The Immune System in Children with Malnutrition—A Systematic Review

Background Malnourished children have increased risk of dying, with most deaths caused by infectious diseases. One mechanism behind this may be impaired immune function. However, this immune deficiency of malnutrition has not previously been systematically reviewed. Objectives To review the scientific literature about immune function in children with malnutrition. Methods A systematic literature search was done in PubMed, and additional articles identified in reference lists and by correspondence with experts in the field. The inclusion criteria were studies investigating immune parameters in children aged 1–60 months, in relation to malnutrition, defined as wasting, underweight, stunting, or oedematous malnutrition. Results The literature search yielded 3402 articles, of which 245 met the inclusion criteria. Most were published between 1970 and 1990, and only 33 after 2003. Malnutrition is associated with impaired gut-barrier function, reduced exocrine secretion of protective substances, and low levels of plasma complement. Lymphatic tissue, particularly the thymus, undergoes atrophy, and delayed-type hypersensitivity responses are reduced. Levels of antibodies produced after vaccination are reduced in severely malnourished children, but intact in moderate malnutrition. Cytokine patterns are skewed towards a Th2-response. Other immune parameters seem intact or elevated: leukocyte and lymphocyte counts are unaffected, and levels of immunoglobulins, particularly immunoglobulin A, are high. The acute phase response appears intact, and sometimes present in the absence of clinical infection. Limitations to the studies include their observational and often cross-sectional design and frequent confounding by infections in the children studied. Conclusion The immunological alterations associated with malnutrition in children may contribute to increased mortality. However, the underlying mechanisms are still inadequately understood, as well as why different types of malnutrition are associated with different immunological alterations. Better designed prospective studies are needed, based on current understanding of immunology and with state-of-the-art methods.


Introduction
Malnutrition in children is a global public health problem with wide implications. Malnourished children have increased risk of dying from infectious diseases, and it is estimated that malnutrition is the underlying cause of 45% of global deaths in children below 5 years of age [1][2]. The association between malnutrition and infections may in part be due to confounding by poverty, a determinant of both, but also possibly due to a two-way causal relationship ( Figure 1): malnutrition increases susceptibility to infections while infections aggravate malnutrition by decreasing appetite, inducing catabolism, and increasing demand for nutrients [3]. Although it has been debated whether malnutrition increases incidence of infections, or whether it only increases severity of disease [3], solid data indicates that malnourished children are at higher risk of dying once infected [2][3][4]. The increased susceptibility to infections may in part be caused by impairment of immune function by malnutrition [5]. The objective of this study was to investigate the associations of different types of malnutrition with immune parameters in children, through a systematic review of the literature.
Since most infections and deaths in malnourished children occur in low-income settings, the organisms causing disease are rarely identified. Therefore, little is known about whether these differ from pathogens infecting well-nourished children, and whether malnourished children are susceptible to opportunistic infections. Although opportunistic infections like Pneumocystis jirovecii and severe varicella has been reported in malnourished children [6][7], these studies were carried out before the discovery of HIV, and may represent cases of un-diagnosed paediatric AIDS. More recent studies have found that Pneumocystis jirovecii pneumonia is not frequent in malnourished children not infected with HIV [8]. However, quasi-opportunistic pathogens like cryptosporidium and yeast are frequent causes of diarrhoea in malnourished children [9], and malnourished children have a higher risk of invasive bacterial infections, causing bacterial pneumonia [8], bacterial diarrhoea [10-11], and bacteraemia [12][13][14], with a predominance of gram negative bacteria. Due to the high prevalence of invasive bacterial infections, current guidelines recommend antibiotic treatment to all children with severe acute malnutrition, even though the evidence behind is not very strong [14].
Non-immunological factors may also contribute to increased mortality in malnourished children: reduced muscle mass may impair respiratory work with lung infections [15]; reduced electrolyte absorption from the gut [16] and impaired renal concentration capacity may increase susceptibility to dehydration from diarrhoea [5]; and diminished cardiac function may increase risk of cardiac failure [17]. Thus, immune function may only be one of several links between malnutrition, infections and increased mortality, but most likely an important one.

Definitions of malnutrition
This review considers childhood malnutrition in the sense of under-nutrition, causing growth failure or weight loss, or severe acute malnutrition, either oedematous, or non-oedematous.
Growth failure caused by malnutrition has commonly been defined by low weight-for-age (underweight), length-for-age (stunting), or weight-for-length (wasting) [5]. Generally, older studies diagnosed malnutrition using weight-for-age, while newer studies tend to use weight-for-length. Recently, mid-upper arm circumference (MUAC) has been promoted to diagnose severe acute malnutrition, because of its feasibility and because it predicts mortality risk better than other anthropometric indices [18]. Other definitions of malnutrition include specific micronutrient deficiencies, intra-uterine growth restriction, and obesity, but these conditions are outside the scope of this review.

