Pathways: How Do We Know They Got It Right?

Pathways: How Do We Know They Got It Right?

Peter Paul Yu, MD, FASCO, FACP

Oct 21, 2013

By my watch, I am somewhere over Lincoln, Nebraska, flying back from the ASCO Provider-Payer Initiative (PPI) Meeting. It’s the third annual PPI meeting and the fact that it exists is significant evidence of ASCO's ability to adapt in response to circumstances and of the increasing resources that ASCO is investing in support of clinical practice. Aetna, United Healthcare, Cigna, WellPoint, and other private payers were invited. Ten years ago, the suggestion to expand the scope of payer discussion beyond CMS by hosting meetings with private payers was considered unfeasible. The First Amendment would be solemnly evoked as the Constitutional basis permitting discussions with a federal agency under freedom of speech that did not apply to private payers. To be sure, anti-trust concerns are real whenever payers and providers meet together, and ASCO arranges for a health law professor to be present throughout the PPI meeting to monitor the discussions for this reason.

The focus of this year’s PPI was the role of pathways in oncology practice.  Pathways are commercial products that provide guidance on chemotherapy selection, although they can be used for distilling clinical guidelines in other aspects of cancer care such as use of radiation, molecular testing, or palliative care. Five pathway products were presented, and all follow a similar development process. The starting point is a review of the evidentiary base, which includes published literature, compendium, and guidelines from sources such as NCCN and ASCO. Expert panels are then brought to bear to evaluate the evidentiary base, often a mixture of academic and community oncologists. Considerations of efficacy, toxicity, and sometimes cost are applied to rank order chemotherapy regimens. Integration of these pathway products into electronic medical records would provide useful clinical decision support, and efforts to do this are underway. Developers of pathways do not intend that their pathway products be rigidly used, although some products require that a provider deviating from the pathway provide a reason for so doing. They understand that clinical judgment and patient realities need to be respected.

On the payer side, pathways were viewed favorably and some payers are starting to waive pre-authorization of chemotherapy if a proposed treatment is on a pathway. Some of the private payers were outspoken on their view that universal compliance with pathway recommendations should be expected and demanded to know why oncologists would feel otherwise.

ASCO members, in turn, vigorously argued that a patient-centric approach would at times require clinically legitimate deviation off of a pathway. Thoughtful integration of the pathways into the workflow of oncologists so as to not detract from provider-patient face time is a necessary condition for user acceptance. Of note, one of the pathway developers presented evidence to refute the idea commonly held in some quarters that economic gain drives physician choice of chemotherapy and suggested that physician behavior is far more complex and subject to multiple factors.

The question I posed to the pathway developers was this. How do they know that they have gotten it right? In the end, the pathway product is the result of expert opinion attempting to supplement an incomplete evidence base with a best guess.  There is absolutely nothing wrong with that as a starting point; one needs to drive a stake in the ground somewhere, but it should be a point of departure and not a destination. We need a way to validate that the pathway construction was correct and adjust if not. Without the reassurance that the system design has an inherent ability to correct the course it is plotting, physicians will hesitate to walk down that path. Dr. Larry Norton, speaking about another investigator, once remarked to me that the investigator in question believes his own data. It is a human trait to believe most what we see with our own eyes and touch through our own experiences. And that instinct is not without some merit.

This is precisely what a rapid learning health system can provide us, the reassurance that our own experiences count and contribute to collective wisdom.  Rapid learning health systems value the experiences of all providers and draws learning from the aggregated data of our electronic medical records. If going off pathways still leads to good patient outcomes, we have learned something important and in the end have an improved pathway that others can follow. To get there though we will need to agree upon a new set of outcomes that add to those that journal editors and the FDA value. It is certainly critical to have survival data for drug approval, but this figure has lesser meaning to an individual patient who may be more concerned about control of symptoms and that elusive quality of life assessment. For CancerLinQ to succeed the entire oncology ecosystem will need to contribute data, including in this case pathway developers and payers.  We will need to re-think and accept new concepts such as patient-centered outcomes. There is no point in building CancerLinQ to ossify what we are already doing. More than a registry, CancerLinQ is an opportunity to challenge ourselves to apply ingenuity to innovation and illumination to invention.


The ideas and opinions expressed on the ASCO Connection Blogs do not necessarily reflect those of ASCO. None of the information posted on is intended as medical, legal, or business advice, or advice about reimbursement for health care services. The mention of any product, service, company, therapy or physician practice on does not constitute an endorsement of any kind by ASCO. ASCO assumes no responsibility for any injury or damage to persons or property arising out of or related to any use of the material contained in, posted on, or linked to this site, or any errors or omissions.


