Heterogeneity of Serum Creatine Kinase Activity among Racial and Gender Groups of the Population

Heterogeneity of Serum Creatine Kinase Activity among Racial and Gender Groups of the Population EDWARD T. WONG, M.D., CAMILLA COBB, M.D., MARY K. UMEHARA, MT(ASCP), GARY A. WOLFF, M.S.E.E., L. JULIAN HAYWOOD, M.D., THEODORE GREENBERG, M.D., AND S. T. SHAW, JR., M.D. To develop reference ranges for creatine kinase (CK) appropriate for the patient population served by this hospital, levels of serum CK were measured in 1,537 individuals in our employee population. There was substantial heterogeneity in mean, median, and range of CK levels among the several race/ gender subgroups in the population studied. The race/gender subgroups could be placed into three broad groups: a high CK group, composed solely of black men; an intermediate CK group, consisting of nonblack men plus black women; and a low CK group, comprised of nonblack women. Mean CK level of the high CK group was twice that of the intermediate CK group, which, in turn, was twice that of the low CK group. Differences in mean CK values among the subgroups placed into either the intermediate CK group or the low CK group were not significant when tested with analysis of variance. Therefore, practical reference ranges for these groups are as follows: 52-520 U/L for the high CK group; 35-345 U/L for the intermediate CK group; and 25-145 U/L for the low CK group. (Key words: Creatine kinase; Reference ranges) Am J Clin Pathol 1983; 79: 582-586 MEASUREMENTS OF CREATINE KINASE (CK) levels in serum are important aids in the diagnosis of two major categories of disease: acute myocardial infarction and skeletal muscle disorders. 1019 A positive CK test result for diagnosis of these disorders often is defined simply as a CK value greater than the upper limit of the reference range provided by the laboratory. Considering their clinical importance, reference ranges for laboratory tests should be defined carefully. However, determining reference ranges seems to be a disarmingly simple task, when in fact it can be a most difficult problem. In practice, reference ranges for laboratory tests often have been determined by performing the tests on a small number of volunteers (often recruited from the laboratory staff) and computing the mean and standard deviation (SD) of the results. 7 Under the assumption that the results of the test are Gaussian in distribution, the reference limits then would be set at the mean ± 2 SD, which are assumed to be the boundaries of the central 95% of the test results among normal or healthy individuals in the population at large. Reference ranges determined in this manner have some potentially Received July 12, 1982; received revised manuscript and accepted for publication September 10, 1982. Reprints are not available. Departments of Pathology and Medicine, University of Southern California School of Medicine and Los Angeles County-University of Southern California Medical Center, Los Angeles, California serious shortcomings. First, the underlying distribution of values for the tests in the population often is not Gaussian, nor does it necessarily become Gaussian after various mathematical transformations of the data; reference limits set using the mean ± SD thus are misleading when the distribution of values is non-gaussian. A second and perhaps more serious shortcoming is that the population may not be homogeneous with respect to the laboratory test, so that reference limits may not be applicable to members of groups that were represented either inadequately or not at all among the volunteers recruited for determining the reference range. Defining an appropriate reference range for CK has been a particularly difficult problem. The distribution of CK values in populations of presumably healthy people is skewed and continues to be non-gaussian, even after log transformation of the data. Substantial heterogeneity of CK values has been demonstrated when gender and race are taken into account. 6 ' 8 '" 1418,23 In addition, factors, such as muscle mass and body size, have been described as determinants of CK values in serum. 4,24 Being an urban hospital located near the geographic center of a large metropolitan area, both our patient and employee populations contain diverse racial and ethnic groups. To develop appropriate reference ranges for certain biochemical tests for our patient population, we determined results for these tests on a large sample of our employee population. We found major differences in mean CK level among the race/gender subgroups. This study indicates the complexity of providing reference ranges for CK. Materials and Methods During the period of January 1977 through November 1980, blood samples were drawn from newly hired employees of this medical center with their informed 0002-9173/83/0500/0582 $01.05 American Society of Clinical Pathologists 582

