ctDNA’s Prognostic Strength, Low Sensitivity Seen in Studies

GALAXY Study Results Updated

In an afternoon session, Hiroki Yukami, MD, PhD, presented updated results of the GALAXY study, which examined 2998 patients with stage I-IV colorectal cancer who underwent ctDNA surveillance over a median 16.14 months following surgery. ctDNA-positive status was associated with worse DFS (HR, 10.53; P < .0001) in all stages as well as in stage II/III (HR, 12.05; P < .0001). The researchers also distinguished between patients with sustained ctDNA clearance and those with transient ctDNA clearance, in which a positive test occurred after an initial negative result. Recurrences occurred in 7.1% of patients with sustained ctDNA clearance, versus 85.2% of patients with transient clearance (P < .0001) and 89.4% of those with no clearance (P < .0001). “Sustained clearance indicates superior DFS compared to transient or no clearance,” said Dr. Yukami during his presentation.

Of 117 patients treated with adjuvant chemotherapy after testing ctDNA positive, subsequent ctDNA clearance was associated with better DFS (HR, 6.72; P < .0001). There were also better DFS outcomes among patients who saw a greater decline in ctDNA plasma levels after adjuvant chemotherapy (0%-50% versus 50%-100% reduction; HR, 2.41; P = .001).

Aparna Raj Parikh, MD, assistant professor of medicine at Harvard Medical School, served as a discussant for the GALAXY study. She acknowledged that ctDNA is the most powerful prognostic marker in oncology, but to be clinically useful it is necessary to consider its utility at the landmark time point, which is when decisions are made whether to treat with adjuvant chemotherapy. At that time point, the sensitivity of ctDNA is about 48% in the GALAXY study, which Dr. Parikh said is consistent with other data.
 

ctDNA ‘not sensitive enough’

“We know that postoperative ctDNA is only capturing 40%-50% of patients with recurrences in non–stage IV patients in multiple datasets to date. I think it’s really important to keep in mind the sensitivity of the different time points when you’re actually thinking about how to use this in clinic. The first generation of tests are certainly promising, but I would make the argument that these are just not sensitive enough,” said Dr. Parikh.

“Landmark testing is not yet sensitive enough to deescalate care in a patient where chemotherapy would otherwise be indicated, and surveillance testing has not yet demonstrated clinical utility. I think our goal to actually deescalate care would be to try to lower the ctDNA-negative population recurrence risk to akin to stage I patients, with that 5-year DFS of 93%-95%,” Dr. Parikh said.

Dr. Parikh offered some advice on how to use ctDNA outside of a clinical trial setting. She said that positive ctDNA results can help drive the decision to initiate adjuvant chemotherapy in concert with clinical and other factors.

“I’m pretty convinced by the data that ctDNA is prognostic, and though we still need outcomes data, in particular scenarios where I’m thinking of not giving chemotherapy, a positive test may sway me in that direction,” she said. She gave examples such as patients with a single high-risk feature, or a stage III patient with marginal performance status, or an elderly patient with low-risk stage III disease.

Dr. Kasi has financial relationships with Precision Biosensors, Elicio Therapeutics, Bayer, BostonGene, Daiichi Sankyo/AstraZeneca, Delcath Systems, Eisai, Elicio Therapeutics, Exact Sciences, Foundation Medicine, Guardant Health, Illumina, Ipsen, Lilly, MSD Oncology, Natera, NeoGenomics, QED Therapeutics, SAGA Diagnostics, Seagen, SERVIER, Taiho Oncology, Taiho Pharmaceutical, Advanced Accelerator Applications, Boston Scientific, and Tersera. Dr. Tie, Dr. Kim, Dr. Ellison, and Dr. Yukami did not disclose conflicts of interest. Dr. Parikh has financial relationships with Abbvie, Bayer, Biofidelity, CheckMate Pharmaceuticals, CVS, Delcath Systems, Foundation Medicine, Guardant Health, Illumina, Lily, SAGA Diagnostics, Scarce, Seagen, Taiho Oncology, Takeda, UpToDate, and Value Analytics Labs.