The predictive effect of primary tumour location in the treatment of metastatic colorectal cancer: a Canadian consensus statement.

In recently published data, the predictive value of primary tumour location for the treatment of metastatic colorectal cancer with available biologic therapies has been explored. Recognizing the potential effect of those data on clinical practice, we convened a meeting of Canadian experts who treat metastatic colorectal cancer to develop a set of national, evidence-based treatment guidelines based on primary tumour location. This report summarizes the relevant evidence and presents the consensus recommendations of those experts.


INTRODUCTION Purpose
The purpose of the national meeting reported here was to develop a set of national, evidence-based guidelines for the treatment of metastatic colorectal cancer (mcrc) based on primary tumour location (ptl). Here, we summarize the evidence with respect to this topic and present the consensus recommendations of the Canadian mcrc experts who attended the meeting.

Participants
A representative group of gastrointestinal medical oncology experts from across Canada was invited to attend. A group of 12 medical oncologists formed the consensus expert panel, which also received input from 10 additional experts (Table i).

Target Audience
The target audience for this report includes ■ medical oncologists involved in the treatment of mcrc, ■ patient advocacy and education groups such as Colorectal Cancer Canada, and ■ provincial or jurisdictional cancer agencies and funding bodies.

SUMMARY OF THE EVIDENCE
Colorectal cancer is the second leading cause of cancer death in the Western world 1 . Despite improvements in screening and diagnosis, approximately 10%-15% of patients present with synchronous metastatic disease, and 25%-40% of patients originally diagnosed with potentially curable disease will develop metastases 2 . In recent years, several clinical and molecular markers that influence outcomes and guide treatment decisions have been identified. Of those markers, ptl has regained considerable attention. The hypothesis that proximal and distal colorectal cancers represent distinct entities originated in the late 1980s [3][4][5] . Anatomically, the right and left colon arise from different embryonic origins: The proximal colon arises from the midgut and receives its main blood supply by way of the superior mesenteric artery. The distal colon arises from the hindgut and is supplied by way of the inferior mesenteric artery. A variety of histologic 6,7 , genetic [8][9][10][11] , and clinical [12][13][14] differences between the left and the right colon have been described. Several recent studies, including two meta-analyses, have shown that prognosis is worse for patients with metastatic right-sided colon cancer (rcc) than for those with left-sided disease (lcc) [12][13][14] .

Effect of PTL on Therapy with EGFR mAbs
The trials of egfr mAbs in RAS wild-type tumours are summarized in Table ii.

First Line
The egfr mAbs cetuximab and panitumumab have been shown to improve outcomes in mcrc. The randomized phase iii crystal trial studied the combination of cetuximab and folfiri (folinic acid-5-fluorouracil-irinotecan) compared with folfiri alone in the first-line treatment of patients with RAS wild-type mcrc. With the addition of cetuximab, improvements were observed in overall survival (os), progression-free survival (pfs), and the overall response rate (orr) 17,25 .
A study looking at the influence of ptl in the population of the crystal trial was recently published 18 . Baseline characteristics in the patients with lcc were relatively balanced; however, in patients with rcc, several differences between the treatment groups appeared to favour the folfiri arm. Those differences in the folfiri-only group included more patients having an Eastern Cooperative Oncology Group performance status of 0, measurements of the index metastatic lesion that were shorter, fewer patients having received prior adjuvant therapy, and patients receiving more subsequent lines of cancer treatment-factors that were neither adjusted nor controlled for in the analysis.
Patients with lcc receiving cetuximab experienced significant benefits in orr [72.5% vs. 40 The authors reported a significant interaction with respect to pfs and os, but not orr, between ptl and treatment.
The randomized phase iii prime trial showed that the combination of folfox (folinic acid-5-fluorouracil-oxaliplatin) and panitumumab given as first-line treatment was associated with improved os duration and pfs in patients with RAS wild-type mcrc 20 . A retrospective analysis of the prime study based on ptl was recently published by Boeckx and colleagues 26,27 . Hazard ratios were adjusted for BRAF status, prior adjuvant treatment, and baseline Eastern Cooperative Oncology Group performance status. The addition of panitumumab was associated with an improved median os duration in patients with lcc (30.3 months vs. 23.6 months; hr: 0.73; 95% ci: 0.57 to 0.93). However, in patients with rcc, no benefit in os duration was associated with combination treatment (11.1 months vs. 15.4 months; hr: 0.87; 95% ci: 0.55 to 1.37). Similarly, the addition of panitumumab was associated with improved pfs in patients with lcc (12.9 months vs. 9.2 months; hr: 0.72; 95% ci: 0.57 to 0.9), but not in those with rcc (7.5 months vs. 7.0 months; hr: 0.8; 95% ci: 0.50 to 1.26). Similar results for os duration and for pfs were seen when BRAF-mutant tumours were excluded.

