April 10, 2018
5 min read

Clinical pathways: A new target

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

It is now widely accepted that adherence to evidence-based clinical pathways for cancer treatment has the potential to reduce variability, improve patient outcomes and reduce costs.

As the data supporting this concept continue to grow, pathway development has become an industry in its own right — many companies have appeared with pathway software platforms.

Like many other cancer centers, we have had to decide to develop our own pathways in house or look for a commercial product that meets our “must-have” criteria. Among those essential criteria, we include the ability to customize the pathways to our practice, the flexibility to prioritize our clinical trials, integration into our electronic medical record, and compliance with National Comprehensive Cancer Network guidelines.

Impact of NCCN guidelines

Although the NCCN does not have a monopoly on practice guidelines in the United States and beyond, it certainly produces the most widely used and referenced set. They are of particular significance because the NCCN compendium is one of five used by Medicare, Medicaid and private insurance companies to assist in coverage decisions.

John Sweetenham, MD, FRCP, FACP
John Sweetenham

Because NCCN guidelines frequently include the off-label use of cancer therapies, Wagner and colleagues — in a study published last month in BMJ — assessed the frequency and level of evidence for NCCN recommendations compared with FDA approvals and called into question the validity of some of these guidelines.

In the interest of full disclosure, I should point out that my institution is a member of NCCN and that I am a member of the NCCN Board of Directors. That said, I don’t think this has clouded my objectivity on this issue. I should also point out that the opinions expressed in this editorial are mine, and that I am not writing as a representative of NCCN.

The publication by Wagner and colleagues describes a retrospective study of 47 drugs approved for treatment of adult solid tumors and hematologic malignancies by the FDA between 2011 and 2015. These drugs were approved for a total of 69 recommendations.

The authors then searched the NCCN website and identified the recommended indications for the same drugs as of March 2016. They went on to explore the cited references to support the NCCN recommendations and graded these according to the level of evidence.

Perhaps not surprisingly, the NCCN’s recommendations for the use of these drugs were broader than the FDA’s. NCCN guidelines recommended the same 47 drugs for 113 indications, of which 69 overlapped with the FDA and 44 were additional. Expressed another way, the average number of recommendations for these drugs was 0.92 for the FDA compared with 2.4 for the NCCN.


The authors stated that the additional recommendations made by the NCCN are often based on low-quality or no evidence, with only 10 of the additional 44 recommendations being based on evidence from randomized clinical trials. This compares unfavorably with FDA approvals, which were supported by randomized trial data in 58% of cases. The authors implied that the evidence base for NCCN recommendations is too low and that this has resulted in the overuse of costly and ineffective cancer treatments.

Guideline development

At one level, it is difficult to argue with the authors — the data are accurate and I think everyone would acknowledge that NCCN frequently recommends off-label use.

The implication that the process of guideline development is badly flawed seems to me to be unfair for several reasons.

The authors give the impression that FDA approval is the gold standard for appropriate, high-value use of costly new treatments. Even a superficial look at the 47 FDA-approved drugs cited in this study will reveal that the clinical benefits of some of these are minimal. Although significant HRs were observed in randomized trials, OS improvements were minimal in some of these — often measurable in weeks. The true clinical benefit of some of these FDA-approved drugs is, therefore, questionable, and the value proposition has not been clearly demonstrated. Further, despite failure of some of these drugs to live up to early promise, they have retained FDA approval.

It’s also worth pointing out that only 58% of the FDA approvals were based on randomized studies. That’s not a criticism — the FDA has shown its clear commitment to giving patients access to effective new drugs through accelerated approval — as this has been a positive trend, and one we should embrace. Although most of us would accept randomized data as the highest quality, even the FDA does not have an absolute requirement for phase 3 data.

The authors cautioned against the use of low-quality evidence for recommendations and cited the example of high-dose therapy and autologous stem cell transplantation for early-stage high-risk breast cancer, which was initially widely supported but then shown to be ineffective in randomized trials. No one can argue with that; but, if we advocate for prospective randomized trials as the only benchmark for treatment recommendations, as an example, almost all of our practice of stem cell transplantation for hematologic malignancies would be disapproved — there are virtually no randomized trials in this context.

Further, randomized trials for patients with rare cancers are very challenging and often impossible to conduct for reasons of accrual and expense. Without off-label use of some drugs, patients with less common cancers would be denied potentially effective treatments.


Another implied criticism in the paper is the overall process of guideline development, which relies on expert opinion, as well as phase 2 and phase 3 data, which, therefore, opens the door to potential bias. Again, this is a valid point, but it is also worth remembering that NCCN guidelines are developed and updated by more than 1,300 experts with hundreds of years of cumulative experience in cancer treatment; clearly one has to be circumspect about the use of lower levels of evidence, but experience has to be worth something.

An imperfect process

The approval process for new anticancer drugs and their inclusion in guidelines and treatment pathways is a long way from ideal. Clinical guidelines developed by any organization — including NCCN — are not perfect and improving upon these is an iterative process, constantly updated by new data.

The balance between rapid approval with preliminary/inadequate data compared with slower approval after robust trials are complete is tricky. As advocates for our patients, we also need to understand the balance between giving our patients the earliest possible access to potentially life-saving treatments vs. uncontrolled use of high-cost, low-value, ineffective therapy. A rush to recommend new treatments exposes our patients to potential toxicity — including substantial financial toxicity — in the absence of level-one data. In contrast, waiting for results of phase 3 studies can delay approval of highly effective therapy and can potentially deny new treatments to patients with rare diseases, for which randomized trials never happen.

Evidence-based pathways and guidelines are now widely acknowledged to be central to improving outcomes and values in cancer care. Targeting the flaws in the pathway development process is appropriate and is important to ensure that this remains evidence based and rigorous.

Setting the bar too high for approval will run the risk of denying important therapy to some of our patients.


Wagner J, et al. BMJ. 2018;doi:10.1136/bmj.k668.

For more information:

John Sweetenham, MD, FRCP, FACP, is HemOnc Today’s Chief Medical Editor for Hematology. He also is senior director of clinical affairs and executive medical director of Huntsman Cancer Institute at The University of Utah. He can be reached at john.sweetenham@hci.utah.edu.

Disclosure: Sweetenham reports no relevant financial disclosures.