Early Recurrence After Hepatectomy for Colorectal Liver Metastases: What Optimal Definition and What Predictive Factors?

To define and predict early recurrence after surgery, the authors examined 987 patients who underwent curative surgery for colorectal liver metastases. The optimal cutoff point of early recurrence was 8 months after surgery. Age, number of preoperative chemotherapy lines, response to last-line chemotherapy, number of tumors, and carbohydrate antigen 19-9 concentrations were identified as predictors of early recurrence. Salvage surgery significantly improved survival.

Abstract

Background.

The purpose of this study was to determine the optimal definition and elucidate the predictive factors of early recurrence after surgery for colorectal liver metastases (CRLM).

Methods.

Among 987 patients who underwent curative surgery for CRLM from 1990 to 2012, 846 with a minimum follow-up period of 24 months were eligible for this study. The minimum p value approach of survival after initial recurrence was used to determine the optimal cutoff for the definition of early recurrence. The predictive factors of early recurrence and prognostic factors of survival were analyzed.

Results.

For 667 patients (79%) who developed recurrence, the optimal cutoff point of early recurrence was determined to be 8 months after surgery. The impact of early recurrence on survival was demonstrated mainly in patients who received preoperative chemotherapy. Among the 691 patients who received preoperative chemotherapy, recurrence was observed in 562 (81%), and survival in patients with early recurrence was significantly worse than in those with late recurrence (5-year survival 18.5% vs. 53.4%, p < .0001). Multivariate logistic analysis identified age ≤57 years (p = .0022), >1 chemotherapy line (p = .03), disease progression during last-line chemotherapy (p = .024), >3 tumors (p = .0014), and carbohydrate antigen 19-9 >60 U/mL (p = .0003) as independent predictors of early recurrence. Salvage surgery for recurrence significantly improved survival, even in patients with early recurrence.

Conclusion.

The optimal cutoff point of early recurrence was determined to be 8 months. The preoperative prediction of early recurrence is possible and crucial for designing effective perioperative chemotherapy regimens.

Implications for Practice:

In this study, the optimal cutoff point of early recurrence was determined to be 8 months after surgery based on the minimum p value approach, and its prognostic impact was demonstrated mainly in patients who received preoperative chemotherapy. Five factors, including age, number of preoperative chemotherapy lines, response to last-line chemotherapy, number of tumors, and carbohydrate antigen 19-9 concentrations, were identified as predictors of early recurrence. Salvage surgery for recurrence significantly improved survival, even in patients with early recurrence. For better selection of patients who could truly benefit from surgery and should also receive strong postoperative chemotherapy, the accurate preoperative prediction of early recurrence is crucial.

Introduction

Hepatectomy is the gold standard for treatment of colorectal liver metastases (CRLM) and can offer prolonged survival and a chance of cure, with a 5-year survival rate of up to 67% [1–3]. Remarkable advances in surgical techniques, imaging modalities, and postoperative management and the advent of more effective chemotherapy have expanded the indications of surgery for CRLM, even with a large tumor burden or resectable extrahepatic diseases. However, more than half of patients experience recurrence after hepatectomy, and the majority of such recurrences occur within 2 years [4–7].

Although early recurrence has been reported to adversely influence survival after hepatectomy for CRLM, the definition of early recurrence is still ambiguous, and it differs among previous reports, ranging from 6 months to 2 years [8–15]. In addition, recent advances in oncosurgical approaches, consisting of aggressive surgery (for CRLM and also for extrahepatic diseases), perioperative chemotherapy, and repeat surgery for recurrence, have dramatically changed the consensus for treatment of CRLM. Under such circumstances, however, the significance of early recurrence, including the indications for surgery and perioperative chemotherapy, is also unclear.

The aims of this study were to determine the optimal cutoff point for the definition of early recurrence after hepatectomy for CRLM based on survival after initial recurrence and to identify predictive factors associated with early recurrence. Furthermore, the impact of early recurrence on survival and the prognostic factors for patients with early and late recurrence were analyzed.

Patients and Methods

Study Population

From 1990 to 2012 at our institution, 987 patients with CRLM underwent curative surgery (not only for CRLM but also concomitant extrahepatic disease if present). Twelve patients (1.2%) died of postoperative complications and were excluded from this study; 846 patients had a minimum follow-up period of 24 months and were eligible for inclusion. A prospectively maintained database of CRLM was used to identify the patients, and additional information was collected by reviewing each patient’s medical record.

Preoperative Work-Up

Before hepatectomy, all patients underwent routine laboratory and liver function tests and were evaluated with thoraco-abdominal and pelvic imaging (ultrasonography, computed tomography [CT], or magnetic resonance imaging) to accurately stage the disease. Preoperative chemotherapy was administered for patients with initially unresectable CRLM in the conversion setting or patients with synchronous (diagnosed before, during, or within 3 months after colorectal resection) or marginally resectable CRLM in the neoadjuvant setting, as described previously [16]. The response to preoperative chemotherapy was evaluated with CT after every four cycles of treatment, initially according to World Health Organization criteria [17] but more recently using the Response Evaluation Criteria in Solid Tumors [18].

Surgery

The principle of surgery was to remove all detectable lesions with a tumor-free margin in all patients. If tumor-free margins could not be obtained because of contact with major vascular or biliary structures, then resection was still indicated, provided that all tumors could be macroscopically resected [19]. If the removal of all tumors could not be achieved by a single hepatectomy while keeping sufficient future liver remnant volume, then portal vein embolization (PVE), additional local ablation therapy including cryotherapy and radiofrequency ablation, and a two-step approach including two-stage hepatectomy and associating liver partition and portal vein ligation for staged hepatectomy were performed [20–22]. In cases with concomitant extrahepatic disease, sequential resection was considered when both sites were deemed to be resectable.

Postoperative Work-Up

Postoperative complications were graded according to the validated classification criteria described by Dindo et al. [23], and major complications were defined as any complication of grade III or higher. After treatment, all patients underwent regular follow-up measurements of serum carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9), and imaging studies including ultrasonography or abdominal and thoracic CT were performed every 4 months to detect any intrahepatic or distant recurrence. Recurrences were treated surgically only when the overall strategy was considered to be potentially curative.

Statistical Analysis

Categorical variables were compared using the χ² test. The optimal cutoff points for the definition of early recurrence were determined using the minimum p value approach that was calculated using the log-rank test for overall survival (OS) after first recurrence. The cutoff values of continuous variables were determined using receiver operating characteristics analysis for differentiation between early and late recurrence and the minimum p value approach for survival data, including censored cases. Survival analyses were carried out using the Kaplan-Meier method and were compared using the log-rank test. OS was calculated from the date of hepatectomy until death or last follow-up. Disease-free survival (DFS) was calculated from the date of hepatectomy or the last potentially curative surgery for concomitant extrahepatic disease, if present, until the date of recurrence or death. Variables in which the p value for the univariate analysis was <.10 were subjected to either multivariate logistic regression or the Cox regression model by a stepwise backward elimination procedure, as appropriate. All statistical analyses were performed using JMP software (SAS institute, Cary, NC, http://www.sas.com/en_us/home.html) and R version 3.1.1 (http://www.r-project.org), and p values <.05 were considered statistically significant.

Results

Patient Characteristics and Definition of Early Recurrence

There were 502 male and 344 female patients, with a median age of 61 years (range, 28–89). The primary tumor was located in the colon in 634 (75.0%) and in the rectum in 211 (25.0%) patients. Liver metastases were diagnosed synchronously in 505 patients (60.0%), and bilobar distribution was observed in 428 (50.6%). The mean tumor size was 45.0 ± 32.1 mm (median, 36; range, 4–300) at diagnosis and 38.1 ± 27.9 mm (median, 30; range, 0–300) at hepatectomy, and the mean tumor number was 3.8 ± 3.8 (median, 3; range, 1–35) at diagnosis and 3.9 ± 4.1 (median, 2; range, 0–32) at hepatectomy. Concomitant extrahepatic disease was present in 134 patients (15.9%), located in the lung in 42, lymph nodes in 33, peritoneum in 33, other sites in 13, and multiple sites in 13. Preoperative chemotherapy was administered in 691 patients (81.7%), with a median number of 1 line (range, 1–5) and 8 cycles (range, 1–49) (supplemental online Table 1). Biologic agents such as bevacizumab, cetuximab, and panitumumab were used in 175 patients (25.4%). Details of chemotherapy regimens are summarized in supplemental online Table 1. Cumulative OS and DFS at 3, 5, and 10 years were 67.1, 50.2, and 33.4% and 22.1, 17.8, and 16.5%, respectively (Fig. 1).

Figure 1.

OS and DFS for patients with colorectal liver metastases who underwent hepatectomy and had a minimum follow-up of >2 years (n = 846).

Abbreviations: DFS, disease-free survival; OS, overall survival.

Among the 846 patients, recurrence was observed in 667 (78.8%) during the mean follow-up period of 57.6 months (median, 44.2 months). In patients who had recurrence, the optimal cutoff point to define early recurrence was determined to be 8 months after hepatectomy, based on the results of the minimum p value approach for survival after initial recurrence (p = 1.08 × 10⁻¹²) (Fig. 2). When the patients were divided into two groups according to preoperative chemotherapy, the minimum p values approach in patients who did not receive preoperative chemotherapy revealed no significant change compared with those who received preoperative chemotherapy (Fig. 2). In patients who received preoperative chemotherapy, the optimal cutoff point remained 8 months after hepatectomy (p = 4.89 × 10⁻¹³). Based on these results, we defined early recurrence as recurrence that occurred within 8 months after surgery, and further analyses were performed in patients who received preoperative chemotherapy.

Figure 2.

Optimal cutoff point regarding the time to first recurrence for the definition of early recurrence using the minimum p value approach. p values of log-rank test for overall survival after first recurrence were plotted.

Impact of Early Recurrence and Salvage Surgery on Survival in Patients Who Received Chemotherapy

Among the 691 patients who received preoperative chemotherapy, cumulative OS and DFS at 3, 5, and 10 years were 64.7, 49.6, and 30.1% and 19.1, 15.1, and 14.2%, respectively. Recurrence was observed in 562 patients (81.3%); among them, early recurrence (≤8 months) and late recurrence (>8 months) were observed in 252 (44.8%) and 310 (55.2%), respectively. OS in patients with early recurrence was significantly worse than that in those with late recurrence (p < .0001) (Fig. 3).

Figure 3.

Overall survival after hepatectomy for patients with early recurrence (n = 252) and late recurrence (n = 310).

Abbreviation: MST, median survival time.

First recurrence sites and salvage surgery for first recurrence are summarized in Table 1. Of 252 patients with early recurrence, 214 (84.9%) developed single-site recurrence, as did 275 of 310 (88.7%) patients with late recurrence (p = .19). For the first recurrence, salvage surgery was performed more often in patients with late recurrence (58.1%) than in those with early recurrence (40.9%) (p < .0001). OS in patients who underwent salvage surgery for recurrence was significantly better than in those who did not (5-year OS, 59.7 vs. 15.4%, p < .0001) (Fig. 4A). The impact of salvage surgery for recurrence was also confirmed in both groups of patients with early and late recurrence (5-year OS, 38.8 vs. 4.2%, p < .0001; 70.7 vs. 27.6%, p < .0001, respectively) (Fig. 4B, 4C).

Table 1.

First recurrence sites and salvage surgery for recurrence in patients who received preoperative chemotherapy

Figure 4.

Overall survival after hepatectomy according to salvage surgery for recurrence in all patients (A), patients with early recurrence (B), and patients with late recurrence (C).

