ctDNA Scratches the Surface in Colorectal Cancer

Axel Grothey, MD

Although circulating tumor DNA (ctDNA) has been shown to be prognostic in colorectal cancer (CRC), research regarding its predictive capacity is in its infancy, explained Axel Grothey, MD, who added that pending further research, the marker could be used to anticipate treatment response, inform the need for escalated and de-escalated therapy, and potentially even re-exposure to prior therapy.

“ctDNA is definitely a strong prognostic marker. In the context of a predictive marker, we look at the treatment effect. This is where we don’t have good prospective data yet. In order to identify a predictive marker, you need prospective randomized trials, and those have not been conducted yet,” said Grothey.

In an interview with OncLive^®, Grothey, a medical oncologist and director of Gastrointestinal Cancer Research at West Cancer Center and Research Institute, discussed the potential applications for ctDNA in CRC and shed light on the updated analysis from the pivotal IMbrave150 trial in advanced hepatocellular carcinoma (HCC).

OncLive^®:What assays are used to evaluate ctDNA as a marker for minimal residual disease (MRD) in CRC?

Grothey: ctDNA assays break down into 2 different types of assays. One is a tumor-informed assay, meaning we profile the cancer, do whole-exome sequencing, identify a genetic fingerprint of the cancer, and then develop a specific and individualized test for the patient, which is highly sensitive. That is a little more complicated, because we need to first profile the cancer, and then really develop this test kit on a PCR [polymerase chain reaction] basis for the individual patient.

The other test is not tumor informed and is kind of a one-size-fits-all test, where we look at common mutations in cancer methylation markers, mutations like APC for colon cancer, RAS mutations, etc. [While it] is not tumor informed, it’s easier to use, because it’s really kind of an off-the-shelf test. It’s probably not as sensitive as the [other] assay if it’s [not] tumor informed and not specific to individual patients. However, both of these kind of tests are available right now.

A study published in JCO Precision Oncology demonstrated that the Signatera assay detected esophageal cancer recurrence approximately 1 year before current standard assays. What are your thoughts on that assay?

The Signatera assay is one of those tumor-informed tests, which really tests biologic principle. We have a lot of data in colon cancer, but it should be true for bladder cancer, breast cancer, lung cancer, esophageal cancer, etc. What we really try to do after resection or after biopsy of a cancer is profile the tumor and then see whether there’s still some kind of remnants of this tumor circulating somewhere in the patient’s blood.

We can use this test for MRD diagnosis, but also to monitor our treatment approaches because we know that the amount of DNA that we find in the blood really correlates with the amount of cancer in patients, and we can look at treatment responses.

As it stands, is ctDNA a prognostic as well as a predictive marker?

In oncology, we make a lot of treatment decisions based on prognostic markers. We treat, for instance in the adjuvant setting, higher-risk patients, like [patients with] T4N2 colon cancer, patients who have a higher stage of disease, [vs those] patients with stage II disease. We use prognostic markers to make treatment decisions. Now we don’t have the predictive value. Does a positive test really correlate with a certain treatment response? That’s the missing link that we need to feel very sure.

Currently, is ctDNA being used to guide treatment escalation or de-escalation strategies in practice?

If I have a patient with stage II colon cancer and is low risk, where I normally do not offer adjuvant therapy and I see that their ctDNA after surgery is positive, I know this patient has a very high risk of recurrence: it’s virtually 100% within the next 3 years. My tendency is to offer these patients adjuvant therapy because I know that we can convert patients from positive to negative, which is a prerequisite, more or less, for cure. We have seen more emerging data now that the assay negativity can actually be longer lasting, which will hopefully translate into cure at some point.

De-escalation is not as easy to do because we have a high specificity for recurrence, but we don’t have a lot of sensitivity in the assay [to say that] patients won’t recur. When I have one assay [that shows negativity] after resection, in a [patient with] stage III colon cancer in whom I would normally offer adjuvant treatment, I cannot be sure that [surgery alone] will translate into cure.

If at all, we need to do sequential testing, but then we lose the time window, which we normally consider for adjuvant therapy. Therefore, this de-escalation approach is currently being tested in various clinical trials in Australia, Japan, and the United States, which will allow us to in a controlled fashion de-escalate treatment, and then only use adjuvant therapy over time when we do sequential analysis when the assay turns positive, then we will start delayed adjuvant therapy to salvage the situation.

We need to be very sure that we’re not harming patients, because withholding treatment is much more difficult than offering treatment where we would normally not offer treatment based on a prognostic marker.

