Report from the 19th annual Western Canadian Gastrointestinal Cancer Consensus Conference; Winnipeg, Manitoba; 29-30 September 2017.

The 19th annual Western Canadian Gastrointestinal Cancer Consensus Conference (wcgccc) was held in Winnipeg, Manitoba, 29-30 September 2017. The wcgccc is an interactive multidisciplinary conference attended by health care professionals from across Western Canada (British Columbia, Alberta, Saskatchewan, and Manitoba) who are involved in the care of patients with gastrointestinal cancer. Surgical, medical, and radiation oncologists; pathologists; radiologists; and allied health care professionals participated in presentation and discussion sessions for the purpose of developing the recommendations presented here. This consensus statement addresses current issues in the management of colorectal cancer.


Participants
The wcgccc welcomes medical oncologists, radiation on cologists, surgical oncologists, pathologists, radiologists, gastroenterologists, and allied health professionals from western Canada who are involved in the care of patients with gastrointestinal malignancies (Table i).

Target Audience
The recommendations presented here are targeted to health care professionals involved in the care of patients with colorectal cancer (crc).

Basis of Recommendations
The recommendations are based on presentation and dis cussion of the best available evidence. Where applicable, references are cited.

QUESTION 1
What is the optimal duration of oxaliplatinbased adjuvant chemotherapy in nodepositive colon cancer?

Recommendations
■ For T1-3N1 disease, 3 months of capecitabine and oxaliplatin (capox) treatment is a reasonable option. If using 5fluorouracil (5fu), leucovorin, and oxaliplatin (folfox), 6 months should remain the standard. ■ In "high risk" stage iii (T4 or N2 disease), 6 months of oxaliplatinbased treatment is the standard of care.

