Oncologists Are ‘Sacrificing Common Sense’ in Pursuit of Innovation

As a young oncologist, I have come of age in an era of incredible optimism in cancer care. But with this I have also noticed a number of troubling trends. I fear that in our pursuit of innovation, we are sacrificing common sense, forgetting what actually matters to our patients. Instead, we celebrate the number of drug approvals as a marker of success, as if it meant anything in itself. We have become so focused on the road that we have forgotten our destination.

Cancer patients don’t just need more drugs; they need ones that give them better and longer lives. However, when it comes to a drug that we like, or have received industry money to like, we seem to think that testing it in a randomized controlled trial (RCT) is unethical. And if it fails in an RCT, we think there was some problem with the trial rather than with the drug. Instead of developing better drugs, we believe in developing innovative trial designs that can show every drug as beneficial (eg, noninferiority trials).

It’s true that I have complained about this issue already. But its importance merits repetition. When you are swimming in the sea of hype alongside top oncologists who celebrate every new drug, even the failed ones, it is easy to forget that data must be viewed critically. It is also vital to remember that encouraging good practices is as important as discouraging bad ones.

With that in mind, I’d like to praise the recently published GOG-0213 trial, which tested the efficacy of secondary surgical cytoreduction in recurrent ovarian cancer with a platinum-free interval of 6 months or more. These researchers have done the commendable (and long overdue) work of critically assessing something that had become accepted practice, despite having never before been tested in an RCT. To date, many women have already undergone secondary cytoreductive surgery, an approach recommended by the National Comprehensive Cancer Network guidelines. I can only imagine how difficult the recruitment was, given doctors’ fears that their patients might be randomly assigned to the no-surgery arm receiving chemotherapy alone.

Results from the GOG-0213 trial showed that, compared with chemotherapy alone, surgical cytoreduction actually had a detrimental effect on survival (hazard ratio, 1.29; 95% confidence interval [CI], 0.97-1.72), corresponding to a 14-month shortening of median overall survival (50.6 vs 64.7 months). This difference was not statistically significant (P = .08). Looking at the confidence interval, however, even the most optimistic take here would be a reduction in the hazard of death by a mere 3%, whereas the most pessimistic take would be an increase by 72%.

Consider how astonishing that 14-month median survival difference really is. It’s rare to see an overall survival difference of more than a year in oncology—more so for a detrimental effect.

When Practice-Changing Data Meet Reluctant Clinicians

Ideally, such results would teach us some humility and lead to the kind of introspection that changes our practice. We might pledge not to fall into the “it makes sense” bias again and take an oath against recommending any invasive intervention—drugs, surgery, radiation, whatever—unless we know for sure that it will help the patient.

But I heard some very different arguments when these results were published. There were contentions that certain subgroups would still benefit from surgery, criticism of the trial’s patient selection, and declarations from some oncologists that they would continue to offer surgery to these patients. One oncologist told me, “Bishal, if this was a drug trial that showed a hazard ratio of 0.85, with a 95% CI of 0.83-1.10, you wouldn’t say that the drug is beneficial. So how can you say here that surgery is harmful? There was no statistical significance.”

We need to return to a fundamental concept of medicine: Interventions are supposed to benefit patients, not harm them. In fact, the threshold to not recommend a practice is justifiably lower when it may harm a patient. Why would anyone want an intervention when the evidence for harm wasn’t conclusive? Let’s remember that the difference in median survival is 14 months shorter with surgery. For an intervention as invasive as surgery (patients’ quality of life deteriorated significantly during the first few weeks after surgery), there should be undisputed evidence of benefit. Instead, there is zero evidence of a benefit from surgery for these patients. Whatever evidence we have points toward harm, and it’s substantial harm.

Building Better Clinical Trials in Oncology

GOG-0213 received funding from the National Cancer Institute and was under the sponsorship of the academic Gynecologic Oncology Group. That this background led to a truly practice-changing trial should not come as a surprise. We published a 2019 research paper highlighting the importance of public funding, academic sponsorship, and non-drug trials in improving the survival of lung cancer patients. Such trials answer clinically important questions. They are also written by the authors themselves, leading to more thoughtful manuscripts with less spin than manuscripts for industry-sponsored trials traditionally authored by medical writers.

GOG-0213 is not a flawless trial. The fact that it used CA-125 levels to assess progression certainly caught my eye. We have hard evidence that routine assessment of CA-125 is not helpful, and it has even been referred to as “The Fatal Attraction of Testing.” Even in this forward-thinking trial, getting rid of an ineffective routine practice proved difficult.

But the GOG-0213 investigators still deserve praise for all the hard work they have done in questioning a dogmatic practice with a publicly funded RCT. They rightly chose overall survival as the primary endpoint while also including quality of life as an endpoint, which in nearly half of oncology drug clinical trials is either not assessed or not reported.

GOG-0213 is also a lesson on the need to exercise caution with surrogate endpoints in oncology. The 0.82 hazard ratio for progression-free survival (PFS) hints toward benefit in this trial (95% CI, 0.66-1.01), as does the median PFS difference of 2.7 months (18.9 vs 16.2). But as we’ve seen, this seemingly beneficial PFS is masquerading the possible substantial detriment seen in the primary endpoint of overall survival. For a moment, imagine the harm caused if the GOG-0213 investigators published only PFS data and overall survival remained unreported. What scares me is that there are many cancer drugs with only PFS data reported but with no overall survival data yet reported, and which we nonetheless feel comfortable using every day in the clinic.

Preventing harm to patients with cancer is as important as providing a new therapy to them. GOG-0213 stands alongside the similarly publicly funded, academia-sponsored LION trial in offering important, apparently negative results in gynecologic oncology, potentially saving thousands of women with ovarian cancer from unnecessary surgery.

For trials like these to truly have an impact, we must actually change our practice on the basis of their results. If data go against our hypothesis, we should question the hypothesis, not the data.

Bishal Gyawali, MD, PhD, is an assistant professor in the Department of Public Health Sciences, a scientist in the Division of Cancer Care and Epidemiology, and a clinical fellow in the Department of Medical Oncology at Queen’s University in Kingston, Ontario, Canada, and is also affiliated faculty at the Program on Regulation, Therapeutics, and Law in the Department of Medicine at Brigham and Women’s Hospital in Boston. His clinical and research interests revolve around cancer policy, global oncology, evidence-based oncology, financial toxicities of cancer treatment, clinical trial methods, and supportive care.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube