Advanced Intrahepatic Cholangiocarcinoma : Post Hoc Analysis of the ABC-01 ,-02 , and-03 Clinical Trials

Background: The incidence of intrahepatic cholangiocarcinoma (iCCA) is increasing. The aim of the study was to provide reference survival data for patients with advanced iCCA treated with first-line cisplatin-gemcitabine chemotherapy (current standard of care). Methods: Individual data from patients with iCCA recruited into the prospective, random assignment Advanced Biliary Tract Cancer (ABC)-01, -02, and -03 studies were retrieved. The prevalence and survival of liver-only iCCA was also assessed. Survival analysis was performed using univariate and multivariable Cox regression. All statistical tests were two-sided. Results: Of 534 patients recruited into the ABC-01, -02, and -03 studies, 109 (20.4%) had iCCA. Most patients (n1⁄486, 78.9%) had metastatic disease at the time of recruitment; 52 patients (47.7%) had liver-only disease. Following random assignment, 66 (60.6%) iCCA patients received cisplatin and gemcitabine. The median progression-free and overall survival (OS) were 8.4 months (95% confidence interval [CI] 1⁄4 5.9 to 8.9 months) and 15.4 months (95% CI 1⁄4 11.1 to 17.9 months), respectively. Of these 66 patients, 34 patients (51.5%) had liver-only disease. Following chemotherapy, 30 (45.5%) and 21 (31.8%) were progression-free at 3 and 6 months from chemotherapy commencement, respectively. The median OS for patients with liver-only iCCA at diagnosis and after 3 and 6 months of chemotherapy was 16.7 months (95% CI 1⁄4 8.7 to 20.2 months), 17.9 months (95% CI 1⁄4 11.7 to 20.9 months), and 18.9 months (95% CI 1⁄4 16.7 to 25.9 months), respectively. Multivariable analysis confirmed that iCCA had a longer OS compared with other non-iCCA biliary tract cancers (hazard ratio 1⁄4 0.58, 95% CI 1⁄4 0.35 to 0.95, P value 1⁄4 .03); liver-only iCCA patients also showed longer OS even though findings did not reach statistical significance (hazard ratio 1⁄4 0.65, 95% CI 1⁄4 0.36 to 1.19, P value 1⁄4 .16). Conclusions: Patients diagnosed with advanced iCCA have a better OS compared with other biliary tract cancers; a similar trend was identified for patients diagnosed with liver-only iCCA. These findings are to be considered for future clinical trial design. Biliary tract cancers (BTCs) include cholangiocarcinoma (intrahepatic [iCCA], hilar and distal), gallbladder, and ampullary carcinoma (1). BTCs are usually diagnosed in patients aged 50 to 70 years (2), and prognosis is poor (1,3,4). Although considered rare (2), their incidence is increasing due to a clinically significant increase in diagnosis of iCCA (5–7). Systemic chemotherapy is the only treatment approach that demonstrated survival benefit in randomized phase III studies for advanced BTC (8,9). Cisplatin and gemcitabine is currently accepted as a reference first-line treatment in advanced BTC in many countries (8) based on the Advanced Biliary Tract Cancer (ABC)-02 phase III clinical trial (10). These findings were confirmed in a Japanese randomized phase II study (BT22 study) (11), with no quality-of-life detriment in the combination arm identified (12). Other first-line chemotherapy options are under development (13–16). The role of second-line chemotherapy A R T IC LE Received: December 12, 2018; Revised: March 26, 2019; Accepted: March 19, 2019 © The Author(s) 2019. Published by Oxford University Press. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 1 JNCI J Natl Cancer Inst (2020) 112(2): djz071 doi: 10.1093/jnci/djz071 First published online May 11, 2019

Systemic chemotherapy is the only treatment approach that demonstrated survival benefit in randomized phase III studies for advanced BTC (8,9).Cisplatin and gemcitabine is currently accepted as a reference first-line treatment in advanced BTC in many countries (8) based on the Advanced Biliary Tract Cancer (ABC)-02 phase III clinical trial (10).These findings were confirmed in a Japanese randomized phase II study (BT22 study) (11), with no quality-of-life detriment in the combination arm identified (12).Other first-line chemotherapy options are under development (13)(14)(15)(16).The role of second-line chemotherapy remains unclear (17)(18)(19)(20)(21)(22) and is suitable for only approximately 15% of patients due to rapidly worsening performance status (19).Results of a randomized phase III study in this setting are awaited (23).
This post hoc analysis aimed to provide reference survival data to inform the design, sample size calculation, and feasibility of future studies exploring the role of systemic (including targeted) therapies and LDT in advanced iCCA.Potential trial designs together with factors to consider in designing such studies are discussed.

