Maurie Markman, MD
"Uncertainty" is a difficult concept to address in clinical medicine in general and specifically in oncology. Consider the surgeon who meets the family of a patient after finishing the resection of what appears to be a large but localized high-grade non–small cell primary lung cancer. The surgeon might say, “I was able to remove the cancer.” Alternatively, the surgeon might say, “I removed what I could see and, while it is reasonable to be hopeful the cancer has not spread, there is unfortunately a high statistical likelihood that in a relatively short period of time metastatic disease will be revealed in one or multiple locations. I am sorry, but we simply do not know with the desirable degree of certainty what exactly will happen or when it will occur, but the odds are not favorable.”
Or consider the medical oncologist’s dilemma when reporting the results of a CT scan to a patient with an advanced abdominal malignancy who has exhibited a most impressive response to cytotoxic chemotherapy. The oncologist might say that “there is no evidence of disease” or “the cancer is in remission.” Alternatively, the oncologist could report that “the scan reveals no definite abnormalities, but the sensitivity of this test does not permit an examination for the presence of very small-volume macroscopic or persistent microscopic disease. Realistically, at this time we do not know if there is residual cancer, although statistically that is likely to be the case. Only with careful follow-up will we be able to determine if this cancer will exhibit a long-term remission.”
Of course, there is no reason for the alternative words in these two scenarios to be spoken at this exact point in time in the patient’s cancer journey, but they do express the reality of the situations.
Unfortunately, the myth of “certainty” permeates the entire realm of clinical medicine. Consider, for example, governmental regulatory or third-party payer coverage language where expressions such as “medical necessity” or “safe and effective” are used as the basis for defining clinical utility. Such terminology implies absolute determination that the drug, procedure, or approach “is medically necessary” or “is safe and effective.” But it would be far more realistic to acknowledge that the benefits versus the risks of oncologic therapeutics are overwhelmingly relative to a specific clinical setting and recommendations should be based on both objective data and the subjective judgment of a thoughtful clinician.
Further, as noted in a most provocative and highly relevant commentary discussing what has been labeled “gray-zone medicine,” the authors state1
“One fundamental problem may be a misguided perspective that health care is a binary world in which interventions are either effective or ineffective, appropriate or inappropriate. In truth, there are large gray zones in which an intervention is neither clearly effective nor clearly ineffective—zones where benefits are unknown or uncertain and value may depend on patients’ preferences and available alternatives.”
Confronting the Data Gap
“Uncertainty” is a routine dilemma when discussing a prognosis with a patient with cancer and his or her family. The prognosis is, at best, a statistical probability—assuming the available objective data are somewhat representative of the individual patient. Unfortunately, this assumption can often be quite problematic.
Consider, for example, a newly diagnosed patient with stage IIIC high-grade ovarian cancer who inquires about the probability of whether she will survive at least 5 years. An examination of published survival statistics suggests an appropriate response would be less than 50%. But what happens to that statistically anticipated survival figure, as uncertain as these population-based figures are in defining individual outcomes, if that particular woman has already survived for several years without clinical evidence of progression? A recently reported analysis has suggested, not surprisingly, that the likelihood for her continuing to remain disease free substantially improves compared with the baseline assessment.2
Yet it is uncertain if such data are routinely communicated to patients who may incorrectly continue to believe that the initial prognostic assessment remains accurate in their particular situation.
Uncertainty is also uncomfortable when addressing a patient with cancer regarding recommendations that must be made with data that are far less than perfect. One of the major attractions of the randomized trial may be that an outcome with P
<.05 permits some clinicians to declare with an inappropriate or unreasonable degree of medical certainty that “Regimen A” is superior to “Regimen B” for a specific individual.