Summary of Evidence
Treatment with 6 months of fluoropyrimidine-oxaliplatin has been the standard adjuvant treatment for stage iii colon cancer since the landmark mosaic trial 1 showed that the addition of oxaliplatin to 5fu improved 3year diseasefree survival (dfs) and 10year overall survival (os) 1,2 . However, oxaliplatin is associated with cumulative dosedependent neurotoxicity. Nearly all patients experience some degree of peripheral neuropathy during treatment, which can lead to functional impairment in nearly 45% of patients (31.4% grade 2 and 12.5% ≥grade 3). Peripheral neuropathy can persist long after treatment is complete and in many cases can be permanent, with 30% of patients continuing to experience neuropathy 12 months after completion of adjuvant therapy 3 . The International Duration Evaluation of Adjuvant Chemotherapy (idea) collaboration was designed to assess whether, to minimize toxicity, the duration of oxaliplatin based adjuvant chemotherapy could be reduced to 3 months from 6 months without compromising treatment efficacy 4 . A prospectively planned pooled analysis, idea considered six concurrent international trials involving 12,834 patients, including tosca, scot, idea France, horg, C80702, and achieve. The primary endpoint was the noninferiority of 3 months compared with 6 months of oxaliplatinbased adjuvant chemotherapy (capox or folfox) in stage iii colon cancer, with an upperbound noninferiority margin of 12% for dfs (hr: 1.12). All stage iii colon cancer patients randomized in the trials were in cluded in the primary analysis, but several of the trials included a second randomization to address additional research questions. For example, the Alliance C80702 trial included a randomization to compare celecoxib with placebo after completion of folfox, and the tosca trial had a secondary randomization to bevacizumab, which was eventually dropped. In addition, some trials included stage ii patients (tosca, scot, and horg), and scot also included rectal cancers, although the primary analysis focused on patients with stage iii colon cancer.
At the American Society of Clinical Oncology 2017 annual general meeting, the primary analysis of the idea collaboration was shown to be negative because the upper bounds of the confidence interval passed the acceptable prespecified noninferiority margin (dfs hr: 1.07; 95% ci: 1.00 to 1.15) 4 . The 3year dfs was 74.6% for the 3month course compared with 75.5% for the 6month course. De pending on treatment received (capox or folfox), grade 2 or greater peripheral neuropathy was reduced to 15%-17% for 3 months of therapy compared with 45% -58% for 6 months of therapy (p < 0.001).
The idea collaboration included multiple preplanned subgroup analyses. Specifically, the authors planned to assess subgroups stratified by N stage, T stage, chemo therapy regimen (folfox vs. capox), age (<70 years vs. >70 years), and tumour sidedness. When comparing outcomes based on regimen, patients who received folfox did not meet the cutoff for noninferiority (hr: 1.16; 95% ci: 1.06 to 1.26); those who received capox did (hr: 0.95; 95% ci: 0.85 to 1.06). The difference in outcome based on regimen was confirmed with an interaction test (p = 0.0051). Those results were driven largely by the scot trial, which con tributed more than half the capoxtreated patients to the full analysis and which also met its primary endpoint of demonstrating that 3 months of adjuvant oxaliplatinbased therapy was noninferior to 6 months of therapy (hr: 1.006; 95% ci: 0.909 to 1.114; p = 0.012) 5 .
Since the presentation of those data, there has been debate about how to interpret the stratification based on treatment regimen given that patients were not ran domized according to type of chemotherapy. There was no suggestion that the patients who received capox had more favourable baseline or disease characteristics. The increased efficacy of a capecitabinecontaining regimen might be related to the continuous administration of capecitabine for 2 of 3 weeks, to the earlier exposure to higher doses of oxaliplatin, or perhaps to something in herent to the pharmacology of capecitabine.
Similarly, in the x-act trial, adjuvant capecitabine and 5fu were shown to be equivalent for dfs (p < 0.0001) and os (p = 0.0001), but a trend toward superiority for capecitabine was observed for both endpoints (p = 0.068 and p = 0.060 respectively) 6 . Other data suggest that 3 months of continuous exposure to 5fu is superior to 6 months of bolus 5fu 7 , and retrospective data suggest that capox might be more active than folfox in stage iii colon cancer 8 .
With respect to the subgroup analyses based on T and N stage, no single T or N stage met the noninferiority criteria; however, the authors of the idea collaboration performed an unplanned analysis of lowrisk (T1-3N1) compared with highrisk (T4 or N2) groups. In that analysis, which pooled the capox and folfox groups, the result in the lowrisk group met the criterion for noninferiority (hr: 1.01; 95% ci: 0.90 to 1.12), but the result in the highrisk group was clearly inferior (hr: 1.12; 95% ci: 1.03 to 1.23). When the authors assessed those risk groups stratified by type of chemotherapy, neither capox nor folfox was noninferior in the highrisk group. In the lowrisk group, capox (hr: 0.85; 95% ci: 0.71 to 1.01), but not folfox (hr: 1.10; 95% ci: 0.96 to 1.26), was noninferior. The statistical plan for the trial did not include those low and highrisk groups as part of the prespecified analysis, and so the post hoc nature of those results should be interpreted with caution. The large sample size and pragmatic nature of the groupings do, however, provide support for the interpretation.
Taking into account all the evidence, the results of the idea pooled analysis support a riskbased approach to patients with nodepositive colon cancer. For patients with highrisk disease (T4 or N2), 6 months of oxaliplatinbased adjuvant therapy remains the standard of care. For patients with lowrisk disease (T1-3N1), 3 months of adjuvant capox treatment is a reasonable option to consider. If using folfox, 6 months of therapy remains the standard of care for all risk groups.

QUESTION 2
What is the optimal timing of surgery after neoadjuvant chemoradiation therapy in patients with clinical stage ii or iii rectal cancer?

Recommendations
■ The optimal timing of surgery after neoadjuvant chemoradiation therapy in patients with clinical stage ii and iii rectal cancer is not known.
■ Surgery should be considered between 6 and 10 weeks after completion of chemoradiation in patients with clinical stage ii or iii rectal cancer.

