912
December 1975 The Journal of P E D I A T R I C S
Serum complement profiles in infants and children The purpose of this study was to examine normal serum values for the complement components, C3, C4, and C5, and total hemolytic complement (CH50) activity in 163 healthy infants and children, in the age range from birth through 14 years. There were statistically significant relationships of Ca, C,, and C5, but not CH50 with age. None of the complement components or CHSO could be differentiated by sex or race. Tolerance limits for high and low values were projected for each complement measurement for 75%, 90%, and 95% of the general population. Our data confirm the differences in complement levels between children and adults and demonstrate a wide range of values within each group, reflecting the biologic variability of complement measurements. These results emphasize the importance of establishing normal pediatric values in any laboratory that measures complement profiles in various diseases of childhood
M. E. Norman, M.D.,* E. P. Gall, M.D., A. Taylor, L. Laster, Ph.D., and U. R. Nilsson, M . D . , Philadelphia, Pa.
M EA SU RE ME N Wof total hemolytic complement activity (CH50) in human serum remains the most commonly employed screening method for testing the integrity of the complement sequence. A variety of techniques 1, ~ have been employed in determining CH50 in normal adults, 3-r adults and children with glomerulonephritis . . . . . . . . and collagen vascular disease, s-~, 10. 11 and normal newborn infants?~, 13 Complement profiles, including measurements of individual components by hemolytic assay and immunochemical quantitation, have been reported in From the Clinical Immunology Laboratory, The Children's Hospital of Philadelphia, Departments of Pediatrics and Medicine, University of Pennsylvania School af Medicine, and Division of Biometrics, University of Pennsylvania School of Dental Medicine. Supported by the Clinical Immunology Laboratory, The Children's Hospital of Philadelphia; PUblic Health Service Training Grant A I 00154; Veterans" Administration Hospital, Philadelphia; Barsumian Memorial Fund; John A. Hartford Foundation; P H S Grant 5 R01 A M 13, 515; and RCDA 5 K 0 4 A I 50 198. A portion of this work was submitted in abstract form to the 1974 meeting of the Society for Pediatric Research. *Reprint address: Room 806.5, The Children's Hospital of Philadelphia, 34th St. and Ovic Center Blvd., Philadelphia, Pa. 19104.
Vol. 87, No. 6, part 1, pp. 912-916
adults,3, lo. 1, but there is insufficient data in infants and children to establish a range of normal values. This study investigates the range of normal serum values for the third, fourth, and fifth components of complement (C3, C,, and CO and CH50 in 163 healthy infants and children and examines their relationship with age, sex, and race.
MATERIALS A N D M E T H O D S Normal children. Specimens o f cord blood and peripheral blood were obtained from 163 healthy infants and children, from birth through age 14 years, during the course of routine medical evaluations. There were 88 boys and 75 girls; 48 boys and 42 girls were black. Age groupings were arbitrarily defined and are listed in Table I. Children with fever, signs or symptom s of infection, or chronic illness were excluded from this study. A l l sera were collected under standardized conditions, and all studies were performed within two weeks of collection, after storage at - 7 0 ~ C. Informed consent was obtained from the parents in all cases, and in age groups six and seven, from the children themselves. CHS0 assay. The method of Mayer was employed for the CH50 titration and calculation of the CH50 titer, 1 with the following modification: the reaction mixture was incubated for 30 minutes at 37 ~ C and then immediately chilled to 4 ~ C prior to centrifugation and optical density readings. The results were highly reproducible.
Volume 87 Number 6, part 1
Serum complement profiles
240
-
2O0
160
~
9 13
9
o, 9
it O0
--~@
.
120
II
80
0
9
40
Cord Blood
Doys
I-5 Mos.
6-23 2-5 M o s . Yrs.
Yrs.
10-14 Yrs.
Age Groups Fig. 1. Scattergram plot of individual CH50 units for the entire study population. 0 = white females, A = white males, 9 = black females, 9 = black males. Mean values are indicated by the horizontal bars.
C3, C4, and C~ assays. Immunochemical quantitations of Ca, C4, and C5 were performed using the modified single radial immunodiffusion technique described by Mancini and colleagues. ~ C3 determinations were performed in immunoplates purchased commercially from Hyland, Inc. (Division of Travenol Laboratories, Inc., Costa Mesa, Calif.). For C4 and C5 determinations immunoplates were prepared incorporating monospecific rabbit antihuman C4 prepared according to the method of Muller-Eberhard and Biro, 1~ and monospecific goat antihuman C5 prepared according to the method of Nilsson and Muller-Eberhard. 17 Four per cent epsilon amino caproic acid was added to the agar for measurement of C~. A reference serum pool was prepared from 22 normal adult volunteers and was used throughout the study. The levels of C3, C4, and C~ in the pool had previously been determined by radial immunodiffusion in the laboratory of Dr. Hans Muller-Eberhard, using purified complement components for the standards. In each of our experiments a standard curve was constructed for each component determined from serially diluted aliquots of the serum pool. All experiments were performed in duplicate. Evaluation of techniques. The effects of day-to-day variability, 16 weeks of storage at - 7 0 ~ reproducibility, and repeated freezing and thawing were tested to determine their influence on the complement measurements. Statistical methods. Regression analyses of individual complement components and CH50 on age, sex, and race
Table I. Distribution of subjects by age groups
Age group
Range
No.
