Erlotinib

Clinical Relevance of EGFR- or KRAS-mutated Subclones in Patients With Advanced Nonesmall-cell Lung Cancer Receiving Erlotinib in a French Prospective Cohort (IFCT ERMETIC2 Cohort – Part 2)

Detecting driver mutations belongs now to the best practices in advanced/metastatic nonesmall-cell lung cancer. New molecular techniques are highly sensitive. In nonesmall-cell lung cancer treated with erlo- tinib (n [ 228), we report that EGFR- and KRAS-mutated subclones had a prognostic value, but not minor KRAS-mutated subclones. Molecular techniques must be sensitive but not under 1% of mutated tumor cells.
Introduction: Evaluation of EGFR Mutation status for the administration of EGFR-TKIs in non-small cell lung Carci- noma (ERMETIC) was a prospective study designed to validate the prognostic value of EGFR/KRAS mutations in patients with advanced nonesmall-cell lung cancer (NSCLC), all receiving a first-generation tyrosine kinase in- hibitor, erlotinib. ERMETIC2 was an ancillary project evaluating the clinical value of common EGFR/KRAS- mutated subclones regarding prognosis using highly sensitive molecular detection methods. Materials and Methods: Tumor samples from 228 patients with NSCLC (59% adenocarcinoma, 37% women, and 19% never/ former smokers) were available for reanalysis using alternative highly sensitive molecular techniques. A multi- variate Cox model was used for prognostic analysis. Results: Using alternative highly sensitive techniques, 16 EGFR and 51 KRAS supplementary mutations were newly identified, all still exclusive, leading to an overall rate of 12.3% (n ¼ 28) and 33.3% (n ¼ 76), respectively. Using real-time polymerase chain reaction (hybridization probe),
they were significantly associated with progression-free survival (P ¼ .02) and overall survival (OS) (P ¼ .01), which were
better for EGFR-mutated patients for progression-free survival (hazard ratio [HR], 0.46; 95% confidence interval [CI], 0.28-0.78) and OS (HR, 0.56; 95% CI, 0.31-1), and worse for KRAS mutations and OS (HR, 1.63; 95% CI, 1.09-2.44). Using the most sensitive technique detection for KRASeclamp polymerase chain reactioneKRAS mutated subclones did not impact OS. Conclusions: KRAS and EGFR mutations were detected in higher.

Introduction
Epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs), erlotinib, gefitinib, and afatinib, are authorized worldwide as first-line treatment for patients with advanced or metastatic nonesmall-cell lung cancer (NSCLC) with EGFR-acti- vating mutations in their tumor.1-3 Recently, the third-generation EGFR-TKI, osimertinib, has been validated as the standard of care for patients with T790M-positive NSCLC in whom disease had progressed during first-line EGFR-TKI therapy.4 And more recently, osimertinib showed efficacy superior to that of standard first-/second-generation EGFR-TKIs in the first-line treatment of EGFR mutation-positive advanced NSCLC.5
Molecular tumor testing is actually mandatory for selecting first- line treatment in patients with advanced or metastatic NSCLC.6-9 Yet no EGFR mutation assay is currently specifically recommended by the United States Food and Drug Administration, the European Agency for the Evaluation of Medicinal Products, or the European Society of Medical Oncology to inform treatment decisions.10,11 Direct sequencing has for many years been considered the gold standard for testing, yet its sensitivity can limit its use to routinely somatic tumor testing, and alternative more sensitive molecular methods, targeted or not, have often since replaced this approach.12 Furthermore, next-generation sequencing (NGS) or droplet digital polymerase chain reaction (ddPCR) are developed in molecular platforms. All these highly sensitive molecular methods should be ableto detect mutated subclones (5% of mutated cells) to minor subclones (< 1% of mutated cells), thus raising the question of the prognostic/ predictive value of such mutated different subclones. Although KRAS is not a target for therapy, lung cancer molecular analyses often test for EGFR as well as KRAS mutations. In the Caucasian NSCLC population, EGFR and KRAS mutations are found in 11% and 29% of the patients, respectively.3,7 These mutations are often mutually exclusive, hence why some authors have even proposed a step-by-step algorithm using KRAS mutation testing as the first step to rule out the presence of an EGFR mutation, in case of using targeted alternative molecular methods. Furthermore, KRAS mutations appear to affect an heterogeneous population with different prognostic/ predictive values depending on the type of nucleotide base substitution, regarding EGFR-TKI treatment.13,14 Finally, several studies have distinguished patients with EGFR mutations from those with non- EGFR mutated tumors, yet including KRAS-mutated tumors. In our study, we evaluated what impact the mutation detection threshold has on the prognostic value of erlotinib efficacy.The ERMETIC (Evaluation