To distinguish familial differences from sex-related differences in the clinical manifestations of systemic lupus erythematosus (SLE).
To distinguish familial differences from sex-related differences in the clinical manifestations of systemic lupus erythematosus (SLE).
A total of 372 affected individuals from 160 multiplex SLE pedigrees were analyzed. Twenty-five of these pedigrees contained at least 1 affected male relative. Comparisons of the presence of each of the 11 1982 American College of Rheumatology criteria for SLE were made between female family members with affected male relatives and those without affected male relatives, using Fisher's exact tests.
The presence of renal disease was significantly increased in female family members with an affected male relative when compared with those with no affected male relative (68% and 43%, respectively; P = 0.002). This trend remained after stratifying by race and was most pronounced in European Americans. A familial basis for differences in hematologic and immunologic manifestations was also suggested, while arthritis and dermatologic features appeared to be most influenced by sex.
Our results demonstrate that the increased prevalence of renal disease previously reported in men with SLE is, in large part, a familial rather than sex-based difference, at least in multiplex SLE families. Distinguishing familial from sex-related differences may facilitate efforts to understand the genetic and hormonal factors that underlie this complex autoimmune disease.
Systemic lupus erythematosus (SLE) is a chronic, heterogeneous autoimmune disorder. The origin of SLE is multifactorial and most likely involves complex interactions between genetic, environmental, and hormonal factors. SLE is at least 9 times more prevalent in women, but the underlying cause of this sex bias is not clearly established. Clinical differences between male and female lupus patients have been examined, and several distinctions have been noted. A higher prevalence of renal dysfunction in affected men has been noted in numerous studies (1–7). This manifestation is of particular interest because many studies have suggested that renal disease is a sign of poor prognosis (8–14), as is male sex (10, 13, 15). Other clinical differences between male and female patients have been observed, although results vary between studies. An increased prevalence of malar rash (1), discoid rash (7), serositis (6, 16), pleuritis (4), central nervous system involvement (1), and thrombocytopenia (4) has been reported in male lupus patients. Conversely, increased rates of arthritis (7, 17), leukopenia (17), and photosensitivity (6) have been described in female patients.
Differences in serologic profiles may also exist between male and female SLE patients. Female patients tend to have more anti-Ro antibodies than do male patients (7, 17), whereas male patients have more anti–double-stranded DNA (5) and anti-Sm antibodies (4). Differences in antinuclear antibody (ANA) patterns have also been observed; female patients tend to have more homogeneous patterns, while male patients have more nucleolar patterns (17). No sex differences have been shown for serum immunoglobulin levels (18–21), although there is a non–statistically significant trend toward higher IgA deficiency and increased prevalence of IgM in male patients (22).
In previous studies, clinical manifestations were compared between male and female SLE patients who were unrelated. Because large collections of families were not available, it was not possible to differentiate between true sex differences and familial effects. The collection of multiplex SLE families for genetic studies presents a unique opportunity to discern potential sex and familial differences in disease manifestation. The aim of this analysis was to examine clinical differences between SLE multicase families with and without affected male relatives, using an approach capable of distinguishing the effects of sex from familial manifestations, in an effort to further identify genetically distinct SLE subsets.
Recruitment. Patients and pedigrees were recruited as previously described (23). All affected individuals met at least 4 components of the American College of Rheumatology (ACR) 1982 revised criteria for SLE (24), and the relationship between affected individuals within each pedigree was potentially informative for genetic linkage analysis.
We first examined sex differences in clinical and serologic manifestations in 295 individuals from 126 pedigrees that were multiplex for SLE (data not shown), and then extended the analysis by including an additional 77 individuals from 34 pedigrees. There were no significant differences between the first 126 pedigrees and the 34 new pedigrees with respect to demographics, clinical manifestations, or serologic variables and their respective sex differences. Thus, we report the results for the total sample, which included 372 affected individuals from 160 families that were multiplex for SLE. For each pedigree, race was self-reported. Age at onset of SLE was determined using the date the patient first met at least 4 of the ACR criteria. Each SLE patient was grouped into 1 of 3 categories: 1) male, 2) female with an affected male relative, or 3) female without an affected male relative.
