NEW YORK – Data continues to collect supporting the potential of circulating tumor (ctDNA) testing to differentiate patients cured after early-stage cancer surgery from those likely to recur, with new lung cancer research published this month.
The study, led by investigators at China's Guangdong Lung Cancer Institute and published in Cancer Discovery, also took a first stab at establishing a clinically useful cutoff point for serial ctDNA testing, concluding that the likelihood of disease relapse becomes miniscule for high-risk stage II/III patients after 18 months without a positive result.
Oncologists have suggested that a cutoff point like this — for lung cancer but potentially also in other tumor types — could help clinicians feel more comfortable using ctDNA results to inform de-escalation of adjuvant therapy in those who don't need it. It could also calm fears of outsized healthcare costs associated with limitless post-surgical surveillance.
Writing in a commentary piece, Alexander Frankell and Mariam Jamal-Hanjani of University College London said that the field is only beginning to grapple with how and when to act on the basis of post-surgical ctDNA findings.
Ideally, a single test immediately after curative-intent surgery — what the field has come to call a "landmark" time point — could on its own predict recurrence. But Frankell and Jamal-Hanjani wrote that previous studies in lung cancer have shown that some patients can test negative for months or even years before they eventually suffer a recurrence. The same has also been observed in other tumor types.
In their study the Guangdong group tracked ctDNA from 261 patients with stages I–III non-small cell lung cancer, using a 338-gene panel design that researchers described as similar to CAPP-seq, and which maximized the coverage of mutations most likely to occur in these tumors.
The group's minimal residual disease (MRD) testing method "is completely different from the ctDNA test that we [have] used before in the advanced-stage patients," the study's corresponding author Yi-Long Wu wrote in an email.
As of the end of 2021, the group had analyzed a total of 652 postoperative blood samples — between one and six per patient — with a median follow-up time of 19.7 months.
The authors reported that only six patients — about 3 percent of those with longitudinal undetectable MRD — recurred, resulting in a negative predictive value of nearly 97 percent. In contrast a single landmark test missed a significant number of patients who went on to recur during the study period, reaching only about 36 percent sensitivity.
When the investigators looked at the pattern of these ctDNA-positives, they found that the majority occurred up to 18 months, with the risk of emerging ctDNA and clinical relapse dropping after that point.
Much of the excitement about MRD testing has focused on its positive predictive value, the potential to identify patients destined to recur, and the idea of treating them with adjuvant therapies to forestall this recurrence — possibly to cure them, or to otherwise improve their life expectancy and/or quality.
But researchers are also increasingly discussing the negative predictive value of ctDNA MRD tests, or the potential for negative ctDNA results to help reduce unnecessary overtreatment in patients cured by surgery alone. According to Wu and colleagues, this is a higher hurdle for clinicians hesitant to deny patients potentially lifesaving treatment.
Reflecting on the question of whether patients with undetectable MRD might be guided to avoid adjuvant therapy, Wu and his colleagues cited the phase III randomized trial IMvigor010 which compared Genetech's Tecentriq (atezolizumab) to observation after surgical resection for urothelial cancer. In that trial, only MRD-positive patients benefitted from adjuvant atezolizumab.
In the group's lung cancer study 42 percent of the cohort received peri-operative therapy and an exploratory analysis appeared to reflect a similar potential predictive value for MRD. However, the authors wrote "because of the limited sample size and slightly unbalanced baseline features, it is difficult to draw a reliable conclusion."
Frankell and Jamal-Hanjani similarly cautioned that the results are from a limited number of patients with incomplete follow-up monitoring, so future prospective studies will be needed to confirm whether the 18-month timepoint, or any reliable, reproducible cutoff, will hold true.
"Additional statistical power from even larger cohorts, with an increased number of relapse events, is likely to be needed both to provide further insights and greater certainty. A longer period of follow-up monitoring, such as 5 years in the adjuvant setting for patients with NSCLC, would also allow the majority of DFS events to have occurred," they wrote.
In his email, Wu agreed but said the team believes the results are spurring follow-up efforts that have potential to inform clinical practice in the future, including several prospective randomized MRD trials now underway within the Chinese Thoracic Oncology Group.
He and his team are also continuing to follow the recurrence-free patients in their current study, monitoring them every 3-6 months. "We hope after 3 to 5 years follow up, we can give a solid answer for this question," Wu wrote.
The Guangdong group is focused on lung cancer, but the concept of a likely-cured cutoff point could also apply to other cancer types where MRD is being actively investigated. Most likely, it wouldn't be the same timepoint across different tumors, Wu said, as each cancer has its own biological features and survival statistics.
"We [do] not expect the same thing in other cancer types, but we do hope more attention should be given for the negative predictive value of MRD," he said. "In other cancer types some studies have showed similar results but all were retrospective studies."