Decision support is only lightly touched upon in the academic oncology papers that are published almost daily suggesting a potential change in paradigms for management in a particular clinical setting.
Maurie Markman, MD
When will the increasingly critical need for meaningful decision-support in oncology be recognized? Unfortunately, decision support is only lightly touched upon in the academic oncology papers that are published almost daily suggesting a potential change in paradigms for management in a particular clinical setting.
The data from real-world patient populations included in these manuscripts often challenge conclusions reached in “gold-standard” randomized clinical trials, whose shortcomings are often that patients are selected based on homogenous clinical characteristics, and the elderly or those with common and highly clinically relevant comorbidities have been either intentionally or unintentionally excluded.
An excellent example of the complexity facing oncologists in their routine decision making is provided by considering the potential clinical utility of aspirin, one of the oldest pharmaceutical agents employed in medicine. For more than a decade, investigators have inquired whether the use of this drug may prevent cancer or meaningfully slow the progression of an established malignancy.
A recent report examining existing preclinical and clinical data associated with aspirin and endometrial cancer risk reduction highlighted the solid biological and clinical rationale for use of aspirin but also substantial existing differences in outcomes and conclusions from various groups.1 The investigators noted the potential relevance of drug dose and schedule as well as the impact of obesity in relation to a favorable or unfavorable influence of aspirin. Imagine the number of factors that clinicians (cancer specialists or primary care physicians) would need to consider when electing to recommend the administration of this simple OTC medication in a woman with known endometrial cancer or where risk reduction was the goal of therapy.
Further evidence of the complexity of clinical decision making associated with aspirin administration is found in a detailed case discussion by 2 physicians (one focused on cancer and the other on cardiovascular risk reduction) of the benefits versus harm in a hypothetical patient with a known increased risk of gastrointestinal bleeding.2 In the case, the heightened risk of bleeding was not determined to override the potential benefit in this hypothetical patient with established adenomatous polyps, but the risk of bleeding was determined to not be justified if the goal was potential protection of the heart. Again, how are conscientious but busy clinicians going to be able to know the literature for all cancer and noncancer medications (including recent publications and/or modifications to existing guidelines) and analyze this massive quantity of data to optimally manage each individual patient in their practice?
It also needs to be acknowledged that the real world of oncology includes a high percentage of older individuals and those with comorbidities where polypharmacy (the simultaneous use of multiple pharmaceutical agents) is common. The potential for interactions associated with a less-than-favorable outcome is a genuine concern. For example, in a recent report in the ovarian cancer literature, investigators found that individuals taking multiple medications experienced a statistically significant greater incidence of both hematologic and nonhematologic grade III/IV toxicity.3 How will physicians optimally manage patients taking multiple medications, delivered for the management of their cancer or for other conditions, to prevent or at least minimize the risk of an adverse treatment-associated outcome?
During the past several years investigators have explored large well-constructed databases with the goal to devise specific algorithms to optimize patient management in an increasingly costconscious environment. For example, employing data from almost 1 million women who underwent cervical cancer screening within the Northern California Kaiser Permanente health system from 2003 to 2014, investigators suggested the potential for modifications in screening intervals based on an individual woman’s prior test results.4 It would be difficult to envision implementing such an individualized approach to cancer screening or multiple other patient-specific disease and symptom management issues in the absence of a simple, yet robust, clinical decision-support strategy.
Finally, the efficacy and toxicity of an increasing number of pharmaceutical agents is partly defined by individual germline variants. Although the field of pharmacogenomics is in its infancy, it is virtually certain that, over the next decade, knowledge of certain relevant genetic variants in individual patients will become an essential component of optimal clinical care, including within the sphere of oncology.5
And, of course, the giant in the room in any discussion of decision support in oncology is the paradigm-changing clinical investigative and clinical care process known as precision cancer medicine. Research results published in high-impact peer-reviewed journals continue to document the increasing impact of this approach to cancer management. Such papers may define a novel patient subgroup that now requires a different approach to disease management compared with another population whose cancers originated from the same organ site and whose members previously were treated in a similar manner.
While precision cancer medicine is not itself a novel concept and, in fact, has been a component of cancer management for decades (eg, hormone receptor—positive or –negative breast cancer), treatment of multiple malignances is now defined—at least in part—by the presence or absence of specific molecular characteristics (eg, EGFR mutation—positive non– small cell lung cancer; BRAF mutation—positive metastatic melanoma). It is virtually certain that optimal cancer management in the future will require knowledge of an ever-increasing number of patient subtypes, likely based as much on unique molecular profiles as on the site of origin of the cancer.
For the reasons highlighted in this commentary, it is critical that the clinical cancer research community devote time, effort, and funding support to help create the tools essential to help clinicians interpret and incorporate into practice the overwhelming amount of real-world data that will make delivering optimal cancer management an increasingly complex task.