I concede that I haven’t thought as deeply about this as ethicists, but I strongly suspect that the cost/benefit calculation here is way over-cautious if you think the theoretical induced harm is remotely close to the benefits of publishing.
The history of science is already full of self-experimenters, so at the margin publishing is unlikely to move the needle.
Furthermore, patients with cancer diagnoses are already extremely motivated to try whatever experimental treatments the FDA will permit; self-experimentation is already supply-constrained (of experiment opportunities) and there is excess demand. Again fuzzy concerns about population-level harms overrule individuals’ rights to seek treatments for their fatal diagnoses.