Available at journal homepage:

Transcription

1 European Journal of Cancer (212) 48, Available at journal homepage: Addition of cetuximab to chemotherapy as first-line treatment for KRAS wild-type metastatic colorectal cancer: Pooled analysis of the CRYSTAL and OPUS randomised clinical trials Carsten Bokemeyer a,g,, Eric Van Cutsem b,g, Philippe Rougier c, Fortunato Ciardiello d, Steffen Heeger e, Michael Schlichting e, Ilhan Celik e, Claus-Henning Köhne f a University Medical Center Hamburg-Eppendorf, Hamburg, Germany b University Hospital Gasthuisberg, Leuven, Belgium c European Hospital George Pompidou, APHP and UVSQ, Paris, France d Division of Medical Oncology, Second University of Naples, Naples, Italy e Merck KGaA, Darmstadt, Germany f Klinikum Oldenburg, Oldenburg, Germany Available online 23 March 212 KEYWORDS Cetuximab KRAS BRAF Colorectal cancer Abstract Background: The CRYSTAL and OPUS randomised clinical trials demonstrated that adding cetuximab to first-line chemotherapy in patients with KRAS wild-type metastatic colorectal cancer (mcrc) significantly improved treatment outcome compared with chemotherapy alone. The objective of this pooled analysis was to further investigate these findings in patients with KRAS wild-type tumours using extended survival data and following an enhancement in the ascertainment rate of KRAS and BRAF tumour mutation status from these studies. Methods: Pooled individual patient data from each study were analysed for overall survival (OS), progression-free survival (PFS) and best overall response rate (ORR) in patients evaluable for KRAS and BRAF mutation status. Treatment arms were compared according to mutation status using log rank and Cochran Mantel Haenszel tests. Results: In 845 patients with KRAS wild-type tumours adding cetuximab to chemotherapy led to a significant improvement in OS (hazard ratio [HR].81; p =.62), PFS (HR.66; p <.1) and ORR (odds ratio 2.16; p <.1). BRAF mutations were detected in 7/8 evaluable tumours. No significant differences were found in outcome between the treatment Corresponding author: Address: Dept. of Oncology, Hematology, BMT with section pneumology, Hubertus Wald Tumorzentrum University Cancer Center Hamburg, University Medical Center Hamburg-Eppendorf, Martinistr. 52, 2246 Hamburg, Germany. Tel.: +49 () ; fax: +49 () address: c.bokemeyer@uke.de (C. Bokemeyer). g They contributed equally to this work /$ - see front matter Ó 212 Elsevier Ltd. All rights reserved. doi:1.116/j.ejca

2 C. Bokemeyer et al. / European Journal of Cancer 48 (212) groups in these patients. Prognosis was worse in each treatment arm for patients with BRAF tumour mutations compared with those with BRAF wild-type tumours. Conclusion: Analysis of pooled data from the CRYSTAL and OPUS studies confirms the consistency of the benefit obtained across all efficacy end-points from adding cetuximab to firstline chemotherapy in patients with KRAS wild-type mcrc. BRAF mutation does not appear to be a predictive biomarker in this setting, but is a marker of poor prognosis. Ó 212 Elsevier Ltd. All rights reserved. 1. Introduction Activating mutations of the KRAS gene, which encodes a downstream effector of ligand-stimulated epidermal growth factor receptor (EGFR)-signalling, have been reported to occur in 35 4% of colorectal cancers (CRC). 1,2 Retrospective analysis of single-arm studies initially suggested that the activity of the EGFR-targeting monoclonal antibody cetuximab in the treatment of metastatic colorectal cancer (mcrc) was limited to patients whose tumours were wild-type at codons 12 and 13 of the KRAS gene. 2 6 This observation was confirmed in the first-line setting in retrospective analyses of the randomised phase III CRYSTAL (Cetuximab Combined with Irinotecan in First-Line Therapy for Metastatic Colorectal Cancer) 7 and randomised phase II OPUS (Oxaliplatin and Cetuximab in First-Line Treatment of mcrc) 8 clinical trials, in which cetuximab was combined with standard first-line irinotecan- and oxaliplatin-based chemotherapy regimens, respectively. An updated analysis of the CRYSTAL trial confirmed that the addition of cetuximab to a combined chemotherapy regimen of infusional 5-fluorouracil, leucovorin and irinotecan (FOLFIRI) significantly improved response rate, progression-free survival (PFS) and overall survival in the first-line treatment of patients with mcrc compared with FOLFIRI alone, 9 with a significant interaction between tumour KRAS mutation status and treatment effects. Similarly, the updated analysis of the OPUS study confirmed that tumour KRAS mutation status was predictive for clinical outcome (PFS and tumour response) in patients receiving cetuximab combined with infusional 5-fluorouracil, leucovorin and oxaliplatin (FOLFOX-4) as first-line treatment for mcrc. 1 Retrospective analyses have also suggested that mutations in the BRAF gene, the product of which is a downstream effector of KRAS in the MAPK pathway, 11 could also be predictive for the efficacy of cetuximab treatment in pretreated mcrc These studies have raised the possibility that knowledge of the BRAF tumour mutation status in patients with KRAS wildtype tumours may facilitate the further tailoring of EGFR-targeted treatment to those mcrc patients most likely to derive a clinical benefit. Although not definitive, the expanded analyses of the CRYSTAL study suggested that BRAF tumour mutation did not appear to be a predictive biomarker for the addition of cetuximab to FOLFIRI in the first-line treatment of KRAS wild-type mcrc. 9 Meta-analyses of randomised studies form an important component of evidence-based medicine. Firstly, analyses of pooled individual patient data can, through increasing the statistical power of analysis, provide a more precise estimate of the overall treatment effect. Secondly, they can substantiate the effects for sub-populations of interest, and highlight or confirm differences between treatments. Such analyses therefore, may provide a more objective appraisal of treatment benefit than the consideration of individual studies. 17,18 The objective of the current pooled analysis was to investigate the association between tumour KRAS and BRAF mutation status and the efficacy of adding cetuximab to standard first-line chemotherapy in the CRYSTAL and OPUS clinical trials. 9,1 2. Patients and methods 2.1. Patient population The primary analysis population of the CRYSTAL study comprised 1198 patients who received FOLFIRI plus cetuximab or FOLFIRI alone. 7 The intention to treat population (ITT) of the OPUS study comprised 337 patients who received either FOLFOX-4 plus cetuximab or FOLFOX-4 alone. 8 The key eligibility criteria were very similar for the two studies. Patients were eligible for inclusion if they: were P18 years old; had a histologically confirmed, first-occurrence of a non-resectable, EGFR-expressing, metastatic adenocarcinoma of the colon or rectum; had at least one radiologically measurable lesion in a non-irradiated area; had a life expectancy of P12 weeks, an Eastern Cooperative Oncology Group (ECOG) performance status 62 and had adequate hepatic, renal and bone marrow function. Patients were ineligible if they were pregnant, or had a history of previous exposure to EGFR-targeted therapy, previous irinotecanbased (CRYSTAL) or oxaliplatin-based (OPUS) chemotherapy or any other chemotherapy (except adjuvant treatment terminating 66 months prior to the start of study treatment) for CRC, or uncontrolled severe organ or metabolic dysfunction. They were also ineligible if they had received radiotherapy, surgery (excluding prior

3 1468 C. Bokemeyer et al. / European Journal of Cancer 48 (212) diagnostic biopsy) or any investigational drug in the 3 days before the start of study treatment (CRYSTAL) or randomisation (OPUS). Both studies were approved by local institutional independent ethics committees and were conducted in accordance with the Declaration of Helsinki. All patients provided written, informed consent prior to study entry Randomisation, treatment and study design In both studies, randomisation (1:1) to standard chemotherapy plus cetuximab or chemotherapy alone was carried out using a stratified permuted-block procedure, with ECOG performance status ( and 1 versus 2) as a stratification factor. In the CRYSTAL study, region (Western Europe versus Eastern Europe versus outside Europe) was an additional stratification factor. Patients in the CRYSTAL study received on day 1 of a 14-day treatment cycle, cetuximab (initial dose 4 mg/m 2 infused over 2 h, and 25 mg/m 2 weekly over 1 h thereafter) followed after 1 h by