Comprehensive meta-analysis of COVID-19 mortality rates for 22 cancer subtypes from the Reboot: COVID-Cancer Project, an interactive resource with aggregated data from 21,839 cancer patients

ABSTRACT

Introduction: The COVID-19 pandemic continues to disproportionately impact people with cancer. Mortality estimates among cancer patients vary and are influenced by numerous factors including cancer type, treatment, disease stage, and patient demographics. To date, attempts to explore these associations have been limited by small cohorts. Methods: Here, we present a meta-analysis of data available through the Reboot COVID-Cancer Project, a living and freely available resource that includes published clinical studies that report outcomes for cancer patients with COVID-19. Studies were identified using targeted search queries in PubMed, MedRxiv, BioRxiv, and the SSRN eLibrary, followed by rule-based approaches and extensive manual validation and data extraction. The data is updated monthly and can be explored through an interactive dashboard as well as downloaded. Case fatality rates (CFR;the number of deaths per 100 confirmed cases during the study period) were calculated using a random-effects model. Study heterogeneity and sample size bias was assessed using the Egger regression test. Results: As of December 18, 2020, the resource contained 225 publications comprising 21,839 cancer patients with COVID-19. Of these, there was sufficient sample size to quantify CFRs for 22 cancer types across 19,147 patients from 109 publications. The pooled CFR among all cancer patients was 27% (95% CI 25-30%). For solid tumors and hematological malignancies, the CFRs were 23% (95% CI 20-25%) and 30% (95% CI 27-33%), respectively. Within solid tumors, patients with lung (CFR 32%, 95% CI 27-36%), prostate (CFR 30%, 95% CI 17-43%), and central nervous system (CFR 27%, 95% CI 18-36%) malignancies had relatively high CFRs, whereas patients with breast (CFR 10%, 95% CI 7-14%) and thyroid (CFR 5%, 95% CI 1-11%) malignancies had relatively low CFRs. Among patients with hematological malignancies, CFRs ranged from 10% (95% CI 3-18%) in patients with chronic myelogenous leukemia to 39% (95% CI 20-57%) in patients with acute myeloid leukemia. Discussion: We observed significant heterogeneity of COVID-19 CFRs between cancer subtypes. This may in part reflect differences in patient demographics, treatment history, or disease state. Subtype-specific analysis can stratify cancer patients by risk for COVID-19 mortality and advise management strategies. The Reboot COVID-Cancer Project provides an accessible means to evaluate subtype-specific COVID-19 fatality rates on a by-publication basis.