Sunday, February 1, 2015

Is the Learning Healthcare System Almost Suddenly Here? From ASCO and MolDX?

2007:  We starting hearing the phrase "learning healthcare system," used as the title of an Institute of Medicine report (here) and rapidly entering the health policy vernacular.  In recent years, the idea has particularly circulating in the context of cancer care, where it's been called the innovative relief valve for "a cancer care system in crisis" (for this 2013 U.S. IOM report, here; for the same idea in Canada, 2015, here.)

However, from one source or another, in 2015 and 2016, the big bang for the learning healthcare system may finally be ready to occur.  For a peek-a-boo, see the Med-C website (here).  Details after the break.

Two Recent Article Set the Stage

  • Feeley TW et al. (2014)  Improving the quality of cancer care in America through health information technology.  J Am Med Inform Assoc 21:772-5.  (here)
    • Starts from the 2013 IOM report on new horizons for better cancer care.  Discusses the learning path driven by healthcare IT.  Today's systems are "incompletely implemented, have functional deficiencies, and are not integrated in a way that creates a true learning healthcare system."  But great advances are possible - "clinicians, professional organizations, government, and the IT industry will have to partner..."
  • Abernethy A et al. (2014)  Turning the tide against cancer through sustained medical innovation: the pathway to progress.  Clin Cancer Res 20:1086-6.  (here; see also here.)
    • Cancer is 200 diseases; the promise of personalized medicine; but  "these extraordinary opportunities are occurring in an environment of intense pressure to contain rising healthcare costs."  Cancer is becoming so information-rich it provides a unique opportunity to create an "increased capability to integrate clinical research and care to enable continuous learning."
    • Abernethy is an authority on scale and reliability in cancer data, authoring a major 2010 AHRQ study (press here, report here)and many other papers (e.g. here and here).
    • In July 2014, she became clinical research director of Flatiron, a health IT startup that has raised over $130M (here).  One of the lead investors is Google (here).
One might still wonder when the traction will occur, but several additional documents provide clues that it may be sooner rather than later.

Four Additional Papers Lay the Tracks

March 2014: The Tapestry Networks White Paper

  • Last spring the consulting group Tapestry Networks, which convenes and organizes stakeholders, published a white paper about a two-day workshop, "Sustainable Predictive Oncology Therapeutics and Diagnostics [SPOT/Dx]." Tapestry White Paper, here.
    • The March 31, 2014, symposium included a progress update on the ASCO "CancerLinQ" project, to "provide patients and clinicians with the information and decision support to make well-informed medical decisions."  The project has its own website, here
    • The meeting reports that "CancerLinQ successfully demonstrated proof of concept through a breast cancer prototype that comprised 30 different oncology practices and approximately 170,000 breast cancer patient records. The prototype successfully generated detailed clinical practice data to a number of FDA queries. ASCO has committed to fund the full CancerLinQ system and recently released a request for proposals to solicit vendors to build the platform."
    • The program also provided an early description of another program called "MED-C." 
    • MED-C was presented as a program in draft form.  It envisioned a "multi-stakeholder consortium to identify and generate clinical utility evidence."  The basic model was framed as:   "Payers would reimburse qualified labs for a multimarker diagnostic screen for promising biomarkers preselected by the consortium’s leadership committee. Patients and doctors would commit to following the designed protocols and to tracking outcomes. Pharma would provide the targeted agents.

April 2014: Richard Schilsky's Nature Review Paper

  • About the same time, in the April issue of Nature Reviews Clinical Oncology, ASCO's Richard Schilsky presented ASCO's plans in more detail for an international readership.  (Nat Rev Clin Oncol 11:432-438, here.)  
  • Schilsky describes cancer genomics as rapidly leading to "new paradigms" in cancer care, but as "oncologists increasing struggle with decisions about what tests to order", a much more organized system is required than what we  have today.  The sequence of events starts with the biopsy (volume, tissue handling), defining the molecular workup, transforming the molecular analysis through clinical interpretation, and making the therapy match (or clinical trial match).  
  • Schilsky proposed a system where genomic data is obtained; the patient entered in a nationwide registry like CancerLinQ, and in an interesting spin, an "honest broker" or oncology/genomic panel helps direct the use of drugs.  Outcomes are studied in a registry.
  • He discusses the scale and scope of implementation, challenges, and ways to mitigate the challenges,

