The scenario is best exemplified by the comorbidities associated with HIV infection (472). Although HIV is most commonly used as the example of comorbidity, the recent analysis of the GBD clearly indicates that there are various colliding epidemics of infectious diseases and NCDs occurring particularly in low-/middle-income settings (1). Other examples of comorbid conditions include latent infections (e.g., mycobacteria, cytomegalovirus), chronic helminth infections (e.g., geohelminths, schistosomes, filariae) maintained by regular exposure, chronic viral infections (HIV; hepatitis A, B, and C), with frequent, superimposed acute respiratory, gastrointestinal, and urogenital tract infections. In childhood, diarrhea may occur 6–8 times/y.

The inflammatory drivers are multiple, especially because infection exposure occurs from birth onward. These patterns of comorbidity may be complex. Helminth infections are often established in early childhood, and the prevalence and intensity peaks in childhood, with persistent infection exposure through adulthood. Along with effects such as EE discussed below, these infections give rise to APRs during the tissue-phase of the primary infection or during reinfection, whereas stable, low-intensity infections may not be associated with elevated levels of acute phase reactants.

The chronicity of effect depends on the nature of the infection, including its transmission patterns. Some examples include the following:

• ● In holoendemic malaria areas with year-round transmission, a large proportion of the population, including pregnant women, may experience asymptomatic parasitemia, concurrent elevation of serum APRs, and anemia of chronic inflammation.

• ● Overt tuberculosis, which affects 11 million people/y, causes a massive APR; but even latent tuberculosis, which affects one-third of the world’s population, has been associated with a slightly increased APR (H Friis, University of Copenhagen, personal communication, 2012).

• ● HIV continues to be a significant problem, with >34 million people now living with HIV/AIDS; 3.4 million of them are <15 y old (473). Current knowledge with regard to their nutritional needs was recently reviewed (474). Comorbid conditions (both infectious and NCDs) continue to be a common concomitant of HIV infection and may impair ability to produce an APR, thereby further increasing the risk of common and opportunistic infections that drive an APR.

The use and interpretation of nutritional biomarkers must take into account these age-specific and potentially multiple effects of infection. This also highlights the importance of clear clinical definitions of subject categories in studies of nutritional biomarkers.

b. Chronic malaria and parasitic infections.

In 2010, there were 219 million cases of malaria, with ∼80% of cases limited to 17 countries, most prominently in Africa with an estimated 660,000 deaths (with an uncertainty range of 490,000–836,000), mostly among African children (475). Not only does this represent a significant global health concern but its unique characteristics offer specific insights into the nature and implications of chronic exposure.

Many children, adults, and pregnant women living under conditions of perennial malaria transmission remain susceptible to malaria parasitemia. This is often asymptomatic due to the partial acquisition of malarial immunity. Chronic parasitemias lasting several months are common in such individuals unless treated with antimalarial drugs. Age differences in patterns of malaria immunity relate to this chronicity of exposure. Hepatosplenomegaly occurs with persistent infections and a chronic inflammatory state develops, often with hypergammaglobulinemia, which is attributable to reticuloendothelial and lymphoid hyperplasia. In a minority of individuals this may result in hyperreactive malaria splenomegaly with enormous splenic enlargement (476). Chronic exposures to other parasitic infections (e.g., schistosomiasis) may also cause hepatosplenomegaly, and these coexposures may further exacerbate chronic inflammation. Pathologic hepatic disease in childhood can lead to anemia, RBC production failure, and growth impairment. The association between hepatosplenomegaly, secondary to these chronic exposures, and cytokine responses has been little studied (477). Proinflammatory mechanisms are important, but there is little research relating these to nutritional correlates and biomarkers and their potential interactions (478). The following sections describe what is known about the impact of infection, NCDs, and certain physiologic states on nutrition with specific examples provided within each category.

c. Intestinal barrier integrity.

Any discussion of the interaction between infection, nutrition, and health must include an appreciation of the expanding understanding of the critical roles played by the gastrointestinal environment. The importance of gastrointestinal integrity is demonstrated by numerous reports documenting a relation between increased intestinal permeability and poor growth in low-income countries (479–481). In addition, the gastrointestinal environment, including the critical influence of the gut microbiome, plays an integral role as a barrier protecting the host from invasion by pathogenic organisms found in our environment. Key features of and approaches for assessing gastrointestinal function/integrity are highlighted in Text Box 32.

Although much has been learned about the interactions between the gut microbiome, immune system, and nutrition (283), the association of measurements of inflammatory markers of gut wall integrity with nutritional profiles has not been studied extensively. This is an important research gap because enteric infections are common causes of child morbidity and mortality, particularly in low-resource settings; thus, identification of relevant biomarkers for assessing risk is a priority.

d. Environmental enteropathy.

EE, also called tropical enteropathy, is a chronic subclinical condition caused by persistent fecal-oral contamination that results in colonization of the small intestine, destruction of intestinal villi, and inflammation (490, 491). The relations between EE, malnutrition, and development are complex and include impaired cognition and behavioral development in infants and young children consequent to the loss of essential nutrients integral to brain development (492, 493). EE has been implicated as a cause of stunting or chronic malnutrition; as much as 40% of stunting among under-5 children is attributed to EE in sub-Saharan Africa (494). Not only are growth and development adversely affected by EE, it also causes impaired immunologic responses to oral vaccines (495).

The epidemiology of EE is not well described, because diagnostic criteria are lacking, as are reliable point-of-care biomarkers. The intestinal permeability test using lactulose and mannitol has been used extensively to assess the integrity of intestinal barrier function, but it is limited due to its sensitivity and specificity. Candidate biomarkers currently being tested include neutrophil myeloperoxidase, fecal neopterin, citrulline, Regeneration gene 1A, Regeneration gene 1B, antibody to bacterial cell wall LPS core oligosaccharide (EndoCab antibody), and zonulin (see above section on intestinal barrier integrity) (491).

Immunologic manifestations of EE are outlined in Text Box 33.

