Complement (C3), nutrition, and infection*

Bulletin of the World Health Organization, 57 (1): 113-121 (1979) Complement (C3), nutrition, and infection * A. A. KIELMANN 1 & L. M. CURCIO 2 Complement (C3) was determined and related to various parameters oJ nutritional status and past infectious disease experience in a group of 53 rural preschool children in North India. Mean complement level was 25% lower than in an age-matched European reference population. Low complement (C3) levels were associated mainly with children who were both stunted and wasted, as well as with those who had experienced frequent purulent skin infections in the past. Malnutrition has been shown conclusively to impair nonspecific cell-mediated and specific humoral defence mechanisms, resulting in increased incidence and prolonged duration of infectious diseases and ultimately in high childhood mortality. The complement system is one of the major com- ponents of the defence mechanisms against infection; knowledge about complement function in malnutri- tion is, therefore, of critical importance to the understanding of disease patterns in poorer countries. Since Smythe et al. (14) originally demonstrated impairment of complement function in children with severe PEM (protein-energy malnutrition), several investigations have shown that C3 levels are significantly reduced in malnourished children (2, 12, 13), but rapidly increase following nutritional rehabilitation (3, 11). Complement levels have also been found to be significantly reduced during acute infectious diseases, chronic tropical parasitoses, and autoimmune disease (1, 4, 9). Most studies concerned with nutritional effects on complement have been done on children in severe states of undernutrition, i.e., third degree according to the Gomez classifica- tion, frank marasmus, or kwashiorkor. Recently we have shown that complement (C3) levels were significantly reduced even in borderline under- nourished children free from infection at the time * This study was part of a World Health Organization sponsored collaborative investigation into the effects of malnutrition on immunity. The analysis of this part of the study was supported by World Bank Grant No. 671-38 and by the United States Agency for International Development (grant no. AID/TA-C-1450). ' Assistant Professor, Department of International Health, the Johns Hopkins University School of Hygiene and Public Health, Baltimore, MD 21205, USA. I Programmer/Analyst, Department of International Health, the Johns Hopkins University School of Hygiene and Public Health, Baltimore, MD 21205, USA. of examination (7). Since our earlier publication and largely in response to it, a number of questions have arisen. Was weight-for-age the appropriate indicator for measuring the level of nutrition or would another anthropometric or biochemical parameter, for instance weight-for-height or haemo- globin concentration, have yielded better results? It would similarly be important to know whether any specific nutrient deficiency state-protein deficiency, energy deficiency, or other nutrient deficiency (iron, folic acid)-was principally responsible for the observed reduction in C3 levels. Though other investigators had stressed the effects of acute illness on complement levels, little was known about whether, past illness had affected complement levels and whether illness and nutritional levels interacted in their effects on complement synthesis and utilization. Lastly, we wanted to re-examine our data to determine more precisely the inter- relationship between nutritional levels and comple- ment (C3). METHODS The study was done at the former Rural Health Research Center of the Johns Hopkins University, Department of International Health, at Narangwal, Punjab, North India. Blood samples were obtained from 53 children belonging to the study population of a larger longitudinal investigation into interactions of nutrition and infection (8). All children were normally active and free from intercurrent infections. The children were weighed to the nearest 20 g; height, arm circumference, and biceps skinfold thickness were determined according to Jelliffe (6). Complement (C3) levels were determined by radial diffusion (10); haemoglobin was determined colori- metrically (15); total serum protein was measured 3781 - 113 A. A. KIELMANN & L. M. CURCIO by refractometer. Information on past morbidity incidence was available for all 53 children who had been examined physically at weekly intervals for a total of 13 144 days (mean: 248 days per child). Since these were most likely to have triggered the complement system, we examined mainly the records concerning bacterial infections, fever, eye infections, and purulent skin infections. Fever was chosen as a general indicator of the occurrence of past infectious disease, purulent skin infections and eye infections because they are readily identifiable conditions usually (purulent skin infection) or frequently (eye infection) caused by bacteria. RESULTS Complete anthropometric data (weight, height, arm circumference, and skinfold thickness) and data on past morbidity (fever, eye infections, skin infections) were available for 53, haemoglobin levels Table 1. Mean age, complement (C3), nutritional status, and past morbidity incidence of study population Characteristic n Mean S.D. Range age (months) complement (C3) a [mg/litre] C3-for-age b [percent] weight-for-age [% of Harvard median] weight-for-age [% of Narangwal median] height-for-age [% of Harvard median] height-for-age [% of Narangwal median] weight-age c [percent] height-age^ [percent] weight-for-height [% of exp. Harvard weight] arm circumference [% of Wolanski Standard] skinfold thickness [mm] arm muscle circumference [cm] haemoglobin [g/litre] total serum protein [g/litre] whole blood folate [pg/litre] fever incidence d fever duration [days] fever prevalence e skin infection incidence skin infection duration [days] skin infection prevalence eye infection incidence eye infection duration [days] eye infection prevalence 53 53 53 53 53 53 53 53 53 53 53 53 53 44 43 18 53 53 53 53 53 53 53 53 53 14.2 1097 76.8 72.1 94.4 91.2 98.6 68.3 77.2 85.0 78.5 6.6 10.2 89 76 178 1.2 3.3 3.9 0.2 4.5 0.2 1.0 8.1 9.2 7.49 259.6 18.00 14.40 19.43 7.20 4.41 16.71 17.15 9.20 9.50 2.58 1.19 17.1 3.7 111.1 0.11 1.44 2.8 0.04 6.13 3.9 0.10 5.0 8.0 4.0 - 38.0 500 - 1581 33.3 - 117.0 38.0 - 98.0 51.5 - 127.4 72.0 - 103.0 97.5- 109.7 31.9- 111.5 30.7 - 107.9 55.8 - 106.3 55.0 - 94.0 3.1 - 13.5 7.6- 14.5 57 - 120 60- 79 36- 450 0- 3.6 1 - 9.0 0- 10.1 0- 1.3 1 - 24.0 0- 15.8 0- 3.3 1 - 23.5 0- 35.2 a Adjusted for batch differences by linear regression (unadjusted mean = 870 mg/litre). b C3 level of study child as percentage of C3 level of age-matched European children. c Percentage of expected age by Harvard growth standards. d Number of episodes per 100 child days of observation. e Number of days ill per 100 child days of observation. 114 COMPLEMENT (C3), NUTRITION, AND INFECTION 115 Height-for-age (% of Boston Median) 90% NORMAL STUNTED e. Mean - 81.9 Mean = 86.0 Co s.d. = 13.38 s.d. - 0 n 13 n 1 14 0 ° 1 90%_ I'0 WASTED WASTED-AND-STUNTED ~~~ ~~Mean - 78.9 Mean - 69.2 s.d. 19.06 s.d. - 18.88 n 23 n 16 Fig.1. Complement (C3)-for-age in children of differing nutritional status. or 44, and whole-blood folate for 18 of the children on whom complement (C3) was determined. The means, standard deviations, and ranges of these characteristics of the study children are shown in Table 1. The mean adjusted (for batch differences) comple- ment (C3) level was 1100 mg/litre, 25% lower than in an age-matched European reference population. Although the mean weight-for-age corresponded to second degree malnutrition according to the Gomez classification, it was only slightly below the median of the total Narangwal child population. Mean height was below the third percentile of the Boston data and at the 50th percentile of all Narang- wal children, respectively. The anthropometric indices suggest chronic, primarily energy under- nutrition resulting in stunting and wasting. Total serum protein concentration was 76.0 g/litre. Mean haemoglobin and whole blood folate levels were low at 89.0 g/litre and 178 pg/litre, respectively. The incidences of fever, eye infections, and skin infections were high at 1.2, 1.0, and 0.3 episodes per 100 child- days of observation, respectively, but did not differ from the incidence rates of these conditions for