1 2 3 4 5 6 7
Cord blood 2-7 days 1-5 mo 6 mo-2 yr 2-5 yr 6-9 yr 10-14 yr
24 20 16 25 27 28 23
Table H. Tolerance intervals for CH50
I Lowvalue P* (N) 0.75 P (N) 0.90 P (N) 0.95
102t 84 73
Mean 143 143 143
] Highvalue 184 202 213
*P (N) = percentage of the generalpopulation. t102 = U/ml. were performed according to published techniques. TM In addition tolerance intervals were projected (i.e., high and low values) for each complement measurement for 75%, 90%, and 95% of the general population within 95% confidence limits. 19 When there were significant relationships of complement levels with age, these tolerance intervals were projected simultaneously for all age groups. 2~
9 14
Norman et al.
The Journal of Pediatrics December 1975
Table HI. Group mean values and tolerance intervals for C~, C,, and C~ C~*
groupge A
Mean
C~*
P (N) 0.75
P(N)~75
P((N)~90
P(N)~95
Low I High
Low [ High
Low I High
Mean
Low I High
P (N) 0.90
P(N) 0.95
Low
High
Low
High
18 70 116 156 184 201 215
1,054 1,066 1,082 1,106 1,142 1,187 1,237
0 0 36 76 104 121 135
1,134 ~ 1,146 1,162 1,186 1,222 1,267 1,317
I
1 2 3 4 5 6 7
1,089 1,139 1,188 1,238 1,287 1,336 1,386
719 788 851 908 954 989 1,022
1,459 1,490 1,525 1,568 1,620 1,683 1,750
596 665 728 785 831 867 899
1,582 1,613 1,648 1,690 1,743 1,805 1,873
520 589 652 709 755 790 823
1,658 1,689 1,724 1,767 1,819 1,882 1,949
536 568 599 631 663 694 726
147 199 245 285 313 330 343
925 937 953 977 1,013 1,058 1,109
*Results expressedin/~g/ml. tSee Table I. RESULTS Study population. In Table I the patients are divided by age groups, indicating nearly equal numbers of children in each group except Group 3 (1 to 5 months). The malefemale and black-white distributions paralleled those found for the entire population except for Group 3 (4 white, 12 black children). CHS0. The regression analysis demonstrated no statistical differentiation by CH50 levels for age, sex, or race (Fig. 1). The mean values and tolerance intervals pr Ojected from the entire study population are shown in Table II. As expected, the width of the intervals increase as the percentage of the general population for which our data are projected is increased. Complement components. There were statistically significant regressions of C3, C4, and C5 on age (p < 0.01 for each component), but no differentiation by sex or race. Interestingly, the regression analysis revealed an inverse relationship of C~ with age in contrast to Ca and C4. The projected individual age group mean values and tolerance limits for each component are presented in Table III. There is a wide range of values around the mean for each component and each age group, reflecting significant biologic variability of complement levels independent of age. Evaluation of techniques. There were no consistent patterns of loss or gain of CH50 activity in any of the specimens tested, in either the day-to-day variability or storage experiments. Excellent reproducibility was noted between triplicate determinations of six serum specimens (mean difference between high and low values: + 12.5 U/ml), and freezing and thawing 0 to 2 times demonstrated no significant reduction in CH50 titers, DISCUSSION The purpose of this study was to examine the range of normal serum values for a series of Complement measure-
ments that comprise a complement profile suitable for routine clinical use in the pediatric population. We selected CH50 as the best single measurement of the overall integrity of the complement sequence, expressed in functional (e.g., hemolytic) terms. Immunochemicai quantitation of the complement component proteins, C3, C4, and C5, was chosen because it is a relatively simple technique permitting a distinction to be made between activation (consumption) of these proteins via the classic and alternate pathways. ~I Taken together, these measurements provide an effective screening procedure for both the inherited and acquired disorders of the complement system.22 An abnormality of the screening profile should be followed by analyses of individual complement components employing hemolytic titrations and other available in vitro functional assays, such as chemotaxis and phagocytosis, t o monitor the role of complement components in host defense. 2.... CH50 levels have been reported to be reduced in randomly chosen newborn sera when compared to normal adult sera 24 and in newborn' sera when paired with maternal sera. 1~...... However, Sawyer and associates ~r demonstrated that CH50 in cord sera were significantly lower than paired maternal sera only in infants of low birth weight, but that CH50 levels were significantly higher in maternal than in nonpregnant adult sera, suggesting a possible explanation of the depressed newborn-maternal ratios of CH50 previously reported. We did not examine the relationship o f CH50 levels in paired newborn-maternal sera, but studied the relationship of CH50 to age and found no correlation. The presence of normal CH50 levels in newborn infants probably reflects adequate fetal synthesis of the component(s) that are rate limiting for total hemolytic activity. Sawyer and colleagues 2~ have recently reviewed the arguments supporting accelerated complement synthesis in the late weeks of gestation.
