Contribute Try STAT+ Today

The rapid pace of new cancer drugs approved by the FDA, 16 in 2018 alone, is — on the surface — good news for people with cancer. But some of these approvals were based on clinical trials with substandard control arms, calling into question the effectiveness of some new drugs.

Drug approvals generally require randomized clinical trials in which an investigational agent is compared to a control, usually an established treatment. If the investigational agent is superior to the control, FDA approval is likely.

The choice of the treatment received by participants in the control arm is essential. It should be the current best standard of care. But it isn’t uncommon for the established agent chosen as the standard of care to be outdated, or something that practicing oncologists rarely or never use. When such “straw man” comparators are used, the newly approved drug wasn’t really compared to the standard of care. That can leave oncologists and patients wondering if a new drug is better than existing treatments, which may also be less expensive.


To measure this phenomenon, I led an independent review of all randomized clinical trials that culminated in FDA approval of new cancer drugs between January 2013 and July 2018. With my colleagues Bassam Sonbol and Vinay Prasad, we first reviewed guidelines for cancer care, such as those from the National Comprehensive Cancer Network and review articles on cancer written by experts in the field. We chose those that had been published at least one year before participants began enrolling in each of the seminal clinical trials leading to FDA approval.

We then compared the recommended standard therapy to the comparison therapy chosen for the various trials. As we reported Thursday in JAMA Oncology, among 96 cancer drugs approved by the FDA, 16 (17%) had been tested against something arguably inferior to the best available care. (The 16 drugs and the indications for which they were tested are in the table below.)


To give clinical trial organizers the benefit of the doubt, we re-ran the analysis using guidelines from two years before enrollment began, and still showed that 15% of trials had used suboptimal control arms.

The take-home message from our analyses is that FDA approval doesn’t automatically make a new drug better than treatments doctors are currently using. Take, for example, the case of ibrutinib, a drug the FDA approved for treating chronic lymphocytic leukemia. The treatment chosen for the control arm of the RESONATE-2 trial was chlorambucil, a drug that had repeatedly been beaten by alternatives and isn’t commonly used by practicing hematologists.

In our analysis, most of the approvals based on randomized controlled trials with suboptimal control arms received regular or full FDA approval, and so don’t require additional trials to verify their clinical benefit. That’s a problem: if an experimental agent has not been proven to be superior to the established standard of care, clinicians are potentially offering patients a treatment that may be no better than the standard of care — or even inferior to it — usually at a higher cost and with alternate or unknown side effects.

Another concern is that suboptimal control arms may more quickly accrue events such as progression of disease or death, making experimental agents look better sooner and hastening them to market.

To be sure, choosing a treatment for the control arm of a cancer trial, or any trial for that matter, can be somewhat subjective. So its up to the FDA to set the bar for drug companies designing clinical trials for what constitutes standard-of-care treatment.

The choice of therapies the FDA considers as standard treatment should reflect the reality of clinical practice — namely that the standard of care changes rapidly — and routinely raise the bar that new investigational agents must clear. Better choices for control arms would help ensure that newly approved drugs are truly superior to the standard of care, and allow these drugs to gain wider acceptance by the oncology community. Otherwise, the established standard of care could be just as effective and less toxic as a new drug, but we may never know with substandard control arms.

People being treated for cancer must often withstand physical and financial toxicity. It is only right that any new treatments offered to them must have been proven to be better than what is already available.

Talal Hilal, M.D., is a hematology/oncology fellow at the Mayo Clinic.

Drugs (in alphabetic order) that were approved by the FDA following randomized controlled trials that our analysis showed employed suboptimal control arms:

Drug name Indication
afatinib first line metastatic non-small-cell lung cancer with EGFR exon 19 deletions or L858R mutations
apalutamide nonmetastatic castration-resistant prostate cancer
bosutinib first-line chronic chronic myeloid leukemia
brentuximab vedotin previously treated CD30-expressing mycosis fungoides
brigatinib metastatic ALK-positive non-small-cell lung cancer after brigatinib
cabozantinib advanced renal cell cancer following one line of therapy
ceritinib first line metastatic ALK-positive non-small-cell lung cancer
everolimus advanced, non-functional, neuroendocrine tumors
ixazomib (+ lenalidomide + dexamethasone) multiple myeloma after at least one line of therapy
obinutuzumab (+ bendamustine) follicular lymphoma following relapse after rituximab
obinutuzumab (+ chlorambucil) previously untreated chronic lymphocytic leukemia
olaparib germline BRCA-mutated triple negative breast cancer
olaratumab first-line metastatic soft tissue sarcoma
pembrolizumab metastatic urothelial cancer with progression on platinum
ramucirumab advanced gastric or esophagogastric junction adenocarcinoma with progression on platinum
venetoclax chronic lymphocytic leukemia patients who received at least one line of therapy
  • There is another problem discovered here – If substandard care is being used for the cancer patients in the control arm, then this also highly unethical, possibly even criminal. I would urge Dr. Talal to take his research a bit further and actually report the physicians who used these outdated drugs on cancer patients for their control arm, which likely resulted in many more unnecessary deaths of their patients in the control arm, to their respective state boards and the FBI.

  • Thank you to the authors for distilling detailed, deep analyses to succinct and thought-provoking takeaways. The single sentence from Mark Hochhauser is a 2 x 4 right between the eyes. The system wants to have more treatments available to show that the system is helping people. Developers of treatments need to reduce the risk of clinical trial failure and increase the odds of positive results. When these underlying rip currents carry us far from shore, we drown. The authors and others are pointing out the dangerous conditions and paddling out to help.

  • Having spent many years on a hospital IRB, I wonder how the control arms were explained to prospective subjects in the consent process and consent forms.

Comments are closed.