A comparison of the simultaneous , liver-ﬁ rst , and colorectal-ﬁ rst strategies for surgical treatment of synchronous colorectal liver metastases at two major liver-surgery institutions in Sweden

Background: The optimal treatment strategy for patients with synchronous colorectal liver metastases (CRLM) is unclear. The aim of this study was to compare the outcome of the simultaneous , liver-ﬁ rst , and colorectal-ﬁ rst surgical approaches. Methods: All consecutive patients who had been resected with curative intent for CRLM were included. A Cox regression model was constructed, and an intention-to-treat analysis was performed between the liver-ﬁ rst and the simultaneous approaches, after propensity score matching. Results: 658 patients were included in the analysis. 92 patients had a simultaneous resection, 163 patients had liver-ﬁ rst , and 403 patients had a colorectal-ﬁ rst approach. Overall survival was 54.9 months (95% CI 39.2 – 70.4) in the liver-ﬁ rst group, 54.5 months (95% CI 46.8 – 62.3) in colorectal-ﬁ rst group, and 59.6 months (95% CI 42.2 – 77.0) in the simultaneous group (log-rank p =0.850). In the matched cohort there were no differences in Clavien-Dindo 3a (p = 0.992) or 3b and greater (p = 0.999). Median overall survival was for liver-ﬁ rst group 42.2 months (95% CI 26.3 – 58.2), and for the simultaneous group 56.2 months (95% CI 47.1 – 65.4) (strati ﬁ ed log-rank p = 0.455). Conclusion: A simultaneous approach was not associated with worse overall survival or morbidity compared to a liver-ﬁ


Introduction
Colorectal cancer is the second most common cause of cancerrelated deaths in the Western world. The most common site of metastasis is the liver. Approximately 15-25% of patients have colorectal liver metastases (CRLM) at the time of diagnosis. [1][2][3] The optimal treatment strategy for patients with synchronous CRLM is unclear. Many of these patients often present with initially unresectable disease, 4 and have worse cancer biology, 4 with shorter disease-free survival and overall survival. 5,6 In brief, there are three established treatment strategies for patients with CRLM where curative intent is deemed feasible. These include, the simultaneous approach or a staged approach (either liver-first, or colorectal-first approach).
In the simultaneous approach the liver and bowel resections are performed at the same time. Whereas in a staged approach, either the primary cancer is resected first (colorectal-first approach) or the liver metastases (liver-first), with a period of recovery between the two resections. In the only published randomized controlled trial on the topic, colorectal-first approach was compared to the simultaneous approach. 7 Inclusion criteria were resectable synchronous CRLM, and primary endpoint was major postoperative complications. The study could not find any differences in complication rates between the groups. Given the complexity of patients with synchronous colorectal liver metastases, some patients never receive all the intended treatment plan. In a study by Sturesson et al., roughly two-thirds of all patients completed both the liver and colorectal resections. 8 Arguments for the colorectal-first approach, include early resection of the source of metastases (primary cancer), as well as a reduction of the risk of bowel obstruction, perforation or bleeding. Since these patients often commence chemotherapy treatment, this may help to identify those patients who progress during treatment, and who may not benefit from a liver resection. 4,6,9 Arguments for the liver-first approach include early resection of liver metastases, which are known to be the drivers of prognosis. In addition, many patients with synchronous CRLM have an advanced hepatic involvement where resectability is borderline, or where surgery has been made possible thanks to chemotherapy. In these instances, the liver-first approach seems favorable, since it makes use of the window of opportunity for curability. 10 Benefits of the simultaneous approach include definitive treatment of both the primary cancer and the metastases at the same time; an important incentive given that about one-third of patients fail to receive the intended treatment. 8 Additionally, it may entail only one anesthetic induction and overall likely shorter hospital length of stay. 11 Prima facie, the simultaneous approach appears attractive; with a potential to offer prolonged survival, shorter hospital stay, and reduced healthcare costs. The aim of the present study was to examine overall survival and morbidity and mortality across the simultaneous, liver-first, and colorectal-first approaches.