Second Line
In the second-line setting, the combination of chemotherapy and panitumumab was studied in the 20050181 trial 21,22 . The combination of folfiri and the egfr mAb was associated with improved orr and pfs (os duration was not significantly different). The analysis by tumour location did not uncover a significant difference in os duration between treatments for either lcc or rcc (lcc hr: 0.96; 95% ci: 0.74 to 1.23; rcc hr: 1.14; 95% ci: 0.68 to 1.89). Similar results were observed for pfs 28 . Exclusion of patients with BRAFmutant tumours did not significantly affect those results. Moreover, an analysis of ptl in the piccolo trial (irinotecan plus panitumumab as second-or third-line treatment) did not demonstrate an interaction between panitumumab treatment effect and ptl 23,29 .

Third Line
The randomized phase iii ncic co.17 trial demonstrated an improvement in os duration by 4.7 months with the use of single-agent cetuximab compared with best supportive care in patients with KRAS wild-type, chemotherapyrefractory disease 24 . Brulé et al. 14 undertook a retrospective analysis of that study to examine outcomes according ptl. They showed that, compared with best supportive care in patients with lcc, treatment with cetuximab was associated with significantly improved os duration (

Effect of PTL on Therapy with Bevacizumab
The randomized phase iii max trial from the Australasian Gastrointestinal Trials Group (Table iv) evaluated capecitabine as a single agent or in combination with bevacizumab, with a third arm adding mitomycin C in the first-line treatment of mcrc. In the overall population, the combination of capecitabine and bevacizumab was associated with significantly prolonged pfs and a trend toward improved os and orr 30,31 . A subgroup analysis of the trial presented at the European Society for Medical Oncology meeting in 2014 showed hrs for pfs of 0.82 in rcc (95% ci: 0.54 to 1.22) and 0.51 in lcc (95% ci: 0.4 to 0.63), with p = 0.10 for interaction. Similarly, interaction tests for tumour location and bevacizumab treatment were nonsignificant for os duration, pfs, and orr in the first-line NO16966 trial (capecitabine-oxaliplatin or folfox ± bevacizumab) and the AVF2107g trial (folfiri ± bevacizumab) 32 .

Impact of PTL on Therapy with Either EGFR mAbs or Bevacizumab
Three prospective randomized trials have examined the benefit of an egfr mAb or bevacizumab combined with chemotherapy in the first-line setting ( Table v). Results from the combined Cancer and Leukemia Group B (calgb) and Southwest Oncology Group (swog) 80405 trial showed   no difference in os duration or pfs with chemotherapy (folfiri or folfox) plus cetuximab or bevacizumab as firstline therapy 35 . A retrospective analysis of ptl in that trial was first presented at the 2016 American Society of Clinical Oncology annual meeting 36 and updated in a recently published meta-analysis 28 . Those analyses showed that, in patients with lcc, treatment with cetuximab was associated with a median os duration of 39.3 months; treatment with bevacizumab was associated with a median os duration of 32.6 months (hr: 0.77; p = 0.04). However, in patients with rcc, treatment with cetuximab was associated with a median os duration of only 13.9 months; treatment with bevacizumab was associated with an os duration of 29.2 months (hr: 1.36; p = 0.10). The ptl was observed to be similarly associated with pfs: patients with lcc experienced a nonsignificant improvement in pfs (hr: 0.84; p < 0.15) with cetuximab, and patients with rcc experienced worse pfs (hr: 1.64; p = 0.006) with cetuximab.
The fire-3 trial demonstrated that, in the first-line setting, compared with folfiri plus bevacizumab, folfiri plus cetuximab was associated with improved os 33 . A retrospective analysis looking at the effect of ptl in that trial was recently published in JAMA Oncology 18 . Among patients with lcc, the median os duration was significantly longer for those treated with folfiri plus cetuximab than for those treated with folfiri plus bevacizumab (38.3 months vs. 28.0 months; hr: 0.63; p = 0.002). Conversely, the median os duration was shorter (although not statistically significantly so) in patients with rcc treated with first-line cetuximab than in those treated with bevacizumab (18.3 months vs. 23.0 months; hr: 1.31; p = 0.28). The orr and pfs did not appear to differ between the treatments in either lcc or rcc.
Finally, in the peak trial, panitumumab was compared with bevacizumab, both in combination with folfox; an improvement in pfs was observed in patients treated with panitumumab 34 . In addition, chemotherapy plus panitumumab yielded a statistically nonsignificant improvement in os duration.
The effect of ptl in that trial was recently published 27 . After the hrs were adjusted for BRAF status, prior adjuvant treatment, and baseline Eastern Cooperative Oncology Group performance status, the investigators showed that, in patients with lcc, a statistically nonsignificant improvement in median os duration was associated with panitumumab (43.4 months vs. 32.0 months with bevacizumab; hr: 0.77; 95% ci: 0.46 to 1.28; p = 0.31). In patients with rcc, no clear signal of median os duration benefit was evident (17.5 months vs. 21.0 months; hr: 0.67; 95% ci: 0.30 to 1.50; p = 0.32). A nonsignificant trend toward improvement in pfs was observed for patients with lcc treated with panitumumab (14.6 months vs. 11.5 months; hr: 0.68; 95% ci: 0.45 to 1.04; p = 0.07). In patients with rcc, the pfs associated with the two treatments did not differ (8.7 months with panitumumab vs. 12.6 months with bevacizumab; hr: 1.04; 95% ci: 0.50 to 2.18; p = 0.90). The foregoing results did not change significantly when patients with BRAF-mutant disease were excluded from the analysis. The small number of patients with rcc in the trial (n = 36) likely limited the analysis by ptl.
Table vi summarizes the effect of ptl in trials with bevacizumab and egfr mAbs.

Meta-Analyses of the Effect of PTL
Several recently published meta-analyses have evaluated the predictive effect of ptl (Table vii). Arnold and colleagues 28 recently published the results of summary data from six randomized trials of tumour sidedness and egfr mAb treatment in patients with advanced crc. They used a standard fixed-effects model to calculate the hrs for os duration and pfs, and the or for orr. The six trials included in the pooled analysis were crystal, fire-3, peak, prime, 20050181, and calgb/swog 80405. The hrs were adjusted for covariates, but the covariates adjusted for were different in each study. The results showed that chemotherapy (either folfox or folfiri) plus an egfr mAb (cetuximab or panitumumab) was associated with improved os duration (hr: 0.75; 95% ci: 0.67 to 0.84) in patients with lcc, but not in patients with rcc (hr: 1.14; 95% ci: 0.88 to 1.47). The hr for interaction between ptl and os duration was 1.53 (p < 0.001). In terms of orr, lcc had an or of 2.12 (95% ci: 1.77 to 2.55) with egfr mAb treatment. Interestingly, an improved orr was also seen for rcc, although it was not statistically significant (or: 1.47; 95% ci: 0.94 to 2.29). A trend toward a greater benefit of chemotherapy plus egfr mAb was observed in patients with lcc (p value for interaction: 0.07). Holch et al. 12 performed a meta-analysis of the relevant first-line trials. It included an analysis of the two trials that compared chemotherapy with or without an egfr mAb, crystal and prime, to examine the predictive implications of ptl. A significant benefit of first-line egfr mAb treatment for os duration, pfs, and orr was observed only in patients with RAS wild-type lcc (os hr: 0.69; p < 0.0001; pfs hr: 0.65; p < 0.0001; orr or: 2.45; p < 0.00001). The Holch meta-analysis 12 also included a separate examination of the three trials comparing chemotherapy plus bevacizumab with chemotherapy plus egfr mAbs: calgb/swog 80405, fire-3, and peak. In patients with RAS wild-type lcc, the analysis revealed a significant benefit associated with egfr mAb treatment with respect to os duration and orr, but not pfs. In rcc, improved pfs and os duration were associated with bevacizumab treatment, but only the pfs benefit was statistically significant (pfs hr: 1.53; 95% ci: 1.16 to 2.01; p = 0.003; os hr: 1.3; 95% ci: 0.97 to 1.74; p = 0.081). The or for orr favoured egfr mAb treatment in rcc (or: 1.2; 95% ci: 0.77 to 1.22; p = 0.432). In addition, a significant interaction between ptl and treatment was observed for os duration and pfs, but not for orr.
Li et al. 37 published a meta-analysis of primary tumour site and anti-egfr mAb benefit in mcrc. The analysis of the prognostic value of ptl included eleven studies, but the analysis of the predictive effect was limited to just two randomized controlled trials (crystal and ncic co.17). The results showed that, with respect to os duration, the addition of an anti-egfr mAb was associated with a significant improvement in lcc (hr: 0.60; 95% ci: 0.47 to 0.77; p < 0.0001), but not in rcc (hr: 0.87; 95% ci: 0.54 to 1.40; p = 0.56). Similarly, pfs was improved with the addition of cetuximab in lcc (hr: 0.38; 95% ci: 0.22 to 0.67; p = 0.0008), but not in rcc (hr: 0.79; 95% ci: 0.52 to 1.19; p = 0.26). Significant interaction between tumour location and treatment was observed for os duration and pfs (p = 0.0002 for os and pfs alike).

Limitations
Several limitations apply to the interpretation of the analyses presented in this review. First, the analyses of ptl from the randomized trials are retrospective in nature, potentially resulting in selection, systematic, and random bias. Second, the trials included in the meta-analyses are heterogeneous with respect to study phase, type of RAS testing performed, line of treatment, and treatment arms. Third, the metaanalyses did not analyze individual patient data. In addition, although most trials defined lcc as beginning at the splenic flexure, the calgb 80405 and Australasian Gastrointestinal Trials Group max trials excluded the transverse colon in their analysis. Furthermore, there is evidence to show that, even within rcc and lcc, significant heterogeneity is evident with respect to molecular characteristics 34,37 , such that classification by side could be oversimplifying a more complex interaction between tumour and treatment. Finally, to date, no data are available about the effect of ptl in the context of triplet therapy (that is, folfoxiri) combined with either bevacizumab or an efgr mAb.

DATA SUMMARY AND IMPLICATIONS FOR PRACTICE
The foregoing section has summarized the evidence for the predictive value of ptl with respect to treatment with anti-egfr mAbs and bevacizumab in patients with RAS wild-type mcrc. Despite the limitations already noted, it is important that current guidelines reflect the best available evidence.
For patients with RAS-mutant mcrc, there is no benefit of egfr mAb treatment as a single agent or combined with standard chemotherapy in either lcc or rcc. In patients with RAS wild-type mcrc, treatment should take into account the location of the primary tumour.

Basis of Recommendations
A draft of the evidence summary was provided to participants before the consensus meeting. All recommendations are based on a structured presentation and discussion of the best available evidence.

Preamble
The recommendations that follow are intended for patients who are fit and eligible for the doublet chemotherapy and biologic combinations being proposed. The recommendations provide a basis for discussion with patients about the management options for their mcrc and informed decision-making by patients about their care. Individual treatment plans will depend on appropriate patient selection and a complete discussion of the risks and benefits of proposed therapies with individual patients. Although improving os duration is an important goal of treatment, efforts to maximize quality of life by effective management of regimen-related toxicities and provision of appropriate supportive measures are essential.
Heterogeneity with respect to KRAS or RAS testing was observed in the evidentiary studies, but the recommendations provided here have been extrapolated to include only patients with wild-type tumours on extended RAS testing (KRAS or NRAS exons 2-4). In addition, although heterogeneity in the definition of right-sided and left-sided tumours was also observed, most studies defined rcc as arising from the cecum up to, but not including, the splenic flexure; and lcc as arising from the splenic flexure to the rectum. That definition is therefore the one that we propose in the implementation of the recommendations presented here.

First-Line Treatment
In the first-line setting, retrospective analyses of four of five trials show, for RAS wild-type lcc tumours, a statistically significant benefit in os duration associated with chemotherapy plus an egfr inhibitor compared with chemotherapy plus bevacizumab or chemotherapy alone. Furthermore, the results of two meta-analyses (including one focused on firstline studies) confirmed that association. In the analyses, the benefit of egfr mAb treatment in lcc is consistent, regardless of the egfr mAb or the chemotherapy backbone used. Conversely, in patients with RAS wild-type rcc tumours, the addition of egfr mAb to chemotherapy alone was not associated with an os duration benefit. In the meta-analysis of first-line trials comparing chemotherapy plus egfr mAb with chemotherapy plus bevacizumab for patients with rcc tumours, chemotherapy plus bevacizumab was associated with a statistically significant benefit in pfs and a statistically nonsignificant trend toward improved os duration. T he consensus meet i ng t herefore made t hese recommendations: 1. (a) In patients with RAS wild-type lcc, standard chemotherapy (folfox or folfiri) in combination with an egfr mAb (cetuximab or panitumumab) is recommended in the first-line setting. (b) In patients with RAS wild-type rcc, first-line egfr mAbs are not recommended. The combination of bevacizumab plus standard chemotherapy remains the standard of care for these patients. (c) Extended RAS testing should be available in a timely manner to allow for the appropriate selection of a biologic for first-line treatment decisions.

Second-Line Treatment
Data with respect to the use of egfr mAb in the second-line setting did not clearly demonstrate outcome improvements based on ptl. In the piccolo trial, tumour side was not a predictive factor for benefit with panitumumab. Moreover, the 20050181 trial did not demonstrate statistically significant differences between the folfiri plus panitumumab arm and the folfiri arm in right-or left-sided tumours. T he consensus meet i ng t herefore made t hese recommendations:

Third-Line Treatment
In the third-line setting, the only analysis of ptl comes from the co.17 trial. That analysis clearly showed that, in lcc, a benefit was associated with cetuximab compared with best supportive care. However, patients with rcc did achieve a 3-month improvement in os duration with cetuximab (although that gain was not statistically significant), with no difference in pfs.
The consensus meeting therefore made this recommendation: 3. All patients with RAS wild-type disease who have not previously been treated with an egfr mAb should be offered one.

Tumour Response
In contrast to the differential benefits associated with egfr mAbs by sidedness for os duration, egfr mAbs appear to produce a superior orr in both lcc and rcc, as shown in the Holch et al. 12 meta-analysis and the Arnold et al. 28 pooled analysis. That observation could suggest that, if tumour shrinkage is the primary initial goal of treatment (for example, in patients with high-bulk, symptomatic, unresectable disease), an egfr mAb combined with chemotherapy might be more effective than bevacizumab combined with doublet chemotherapy. However, the use of an egfr mAb with such a goal in rcc would have to be balanced with the effect on long-term outcomes (pfs and os duration). Given the conflicting data concerning the use of egfr mAb-based therapy in conjunction with surgical resection, the orr evidence is also insufficient for a recommendation favouring egfr mAbs in patients for whom conversion therapy for resection is the primary goal [38][39][40] . The selective use of biologics based on ptl for patients requiring maximal response based on the observed orr data requires further prospective study. The consensus meeting therefore made this recommendation: 4. At this time, in cases in which tumour response is the primary goal of therapy, the evidence is insufficient for the selective use of egfr mAbs based on ptl.

Future Research
The consensus meeting made these recommendations with respect to future research in this area: 5 (a) Primary tumour location should be factored into the design of future clinical trials in the treatment of RAS wild-type mcrc. (b) Given that ptl is a surrogate for more complex biologic mechanisms, ongoing research should seek to understand the patient-and tumourrelated factors that underlie the observed differential benefits of biologics based on ptl.

CONCLUSIONS AND PRACTICAL CONSIDERATIONS
In view of the opinions of the expert panel members, the results of the retrospective analyses discussed earlier in this article indicate that, although tumour sidedness might be a surrogate for more complex molecular mechanisms, it currently represents a practical predictive and prognostic biomarker that should guide treatment in patients with RAS wild-type mcrc.
Finally, the panel acknowledged that some practical considerations would accompany the consensus statements. The most important consideration is that implementation of the recommendations would necessitate a change in the current drug funding model for most provinces. Specifically, patients would require access to funded egfr mAbs in combination with chemotherapy in the first-line setting and to funded bevacizumab in the second-line setting (if they have not previously been treated with bevacizumab).

Scenario Recommendation
First line ■ For patients with left-sided disease, standard chemotherapy (FOLFOX or FOLFIRI) in combination with an EGFR mAb (cetuximab or panitumumab) is recommended in the first-line setting.
■ For patients with right-sided disease, first-line EGFR monoclonal antibodies are not recommended. The combination of bevacizumab with standard chemotherapy remains the standard of care for these patients.
■ Extended RAS testing should be available in a timely manner to allow for the appropriate selection of a biologic for firstline treatment decisions.
Second line ■ There is no evidence to recommend the selective use of EGFR monoclonal antibodies in the second-line setting based on the location of the primary tumour.
■ Patients who have not been treated with bevacizumab in the first line should be offered bevacizumab in combination with standard chemotherapy.
Third line ■ All RAS wild-type patients who have not previously been treated with an EGFR monoclonal antibody should be offered one.
Tumour response ■ At this time, there is insufficient evidence for the selective use of EGFR monoclonal antibodies based on primary tumour location if tumour response is the primary goal of therapy.
Future research ■ Primary tumour location should be factored into the design of future clinical trials in the treatment of RAS wild-type mCRC.
■ Given that primary tumour location is a surrogate for more complex biologic mechanisms, ongoing research should try to understand the patient-and tumour-related factors that underlie the differential benefit of biologics that have been observed based on primary tumour location. FOLFOX = folinic acid-5-fluorouracil-oxaliplatin; FOLFIRI = folinic acid-5-fluorouracil-irinotecan.