The Essential But Often Complex Role of Numbers and Measurements in Oncology

It is fair to state without the risk of disagreement that today it would be essentially impossible to imagine a discussion of cancer management excluding clinically relevant critical numbers and measurements. From patient age and weight, to laboratory values, tumor stage, duration of treatment, and survival, we regularly record objective clinical features, express the details of planned therapy, and define ultimate outcomes based on a variety of numerical values and relevant measures of results.

It is recognized that some measurements fluctuate regularly, such as blood pressure, and others change at a slower rate (e.g., white blood cell count following cytotoxic chemotherapy) or may increase quickly and then not at all (e.g., height). Certain measurements are recognized as being independently associated with health and disease (e.g., morbid obesity, substantially elevated serum creatinine, lactate dehydrogenase, and fasting glucose), whereas others provide reassurance (e.g., normalization of serum CA-125 or beta-HCH following cytotoxic chemotherapy in advanced ovarian cancer and metastatic germ cell tumors, respectively).

What has been described above is neither complex nor controversial. However, for some measurements, the claimed objectivity for an intended use may struggle to satisfy the requirement of being both simple to understand and noncontroversial.

In the opinion of this commentator, it would be difficult to find a more consequential example of the contrast in the objective interpretation of measurements in oncology than the two widely proclaimed primary goals of antineoplastic drug therapeutic trials: to improve overall survival (OS) and/or improve quality of life (QOL), “because they directly measure longer and better lives, respectively, which matter the most to both patients and physicians.”1 And like “mom and apple pie” it is difficult to argue against improved survival and QOL as being core components of what clinical oncologists hope to achieve on behalf of their patients.

But we now arrive at the concern and the striking contrast between the two stated objectives. As regards “OS” and ignoring for the sake of the current discussion the effect of clinical factors which may interfere with its interpretation, it is difficult to imagine a more objective measure in an interventional cancer trial. Whether one considers the study start date to be the day the consent is signed or the day initial therapy is delivered, the end point is the date of death. In fact, if necessary, this date can be confirmed by viewing the official death certificate.

Further, simple statistical tools are available in an analysis of survival to deal with the fact that some (and hopefully many!) research subjects are alive (“censoring”) at the time investigation is conducted into a trial’s outcome. Whether observed to be “months” or increasingly found to be “years,” the measurement of OS of a study population in an interventional trial should unquestionably be objective, accurate, and noncontroversial.

Now contrast this measurement to the second stated standard goal for declaring a positive trial outcome: improving QOL among research subjects in a randomized phase 3 study. Space available in this commentary does not permit an adequate critique of this clinical parameter, but I will briefly summarize a number of issues to make the point regarding the striking contrast in objectivity.

First, what exactly is “QOL”? If a definition cannot be agreed upon by relevant stakeholders (including clinicians, academic experts, and most importantly patients as well as their families) how can one hope to develop a specific strategy to measure improvement?

Can QOL realistically be represented by research subject responses to a simple series of survey questions focused specifically on one or a few clinically meaningful symptoms (e.g., pain or fatigue), or should an inquiry into QOL be more general (e.g., asking patients to rate their general feeling of well-being on a scale of 1 to 10)? Should the evaluation be designed to focus on symptoms commonly observed in a particular tumor type (e.g., shortness-of-breath in lung cancer), toxicities associated with a class of drugs being delivered in the trial (e.g., symptoms of peripheral neuropathy with platinum agents), or the specific symptoms of disease and/or adverse effects (AEs) experienced by a patient?

What if some signs/symptoms improve, but others deteriorate? How does one score this scenario in a QOL evaluation? At what point in the cancer journey, during and following therapy should QOL be evaluated for the purpose of labeling this end point as revealing improvement or failing to achieve this goal? What is the justification for this decision? Recognizing that many effective antineoplastic agents have AEs, how do these tools differentiate between treatment toxicity and cancer-related symptoms? Does this distinction even matter to a patient?

And finally, what is the definition of “improved” versus “stable” or “not improved?” Why was it selected, and what process was employed to confirm agreement among the above noted stakeholders that the measurement tool accurately captures the clinical meaning of QOL?

The point of the preceding discussion is to acknowledge that while essential in evidence-based cancer medicine, the development of objectively valid and clinically meaningful outcome measures may be a complex process, requiring considerable compromise and a thoughtful understanding of the flaws and just how arbitrary certain measurements may be that are routinely employed in clinical, regulatory, financial and public health decision-making.

An additional point to acknowledge, which will be expanded upon in a subsequent commentary, is the critical need for careful interpretation of a variety of clinical guidelines that employ discrete numerical criteria. Simply stated, insuring optimal value in patient care mandates thoughtful analysis rather than rigid adherence.

Consider, for example, the long standing widely applied and accepted definition of platinum-resistant versus platinum-sensitive recurrent ovarian cancer.2 Patients with platinum-resistant disease are commonly considered to be those whose cancers failed to respond to a regimen containing this class of drugs or where the tumor recurs within 6 months of the completion of therapy. Platinum sensitive-disease is disease that has recurred after 6 months following the completion of such treatment. The purpose of this stratification or restriction of eligibility in a clinical trial is well understood to be an attempt to create a reasonably clinically defined homogenous population as regards underlying residual platinum sensitivity.

However, it is critical to appreciate that while appropriate from the perspective of optimizing the ability to interpret trial results, this stratification is quite arbitrary. This is due to the fact the presence or absence of a meaningful degree of responsiveness to retreatment with a platinum agent is a continuum with the statistical likelihood of persistent clinically relevant sensitivity increasing the longer the tumor has not been exposed to a platinum agent.

More to come on the topic of the complexity and limitations of measurements and numerical parameters in oncology in the next commentary.

References

1. Mittal A, Kim MS, Dunn S, et al. Frequently asked questions on surrogate endpoints in oncology-opportunities, pitfalls, and the way forward. EClinicalMedicine. 2024;76: 102824.doi:10.1016/j.eclinm.2024.102824
2. Markman M. Pharmaceutical management of ovarian cancer: current status. Drugs. 2019;79(11):1231-1239. doi:10.1007/s40265-019-01158-1