Abbreviation: MST, median survival time.

Predictive Factors for Early Recurrence

Table 2 shows univariate and multivariate analyses of the factors related to early recurrence. According to univariate analysis, age, primary tumor stage, bilobar distribution of liver metastases, preoperative chemotherapy cycles and lines, response to last-line chemotherapy, tumor number and size at hepatectomy, CEA and CA19-9 at hepatectomy, PVE, major hepatectomy, two-step approach, surgical margin status of liver metastases, and concomitant extrahepatic disease were found to be related to early recurrence (p < .10). Multivariate logistic regression analysis revealed that age ≤57 years (relative risk [RR] 1.79, p = .0022), >1 preoperative chemotherapy line (RR 1.58, p = .03), progression of disease during last-line chemotherapy (RR 2.18, p = .024), >3 tumors at hepatectomy (RR 1.81, p = .0014), and CA19-9 >60 U/mL at hepatectomy (RR 2.21, p = .0003) were independent predictive factors for early recurrence. The patients were divided into two groups according to period of treatment: early (1990–2000, n = 244) and late (2001–2012, n = 318). Early recurrence was observed in 105 patients (43.0%) in the early period and 147 patients (46.2%) in the late period, respectively (p = .45). In the early period, multivariate analysis revealed that age ≤57 years (RR 1.88, p = .033), >3 tumors at hepatectomy (RR 3.09, p = .0001), and CA19-9 >60 U/mL at hepatectomy (RR 2.83, p = .0015) were independent predictive factors for early recurrence. On the other hand, in the late period, multivariate analysis revealed that age ≤57 years (RR 1.67, p = .038), CA19-9 >60 U/mL at hepatectomy (RR 2.42, p = .0021), presence of extrahepatic disease (RR 1.94, p = .034), and two-step approach (RR 2.36, p = .014) were independent predictive factors.

Table 2.

Univariate and multivariate analyses of predictive factors of early recurrence in patients who received preoperative chemotherapy

Prognostic Factors of Survival After Hepatectomy

According to univariate analysis, 13 variables, including age, primary node stage, number of preoperative chemotherapy cycles, use of biologic agents, response to last-line chemotherapy, tumor size at hepatectomy, CEA and CA19-9 at hepatectomy, PVE, red blood cell transfusion, surgical margin status of liver metastases, early recurrence, and salvage surgery for recurrence, were related to OS after hepatectomy (supplemental online Table 2). Multivariate Cox regression analysis identified primary node stage I or II (hazard ratio [HR] 1.46, p = .003), no use of biologic agents (HR 1.56, p = .0021), CA19-9 >100 U/mL at hepatectomy (HR 1.86, p < .0001), PVE (HR 1.48, p = .017), early recurrence (HR 2.69, p < .0001), and no salvage surgery for recurrence (HR 3.20, p < .0001) as independent prognostic factors of survival after hepatectomy (Table 3).

Table 3.

Multivariate Cox regression analyses of factors related to survival after hepatectomy in patients who received preoperative chemotherapy

Subgroup Analyses in Patients With Early and Late Recurrence

Patients were divided into two subgroups according to early recurrence (n = 252) and late recurrence (n = 310), and prognostic factors of survival after hepatectomy were analyzed. The results of univariate analyses are summarized in supplemental online Table 3. In patients with early recurrence, multivariate analysis revealed that primary node stage I or II (HR 1.56, p = .011), no use of biologic agents (HR 1.80, p = .0013), CEA >50 ng/mL at hepatectomy (HR 2.10, p = .0004), and no salvage surgery for recurrence (HR 3.29, p < .0001) were independent prognostic factors for survival (Table 3). On the other hand, in patients with late recurrence, multivariate analysis revealed that CA19-9 >100 U/mL at hepatectomy (HR 2.01, p = .0017) and no salvage surgery for recurrence (HR 3.48, p < .0001) were independent prognostic factors for survival (Table 3).

