Introduction
Systemic sclerosis (scleroderma, or SSc) is a heterogeneous, multisystem connective tissue disease that arises as a consequence of a complex interplay of altered immunologic processes involving vascular endothelial cell damage and excessive activation of fibroblasts, culminating in skin sclerosis and fibrotic changes of affected visceral organs.
The association between SSc and increased cancer risk has been reported in several case series, although this association has not been consistently demonstrated across all studies[
1]. Researchers who have conducted recent meta-analyses have reported increased numbers of solid organ malignancies in SSc patients, including lung cancers but not breast cancers. Investigators in other studies, however, have suggested that lung cancers are considered to be common in patients with SSc[
1‐
6]. The significant heterogeneity and variability in the cancer subtypes reported in these studies may be a consequence of limited patient numbers, patient characteristics and study design. Moreover, the actual role of SSc in the development of cancer remains unclear. For example, it is debatable if SSc patients with severe interstitial lung disease have an increased risk of developing lung cancer.
Few studies have examined the temporal relationship of cancers to the onset of SSc, and the current literature that links breast cancer to the onset of SSc remains contradictory[
7,
8]. The role of autoantibodies and malignancies in other autoimmune rheumatic diseases, however, has been well-described, such as in large population-based studies for inflammatory myositis. Anti-MJ/NXP2 and anti-p155/140 antibodies have been described in cancer-associated myositis. With regard to the latter, its high negative predictive value is potentially helpful to rule out the presence of occult malignancy in patients with dermatomyositis[
9,
10]. In contrast, only a few small case series have indicated that the presence of hallmark SSc-specific antibodies, anticentromere antibodies (ACAs) and anti-Scl-70 antibodies, correlates with cancer in SSc, but this finding has not been consistently replicated in other studies. Shah
et al. recently reported that anti-RNA polymerase III (anti-RNAP) antibodies may be associated with malignancy in a small cohort of five patients with early-stage SSc[
8]. Airò
et al. also described similar clustering of cancer associated with the onset of SSc in a small sample of patients[
7]. However, this observation has yet to be confirmed independently in a large cohort of patients. A better understanding of this relationship between antibody specificity and cancer in SSc is necessary to optimise follow-up and surveillance of these patients.
We evaluated the frequency of cancer subgroups in a large, single-centre UK cohort of patients with SSc. The demographic and clinical characteristics of this cohort were compared with matched cases without a history of cancer. The risk of cancer was assessed across patients with different autoantibody status, and the temporal relationship between cancer diagnosis and clinical onset of SSc was explored.
Methods
We conducted a retrospective study of patients attending our centre. All of our patients met the classification criteria of LeRoy as having limited cutaneous SSc (lcSSc) or diffuse cutaneous SSc (dcSSc)[
11‐
13]. Approval for this study was obtained from the London Hampstead National Research Ethics Services Committee, and all individuals consented to use of their clinical and laboratory data for the purposes of research.
To compare the patients with vs. those without cancer, relevant demographic and clinical data related to cancer subtypes and time of diagnosis were retrieved from the Royal Free Hospital integrated patient records, and further information was obtained from our large SSc research database, as well as, if appropriate, from referring physicians or primary practitioners. Our database comprises information on patients with a median follow-up of 12.8 years (mean = 13.7 years; range = 0 to 52.5 years). All SSc patients are seen in our research centre at least annually. The dates of the most recent visit were recorded for all patients and were used for censoring in the analysis of time to development of cancer.
SSc onset was defined by skin sclerosis or first SSc-related internal organ manifestations other than Raynaud’s phenomenon (RP). Patient demographics and key SSc characteristics, including disease duration from the onset of both first non-RP and first RP symptoms in addition to modified Rodnan skin score were collected at baseline. Clinical details pertaining to major internal organ involvement for lung fibrosis, pulmonary arterial hypertension, renal crisis and significant gastrointestinal (GI) involvement were evaluated using consensus definitions as outlined in previously published studies[
14].