Severe Acute Malnutrition
Two forms of severe acute malnutrition in children exist: nonoedematous malnutrition, also known as marasmus, characterized by severe wasting and currently defined by weight-for-length zscore ,23 of the WHO growth standard, or MUAC ,11,5 cm; and oedematous malnutrition defined by bilateral pitting oedema ( Figure 2) [19]. Kwashiorkor refers to a form of oedematous malnutrition, the fulminant syndrome including enlarged fatty liver, mental changes as well as skin and hair changes [20]. The term ''marasmic kwashiorkor'', has been used to describe children The Immune System in Children with Malnutrition PLOS ONE | www.plosone.org with both wasting and oedema [21]. It is still unknown why some children develop oedematous malnutrition, and unclear whether this form of malnutrition is associated with a different degree of immune deficiency.

Materials and Methods
A systematic literature search was carried out in PubMed using combinations of the search terms related to malnutrition and immune parameters. The full search strategy and the search terms used are described in Figure 3.
Inclusion criteria were: studies presenting original clinical data regarding immune parameters in children, aged 1-60 months, where a comparison was made, either between malnourished and well-nourished children, or between malnourished children before and after nutritional rehabilitation. Exclusion criteria were studies of children with another primary diagnosis such as cancer, congenital heart disease or endocrine disease. Studies were accepted where children had co-morbid infections, since this is typically seen in malnourished children. Articles by RK Chandra were excluded, due to concerns about possible fraud [22]. Studies published in peer-reviewed scientific journals, as well as in books were included. Only articles in English were included.
The search was carried out in August 2013, and updated in December 2013. The search results were sorted by MJHR, based on titles, abstracts or full-text-articles. Additional literature was obtained from reference lists, text books and by personal communication with experts.
For data retrieval, studies were sorted according to whether they investigated barrier function (skin and gut), innate immunity or acquired immune system, and listed in tables based on the specific immune parameter studied. Some studies were included in more than one table. The following data was extracted from each article: year and country, number and age range of malnourished and well-nourished participants, type of malnutrition and whether included children fulfilled WHOs current diagnostic criteria for severe acute malnutrition, whether infections were present, immune parameter studied, methods used, how the parameter was associated with malnutrition, and whether children with oedematous and non-oedematous malnutrition were differentially affected.
The results of the included articles were summarized for each immune parameter. Due to the heterogeneous nature of study designs, participants and outcomes, it was not meaningful to synthesize the results in a meta-analysis. The main potential bias was presence of infection. For this reason, presence and effect of infection was considered for each study as well as for each outcome. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guideline was followed, except for the items relating to meta-analysis (Checklist S1).

Results
The search in PubMed yielded 3402 articles. By contacting experts in the field, an additional 631 papers were obtained. Reference list of all papers read were screened for relevant papers not included in the initial search. Of all the screened papers, 245 met the inclusion criteria ( Figure S1). Another 49 articles were identified which, in addition to children 1-60 months old, also included older children. These studies were not included in the main analysis, but used in a sensitivity analysis in which all studies were included. The result of this additional analysis was essentially similar to the results obtained with studies only including children less than 60 month (results not shown). The studies were published between 1957-2014, mainly in the 1970s and 1980s. Only 33 studies were published after 2003 ( Figure 4). The studies included 29 prospective studies that compared malnourished children to themselves after nutritional recovery, and 216 cross-sectional studies. Of the cross-sectional studies, 51 were community-based, comparing immune parameters in children according to nutritional status. The remaining 165 cross-sectional studies compared hospitalised malnourished children to well-nourished children, often recruited outside the hospital. In 53 studies, all children fulfilled WHOs diagnostic criteria for severe acute malnutrition [23]. The vast majority of these studies included children with oedematous malnutrition, while only two studies included children with non-oedematous malnutrition based on the new WHO growth standard.
The results of each immune parameter are summarized in Table 1, and the results of individual articles are summarized in Tables S1-14.

Epithelial barrier function
The barrier function of the skin and mucosal surfaces is considered the first-line defence of the immune system, upheld by the physical integrity of the epithelia, anti-microbial factors in secretions (e.g. lysozyme, secretory IgA and gastric acidity) and the commensal bacterial flora [24].
Of the articles describing barrier function in malnourished children, six described skin structure and function, 21 described structure and permeability of intestinal mucosa, 19 protective factors in secretions and 11 the microbial flora colonizing mucosal surfaces.
