Nonanatomic resection is not inferior to anatomic resection for primary intrahepatic cholangiocarcinoma: A propensity score analysis

Whether anatomic resection (AR) achieves better outcomes than nonanatomic resection (NAR) in patients with primary intrahepatic cholangiocarcinoma (ICC) is unclear. Data were retrieved for all consecutive patients who underwent liver resection for primary ICC from January 2007 to July 2017. The prognoses of the patients without direct invasion to contiguous organs or extrahepatic metastasis who underwent AR or NAR were compared. 85 patients underwent AR, and 65 patients underwent NAR. operation time were slightly decreased in the NAR group. The risk of Clavien-Dindo classification (CDC) IV in the AR group was significant higher than that in the NAR group. Cox regression analysis showed lymph node metastasis and adjuvant therapy were significant prognostic factors for overall survival (OS) and disease-free survival (DFS), respectively. After 1:1 propensity score matching (PSM), 29 pairs of patients were compared. The survival curves showed the NAR group had slightly improved DFS and OS than the AR group before and after matching. Thus, we conclude NAR was not inferior to AR in improving the survival outcomes for patients with primary solitary ICC lesions without direct invasion to contiguous organs or extrahepatic metastasis. Furthermore, patients may benefit from NAR.

prognoses of patients between the AR and NAR groups with primary solitary ICC lesions without direct invasion to contiguous organs or extrahepatic metastasis using a one to one propensity score matching (PSM) analysis.

Results
Baseline patient and Clinicopathological characteristics. Patients were followed until the date of death or the final date of the study, December 30, 2017. A total of 150 patients underwent hepatectomy for primary solitary ICC lesions without direct invasion to contiguous organs or extrahepatic metastasis. These patients were followed for 3 to 107 months (median 12 months). The majority of patients were male (n = 93, 62.0%). More than half of the patients presented with a tumour size ≤5 cm (n = 80, 53.3%). Of note, half of the patients were positive for hepatitis B virus surface antigen (HBsAg) (n = 76, 50.7%), and presented with cirrhosis (n = 83, 55.3%). Only 9.3% patients had intrahepatic biliary stones (n = 14). Specifically, vascular invasion, periductal invasion, and perineural invasion were only observed in 17.3% (n = 26), 2.7% (n = 4), and 4.7% (n = 7) of tumours, respectively. Poorly differentiated lesions were found in 34.7% (n = 52) of patients, metastatic nodal disease was observed in 12.7% (n = 19) of these patients, and a positive resection margin was found in 8 (5.3%) patients. Regarding liver function, as defined by the Child-Pugh classification, 131 (87.3%) patients were class A, and 7 (4.7%) were class B; Of these 150 patients, 85 (56.7%) underwent AR, and 65 (43.3%) underwent NAR. Patients undergoing NAR were younger, more likely to be positive for HBsAg (NAR vs. AR; 60.0% VS 43.5%, P = 0.046), and more often presented with cirrhosis (NAR vs. AR; 66.2% VS 47.1%, P = 0.02). NAR was more frequently performed among patients who had small tumors. Patients undergoing NAR were also more likely to have advanced disease characterized by poor tumor differentiation (Table 1). Interestingly, patients who underwent a NAR were more likely to have a microscopically positive margin than patients who underwent AR (p = 0.001).
Association of surgical procedure with Short and Long-Term Outcomes. There were no difference of Operation time, intraoperative blood loss, intra-and postoperative transfusion, and the length of postoperative hospital stay between the two groups (  Fig. 1), respectively. The 1-, 3-, and 5-year OS rates were 75.2%, 47.0%, and 25.7% in the NAR group and 65.0%, 35.4%, and 29.1% in the AR group (P = 0.477, Fig. 2), respectively. A Cox regression model was performed to identify risk factors associated with DFS and OS of these 150 patients. On multivariable analysis, adjuvant therapy was associated with better DFS (HR 0.757, [95% CI, 0.605-0.949], p = 0.016) (Table 3), lymph node metastasis was associated with worse OS (HR 1.972, [95% CI, 1.039-3.743], p = 0.038) ( Table 4). To better account for any residual confounding due to case mix, a sensitivity analysis investigating the association of surgical procedure with DFS and OS was performed within a matched patient cohort. Patients were matched based on propensity scores obtained from a logistic regression model accounting for patient, operation and tumour characteristics. Based on this 1:1 propensity score matching (PSM) approach, 29 NAR patients and 29 AR patients were matched (  Fig. 4)).