Summary of Evidence
In the German rectal cancer study that demonstrated a benefit with neoadjuvant chemoradiation, surgery took place 6 weeks after completion of the chemoradiation 9 . However, studies have suggested that a pathologic com plete response (pcr) after neoadjuvant chemoradiation is associated with improved dfs and os 10,11 . A recent meta analysis suggested that increasing the interval between completion of neoadjuvant chemoradiation and surgery to more than 6-8 weeks significantly increased the rate of pcr without any increase in detrimental outcomes 12 .
Since publication of that metaanalysis, other large ret rospective series have also demonstrated an increase in pcr rates with a longer interval from completion of chemoradiation to surgery [13][14][15] . Peak pcr rates were seen when surgery occurred between 9 and 12 weeks after com pletion of neoadjuvant therapy, although one study noted an increase in circumferential radial margin involvement if surgery occurred more than 12 weeks after completion of the chemoradiotherapy [13][14][15] . The greccar6 trial is the only prospective random ized controlled trial that was designed to evaluate the pcr rate with an increased interval between completion of neoadjuvant therapy and definitive surgery 16 . That multi centre trial randomized patients with clinical T3-4 or Tx nodepositive tumours of the mid or lower rectum to sur gery either 7 or 11 weeks after completion of neoadjuvant chemoradiation. No difference in the primary endpoint of pcr was observed between the two groups (15.0% in the 7week group vs. 17.4% in the 11week group, p = 0.5983). In both groups, the pcr rate was as high or higher than the pcr rates reported in retrospective studies. In terms of secondary outcomes, complete total mesorectal exci sion specimens were obtained more often in the 7week group (90% vs. 79%, p = 0.0156). Although no difference in anastomotic complications was observed, differences in medical complications were evident, mostly because of more urinary complications in the 11week group. A longer wait time between completion of neoadjuvant chemoradi ation and surgery was not associated with an increase in synchronous metastatic disease (2.3% in the 7week group vs. 1.5% in the 11week group, p = 1.00). The results of the study suggest that waiting longer between the completion of neoadjuvant therapy and surgery is not associated with an improvement the pcr rate and might be associated with increased morbidity; however, the results are far from in disputable. The optimal timing for surgery after completion of neoadjuvant chemoradiation for stages ii and iii rectal cancer remains unclear. Based on the data, surgery should be considered between 6 and 10 weeks after completion of chemoradiation.

QUESTION 3
What is the current role for nonsurgical management of rectal cancer after a complete clinical response (ccr) to neoadjuvant chemoradiation therapy?

Recommendations
■ The standard approach for patients after a ccr is definitive surgical resection. ■ If being considered for a nonsurgical approach, patients should be considered for a clinical trial if available. ■ In patients who do not undergo resection, an inten sive surveillance strategy is required. The case must be presented at a multidisciplinary case conference.

Summary of Evidence
Patients with clinical T3-4 or nodepositive rectal cancer are often treated with upfront chemotherapy and radia tion, followed by radical surgery and further postoperative chemotherapy 9,17 . A ccr after neoadjuvant chemoradio therapy occurs in 20%-30% of cases 18,19 ; however, rates of pcr after neoadjuvant chemoradiotherapy range from 10% to 20% 20,21 . Generally, patients who experience a pcr in response to induction therapy have a good longterm outcome 21 . The question therefore arises whether selected patients with a ccr might be spared surgical resection. A number of institutional reports provide insight into the longterm out comes of patients managed using a nonsurgical approach. The HabrGama group in Brazil published data for patients who received longcourse neoadjuvant chemoradiation for rectal cancer, 20% of whom had clinical T2N0 disease at baseline. Those who experienced a ccr, as determined by physical examination, endoscopy, and crosssectional imaging, were managed with close observation 22 . At almost 5 years of followup, recurrence rates were low (13%), and although 5% of recurrences were local, all were able to be managed with salvage surgery 23 . In patients managed with close observation, the 5year os was 93%, and the dfs was 85%. However, a recent publication showed that local recurrencefree survival was significantly worse in patients with baseline clinical T3-4 tumours than in those with clinical T2 tumours (69% vs. 96%, p = 0.009) 24 .
Other small case series have also shown promising out comes for patients managed with a nonsurgical approach. In a series of Dutch patients with a ccr (determined clin ically and endoscopically, and with magnetic resonance imaging and biopsy) who underwent close observation, the 3year local regrowthfree survival was 86% 25 . Patients from the Memorial Sloan Kettering Cancer Center who experienced a ccr were compared with a group of control patients who experienced a pcr: recurrence rates were low in both groups 26 . A propensity score-matched analysis of patients in Manchester who underwent chemoradiother apy followed by observation revealed no difference in the 3year os between those who underwent observation and those who underwent surgical resection. Of those who were observed, 34% experienced local regrowth, with 88% of those cases being amenable to surgical resection 27 .
A number of factors make it difficult to select can didates for nonoperative management from among the patients who achieve a ccr. The definition of a ccr differs in various reported series, and it is difficult to determine which patients with a ccr will also experience a pcr. In pa tients who experience a ccr as assessed by digital rectal ex amination and proctoscopy, up to 75% might have residual tumour found on pathology examination 19 . Currently no imaging modality has been shown to reliably predict pcr [28][29][30][31][32][33] . Also, in patients who do not undergo standard surgical resection, there is a possibility that curative sur gery might not be an option upon recurrence 34 .
Although the data for nonoperative management are promising, prospective clinical trials are necessary before such an approach can be recommended in patients who experience a ccr in response to neoadjuvant chemora diotherapy. The results of the prospective NCT01047969 trial (https://clinicaltrials.gov/ct2/show/NCT01047969) will add further information. For now, surgical resection remains the standard of care. If a patient declines surgical intervention or is medically unfit for major surgery, their case should be discussed in a multidisciplinary case con ference. Ideally, nonoperative approaches should occur in the setting of a prospective clinical trial, if available. In patients who opt for a nonsurgical approach, close and fre quent monitoring should occur. No guidelines discussing how to follow those patients have been published; however, an intensive surveillance strategy should include frequent history, physical examination, endoscopy, and cross sectional imaging (computed tomography and magnetic resonance imaging).