People living in communities in poor conditions of hygiene and sanitation are not equally affected by EE, which implies the need to discover underlying factors, such as genetic predisposition, amount of exposure to pathogen, role of pre-existing malnutrition, as well as the role of immune system (490). Characterization of EE in at-risk individuals is clearly important for the interpretation of nutritional biomarkers.

a. The double burden of malnutrition.

Developed nations in the 1980s saw an upsurge in overweight and obesity among adolescents. Now, LMICs are facing the same problem of an increasing prevalence of overweight/obesity. In addition, these same settings are facing the added trend of overweight and obesity coexisting with undernutrition (associated with deficits of calories, specific single or multiple nutrients) in the same individuals, households, or communities—a condition called the double burden of malnutrition (6). This double burden is occurring in many settings, but recent data from Bangladesh offer a graphic example of the problem. In Bangladesh, where childhood stunting prevalence is one of the highest in the world (41% with height-for-age z scores ≤2, and 24% of women chronically malnourished with a BMI <18.5), overweight and obesity affect 17% women of childbearing age (507).

Among the challenges presented by the double burden are how best to develop and roll out programs to address either under- or overnutrition in settings in which both coexist. Clinically, the concern becomes how best to assess not only nutritional status broadly but how best to address potential interactions of over- and undernutrition with inflammation (i.e., with the APR). As will be outlined below, obesity (and its comorbidities) has an array of effects on the inflammatory response and vice versa. Add to that the impact of anomalies in the status of single or multiple micronutrients, and one can envision a complex scenario that can affect clinical accuracy and the development of effective interventions. Clearly, this complexity outlines an urgent research agenda.

The role and impact of selected nutrients on immune function and the inflammatory response were covered by WGs 1 and 2 and are covered in additional detail in Supplemental Table 9. The following is a brief discussion of the relations between overweight/obesity individuals and inflammation.

b. Overweight and obesity are chronic inflammatory states.

Overweight and obesity are inflammatory in nature because adipose tissue releases a number of inflammatory mediators (508). The main source of the mediators is believed to be macrophages that infiltrate adipose tissue, although adipocytes also contribute to the inflammatory process. Chronic low-grade inflammation is also a risk factor for type 2 diabetes and the metabolic syndrome (a complex of symptoms that include a combination of obesity, hypertension, impaired fasting glucose, or dyslipidemia). In obese subjects, compared with those with normal weight, there is an upregulation of systemic indicators of inflammation such as total leukocyte count, serum APRs, proinflammatory cytokines and chemokines, soluble adhesion molecules, and prothrombotic mediators. Although there may be an overlap between obese and nonobese states, obese individuals may have a severalfold increase in serum concentrations of TNF, IL-6, MCP-1, IFN-γ–induced protein 10, IL-18, macrophage migration inhibitory factor (MIF), and regulated on activation, normal T cell expressed and secreted (RANTES).

These chronic inflammatory mediators occurring in overweight and obesity are considered to be involved in insulin resistance and the metabolic syndrome. Adipose tissue is in fact a large endocrine organ secreting the proinflammatory hormone leptin and the hormones visfatin and resistin, which cause insulin resistance (509–511). In addition, it produces APRs, cytokines, chemokines, growth factors, components of the alternative complement system, and RBP. Ectopic deposition of adipose tissue may also occur in the absence of obesity and causes inflammation. This is particularly important in causing cardiac pathologies if the deposit is in the heart. This perhaps underlies the “thin-fat” hypothesis of increased atherosclerotic heart disease among Indians who have increased visceral fat deposition that may not be associated with obesity (512). The concentration of inflammatory mediators in obesity is reduced with increased physical activity (513, 514), although whether the reduction is due to physical activity itself or to reduced body weight needs to be clarified.

c. Inflammation and nutrition in obstructive lung diseases.

Persistent inflammation seen in many chronic diseases is associated with increased morbidity and mortality (515–518). Moreover, therapies aimed at reducing inflammation were found to lower overall mortality among healthy people with elevated CRP (519), reduce progression of atherosclerosis in patients with coronary artery disease (520), and improve exercise capacity (521) in patients with COPD. The topic of how inflammation affects nutrition in chronic NCDs or vice versa, however, is less well studied. In this section, obstructive lung disease (OLD) is discussed as an example of the bidirectional relation that may exist between inflammation and nutrition in chronic diseases, in order to identify potential gaps in our understanding (characteristics of OLD are shown in Text Box 34). In Figure 1, a schematic is provided of the bidirectionality of this relation identifying potential nutritional contributors and outcomes. A range of nutritional disturbances is considered, including obesity, weight loss, muscle wasting, adipose tissue accumulation, and specifically, micronutrients and vitamins A and D (Figure 1).

In chronic conditions such as asthma, inflammation and nutritional disturbances are often related to epiphenomena such as environmental exposures to tobacco smoke, air pollution, or airborne allergens. Evidence that inflammation leads to worse nutrition is reported in studies linking increased TNF production (527–530) and higher resting energy expenditure (531) with being underweight in patients with COPD. Moreover, the systemic inflammation may also be directly responsible for muscle wasting and cachexia seen in some patients with COPD (532–535). It was also reported that high-sensitivity CRP was a marker for impaired energy metabolism, lower functional capacity, and a lower St. George’s Respiratory Questionnaire score in patients with COPD (536). Previous studies showed that a low fat-free mass assessed by bioelectrical impedance analysis is an independent predictor of mortality irrespective of fat mass (534).

More recently, a placebo-controlled randomized trial in patients with COPD found that nutritional supplementation, in combination with a short course of anabolic steroids vs. nutritional supplementation alone, enhanced gain in fat-free mass and respiratory muscle function in depleted patients with COPD over an 8-wk period (537); however, a post hoc analysis did not identify a survival benefit from the intervention (538). Hence, more evidence is needed to determine if the prevention of weight loss in patients with COPD is a modifiable endpoint with nutritional interventions, and if such interventions lead to a measureable survival benefit.

a. Oxidative stress in pregnancy and in prematurity.