the total Narangwal child population. In a first step of the analysis, study children were grouped according to Waterlow's (16) classification of levels of nutrition into " normal ", "stunted ", " wasted ", and " wasted-and-stunted" by using 90% of the Boston median curves as cut-off points for both height-for-age and weight-for-height. According to these criteria, 13 (24.5 %) of the children were of " normal " nutritional status, 1 (1.7 %) was " stunted ", 23 (43.4%) were " wasted ", and 16 (30.2%) were "wasted-and-stunted ". The distribu- tion of these children as well as their mean com- plement (C3)-for-age values a are shown in Fig. 1. Since only one child was "stunted" according to preset criteria, little can be said about mean C3-for-age in this group. The result seems to suggest that a Complement (C,)-for-age, determined by dividing the C, level of the study children by the mean C, level of an age- matched European reference population, was used to (1) eliminate need to correct for age; and (2) establish points of reference in the absence of concrete knowledge about " adequate C, " levels. 116 A. A. KIELMANN & L. M. CURCIO Table 2. Simple correlation coefficients between complement (C3) and anthropometric and morbidity variables Statis- Complement Log of C3 C3-for-age Log of tics (C3) Lo fC 3frae C3-for-age log (weight-for-age a - 1) 3 r 0.444 0.470 0.474 0.495 n 53 53 53 53 P < 0.001 0.001 0.001 0.001 log weight-for-agea r 0.388 0.415 0.429 0.448 n 53 53 53 53 P < 0.01 0.01 0.01 0.01 log weight-for-heighta r 0.324 0.352 0.350 0.379 n 53 53 53 53 P < 0.05 0.01 0.01 0.01 log percent Wolanski arm circumference-for-age r 0.319 0.331 0.371 0.377 n 53 53 53 53 P < 0.05 0.05 0.01 0.01 log arm-muscle circumference b (cm) r 0.307 0.315 0.303 0.308 n 53 53 53 53 P < 0.05 0.05 0.05 0.05 log weight-age c r 0.153 0.164 0.267 0.262 n 53 53 53 53 P < n.s. n.s. 0.10 0.10 skinfold thickness (mm) r 0.191 0.207 0.188 0.210 n 53 53 53 53 P < n.s. n.s. n.s. n.s. log height-for-age a r 0.168 0.191 0.243 0.253 n 53 53 53 53 P < n.s. n.s. 0.10 0.10 log height-age C r 0.111 0.117 0.226 0.214 n 53 53 53 53 P< n.s. n.s. n.s. n.s. skin infection incidence (no. of episodes/100 days observed) r - 0.210 - 0.198 - 0.244 - 0.220 n 53 53 53 53 P < n.s. n.s. 0.10 n.s. log eye infection incidence r - 0.192 - 0.191 - 0.120 - 0.125 n 53 53 53 53 P < n.s. n.s. n.s. n.s. log fever incidence r 0.150 0.145 0.195 0.180 n 53 53 53 53 P < n.s. n.s. n.s. n.s. a As percentage of Harvard median. b Gurney & Jelliffe (5). c Percentage of expected age by Harvard growth medians. children in the wasted-and-stunted category have the largest (and significant, P< 0.01) deficit in com- plement-for-age. Wasted children had a slight reduc- tion in mean C-for-age but this difference was not statistically significant. In a second step, complement (C3) and C3-for-age levels were correlated with the remainder of the anthropometric or morbidity variables. Complement (C3) levels and C,-for-age yielded statistically significant positive correlations with most of these (Table 2). Scattergrams of complement (C3) or C3-for-age against nutritional status sug- gested nonlinearity of most correlations except those in which nutritional status was measured by skinfold thickness or weight-for-age.a Linearization was done through logarithmic or exponential trans- formation. Use of C3-for-age instead of C8 level, as well as the logarithm of both C3-for-age and the logarithm or the exponential of the anthropometric variables, improves the correlations. High correlation coefficients were obtained with weight-for-age (r = 0.495), weight-for-height (r = 0.379), and arm circumference-for-age (r = 0.329); arm muscle cir- cumference showed larger correlation coefficients a Percentage of expected age by Harvard weight standards. COMPLEMENT (C3), NUTRITION, AND INFECTION 117 either C3 level (r = 0.315) or C3-for-age (r = 0.308) than skinfold thickness (r = 0.207; 0.210). Skin and eye infection incidence gave negative correlations with C3 level or C3-for-age. Statistical significance was reached only between C3-for-age and skin infection. Fever showed a positive though statistically not significant correlation with C3 level and C3-for-age. None of the