Hepatic disease control in patients with intrahepatic cholangiocarcinoma correlates with overall survival

The role of locoregional therapy compared to systemic chemotherapy (SYS) for unresectable intrahepatic cholangiocarcinoma (IHC) remains controversial. The importance of hepatic disease control, either as initial or salvage therapy, is also unclear. We compared overall survival (OS) in patients treated with resection, hepatic arterial infusion pump (HAIP) chemotherapy, or SYS as it relates to hepatic recurrence or progression. We also evaluated recurrence after resection to determine the efficacy of locoregional salvage therapy.


| INTRODUCTION
Intrahepatic cholangiocarcinoma (IHC) comprises approximately 10% of all liver cancer cases, and its incidence has steadily increased over time. [1][2][3] Most patients present with locally advanced or metastatic disease at diagnosis and are treated with systemic chemotherapy (SYS) alone. Unfortunately, standard cytotoxic regimens have limited effectiveness, and while recent studies have demonstrated encouraging results with targeted therapies, these agents are only effective in a small subset of patients with specific genomic alterations. [4][5][6][7] Resection offers the best potential for long-term disease control but is only possible in onethird of patients, and ultimately >60% of resected patients will recur. 8,9 Failure to achieve hepatic disease control, due to either recurrence after resection or progression while on treatment for advanced disease, is common in IHC. As a result, liver-directed therapies may be appropriate and potentially effective in a large subgroup of patients. 10 In a recently completed Phase II study of hepatic artery infusion pump (HAIP) floxuridine (FUDR) combined with systemic gemcitabine and oxaliplatin (GEM/OX) in patients with unresectable IHC (N = 38), we reported a median overall survival (OS) of 25.0 months despite a 47% incidence of lymph node metastasis. 11 Conversely, the 3-year OS of liver-only disease in patients treated with gemcitabine and cisplatin alone is 3%. 12 However, the effectiveness of locoregional therapy when compared to SYS alone in patients with advanced disease has yet to be compared in a randomized clinical trial and remains controversial. Likewise, the role of locoregional treatment on intrahepatic recurrence following resection has not been established.
The goal of this study was to investigate the impact of hepatic disease control in IHC and its implications for OS. We analyzed a large, single-institution series of patients with biopsy-proven IHC treated by resection, HAIP FUDR, and SYS alone. We hypothesized that control of hepatic disease would be associated with improved survival, both in patients with liver-limited resectable and unresectable disease (without distant organ metastasis). Additionally, we analyzed a separate cohort of patients with liver-limited recurrence after curative resection to evaluate the impact of liver-directed locoregional therapy as salvage therapy in this patient population.

| Patients
This was a retrospective analysis of a prospectively maintained database of patients with biopsy-proven IHC from a single institution treated between 2000 and 2017 with either curative-intent resection, HAIP (with or without SYS), or SYS alone. Approval from the Institutional Review Board at Memorial Sloan Kettering Cancer Center (MSKCC) was obtained prior to data collection (protocol #16-698).
The operative approach and outcomes in patients with resectable IHC have been previously described. 13,14 The technical details of HAIP placement and outcomes from this therapy have similarly been described. 11,[15][16][17] A portal lymphadenectomy or targeted excisional lymphadenectomy in the porta hepatitis was performed in the HAIP and resected cohorts at the surgeon's discretion. Multifocal hepatic disease was defined as the presence of intrahepatic metastases or satellite nodules, either radiographically, intraoperatively, or on final pathology. Locally advanced unresectable disease was defined as anatomically unresectable disease due to insufficient future liver remnant or inability to obtain a margin negative resection. Tumor size was defined as the largest diameter of the tumor in the pathology specimen for resected patients, or the largest diameter seen on imaging for patients who were treated with HAIP. Patients who received HAIP but ultimately underwent curative-intent resection were analyzed in the HAIP cohort on an intention-to-treat basis. The impact of locoregional salvage therapies in patients with liver-limited recurrence was analyzed in the resection cohort. patients treated with HAIP versus SYS (HR 0.65; 95% CI 0.46-0.91; p = 0.01), which correlated with improved OS (HR 0.59, 95% CI 0.43-0.80; p < 0.001).

Conclusion:
In patients with liver-limited IHC, hepatic disease control is associated with improved OS, emphasizing the potential importance of liver-directed therapy.

