Prostate cancer in systemic lupus erythematosus

Abstract

Our research objective was to estimate prostate cancer risk in systemic lupus (SLE), relative to the age-matched general population. A progressive literature review was performed to identify SLE cohort studies with cancer registry linkage for cancer ascertainment. Data were pooled from four studies of large SLE cohorts who met these criteria. The total number of prostate cancers observed was derived by pooling the incident cases across all studies. The total expected number of prostate, derived from applying appropriate general population cancer incidence data to the observed number of patient-years of follow-up for each study, was similarly determined. The parameter of interest was the standardized incidence ratio (SIR), the ratio of observed to expected malignancies.

The four studies together provided a pool of 6,068 male SLE patients observed for a total of 38,186 patient years (mean 6.3 years). Within these subjects, 80 prostate cancers observed. In each contributing study, the number of cancers expected far exceeded that observed. The pooled SIR estimate for prostate cancer risk in males with SLE, compared to the general population, was 0.72 (95% CI 0.57, 0.89).

These data suggest a decreased risk of prostate cancer in SLE; more definite conclusions require additional data. Since alterations in androgen pathways can potentially alter prostate risk, a lower risk of prostate cancer in SLE could possibly be due to low hypoadrenergic states which some believe may occur in men with SLE; underlying genetic factors could also be at play. Further study of these issues in large cohorts is needed.

Recent evidence linking systemic lupus erythematosus (SLE) and lymphoma risk has largely eclipsed detailed evaluations of other types of cancer in SLE. As well, since SLE is a disease predominantly of women, cancers specific to the males (of which prostate cancer is by far the most common) have gained little attention. In our earlier work of cancer risk in a large SLE cohort, we noted a trend towards fewer prostate cancers in men with SLE, compared to the general population(1). However, since SLE is a female-predominant disorder, pooled data are required to generate precise estimates regarding outcomes in males. The objective of this meta-analysis was thus to estimate the risk of prostate cancers in men with SLE, relative to an age, calendar-year, and geographically matched general population.

Methods

We performed a literature review of published peer-reviewed articles documenting cancer occurrence in males with SLE, focussing on observational cohort designs. As others have pointed out, standard systematic literature searches do not efficiently identify key articles of observational design, so as per Furlan et al, we completed a progressive literature search(2). This entails identifying key articles from existing reviews of the literature, and employing search terms (in Medline, MeSH terms) from these articles to locate additional citations of interest. Progressive search strategies locate essentially 100% of articles from traditional systematic literature searches.

For our analyses, we reviewed all cohort studies since 1995 that provided basic descriptive information (cohort assembly, demographics, person-time of observation), and cancer outcomes. Several studies of cancer in SLE published since 1995 have come from centres that later provided data to the international multi-centre cohort study of cancer in SLE, published in 2005. The early single centre reports were excluded from our current pooled analyses, since they were essential sub-set analyses of the larger multi-centre sample. Studies that did not include male SLE patients were also excluded. The criteria for inclusion were thus as follows: studies of cancer occurrence in SLE cohorts not otherwise replicated, that included males, where cancer registry linkage had been performed to ascertain cancer occurrence.

For our primary analyses, we obtained the total number of incident prostate cancers by summing the number of events across all four studies. Similarly, the total expected number of prostate cancers, estimated in each study by applying appropriate general population cancer incidence data to the observed number of patient-years of follow-up, was derived by pooling the estimates across studies. We then calculated the primary parameter of interest, the standardized incidence ratio (SIR), as the ratio of observed to expected malignancies.

Since SIRs for cancer may differ across centers, we also fit a random-effects hierarchical model allowing differences among studies, rather than assuming a single fixed rate across all. SIR estimation using this hierarchical modeling represents a compromise between the pooling of data across sites (our primary analysis, which assumes no variation in cancer experience from one center to the next) versus independent estimates for each center (the other extreme, which would preclude estimation of a single SIR). We used the Gibbs sampler as implemented with WinBUGS 1.4 software to estimate the model parameters, with 95% credible intervals (CrI)(3).

Results

The progressive literature review identified 11 potential candidates for review. After excluding six studies(4–9), that were essentially sub-set samples from the multi-centre cohort study of cancer occurrence published in 2005(1), and one study which focused only on females with SLE(10), we were left with four large cohort studies that met our criteria for inclusion: the larger international cohort study, a study from Sweden(11), one from Denmark(12), and one from the United States(13). The four studies together (Table 1) resulted in a pool of 6,068 male SLE patients observed for a total of 38,186 patient years (mean 6.3 years). In total, there were 80 prostate cancers observed over this time. The total number of cancers expected exceeded that observed, with an over-all SIR of 0.72 (95% CI 0.57, 0.89) for prostate cancer (Table 2). This suggests an almost 30% reduction in prostate cancers for men with SLE, compared to an age-matched, geographically appropriate general population. Our random-effects model produced very similar findings, with an SIR of 0.73 (95% CrI 0.58, 0.89).

Table 1.

