Prognostic and Predictive value of EGFR Tyrosine Kinase Domain Mutation Status and Gene Copy Number for Adjuvant Chemotherapy in Non-Small Cell Lung Cancer

Abstract

Purpose

Non-small cell lung carcinoma (NSCLC) patients with EGFR mutations may have a more favorable prognosis and greater response to chemotherapy. The effect of EGFR mutation and gene copy on early stage NSCLC patients receiving adjuvant chemotherapy has not been reported.

Patients and Methods

Tumor samples from NCIC CTG JBR.10, an adjuvant trial of vinorelbine/cisplatin (ACT) versus observation (OBS), were analysed for EGFR mutation by multiple sensitive methods and copy number by fluorescent in-situ hybridization (FISH). Their prognostic and predictive roles were explored in correlation with survival.

Results

Mutation results were available in 221 OBS, 215 ACT and FISH results in 159 OBS, 163 ACT patients. Mutations were identified in 43 (27 OBS, 16 ACT) patients (36 sensitizing exon-19 deletions or L858R mutations). Compared to wild type (WT), sensitizing mutations were not significantly prognostic in OBS patients (HR 0.79, 95%CI 0.38-1.63, p=0.53). Although the presence of sensitizing mutations resulted in relatively greater benefit in ACT patients (HR 0.44, 95%CI 0.11-1.70, p=0.22) compared to WT patients (HR 0.78, 95%CI 0.58-1.06, p=0.12), this quantitative difference was not significant (interaction p=0.50). Similarly, high EGFR copy was neither significantly prognostic, nor predictive, although quantitatively it was associated with greater benefit from ACT.

Conclusions

Trends towards longer survival and a greater benefit from chemotherapy were observed in patients with exon 19/21 mutations and high EGFR copy although the differences were not statistically significant. The interpretation of the results was limited by the low EGFR mutation rate in this study of mainly Caucasian patients.

INTRODUCTION

Early stage non small cell lung carcinoma (NSCLC) patients are treated surgically with curative intent; however 30-65% develop recurrence and die of their disease.¹ Recent trials using cisplatin-based doublet chemotherapy as well as meta-analyses have demonstrated significant survival benefits with post-operative chemotherapy^2-5. JBR.10, a North American inter-group study led by the NCIC Clinical Trials Group (CTG), randomized patients with completely resected stage IB-II NSCLC to receive adjuvant chemotherapy with cisplatin/vinorelbine or observation alone. Chemotherapy treated patients derived significant survival benefit (Hazard Ratio [HR] 0.70, p=0.03).^5,6 It is recognized that certain clinical and pathological factors such as stage, age, sex, tumour histology and differentiation grade are prognostic of outcome in NSCLC.⁷ At the molecular level, although a large number of markers have been investigated for their prognostic value,⁸ few have been reported to predict a differential effect of adjuvant chemotherapy on survival.⁹ Currently, apart from stage, neither prognostic nor predictive markers are used routinely to select NSCLC patients for adjuvant chemotherapy.

Greater than 60% of NSCLC tumors express epidermal growth factor receptor (EGFR) protein.¹⁰ The discovery that EGFR tyrosine kinase domain mutations are strongly associated with greater sensitivity of NSCLC to EGFR tyrosine kinase inhibitors (TKIs), and that mutations are more common among patient subgroups who demonstrate response to these drugs, suggests that this or other molecular markers may be used to determine patient subgroups most likely to benefit from EGFR TKI therapy.^11,12 Recent studies have shown that patients whose tumors contain EGFR activating mutations on exons 19 and 21 also demonstrate higher response rates to chemotherapy.^13,14 EGFR gene copy gain also has been shown to be a predictive marker for response and survival benefit with EGFR TKI therapy,^15,16 but has not been shown to be associated with a differential response to chemotherapy.^14-18 We report here our analyses of EGFR TK domain (TKD) mutation and gene copy number status in patients from the JBR.10 trial, and correlate these markers with prognosis in the untreated control arm as well as survival benefit from chemotherapy.

