Abstract
Objective:
This study investigates cancer risk in idiopathic inflammatory myopathy (IIM) subgroups relative to the general population.
Methods:
We conducted a single center, retrospective cohort study of IIM patients and malignancy. Myositis-specific and -associated autoantibodies were determined by Euroimmun line blot, ELISA and immunoprecipitation. The Surveillance, Epidemiology, and End Results registry was used to compare the observed number of cancers to that expected in the general population, defined as a standardized prevalence ratio (SPR), adjusted for calendar year, age, sex, race, and ethnicity.
Results:
1172 IIM patients, 203 (17%) with a cancer history, were studied. From cohort enrollment over a median follow-up of 5.2 years, the observed number of IIM patients diagnosed with cancer was increased 1.43-fold, SPR 1.43 (95%CI 1.15–1.77), p=0.002. Within 3 years of IIM-symptom onset, an increased SPR was observed for anti-TIF1γ-positive patients for ovarian and breast cancer (ovarian SPR 18.39 (95%CI 5.01–47.08), p<0.001; breast SPR 3.84 (95%CI 1.99–6.71), p<0.001). While anti-TIF1γ expectedly had a significantly elevated SPR, only 55% (36/66) of all cancers within 3 years of DM-onset were observed in TIF1γ-positive patients. Other MSAs including anti-Mi2, -SAE, and -NXP2 accounted for the majority of the remaining cancer diagnoses, 17/66 (26%), within 3 years of DM-onset. No cancer association, positive or negative, was observed for patients with anti-synthetase, anti-MDA5, or anti-HMGCR antibodies.
Conclusion:
In a tertiary referral center population, anti-TIF1γ was most strongly associated with breast and ovarian cancer. Patients with anti-synthetase, anti-MDA5, or anti-HMGCR antibodies had the same cancer risk as the general population.
Keywords: autoantibodies, myositis, dermatomyositis, epidemiology
Graphical Abstract
INTRODUCTION
The association between specific subgroups of idiopathic inflammatory myositis (IIM) and contemporaneous cancer has long been recognized(1,2). The initial signal of such an association originated from countries with nationalized healthcare systems and cancer registries. However, the majority of these early studies lacked data on myositis-specific autoantibodies (MSAs), a key filter for addressing myositis disease heterogeneity(3). Over the past decade, MSAs, specifically anti-TIF1γ and anti-NXP2, have emerged as crucial tools that facilitate stratification for the occurrence of contemporaneous cancer(4,5). While data suggest that patients with high-risk autoantibodies, in particular anti-TIF1γ and anti-NXP2, should be assessed thoroughly for cancer, several factors limit targeted cancer surveillance: (i) only a subgroup of patients with high-risk autoantibodies are diagnosed with cancer(6); (ii) it remains unknown whether contemporaneous cancer sites differ among types of MSAs; and (iii) IIM patients without these autoantibodies may also have contemporaneous cancer(7,8). Consequently, the prevailing practice is to regularly assess the majority of IIM patients for cancer.
Prior studies focusing on this problem have often been limited by lack of a general population reference group, inadequate sample size for MSAs with low prevalence, imperfect cancer ascertainment, and suboptimal sensitivity and specificity of autoantibody assay techniques. The use of a general population reference group in particular is crucial for determining excess cancer risk, as it is this excess risk that would warrant intensive cancer surveillance strategies.
In this context, we investigated whether the number of IIM patients diagnosed with cancer differs from that expected based on the general population by IIM subgroup and autoantibody type. We used data from a large, single-center, tertiary referral IIM cohort with systematic and comprehensive autoantibody profiling and cancer ascertainment. Reference cancer rates were obtained from the US Surveillance Epidemiology, and End Results Program. The goal of this study was to provide new data for informing cancer surveillance strategies in IIM patients.
METHODS
Patients
The Johns Hopkins Myositis Cohort was retrospectively reviewed from 2003–2020 for patients who met any of the following criteria: (i)Probable or definite dermatomyositis (DM)/polymyositis (PM) by Peter and Bohan(9), (ii)Immune-mediated necrotizing myopathy (IMNM) by the 2003 ENMC Criteria(10), (iii)Patients with a classic DM rash (Gottron’s or heliotrope) and consistent histopathology on skin biopsy to capture those with amyopathic disease. Patients with clear diagnoses of muscular dystrophy, metabolic or mitochondrial myopathy, or inclusion body myositis (IBM) were excluded, even if they technically met criteria for probable/definite PM. All patients were >18 years old at the time of first IIM symptom. Given that the majority of patients were diagnosed with IIM prior to 2017, we could not assess the percentage of patients who met the ACR/EULAR 2017 Classification Criteria for IIM(11). A flowsheet of patient selection for the analysis can be found in the Supplemental Materials Figure 1.
Symptom onset was defined as the first patient-reported IIM symptom limited to the following: arthralgia, dyspnea, myalgia, weakness, or rash consistent with DM. For each patient, the end of follow-up was the date of last visit/patient contact, the first cancer after IIM onset, or 12/31/2020, whichever came first. The Johns Hopkins IRB approved the protocol, and all patients provided informed consent.
Autoantibody assays
As part of the study, all patients underwent Euroimmun line blot testing (EUROLINE Autoimmune Inflammatory Myopathies 16 Ag IgG, Lubeck, Germany) for myositis-specific and associated autoantibodies. Based on prior published data comparing Euroimmun results to Clinical Laboratory Improvement Amendments (CLIA)-certified laboratories and in-house immunoprecipitation (IP) assays, we used the following thresholds to define antibody positivity in the Euroimmun assay: The cut-off of 15/+ was used to define anti-Jo-1, anti-Mi2, anti-PM/Scl and anti-MDA5-positive sera. For anti-PL-12, -PL-7, -OJ, -EJ, -SRP, -Ku, -SAE, and -Ro52, ≥36/++ was used as the cutoff for a positive antibody status. The use of a higher threshold for several of the autoantibodies was defined in prior research demonstrating a lower cut-off (≥15/+) results in false positives (that is, positive results by Euroimmun with an inconsistent clinical phenotype and/or negative testing by immunoprecipitation)(12–19). Anti-TIF1γ was considered positive by two thresholds: >32 units per manufacturer (MBL), and >6 units (3 standard deviations above the mean of a healthy control cohort, as previously described(14)). Anti-HMGCR autoantibodies were assessed by ELISA (Inova Diagnostics, San Diego, USA). Additional methodology regarding autoantibody assays, including immunoprecipitation for anti-NXP2 and the handling of discordant results for multicomponent complexes (e.g. PM75 and PM100), can be found in the Supplemental Materials.