QUESTION 4
What molecular tests should now be the standard of care for patients with newly diagnosed crc?

Recommendations
■ Mismatch repair (mmr) testing should be performed in all crc patients for Lynch syndrome ascertainment and for predictive and prognostic factors. ■ Extended RAS and BRAF testing should be performed in patients with metastatic disease being considered for therapy. ■ Other biomarkers currently remain investigational.

Summary of Evidence
In crc, molecular testing can inform tumour classification, pathogenesis, prognosis, and prediction of response to specific therapies. Tumour mismatch repair (mmr) protein status, microsatellite instability (msi) status, and BRAF and KRAS/NRAS mutation status are the most commonly sought molecular tests for crc.
Testing for mmr allows for the identification of mmr deficient (dmmr) crc, either hereditary (Lynch syndrome secondary to germline mutations) or sporadic in nature. In crc, dmmr produces msi through dna replication errors, and msihigh (msih) crcs are associated with improved dfs and os 35 . Unlike Lynch syndrome-associated crc, sporadic dmmr tumours are associated with BRAF V600E mutations and typically present as rightsided tumours in older female patients 36,37 . Although clinical features and some tumour morphology features can help to identify dmmr and msih tumours, those approaches lack sensitivity. Immunohis tochemical (ihc) testing for mmr expression has become widespread in recent years and is currently used primarily for screening for Lynch syndrome. The standard 4antibody mmr ihc panel incorporates MLH1, MSH2, MSH6, and PMS2 testing and has sensitivity and specificity exceeding 90% for the detection of dmmr tumours 38 . Mismatch repair ihc testing is highly concordant with polymerase chain reaction-based msi testing 38,39 .
In select cases of earlystage crc, mmr status can provide information that assists in decisionmaking for adjuvant chemotherapy. Retrospective analyses suggest that singleagent fluoropyrimidine therapy might be inef fective in dmmr or msih crcs [40][41][42] ; mmr or msi status could therefore influence whether adjuvant therapy consists of observation alone, singleagent fluoropyrimidine therapy, or oxaliplatinbased doublet chemotherapy after surgical resection for stage ii colon cancers. Furthermore, recent data suggest that mmr and msi status can identify patients with metastatic crc who might benefit from immunother apy (see Question 5).
Selective, agedbased mmr testing is advocated as a costeffective process for Lynch syndrome screening, but only universal testing is able detect all sporadic dmmr or msih cancers 43 . Testing for mmr is therefore recommended in all patients with crc, because it helps to identify Lynch syndrome and provides prognostic 44 and predictive information.
In crc, BRAF V600E mutations are common; they are identified in approximately 10% of all crcs, 60% of spo radic dmmr tumours, and approximately 8% of advanced crcs [45][46][47][48] . A BRAF V600E mutation in metastatic crc is associated with worse progressionfree survival (pfs) and os, and a decreased response to antiegfr agents 37,46,47,49 . More than half of all crcs harbour mutations in KRAS, NRAS, and BRAF; only a small proportion of crcs have mutations in both RAS and RAF 50 . Improved pfs and os are expected for patients with wildtype (wt) KRAS and BRAF metastatic crcs treated with egfrtargeted therapies 51 . In the setting of mlh1-deficient tumours, the presence of a BRAF mutation excludes a Lynch syndrome-related cancer 44 . Testing for BRAF V600E and KRAS mutations can be performed by polymerase chain reaction. Recently, ihc clones that detect the BRAF V600E protein product have become commercially available. When fully optimized, ihc with the VE1 clone has sensitivity and specificity exceeding 90% for the detection of the BRAF V600E protein product 52 . Because the tests provide information about response to antiegfr therapies and screen for Lynch syndrome, it is recommended that patients with metastatic crc undergo testing for KRAS/NRAS and BRAF 44 . Other biomarkers remain investigational at this point in time.