Methods
Between 2005 and 2015, 703 consecutive patients with synchronous CRLM treated with liver resection with curative intent were identified at Karolinska University Hospital (Stockholm) and Uppsala University Hospital. Since data were retrieved from local liver-surgery data bases all patients had received liver surgery, but not necessarily the colorectal resection. Consequently, a few of the liver-first patients in the data bases never had all the intended treatment (both the colorectal and liver resections). Therefore, to create three comparable groups, the initial analysis included only patients that had received both the liver resection and the primary cancer operation. Forty-five patients in the liverfirst group were thus excluded since they never received a colorectal resection. A flow-diagram of all patients is provided in Fig. S1 (Supplementary Material). In order to make an intentionto-treat analysis a subgroup analysis was performed that compared the simultaneous and liver-first approaches. This analysis included all liver-first patients regardless of whether they had received both the liver-resection and the colorectal resection. Demographic and clinicopathological data were collected, including data on age, sex, center of surgery, body mass index (BMI), and date of eventual death. Data were also collected on tumor characteristics, such as location of tumor, R0/R1 resection margin after liver surgery, maximum size and number of metastases, type of colorectal and liver surgery (minor, major or extended) were collected. The study was approved by the regional Ethical Review Boards (Dnr 2010/1872-31, 2018/086).

Pre-operative assessment and treatment
All patients were discussed in both colorectal and hepatobiliary multidisciplinary treatment (MDT) conferences, with oncologists, radiologists, hepatobiliary and colorectal surgeons present. Treatment strategy decisions were made during MDT conferences and individually tailored for each patientwhich involve both assessing the patient's operability and the tumors' resectability. If a patient required an extensive liver resection with an increased risk of post-operative liver failure; a volumetric assessment of future liver remnant (FLR) was calculated based on radiological imaging. If FLR was deemed insufficient based on the patient's body weight, either portal vein embolization (PVE), or associating liver partition and portal ligation for staged hepatectomy (ALPPS), was considered. At our centers PVE was the standard technique used for augmenting FLR up until 2012, thereafter ALPPS has been used in selected patients.
The treatment for synchronous CRLM is highly complex, with often several competing potential treatment strategies. To determine, which treatment strategy to use several factors were assessed. Are the liver metastases resectable or potentially resectable with chemotherapy? What type of liver surgery is required (extensive or minor)? Is the primary cancer advanced, and is there a high risk of anastomotic leakage, or is the tumor easily resectable? Does the patient suffer from symptoms from the primary cancer, and will the patient benefit from chemotherapy?
Evidently, quite few patients will be candidates for the simultaneous approach, since this strategy is often only warranted in medically fit patients, with less extensive liver metastases, and a potentially easily resectable primary cancer. Liver-first approach was often considered in patients with extensive liver metastases in need of downstaging chemotherapy with an asymptomatic primary cancer; and in patients with a rectal cancer whom received chemoradiation. Colorectal-first was mainly considered in patients with an advanced and/or symptomatic primary cancer, with either limited or extensive liver metastases. In general, most patients received peri-operative chemotherapy, that comprised chemotherapy three months before and three months after surgery. 12 During neo-adjuvant chemotherapy, contrast-enhanced computed tomography (CE-CT) scans of the liver and thorax were performed every 3-4 cycles to assess response. 13 If no progression of the disease was radiologically evident, and the liver metastases were still deemed resectable a laparotomy was performed four or five weeks after the last course of systemic chemotherapy.

Follow-up
Patients were followed-up at the hepatobiliary and colorectal outpatient clinics four to six weeks after surgery. Thereafter, patients were followed-up according to the national guidelines for colorectal cancer with liver metastases, which as a minimum entail a yearly contrast-enhanced computed tomography of the abdomen and thorax.