Skin. Skin barrier has mostly been studied in children with oedematous malnutrition, who may develop a characteristic dermatosis, characterized by hyper-pigmentation, cracking and scaling of the epidermis, resembling ''peeling paint'', providing a potential entry port for pathogens [25].
Six articles assessed barrier and immune function of the skin in malnourished children (table S1). Two articles describing histology reported atrophy of skin layers, but did not describe cutaneous immune cells [26][27]. Four articles described the ''cutaneous inflammatory response'': They made small abrasions in the skin, and placed microscopy slides over the sites. Similar or higher numbers of white blood cells migrated onto slides in malnourished children, predominantly granulocytes and a lower proportion of monocytes and macrophages [28][29][30][31]. This pattern was noted to resemble a neonatal immature immune response [30]. All four articles found this pattern in patients with oedematous malnutrition, while one study found that the response of non-oedematous children resembled that of well-nourished [30].
Two articles described immune cells in small intestinal biopsies from malnourished children in Gambia and Zambia: both reported increased lymphocyte infiltration, more T-cells, and cells expressing HLA-DR in malnourished children compared to English children [37][38]. However, it was similar to Gambian well-nourished children [38], and unaltered by nutritional recovery [37]. Both well-nourished and malnourished Gambian children had high levels of intestinal cytokine expression, but malnourished children had an increased ratio of cells expression pro-inflammatory to regulatory cytokines, compared to the wellnourished Gambian children [38].
The colon was only described in one article, reporting increased vascularity, atrophy of the mucosa and a tendency to rectal prolapse in children with oedematous malnutrition [52].
Four articles compared the intestine of children with oedematous and non-oedematous malnutrition: one study from South Africa found that the histological changes were most severe in those with oedema [40]. Two articles from Chile found that children with non-oedematous malnutrition had a thinner mucosa, whereas children with oedema had more villous atrophy and more cellular infiltration [35][36]. In contrast, a more recent study from Zambia found higher numbers of T-cells and cells expressing HLA-DR in the intestines of children with non-oedematous than oedematous malnutrition, while the intestines of oedematous children were deficient in sulphated glycosaminoglycan [37].
SIgA in saliva, tears and nasal washings was frequently reduced in severely malnourished children [54][55] [57][58]. One article from Egypt reported increased levels in children with oedematous malnutrition [56], but may have overestimated sIgA, since saliva flow was reduced in malnourished children, and sIgA was expressed as g/l, whereas other articles expressed it as sIgA as % of protein content. Studies of sIgA in duodenal fluid showed conflicting results [57] [59], as did studies quantifying sIgA in small intestinal biopsies [39][40] [64]. The sIgA content of urine was increased or normal in severely malnourished children [60][61]. In mild to moderately underweight children, inconsistent results were found for sIgA in tears [63] and saliva [53][54] [62][63].
Tear lysozyme content was found to be reduced in malnourished children [54] [63], while saliva lysozyme was unaffected [53][54]. Gastric acid secretion was consistently reduced in severely   malnourished children [65][66][67][68], and higher pH was associated with bacterial colonization of the stomach [65]. Microbial colonization. Microbes colonizing skin and mucosa may protect against infections by competing with pathogens, by producing specific antimicrobial substances, and by stimulating host immune function [69]. Despite much recent interest in the subject, of 11 articles describing the micro-flora in malnourished children, only four were published during the last ten years (table S4). All found malnourished children to host a different flora from well-nourished children. Their mouths and throats contained more yeast [70][71][72], and their stomach and duodenum, which in healthy children is considered to be almost sterile, contained a large number of microorganisms [72][73][74][75]. Although one study found similar degree of small intestinal bacterial overgrowth in diarrhoeal patients with and without malnutrition [75], another found more small intestinal bacteria in malnourished than in wellnourished children with diarrhoea [72]. While gram positive cocci predominated in the small intestine of well-nourished children, malnourished children hosted more gram negative bacteria [65] and yeast [74].
The colonic flora, containing the vast majority of commensal bacteria, was described by sequencing bacterial DNA from stool samples in four recent articles, which consistently found that the pattern of bacteria was different in malnourished and wellnourished children [76][77][78][79]. More bacteria with pathogenic potential were found in the malnourished children [77][78], and their flora was less mature [79] and less diverse [76] [78]. A twin study from Malawi suggested that micro-flora pattern could also play a role in developing malnutrition [76]. No articles have so far reported whether the intestinal flora is different in children with oedematous and non-oedematous malnutrition.

