Multiomic Screening Modality Shows Promise in Early-Stage CRC

Girish Putcha, MD, PhD, chief medical officer and clinical laboratory director at Freenome

Girish Putcha, MD, PhD

A multiomic blood-based screening assay showed higher sensitivity and specificity versus singular tumor-based assays in detecting early-stage colorectal cancer (CRC), according to preliminary findings from the AI EMERGE trial (NCT03688906) presented at the 2020 Gastrointestinal Cancers Symposium.

The assay, which combines tumor- and immune-derived signals from cell-free DNA, epigenetic, and protein biomarkers, is thought to provide a more comprehensive profile of an individual’s microenvironment and possess the ability to overcome some of the challenges faced with current screening modalities.

Among 591 participants with an average screening risk, investigators identified 43 cases of CRC and 548 cases of colonoscopy-confirmed CRC-negative controls. The samples in each cohort were well balanced. In the CRC cohort, 56% of participants were male and the median age was 62 years. In the control cohort, 54% of participants were male and the median age was 59 years. In terms of staging, 53% of patients had stage I/II disease, 34.5% had stage III/IV disease, and 12.5% were unknown.

Each sample was analyzed via whole-genome sequencing, bisulfite sequencing, and protein quantification. For CRC, the average sensitivity and specificity of the multiomic assay were both 94% in those with early-stage disease (n = 17). The average specificity was 91% in those with late-stage CRC (n = 11) with a specificity of 94%.

“In our comparisons with 2 other CRC screening approaches, mutation detection testing via circulating tumor DNA (ctDNA) and carcinoembryonic antigen (CEA)-based classifiers, we [also] saw better sensitivity [with our blood-based test],” said lead study author Girish Putcha, MD, PhD.

In an interview with OncLive, Putcha, chief medical officer and clinical laboratory director at Freenome, provided insight on the study results and highlighted the advantages of using a multiomic, blood-based assay in CRC screening.

OncLive: Could you provide background on this trial?

Putcha: I presented the initial results from the AI EMERGE trial, [which evaluated a blood-based screening test] in average-risk patients [determined to be eligible for screening]. These individuals were asymptomatic and were due for CRC screening. We're trying to establish another method of screening for CRC because some of the challenges with the current screening modalities [are prohibitive]. Approximately one-third of eligible individuals are not getting screened. By providing a blood-based screening test that will allow those individuals to get screened, we hope to improve adherence and disease-related mortality.

What distinguishes this assay from other assays in CRC?

Our approach is unique in that we combined tumor-derived and nontumor-derived immune signals to [capture] the biological heterogeneity of the disease as well as the evolution of the disease. By using a multiomics approach, we hope to [identify] late-stage disease where there is plenty of tumor-derived signals, as well as early-stage disease, where there is far less tumor-derived signals through the use of nontumor-derived signals. We're focused on the immune system because it is designed to detect foreign [signals] and amplify [those] signals.

What was the design of the study?

We included patients with CRC as well as those who were colonoscopy-confirmed CRC negative. We analyzed patients with whole-genome sequencing, bisulfite sequencing, and protein quantification methods. Then we did 4-fold cross validation and reported model performance. We compared our multiomics test to a lead-in fecal immunochemical test (FIT), a ctDNA-based mutation detection test that is commercially available, as well as to a common tumor protein called carcinoembryonic antigen.

What were the results of the study?

In [patients with] stage I/II CRC, we saw a 94% sensitivity and specificity rate [with the assay]. In [those with] stage III/IV CRC, the sensitivity rate was 91% sensitivity and [the specificity rate was] 94%. The performance [of the test] was similar in both distal and proximal lesions, which is really critical, as well as in comparison with a lead-in FIT-based test.

We asked individuals who gave us blood samples whether they would also give us a stool sample so that we could do the comparison. Only 52% of individuals who gave us a blood sample also gave a stool sample for testing, which highlights some of the challenges with stool-based testing and patient willingness to undergo such testing.

However, according to the matched samples, [the blood-based test] had better sensitivity versus FIT testing at 100% versus 67%, respectively, at a specificity of 96%. The performance of FIT in our research was exactly what we would expect in a prospective study, such as in the DeeP-C study, which showed about a 74% sensitivity at 95% specificity. This reassures us that our results are representative of what we should expect to see in our validation study that will be launching later this year [after] registration with the FDA.

What is the key takeaway from this research?

The take-home message is that these results suggest the possibility of having a blood-based test for screening in CRC. The sensitivity with [the blood-based test] was approximately 90% versus 47% with the ctDNA-based assay, and the specificity was 100% versus 75%, respectively. The sensitivity for [the blood-based test] was 91% versus 31% with the CEA-based assay with a specificity of 94% for both assays. This shows that the multiomic approach appears to hold real promise.

Many different liquid biopsy approaches [are being used] for screening as well as in the adjuvant setting. [These tests can be difficult to interpret in the adjuvant setting] because you're looking at patients after they have had surgery, when their disease burden isn't much less, to see if they may recur. As such, I'm really interested to see some of the work that is happening in that area because it has implications for the kinds of work that we're doing.

Putcha G, Liu T-Z, Ariazi E, et al. Blood-based detection of early-stage colorectal cancer using multiomics and machine learning. J Clin Oncol. 2020;38(suppl 4; abstr 66). doi: 10.1200/JCO.2020.38.4_suppl.66.