Differences in management and outcome for colon and rectal carcinoma with synchronous liver metastases: a population‐based cohort study

Abstract Aim Surgical treatment of colorectal cancer with synchronous colorectal liver metastases (SCRLM) can follow three different strategies with regard to the timing of liver resection. The aim of this study was to describe the selection of surgical strategy, focusing on differences between colon and rectal cancer with SCRLM, postoperative morbidity/mortality and survival. Method This was a retrospective population‐based study of patients with SCRLM registered in the Swedish Colorectal Cancer Registry in the Stockholm/Gotland region during 2010–2017 and treated with surgical resection of the primary tumour and liver metastases (LM). Patients were followed for 5 years or censored at 22 November 2018. Results A total of 238 patients met the inclusion criteria during the study period. Patients with rectal cancer were treated with the ‘liver first’ strategy in 70% of cases, whereas the main treatment strategies for colonic tumours were ‘simultaneous resection’ (44%) and ‘primary first’ (37%). Rectal cancer had a superior 5‐year survival rate compared with colon tumours with SCRLM (62 vs. 47%; p = 0.033). There was no difference in survival between treatment strategies irrespective of primary tumour location. Postoperative complications occurred most commonly among rectal tumours treated with simultaneous resection (p = 0.024). Conclusion Patients with rectal cancer and SCRLM were more often treated with the ‘liver first’ strategy than patients with colon cancer. Patients with rectal cancer and SCRLM where both primary tumour and LM were operated on had significantly better survival than corresponding patients with colon cancer.

ablation (RFA) of liver metastases has become an additional modality for selected patients and is applicable at several timepoints during treatment [5]. However, the optimal timing of surgery is still controversial, with the option of resecting the 'primary first' (the classical strategy), the 'liver first' or 'simultaneous resection'.
In the absence of conclusive international or regional guidelines, surgical strategy differs among patients and is decided at multidisciplinary team (MDT) conferences. The liver first strategy has been described to be suitable for rectal cancer and patients with a high liver tumour burden [6,7]. The preoperative treatment and treatment response differ in important aspects between colon and rectal cancer; rectal cancer patients often receive neoadjuvant (chemo) radiotherapy and have a greater tumour regression in response to therapy [8]. However, in the Swedish Colorectal Cancer Registry (SCRCR) the reported long-term survival is equivalent for colon and rectal cancer [9,10].
The present study aimed to describe the selection of surgical strategy for colorectal cancer with SCRLM in the Stockholm/Gotland region where both the primary tumour and liver metastases were resected. The study focuses on differences between colon and rectal cancer, postoperative morbidity/mortality and overall survival.

Study design and setting
This was a population-based cohort study that included all patients treated with surgery for colorectal cancer and SCRLM in the Stockholm/Gotland region (2.4 million inhabitants). The incidence of SCRLM in the region during the study period (2010-2017) was 17.3% of the total number of patients with colorectal cancer (1498/8673) (see the SCRCR database [11]). Exposure was the location of the primary tumour (colon versus rectum) and 5-year overall survival was the primary outcome. Colorectal cancer surgery is performed at five units and liver resections are centralized to Karolinska University Hospital. Patients were identified from the prospectively maintained national SCRCR database with a validated coverage of over 99% of colorectal cancer patients in the Stockholm/Gotland region [9,10,12]. The SCRCR is linked to the Swedish Tax Agency to obtain valid data on date of death. The National Quality Registry for Liver and Biliary Cancer (SweLiv), incorporating all patients in Sweden with primary and secondary malignancies of the liver, was used to identify patients who had resection of liver metastases. Ethical approval was obtained from the regional Swedish Ethical Review Authority (EPN 2018/925-31/2). This study has been reported in accordance with STROBE guidelines.

Participants
All patients registered in SCRCR/SweLiv and treated with segmental bowel resection and surgery for liver metastases in the Stockholm/Gotland Region from January 2010 to December 2017 were included. Patients with extrahepatic metastases, emergency bowel surgery or RFA of the liver metastases were excluded. The multimodal treatment strategy was discussed at MDT conferences and individualized. The categories of surgical strategy were defined according to the timing of surgery as 'primary first', 'liver first' or 'simultaneous resection'. Chemotherapy could be given before the first resection (neoadjuvant) and perioperatively (before and after the liver resection); radiotherapy for rectal cancer was given prior to rectal resection.

