Pattern of progression of intrahepatic cholangiocarcinoma: Implications for second‐line clinical trials

Abstract Background Intrahepatic cholangiocarcinoma (iCCA) is the second most frequent liver cancer. The overall survival of iCCA and other biliary tract cancers (BTC) remains poor. Recently, the ABC‐06 trial reported the superiority of FOLFOX vs clinical observation as a second‐line treatment. Still, the survival benefit was less than expected. We hypothesized that the pattern of progression of iCCA can drive post‐progression survival (PPS), similar to hepatocellular carcinoma. Methods Multicentre retrospective evaluation of consecutive iCCA patients who progressed after frontline systemic treatment with gemcitabine as monotherapy or in combination with platinum. Radiological assessment of progression was evaluated according to RECIST 1.1. The progression pattern was divided according to the presence/absence of new extrahepatic lesions (NEH). Results We included 206 patients from 5 centres. The median OS was 14.1 months and its independent predictors (hazard ratio [HR], 95% confidence interval [CI]) were previous surgery 0.699 [0.509‐0.961], performance status >2.445 [1.788‐3.344], permanent first‐line discontinuation 16.072 [5.102‐50.633], registration of ascites 2.226 [1.448‐3.420] or bilirubin >3 mg/dl 3.004 [1.935‐4.664] during the follow‐up, and disease progression 2.523 [1.261‐5.050]. The appearance of NEH independently predicted OS 2.18 [1.55‐3.06] in patients with radiological progression. Amongst 138 patients eligible for second‐line treatment, PPS was 16.8 and 5.9 months in cases without and with NEH, respectively (P = .001). Progression owing to NEH lesions was an independent predictor of PPS 1.873 [1.333‐2.662], together with performance status, time to progression to the frontline treatment, bilirubin >3 mg/dl and ascites. Conclusions PPS of iCCA is influenced by progression pattern, with important implications for second‐line trial design and analysis.


| INTRODUC TI ON
Intrahepatic cholangiocarcinoma (iCCA) is the second most frequent primary liver cancer, following hepatocellular carcinoma (HCC). 1 Similar to HCC, iCCA can be diagnosed in patients with pre-existing chronic liver disease during surveillance. 2 Currently, iCCA is responsible for 20% of liver-related deaths. 3  The search for prognostic tools for iCCA is a hot topic. 12,13 The determinants of post-progression survival (PPS) after a frontline therapy for iCCA have been scantly investigated. 14 The pattern of radiological progression has been shown to be an important driver of PPS in HCC progressing to sorafenib 15,16 and in extrahepatic malignancies. [17][18][19] In the case of HCC, the pattern of progression is now recognized as one of the critical elements in post-sorafenib trials and has been explicitly investigated in RCTs of second-line agents. 20,21 We hypothesized that the pattern of progression might also have a role in determining the PPS of iCCA, with implications in understanding some findings of the ABC-06 trial, optimizing the design of future second-line trials and informing patients in clinical practice.

| Outcomes and assessments
OS and time to progression (TTP) were calculated from the start of the frontline systemic treatment to death and disease progression, respectively. Radiological evaluation of response was done by computed tomography (CT) scan according to the response evaluation criteria in solid tumours (RECIST) v.1.1. 22 We categorized the type of progression as proposed by Reig et al. 15 Briefly, all patients with the appearance of new metastatic lesions were classified as progressors owing to new extrahepatic lesions (NEH). The remaining patients with radiological progression (ie ≥20% increase in tumour size against a known baseline lesion and/or new intrahepatic lesion) were considered to have a pattern of progression without NEH.
Radiology assessment was blinded to the evolution and outcome. Patients who died before the first imaging assessment were classified as progressors. PPS was measured from the date of detecting progression at radiology until the date of death or last follow-up.
The predictors of OS were determined in the whole cohort. We also assessed the impact of progression pattern on OS and PPS in patients with radiological progression and performed subanalyses of patients who had preserved PS and liver function at the time of progression and were therefore a candidate to second-line treatment.
Given the heterogeneity of the enrolment criteria of second-line iCCA and BTC trials, we simulated three scenarios at the progression: (1) patients with PS 0/1, bilirubin <3 mg/dl, no clinically relevant ascites, first-line treatment with any gemcitabine-containing regimen (similar to the pemigatinib and ivosidenib trials 23,24 ); (2) same characteristics but limited to patients who received gemcitabine in combination with any platinum agent (simulating the regorafenib REACHIN trial and the bintrafusp alfa INTR@PID BTC 047 study) 25 In the latter scenario, we also verified the role of sensitivity to platinum, as it was recognized as a prognostic factor after progression to GEMCIS in the ABC-06 trial. 11 Platinum sensitivity was defined as 'sensitive' (progression after 90 days of day 1 of the last cycle of frontline GEMCIS), resistant (progression within the first 90 days after completion of day 1 of the last cycle of frontline GEMCIS) or refractory (progression during frontline GEMCIS) 11 Finally, we verified whether the information about the pattern of progression could refine the prognostic ability of the Association des Gastro-Entérologues Oncologues (AGEO) CT2BIL score. 14

| Treatments
GEMCIS and GEMOX (gemcitabine-oxaliplatin) doublets were prescribed as the standard of care. Similar to other European centres, GEMOX was used as an equivalent to GEMCIS, owing to better tolerance and simpler outpatient administration. Management of toxicities included gemcitabine and platinum dose reductions or permanent discontinuation. GEM monotherapy was prescribed only in patients who had a contraindication to platinum at the baseline or who permanently discontinued platinum.