three biochemical indicators of nutrition status-haemoglobin, total serum protein, and whole blood folate-showed any statistically significant correlation with either C3 levels or C3-for- age. To facilitate interpretation of results, C3-for-age rather than its logarithm was used in subsequent multiple regression analyses. Multiple regression with nutritional status a (continuous) and pustular skin infection (categorical) b as independent variables and C3-for-age as a dependent variable suggested a positive effect of the nutritional status and a negative effect of skin infection (Table 3). According to this equation (model A), previous pustular skin infection resulted, on the average, in a reduction of 11 percentage points in C3-for-age. To clarify the role of incidence and duration of pustular skin infection on comple- ment, the regression of C3-for-age on nutritional status a skin infection incidence, duration, and prevalence (model B) was determined. The regression resulted in a highly significant equation explaining 37% of the variance of C3-for-age a Expressed as the complement of the ratio: weight/ Harvard median weight at given age raised to the third power. b x = 1 if past pustular skin infection was present; zero if not. Table 3. Partial values for variable in multiple regression model A a (complement (C3)-for-age as dependent variable) Variable b b Standard Partial F perror (b) r2 (1.51) p nutritional status 185.2 42.7 0.22 18.8 < 0.005 skin infection - 11.0 4.2 0.09 6.8 < 0.025 (constant) 89.3 a Multiple R 2 = 0.317; SY.x = 15.2; F(2,50) = 11.61; P < 0.005for the total model. b Nutrition status = (weight-for-age- 1) 3; skin infection variable value is I if it was present in the past; zero if it was absent. (Table 4). According to the model, an increase of 0.1 units of skin infection incidence was associated with a fall of approximately 3.6 units of C3-for-age, keeping nutritional status and skin infection duration constant. An increase in 1 unit of skin infection duration changed C3-for-age by 1.2 units. The combined effects of duration of skin infection and incidence on C3-for-age were subadditive. Their coefficients were both negative while that of their product (skin infection prevalence) was positive. The interaction term (prevalence = incidence x duration) thus " balanced off" the main effects. In Fig. 2, actual values of C3-for-age have been plotted, together with the curves derived from the exponen- tial equation over the examined range. The curve suggests a high sensitivity of C3 levels to nutritional status at low levels of nutrition (< 60% of the Harvard weight median) after which the complement levels formed a plateau at approximately 70-80% of the Harvard weight median. Purulent skin infec- tion of low (x-1 standard deviation) incidence and Table 4. Partial values for variables in multiple regression model B a (complement (C3)-for-age as dependent variable) Variable b b Standard Partial F(1,51 P error (b) r (,52 nutritional status 193.1 42.3 0.224 20.8 < 0.005 skin infection incidence - 36.2 14.7 0.060 6.1 < 0.025 skin infection duration - 1.2 0.5 0.066 5.0 < 0.05 skin infection prevalence 3.5 1.5 0.023 5.6 < 0.05 (constant) 90.1 a Multiple R 2 = 0.372; Sy.x = 14.9; F(4,48) = 7.1 ; P < 0.005 for the total model.b Nutrition status = (weight-for-age- 1) 3; skin infection incidence = no. of episodes/100 days of observation; skin infection duration = mean no. of days ill/episodes; skin infection prevalence = incidence x prevalence. 118 A. A. KIELMANN & L. M. CURCIO 120 110 . 100 * - 0* 0 90 (* 0 zero incidence o \ zero duration Y\f _ low incidence - \low duration } 80* / /{,,_ OO c("galduration o: * / / o 7_.7(lowinc.; high dur. I,, t 9zJ ~thigh inc.; high dur. j'70 01/,. H // ~~~~~~~~~~~~~~~~0 60 _,? 