Tissue lipid peroxidation is increased in healthy pregnancy, but gestational increases in lipid peroxidation by-products are aggravated in overweight, obese, and diabetic pregnant women, and oxidative stress is an important consequence of pregnancy obesity that can directly affect micronutrient status. Oxidative stress associated with increased ROS and/or reactive nitrogen species has also been linked to poor reproductive outcomes (e.g., idiopathic infertility, poor semen quality, and low-quality oocytes) (619) as well as poor implantation and increased endometriosis via placental ischemia/hypoxia-mediated defects (620). Thus, it has been suggested that early-onset pre-eclampsia, spontaneous miscarriage, and intrauterine growth restriction may represent a spectrum of placental oxidative stress–related disorders (620). The role of oxidative stress biomarkers (F2-isoprostanes, oxidized low density lipoprotein, malondialdehyde, hyperhomocysteinemia, NO, peroxynitrites, 8-oxodeoxyguanosine, reduced glutathione/oxidized glutathione, 4-hydroxy-2-nonenal, tyrosine nitration) and their nutritional modification in relation to pregnancy pathogenesis requires further investigation (619).

There is good evidence that oxidative stress is associated with preterm birth complications due to an underdeveloped and highly stressed antioxidant system (621, 622). Because the activity of key antioxidant enzymes only matures in the last weeks of gestation, preterm infants have a developmentally immature antioxidant system showing generally low activity of superoxide dismutase, catalase, and glutathione peroxidase enzyme activities, thus predisposing to oxidative stress. Moreover, preterm infants demonstrate a relatively high activity of the xanthine/xanthine oxidase system that can aggravate oxidative stress via production of superoxide anion. Membrane α-tocopherol concentrations and plasma and tissue carotenoid concentrations are lower in premature infants than in full-term infants due to lack of placental transfer of lipophilic antioxidants during the late gestation (623). There are also indications of problems with the intestinal absorption of fat-soluble antioxidant micronutrients, such as tocopherol, in premature infants, which might relate to the immaturity of the intestine and limited absorption of these vitamins. Because of the current conflicting findings, further research is needed to establish whether preterm or term human milk, both of which contain a range of hydrophilic and lipophilic antioxidants, provides better protection against oxidative stress than formulas fortified with a few antioxidants.

Preterm oxidative stress and metabolic programming of the fetus may propel the fetus toward obesity and insulin resistance.Maternal obesity also contributes to an environment in which the embryo and fetus may be exposed to oxidative stress, because the expression of genes related to oxidative stress are elevated in the placenta of obese women. Lipid peroxides, oxysterols, and oxidized lipoproteins can act as ligands for nuclear receptors such as liver X receptor, which is involved in important metabolic pathways, and lead to metabolic programming of the offspring toward insulin resistance and future obesity (624, 625). Pregnancy obesity is also reported to impair fetal iron status, which may be linked to hepcidin homeostasis (626).

b. Acute and chronic infection in pregnancy.

It has been well established that pregnant women are more susceptible to some infections, and for this reason they represent a special population group (627). Systemic immune responses in pregnancy are modulated rather than suppressed (628) and are influenced by placental cytokine networks (Figure 2); these changes include both proinflammatory and anti-inflammatory profiles (613). In addition the adaptive immune response is downregulated whereas innate immunity is enhanced. As a result, pregnant women are more susceptible to parasitic infections and to severe bacterial and viral infections. This leads to distinct inflammatory pathways with maternal, placental, and fetal consequences (630). Specific pathways are described in the following sections, indicating biomarkers that are induced or altered by inflammation, and indicating relations with nutritional status and biomarker profiling.

Important outcomes of pregnancy infection are preterm birth and low birth weight, both of which have important nutritional characteristics that effect biomarker profiles. These adverse pregnancy outcomes may result from acute infection late in pregnancy or from chronic infection with persistent low-grade inflammation from early gestation. Bacterial vaginosis and chorioamnionitis may be nutritionally modulated, and maternal iron status could be important, with the mucosal immune biomarker lactoferrin in influencing infection profiles (629).

Other chronic pregnancy infections, such as soil-transmitted helminths, contribute to maternal nutritional anemias, which are associated with similar adverse pregnancy outcomes (631–633). Low-grade bacteremia and/or cytokines generated within oral tissues have also been associated with increased risk of preterm birth and neonatal morbidity (634). In the context of low-grade inflammation, regardless of cause, extent, and location, there is evidence for an impact on gastrointestinal morphology and function, which can have nutritional implications (635). Thus, the effects of both acute and chronic infections on nutritional status raise the possibility that nutritional interventions, in addition to infection control, could modulate pregnancy outcomes, especially in developing countries where an underlying prevalence of undernutrition is common and where exposure to pregnancy infections such as malaria is often high (632). Nevertheless, few studies have examined how pregnancy inflammation/infection itself directly modulates nutritional status or biomarker profiles of the mother or the infant. Even the utility of CRP as an indicator of inflammation/infection during pregnancy has been questioned (636). Studies usually describe how infection and inflammation are associated with nutritional deficiencies [e.g., iron, vitamin B-12, folate, and vitamin D for bacterial vaginosis (637–639) and anemia or obesity for stillbirths (640)].

c. Fetal and infant infection.

The immunologic responses by the maternal, placental, and/or fetal immune systems can affect fetal development (641) and perinatal survival. The question of whether these adverse effects are due to a direct effect of the pathogen (viral, bacterial) or are secondary through maternal responses to inflammation/cytokines requires further investigation. Some investigators consider that impaired offspring TLR function is associated with increased susceptibility to infections and to preterm labor (613).

Several infections are known to affect fetal growth. An important example in humans is Plasmodium falciparum malaria, which, after placental parasite sequestration, induces specific maternal endocrine, immunologic, and hematologic responses that are associated with fetal growth restriction and preterm birth (642). Infants born to these mothers exposed to malaria in pregnancy have increased infant anemia and mortality risk (643). Another example encompasses the range of gastrointestinal nematodes, which are mostly chronic maternal infections, and which have been associated with stunting in the offspring (644). At least in an animal model, these infections in pregnancy may lead to offspring stunting by causing inflammation, and this has been associated with increased cortisol concentrations and altered cytokine profiles (645).