Definition of clinical criteria. Sources of data included medical records and patient interviews (or an interview of a patient surrogate in 7 rare instances in which the patient was unavailable). A dichotomous score for each ACR criterion was given as follows. A score of 1 was given if the medical record or patient self-report supported the presence of the criterion. Only on rare occasions was a score of 1 given solely on the basis of patient self-report. A score of 0 was given if the evaluator doubted the criterion was present, if there was absolutely no evidence for the criterion's presence, or if data were missing for the manifestation.
The original ACR criteria were further collapsed into 8 major classes; a score of 1 was given if at least 1 subcriterion equaled 1, otherwise a score of 0 was given. For example, the dermatologic score equaled 1 if at least 1 of the common features, i.e., malar rash, discoid rash, photosensitivity, or oral ulcers, was present. Similarly, the renal score was obtained from the proteinuria and cellular casts scores, the cardiopulmonary score from the pericarditis and pleuritis scores, the neurologic score from the seizures and psychosis scores, the hematologic score from the hemolytic anemia, leukopenia, lymphopenia, and thrombocytopenia scores, and the immunologic score from the lupus erythematosus cell, anti-DNA, and anti-Sm values. The presence of ANA and arthritis defined individual classes of criteria.
Serology. Autoantibodies were assayed using standard methods. Anti–native DNA was detected by immunofluorescence on a Crithidia luciliae substrate (Helix Diagnostics, West Sacramento, CA) and was recorded as either positive or negative. Precipitating antibodies, such as anti-Sm, anti–nuclear RNP, anti-Ro, anti-La, and anti-P, were detected by immunodiffusion and recorded as positive or negative. Antibodies to cardiolipin, herein referred to as IgM, IgG, and IgA, were detected by enzyme-linked immunosorbent assay; titers ≥1:20 were considered positive. Tests for a functional classical complement cascade were assessed using lysis of sheep erythrocytes in a CH50 assay for affected individuals. If the CH50 titer was less than half of the corresponding “normal” titer, the result was considered to be abnormal. ANAs were detected using standard immunofluorescence staining. ANA titers were considered to be significantly positive if the ratio was ≥1:120.
Data analysis. For each clinical criterion and serologic titer, comparisons between female family members with affected male relatives and those without affected male relatives and between all male and all female members of the pedigrees were made using Fisher's exact tests. For the 8 grouped clinical criteria (excluding ANA and arthritis), a difference was considered statistically significant if, for a particular clinical manifestation, the percentage displaying the manifestation was significantly different, at α < 0.007 (using a Bonferroni correction), among female members with an affected male relative compared with female members without affected male relatives. If the difference in a major characteristic was significant, then its subcharacteristics were tested at the α = 0.05 level. All other comparisons were tested at the α = 0.05 level and considered to be exploratory in nature.
Presence of renal disease was analyzed after stratification by race and by considering the correlated nature of familial data. Multiple logistic regression (25) was performed in the largest ethnic subset to test the effect of the presence of an affected male relative on the probability of having renal disease, adjusting for disease duration and age at onset of SLE. Because our analysis included related individuals and the observations were therefore correlated, the multiple logistic regression was embedded in a generalized estimating equation (GEE) framework (26) to account for correlations within families. This analysis was performed using the statistical software program PROC GENMOD in SAS (27). We also computed a concordance odds ratio to determine if renal disease clusters in families; to allow for differing prevalences of renal disease in subgroups, we stratified the families by race and presence of affected male relatives and used a Mantel-Haenszel–type approach (28). An odds ratio significantly higher than 1 provided evidence of familial clustering of renal disease among SLE-affected individuals.
