When I was diagnosed with advanced prostate cancer in 2009, I ran to the Internet to learn as much as I could about it, its treatment, and of course, its implications. Article after article described how prostate cancer was invariably linked to an elevated PSA. I had been diligent about monitoring my PSA annually for the previous 7 years and it had been a steady 1.5 every year. And it was still 1.5 when I was diagnosed. What was going on? How could I have advanced cancer when my PSA had never changed?
After hours searching abstracts of obscure academic papers online, I finally found one that indicated that about 3 in 100 men are diagnosed with an aggressive (and mine was aggressive) version of prostate cancer that does not increase PSA. While it was gratifying to know that while my case was unusual, it had been a stressful and frustrating search to uncover this information. Why couldn’t I have gone to a centralized repository or database and found this out sooner and more easily?
A major stir occurred at the 2014 annual meeting of the American Society of Clinical Oncology (ASCO) when a study was presented proving that men with metastatic prostate cancer, who received the standard androgen deprivation therapy (ADT) together with the chemo agent docetaxel, outlived those given ADT alone by about 17 months. In the world of metastatic cancer, that’s an astounding advantage. Up to now, “ADT only” has been the “standard of care,” with chemo used only after ADT has stopped working.
I had watched my friend, Bill, die of aggressive prostate cancer in 2011, while his oncologist kept saying, “Chemo won’t help” and kept him on the ADT-only regimen because that was “the standard of care.” Now we find out that chemo given sooner may have enabled him live longer. (Of course, when it comes to cancer, we can’t say for certain it would have helped.)
Docetaxel’s been around a long time. So has ADT. Why are we discovering just in 2014 that doing chemo with ADT rather than waiting until ADT fails is quite beneficial? One (but not the only) reason is that clinical trials often take years to complete. However, I’m sure there were oncologists out there thinking outside the box, who may actually treated their patients with the ADT-chemo combination. But if they did, we never heard about them because their patients weren’t enrolled in a formal trial. What if we had known there had been several patients, each at a different cancer center or oncologist’s office scattered around the country that had received a similar treatment and had similar outcomes? Would we have found hints of the ADT-chemo advantage sooner?
While these two incidents–my personal experience of cancer without a PSA rise and the recent ADT-chemo study–may seem unrelated, they one big thing in common. It is the “knowledge gap” created by what I call “cancer data silos.” Even though hundreds of thousands of cancer patients are being treated at any given time, there is little cross-fertilization of diagnostic knowledge and outcomes at the individual patient level.
Cancer treatment “standard of care” protocols are driven by clinical trials such as the ADT-chemo study just completed. But just 4% of cancer patients participate in clinical trials. That means the results of 96% of cancer treatments actually performed on patients are unavailable beyond the walls of the local oncologist’s office or cancer center. Invaluable information that could be useful to treat other patients with the same set of conditions is hidden away in individual data silos.
As ever-increasing numbers of drugs and protocols come online, which one to use and what order in which to use them is becoming an enormous issue. And because there are so many “oncotypes” for each cancer (at least 27 for prostate cancer alone), patients are going to respond differently to each drug and each sequence of drugs. Do we have to wait around for for years-long, small-sample clinical trials evaluating all these treatment permutations in order to define what will inevitably become a growing multiplicity of “standards of care?”
Access to that 96% is evolving from merely desirable to essential.
Obviously, there are obstacles to tearing down these silos and sharing information. Privacy issues (HIPAA) require “anonymizing” data, i.e., removing all information that would identify the patient involved. The data structures of electronic medical records (EMR) systems used by different treatment centers are typically incompatible. Then there are the inevitable bureaucratic turf wars and politics (yes, even in cancer treatment…) that make organizations reluctant to share their data with others.
Nevertheless, extracting even a portion that 96% of patient data out of the silos and getting it into the hands of medical professionals is the ambitious goal of Flatiron Health, located in New York City. The company is developing an anonymized database of patient outcomes that is available in the proverbial “cloud.” According to a recent article in the Wall Street Journal, so far they’ve got specific data about more than 550,000 cancer cases in their database.*
More than 200 cancer centers in the US are using and/or testing the Flatiron data. Right now, the company’s OncoAnalytics(tm) database contents are available only to doctors and medical facilities. Their web site implies it will eventually be available to patients, as well.
The Flatiron product is exciting because it will an effective way to free up cancer care from its 20th century sole reliance on lengthy clinical trials as the exclusive source of documented patient outcome data. Creative clinicians and willing patients now will have an additional tool to gather relevant information, encouraging outside-the-box thinking when it comes to cancer diagnosis and treatment.
As the owner of an advanced cancer that could return at any point, the idea of potential treatment alternatives confined neither to silos nor boxes, but which are based on the outcomes of more than 4% of the individuals who came before me, is both exciting and comforting.
*http://online.wsj.com/articles/ad-tech-entrepreneurs-build-cancer-database-1403134613? (Article is behind the WSJ paywall.) Accessed 6/24/14