K E Y W O R D S
bile duct cancer, biliary neoplasm, cholangiocarcinoma, hepatic artery pump, locoregional therapy, regional chemotherapy

| Statistical analysis
Continuous variables are described with median and range and compared using Wilcoxon rank sum test. Categorical variables are described with count and percentage and compared using Fischer's exact test. OS was defined as time from initial treatment initiation (resection, HAIP placement, or initiation of SYS) to the date of death or censored at the date of last follow-up. Progression-free survival (PFS) was measured from initial treatment initiation to time of first progression. A competing risk framework was used to analyze PFS outcomes in which death without progression was treated as a competing risk. Site-specific PFS treated recurrence at a different site as a competing risk. PFS outcomes were visualized using cumulative incidence function (CIF) curves and compared using Fine and Gray test. OS was visualized using the Kaplan-Meier method, and the log-rank test was used to assess differences between treatment groups.
Propensity score matching (PSM) was used to assess HAIP versus SYS. A propensity score model was built using logistic regression and included age, sex, race, lymphadenopathy, multifocality, and tumor grade. Patients were matched by propensity scores using a 1:1 greedy matching algorithm with a 0.25 (04.16%) standard deviation caliper. Balance of factors used in the PSM was assessed using Wilcoxon sign rank test for continuous variables, McNemar's test for categorical variables with two levels, and GEE model for categorical variables with >2 levels to account for correlation between matched pairs. Time to event outcomes were compared between matched groups using Cox proportional hazards models for OS and Fine and Gray models for PFS outcomes, with a robust sandwich estimator to account for matching.
Within the resected cohort, Cox proportional hazard models with time-dependent covariates were used to estimate the effect of recurrence on OS. An exploratory subgroup analysis on resected patients that recurred was performed to evaluate locoregional salvage therapy in patients with liver-limited recurrence. p < 0.05 was considered statistically significant. Data were analyzed using SAS Version 9.4 (SAS Institute) and R 4.0.2 (R Foundation for Statistical Computing).

| RESULTS
From 2000 to 2017, 714 consecutive patients were treated at MSKCC for IHC (Table 1). Of these, 219 patients (30.7%) presented with resectable disease, 316 (44.3%) presented with locally advanced, and 179 (25.1%) presented with metastatic disease ( Figure 1A). It is our practice to offer resection or HAIP placement to patients with either no evidence of cirrhosis, or well-compensated Child's A cirrhosis without portal hypertension. Similarly, patients treated with SYS were required to have adequate liver function. In our cohort, only 32/288 (11.1%) of patients treated with SYS alone had underlying cirrhosis. Patients with advanced underlying liver parenchymal disease, portal hypertension, or decompensated cirrhosis were typically not candidates for aggressive tumor-directed therapy of any time. The percentage of patients presenting with unresectable disease and treated with SYS alone has increased in recent years, notably when splitting the cohort between 2011-2017 and 2000-2010 (49.6% vs. 25.4%, p < 0.001) ( Figure 1B). The median follow-up for the cohort was 31.8 months. In the overall cohort, 150 (21%) patients presented with extrahepatic disease, of whom 145 (96.7%) were treated with SYS alone. Gemcitabine plus cisplatin was the most common first-line systemic regimen (n = 127, 44.1%) in SYS-only patients. HAIP patients received either FUDR alone (n = 53, 30.1%) or FUDR combined with either gemcitabine (n = 27, 15.3%), irinotecan (n = 37, 21.0%), or gemcitabine plus oxaliplatin (n = 46, 26.1%).
In HAIP patients, 55 (32.0%) received SYS prior to HAIP. Of these patients, 21 (38.2%) progressed on systemic therapy prior to HAIP placement, 24 (43.6%) had stable disease, 3 (5.5%) responded to systemic therapy but remained unresectable, and 5 (9.1%) developed doselimiting toxicity. One patient had a HAIP placed but was unable to receive liver-directed therapy; this patient was included in the HAIP cohort for analysis. PSM was performed adjusting for age, sex, race, lymph node metastasis, multifocal liver disease, and tumor grade (Table 3 and Figure S2). After matching, HAIP was associated with improved hepatic PFS with HAIP (HR 0.65, 95% CI 0.46-0.91; p = 0.012) ( Figure 3A Finally, we analyzed the impact of liver progression and extrahepatic progression on OS in this propensitymatched cohort. Progression in the liver, either alone or combined with extrahepatic recurrence/progression, was associated with a 10-fold increased risk in mortality (compared to eightfold for extrahepatic disease alone) ( Figure S3).

| DISCUSSION
In this study, we demonstrate the importance of hepatic disease control to outcome in patients with IHC. For those eligible, curative-intent resection provides the best hepatic disease control and survival. Although the median OS for curative-intent resection in our series was encouraging at 48.1 months, intrahepatic recurrence is common, with a cumulative incidence of 41% at 2 years. For those patients with unresectable disease, HAIP-delivered chemotherapy was associated with a 10.9-month survival advantage compared to SYS alone. Moreover, recurrence or progression in the liver, either alone or simultaneously with extrahepatic recurrence or progression, was associated with a 10-fold increased risk of mortality (compared to eightfold for extrahepatic disease alone). In a PSM cohort, patients treated with HAIP had a 35% lower risk of simultaneous intra-and extrahepatic progression and a 44% lower risk of isolated intrahepatic progression (compared to those treated with SYS alone). This was accompanied by a 9.4-month survival difference in favor of HAIP between the two cohorts. Lastly, in resected patients with isolated LR, treatment with liver-directed therapy was associated with a 15.4-month improvement in median survival. Taken together, these data demonstrate both the adverse impact of intrahepatic recurrence/progression on survival, as well as the ability of HAIP to provide better hepatic disease control and improved survival compared to treatment with SYS alone. In a systematic review and meta-analysis by Choi et al. the authors identified four series describing early intrahepatic recurrence (defined as either 6, 12, or 24 months after resection) and noted a 49.6% 24-month intrahepatic recurrence rate, similar to our finding of 41%. 18 Unfortunately, the majority of patients will expereince liver recurrence, with most reports describing >50% isolated liver or simultaneous liver and extrahepatic T A B L E 2 Characteristics of patients with liver recurrence only after resection. recurrence. 9,18,19 In the same report, the authors pooled prognostic factors and found that locoregional therapy in the form of adjuvant chemoradiation was the only factor associated with a decreased risk of intrahepatic recurrence ( between those who did not receive treatment and those who did (20 vs. 46 months, respectively). 20 In this same series, 84.6% of recurrences involved the liver, with 64.1% isolated intrahepatic and 20.5% simultaneous intra-and extrahepatic recurrence.