Discussion

In the current study, 79% of the patients who underwent curative hepatectomy developed recurrence. The optimal cutoff point of early recurrence was determined to be 8 months after hepatectomy based on the results of the minimum p value approach for survival after initial recurrence, and the impact of early recurrence on survival was higher in the patients who received preoperative chemotherapy (Fig. 1). In patients who received preoperative chemotherapy, recurrence was early (≤8 months) in 45% of the patients with recurrence, and age, number of preoperative chemotherapy lines, response to last-line chemotherapy, number of tumors, and CA19-9 at hepatectomy were identified as independent predictive factors for early recurrence.

Although many previous studies reported early recurrence to be a prognostic factor of survival after hepatectomy for CRLM, the definitions of early recurrence were not based on statistically determined values and thus differed between the reports, ranging from 6 months to 2 years [8–15]. To our knowledge, this is the first study that determined the optimal cutoff point to define early recurrence, based on the minimum p value for survival after initial recurrence. As shown in Figure 2, the cutoff points of 6, 12, and 24 months did not seem to be adequate to define early and late recurrence. Of note, the impact of early recurrence on survival was not demonstrated in the patients who did not receive preoperative chemotherapy, compared with those who did receive it (Fig. 2). These findings suggest that patients who received preoperative chemotherapy were likely to demonstrate more severe metastatic disease and poorer tumor biology and that special caution is required for the follow-up of patients who received preoperative chemotherapy to detect any early recurrences.

The current study demonstrated that OS in patients with early recurrence was significantly worse than in those with late recurrence. Therefore, it is very important to predict early recurrence in patients with CRLM who have undergone hepatectomy. Previous studies have reported several predictive factors of early recurrence after hepatectomy: primary tumor stage III or IV [14], synchronous metastases [14], response to preoperative chemotherapy [14], number of tumors (>2 [10], >3 [14], >7 [9, 13]), concomitant use of radiofrequency [14], no use of intraoperative ultrasonography [14], R1 resection [14], postoperative morbidity [11], and postoperative chemotherapy [14, 15]. In our study, multivariate logistic regression analysis identified 5 factors, including age ≤57 years, >1 chemotherapy cycle, progression disease during last-line chemotherapy, >3 tumors, and CA19-9 >60 U/mL, as independent predictors of early recurrence. Of note, younger patients developed early recurrence more frequently than older patients after hepatectomy for CRLM (Table 1). It is known that a younger age for patients with colorectal cancer is associated with advanced stage and higher recurrence rate, even though the oncological outcome is similar to that of older patients [24–26]. Similarly, in the current study, even though younger age was one of the independent predictors of early recurrence, it did not influence survival. Although presenting with more aggressive disease with a risk of early recurrence, young patients with CRLM do not have worse prognosis after hepatectomy.

Recent expansion of the indications for surgery is not only due to the development of surgical techniques but also to more effective chemotherapy. Although these expanded criteria have led to an increase in the potentially resectable pool of patients, they can also lead to an increased risk of early recurrence. The benefit of hepatectomy for patients who eventually develop early recurrence therefore remains questionable because of their poorer prognosis. In the current study, two chemotherapy-related factors, >1 chemotherapy line and progression of disease during last-line chemotherapy, were identified as independent predictors of early recurrence, suggesting that the patients with CRLM who show resistance to chemotherapy have a higher risk of early recurrence.

Although postoperative chemotherapy is a possibility that can reduce the risk of early recurrence, the current study failed to demonstrate the benefit of postoperative chemotherapy to prevent early recurrence. Of course this is thought to be due, to no small extent, to the decision bias of whether adjuvant chemotherapy was needed. Nevertheless, at present, there is no clear method to prevent early recurrence. Therefore, it would be of great importance to predict early recurrence preoperatively, especially to tailor the type of postoperative chemotherapy. Accordingly, it has been recently suggested that intra-arterial hepatic chemotherapy could be more effective to prevent recurrence [27]. The predictive factors identified in this study were all available preoperatively, and they are thought to be helpful for the selection of patients who can benefit from hepatectomy.