Vol. 79 No. 5 HETEROGENEITY OF CREATINE KINASE 583 consent. The employees were seen at the employee health service center at the time of their pre-employment physical examinations; they were screened by the nursing staff on the basis of their responses to health history questionaires and the results of their physical examinations to eliminate those with significant illnesses. Volunteers for this study were asked about their participation in strenuous physical activities, such as running or weightlifting. For purposes of determining reference ranges for CK, values for subjects who admitted to such activities were excluded from the CK data base, because physical exertion is known to produce substantial elevations of serum CK levels. 5-615 ' 24 Any intramuscular or subcutaneous immunization procedures were performed after the blood samples were taken. While these precautions were taken to eliminate values from subjects with extraneous causes of elevated CK, the possibility of having included a few CK values elevated because of strenuous physical exertion, trauma, and other causes, could not be excluded entirely. Blood samples were sent to the clinical chemistry laboratory of the medical center, where they were processed and analyzed in the same manner as for patient samples submitted to the laboratory. A battery of 19 tests, including CK, was performed using two Technicon SMA 12/60 analyzers with separate but slightly overlapping test configurations (until September 1977) or the Technicon SMAC analyzer (after September 1977). CK analysis was performed using Technicon reagents by a modification of the method of Siegel and Cohen. 21 In this method, mercaptoethanol is added to the substrate to maximize the CK activity, and the analysis is performed at 37 C. Results The distributions of CK and of log CK values found in the overall population of healthy employees in the study are shown in Figures 1 and 2. Neither the CK nor the log CK values conform to a Gaussian distribution. Based on parametric statistics for CK values on the overall population (mean CK ± SD = 111.8 U/L ±215.9 U/L), the reference range could be set at 0 to 543 U/L, representing the mean ± 2 SD. However, these limits were excessively wide, including 99.3% of the overall population, because the SD was increased by the skewness of the distribution as well as by some outliers with very high CK levels. Use of parameters of the distribution of log CK produced a more narrow reference range, 24-301 U/L (mean ± 2 SD), values which corresponded to the 1.3 and 96.3 percentiles of the population, respectively, and thus were not the boundaries of the central 95%. Since the distribution of log CK was non-gaussian because of significant kurtosis and skewness (P < 0.001 for both kurtosis and skewness), a non- 30r 20 10 m T h - T r - r ^ 100 200 300 400 500 Creatine Kinase (CK), U/L FIG. I. The frequency distribution of creatine kinase values on the overall population of healthy employees plotted on an arithmetic scale of CK values. Because of the extreme skewness of the distribution, the abscissa has been truncated at 500 U/L; thus, 12 individuals with CK levels ranging from 502 to 6755 U/L have been excluded. parametric approach based on percentiles of the distribution would seem to be a preferable means of establishing a reference range. Using the 2.5 and 97.5 percentile values for CK in the overall population, the reference range for CK would be 28-358 U/L. However, even the latter limits would be inappropriate, since they ignore the differences that existed among the race/gender subgroups. Characteristics of the race/gender subgroups within the overall population of healthy employees are shown in Table 1. The values for the mean, median, SD, and range of CK values seen in each subgroup are shown in 40 30-20 10 3-10 20 50 100 200 500 1000 2000 4000 Creatine Kinase (CK), U/L FIG. 2. The frequency distribution of creatine kinase values on the overall population of healthy employees plotted on a logarithmic scale of CK values.