Statistical analysis
Descriptive statistics are expressed as median (range) and interquartile ranges (IQR), while the categorical variables are presented with proportions and percentages. Categorical variables were compared using the Chi-square test, for continuous variables either the Kruskal-Wallis test (if more than two groups) or Mann-Whitney U test (if two groups) were used. When more than two cells were compared for categorical variables, a z test for column proportions for each row in the Chi Square contingency table, was calculated. Thereafter the z tests were adjusted using Bonferroni correction. Survival analysis was performed using the Kaplan-Meier method, and the proportional-hazards assumption was assessed. If the proportional-hazards assumption was not violated, the log-rank test was used to assess difference between groups. When the proportional hazards assumption was violated, the Peto-Peto test was used to assess differences. 14 The proportional hazards assumption was tested based on Schoenfeld residuals. Overall survival was defined as the time period between date of liver surgery and death from any cause. Time was censored at the last follow-up for patients that were still alive. To compare survival differences between groups after propensity score matching, the stratified log-rank test was used. 15 In the matched cohort a paired t test was used for continuous variables, and McNemar or McNemar-Bowker for categorical variables. 16 Median follow-up period was calculated using reversed Kaplan-Meier method. 17 In the Cox model, backward elimination was used to assess the relationship between relevant clinico-pathological variables, and overall survival. Entry level in the multivariable analysis was set at a p = 0.05, and removal at 0.10. A subgroup intention-to-treat analysis was performed that compared simultaneous and liverfirst resections. To minimize imbalances between the groups propensity score matching was used in a 1:1 ratio, using a multivariable logistic regression model. The variables included in the model were age, ASA group classification, primary tumor stage, primary nodal status, type of liver resection (minor versus major/extended) and number of colorectal liver metastases. The caliper was set equal to 0.020, and matching was performed without replacement. The total propensity scores of each group were graphically compared with Kernel density distribution plots. The degree of imbalance between each covariate was assessed with standardized mean difference (SMD), which is the difference in the mean of a variable between two groups divided by the estimated standard deviation of that variable. 16 Values of p < 0.05 were considered statistically significant. Data analyses were performed using IBM SPSS Statistics Version 28, 2021, and STATA/SE version 15.1 (StataCorp, College Station, Texas, USA).

Results
A total of 658 consecutive patients were included in the analysis. Ninety-two patients were treated with the simultaneous approach, 163 patients with liver-first, and 403 patients with the colorectal-first approach. Baseline characteristics of all patients included in the study are provided in Table 1. Median age at liver surgery for the whole cohort was 65.3 years; 61.1% of the patients were men, and median Body Mass Index (BMI) was 25.1 kg/m 2 . In the aforementioned variables there were no differences between the groups.
Differences between the three treatment groups were discerned in the location of the primary cancer (p=<0.001). In the liver-first group, 66.3% (n = 108) had a primary rectal cancer, compared to 26.1% (n = 24), and 27.6% (n = 107) in the simultaneous and colorectal-first groups, respectively. Right-sided colon cancer in the simultaneous group occurred in 35.9% (n = 33), compared to 23.3% (n = 94) in the colorectal-first group, and 5.5% (n = 9) in the liver-first group. In addition, there was a difference in the number of liver metastases between the groups (p = 0.018). In the simultaneous group, 70.7% (n = 65) of patients had one or two liver metastases, compared to 47.9% (n = 78) of patients in the liver-first group. In the latter group, 21.4% (n = 86) had five or more metastases. Overall, most patients had a primary cancer staged T2-3 (62.3%), or T4 (24.2%). Lymphatic spread (N1/N2) occurred in more than fifty percent of cases in the simultaneous and colorectal-first groups, compared to 38.6% in the liver-first group. Overall, roughly 90% of patients belonged to either ASA classification 2 or 3. Only four patients were classified as ASA 4.
A post-operative complication, classified as Clavien-Dindo 3a occurred in 15.2%, 12.3%, and 16.6% in the simultaneous, liver-first, and colorectal-first groups, respectively (Table 2). In the simultaneous group 15.2% suffered from a Clavien-Dindo 3b or greater compared to 9.8% in the liver-first group. There were no differences in Clavien Dindo 3a (p = 0.429), Clavien-Dindo >3a (p = 0.334), nor mortality (p = 0.686). The overall 90-day mortality rate for all treatment groups, after liver-surgery (or colorectal and liver surgery combined, as is the case in the simultaneous group) was 1.2%.  For each patient a tumor burden score was calculated, which takes into account the number of tumors and the largest size of tumor. 18 Furthermore, patients were stratified as low risk, medium risk, and high risk, according to a Composite Score, a validated predictive tool to calculate overall survival, recently published by the authors of the present study. 19 The Composite Score takes into account age at surgery, c-reactive protein, serum albumin levels, embryonic origin of primary cancer, and whether it was an extended liver resection (>4 liver segments). There were no differences in distribution of either tumor burden score (TBS) 18 or Composite Score between the treatment groups. A majority of patients belonged to TBS 2, or medium-risk according to the Composite Score. In the liver-first and simultaneous groups approximately, three per cent belonged to highrisk, whereas six percent of patients in colorectal-first belonged to this risk category ( Table 2). Overall survival according to the Composite Score is provided in Fig. S2, Supplementary Material.

