The participants in the study were 620 patients (181 men and 439 women; 351 with PM and 269 with DM; 427 Caucasians, 130 blacks and 63 of other races) of the initial 800 consecutive individuals with myopathies who were evaluated in natural history and therapeutic research protocols in the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) between 1985 and 2005. All subjects signed informed consent for NIAMS institutional review board approved protocols. Most subjects were referred for recommendations for therapy, possible inclusion in therapeutic trials or for diagnostic purposes. Selection criteria for this study included adult patients with the diagnosis of PM or DM on the basis of the exclusion of other forms of myopathy and meeting probable or definite Bohan and Peter criteria [9], and who had results for at least 15 of the 20 laboratory tests commonly assessed in myositis patients at the NIH. All laboratory tests were performed in the NIH Department of Laboratory Medicine of the NIH Clinical Center and data from the first time that at least 15 concurrent laboratory test values were available were used. Subjects with criteria for possible myositis (n=35), muscular dystrophies (n=29), inclusion body myositis (n=28), cancer-associated myositis (n=9), juvenile myositis (n=5) metabolic myopathies (n=4), PM or DM but fewer than 15 concurrent laboratory tests available (n=27), and undefined myopathies (n=43) were excluded from study.

Data collected in this retrospective cohort study included demographics (age, gender and race), clinical information (the presence of Gottron’s papules or sign or Heliotrope rash to define DM), and 20 laboratory tests commonly performed on myositis patients at the NIH. The laboratory tests included were serum levels of albumin, creatinine, aldolase, CK (total level and isoenzymes CK-BB, CK-MB, CK-MM), quantitative immunoglobulins (IgG, IgM and IgA levels), erythrocyte sedimentation rate (ESR), total white blood cell count (WBC) and differential (including percent basophils, lymphocytes and polymorphonuclear leukocytes), hematocrit, aspartate aminotransferase (AST), alanine aminotransferase (ALT), LD, and myoglobin. When the normal reference ranges changed during the study period, laboratory results were expressed in relation to the upper limits of the normal range and recalculated based upon the current normal range.

Data analyses were performed using Statistical Package for the Social Sciences (SPSS) version 10.0 (SPSS, Inc., Chicago IL.), following the SPSS Base 10.0 for Windows User's Guide (SPSS Inc., Chicago IL.), and Stata version 11 (Austin, TX). Because analyses using the Shapiro-Wilk test showed that all 20 of the laboratory measurements were not normally distributed, the summary measures reported are medians and interquartile ranges. The proportions of subjects with abnormally high and low values were calculated based on the normal ranges established for healthy individuals at the NIH Department of Laboratory Medicine (established to cover 95% of normal individuals). Due to the non-normality of the distribution of measurements, the primary analyses to assess possible differences in laboratory tests in demographic groups (i.e., male and female gender; Caucasian, black, and other races; age tercile groups of <37, 37-50, and >50 years) and in clinical groups (i.e., PM and DM), were performed via the nonparametric Wilcoxon rank sum test (for two groups), the Kruskal-Wallis test (for the three unordered racial categories), or Kuzick’s nonparametric trend test (“nptrend” test in Stata) for the three ordered age categories.

Correlations of selected laboratory values used Spearman's rank correlation to accommodate the non-normality of the values. For this study we defined Rho correlations as falling into four categories: 0.7-1.0 = high correlation; 0.5- <0.7 = moderate correlation; 0.3- <0.5 = low correlation; and <0.3 = little correlation [10].

Additional analyses used multiway ANOVA or prediction intervals based on linear regression models. For these analyses either logarithms of the raw values were used (14 of the parameters) or the inverse hyperbolic sine transformed-values were used (for six parameters: basophils, lymphocytes, Polys, ESR, and the three CK isoenzyme values; this transformation is very similar to a log transformation, but is more flexible in allowing 0, or even negative, values). As indicated by the Shapiro-Wilk test and other confirmatory tests, these transformations greatly reduced the skewness of the distributions and resulted in more Gaussian-like distributions. The multiway ANOVA analyses were used to confirm the primary univariate group comparisons, by including in each analysis gender, race, and diagnosis. In assessing the effect of age, all three factors of gender, race, and diagnoses were taken into account.

Because this was an exploratory study, a two-sided p value of <0.05 was considered significant for these analyses, with no correction for multiple comparisons.

As expected, many of the IIM patients frequently had various enzyme values above the normal limits (Table 1). The abnormal laboratory results included skeletal muscle markers such as aldolase, CK, CK-MB, LD, myoglobin, and transaminases (ALT and AST), as well as acute and chronic inflammatory markers (WBC count, polymorphonuclear leukocytes, ESR, and immunoglobulins IgG and IgM).

In contrast, a number of other test results were frequently below the normal range. Of these, low CK-MM levels, lymphocyte counts, and creatinine levels have been previously reported in patients with IIM [5, 11, 12]. Low hematocrit and serum albumin levels, however, have not been systematically described in myositis.

CK levels were significantly higher in men compared to women, in blacks compared to Caucasians and in PM compared to DM (Fig. 1), as previously described [13]. CK-MB levels were also significantly higher in men compared to women with IIM. The median aldolase levels were also above normal in all groups and they were significantly higher in men, blacks and in patients with PM. Age also seemed to impact the aldolase levels. There was a significant trend for values to go down in older age groups primarily due to the oldest age tercile (>50 years) being significantly lower than the other two terciles. Similar age-related trends were seen in all three skeletal muscle markers (CK, CK-MB and aldolase; Fig. 1).