Infants of mothers with pregnancy inflammation and/or infection have altered systemic immune exposures. A fetal inflammatory response syndrome is described in which infants born after placental infection have elevated inflammatory cytokine profiles (628, 646), despite absence of cultivable microorganisms. Constituents in breast milk help protect the infant from enteric infections (e.g., lactoferrin), whereas inflammatory factors, such as in subclinical and clinical mastitis and which influence postnatal infection transmission via breast milk (e.g., HIV), may directly influence infant mucosal immunity as well as the infant’s microbiome. This is an area for further research.

d. Mastitis.

Mastitis, an inflammatory condition of the breast tissues, may or may not be accompanied by infection (647, 648).Milk stasis and infection are the 2 majors causes of mastitis (647, 649), but the concept of mastitis as used in contemporary empirical studies remains inconsistent, perhaps due to the disparate views of its etiology (650). The details of the inflammatory response have been best described in cows (651).

In humans, subclinical mastitis (SCM) is usually defined by the absence of external inflammatory symptoms and one or more of the following indicators: reduced milk secretion, increased sodium-to-potassium (Na⁺:K⁺) ratio, elevated IL-8 concentrations, elevated milk leukocyte or somatic cell count, and/or a high bacterial count in a milk sample. Other ancillary tests include the following: milk microbiological cultures, N-acetyl-β-d-glucosaminidase activity, pH, lactose content, electrical conductivity, flow measurements, and quantification of APPs (648, 652). Some suggest that somatic cell and IL-8 counts are more informative indicators of mammary inflammation than sodium content (653). Others suggest microbial identification techniques based on pyrosequencing of the 16S ribosomal RNA gene to obtain a description of the human-milk microbiome (654), as a useful diagnostic tool to assess pathogen persistence (655). Metagenomic and transcriptomic analysis of milk samples from healthy and mastitis-affected women are currently underway. Preliminary results indicate that human mastitis may be characterized by a mammary bacterial dysbiosis, a process in which the population of mastitis agents increases whereas other bacteria, normal or commensal mammary microbiota, decrease (648). How these responses relate to, or result from, maternal nutritional factors is little studied.

Mastitis enhances vertical mother-infant transmission of infections, particularly HIV (656). Prompt diagnosis and treatment prevents lactation failure, recurrent mastitis, and breast abscess (655), but the source of entry of pathogenic bacteria remains uncertain. In contrast to mastitis, systemic inflammation, although unconfirmed, may be the cause of SCM. SCM may affect the health of the newborn (657), because it may alter milk composition through changes in intramammary inflammation and leakage, changes in the metabolic activity of the mammary epithelial cells, and inflammation-mediated changes in nutrient transport (648, 652).

Mastitis has been associated with oxidative stress (658) and changes in cytokine gene and protein expression associated with adaptive immunity (659). In some conditions, supplementation with vitamins A, D, and E and copper and selenium was protective against mastitis (652), but others reported in HIV-infected women that antioxidant micronutrient supplementation did not lower (660), and might even increase the risk of SCM (661).

e. Polymorphisms and epigenetic effects associated with infection, inflammation, oxidative stress, and pregnancy outcomes.

Polymorphisms in the TLR genes have been associated with both increased susceptibility to infections, in particular bacterial vaginosis, and with preterm labor (613). Some superoxide dismutase and catalase polymorphisms have been related to the development of prematurity complications. It was suggested that single nucleotide polymorphisms involved in the generation of ROS and reactive nitrogen species, or free radical scavenging proteins, might be used to identify preterm infants who might benefit from specific antioxidant prevention strategies against free radical–mediated diseases (662). More recently, the emerging role of oxidant molecules in fetal programming was reviewed (663). It is now recognized that genomewide association studies only explain a small proportion of the known heritability of phenotypes. Future studies based on higher definition exome chips that can identify highly penetrant but rare variants, using full genome sequencing to better examine regulatory regions, and studies of genone-epigenome interactions, will better illuminate these issues.

Basic principles of epigenetics were highlighted by WG 1 (Text Box 10). More specifically, in pregnancy, the epigenetic effects of maternal infection, oxidative stress, and nutrition exposures on fetal immune response genes have been little studied (663). Nutritional interactions may be critical because oxidative stress has been associated with epigenesis (663). Changes in methylation patterns can be directed or stochastic in nature. Animal experiments clearly show that changes in maternal methyl-donor supply (choline, methionine, folate, and vitamin B-12) can influence the offspring phenotype (664). Emerging data from humans are confirming such associations and, importantly, indicate that maternal nutritional exposures before and during the earliest stages of pregnancy can be critical (665, 666). Folic acid supplements in pregnancy have been associated with a slightly increased risk of wheeze and lower respiratory infection in children up to 18 mo, which may relate to methyl donors in the maternal diet during pregnancy that influence respiratory health in children, consistent with epigenetic mechanisms (667).

Similarly, the consequences of cigarette smoking in pregnancy on these important aspects of early development may have an epigenetic basis (668), and may influence nutrition in children, including risk of obesity and stunting (669).The periods of gestational susceptibility of the epigenome need to be defined precisely, particularly in relation to peak gestational periods of exposure or infection risk. An example of an infectious disease exposure is the well-established peak P. falciparum malaria prevalence at 13–16 wk gestation in primigravidae living in high malaria transmission areas (670).

f. Profiling nutritional and inflammatory biomarkers in mothers and infants in relation to clinical outcomes.

Profiling of nutritional and inflammatory biomarkers in mothers and neonates at birth should be emphasized to allow identification, through longitudinal studies, of baseline characteristics related to subsequent infant immunologic responsiveness, infection risk, nutritional status, and vaccine efficacy. In particular, there is a need for a better understanding of mechanisms affecting iron acquisition of pathogens causing pregnancy and perinatal infections, because this could have implications for maternal iron supplementation (629). Two recent epidemiologic studies showed that iron status predicted malaria risk in very young children. Tanzanian infants recruited at birth and who developed ID during follow-up had significantly decreased odds of subsequent parasitemia (23% decrease; P < 0.001) and subsequent severe malaria (38% decrease; P = 0.04) (671). ID was also associated with 60% lower all-cause mortality (P = 0.04) and 66% lower malaria-associated mortality (P = 0.11) (671). Likewise, malaria risk was predicted by iron status in Malawian preschool-aged children, suggesting that ID protects against malaria parasitemia and clinical malaria in young children (672). Children with ID at baseline had a lower incidence of malaria parasitemia and clinical malaria during a year of follow-up [adjusted HRs (95% CI): 0.55 (0.41, 0.74) and 0.49 (0.33, 0.73), respectively]. Whether these infection risks are definable using birth/neonatal nutritional and/or inflammatory profiling is uncertain. The extent to which invasive bacterial disease is present may also be a risk and unclear (673).