QUESTION 5
What is the role of immunotherapy in patients with met astatic crc?

Recommendations
■ Use of a PD1 inhibitor (nivolumab or pembrolizumab) is a reasonable option in patients with stage iv msih or mmrdeficient crc after treatment failure with, or intolerance to, f luoropyrimidine, oxaliplatin, and irinotecan. ■ In mmrproficient tumours, singleagent nivolumab or pembrolizumab has been shown to be ineffective and should not be used.

Summary of Evidence
The distinct subset of msih crcs is distinguished by a hypermutable state, increased neoantigen load, and tumourinfiltrating lymphocytes. Although such tumours constitute only 3%-4% of all metastatic crcs, the foregoing characteristics suggest that they might respond to check point inhibitors. In 10 patients with dmmr crcs, Le et al. 53 demonstrated that the PD1 inhibitor pembrolizumab resulted in an overall response rate (orr) of 40% and a disease control rate (dcr) of 90%. Responses tended to be quick and prolonged. Pembroli zumab was well tolerated, with few grade 3 or 4 toxicities re ported. In patients with mmrproficient tumours, the response rate was 0%. Those results were confirmed in the keynote164 study, which showed an orr of 28% and a dcr of 51% in 61 patients with msih tumours treated with pembrolizumab 54,55 .
The PD1 inhibitor nivolumab has also shown prom ising activity. The CheckMate 142 study was a nonran domized phase ii trial of nivolumab in heavily pretreated patients with msih crc 56 . Patients in the first arm received nivolumab 3 mg/kg every 2 weeks, and patients in the sec ond arm received a combination of nivolumab 3 mg/kg plus ipilimumab 1 mg/kg every 3 weeks for 4 cycles, followed by nivolumab 3 mg/kg every 2 weeks until progression. In the 74 patients who received nivolumab monotherapy, the orr and dcr were 36% and 74% respectively 56 . The therapy was welltolerated, with few grade 3 or 4 toxicities. Recently, the results of the combination therapy arm were published, showing an impressive 55% orr and 88% dcr 57 . Grade 3 or 4 toxicities occurred in 32% of patients, with diarrhea, fatigue, and pruritus being the most frequent adverse events reported. The response rates with combination immunotherapy appear to be higher, but because of the nonrandomized nature of the trial, a direct comparison between the two arms is not possible.
In 2017, pembrolizumab and nivolumab were both approved by the U.S. Food and Drug Administration for patients with msih or dmmr metastatic crc experiencing progression after treatment with a fluoropyrimidine, ox aliplatin, and irinotecan. At the time of the present review, no hc Notice of Compliance had been issued for pembroli zumab or nivolumab in this setting.
To summarize, a small (3%-4%) subset of crc patients appears to derive substantial benefit from checkpoint inhibitors. The benefits are marked and therefore PD1 inhibitors should be offered to this select group of patients.

QUESTION 6
What are the indications for radioembolization in patients with metastatic crc?

Recommendations
■ There is no survival benefit with the use of radioem bolization in patients with metastatic crc, and its use should not be considered in the firstline setting unless part of a clinical trial.