Cancer Ascertainment
Our center’s general practice is to obtain at least one CT chest, abdomen and pelvis scan upon IIM diagnosis, and to ensure patients maintain age and sex-specific cancer screening throughout follow-up. Every patient was contacted by research staff by either phone or online survey to update their cancer status and assess cancer risk factors using a systematic script. If patients could not be reached, two sources were required to consider a patient “cancer-free” (see Supplement). For information on cancer site, grade, and stage (American Joint Committee on Cancer, AJCC), information was abstracted directly from pathology reports and oncology notes for the majority (80%) of patients with cancer history. The remaining 20% of patients had cancer information abstracted directly from rheumatologist or primary care physician records. In patients with multiple cancers, the most proximal cancer to IIM-symptom onset was included for analyses. Non-melanoma skin cancers (basal and squamous cell carcinomas) were excluded from all analyses.
Statistical Analyses
Due to the inherent difficulties in determining whether a clinically diagnosed cancer truly emerged before or after IIM symptom onset, we considered all cancers diagnosed to be either prevalent or a mixture of prevalent and incident relative to IIM-symptom onset. Epidemiologically, a cancer that is diagnosed one month after IIM symptom onset would be viewed as incident. However, biologically, this cancer likely existed well before IIM symptoms began, given that the pathogenesis of many cancers takes years, sometimes decades, to develop. Thus, the cancer could also be referred to as ‘prevalent’ as it was likely already present prior to IIM symptom onset, only not formally diagnosed. Therefore, we estimated standardized incidence ratios, but refer to them as standardized prevalence ratios (SPRs), as the measure to assess whether cancer diagnosis differed from that expected in the general population over the follow-up time. SPRs were determined by comparing cancer occurrence in our IIM cohort to that of the Surveillance, Epidemiology, and End Results (SEER) registry, a nationally representative sample of the US population(23). This registry allows for adjusting for age, sex, race, ethnicity, and calendar year when comparing cancer incidence rates. All SEER analyses included in situ malignancies for all cancer sites. Using these two cohorts, the observed/expected period prevalence (SPR) of cancer was calculated in three time windows: Time of entry into the Johns Hopkins Myositis Cohort onwards, time from first IIM-symptom onwards, and within ±3 years of first IIM symptom. Patients with a history of cancer prior to the analysis window (−3 years from IIM-symptom onset, IIM-symptom onset, or cohort enrollment) were excluded from analysis. A Bonferroni correction was applied for multiple comparison testing. To find the 95% confidence limits, we followed standard procedure(24).
RESULTS
Patient demographics, IIM subgroup, and Autoantibody Prevalence
Patient demographics, IIM subgroup, autoantibody data, other clinical characteristics, and IIM-cancer interval for the 1172 patients in the cohort are reported in Table 1. The median age of IIM-symptom onset in our cohort was 49 years (IQR 38–59), and 71% were female. 69% of the cohort was White and 20% were Black. The time from IIM-symptom onset to cohort enrollment was a median of 1.6 yrs (IQR 0.7–3.8), and median follow-up duration was 5.2 yrs (IQR 2.7–8.7). Regarding IIM subgroup, 727 patients had DM, and 445 had PM, of which 109 were overlap conditions (59 systemic sclerosis, 14 rheumatoid arthritis, 18 Sjogren’s syndrome, and 18 systemic lupus erythematosus). A total of 240 patients had antisynthetase syndrome (ASyS) and 259 patients had IMNM. Anti-TIF1γ was the most common myositis-specific antibody (n=270/726 (37%) using an ELISA cutoff of >6 units), followed by anti-Jo-1 (n=165/1172, 14%) and anti-HMGCR (n=160/1172, 14%). Anti-Ro52 was the most common autoantibody overall (myositis-specific or associated, n=304/1172, 26%).
Table 1. Patient demographics, IIM subgroup, autoantibody data, and IIM-cancer interval.
Of 203 patients with a history of cancer, 199 had a confirmed diagnosis date.
