Experts Provide Real-World Advice for Molecular Testing in Gastrointestinal Malignancies

Johanna C. Bendell, MD

Differences in germline and somatic BRCA mutations in pancreatic cancer, the role of molecular testing in hepatocellular carcinoma (HCC), and treatment options for patients with BRAF-mutant, microsatellite instability–high (MSI-H) colorectal cancer (CRC) were among some of the topics that were discussed during the Peer Exchange portion of the Institutional Perspectives in Cancer (IPC) webinar on gastrointestinal (GI) malignancies, moderated by Johanna C. Bendell, MD, chief development officer and director of the GI Cancer Research Program at Sarah Cannon Research Institute. 

Bendell was joined by:

• Andrea Wang-Gillam, MD, PhD, of Washington University School of Medicine in St. Louis
• Tanios S. Bekaii-Saab, MD, of Mayo Clinic
• Axel Grothey, MD, of West Cancer Center and Research Institute
• Yelena Y. Janjigian, MD, of Memorial Sloan Kettering Cancer Center

Bendell: Significant changes have affected how we treat patients, particularly regarding how we should look at molecular profiling and how we should test patients with GI malignancies. In pancreatic cancer, does it matter if patients have a germline BRCA mutation versus a somatic BRCA mutation? What biomarkers should we be testing for?

Wang-Gillam: Germline BRCA1/2 mutations [are] validated for PARP inhibitors based on the POLO trial. Per NCCN [National Comprehensive Cancer Network] guidelines, [some] insurers will pay for PARP inhibitors for [patients with germline mutations]. Somatic mutations are slightly different. Some studies suggest that somatic BRCA1/2 mutations, and even ATF2 mutations, have some response to PARP inhibitors. However, studies have varied and [shown] inconsistent results. If we go by the book, BRCA1/2 mutations should be germline, but somatic mutations [should] be further studied and have been explored in various trials. 

We should look for various [biomarkers] because more information can help us. In the clinic, I normally prefer to do next-generation sequencing [NGS] from tissue if possible. If not, we can get liquid biopsy to look at cell-free DNA. We could find something interesting [from liquid biopsy], such as NTRK fusions, or BRAF, ATM, ATR, or CHEK2 mutations. Those can be helpful in determining [eligibility] for clinical trials, especially as some patients with ATM mutations may be sensitive to ATR inhibitors with chemotherapy. More information [is better] because you don’t want to miss anyone who has MSI-H [MSI-H] disease. 

Bendell: Would you test for MSI and BRCA mutations as the bare minimum for patients with pancreatic cancer?

Wang-Gillam: I do more than that, but it is a similar concept. I send my patients for the full panel because we are justified to [test for] NTRK fusions because NTRK inhibitors are tissue agnostic. If I’m testing for [NTRK], I could potentially test the whole panel. I do MSI-H, BRCA, as well as at least NTRK.

Bekaii-Saab: If you have a target that is present in 1 in every 200 patients, consistent testing will show it in 1 in every 200 patients. If you [test] every third patient, it won’t just be multiplied by 3, it will become 1 in every 10,000^th patient or really luck of the draw. We know these patients [with pancreatic cancer] don’t do well on chemotherapy. They do fine for a while, and then they progress. However, most patients with MSI-H [disease] and NTRK fusions do fantastic [with appropriate therapy].

Somatic BRCA [mutations] are important, and there are a number of other [targets] we are looking for. Also, some studies are looking at HER2- or BRAF-targeted strategies. They are rare mutations, present mostly in the KRAS wild-type [population]. KRAS G12C mutations are present in a couple percent of patients, maybe even less than that. [My advice is to] consistently test and test broadly if you can.

Bendell: Why wouldn’t NGS pick up a germline mutation?

Wang-Gillam: A majority of [molecular tests] depend on how they gain mutations or [rather the fact that] they are not engaged to test for germline mutations. However, some of the new panels are able to do it. To be able to test for germline mutations has a lot of social implications, so it needs to meet a high standard. Some of the NGS panels tell you the germline [mutational status] but is not [traditional] germline testing. If you want to get a full germline panel, you need to move forward with certain tests. If patients truly have BRCA-positive disease, that has a lot of implications. The patient may need genetic counseling, so that test has very high standards.

Janjigian: With pancreatic cancer, when we are looking at BRCA alterations, it is important to differentiate between germline versus somatic [mutations]. Some of the commercial assays don’t differentiate, particularly FoundationOne, because there is not consent for it. Even if you have a BRCA-mutated tumor, after resistance to platinum-based therapy, the data suggest it is highly unlikely that PARP inhibitors will help. We sometimes see these desperate efforts to control the disease with PARP inhibitors in platinum-resistant disease, but it is not all that helpful. 

Bendell: Is there a role of molecular profiling in HCC? Is PD-L1 status associated with response to checkpoint inhibitors? 

Bekaii-Saab: The clear answer is probably not at this stage [in HCC], but maybe eventually. Some specific targets appear to be mildly interesting, such as FGF19, that may have studies [evaluating] them. For the most part, MSI-H is almost never present [in HCC]. Frankly, we do not find targets of interest in HCC.

However, phase 2 studies have integrated biomarker [analyses], and PD-L1 staining has not been shown to have any predictive value for response. It is unclear which patients are likely to respond [to therapy] or not; there isn’t a good biomarker in HCC. For advanced disease, get a biopsy, but don’t get NGS.

Janjigian: In terms of HCC biology and targeted alterations, NGS may be limited, but it may help if we are able to validate the sequence of treatments. Data from James Harding, MD, [of Memorial Sloan Kettering Cancer Center], suggest that if you have Wnt activation on NGS, those patients may not do as well, or their disease control rate may be lower with checkpoint inhibitors. Again, it may not be a one-target, one-drug approach, but it may help you get a sense of tumor biology to sequence [treatment].

Bendell: Is anyone doing esophagogastroduodenoscopies [EGDs] to look for varices before testing patients with GI malignancies?

Bekaii-Saab: No. If you want to do it, you have to do it for sorafenib [Nexavar], lenvatinib [Lenvima], cabozantinib [Cabometyx], and ramucirumab [Cyramza].

Grothey: You can also look at varices on scans; your radiologist will give you some indication of whether they see varices and signs of portal hypertension. There is selective use of EGDs, but screening with every patient isn’t justified.

Bendell: We talked about patients with BRAF-mutant, MSI-H CRC. Does MSI status supersede the presence of BRAF mutations in terms of treatment selection?

Grothey: It depends on the stage. In early-stage disease, we know that the presence of MSI can offset some of the poor prognosis [BRAF mutations confer]. However, without the use of PD-L1 antibodies, BRAF-mutant tumors do poorly in the metastatic setting. For BRAF-mutant, MSI-H tumors, we have 2 approaches. I prefer immunotherapy up front because the durability of response is potentially higher, and it is better tolerated overall. However, I have had patients who did not tolerate pembrolizumab [Keytruda] based on significant diarrhea but did respond to the BEACON triplet [encorafenib (Braftovi), binimetinib (Mektovi), cetuximab (Erbitux)] and later the doublet [of encorafenib plus cetuximab]. [The patient] was one of the outliers who had longer benefit [with the doublet].

Bendell: Gastroesophageal cancers are being split into squamous and adenocarcinoma histologies. How does treatment differ for these subtypes?

Janjigian: For patients with adenocarcinoma, we have moved away from giving carboplatin plus paclitaxel because the literature suggests that [this regimen] is not good [for these patients]. For our patients with adenocarcinomas, we typically use capecitabine plus oxaliplatin.