METHODS

Patients and tissue

This molecular correlative study was approved by the Research Ethics Board of the University Health Network. JBR.10 compared the effect of adjuvant vinorelbine/cisplatin to observation alone in 482 patients with completely resected T2N0, T1-2N1 NSCLC.⁵ Patients provided written consent for the study and for RAS mutational analyses (RAS mutation status was a stratification parameter) and 445 patients consented to tumour banking for future studies (Supplementary Figure 1S). These included snap-frozen tissue with or without formalin-fixed, paraffin embedded (FFPE) blocks from 171 patients, FFPE blocks only from 161 patients, and 10 unstained slides only from 119 patients. Tissue microarrays (TMA) were constructed from 332 cases with FFPE blocks.

Analysis for EGFR tyrosine kinase domain mutations

The methods for isolation of genomic DNA and analysis of mutations on EGFR exons 19 and 21 have been reported previously.¹⁵ For samples that were estimated by histology to have <50% tumor cellularity, we performed macro-dissection of unstained sections to enrich for tumor DNA. DNA samples were analyzed first by direct sequencing and positive results were confirmed by repeat sequencing of an independent PCR product. Negative cases were screened secondarily using the higher sensitivity fragment length analysis method to detect exon 19 deletion and L858R mutations.¹⁹ Newly identified deletion/mutations were confirmed by the Scorpion Amplified Refractory Mutation System (DxS, Manchester, UK). Samples were classified as failed or indeterminate if repeated analyses failed or were unable to confirm mutations on exons 19 or 21.¹⁵

EGFR gene copy analysis

EGFR gene copy number was evaluated by fluourescent in-situ hybridization (FISH), as described previously using the Vysis LSI EGFR probe labelled with Spectrum Orange, and the CEP7 chromosome 7 centromere (7p11.1 through q11.1) probe labelled with Spectrum Green.¹⁶ This was conducted using tissue microarrays (TMA) constructed from 332 cases with tumor FFPE blocks. In 18 cases for which TMA cores showed heterogeneity with the presence of both high and low copy number cores, we repeated the analysis using full sections. To validate TMA results further, full section analyses were also conducted on 24 randomly chosen cases with EGFR amplification and 22 cases with low copy number (trisomy and low polysomy). Six FISH categories were used to define EGFR FISH status.¹⁵ Tumours with high EGFR gene copy (“High EGFR copy”) included high polysomy or amplification and all other categories were classified as “Low EGFR copy”.

Statistical analyses

Based on a pre-specified statistical analysis plan, exploratory analyses were performed to characterize relationships between marker levels and baseline clinical characteristics and survival.⁴ Chi-square or Fisher’s exact test was used to test association between marker levels and baseline factors; Kaplan-Meier product-limit method and the log-rank test were used to estimate and test overall survival distributions and their difference between marker levels and treatment arms, and multivariate Cox regression models were used to verify the prognostic and predictive effects of markers on survival while adjusting for baseline factors and potential prognostic factors including sex, age, performance status, stage, histology, smoking status, baseline anemia, type of resection, serum lactate dehydrogenase, p53 mutation status and p53 immunohistochemistry. All reported p-values are two-sided, and a level of 5% (p=0.05) was considered statistically significant. To be consistent with our other mutational analysis reports, survival analyses are based on the original survival analysis of the trial.

RESULTS

EGFR tyrosine kinase domain mutation

Altogether, 445 patients who consented to future molecular studies had DNA available for mutation analysis. Among these, failed or indeterminate results were obtained in only nine patients. Baseline stratification and potential prognostic factors for 436 patients with and 46 patients without EGFR mutation results are shown in Supplemental Table 1S. More patients with mutation data were ever-smokers (94.5% vs 84.8%, p=0.027); there were no other significant differences between the populations. In patients with mutation results, the survival benefit from chemotherapy was similar to that of the overall study (HR 0.77, 95% CI 0.57-1.03, p=0.08).