Prognostic clinicopathological factors
In the univariable Cox regression analysis age over 70 years (HR = 1.43, 95% CI 1. 16 A subgroup intention-to-treat analysis of the simultaneous and liver-first approaches. To compare the simultaneous approach to the liver-first approach in an intention-to-treat analysis, a propensity score matching was performed. In this analysis all liver-first patients were included, that is, also those who for various reasons never had the intended colorectal resection (Table 4). Before matching, a comparison of overall survival between liver-first simultaneous approach is provided in Fig. S3 (Supplementary Material). After matching in a 1:1 ratio, two groups with 58 patients in each  Table 4. After matching no differences could be discerned in the matched variables, that is, age, number of liver metastases, ASA classification, type of liver resection, primary cancer T and N stages.

Discussion
The present study describes a more than a decade experience in two large hepatobiliary centers of three different treatment strategies for synchronous CRLM. The study compares the outcome for patients treated with the simultaneous, liver-first, and colorectal-first approaches, and who had all the intended surgical treatment. Overall survival for patients who underwent a simultaneous approach were not found to live shorter than patients who underwent either the liver-first or colorectal-first approaches (Fig. 1). Moreover, in a subgroup intention-to-treat analysis between the liver-first and the simultaneous groups, overall survival for patients in the simultaneous group, was found to be similar to patients in the liver-first group (Fig. 2). Morbidity and mortality were not found to be any higher in the simultaneous group compared to a staged approach ( Table 2).
Determining the best surgical strategy for each individual patient with CRLM is complex. A number of important factors need to be considered such as location and extent of primary tumor and liver metastases, patient performance status and presence of symptoms, and underlying co-morbidities. All patients are not suitable for all treatment options. To exemplify, at our institutions a simultaneous resection is primarily a potential strategy in the setting of a fit patient, with limited liver metastases (with the aim of parenchymal-sparing approach), with a low-risk primary cancer. By contrast, a patient with borderline resectable liver metastases that require conversion chemotherapy, and a high-risk primary rectal cancer, is more likely to benefit from a liver-surgery approach. Whereas, a patient with a symptomatic primary in the form of bleeding or signs of obstruction, may benefit from a colorectal-first approach. Hence, since patient selection between the treatment strategies differs from the outstart, any comparison of morbidity, mortality and overall survival, between these treatment strategies is subject to a risk of bias. A testament to the complexity involved in determining treatment strategy and the heterogeneity of patients, is the recent multicenter randomized controlled study (RCT) published by Boudjema et al. 7 This is the first RCT on the topic, which randomized patients with synchronous initially resectable CRLM at ten French tertiary centers to either the simultaneous or colorectalfirst approaches. The primary endpoint was major complications within 60 days following surgery, and secondary outcome overall survival and disease-free survival. After ten years of accrual only 220 patients were assessed for eligibility, and 105 patients were deemed suitable to be randomized. The final analysis included 39 patients in the simultaneous group, and 46 patients in the colorectal-first group. The study found no differences in major complications between the two treatment groups, but patients in the simultaneous group had a shorter hospital stay compared to the colorectal-first group (12 versus 17 days, p = 0.002). Median overall survival in the simultaneous group was 5.9 years, compared to 3.9 years in the colorectal-first approach (p = 0.07).