| Demographics and Clinical Characteristics | N = 1172 |
|---|---|
| median (IQR)/% | |
| Age at IIM onset, median (IQR) | 49 (38–59) |
| Age at Cohort Enrollment, median (IQR) | 53 (43–62) |
| Time from IIM onset to Cohort Enrollment (IQR), yrs | 1.6 (0.7–3.8) |
| Follow-up duration in Cohort (IQR), yrs | 5.2 (2.7–8.7) |
| Sex (Female) | 828 (71) |
| Race | |
| Asian | 40 (3.4) |
| Black | 238 (20) |
| White | 806 (69) |
| Declined/Unknown | 29 (2.5) |
| Other | 63 (5.3) |
| IIM Subgroup | |
| Dermatomyositis | 727/1172 |
| Polymyositis | 445/1172 |
| Immune-mediated necrotizing myopathy | 259/1172 |
| Antisynthetase syndrome | 240/1172 |
| Overlap myositis | |
| Systemic sclerosis | 59 |
| Rheumatoid arthritis | 14 |
| Sjogren’s syndrome | 18 |
| Systemic lupus erythematosus | 18 |
| Myositis-specific/associated autoantibodies | |
| TIF1γ (MBL>32) | 171/726* |
| TIF1γ (MBL>6) | 270/726* |
| NXP2 (IVTT IP) | 80/725* |
| SAE (EBO) | 19/1172 |
| Mi2 (EBO, ELISA) | 75/1172 |
| MDA5 (EBO) | 68/1172 |
| Jo1 (EBO) | 165/1172 |
| PL12 (EBO) | 33/1172 |
| PL7 (EBO) | 35/1172 |
| OJ (EBO) | 7/1172 |
| EJ (EBO) | 8/1172 |
| SRP (EBO) | 76/1172 |
| HMGCR (IP, ELISA) | 160/1172 |
| PM-Scl (EBO) | 41/1172 |
| Ku (EBO) | 26/1172 |
| Ro52 (EBO) | 304/1172 |
| Cancer (ever) | 203 (17) |
| Cancer (within 5 years of IIM onset) | 123 (11) |
| Cancer (within 3 years of IIM onset) | 93 (8) |
| Cancer (within 1 year of IIM onset) | 48 (4) |
| Cancer after cohort enrollment | 92/203 |
While 727 patients had DM, sera ran out before TIF1γ and NXP2 autoantibodies could be assayed (n=1 and 2, respectively). Thresholds for autoantibody positivity can be found in the Methods section.
Of 1172 patients, 203 (17%) had a history of cancer. Of the 203 cancers, cancer diagnosis dates were confirmed in 199, of which 123 (62%) were diagnosed within ±5 years of IIM-symptom onset, 93 (47%) within ±3 years, and 48 (24%) within ±1 year.
Cancer Stage, Site, and Risk Factors
Cancer types and locations are listed in Table 2, along with cancer risk factors. Of the 203 cancers, stage information was available for 151 (74%). The majority of cancers were diagnosed either in situ or stage 1 (27 and 56, respectively), accounting for 55% of the cancers with known staging. Breast cancer was the most common (23.6%), followed by prostate (9.5%), melanoma (8.5%), lung (7.5%), and cervical/uterine (7%). Additional data on cancer risk factors can be found in the Supplemental Materials.
Table 2. Details on cancer type, stage, and risk factors.
Data reported on 199 of the 203 cancers with complete data on diagnosis date. Non-melanoma skin cancer included basal or squamous cell carcinoma. Cancer risk factors were obtained directly from telephone interviews with patients and the electronic medical record. Miscellaneous cancers included anal (n=2), stomach (n=2), testis (n=2), vulva (n=2), and one of each of the following: esophagus, gallbladder, larynx, peritoneum, small intestine, eye, soft tissue, and 8 which were unclassified.
| Cancer Type (n=199) | N (%) |
|---|---|
| Breast | 47 (23.6%) |
| Prostate | 19 (9.5%) |
| Melanoma | 17 (8.5%) |
| Lung | 15 (7.5%) |
| Cervical/Uterine | 14 (7.0%) |
| Thyroid | 12 (6.0%) |
| Ovarian | 10 (5.0%) |
| Kidney/renal pelvis | 10 (5.0%) |
| Colorectal | 9 (4.5%) |
| Lymphoma | 8 (4.0%) |
| Oral cavity/pharynx | 8 (4.0%) |
| Leukemia/Myeloma | 3 (1.5%) |
| Other endocrine (non-thyroid) | 3 (1.5%) |
| Miscellaneous | 24 (12.1%) |
| AJCC Stage at Diagnosis | |
| 0 (in situ) | 27 |
| 1 | 56 |
| 2 | 28 |
| 3 | 21 |
| 4 | 19 |
| Unknown/Unstaged | 52 |
| History of Smoking | 418/1172 (36%) |
| History of regular EtOH use | 366/1172 (31%) |
| # of drinks/week (median, IQR) | 3 (1–5) |
| History of statin exposure | 358/1172 (30%) |
| BRCA Positive | 15/77 (19%)* |
| Age of menarche (median, IQR) | 13 (12–13) |
| Parity (median, IQR) | 2 (0–2) |
| Age of menopause (n=552 in year 2020) | 48 (IQR 38–51) |
| History of breastfeeding | 299/828 (36%) |
| History of oral contraceptive use | 411/828 (50%) |
| History of hormone replacement therapy | 126/828 (15%) |
77 of the 1172 patients were tested for BRCA mutations
Time from Cohort Enrollment to End of Follow-up
SPRs for cancers diagnosed in the time window from cohort enrollment to end of follow-up are shown in Table 3. For all cancers combined, the observed number of cancers diagnosed in IIM patients was larger than expected, SPR 1.43 (1.15–1.77), p=0.002, which was statistically significant (p<0.003) after adjustment for multiple comparisons. This same pattern was noted for all three IIM types - DM, PM and IMNM (all p<0.05). By autoantibody strata, SPRs for cancer were consistently above 1.0 for the DM-specific autoantibodies including anti-TIF1γ, -Mi2, -NXP2, and -SAE (Table 3).
Table 3. Observed and expected numbers of cancer diagnosed from Johns Hopkins Cohort Enrollment onwards.