Forty-five mutations were found in samples from 43 patients (9.9%), with two tumours having two separate mutations (Del E746_750, I744V; G735S, L858R). Female patients were more likely to have mutations (13.4% vs 8.0%, p=0.07), as were lifetime never-smokers (40% vs 8.3%, p<0.001) and patients with adenocarcinoma (13.9% vs 4.9% for squamous and 7.0% for other histologies, p=0.01) (Table 1). Each case of squamous cell carcinoma with mutation was reviewed histologically and the diagnosis was confirmed for all. Among non-white ethnicity patients, there were 10 native Hawaiian/Pacific islanders, who demonstrated a mutation rate (40%) comparable to rates reported among East Asian patients (Table 2).²⁰

Table 1.

Baseline clinical and pathological demographics according to EGFR gene status

Variable	Demographic	EGFR wild type (n = 393)	EGFR mutant (n = 43)	P-value	EGFR low gene copy (n=122)	EGFR high gene copy (n=200)	P-value
Sex	Male	264 (92.0%)	23 (8.0%)	0.072	94 (44.5%)	117 (55.5%)	0.001
	Female	129 (86.6)	20 (13.4%)		28 (25.2%)	83 (74.8%)
Histology	Squamous	154 (95.1%)	8 (4.9%)		43 (36.4%)	75 (63.6%)
	Adenocarcinoma	199 (86.1%)	32 (13.9%)	0.010	70 (40.5%)	103 (59.5%)	0.444
	Other	40 (93.0%)	3 (7.0%)		9 (29.0%)	22 (71.0%)
Smoking	Never smoked	12 (60.0%)	8 (40.0%)		2 (15.4%)	11 (84.6%)
	Ever smoked	378 (91.7%)	34 (8.3%)	<0.001	119 (38.8%)	188 (61.2%)	0.142
	Missing	3 (75.0%)	1 (25.0%)		1 (50.0%)	1 (50.0%)
Ethnicity	White	175 (89.7%)	20 (10.3%)		43 (35.2%)	79 (64.8%)
	Others**	21 (77.8%)	6 (22.2%)	0.102	1 (8.3%)	11 (91.7%)	0.103
	Unknown	197 (92.1%)	17 (7.9%)		78 (41.5%)	110 (58.5%)

^*p-value of Chi-square test without missing values;

^**Four of 10 pacific islander patients had EGFR mutations were

Table 2.

EGFR mutation rates among 27 patients of different non-white ethnicity

	Mutant	Wild-type	Total
Hispanic/Latino	0	1 (100%)	1
African American	2 (17%)	10 (83%)	12
Native Hawaiian/Pacific Islander	4 (40%)	6 (60%)	10
American/Alaska Native	0	3 (100%)	3
Other	0	1 (100%)	1

When considering all EGFR mutations, 27 patients in the observation arm with tumours that had mutations (Figure 1A) showed numerically (but not-significantly) longer overall (adjusted HR 0.68, 95%CI 0.34-1.36, p=0.28) and relapse-free (adjusted HR 0.83, 95%CI 0.44-1.56, p=0.56) survivals compared to wild-type patients (Table 3). Among these 27 patients, 22 had exon 19 deletions or L858R mutations that are known to confer sensitivity (sensitizing mutations) to both EGFR TKI therapy and chemotherapy (Supplemental Table 2S). The adjusted survivals of these 22 patients with sensitizing mutations also were longer but not significantly different compared to wild type patients (overall survival HR 0.73, 95%CI 0.35-1.52, p=0.41) (Table 3, Figure 1B).

Figure 1.

Impact of EGFR mutations on overall survival.

Table 3.

Impact of EGFR status on the survival of patients in the observation arm by multivariate analysis¹

	Relapse Free Survival			Overall Survival
	Hazard Ratio	95% Confidence Interval	P value	Hazard Ratio	95% Confidence Interval	P value
EGFR tyrosine kinase domain mutation status
Wild-type (n=194)	1.0	-	-	1.0	-	-
All mutations (n=27)	0.83	0.44-1.56	0.56	0.68	0.34-1.36	0.28
Exon-19 deletions + L858R mutation (n=22)	0.92	0.48–1.76	0.79	0.73	0.35-1.52	0.41
EGFR gene copy number
Low copy (n=60)	1.0	-	-	1.0	-	-
High copy (n=99)2	0.94	0.60-1.48	0.79	0.74	0.46-1.19	0.22
Amplification (n=29)3	1.24	0.69-1.23	0.48	1.10	0.58-2.05	0.78

¹Multivariate adjustment includes other potential prognostic factors including sex, age, performance status, stage, histology, smoking status, baseline anemia, resection, serum LDH, p53 mutation status and immunohistochemistry.