FOLFIRI (irinotecan 18 mg/m 2, day 1, infused over 3 9 min; leucovorin 2 mg/m 2 L-form, or 4 mg/m 2 racemic, infused over 2 h, followed by 5-fluorouracil, as a 4 mg/m 2 intravenous bolus then a 24 mg/m 2 46-h continuous infusion) or FOLFIRI alone. Patients in the OPUS study received on day 1 of a 14-day treatment cycle, cetuximab (as in the CRYSTAL study) followed after 1 h by FOLFOX-4 (oxaliplatin 85 mg/m 2, day 1, infused over 2 h; leucovorin 2 mg/m 2, infused over 2 h, followed by 5-fluorouracil, as a 4 mg/m 2 intravenous bolus then a 6 mg/m 2 infusion over 22 h, days 1 and 2) or FOLFOX-4 alone. Treatment was continued in both studies until disease progression or the occurrence of unacceptable toxicity Assessments and study end-points For each study, a radiological assessment of response was carried out every eight weeks until disease progression or withdrawal for any reason. The primary end-point was PFS in the CRYSTAL study and, best overall response in the OPUS study, respectively. In each case, these end-points were assessed, according to modified World Health Organization criteria by an independent review committee. Secondary end-points included overall survival, best overall response (CRYSTAL), PFS (OPUS) and safety. In both studies, retrospective subgroup analyses were designed to investigate associations between these efficacy end-points and the status of molecular biomarkers including the KRAS and BRAF mutation status of the tumours. Efficacy was reported for subsets of patients from both studies according to tumour KRAS and BRAF mutational status KRAS and BRAF mutation analysis DNA was extracted from formalin-fixed paraffinembedded (FFPE) tumour tissue and KRAS mutations at codons 12 and 13 of exon 2 and BRAF (V6E) mutations were assessed in both studies using a polymerase chain reaction clamping and melting curve technique, as previously described. 7, Statistical methods and considerations The data cuts-offs used for statistical analysis were for the CRYSTAL study, 27th July 26 for PFS and best overall response where the independent blinded review was completed for confirmatory analyses, 31st May 29 for overall survival and 28th August 29 for the updated KRAS analysis, and for OPUS, 1st March 27 for best overall response and PFS, 3th November 28 for overall survival and 12th August 29 for the updated KRAS analysis. The current analysis was performed on pooled individual patient data from both studies. A Cox s proportional hazards model and a logistic regression model stratified by the common randomisation strata (ECOG performance status) and adjusted by the study effect was used to estimate the hazard ratios (HRs) or odds ratios for the treatment effect in relation to progression-free and overall survival, and best overall response, respectively. Treatment groups were compared using stratified log rank tests for progression-free and overall survival and a stratified Cochran Mantel Haenszel (CMH) test for best overall response. Hazard and odds ratios are presented for chemotherapy plus cetuximab versus chemotherapy alone. Inter-study heterogeneity was assessed by likelihood ratio test on study treatment interaction in stratified Cox s proportional hazards model. A sensitivity analysis was performed using a random/fixed study effect model. 3. Results The population of patients from the CRYSTAL and OPUS studies used in this pooled analysis is shown in Fig. 1. In this analysis 1378/1535 patients (9%) from both studies were evaluable for KRAS tumour mutation status and 845 patients (61%) had KRAS wild-type tumours (666/163 evaluable patients from the CRYS- TAL study and 179/315 evaluable patients from the OPUS study). The pooled analysis was performed on this combined population of patients with KRAS wildtype tumours. BRAF tumour mutation status was evaluated in 8/845 (95%) patients with KRAS wild-type tumours and BRAF mutations were detected in 7 (9%) of these patients (Table 1). The baseline characteristics in the pooled population with KRAS wild-type tumours and in patients with KRAS wild-type tumours