| Follow-up evaluations
Clinical and laboratory assessments were done before each chemotherapy administration. Radiology tumour evaluation was usually performed first at week 8 and afterwards approximately every 12 weeks.

| Statistical analysis
Categorical variables are described as frequencies and percentages and continuous variables as median and the interquartile range (IQR). Times to event data were estimated by Kaplan-Meier curves.
Fisher's exact test was used to compare categorical variables. The Mann-Whitney method was used to compare ordinal and continuous variables.
Moreover, we also assessed the impact of registering the following categorical variables (yes/no): total bilirubin ≥3 mg/dl, the appearance of palpable ascites, dose reduction, permanent discontinuation of systemic drugs, deterioration of PS (using PS 0 as a reference) and changes in CA19-9 (>37 IU/ml). All statistics involving evolutionary events were done by time-dependent covariate analyses. 15 Cox univariate and multivariate regression models with time-dependent covariates were used to estimate hazard ratios (HR). Statistical analyses were performed using SPSS version 24.0 and Stata version 16.0.

| Ethics
The Ethics Committee approved the study (protocol 78/2017/O/ OSSN), which was conducted according to the 1975 Declaration of Helsinki. Considering the retrospective design and the unfavourable prognosis of the investigated disease, the Ethics Committee waived the need for informed consent for deceased patients and for patients whose clinical conditions had worsened to a point in which they were not able to sign a valid consent. All of the remaining patients provided written informed consent for this study.

| Baseline parameters
Demographic, clinic and laboratory characteristics are summarized in Table 1. Seventy-eight patients (37.8%) had received prior surgery, and 30 (14.6%) had previously received adjuvant systemic treatment. Amongst the remaining evolutionary events, bilirubin >3 mg/dl occurred in 18 (8.7%) patients and palpable ascites was documented in 10 (4.9%) cases. The most frequent causes of hyperbilirubinemia were biliary obstruction (n = 10), followed by liver failure (n = 6), and cholangitis (n = 2). Imaging evidence of peritoneal carcinomatosis and/or positive cytology of the peritoneal fluid was found in the majority (n = 7) of ascitic patients.

| Radiological response
The best radiological response was progressive disease, stable dis-

| Predictors of OS
The univariate analysis of the whole population identified the following 5 baseline predictors of survival: previous surgery, multinodular disease, nodal involvement, ECOG-PS and GEM monotherapy (Table   S1). The multivariable Cox regression restricted them to ECOG-PS and prior surgery (Table 2).
Afterwards, we analyzed whether changes in the evolutionary covariates during the treatment had any impact on OS, with the statistical methodology that properly takes into account both baseline and evolutionary parameters. 15,27 The time-dependent analysis identified 4 additional possible predictors of survival: registration of hyperbilirubinemia and ascites, permanent first-line discontinuation, radiological progression. Because of GEM monotherapy, having resulted related to survival in the previous univariable analysis, we also included permanent platinum discontinuation (considering the time to event = 0 days in monotherapy patients) in a multivariable model.
The multivariable Cox analysis, confirmed registration of ascites or bilirubin >3 mg/dl, definitive first-line discontinuation and radiological tumour progression.
The baseline and evolutionary predictors were maintained after the exclusion of the 28 patients without radiological tumour progression. Moreover, progression due to new extrahepatic lesions was found to be an independent OS predictor in patients with radiological tumour progression (HR 1.878, 95% CI 1.350-2.612, P < .001).

| Post-progression survival
The post-progression analyses included 178 patients. Twenty-eight patients were excluded as they died before the first imaging followup (n = 24) or because they had not progressed at the time of the database lock (n = 4). The median PPS in patients with radiological progression was 9.0 months (95% CI 6.7-11.3). Ascites, hyperbilirubinaemia and ECOG-PS at the time of progression, together with TTP and progression owing to new extrahepatic lesions, were independent predictors of OS. (Table 3).

| Post-progression survival in potential candidates for second-line trials
We simulated different scenarios reflecting the enrolment criteria of the iCCA/BTC clinical trials ( Table 3) In the multivariable model, both platinum sensitivity (sensitive vs refractory/resistant) and pattern of progression were independent predictors of PPS (Table S2).

| DISCUSS ION
We demonstrated that progression correlated with survival, but also that specific patterns of progression implied different PPS and OS.    In conclusion, we found that the iCCA radiological pattern of progression influences the post-progression outcomes, both in a pure population of progressors and in candidates to second-line trials. This information can refine the information deriving from existing prognostic tools and have repercussions on the design of future clinical trials.

ACK N OWLED G EM ENTS
None.

CO N FLI C T O F I NTE R E S T S
The authors declare that the following financial interests/personal relationships may be considered as potential competing interests:

TR I A L R EG I S TR ATI O N N U M B E R
Not applicable.

PATI ENT CO N S ENT S TATEM ENT
Considering the retrospective design and the unfavourable prognosis of the investigated disease, the Ethics Committee waived the need for informed consent for deceased patients and for patients whose clinical conditions had worsened to a point in which they were not able to sign a valid consent. All of the remaining patients provided written informed consent for this study.

PE R M I SS I O N TO R E PRO D U CE M ATE R I A L FRO M OTH E R S O U RCE S
No material has been reproduced from other sources.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available because of privacy or ethical restrictions.