Fig. (1). Box plots showing the median, 25th and 75th percentiles and highest and lowest values for CK, CK-MB and aldolase in all myositis patients and differences among groups. The shaded area depicts the normal range.* *Abbreviations: Dx, diagnosis; DM, dermatomyositis; PM, polymyositis; B, black; W, white, Caucasian; O, other; rest per Table 1.

Fig. (2). Box plots showing the median, 25th and 75th percentiles and highest and lowest values for AST, ALT and LD in all myositis patients and differences among groups. The shaded area depicts the normal range*. *Abbreviations: per Table 1 and Fig. (1).

Fig. (3). Box plots showing the median, 25th and 75th percentiles and highest and lowest values for WBC count, lymphocytes and ESR in all myositis patients and differences among groups. The shaded area depicts the normal range*. *Abbreviations: per Table 1 and Fig. (1).

Fig. (4). Scatterplots and the 99% and 95% prediction intervals for the relationships among CK and aldolase (A), ALT (B), AST (C) and LD (D) levels*. *Abbreviations as in Table 1 and Figs. (1-3).

The median values tended to be borderline high for both AST and ALT in most groups, while above normal for LD in all groups (Fig. 2). There were statistically significant differences in gender, race and clinical diagnosis groups for AST and ALT. Men had higher levels of these enzymes than women, as has been described in healthy populations [14]. On the other hand, LD was significantly higher only in blacks and in patients with PM. Higher serum LD in blacks compared to Caucasians is consistent with the observation that enzymes catalyzing reactions in phosphagenic (CK) and glycolytic (e.g., LD) metabolic pathways have significantly higher activities (by 30-40%) in skeletal muscle biopsy tissue of Africans compared to Caucasians [15].

Although all subgroup medians fell within the respective normal reference intervals, there were several statistically significant differences for WBC count, percent lymphocytes, and ESR (Fig. 3). WBCs were higher in males than females and in Caucasians than blacks, while lymphocytes were lower in males than females and tended to decrease with age. This conflicts with data in other populations describing higher WBC counts in women [16]. ESR levels were higher in females compared to males and in blacks compared to Caucasians, as previously reported for healthy populations [17].

The median values were within the respective reference intervals for serum albumin and hematocrit, but they were at the low end of the reference interval for serum creatinine, consistent with the expected loss of muscle mass in myositis patients. On the other hand, serum albumin levels were significantly higher in Caucasians than blacks and significantly lower in PM compared to DM patients. Similar to a healthy population, serum creatinine levels were significantly higher in men than women with IIM, likely due to the larger muscle mass in men. Creatinine levels were higher in those older than 50 years, possibly due to declining renal function. The hematocrit was significantly higher in males compared to females with IIM, as described in healthy populations [18].

In order to assess the relative contribution of diagnosis, age, gender and race to the differences in certain laboratory measures, we compared P-values based on the log values of laboratory measures. These analyses showed that CK and aldolase were most different among the diagnostic groups (P = 1.2 X 10^-17 and P = 5.3 X 10^-9, respectively). Nonetheless, CK and aldolase also differed among races (P = 4.1 X 10^-9 and P = 0.0086, respectively), and by gender (P = 0.002 and P = 0.005, respectively), but both of these were less significant than the differences among diagnostic groups. Hematocrit differed most between males and females (P = 8.4 X 10^-13), but less so among races (P = 0.00008), age groups (P = 0.001) and diagnostic groups (P = 0.035).

The parametric multiway ANOVA analyses, always taking into account the factors of gender, race, age and diagnosis, confirmed all of the statistically significant univariable results, with confirmatory P-values <0.05, except in the following 3 cases where the multivariable P-value did not reach significance: (1) the effect of age on lymphocytes (P = 0.082); (2) the effect of gender on WBCs (P = 0.088); and (3) the effect of gender on lymphocytes (P = 0.061) were not significant. The number and magnitude of these changes in P-values are quite minor, and serve to strongly confirm the primary univariable results.

A number of significant correlations among the laboratory tests studied were identified in the IIM (see Table 2 where only correlations greater than 0.5 in absolute value are shown). Moderate to high correlations were observed among aldolase, ALT, AST, CK, LD, myoglobin, Polys, and WBC. Muscle enzymes and muscle breakdown products all correlated highly, as expected. Aldolase correlated significantly with 14 of the other laboratory tests and particularly strongly with ALT, AST, CK, LD, and myoglobin. A moderate correlation was found between Polys and WBCs. Lower correlations were found among the other laboratory tests.

Fig. (4) displays scatter plots for CK levels with aldolase, ALT, AST and LD values for all patients. The plots show the 95% and 99% prediction intervals of CK with each of these four tests, thus providing guidance to clinicians of when a test result may be considered to be outside the expected range for patients with myositis and when further evaluation for liver disease may be warranted. Although only limited data were available in this retrospective study, the medical records of the 17 individuals who had ALT, AST or LD levels higher than the 99% prediction interval were examined to assess the possibility of liver disease. Of these 17 individuals, four had documented liver disease (one each with fibrotic liver disease of unknown origin, non-alcoholic steatohepatitis, hepatomegaly of unknown cause and hepatitis secondary to blood transfusion).