#133 - Vinay Prasad, M.D., M.P.H: Hallmarks of successful cancer policy

KEYPOINTS

• 01

  Medical reversal: widely used practices that turn out not to work

  Prasad, a haematologist-oncologist and meta-researcher at UCSF and author of Ending Medical Reversal and Malignant, came to the idea as a resident. He noticed practices that ran against the best evidence. As a resident he managed stent patients after the procedure and saw the complications. Meanwhile the large COURAGE trial had found that stents in stable disease didn't reduce heart attacks or extend life. He and colleagues called this pattern 'medical reversal': plausible biology plus observational data, later overturned by randomised trials.

• 02

  Sham-controlled trials separate the procedure from the ritual

  Arthroscopic knee surgery beats physical therapy or ibuprofen, but not a sham operation in which the surgeon inserts the scope and does nothing. The same holds for cement injections into osteoporotic vertebral fractures, some shoulder procedures and, Prasad believes, stenting for chronic stable angina (the ORBITA trial). Much of the benefit comes from reassurance and the patient's belief that something was fixed.

• 03

  Stents are 'seductive' because both patient and doctor feel rewarded

  Patients come back grateful, convinced their life was saved and that shovelling the driveway hurts less. That feedback reinforces the procedure even when trials show no hard benefit. Prasad says this disconnect between what patients think a treatment does and what doctors know it does recurs across medicine.

• 04

  Structural problem 1: huge costs for small gains

  In an analysis by Fojo and colleagues, 71 consecutive drugs approved for solid cancers improved survival by an average of 2.1 months. Prasad says two months can matter a great deal to someone, but the result should come with a lower price. Attia recalls a drug that added 0.7 months, about 20 days, at roughly $38,000. Compared with 1960 or 1970, someone with metastatic breast cancer in 2020 is not dramatically better off, Prasad argues, while surgical outcomes like the Whipple procedure and the move away from radical mastectomy have improved.

• 05

  Structural problem 2: approvals rest on weak surrogate endpoints

  Progression-free survival and response rate often stand in for living longer or better. 'Progression' means the tumour has grown 20% from its smallest size. At 19% the disease is 'stable', at 21% it has progressed: an arbitrary cut-off that often doesn't track how patients feel. The 30% shrinkage needed to count as a 'response' goes back to a 1976 Mayo Clinic exercise in which doctors measured marbles through foam rubber. Prasad questions why so much weight is put on drugs with 10–20% response rates.

• 06

  Structural problem 3: the people choosing don't pay

  Patients and doctors decide while insurers, governments and employers pay. Prasad compares it to a Saudi friend who left his air conditioning running all summer because energy was subsidised. When someone else pays, cost stops affecting choices. In the UK, NICE declined drugs above a cost-effectiveness threshold, which Attia recalls as about £100,000.

• 07

  Hallmark 1: independence from industry money

  Patient advocacy groups are often heavily industry-funded. The most common employer for staff leaving the FDA is industry. Prasad calls expert oncologists, many of whom take payments, 'probably the biggest offender'. He thinks the profession won't police itself, because the most influential doctors receive the most payments. He would have guidelines written by experienced oncologists with no financial ties.

• 08

  Hallmarks 2–3: evidence and relevance

  Drugs should prove they help patients live longer or better, against the right control group. Trials should also enrol patients who look like the people who will actually get the drug. Trial participants tend to be younger and fitter, so real-world success rates are lower because the drug is less tolerable and response is poorer.

• 09

  Hallmark 4: affordability

  Prasad groups possible fixes into buckets. One he highlights is letting Medicare decline to pay for drugs. At the moment it must cover drugs recommended at level 2A or above by guidelines such as the NCCN's, whose authors are often industry-funded. He notes that some states asked HHS for permission to experiment with not covering every drug.

• 10

  Hallmarks 5–6: possibility and agenda

  Possibility means funding more bold science. The NIH budget of about $30 billion a year is 'not nearly enough', grant cycles favour incremental work, and Prasad suggests running randomised experiments on how grants are awarded. Agenda means looking at the whole trial landscape. Many trials test 'Coke, Pepsi and 18 other Cokes', near-identical low-promise drugs, and at p<0.05 some will come out positive by chance.