Serum was routinely analysed in all SSc patients by indirect immunofluorescence, counterimmunoelectrophoresis and enzyme-linked immunosorbent assay to identify SSc-specific autoantibodies (ACAs, anti-RNAP and anti-Scl-70) at the first visit, and these analyses were repeated annually during the clinic visit. This routine ensures complete data capture for antibody reactivity and each patient has the same antibody reactivity during the disease course in our cohort[
15,
16]. This is consistent with published studies that indicated that these antibodies are usually mutually exclusive. Although the presence of these antibodies was not studied prior to the onset of SSc in our cohort, their specificities remained constant throughout the disease course[
17]. The Hep2 immunofluorescence pattern was assessed by an experienced clinical scientist to inform further characterization of antibodies, including anti-RNAP antibodies, as described previously[
17,
18].
The UK NHS National Care Records Service was used to verify the vital status of patients with cancer who were otherwise lost to follow-up. Statistical methods for contingency tables (χ
2 test or Fisher’s exact test) and time-to-event data were employed. For proportions, 95% confidence intervals (95% CI) were calculated according to Wilson’s method. Differences between SSc patients with vs. without a history of malignancy were assessed by means of Kaplan-Meier curves and the logrank test. Multiple regression analysis (logistic and Cox) was undertaken to assess the impact of autoantibody status on cancer risk (adjusted for age and sex distribution). To include the development of cancers prior to SSc onset in the analysis, these event times were rescaled by subtraction of the smallest observed event time (that is, -22 years). This step does not affect the characteristics of semiparametric Cox regression or the logrank test. For logistic regression, only the development of cancers within 36 months of SSc onset (that is, before and after onset), 36 to 60 months around SSc onset and 60 to 120 months around SSc onset were considered. Patients for whom there was insufficient follow-up were excluded, and event times beyond the preset limits were censored. Similar logistic regression analysis was undertaken for breast cancers. Hazard ratios (HRs), odds ratios (ORs) and the corresponding 95% CIs were calculated. In order to guard against type I error inflation due to multiple testing, only
P-values below 0.01 were considered statistically significant[
19]. For the purpose of data analysis, we defined the onset of SSc as the date of appearance of the first non-RP symptoms. Further analysis was undertaken using the onset of RP as a proxy for the date of SSc onset. Statistical analysis was performed using the following software: SPSS (IBM, Armonk, NY, USA), Stata (StataCorp, College Station, TX, USA; command ci) and Excel (Microsoft, Redmond, WA, USA).
Discussion
There is emerging evidence from recent case series and epidemiological studies that SSc is associated with an increased risk for various malignancies[
1,
20]. In this large retrospective registry-based cohort study, we have shown that, compared with patients without cancer (
N = 2,023), SSc patients with cancer (
N = 154) were more frequently positive for anti-RNAP antibodies than the other hallmark SSc-specific antibodies, ACA or anti-Scl-70 antibodies. This study confirms that positivity for anti-RNAP antibodies is associated with at least a twofold increased HR for cancers that occurred before or after onset of SSc compared to those without anti-RNAP antibodies. The association remained significant for anti-RNAP antibodies and the development of cancers that occurred after the onset of SSc. In contrast, no significant differences in cancer frequency were demonstrated across disease subsets or between the sexes. Ethnicity data were not available, so this factor could not be considered in the analyses.
Importantly, the results of our present study demonstrate that positivity for anti-RNAP antibodies conferred sixfold higher odds for developing cancer within 36 months of SSc onset compared to those without this antibody, highlighting the close temporal association in this subgroup. This result supports the observation in an earlier study by Shah
et al., who found that patients with anti-RNAP antibodies developed SSc within 2 years of cancer onset in a smaller cohort of 23 patients[
8]. Airò
et al. also reported a clustering of cancers with SSc onset in a small sample of patients with anti-RNAP antibodies[
7]. In our present study, this temporal relationship was noted in particular among SSc patients with breast cancer, supporting the observation recently reported by Launay
et al.[
21]. The temporal relationship between cancer and SSc reported herein is similar to that of adenocarcinomas in dermatomyositis patients, and this association also extends, albeit with lower risk, to patients with two other autoimmune rheumatic diseases: polymyositis and systemic lupus erythematosus[
22‐
24]. In this regard, this association suggests that the presentation of malignancy and the onset of SSc may be mechanistically linked and that the confluence of SSc and cancer also indicates that some cases of SSc might represent a paraneoplastic syndrome in which the immune system is responding to cancer.