In pregnant women, host iron status may affect maternal and/or neonatal infection risk, potentially contributing to neonatal death (629). Iron acquisition mechanisms of pathogens causing stillbirth, preterm birth, and congenital infection are described but not understood in relation to host iron status. There is in vitro evidence that iron availability influences the severity and chronicity of infections that cause these clinical outcomes. However, in vivo, the risk is unknown because relevant studies of maternal iron supplementation did not assess infection risk (629). Caution with maternal iron supplementation may be indicated in iron-replete women who have high infection exposure, but the challenge in distinguishing iron-replete and iron-deficient women remains. The investigation of infection risk in relation to iron status in mothers and infants and the relevance of nutritional biomarker profiling is certainly an area in need of further research.

a. Nutrition and pharmacology: overview.

In the face of pandemic infection and the explosion of NCDs, the use of available single- and multiple-drug regimens will be expected to increase in all settings. Similarly, efforts to ameliorate hunger and more specifically single- and multiple-micronutrient insufficiencies continue to be a high global health priority. The potential clash of these public health interventions demands attention. Numerous examples of the potential for adverse effects of nutrient-drug interactions have been explored previously in the context of conditions such as cancer (674) and HIV (216, 675).

The role of nutrition in all aspects of pharmacology (pharmacokinetics and pharmacodynamics) was been outlined previously (216). Essentially, nutrients and drugs share common pathways and mechanisms of absorption, metabolism, transport, and elimination. Consequently, the potential exists for either of these to affect the other. The implications of these potential interactions are significant when one appreciates the magnitude and complexity of the global health reality, i.e., that major infectious diseases and NCDs are treated with the available single- (and more often multiple-) drug regimens. As noted throughout this report, these conditions continue to be highly prevalent in LMICs, subject in many cases to the impact of food insecurity and all aspects of the malnutrition continuum, including exposure to large-scale, high-dose micronutrient interventions. Thus, a need exists to recognize the potential for adverse outcomes consequent to the bidirectional relation between nutrients and drugs.

Aside from the well-established mechanisms of interactions outlined in Text Box 38, recent studies have highlighted another route, via genetic induction of key enzymes within critical pathways of drug metabolism.

For example, 1,25(OH)₂D was observed to regulate genes responsible for the production of enzymes (including CYP3A4) responsible for detoxification in the intestine (676). This study, which was followed by numerous others, highlights an important and underappreciated role for vitamin D in drug metabolism through the induction of gene expression of key drug-metabolizing enzymes. This phenomenon should be evaluated closely, particularly in light of the expanding interest in vitamin D supplementation to boost immune function and for use in numerous conditions including HIV and other infections. Several of these types of interactions, including a similar enzyme-nutrient interaction involving vitamin A, were highlighted by WG 1.

The following is a brief coverage of specific aspects of these relationss as pertains to inflammation, nutrition, and health.

b. Steroids.

Numerous studies in humans have documented effects on inflammation of various drug exposures, although how these changes translate into effects on nutritional status or nutrient status or assessment (i.e., biomarker performance or interpretation) is less well defined. A basic categorization into steroidal and nonsteroidal compounds might be used for considering the influence of drug exposures. One of the most extensively documented compounds is the human glucocorticoid cortisol (or hydrocortisone). Various synthetic glucocorticoids are used in therapy and affect the immune system, with specific clinical consequences. . Glucocorticoids antagonize TNF-stimulated lipolysis and reduce resistance to the antilipolytic effect of insulin in human adipocytes (677). The extent to which this occurs may reduce the potentially deleterious effects of excess lipolysis and contribute to fat deposition in obesity, as well as inducing appetite leading to weight gain. Other effects include the inhibition of prostaglandin synthesis at the level of phospholipase A₂ and cyclo-oxygenase isomerase, potentiating the anti-inflammatory effect (678). Glucocorticoids act nonselectively and may impair normal anabolism, leading to negative nitrogen balance, altered amino acid profiles, and muscle breakdown.

c. Nonsteroidal compounds.

Several anti-inflammatory drugs fall into this category, including aspirin, statins, and flavones, all of which may have an adjunctive effect on chronic inflammation. Aspirin can attenuate lipid-induced insulin resistance in healthy men, unrelated to changes in inflammatory markers (679). Aspirin is also commonly associated with gastrointestinal bleeding, potentially leading to mucosal injury and ulceration (680), and oxidative stress associated with released iron and inflammation, a scenario that can contribute to altered nutrient/micronutrient absorption and should be considered in clinically assessing nutritional status.

Reduced inflammation is also an effect of statins, which, in the NHANES 1999–2004, were associated with lower CRP and white blood cell (WBC) counts (681). These multiple effects indicate that drug exposures should be considered in the assessment of nutritional homeostasis. A detailed review of these effects is required to improve understanding of the consequences of the potentially wide variety of drug exposures on nutritional biomarkers in various clinical situations.

a. WBCs.

Leukocytosis, an increase in the number of WBCs, has long been recognized as a nonspecific marker of systemic infection, tissue damage, allergic reaction, malignancy, or physiologic processes. Leukocytosis is classified according to the component of WBCs that contribute to an increase in the total number of WBCs; types include increased neutrophil count, lymphocyte count, monocyte count, eosinophilic granulocyte count, basophilic granulocyte count, or blastocyte (immature) cells. WBC count is routinely used clinically as an initial screening marker in patients with fever. Although most bacterial infections cause neutrophilia (≥10,000/mm³), the WBC and neutrophil counts alone are not sensitive or specific enough to accurately predict bacterial infection. Automated cell counter analyzers, common in many clinical laboratories, provide precise information on WBC counts as part of a complete blood count profile using only a small volume of freshly collected anticoagulated whole blood. Cutoff values for elevated WBC counts are age-specific and vary greatly (>11,000/μL in adults).

b. CRP.

