‘How Could We Fail so Miserably?’ Real-World Evidence Offers Sobering Reality
We are doing an unconscionably poor job of delivering appropriate targeted therapies to our patients. That’s the only conclusion I can draw, based on a recent report by Foundation Medicine and Flatiron Health.
The analysis used “real-world evidence” of cancer outcomes—in this case, among patients with non–small cell lung cancer (NSCLC) who underwent next-generation sequencing (NGS) testing.
Unlike the report’s authors, I won’t bury the lede: Fewer than half of patients with a National Comprehensive Cancer Network (NCCN)-recognized driver mutation received a targeted therapy.
The paper describes results from a database of 28,998 patients from 285 US oncology practices, 85% of which were community-based settings, and included 4064 patients with NSCLC. Rather than being a highly enriched, skewed population of younger never-smokers, this group was a fair representation of the general NSCLC population: a median age of 66, 78% with a smoking history, 52% women, 77.6% with nonsquamous histology, and 21.4% with an EGFR, ALK, or ROS1 mutation. At the time of analysis, whether at initial diagnosis or subsequent relapse/recurrence, advanced disease was present in 86.7% of patients over the median 34 months of follow-up. The data on both patient characteristics and treatments administered were very comprehensive.
The study provides benchmark results for the molecular genomics patterns in a much larger cohort of NSCLC than can be described in even the largest institutional database, without many surprises.
But what I think was most interesting—and by that, I mean appalling—is that only 64.3% of patients with an EGFR mutation (405 of 630 patients) and only 70.1% of patients with an ALK rearrangement (75 of 107 patients) received the appropriate targeted therapy at any time for their advanced NSCLC. When tallying the use of targeted therapy among the wider array of driver mutations specified by the NCCN (N = 1260), only 48.3% of patients (n = 609) ever received a targeted therapy—meaning that a recommended targeted therapy was delivered to only 32.9% (179 of 543) of patients with a less common targetable mutation.
These results do not speak to the huge population of patients who never had molecular marker testing, whether because they had insufficient tissue or because it simply wasn’t tested. We are clearly missing many patients who fly under the radar.
But that makes this a more egregious omission, because these are the patients who already had testing that revealed a biomarker associated with an effective therapy. They already hit the ground-rule double, if not a home run. They just never ran the bases.
How could we fail so miserably? One third of patients who qualify for well-established targeted therapies, and two thirds of those with less common but still evidence-based options highlighted by the NCCN, are not receiving the benefit of some of our most effective treatments for advanced NSCLC.
I have a hard time even explaining these haunting findings. Yes, some patients had a rare mutation that may not have been the most responsive subtype, such as an uncommon EGFR mutation, but that would account for only a very small minority. Perhaps some could not afford the high cost of a targeted therapy, but there are many copay assistance or free drug programs, and this cannot account for a significant fraction. Some may have refused treatment altogether, while a few may have had severe contraindications, though it is difficult to envision a patient in whom the risk of these treatments would exceed the anticipated benefit. Could the siren song of direct-to-consumer advertising and mass-media hype around immunotherapy have led many to favor this strategy, even though it is a fool’s errand for these particular patients?
There are other potential contributing explanations. But if they help explain this gap in treatment, they are even more of an indictment of our healthcare system. It’s likely that some patients died before starting targeted therapy, though I hope this wasn’t a common explanation. It is possible that some oncologists weren’t getting NGS reports or received them weeks after starting patients on initial chemotherapy, eventually forgetting about their patient’s driver mutation by the time of progression.
Because the patient data were de-identified, there is no way to get further information from the treating physicians. But regardless of the reasons, the main conclusion is inescapable: Far too many patients with driver mutations are dying without receiving treatments that are among the most effective that we have for lung cancer, or many other malignancies, for that matter. As an oncology community, we should take a hard look in the mirror and reflect on how deeply we are failing these patients.
What good is it to be able to cure cancer if we routinely fail to deliver that cure?
I welcome your thoughts on whether there’s something I’m missing—please comment. I cannot fathom how an oncologist could send for molecular testing, receive a result that is a clear and compelling mandate to initiate a well-defined optimal therapy, and then have a very significant minority of patients miss that screaming opportunity.
These results should also remind us of the wisdom of focusing our attention on executing the proven interventions already available to us, rather than constantly developing ever more complex treatment algorithms. These advances need to actually reach our patients, a concept that has been cleverly coined a “cancer groundshot,” in contrast to the gravitational pull of the famed “cancer moonshot.” What good is it to be able to cure cancer if we routinely fail to deliver that cure?
The challenge of disseminating the best treatments is what led me to found a cancer education nonprofit for patients, Global Resource for Advancing Cancer Education (GRACE), more than a decade ago, featuring free, credible information from physician experts for the lay public. If our patients knew what their oncologists should know, those with a driver mutation would never miss the opportunity for targeted therapy. However we achieve it, we need to ensure that at least one person in the exam room knows what the optimal treatment is and then acts on that knowledge.