Characteristics of the four recent large cohort studies of malignancy in systemic lupus erythematosus (SLE) which included males

First author	Year	Population, cohort	Mean person-years follow-up	N, Males	Person-years, Males	Over-all SIR (95% CI)*
Bernatsky	2005	Multi centre clinical	6.0	940	5,640	1.15 (1.05–1.27)
Bjornadal	2002	Sweden, Administrative data-based	8.8	1,514	13,323	1.25 (1.14–1.37)
Mellemkjer	1997	Danish, Administrative data-based	6.8	269	1,829	1.30 (1.06–1.58)
Patel	2008	USA, Administrative data-based	5.2	3,345	17,394	1.14 (1.07–1.20).
Total			5.8	6,068	38,186

^*SIR=standardized incidence ratio, ratio of observed to expected cancers. CI=confidence interval. Presented for both males and females since not all studies provided estimates stratified by sex.

Table 2.

Prostate cancers in systemic lupus erythematosus (SLE)

First author	O	E	SIR*	95% CI
Bernatsky	8	11.1	0.72	0.31	1.42
Bjornadal	28	36.5	0.77	0.51	1.11
Mellemkjer	1	1.7	0.59	0.01	3.28
Patel	43	62.2	0.69	0.50	0.93
Total	80	111.5	0.72	0.57	0.89

^*SIR=standardized incidence ratio, the ratio of observed (O) to expected (E) cancers.

This finding was specific for prostate cancer in that there was no decrease in the over-all cancer SIR (Table 1) for the SLE populations studied (in fact, the over-all SIRs suggest a slight increased cancer risk over-all, in each of the four studies).

Discussion

Prostate cancer is the most common non-skin cancer in the United States, affecting 1 in 6 men(14). We present evidence that prostate cancer is decreased in men with SLE, compared to an age-matched, geographically appropriate comparator population. These results are of interest given the putative role of hormones in the pathogenesis of SLE in males. In particular, there is some evidence that males with SLE show a tendency towards low testosterone levels, as compared to males without SLE(15). Androgens mediate cell proliferation in prostate tissue and are thus important in the physiology of prostate cancer. Alterations in androgen pathways that render an individual relatively hypoandrenergic can potentially alter prostate risk, although the relative importance of these pathways, including the genetic polymorphisms that affect them, still requires more study(16).

Genetic factors are also worth considering. One genetic common variant (single nucleotide polymorphisms, SNPs) linked with prostate cancer (rs6983267, on chromosome 8q24) is also associated with ovarian cancer (and co-incidentally is in the same region as a breast cancer-related locus)(17). Both breast and ovarian (as well as endometrial) cancers appear to be decreased in women with SLE, compared to the general population(1). The completion of genome-wide association studies (GWAS) in unselected SLE patients may help in determining whether these specific cancer-related SNPs are in fact decreased in SLE.

Important potential limitations of this work exist. We did not have information about prostate cancer risk stratified by race; all studies were drawn from Caucasian-predominant populations. However, race-specific SIRs for cancers over-all have been published using data from the international multi-centre SLE cohort(18). Those analyses demonstrated similar SIRs for cancer risk over-all in black and white patients with SLE. Unfortunately since men represent only 10% of SLE subjects, precise race-specific estimates for prostate cancer alone could not be generated. The race-specific multi-centre analyses did demonstrate that SIR estimates for breast cancer in women with SLE were decreased similarly in both black and white patients. So, available data suggest that altered cancer risk in SLE (relative to the general population) occurs similarly for blacks and whites. However, we cannot comment specifically on prostate cancer. Since the multi-centre study is being updated, with twice as many patient-years as previously, future data may help resolve the issue(19).

No one has examined the issue of prostate cancer screening in males with SLE. However, data do suggest that women with SLE are less likely to undergo certain types of cancer screening(20). Thus theoretically, it may be that a lower prostate cancer risk could in part be related to less screening. However, comprehensive prostate cancer guidelines were not in place for most of the calendar years over which the data was collected; moreover, most men even in the general population are not compliant with recommended cancer screening(21). Thus, it seems unlikely that differences in screening explain all of the observed findings.

Finally, regarding the effects of medications on cancer risk, the published studies we examined did not provide details about these exposures. Some observational studies have a potential protective effect of non-steroidal anti-inflammatory drugs (NSAIDs, often used in rheumatic diseases) and prostate cancer risk; however, in a recent large prospective cohort study, NSAID use was not associated with prostate cancer risk(22). In sub-analyses of the multi-centre lupus cohort, there was a trend towards a protective role of NSAIDs and aspirin for cancer risk over-all, but again, the low numbers of men preclude much precision with respect to prostate cancers(23).

In summary, these pooled data suggest a decreased risk of prostate cancer in SLE; more definite conclusions require additional data. Since alterations in androgen pathways can potentially alter prostate risk, a lower risk of prostate cancer would be in keeping with earlier data suggesting hypoadrenergic states in men with SLE; underlying genetic factors could also be at play. Further study of these issues in large cohorts is needed.