Discussion
To date, partial hepatectomy remains the mainstay of curative treatment for ICC 16,17 . Unfortunately, prognosis after partial hepatectomy is unsatisfactory, with a high incidence of locoregional recurrence and/or extrahepatic metastasis and a low 5-year OS rate [18][19][20] . In the 1980 s, Muccuchi et al. 5 proposed AR in hepatectomy for HCC. Previous studies [8][9][10][11][12] showed the benefit of AR for single HCC lesions. Both ICC and HCC arise in the hepatic parenchyma. However, the extent of hepatic resection and its impact on outcomes among patients with ICC has not been well examined. Thus, whether the application of AR or NAR can improve the prognosis of ICC patients is worthy to investigate. Selection biases are inevitable in retrospective comparisons of postoperative outcomes in patients who underwent AR and NAR for ICC. To better account for any residual confounding due to case mix, we adjusted the clinicopathological characteristics of patients to eliminate as much selection bias as possible by excluding patients with multiple lesions, adjacent organ invasion, and extrahepatic metastasis. Furthermore, the clinicopathological characteristics of patients were adjusted by 1:1 PSM.
In the current study, patients who underwent AR tended to have larger tumors, the NAR group had more patients with positivity of HBsAg and liver cirrhosis. Previous study 6 had pointed that the ideal surgical approach to hepatic resection of primary liver carcinoma should optimize locoregional control, yet preserve as much nontumorous hepatic parenchyma in the FLR. Theoretically, wider margins may offer a better chance of disease control, whereas a more limited resection allows for the preservation of hepatic parenchyma and mitigation of the risk for liver insufficiency 12,21,22 . Thus, patients with large tumors may, however, simply be more likely to undergo AR versus NAR as larger tumors may necessitate major resections to achieve an negative margin.
As mentioned above about the definition of AR and NAR. We can see the extent of liver resection is greater in AR than in NAR. Theoretically, the risk of postoperative liver failure is higher in AR, especially in patients who have comorbidities such as liver cirrhosis, the advantage of NAR is related to the postoperative liver function reserve. Decreases in albumin, prothrombin time, and cholinesterase during the first year after surgery were reported to be lower in patients who underwent NAR than in patients who underwent AR 23 . Thus, patients with hepatic tumours have concomitant injury to the nonmalignant liver parenchyma, a smaller range of resections can result in better restoration and/or preservation of the liver function [24][25][26] . Meanwhile, for patients with high risk of liver failure due to the low FLR, portal vein embolization can be performed to induce compensatory hypertrophy of the remnant liver and thus increase the safety of major hepatectomy 27 .
