NEW YORK – Predicting which patients with EGFR-mutated non-small cell lung cancer will derive benefit from targeted therapy after surgery may be greatly improved by looking beyond the single EGFR biomarker, according to a recent study published in Nature Communications.
While the advent of EGFR inhibitors — including gefitinib (AstraZeneca's Iressa), erlotinib (Roche's Tarceva), and more recently, osimertinib (AstraZeneca's Tagrisso) — has brought lifesaving survival benefit to certain NSCLC patients, the benefit has not extended to all treatment settings, nor to all patients with tumors harboring EGFR mutations. In fact, in the adjuvant treatment setting, between 19 and 40 percent of EGFR-mutated NSCLC patients who receive one of these treatments end up experiencing disease relapse, even if the treatment initially showed encouraging disease-free survival improvement.
Using next-generation sequencing panels from the Nanjing, China-based firm Geneseeq, researchers from Guangdong Provincial People's Hospital in China, in collaboration with Geneseeq, conducted a retrospective analysis of the Phase III ADJUVANT-CTONG1104 study analyzing patients' molecular profiles alongside their outcomes on EGFR TKIs. The key to homing in on which EGFR-mutated NSCLC patients are most likely to benefit from EGFR TKIs versus standard-of-care chemotherapy agents, according to their findings, could be testing patients for a collection of molecular biomarkers — that is, a predictive genomic signature — rather than EGFR alone.
"EGFR-mutated lung cancer shows a high degree of genetic heterogeneity, so different concurrent molecular mutations are common," explained Hua Bao, Geneseeq's director of research and development and an author on the Nature Communications paper. "So, our hypothesis was that concurrent genomic alterations might contribute to differential response to adjuvant treatment and that a single gene might not explain all the variance of differential clinical benefit."
The clinical trial that Bao and his team considered as a basis to test this hypothesis, ADJUVANT-CTONG1104, pitted gefitinib against standard-of-care cisplatin and vinorelbine chemotherapy as treatment for stage II to IIIA EGFR-mutated NSCLC patients following surgery. The study randomized 222 patients to one or the other and measured disease-free survival as a primary endpoint.
Generating predictive MINERVA scores
To conduct their retrospective study, the researchers first took 171 patient samples available from the ADJUVANT-CTONG1104 study and sent them to Geneseeq's lab, where they analyzed each sample with Geneseeq's NGS-based Geneseeq Prime Panel of 422 cancer-related genes.
Pairing these NGS results with the clinical outcomes these patients had experienced on the ADJUVANT-CTONG1104 study, the researchers then zeroed in on five biomarkers beyond EGFR that played into patients' responses to one treatment type or the other. These biomarkers included TP53 exon 4/5 mutations, RB1 alterations, copy number gains of NKX2-1, copy number gains of CDK4, and copy number gains of MYC.
After selecting these five biomarkers, they then came up with a score, dubbed the Multiplegene Index to Evaluate the Relative benefit of Various Adjuvant therapies, or "MINERVA." The score is designed to categorize patients into three predictive groups based on these biomarkers and preferred treatment response: one being "Highly TKI-Preferable," one being "TKI Preferable," and one being "Chemotherapy Preferable."
On a genomic level, patients whose NGS testing results stratified them into the Highly TKI-Preferable group had tumors enriched with copy number gain of NKX2-1, CDK4, and MYC as well as TP53 exon 4/5 missense mutations. Patients in the Chemo-Preferable group, meanwhile, were more likely to have RB1 alterations. Patients that fell into neither category were stratified to the TKI-Preferable subgroup, which in most cases meant that their tumors lacked any of the five predictive biomarkers but in some cases harbored both NKX2-1 and RB1 alterations, which canceled each other out in terms of treatment response to either TKI or chemo.
The Chemotherapy Preferable subgroup — which comprised patients for whom chemotherapy actually resulted in preferable disease-free survival versus TKIs — was particularly notable to the researchers. These patients, based on the status of these five additional biomarkers in their tumors — particularly RB1 alterations — had a lower possibility of benefiting from gefitinib despite having the crucial EGFR mutation the drug is meant to target.