Robert E. Fisher, MD

Oct, 22 2013 11:53 PM

      Peter, your essay resonates with what I have witnessed in my career in the community.  However, at some level, I think that society is asking us to accept "a safe and reasonable outcome," in place of the potentially "best outcome" for our patients.  We have come to refer to, and accept the NCCN algorithms that become a standard weekly reference in routine Oncology practice.  The list of NCCN  published expert clinicians seem beyond reproach to a doctor such as I in the community.  I think we impart the same sense of trust in the development of pathways, also.  Having been in a USOncology practice for 8 years, I have seen the collective wisdom of streamlined, (hopefully) evidence-based care incorporated into our pathway treatment choices.  Our group in Colorado allows the clinician to present a rationale for an off-pathway treatment choice, followed by a vote among peers to approve or disapprove the off-pathway treatment.
        Could one weight the clinical evidence differently and come up with a different treatment standard?  I am certain of that.  Is pathway creation a melding of personalities and medicine by committee?  I am certain of that, too. But, at a time when the basic and clinical science seems to be growing exponentially, isn't there some comfort for us and for society to know that some of the rationale for treatment planning has moved into the minds of clinical experts in our profession?  
         Pathway treatment planning may be one of the means by which Hematologists/Oncologists can maintain our autonomy within our profession, as society becomes more aware of the limitations of our financial resources for health care, and as more $ 100,000 a year "medical breakthroughs" arrive, few of which appear  truly "curative."
          I favor new understanding and new outcomes from CancerLinQ, but for the current time, I think that expert-panel clinical pathways help most of us to provide high quality, evidence-based care. 

Rob Fisher, MD
Rocky Mountain Cancer Centers@ Longmont, Colorado 

Peter Paul Yu, MD, FASCO, FACP

Oct, 23 2013 3:19 AM

Hi Rob,

I completely agree with your comments.  We are in a time where consensus judgment and careful weighing of the entire evidence base should  guide what we do and the choices we make.  And, consideration of the cost of care is a part of that.  With medical costs accounting for an estimated third of personal bankruptcies (most of whom had health insurance of some sort) we are responsible for making sure that our patients understand the costs of care options.

The recently initiated NCCN-McKesson pathways offer to do just that.

What I was am trying to do is to drive the discussion beyond that, to look into the future and see how we can do even better in designing the pathways which will impact both what our patients pay and how oncologists are re-imbursed.


Frederick E. Millard, MD

Oct, 24 2013 9:31 PM

Hello Peter,

Part of the dynamic with pathways at present is the focus on personalized/precision medicine, and the perception in some quarters that the two approaches conflict.  I don't personally see why pathways can't incorporate genomic/biomarker driven treatment selection, but this requires evidence about outcomes using the precision medicine approach that is thin on the ground at present.  We all know the remarkable successes such as EGFR/erlotinib and EML-ALK/crizotinib in NSCLC, but in the general oncology population routine use of tumor genome sequencing very often gives information of uncertain value.  The technology, as always, proceeds ahead of the actual knowledge of how and when to best use these potentially powerful methods.  Nonetheless, a certain aura of evangelism accompanies the precision medicine discussions at times.  I wonder if you can comment on how CancerLinQ specifically can contribute to this issue? 

Fred Millard
Moores UCSD Cancer Center

Peter Paul Yu, MD, FASCO, FACP

Oct, 26 2013 6:55 PM

Hi Fred,

As technology increases the speed and lowers the cost of sequencing, physicians are being exposed to marketing efforts to order these tests routinely on cancer patients without good knowledge about how to use the test results and absent established insurance coverage for the targeted agents available for the genetic targets listed on the test results.  Detection of a mutation or altered gene expression alone does not mean that this gene is the critical driver of carcinogensis because of a variety of reasons one might imagine.  Geneticaly unstable cancers may acquire mutations that are background noise, cross talk between signal transduction pathways may circumvent attempts at growth signal blockade with targeted therapies or heterogeneity among metastatic clones may mean that the sequencing findings relate to only the biopsied site and not other sites of metastases.

If computational biology was combined with tumor sequencng and expression data, we might be able to find common genetic alteration models that the majority of organ defined cancers fall into and enable us to construct cancers models consisting of set patterns of genetic disruption that enable prediction of combination of targeted agents that would be effective against the cancer in question.  Alternatively, computational biology may be able to mechanistically map out signal transduction pathway cross talk and provide rationale guidance on which targeted agents should be combined.

I do not see CancerLinQ playing a direct role in this, certainly not in it's earliest versions.  The science isn't there yet as you point out.  Furthermore the regulatory apparatus such as FDA policy allowing more flexible and adoptive study design and the pharmaceutical industry willingness to provide drugs to be used in combinational studies is lacking.  But, if such a day came, and it might, aggregating patient data would allow so called weaker signals to be picked up. In particular, non genetic, clinical factors that might help determine response such as prior treatment exposure, family cancer history, diet or lifestyle factors, drug-drug interactions could be uncovered through Big Data analytic approaches.

I am reasonably sure that the evolution of CancerLinQ will be dependent upon on what is going on in the cancer world and that CancerLinQ to be successful will need to adopt to the opportunites and needs that arise at any given moment.  Conversations such as these are very important to have and to share.




Back to Top