5g4 WONG ET AL. A.J.C.P. May 1983 Table 1. Characteristics of the Race/Gender Subgroups in the Population of Healthy Employees Studied Age (yr) Weight (lb) Height (in) Subgroups Number Men Black White Hispanic Asian Other Women Black White Hispanic Asian Other 196 164 118 70 11 317 283 219 143 16 28.9 ± 8.1 34.1 ± 10.1 33.9 ± 12.0 31.6 ± 8.0 32.6 ± 8.4 31.4 ± 9.9 29.9 ± 8.9 32.2 ± 10.9 31.4 ± 7.6 31.3 ± 6.2 18-57 20-64 17-63 22-56 23-48 17-61 17-60 18-61 20-54 21-46 162.9 ± 27.4 170.1 ± 25.0 159.1 ± 24.2 147.3 ± 19.9 160.8 ± 23.2 146.9 ± 27.5 136.3 ±25.8 134.2 ±26.6 114.9 ± 16.3 142.3 ± 24.8 128-242 117-240 101-219 106-205 131-213 90-228 89-260 89-279 76-177 111-209 69.5 ± 3.0 69.4 ± 3.0 66.7 ± 2.6 66.8 ± 2.5 67.0 ± 3.5 64.5 ± 2.7 65.0 ± 2.8 62.2 ± 2.5 61.7 ±2.1 63.9 ± 3.1 60-77 58-78 59-72 62-72 59-72 57-74 52-73 54-71 55-67 58-69 Table 2; reference ranges based on the 2.5 and 97.5 percentile values for the distribution of CK in each subgroup also are shown. From the data in Table 2, it appeared that the race/gender subgroups could be placed into three broader groups: a high CK group, composed solely of the black men; an intermediate CK group consisting of all nonblack men plus black women; and a low CK group, comprised of all the nonblack women. The validity of placing these subgroups into the three broader groups was tested by analysis of variance (AN- OVA). 1 Differences between the race/gender subgroups placed into the intermediate CK group and into the low CK group were not significant by ANOVA. Thus, the problem of providing clinically useful limits for CK may be simplified to the limits for these three groups, as shown in Table 3. Discussion Significantly higher serum CK levels in black men and women have been reported previously. " 14 The present study on healthy adults confirms and extends these findings concerning racial differences in serum CK levels. In our study, the mean serum CK in black men was more than twice the mean serum CK in white men, and the mean CK value in black women was almost twice that in white women. Two additional racial groups were included in our study, Hispanics and Asians. Hispanic and Asian men had CK values similar to those found in black women and white men, permitting these four subgroups to be placed into the intermediate CK group. Hispanic and Asian women had CK values similar to white women, permitting these three subgroups to be placed into the low CK group. The reasons for racial differences in serum CK levels are not known. Meltzer and Holy examined diet and level of physical activity as causes of black/white differences in serum CK but failed to find significant differences in these factors between blacks and whites of the same sex. 14 As a result of numerous studies, it is well established that, in normal persons, serum CK levels in men are significantly higher than those in fe- Table 2. Creatine Kinase Values on Race/Gender Subgroups of the Healthy Employee Population High CK Group Black men Intermediate CK Group Hispanic men Asian men White men Black women Other men Low CK Group Asian women Hispanic women White women Other women 237.8 ± 492.1 149.7 ± 362.9 116.5 ± 50.0 113.8 ± 86.0 109.3 ± 71.9 116.9 ± 54.0 69.3 ± 69.2 ± 64.6 ± 78.9 ± 26.3 30.6 34.0 36.2 Median 165 100 106 89 89 115 65 65 59 72 Minimum 15 14 53 16 24 71 20 5 5 15 2.5 52 29 53 30 35 27 29 18 Percentiles 97.5 520 347 230 370 290 127 145 144 Maximum 6755 3971 348 637 588 265 199 255 265 153

Vol. 79 No. 5 HETEROGENEITY OF CREATINE KINASE 585 Table 3. Practical Reference Limits Based on the Boundaries of the Central 95% of the Distribution of CK Values within the High, Intermediate, and Low CK Groups Reference High CK Group 52-520 (Black men) Intermediate CK Group 35-345 (Nonblack men and black women) Low CK Group 25-145 (Nonblack women) males. 6 ' 8 ' 618 ' 23 ' 24 Since skeletal muscle is the largest reservoir of CK in the body, male-female differences in serum CK generally are attributed to the relatively higher muscle mass in men than that in females. 