Study Design
A post hoc analysis of patient data collected as part of the prospective ABC-01 (48), -02 (10), and -03 (49) clinical trials was performed.These studies explored the role of first-line systemic chemotherapy in advanced BTCs (cisplatin and gemcitabine vs gemcitabine [ABC-01 and ABC-02] and cisplatin-gemcitabinecediranib vs cisplatin-gemcitabine-placebo [ABC-03]).All patients had provided written consent for participating in the above-mentioned trials, published elsewhere (10,48,49).The studies were sponsored by the University College of London and coordinated by the Cancer Research UK and UCL Cancer Trials Centre, which facilitated access to anonymized individualpatient data.
Patients diagnosed with iCCA were evaluable for this post hoc analysis.Clinical data, including demographics, baseline tumor markers, complete blood count, renal or liver profile, treatment characteristics, and response and survival data, were retrieved.The best radiological response achieved by each patient was classified based on the version of Response Evaluation Criteria in Solid Tumours (RECIST) employed in each study: RECISTv1.0(50) for ABC-01/-02, RECISTv1.1 (51) for ABC-03.
The primary aim of this analysis was to provide reference overall survival (OS) data of patients diagnosed with iCCA treated with first-line cisplatin and gemcitabine chemotherapy for future prospective studies.The subgroup of patients diagnosed with iCCA who were potentially eligible for LDT (defined as patients with liver-only disease) was analyzed separately.Secondary objectives included progression-free survival (PFS), description of demographic data of patients diagnosed with iCCA, and assessment of the frequency of iCCA patients with liver-only disease.Suitability for LDT at 3 and 6 months required meeting the above-mentioned criteria for LDT and being progression-free at 3 and 6 months following chemotherapy commencement, respectively.

Statistical Analysis
All eligible patients were included in the analysis.All patients diagnosed with iCCA were included for a summary of baseline characteristics.For survival analysis, only patients treated with the combination of cisplatin and gemcitabine were included (current standard of care for good-performance status patients).Because the addition of cediranib to cisplatin and gemcitabine did not result in a statistically significant impact on survival in the ABC-03 study (49), patients receiving both cediranib and placebo were included in the survival analysis of patients treated with cisplatin and gemcitabine.
PFS and OS were measured as the time from random assignment to progression or death (PFS) or death of any cause (OS).Patients who did not experience a PFS or OS event were censored at the date of last follow-up.Calculations of PFS and OS using as a starting point 3 and 6 months from random assignment were performed to identify patients potentially suitable for LDT at 3 and 6 months, respectively.Survival analysis was performed with the Kaplan-Meier method and Cox regression (univariate and multivariable analysis including variables statistically significant in the univariate analysis, defined as P value <.05); Ph test was used to test for the proportionalhazards assumption.For identification of prognostic factors, derived hazard ratio (HR), 95% confidence intervals (CI), and P values were reported.Stata v.12 software was employed.P values <.05 were considered statistically significant.All statistical tests were two-sided.

Study Population
Data from a total of 534 patients was retrieved (86, 324 and 124 patients from the ABC-01, -02, and -03 studies, respectively).Although the ABC-02 clinical trial reported a total of 410 patients, 86 were patients previously recruited into the ABC-01; such patients were included only once in this study.Thus, 324 patients from ABC-02 were eligible for this post hoc analysis.Of the whole population of eligible BTC patients, 318 (59.6%) were diagnosed with cholangiocarcinoma: 109 (20.4% of the whole population) had iCCA.By the end of follow-up, 82.9% of the whole population had died.The results are provided according to the cohorts specified in Supplementary Figure 1 (available online).

Whole Population of Patients Diagnosed with BTC
Baseline characteristics are shown in Table 1.Estimated median OS (regardless of type of chemotherapy administered; Supplementary Figure 1, CohortA, available online) for the whole population of 534 patients was 10.3 months (95% CI ¼ 8.8 to 11.7 months).When only patients receiving cisplatin and gemcitabine were included, the median PFS and OS were 7.9 months (95% CI ¼ 6.8 to 8.4 months) and 12.2 months (95% CI ¼ 10.7 to 13.6 months) (Figure 1A), respectively.

