Dana-Farber Embraces Pathways in Clinical Practice

OncologyLive, Vol. 19/No. 21, Volume 19, Issue 21

Partner | Cancer Centers | <b>Dana Farber</b>

David M. Jackman, MD, the medical director of Clinical Pathways at Dana-Farber, discusses the benefits of a clinical pathways system.

David M. Jackman, MD

David M. Jackman, MD

Senior Physician

Medical Director of Clinical Pathways

Dana-Farber Cancer Institute

Assistant Professor of Medicine

Harvard Medical School

Boston, Massachusetts

Although a myriad of changes have occurred in oncology during the last decade or so, I’ll explain the development of our Clinical Pathways system by focusing on 3 areas: complexity, cost, and setting.

First, with continued advances in cancer biology and drug discovery, we have made significant strides in anticancer therapy. Many of these advances, however, apply to just a narrowly defined population. A drug such as osimertinib (Tagrisso), for example, is not a broadly effective lung cancer therapy. It is specifically indicated in patients with metastatic non—small cell lung cancers that harbor a specific mutation in a particular gene: an EGFR exon 19 deletion or an exon 21 L858R mutation. If the same patient had a slightly different mutation—perhaps an L861Q mutation—in that same gene, we might instead recommend afatinib (Gilotrif), a different EGFR inhibitor. Scale this level of complexity across dozens of different cancers and hundreds of different stages and settings and you begin to see an unfathomable level of granularity needed to care for patients appropriately.

Second, consider the quickly rising cost of treatment—to the system, patients, and their families. At the beginning of this decade, an estimate from Mariotto et al posited that cancer care might approach $200 billion annually in the United States by 2020.2

Costs are rising exponentially, yet our ability to access and understand these costs at the point of care is becoming increasingly difficult. If we are going to task providers and practices with responsible resource management, we need to give them better tools with which to do so.

Developing Customized Content

Third, due to any number of factors, cancer care is increasingly being delivered through larger centers, networks, and consortiums. All stakeholders, particularly patients, need to trust that they can get the same care from any provider at any location within an organization and that the care they receive will adhere to evidence-based standards.When we consider these trends, it becomes apparent how Clinical Pathways might be used to address each of these. The executive leadership at Dana-Farber recognized these trends several years ago and committed to developing and using Clinical Pathways to address these and other concerns.

Early on, we decided to develop fully customized content for our medical oncology and radiation oncology pathways. Every recommendation on our pathways system is the result of DanaFarber disease experts making specific choices about patients in a particular setting.

Not surprisingly, all of this takes an incredible amount of thought, time, and effort. We bring each group together several times a year to discuss and debate the merits of newly approved drugs and recently presented studies, along with the lessons learned in the course of our own clinical practices and lab efforts. Each session includes doctors, nurses, and pharmacists with expertise in that specific disease (Figure).1

Integrating Pathways into Practice

Furthermore, our pathways team helps bring uniformity and structure to the process, introduces costs to the discussion, and provides a focused data analysis that not only includes overall on-pathway rates but also addresses how often physicians went off pathway in each particular branch, what alternative regimens they used, and why they did so.Where do these pathways fit in modern cancer care? I think they are attempting to achieve multiple goals. Dana-Farber’s pathways program fundamentally aims to define the best care we can provide. Patient outcomes are paramount as we consider our treatment recommendations, with added focus on the potential toxicities of therapy. Cost considerations also need to play a role. If various therapies are essentially equivalent in efficacy and toxicity, then we absolutely should seek to select the least expensive of these options.

Once we have defined optimal care, our pathways program implements this guidance across a web-based, decision-making platform. This helps ensure that patients get the same level of expert care at every exam room, infusion chair, and radiation table throughout our institution and across our network.

Next, the platform captures the decision that was made, including all the navigated driving factors. This level of information is exactly the type that institutions might use to work with payers to try to reduce the time and effort spent on the prior authorization process. It is also the kind of information that can be used in the aggregate to better understand the quality of care that we are delivering, sometimes pointing out areas where our pathways may need refining.