0~~~~~~~~l/i 0 0~~~~~~~~~ 50~~~~~~~ I °,40 i 30 10 20 30 40 50 60 70 80 90 100 Percentage of Harvard Weight-for-Age Standard Fig. 2. The effects of nutritional status and past purulent skin infection on complement (C3)-for-age. low duration (1 day) a had the least depressing effect on C3-ratio; high incidence and high duration (x + 1 standard deviation) or low incidence and high duration appeared to result in a maximum decrease of C3-for-age, while high incidence and low duration occupied an intermediate position (Fig. 2). Examination of the residual plot shows a symmetrical distribution of residuals around zero suggesting homoscedasticity and linearity of the (linearized) a First value greater than zero when ranked in order of magnitude. function over the examined range. Suggested effects of purulent skin infection incidence and duration at both high and low levels of nutrition on complement levels (Table 4) were further examined in Fig. 3. According to the two models, a child of high nutritional status (90% Harvard weight median) and maximum purulent skin infection incidence and duration still has a level of complement (C3) of more than 70% of his/her European counterpart. A child of poor nutritional status (50% Harvard weight median) at maximal skin infection incidence and duration has only about 50% of the C3 level of COMPLEMENT (C3), NUTRITION, AND INFECTION HIGH WEIGHT-FOR-AGE STATUS LOW WEIGHT-FOR-AGE STATUS Purulent Skin Infection Duration (days) 119 T2 4T 6 8 10 12 i4 1 18 minimum mean (s) mean (i) + 2 s.d. Fig. 3. The effects of purulent skin infection incidence and duration on complement (C3)-for-age at high (90 %) and low (50 %) weight-for-age status. w 0 o aa E; z 8 86 85 84 83 82 81 80 79 78 77 76 75 74 61 60 59 58 57 56 55 54 53 52 51 50 49 48 y * * Sf~1 120 A. A. KIELMANN & L. M. CURCIO European children of his/her age; one with low incidence but high duration has less than 50%. DISCUSSION Deleterious effects of undernutrition on comple- ment (C3) levels were demonstrated some time ago. The results of this investigation may possibly help in understanding the nature of the interrelationship between nutritional status, C3, and infectious disease. From our results it seems that maximum C3 levels were reached at approximately 70-80% of the Harvard weight median, corresponding to first degree malnutrition on the Gomez scale. Below that level, serum C3 concentration was extremely sensitive to nutritional fluctuations. Thus in the interval between 40 and 50% of the Harvard weight median, the mean C3 concentration rose by approximately 60%, whereas between 70 and 80% of the Harvard median it rose by less than 10%. While we presume that the specific patterns that we observed between nutrition and C3 levels apply mainly to children living in the Punjabi ecosystem, the shape of the curve is probably similar for most child communities, with the pre- valence of specific infectious diseases and the nature of undernutrition determining the nutritional threshold at which maximum complement (C3) levels are maintained. Use of Waterlow's classifica- tion of nutritional status further helped to identify wasted-and-stunted children as the ones with mini- mum C3 levels and, therefore, most vulnerable to infections. The commonly observed excess mortality in this group is thus not surprising. In our children, complement (C3) was determined more by the immediate level of nutrition as measured by weight-for-age or weight-for-height than by measures of long-term nutrition such as height. Surprisingly arm-muscle circumference, an accepted indicator of lean body muscle mass, showed a lower correlation with C3 than weight-for-age. This is probably because Punjabi children get relatively high amounts of protein in their dieta and, even a On the average, 8 % of energy was found to come from high-quality protein. though total food intake may be low, specific protein deficiency was not the limiting factor in C3 production. A history of purulent skin infections was associated with a significant reduction in C3 concentration. From our data it seems that both frequency and duration of skin infection episodes determined the extent of the reduction. Several mechanisms may explain these results: (a) complement levels do not return to normal even in disease-free intervals in children with frequent skin infections, suggesting that the time required for C3 repletion is longer than the mean duration of the disease-free interval; (b) even though they showed no signs of overt infection, the majority of children with a high prevalence of past skin infection were suffering from subclinical infections at the time of the survey, suggesting remission and exacerbation