Different from previous studies 12, 28 , the present study showed that the intraoperative bleeding and intraoperative blood transfusion had no significant difference between groups before and after matching, this may be due to AR are often performed under inflow occlusion of the liver by a Pringle maneuver 28 in order to minimize bleeding during the transection phase. In addition, operation time were slightly decreased in the NAR group, but there were no significant difference between groups. Interestingly, in our current analysis, we found NAR group had more patients with tumour lesions located in the right liver than AR group, generally, the time spent at resection of the right liver tumour is longer than that of the left liver tomour, this might be explained by the fact that the right liver resection was more difficult than the left. That is, operation time in NAR group was less than in AR group. Furthermore, hepatic resection can be associated with a certain degree of peri-operative morbidity and mortality. In the current analysis, the length of hospital stay and the overall incidence of complications were largely comparable among patients who underwent AR versus NAR. Meanwhile, there were no significant differences of CDC I, CDC II, CDC III or CDC V between groups. However, notably, the risk of CDC IV in the AR group was significant higher than that in the NAR group. In addition, the incidence of postoperative hepatic failure, bile leakage and intra-abdominal bleeding as well as perioperative mortality did not differ between the two groups. Additionally, the occurrence of a postoperative complication has previously been reported to be an independent predictor of worse long-term outcomes 29 . However, we didn't find the similar result, furthermore, a previous study 30 found that the impact of major complications on survival was primarily in the immediate post-operative period rather than in the long term. Thus, these data suggest that the overall perioperative outcomes can be superior for NAR. Regarding the clinicopathological factors affecting OS and DFS in patients with macroscopically solitary ICC lesions, univariate analysis showed that positivity of HBsAg was associated with worse DFS and OS. Thus, positivity for HBsAg was a risk factor for DFS and OS. However, in disagreement with a previous study 31 , this study showed antiviral therapy for HBV-related ICC patients didn't improve survival outcomes. This maybe due to the comparable liver function or stable low level of virus reproduction. Consistent with previous studies 31, 32 , vascular invasion was associated with worse OS, However, in the current study, the two groups had no significant difference of vascular invasion, thus, whether the AR is benefit for ICC patients with vascular invasion needs further AR (n = 85) NAR (n = 65) P value study. Meanwhile, the results of our Cox regression analysis showed lymph node metastasis and adjuvant therapy were significant prognostic factors for OS and DFS, respectively. Similar to previous study 33,34 , lymph node metastasis was a risk factor for survival, however, there is no clear consensus on whether lymph nodes should be dissected 35,36 , and this aspect requires further research. Whilst the present study found the adjuvant therapy may improve the prognosis. Furthermore, previous study 37 suggested that the neoadjuvant/adjuvant therapies may improve the prognosis of ICC patients especially for those with positive surgical margin. A meta-analysis 38 demonstrated that adjuvant therapy benefited biliary cancer patients with node-positive or R1 disease. In this study, the survival curves showed the NAR group had slightly improved DFS and OS than the AR group before and after matching, yet with no significant statistically differences. Similarly, a Japanese retrospective study 39 had concluded that in HCC patients with impaired liver functions, limited liver resection without tumor  exposure may provide longer tumor-free and overall survival. That is, the NAR had better survival outcomes than the AR especially for the patients with liver cirrhosis. In fact, several previous studies 23, 40 have noted that preservation of liver parenchyma should take priority, and minor hepatectomy can provide equal OS and DFS compared with major hepatectomy for small HCC. In addition, some previous studies 41,42 found that HCC and ICC tended more often to have associated micro-metastasis and disseminated via portal venous branches. Thus, local recurrence (LR) may be due to residual intrahepatic metastasis spreading through the portal venous system, which cannot be detected before and during surgery. AR based on the Couinaud system requires a complete removal of at least one Couinaud segment   containing the tumour. This can completely eliminate the entire unilobar portal venous drainage of the involved lobe of the liver to decrease the LR. Pawlik T. M, et.al. 30 found that patients undergoing major resection had similar outcomes to patients undergoing minor resection, as long as the surgical margin was wider than 5 mm. In contrast, when the surgical margin was narrow (1-4 mm), major resection was associated with a decrease in tumor recurrence compared with minor resection. Therefore, that is, both surgical margin and resection volume can impact the survival outcomes. The current study found the NAR group had more patients with positive surgical margin, however, survival analysis showed that the surgical margin had no significant effect on DFS or OS before or after matching. Generally, the distance between surgical margin and lesion is greater in AR than in NAR; thus, theoretically, the NAR group had a higher potential of positive surgical margin than the AR group. Furthermore, previous studies 43,44 reported that the residual tumor on the resection margin may grow and spread in a much more aggressive pattern, which may result in the early recurrence and poor OS. A systematic review and meta-analysis 45 found that Patients with negative surgical margin had significantly favorable OS and progression-free survival after surgical resection for ICC. Similarly, Spolverato G, et al. 46 had conclued, for patients undergoing resection of ICC, a positive margin was associated with an inferior long-term outcome. In addition, there was an incremental worsening DFS and OS as margin width decreased from 1 cm. Thus, one possible explanation for the results of the current study is that a higher application rate of adjuvant treatment in the NAR group. And as previous study 37 mentioned, the patients with positive margin may benefit more from adjuvant therapy. In other words, the adjuvant therapy may improve the suvrvival of ICC patients especially for those with NAR. However, more studies are needed to validate these findings.
