#114 – Eileen White, Ph.D.: Autophagy, fasting, and promising new cancer therapies

Attia: 'I used to do a thing where I fasted seven days every quarter... I began to ask the question which is okay is seven days a quarter the right dose? I'm convinced that it's a big bolus of autophagy but is it frequent enough? What about three days every month — that's about the same number of total days fasted but it's more frequent but it's probably less potent.' He eventually switched to 3-day monthly fasts after observing that all key metabolic markers that appeared at day 7 also appeared at day 3. The optimal protocol is the single most important unanswered translational question in this field.

Attia systematically checks blood before and after each 7-day fast: 'glucose plummets, insulin becomes unmeasurable, uric acid goes through the roof along with beta-hydroxybutyrate... the ratio of T3 to reverse T3 changes by four to six fold which means you basically shut off the metabolism, explaining why you become incredibly cold intolerant.' He describes two mechanisms for the uric acid rise: purine release from nucleic acid catabolism (a potential autophagy-flux surrogate) and competitive inhibition of uric acid renal excretion by BHB sharing the same transporter.

White's mouse experiment: engineer adult mouse with inducible autophagy knockout, give mouse KRAS-driven lung cancer, then delete the essential autophagy gene. 'The answer was the tumor died first.' This is the proof-of-principle that preferential tumor killing is achievable. Her co-founded company focuses on developing small molecules against specific autophagy enzymes rather than the blunt HCQ approach. The emerging combinatorial strategy: autophagy inhibitor + immune checkpoint blockade, because autophagy loss promotes inflammation that makes cold tumors hot.

White's lab identified glucuronic acid as a metabolite that accumulates when autophagy is inhibited — making it a potential negative proxy: if glucuronic acid is low, autophagy may be active. Attia's proposed machine learning approach: 'You take an unbiased view... it is gonna be much more complicated, it is not gonna be a regression model based on five things we know.' The reference gold standard in such a study would be the muscle biopsy LC3 flux assay.

White: 'A main function of autophagy in tissue homeostasis is to preserve cellular function to be normal, to prevent chronic damage and inflammation in tissues that are susceptible to cancer as a result of chronic damage and inflammation.' Attia frames it as: autophagy is highly protective against 3 of the 4 biggest killers (cancer prevention, Alzheimer's, metabolic disease), which are collectively responsible for 82% of deaths over 50.

White explains: 'In the fed state the cancer cells already have elevated autophagy flux and when you fast them it does go up but it only so high it can go.' Blocking autophagy in genetically engineered mouse models with KRAS-driven lung or pancreatic cancer causes the tumor to die first — before normal tissue is meaningfully damaged — creating a therapeutic window. The liver-specific autophagy knockout produces fatty liver and benign hepatoma through cycles of damage and chronic inflammation, illustrating how loss of autophagy in normal tissue is itself oncogenic. The implication: stimulate autophagy before cancer develops (fasting, metabolic health), then consider inhibiting it therapeutically once cancer is present.

White: 'KRAS-driven lung cancer and pancreatic cancer are extraordinarily autophagy-dependent.' The b-RAF V600E mutation in melanoma and lung cancer is also very sensitive. APC-deficient colon cancer and prostate cancer are sensitive. Cancers along this spectrum have a different sensitivity level — BRAF-driven cancers are MORE sensitive than KRAS-driven lung cancer. The common thread is constitutive pro-growth MAP kinase signaling that creates a high metabolic demand the cell can only sustain via continuous autophagy-mediated recycling.

White: 'If inhibiting autophagy activates the immune response and can facilitate people who wouldn't respond to immune checkpoint blockade to respond, then that would be critically important to do.' In short-term experiments, T-cells from autophagy-deficient mice appear 'more anti-tumorigenic' — consistent with the hypothesis. Long-term immune cell function does eventually degrade without autophagy, but for the finite duration of cancer therapy this may not be a limiting factor.

The tissue hierarchy of autophagy dependency: brain (most sensitive — post-mitotic neurons cannot dilute damage by division), liver (sensitive — loses autophagy → steatosis and Mallory body protein aggregates, but regenerates for a while), lung (appears relatively insensitive phenotypically). The neonatal starvation period — between placental cutoff and first suckling — triggers a massive autophagy induction in mammals; autophagy-deficient newborns failed even with force-feeding and died within 24 hours regardless.

The proposed research design: take 100 volunteers, fast them for varying periods (3–10 days), collect muscle biopsies at each time point for LC3 flux quantification alongside comprehensive blood metabolomics. Use machine learning on the metabolome to identify an autophagy signature that predicts biopsy-confirmed flux. Attia: 'Metabolites as surrogate markers for the consequences of autophagy.' White suggests the mouse model comparison — wild-type fasted vs. autophagy-deficient fasted — would first identify which metabolic changes are autophagy-dependent, narrowing the search.

White's research focus: 'Defining at the molecular level the functional requirements for autophagy in individual cancers and this involves understanding the metabolic role of autophagy — why one cancer needs autophagy more than another.' The combinatorial future is autophagy inhibitor + immune checkpoint blockade, using autophagy inhibition to generate inflammation that makes the immune system see the tumor.