Volume 87 Number 6, part 1
Serum complement profiles
C~*
Mean 67 65 64 62 61 59 58
P(N) G~
P(N) 0.90
P(N) G95
Low [ H@h
Low I H@h
Low I H@h
46 46 45 44 ~ 40 38
87 84 82 80 79 78 78
40 39 38 37 36 33 31
93 91 89 87 86 85 84
35 35 34 33 31 29 27
98 95 93 91 90 89 89
There have been several published reports of serum C~ complement levels in healthy infants and children, 28, 29 and Fireman and associates,12 Kohler, 18 and Sawyer and associates~' compared complement component levels in paired newborn-maternal sera. There are few published data, however, on the relationship of complement component levels to age, sex, or race throughout the pediatric age range. Coker and colleagues ~8 found no correlation of C~ levels with age in 60 normal children, but Fireman and colleagues TM and Kaufman and colleagues ~9 reported increasing serum concentrations of C~ proportionally with age, from birth to one and two years, respectively. Fireman's data demonstrated a progressive increase in group mean C~ and C5 levels from birth to 12 months; we have confirmed and extended those observations for Ca, showing a continued rise in mean Ca levels throughout the pediatric age range. However, in our study population, group mean C5 levels were inversely related to age. The reasons for this are not dear, although Kohler 2~ reported that the onset of synthesis of C~ precedes that of Ca and C~ in the human fetus. C~ achieved normal adult levels by 9 months of age in Fireman's series but continued to rise in our population. One possible explanation for the discrepancies between our own and previously published results is that we studied larger numbers of children in each age group. Although our data demonstrated a statistically significant relationship of C3, C,, and C~ (but not CH50) with age, the differences between means of the various age groups were small, in view of the wide range of variability of individual Ca, C,, and C~ levels. It is obvious that in comparing results between laboratories, the size and distribution of the study population are as important as differences in methodology in explaining conflicting data. Our data on the techniques of complement measurement reiterate that meticulous attention to the details of specimen collection, processing, and test procedures
9 15
minimized the contribution of technical factors to the variability of the results. The choice of tolerance intervals to express our data was based on the following question: Given a sample of individuals drawn from the general pediatric population, what would the range or values be if the data were projected for 75%, 90%, or 95% of all normal children? We have presented these data for C3, C,, and C5, projecting with 95% confidence limits what the high and low values would be around the estimated group means that incorporated 75%, 90%, and 95% of all normal children. For CH50, similar data were presented as a single set of values for the total study population. In differentiating normal from abnormal complement values, the physician must decide the relative importance of false positive and false negative results. This depends on how complement tests are being used in a particular laboratory. For example, data from a sample of a normal population that projects a tolerance interval for 75% of that population would tend to minimize false normal results, whereas an interval projected for 95% of that population would tend to minimize false abnormal results. The wide range and large standard deviations within each age group probably reflect inherent biologic variability of complement levels in children, for reasons which are not entirely dear. Our data on quantitative measurements of C3 and C~ appear to have a more stable relationship than C, to the projected tolerance intervals. The inherent variability in C4 resulted in a poor fit between the actual data obtained and the theoretical data projected in Table III. In our view, this limits the diagnostic usefulness of quantitative C, levels in children. The authors feel compelled to caution pediatricians to utilize measurements of CH50 and individual complement components only when they have been compared to normal pediatric values (age matched for C~, C~, and C~) established in their own laboratory. Our current practice is to employ a tolerance interval of 90% in determining abnormally elevated or reduced complement levels. In any one patient who is followed longitudinally, a change in the complement level, although remaining within the high and low values for 90% of the population, may be as significant as the measurement of a single value outside this range. The authors gratefully acknowledge Dr. Burton Zweiman, who made helpful suggestions throughout the course of the study, and the Chestnut Hill Pediatric Group (Pennsylvania), who allowed us to study their patients. REFERENCES
1. Kabat EA, and Mayer MM: Experimental immunochemistry, ed. 2, Springfield, Ill, 1961, Charles C Thomas, Publisher, chap 4. 2. Kent JA, and Fife EH: Precise standardization of reagents
9 16
N o r m a n et al.