CRP is an APP produced in response to acute injury, infection, or other inflammatory stimuli. Because CRP concentration increases within 4–6 h of an insult and is a measure of underlying systemic inflammation, it has numerous clinical uses including detecting bacterial infection, autoimmune disease, and malignancy, as well as measuring response to treatment (714). CRP may also remain elevated in the presence of a continued, chronic inflammatory process [e.g., atherosclerosis, rheumatoid arthritis, and obesity (715)].

Clinically, 5 mg/L originally defined the limit of detection of the method of analysis, and 10 mg/L was accepted as the upper level of the normal range. To identify subclinical inflammation in apparently healthy people in population studies and in research, 5 mg/L has generally become accepted as the upper limit of the normal range (716). CRP is used (in combination with other risk factors) to define risk groups for CVD.The American Heart Association and the CDC defined CVD risk groups as follows: low-risk CRP, <1.0 mg/L; average risk, 1.0–3.0 mg/L, and high risk, >3.0 mg/L (717).

Small differences in normal CRP concentrations have been found relative to race, age, and sex (718–720); however, because CRP responses to inflammation are usually large, and these differences are small, they received little attention until the introduction of the new high-sensitivity CRP assay that allows measurement to concentrations of 0.3 mg/L, compared with 3 mg/L for the modern conventional assay. Population data from NHANES 1999–2002 indicate that concentrations of CRP are higher among women than amoung men (median: 2.7 vs. 1.6 mg/L, respectively) and increase with age [median: 1.4 mg/L (20 – 29 y) vs. 2.7 mg/L (>80 y)] but vary less across categories of race or ethnic background (719). Later work by Cartier et al. (720 found that if CRP concentrations are adjusted for subcutaneous adipose tissue, the sex differences disappear, suggesting the greater subcutaneous fat in women to be responsible for the higher CRP concentrations.

The preanalytical stability of CRP in serum or plasma is excellent (11 d at room temperature, 2 mo refrigerated, 3 y frozen at −20°C) (721). No significant difference has been observed between samples collected either while fasting or nonfasting (722) and diurnal variation of CRP is negligible (723). Most CRP assays provide comparable results (724, 725), probably due to the availability of international reference materials and external quality assessment (EQA) programs to monitor the degree of variability and bias in laboratory assays (e.g., UK National EQA Service, US College of American Pathologists).

c. AGP.

AGP is a slower reacting positive APP (see Table 10) (24). At birth, concentrations are low and increase slowly, reaching adult values by 10 mo of age (726).

The methods of analysis are similar to those for CRP. There is a certified human serum reference material [ERM-DA470/IFCC (727)] for calibration and since 2008 it has been possible to standardize methods against this. There is an EQA scheme available, which is run by Randox Laboratories Ltd. UK (728).

AGP is not commonly used in clinical situations but is used for assessing longer-term (>5 d) or subclinical inflammation in population studies. With the onset of infection or trauma, changes in plasma nutrient concentrations closely parallel those of CRP. As clinical symptoms disappear, CRP concentrations fall sharply, but nutrient concentrations (e.g., iron, vitamin A) do not respond as rapidly. Therefore, AGP, which remains elevated longer than CRP and may take 3–5 d to reach a plateau, is often used to monitor inflammation during this time period (24).

The combination of CRP and AGP has been used to detect those who have only recently been infected and who are not yet showing clinical evidence of disease (increased CRP only) and those who have recovered and are convalescing (increased AGP with or without an increased CRP). However, as mentioned earlier, the changes in CRP and AGP have been described on the basis of acute tissue injury secondary to surgical procedures, so it is unclear how CRP and AGP act in response to an infectious agent (24).

d. ESR.

The ESR is defined as the rate at which erythrocytes suspended in plasma settle when placed in a vertical tube (mm/h) and is a nonspecific measure of inflammation that primarily reflects plasma viscosity. It is often assessed in low-resource settings with the use of an upright tube, but automated analyzers are also available. Despite its wide use, ESR has a number of disadvantages compared with other biomarkers (e.g., CRP), including slow changes in response to treatment and false elevation secondary to anemia. ESR is probably most useful clinically when markedly elevated (>100 mm/h), indicating infection or neoplasm (729).

e. Albumin.

Albumin is most commonly used as a marker of nutritional status (adequate protein intake) and also acts as a negative APP. The complex mechanisms responsible for serum albumin control render the interpretation of serum albumin concentrations challenging (730). Unlike prealbumin or transthyretin, which has a short half-life (∼2 d) and therefore reflects recent dietary intake, albumin has a long half-life (∼20 d); however, albumin is affected by several other factors and therefore has little value in the assessment of monitoring of nutritional status (731). Distribution ranges have been shown to differ on the basis of age and sex (732). A normal concentration is 3.5–5 g/dL (732).

f. PCT.

PCT is the prohormone of calcitonin. It has been identified as a biomarker of inflammation with potential for distinguishing bacterial from viral infections and correlates well with clinical severity (733, 734). Compared with either CRP or WBCs, PCT has been shown to be a better marker for identifying patients with invasive bacterial infections (735). In healthy individuals, circulating concentratons of PCT are generally very low but can increase by thousands-fold within 4 to 6 h in response to systemic infection (peaking as early as 12 to 24 h after onset). By contrast, CRP concentrations increase slowly during the first 6–12 h, peaking 48 to 72 h after infection onset. PCT also shows a dose-response relation, with higher concentrations correlating with increased severity of infection (735). The cutoff point for PCT varies widely and is most commonly ≥0.5 μg/L but can be as low as ≥0.12 μg/L (736). Multiple methods for PCT measurement are available from serum or plasma, including rapid, manual, and automated immunometric assays. The choice of assay and cutoffs depends on the intended clinical use (737).

g. TNF-α.