November 2014: The Brookings Institute/Friends of Cancer Research White Paper

  • On November 21, 2014, Friends of Cancer Research and Brookings Institute, allied with ASCO and Komen for the Cure, held a conference, leaving behind a lengthy online deck and a substantial white paper Issue Brief (here and here).
  • The Issue Brief "Improving Evidence Developed from Population-Level Experience with Targeted Agents"  is authored by Mark McClellan, Brookings Institution, along with authors from ASCO, Palmetto MolDX, Wellpoint, Genentech, Foundation Medicine, and the FDA.  
  • Several pages are devoted both to the ASCO TAPUR study and the proposed MolDX-related MED-C program.
  • TAPUR (see here) is described as: 
    • "A prospective, observational, non-randomized clinical trial that aims to describe the performance (both safety and efficacy) of commercially available, targeted anticancer drugs prescribed for treatment of patients with advanced cancer that has a potentially actionable genomic variant. The study also aims to simplify patient access to approved targeted therapies that are contributed to the program by collaborating pharmaceutical companies, catalogue the choice of genomic profiling test by clinical oncologists and learn about the utility of registry data to develop hypotheses for additional clinical trials." 
  • MED-C is described as in the Tapestry white paper, but is now further envisioned as a non profit corporation serving multiple stakeholders.  
    • Although the MolDX recent proposed LCD for NGS tests in lung cancer (here) does not mention MED-C, that LCD requires a complex test-drug-registry-payment coalition...lashed together in a coverage with evidence development framework...which could be carried out via MED-C.
    • Med-C has a public website as well, here (accessed 2/23/2015).
In a table, the Brookings/FOCR white paper compares TAPUR and MED-C:

December 2014:  BIOCENTURY Trade Press

  • Finally, the December 22, 2014 issue of BioCentury contains a detailed article on TAPUR and MED-C, by Ted Usdin.   The article covers the same general ground as the white paper, but with additional insight and analysis, and additional interviews with key stakeholders.  (See here.)

 Why It Matters...Why It's Real This Time...

Whether TAPUR, MED-C, the NCI intiative for targeted NGS trials in lung cancer (LungMAP, here; also here and here) or another initiative really gets the most traction isn't the key thing.  The volume of activity and the number of directions it's coming from... from investments in Flatiron to Friends of Cancer Research to Medicare MolDX to NCI...suggests that many stakeholders are seeing the same constellation of factors lining up.   A rising number of targeted oncology drugs are available.  Patients getting NGS gene sequencing panels will soon number in the tens of thousands.   EHR's, even if imperfect, are real now.   The "transaction costs" for getting and coordinating data are lower.   Finally, the values for pharma are easily imagined.   Although it's not going to be label-quality data, signals from scattered patients nationwide with certain genomic fingerprints and different tumors may flag unexpectedly clear early responses from certain drugs.   These drugs can be lifted out and targeted for  study and information-gathering that will result first in NCCN compendium listing and later FDA labeling.  Equally importantly, the flow of otherwise routine clinical data coupled with genomic informatics will flag drug/gene combinations that don't seem to work at all - and which can be deprioritized from study.   Since drugs are cleared (whether by NCCN or FDA) on a drug-by-indication basis, there is a constrained level of damage from the data being public.   If there's evidence that Drug X is 80% likely to be quite effective in Cancer X, the existing pharma still has a sole source ability to produce Drug X.

There is a lot of value to be generated here, and the missing link is the ability to coordinate and harvest it - problems that should be tractable.  There is so much value here, and it is so close to the surface, that somebody will figure out how to make it work, and sooner than we might have expected just a few years ago.   A train is rumbling around the bend, and from one track or another, it's going to be here soon.   2/23/2015

For an only-lightly selected list of 30-odd recent PUBMED citations on "real world evidence," see here.

For related work, not necessarily in cancer, see also the "National Patient Centered Clinical Research Network," see "About PCORnet" here, and a listing of a dozen-plus patient centered research networks, such as DuchenneConnect Patient-Report Registry Project, see here.

June 2015 update.  For Genomeweb subscription stories on the June launches of TAPUR and NCI-MATCH, see here and here.