Resectability and resection rates of colorectal liver metastases according to RAS and BRAF mutational status: prospective study

Aki Uutela* , Arno Nordin, Emerik Osterlund, Päivi Halonen, Raija Kallio, Leena-Maija Soveri, Tapio Salminen, Annika Ålgars, Ari Ristimäki, Ali Ovissi, Annamarja Lamminmäki, Timo Muhonen, Juha Kononen, Raija Ristamäki, Eetu Heervä, Hanna Stedt, Kaisa Lehtomäki, Soili Kytölä, Jari Sundström, Markus J. Mäkinen, Lasse Nieminen, Teijo Kuopio, Mauri Keinänen, Pia Osterlund and Helena Isoniemi ; on behalf of the RAXO Study Group


Introduction
Resection of colorectal liver metastases (CRLMs) improves survival and may lead to cure. Resectability rates can be improved with conversion therapy 1,2 .
RAS and BRAF mutations are found in 50 and 5-20 per cent of tumours respectively in patients with metastatic colorectal cancer 3,4 . These mutations limit systemic therapy alternatives 5 and have been associated with worse outcomes in patients with CRLMs 6 . Patients with the BRAF V600E mutation clearly have shorter median survival, but some may survive without recurrence 7 .
Multidisciplinary teams (MDTs) have emerged to facilitate cooperation between medical specialties to ensure optimal care for the patient [8][9][10][11] . The aim of this study was to evaluate how RAS and BRAF mutational status affected resectability and conversion assessments performed by local hospitals and by a centralized MDT, and how this information could be used to improve resection rates and survival in patients with CRLMs.

Study design
RAXO was a prospective, investigator-initiated, nationwide Finnish study (NCT01531621, EudraCT 2011-003158-24) that included 1086 patients with metastatic colorectal cancer between 2012 and 2018. The main protocol 12 , liver metastases group 13 , and RAS/BRAF mutations in studies of metastatic colorectal cancer 14 have been published previously. This substudy included patients with known RAS/BRAF status and CRLMs. Further details are available in the supplementary material. Patients with non-V600E BRAF mutations were excluded. The patients were assessed as having liver-only metastatic disease or liver and extrahepatic disease at the time of inclusion in the study. The central MDT at Helsinki University Hospital tertiary centre evaluated each patient's technical resectability as described previously 12,13 and in the supplementary material. The mutational status was mostly known to the local team, but only occasionally to the central MDT. Patients were classified into the following resection outcome groups: R0-1, R2/local ablative therapy (LAT) or systemic therapy only. The study was approved by the Ethics Committee at Helsinki University Hospital and all patients provided written informed consent. Statistical methodology is presented in the supplementary material.
Upfront resectability and conversion rates in the central assessment of 354 patients with liver-only and 318 with liver and extrahepatic metastases are shown in Fig. 1. In the liver-only group, the central MDT considered the metastases to be upfront resectable in 48.0, 45.5, and 27.3 per cent of patients with RAS&BRAFwt, RASmt, and BRAFmt tumours respectively. Conversion rates for the borderline or unresectable liver-only group were 48. 4, 39.5, and 25.0 per cent respectively. Conversion rates for patients with initially borderline liver-only CRLMs were 78.9 per cent for RAS&BRAFwt (reference), 81.5 per cent for RASmt (OR 1.17, 95 per cent c.i. 0.42 to 3.32), and 40.0 per cent for BRAFmt (OR 0.18, 0.04 to 0.79) subgroups. The overall R0-1 resection rates for patients with liver-only CRLMs were 67.5 per cent for those with RAS&BRAFwt tumours (reference), 51.2 per cent for patients with RASmt tumours (OR 0.51, 0.32 to 0.80), and 31.8 per cent for those with BRAFmt tumours (OR 0.22, 0.09 to 0.60). The influence of tumour location on conversion is shown in Table S2.
Patients with liver and extrahepatic RAS&BRAFwt and RASmt metastases had similar upfront resectability and conversion rates. There was, however, a difference in R0-1 resection rates between the RAS&BRAFwt (12.6 per cent; reference) and RASmt (4.9 per cent; OR 0.35, 0.15 to 0.87).
When patients with liver-only disease were considered to have upfront resectable tumours by the central MDT, the local hospital underestimated resectability in 39, 41, and 83 per cent for RAS&BRAFwt, RASmt, and BRAFmt tumours respectively (Fig. 2a). If the central MDT considered a patient to have borderline resectable disease, 16, 15, and 0 per cent respectively of the local assessments were scored as never resectable.
Among patients with liver and extrahepatic metastases considered upfront resectable by the central MDT, the local teams underestimated resectability in 69 and 53 per cent of those with RAS&BRAFwt and RASmt tumours respectively (Fig. 2b). The rate of underestimation for borderline liver and extrahepatic metastases was 31 per cent for RAS&BRAFwt and 50 per cent for RASmt tumours. Reasons for not resecting technically resectable metastases are listed in Table S3.
Forty-two patients (6.3 per cent) had CRLMs that the local team considered never resectable. These were considered upfront or borderline resectable in central assessment, and 28 became technically resectable. Nine of these patients underwent resection with curative intent, including six with RAS&BRAFwt tumours (5 with liver-limited and 1 with liver and extrahepatic metastases) and three with RASmt tumours (2 liver-limited, and 1 liver and extrahepatic metastases).
Median overall survival (OS) after the first resection of metastases for 197 patients with liver-only metastases who underwent R0-1 resection was 82, 73, and 28 months according to RAS&BRAFwt (reference), RASmt (HR 1.55, 0.91 to 2.65), and BRAFmt status (HR 7.24, 2.38 to 22.00) respectively (P < 0.001). Corresponding 5-year OS rates were 68, 60, and 0 per cent respectively (Fig. S1). For 22 patients with RAS&BRAFwt and RASmt status who underwent R0-1 resection of liver and extrahepatic metastases, median OS was 79 and 71 months respectively (P = 0.847). Corresponding 5-year OS rates were 79 and 88 per cent. Recurrence-free survival, survival In multivariable analysis of risk factors for OS, assessment as unresectable by the central MDT appeared to be a strong risk factor. The second most notable factor was mutational status ( Table 1).