4 C. Bokemeyer et al. / European Journal of Cancer 48 (212) Randomised and treated pooled populations from the CRYSTAL and OPUS studies n=1535 Chemotherapy plus Cetuximab n=768 Chemotherapy alone n=767 Patients discontinued: Reasons, n Disease progression 521 Adverse event 65 Death 34 Withdrawal of consent 3 Symptomatic deterioration 15 Non-compliance 14 Lost to follow-up 2 Other 79 Missing 8 Patients discontinued: Reasons, n Disease progression 52 Adverse event 54 Death 29 Withdrawal of consent 36 Symptomatic deterioration 17 Non-compliance 15 Lost to follow-up 5 Other 88 Missing 3 KRAS evaluable population n=689 KRAS evaluable population n=689 KRAS population n=1378 KRAS wild-type n=398 KRAS wild-type/braf wild-type n=349 KRAS wild-type/braf mutant n=32 KRAS/BRAF population n=8 KRAS wild-type n=447 KRAS wild-type/braf wild-type n=381 KRAS wild-type/braf mutant n=38 Fig. 1. Patient disposition in the pooled analysis. according to BRAF mutation status were generally balanced between the treatment arms and across these populations. In the pooled analysis of the 845 patients with KRAS wild-type tumours, the addition of cetuximab to first-line chemotherapy was associated with a significant increase in the chance of response compared with chemotherapy alone (odds ratio for stratified analysis 2.16; p <.1; Fig. 2A and Table 2), and a significant reduction in the risk of disease progression compared with chemotherapy alone (HR for stratified analysis.66; p <.1; Figs. 2B, 3A and Table 2). The risk of death was also significantly lower in patients who received chemotherapy plus cetuximab compared with those receiving chemotherapy alone (HR for stratified analysis.81; p =.62; Figs. 2C, 4A and Table 2). The treatment effect did not vary across the two studies for any of the efficacy end-points studied, as indicated by the non-significant p-values in the tests for heterogeneity (Table 2). Also, the results of the sensitivity analysis for fixed or random effects models were essentially similar thus confirming the robustness of the given results (Fig. 2A C). In patients with KRAS wild-type and BRAF wildtype tumours, the addition of cetuximab to chemotherapy led to a significant improvement in all the efficacy end-points examined and the treatment effect did not appear to vary across the two studies (Table 2, and Figs. 3B and 4B). In the 7 patients with KRAS wildtype and BRAF mutant tumours, a poorer prognosis was observed in both treatment arms when compared with those patients with KRAS wild-type and BRAF wild-type tumours (Table 2, Figs. 3B and 4B). In this group of patients, those receiving chemotherapy plus cetuximab (n = 32) demonstrated improved tumour responses (odds ratio 1.6), PFS time (HR.67) and survival time (HR.62), compared with patients receiving chemotherapy alone (n = 38), although these differences were not statistically significant, and a significant heterogeneity p-value was observed for survival time (Table 2).

5 147 C. Bokemeyer et al. / European Journal of Cancer 48 (212) Table 1 Patient and disease characteristics in patients with KRAS wild-type tumours according to BRAF mutation status. KRAS wt KRAS wt/braf wt KRAS wt/braf mt (N = 845) (N = 73) (N = 7) + cetuximab + cetuximab + cetuximab (n = 447) (n = 398) (n = 381) (n = 349) (n = 38) (n = 32) Sex Male 266 (6) 238 (6) 228 (6) 214 (61) 22 (58) 17 (53) Female 181 (4) 16 (4) 153 (4) 135 (39) 16 (42) 15 (47) Age, years Median (range) 59 (19 84) 61 (24 79) 59 (19 84) 61 (24 79) 59 (25 75) 64 (34 79) <65 years 297 (66) 246 (62) 254 (67) 216 (62) 25 (66) 18 (56) P65 years 15 (34) 152 (38) 127 (33) 133 (38) 13 (34) 14 (44) Region Western Europe 197 (44) 179 (45) 164 (43) 147 (42) 22 (58) 23 (72) Eastern Europe 181 (41) 163 (41) 167 (44) 157 (45) 11 (29) 5 (16) Outside Europe * 69 (15) 56 (14) 5 (13) 45 (13) 5 (13) 4 (13) ECOG PS (95) 379 (95) 363 (95) 332 (95) 34 (89) 32 (1) 2 24 (5) 19 (5) 18 (5) 17 (5) 4 (11) Liver metastases only Yes 95 (21) 93 (23) 85 (22) 8 (23) 4 (11) (31) No 352 (79) 35 (77) 296 (78) 269 (77) 34 (89) 22 (69) Number of metastatic sites (83) 344 (86) 321 (84) 35 (87) 25 (66) 25 (78) >2 71 (16) 48 (12) 55 (14) 39 (11) 12 (32) 6 (19) missing 6 (1) 6 (2) 5 (1) 5 (1) 1 (3) 1 (3) Previous adjuvant chemotherapy Yes 9 (2) 85 (21) 8 (21) 79 (23) 6 (16) 5 (16) No 357 (8) 313 (79) 31 (79) 27 (77) 32 (84) 27 (84) Data are n (%) unless otherwise stated. wt = wild-type; mt = mutant; = chemotherapy; ECOG PS = Eastern Cooperative Oncology Group Performance Status. * Patients from outside of Europe were recruited in the CRYSTAL study only. 4. Discussion The objective of this pooled analysis of individual patient data from the CRYSTAL and OPUS studies was to increase the statistical power of the efficacy analysis in those patients with mcrc who were most likely to benefit from treatment with chemotherapy plus cetuximab compared with chemotherapy alone, and thereby to more accurately evaluate the benefit of adding cetuximab to standard first-line treatment. These clinical studies were deemed to be comparable in that cetuximab was combined, in each case, with standard first-line regimens (FOLFIRI and FOLFOX-4) of essentially equivalent efficacy. 