Tuesday, October 13, 2026 Evidence-led medical research news
The First Cohort
Medical research, with the receipts

Lena Marshall

136 stories · Page 3 of 6

Drugs & FDA

A Pill Doubled Survival in Advanced Pancreatic Cancer: Six Months to Thirteen, at $39,800 a Month

The best result advanced pancreatic cancer has seen in a generation, and the FDA cleared it in about five weeks. For patients who have run out of chemo, it turns a median of six months into thirteen, at a list price of $39,800 a month.

Investigations

A Third of England's COPD Patients Aren't Diagnosed Until a Hospital Admission

Across 1.7 million patients and thirteen conditions, being diagnosed only after an emergency tracked with far higher odds of dying within a year, up to sevenfold for coeliac disease, in a health service whose own records held the story for two decades.

Cancer

A New Cancer 'Weapon' That Has Only Ever Slowed Tumors in Mice

Stanford's tissue-resident natural killer cells slowed melanoma and head-and-neck tumors in mice. John Sunwoo warned against reading too much into that. The coverage read plenty into it anyway.

Infectious Disease

The Anti-Mosquito Industry Sells a Fix for Something Your Skin Chemistry Settled Years Ago

One volunteer drew 100 times more mosquitoes than another, a difference set by skin acids that hold steady for years. The wristbands, ultrasonic gadgets and vitamin-B patches sold to stop the bites show no measurable protection.

Brain & Aging

The Vitamin D Trials That Found No Brain Benefit Never Tested This Dose

Fifty-four older adults with failing sleep and early memory loss, and the ones taking 5,000 IU of vitamin D a day scored 13 percent higher on a standard cognitive test. The major null trials never tested that dose.

Infectious Disease

Arizona State's AI Flags Healthy People as 'Weak' Vaccine Responders Without Reporting How Often It's Right

Arizona State's AI sorts roughly one in twenty healthy people into the 'weak responder' column. The clinical payoff, in the lead author's own words, is more doses. Nobody reported how often the model is right.

Nutrition & Food

A PETA-Tied Doctors Group Is Suing to Kill RFK Jr.'s New Food Pyramid

The Physicians Committee says a secret, industry-tied panel redirected the 2025-2030 dietary guidelines. The group crying capture spent years inside PETA's funding orbit.

Drugs & FDA

A $2.7 Million Gene Therapy, Approved Because Patients Needed Less Cornstarch

Ultragenyx's Genglycos won accelerated approval on a surrogate: a 31 percent cut in daily cornstarch. But the disease is low blood sugar, and by the company's own numbers the treated group spent slightly more time below 70 mg/dL, while liver enzymes spiked in 71 percent and every patient needed ster

Investigations

Fauci's Top Adviser Admitted to Hiding COVID Emails and Taking Wine for a Natural-Origins Commentary

David Morens pleaded guilty to conspiring to run NIH business through Gmail to dodge FOIA, and to taking wine as an illegal gratuity tied to a commentary declaring COVID had natural origins. Fauci was not charged.

Cancer

UC Irvine's 'Framework' for Safer CAR-T Is a Hypothesis Essay, Not a Study

The senior author consults for Kite Pharma, whose Yescarta accounted for all 6 grade-5 neurotoxicity deaths in the cited review's real-world data, the exact harm the paper proposes to treat.

Infectious Disease

Eight Infant Macaques Showed No HIV Rebound for a Year. The Third Drug Comes From a Convicted-Fraud Biotech.

OHSU's triple regimen left no detectable virus in all eight infant macaques a year after treatment stopped, even after the harshest rebound test. Two of the authors hold a financial stake in the company behind the experimental third drug.

Cancer

The Cells That Heal Wounds Are Feeding Breast Cancer Its Nerves

University of Oklahoma researchers blocked the protein in mice and triple-negative tumors grew far less. In patients, all they have is a correlation and a drug, already sold for something else, that has never been tried against breast cancer.