Person-years, expected and observed cancers, and SPR for IIM strata – both IIM subgroup and autoantibody strata.
| Cancer site | IIM Subgroup/Antibody | Person-years | No. observed | No. expected | SPR (95% CI) | p-value | N |
|---|---|---|---|---|---|---|---|
| All | All | 6464 | 88 | 61.4 | 1.43 (1.15–1.77) | 0.002** | 1059 |
| DM | 4031 | 50 | 35.9 | 1.39 (1.03–1.84) | 0.029 | 655 | |
| PM | 2433 | 38 | 25.5 | 1.49 (1.05–2.04) | 0.025 | 404 | |
| IMNM | 1366 | 26 | 15.7 | 1.66 (1.08–2.43) | 0.021 | 231 | |
| TIF1γ>6 | 1459 | 23 | 13.5 | 1.70 (1.08–2.56) | 0.023 | 236 | |
| TIF1γ>32 | 897 | 16 | 8.2 | 1.96 (1.12–3.18) | 0.020 | 149 | |
| EJ|Jo1|PL7|PL12|OJ | 1409 | 16 | 12.4 | 1.29 (0.73–2.09) | 0.380 | 226 | |
| Mi2 | 419 | 7 | 4.6 | 1.51 (0.61–3.12) | 0.370 | 66 | |
| NXP2 | 404 | 6 | 3.5 | 1.71 (0.63–3.71) | 0.290 | 70 | |
| SAE | 105 | 4 | 1.2 | 3.38 (0.92–8.66) | 0.064 | 17 | |
| MDA5 | 392 | 3 | 2.9 | 1.03 (0.21–3.00) | 0.990 | 64 | |
| Mi2|NXP2|SAE | 928 | 17 | 9.3 | 1.82 (1.06–2.92) | 0.030 | 153 | |
| HMGCR | 888 | 16 | 11.8 | 1.35 (0.77–2.19) | 0.290 | 142 | |
| SRP | 361 | 2 | 2.8 | 0.71 (0.09–2.55) | 0.920 | 68 | |
| PM-Scl | 261 | 0 | 2.1 | 0.00 (0.00–1.78) | 0.250 | 41 | |
| Ku | 126 | 1 | 1.3 | 0.79 (0.02–4.40) | 0.990 | 23 |
N=number of patients in strata. Green and orange shaded bars are used to indicate p-values less than the standard and Bonferroni-corrected values of 0.05 and 0.003 (0.05/16), respectively.
The observed number of oral cavity/pharynx cancers (SPR 5.59 [2.25–11.51] p<0.001) was significantly greater than expected. SPRs for both thyroid cancer (3.04 [0.99–7.09], p=0.05) and lymphoma (2.71 [0.99–5.90], p=0.05) were notably greater than 1.0 and approached statistical significance. Among DM patients, SPRs for several cancer sites approached statistical significance: ovarian (SPR 5.04 [1.04–14.73], p=0.045, lymphoma (SPR 3.91 [1.27–9.12], p=0.020), and thyroid (SPR 3.69 [1.01–9.54], p=0.049).
The number of ovarian cancer diagnoses was greater than expected in patients with anti-TIF1γ autoantibodies using a cutoff of >32 U by ELISA (SPR 19.99 [4.12–58.43], p=0.001). Of note, in patients who were negative for anti-TIF1γ autoantibodies (as defined by either <7 U or <33 U by ELISA), no increase in cancer SPR was noted. Similarly, patients positive for anti-synthetase or anti-HMGCR antibodies had no increase or decrease in SPR for any cancer type.
Time from IIM-symptom onset onwards
Similar patterns were observed when using time from IIM-symptom onset to the end of follow-up rather than time from cohort enrollment to the end of follow-up, with some notable differences: The SPR for cancer appeared to be larger in those with DM (SPR 1.68 [1.33–2.08], p<0.001) or IMNM (SPR 1.56 [1.08–2.18], p=0.019), and in patients with DM-specific autoantibodies [Supplemental Table 1]; a significant p-value after Bonferroni correction was <0.003.
Time from 3 years before to 3 years after IIM-symptom onset
To better understand the association between IIM and cancer, we studied the 6-year time window surrounding IIM-symptom onset (Table 4). Cancer was most common among patients with DM, in particular among those with anti-TIF1γ antibodies. However, only 38% (25/66) or 55% (36/66) of the cancers diagnosed in patients with DM occurred in anti-TIF1γ-positive patients (using TIF1γ ELISA cutoffs of >32 U or >6 U, respectively). Other DM-specific MSAs including anti-Mi2, -SAE, and -NXP2 were accounted for the majority 17/66 (26%) of the remaining cancer diagnoses within 3 years of DM-onset. The expected and observed cancers, stratified by IIM subgroup and autoantibody from −3 years from IIM symptom onward, are shown in Figure 1. Individual cancer types over the same time period are shown in Supplementary Figure 2. Within 3 years of IIM-symptom onset, several cancer sites were enriched in IIM patients, including breast (SPR 2.10 [1.35–3.13], p<0.002) and ovarian (SPR 7.65 [2.81–16.65], p<0.001), both of which were statistically significant (p<0.003) after adjustment for multiple comparisons.
Table 4. Observed and expected cancer diagnosed from −3 to +3 years surrounding IIM-symptom onset.
Person-years, expected and observed cancers, and SPR for IIM strata – both IIM subgroup and autoantibody strata.
| Cancer site | IIM Subgroup/Antibody | Person-years | No. observed | No. expected | SPR (95% CI) | p-value | N |
|---|---|---|---|---|---|---|---|
| All | All | 6576 | 94 | 44.7 | 2.10 (1.70–2.57) | <0.001** | 1125 |
| DM | 4084 | 66 | 26.3 | 2.51 (1.94–3.20) | <0.001** | 701 | |
| PM | 2492 | 28 | 18.4 | 1.52 (1.01–2.19) | 0.045 | 424 | |
| IMNM | 1432 | 22 | 11.3 | 1.94 (1.22–2.94) | 0.006 | 244 | |
| TIF1γ>6 | 1504 | 36 | 10.3 | 3.48 (2.44–4.82) | <0.001** | 262 | |
| TIF1γ>32 | 950 | 25 | 6.4 | 3.88 (2.51–5.73) | <0.001** | 167 | |
| TIF1γ<=6 | 2532 | 30 | 15.7 | 1.91 (1.29–2.73) | 0.002** | 431 | |
| TIF1γ<=32 | 3086 | 41 | 19.6 | 2.09 (1.50–2.84) | <0.001** | 526 | |
| EJ|Jo1|PL7|PL12|OJ | 1379 | 10 | 8.8 | 1.14 (0.55–2.10) | 0.760 | 233 | |
| Mi2 | 407 | 8 | 3.2 | 2.52 (1.09–4.96) | 0.032 | 71 | |
| NXP2 | 423 | 5 | 3 | 1.65 (0.54–3.85) | 0.380 | 72 | |
| SAE | 109 | 4 | 1 | 4.04 (1.10–10.34) | 0.037 | 19 | |
| MDA5 | 388 | 3 | 2.2 | 1.37 (0.28–4.01) | 0.750 | 66 | |
| Mi2|NXP2|SAE | 938 | 17 | 7.1 | 2.40 (1.40–3.84) | 0.002** | 162 | |
| HMGCR | 878 | 13 | 8.4 | 1.54 (0.82–2.63) | 0.176 | 150 | |
| SRP | 425 | 6 | 2.3 | 2.64 (0.97–5.75) | 0.057 | 73 | |
| PM-Scl | 246 | 0 | 1.2 | 0.00 (0.00–3.11) | 0.610 | 41 | |
| Ku | 143 | 3 | 0.9 | 3.17 (0.65–9.26) | 0.141 | 25 |
N=number of patients in strata. Green and orange shaded bars are used to indicate p-values less than the standard and Bonferroni-corrected values of 0.05 and 0.00278 (0.05/18), respectively.