We examined the presence of clinical criteria and serologic findings in 372 SLE-affected individuals. Of these, 27 were male, 38 were female with affected male relatives, and 307 were female without affected male relatives. These individuals comprised 160 pedigrees, 25 of which included affected male relatives. The number of pedigrees in each racial group is shown in Table 1. Pedigrees with and without affected male members were relatively evenly distributed among the racial groups. There was no statistically significant difference in the age at onset of SLE between male and female family members; the mean ages at onset were 34.89 years, 33.29 years, and 33.03 years for male members, female members with affected male relatives, and female members without affected male relatives, respectively. Disease duration was also compared and revealed no significant difference (P = 0.5). Differences in the mean age at onset of SLE and disease duration remained insignificant when the data set was stratified by race (P = 0.8 and P = 0.9, respectively).
|Race||No. of individuals||No. (%) of pedigrees|
|With affected males (n = 25)||Without affected males (n = 135)||Total (n = 160)|
The percentages of male members, female members with affected male relatives, and female members without affected male relatives meeting each of the ACR criteria are shown in Table 2. In comparing the affected individuals in the SLE pedigrees according to the presence of each of the grouped criteria, there was a significant difference in the presence of renal disease between the female individuals with affected male relatives and those without affected male relatives; 68% of female members with affected male relatives had renal disease, compared with 43% of female members without affected male relatives (P = 0.002). This contrast was also evident with regard to the presence of both proteinuria (P = 0.006) and cellular casts (P = 0.002). Renal manifestations, proteinuria, and cellular casts were found more often in all of the male members of the pedigrees compared with the unrelated women (Table 2). These differences suggest that the presence of renal disease is a familial difference rather than a sex difference in multiplex SLE families. A suggestive, but nonsignificant, contrast was also present for overall hematologic dysfunction; 84% of the female relatives of affected male members had some type of hematologic dysfunction, compared with 67% of female members without affected male relatives (P = 0.012). For each of the hematologic characteristics, the men had a higher prevalence than did the women without an affected male relative.
|Clinical manifestation||With affected males||Female members with no affected males (n = 307)|
|Males (n = 27)||Females (n = 38)|
|Positive lupus erythematosus cell||11.1||15.8||10.7|
Suggestive differences between the affected men and all of the affected women were present for arthritis and hemolytic anemia; 85% of male family members had arthritis, compared with 97% of the female members with affected male relatives and 96% of the female members without affected male relatives (P = 0.033). Male members appeared to have a higher frequency of hemolytic anemia than did female members; 30% of the men met this criterion compared with 16% of the women with affected male relatives and 12% of the women without affected male relatives (P = 0.023). Although not statistically significant, the dermatologic manifestations appeared to segregate according to sex, with malar rash, photosensitivity, and oral ulcers being more common in the women and discoid rash found more frequently in the men. None of the above results changed appreciably when adjusted for age at onset of SLE and disease duration in the analysis.
Renal disease was further evaluated in these multiplex SLE pedigrees. It is well known that renal disease is more prevalent in African Americans than in European American SLE patients (18, 29). Therefore, we examined the relationship between presence of renal dysfunction and pedigree type by stratifying by race. These results are shown in Table 3. The increased prevalence of renal disease in female family members with affected male relatives versus female family members without affected male relatives was most clear in the European American pedigrees (P = 0.019).This difference appeared to be consistent across races, suggesting that the increased prevalence of renal disease in the women with affected male relatives was a familial difference that is not explained by race. Interestingly, African American men showed considerably less renal disease than did either their female relatives, any of the affected women in general, or European American men. However, the sample sizes were small and these comparisons were not statistically significant.
|With affected males||Female members with no affected males|
|Sample size, n||17||23||173|
|Overall renal dysfunction||52.9||56.5†||31.8|
|Sample size, n||7||9||113|
|Overall renal dysfunction||42.9||88.9||56.6|
|Sample size, n||1||3||2|
|Overall renal dysfunction||100.0||66.7||50.0|
|Sample size, n||2||3||12|
|Overall renal dysfunction||100.0||100.0||66.7|
The difference between the female family members with and those without affected male relatives among the European Americans, the largest ethnic subgroup, remained significant (P = 0.0062) after adjusting for age at onset of SLE and disease duration, using logistic regression with GEE to account for correlation within families. In multicase European American families, a female family member with an affected male relative was 3.46 times more likely (95% confidence interval 1.42–8.43) to develop renal disease than was a female family member without an affected male relative. In addition, the odds ratio for concordance of renal disease in the total sample was 2.34 (Z score 3.07, P = 0.001) after adjusting for ethnicity and presence of affected male relatives, demonstrating that renal disease does cluster within families.