Discussion
With the help of centralized multidisciplinary assessment, high resectability, conversion, and resection rates are achievable for patients with RAS&BRAFwt and RASmt CRLMs. Selected patients with unfavourable BRAF mutation or with extrahepatic metastases may even undergo potentially curative resection.
Prospective studies of highly selected patients, often with RASwt tumours, have reported a conversion rate of 44-64 per cent and secondary resection/LAT rates of 44-61 per cent for patients with initially borderline or unresectable CRLMs 2,15-17 . In patients who also underwent hepatic artery infusion as induction therapy, conversion and secondary resection/LAT rates were 32 per cent for patients with RAS&BRAFwt tumours, 39 per cent for those with RASmt lesions, and 0 per cent for those with BRAFmt disease 18 . In a retrospective neoadjuvant therapy response assessment 19 , disease in up to 53 per cent of patients, mostly with RASwt tumours, was considered resectable upfront or after conversion, but only 29 per cent of these patients actually underwent resection. The present prospective study has shown comparable secondary resection/LAT rates for patients with liver-only RAS&BRAFwt and RASmt metastases, and a secondary resection/LAT rate as high as 25 per cent for patients with BRAFmt tumours.
In the literature, resectability and resection rates range from 18 to 71 per cent and from 16 to 54 per cent respectively for patients with liver-only metastases 1,9,19,20 . This study has shown that the chance of curative resection is highest for liver-only RAS&BRAFwt metastases, then RASmt metastases. Even for patients with tumours harbouring a BRAF mutation, a resectability rate of 45 per cent and corresponding resection rate of 32 per cent provided at least a chance of prolonged survival. Other groups have reported up to 44 per cent resectability, but resection rates of only 5-11 per cent for patients with multiorgan metastases 19,20 . The present study has shown that such patients are indeed less likely to undergo resection. After conversion therapy, curative resection of all diseased organs could still be completed in 13 and 5 per cent of patients with RAS&BRAFwt and RASmt tumours respectively.
Reported disagreements between surgeons assessing resectability of 35-52 per cent, including 7-11 per cent major disagreements, have stressed the importance of a multidisciplinary team 9,15 . Disagreement between local teams and central MDT was considerable in the present study, with most major disagreements relating to borderline resectable or extrahepatic disease, suggesting that patients with more advanced disease could benefit even more from centralized MDT assessment. In multivariable survival analysis, central MDT assessment was associated with survival. This indicates that the central MDT is capable of including important clinical and radiological information in their decision, and underlines the potential additional value of multidisciplinary assessment of patients with liver-only or liver-dominant CRLMs.
Collaborators investigators, study personnel, and hospitals that participated in this study. The authors acknowledge C. Österlund for preparation of figures and S. Field, medical writer at Meducom BV, for English language editing of the manuscript. C.Ö. and S.F. were compensated for their support by grants. This paper reports subgroup study results of the preregistered RAXO study (NCT01531621, EudraCT 2011-003158-24). The preregistration can be accessed at https://clinicaltrials.gov/ct2/ show/NCT01531621. Understanding of the significance of molecular pathology has increased greatly since the RAXO study was originally conceived. The analysis plan for molecular pathology has been developed partially during the course of the study, and can be found in the RAXO Protocol version 3.2, dated 7 May 2017 12 .