²High copy includes high polysomy and amplification; survival compared to low polysomy or less copy number status

³Survival compared to all patients without amplification

Both patients with wild-type and mutated EGFR (Table 4, Figure 1C) had survival benefit from adjuvant chemotherapy (wild-type HR 0.76, 95%CI 0.56-1.04, p=0.08; mutants HR 0.65, 95%CI 0.20-2.14, p=0.48). The patients with sensitizing mutations appeared to derive even greater (although statistically non-significant) survival benefit from adjuvant chemotherapy (HR 0.44, 95%CI 0.11-1.70, p=0.22, interaction p=0.50) (Table 4, Figure 1D). However, multivariate Cox regression analysis failed to detect significant interaction between chemotherapy with all-type mutations or sensitizing mutations, for either overall survival (p=0.99 and p=0.53) or relapse-free survival (p=0.43 and p=1.00).

Table 4.

Survival benefit from adjuvant chemotherapy according to EGFR gene status

	Relapse-Free Survival				Overall Survival
	Hazard Ratio3	95% Confidence Interval	P value	P Inter-action	Hazard Ratio3	95% Confidence Interval	P value	P Inter-action
EGFR tyrosine kinase domain mutation status
Wild-type (n=393)	0.65	0.48 - 0.89	0.006		0.76	0.56 - 1.04	0.08
All mutations (n=43)	0.78	0.30 - 2.00	0.60	0.62	0.65	0.20 - 2.14	0.48	0.86
Exon-19 deletions + L858R mutation (n=36)1	0.58	0.21 – 1.60	0.29	0.98	0.44	0.11 - 1.70	0.22	0.50
EGFR gene copy number
Low copy (n=122)	0.73	0.44 – 1.22	0.23		0.96	0.58 – 1.60	0.88
High copy (n=200)2	0.58	0.37 – 0.90	0.015	0.51	0.73	0.47 – 1.14	0.17	0.47
Amplification (n=67)3	0.45	0.21 – 0.97	0.036	0.26	0.67	0.32 – 1.41	0.29	0.32

¹Sensitizing mutations included only exon-19 deletions and exon-21 L858R mutation; Interaction with chemotherapy was between these EGFR sensitizing mutant with wild type patients

²High copy includes high polysomy and amplification, as defined by the University of Colorado EGFR scoring system

³Interaction with chemotherapy was between patients with EGFR gene amplification and wild type patients

EGFR Copy number

FISH results were obtained in only 322 patients for whom we had paraffin blocks to construct TMAs. The baseline stratification and potential prognostic factors are shown for patients with and without EGFR FISH results in Supplemental Table 3S. More patients with FISH data were stage IB (51.6% vs 33.1%, p<0.001), over age 65 (35.4% vs 25.6%, p=0.03) and ever-smokers (95.3% vs 90.0%, p=0.06); there were no other significant differences between the populations. There was a non-significant difference (interaction p=0.14) in survival benefit from chemotherapy between patients with FISH results (HR 0.83, 95%CI 0.59-1.16, p=0.28), and patients without FISH data (HR 0.51, 95%CI 0.530-0.86, p=0.01). Female patients were more likely to have high EGFR copy (74.8% vs 55.5%, p=0.001), but there were no significant differences in histological subtype, smoking history or ethnicity between the groups (Table 1).

To verify the validity of FISH results obtained from TMA analysis, we conducted independent FISH analyses on whole tumour sections of 24 cases with EGFR amplification in TMA cores and 22 cases with EGFR low copy. Discrepancy was found in only 1 of 24 (4.2%) amplified and in 1 of 22 (4.5%) low copy cases (Supplemental Table 4S). In 18 cases with TMA showing heterogeneous cores, some of which had high polysomy and/or amplification, full section analysis resulted in only 2 cases that were subsequently scored as low copy number tumours (Supplemental Table 4S).