for complement fixation, Am J Trop Med Hyg 12:103, 1963. 3. Kohler PF, and ten Bensel R: Serial complement component alterations in acute glomerulonephtitis and systemic lupus erythematosus, Clin Exp Immunol 4:191, 1969. 4. Schur PH, and Sandson J: Immunologic factors and clinical activity in systemic lupus erythematosus, N Engl J Med 278:533, 1968. 5. Petz LD, Sharp GC, Cooper NR, and Irvin WS: Serum and cerebral spinal fluid complement and serum autoantibodies in systemic lupus erythematosus, Medicine 50:259, 1971. 6. Fischel EE, and Gajdusek DC: Serum complement in acute glomernlonephtitis and other renal diseases, Am J Med 12:190, 1952. 7. Gabriel R, Glynn AR, and Joekes AM: Raised complement in nephritis: prognostic significance, Lancet 2:55, 1972. 8. West CD, McAdams JA, McConville JJ, Davis NC, and Holland NH: Hypocomplementemic and normocomplementemic persistent (chronic) glomerulonephritis; clinical and pathologic characteristics, J PEDIATR67:1089, 1965. 9. Gewurz H, Page AR, Picketing RJ, et al: Complement activity and inflammatory neutrophil exudation in man, Int Arch Allergy 32:65, 1967. 10. Lewis EJ, Carpenter CB, and Schur PH: Serum complement component levels in human glomerulonephritis, Ann Intern Med 75:555, 1971. 11. Wedgewood RJP, and Janeway CA: Serum complement in children with collagen diseases, Pediatrics 11:569, 1953. 12. Fireman P, Zuchowski DA, and Taylor PM: Development - of human complement system, J Immunol 103:25, 1969. 13. Kohler PF: Quantitative comparison of complement in the mother and newborn, Fed Proc 27:491, 1968. 14. Kohler PF, and Muller-Eberhard HJ: Immunochemical quantitation of the third, fourth, and fifth components of human complement: concentrations in the serum of healthy adults, J Immunol 99:1211, 1967. 15. Mancini G, Carbonara HO, and Heremans JF: Immunochemical quantitation of antigens by single radial immunodiffusion, Immunochemistry 2:235, 1965.
The Journal of Pediatrics December 1975
16. Muller-Eberhard HJ, and Biro CE: Isolation and description of the fourth component of human complement, J Exp Med 118:447, 1963. 17. Nilsson UR, and Muller-Eberhard HJ: Isolation of fllrglobulin from human serum and its characterization as the fifth component of complement, J Exp Med 122:277, 1965. 18. Johnston J: Econometric methods, New York, 1963, McGraw-Hill Book Company, Inc. 19. Eisenhart C, Hasting MW, and Wallis WA: Techniques of statistical analysis, New York, 1947, McGraw-Hill Book Company, Inc, chap 2. 20. Miller RG: Simultaneous statistical inference, New York, 1966, McGraw-Hill Book Company, Inc. 21. Colten HR, and Rosen FS: In Stiehm ER, and Fulginiti VA, editors: Immunologic disorders in infants and children, Philadelphia, 1973, WB Saunders Company, chap 7. 22. Rosen FS, and Alper CA: In Stiehm ER, and Fulginiti VA, editors: Immunologic disorders in infants and children, Philadelphia, 1973, WB Saunders Company, chap 17. 23. Miller ME: In Stiehm ER, and Fulginiti VA, editors: Immunologic disorders in infants and children, Philadelphia, 1973, WB Saunders Company, chap 9. 24. Eward RA, Williams JH, and Bowden DH: Serum complement in the newborn, Vox Sang 6:312, 1961. 25. Fishel CW, and Pearlman DS: Complement components of paired mother-cord sera, Proc Soc Exp Biol Med 107:695, 1961. 26. Kohler PF: Maturation of the human complement system, J Clin Invest 52:671, 1973. 27. Sawyer MK, Forman ML, Kuplic LS, and Stiehm ER: Developmental aspects of the human complement system, Biol Neonate 19:148, 1971. 28. Coker SB, Nunnery AW, and Wenzl JE: Determination of file/iliA globulin in children, Southern Med J 65:61, 1972. 29. Kaufman HS, Frick OL, and Fink D: Serum complement (fl~c) in young children with atopic dermatitis, J Allergy 42:1, 1968.