Variables and data sources
Epidemiological and oncological data were retrieved from SCRCR and SweLiv, respectively. Information was completed by medical chart review and postoperative complications with a Clavien-Dindo grade of 3a or above were included. The liver tumour burden score (TBS) was calculated from the number of metastases and the size of largest metastases [TBS 2 = (maximum tumour diameter in cm) 2 + (number of lesions) 2 ], previously described by Sasaki et al. [13]. The TBS score was categorized using <3, 3 to <9, ≥9 points as cut-offs [12]. Overall survival was calculated from the date of diagnosis and death from any cause was registered as 'event'. Follow-up was censored after 5 years or at 22 November 2018.

Statistical methods
Register data were analysed using Stata software version 15 Uni-and multivariable Cox regression models, stratified for the binary covariate age over 75 years, were applied to analyse the effect of covariates on overall survival. Conservative use of chemotherapy above the age of 75 and death unrelated to colorectal cancer or its

What does this paper add to the literature?
This population-based study is the first study to date that indicates important differences in both surgical strategy and survival for colon and rectal cancer with synchronous liver metastases. It reveals that rectal cancer with synchronous liver metastases has superior survival with liver first as the 'standard' surgical strategy. treatment were the reasons to stratify the Cox regression models.
Apart from location of primary tumour and surgical strategy, covariates that changed point estimates by more than 10% were included in the multivariable model. The proportional hazards assumption was assessed using a log-log plot of survival and Schoenfeld residuals, and no evidence was found to reject the proportional hazard assumption.

Participants and baseline characteristics
A total of 238 patients had surgery for both primary tumour and liver metastases in the Stockholm/Gotland region between 2010 and 2017. A flow chart of inclusion and surgical strategy is presented in Figure 1. Liver first was the surgical strategy in 39%, with simultaneous resection and primary first being less common, at 33% and 28%, respectively. Descriptive data on patient characteristics and cancer stage are presented in Table 1. The primary tumour location was the colon in 61%. Gender and American Society of Anesthesiologists (ASA) score were comparable between patients with their primary tumour in the colon and the rectum. Clinical TNM classification showed comparable categories for the primary tumour (p = 0.521), regional lymph nodes were negative in 33% of patients with colon cancer compared with 15% with rectal cancer (p = 0.001). In patients with a primary colon cancer, 75 (52%) of tumours were located on the right side or transverse colon and the remaining 69 (48%) were left-

Surgical strategy
Data on surgical strategy and the use of neoadjuvant chemo-and/or radiotherapy for colon and rectal primary tumours with SCRLM are presented in Table 2. The selection of surgical strategy differed significantly between primary colon and rectal tumours. A liver first strategy was applied in 70% of rectal cancer patients whereas patients with a primary colon tumour most commonly underwent a simultaneous resection (44%) or primary first strategy (37%) (p < 0.001) ( Data on the use of chemotherapy for primary colon and rectal tumours, respectively, are shown in Table 2. Patients received 5-fluorouracil (5-FU)-based chemotherapy, in general combined with oxaliplatin. Only a few (15%) received bevacizumab, an antivascular endothelial growth factor therapy in combination with 5-FU-based chemotherapy. The median total number of cycles was 12 for both groups (p = 0.103). Patients with rectal cancer received neoadjuvant chemotherapy more often than colon cancer patients (76% vs. 50%, respectively; p < 0.001). Neoadjuvant chemotherapy was given more often to liver first than to primary first and simultaneous resection patients.

Short-term postoperative outcomes
Length of stay (LoS) and other short-term postoperative outcomes are presented in Table 3.
Median LoS including both surgical procedures (when staged) for colon cancer patients was 13 days compared with 16.5 days for rectal cancer (p < 0.001). Postoperative morbidity, mortality, ICU stay and readmission rate did not differ between both groups, with an overall complication rate (Clavien-Dindo grade 3a or above) of 32% and 90-day mortality of 1.3%. The complication rate stratified on surgical strategy was higher for simultaneous resection (53%) than primary first (8%) and liver first (33%) in patients with rectal cancer (p = 0.033), but similar for the three surgical strategies in patients with colon cancer (p = 0.500) ( Table S1 in the online Supporting Information). A defunctioning stoma was more frequently used in colon cancer patients (60%) treated with the liver first strategy compared with other strategies (p < 0.001) (Table S1).