Summary of Evidence
The prognosis of patients with crc and liver metastasis is determined largely by the degree of liver infiltration by tumour. The rationale for, and use of, liverdirected therapies such as surgical resection or ablation for the treatment of resectable or potentially locally curative liver metastases was, despite high rates of recurrence within the liver, established in populationbased cohorts and randomized controlled trials 58,59 . A recent randomized phase ii trial of patients with crc and unresectable liver limited metastases revealed that, at a median followup of 9.7 years, compared with chemotherapy alone, che motherapy plus local treatment with radiofrequency ablation, with or without resection, achieved a signifi cant improvement in os (hr: 0.58; 95% ci: 0.38 to 0.88, p < 0.01) 60 . However, that study had methodology limita tions that make the results difficult to interpret. The study was initially designed as a phase iii trial, with os as the primary endpoint. But accrual was slower than expect ed, and the study was amended to a phase ii trial with a smaller sample size and a primary endpoint of a 30month os rate exceeding 38% in the combinedmodality arm. Secondary endpoints included pfs and os. At the time of the initial report, a statistically significant difference in pfs was observed between the two arms (16.8 months for patients in the combinedmodality arm vs. 9.9 months in the chemotherapyonly arm, p = 0.025), but no difference in os 61 . Systemic treatment initially consisted of folfox4.
After 2005, bevacizumab was added, and in both arms, according to protocol, chemotherapy was required to be given for 6 months only. Although the results are encour aging, the study sample consisted of highly selected pa tients who experienced an impressive median os in both groups (45.6 months for the combinedmodality group vs. 40.5 months in the chemotherapyonly arm), and the use of locoregional therapies in patients with incurable crc remains unclear. The use of selective internal radiation therapy (sirt) that uses resincoated microspheres loaded with the thera peutic radioisotope yttrium90 has demonstrated encour aging results in terms of dcr 62 and depth of response 62,63 . Those results have been reiterated in small studies in the third line or in chemorefractory disease 64 , in the second line 65 , and in the earlier first line 66 for patients with liver dominant metastatic crc. The sirflox and foxfire family of studies was designed as a series of multinational trials to determine whether, compared with systemic chemothera py alone, radioembolization with yttrium90 resin micro spheres (sirSpheres: Sirtex Medical, Sydney, Australia) in combination with systematic chemotherapy resulted in an improvement in os in the firstline treatment of patients with metastatic crc with unresectable liver metastases 67 . Despite a higher proportion of patients achieving a tumour response and improved liverspecific pfs, the aggregated studies failed to achieve the primary outcome of improve ment in os with the addition of sirt to chemotherapy (hr: 1.04; 95% ci: 0.90 to 1.19; p = 0.61). The groups showed no difference in overall pfs or the rate of conversion to a sur gically resectable scenario. Notably, rates of conversion to surgical resection were lower than expected in both arms, possibly related to the high incidence of an intact primary (17% in the sirtpluschemotherapy arm vs. 16% in the chemotherapyonly arm) and the presence of extra hepatic metastatic disease 63 . The most common grade 3 or 4 adverse event was neutropenia, which occurred in 37% of patients who received sirt plus chemotherapy and in 24% of patients who received only chemotherapy. Increased rates of thrombocytopenia, abdominal pain, and fatigue were also seen with sirt plus chemotherapy compared with chemotherapy alone. Technical considerations related to the radioactivity dosing schema, biomarkers, catheter placement, and site of actual administration were not fully explored; however, by nature, the sirt procedure requires a high degree of technical expertise. Adverse events of gastrointestinal toxicities might have been attributable to a lack of experience on the part of the interventional radiologists performing the procedure in an earlier phase of the study, with no rollin period 68 .
A post hoc analysis of the sirflox and foxfire studies suggests that patients with rightsided primary tumours experienced an os benefit with sirt plus chemotherapy compared with chemotherapy alone (median os: 22 months vs. 17 months). The presence of a rightsided primary has been associated with worse prognosis 58,69 and therefore warrants further investigation.
Based on the data, the addition of sirt to chemother apy in the firstline treatment of unselected patients with metastatic crc is not recommended. Outside of a clinical trial setting, sirt should not be used in the firstline setting of metastatic crc with liver metastases. If sirt is being considered in the chemotherapyrefractory setting, the patient's case should be discussed in a multidisciplinary case conference setting.

QUESTION 7
Does sidedness matter in advanced colon cancer?

Recommendations
■ Rightsided tumours are genomically different from leftsided tumours. That difference can affect the use of biologic and chemotherapy strategies. Prognosis is poorer in rightsided tumours, which might not respond well to egfrdirected therapies.