Whole iCCA Population: Baseline Characteristics and Chemotherapy Treatment
The characteristics of all patients diagnosed with iCCA (109 patients; Supplementary Figure 1 Of the 109 patients diagnosed with iCCA, 19, 61, and 29 patients were treated within the ABC-01, -02, and -03 clinical trials, respectively.Following study entry, 43 (39.5%) and 66 (60.6%) patients received gemcitabine or cisplatin and gemcitabine combination, respectively.Of the 66 patients who received cisplatin and gemcitabine chemotherapy (Supplementary Figure 1, CohortC, available online), 14 also received cediranib as per the ABC-03 study protocol (49).Within these 66 patients (Supplementary Figure 1, CohortC, available online), one patient (1.5%) achieved a complete radiological response.In addition, 15 (22.7%) and 26 (39.4%) patients achieved a partial response and stable disease as best response, respectively.Eleven patients (16.7%) had progression and 13 did not undergo radiological assessment.Supplementary Table 1 (available online) summarizes absolute responses achieved at each time point explored; the response rate for the iCCA population was similar to that for patients with non-iCCA BTC (Supplementary Table 2, available online).

Subgroup of Patients with iCCA and Liver-Only Disease
Fifty-two patients diagnosed with iCCA had liver-only disease (52 of 109, 47.7%); Table 1 includes the baseline characteristics for this patient population (Supplementary Figure 1, CohortD, available online).Of the 66 patients diagnosed with iCCA receiving cisplatin and gemcitabine chemotherapy (Supplementary Figure 1, CohortC, available online), 34 patients (51.5%) had liver-only disease at diagnosis (Supplementary Figure 1, CohortE, available online).Of these 34 patients, 7 patients (20.6%) achieved a partial response and 17 (50%) had stable disease as best response, which accounted for a disease control rate of 70.6%.Of the remaining 10 patients, 7 (20.6%)had progression as best response and 3 had no radiological assessment of response due to clinical progression (Supplementary Table 1, available online).The response rate was similar to that for patients with non-iCCA BTC (Supplementary Table 2, available online).Within these 34 patients following chemotherapy with cisplatin and gemcitabine, 30 (45.5%) and 21 (31.8%) were progression-free (liver-only disease) at 3 and 6 months, respectively.Supplementary Figure 2 (available online) summarizes the adapted consort diagram.Patients diagnosed with  *Median PFS, OS, and survival rates were calculated from the time of starting palliative chemotherapy.PFS and OS were also measured from 3 and 6 mo of starting chemotherapy to inform all possible trial design sample size calculations (such information can be found in Supplementary Table 4, available online).iCCA patients at risk [number of events] for each time point are as follows for PFS (at 3 mo: 66 [16]; at 6 mo: 50 [9]; at 9 mo: 40 [15]; at 12 mo: 23 [8]; at 18 mo: 15 [8]) and OS (3 mo: 64 [6]; 6 mo: 57 [7]; 9 mo: 50 [7]; 12 mo: 41 [4]; 18 mo: 36 [14]).Data for additional time points (mo 24, 30, and 36) can be found in Supplementary The median PFS and OS for the subgroup of patients who received cisplatin and gemcitabine diagnosed with liver-only iCCA (Supplementary Figure 1, CohortE, available online) at study entry and following 3 and 6 months from starting chemotherapy were 8.4 months (95% CI ¼ 5.9 to 10.02 months) and 16.7 months (95% CI ¼ 8.7 to 20.2 months), 8.6 months (95% CI ¼ 7.6 to 11.3 months) and 17.9 months (95% CI ¼ 11.7 to 20.9 months), and 11.1 months (95% CI ¼ 8.5 to 2.9 months) and 18.9 months (95% CI ¼ 16.7 to 25.9 months), respectively (Table 2; Supplementary Table 4, available online).Table 2 provides further PFS and OS rate information at different time points explored for this patient population (data for additional time points can be found in Supplementary Table 4, available online).No statistically significant differences in PFS (measured from study entry) were identified between liver-only iCCA (median ¼ 8.4 months, 95% CI ¼ 5.9 to 10.0 months) and non-iCCA BTC patients (median 7.9 months, 95% CI ¼ 6.5 to 8.4 months); log rank P = .37.

Prognostic Factors
OS was shorter in patients diagnosed with iCCA and treated with cisplatin and gemcitabine who had higher serum carcinoembryonic antigen (CEA) levels at baseline.This factor was an independent prognostic factor on multivariable analysis adjusted for other variables statistically significant in the univariate analysis (Table 3).
Multivariable analysis confirmed that higher platelet count and high CEA at baseline were associated with shorter OS in the population of patients with iCCA treated with cisplatin and gemcitabine with liver-only disease (Table 3).Although other factors affected OS in the univariate analysis, none were independently prognostic in the multivariable analysis.