19,2 In addition, eligibility criteria for the two studies corresponded very closely and randomisation stratification factors were similar. The outcome of the analysis of pooled patient data from the CRYSTAL and OPUS studies, confirms the observations from the separate studies, 9,1 that the addition of cetuximab to standard chemotherapy for patients with KRAS wild-type tumours significantly improves best overall response (odds ratio 2.16; p <.1), PFS (HR.66; p <.1) and overall survival (HR.81; p =.62) compared with those patients receiving chemotherapy alone. The treatment effect did not vary across studies as indicated by the non-significant p-values when testing for heterogeneity, and the results of the sensitivity analysis for fixed and random effects models were similar, confirming the robustness of the data. BRAF mutations have been described in up 14% of previously treated, chemorefractory patients with KRAS wild-type mcrc, 12 and are reported to be associated with resistance to treatment with EGFR-targeting monoclonal antibodies. These findings come primarily from a series of retrospective analyses of single-arm studies of pretreated patients with chemorefractory disease in which a chemotherapy-alone control arm was not investigated ,16 The current pooled analysis of the updated CRYSTAL and OPUS studies has provided the opportunity to further investigate this association in patients with KRAS wild-type tumours treated with cetuximab and first-line chemotherapy in randomised trials. BRAF mutations were detected in 9% of evaluable patients with KRAS wild-type tumours. While some benefit was observed in patients with KRAS wild-type and BRAF mutant tumours receiving chemotherapy plus cetuximab compared with chemotherapy alone, these differences were not statistically significant, and in the survival analysis, the treatment effect appeared to vary significantly between the studies as indicated by the heterogeneity test. This finding may reflect the

6 C. Bokemeyer et al. / European Journal of Cancer 48 (212) Table 2 Efficacy data of patients with KRAS wild-type tumours according to BRAF mutation status. KRAS wt KRAS wt/braf wt KRAS wt/braf mt (N = 845) (N = 73) (N = 7) + cetuximab + cetuximab + cetuximab (n = 447) (n = 398) (n = 381) (n = 349) (n = 38) (n = 32) Tumour response ORR Odds ratio a [95% Confidence Interval (CI)] [ ] [ ] [ ] p-value b <.1 <.1.46 (Heterogeneity p-value) c (.56) (.59) (.17) PFS Median PFS [95% CI] [ ] [ ] [7.4 9.] [ ] [ ] [ ] Hazard ratio a [95% CI] [.55.8] [.52.79] [ ] p-value b <.1 <.1.23 (Heterogeneity p-value) c (.33) (.34) (.38) OS Median OS [95% CI] [ ] [ ] [ ] [ ] [ ] [ ] Hazard ratio a [95% CI] [.69.94] [.71 1.] [ ] p-value b (Heterogeneity p-value) c (.7) (.7) (.48) Data for overall response rates are %. Data for median progression-free survival and overall survival are in months. wt = wild-type; mt = mutant; = chemotherapy; ORR = best overall response rate; PFS = progression-free survival; OS = overall survival. a Hazard ratios <1 for overall and progression-free survival and odds ratios >1 for best overall response indicate a benefit for the addition of cetuximab to chemotherapy compared with chemotherapy alone. b Likelihood ratio test on treatment effect in stratified Cox s proportional hazards or logistic regression model. c Likelihood ratio test on study treatment interaction in stratified Cox s proportional hazards or logistic regression model. data from the expanded OPUS study. In this study the sample size of patients with KRAS wild-type and BRAF mutant tumours was small (n = 11), however six patients receiving FOLFOX-4 plus cetuximab (two of whom demonstrated a partial response to treatment) experienced prolonged survival compared with five receiving FOLFOX-4 alone (median 2.7 versus 4.4 months). 