Final analysis of the randomised PEAK trial: overall survival and tumour responses during first-line treatment with mFOLFOX6 plus either panitumumab or bevacizumab in patients with metastatic colorectal carcinoma

Purpose To report planned final overall (OS) and progression-free survival (PFS) analyses from the phase II PEAK trial (NCT00819780). Methods Patients with previously untreated, KRAS exon 2 wild-type (WT) metastatic colorectal cancer (mCRC) were randomised to mFOLFOX6 plus panitumumab or bevacizumab. The primary endpoint was PFS; secondary endpoints included OS, objective response rate, duration of response (DoR), time to response, resection and safety. Treatment effect by tumour RAS status was a prespecified objective. Exploratory analyses included early tumour shrinkage (ETS) and depth of response (DpR). Results One hundred seventy patients had RAS WT and 156 had RAS WT/BRAF WT mCRC. Median PFS was longer for panitumumab versus bevacizumab in the RAS WT (12.8 vs 10.1 months; hazard ratio (HR) = 0.68 [95% confidence intervals (CI) = 0.48–0.96]; p = 0.029) and RAS WT/BRAF WT (13.1 vs 10.1 months; HR = 0.61 [95% CI = 0.42–0.88]; p = 0.0075) populations. Median OS (68% OS events) for panitumumab versus bevacizumab was 36.9 versus 28.9 months (HR = 0.76 [95% CI = 0.53–1.11]; p = 0.15) and 41.3 versus 28.9 months (HR = 0.70 [95% CI = 0.48–1.04]; p = 0.08), in the RAS WT and RAS WT/BRAF WT populations, respectively. Median DoR (11.4 vs 9.0 months; HR = 0.59 [95% CI = 0.39–0.88]; p = 0.011) and DpR (65.0 vs 46.3%; p = 0.0018) were improved in the panitumumab group. More panitumumab patients experienced ≥30% ETS at week 8 (64 vs 45%; p = 0.052); ETS was associated with improved PFS/OS. No new safety signals occurred. Conclusions First-line panitumumab + mFOLFOX6 increases PFS versus bevacizumab + mFOLFOX6 in patients with RAS WT mCRC. Electronic supplementary material The online version of this article (doi:10.1007/s00384-017-2800-1) contains supplementary material, which is available to authorized users.


Introduction
First-line treatments for patients with metastatic colorectal cancer (mCRC) comprise chemotherapy combined with an epidermal growth factor receptor inhibitor (EGFRI), Electronic supplementary material The online version of this article (doi:10.1007/s00384-017-2800-1) contains supplementary material, which is available to authorized users. panitumumab or cetuximab, or the vascular endothelial growth factor inhibitor (VEGFI) bevacizumab [1]. Confirmation of wild-type (WT) tumour RAS status is essential ahead of prescribing EGFRIs [1,2], the importance of evaluating tumour RAS status ahead of bevacizumab treatment is less clear [3,4]. In the EU, panitumumab is indicated in combination with FOLFOX or FOLFIRI for the first-line treatment of patients with RAS WT (no mutations in exons 2, 3 and 4 of KRAS and NRAS) mCRC [5]. In the USA, panitumumab is indicated in KRAS (exon 2) WT mCRC combined with FOLFOX for first-line treatment, and as monotherapy following disease progression after prior treatment with fluoropyrimidine, oxaliplatin and irinotecancontaining chemotherapy [6]. Bevacizumab is also indicated in the EU for the treatment of patients with mCRC in combination with fluoropyrimidine-based chemotherapy [7] and in the USA with intravenous 5-fluorouracil-based chemotherapy for first-or second-line mCRC treatment and with fluoropyrimidine-irinotecan-or fluoropyrimidine-oxaliplatinbased chemotherapy for second-line treatment after progression (PD) on a first-line bevacizumab-containing regimen [8].