Figure 1. Observed and Expected Cancers stratified by IIM subgroup and myositis-specific autoantibodies over time.
In each graph, the x-axis reflects time from myositis symptom onset (defined as time zero). The top row depicts the observed number of cancers over time (blue line) in comparison with the number of cancer cases expected based on SEER data (red line). The (#) corresponds to the total number of cancers diagnosed over follow-up. The middle row depicts the associated SPR binned in the six-year period −3 to +3. Yellow and orange shaded bars are used to indicate SPR in six-year periods with p-values less than the standard and Bonferroni-corrected values of 0.05 and 0.00139 (0.05/36), respectively. The bottom row shows the observed cumulative probability of a cancer diagnosis over time (blue line) versus that expected based on the general population using SEER data (red line).
SPRs for several other cancer sites approached statistical significance, including cancers of the kidney/renal pelvis (4.13 [1.34–9.63], p=0.016), thyroid (SPR 4.29 [1.58–9.35], p=0.006), and cancers of the oropharynx (SPR 4.38 [1.19–11.21], p=0.028) [Supplemental Table 2]; the significance threshold for multiple comparisons was p<0.004.
Anti-TIF1γ-positive Patients and Association with Specific Cancer Types
Given the known association between anti-TIF1γ and contemporaneous cancer, we explored the site-specific cancer diagnoses among anti-TIF1γ positive patients in the 6 years surrounding IIM-symptom onset (Table 5). Using the >6-unit ELISA cut-off, breast and ovarian cancers were most enriched, whereas using the >32-unit ELISA cut-off, only ovarian cancer achieved statistical significance (p<0.004 after adjustment for multiple comparison testing), although trends were present for breast, lymphoma, and cancers of the oropharynx. Within 6 years surrounding IIM-onset, no patients with anti-TIF1γ antibodies were diagnosed with prostate cancer, leukemia, myeloma, or renal cancer. Supplementary Figure 3 shows the numbers of observed and expected cancers by site among anti-TIF1γ-positive patients stratified by cancer type from −3 years from IIM-onset onwards.
Table 5. Observed and expected cancer diagnosed from −3 to +3 years surrounding anti-TIF1γ-DM symptom onset.
Person-years, expected and observed cancers.
| Cancer site | No. observed | No. expected | SPR (95% CI) | p-value |
|---|---|---|---|---|
| All | 94 | 44.7 | 2.10 (1.70–2.57) | <0.001** |
| Anti-TIF1γ>6, Person-years: 1504, Total Patients: 262 | ||||
| All | 36 | 10.3 | 3.48 (2.44–4.82) | <0.001** |
| Breast | 12 | 3.1 | 3.84 (1.99–6.71) | <0.001** |
| Lung | 2 | 0.9 | 2.14 (0.26–7.73) | 0.48 |
| Colorectal | 1 | 0.7 | 1.47 (0.04–8.19) | 0.99 |
| Melanoma | 3 | 0.8 | 3.83 (0.79–11.19) | 0.09 |
| Ovarian | 4 | 0.2 | 18.39 (5.01–47.08) | <0.001** |
| Cervical/uterine | 2 | 0.7 | 2.97 (0.36–10.72) | 0.29 |
| Prostate | 0 | 0.5 | 0.00 (0.00–7.05) | 0.99 |
| Kidney/renal pelvis | 0 | 0.2 | 0.00 (0.00–14.86) | 0.99 |
| Lymphoma | 2 | 0.4 | 5.36 (0.65–19.37) | 0.109 |
| Thyroid | 2 | 0.4 | 5.32 (0.64–19.22) | 0.11 |
| Oral cavity/pharynx | 2 | 0.2 | 11.06 (1.34–39.94) | 0.029 |
| Leukemia or myeloma | 0 | 0.3 | 0.00 (0.00–12.54) | 0.99 |
| Anti-TIF1γ>32, Person-years: 950, Total Patients: 167 | ||||
| All | 25 | 6.4 | 3.88 (2.51–5.73) | <0.001** |
| Breast | 7 | 2 | 3.45 (1.39–7.11) | 0.01 |
| Lung | 2 | 0.6 | 3.56 (0.43–12.86) | 0.22 |
| Colorectal | 0 | 0.4 | 0.00 (0.00–8.99) | 0.99 |
| Melanoma | 3 | 0.5 | 5.98 (1.23–17.48) | 0.029 |
| Ovarian | 3 | 0.1 | 20.88 (4.31–61.03) | <0.001** |
| Cervical/uterine | 2 | 0.4 | 4.60 (0.56–16.62) | 0.142 |
| Prostate | 0 | 0.3 | 0.00 (0.00–12.56) | 0.99 |
| Kidney/renal pelvis | 0 | 0.1 | 0.00 (0.00–24.71) | 0.99 |
| Lymphoma | 2 | 0.2 | 8.70 (1.05–31.44) | 0.045 |
| Thyroid | 1 | 0.2 | 4.12 (0.10–22.97) | 0.43 |
| Oral cavity/pharynx | 2 | 0.1 | 18.47 (2.24–66.71) | 0.011 |
| Leukemia or myeloma | 0 | 0.2 | 0.00 (0.00–21.00) | 0.99 |
Green and orange shaded bars are used to indicate p-values less than the standard and Bonferroni-corrected values of 0.05 and 0.00385 (0.05/13), respectively.