December 1975 The Journal of P E D I A T R I C S
Serum complement profiles in infants and children The purpose of this study was to examine normal serum values for the complement components, C3, C4, and C5, and total hemolytic complement (CH50) activity in 163 healthy infants and children, in the age range from birth through 14 years. There were statistically significant relationships of Ca, C,, and C5, but not CH50 with age. None of the complement components or CHSO could be differentiated by sex or race. Tolerance limits for high and low values were projected for each complement measurement for 75%, 90%, and 95% of the general population. Our data confirm the differences in complement levels between children and adults and demonstrate a wide range of values within each group, reflecting the biologic variability of complement measurements. These results emphasize the importance of establishing normal pediatric values in any laboratory that measures complement profiles in various diseases of childhood
M. E. Norman, M.D.,* E. P. Gall, M.D., A. Taylor, L. Laster, Ph.D., and U. R. Nilsson, M . D . , Philadelphia, Pa.
M EA SU RE ME N Wof total hemolytic complement activity (CH50) in human serum remains the most commonly employed screening method for testing the integrity of the complement sequence. A variety of techniques 1, ~ have been employed in determining CH50 in normal adults, 3-r adults and children with glomerulonephritis . . . . . . . . and collagen vascular disease, s-~, 10. 11 and normal newborn infants?~, 13 Complement profiles, including measurements of individual components by hemolytic assay and immunochemical quantitation, have been reported in From the Clinical Immunology Laboratory, The Children's Hospital of Philadelphia, Departments of Pediatrics and Medicine, University of Pennsylvania School af Medicine, and Division of Biometrics, University of Pennsylvania School of Dental Medicine. Supported by the Clinical Immunology Laboratory, The Children's Hospital of Philadelphia; PUblic Health Service Training Grant A I 00154; Veterans" Administration Hospital, Philadelphia; Barsumian Memorial Fund; John A. Hartford Foundation; P H S Grant 5 R01 A M 13, 515; and RCDA 5 K 0 4 A I 50 198. A portion of this work was submitted in abstract form to the 1974 meeting of the Society for Pediatric Research. *Reprint address: Room 806.5, The Children's Hospital of Philadelphia, 34th St. and Ovic Center Blvd., Philadelphia, Pa. 19104.
Vol. 87, No. 6, part 1, pp. 912-916
adults,3, lo. 1, but there is insufficient data in infants and children to establish a range of normal values. This study investigates the range of normal serum values for the third, fourth, and fifth components of complement (C3, C,, and CO and CH50 in 163 healthy infants and children and examines their relationship with age, sex, and race.
MATERIALS A N D M E T H O D S Normal children. Specimens o f cord blood and peripheral blood were obtained from 163 healthy infants and children, from birth through age 14 years, during the course of routine medical evaluations. There were 88 boys and 75 girls; 48 boys and 42 girls were black. Age groupings were arbitrarily defined and are listed in Table I. Children with fever, signs or symptom s of infection, or chronic illness were excluded from this study. A l l sera were collected under standardized conditions, and all studies were performed within two weeks of collection, after storage at - 7 0 ~ C. Informed consent was obtained from the parents in all cases, and in age groups six and seven, from the children themselves. CHS0 assay. The method of Mayer was employed for the CH50 titration and calculation of the CH50 titer, 1 with the following modification: the reaction mixture was incubated for 30 minutes at 37 ~ C and then immediately chilled to 4 ~ C prior to centrifugation and optical density readings. The results were highly reproducible.
Volume 87 Number 6, part 1
Serum complement profiles
240
-
2O0
160
~
9 13
9
o, 9
it O0
--~@
.
120
II
80
0
9
40
Cord Blood
Doys
I-5 Mos.
6-23 2-5 M o s . Yrs.
Yrs.
10-14 Yrs.
Age Groups Fig. 1. Scattergram plot of individual CH50 units for the entire study population. 0 = white females, A = white males, 9 = black females, 9 = black males. Mean values are indicated by the horizontal bars.
C3, C4, and C~ assays. Immunochemical quantitations of Ca, C4, and C5 were performed using the modified single radial immunodiffusion technique described by Mancini and colleagues. ~ C3 determinations were performed in immunoplates purchased commercially from Hyland, Inc. (Division of Travenol Laboratories, Inc., Costa Mesa, Calif.). For C4 and C5 determinations immunoplates were prepared incorporating monospecific rabbit antihuman C4 prepared according to the method of Muller-Eberhard and Biro, 1~ and monospecific goat antihuman C5 prepared according to the method of Nilsson and Muller-Eberhard. 17 Four per cent epsilon amino caproic acid was added to the agar for measurement of C~. A reference serum pool was prepared from 22 normal adult volunteers and was used throughout the study. The levels of C3, C4, and C~ in the pool had previously been determined by radial immunodiffusion in the laboratory of Dr. Hans Muller-Eberhard, using purified complement components for the standards. In each of our experiments a standard curve was constructed for each component determined from serially diluted aliquots of the serum pool. All experiments were performed in duplicate. Evaluation of techniques. The effects of day-to-day variability, 16 weeks of storage at - 7 0 ~ reproducibility, and repeated freezing and thawing were tested to determine their influence on the complement measurements. Statistical methods. Regression analyses of individual complement components and CH50 on age, sex, and race
Table I. Distribution of subjects by age groups
Age group
Range
No.