In addition to RAS mutations, other biomarkers have potential predictive and prognostic importance in mCRC. BRAF mutation status has emerged as a strong prognostic marker [12] but there is a lack of evidence to support best approaches to treatment for BRAF mutant (MT) mCRC [1,2]. Clinical trials may, therefore, be the best option for some patients. Alternatively, a small population sub-analysis from the TRIBE trial suggested a potential OS benefit for patients with BRAF mutations receiving bevacizumab + FOLFOXIRI (n = 16) versus bevacizumab + FOLFIRI (n = 12) [13], and so, this combination may be an option for patients with good performance status.
Here, we report updated OS and PFS from a planned final analysis of PEAK data (RAS WT and RAS WT/BRAF WT populations). Exploratory analyses of tumour assessments beyond RECIST were also performed, focussing on tumour dynamics (early tumour shrinkage (ETS), depth of response (DpR) and changes in tumour load over time (RAS WT population only)).

Material and methods
Study design and patients PEAK (NCT00819780) was a phase II, randomised (1:1), open-label first-line study in patients with previously untreated KRAS exon 2 WT mCRC. Treatment continued until PD, unacceptable toxicity, death, consent withdrawal, or investigator decision. The trial was conducted in compliance with the Declaration of Helsinki. The study protocol was approved by an independent ethics committee at each participating study centre. All patients provided informed consent before any study procedures were performed.
Key eligibility criteria for PEAK (primary analysis) included age ≥18 years; an Eastern Cooperative Oncology Group performance score of 0/1; histologically/cytologically confirmed mCRC with unresectable metastatic disease; KRAS exon 2 (codons 12 and 13) WT tumour status; one unidimensionally measurable lesion of ≥20 mm (per modified RECIST guidelines); and no prior chemotherapy, anti-EGFR therapy or bevacizumab therapy for mCRC.

Extended RAS analyses and BRAF mutation analyses
An extended RAS analysis was prespecified in PEAK [9]. In brief, banked tumour samples for patients with KRAS exon 2 WT tumours were tested for prespecified mutations in NRAS exon 2 (codons 12 and 13), and KRAS and NRAS exons 3 (codons 59 and 61) and 4 (codons 117 and 146). Mutations were also assessed in exon 15 (codon 600) of BRAF (exploratory analysis). Mutation status was assessed using bidirectional Sanger sequencing and WAVE-based SURVEYOR ® scan kits (Transgenomic; Omaha, NE, USA), as reported previously [12].

Efficacy assessments
Endpoints reported here include PFS, OS, ORR, DoR, time to response (TTR), resection rates, DpR, change in tumour load versus baseline (over time), and ETS. Tumour response was assessed by the investigator every 8 weeks (±7 days) using modified RECIST (version 1.0; there was no independent review). ETS was defined as the proportion of patients with tumour shrinkage at week 8; two cutoff values were used, ≥30 and ≥20%. Tumour shrinkage was defined as the change in sum of the longest diameters (SLD) of all target lesions. DpR was defined for each patient as the percentage of tumour shrinkage at nadir or progression. DpR has a positive value for tumour reduction, a negative value for tumour growth and is zero for no change. If shrinkage occurred during treatment, then DpR was the greatest percentage tumour shrinkage observed at nadir versus baseline. In all other cases, DpR was defined as the percentage of tumour change at progression. If progression did not occur, DpR was not defined as PFS could not be defined. Changes in tumour load were calculated as mean (95% CI) percentage change from baseline in the SLD of all target lesions over time.

Safety assessments
Patients were followed-up for safety until 30 days after the last dose of study drug. Adverse events (AEs) were graded using Common Terminology Criteria for Adverse Events (version 3.0), with modifications for specific skin-and nail-related toxicities.

Statistical analyses
The primary endpoint in PEAK was PFS; secondary endpoints included OS, ORR, DoR, TTR, resection rates, and safety (all reported here for the RAS WT population). Sixmonth PFS and 2-year OS rates by resection status were calculated by treatment (exploratory analysis). Exploratory analyses of DpR, changes in tumour load versus baseline and ETS, were performed for the RAS WT population; median PFS and OS were calculated for patients with/without ≥30 and ≥20% ETS, by treatment and overall. Top-line efficacy results (PFS, OS and ORR) were also reported for the RAS WT/BRAF WT and RAS WT/BRAF MT populations.
The treatment effect HR for panitumumab relative to bevacizumab combined with mFOLFOX6 and the associated 95% CIs were estimated from a Cox model stratified by previous adjuvant oxaliplatin therapy. Kaplan-Meier curves were generated for all time-to-event endpoints.
The primary analysis of PEAK was performed in 2012 when the target of~168 PFS events was achieved [9]. This PFS event goal was expected to be achieved with recruitment of approximately 280 patients. The current final analysis was planned to occur 36 months after the last patient enrolled. No formal hypothesis testing was planned; all reported p values are descriptive.
The prespecified extended RAS analysis included all intentto-treat (ITT) patients with RAS data available. The extended RAS statistical analysis plan was developed before the data analysis was conducted and no biomarker testing results were available to the study team before plan approval.
Baseline demographics and disease characteristics were generally similar between treatments in both the RAS WT and RAS WT/BRAF WT populations ( Table 1). The mean (standard deviation) actual follow-up time was 134.1 (73.8) weeks in the panitumumab arm and 115.7 (71.0) weeks in the bevacizumab arm. Overall, the median duration of panitumumab treatment was 14.0 cycles in both the RAS WT and RAS WT/BRAF WT populations. The median duration of bevacizumab treatment was 11.0 and 11.5 cycles in the RAS WT and RAS WT/BRAF WT populations, respectively.
Mortality rates in the RAS WT population at the time of analysis were 65 versus 71% for the panitumumab + mFOLFOX6 versus bevacizumab + mFOLFOX6 groups, respectively ( Table 2). In the RAS WT population, median OS was 36.9 versus 28.9 months in the panitumumab versus bevacizumab groups, respectively (HR = 0.76 (95% CI = 0.53-1.11)); p = 0.15) (Fig. 2b). Although some patient subpopulations were small, the trend (not statistically significant) towards improved OS in the panitumumab group was generally observed across subpopulations defined by baseline covariates (Online Resource Fig. A1). Similar proportions of patients in the panitumumab and bevacizumab groups subsequently crossed over to anti-VEGF and anti-EGFR therapies, respectively (Table 2).
In patients with RAS WT mCRC, median TTR was numerically shorter, median DoR longer and median DpR greater in the panitumumab + mFOLFOX6 versus bevacizumab + mFOLFOX6 group (Table 3). Resection rates and outcomes following resection were similar between treatments for the RAS WT population, although sample sizes were small for this analysis ( Table 3).
The mean percentage change from baseline in tumour load in the RAS WT population appeared to favour panitumumab at all measured time points, although the CIs were wide (Fig. 3). Figure 3 shows tumour data at each scheduled visit; following PD, patients were only followed up for survival and no further CT scans were analysed.
Overall, 154 patients with RAS WT mCRC had tumour shrinkage data available at baseline and week 8. Compared with the bevacizumab + mFOLFOX6 arm, more patients in the panitumumab + mFOLFOX6 group experienced ≥30% ETS (Table 4). Numerically, more patients in this group also had ≥20% ETS. For those achieving ≥30 or ≥20% ETS, numerically longer PFS and OS were seen in the panitumumab versus bevacizumab arm ( Reason for ending panitumumab: Adverse event n=19 Full or partial consent withdrawn n=9 Disease progression n=36 Administrative decision n=4 Death n=3 Protocol-specified criteria n=9 Other n=4 Reason for ending bevacizumab: Protocol deviation n=1 Non-compliance n=2 Adverse event n=19 Full or partial consent withdrawn n=5 Disease progression n=34 Administrative decision n=3 Death n=5 Protocol-specified criteria n=10 Other n=1 Still receiving intervention at data cut-off n=0 Discontinued intervention n=80 Bevacizumab + mFOLFOX6 Allocated to intervention n=82 Received allocated intervention n=80 Did not receive allocated intervention n=2 RAS wild-type: n=170 ITT, intent-to-treat

Discussion
The efficacy results from this final analysis of PEAK were consistent with the results of the primary analysis of this trial [9]. Patients with RAS WT tumours receiving panitumumab + mFOLFOX6 had longer PFS (12.8 vs 10.1 months, p = 0.029) and DoR (11.4 vs 9.0 months, p = 0.011) compared with those receiving bevacizumab + mFOLFOX6. ORR (65 vs 60%, p = 0.86) and resection rates (14 vs 11%) were similar between treatments. Exploratory analyses of DpR (median 65 vs  [10,11,14,15], ETS appeared to be associated with improved PFS and OS, irrespective of first-line treatment received. A limitation to this study is that radiologic assessment was performed by the investigator and no independent radiologic review was performed. In the RAS WT population, median OS was 8 months longer in the panitumumab group (36.9 vs 28.9 months; HR = 0.76 [95% CI = 0.53-1.11]; p = 0.15). Median OS was further extended in patients with RAS WT/BRAF WT tumours receiving panitumumab (41.3 months), but remained the same as for the RAS WT population in RAS WT/BRAF WT patients receiving bevacizumab (28.9 months). This is of note because patients with BRAF MT mCRC have a poor prognosis [12] and there is a small imbalance in the number of patients with BRAF MT tumours in this study (panitumumab arm n = 11; bevacizumab arm n = 3). However, the primary endpoint in PEAK was not OS and this trial was not powered to detect an OS increase, which may explain why the observed difference between arms was not statistically significant. The overall findings from PEAK nonetheless suggest an OS benefit for EGFRIs + chemotherapy versus bevacizumab + chemotherapy, which is consistent with results from FIRE-3 [16] and a congress presentation of limited data for the FOLFOX arm of the CALGB/SWOG 80405 study (full publication not yet available) [17]. This is also in line with results from two recent meta-analyses of these trials, which included primary analysis data from PEAK. Both meta-analyses demonstrated significantly improved ORR and OS with first-line anti-EGFR compared with anti-VEGF therapy in patients with RAS WT mCRC [18,19]. The HR (95% CI) for OS was 0.77 (0.63-0.95) in favour of the EGFRI in both analyses. Similar results were also observed in a meta-analysis including seven first-line trials comparing panitumumab or cetuximab plus chemotherapy with chemotherapy alone or chemotherapy plus bevacizumab [20]. The median OS of 41.3 months achieved in patients with RAS WT/BRAF WT mCRC receiving panitumumab in PEAK, compares well with other data for first-line biologics combined with doublet chemotherapy [12,13]. Data have recently been reported for triplet chemotherapy combined with biologics in several studies including TRIBE [13,21] (bevacizumab), TRIP [22] and VOLFI [23] (panitumumab) and MACBETH [24] (cetuximab), indicating that FOLFOXIRI may offer superior efficacy over doublet chemotherapy in the first-line setting. However, improved outcomes are achieved at the expense of increased grade 3/4 toxicity, including neutropenia, diarrhoea, stomatitis and neurotoxicity, which may limit the use of such regimens in clinical practice.
Although the OS results from PEAK are now more mature (68 and 66% deaths in the RAS WT and RAS WT/BRAF WT populations, respectively), these results need to be considered in context of the fact that PEAK is a phase II trial. There are several possible reasons for the prolonged OS observed. For example, the improved DoR in the panitumumab + mFOLFOX6 arm could be linked with the improved PFS and numerically improved OS seen in this arm. Likewise, the improved DpR, ETS and tumour load changes in the panitumumab versus bevacizumab arm could also play a role  in improving long-term outcomes, as has been suggested in other reports [11,25]. It is unlikely that the numeric, betweentreatment OS differences were driven by resection, as resection rates in patients with RAS WT tumours were balanced between groups. It is also unlikely to be driven by differences in tumour side as the proportion of patients with left-and right-sided disease was well-balanced between treatment groups. As was suggested in FIRE-3 [26], the impact of treatment sequence could also be important, given that the proportions of patients switching biologic approach were balanced  between arms (i.e. EGFRI to VEGFI or VEGFI to EGFRI). Preclinical data suggest that first-line EGFRI treatment may sensitise tumours to subsequent therapy, whereas first-line VEGFI may desensitise tumours [27][28][29]. This provides a biological rationale for first-line EGFRI therapy followed by second-line VEGFI therapy.
