cfDNA Assay Shows Predictive Value in Detecting Cancer, Tissue of Origin

David D. Thiel, MD

David D. Thiel, MD

A cell-free DNA (cfDNA) test demonstrated the potential to detect cancer and predict tissue of origin in patients with a suspicion of cancer, according to findings from the Circulating Cell-free Genome Atlas (CCGA) study presented at the 2020 AACR Virtual Annual Meeting I.

Late cancer detection is one of the contributing factors toward advanced cancer diagnoses and cancer-related mortality. In particular, 30% of patients with breast cancer present with regional or distant metastases at diagnosis, as do roughly 55% of colorectal cancers and about 75% of lung and bronchial cancers.

“Many cancers are detected too late. The large percentage of breast, colorectal, and lung cancers are diagnosed simultaneously with metastatic disease. The detection of cancer prior to the development of metastatic disease can improve 5-year survival,” said David Thiel, MD, chair of the Department of Urology at Mayo Clinic.

The chance of 5-year cancer-specific survival decreases from 89% in patients who are diagnosed early to 21% in those who are diagnosed late. This is especially true in lung cancer, where the 5-year cancer-specific survival drops from 56% to 5%, in patients who are diagnosed early versus those who are diagnosed late, respectively.

Thiel and coinvestigators proposed that a multi-cancer early detection test could decrease cancer-related mortality. The development of the assay is underway in multiple clinical trials, including the CCGA study.

A total of 15,254 participants with and without cancer were enrolled across 142 sites in the prospective, longitudinal, case-controlled, observational study. Blood samples were collected from all participants, and tissue samples were collected from all participants with cancer. The follow-up was 5 years and consisted of vital status and cancer status. To be eligible for enrollment in the study, participants had to have a confirmed pathologic diagnosis of cancer or a high suspicion of cancer. Altogether 50 different cancer types were tested.

The initial analysis showed that the assay had 99% specificity and 90% accuracy in predicting the tissue of origin.

The objective of the CCGA study was to determine the best performing assay to move forward in development. To identify which assay was the highest performing and make a selection for further validation, the study was broken down into 3 prespecified substudies.

In substudy 1, discovery analyses were conducted with 3 independent assays. In substudy 1, a total of 1,785 patients were included in the training portion, and 1,015 patients were included in the validation portion. Three comprehensive sequencing strategies were compared in this substudy to help determine the best assay for further development. These approaches included targeted sequencing, whole-genome sequencing, and whole-genome bisulfite sequencing.

Whole-genome bisulfite sequencing performed better than the 2 other approaches, and the chosen approach for further development was a methylation-based assay.

Substudy 2 was carried out to refine the chosen assay with 3,113 patients in the training portion and 1,354 patients in the validation portion. Substudy 3, which is ongoing, is the final validation study and will include 5,000 patients.

A subgroup analysis evaluated the test performance in patients with high clinical suspicion (HCS) of cancer. These participants underwent clinical and/or radiological assessment, planned biopsy, or surgical resection to establish a definitive diagnosis within 6 weeks after their first blood draw. In total, 303 patients were included in the subgroup analysis.

Of those who had clinically confirmed cancer, 164 were included in the training group and 75 were included in the validation group. Among participants who did not have cancer, 49 were included in the training group and 15 were included in the validation group. Of the patients with cancer, 20 cancers ranging from stage 1 to 4 were identified, and 1 patient with stage 3 disease of unknown origin was identified in the validation group.

The specificity observed in substudy 2 and the HCS subgroup analysis was comparable in the training and validation groups. In substudy 2, the test achieved 99.8% specificity (95% CI, 99.4-99.9%) in the training group and 99.3% (95% CI, 98.3-99.8%) in the validation group. The subgroup analysis showed 100% specificity (95% CI, 92.7-100%) in the training group and 100% (95% CI, 78.2-100%) in the validation group.

“The high specificity suggests that the false positive rate was not elevated in individuals with high risk compared with the enrolled population,” said Thiel.

The CCGA substudy 2 and the HCS subgroup also showed comparable sensitivity with the test. In the CCGA substudy 2, the sensitivity of the assay was 40.2% (94% CI, 32.7-48.2%) in the training group. The sensitivity in the validation group was 46.7% (95% CI, 35.1-58.6%). In the HCS subgroup, the sensitivity was 50.4% (95% CI, 49.5-59.2%) in the training group and 59.3% (95% CI, 45.7-71.9%) in the validation group. Thiel noted that kidney cancer was excluded from the subgroup analysis, whereas the CCGA study included all cancers.

In the HCS subgroup analysis, the accuracy of predicting tissue of origin differed between patients in the training group (85.5%; 95% CI, 74.2-93.1%) versus those in the validation group (97.1%; 95% CI, 85.1-99.9%). However, the precision of detecting specific cancers types, such as lung cancer, ovarian cancer, and colorectal cancer, were precise 100% of the time in both the training and validation groups.

Moreover, the accuracy of the test in predicting the tissue of origin in the CCGA substudy 2 and the HCS subgroup were comparable.

Thiel concluded that early detection with the targeted methylation-based multi-cancer assay could potentially accelerate the diagnostic workup in patients with high suspicion of cancer. The test is undergoing further evaluation in the STRIVE study (NCT03085888) and the SUMMIT study (NCT03934866).

Thiel DD, Chen X, Kurtzman KN, et al. Prediction of cancer and tissue of origin in individuals with suspicion of cancer using a cell-free dna multi-cancer early detection test. Presented at: 2020 AACR Virtual Annual Meeting I; April 27-28, 2020. Abstract CT021.

<<< 2020 AACR Annual Meeting