of EGFR Mutation status for the administration of EGFR-TKIs in non-small cell lung Carcinoma) study, designed and supported by the French Collaborative Thoracic Cancer Intergroup, funded by the French National Cancer Institute, reported that formaldehyde-fixed paraffin-embedded specimens may be suitable sources for DNA analysis by means of genomic Sanger sequencing, providing rigorous preanalytical quality control standards are respected.9,12 Since that publication, ERMETIC centers have switched to alternative molecular methods. ERMETIC2 consisted of a nationwide technological evaluation of this new EGFR/KRAS testing using NSCLC cell line DNA with various allele proportions. We demonstrated that the best threshold of mutation detection was obtained using allele-specific amplification- based technologies, with cutoff values of 5% and 1% for clamped PCR with peptide nucleic acid (PNA).15 We report the prognosis- based clinical impact of this new strategy on 228 ERMETIC pa- tients with available tumor samples reanalyzed using these techniques.The ERMETIC prospective observational study included 522 patients with advanced NSCLC, either newly treated with erlotinib or before erlotinib administration.6 A preliminary study revealed that EGFR and KRAS mutations identified using Sanger direct sequencing were independent markers of outcome in this population.6 A sub- group of 228 patients provided sufficient samples for a second round of common EGFR and KRAS mutation screening using the alterna- tive molecular methods selected after ERMETIC2 e part 1.12The methods used were previously described.12 Briefly, we used fragment analysis for EGFR exon 19 assessment, and targeted mo- lecular techniques based on allele-specific amplification: probe-specific detection, TaqMan assay for EGFR exon 21 L858R mutation, and hybridization probe (SH assay), without PNA (KRAS_SH), or with PNA as clamp-PCR strategy (KRAS_PNA) for KRAS mutations. Endpoint definitions were as previously defined.6 Survival rates were estimated using the Kaplan-Meier method with 95% confidence intervals (CIs). Impact on survival was quantified using Cox models and hazard ratios with 95% CIs. Variables with a P- value < .20 in univariate analysis were included into the multi- variate analysis. A backwards selection process was undertaken with the final model, including all variables with P-values < .05. Results The population consisted of patients with samples available for both EGFR and KRAS mutation analyses, excluding patients with insufficient material for simultaneous analysis of both genes or with non-amplifiable samples. Clinical characteristics of the patients (Table 1) did not differ from those with sample not available for this second part of the study (data not shown).When tumors (n ¼ 228) were tested by Sanger direct sequencing, the overall mutation rate was 16.2% (12 [5.3%] and 25 [11.0%] EGFR and KRAS mutations, respectively). When the same tumors (n ¼ 228) were tested by alternative molecular methods, the overall mutation rate was 45.6% (28 [12.2%] and 76 [33.3%] EGFR and KRAS mutations, respectively), with 67 new mutations found (16 in EGFR and 51 in KRAS ), including 60.8% (31/51) for KRAS mutations identified using the clamp-PCR strategy (Tables 2A and 2B). Details of EGFR and KRAS mutations by techniques are described in Table 3. No differences in OS or PFS were observed between the initial and reanalyzed ERMETIC populations (data not shown). Patients with KRAS-mutated tumors were categorized according to the mutation detection method, SH assay with or without clamp-PCR strategy (KRAS_SH vs. KRAS_PNA). The median OS was 15 months (95% CI, 4.7-28.4 months), 6.7 months (95% CI, 2.1-9.2 months), 5.3 months (95% CI, 3.9-8.1 months), and 2.7 months (95% CI, 2.1-9.2 months) for EGFR-mutated tumors, KRAS_PNA- mutated tumors, EGFR/KRAS wild-type (WT) tumors, and KRAS_SH-mutated tumors, respectively (P ¼ .0018) (Figure 1A). The median PFS was 9.3 months (95% CI, 2.6-15.3 months), 2.8 months (95% CI, 1.5-3.1 months), 2.3 months (95% CI, 2.0-2.6 months), and 1.6 months (95% CI, 0.9-2.5 months) for patients with EGFR mutated, KRAS_PNA-mutated, WT, and KRAS_SH- mutated tumors, respectively (P ¼ .0007) (Figure 1B). The 1-year survival rates were 60.7% (95% CI, 42.4%-76.4%), 25.8% (95% CI, 18.9%-34.2%), 23.5% (95% CI, 12.4%-40%), and 16.7% (95% CI, 8.3%-30.6%) for these 4 groups, respectively. The clin- ical characteristics significantly associated with prognosis (OS, PFS) were the same as for the initial population (Table 4 and data not shown). After adjusting for clinical factors, multivariate analysis of mutation status remained significantly associated with OS (P ¼ .01) and PFS (P ¼ .02), which were better for EGFR-mutated patients for PFS (hazard ratio [HR], 0.46; 95% CI, 0.28-0.78) and OS (HR, 0.56; 95% CI, 0.31-1), and worse for KRAS mutations for OS (HR, 1.63; 95% CI, 1.09-2.44) (Table 4). Using the most sensitive technique