Prognostic Value of Carbohydrate Antigen 19‐9 and the Surgical Margin in Extrahepatic Cholangiocarcinoma

Abstract Aim The prognostic value of the perioperative carbohydrate antigen 19‐9 (CA19‐9) value and the prognostic relationship between the CA19‐9 value and the surgical margin in extrahepatic cholangiocarcinoma (EHCC) have not been fully discussed. Methods A total of 390 patients who underwent curative resection for EHCC between 2002 and 2018 were retrospectively analyzed. According to the perioperative CA19‐9 value, patients were divided into three groups: preoperative normal (Normal, n = 178), preoperative high and postoperative normal (Normalization, n = 155), and preoperative high and postoperative high (Nonnormalization, n = 57). Survival was analyzed according to the perioperative CA19‐9 value and surgical margin. Results The optimal cutoff value of CA19‐9 was 37 U/mL. Overall survival (OS) was significantly stratified according to the perioperative CA19‐9 value. The 5‐y OS rates in the Normal, Normalization, and Nonnormalization groups were 53%, 38%, and 23%, respectively (P < .001). Although the locoregional recurrence rate was comparable among the groups, the Normal group exhibited distant recurrence less frequently in comparison to the other groups. In the Normal group, the margin status had a significant impact on the OS (surgical resection with a negative margin [R0], 59% vs a microscopically positive margin [R1], 7% at 5‐y, P < .001). In contrast, in the Normalization and Nonnormalization groups, the OS rate of the R0 and R1 resection groups did not differ to a statistically significant extent. Conclusion The perioperative CA19‐9 value was related to the prognosis of resectable EHCC. A preoperative CA19‐9 value of ≥37 U/mL reflected systemic disease. R0 resection did not affect the survival in this patient group.

tive CA19-9 values were not measured were also excluded. Biliary drainage was routinely performed for patients with jaundice, mainly via an endoscopic approach. Portal vein embolization was performed when the future liver remnant was judged to be insufficient. 23 Neoadjuvant treatment was not performed. Adjuvant treatment was only performed in exceptional cases. Patients who participated in the BCAT trial 24 received adjuvant gemcitabine chemotherapy, and those who participated in the ASCOT trial 25 received adjuvant S-1 chemotherapy. Some patients who exhibited a microscopically positive margin (R1) at the hepatic bile duct received 5-FU or S-1 chemotherapy and radiotherapy; the decision on the administration of adjuvant chemoradiotherapy was left to the patient, after providing them with sufficient information to give their informed consent. 26 This study was approved by our Institutional Ethics Committee (approval number J2020-87-2020-1-3).

| Evaluation of CA19-9
The preoperative CA19-9 value was usually measured within 2 weeks before the day of surgery, after the resolution of jaundice and cholangitis. The postoperative CA19-9 value was usually measured at 2 weeks after the day of discharge. The institutional cutoff value of CA19-9 was 37 U/mL, according to the standard reference value. 27 In the present study, the cutoff values of CA19-9 were determined by a minimum P-value analysis, 28 which was performed to identify the preoperative and postoperative CA19-9 values that were associated with the best overall survival (OS).
Eligible patients were divided to three groups according to their perioperative CA19-9 value: preoperative normal (Normal group), preoperative high and postoperative normal (Normalization group), and preoperative high and postoperative high (Nonnormalization group).

| Surgery and pathology
The standard surgical procedure in the author's institution was hepatectomy with extrahepatic bile duct resection for perihilar cholangiocarcinoma and pancreatoduodenectomy for distal cholangiocarcinoma. 29 The regional lymph nodes were dissected in all patients. Paraaortic lymph node sampling was performed, but surgical resection was typically performed if the intraoperative frozen section diagnosis yielded a positive result. When necessary, HPD and/or VR were aggressively performed to achieve R0 resection. Postoperative complications were graded according to the Clavien-Dindo classification. 30 Pathological examinations were performed in accordance with the International Union Against Cancer (UICC) TNM classification 7th edition. 31 In the present study, carcinoma in situ at the ductal margin was defined as R0, because it did not affect OS. 1

| Postoperative follow-up
The median follow-up period of the censored patients was 47 mo in the present study. The site of recurrence was confirmed based on radiologic or histologic evidence. Locoregional recurrence was specifically defined as a local ill-defined mass at the site of choledochojejunostomy, the hepatic artery, or the portal vein, accompanied by positive positron emission tomography findings, increased tumor marker levels, and an increase in size over time on serial imaging performed to detect disease progression. 26

| Statistical analyses
Continuous data were described as the median and interquartile range and were compared using the Mann-Whitney U-test. Categorical variables were compared using Fisher's exact test. The cutoff values of continuous variables according to OS were determined based on a minimum P-value analysis. 28 Survival curves were generated using the Kaplan-Meier method, and differences were compared by a log-rank test. A Cox proportional hazards model, with stepwise backwardforward selection, was used for a multivariate analysis. Two-sided P < .05 were considered statistically significant. All statistical analyses were performed using the R software program (v. 4.0.3; The R Foundation for Statistical Computing, Vienna, Austria). In the minimum P-value analysis, the optimal cutoff value of preoperative and postoperative CA19-9 was determined to be 37 U/mL; this was the same as the standard cutoff value ( Figure 1). The patients were grouped according to their perioperative CA19-9 values as follows: Normal group, n = 178 (46%); Normalized group, n = 155 (40%); and Nonnormalized group, n = 57 (15%). Table 1 shows the patient characteristics according to the perioperative CA19-9 values. The median preoperative CA19-9 values of the Normal, Normalization, and Nonnormalization groups were 16, 104, and 311 U/mL, respectively, and each difference was statistically significant. The median postoperative CA19-9 values of the Normal, Normalization, and Nonnormalization groups were 10, 13, and 75 U/mL, respectively, and each difference was statistically significant. Distal cholangiocarcinoma was more frequently observed in the Normal group in comparison to the other groups. The R0 resection rates of the Normal, Normalization, and Nonnormalization groups were 90%, 87%, and 79%, respectively. R1 at the ductal margin was observed in 4%, 7%, and 9% of cases, respectively. R1 at the radial margin was observed in 6%, 8%, and 12% of cases, respectively. Two patients in the Normalization group had R1 at both the ductal and radial margins. Nonnormalization groups were 53%, 38%, and 23%, respectively, and each difference was statistically significant (P < .001). The same tendency was also observed when patients with distant metastasis were excluded from the analysis ( Figure S1). Moreover, a similar tendency was observed when patients were divided into perihilar cholangiocarcinoma ( Figure S2) and distal cholangiocarcinoma ( Figure S3) groups. Table 2 shows the sites of recurrence according to the perioperative CA19-9 values. The locoregional recurrence rate did not differ significantly among the groups (Normal group, 15%; Normalization group, 13%; Nonnormalization group, 11%). The frequency of distant recurrence in the Normal group was significantly lower in comparison to the other groups (Normal group, 38%; Normalization group, 54%; Nonnormalization group, 65%). The same tendency was also observed when patients with distant metastasis were excluded (Table S1). Figure 3 shows the OS according to the surgical margin status in each of the groups. In the Normal group, OS was significantly better in patients who received R0 resection than in those who received R1 resection (59% vs 7% at 5-y, P <.001). In the Normalization and Nonnormalization groups, the OS of the patients who received R0 resection was not significantly different from that in patients who received R1 resection. The same tendency was also observed when patients with distant metastasis were excluded ( Figure S1).

| RE SULTS
Moreover, a similar tendency was observed when patients were divided into perihilar cholangiocarcinoma ( Figure S2) and distal cholangiocarcinoma ( Figure S3) groups.

F I G U R E 1
The optimal cutoff value of preoperative and postoperative CA19-9 for overall survival was determined to be 37 U/mL. CA19-9, carbohydrate antigen 19-9 The multivariate analysis revealed that both the preoperative CA19-9 value (P = .018) and the postoperative CA19-9 value (P < .001) were independently associated with OS (Table 3).

| D ISCUSS I ON
The present study showed the utility of the perioperative CA19-9 value and the prognostic relationship between the perioperative CA19-9 value and the surgical margin in a relatively large cohort of patients with resectable EHCC. The optimal cutoff value was determined to be 37 U/mL. Patients in whom the preoperative CA19-9 value was <37 U/mL showed a good prognosis, and greatly benefited from R0 resection. The preoperative CA19-9 value of ≥37 U/ mL reflected the presence of systemic disease. R0 resection did not affect survival in this patient group.
Hepato-biliary-pancreatic surgeons have made efforts to achieve complete eradication of EHCC through extended hepatectomy, VR, and HPD, approaches that require highly sophisticated surgical skills and perioperative management approaches. [6][7][8][9][10][11]29 In the present study, VR and HPD were performed for 103 patients (26%) and 81 patients (21%), respectively. However, some patients did not benefit from R0 resection, despite receiving aggressive surgery, and showed high morbidity and mortality. A few studies reported the prognostic relationship between CA19-9 and R0 resection in biliary tract cancer among patients with normalized and nonnormalized CA19-9 values. 5,19,21 However, the significance of R0 resection in patients with normal CA19-9 values has not been reported. The present study showed that patients with preoperative CA19-9 values of <37 U/mLbut not those with values of ≥37 U/mL-benefited from R0 resection.
It is known that CA19-9 influences the prognosis of biliary tract cancer; the standard cutoff value is 37 U/mL. 5,19,20,27 However, the optimal cutoff value of CA19-9 for EHCC remains controversial.
Wang et al 18 reported a cutoff value of 150 U/mL, but their sample size was very limited. Although Lee et al 21 reported a cutoff value of 300 U/mL, preoperative CA19-9 >300 U/mL did not remain a significant factor in their multivariate analysis. Based on the statistical analysis of the present study, the cutoff value of 37 U/mL was found to be appropriate. Moreover, the multivariate analysis revealed that preoperative and postoperative CA19-9 values of ≥37 U/mL were both independent prognostic factors. Therefore, the standard cutoff value of 37 U/mL was found to be the optimal cutoff value for CA19-9.
Recently, the usefulness of the perioperative change in CA19-9 in patients with resectable biliary tract cancer has received attention. 5