When Shooting at the Moon, Take Careful Aim Before Firing
February 23rd, 2016
Posted by Luis Gutierrez
During his final State of the Union address, President Obama announced he is charging Vice President Biden with leading a new “moonshot” program to accelerate the pace of improvements in cancer care. While the goal is certainly laudable, I share the “sense of déjà vu” that was eloquently expressed by Vinay Prasad in his recent editorial1 in which he recounts the limited progress seen in the Nixon-era War on Cancer, and even more recently in the G.W. Bush era, when then-NCI Director Andrew Van Eschenbach testified that we could rid the world of cancer by 2010 for just $600 million per year.
Dr. Prasad goes on to rightly point out that an overly-onerous FDA is not the issue. To the contrary, the issue is that the 71 drugs FDA has approved between 2002 and 2014 for treatment of solid tumors deliver a median improvement of survival time of just 2.1 months. The FDA can’t set the bar much lower than that.
Instead, Dr. Prasad concludes that because “scientific discovery is hard to predict and happens in serendipitous and unexpected ways… A serious moonshot would require funding science broadly, consistently, and in steadily increasing amount.” That sounds nice, but flies in the face of global fiscal realities. We cannot simply give ever-more funding to everyone.
Taking careful aim
We need to decide where and how to target the funding. Put another way, we cannot afford to shoot in all directions in the hopes that one shot will hit the moon; we need to take aim more carefully before shooting.
Recognizing that so much of medical research occurs in silos — be they academic, governmental, or commercial – a key initial component of the “moonshot” program is an initiative to encourage, and even require, more transparent sharing and pooling of research results to enable the application of “big data” analytic techniques to identify which treatment approaches are most likely to prove effective at the individual level by analyzing thousands, if not millions, of cases that are superficially similar (e.g., “Stage III breast cancer”) but may be quite distinct at the molecular level (e.g., “ER-/PR-/HR- Stage III Breast Cancer in the context of three well-characterized gene mutations and an over-expression of the mTOR pathway.”).
While data-pooling mandates threaten to upset the status quo for those in all three sectors who have historically guarded the data they generate, the pooling of such data is an essential prerequisite for enabling truly “precise” medicine. In short, I strongly support efforts to use the collective experience and outcomes of patients that have come before to help “aim” our therapeutic armamentarium more accurately.
As just one recent example, Piero Dalerba and colleagues recently reported2 on a collaboration with a large research consortium funded by the National Cancer Institute to gain access to tissue and to clinical trial results on an individual patient basis. They found that a small proportion (four percent) of colon cancers did not express the key biomarker of interest (CDX2) and that when the effect of adjuvant chemotherapy was assessed, nearly all the benefit from adjuvant treatment was within the biomarker-negative group.
As interesting as this result is, an accompanying op-ed by Dan Longo and Jeffrey Drazen3 has created quite a dust-up4 over the issue of data sharing by characterizing those who pool study results as “research parasites,” a term they have subsequently toned down in a follow-up piece5 affirming their commitment to collaborative data-sharing.
Here’s what I think is getting lost. Dr. Dalerba and colleagues should be applauded for spending more time investigating the key question of precisely who benefits from a given therapeutic intervention, in this case adjuvant chemotherapy.
While it still needs to be replicated in a prospective study, their retrospective analysis indicates that as few as 4% of those with colon cancer – those who are CDX2-negative — benefit from adjuvant therapy. Put another way, the remaining 96 percent do not benefit from it. With a result like that, it makes a lot of sense to ascertain CDX2 expression status before initiating the adjuvant therapy.
Create actionable information
Popper and Company’s mission is to help all members of the healthcare ecosystem to create actionable information from data like this to improve healthcare outcomes. Getting back to the rather unimpressive 2.1 month median increase in survival time among the 71 cancer drugs approved between 2002 and 2014, could it be that we simply are not aiming with sufficient precision, and thus missing the target more often than hitting it, all in the hopes of getting as broad an indication as possible? After all, drug company revenues are driven by the number of doses given (or “bullets fired”) rather than number of patients benefiting from therapy (“targets hit”).
We’re all for shooting at the moon, but we strongly advocate spending at least as much time, effort, and money in aiming (that is, diagnostics), as we do in firing (new and generally expensive therapeutics).
- Prasad, Vinay, “What a True Cancer Moonshot Entails.” The Washington Post. January 31, 2016. P. A23. https://www.washingtonpost.com/opinions/why-a-cancer-moonshot-is-unlikely-to-find-us-a-cure/2016/01/29/08cc66dc-c545-11e5-8965-0607e0e265ce_story.html
- Dalerba, Piero, et.al. “CDX2 as a Prognostic Biomarker in Stage II and Stage III Colon Cancer.” New England Journal of Medicine. January 21, 2016. 374: 211-222. http://www.nejm.org/doi/full/10.1056/NEJMoa1506597
- Longo, Dan and Drazen, Jeffrey. “Data Sharing.” New England Journal of Medicine. January 21, 2016. 374: 276-277. http://www.nejm.org/doi/full/10.1056/NEJMe1516564
- Allen, Arthur. “Biden’s Cancer Bid Exposes Rift Among Researchers.” Politico. January 31, 2016. http://www.politico.com/story/2016/01/joe-biden-cancer-researchers-rift-218465
- Longo, Dan and Drazen, Jeffrey. “Data Sharing and the Journal.” New England Journal of Medicine. January 25, 2016. http://www.nejm.org/doi/full/10.1056/NEJMe1601087
Like this content? Don’t miss an update.