Innate immune system
The innate immune system delivers an unspecific response relying on leukocytes (like granulocytes, monocytes and macrophages), as well as soluble factors in blood (like acute phase proteins and the complement system) [24]. Of the articles describing innate immune response, 38 described number and function of leucocytes, 25 acute phase proteins and 24 complement components and activity.
Two studies from Nigeria and one from Ghana found no difference in the mean percentage of natural-killer-cells among malnourished or well-nourished children [94][95][96], although two reported that more malnourished children had abnormally low numbers of natural-killer cells. In Zambia, levels of dendritic cells were lower in blood from malnourished children before nutritional rehabilitation than after, and elevated inflammation markers were associated with a paradoxical lower level of dendritic cell activation. This was associated with endotoxin levels in the blood, and was interpreted as a type of immune-paralysis, related to inflammation and bacterial translocation [49]. Unfortunately, it was not assessed whether this was different from well-nourished children with severe infections.
Chemotaxis of granulocytes was reduced in malnourished children in three of five studies [80] [83] [97][98][99], and one study found a diminished ability to adhere to foreign material [100]. Results for phagocytosis were mixed: five of 12 studies found that leukocytes of malnourished children had reduced ability to ingest particles or bacteria [81] [108][109]. Neutrophils may kill microorganisms by producing reactive oxygen compounds; assessable by the Nitroblue Tetrazolenium (NBT) test, which, however, gave inconsistent results in malnourished children [83] [105] [110][111][112][113][114]. It has been hypothesized, that reactive oxygen production is involved in the pathogenesis of oedematous malnutrition [115]; however, the NBT test results did not show any clear pattern in children with oedematous compared to non-oedematous malnutrition.
One study found the levels of enzymes, like alkaline and acid phosphatase, to be increased in leukocytes from children with malnutrition [116]. More leukocytes of malnourished children were found to have markers of apoptosis (CD95) [92], and signs of DNA damage [117][118].
No articles have yet described the expression of patternrecognition molecules, like Toll-like receptors in malnourished children, although these are fundamental to the function of the innate immune system.
Acute phase response. Acute phase responses is induced by infection or trauma, and mediated by cytokines like IL-6 and TNF-a. It involve temporal suppression of acquired, and amplification of innate immune responses, with secretion of positive acute phase proteins (APP) like C-reactive protein (CRP), serum-amyloid-A (SAA), complement factors, a-1-acid-glycoprotein or ferritin [119], while levels of other proteins are reduced, as albumin, pre-albumin, transferrin, a -2-HS-glycoprotein, and afetoprotein. These are sometimes called 'negative acute phase proteins', although it is not clear whether their reduced level are due to active down-regulation, or because of competition with production of positive acute phase proteins. Twenty-four articles described the levels of acute phase proteins in malnourished children with or without infection (table S6).
Acute phase response to a controlled stressor. Four articles described the acute phase response induced by a vaccine. Two reported a normal [140] or increased [141] febrile response to measles vaccine in malnourished children. In another study, a similar rise in APP was seen in malnourished and well-nourished children [137], in response to a diphtheria-pertussis-tetanusvaccination, but the increase in APP was greater when the vaccination was repeated after nutritional rehabilitation. The same was found for the febrile response to a repeated vaccine in malnourished children [142]. Since no repeated vaccine was given to well-nourished children, it is unknown whether they would also have had a stronger response to the second dose.
Complement. The complement system consists of plasma proteins secreted by the liver that, upon activation, react to recruit immune cells, opsonize and kill pathogens [24]. Three main pathways activate the complement system: the classical pathway, the alternative pathway and the lectin pathway [143], with the complement protein C3 playing a central role in all three pathways.
Twenty-four articles described levels or in-vitro activity of complement proteins (  [156], or reduced only in oedematous, but not in non-oedematous malnutrition [157]. Alternative pathway activity was reduced in two studies [145] [156] and unaffected in one [146]. General opsonic activity of serum was reduced in one study [156]. No articles reported the activity of the lectin pathway.
Both reduced production and increased consumption may explain the reduced levels of complement factors. Complement components are produced by the liver, and their levels correlated with albumin levels, the production of which is also impaired in malnutrition [158]. However, increased consumption is also supported by one study showing high levels of C3d, a by-product after activation of C3, in malnourished children, most pronounced in oedematous malnutrition [148].

Acquired immunity
Acquired immunity is characterized by specialized cellular and antibody-mediated immune responses, generated by T-and Blymphocytes reacting with high specificity towards pathogens and creating long-lasting immunological memory. The acquired immune system also orchestrates tolerance to self and other nonpathogenic material like gut bacteria [24]. Of the articles describing acquired immunity, 12 described the thymo-lymphatic system, 21 delayed-type hypersensitivity responses (DTHR), 58 lymphocyte subsets in blood, 32 immunoglobulins in blood, 35 vaccination responses and 35 cytokines.
