Clinical outcomes of neoadjuvant chemotherapy for resectable colorectal liver metastasis with intermediate risk of postoperative recurrence: A multi‐institutional retrospective study

Abstract Aims Risk‐scoring systems for colorectal liver metastasis (CRLM) after hepatectomy allow prognoses to be predicted preoperatively. We investigated the clinical outcomes of neoadjuvant chemotherapy for resectable CRLM according to patient risk status, aiming to determine the subgroup of patients who could benefit from neoadjuvant chemotherapy. Methods In this multi‐institutional retrospective analysis, the preoperative risk score was calculated from six previously reported factors: synchronous metastases, primary lymph node positivity, tumor number, largest tumor diameter, extrahepatic metastasis, and the preoperative carbohydrate antigen 19–9 level. Patients were divided into three groups according to their risk scores: low risk (score = 0), intermediate risk (score 1–10), and high risk (score ≥11). Overall and recurrence‐free survival curves were calculated using the Kaplan–Meier method. After propensity‐score matching in the intermediate‐risk group, we compared clinicopathological features and outcomes. Results There were 318 cases, from 20 institutions. The preoperative risk score could be calculated in 277 cases. There were 34, 192, and 51 patients in the low‐, intermediate‐, and high‐risk groups, respectively. Intermediate‐risk group patients who received neoadjuvant chemotherapy had significantly better recurrence‐free survival than that of patients without neoadjuvant chemotherapy (P = .0453). After propensity‐score matching in the intermediate‐risk group, the recurrence‐free survival rate was better in patients who received neoadjuvant chemotherapy (P = .0261). But the overall survival rate was not improved after the matching. Conclusion Neoadjuvant chemotherapy for resectable CRLM might prolong the recurrence‐free survival period for intermediate‐risk patients with preoperative risk scores in the range of 1–10, but the overall survival was not improved by neoadjuvant chemotherapy.


| INTRODUC TI ON
Colorectal liver metastasis (CRLM) is a major cause of death and case numbers are increasing worldwide. Hepatic resection is the only treatment for CRLM with curative potential, and the 5-y survival rate after hepatectomy is reported to be around 50%. 1,2 However, the recurrence rate after hepatectomy is still high, and the most frequent site of tumor relapse is the remnant liver. [3][4][5] In recent years, neoadjuvant chemotherapy has been applied to several types of malignancy and has improved patients' prognosis. 6 Neoadjuvant chemotherapy has been shown to prolong recurrencefree survival or overall survival compared with upfront surgery by suppressing tumor micrometastasis and increasing surgical resectability. 7,8 However, neoadjuvant chemotherapy has the disadvantages of tumor progression and liver toxicity, including sinusoidal obstruction and steatosis. 9,10 For CRLM, Nordlinger et al reported that perioperative chemotherapy with 5-fluorouracil/folinic acid and oxaliplatin (FOLFOX4) increased progression-free survival compared with hepatic resection alone. However, the final analysis demonstrated that perioperative chemotherapy did not benefit patients' survival. 6,11 The European Society for Medical Oncology (ESMO) consensus guidelines indicate that perioperative treatment might not be necessary in cases diagnosed as resectable CRLM with favorable prognosis, but that perioperative chemotherapy should be administered to patients with resectable CRLM when the prognosis is unclear or unfavorable. 12 Thus, the major issue is how to select patients with resectable CRLM for neoadjuvant chemotherapy or surgery alone.
Several risk-scoring systems for CRLM after hepatectomy have been developed recently. [13][14][15] Patients' prognosis after hepatectomy for CRLM can be predicted preoperatively. We therefore investigated the clinical outcomes of neoadjuvant chemotherapy for resectable CRLM according to risk status after hepatectomy in a large number of patients recruited for a multi-institutional retrospective analysis, with the objective of determining the subgroup of patients who could benefit from neoadjuvant chemotherapy for CRLM.

| Patient enrollment
This study was a multi-institutional retrospective analysis. The  Figure 1A.

| Data collection
The following clinicopathological data were collected: sex, age, Eastern Cooperative Oncology Group (ECOG) performance status, carcinoembryonic antigen (CEA), and carbohydrate antigen 19-9 (CA19-9) before treatment of the metastatic hepatic lesion, primary lesion location, presence of preoperative treatment for primary in patients who received neoadjuvant chemotherapy (P = .0261). But the overall survival rate was not improved after the matching.

| Assessment of preoperative risk score
The preoperative risk score was calculated from six preoperative factors based on the nomogram for disease-free survival, as previously reported. 15 The six factors were the following: synchronous metastases (3 points), primary lymph node positive (3 points), number of tumors 2-4 (4 points) and ≥5 (9 points), largest tumor diameter over 5 cm (2 points), extrahepatic metastasis at hepatectomy (4 points), and preoperative carbohydrate antigen 19-9 level over 100 (4 points). The total points of the preoperative score ranged from 0 to 25. The patients were divided into three groups according to the preoperative risk score: a low-risk group (score = 0), an intermediaterisk group (score 1-10), and a high-risk group (score ≥11). balance, and SMD between 0.2 and 0.8 indicated the medium differences. SMD >0.8 meant considerable differences. Analyses were performed using the EZR software program, which is a graphical user interface for R. 21

| Statistical analysis
Statistical analyses were performed according to the prescribed protocol of the clinical trial. All data are expressed as mean ± standard deviation. Statistical differences between the groups were analyzed using Student's t-test for continuous variables and the chi-square test for other variables. Overall survival and disease-free survival curves were computed using the Kaplan-Meier method, and differences between survival curves were compared using the log-rank test. All statistical analyses were conducted using JMP Pro (v. 15.1.0; SAS Institute, Cary, NC, USA). All P values < .05 were considered statistically significant.