Summary of Evidence
Cancers of the right side of the colon (cecum, ascending colon, hepatic flexure, and transverse colon) differ on a molecular level from cancers of the left side of the colon (splenic flexure, and descending and sigmoid colon). The molecular differences between right and leftsided crcs might be related to their different embryologic origins, given that the right colon originates from the midgut, and the left colon develops from the hindgut. It is possible that tumour location is simply a marker for molecular biology. BRAF and KRAS mutations and hyper methylation occur more commonly in rightsided tumours; leftsided tumours are more likely to be nonmutated 70,71 . Gene expression profiling has led to the identification of 4 consensus molecular subtypes (cms) of crc 72 : ■ cms1 tumours tend to display msi, CpG island methyla tor phenotype, hypermethylation, and BRAF mutation. ■ cms2 (canonical type) tumours have high somatic copy number alterations and Wnt and Myc activation.
■ cms3 tumours have low CpG island methylator pheno type and somatic copy number alterations, together with mixed msi status. ■ cms4 (mesenchymal type) tumours are associated with high somatic copy number alterations, angiogenesis, stromal infiltration, and activation of transforming growth factor β.
The cms type appears to be associated with tumour location and prognosis. The cms1 type tends to occur more frequently in women with rightsided tumours and are associated with worse prognosis upon relapse. In contrast, cms2 tumours tend to be leftsided and associated with better prognosis. The cms4 tumours are associated with worse relapsefree survival and os.
Retrospective analyses of randomized clinical trials confirm that tumour sidedness is associated with progno sis, even in RAS wt tumours. Among the 370 patients from the crystal and fire3 trials who had RAS wt metastatic crc, pfs and os were superior for the patients with leftsided tumours compared with the patients with rightsided tu mours, although the number of patients with rightsided tumours was small in both studies (33 and 28 respective ly) 73 . A number of other retrospective analyses 74,75 and metaanalyses 74-76 support the finding that tumour sided ness is prognostic.
Furthermore, recent data 75,76 suggest that tumour sidedness might be predictive, because patients with RAS wt leftsided tumours experience superior survival when an egfrdirected antibody is added to chemotherapy. The Cancer and Leukemia Group B/swog 80405 study compared the addition of cetuximab or bevacizumab to 5fubased chemotherapy, showing no difference in median os (30 months with chemotherapy-cetuximab vs. 29 months with chemotherapy-bevacizumab; hr: 0.88; 95% ci: 0.77 to 1.01; p = 0.08) 77 . However, when the data were retrospectively examined to assess the effect of primary tumour location in patients with RAS wt, those with leftsided tumours experienced a greater os benefit from chemotherapy-cetuximab than from chemotherapybevacizumab (median os: 36 months vs. 31.4 months), and those with rightsided tumours experienced better survival with chemotherapy-bevacizumab than with chemotherapy-cetuximab (median survival: 24.2 months vs. 16.7 months) 78 . A metaanalysis of randomized con trolled trials has since confirmed that, in patients with RAS wt leftsided colon cancer, survival is improved with the use of egfrdirected therapy plus chemotherapy compared with the use of bevacizumab plus chemotherapy. In con trast, patients with rightsided colon cancer do not receive the same benefit from egfrdirected therapy and seem to do better with chemotherapy-bevacizumab 76 . Although results have consistently suggested that tumour sidedness is predictive of response to egfrdirected therapy plus chemotherapy, the analyses have all been retrospective in nature and should be interpreted with caution. Prospective data are eagerly awaited.

ACKNOWLEDGMENTS
The wcgccc organizing committee thanks all meeting participants for their contributions to the development of this consensus statement. In addition, the committee thanks the meeting sponsors and Pascale Daigneault of Buksa Strategic Conference Services for support in organizing the meeting.
The 2017 wcgccc received unrestricted educational grants from Hoffmann-La Roche, Shire, Amgen Canada, Eli Lilly Canada, Celgene, Merck, ipsen Biopharmaceutical Canada, btg Inter national Canada, and Taiho Pharma Canada. During the entire process, the sponsors had no influence whatsoever over the de velopment of the guidelines, and they did not review or read the guidelines before submission. No author was compensated for their work on this article.