Discussion
There is an urgent need for additional therapies for patients with BTCs.Patients with iCCA represent a specific subgroup for whom novel targeted therapies and LDT are emerging as promising therapeutic options.IDH and FGFR inhibitors have been tested in early-phase clinical trials, and phase III studies aimed at evaluating their efficacy are ongoing (9).In addition, current evidence supporting LDT in iCCA is of limited quality [category C; as per "Standards, Options and Recommendations" guidelines (52)], and therefore phase III randomized studies evaluating the impact of adding LDT to the current standard of care (cisplatin and gemcitabine) chemotherapy are planned to confirm previously suggested benefits.This post hoc analysis of the prospective ABC-01 (48), -02 (10), and -03 (49) clinical trials explored the outcome of patients diagnosed with iCCA who were treated with cisplatin and gemcitabine chemotherapy in order to inform the design of such studies.
Patients diagnosed with iCCA had a prolonged OS compared with the pooled patients with BTCs and non-iCCA, which makes them an attractive subgroup for development of further treatment approaches.A similar trend was identified when patients diagnosed with iCCA with liver-only disease (therefore suitable for LDT) were analyzed.Survival may have been underestimated by the inclusion of patients with Eastern Cooperative Oncology Group (ECOG) performance status 2.
Various findings support the fact that this prolonged OS is reflective of a different natural history rather than a better response to palliative chemotherapy (53).Firstly, the median PFS for cisplatin and gemcitabine-treated iCCA patients (8.4 months) was similar to the one achieved in the ABC-02 clinical trial (mixed population of BTC) [8.0 months (10)].Secondly, this difference in survival is unlikely to be related to a stage shift, because a similar percentage of patients with metastatic disease was identified within all patient subgroups (76.8%) and within the iCCA subgroup (78.9%).Thirdly, differential molecular findings described within iCCA, such as FGFR translocations and IDH mutations, have been suggested to affect survival (9).As an example, FGFR translocations [present in around 11%-45% of iCCAs ( 9)] in iCCA have been identified as a marker for more indolent behavior and better outcome (33,34).
Regarding the potential role of LDT in patients with iCCA, approximately one-half of the patients with iCCA in the current analysis had liver-only disease and would therefore be suitable for LDT.This supports the feasibility of recruiting to future prospective studies exploring LDT in this population of patients.Potential trial designs for incorporating LDT into the treatment management of patients with iCCA are summarized in Figure 2. Options could include the incorporation of LDT before any palliative chemotherapy (Figure 2, Option A), during chemotherapy (Figure 2, Option B), or at the end of 6 months of chemotherapy (Figure 2, Option C).
This study provides data on LDT suitability (defined as presence of liver-only disease) and survival at different time points (baseline, 3 months, and 6 months) to explore which time point would be more feasible to be explored.From the patient recruitment point of view, these results suggest that all three time points would be adequate for introduction of LDT into patients' pathway, although there is a progressive drop in the number of eligible patients.OS increases progressively between these 3 groups, as can be expected due to an immortality bias (54).Due to the longer survival of the patient population who would be progression-free at 6 months from starting palliative chemotherapy (and therefore eligible for LDT at this point), a study incorporating LDT at this time point would require prolonged follow-up with associated increased cost.
Few ongoing prospective studies are currently exploring the role of novel forms of LDT in iCCA, mainly focused on RE (Table 4).A phase II single-arm prospective study in Hong Kong is recruiting 30 patients with iCCA for treatment with RE followed by standard chemotherapy (equivalent to Figure 2, Option A) (NCT02167711).A similar design has been followed by the randomized phase II SIRCCA clinical trial (NCT02807181), which is investigating the use of the cisplatin and gemcitabine combination with and without RE as first-line treatment in patients with advanced iCCA.An innovative approach with concomitant chemotherapy and RE (RE will be administered on day 3 or 4 in combination with cisplatin and gemcitabine in cycles 1 and 2) is being explored in another phase I study (NCT02512692).One study is currently recruiting patients diagnosed with iCCA to explore other LDT approaches such as chemosaturation (NCT03086993).
Although the above aspects will inform sample size calculation and trial design, prognostic factors in patients with iCCAs treated with systemic chemotherapy should be considered for Downloaded from https://academic.oup.com/jnci/advance-article-abstract/doi/10.1093/jnci/djz071/5488118 by guest on 27 June 2019   values are two-sided.‡For the purpose of survival analyses, CEA and Ca19.9 variables were modified to obtain a clinically meaningful HR; CEA is presented as CEA/10 and Ca19.9 as CA19.9/10.000.Therefore, HR represents increased risk of death by every increase of 10 units and 10 000 units of CEA and Ca19.9, respectively.patient stratification.In a large series of patients with BTC treated with cisplatin and gemcitabine, the impact of factors such as hemoglobin, disease status, bilirubin, and neutrophils on both PFS and OS was identified and validated (61).The current post hoc analysis did not confirm these results, probably due to limitations of sample size in the subgroup analysis presented.Based on the current results, factors such as ECOG performance status and baseline CEA should be considered at the time of stratification for patients diagnosed with iCCA.In addition, retrospective studies including patients treated with LDT (specifically with RE) have identified multiple factors affecting survival that should be considered at the time of study    (42,55,57), bilobar disease (55), and liver tumor burden (56,62).The presence of portal vein thrombosis has been reported to have varying impacts on prognosis in different studies (42,55).The presence of extra-hepatic disease, if low volume, is not necessarily required as an exclusion criterion (58).Most of these previously described prognostic factors were not explored in the current analysis due to lack of available information and should be considered at the time of RE study design.A strength of this work is the fact that all data had been previously produced and quality assured as part of prospective clinical trials.This provides robustness to the results.In Regarding the limitations of this post hoc analysis, it is worth mentioning that the sample size was modest, because the analysis was focused on a small subpopulation of patients, which could have limited the survival analysis (particularly the Cox regression for identification of prognostic factors when multiple covariates were included).Some patients had received cediranib as part of the ABC-03 study, but this was not expected to affect the patients' outcomes (49).In addition, the actual percentage of patients suitable for LDT may have been under-or overestimated due to the lack of information for assessing the above criteria, such as tumor spread pattern and technical problems (ie, liver-lung shunt).In fact, only patients with liver-only disease were classified as suitable for LDT, excluding patients with liver-predominant disease.
In summary, the magnitude of benefit described in some of the studies focusing on iCCA is within the range that would be considered statistically significant if there was no knowledge of the survival of this patient cohort.This post hoc analysis demonstrates that patients with iCCA have a better outcome than other patients with BTC, and these survival figures should be considered at the time of future study design in this patient population.In addition, close to one-half of the patients diagnosed with iCCA are likely to have liver-only disease and therefore may be suitable for approaches involving LDT.