48 '"' 8 Differences in skeletal muscle mass/total body mass ratios also have been invoked to explain racial differences in CK levels. Some investigators have suggested that blacks have higher skeletal muscle mass/total body mass ratios than whites." Olerud and co-workers reported that among Marine Corps recruits who developed acute exertional rhabdomyolysis, blacks had significantly higher peak serum CK, lactate dehydrogenase, aspartate aminotransferase, and myoglobin levels than did nonblacks. 17 Since all of these are constituents of skeletal muscle, Olerud suggested that blacks have a relatively higher skeletal muscle mass than whites. However, it is doubtful that relative muscle mass by itself is sufficient to explain the magnitude of the differences in CK levels between blacks and nonblacks. Meltzer and Holy suggested that racial differences in CK levels may be the result of genetic differences in permeability of the sarcolemma membrane to CK, variations in clearance rates of CK from the vascular compartment, variations in CK activity as measured with in vitro assays, or the presence of inhibitors or activators of CK. 14 The studies of Meltzer and colleagues on monozygotic and dizygotic twins indicated that serum CK activity is at least partially under genetic control. 1213 It is important to recognize significant racial differences in CK activity, particularly when considering diseases or conditions that are associated with mild to moderate elevations in CK values. Examples would include the use of serum CK levels in screening for carriers of Duchenne's muscular dystrophy 3 or for carriers of malignant hyperpyrexia 2 and even at times in the diagnosis of acute myocardial infarction. While acute myocardial infarction often is associated with pronounced elevations of serum CK levels, it may be associated with only mild to moderate elevations, depending in part on the time that blood is drawn for CK determination in relation to the onset of symptoms and on individual differences in baseline CK levels. Failure to consider racial and gender differences in CK activity may lead to a "positive" interpretation of CK results, which actually may represent normal levels. These falsepositive results could lead to needless and costly diagnostic procedures or specific therapies that may be time consuming or even harmful to the patient. Despite the known heterogeneity of CK values among racial and gender groups, recent studies using measurements of total CK as part of the criteria for diagnosing acute myocardial infarction have persisted in using a single limit to define a positive test result for CK, with the cut-off point being the upper limit of the reference range for CK in their laboratories. Thus, Schroeder and associates defined a positive total CK result as >200 U/ L 20 ; Strauss and Roberts used > 145 U/L 22 ; and Irvin and colleagues used > 130 U/L. 9 These studies did not indicate the manner in which their laboratories determined their reference ranges for CK. As we pointed out previously, many laboratories have determined their reference limits by performing the test on a small number of volunteers, often recruited from the technical staff of the laboratory, which results in a biased or at least a restricted group composed principally of women and containing relatively few representatives of nonwhite racial groups. If we were to base the reference range for CK on volunteers from our laboratory's technical staff (the composition of which is closest to that of our low CK group), we might well define a positive CK result as a value greater than 145 U/L, similar to the definitions used in the cited studies. However, a high proportion of healthy individuals in this study would have "positive" CK levels: 56.6% of black men, 20.1% of white men, 20.0% of Asian men, 19.5% of Hispanic men, and 17.7% of black women. On the other hand, using 358 U/L (the upper boundary of the central 95% of the CK values on the overall population studied) to define a positive CK test result, the definition would be approximately correct for the intermediate CK group (black women plus nonblack men) but would not be appropriate for the low CK or the high CK groups. Therefore, we conclude that it is imperative to use definitions of positive CK test results based on reference limits that take into account the major differences in CK levels that exist among racial and gender groups of healthy individuals. References 1. Dixon WJ: P7D. Description of groups (strata) with histograms and analysis of variance. BMDP Biomedical Computer Programs, P-Series, 1979. Edited by WJ Dixon, MB Brown. Berkeley, University of California Press, 1979, pp 185-198 2. Ellis FR, Keaney NP, Harriman DGF, et al: Screening for malignant hyperpyrexia. Br Med J 1972; 3:559-561