1 2 3 4 5 6 7
Cord blood 2-7 days 1-5 mo 6 mo-2 yr 2-5 yr 6-9 yr 10-14 yr
24 20 16 25 27 28 23
Table H. Tolerance intervals for CH50
I Lowvalue P* (N) 0.75 P (N) 0.90 P (N) 0.95
102t 84 73
Mean 143 143 143
] Highvalue 184 202 213
*P (N) = percentage of the generalpopulation. t102 = U/ml. were performed according to published techniques. TM In addition tolerance intervals were projected (i.e., high and low values) for each complement measurement for 75%, 90%, and 95% of the general population within 95% confidence limits. 19 When there were significant relationships of complement levels with age, these tolerance intervals were projected simultaneously for all age groups. 2~
9 14
Norman et al.
The Journal of Pediatrics December 1975
Table HI. Group mean values and tolerance intervals for C~, C,, and C~ C~*
groupge A
Mean
C~*
P (N) 0.75
P(N)~75
P((N)~90
P(N)~95
Low I High
Low [ High
Low I High
Mean
Low I High
P (N) 0.90
P(N) 0.95
Low
High
Low
High
18 70 116 156 184 201 215
1,054 1,066 1,082 1,106 1,142 1,187 1,237
0 0 36 76 104 121 135
1,134 ~ 1,146 1,162 1,186 1,222 1,267 1,317
I
1 2 3 4 5 6 7
1,089 1,139 1,188 1,238 1,287 1,336 1,386
719 788 851 908 954 989 1,022
1,459 1,490 1,525 1,568 1,620 1,683 1,750
596 665 728 785 831 867 899
1,582 1,613 1,648 1,690 1,743 1,805 1,873
520 589 652 709 755 790 823
1,658 1,689 1,724 1,767 1,819 1,882 1,949
536 568 599 631 663 694 726
147 199 245 285 313 330 343
925 937 953 977 1,013 1,058 1,109
*Results expressedin/~g/ml. tSee Table I. RESULTS Study population. In Table I the patients are divided by age groups, indicating nearly equal numbers of children in each group except Group 3 (1 to 5 months). The malefemale and black-white distributions paralleled those found for the entire population except for Group 3 (4 white, 12 black children). CHS0. The regression analysis demonstrated no statistical differentiation by CH50 levels for age, sex, or race (Fig. 1). The mean values and tolerance intervals pr Ojected from the entire study population are shown in Table II. As expected, the width of the intervals increase as the percentage of the general population for which our data are projected is increased. Complement components. There were statistically significant regressions of C3, C4, and C5 on age (p < 0.01 for each component), but no differentiation by sex or race. Interestingly, the regression analysis revealed an inverse relationship of C~ with age in contrast to Ca and C4. The projected individual age group mean values and tolerance limits for each component are presented in Table III. There is a wide range of values around the mean for each component and each age group, reflecting significant biologic variability of complement levels independent of age. Evaluation of techniques. There were no consistent patterns of loss or gain of CH50 activity in any of the specimens tested, in either the day-to-day variability or storage experiments. Excellent reproducibility was noted between triplicate determinations of six serum specimens (mean difference between high and low values: + 12.5 U/ml), and freezing and thawing 0 to 2 times demonstrated no significant reduction in CH50 titers, DISCUSSION The purpose of this study was to examine the range of normal serum values for a series of Complement measure-
ments that comprise a complement profile suitable for routine clinical use in the pediatric population. We selected CH50 as the best single measurement of the overall integrity of the complement sequence, expressed in functional (e.g., hemolytic) terms. Immunochemicai quantitation of the complement component proteins, C3, C4, and C5, was chosen because it is a relatively simple technique permitting a distinction to be made between activation (consumption) of these proteins via the classic and alternate pathways. ~I Taken together, these measurements provide an effective screening procedure for both the inherited and acquired disorders of the complement system.22 An abnormality of the screening profile should be followed by analyses of individual complement components employing hemolytic titrations and other available in vitro functional assays, such as chemotaxis and phagocytosis, t o monitor the role of complement components in host defense. 2.... CH50 levels have been reported to be reduced in randomly chosen newborn sera when compared to normal adult sera 24 and in newborn' sera when paired with maternal sera. 1~...... However, Sawyer and associates ~r demonstrated that CH50 in cord sera were significantly lower than paired maternal sera only in infants of low birth weight, but that CH50 levels were significantly higher in maternal than in nonpregnant adult sera, suggesting a possible explanation of the depressed newborn-maternal ratios of CH50 previously reported. We did not examine the relationship o f CH50 levels in paired newborn-maternal sera, but studied the relationship of CH50 to age and found no correlation. The presence of normal CH50 levels in newborn infants probably reflects adequate fetal synthesis of the component(s) that are rate limiting for total hemolytic activity. Sawyer and colleagues 2~ have recently reviewed the arguments supporting accelerated complement synthesis in the late weeks of gestation.