detection for KRASeclamp PCReKRAS mutated sub- clones did not impact OS. EGFR mutation significantly decreased the risk or death by 44%, and the risk of progression or death by 54% in patients treated with erlotinib. KRAS mutations detected by SH (KRAS_SH) significantly increased the risk of death, by 63%.Conversely, KRAS mutations detected by clamp-PCR strategy (KRAS_PNA) did not increase the risk of death.No prognostic value was related to the alteration type (transition/ transversion) or mutation location (codon 12 or 13) among the 76 KRAS-mutated patients. Discussion The prognostic or predictive value afforded by driver-mutated subclones and minor sub-clones in NSCLC and other cancer types is still open to debate. With the development of high throughput and extremely sensitive methods, such as NGS, clamp- based PCR, or ddPCR, establishing a cutoff is now mandatory. The clinical value of low allele frequency detection needed to be assessed regarding 2 issues: Can it rescue mutation testing for small biopsies with low tumor-cell content and high stromal component? Does it have any clinical value? This study was designed to reanalyze paraffin-embedded NSCLC tumor samples using alternative molecular techniques currently employed in France and many laboratories worldwide thanks to their cost-efficiency for analysis of recurrent genetic alterations, requiring low amounts of DNA from formaldehyde-fixed paraffin-embedded samples. We described 16 and 51 new EGFR and KRAS mutations, respectively, after reanalysis. In the ERMETIC initial population, EGFR status impacted both PFS and OS, whereas KRAS status only impacted OS.9,12 Similar results were observed for EGFR mutation in the reanalyzing study. In contrast, though, KRAS status analyzed by PCR using hybridization probes remained predictive of OS, the more sensitive clamped PCR method that identified low mutated subclones failed to impact prognosis, raising the question of these minor sub- clones clinical relevance for patient care. The number of newly-detected EGFR mutations in our study proved relatively small, suggesting that EGFR mutations are present in the majority of tumor cells or associated with an amplification of the mutated allele in NSCLC, as previously described.16,17 We demonstrated that, by using sensitive methods, we may be able to reattempt detection of an EGFR alteration, an important capability for treatment decisions. All our patients receiving EGFR-TKIs were correctly treated, although the mutation was not identified in the initial ERMETIC study; hence, why no modification of EGFR prognostic value was observed between initial and reanalyzing ERMETIC studies. In the latest study, testing was performed using methods with a 10% to 5% detection threshold for fragment analysis of EGFR exon 19 deletions and for PCR using TaqMan probes of EGFR L858R mutations.15 Therefore, we deduce that the new EGFR-mutated cases were related to the tumor cell content being low, initially under the level of detecting EGFR mutation by direct sequencing. The unpredictable variability in EGFR copy number and therefore in EGFR WT/mutant allelic ratio justifies using sensitive methods to identify patients with EGFR-mutated tumors.The situation is probably more complex for KRAS mutations in the setting of NSCLC,7,13 with KRAS-mutated subclones previously described in NSCLC.18 All the 25 KRAS mutations detected by direct sequencing in the initial population were detected by alternative molecular techniques in the reanalyzing population (internal positive controls). Among the remaining cases, 45 (19.7%) were positive for KRAS mutations using PCR with hybridization probes, and 76 (33.3%) were detected using more sensitive clamped PCR (no cases with concomitant EGFR mutation). However, this increased sensi- tivity did not detect any minor subclones as having prognostic impact. Our results suggest that patients with KRAS-mutated sub- clones, using 1% as a cutoff (clamp-PCR), behave as with WT mutations. Such absence of clinical relevance of KRAS-mutated minor subclones was previously described in advanced colorectal cancer treated with anti-EGFR therapy.19 In NSCLC, it appears that such sensitive methods are not necessarily useful, achieving approximately 1% detection rate for KRAS mutations. Conclusion Highly sensitive molecular methods increased the number of EGFR and KRAS mutations in NSCLC tumors. For common EGFR mutations, this increase is lower and correlated with classical prognostic values (OS, PFS) in first-line EGFR-TKI-treated patients with NSCLC. For KRAS mutation, detection of mutated subclones(5%) is associated with survival (OS) but not the minor subclones (< 1%). Our study demonstrated that if more sensitive techniques could detect new mutated cases, it is not necessary to have a too low cutoff for such analysis. Threshold cutoff for mutation analysis must be taken into account for new Erlotinib molecular techniques such as NGS or ddPCR.