Can autophagy treat cancer

The honest answer is that autophagy modulation is a real biological lever, but the evidence that it can treat cancer in humans is thin, conflicted, and nowhere near ready for the clinic.

Autophagy is the cell's self-cleaning process: it degrades damaged parts and recycles them. In cancer, it plays a double role. Early on, it suppresses tumors by clearing damaged mitochondria and protein aggregates that could drive mutations. Later, once a tumor is established, cancer cells hijack autophagy to survive stress from chemotherapy, low oxygen, and nutrient shortage. So the question is not whether autophagy works, but whether you want to boost it or block it, and in which patient, at which stage.

The evidence for treating cancer by modulating autophagy is mostly preclinical. A 2015 study identified TMEM33 as a stress-inducible protein that drives the unfolded protein response, increasing CHOP and downstream death signals in breast cancer cells [1]. That is a mechanism, not a treatment. A 2025 review confirms that ER stress and the UPR can push cells into apoptosis, pyroptosis, and ferroptosis, with CHOP as a central switch [2]. But both are cell and animal work. No large randomized trial in humans shows that any autophagy drug, whether an inducer like rapamycin or an inhibitor like hydroxychloroquine, improves overall survival in cancer.

The clinical data that exists is mostly small phase 1 and 2 trials of hydroxychloroquine combined with chemotherapy. Results are mixed. Some show modest responses in pancreatic and brain cancers; most show no survival benefit. The trials are almost all manufacturer-funded or academic with industry ties, and none compare autophagy modulation against a genuinely untreated control. The long-term safety data is absent. Boosting autophagy systemically could suppress immune surveillance; blocking it could accelerate tumor growth in early disease. Both risks are real and unmeasured.

My confidence is high that this is not a proven treatment. The mechanism is plausible, the human data is weak, and the field has not run the trials that would settle it. If you are considering this for yourself or a family member, the evidence does not support it outside a clinical trial.