The number of examples of genomic biomarker-guided therapy in oncology continues to grow. In non-small cell lung cancer (NSCLC), for example, current NCCN guidelines recommend testing of EGFR, ALK, ROS1, and BRAF as part of broad molecular profiling, which is considered “a key component of the improvement of care of patients with NSCLC.” Despite this, in 2017, we reported that less than 10 percent of patients with advanced NSCLC for whom testing was recommended actually received it (Clin Lung Cancer 2017;18(6):651-659). Only a small minority of the omissions were explained by lack of sufficient tissue for testing.
ASCO and NCCN guidelines similarly endorse genomic biomarker testing for patients with metastatic colorectal cancer (mCRC) and have done so since 2009 (J Clin Oncol 2017;35(13)1453-1486). Specifically, the current guidelines call for RAS (KRAS and NRAS) and BRAF testing in all patients with metastatic disease. In tumors that are RAS and BRAF wild-type, HER2 amplification testing is also advised. All newly diagnosed patients are also recommended to have universal mismatch repair or microsatellite instability testing. The guidelines have expanded to cover additional exons and genes over time, but the fact remains that some degree of molecular testing has been advised for more than a decade.
The results of these molecular tests are important for therapy selection: 1) patients whose tumors have RAS or BRAF mutations or HER2 amplification do not benefit from treatment with the anti-EGFR antibodies cetuximab and panitumumab; 2) patients whose tumors have BRAF V600E mutations or HER2 amplification benefit from specific therapies directed against these targets in combination with other agents; and 3) patients whose tumors are MSI-high are more likely to respond to immune checkpoint inhibitors.
Given increasing awareness among oncologists and the public about the key role precision medicine plays in oncology, we hypothesized that we would find evidence of improved testing rates compared to older data. We investigated a real-world cohort of patients with metastatic colon cancer (mCC) diagnosed between 2013 and 2017 and assessed the extent to which guideline-recommended testing was completed. The disappointing results were in a study by Gutierrez et al. in JCO Precision Oncology (2019; doi: 10.1200/PO.19.00274).
We conducted a retrospective review of the COTA Real World Data database and identified 1,497 patients with mCC diagnosed between 2013 and 2017. Seventy-seven percent of patients received care at an academic center (1,152) and 23 percent at a community site (345). At the time of the project, the COTA database included detailed clinical information abstracted and de-identified from the electronic health records (EHRs) of 23 practices including 258 oncologists across the U.S. Trained COTA abstractors searched structured, unstructured, and scanned document data available anywhere in the EHR for mention of biomarker testing. Rigorous quality assurance was conducted to ensure that rates of human error were extremely low. Patients were classified as “tested” if available biomarker data indicated relevant testing, regardless of method, vendor, or test completeness, at any point in the cancer journey. We also assessed alignment of testing to the version of the NCCN guidelines that were in effect at the time.
Biomarker testing consistent with guidelines was observed in only 601 patients (40%). Although appropriate testing was more likely among patients treated at academic centers as compared to community settings, rates were poor across the board (44% vs. 29%; p<0.001). We observed modest increases over time in the proportion of patients tested for individual biomarkers. However, with the expansion of guideline recommendations that occurred during the study period, the proportion of patients who received all guideline-advised tests in 2017 was in fact lower than it was in 2013. We observed increasing usage of next-generation sequencing panels as the testing methodology during the course of the study. Of the 177 patients in this cohort who were treated with anti-EGFR therapies, only 50 (28%) had guideline-concordant testing for RAS and BRAF. We also identified 12 patients (7%) who received cetuximab or panitumumab despite the presence of RAS mutations.
The primary limitation of this study was its reliance on retrospectively gathered data which was obtained in the course of routine clinical care. There are likely some patients who did in fact have biomarker testing, the results of which were never recorded in the EHR data. Because our approach was to completely scour the record, looking at structured and unstructured material as well as scanned documents, we do not believe that this type of misclassification was a major factor, but it is difficult to rule out without supplementary claims information.
Key Takeaways & Future Directions
There has been significant debate in the academic literature and lay press regarding the appropriateness of large-scale use of next-generation sequencing panels in patients with cancer (Expert Rev Mol Diagn 2019;19(2):89-93; Chicago Tribune, Is ‘precision medicine’ the answer to cancer? Not precisely; Sept. 19, 2018). One primary concern relates to cost effectiveness, which has been reported to be moderate even in a setting like advanced NSCLC (JCO Clin Cancer Inform 2019; doi: 10.1200/CCI.19.00002). There is, however, no debate about the utility of specific molecular biomarker tests as a means of guiding therapy for a growing number of malignancies, including mCRC.
What is the significance of under-testing? The most common risk in this setting is the possibility of administering anti-EGFR therapy to patients who are unlikely to benefit. These agents cause severe infusion reactions in up to 3 percent of patients, in addition to substantially higher rates of potentially severe skin and gastrointestinal toxicity. Furthermore, the anti-EGFR monoclonal antibodies are sufficiently expensive that the cost of testing for all 1,497 patients would be exceeded by inappropriate cetuximab or panitumumab therapy for 1 year in fewer than 5 percent of patients for whom it may have been inappropriately administered.
All of this information begs the key question—why? Why do fewer than half of patients who should be tested fail to have testing done 10 years after routine testing became a standard of care? This particular question was not addressed in the current study, but there are a number of potential explanations.
One is that the Centers for Medicare and Medicaid Services signed off on approval of the first ever next-generation sequencing panel test for cancer (FoundationOne CDx-F1CDx) less than 2 years ago, in March 2018. The extent to which commercial payers have followed suit is difficult to ascertain, as is the coverage status of other commercial panels which provide similar information. That said, single-gene tests have been readily available and reimbursed by payers for many years, so the extent to which NGS reimbursement challenges are a real explanation for low testing rates is probably limited.
Another barrier to testing includes limited tissue availability, an issue that will become less problematic as blood-based tests, so-called “liquid biopsies,” are increasingly available. Some physicians may be reluctant to test because of prolonged turnaround times in selected scenarios, and there may be appropriate reluctance to test in circumstances where patients are not well enough for therapy.
Finally, genomic testing has become increasingly complex, with new markers and commercial test offerings emerging at a dizzying pace. Particularly in resource-limited settings, oncologists may have low comfort levels with the scientific principles relevant to targeted therapy decision-making, not to mention the various commercial testing panels and their associated logistical, financial, and report idiosyncrasies.
Whatever the precise causes of under-testing, our study highlights that a great deal of work remains to be done. Until high-quality, affordable genomic testing is available widely to appropriate patients, it will remain very challenging to ensure that patients with metastatic CRC, NSCLC, and a growing number of other malignancies actually receive the right treatment. Further study is urgently needed to elucidate the causes of under-testing and to examine potential solutions.
ANDREW NORDEN, MD, MPH, MBA, is a neuro-oncologist who has been involved in the design and execution of dozens of clinical trials and observational studies. An expert in real world data and evidence, he serves as Chief Medical Officer at COTA Healthcare. CK WANG, MD, is a medical oncologist who is experienced in cancer program development and accreditation. An expert in clinical care delivery and guidelines, Wang serves as Senior Medical Director at COTA Healthcare. DANIEL LANE, PHARMD, PHD, MBA, is a pharmacist and experienced health outcomes researcher with expertise in generating real-world evidence from product or service conceptualization through real-world adoption and utilization. He serves as VP of Research at COTA Healthcare.