of a chronic condition rather than new infection. Unfortunately, our sample size was too small to examine interaction patterns between levels of nutrition and skin infections and their combined effects on complement (C3) levels. Thus we had only very few children with a high level of both nutrition and skin infection prevalence and even fewer with a low level of both nutrition and disease prevalence. While it seems fairly clear from the results that the underweight child (< 70% of the Harvard weight median) with a high prevalence of skin infection seems to have minimal C3 reserves, the effects of infection on C3 synthesis at high levels of nutrition remain unclear. Complex statistical manipulations of relatively few samples is a less reliable and less satisfying method of identifying possible variable interactions than examination of contingency tables with a large enough sample in each cell to permit simple Chi- square analysis or analysis of variance. Under field conditions, however, blood samples are not easy to come by and statistical approximations help identify trends. Because of the limited number of samples in our study, our conclusions, therefore, can be but tentative and require verification by means of a more extensive follow-up study. ACKNOWLEDGEMENTS The authors wish to acknowledge the valuable assistance of Dr 1. S. Uberoi, Dr R. K. Chandry, Dr Vijay L. Mehra, and Mrs Martha S. List. COMPLEMENT (C3), NUTRITION, AND INFECTION 121 RtSUMt COMPLEMENT (C3), NUTRITION ET INFECTION Les niveaux de complement (C3) ont ete determines pour un groupe de 53 enfants ag6s de 4 a 38 mois et appartenant A la communaute villageoise de Narangwal (Pendjab, Inde), ou etait alors installe le Centre de Recherches en Sante rurale du Departement de la Sante internationale de 1'Universit6 Johns Hopkins. A 1'6poque de 1'examen, tous les enfants pr6sentaient un developpe- ment moteur normal et 6taient exempts d'infection. Les niveaux de compl6ment (C3) se sont reveles, A ladite 6poque, en corr6lation positive - mais non lin6aire - avec la position par rapport A la norme de poids en fonc- tion de 1 'age. Dans les cas de niveau nutritionnel mediocre, A un faible accroissement du pourcentage de la norme poids/age correspondait une forte elevation de C3; au- dessus de 70-80% de la norme, le niveau de C3 atteignait un plateau se situant a environ 90% de ce niveau dans une population de r6ference europeenne d'age corres- pondant. Les niveaux de compl6ment etaient particulie- rement bas chez les enfants presentant a la fois une insuffisance de poids (maigreur) et de taille (nanisme). Les 6pisodes ant6rieurs d'infection cutan&e purulente exereaient sur le niveau de C3 des effets negatifs pro- portionnels a leur frequence et a leur dur6e. En revanche, on n'a pas constate de relation entre ce niveau et la concentration d'h6moglobine, les prot6ines seriques totales ou la concentration d'acide folique dans le sang. REFERENCES 1. ANDERSON, R. 1. & BUCK, A. A. Zeitschrift fur Tropenmedizin und Parasitologie, 24: 447-456 (1973). 2. CHANDRA, R. K. Journal ofpediatrics, 81: 1194-2000 (1972). 3. COOVADIA, H. ET AL. American journal of clinical nutrition, 27: 665-669 (1974). 4. ECKER, E. E. ET AL. Journal of clinical investigation, 25: 800-808 (1946). 5. GURNEY, J. M. & JELLIFFE, D. B. American journal of clinical nutrition, 26: 912-915 (1973). 6. JELLIFFE, D. B. The assessment of the nutritional status of the community. Geneva, World Health Organization, 1966 (Monograph Series, No. 53). 7. KIELMANN, A. A. ET AL. Bulletin of the World Health Organization, 54: 477-483 (1976). 8. KIELMANN, A. A. ET AL. American journal of clinical nutrition (in press). 9. KOHLER, P. F. & TEN BENSEL, R. Clinical and experimental immunology, 4: 191-202 (1969). 10. MANCINI, G. ET AL. Immunochemistry, 2: 235-254 (1965). 11. MCFARLANE, H. ET AL. British medical journal, 4: 268-270 (1970). 12. NEUMANN, C. ET AL. American journal of clinical nutrition, 28: 89-104 (1975). 13. SIRISINHA, S. ET AL. Lancet, 1: 1016-1020 (1973). 14. SMYTHE, P. M. ET AL. Lancet, 2: 939-943 (1971). 15. VARLEY, M. Practical chemical biochemistry, 4th ed. London, White Friars Press, Ltd, 1976. 16. WATERLOW, J. C. British medical journal, 4: 88-90 (1974).