Systematic Review Spotlights Potential Markers for ctDNA Use in CRC

Sander Bach, a research associate in the Department of Surgery at Amsterdam Universitair Medische Centra

Sander Bach

Interest in circulating tumor DNA (ctDNA) has grown in oncology, said Sander Bach, a research associate in the Department of Surgery at Amsterdam Universitair Medische Centra. However, inconsistent literature regarding optimal ctDNA candidate markers in colorectal cancer (CRC) warrants further investigation.

To that end, a systematic review was conducted to parse out the accuracy of ctDNA for early detection, prognostication, and monitoring of CRC by analyzing mutations, chromosomal copy number alterations, and methylation in plasma or serum.

The review, which was published in Journal of the National Cancer Institute Cancer Spectrum, included 84 previously published studies of patients with treatment-naive CRC. The results showed that for CRC detection, the sensitivity of ctDNA mutation analysis was 75% compared with up to 96% with copy number alteration analysis. Septin 9 hypermethylation analysis demonstrated 100% sensitivity and 97% specificity for CRC detection.

In terms of prognostication, KRAS mutations that were discovered via ctDNA were associated with oncologic outcomes and appeared to be predictive of response to EGFR inhibition.

Regarding monitoring, sequential ctDNA analysis of KRAS mutations demonstrated promise pertaining to relapse or treatment resistance.

"Research output is still very small, but if we wait a few years and sharpen our skills, it will be interesting," said Bach. "We hope our review helps other researchers understand what evidence is available regarding aberrations and ctDNA, and I believe we succeeded in doing that."

In an interview with OncLive, Bach discussed the rationale for conducting this systematic review in CRC, key inclusion/exclusion criteria, and the potential clinical implications of these findings.

OncLive: What is the role of ctDNA in CRC, and what was the rationale for conducting this review?

Bach: In our review, we looked at all of the literature that is available to answer that question. In our paper, we identified 3 objectives in the clinic that could use ctDNA to improve patient outcomes.

The first is to detect CRC at an early stage. That is important because we know that the earlier a cancer is detected, the better a patient's prognosis is. Another [objective] was [to establish] the use of ctDNA to prognosticate the kind of therapy patients will likely benefit from and clearly communicate what they can expect from therapy. ctDNA could help with that.

The other area we could use ctDNA in is more advanced stages of CRC to monitor disease after systemic chemotherapy or targeted therapy. This is applicable in the advanced stage and after primary curative surgical resection where we could detect minimal residual disease (MRD). ctDNA is a good option to [detect MRD] because it is a noninvasive, easy-to-do procedure. Additionally, ctDNA could implicate certain additional therapies [in a patient's treatment course].

What information was included in the review?

Essentially, we looked at patients with CRC who did not have any form of treatment prior to ctDNA sampling in the plasma. We did that because systemic therapy and surgical resection can influence the ctDNA profile.

Then we looked at the detection of ctDNA through various forms of detection methods, including DNA methylation, DNA mutations associated with CRC, and chromosomal copy number alteration, which is more novel and hadn't been [done] in a formal review. Of course, we have known for a long time that [these alterations] happen in CRC, but there is not much literature about these types of aberrations in patients' ctDNA.

To understand the effectiveness of [detecting these aberrations in the ctDNA], we looked at a lot of literature. We started out with more than 1000 articles. At the final qualitative synthesis, we [included] 84 articles. From those 84 articles, we looked at sensitivity and specificity of ctDNA.

What factors determined which studies were included in the review?

We selected [studies that included] treatment-naive patients. That was one of the most important [criteria], and we excluded a lot of articles because they did not comply with that. We excluded a lot of articles because they were not in English or other languages we could [not] understand.

We found a lot of articles that were not looking at ctDNA in plasma or serum, or other kinds of aberrations that can be found in the blood of patients with cancer, such as microRNAs. Those studies were excluded.

Additionally, we only included text articles and complete research papers and excluded abstracts and posters from conferences.

[Once we had compiled these studies], we looked at the risk of bias of each article according to the well-known QUADAS-2 tool. That is a validated way to score for the risk of bias. We had 134 articles left [before using the QUADAS-2], so we used [the tool] to exclude more studies that scored high for bias from the final assessment.

In the end, we had compiled 84 papers that had a low risk of bias and were therefore likely believable results and high-quality research.

What did the results of the review demonstrate?

The most evidence we found for early detection of CRC is with ctDNA and the hypermethylation modality. We found the most papers on this subject. [We also saw] a high sensitivity in which methylation [of] Septin 9, which is currently a commercially available assay, was the most promising with a very high sensitivity around 90% in some papers. It also showed very high specificity.

DNA mutation for the detection of CRC was worse when compared with methylation. The sensitivities were at best around 50%. However, most of them were lower.

The use of copy number variation for the detection of CRC was actually quite good, but there is little evidence on this subject. In our study, there were only 3 papers that used copy number variation aberration detection in treatment-naive patients. Oftentimes, [it was used] on patients who already underwent therapy and therefore had to exclude those articles. It looks promising, but there is not much evidence to show that it works. More research need to be done to confirm these findings.

For the second modality, prognostication, we found that mutations are associated with oncological outcomes and can predict the risk of response to specific therapies. This is an interesting fact. There was less known about methylation in this regard and even less for copy number variation.

For monitoring, DNA mutations look promising. Some studies showed promise for detection of relapses [due to] treatment resistance.

In summary, for cancer detection, DNA hypermethylation is the most promising, according to our review. Regarding prognostication and monitoring, there is evidence for DNA mutations. As for the other modalities, we need more research before we can say that this would be interesting to look at in prospective clinical trials.

What are the next steps regarding research with ctDNA?

A lot of research is being done regarding ctDNA, which is one of the reasons we conducted this study. In our lab, we are doing a lot of ctDNA research in this patient population.

When we started this project, we didn't have what we thought was very good literature looking at all of the types of aberrations that you can look at with ctDNA.

Prospective clinical trials are ongoing that are looking at ctDNA as a way to stratify patients for additional therapy. Those results are beginning to come out, but the next couple of years will show the power that ctDNA has in this field.

I am hopeful [that we will be able to detect] copy number variations in the blood. In theory, it has a lot of advantages because we need less blood, less DNA in the blood, and it is more sensitive than methylation, which only looks for mutations in a very small part of the DNA.

Bach S. Sluiter NR, Beagan JJ, et al. Circulating tumor DNA analysis: clinical implications for colorectal cancer patients. A systematic review. JNCI Cancer Spectr. 2019;3(3):pkz042. doi:10.1093/jncics/pkz042