A number of retrospective studies have been published, that compare simultaneous versus staged resection. [20][21][22] In a study by Slesser et al., no differences in post-operative complications or mortality, nor three-year overall survival were found between the simultaneous and staged approaches. Median length of hospital stay was shorter in the simultaneous group compared to staged resection (14 days compared to 18.5 days, p = 0.03). 20 In a more recent study by Abelson et al., similar results were found. There was no difference in major complication rates between simultaneous and staged approach, and there was a shorter length of hospital stay and reduced healthcare costs. 21 In the present study, which included hospital stay after liver surgery only, patients in the simultaneous group had a median hospital stay of 12 days compared to 10 days in the staged approach. One limitation of this comparison, however, is that it does not include the length of hospital stay after the primary cancer operation. There were no differences in R0/R1 resection margins (p = 0.111) between the treatment groups. In the present study R0 resection margin was obtained in 74.5% of patients (n = 490), and R1 resection in 15.8% (n = 104) of the patients, which is similar to the results reported by Silberhumer and co-workers 23 .
Similar to the results of the present study Mayo et al., 22 reports no differences in morbidity, mortality or overall survival between simultaneous and staged approaches. Furthermore, several reviews and meta-analyses have been published. 11,[24][25][26][27] Three of these meta-analyses report a lower rate of morbidity in the simultaneous group compared to patients undergoing staged resection, [25][26][27] whereas two could not discern any difference. 11,24 Overall survival was similar across all studies and treatment strategies, whereas total length of hospital stay was found to be shorter among patients in the simultaneous group. 11,24,27 As expected patients with more extensive metastatic liver disease, that required a major hepatectomy were less commonly offered a simultaneous approach, 11,25,26 and staged resection appeared to be reserved to patients with advanced age and locally advanced rectal disease. 24 The present study also found that patients in the simultaneous group had fewer liver metastases, and underwent less extensive liver surgery. Strengths of the present study include its large study population. Moreover, it is, to our knowledge, one of the largest studies, to examine all three treatment strategies, without grouping colorectal and liver-first into a staged approach. Furthermore, this study provides the longest reported median follow-up period of 104 months (IQR 97-112 months). In addition, to address baseline imbalances between the simultaneous and liver-first approaches, a subgroup intention-to-treat analysis was performed after propensity score matching. After matching overall survival, morbidity and mortality, were similar in the simultaneous group compared to the liver-first approach.
Limitations of the present study include its retrospective design. Furthermore, since the analysis relies on surgery-based databases at two hepatobiliary tertiary centers, we were not able to perform an intention-to-treat analysis of all patients (including colorectal-first). We have only been able to analyze patients that have been referred to our MDT conference, and have had liver surgery. Patients who underwent a colorectal-first approach, but whom, for various reasons, such as disease progression, side-effects of the chemotherapy, post-operative complications after treatment of the primary cancer) never became subject to a liver resection, could therefore not be included in the analysis.
Patients with synchronous CRLM are a heterogenous group that can be treated with a staged approach (liver-first, or colorectal-first) or a simultaneous approach. The choice of treatment strategy is highly complex, and patients benefit from an individualized approach. The present paper could not find that the simultaneous approach was any worse than a staged approach in terms of morbidity, mortality or overall survival.