Genetic Testing for Lung Cancer: Reflex vs. Clinical Selection

Interest in activating “driver” mutations in tyrosine kinase genes as therapeutic targets in non-small cell lung cancer (NSCLC) has peaked recently due to the discovery that EGFR tyrosine kinase inhibitors (TKIs) produce higher response rates, longer progression free survival (PFS) times, less toxicity, improved symptom control, improved quality of life, and greater convenience compared to cytotoxic chemotherapy in the first line treatment of advanced NSCLC patients harboring EGFR activating mutations (1–4). A more recent study showed that advanced NSCLC patients whose tumors harbor EML4-ALK fusion genes, who are treated with an ALK TKI termed crizotinib, have high response rates and long progression free survival times that are similar to those of NSCLC patients harboring EGFR mutations treated with erlotinib (5,6).

Recent genomic studies in adenocarcinoma of the lung identified other mutations in tyrosine kinase genes including KRAS, NRAS, BRAF, PI3KCA and others in frequencies exceeding 1% (7–15). These findings raised the possibility that other mutation specific TKIs could produce similar benefits in NSCLC patients whose tumors harbor these specific “driver” mutations. BRAF is one of the genes identified in these early reports with studies indicating the presence of BRAF mutations in 0% to 17% of tumors and cell lines. In these reports, 20 of 805 NSCLC tumors harbored BRAF mutations (2.5%) and many of these mutations were not the typical V600E mutations observed in malignant melanoma (8–15). Lung cancer cell lines had even higher rates of BRAF mutations with rates of 2% to 17% (14, 15). Preclinical model studies demonstrated that activating BRAF mutations were sufficient for the development of lung adenocarcinomas in mice (16, 17).

V600E BRAF mutations in malignant melanoma were shown to be present in an even higher percentage of cases (50–70%) (13, 18). Several studies reported that melanomas harboring BRAF mutations are dependent on MEK/ERK signaling (17, 19, 20). PLX4032, a potent inhibitor of oncogenic BRAF, was developed through structure guided discovery and selectively blocked RAF/MEK/ERK signaling in BRAF mutant cells and caused regression of BRAF mutant tumor xenografts (21). In a phase 1 study, PLX4032 inhibited ERK phosphorylation and produced an 81% response rate in melanoma patients with BRAF mutant tumors (22). These findings led to an extension phase of the PLX4032 study in 32 melanoma patients whose tumors had V600E BRAF mutations. There were 24 objective partial responses and 2 complete responses and the median PFS exceeded 7 months (23). Subsequent studies showed that acquired resistance to PLX 4032 developed by mutually exclusive PDGFR beta up-regulation and NRAS mutation and not through secondary mutations in BRAF (24).

In this issue of the JCO, Paik et al. report finding BRAF mutations in 18 of 673 (3%) adenocarcinomas of the lung (25). Although the frequency of BRAF mutations in NSCLC is lower than in melanoma, it is high enough to warrant evaluation BRAF specific inhibitors in this population. The 3% figure is in keeping with the 2.5% observed in other studies and the finding that half of the mutations were non-V600E is also consistent with previous reports. As the authors point out, a 3% rate could mean as many as 5600 new cases annually in the US and 35,000 annually worldwide. However, it may be imperative to distinguish between the V600E and non-V600E mutations. If there are differences in therapeutic response to current BRAF inhibitors between V600E and non-V600E, only about 1.5% of all NSCLCs may have each of these classes of BRAF mutation. Cell lines such as NCIH1755 that has a non-V600E, G469A BRAF mutation has been reported to be resistant to PLX4032 (21). Cells with non-V600E BRAF mutations may also be resistant to other V600E BRAF inhibitors such as GSK2118436. However, tumor cells with other BRAF mutations are sensitive to downstream pathways inhibitors such as MEK inhibitors (20). Thus, it is likely that it will be critical to determine the exact type of BRAF mutation.

Paik et al. also report association between clinical characteristics, BRAF mutation, and survival outcome. Unlike other oncogenic mutations reported to date, they found BRAF mutations occurred most often in former or current smokers compared to never smokers. It should be noted that previous studies have identified BRAF mutations in non-smokers (26, 27). Although EGFR activating mutations were initially identified in a subset of patients that shared clinical and histologic characteristics, a recent comprehensive study has shown that approximately 39% of all EGFR mutations occur in current and former smokers (28); thus basing clinical testing on unreliable clinical characteristics can deny patients the benefit of targeted therapy. The authors did not find a sex-specific association with BRAF mutations. One major limitation of this study is that only adenocarcinomas were evaluated for BRAF mutations. Previous studies have shown the presence of both exon 11 and exon 15 BRAF mutations in squamous cell carcinoma of the lung (26). There is poor concordance on histologic diagnosis, even among expert pathologists (29). Therefore, testing based on histology cannot be recommended. The survival outcome seems to be slightly superior compared to patients without such driver mutations and similar to the survival of patients with EGFR mutations but the number of patients was quite small and there was no uniformity of treatment.

One of the major implications of the association with clinical features is to determine a priority for molecular testing. Many argue that it may be most cost effective to test only the patients most likely to have a specific mutation. However, if one conducts serial analysis of each mutation individually, it may take a long time to rule out the presence of any of the mutations. Enrichment strategies may make sense when trying to identify patients for clinical trials with limited resources to screen large numbers of patients (30); however, newer technologies such as SNapshot and Sequenom allow testing of all of these mutations at the same time and with little additional expense compared to testing single mutations. Furthermore, there are reports of NSCLC patients with more than one oncogenic driver mutations, (e.g., EML4-ALK and EGFR or KRAS mutations) (31, 32). Patients with BRAF (or KRAS) mutations can also have inactivating mutations in the tumor suppressor gene, LKB1, and this may influence clinical benefit from BRAF TKIs (33). As more driver mutations are identified, we can expect the complexity of overlapping mutations to increase. This multiplexed approach is clearly practicable as the NCI-funded Lung Cancer mutation consortium is currently conducting simultaneous genetic testing for 10 oncogenic changes for which there are specific therapies. For example, there is a trial of GSK 2118436 for patients with BRAF V600E mutations and a trial of the MEK inhibitor GSK1120212 in patients with non-V600E BRAF mutations.

This study of mutation testing in NSCLC patients and others measuring other genetic changes, signal a new era of personalized medicine for patients with advanced lung cancer where is will be imperative to match the specific mutations of a patients tumor with a specific therapy. One challenge will be obtaining adequate tumor material at the time of diagnostic biopsies to ensure successful testing. Another challenge will be developing testing platforms that can analyze for the presence of somatic mutations, gene fusions, or other genetic changes simultaneously. It is likely that routine simultaneous (reflex) testing of multiple markers will be conducted on all patients prior to initiation of therapy, irrespective of their clinical features.