1 Given the small sample size, this difference should be interpreted cautiously. In contrast in the expanded CRYSTAL study, in patients with KRAS wild-type tumours harbouring BRAF mutations, no significant difference was found between 26 patients receiving FOLFIRI plus cetuximab, compared with 33 patients receiving FOLFIRI alone (HR.98; 95% Confidence Interval (CI) ). 9 Overall, the findings from the pooled analysis support those from the expanded CRYSTAL analysis and suggest that BRAF mutations appear not to be predictive for resistance to cetuximab in combination with standard first-line chemotherapy in patients with KRAS wild-type mcrc. In the pooled analysis, patients with KRAS wild-type tumours harbouring BRAF mutations were found to have poor prognosis across all outcome measures in both treatment arms compared with those with KRAS and BRAF wild-type tumours. These findings further indicate that BRAF tumour mutation appears to be a strong negative prognostic marker for patients with KRAS wild-type mcrc treated with cetuximab in combination with chemotherapy or with chemotherapy alone. These data support those from the phase III CAIRO2 study in which patients with KRAS wild-type mcrc harbouring BRAF mutations and treated with an oxaliplatin-based chemotherapy in combination with the anti-vascular endothelial growth factor monoclonal antibody bevacizumab had a significantly shorter overall survival which was independent of treatment with cetuximab. 21 Furthermore, recent findings reported from large randomized trials demonstrate BRAF tumour mutations to be associated with poor survival compared with BRAF wild-type tumours in mcrc cancer patients treated with 5-FU-based chemotherapy 22 and in stage II and III colon cancer patients treated with 5-FU-based adjuvant chemotherapy. 23 Collectively these findings suggest that BRAF mutation status could be considered as a stratification factor for future trials investigating mcrc patients receiving EGFR-targeting agents with standard chemotherapy or combinations of standard chemotherapy alone. In summary, this pooled analysis confirmed the consistency of the benefit for all efficacy end-points when cetuximab is added to standard first-line chemotherapy regimens in patients with mcrc whose tumours are wild-type for KRAS. The data further suggest that BRAF mutation is not a predictive biomarker for cetux-

7 1472 C. Bokemeyer et al. / European Journal of Cancer 48 (212) A Study Population Analysis Odds ratio [95% CI] Pooled analysis KRAS wild-type (n=845) Stratified 2.16 [ ] KRAS wild-type (n=845) Unstratified 2.14 [ ] Fixed effects 2.16 [ ] Random effects 2.16 [ ] CRYSTAL KRAS wild-type (n=666) Stratified 2.7 [ ] OPUS KRAS wild-type (n=179) Stratified 2.55 [ ] Benefit under alone Benefit under + cetuximab B Study Population Hazard ratio [95% CI] Pooled analysis KRAS wild-type (n=845) Stratified.66 [.55.8] KRAS wild-type (n=845) Unstratified.65 [.54.79] Fixed effects.67 [.5.81] Random effects.67 [.55.81] CRYSTAL KRAS wild-type (n=666) Stratified.7 [.56.87] OPUS KRAS wild-type (n=179) Stratified.57 [.38.86] Benefit under + cetuximab Benefit under alone C Study Population Analysis Hazard ratio [95% CI] Pooled analysis KRAS wild-type (n=845) Stratified.81 [.69.94] KRAS wild-type (n=845) Unstratified.8 [.69.93] Fixed effects.81 [.69.94] Random effects.81 [.69.94] CRYSTAL KRAS wild-type (n=666) Stratified.8 [.67.95] OPUS KRAS wild-type (n=179) Stratified.85 [ ] Benefit under + cetuximab Benefit under alone Fig. 2. Analysis of efficacy and outcome in the pooled data from patients with KRAS wild-type tumours. Shown are forest plots of odds ratios for ORR (A) and hazard ratios for PFS (B) and OS (C) in the pooled population of patients with KRAS wild-type tumours unstratified or stratified by ECOG performance status. Odds ratios are for chemotherapy + cetuximab/chemotherapy, hazard ratios are for chemotherapy + cetuximab/ chemotherapy. Odds ratios (A) and hazard ratios (B and C) from fixed and random effect models and from the individual CRYSTAL and OPUS studies are shown. Size of the filled circle is proportional to size of the subgroup population. = chemotherapy (FOLFIRI/FOLFOX-4).