The origin of TNF in chronic inflammation is mainly the infiltrated macrophages in adipose tissue (738). Concentrations of TNF are 7.5 higher in obese than in lean subjects (739). Increased concentrations have also been observed in smokers; in patients with type 2 diabetes, metabolic syndrome, and atherosclerosis; and in aging (739–741). TNF promotes endothelial activation, increasing vascular permeability; decreases insulin sensitivity; and contributes to the progression of atherosclerosis (742–744). Plasma concentrations of TNF predict the risk of myocardial infarction (745). The most common available methods for TNF measurement are ELISA and radioisotope-labeled immune assays. Cutoffs are not standardized, and the reference range can vary from 0.55 to 2.53 pg/mL (746).

h. IL-6.

IL-6 has been classified as both a pro- and anti-inflammatory cytokine (747). It inhibits TNF production (748) and also stimulates the production of IL-1 receptor antagonist, IL-10, and soluble TNF receptors, which have anti-inflammatory effects (747, 749). IL-6 is released from contracting muscles during exercise, inducing lipolysis and fat oxidation and is involved in glucose homeostasis (689). The role of IL-6 in insulin resistance remains controversial. IL-6 can be measured by ELISA. Although IL-6 precedes the increase in CRP after exposure to bacterial infection, it has a very short half-life. Cutoff points vary in the literature from 25 to 150 ng/L (750, 751).

i. Conclusions.

In summary, numerous biomarkers of inflammation are available to detect acute, subclinical, and chronic inflammation in both clinical and public health settings. Table 11 summarizes the most commonly used biomarkers of inflammation. Text Box 42 contains a summary and some suggestions with regard to the available biomarkers of inflammation.

a. Overview.

The myriad of causative factors and types of inflammation that need to be considered in the context of nutritional assessment were covered in the previous WG reviews and are briefly summarized in Text Box 43. Similarly, basic concepts with regard to the nature of the nutrient-inflammation relations were also covered previously by the INSPIRE WG reviews. The following sections cover the key features and challenges to interpretation presented by the inflammatory response of biomarkers of specific nutrients. The nutrients included below are those that have been addressed in the previous WG reviews, with a particular focus on the 6 nutrients that were covered under the BOND project: iron, zinc, iodine, vitamin A, folate, and vitamin B-12. These nutrients were selected for their public health importance and because the relations described represent the range of issues that the community needs to address when choosing and interpreting a nutrient or nutrients for clinical or population surveillance purposes. Vitamin D and hemoglobin are also covered because of their obvious importance to global health. The emphasis on these nutrients does not imply that similar effects are absent with other nutrients.

b. Iron.

ID remains the most common micronutrient deficiency in the world, but the distinction of nutritional ID is often obscured by inflammation because the normal control of iron metabolism is disrupted by the primary mediators of the APR (i.e., TNF and IL-1) (752).

As outlined in Text Box 44 and reviewed in detail elsewhere [(753) and BOND iron review], a myriad of biomarkers exist that can be applied to the assessment of iron nutrition.

The BOND iron review can be found on the BOND website (30). The complexity of iron assessment led the WHO/CDC technical consultation on the assessment of iron status at the population level to compare the theoretical advantages for population-based surveys of the following 5 indicators of iron status: hemoglobin, serum ferritin, soluble transferrin receptor (sTfR), ZPP/erythrocyte protoporphyrin (EP), and mean cell volume (MCV) (26). The WHO/CDC consultation recommended serum ferritin as the single best indicator in areas where inflammation is less common and measuring sTfR in areas where inflammation is prevalent.

Hepcidin is a promising iron biomarker, because it is suppressed during ID and anemia but stimulated by inflammation, infection, and iron overload (754). Although hepcidin may be a sensitive marker for iron utilization and absorption, few data are available on hepcidin concentration and its relation to established iron markers in population studies. The relative strengths/weaknesses of other biomarkers of iron were previously reviewed (753).

c. Vitamin A.

VAD remains a significant public health problem, primarily in preschool-aged children and pregnant women (770). The epidemiology of VAD was discussed in the WG 2 review and indicators of vitamin A have been reviewed extensively (BOND review) (30). Text Box 47 provides a summary of currently available biomarkers of vitamin A.

d. Zinc.

The importance of zinc to health and, more specifically, the immune system and the APR was described by WGs 1 and 2 and reviewed recently (371). With specific regard to zinc assessment, the current view of the international nutrition community as represented by the International Zinc Nutrition Consultative Group and WHO is that serum or plasma zinc concentration is the primary indicator for use in the assessment of zinc status of a population (375). Potential factors influencing the interpretation of serum/plasma zinc concentrations include age, sex, time of day of the blood collection, and the fasting status of the individual (371). Cutoffs have therefore been established for different age and sex groups and blood sampling conditions (777). In addition to plasma/serum zinc, a number of other biomarkers (listed in Text Box 48) have been identified and are at various stages of viability for either clinical or population-based use. With the exception of plasma/serum zinc, most are not widely used either clinically or for population-based surveillance.

e. Folate.

The role of folate in public health has been reviewed extensively (783). Of the currently available biomarkers, the most commonly used are serum folate, RBC folate, and plasma homocysteine concentrations. Text Box 49 provides a list of available biomarkers for assessing folate.

f. Vitamin B-12.

The role of vitamin B-12 in health and disease was previously reviewed (785, 786). The available biomarkers of vitamin B-12 were also recently reviewed (785) and are summarized in the Text Box 50.

g. Iodine.

Iodine is a trace element required by the thyroid gland to produce thyroxine and tri-iodothyronine, thyroid hormones necessary for multiple processes related to normal growth and development. The biology including a review of all available biomarkers (Text Box 51) of iodine was recently completed (788).

h. Vitamin D.

