Vinayak K. Prasad

“Censoring Measuring response rate and progression offers more problems than measurement error. We may not be considering the right denominator of patients. In 2017, the FDA approved the first cellular cancer therapy, called tisagenlecleucel (Kymriah, Novartis), or CAR-T, for short. A CAR-T is a chimeric antigen receptor T-cell, basically a genetically modified cell taken from a patient that is trained to attack cancer cells and then placed back in the patient. In the data submitted to the FDA, 88 patients had the cells removed, but 18% (16/88) did not receive the cells because some patients died and some patients’ cells could not be manufactured.17 Unfortunately, the FDA excluded these patients from the denominator and assessed response only in patients who got the cells. This violates a principle called intention to treat, that is, you should judge a drug based on all patients allocated to get it, irrespective of whether or not they received it. Why? Because therapies that take a long time to give (this CAR-T took approximately 22 days to make) may exclude the sickest patients who die while waiting, thus distorting their benefit. In fact, if I have a patient in my office and we decide to treat with tisagenlecleucel, the response rate from the package overestimates her chances of success, as I am unsure she will live long enough to receive the cells.”
― Vinayak K. Prasad, Malignant: How Bad Policy and Bad Evidence Harm People with Cancer