DISCUSSION
In a retrospective cohort of myositis patients from a tertiary care facility, we found that contemporaneous occurrence of cancer and site-specific cancers in IIM patients relative to the general population varied greatly depending on IIM subgroup and autoantibody specificity. While anti-TIF1γ antibodies were most strikingly associated with contemporaneous cancer, this finding was also significant for other DM-specific autoantibodies (anti-Mi2, -SAE, and -NXP2). In contrast, cancer occurrence was not greater than expected in patients with antibodies against HMGCR, MDA5, and the anti-synthetases. Cancer sites most enriched in our cohort include ovarian, breast, and oropharyngeal cancers. Taken together, these new findings suggest that intensive cancer surveillance approaches may not be needed in all IIM types, and may help to inform evidence-based clinical guidelines for cancer assessment in newly diagnosed IIM patients.
Our study addresses several limitations present in earlier research. The current study cohort consisted of 1172 IIM patients – a large number compared to the average IIM cohort size of ~300 patients, as reported in a recent meta-analysis of IIM-cancer studies(25). Comprehensive, systematic serotyping was performed on all sera, using a combination of myositis-specific and associated antibody assays. Importantly, all 1172 patients in our cohort were contacted directly to ascertain their cancer status. Rather than performing within cohort comparisons, the general US population was used as a comparator group to determine excess cancer risk that may warrant more intensive cancer surveillance strategies. These aspects of our study design strengthen its conclusions.
In our cohort, breast and ovarian cancer were most enriched in patients with anti-TIF1γ antibodies. Similarly, cancer (all types) was enriched in anti-Mi2, -SAE, and -NXP2-positive patients, but not anti-MDA5-positive patients. These findings confirm those by others, who used as a reference population both IIM patients(26) and the general population(8,27). Our study also highlights an increase in oropharyngeal cancer in IIM patients, as has been noted by others(28). Due to the small number of oropharyngeal cancers (n=8), we were unable to conclude whether this reflects a consequence of long-term immunosuppression or other risk factors such as alcohol and tobacco exposure.
Of equal importance to the MSAs with higher cancer SPRs are those with no increase in cancer occurrence compared to the general population, including anti-HMGCR, anti-MDA5 and the antisynthetase autoantibodies. Prior studies looking within IIM cohorts have suggested that patients with antisynthetase antibodies are protected from cancer(25). Of note, our new findings indicate that these patients do not have a lower risk of cancer, but rather one that is equivalent to the expected rate in the general population. Our data suggest that age, sex and risk factor appropriate cancer detection strategies are indicated for patients with anti-synthetase antibodies, and that intensive cancer surveillance strategies, such as those often used in patients with IIM, may not be justified and may unintentionally lead to excessive healthcare cost, radiation exposure, and downstream harms of false positive testing.
Our reporting of cancer SPRs within three distinct IIM time windows provides insight into different questions pertaining to the cancer-IIM relationship. The time from cohort enrollment onward is the most rigorous from an epidemiologic perspective, and reflects the experience of the patient population referred to a large academic medical center. It is noteworthy that in this time window, all IIM types had a significantly greater number of cancer cases observed than expected, in particular ovarian and cancers of the oropharynx. The lack of significantly increased SPRs for individual autoantibodies may reflect the exclusion of cancer events that occurred prior to cohort enrollment, as the median disease duration for patients upon cohort entry was 1.6 years.
The six-year period surrounding IIM-symptom onset offers a view into the possible etiologic relationship between cancer and IIM, given this is the time period where the largest number of cancers are diagnosed. Specific subgroups of IIM (e.g. anti- TIF1γ-positive DM) may be a manifestation of an anti-tumor immune response gone awry, referred to as ‘cancer-induced autoimmunity’(29), which could account for the often-observed Gaussian distribution of cancer diagnoses around IIM onset/diagnosis. Cancer-induced autoimmunity has been demonstrated in another rheumatic disease, anti-RNA polymerase III-positive systemic sclerosis (SSc)(30), where a serologically-defined subgroup of SSc patients develop contemporaneous cancer(31). Our findings suggest that DM-specific autoantibodies (anti-TIF1γ, -NXP2, -SAE, and -Mi2) may be biomarkers of a similar process. In our cohort, there was no evidence that anti-MDA5-positive patients have an increase in contemporaneous cancer.