8 C. Bokemeyer et al. / European Journal of Cancer 48 (212) A Events Median PFS, months [95% CI] (n=447) [ ] + cetuximab (n=398) [ ] Kaplan-Meier estimate At risk + cetuximab + cetuximab Time (months) B KRAS wt population 1..9 Events Median PFS, months [95% CI] BRAF mt (n=38) [ ] BRAF mt +cetuximab (n=32) [ ] BRAF wt (n=381) [7.4-9.] BRAF wt +cetuximab (n=349) [ ] KRAS wt population BRAF mt BRAF mt +cetuximab BRAF wt BRAF wt +cetuximab KRAS wt population, at risk BRAF mt +cetuximab BRAF wt BRAF wt +cetuximab Fig. 3. Progression-free survival in the analysis of pooled data from patients with KRAS wild-type tumours. Shown are Kaplan Meir plots of PFS of the pooled patient population: according to treatment arm (A) and according to treatment arm and tumour BRAF mutation status (B). = chemotherapy (FOLFIRI/FOLFOX-4). imab in combination with chemotherapy but is a negative prognostic biomarker in this setting. Role of the funding source The legal sponsor of the study was Merck KGaA, Darmstadt, Germany. The sponsor was responsible for data management and statistical analysis. The clinical study was designed in collaboration with the study coordinating investigators CB (OPUS, [N12534]) and EVC (CRYSTAL, [N15412]). The subgroup analysis described in the current manuscript was carried out by the sponsor. The drafting of this manuscript was commissioned by the sponsor. The corresponding author (and all authors) had full access to all study data, and all authors made the final decision to submit the manuscript for publication.

9 1474 C. Bokemeyer et al. / European Journal of Cancer 48 (212) A (n=447) Events 359 Median OS, months 19.5 [95% CI] [ ] + cetuximab (n=398) [ ] Kaplan-Meier estimate cetuximab At risk Time (months) cetuximab B KRAS wt population Events Median OS, months [95% CI] BRAF mt (n=38) [ ] BRAF mt +cetuximab (n=32) [ ] BRAF wt (n=381) [ ] BRAF wt +cetuximab (n=349) [ ] KRAS wt population BRAF mt BRAF mt +cetuximab BRAF wt BRAF wt +cetuximab KRAS wt population, at risk BRAF mt BRAF mt +cetuximab BRAF wt BRAF wt +cetuximab Fig. 4. Overall survival in the analysis of pooled data from patients with KRAS wild-type tumours. Shown are Kaplan Meier survival plots of overall survival for the pooled patient population: according to treatment arm (A) and according to treatment arm and tumour BRAF mutation status (B). = chemotherapy (FOLFIRI/FOLFOX-4). Conflict of interest statement Carsten Bokemeyer has received honoraria for advisory board meetings and for educational lectures, and research funding from Merck KGaA; Eric Van Cutsem has received consultancy from Merck KGaA and Roche, and research funding from Merck KGaA, Roche and Amgen; Philippe Rougier has participated at advisory boards for and received honoraria from Merck KGaA and sanofi aventis, received a research grant and honoraria from Roche, and participated in Pfizer advisory board meetings; Fortunato Ciardiello has received honoraria and research funding from Merck KGaA, Astra Zeneca and Bayer; Steffen Heeger, Ilhan Celik and Michael Schlichting are employees of Merck KGaA; Claus-Henning Köhne has received research funding and payment for lectures including speakers bureaus.

