Immunotherapy and Cannabis: A Harmful Drug Interaction or Reefer Madness?

Abstract

AbstractA retrospective (N=140) and a prospective (N=102) observational Israeli study by Bar-Sela and colleagues about cannabis potentially adversely impacting the response to immunotherapy have together been cited 191 times including by clinical practice guidelines. There have also been reports on PubPeer outlining unverifiable information in their statistics and numerous discrepancies calculating percentages. This report attempted to replicate the data-analysis including non-parametric statistics. Table 1 of the corrected prospective report contained 22 p-values but only one (4.5%) could be verified, despite the authors being transparent about the N and statistics employed. Cannabis users were significantly (p < .0025) younger than non-users but this was not reported in the retrospective report. There were also errors in percentage calculations (e.g. 13/34 reported as 22.0% instead of 38.2%). Overall, these observational investigations, and especially the prospective, appear to contain gross inaccuracies which could impact the statistical decisions (i.e. significant findings reported as non-significant or vice-versa). Although it is mechanistically plausible that cannabis could have immunosuppressive effects which inhibit the response to immuno-therapy, these two reports should be viewed cautiously. Larger prospective studies of this purported drug interaction that account for potential confounds (e.g. greater nicotine smoking among cannabis users) may be warranted. 198 / 200 wordsSimple SummaryTwo Israeli studies about medical marijuana potentially interfering with immunotherapies like nivolumab for cancer treatment have received substantial attention. However, there have been anonymous but detailed concerns about these reports on PubPeer. This team attempted to verify the data-analysis and statistics of these two reports and the published correction. Many findings including some that could impact the statistical conclusions could not be verified. Of the 22 statistical tests on Table 1 of the prospective report, six could not be repeated using the same statistics and with the provided N. The p-value on 15 corresponded with that of a different statistical test than was listed in the methods. Re-analysis also identified some previously unreported significant differences (e.g. age) between cannabis users and non-users at baseline. Further study of the safety of immunotherapy and cannabis combination may be warranted using patient groups that have been matched on key demographic and medical variables. 150 / 150 wordsGraphical abstract