There are important limitations to our work. As a single center tertiary referral population, the generalizability of our results may be limited. Additional myositis centers and/or consortia are needed to validate these findings, especially as some of the associations observed are derived from relatively few cancer cases. For our analyses focusing on the time windows ‘IIM-onset onward’ and ‘3 years before to 3 years after IIM-symptom onset’, we included person-time/cancers that occurred prior to enrollment into our myositis cohort. In these analyses, some patients in particular IIM subgroups (e.g. anti-TIF1γ) may have developed cancer but never entered our cohort (due to incapacity or death), whereas other patients in the same subgroup who did not have cancer/less advanced cancer could be preferentially enrolled. This may have the effect of falsely lowering cancer SPR. Given this potential bias, our results are likely more applicable to tertiary centers who care for a similar patient population (e.g. a mix of new IIM patients and referrals for second opinions of established IIM patients). While we examined cancer risk around the time of IIM-onset to better approximate what may be observed in community practice, we recognize community practitioners may see a different patient population, and thus further studies are warranted in this clinical setting. We also believe these analyses have value, in that they allow for hypothesis generation about patient subsets that may have an underlying mechanism of cancer-induced autoimmunity. Another limitation of our study is the risk of autoantibody misclassification, since antibody assays were performed upon cohort entry, after patients may have received immunosuppressive therapy. Given reports of decreasing autoantibody titers over time (32–33), it is possible that some patients have been classified as autoantibody “negative” when in fact they were positive at the time of IIM onset. However, given the relatively short duration between IIM onset and cohort enrollment (median time of 1.6 years), we believe the risk of misclassification is low. In addition, diagnostic suspicion bias may influence the number of cancers detected in IIM patients, and may artificially inflate the SPR. This may be the case given the large percentage of cancers in our cohort that were in situ or stage I at diagnosis. Future work is warranted to determine whether this reflects a diagnostic suspicion bias, or whether the large number of low-stage cancers suggests an anti-tumor response preventing late-stage/aggressive cancer presentations.
In conclusion, we report varying cancer associations within a large, comprehensively and systematically phenotyped cohort of IIM patients compared to the general population. Both IIM subgroup and MSAs provide value to stratify for the occurrence of contemporaneous cancer. Whereas DM patients have the strongest association with cancer, large differences exist in SPRs (anti-TIF1γ vs anti-MDA5). Patients with anti-HMGCR or antisynthetase antibodies are diagnosed with cancer at approximately the same rate as the general population. This data may help inform evidence-based clinical guidelines for cancer detection in newly diagnosed IIM patients, and suggests not all patients with IIM require an aggressive assessment for cancer.
Supplementary Material
Funding:
Funding for these studies was provided the Jerome L. Greene Foundation, the Donald B. and Dorothy L. Stabler Foundation and the Huayi and Siuling Zhang Discovery Fund. This study was supported in part by NIH grants P30-AR070254, R01 AR-073208 (to TI, AR, LCR, AAS), 1K23AR075898 (to C.M.), K23AR0739 (to J.J.P), 5T32AR048522-15 (to CR), K24 AR080217 (to AAS), and P30 CA006973. ALM is supported by the Intramural Research Program of the National Institute of Musculoskeletal and Skin Diseases of the National Institutes of Health.
REFERENCES
- 1.Sigurgeirsson B, Lindelof B, Edhag O, Allander E. Risk of cancer in patients with dermatomyositis or polymyositis. A population-based study. N Engl J Med. 1992;326(6):363–7. [DOI] [PubMed] [Google Scholar]
- 2.Hill CL, Zhang Y, Sigurgeirsson B, Pukkala E, Mellemkjaer L, Airio A, et al. Frequency of specific cancer types in dermatomyositis and polymyositis: a population-based study. Lancet. 2001;357(9250):96–100. [DOI] [PubMed] [Google Scholar]
- 3.Betteridge Z, McHugh N. Myositis-specific autoantibodies: an important tool to support diagnosis of myositis. J Intern Med. 2016;280(1):8–23. [DOI] [PubMed] [Google Scholar]
- 4.Chinoy H, Fertig N, Oddis CV, Ollier WE, Cooper RG. The diagnostic utility of myositis autoantibody testing for predicting the risk of cancer-associated myositis. Ann Rheum Dis. 2007;66(10):1345–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Fiorentino DF, Chung LS, Christopher-Stine L, Zaba L, Li S, Mammen AL, et al. Most patients with cancer-associated dermatomyositis have antibodies to nuclear matrix protein NXP-2 or transcription intermediary factor 1gamma. Arthritis Rheum. 2013;65(11):2954–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Trallero-Araguas E, Rodrigo-Pendas JA, Selva-O’Callaghan A, Martinez-Gomez X, Bosch X, Labrador-Horrillo M, et al. Usefulness of anti-p155 autoantibody for diagnosing cancer-associated dermatomyositis: a systematic review and meta-analysis. Arthritis Rheum. 2012;64(2):523–32. [DOI] [PubMed] [Google Scholar]
- 7.Legault D, McDermott J, Crous-Tsanaclis AM, Boire G. Cancer-associated myositis in the presence of anti-Jo1 autoantibodies and the antisynthetase syndrome. J Rheumatol. 2008;35(1):169–71. [PubMed] [Google Scholar]
- 8.Yang H, Peng Q, Yin L, Li S, Shi J, Zhang Y, et al. Identification of multiple cancer-associated myositis-specific autoantibodies in idiopathic inflammatory myopathies: a large longitudinal cohort study. Arthritis Res Ther. 2017;19(1):259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Bohan A, Peter JB. Polymyositis and dermatomyositis (first of two parts). N Engl J Med. 1975;292(7):344–7. [DOI] [PubMed] [Google Scholar]
- 10.Hoogendijk JE, Amato AA, Lecky BR, Choy EH, Lundberg IE, Rose MR, et al. 119th ENMC international workshop: trial design in adult idiopathic inflammatory myopathies, with the exception of inclusion body myositis, 10–12 October 2003, Naarden, The Netherlands. Neuromuscul Disord. 2004;14(5):337–45. [DOI] [PubMed] [Google Scholar]