Results from the comparison of serologic characteristics are shown in Tables 4, 5, and 6. There were no significant differences in any of the laboratory variables between female family members with and those without affected male relatives. However, differences in the CH50 values approached statistical significance (Table 4), with 31.6% of female relatives of affected male members having abnormal CH50 levels compared with 19.9% of female members without affected male relatives (P = 0.061). IgM and IgG anticardiolipin antibody titers were available for only 225 individuals, and these results are detailed separately (Table 5). No IgA anticardiolipin antibodies were detected (data not shown). In Table 6, ANA patterns are shown only for patients who displayed positive ANA titers. The only sex difference observed was for the nuclear homogeneous ANA pattern; female family members overall had a higher prevalence of this type of pattern than did the male members (P = 0.032).
|Serologic manifestation||With affected males||Female memberswith noaffected males(n = 307)|
|Males (n = 26)||Females (n = 38)|
|Abnormal levels of CH50||7(26.9)||12(31.6)||61(19.9)|
|Serologic manifestation||With affected males||Female members with noaffected males (n = 180)|
|Males (n = 18)||Females (n = 27)|
|Abnormal Ig levels (IgG or IgM)||4(22.2)||3(11.1)||19(10.6)|
|Pattern||With affected males||Female memberswith noaffected males(n = 273)|
|Males (n = 21)||Females (n = 37)|
|Mitotic spindle apparatus||0||1(2.7)||2(0.7)|
SLE is a complex disorder with diverse manifestations. Differences in the clinical manifestations between male and female SLE patients have been widely reported. Still, the question remains as to whether these observed differences are truly sex-based or actually familial. Examination of clinical and laboratory manifestations in our collection of multiplex SLE families with affected male relatives provided a unique opportunity to address this question. Our data suggest that, for several clinical features of SLE, differences between male and female members of such families may indeed be familial rather than sex-based, most notably in the development of renal dysfunction.
In the present study, we examined clinical and serologic manifestations in 372 individuals affected with SLE. We found a notable increase in the prevalence of renal manifestations in male members versus unrelated female members, which is consistent with the results from previous studies (1–7). Importantly, we also found a statistically significant increase in renal manifestations in their affected female relatives compared with female members with no affected male relative. Further examination of the presence of proteinuria and cellular casts supported this conclusion, and stratification by race suggested that the difference is not explained by ethnicity. These results suggest that the increased prevalence of renal disease in men is, to a substantial degree, likely to be due to a familial difference rather than being solely sex-based. It should be noted, however, that the present study had insufficient power (due to small sample sizes) to detect small or moderate differences in the prevalence of renal disease between SLE-affected male members and their SLE-affected female relatives, so we cannot exclude the possibility that some of the difference was due to sex. We also stress that these results can be extended only to SLE multiplex families and may not apply to nonfamilial cases of SLE.
Overall, hematologic dysfunction also appears to be familial, particularly if one considers that the raw percentages for each manifestation were increased in all affected members from families with an affected male relative versus unrelated affected female members. However, the difference in the presence of hemolytic anemia was statistically significant between all affected male members and all affected female members, suggesting that an influence of sex may be important for this particular manifestation. Finally, there were no statistically significant differences in autoantibody profiles between female family members with and those without affected male relatives; however, the families with an affected male relative did show a trend (although not significant) toward abnormal CH50 results. Because abnormal CH50 levels are associated with a greater prevalence of renal disease (30, 31), this trend further supports the notion of a disproportionate amount of renal dysfunction in families with affected male relatives.