Locoregional therapy (n = 30) No locoregional therapy (n = 35) p-Value
Yamashita et al. observed a higher risk of liver failure at the time of death in patients treated with chemotherapy compared to patients treated with radiation or resection. 21 Additionally, local therapy was the sole predictor of death without liver failure on multivariable analysis. 21 Consistent with prior studies, rates of intrahepatic progression and recurrence are significant; however, locoregional therapy (either in the form of HAIP or otherwise) is associated with both decreased hepatic disease recurrence/progression and survival. The benefit of local therapy may also extend to hepatocellular carcinoma (HCC). Ablation may convey some benefit for tumors <3 cm and multifocal disease; however, there is an increased risk of local recurrence compared to resection for larger tumors. 22 However, almost two decades ago, TACE was demonstrated to confer a survival benefit compared to supportive treatment for patients with "intermediate stage" tumors (Barcelona Clinic Liver Cancer [BCLC] B). 23,24 More recently, a Phase III trial of HAI with fluorouracil, leucovorin, and oxaliplatin (FOLFOX) demonstrated a 7-month survival benefit (23.1 vs. 16.1 months) compared to TACE for advanced, unresectable HCC. 25 Although resection provides the best potential for a long-term cure, disease-specific factors, such as lymph node metastases and tumor multifocality (either satellite lesions or intrahepatic metastases), can significantly impact post-resection survival. [26][27][28] We recently demonstrated that for patients with lymph node metastases, there was no difference in survival between resection or HAIP (19.7 vs. 18.1 months, respectively). 29 Similarly, using a multiinstitutional cohort of patients from 12 centers who underwent resection of multifocal IHC, there was no difference in survival compared to HAIP (18.9 vs. 20.3 months, respectively). Unfortunately, due to the rarity of IHC, available literature on other locoregional therapies is heterogeneous and of insufficient quality to generate high-level recommendations. However, in a systematic review and pooled analysis by Edeline et al. the authors report that the pooled mean OS for intra-arterial therapy, including selective internal radiation therapy, TACE, and HAIP, was 25.2 months for first-line treatment with concomitant systemic chemotherapy. 30 Thus, while resection remains the optimal treatment, liver-directed chemotherapy may be as efficacious as resection at controlling liver disease in the case of adverse disease-specific factors.
There are several limitations to this study, including its retrospective and single-institution nature. Although the design employed was designed to limit selection bias, this can only be truly controlled by prospective randomization, and there may be other factors not in the cohort matching that might have influenced clinicians not to go with HAIP. The median OS of 10.9 months for SYS treated patients is lower than the landmark analysis of the ABC-01, -02, and -03 trials which demonstrated a 15.4-month median OS for liver-limited disease. 12 However, clinical trials result in a highly selected cohort of patients and similarly, in the Phase II trial of HAI-FUDR for unresectable IHC conducted at our institution, the median OS was 25.0 months. 11 Additionally, 31.3% of patients in the HAIP cohort received SYS prior to initiation of HAIP, which may have selected for patients with stable disease on systemic therapy. However, 32 were included in the PSM survival analysis and 11 were started on HAIP in the context of progression of disease, suggesting favorable disease biology did not artificially inflate survival data. Finally, although these data suggest an association of liver progression or recurrence with OS, the exact cause of death was not determined. Liver progression can lead to death in a number of ways, including hepatic failure, and uncontrolled cholangitis/sepsis, and while one or more of these mechanisms were likely involved, the precise mechanism cannot be ascertained in a retrospective analysis.
In conclusion, these data demonstrate the adverse impact of loss of liver disease control and the ability of liver-directed chemotherapy to prevent intrahepatic recurrence/progression. These data demonstrate that liver-specific treatment with locoregional therapy is an important consideration in addition to treatment with cytotoxic regimens in patients with IHC. Previous Phase II clinical trials have demonstrated the safety and potential efficacy of HAI FUDR in combination with gemcitabine and oxaliplatin for unresectable IHC. 31 Our institution has an ongoing prospective trial that will compare this combination to gemcitabine and oxaliplatin alone (NCT04891289) to both determine the survival benefit of liver-directed chemotherapy and better define factors associated with IHC pathogenesis and outcome.