- 11.Lundberg IE, Tjarnlund A, Bottai M, Werth VP, Pilkington C, de Visser M, et al. 2017 European League Against Rheumatism/American College of Rheumatology Classification Criteria for Adult and Juvenile Idiopathic Inflammatory Myopathies and Their Major Subgroups. Arthritis Rheumatol. 2017;69(12):2271–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Albayda J, Mecoli C, Casciola-Rosen L, Danoff SK, Lin CT, Hines D, et al. A North American Cohort of Anti-SAE Dermatomyositis: Clinical Phenotype, Testing, and Review of Cases. ACR Open Rheumatol. 2021;3(5):287–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Mecoli CA, Albayda J, Tiniakou E, Paik JJ, Zahid U, Danoff SK, et al. Myositis Autoantibodies: A Comparison of Results From the Oklahoma Medical Research Foundation Myositis Panel to the Euroimmun Research Line Blot. Arthritis Rheumatol. 2020;72(1):192–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Fiorentino DF, Gutierrez-Alamillo L, Hines D, Yang Q, Casciola-Rosen L. Distinct dermatomyositis populations are detected with different autoantibody assay platforms. Clin Exp Rheumatol. 2019;37(6):1048–51. [PMC free article] [PubMed] [Google Scholar]
- 15.Cavazzana I, Fredi M, Franceschini F. Semiquantitative analysis of line blot assay for myositis-specific and myositis-associated antibodies: a better performance? Ann Rheum Dis. 2019. [DOI] [PubMed] [Google Scholar]
- 16.Cavazzana I, Fredi M, Ceribelli A, Mordenti C, Ferrari F, Carabellese N, et al. Testing for myositis specific autoantibodies: Comparison between line blot and immunoprecipitation assays in 57 myositis sera. J Immunol Methods. 2016;433:1–5. [DOI] [PubMed] [Google Scholar]
- 17.Mahler M, Betteridge Z, Bentow C, Richards M, Seaman A, Chinoy H, et al. Comparison of Three Immunoassays for the Detection of Myositis Specific Antibodies. Front Immunol. 2019;10:848. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Hamaguchi Y, Kuwana M, Takehara K. Performance evaluation of a commercial line blot assay system for detection of myositis- and systemic sclerosis-related autoantibodies. Clin Rheumatol. 2020;39(11):3489–97. [DOI] [PubMed] [Google Scholar]
- 19.Tansley SL, Li D, Betteridge ZE, McHugh NJ. The reliability of immunoassays to detect autoantibodies in patients with myositis is dependent on autoantibody specificity. Rheumatology (Oxford). 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mecoli CA, Gutierrez-Alamillo L, Yang Q, Sampedro M, Woods A, Hummers LK, et al. PM-Scl and Th/To in systemic sclerosis: a comparison of different autoantibody assays. Clin Rheumatol. 2021;40(7):2763–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Espinosa-Ortega F, Holmqvist M, Alexanderson H, Storfors H, Mimori T, Lundberg IE, et al. Comparison of autoantibody specificities tested by a line blot assay and immunoprecipitation-based algorithm in patients with idiopathic inflammatory myopathies. Ann Rheum Dis. 2019;78(6):858–60. [DOI] [PubMed] [Google Scholar]
- 22.Tansley SL, Snowball J, Pauling JD, Lissina A, Kuwana M, Rider LG, et al. The promise, perceptions, and pitfalls of immunoassays for autoantibody testing in myositis. Arthritis Res Ther. 2020;22(1):117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence - SEER Research Data, 9 Registries, Nov 2020 Sub (1975–2018) - Linked To County Attributes - Time Dependent (1990–2018) Income/Rurality, 1969–2019 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, released April 2021, based on the November 2020 submission. .
- 24.Liddell FD. Simple exact analysis of the standardised mortality ratio. J Epidemiol Community Health. 1984;38(1):85–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Oldroyd AGS, Allard AB, Callen JP, Chinoy H, Chung L, Fiorentino D, et al. A systematic review and meta-analysis to inform cancer screening guidelines in idiopathic inflammatory myopathies. Rheumatology (Oxford). 2021;60(6):2615–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Oldroyd A, Sergeant JC, New P, McHugh NJ, Betteridge Z, Lamb JA, et al. The temporal relationship between cancer and adult onset anti-transcriptional intermediary factor 1 antibody-positive dermatomyositis. Rheumatology (Oxford). 2019;58(4):650–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Monseau G, Landon-Cardinal O, Stenzel W, Schoindre Y, Mariampillai K, Barete S, et al. Systematic retrospective study of 64 patients with anti-Mi2 dermatomyositis: A classic skin rash with a necrotizing myositis and high risk of malignancy. J Am Acad Dermatol. 2020;83(6):1759–63. [DOI] [PubMed] [Google Scholar]
- 28.Dani L, Ian Che W, Lundberg IE, Hellgren K, Holmqvist M. Overall and site-specific cancer before and after diagnosis of idiopathic inflammatory myopathies: A nationwide study 2002–2016. Semin Arthritis Rheum. 2021;51(1):331–7. [DOI] [PubMed] [Google Scholar]
- 29.Shah AA, Casciola-Rosen L, Rosen A. Review: cancer-induced autoimmunity in the rheumatic diseases. Arthritis Rheumatol. 2015;67(2):317–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Joseph CG, Darrah E, Shah AA, Skora AD, Casciola-Rosen LA, Wigley FM, et al. Association of the autoimmune disease scleroderma with an immunologic response to cancer. Science. 2014;343(6167):152–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Igusa T, Hummers LK, Visvanathan K, Richardson C, Wigley FM, Casciola-Rosen L, et al. Autoantibodies and scleroderma phenotype define subgroups at high-risk and low-risk for cancer. Ann Rheum Dis. 2018;77(8):1179–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Shimizu K, Kobayashi T, Kano M, Hamaguchi Y, Takehara K, Matsushita T. Anti-transcriptional intermediary factor 1-γ antibody as a biomarker in patients with dermatomyositis. J Dermatol. 2020;47(1):64–68. [DOI] [PubMed] [Google Scholar]
- 33.Pinal-Fernandez I, Mecoli CA, Casal-Dominguez M, Pak K, Hosono Y, Huapaya J, Huang W, Albayda J, Tiniakou E, Paik JJ, Johnson C, Danoff SK, Corse AM, Christopher-Stine L, Mammen AL. More prominent muscle involvement in patients with dermatomyositis with anti-Mi2 autoantibodies. Neurology. 2019;93(19):e1768–e1777. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.