In the current study, salvage surgery for recurrence had great impact on survival after hepatectomy. One of the reasons for the poorer prognosis of patients with early recurrence seems to be that fewer patients with early recurrence could receive salvage surgery compared with those with late recurrence (40.9 vs. 58.1%, p < .0001). However, even for patients with early recurrence, salvage surgery had a positive effect on survival, with 5-year OS of 38.8%, which was significantly better than that of the patients who did not undergo salvage surgery (5-year OS 4.2%). These findings suggest that surgeons and oncologists should keep in mind the possibility of salvage surgery when recurrence is observed, even if it occurs within 8 months.

In comparison with CEA, so far there have been fewer reports demonstrating the significance of CA19-9 as a biomarker of recurrence or survival in patients with CRLM [28–31]. In the current study, CA19-9, not CEA, was found to be an independent predictive factor of early recurrence, and it was also an independent prognostic factor of survival after hepatectomy for CRLM. Interestingly, however, when the patients were divided into two groups according to the time to first recurrence, CA19-9 was an independent prognostic factor of survival in patients with late recurrence but not in those with early recurrence. Instead of CA19-9, CEA was found to be an independent prognostic factor of survival in patients with early recurrence. These findings suggest that CA19-9 is a good predictor of early recurrence and survival after hepatectomy. However, when early recurrence develops after hepatectomy, CEA is a better marker to predict survival. Although CA19-9 is not typically used worldwide, we recommend measurement of CA19-9 levels in combination with CEA during the course of treatment in patients with CRLM.

Of note, the use of biologic agents with preoperative chemotherapy was an independent prognostic factor for better survival in patients of the whole cohort and in those with early recurrence, suggesting that better control of metastatic disease before surgery could be associated with better outcome in cases of early recurrence (Table 3). Although the effect of preoperative use of biologic agents for the survival of patients with late recurrence is still unclear, the preoperative use of biologic agents might be recommended in view of the prolonged survival observed for those who may develop early recurrence.

Conclusion

The current study determined the optimal cutoff point of early recurrence as 8 months after hepatectomy for CRLM, and the impact of early recurrence on survival was higher in patients who received preoperative chemotherapy. Five factors, including age, number of preoperative chemotherapy lines, response to last-line chemotherapy, number of tumors, and CA19-9, were identified as predictors of early recurrence. For better selection of the patients who could truly benefit from surgery and should also receive strong postoperative chemotherapy, the accurate preoperative prediction of early recurrence is crucial. Salvage surgery should therefore be proactively considered even for cases of early recurrence.

See http://www.TheOncologist.com for supplemental material available online.

This article is available for continuing medical education credit at CME.TheOncologist.com.

Author Contributions

Conception/Design: Katsunori Imai, René Adam

Provision of study material or patients: Katsunori Imai, Carlos Castro Benitez

Collection and/or assembly of data: Katsunori Imai, Marc-Antoine Allard, Carlos Castro Benitez

Data analysis and interpretation: Katsunori Imai, Marc-Antoine Allard, Carlos Castro Benitez, Eric Vibert, Antonio Sa Cunha, Daniel Cherqui, Denis Castaing, Henri Bismuth, Hideo Baba, René Adam

Manuscript writing: Katsunori Imai, Marc-Antoine Allard, Carlos Castro Benitez, Eric Vibert, Antonio Sa Cunha, Daniel Cherqui, Denis Castaing, Henri Bismuth, Hideo Baba, René Adam

Final approval of manuscript: Katsunori Imai, Marc-Antoine Allard, Carlos Castro Benitez, Eric Vibert, Antonio Sa Cunha, Daniel Cherqui, Denis Castaing, Henri Bismuth, Hideo Baba, René Adam

Disclosures

Antonio Sa Cunha: Merck, Sanofi, Takeda (C/A). The other authors indicated no financial relationships.

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board