This study has several limitations, the main limitation is its retrospective design. And unmeasured differences between the 2 groups could potentially lead to confounding effects. The results are also limited by the small sample size. Thus, randomized controlled trials are needed to further determine whether NAR is superior to AR.
These findings suggest that ICC patients may receive more benefits from NAR than that from AR, especially for patients with liver cirrhosis due to the lower risk of CDC-IV such as liver failure. Thus, we conclude NAR was not inferior to AR in improving the survival outcomes for patients with primary solitary ICC lesions without direct invasion to contiguous organs or extrahepatic metastasis. Furthermore, patients may benefit from NAR.

Data Sources and Study Population.
Data for all consecutive patients who underwent liver resection for primary solitary ICC lesions at West China Hospital of SiChuan University, China, from January 2007 to July 2017 were retrieved from a prospective database for this study. The Institutional Review Board of West China Hospital of SiChuan University approved this study, confirm that the study was performed in accordance with relevant guidelines/regulations and informed consent was obtained from all patients and/or their legal guardian/s. A total of 150 patients underwent initial, curative hepatectomy for histologically confirmed primary solitary ICC lesions without direct invasion to contiguous organs or extrahepatic metastasis. The resection margin was ascertained based on final pathology. Six patients died of some non-tumour-related causes during follow-up -1 patient each died of intracerebral bleeding, severe pneumonia, gastrointestinal bleeding, acute cardiac infarction, liver failure, and car accident. Data on standard demographic, perioperative clinicopathological, and tumour-related characteristics were collected. Factors compared in the AR and NAR groups included patient age and sex, positivity for HBsAg, intrahepatic biliary stones, Child-Pugh classification, carcinoembryonic antigen (CEA) level, carbohydrate antigen 19-9 (CA19-9) level, blood loss, blood transfusions, and complications. Tumour characteristics were based on final pathology reports, including liver cirrhosis, tumour size, tumor location, vascular/perineural/biliary/ adjacent organ invasion, lymph node metastasis, and histological grade. Tumor size was defined as the largest diameter (in centimeters) for the tumor within the resected specimen. In addition, information on surgery, adjuvant therapy, such as adjuvant chemotherapy, radiofrequency ablation, transcatheter arterial chemoembolization, and re-operation was also collected for each patient. Margin status were determined from the final post-operative pathological report.
Follow-up. After discharge, the patients were followed-up every 3 months for up to 2 years after the initial operation and every 6 months thereafter. Recurrence was defined as the appearance of a new lesion that exhibited radiologic features compatible with ICC. DFS was defined as the interval between the date of operation and the date of diagnosis of first recurrence or last follow-up. OS was measured from the date of operation to the date of death or last follow-up. The date of last follow-up and survival status were collected for all patients.
Statistical analyses. Continuous data are expressed as medians and ranges. Qualitative variables are expressed as frequencies (percentages). The Student t test or Mann-Whitney U test was used for intergroup comparisons of quantitative variables as appropriate, whereas the χ2 test or Fisher exact test was used to compare categorical data. Two-sided P-values of <0.05 were considered statistically significant. A survival analysis was conducted using the Kaplan-Meier product-limit method, and the significance of differences between survival curves was determined using the log-rank test. Multivariate comparisons of survival distributions were carried out using Cox proportional hazard models. To eliminate selection bias, patients were matched 1:1 by propensity score 47,48 based on their clinicopathological characteristics. All statistical analyses were performed using SPSS 23.0 (SPSS Inc., Chicago, IL).

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.