New Discoveries and Honors in Cancer Research

Read the latest cancer research and recognition from the members of the Damon Runyon scientific circle.
July 2, 2021
Avoiding selection bias in cancer clinical trials

Selection bias occurs when those chosen to participate in a study are not representative of the target population, limiting how much we can trust the study results. When that study is a clinical trial for a new cancer therapy, selection bias among the participants means that a drug that achieves positive outcomes for the sample may not work as well in the general population. It is concerning, then, that patients who need therapy urgently are sometimes underrepresented in clinical trials because they cannot delay treatment for the time it takes to undergo screening, consent, baseline assessment, and other enrollment requirements. Such underrepresentation has been particularly problematic in recent trials for treatment of diffuse large B-cell lymphoma (DLBCL), an aggressive blood cancer.

In order to quantify this selection bias, researchers have come up with a metric known as the diagnosis-to-treatment interval (DTI), which measures treatment urgency among trial participants. DTI, however, is not an ideal metric for selecting trial participants, as non-biological factors like access to medical care also influence the amount of time between diagnosis and treatment. Finding a biological basis for DTI would offer a more objective measure of clinical urgency, and thus be more useful in mitigating selection bias.

In order to quantify this selection bias, researchers have come up with a metric known as the diagnosis-to-treatment interval (DTI), which measures treatment urgency among trial participants. DTI, however, is not an ideal metric for selecting trial participants, as non-biological factors like access to medical care also influence the amount of time between diagnosis and treatment. Finding a biological basis for DTI would offer a more objective measure of clinical urgency, and thus be more useful in mitigating selection bias.

At Stanford University School of Medicine, Damon Runyon alumni Ash Alizadeh, MD, PhD, and David Kurtz, MD, PhD, have taken up this challenge, testing whether circulating tumor DNA (ctDNA) reflects urgency as well as DTI, such that it could be used as an alternative metric. Since DTI is meant to measure disease burden, it stands to reason that a patient’s ctDNA level prior to treatment would give similar information. Indeed, Dr. Alizadeh and Dr. Kurtz found that in DLBCL patients, ctDNA has as strong an association with disease stage, tumor size, and prognosis as does DTI. Furthermore, they found that high pretreatment ctDNA levels predicted short DTI.

These findings mean that researchers designing clinical trials can use ctDNA levels (rather than DTI) as an objective measure of disease burden in prospective participants, to avoid selection bias against patients with advanced-stage disease. From more inclusive studies come more generalizable results—and efforts to include DLBCL patients in urgent need of therapy will result in better treatments.

This study was published in the Journal of Clinical Oncology.