Whether clustering of renal disease (and possibly hematologic and immunologic disorders) is due to common environmental or genetic factors (or both) will require further investigation. However, there is an abundance of evidence to support the potential for a genetic contribution to the disease in these families, from both murine and human studies. Previous studies have examined familial aggregation of SLE in pedigrees containing affected male members (32–34). The report by Lahita et al (34) describes 4 unique pedigrees containing at least 2 affected male members each. Only 1 pedigree had a female diagnosed with SLE, although lupus-like symptoms were present in additional family members. The men in these families had similar, although not identical, disease expression, including renal manifestations, which has interesting parallels with the BXSB mouse model of SLE (35). In this murine model, male mice with lupus die early from immune glomerulonephritis. Hormones do not appear to influence disease expression. The genetics of this model include a single gene on the Y chromosome, Yaa, which accelerates the onset and severity of autoimmunity in males in the context of at least 5 additional autosomal susceptibility loci (36).
If families with affected male relatives truly are a unique genetic subset, in-depth analysis of individuals in these families could facilitate mapping for SLE-susceptibility genes. The importance of genetic factors in SLE is widely recognized and supported by an estimated 66% heritability (37) and evidence of familial aggregation (37, 38). Higher concordance rates between monozygotic twins (24%) compared with dizygotic twins (2%) are also consistent with potential genetic determinants (39, 40).
Genetic linkage of SLE to >40 unique chromosomal regions has been identified across multiple studies (23, 41–44). However, results from these 5 genome scans differ widely from each other, and often are not replicated by individual studies. This may be due, in part, to differences in sampled proportions of subtypes of lupus that result from underlying genetic heterogeneity. Identification of clinical subtypes of lupus will help allow researchers to explicitly control for genetic heterogeneity when searching for genes. Of possible relevance to this study, deficiencies in early components of the complement cascade constitute strong, but rare, risk factors for susceptibility to SLE. In such cases, the male:female ratio is nearly equal (45). Given the propensity for abnormal CH50 levels (which is a measure of global complement component function) in our pedigrees with affected male relatives, analyses of candidate genes such as C1q, C2, C4, and related genes may provide at least a partial explanation for genetic susceptibility to SLE in this subset of families. Analysis using a conditional-logistic affected relative pair analysis for linkage (46) restricted to the 25 families with affected male relatives showed tentative evidence for linkage (logarithm of odds 1.689) on chromosome 15 (data not shown). A more conclusive linkage analysis will require a larger number of such families.
A genetic component in families with affected male relatives may also directly affect known hormonal factors in SLE. Sex steroids affect the maturation of organ systems, including the immune system (47), and are thought to play a role in autoimmune disorders. Feminizing steroids are known to exacerbate SLE by stimulating the immune response (48, 49). Estrogen has been associated with decreased levels of T cells and natural killer cells, and increased levels of B cells (47, 50). Estradiol has been found to accelerate B cell activation (51). Prolactins accelerate autoimmunity (47), by exerting an influence on cytokine levels (52). Androgens are thought to be protective against SLE, but are found in decreased levels in women with SLE (47). Differences in hormone levels or metabolism between lupus patients and healthy individuals may have a genetic basis, and may, in part, explain particular differences in disease expression that vary between the sexes. Collection of additional data regarding hormone (e.g., testosterone or prolactin) levels, impotence, hypogonadism, and feminization in the families presented herein will certainly be of interest in future studies and may provide further insight into the underlying pathogenetic mechanisms related to lupus in men as well as their affected female relatives.
In summary, we have shown that female SLE patients in families with affected male relatives have significantly more renal disease than do female SLE patients without affected male relatives. Our results suggest that the increased prevalence of renal manifestations is familial, rather than sex-based as previously reported, at least in multiplex SLE families. This increased prevalence of renal disease in families with affected male relatives indicates that this group of pedigrees may be a plausible subset for further genetic analysis, potentially facilitating the search for SLE-susceptibility genes. Furthermore, the results of this study underscore the importance of distinguishing familial from sex-related differences when evaluating potential genetic and hormonal factors that underlie this complex autoimmune disease.
Data on 101 pedigrees were obtained from the Lupus Multiplex Registry and Repository (see http://omrf.ouhsc.edu/lupus). Some of the results were obtained using the program S.A.G.E., supported by US Health Service resource grant RR-03655 from the National Center for Research Resources.