586 WONG ET AL. A.J.C.P. May 1983 3. Gale AN, Murphy EA: The use of serum creatine phosphokinase in genetic counselling for Duchenne muscular dystrophy. II. Review of methods of assay and factors which may be relevant in the interpretation of serum creatine phosphokinase activity. J Chronic Dis 1979; 32:639-651 4. Garcia W: Elevated creatine phosphokinase levels associated with large muscle mass. Another pitfall in evaluating clinical significance of total serum CPK activity. JAMA 1974; 228:1395-1396 5. Goode DJ, Meltzer HY: Effects of isometric exercise on serum creatine phosphokinase activity. Arch Gen Psychiatry 1976; 33:1207-1211 6. Griffiths PD: Serum levels of ATP: creatine phosphotransferase (creatine kinase). The normal range and effect of muscular activity. Clin Chim Acta 1966; 13:413-420 7. Harwood SJ, Cole GW: Reference values based on hospital admission laboratory data. JAMA 1978; 240:270-274 8. Hughes BP: A method for estimation of serum creatine kinase and aldolase activity in normal and pathological sera. Clin Chim Acta 1962; 7:597-603 9. Irvin RG, Cobb FR, Roe CR: Acute myocardial infarction and MB creatine phosphokinase. Arch Intern Med 1980; 140:329-334 10. Lott J A, Stang JM: Serum enzymes and isoenzymes in the diagnosis and differential diagnosis of myocardial ischemia and necrosis. Clin Chem 1980; 26:1241-1250 11. Meltzer HY: Factors affecting serum creatine phosphokinase levels in the general population: The role of race, activity and age. Clin Chim Acta 1971; 33:165-172 12. Meltzer HY, Belmaker R, Wyatt RJ. Pollin W, Cohen S: Serum creatine phosphokinase (CPK) activity in monozygotic twins discordant for schizophrenia. Heritability of serum CPK activity. Compr Psychiatry 1969; 17:469-475 13. Meltzer HY, Dorus E, Grunhaus L, Davis JM, Belmaker R: Genetic control of human plasma creatine phosphokinase activity. Clin Genet 1978; 13:321-326 14. Meltzer HY, Holy PA: Black-white differences in creatine phosphokinase (CPK) activity. Clin Chim Acta 1974; 54:215-224 15. Nuttall FQ, Jones B: Creatine kinase and glutamic-oxaloacetic transaminase activity in serum. Kinetics of change with exercise and effect of physical conditioning. J Lab Clin Med 1968; 71:847-854 16. Nuttall FQ, Wedin DS: A simple, rapid colorimetric method for determination of creatine kinase activity. J Lab Clin Med 1966; 68:324-332 17. Olerud JE, Homer LD, Carroll HW: Incidence of acute exertional rhabdomyolysis. Arch Intern Med 1976; 136:692-697 18. Rosalki SB: An improved procedure for serum creatine phosphokinase determination. J Lab Clin Med 1967; 69:696-705 19. Rosalki SB: Enzymes in diseases of skeletal muscle. The Principles and Practice of Diagnostic Enzymology. Edited by JH Wilkinson. London, Edward Arnold (Publishers), Ltd., 1976, pp 263-302 20. Schroeder JS, Lamb IH, Hu M: Do patients in whom myocardial infarction has been ruled out have a better prognosis after hospitalization than those surviving infarction? N Engl J Med 1980;303:1-5 21. Siegel AL, Cohen PS: An automated determination of creatine phosphokinase. Automation in Analytical Chemistry, Technicon Symposia, 1966. White Plains, New York, Mediad Inc., 1967, pp 474-476 22. Strauss HD, Roberts R: Plasma MB creatine kinase activity and other conventional enzymes. Comparison in patients with chest pain and tachyarrhythmias. Arch Intern Med 1980; 140:336-339 23. Thompson WHS: Determination and statistical analyses of the normal ranges forfiveserum enzymes. Clin Chem Acta 1968; 21:469-478 24. Wilkinson JH, Steciw B: Evaluation of a new procedure for measuring serum creatine kinase activity. Clin Chem 1970; 16:370-374