Shah
et al. also detected enhanced expression of nucleolar RNAP exclusively in the tumoural tissues of SSc patients with anti-RNAP antibodies in whom there was a close temporal relationship between the onset of cancer and SSc[
8]. RNAP is critical for regulation of sustained cellular protein synthesis and is therefore a fundamental determinant of normal cellular growth. Recently, strong evidence has implicated abnormal RNAP activity in cancer cells from breast and lung carcinomas[
25] and in fibroblasts transformed by Simian virus 40 or other polyomavirus[
26,
27]. It is postulated that repression of tumour suppressors p53 and retinoblastoma[
28] and/or activation of oncogene product c-Myc[
29] may lead to enhanced RNAP activity in malignancy. However, the biological basis for an association between specific autoantibody subtypes against NRAP and malignancy in the context of SSc is unclear. The presence of anti-RNAP antibody may initiate an antitumour immune response that, in the appropriate setting, may cross-react against specific host tissue, resulting in target tissue damage. Proof supporting this hypothesis may provide some insight into the fundamental mechanistic aspects of pathogenesis and highlight the cancer–autoimmunity interface in this particular subset of SSc as a paraneoplastic syndrome.
It is disputed whether SSc onset should be defined from the development of RP or from the first non-RP symptoms[
30]. For that reason, the duration of RP prior to SSc onset was considered, and repeat analysis indicated that the association between anti-RNAP antibodies and malignancy remained significant. This finding in our study is not surprising, given that we have previously shown in an independent cohort that the duration of antecedent RP differs substantially between SSc-specific antibodies with the shortest interval between onset of RP and first non-RP features for patients with anti-RNAP antibodies (average of 1.7 years) and longest for ACA-positive patients (average of 10.8 years)[
31].
Different types of cancers have been reported among SSc patients, including cancers that are common in the general population, such as breast and lung cancers. In agreement with this, the leading cancer subtypes in our patient cohort were breast cancer (42.2%), followed by haematological malignancies (12.3%), GI and gynaecological cancers (11% each), and lung cancer (10.4%). Other authors, however, have reported lung cancer[
32‐
35] and GI malignancies[
36,
37] as the most common cancer subtypes associated with SSc. The variability in the reported incidence of malignancies and common cancer subtypes associated with SSc in these other studies might be due to differences in study design, case ascertainment, cancer prevalence in the general population studied, ethnicity and gender distribution. These factors may account for the high standardized incidence ratios reported for some cancer subtypes, including lung cancer. The magnitude of absolute risk for these individual cancers is likely to be low[
32,
38,
39]. For example, the recent meta-analysis reported by Bonifazi
et al.[
20] included a large number of patients from a heterogeneous cohort and case–control retrospective studies, and the authors did not find an increased risk of breast cancers, although the results of some other studies have implicated a strong temporal clustering of breast cancer with SSc onset[
40,
41].
The association of breast cancer with SSc may be attributed to the potential role of sex hormones in disease development of SSc and breast cancer. Increased levels of prolactin and low dehydroepiandrosterone in patients with SSc[
42,
43] and breast cancer[
44] lend further support to this association. It is also noteworthy that the heterogeneity of SSc may preclude a definitive conclusion regarding the association of SSc and cancer.
A variety of potential reasons have been postulated for the apparent association of SSc and cancer. These include shared risk factors such as gender distribution, exposure to immunosuppressive drugs and possible shared genetic backgrounds of both disorders. The precise pathogenic mechanisms that cause and maintain an autoimmune response in patients with cancer has not been fully elucidated. Autoimmunity in cancer may occur as a consequence of an abnormal self-antigen expression and/or mutated antigens, as well as of tumour-derived biological substances produced by cancer cells. In addition, the consequences of chronic inflammation and fibrotic processes may damage tissue, cells and DNA, leading to altered immune responses and the development of malignant transformation, triggering the development of lung cancer or cutaneous malignancies. Dysregulation of molecular signalling pathways involving transforming growth factor β and SMAD pathways have been reported to induce fibrosis and also tumourigenesis[
45]. In a recent study, the investigators suggested that both DNA methylation and histone modification may contribute to excessive synthesis of extracellular matrix proteins in SSc and that an imbalance may lead to autoimmune diseases and also cancer[
46].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
PM, VO and CD conceived and designed the study. PM, CF, CC, CD, AS and VO were involved in data acquisition. MH made particular contributions to the statistical analyses and interpretation of data. All authors contributed to the analysis and interpretation of the data and jointly wrote and approved the final manuscript.