Confronting the Giant in the Room: Cancer Management Needs Real Decision-Support Tools

The acceleration in our understanding of the molecular basis of cancer and mechanisms of drug resistance has energized fundamental and applied research efforts by both academic and industry scientists, leading to the development, clinical testing, and regulatory approval of a bewildering number of novel therapeutic agents and additional indications for previously approved drugs. Furthermore, the FDA has authorized a number of therapies for delivery based on the demonstrated presence of specific molecular targets rather than the site of origin of a particular malignancy, and it is highly likely that this regulatory trend will continue, if not increase in speed.

Given the heavy workload of oncologists—who must balance their clinical duties with an ever-growing list of administrative tasks for patients, payers, regulators, and employers—it would be reasonable to expect a major effort to develop robust decision-support tools. Ideally, these tools would be widely available, low-cost, or free for providers, relatively simple to use, easily integrated into diverse electronic medical platforms, routinely accepted by payers, and rapidly updated to reflect new regulatory approvals or peer-reviewed evidence.

However, to summarize the actual status of such endeavors, baseball legend Yogi Berra perhaps said it best: “When you come to a fork in the road, take it.” This translates to: “It is someone else’s problem/responsibility to help oncologists know the optimal road to take, which option is best for their patient at a particular time in their cancer journey, and which is the best among the increasing number of therapeutic approaches, all with consideration of relevant comorbidities and individual patient wishes.”

Berra also noted, “The future ain’t what it used to be.” No, it is not. In fact, in the not-so-distant past, antineoplastic drug development proceeded at a relatively slow, deliberate pace, allowing conceptually relatively simple randomized trials in a given setting to be designed, conducted, completed, and reported (at least at national/ international meetings) over an extended period. Of course, although there were exceptions to that stated above, this generalized drug development paradigm permitted the thoughtful design of a subsequent series of studies, including, critically, the assignment of an appropriate, widely accepted standard-of-care (SOC) control arm in phase 3 randomized studies.

Plus, among the cytotoxic drugs in routine clinical or investigative use, there were (with notable exceptions) objectively limited differences in the toxicity profiles of the agents included in therapeutic studies. As a result, regardless of the presence of comorbidities, the options for management focused far more on dose reductions or schedule modifications rather than on the possible selection of entirely different classes of drugs with vastly different risks for adverse effects.

Today, multiple randomized trials may be conducted for a given indication. This has led to an increase in the problematic concern that the study’s control arm is simply outdated by the time accrual is completed, and results may be reported due to the documented inferiority of this regimen compared with what has in the interim become a new SOC. Although a strong argument can be presented that it would certainly be very unfair for a pharmaceutical company that has spent millions of dollars conducting a multiyear, multicenter, multination, randomized study to be denied regulatory approval because “the goalposts have been moved,” it now becomes the treating clinician’s burden to interpret optimal care, potentially without the benefit of valid comparisons.

One must add to this discussion the dilemma of antineoplastic drug approvals that are increasingly (and in the opinion of this commentator, appropriately) based on nonrandomized phase 2 trial data for which there has been no direct comparator, making it difficult to assign a place for the new agent/strategy within a group of possible patient options. The complexity is only heightened if the potential choices are associated with meaningfully different toxicity profiles.

Finally, the era of precision medicine has permitted options to be defined on a more individual patient basis rather than simply the general characteristics, such as tumor type and stage, and number of prior treatment regimens. Consider, for example, a patient with ovarian cancer whose disease is platinum-resistant. Biomarker testing has revealed the overexpression of HER2 and folate receptor α and an elevated tumor mutational burden. Based on current FDA approvals, this patient would be eligible to receive 1 of 2 different antibody-drug conjugates directed at the different molecular targets, as well as treatment with a checkpoint inhibitor. What should the oncologist recommend to this patient?

Or consider just one more example of the many that might be provided of the dizzying array of decision points in oncologic management. It was not that long ago when patients with chronic myeloid leukemia were treated with the chemotherapy agents busulfan or hydroxyurea, interferon-α, and bone marrow transplantation. It was fairly simple for a trainee to learn and for a treating oncologist/hematologist to remember. But today, we have 3 generations of tyrosine kinase inhibitors, with 6 different drugs in this class, 5 approved by the FDA for first-line treatment and 5 approved after initial disease progression.1 For the patient being seen in your office today, what is the optimal choice for first-line treatment, after progression, or in case of drug toxicity? And when should allogeneic bone marrow transplantation be considered in current management?

It is important to note that the preceding discussion focused entirely on the recognized toxicity profiles of the available antineoplastic drugs. But how should specific patient-related factors influence physician recommendations for curative or palliative approaches to treatment? The list here is long and includes current and past medical conditions (hypertension, diabetes, obesity, renal function, history of myocardial infarction or transient ischemic attack, etc), as well as the potentially multiple medications prescribed by other physicians involved in a patient’s care that may interact with the proposed therapeutic strategy. In conclusion, it is important to note that there exist today both proprietary and publicly available pathways and guidelines that surely provide some level of meaningful support to treating oncologists. Although the purpose of these individual efforts may vary, and efforts to maintain their relevance should be lauded, one must ask if we cannot do much better in the development of robust clinical decision-support tools. Is it possible that this endeavor can somehow become a priority for the oncology community? One can only hope.

Reference

1. Jabbour E, Kantarjian H. Chronic myeloid leukemia: a review. JAMA. 2025;333(18):1618-1629. doi:10.1001/jama.2025.0220