Background/Purpose: Many studies of cancer risk in SLE are limited by small sample size or use of administrative data, which rely on billing code diagnoses instead of clinical data. No studies to date focused on incident SLE. We studied cancer risk in the largest-ever cohort of clinically confirmed incident SLE patients.

Methods: Patients meeting ACR criteria for new-onset SLE (within 15 months of diagnosis) were enrolled into the SLICC Inception Cohort, across 32 centres. Patients are followed yearly using a standard protocol, with detailed data collection including SLE Disease Activity Index-2000 (SLEDAI-2K) and damage, and drugs in the past year. New cancer diagnoses are recorded by the examining physician at the annual study visit, and confirmed with chart review including pathology reports.

Multivariate proportional hazard regression was performed, using baseline variables for demographics (age at SLE onset, sex, race/ethnicity), and time-dependent variables for drugs (corticosteroids, anti-malarial drugs, immunosuppressive drugs), smoking, and SLEDAI-2K. As well as cancer over-all, we evaluated risk factors for the most common cancer types.

Results: Of 1848 new-onset SLE patients enrolled between 1999-2011, 1668 had at least one follow-up; these were the sample for the current analysis. End date was the first of death, last visit, or end of study interval for this analysis (Aug. 2015). Baseline demographics are shown in Table 1.

Over 14,215 years (mean 8.5 years) there were 60 cancers (incidence 4. 2 events per 1,000 patient-years). This included 12 breast cancers, 9 non-melanoma skin, 7 lung, 6 hematological, 5 melanoma, 5 prostate, 3 cervical, 3 renal, 2 gastric, 2 head and neck, 2 thyroid, and one each rectal, sarcoma, thymoma, and uterine.   Almost half of the cancer cases (including all of the lung cancers) were associated with baseline  smoking, versus only one-third of those patients who did not develop cancer. Univariate analyses of all cancer types suggested a higher risk of cancer among patients of white race/ethnicity and among those with the highest quartile of disease activity at cohort entry.

However, the multivariate proportional hazard regression indicated that among SLE patients, the over-all cancer risk was related primarily to male sex and older age at SLE diagnosis. In those analyses, the effect of race/ethnicity was not clearly evident, and the point estimate for highest quartile of disease activity actually reversed to suggest a non-significant trend towards lower cancer risk.

In the multivariate analyses specifically for breast cancer, age at SLE diagnoses remained a risk factor, and anti-malarials were associated with a decreased risk. This effect of anti-malarials was not clearly seen for any other cancer type. For non-melanoma skin cancer, both age at SLE diagnosis and cyclophosphamide were strongly linked with risk.

Conclusion: This is the first large, multicentre cohort study to clearly show how different cancer types in SLE are associated with specific risk factors.  Additional follow-up may allow additional determination of the possible effects of disease activity and drugs on cancer subtypes.

« Back to 2019 ACR/ARP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/cancer-risk-in-a-large-inception-sle-cohort-effects-of-age-smoking-and-medications/