Thymus. The thymus gland is the central lymphatic organ in the acquired immune system, where maturation and proliferation of T-lymphocytes take place. The thymus is large at birth and undergoes gradual involution after childhood [159], with diminished output of T-lymphocytes [160].
Eight articles reported thymic size measured by ultrasound, in relation to nutritional status [91] [167][168][169][170][171][172][173] (table S9). Five of these studied children with severe malnutrition and found severe thymic atrophy [91] [167][168][169][170], reversible with nutritional rehabilitation, although thymic size did not reach normal levels as fast as anthropometric recovery [91] [170]. Thymic size was also measured by ultrasound in cohorts of children to determine patterns of thymic growth [159] [171], in a vaccination trial in Guinea Bissau [172] and in a pre-natal nutritional supplementation trial in Bangladesh [171]. These studies confirmed that thymus size was associated with nutritional status, even in mild malnutrition. Breastfed children often had a larger thymus than artificially fed children [174], possibly explained by IL-7 in breast milk [175], and children with a large thymus were found to have a higher chance of surviving than those with a small thymus [172] [176].
Other lymphatic tissue. Six articles reported investigations of other lymphatic tissue. Four autopsy studies found atrophy of lymph nodes, spleen, tonsils, appendix and Peyer's patches, although not as pronounced as in the thymus. Histology revealed a reduction in germinal centres and depletion of lymphocytes from para-cortical areas [161] [163][164][165]. Two studies in living children also found that the tonsils were smaller in malnourished than in well-nourished children [163] [177].  [191][192]. Conflicting results were found for DTHR in children with different types of severe malnutrition: Three studies found most impaired response in oedematous malnutrition [179] [181] [191], while one found that it was worst in non-oedematous malnutrition [184], and two studies found similar responses [186] [187].
Lymphocytes in blood. Fifty-eight articles reported either total numbers of lymphocytes or lymphocyte subsets in blood (  [199].
Three studies found that children with oedematous malnutrition had more atypical lymphocytes in blood, resembling plasma cells [81] [87] [93]. Other indicators of functional differences were higher density [200], different pattern of gene expression [201], and more markers of apoptosis in lymphocytes of malnourished children [92] [202].
CD4+ lymphocytes. With assessment of CD4 counts becoming widely available, it has been investigated whether the number of CD4+ lymphocytes was affected by malnutrition. In children without HIV, two of four studies using monoclonal antibodies and microscopy found reduced levels of CD4+ lymphocyte in malnourished children [ [212]. Bacterial infections were noted to reduce the CD4-count [86]. For malnourished children infected with HIV, it was hoped that renutrition alone could increase their level of CD4+ lymphocytes. However, a study from Zambia found that CD4 counts declined during nutritional rehabilitation in HIV-infected malnourished children without anti-retroviral treatment [198]. Thus, a low level of CD4+ lymphocytes can probably not be attributed to malnutrition, regardless of whether the child has HIV or not.
Three studies noted that level of CD4+ lymphocytes were higher in children with oedematous than with non-oedematous malnutrition [91] [198] [220], and several studies have noted that children with HIV were less likely to develop oedematous malnutrition [198] [220] [221], suggesting that some level of CD4+ lymphocytes could be required to develop the syndrome.
Activation markers on T-lymphocytes. Most flow cytometry studies assessing surface markers on T-lymphocytes have been carried out in Mexico, all comparing malnourished infected children with similarly infected well-nourished children. Malnourished children were found to have fewer effector T-lymphocytes, identified as cells lacking the ''naïve'' markers CD62L and CD28 [90], fewer activated T-lymphocytes, with the markers CD69 and/ or CD25 [212] [222] [223], and fewer memory T-lymphocytes identified by the marker CD45RO+ [86]. In contrast, a study from Ghana found similar numbers of activated T-lymphocytes, identified by HLA-DR, in malnourished and well-nourished children [94].
Antibody levels. Thirty-two articles described immunoglobulins in blood of malnourished children (table S12). Nineteen of 27 studies found no difference in IgG antibodies or total c-globulin between malnourished and well-nourished children [ [238][239]. IgD, present in low amounts in healthy children, was elevated in children with malnutrition in two studies [130] [233], or elevated in oedematous but not non-oedematous malnutrition [179], while one study found that it was similar to well-nourished children [82].
