Immunotherapy-Related Hematological Adverse Events: Incidence and Outcomes

Introduction: Immunotherapy (IO) is frequently used in the treatment of several malignancies and has been associated with various off-target manifestations of autoimmunity. Hematological immune-related adverse events (heme irAEs) are relatively uncommon, but potentially life-threatening complications that can be challenging to diagnose as both cancer and cancer-directed therapies may cause cytopenias. We sought to determine the incidence, treatment patterns and outcomes associated with heme irAEs in individuals with cancer treated with IO from a large database.

Methods: We used a de-identified national health research network, TriNetX, with data sourced from 76 health care organizations (HCOs) located in the United States with over 107 million patients. We identified all adults (≥ 18 years) with a diagnosis of neoplasm (ICD-10: C00-D49), and documented use of IO including pembrolizumab, nivolumab, atezolizumab, avelumab, durvalumab and ipilimumab, between 2015-2023. Heme irAEs were defined as immune thrombocytopenia (ITP; ICD-10: D69.3), autoimmune hemolytic anemia (AIHA; ICD-10: D59.0, D59.8, D59.10, D59.19, D59.11, D59.12, or D59.13), and pure red cell aplasia (PRCA; ICD-10: D60, or D61.01) occurring after IO initiation. We collected demographic, clinical and treatment characteristics from our baseline cohort. We further compared all-cause mortality and all-cause hospitalization outcomes between individuals with heme irAEs vs. without heme irAEs. To account for potential confounders and sample size differences between cohorts, we performed 1:1 greedy nearest neighbor propensity score matching, based on patient demographics, cancer type, radiation therapy and bone metastasis. For continuous data, we performed independent t-tests or one-way ANOVA. For categorical data, we performed chi-square tests. To evaluate survival probabilities, we used a Kaplan-Meier analysis and compared hazards ratios. All tests were two-tailed with an alpha level of .05.

Results: We identified a total of 83,693 individuals with neoplasms who used IO since 2015. Of those 435 (0.52%) had a diagnosis of heme irAEs: 253 (58.2%) ITP, 167 (38.4%) AIHA, 29 (6.7%) PRCA. Individuals who developed heme irAEs had a median age of 64 years old (21-87), were predominantly male (51.7%), white (77.9%) and had a history of cancer of the respiratory system (40.9%) (Table). Also, 52.9% of them received radiation therapy and 36.1% had associated bone metastasis. Most frequently used IO were PD-1 inhibitors, pembrolizumab (52.6%) and nivolumab (39.1%). Overall, the majority (97.2%) of individuals with heme irAEs were treated with glucocorticoids. Additionally, patients with ITP were treated with IVIG (15.4%), Romiplostim (11.1%), Rituximab (9.1%), and Eltrombopag (8.3%), less commonly with MMF and cyclosporine. Patients with AIHA received in addition to steroids, IVIG and Rituximab. Only a small number of patients with a diagnosis of PRCA required cyclosporine. Altogether, individuals who developed heme irAEs had a higher mortality rate than individuals without heme irAEs (57.4% vs 46.5% HR 1.57 95% CI 1.30-1.89, p value <0.001) (Figure). There was no statistical difference in the number of all-cause hospitalizations between groups (4.13 vs 4.55, t -0.417 p value 0.67).

Conclusions: In this large national cohort, the real world incidence of heme irAEs was low. However, despite its rarity, development of heme irAEs was associated with increased mortality in individuals with cancer who received IO. Early recognition and prompt intervention are crucial to optimize its management.