The COVID-19 pandemic has brought extraordinary uncertainty for individuals undergoing cancer treatment. Postponement of non-essential medical care was the appropriate initial response to the unexpected and steady rise in cases witnessed in New York City. However, we know from the SARS epidemic experience in 2003 that the harm resulting from these delays in cancer care is likely to be substantial.

Memorial Sloan Kettering Cancer Center (MSKCC) experienced a surge in COVID-19 cases during the peak of the regional epidemic. Like other hospitals in the area, we were faced with personal protective equipment (PPE) and test supply shortages, but critical care resources were not constrained. In the early phase of the New York City epidemic, the first 423 cancer patients diagnosed with COVID-19 at MSKCC were enrolled in the present study. We took a deep dive into understanding the predictors of COVID-19 severity and hospitalization. The case fatality rate in our cohort was 12%, although death was not the main outcome of interest, a key difference from other studies that have reported cancer-related COVID-19 findings. The primary concerns with this outcome measure were ascertainment bias and overestimation of COVID-19 mortality. Instead, we examined severe respiratory illness as a more consistent and less biased measure. At the outset, outcomes were not dissimilar from those reported in hospitalized patients. In our cohort of 423 patients with COVID-19 illness, 40% were hospitalized, and 21% developed severe respiratory illness, including 9% that required mechanical ventilation.

Cancer is a disease of the elderly, and advanced age correlates with adverse COVID-19 outcomes. Our findings confirm the excess risk posed by age and common chronic medical conditions in a cancer cohort. The analysis also indicates that worse outcomes occur in patients with lung cancer, regardless of therapy, and that patients with hematologic malignancies have higher hospitalization rates than other cancer patients. Reassuringly, recent chemotherapy or surgery did not significantly increase the risk of adverse outcomes. We observed a higher risk of severe respiratory illness and hospitalization in patients treated with checkpoint blockade immunotherapy (anti-PD-1, PD-L1, CTLA-4). This effect was independent of underlying cancer and of other risk factors commonly associated with poor outcomes in lung cancer. However, the number of immunotherapy-related events was modest, and the potential for residual confounding in various cancer types still exists. We recommend cautious interpretation of this finding. The role of checkpoint blockade, either alone or in combination with other treatments, needs to be scrutinized in larger tumor-specific cohorts to make definitive conclusions.

The lack of association with systemic chemotherapy, advanced disease, and recent cancer surgery contradicts earlier reports from China. These observations are reassuring as we endeavor to continue delivering the most effective cancer care during this pandemic. Our findings collectively suggest that oncologic therapies can be administered safely with sustained vigilance. However, there is much that remains to be learned, and further evaluation of COVID-19 related risks across various cancer populations is needed. The accurate measure of adverse events and mortality due to COVID-19 remains an important but complex issue. A multitude of considerations, including delays in cancer diagnosis, direct and indirect consequences of deferred or modified therapies, halted clinical trials, and patient perception of risk, will reveal the actual impact of COVID-19 in patients with cancer.