Predicting overall survival
Importantly, while the randomized, prospective ADJUVANT-CTONG1104 study initially showed a disease-free survival benefit among the gefitinib-treated patients, longer follow-up data presented last year showed that the TKI did not likewise result in a significant overall survival benefit. Although the drug extended the time without disease relapse, in other words, many of these patients ultimately experienced recurrence, and as such, their overall survival did not favor gefitinib. After nearly six and a half years of follow-up, the overall survival outcomes in the two trial arms were disappointingly similar.
In Bao and colleagues' retrospective analysis, however, applying the MINERVA scores based on the five-biomarker genomic signatures rather than the EGFR mutation status alone revealed a specific group of patients — the Highly TKI-Preferable group — for whom overall survival benefit did favor gefitinib versus chemotherapy. In this group specifically, the median overall survival for those treated with gefitinib was not yet reached at five years, versus 48.7 months for those treated with chemotherapy.
Although patients in the TKI-Preferable group, unlike those in the Highly TKI-Preferable Group, experienced similar overall survival outcomes with either therapy, Bao explained that they were still considered TKI-Preferable given that the disease-free survival time was improved with gefitinib. Of course, discussions have rippled through the field in the past years about whether a drug that prolongs disease-free survival but not overall survival is really a preferable option for patients. Previous research has pointed to quality-of-life improvements on TKIs versus chemotherapy as well, which stands as an additional consideration, although higher price tags with the targeted TKI may also be a concern.
Validation, next steps
After identifying the five predictive biomarkers beyond EGFR and generating a predictive MINERVA score for adjuvant NSCLC treatment accordingly, Bao and colleagues sought to validate it further by applying it to an independent cohort of EGFR-mutated NSCLC patients enrolled in the EMERGING-CTONG1103 trial. In this trial, erlotinib rather than gefitinib was the TKI pitted against chemotherapy. But even so, the three subgroups defined by the MINERVA scores predicted which patients were most likely to benefit from each treatment arm, lending further validation to the approach and suggesting that it may not be unique to gefitinib, or any one specific EGFR TKI for that matter. This is particularly encouraging given as newer generations of EGFR TKIs such as osimertinib have since entered the treatment landscape for adjuvant NSCLC treatment, but with significant numbers of treatment-resistant or refractory patients for whom better biomarker stratification would bring tremendous value.
"I think our method can be easily transferred to other settings, with other TKIs, or even in not only adjuvant therapy but neoadjuvant therapy," Bao said, mentioning that his team has seen some encouraging initial results in the pre-surgery treatment setting. Perhaps down the line, he added, using the stratification scores could also help predict patient benefit from adjuvant immunotherapy or perhaps with other tumor types.
Given the retrospective nature of the research, Bao emphasized that it will be crucial to validate the subgroup approach in prospective trials before it can be applied to guide NSCLC patients' treatment decisions going forward. "In the future, if we can validate this in a prospective study, when the patients come to the clinic, we can sequence the tissue using a gene panel, find these predictive gene mutations, and predict whether this patient should choose TKI or chemotherapy," he said.
But, particularly for earlier stages of disease in the adjuvant treatment setting, Bao emphasized that these types of studies take significant time. Five-year survival rates tend to be higher for early-stage NSCLC patients, meaning that it could take the better half of a decade to confirm whether treating patients based on these three multi-biomarker-informed EGFR subgroups could ultimately improve survival across the board.
Though it will clearly take time to validate this work, the study's implications could be more immediate for the field's approach to molecularly guided NSCLC treatment and more broadly to single- versus multi-gene biomarker testing. The study is not the first to suggest that considering just one biomarker at a time may be inadequate in terms of predicting treatment outcomes, but it is one of the first to focus on the adjuvant setting, and to do so with a commercial NGS panel, Bao said.
"Right now, patients don't have many options for adjuvant treatment," he said. "If NGS can help, that's a good thing. … Most previous studies have focused on single genes."