As well as comparative efficacy, safety/tolerability and costeffectiveness of the available treatments need to be considered when choosing first-line therapy for patients with RAS WT mCRC. Updated data for the RAS WT population in PEAK suggest that the safety profiles of the two treatments are consistent with previously reported studies and no new toxicities or safety signals were identified. The main tolerability differences were due to known class effects of these treatments, e.g. skin rash/toxicities and hypomagnesaemia with the EGFRI and deep vein thrombosis and hypertension with bevacizumab. Discontinuation rates due to AEs were similar between arms (29 vs 30%, respectively).
Tumour biomarker status is increasingly being used to guide treatment decisions in mCRC and is a rapidly developing area of research. By targeting drugs only to those patients most likely to benefit, predictive biomarkers may improve clinical outcomes and also potentially reduce overall treatment costs. The availability of banked tumour samples from clinical studies is important to permit prospective, prospective-retrospective or retrospective analyses to be performed to test new biomarker hypotheses [30]. Analysis of data from the phase III PRIME study of panitumumab + FOLFOX4 confirmed the importance of RAS testing beyond KRAS in patients with mCRC and was possible due to the availability of RAS data for 90% of the KRAS WT tumour samples [12]. The RAS ascertainment rate in the PEAK study was also high (82%) in comparison to currently available data from other first-line mCRC studies comparing EGFRIs versus VEGFIs (FIRE-3 = 69% [16]; CALGB 80405 = 55% [17]), highlighting the importance of including the availability of paraffin-embedded tumour tissue for biomarker testing as a requirement for study entry. Mandating baseline biopsy and planning the conduct of prospective/prespecified biomarker analyses in clinical trials, can improve the robustness of the study data and future analyses. While tumour RAS status is undoubtedly a valuable predictive marker for response to EGFRIs, it remains important to investigate the roles of other potential biomarkers that could improve patient selection (e.g. epiregulin/amphiregulin, PIK3CA, PTEN). Data from the PRIME study indicated that BRAF mutations are linked with poor outcomes irrespective of treatment received [12]. RAS mutations do not appear to be prognostic for patients with mCRC. In summary, ORR by RECIST is the accepted standard for assessing response in solid tumours, but has limitations in terms of assessing dynamic tumour burden changes. Although ORRs were similar between treatments in PEAK, tumour responses in the panitumumab arm appeared to occur earlier, last longer and be deeper compared with those occurring in the bevacizumab arm in these exploratory analyses. More patients receiving panitumumab + mFOLFOX6 versus bevacizumab + mFOLFOX6 had ≥30% ETS; ETS of ≥30 or ≥20% appeared to be associated with improved PFS and OS, irrespective of treatment received. PFS was also significantly improved for patients with RAS WT mCRC receiving panitumumab + mFOLFOX6 versus bevacizumab + mFOLFOX6. For panitumumab versus bevacizumab, median OS was 8.0 months longer in the RAS WT (36.9 vs 28.9 months) and 12.4 months longer in the RAS WT/BRAF WT (41.3 vs 28.9 months) populations, respectively (neither difference being statistically significant). Based on these data, panitumumab + mFOLFOX6 can be considered an effective first-line treatment for patients with RAS WT mCRC. A phase III trial is warranted to more fully evaluate any potential differences between panitumumab and bevacizumab treatment in the first-line setting.
Statement of human rights PEAK (NCT00819780) was conducted in compliance with the Declaration of Helsinki. The study protocol was approved by an independent ethics committee at each participating study centre.
Informed consent All patients provided informed consent before any study procedures were performed.
Open Access This article is distributed under the terms of the Creative Comm ons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.