Among 159 observation patients (Figure 2, A and B; Table 3), there was no significant difference in overall survival between EGFR high (amplification + high polysomy) and low copy patients (HR 0.74, 95%CI 0.46-1.19, p=0.22). EGFR amplification was also not prognostic (HR 1.10, 95%CI 0.58-2.05, p=0.78).

Figure 2.

Correlation of overall survival outcome to EGFR gene copy number changes. According to the University of Colorado Cancer Center’s scoring criteria (21, 28), low EGFR gene copy number includes four categories (disomy, low trisomy, high trisomy, low polysomy) that are less than high polysomy (<high polysomy), while high EGFR gene copy includes high polysomy and amplification (≥ high polysomy).

In 122 EGFR low copy patients, 60 were on observation while 62 received chemotherapy; there was no apparent overall survival benefit from adjuvant chemotherapy (HR 0.96, 95% CI 0.58-1.60, p=0.88) (Figure 2C, Table 4). In contrast, EGFR high copy patients treated by adjuvant chemotherapy (n=101) had longer survival compared to observation patients (n=99), although the difference was not significant (HR 0.73, 95%CI 0.47-1.44, p=0.17; interaction p=0.47) (Figure 2D, Table 4). Similar survival benefit from chemotherapy was noted in patients with EGFR high copy number but wild-type gene (Figure 2E, HR 0.77, 95%CI 0.48-1.23, p=0.27) and in patients with EGFR amplification only (Figure 2F, HR 0.67, 95%CI 0.32-1.41, p=0.29). The trend toward greater survival benefit from chemotherapy in EGFR high copy patients was stronger for relapse-free survival (Table 4, HR 0.58, 95%CI 0.37-0.90, p=0.015), especially among patients with EGFR amplification (HR 0.45, 95%CI 0.21-0.97, p=0.036), but the interaction p-value remained insignificant (p=0.51 and 0.32, respectively). In multivariate analysis, EGFR copy number was neither a prognostic nor a predictive factor.

DISCUSSION

In this study, we evaluated the prognostic significance of EGFR TKD mutations and EGFR gene copy gains, as well as their potential impact on outcomes after adjuvant vinorelbine/cisplatin in patients from JBR.10, a pivotal randomized study that included prospective collection of tumor samples. Our results support previous reports that untreated NSCLC patients with EGFR mutations have longer survival (HR 0.68, p=0.28) compared to patients with wild-type tumours in adjusted analysis, confirming the prognostic effect identified in patients with advanced disease. In contrast, this effect was not seen for EGFR amplification (overall survival HR 1.07, p=0.80). Exon 19 deletions or exon 21 L858R sensitizing mutations were associated with a trend towards greater benefit from ACT compared to EGFR wild-type patients, although the difference was not significant (HR 0.44 vs 0.76, interaction p=0.5). Similar, but also non-significant effects (interaction p=0.47) were demonstrated for high EGFR gene copy (overall survival HR 0.73, p=0.17), compared to low copy (HR 0.96, p=0.88). Although numbers were small, we also provide the first evidence that Hawaiian NSCLC patients have an EGFR mutation rate comparable to that found in East-Asian patients.

It is only six years since the first reports showing that patients whose tumors contained activating mutations in the TK domain of the EGF receptor, had higher response rates to EGFR TKIs.^11,12 This was confirmed subsequently in larger retrospective studies from randomized trials of both erlotinib and gefitinib.^13-15 However, the randomized trials of single agent EGFR TKI compared to a placebo control arm did not confirm a significant differentially greater overall survival benefit from erlotinib, despite the lower HR for patients with mutations.¹⁵ The lack of significant interaction may have been due to the small number of patients with mutations in each trial. It has also been postulated that the lack of differential benefit was due to the untreated patients with mutations having an inherently better prognosis than those with wild-type tumours.²⁰ The evidence for this was observed first in the TRIBUTE (Tarceva response in conjunction with paclitaxel) trial of chemotherapy with or without erlotinib.²¹ Regardless of therapy received, patients with EGFR mutations in TRIBUTE had significantly better survival compared to those with wild-type tumours. Several subsequent studies involving resected NSCLC or lung adenocarcinoma patients, also have reported better prognosis for patients with EGFR mutations although not all studies have shown significant differences in prognosis in multivariate analysis when other important variables have been included in the models.^22-25 Despite some imbalances in the patient populations studied, the results of the present study, are supportive of the hypothesis that the presence of an EGFR TKD sensitizing mutation confers better overall survival for all stages of NSCLC.