VDD is common in much of the world. The risk or prevalence is higher in dark-skinned people, infants/young children, pregnant women, the elderly, and in women who are covered for religious reasons as well as in northern latitudes, particularly after winter. As outlined by WGs 1 and 2, an expanding body of evidence attests to the importance of vitamin D to immune function and the inflammatory response, with potential therapeutic applications for acute and chronic infectious and inflammatory diseases (790). Of the currently available biomarkers (Text Box 52), 25-Hydroxy vitamin D [25(OH)D] has been accepted as the most reliable indicator, particularly of exposure (791).

i. Summary of impact of inflammation on nutrient biomarkers.

The impact of inflammation on nutrient biomarkers is well described for several key nutrients, such as iron and vitamin A. For many nutrients (e.g., vitamin D or iodine), there appears to be an association between their biomarkers and markers of inflammation. However, it is not at all clear whether this reflects a cause or effect relation. More study is needed to refine our understanding of these relations. In the interim, because of the ubiquitous nature of acute/chronic inflammation and its effects on nutrient biomarkers, WG 4 suggested that the routine assessment of inflammatory status be included along with nutrient assessment. This will enable a more reliable and valid interpretation of the results of micronutrient status assessment of both patients and populations.

The WG recognized that the inflammatory response may be associated with higher or lower nutrient biomarker concentrations (i.e., it will be difficult to determine a cause or effect relation). In addition, the magnitude of effect in different settings and how to account for the effects of individual factors remain important research questions. Table 12 summarizes the effects of inflammation on some of the key nutrient biomarkers used in both clinical and public health settings. Additional details on the assessment, cutoffs, and interpretation of nutrient biomarkers are provided in the BOND reports found on the BOND website (30).

a. Considerations for clinical settings/point of care.

As reviewed by WG 3, infection and inflammation can affect individual nutritional status through several mechanisms, including suppression of appetite, altered nutrient absorption and excretion, increased caloric requirements through altered metabolism, and nutrient-drug interactions. In a study that estimated the magnitude of the effect of inflammation on plasma micronutrient concentrations of hospitalized patients, for each micronutrient the change in plasma concentrations varied markedly from patient to patient (793). The authors concluded that, in the presence of moderate or severe systemic inflammatory response (CRP >20 mg/L), plasma concentrations of selenium, zinc, and vitamins A, B-6, C, and D were clinically uninterpretable. However, it was possible to interpret these micronutrients if the systemic inflammatory response was less severe: for zinc when CRP is <20 mg/L, for selenium and vitamins A and D when CRP is <10 mg/L, and for vitamins B-6 and C when CRP is <5 mg/L. In clinical settings, it may be best to re-evaluate the nutritional status of individuals when the inflammation has resolved. However, certain chronic disease conditions, such as short gut syndrome, feature a high prevalence of subclinical inflammation as well as rapid transit and malabsorption of nutrients, and thus a high risk of multiple-micronutrient deficiencies (794). Similarly, children in resource-poor settings who are exposed to frequent infections with associated anorexia are at risk of long-standing malnutrition of clinical significance. In such settings, regular, frequent monitoring of micronutrient status is needed and should be assessed with measurements of APPs, such as CRP.

b. Considerations for population-level surveys.

Population-level estimates of micronutrient status are important for the implementation, monitoring, and evaluation of nutrition programs, as well as for comparisons between countries. Population-level interventions to address micronutrient deficiencies focus on the prevention of deficiency and shifting the population distribution of a biomarker (or reducing overall prevalence) rather than treating high-risk individuals. The effects of inflammation on population-level estimates of nutrition status are significant and can lead to distorted estimates of micronutrient deficiencies, either overestimating or underestimating the true prevalence, if inflammation is not accounted for in the analysis (357). Furthermore, the extent of the effect likely depends on the prevalence of inflammation and prevalence of the deficiency in the population.

Examples of where accounting for the effect of inflammation on micronutrient prevalence estimates would be helpful include the following:

• ● A country assesses the prevalence of a micronutrient deficiency, finds the prevalence is high, and implements a population-level intervention program, such as fortification, micronutrient powders, or supplementation. Another survey is performed at a later time to see if the prevalence of the micronutrient deficiency has decreased. If the prevalence of APPs differs between the 2 surveys, this could potentially mask an apparently effective intervention program or make an intervention appear to be successful when in fact it was not.

• ● In comparing countries or geographic areas within a country, decisions may be made to start intervention programs and/or allocate resources in areas with a higher prevalence of a micronutrient deficiency. If the level of inflammation differs between geographic areas, this may affect the prevalence of the micronutrient deficiency and result in inappropriate program interventions and/or allocation of resources.

Before deciding on approaches to account for inflammation, issues related to biomarker assessment must first be addressed, as follows:

• ● To assess the prevalence of micronutrient deficiencies, the methods used to collect, store, transport, and analyze the specimens must be performed with high quality. There are generally expert groups that provide guidance on the laboratory tests available, their advantages and disadvantages, laboratory quality assurance programs, and cutoff values to identify abnormal concentrations (e.g., WHO expert groups, International Council for Control of Iodine Deficiency, BOND).

• ● The quality of CRP and AGP measurements is also important, as is ensuring the quality of laboratory methods. One area that is in need of further investigation is to assess the relation between APPs and micronutrient biomarkers, each on their continuous scale (which may need to be transformed), and determine if there are clear cutoffs for the APP or if alternative methods to account for the APPs that do not rely on cutoffs are needed. Currently, one cutoff is used for CRP and/or for AGP and applied equally to all micronutrient biomarkers thought to be affected by inflammation, but this approach would benefit from further investigation.

• ● Some surveys measure only CRP, some only AGP, and some collect both. Strategies on how to deal with a variety of approaches based on these APPs from existing surveys are needed. If resources are available, measuring both CRP and AGP is preferred. If limited, either CRP or AGP should be measured, although further investigation is needed to determine, for each nutrient biomarker, whether CRP alone, AGP alone, or some combination of the 2 can account for inflammation.

c. Review of proposed approaches.

Potential approaches to account for inflammation will be discussed in detail and include ignoring inflammation, exclusion of individuals with inflammation, adjustment of the biomarker cutoff values, correction factors, statistical approaches, and others. Again, the approach will depend on whether applied in a clinical or public health setting and the context or underlying prevalence of inflammation.