Antibody vaccination responses. Thirty-five articles described vaccination responses to a specific antigen (table S13). The articles either reported sero-conversion rates, or antibody titre response. Studies assessing sero-conversion rates in children with severe malnutrition found mixed results: Six of 10 studies found reduced sero-conversion rates in children with severe malnutrition to typhoid [101] [240], diphtheria [101], tetanus [101] [206], tetanus-diphtheria-pertussis (DTP) [234], hepatitis B [241], measles [141] [149] [242] and yellow fever [243][244], and two studies found that sero-conversion was delayed in malnourished children [245] [238]. Ten of 11 studies found that severely malnourished children responded with reduced antibody titres [101] [141] [149] [206] [233][234] [238] [240][241][242] [246], despite some of the studies finding acceptable sero-conversion rates. No study found that children with oedematous malnutrition had a normal antibody response to vaccination. One study from 1964 found improved antibody response to DTP in children with oedematous malnutrition randomized to a high-protein diet [247]. There did not seem to be any specific vaccines whose antibody response was more affected than others by malnutrition, nor was there any pattern in terms of responses to live or dead vaccines.
Cytokines. Cytokines are signal molecules acting locally between immune cells, and sometimes with systemic effects. Thirty-five articles described cytokines in malnourished children (table S14).
Leukotrienes (LT) are not strictly cytokines, but immune modulating molecules derived from long chain polyunsaturated fatty acids. Levels of LTC4 and LTE4 were higher, and LTB4 lower, in children with oedematous than with non-oedematous malnutrition, whose levels were similar to well-nourished [281], and prostaglandin E2 [282] was higher in children with oedematous malnutrition than in well-nourished.

Discussion
We identified and reviewed 245 articles about immune function in malnourished children. Some general problems apply to many of the studies, mostly related to their observational design. For this reason they can only describe associations, not causalities.
First, many studies were done in severely malnourished children from hospital settings, who were ill with infections, making it difficult to disentangle the immunological effect of malnutrition from the effect of infection. This problem has caused some to propose that there really is no immune impairment by malnutrition, and that all alterations seen are due to infections or underlying unknown immune deficiencies, which are also responsible for the poor growth [283]. Enteropathy could be an example of such an ''invisible'' condition, causing both immune deficiency and malnutrition. This hypothesis is difficult to test. However, some studies did try to account for this problem by selecting malnourished children without clinical infections, or by comparing them to well-nourished infected children. In studies from central Africa in the 1970s and 1980s, some malnourished children may have suffered from unrecognized paediatric HIV [284], giving obvious problems for interpretation.
Second, publication bias is a well-known problem, and may have occurred, particularly in older studies, where some small studies showed a dramatic effect.
Third, studies used different diagnostic criteria for malnutrition, making it difficult to determine the children's degree of malnutrition as defined by present-day criteria. While children in 52 of the studies fulfilled WHOs present criteria for severe acute malnutrition, only two diagnosed children based on the new WHO growth reference. Those defined as severely malnourished based on old growth references would most likely also be classified as severely malnourished today, since the new WHO standard tend to classify more children as severely malnourished, while some children then defined as moderately malnourished would be classified as severely malnourished today. The studies including children based on weight-for-age probably included children with stunting and wasting, without differentiating between the two.
Fourth, even using uniform criteria, malnourished children are a heterogeneous group. Anthropometric measurements are only crude markers of body composition, which -among other thingsreflect nutrient deficiencies. It is unknown what specific nutrients were deficient, and to what extent infection contributed. Deficits in lean tissue and fat tissue are plausibly different physiologic conditions, and children appearing similarly malnourished may be so for entirely different reasons, with different immunological consequences. No articles have so far reported reliable measures of body composition, simultaneously with markers of immune function. Probably, the consequence of malnutrition on immune function may also depend on the pattern and load of infections. Although most studies were carried out in low-income settings with high infectious loads, a few were from middle-or highincome countries. This may also contribute to inconsistencies in the results.
In spite of these limitations, common patterns emerge from the studies, summarized below ( Figure 5).

Immune parameters apparently not affected by malnutrition
Total white blood cell and lymphocyte counts in peripheral blood are not decreased in malnourished children, and granulocytes are frequently elevated. Likewise, T-lymphocytes and CD4 counts appear normal in malnourished children, when measured by flow cytometry, the gold standard for characterizing cell subsets. Their levels seem to be determined more by infections than by nutritional state, and do not reflect the degree of malnutrition-related immune deficiency, as high infectious mortality is seen in malnourished children, despite unaffected white blood cell counts [49].
Malnourished children can mount an acute phase response to infections, with elevated CRP and low negative acute phase reactants, and this can also be seen in absence of clinical infection. Thus, based on available evidence, the acute phase response, if anything, seems exaggerated rather than diminished. Levels of IgM and IgG are normal or elevated in malnourished children. Secretory IgA is not consistently lower in duodenal fluid, and frequently elevated in urine.

