We Are Witnessing Cancer's Downfall

Scientific concepts, discoveries, and nature/biological phenomena

Cancer research and the reproducibility crisis

• Despite expanding knowledge of genetic mutations and their roles in cancer, converting basic research into successful clinical outcomes has often been poor.
• Oncology shows a higher clinical trial failure rate than other therapeutic areas.

Landmark findings

• Glenn Begley and Lee Ellis tested 53 landmark cancer papers tied to cancer drug/target foundations; only 6 could be confirmed (~11% reproducibility).
• An independent effort repeated experiments from 23 cancer biology papers; positive findings had a median effect size ~85% smaller upon replication.

---

Oncogenic mutation mechanism: RAS “switch” activation

• A single DNA typo (one wrong letter) can convert a normal gene into a cancer-driving gene.
• The gene family involved is RAS.
• RAS proteins act as growth/division signaling switches:
  • When stuck in the “on” state, cells keep growing and dividing without stopping.

Prevalence claims

• RAS drives about 9 in 10 pancreatic cancers.
• RAS is implicated in about 1 in 5 cancers overall.

---

The “undruggable” problem and the targeted solution: KRAS G12C

Why RAS was hard to target

• The protein surface was described as nearly smooth, lacking an obvious drug binding pocket (“keyhole”).
• After ~4 decades, it was commonly labeled undruggable.

Key advance (2013)

• Jonathan Ostrem and colleagues (UCSF) identified a binding pocket in a specific mutant form: KRAS G12C.
• Early approaches depended on the specific mutation, leaving room for other variants to evade binding.

Strategy shift toward broader activity

• An FDA-approved drug mentioned as tipaxanib uses a helper binding mechanism involving cyclophilin A:
  • It attaches to active RAS and blocks growth signaling.
• Claimed effect: blocks multiple common RAS variants, including relevant forms in pancreatic cancer.

---

Clinical trial outcome: metastatic pancreatic cancer

Trial design

• ~500 patients across 6 countries
• Population: pancreatic cancer that had spread and returned after chemotherapy
• Compared two groups:
  • Tipaxanib
  • Standard chemotherapy

Reported median survival

• Chemotherapy: 6.7 months
• Tipaxanib: 13.2 months (about nearly double)

The speaker describes the result as receiving major acclaim, but argues it isn’t the only reason behind the discussion of “downfall.”

---

Regulatory acceleration: FDA timelines + expanded access

• The FDA accepted the drug application July 22 and approved it August 26 (~35 days).
• While the application was pending, the FDA cleared an expanded access pathway:
  • A legal route allowing dying patients to receive an unapproved drug.
• Reportedly, the program distributed tipaxanib to 2,000+ patients across nearly 50 states.

---

RAS as a cross-cancer driver (beyond pancreatic cancer)

Prevalence claims

• Roughly half of colorectal cancers
• Roughly one third of lung cancers
• The RAS family behind ~20% of all cancers

Development and testing

• The drug is described as being tested in lung cancer, including combination use with other therapies.

---

Immune-based personalized cancer vaccines (neoantigens)

Historical observations leading to immunotherapy

• William Coley (late 1800s):
  • Inspired by an aggressive bone cancer case and deliberate infection experiments (“Coley’s toxin”).
• Follow-on example:
  • In 1976, a TB vaccine was injected directly into tumors of 9 bladder cancer patients, with reported response.
  • The summary notes this approach is still used as standard-of-care for certain high-risk early bladder cancer cases.

Why cancer vaccines were historically unreliable

• Many vaccines targeted self-like shared proteins, which the immune system already tolerates.

Key concept: neoantigens

• Mutations create new protein fragments not found in healthy tissues.
• An immune response can be triggered if these neoantigens are presented on cell surfaces.

The major remaining problem: “not off-the-shelf”

• Each patient’s tumor has a unique mutation set.
• Therefore, an off-the-shelf vaccine often fails.

Methodology for personalized vaccination (outlined)

1. Sequence
   • Sequence the patient’s tumor and healthy blood
   • Subtract shared mutations to identify tumor-specific mutated fragments (candidate neoantigens)
2. Predict antigen presentation
   • Use neural AI networks to predict which fragments the immune system can present on the cell surface
3. Manufacture individualized vaccine
   • Create the personalized vaccine based on predicted neoantigens

Phase three vaccine trial described (August)

• 1,137 patients enrolled
• Condition: melanoma, surgically removed
• Tumor cells analyzed using an AI platform by Moderna
• Result: personalized-vaccine recipients showed improved time to recurrence/death versus control

Expansion of trials

• Mentioned Merck/Moderna trials:
  • 9 trials total
  • Targets: melanoma, lung cancer, kidney cancer, bladder cancer
  • Multiple trials already in phase three

---

Protein structure prediction tool: AlphaFold

• AlphaFold is cited as an enabling technology:
  • It predicts 3D protein structures from amino acid sequences.
  • This helps identify how drug molecules might bind/interact, supporting discovery for previously hard targets.

---

Open question raised

• The speaker flags a broader question in epidemiology/biology:
  • Why some cancers are becoming more common in younger adults?
• A follow-up video is suggested.

---

Researchers and sources featured (named in the subtitles)

• Glenn Begley
• Lee Ellis
• Jonathan Ostrem
• William Coley
• Fred Stein
• Bessie Dashiell
• Zola
• Bessie Dashiell’s surgeon (William Coley) (named above)
• Moderna (company; vaccine developer; referenced for AI platform)
• Merck (referenced as running trials alongside Moderna)
• FDA (regulator; cited for application acceptance/approval and expanded access)