RETRACTED: Impact of Romiplostim on Overall Survival in Chemotherapy Induced Thrombocytopenia

Abstract

BACKGROUND Romiplostim, a thrombopoietin receptor agonist, has been shown in a phase 2 trial to increase platelet count and facilitate chemotherapy administration in the setting of chemotherapy induced thrombocytopenia (CIT). The off-label use of romiplostim for this indication is common and is now a Category 2B recommendation by the National Comprehensive Cancer Network. Whether the administration of romiplostim improves clinical outcomes is not established. There are theoretical deleterious effects of platelet stimulation in cancer including the potential release of tumor stimulatory factors and blunting immune response to tumor cells. The goal of this retrospective study was to assess overall survival in patients with cancer treated with romiplostim compared to a cohort who received chemotherapy and granulocyte-colony stimulating factor (GCSF). The rationale for comparison of GCSF-treated arm was to address potential confounders related to chemotherapy dose reductions, delays, and abbreviated treatment courses due to cytopenias. METHODS To minimize cohort differences, we conducted an exact matched case-control analysis of patients with solid tumor malignancies receiving romiplostim with up to four patients receiving GCSF (without romiplostim). Patients were exact-matched based on a) primary cancer site, b) presence or absence of metastatic disease at the time of drug exposure, c) age (within 5 years), d) time from cancer diagnosis to treatment with romiplostim or G-CSF, 5) number of prior lines of antineoplastic therapy, and 6) comorbidities using the Charleston Comorbidity Illness index (<=6,7-10,>10). Overall survival (OS) was calculated from date of first dose of GCSF or romiplostim administration until death or last contact. Patient characteristics were summarized by frequency (percentage) and medians for categorical and continuous characteristics. OS was evaluated by Kaplan-Meier method and the difference between two treatment groups was determined by log-rank test. All statistical analyses were performed using R. RESULTS A total of 130 patients with solid tumor malignancy were identified as receiving romiplostim for CIT. Following strict exact-matching criteria, the analysis cohorts consisted of 58 patients in the romiplostim cohort and 138 patients in the GCSF cohort. The most common malignancies were Colorectal (18%), Non-Small Cell Lung Cancer (15%), Pancreas (13%). There was a median of 13 months between cancer diagnosis to romiplostim or and 16 months to GCSF and the mean number of prior lines of therapy was 2 in romiplostim and 1.8 in GCSF cohorts. The median comorbidity indices were similar in both groups (9 vs 9) and most patients had metastatic disease at the time of drug exposure (83%). The median overall survival in G-CSF treated patients was 18 months compared with romiplostim 10 months (Log rank P=0.014), Figure 1. In the stratified analysis by number of prior lines (1,2,>2), association with romiplostim remained statistically significant (Log rank P=0.012). Further restricting the analysis to patients who received romiplostim or GCSF during first or second lines of therapy, median overall survival in the romiplostim cohort (N=40) was 12 compared 21 months in the GCSF (N=109) cohort (log rank P=0.057), Figure 2. CONCLUSION In this retrospective exact-matched cohort analysis of patients treated with romiplostim for CIT, survival was inferior compared to patients receiving G-CSF alone. We acknowledge the limitations of single-center retrospective cohort analysis, but these findings highlight the urgent need to establish clinical benefit of thrombopoietin receptor agonists in the management of CIT.