Immune parameters affected by malnutrition
The gut mucosa is atrophied and permeable in malnourished children. This enteropathy also affects well-nourished children in poor communities, but probably most severely in malnourished children. The condition appears similar to tropical sprue described in adults, and the term enteropathy of malnutrition has been replaced by the broader term environmental enteropathy [285]. At present, this condition is thought to result from high pathogen load rather than nutrient deficiencies, and thus primarily a cause of malnutrition, particularly of stunting [286] [287].
Production of gastric acid and flow of saliva is reduced in malnourished children. Secretory IgA is also reduced in saliva, tears and nasal washings from children with severe, but not moderate malnutrition. The small bowel of malnourished children is often colonized with abundant bacteria, and their pattern of commensal flora is altered. Granulocytes kill ingested microorganisms less effectively. Levels of complement proteins are low in blood from malnourished children, particularly in children with oedematous malnutrition, and less in children with moderate malnutrition.
Lymphatic tissue, particularly the thymus, undergoes atrophy in malnutrition in a dose-response fashion: thymic size depends on nutritional status even in milder degrees of malnutrition, and thymus size is a predictor of survival in children.
DTHR is diminished in malnourished children. Lymphocytes of malnourished children are less responsive to stimulation with PHA, fewer are activated and more cells have markers of apoptosis. Plasma IgA is mostly elevated in malnourished children, particular in those with oedema. Children with severe, but not moderate, malnutrition mount a lower specific antibody response to vaccination, although for most children sufficient to obtain protection. The lower titres seen in malnourished may be due to a delay in vaccination response.
Cytokines can be classified as those promoting a Th1 response of predominantly cellular immunity, and those promoting a Th2response of humoral immunity [24]. Although this approach has somewhat been replaced by other classifications [288], it seems useful to describe the profile of malnourished children, whose immune system seems tuned towards a Th2 response, with high IL4 and IL10, and low levels of IL-2, IL-12 and IFN-c. Elevated levels of IL-6 and TNFa may primarily be related to infections, and support the observation that induction of an acute phase response is intact in malnutrition. A more recent classification The Immune System in Children with Malnutrition PLOS ONE | www.plosone.org focuses on whether cytokines are predominantly inflammatory or anti-inflammatory [289]. Malnourished children appear to have high levels of anti-inflammatory cytokines and less clearly affected levels of pro-inflammatory cytokines in blood, in contrast to the predominantly pro-inflammatory cytokine expression in the gut of malnourished children.

Mechanisms
The mechanisms behind these immunological alterations are still not adequately understood. Some explain it by lack of energy and building blocks to synthesize the proteins required [290]. However, lack of building blocks does not explain why some immune parameters seem intact, or paradoxically elevated in malnutrition, such as plasma IgA, acute-phase proteins, leucocytes in blood, and production of Th2 cytokines. If it was simply a matter of lack of building blocks, all parameters of the immune system should be equally affected. The fact that the pattern of cytokines in malnourished children is tuned towards at Th2response fits with their high levels of immunoglobulins, reduction in thymus size and diminished DTHR. Still, the pathophysiology behind this Th2 skewedness remains unexplained.
Infections could obviously contribute to the changes seen, and interactions have been noted between infection and malnutrition in their respective effects on immune parameters [197]. However, although many of the immunological changes appear to be synergistically affected by malnutrition and infections, malnutrition also seems to be independently associated with altered immune function.
Animal studies suggest hormonal factors to be involved in the immune profile of malnutrition. Leptin [291], prolactin [292] and growth hormone [293] all stimulate thymic growth and function, and their levels are low in malnourished children. In support of this, a recent study found that a low leptin level was associated with a higher risk of death in malnourished children [272]. Growth hormone therapy increased thymic size and output in adult HIV patients [294]. In contrast, cortisol and adrenalin induce thymic atrophy in mice [295][296], and cortisol is high in children with malnutrition and other forms of stress. It is plausible that this hormonal interplay is implicated in the immune deficiency in malnourished children.
This hormonal profile is similar to that of an acute phase response, where thymus atrophy also occurs, acquired immunity is temporarily suppressed and innate immunity takes over [296]. This could explain why some malnourished children have elevated positive APP and most have depressed negative APP in absence of clinical infections. Zinc deficiency causes thymic atrophy [297][298], and acute phase responses lower plasma zinc, so zinc status may contribute to the immune deficiency of both malnutrition and acute phase responses.
