NEW YORK – Experts in the European Society for Medical Oncology (ESMO)'s Precision Medicine Working Group have created a tiered framework for classifying biomarkers to guide cancer treatment de-intensification.
The growing recognition that all cancer patients may not derive added benefit from an intense treatment regimen and might instead experience added toxicities, high costs, and reduced quality of life has increased interest in identifying patients who can do just as well with scaled back treatment.
A range of biomarkers including molecular diagnostics are currently in development or on the market for this very purpose — to select patients who may receive the same benefit with a less intense treatment regimen — but some have more supporting data than others.
"We wanted to provide a multistakeholder framework to provide a sense of the level of evidence supporting the decisions that we make in cancer treatment," said Dario Trapani, a medical oncologist at the Dana-Farber Cancer Institute and a co-lead author of the framework published Monday in the Annals of Oncology. "It's a standardized scale for clinicians to communicate with patients … and for policymakers and decision makers to understand the level of evidence" behind a de-intensified treatment strategy.
The framework can be used to guide precision oncology treatment decisions similar to how the ESMO's Scale for Clinical Actionability of Molecular Targets (ESCAT) and Memorial Sloan Kettering's Oncology Knowledgebase (OncoKB) are used.
The ESMO working group, which included clinical, translational, methodological, and public health experts as well as patient representatives, met periodically online to zero in on three categories, or tiers, of biomarker strategies based on available evidence. They created a system that they hope will give oncologists, researchers, and patients, as well as decision makers like regulators and payors, a common language for classifying these de-intensification biomarkers and guiding their use.
The 'gold standard'
The tiered framework considers first and foremost the type of clinical trial used to validate the treatment de-escalation biomarker. Tier A involves biomarkers that have been clinically validated in prospective, randomized, non-inferiority clinical trials. According to the working group experts, these types of trials should be considered the "gold standard" when it comes to validating biomarkers for use in de-escalating therapy.
One example of a Tier A de-intensification biomarker would be the recurrence score determined using Exact Sciences' Oncotype Dx gene expression assay, which has been validated as a method for deciding which early-stage breast cancer patients can do well without adjuvant chemotherapy. The Phase III TAILORx trial established that patients with an Oncotype Dx recurrence score of 0 to 25 had similar five-year disease-free survival regardless of whether they received just hormone therapy or hormone therapy plus chemotherapy. Further exploratory analysis in this trial has suggested that patients age 50 or younger with recurrence scores of 16 and 25 may benefit from the addition of chemotherapy.
Although data like this from large, randomized clinical trials should ideally be the basis of medical decisions, such investigations are resource-intensive, lengthy, and not widely supported. Accordingly, researchers conducting these de-escalation trials often must rely on academic and public funding sources. While this is a challenge, Trapani emphasized the importance of reaching out to drugmakers, suggesting that pharmaceutical firms could still be interested in supporting these trials when a de-intensification strategy involves their therapies.
Single-arm trials with historical controls
One tier below treatment de-intensification biomarkers with "gold standard" clinical evidence in the ESMO framework are biomarkers tested in a single-arm trial, in which a given group is compared against a historical control arm. One biomarker in this tier is minimal residual disease (MRD) to guide chemotherapy reduction among children with acute lymphoblastic leukemia.
In Study ALL10, conducted by the Dutch Childhood Oncology Group, children were stratified according to their MRD levels following initial chemotherapy treatment. Among patients with no detectable MRD, reduced chemotherapy did not jeopardize survival outcomes versus historical controls, whereas patients with intermediate and high MRD levels benefited from the intensified chemotherapy. Given the trial design, the ESMO framework groups MRD for pediatric ALL treatment de-escalation into its Tier B bucket.
Finally, Tier C biomarkers, according to the framework, are backed by retrospective analyses of prospective trials. These analyses are common but carry substantial caveats and may be biased.
That said, as Trapani and colleagues noted, some Tier C biomarkers have received regulatory approval despite the retrospective nature of their validation. For example, in the US, Myriad Genetics' EndoPredict is approved for guiding adjuvant chemo decisions for certain breast cancer patients and included in clinical guidelines, even though the assay falls into ESMO's Tier C category. The multigene assay was validated using retrospective analyses of two Phase III clinical trials.
In comparison, Trapani noted that other competing assays, Oncotype Dx and Agendia's MammaPrint, are in Tier A and Tier B, respectively, based on the level of underlying evidence. Indeed, the Tier A Oncotype Dx test is listed as the "preferred" test in National Comprehensive Cancer Network (NCCN) guidelines.
Within Tier C, the authors went an additional step and created two sub-tiers. Sub-tier C1 includes evidence from trials that were randomized from the onset, and sub-tier C2 includes evidence of trials that were not randomized. Theoretically, Tier C could also include exploratory non-inferiority "re-analyses" of trials that failed to show superiority to begin with.
Trapani and colleagues have defined de-intensification biomarkers as those that can lead to a lower dose or shorter duration treatments, or those that can be used to reduce the number of treatments a patient is getting in a regimen. For a treatment to be considered a de-intensification, it must also have "tangible evidence of improved patient outcomes," including improvements in quality of life, toxicities, or costs.
Cost, health system considerations
Whichever method researchers, drugmakers, and commercial labs use to validate these de-escalation biomarkers, the ESMO working group stressed the importance of considering toxicity and patients' quality of life in studies as well as survival endpoints.
Toxicities, of course, can be financial and physical. With the high cost of intense treatment — especially in cases where there is debate as to whether certain biomarker-defined patients may truly benefit — the cost-effectiveness of treatment intensity versus de-escalation should play into these studies, the working group argues.
Health systems have used quality-adjusted life years (QALYs) as a metric to decide whether to pay for treatments, but "presently, there is no obvious consensus about a threshold of cost-effectiveness," wrote the authors, adding that although the field is unlikely to arrive at one universal metric for economically evaluating biomarker-guided therapy de-escalation alongside clinical evidence, it's possible that specific countries may set their own thresholds based on their gross domestic product or health system budgets.
For these crucial quality-of-life and cost considerations to play a role in guiding treatment de-intensification biomarkers going forward, Trapani noted, the field needs new and better metrics.