Metabolic & Weight

The 123 Percent Death-Risk Number in the Vitamin D Headlines Was Never Tested as a Pill

An observational study of 5,520 older Britons tied belly fat plus low vitamin D to a 123 percent higher death risk. Hand people the actual pill in a randomized trial and mortality drops about 5 percent, with nothing at all for the heart.

Brain & Aging

Bracing for a Fall That Hasn't Happened Yet: The Bargain the Aging Ankle Strikes

Falls kill more older Americans than any other injury, and a study of 107 walkers points at the body's own defense: the aging ankle appears to brace for stability, buying it with a shorter, slower, more tiring stride.

Drugs & FDA

The FDA's Reviewers Called This Melanoma Data 'Not Interpretable.' Nine Days Later, the Agency Approved the Drug.

Twice the FDA rejected Replimune's engineered-herpes melanoma therapy, its reviewers calling the single-arm trial not interpretable. The design never changed. A 10-3 advisory vote came anyway, and nine days after that memo, the accelerated approval.

Metabolic & Weight

Lilly Turned the Receptor On, Amgen Turned It Off, and Both Strip Weight. The Map Came Four Years Late.

Eli Lilly's tirzepatide switches the gut-hormone receptor on. Amgen's MariTide blocks it. Both drive weight loss, and until a publicly funded Cambridge mouse study, no one had cleanly mapped how one receptor works in both directions.

Investigations

Eighteen Years and $250 Million Later, Human Limb Regrowth Is Still a Salamander Trick

A Northeastern lab showed exactly how an axolotl rebuilds an arm from the shoulder down, and humans carry the same genes. We can still regrow a lost fingertip as children. Everything past that we heal by scarring over, and $250 million of Pentagon money has not changed the math.

Brain & Aging

For Twenty Years, Women Were Warned That Estrogen Raised Their Dementia Risk. Stanford's Autopsies Point the Other Way.

A randomized federal trial spent two decades warning that hormone therapy raised older women's dementia risk. A Stanford autopsy study just found the opposite association. The catch: both studied women who started the drug far too late to settle it.

Drugs & FDA

NIH Said Its Cervical-Cancer AI Beat Human Doctors. Six Years Later, Its Own Author Called It Overfitting.

In 2019, NIH announced an algorithm that read cervix photos better than doctors, an AUC of 0.91. Its senior author has since written that the number came from overfitting and that the follow-on projects failed. Nature Medicine is now calling the same idea the 'missing piece.'

Investigations

King's College London Turned Sheep's Wool Into a Bone Scaffold, and Called It a Milestone Before a Single Human Trial

The wool membrane grew more organized bone than collagen in a rat's skull. Collagen still grew more of it, the sample size never made the press release, and no one has put the material in a person.

Brain & Aging

This Week's Autism 'Breakthrough': A Rare Mutation, an Injection Into Mice, and Not One Human Patient

An injection that silences one gene rescued social behavior in mutant mice and a dish of human cells. The headlines called it an autism therapy. No autistic person has received it, and the gene it targets is mutated in a small fraction of cases.

Autoimmune & Inflammation

The Finest Immune Map of the Sarcoid Lung Explains Why Its Targeted Drugs Keep Failing

National Jewish Health charted more than 150,000 immune cells from sixteen sarcoidosis patients and found a disease with too many moving parts for any single drug to switch off. The targeted therapy that got furthest had already failed Phase 3 and erased 83 percent of its maker's value.

Cancer

The Cancer Trials That Buried Vitamin C Never Put It in a Vein

Mayo's famous trials gave advanced cancer patients vitamin C by mouth and found nothing. A randomized trial that delivered it by IV, at a dose pills cannot reach, doubled survival in metastatic pancreatic cancer.

Drugs & FDA

For Years the Muscle Aches Were Called a Nocebo Effect. A McMaster Lab Found Statins Injuring the Muscle Cell Itself.

An estimated 7 to 29 percent of statin users report muscle pain. The flagship trial called most of it a placebo response. In mouse and human muscle cells, a McMaster team has now traced a cholesterol-independent pathway by which statins damage muscle directly.