In HIV infection, persisting subclinical inflammation and immune activation is frequently present, and may be partly responsible for immune deficiency and disease manifestations [299]. Given the frequent finding of elevated acute phase proteins in malnourished children, it seems plausible that a similar state of subclinical inflammation could be involved in both the impairment of immune function, and in the vicious circle of catabolism and deterioration of the nutritional status. However, in spite of elevated acute phase proteins, most studies have reported unaffected or even paradoxically lowered levels of activated Tcell and dendritic cells in malnourished children.
The intracellular receptor, mammalian target of Raptomycin (mTOR), is present in most cells. It responds to concentrations of nutrients in the cell's surroundings, and to other signs of stress, such as hypoxia, enabling the cell to adapt its metabolism to locally available nutrients. Immune cells also use mTOR to regulate their state of activation. Nutrient availability may thereby determine whether an immune cell is activated [300], and whether T-cells differentiate towards a pro-inflammatory or a tolerance-inducing phenotype [301]. Some immune cells may even deplete the microenvironment of certain nutrients, to manipulate the activation of mTOR. Accordingly, the significance of nutrients in the microenvironment expands from simple building blocks to signal molecules. Obviously, this mechanism could be involved in the immunological profile in malnutrition. However, no articles have yet described the activity of mTOR in malnourished children.
A research group working with animal models of malnutrition has proposed a theory called the ''tolerance hypothesis'' [302]. This suggests that the depression of cellular immunity in malnutrition is an adaptive response to prevent autoimmune reactions, which would otherwise occur as a result of catabolism and release of self-antigens. Although adaptive in this sense, it happens at the price of increased susceptibility to infections [303]. However, if this tolerance hypothesis holds true, one would expect to see occasional break-through of auto-immune reactions in malnourished children. Such phenomena have apparently not been studied.
The pathogenesis of oedematous malnutrition is still unknown. Many immune parameters seem affected to a different degree in children with oedematous malnutrition, with higher levels of IgA, higher levels of abnormal antibodies like IgD, poorer vaccination responses and cytokines more skewed towards a Th2-response; their complement levels are lower, which may partly be caused by increased consumption of complement in-vivo. The pattern of leukotrienes is different in children with oedematous compared to non-oedematous malnutrition. This immunological profile resembles that seen in autoimmune diseases such as lupus erythematosus [304][305]. Moreover, elevated immunoglobulins in children with oedematous malnutrition seem to correlate with its unexplainable manifestations, like dermatosis and oedema [231] [233]. It could be speculated whether this syndrome could indeed represent some kind of autoimmune reaction to malnutrition, perhaps resulting from a failure to induce efficient tolerance.

Conclusion
In spite of the prevalence of malnutrition, and its fatal consequences, scientific interest in the immune deficiency of malnutrition seems dwindling, and little research has been carried out on the topic during the last ten years. For this reason, most evidence on the subject relies on immunological methods used 30 to 40 years ago, many of which are no longer in use, and little research has been done with modern methods, and with the present understanding of immunology. Moreover, most studies have looked at isolated aspects of immune function, despite the fact that the parameters are interdependent, and the division into innate and adaptive immune function seems to be a simplification. Thus, our understanding of immune function in malnutrition is still very limited.
This review illuminates the little that we know about the immunological alterations associated with malnutrition, and also points to significant gaps in our knowledge. Future well designed prospective cohort studies should examine how immune parameters are related to morbidity and mortality in malnourished children, with detailed characteristic of nutritional status, preferably body composition, of infections, enteropathy and of low-grade inflammation. When testing nutritional and medical interventions for malnutrition, immune parameters should be included as outcomes. Studies should investigate newer immunological parameters in malnutrition, like expression of innate pattern recognition receptors (as the Toll-like receptor), the lectin pathway of the complement system and mTOR expression and activity. It should be investigated whether a small thymus is associated with lower output of recent thymic-derived T-cells, and how it correlates with hormones like leptin, cortisol, insulin and Insulin Growth Factor-1. Innate and adaptive immune parameters should be assessed simultaneously, taking into account their dynamic interdependency. To understand whether malnutrition is indeed associated with active downregulation of immune reactivity (as formulated in the ''tolerance hypothesis''), the balance between regulatory T-lymphocytes and their counterparts, Th17 lymphocytes should be measured. Finally, prospective studies among children at risk should assess whether immune profiles differ in those who subsequently develop oedematous and non-oedematous malnutrition, and it should be investigated whether children with oedematous malnutrition have markers suggestive of auto-immune or inflammatory diseases. Such studies would reduce our current ignorance on the interplay between malnutrition and infectious diseases.          Checklist S1 PRISMA Checklist.