Aside from the prognostic importance of EGFR TK domain mutations, this study also shows that patients with exon 19 deletions or L858R EGFR sensitizing mutations appear to derive greater benefit from chemotherapy than do patients with wild-type tumours. This result is consistent with several recent studies that have reported higher response rates not only to EGFR TKIs, but also to chemotherapy in both the first-line and second-line advanced NSCLC settings when EGFR mutations are present. In the IPASS (Iressa pan-Asia Study) trial that compared first-line gefitinib to paclitaxel/carboplatin in Asian patients enriched for EGFR mutation, response in patients with mutations was two-fold higher in the chemotherapy arm (47.3% vs 23.5%).¹³ There was also a trend towards longer progression-free survival in the chemotherapy arm for patients with mutations compared to patients with wild-type tumors, although the difference was not statistically significant (HR 0.78, p=0.11). In the INTEREST (IRESSA NSCLC Trial Evaluating Response and Survival against Taxotere) trial that compared second-line gefitinib to docetaxel in unselected patients, those with mutations also demonstrated a two-fold higher response rate to docetaxel than patients with wild-type EGFR (21.1% vs 9.8%), and their median survival on the docetaxel arm was also longer (16.6 vs 6.0 months).¹⁶ In the two large Phase III trials that compared chemotherapy alone to chemotherapy and cetuximab in advanced stage NSCLC, patients with EGFR mutant tumors receiving chemotherapy alone also demonstrated significantly longer survival with no differentially greater benefit seen in patients treated with cetuximab.^{26, 27} These four trials suggest that the presence of an EGFR mutation is predictive of a differential response to chemotherapy. However, due to the absence of untreated control arms in these studies, it was not possible to determine whether the presence of a mutation was predictive of a differential survival benefit from chemotherapy.

There are fewer reports of the predictive effect of EGFR mutation status in the adjuvant chemotherapy setting. Suehisa et al²⁸ assessed mutation status in 187 Japanese patients with adenocarcinoma (~80% stage I) who had undergone surgical resection at their institution. As in our own study, survival of the 79 patients with mutations was superior, but the difference was not significant (HR 0.81, p=0.48). Twenty-five patients with mutations received adjuvant uracil-tegafur. When their survival was compared to that of the remaining 54 patients with mutations who received no adjuvant therapy, no significant survival benefit could be detected (HR 0.52, p=0.28). It is important to note, however, that assignment to receive adjuvant therapy in this report was not random, but rather at the discretion of the treating physicians. Furthermore, baseline patient and tumor characteristics were not provided for patients who did and did not receive adjuvant therapy. The effect of EGFR protein expression, although not EGFR mutation status or copy number, in IALT (international adjuvant lung trial) has been presented in abstract form. ²⁹ No differential survival benefit from adjuvant platinum-based chemotherapy could be identified for any degree of EGFR staining.

Thus, JBR.10 provides the first evidence to suggest that, as in the advanced disease setting, patients with EGFR mutations appear to derive greater survival benefit from post-operative adjuvant platinum-based chemotherapy than do patients with WT EGFR. Because only a small proportion of tumors in JBR.10 had EGFR mutations, our study lacked statistical power. Thus, our observations require confirmation. However, such confirmation will have to come from retrospective analyses of tumor bank specimens from completed trials, as it is unlikely that further adjuvant chemotherapy studies will ever be conducted with a no-treatment arm. Furthermore, in order to have adequate power to detect statistically significant differences, meta-analyses may be required.