Overall survival
The median follow-up time was 42 months.

DISCUSS ION
This population-based study shows that for patients with rectal cancer and SCRLM a liver first strategy was applied in more than two thirds of patients whereas no clear pattern for strategy selection could be detected among patients with colon cancer and SCRLM. Furthermore, patients with rectal cancer were shown to have a significantly improved overall survival compared with the corresponding colon cancer patients with SCRLM, also in a TA B L E 2 Surgical strategy and radio-/chemotherapy for 238 patients with colorectal cancer and synchronous liver metastasis treated with surgical resection of both primary tumour and liver metastases  The surgical strategy chosen for rectal cancer patients with SCRLM was mainly a liver first strategy (70%), which has been suggested to be favourable by a consensus recommendation [7]. The liver first strategy has the benefit of giving rectal cancer patients longer to respond to treatment of the primary tumour and, in the presence of extensive liver disease, starting systemic and surgical treatment without delay for liver metastases [14]. This was reflected in the present study, where patients irrespective of primary tumour location, operated on with a liver first strategy had the highest median liver TBS and received most neoadjuvant chemotherapy (88%).
Liver TBS was shown in the univariate Cox analysis to be a significant predictor of survival. However, there was no difference in median liver TBS between patients with colon and rectal cancer. In Sweden most patients with rectal cancer and SCRLM are also treated with neoadjuvant radiotherapy delivered as a short course (5 × 5 Gy) after chemotherapy and immediately prior to liver surgery. Unlike for rectal cancer, recommendations for one strategy in colon cancer patients with SCRLM are absent [6]. Furthermore, it is difficult to see a clear pattern regarding strategy selection, even when taking ASA score, age, stage of primary tumour and extent of metastatic burden into account. Use of the liver first strategy is still limited for colon cancer patients with SCRLM in the present study and most patients were treated with simultaneous resection (44%) or the primary first strategy (37%).
Simultaneous resection has the benefit of a considerably shorter LoS for both colon and rectal primary tumours, as reported by other studies [15,16]. However, a higher risk of major complications for simultaneous resection has been debated, especially for rectal primary tumours and when the hepatic resection is major [17,18].
Irrespective of the surgical strategy selected, no major differences in postoperative complications were found when analysed for all patients. However, it is noteworthy that the highest rates of Clavien-Dindo complications ≥3a, readmission and reoperation were found in the small number of patients (n = 15) with rectal cancer who underwent simultaneous resection. Comparison of complication rates between strategies may be hampered by the fact that around one third of patients planned for staged resection progress or have complications after the first surgical procedure that prevent the second step of intended surgical treatment [19,20]. Despite these observations regarding complication rates, it appears that simultaneous resection with only one in-patient episode reduces the LoS.
Furthermore, for a fair comparison of complications it should be noted that a pretherapeutic defunctioning stoma, more commonly used in colon cancer patients treated with the liver first strategy, carries a risk of affecting quality of life as well as a risk of complications when it is reversed.
The 5-year overall survival for rectal cancer with SCRLM of 62% is encouraging and comparable to single-centre studies [21,22].
Previous studies have reported an inferior outcome in right-sided colorectal tumours compared with left-sided tumours including rectal tumours [4,23,24], and this has been attributed to a more aggressive tumour biology and later diagnosis [23,25]. In this study, we chose to report on colon and rectal cancer separately because of the dif- Meta-analysis of previous studies has not reported superior survival in any surgical strategy [26], and neither could this be seen in the present study. However, in one randomized controlled trial superior survival was seen for simultaneous resection when analysed for all patients starting treatment, partly explained by disease progression before liver resection in one third of patients assigned to a primary first strategy [27].

CO N FLI C T O F I NTE R E S T
LBO, CB, CV and PJN declare that they have no declarations of interest.

E TH I C A L A PPROVA L
Ethical approval was obtained from the regional Swedish Ethical Review Authority (EPN 2018/925-31/2).

AUTH O R CO NTR I B UTI O N S
PJN conceived of the presented idea and supervised the project.
LBO and CV collected the data, LBO conducted the data analysis and all authors interpreted the data. LBO and CB were responsible for the statistical analysis. LBO, CB, CV and PJN drafted and revised the manuscript. All authors reviewed the manuscript and approved the final version.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.

FU N D I N G I N FO R M ATI O N
No funding to declare.