Volume 87 Number 6, part 1
Serum complement profiles
C~*
Mean 67 65 64 62 61 59 58
P(N) G~
P(N) 0.90
P(N) G95
Low [ H@h
Low I H@h
Low I H@h
46 46 45 44 ~ 40 38
87 84 82 80 79 78 78
40 39 38 37 36 33 31
93 91 89 87 86 85 84
35 35 34 33 31 29 27
98 95 93 91 90 89 89
There have been several published reports of serum C~ complement levels in healthy infants and children, 28, 29 and Fireman and associates,12 Kohler, 18 and Sawyer and associates~' compared complement component levels in paired newborn-maternal sera. There are few published data, however, on the relationship of complement component levels to age, sex, or race throughout the pediatric age range. Coker and colleagues ~8 found no correlation of C~ levels with age in 60 normal children, but Fireman and colleagues TM and Kaufman and colleagues ~9 reported increasing serum concentrations of C~ proportionally with age, from birth to one and two years, respectively. Fireman's data demonstrated a progressive increase in group mean C~ and C5 levels from birth to 12 months; we have confirmed and extended those observations for Ca, showing a continued rise in mean Ca levels throughout the pediatric age range. However, in our study population, group mean C5 levels were inversely related to age. The reasons for this are not dear, although Kohler 2~ reported that the onset of synthesis of C~ precedes that of Ca and C~ in the human fetus. C~ achieved normal adult levels by 9 months of age in Fireman's series but continued to rise in our population. One possible explanation for the discrepancies between our own and previously published results is that we studied larger numbers of children in each age group. Although our data demonstrated a statistically significant relationship of C3, C,, and C~ (but not CH50) with age, the differences between means of the various age groups were small, in view of the wide range of variability of individual Ca, C,, and C~ levels. It is obvious that in comparing results between laboratories, the size and distribution of the study population are as important as differences in methodology in explaining conflicting data. Our data on the techniques of complement measurement reiterate that meticulous attention to the details of specimen collection, processing, and test procedures
9 15
minimized the contribution of technical factors to the variability of the results. The choice of tolerance intervals to express our data was based on the following question: Given a sample of individuals drawn from the general pediatric population, what would the range or values be if the data were projected for 75%, 90%, or 95% of all normal children? We have presented these data for C3, C,, and C5, projecting with 95% confidence limits what the high and low values would be around the estimated group means that incorporated 75%, 90%, and 95% of all normal children. For CH50, similar data were presented as a single set of values for the total study population. In differentiating normal from abnormal complement values, the physician must decide the relative importance of false positive and false negative results. This depends on how complement tests are being used in a particular laboratory. For example, data from a sample of a normal population that projects a tolerance interval for 75% of that population would tend to minimize false normal results, whereas an interval projected for 95% of that population would tend to minimize false abnormal results. The wide range and large standard deviations within each age group probably reflect inherent biologic variability of complement levels in children, for reasons which are not entirely dear. Our data on quantitative measurements of C3 and C~ appear to have a more stable relationship than C, to the projected tolerance intervals. The inherent variability in C4 resulted in a poor fit between the actual data obtained and the theoretical data projected in Table III. In our view, this limits the diagnostic usefulness of quantitative C, levels in children. The authors feel compelled to caution pediatricians to utilize measurements of CH50 and individual complement components only when they have been compared to normal pediatric values (age matched for C~, C~, and C~) established in their own laboratory. Our current practice is to employ a tolerance interval of 90% in determining abnormally elevated or reduced complement levels. In any one patient who is followed longitudinally, a change in the complement level, although remaining within the high and low values for 90% of the population, may be as significant as the measurement of a single value outside this range. The authors gratefully acknowledge Dr. Burton Zweiman, who made helpful suggestions throughout the course of the study, and the Chestnut Hill Pediatric Group (Pennsylvania), who allowed us to study their patients. REFERENCES
1. Kabat EA, and Mayer MM: Experimental immunochemistry, ed. 2, Springfield, Ill, 1961, Charles C Thomas, Publisher, chap 4. 2. Kent JA, and Fife EH: Precise standardization of reagents
9 16
N o r m a n et al.