Despite using two additional sensitive methods (fragment analysis and ARMS), the EGFR mutation rate detected in this cohort was at the lower level of previously reported prevalence rates for EGFR mutations in North American patients with NSCLC. We did not search for other mutations including T790 M in this analysis since they are not known to be associated with a differential response to chemotherapy, and their extremely low frequency would preclude meaningful observations or conclusions in this study. Unexpectedly, the prevalence rate of high EGFR gene copy number was greater than that reported previously in many studies including BR.21 and ISEL,^15,16 but consistent with more recent results of FISH analysis in early stage resected NSCLC, which have reported prevalence rates ranging from 50-74%.^{30, 31}

Several previous studies have reported that high EGFR gene copy number (high polysomy and amplification) is associated with poorer prognosis in advanced NSCLC patients,^14,15 but comparable studies for unselected early stage surgically treated NSCLC patients are limited. Hirsch et al³² reported that in 173 stage I-III resected patients, high EGFR gene copy (balanced polysomy/amplification) was associated with a trend towards shorter survival (HR not reported, p=0.13). Liu et al did not find any significant effect of high polysomy (≥5 copies) as determined by chromogenic in situ hybridization in their study of 164 surgical patients from Taiwan (median survival 56.2 versus 53.4 months). In our study, patients with high gene copy number showed a trend towards better survival compared to low copy number patients, but this appears to be driven mainly by the high polysomy patients, as patients with amplification demonstrated a trend towards poorer survival similar to that reported by Hirsch et al.³² These data may suggest the presence of other good prognostic genes on chromosome 7.

There is little evidence to suggest that EGFR copy number is an important predictor of response to chemotherapy and results from the literature are variable. In the advanced NSCLC first-line setting, Hirsch et al¹⁸ reported response rates to paclitaxel/carboplatin of 29.8% for high copy patients vs 25.4% for low copy patients in the TRIBUTE trial. In IPASS,³³ response rates to the same chemotherapy regimen were higher in high than low copy patients (44.8 vs 26.3%) but this may have been driven by the large subpopulation of patients with mutations within the high copy subgroup. In the INTEREST trial,¹⁴ where the mutation rate was lower, response to docetaxel was similar for high and low copy patients (10.1 vs 7.4%), as were median survivals (7.5 vs 7.7 months). In INVITE (Iressa versus vinorelbine in chemonaive elderly patients with advanced non-small-cell lung cancer), a trial that compared first-line gefitinitib to single-agent vinorelbine in elderly patients,³⁴ response according to EGFR copy number was not reported. However, high copy patients had a non-significant trend toward improved overall survival compared to low copy patients (HR 0.52, 95%CI 0.25-1.10) in the vinorelbine arm, and unexpectedly, high copy patients appeared to derive greater survival benefit from vinorelbine compared to gefitinib (HR 1.61, 95%CI 0.87-3.01, p: not significant).

In the current study, although the differences were not statistically significant, there was a trend to suggest that high copy patients possibly might derive greater benefit from the adjuvant vinorelbine/cisplatin doublet studied in JBR.10. This result would be in keeping with the superior survival seen with single-agent vinorelbine in the INVITE study.³⁴

In summary, the presence of EGFR activating mutations or high gene copy may affect survival benefit from adjuvant chemotherapy. Confirmatory studies might be performed more easily using tumour samples from placebo-controlled adjuvant chemotherapy trials in East Asia where the EGFR mutation rate is significantly higher. In studies from Western countries, it may be necessary to pool results in meta-analyses in order to have an adequate number of subjects with mutations to achieve statistical significance. Meta-analyses may also have the power to isolate the effect of high copy number in patients with EGFR wild-type tumors. Our results may provide some explanation as to why EGFR mutation and gene copy are not as predictive in trials comparing EGFR TKIs to chemotherapy, as the benefit of both therapies appears to be higher in patients with EGFR mutant or high copy tumors. Finally, the clinically meaningful survival benefit from chemotherapy experienced by patients with EGFR mutant tumors in BR.10 suggests that adjuvant chemotherapy should not be withheld form these patients in favour of treatment with an EGFR TKI. This is of particular importance in view of the lack of survival benefit seen with adjuvant gefitinib in the NCIC CTG BR.19 trial.³⁵