e18846 Background: Studies suggest that the immunomodulatory effect of cannabis may impair responsiveness to ICI in cancer patients. These studies were limited by small sample size in a heterogeneous group. Determining the impact of cannabis on cancer control in patients receiving ICI is critical to informing patients and clinicians. We assessed the relationship of cannabis with cancer treatment outcomes using a large national healthcare claims database linked to retail cannabis purchase data. Methods: Using the Forian Inc database from 2017-2022, we identified adults (age > 18 years) with advanced cancer who had ≥ 90 days of continuous health plan enrollment and received ≥ 1 ICI treatment. Point-of-sale purchase records from 18 state cannabis programs defined cannabis consumers (CC; any purchase prior to first ICI to last known follow-up) and cannabis non-consumers (CNC; no known cannabis purchases). We abstracted data on patient demographics, cancer history, cannabis utilization (dates of purchases, routes, strains), and all cancer-directed therapy claims. rwTTD was calculated from time of first ICI to a defining discontinuation event (change in therapy, gap of > 120 days, or end of enrollment in health plan). Results: From a database of nearly 68 million patients, 20,675 patients met the inclusion criteria (374 CC and 20,301 CNC). Baseline characteristics were similar between the two groups (e.g., age, diagnosis type, time to first ICI) (Table 1). Lung, melanoma and renal cancers account for nearly half of all identified diagnoses. By day 30 after ICI initiation, 45% of CC had made at least one cannabis purchase. Median time from advanced cancer diagnosis to first ICI and median follow up time were also similar (Table 1). The difference in median rwTTD between CC and CNC groups was small relative to the overall variation of this outcome (Table 1) and remained similar (88 days) when restricting analysis to CC with first purchase within 30 days of ICI. Conclusions: Cannabis use does not appear to be associated with rwTTD in advanced cancer patients initiating ICI. Formal statistical testing using inverse probability treatment propensity weighting and adjustment for confounding factors (opioids/glucocorticoids) to assess rwTTD and real-world overall survival is ongoing. Well designed, prospective cohort studies of CC and CNC are needed to minimize selection bias in studies of cannabis use. [Table: see text]
Read full abstract