| Patient characteristics
A total of 289 patients with CRLM treated by liver resection were analyzed in this study. Table 1

| Distribution of preoperative risk score
Among the 289 cases, the preoperative risk score could be calculated in 277 cases. In 12 cases, the preoperative risk score could not be calculated because of missing values. The median preoperative risk score was 6; the distribution is presented in Figure 1B. The patients were divided into three groups according to their preoperative risk scores: a low-risk group (preoperative risk score = 0, n = 34), an intermediate-risk group (preoperative risk score 1-10, n = 192), and a high-risk group (preoperative risk score ≥11, n = 51). The characteristics of the groups are summarized in Table 2. The factors of sex, performance status, the location of the primary lesion, and the maximum diameter of the primary lesion did not differ significantly among the three groups. The percentages of cases with high CEA (over 10 ng/mL) and high CA19-9 (over 100 IU/mL) were significantly higher in the high-risk group. In the low-risk group, all cases had metachronous and single liver metastasis without extrahepatic metastasis. In the intermediate-risk group, 78 cases (41%) had synchronous metastasis and multiple liver metastases. The maximum diameter of the metastatic lesion was larger, at 26.3 mm. In the highrisk group, 39 cases (76%) had synchronous metastasis and 48 cases (94%) had multiple liver metastases. The maximum diameter of the metastatic lesion was the largest, at 40.6 mm.

| Long-term surgical outcomes
The recurrence-free survival curves and overall survival curves of the three groups are shown in Figure 2A Figure 2C). The 2-y recurrence-free survival rates were 41% of cases receiving neoadjuvant chemotherapy and 33% of cases without neoadjuvant chemotherapy. The 3-y overall survival rates were 81% in cases receiving neoadjuvant chemotherapy and 79% in cases without neoadjuvant chemotherapy (P = .4551; Figure 2D). In the high-risk group, the patients receiving neoadjuvant chemotherapy had a recurrence-free survival rate similar to that of the patients without neoadjuvant chemotherapy (P = .9287). The patients receiving neoadjuvant chemotherapy had a 3-y overall survival rate similar to that of the patients without neoadjuvant chemotherapy (67% vs 63%, P = .7440; Figure S1C,D).
In each risk group, the overall survival curves between high volume hospital and medium volume hospital aere shown ( Figure S2A-C). In the low-risk group, the 2-y recurrence-free survival rates were 75% in the high volume hospital and 63% in the medium volume hospital (P = 0.2262) and the 3-y overall survival rates were 92% in the high volume hospital and 83% in the medium volume hospital (P = .1698; Figure S2A). In the intermediate-risk group, the 2-y recurrence-free survival rates were 30% in the high volume hospital and 38% in the medium volume hospital (P = .1901) and the 3-y overall survival rates were 82% in the high volume hospital and 78% in the medium volume hospital (P = .1752; Figure S2B). In the high-risk group, the 2-y recurrence-free survival rates were 7.3% in the high volume hospital and 14% in the medium volume hospital (P = .4705) and the 3-y overall survival rates were 71% in the high volume hospital and 58% in the medium volume hospital (P = .6962; Figure S2C). In all risk groups, the 2-y recurrence-free survival rates and the 3-y overall survival rates were not significant between high volume and medium volume hospital.

| Propensity-score matching analysis
In  Table 3. There were no significant differences in characteristics observed between the two matched groups. The short-term surgical outcomes of operative time, blood loss, period of postoperative hospital stay, and morbidity did not differ significantly between the two groups. The recurrence-free survival curves and overall survival curves after propensity-score matching are shown in Figure 3A  overall survival benefit. 6,11 The subsequent exploratory retrospective analysis found that the baseline factors of highly elevated CEA value, performance status, and lower body mass index predicted a benefit of perioperative chemotherapy. 24 30 Based on these results, the authors planned a prospective randomized clinical trial to evaluate the impact of neoadjuvant chemotherapy in high-risk patients with primary resectable CRLM, and that trial is now in progress. 31 Matsumura et al conducted a multicenter, randomized, phase III trial to compare surgery followed by a postoperative FOLFOX regimen with surgery followed by a perioperative FOLFOX regimen plus cetuximab in patients with KRAS wildtype resectable CRLM.
KRAS status was used to stratify patients for perioperative chemotherapy, but there was no significant difference in progression-free survival or overall survival between the perioperative chemotherapy plus cetuximab group and the postoperative chemotherapy group. 32 Ninomiya et al focused on the CRLM grading system that is cur- For the patients in the low-risk group, our results demonstrate that upfront surgery without neoadjuvant chemotherapy might be desirable. A prospective randomized controlled trial is needed for verification.
In conclusion, we found that neoadjuvant chemotherapy for ini-