Eosinophilia in patients with melanoma under treatment with immune checkpoint inhibitor-based regimens.

Abstract

e21528 Background: Immune-related eosinophilia (irEo) has been described as an adverse event of immune checkpoint inhibitors (ICI), but its rate, severity, duration, clinical significance, diagnostic approach, and management have not been analysed. In the present study we analyse these features in patients with melanoma. Methods: We reviewed the files of patients with advanced/metastatic melanoma treated with ICI-based regimens in a tertiary university referral centre, to describe the rate of eosinophilia, along with the baseline disease characteristics, severity, type of treatment, time on treatment, maximum eosinophil count, and symptoms, as well as its diagnostic evaluation and management, course, and prognostic implications. Results: A total of 249 patients were included in the analysis. Eosinophilia was present in 61 (24.5%) and there was no correlation with gender, age, histologic type, stage, or BRAF mutation status. The median time interval from treatment initiation to the onset of eosinophilia was 55 days (10-534), while the median eosinophil count at first presentation was 0.70x109/L (0.50-4.53x109/L), and the maximum eosinophil count was 1.07x109/L (0.50-13.22 x109/L). The rate of eosinophilia was significantly higher in patients treated with nivolumab plus bempegaldesleukin (50.0%), followed by nivolumab plus ipilimumab (21.7%). Symptomatic patients [N = 15, mostly pruritus, rash, arthralgia] and/or patients with hypereosinophilia (eosinophil count > 1.5x109/L, N = 3 at first presentation, N = 20 during the course of eosinophilia) were assessed for organ involvement and for the cause of eosinophilia (CBC with blood smear, biochemistry profile including LDH, and uric acid, cardiac troponin, chest X-ray, and heart ultrasound). Further evaluation was performed according to the clinical scenario and the findings of the initial evaluation. In none of the patients a definitive diagnosis of eosinophilia was reached; thus, all cases were deemed irEo. None of the patients was acutely ill. Patients without findings or hypereosinophilia did not require medical intervention, while the remaining were managed with low-to-medium doses of corticosteroids (N = 2), antihistamines (N = 11), or local treatment (N = 4). Two patients had to delay treatment, and one (max count 13.2x109/L) had to permanently discontinue bempegaldesleukin and continue on nivolumab monotherapy. Eosinophilia relapsed in 21 patients when rechallenged with the same or different immunotherapy regimen. There was a non-significant trend for longer overall survival in patients with eosinophilia (64.5 months vs 33.4 months, p = 0.16). Conclusions: This is the first study evaluating the association between the type of immunotherapy and the rate of eosinophilia in patients with melanoma and analysing the baseline characteristics of the patients as well as their diagnostic evaluation and management, course, and prognosis.