Figure. Key Elements of the Clinical Pathways System1

Comparing Recommendations With Guidelines

Are these the only ways in which pathways can and will have an impact on cancer care? Certainly not. I think that these are some of the main ways in which Clinical Pathways most immediately affect care delivery. However, there are so many other opportunities for pathways to influence care along the full spectrum of the cancer journey.I see cancer guidelines as incredibly important. As someone who has sat on national guidelines committees, is a physician at Dana-Farber, and is part of a member institution of the National Comprehensive Cancer Network, I think that guidelines play a key role in cancer care. Guidelines attempt to address the question, “What can we do?” That is, what is the full list of potential options that are approved and reasonable in a given cancer?

Our pathways, in my opinion, try to take things 1 step further. Clinical Pathways, at least as we have envisioned them at Dana-Farber, attempt to answer a more granular and specific question: not just what we can do but also what we should do. What is the best recommendation that we can make for a patient in a specific setting?

We start by considering the specific type of cancer, histologic subtype, stage, and line of therapy. We drill down to a patient’s specific biomarker—genomic testing, PD-L1 expression testing— and add in that patient’s overall performance status, past therapies received, and any medical comorbidities or personal concerns (about hair loss or neuropathy, for instance).

Considering Genomics and Artificial Intelligence

We also consider the burdens of adverse effects and costs related to various therapies. Based on all that information, our disease experts weigh the most pertinent data—clinical trials, real-world outcomes, own depth of expertise—to recommend the best treatment we can offer in a given scenario.Genomic testing, interpretation, and actionability have become critical and necessary parts of modern cancer care. Although not every patient will have a targetable genomic change, it is critical that every patient who should be tested for genomic mutations undergoes such testing, so as not to miss potential treatment opportunities. Pathways can play an important part in:

  • Identifying which patients should undergo testing and which tests they should receive
  • Helping to separate the genomic wheat from the chaff: Which mutations are driving the cancer, and which are bystander mutations of unclear impact? Of the driver mutations, which are targetable with currently available therapy?
  • Recommending which specific targeted therapy is most appropriate for a given patient, based not only on the genomic profile but also on disease setting, comorbidities, and other key factors

The interface between Clinical Pathways and the larger computer platform on which they live is a much more complex and nuanced question. We are making every effort to develop a platform that can extract key data elements from a user’s electronic medical record, reducing redundant data entry and bringing all those factors needed for oncologic decision making into 1 place.

However, I want to be clear that Dana-Farber’s Clinical Pathways system does not involve artificial intelligence in the sense that a computer is making unverified recommendations based on an algorithm of its own perception of what a human would do. Rather, every decision included in our platform is the product of human beings discussing and debating the most pertinent data for that particular situation.

In the process of implementing the Clinical Pathways platform, we will collect a host of information about how patients were treated and why. Do we intend to learn from that information? Absolutely. Collection and analysis of these data can help us better understand care delivery and find patterns that, without our platform in place, might have gone unnoticed. If we can pair all this with outcomes data, might this real-world evidence help shape future iterations of pathways? It certainly might, but as part of a larger process of human analysis, discussion, and debate of such data.

We are bringing together Dana-Farber’s expertise in cancer research and therapy with Royal Philips’ depth and breadth of knowledge and experience in healthcare informatics. The immediate impact of this collaboration will be a world-class platform, powered by Dana-Farber’s expert, customized content with a commitment to integrating physician workflow to maximize efficiency. The broader impact emanates from Royal Philips’ larger IntelliSpace platform.


  1. Hall JC, Hamilton JM, Jackman DM, et al. Analyzing pathways data to reduce unwarranted variation, understand practice patterns, and update pathways. J Clin Oncol. 2018;36(suppl 30; abstr 300). meetinglibrary.asco.org/record/166529/abstract.
  2. Mariotto AB, Yabroff KR, Shao Y, Feuer EJ, Brown ML. Projections of the cost of cancer care in the United States: 2010-2020. J Natl Cancer Inst. 2011;103(2):117-128. doi: 10.1093/jnci/djq495.