for complement fixation, Am J Trop Med Hyg 12:103, 1963. 3. Kohler PF, and ten Bensel R: Serial complement component alterations in acute glomerulonephtitis and systemic lupus erythematosus, Clin Exp Immunol 4:191, 1969. 4. Schur PH, and Sandson J: Immunologic factors and clinical activity in systemic lupus erythematosus, N Engl J Med 278:533, 1968. 5. Petz LD, Sharp GC, Cooper NR, and Irvin WS: Serum and cerebral spinal fluid complement and serum autoantibodies in systemic lupus erythematosus, Medicine 50:259, 1971. 6. Fischel EE, and Gajdusek DC: Serum complement in acute glomernlonephtitis and other renal diseases, Am J Med 12:190, 1952. 7. Gabriel R, Glynn AR, and Joekes AM: Raised complement in nephritis: prognostic significance, Lancet 2:55, 1972. 8. West CD, McAdams JA, McConville JJ, Davis NC, and Holland NH: Hypocomplementemic and normocomplementemic persistent (chronic) glomerulonephritis; clinical and pathologic characteristics, J PEDIATR67:1089, 1965. 9. Gewurz H, Page AR, Picketing RJ, et al: Complement activity and inflammatory neutrophil exudation in man, Int Arch Allergy 32:65, 1967. 10. Lewis EJ, Carpenter CB, and Schur PH: Serum complement component levels in human glomerulonephritis, Ann Intern Med 75:555, 1971. 11. Wedgewood RJP, and Janeway CA: Serum complement in children with collagen diseases, Pediatrics 11:569, 1953. 12. Fireman P, Zuchowski DA, and Taylor PM: Development - of human complement system, J Immunol 103:25, 1969. 13. Kohler PF: Quantitative comparison of complement in the mother and newborn, Fed Proc 27:491, 1968. 14. Kohler PF, and Muller-Eberhard HJ: Immunochemical quantitation of the third, fourth, and fifth components of human complement: concentrations in the serum of healthy adults, J Immunol 99:1211, 1967. 15. Mancini G, Carbonara HO, and Heremans JF: Immunochemical quantitation of antigens by single radial immunodiffusion, Immunochemistry 2:235, 1965.
The Journal of Pediatrics December 1975
16. Muller-Eberhard HJ, and Biro CE: Isolation and description of the fourth component of human complement, J Exp Med 118:447, 1963. 17. Nilsson UR, and Muller-Eberhard HJ: Isolation of fllrglobulin from human serum and its characterization as the fifth component of complement, J Exp Med 122:277, 1965. 18. Johnston J: Econometric methods, New York, 1963, McGraw-Hill Book Company, Inc. 19. Eisenhart C, Hasting MW, and Wallis WA: Techniques of statistical analysis, New York, 1947, McGraw-Hill Book Company, Inc, chap 2. 20. Miller RG: Simultaneous statistical inference, New York, 1966, McGraw-Hill Book Company, Inc. 21. Colten HR, and Rosen FS: In Stiehm ER, and Fulginiti VA, editors: Immunologic disorders in infants and children, Philadelphia, 1973, WB Saunders Company, chap 7. 22. Rosen FS, and Alper CA: In Stiehm ER, and Fulginiti VA, editors: Immunologic disorders in infants and children, Philadelphia, 1973, WB Saunders Company, chap 17. 23. Miller ME: In Stiehm ER, and Fulginiti VA, editors: Immunologic disorders in infants and children, Philadelphia, 1973, WB Saunders Company, chap 9. 24. Eward RA, Williams JH, and Bowden DH: Serum complement in the newborn, Vox Sang 6:312, 1961. 25. Fishel CW, and Pearlman DS: Complement components of paired mother-cord sera, Proc Soc Exp Biol Med 107:695, 1961. 26. Kohler PF: Maturation of the human complement system, J Clin Invest 52:671, 1973. 27. Sawyer MK, Forman ML, Kuplic LS, and Stiehm ER: Developmental aspects of the human complement system, Biol Neonate 19:148, 1971. 28. Coker SB, Nunnery AW, and Wenzl JE: Determination of file/iliA globulin in children, Southern Med J 65:61, 1972. 29. Kaufman HS, Frick OL, and Fink D: Serum complement (fl~c) in young children with atopic dermatitis, J Allergy 42:1, 1968.