10 C. Bokemeyer et al. / European Journal of Cancer 48 (212) Acknowledgements The authors would like to thank the following from Merck KGaA, Darmstadt, Germany: Christopher Stroh for his contribution in relation to the biomarker analyses, Tobias Haas for carrying out the KRAS and BRAF mutation detection assays and Angela Zubel for critical analysis of the manuscript. Paul Hoban Ph.D. of Cancer Communications and Consultancy Ltd., provided medical writing services on behalf of the study sponsor. References 1. Lea IA, Jackson MA, Li X, et al. Genetic pathways and mutation profiles of human cancers: site- and exposure-specific patterns. Carcinogenesis 27;28: Normanno N, Tejpar S, Morgillo F, et al. Implications for KRAS status and EGFR-targeted therapies in metastatic CRC. Nat Rev Clin Oncol 29;6: De Roock W, Piessevaux H, De Schutter J, et al. KRAS wild-type state predicts survival and is associated to early radiological response in metastatic colorectal cancer treated with cetuximab. Ann Oncol 28;19: Di Fiore F, Blanchard F, Charbonnier F, et al. Clinical relevance of KRAS mutation detection in metastatic colorectal cancer treated by cetuximab plus chemotherapy. Br J Cancer 27;96: Lievre A, Bachet JB, Boige V, et al. KRAS mutations as an independent prognostic factor in patients with advanced colorectal cancer treated with cetuximab. J Clin Oncol 28;26: Lievre A, Bachet JB, Le Corre D, et al. KRAS mutation status is predictive of response to cetuximab therapy in colorectal cancer. Cancer Res 26;66: Van Cutsem E, Kohne CH, Hitre E, et al. Cetuximab and chemotherapy as initial treatment for metastatic colorectal cancer. N Engl J Med 29;36: Bokemeyer C, Bondarenko I, Makhson A, et al. Fluorouracil, leucovorin, and oxaliplatin with and without cetuximab in the first-line treatment of metastatic colorectal cancer. J Clin Oncol 29;27: Van Cutsem E, Kohne CH, Lang I, et al. Cetuximab plus irinotecan, fluorouracil, and leucovorin as first-line treatment for metastatic colorectal cancer: updated analysis of overall survival according to tumor KRAS and BRAF mutation status. J Clin Oncol 211;29: Bokemeyer C, Bondarenko I, Hartmann JT, et al. Efficacy according to biomarker status of cetuximab plus FOLFOX-4 as first-line treatment for metastatic colorectal cancer: the OPUS study. Ann Oncol 211;22: Ji H, Wang Z, Perera SA, et al. Mutations in BRAF and KRAS converge on activation of the mitogen-activated protein kinase pathway in lung cancer mouse models. Cancer Res 27;67: Di Nicolantonio F, Martini M, Molinari F, et al. Wild-type BRAF is required for response to panitumumab or cetuximab in metastatic colorectal cancer. J Clin Oncol 28;26: Laurent-Puig P, Cayre A, Manceau G, et al. Analysis of PTEN, BRAF, and EGFR status in determining benefit from cetuximab therapy in wild-type KRAS metastatic colon cancer. J Clin Oncol 29;27: Loupakis F, Ruzzo A, Cremolini C, et al. KRAS codon 61, 146 and BRAF mutations predict resistance to cetuximab plus irinotecan in KRAS codon 12 and 13 wild-type metastatic colorectal cancer. Br J Cancer 29;11: Siena S, Sartore-Bianchi A, Di Nicolantonio F, Balfour J, Bardelli A. Biomarkers predicting clinical outcome of epidermal growth factor receptor-targeted therapy in metastatic colorectal cancer. J Natl Cancer Inst 29;11: Zlobec I, Bihl MP, Schwarb H, Terracciano L, Lugli A. Clinicopathological and protein characterization of BRAF and K-RAS mutated colorectal cancer and implications for prognosis. Int J Cancer 29;127: Lyman GH, Kuderer NM. The strengths and limitations of metaanalyses based on aggregate data. BMC Med Res Methodol 25;5: Egger M, Smith GD. Meta-Analysis. Potentials and promise. BMJ 1997;315: Tournigand C, Andre T, Achille E, et al. FOLFIRI followed by FOLFOX6 or the reverse sequence in advanced colorectal cancer: a randomized GERCOR study. J Clin Oncol 24;22: Punt CJ. Irinotecan or oxaliplatin for first-line treatment of advanced colorectal cancer? Ann Oncol 25;16: Tol J, Dijkstra JR, Klomp M, et al. Markers for EGFR pathway activation as predictor of outcome in metastatic colorectal cancer patients treated with or without cetuximab. Eur J Cancer 21;46: Richman SD, Seymour MT, Chambers P, et al. KRAS and BRAF mutations in advanced colorectal cancer are associated with poor prognosis but do not preclude benefit from oxaliplatin or irinotecan: results from the MRC FOCUS trial. J Clin Oncol 29;27: Roth AD, Tejpar S, Delorenzi M, et al. Prognostic role of KRAS and BRAF in stage II and III resected colon cancer: results of the translational study on the PETACC-3, EORTC 4993, SAKK 6- trial. J Clin Oncol 21;28: