Real-World Data Shows Increased Incidence of Autoimmune Disease in Survivors of Childhood Lymphoid Malignancies

Introduction With nearly 85% of children diagnosed with cancer becoming long-terms survivors (American Cancer Society, 2024), there is an increased focus on identifying and preventing late treatment-related effects. The list of known long-term effects is extensive and includes secondary malignancies, early cardiovascular disease, and metabolic disorders (COG LTFU Guidelines, 2023). Though there are known associations between autoimmune disease and subsequent development of cancer (Mekinian et al, Rheumatology 2016), the reverse has rarely been assessed, and there are no current recommendations to screen for specific autoimmune diseases in childhood cancer survivors (CCS). One study published by the Adult Life after Childhood Cancer in Scandinavia study group in 2015 reviewed national cancer registries and identified that CCS had a statistically significant increased risk of developing any autoimmune disease as compared to the general population (Holmqvist et al, Ann Rheum Dis 2016). No similar study has been performed in the United States. Methods We conducted this retrospective cohort study utilizing the global database TriNetX, which includes electronic health record data for over 113 million patients. Patients currently >22 years old with a diagnosis of a lymphoid-derived malignancy occurring <21 years of age were selected. This population was defined with the following ICD10 codes: C81-86, C88, and C91. The control cohort consisted of patients 22 and older with at least two visits for general medical encounter and no history of lymphoid-derived malignancies <21 years of age. We conducted 1:1 propensity score matching between the study and control populations to balance for established risk factors for autoimmune disease development, such as: current age, sex, race, and BMI >30. Patients with a history of autoimmune disease prior to their diagnosis of malignancy or general encounter were excluded. Statistical analysis was completed with TriNetX statistics, and the risk of developing autoimmune diseases, p values, and risk ratios (RR) were compared between the two cohorts. Given the greater prevalence of lymphoid malignancies amongst pediatric and adolescent/young adult patients as compared to myeloid malignancies, we chose to focus this analysis on the incidence of autoimmune disease in the lymphoid malignancy population alone. Results 25,133 patients with a history of a pediatric lymphoid malignancy were identified in this dataset. After 1:1 matching with non-CCS controls, 17,782 patients remained. In this CCS population with a history of lymphoid malignancy, the autoimmune diseases identified at increased rates compared to healthy controls were as follows: primary adrenocortical insufficiency (RR 22.1, p< 0.0001), chronic rheumatic heart disease (RR 10.7, p < 0.0001), type 1 diabetes mellitus (RR 6.5, p < 0.0001), systemic lupus erythematosus (RR 5.4, p < 0.0001), autoimmune hemolytic anemia (RR 5.4, p < 0.0001), rheumatoid arthritis (RR 4.1, p < 0.0001), Sjogren syndrome (RR 3.9, p < 0.0001), ulcerative colitis (RR 3.41, p < 0.0001), Crohn's disease (RR 2.7, p < 0.0001), juvenile arthritis (RR 2.5, p = 0.01), sarcoidosis (RR 2.3, p = 0.02), and autoimmune thyroiditis (RR 1.9, p < 0.0001).Limitations of this study include the inherent limitation in using ICD10 codes to identify patients with a specific diagnosis. Additionally, the TriNetX database limits the timeframe for data provided to 20 years from the index event. As such, if a patient was diagnosed with leukemia at the age of 4, TriNetX would only report an autoimmune diagnosis if documented at the age of 24 or younger. As autoimmune diseases more frequently present in adulthood, this study is likely under reporting the incidence. Conclusions This real-world data corroborates previous evidence published from the Scandinavia CCS cohort demonstrating the increased incidence in autoimmune disease in CCS. Given these two separate cohorts now identifying this risk, further studies should be completed using other data platforms to validate this finding with the availability of complete patient data to identify confounding variables. If this finding can be confirmed, next steps would include prospective studies to evaluate this association, assessment of possible biomarkers, and guidelines for screening as part of the long-term follow up guidelines.