What could the implications of ctDNA be on diagnosis and screening?

The use of ctDNA for diagnosis and screening is a brand-new field that’s really fascinating. It [uses not] mutation markers, but methylation markers. There’s a lot of effort right now to develop highly sensitive ctDNA assays to do population screening, and to look at early-stage disease—kind of as a substitute for colonoscopy and pap smears and mammograms that you can substitute by just going to your doctor once a year and have one blood draw that may look for early markers of malignancy in the pre-malignant stage.

This has the potential to change how we really screen for cancer overall. We’re not there yet. Though, the data are very interesting. We have some data in colon cancer, where we benchmark it against colonoscopies, but it will take some time to develop these tests.

Could ctDNA also be used for serial monitoring?

The other aspect of ctDNA that we haven’t talked about yet is using it as a marker to identify patients for rechallenge of certain therapies. We use EGFR antibodies in patients with RAS/RAF wild-type left sided colon cancers, but we’ve learned that there are resistance markers that emerge on treatment. The cancer is being pushed to develop RAS-mutant clones, which we can detect by ctDNA. Those patients do not respond to EGFR antibodies. Now, these clones, these sub-fragments have a half-life. When we take away the selection pressure of EGFR antibodies, the half-life of these remnants and resistant clones is about 3 to 4 months. Over time, we can test for the presence or absence of these clones and rechallenge patients and rechallenge cancers by using EGFR antibodies.

There are some interesting data from Italy, prospectively generated data, that show in the absence of these RAS-mutant clones, by ctDNA likelihood of response to chemotherapy plus EGFR antibodies, is higher. It’s kind of an on-treatment monitoring [tool we can use to] look at markers of secondary resistance, which I believe is one of the missing links that we need to embrace in the continuum of care and the treatment options a patient has over time.

Shifting to HCC, has the updated analysis of the IMbrave150 study confirmed the use of atezolizumab (Tecentriq) and bevacizumab (Avastin) as the frontline standard of care for patients with advanced disease?

Finally, we have a standard of care that’s kind of universally embraced. We played around with TKIs, sorafenib [Nexavar] and lenvatinib [Lenvima] and we had some immunotherapy data, but from single-arm studies. Randomized data really didn’t pay off with nivolumab [Opdivo] and pembrolizumab [Keytruda] in first- and second-line therapy. Finally, we have a combination of a PD-L1 antibody and a VEGF inhibitor. We’ve always assumed that this is not a bad combination, because VEGF inhibitors by themselves have immune modulatory components and activity and activation of dendritic cells.

I look at the atezolizumab/bevacizumab data and see we have a standard of care, at least for now. We might have better immunotherapy combinations. The interesting thing is that the activity is independent of how HCC originated whether it’s hepatitis B, hepatitis B, or hepatitis C, or NASH [non-alcoholic steatohepatitis], or alcoholic HCC. It doesn’t really matter. We have a standard of care, which is pretty well-tolerated overall. People might not realize these studies screen for esophageal varices endoscopically, so that’s something we are embracing right now. It’s not that you can’t do it, but you need to treat these varices and then you have a lower risk of bleeding on atezolizumab/bevacizumab. I believe it establishes a new standard of care, and it’s the benchmark for future comparisons.

Considering pembrolizumab and nivolumab are under re-review by the FDA under their accelerated approvals, has that changed your thoughts on immunotherapy in HCC? Does the announcement suggest that if immunotherapy is going to be used in HCC that it has to be in combination?

I believe that [immunotherapy] has to be used in combination. The data we have for pembrolizumab or nivolumab, the accelerated approvals were based on single-arm phase 2 studies [with] response rates and durability of response. The randomized data have not been intriguing in the first-line [setting], nivolumab vs sorafenib and in the second line pembrolizumab against placebo didn’t really met its primary end point.

On the other hand, we’ve seen data with pembrolizumab and lenvatinib in [HCC]. That was very interesting data. We’ve seen combinations of PD-1/PD-L1 antibodies plus CTLA-4 antibodies, so the future probably not just in HCC but, overall, in immunotherapy in various cancers, is in combination, either with another [checkpoint inhibitor], or in combination with a kinase inhibitor that at least has a VEGF inhibitory component and then something else, a signature effect. This is because I believe we’ll eventually move beyond bevacizumab as a combination partner for atezolizumab, which only [inhibits] VEGF. The data that I’ve seen with lenvatinib, which is also a multikinase inhibitor, including an FGFR inhibitor, are very intriguing. Hopefully, we’ll see this emerge as one of the new standards of care.