NEW YORK – New data presented at the annual meeting of the American Association for Cancer Research this week has further cemented the value of Personalis' ultrasensitive minimal residual disease (MRD) platform in colorectal cancer, with other studies continuing to add evidence for superiority of the approach over more targeted test platforms.
The firm's NeXT Personal assay uses whole-genome sequencing of tumor tissue to develop personalized liquid biopsy panels of up to 1,800 targets which are then used to test patients for circulating tumor DNA (ctDNA) either as a sign of lingering disease after treatment or of later disease recurrence.
One study highlighted at the meeting reported interim results from a clinical trial called VICTORI, which aims to find an optimal time point to test for MRD ctDNA after surgical resection of colon cancer to maximize the ability to predict recurrence.
Posttreatment detection of ctDNA in colorectal cancer patients was one of the first to be explored in terms of solid-tumor MRD, and it has become one of the most promising for clinical translation, with several assays now reimbursed by Medicare. But the ability to use MRD to guide treatment decisions in a way that benefits patients hasn't yet been proven.
Interventional trials are already being conducted, but their success will depend on the ability to detect ctDNA at a time point when clinical decisions about adjuvant chemotherapy are actually taking place.
CtDNA levels after surgery can be incredibly low, just by nature of the stage of the disease, said Emma Titmuss, a bioinformatician at BC Cancer-Vancouver and an investigator on the study, during her presentation on Monday. In addition, she said, "interventions like surgery and chemotherapy can limit the detection of ctDNA by increasing the shedding of DNA from normal cells, which dilutes the signal, making it harder to detect."
Personalis believes that the high sensitivity of its approach could be a boon here, by maximizing the number of individuals whose residual disease gets caught in that window, and the new data from VICTORI suggest that this could be true.
The interim prospective analysis included 71 patients with resectable CRC, 52 with stage I to III disease and 19 with stage IV disease. Liquid biopsies were taken prior to surgery, every two weeks for eight weeks post-surgery, and every three months for a potential three years, and analyzed with the NeXT Personal assay.
All 33 patients with treatment-naïve disease greater than stage I had detectable ctDNA prior to surgery. Of 65 patients whose clinical outcome could be evaluated, 23 experienced a clinical recurrence, with a majority, 87 percent, testing positive for ctDNA within a "landmark" eight-week post-surgical period — the timeframe in which adjuvant chemotherapy is typically administered.
All patients who recurred were ctDNA-positive before their recurrence was detected by imaging, and signs of recurrence showed up in their blood a median of 198 days earlier, including in cases with difficult-to-detect metastatic sites such as the lung.
One patient had ctDNA recurrence 416 days prior to clinical recurrence, and Titmuss said that ctDNA was detected in some samples at frequencies as low as 2 parts per million (ppm). Overall, 70 percent of recurrences were detected with ctDNA levels within what Personalis calls the "ultrasensitive range" of less than 100 ppm.
Notably, other research, including another report at the AACR meeting, has shown that this low level of detection can be beyond the capabilities of narrower MRD approaches. In a separate presentation on Sunday, Charles Swanton, chief investigator of TRACERx and codirector of the CRUK Lung Cancer Centre of Excellence, shared data from the ongoing TRACERx lung cancer study. He disclosed that his team has received research grants from Personalis for some of this work.
Swanton and his colleague began their work with a tumor-informed MRD technology from Natera, which can create personalized panels of up to 16 targets, and then tested another one developed by Invitae (now a part of LabCorp), which allowed for panels of up to 200 markers.
Both technologies could detect ctDNA down to about 80 ppm, Swanton said. But, he added, "with these platforms there were some adenocarcinomas that we just couldn't detect. … We wondered even whether there are some tumors that just don't shed ctDNA at all," he said.
With Personalis, the team had the chance to assess a technology that could push the limit of detection to as low as 2 ppm, and in his AACR presentation, he shared data from his team's analysis of the potential value of that improvement in sensitivity.
"The question is … does it really matter if we can go beyond 80 parts per million in terms of sensitivity? How much clinical performance is gained as assay sensitivity improves?" Swanton said.
This January, TRACERx investigators published a study in Nature Medicine that explored whether the presence of pretreatment ctDNA correlated with patients' risk of relapse. Focusing on the adenocarcinomas in the cohort, the researchers calculated that early-stage patients who tested negative for ctDNA had a 100 percent five-year overall survival rate. In comparison, those who tested positive, even with the barest traces of tumor DNA in their blood, had a much higher risk of recurrence.
Analyzing only samples with ctDNA levels lower than the 80-ppm limit of detection of previously tested assays, the researchers found MRD-positivity remained prognostic for poor overall survival, which suggests that there is a subset of at-risk patients who can be definitively identified only by using an ultrasensitive ctDNA assay.
At AACR, Swanton said that further research has shown that across lung cancers, Personalis' test has shown both higher sensitivity and higher lead time compared to the assays used earlier in TRACERx.
Looking at landmark time points and five-year survival data, the team has found that ctDNA negativity doesn't have as high a negative predictive value as has been shown in other tumor types. Even among ctDNA-negative patients, at least 20 percent recur within five years. Looking at patients who were detected by the Personalis assay below 80 ppm, the researchers could define an intermediate risk population not able to be captured using a higher cutoff.
Swanton noted that 42 percent of patients are in that category, "so that's a much larger proportion of patients we can detect now with an ultrasensitive assay that do much worse than those who don't have ctDNA detected at all."
Another hint of the value of a more sensitive approach in the VICTORI data was in patients whose cancer had metastasized. Looking at cases where metastases were present, Titmuss said that 75 percent of them were detected in the ultrasensitive range, and 75 percent during the landmark time point. This is notable, she added, as the shedding rate of DNA from metastatic sites, especially the lung, can be lower than from a primary tumor.
Meanwhile, all patients in the study who have tested negative for ctDNA remain cancer-free as of the current analysis, which took place at median follow-up time of almost 16 months.
The study continues to enroll more patients, which the authors hope will produce robust enough evidence to guide future prospective randomized studies that incorporate ctDNA as a decision point for clinical management and care, something VICTORI is not itself tackling.
Narrowing in on an ideal testing time point and maximizing sensitivity so that more patients can be detected within it will be crucial for such studies to have the best chance of success, according to leaders in the pharma space who cited interventional trials in other settings, most recently breast cancer, that have faltered due to similar factors preventing the detection of ctDNA MRD from actually leading to effective changes in treatment.