Figure 2 .
Figure 2. Potential trial designs for incorporating liver-directed therapy (LDT) to the management pathway of patients with intrahepatic cholangiocarcinoma (iCCA).

Figure
Figure includes details regarding percentage of patients potentially eligible for LDT (defined as liver-only disease; calculated using the whole iCCA population as reference) and expected outcomes (survival is provided in months) with cisplatin/gemcitabine for each potential scenario.CI ¼ confidence interval; OS ¼ overall survival; PFS ¼ progression-free survival; R ¼ random assignment.
, CohortB, available online) are summarized in Table1.Most of the patients with iCCA had

Table 1 .
Summary of baseline characteristics of all patients and those with iCCA included in ABC-01, -02, and -03 clinical trials *Other treatment included radiotherapy and photodynamic therapy.BTC ¼ biliary tract cancer; ECOG ¼ Eastern Cooperative Oncology Group; iCCA ¼ intrahepatic cholangiocarcinoma; n/a ¼ not applicable; No = number.†Site of disease was missing for some patients.(Figure 1B), respectively.Patients diagnosed with iCCA had statistically significant longer OS compared with those with other BTCs iCCA vs BTC [non-iCCA] Reference group, univariate HR ¼ 0.72, 95% CI ¼ 0.53 to 0.98, P ¼ .04)(Figure 1B); multivariable analysis for OS is shown in Supplementary Table 3 (available online) (adjusted multivariable HR ¼ 0.58, 95% CI 0.35 to 0.95, P ¼ .03).Further detail regarding PFS and OS rates at a number of time points is provided in Table 2 (data for additional time points can be found in Supplementary Table 4, available online).

Table 3 .
Prognostic factors of OS for patients with iCCA treated with cisplatin and gemcitabine The test of proportional hazards assumptions was tested in both multivariable analyses and showed that proportionality assumption in the Cox regression was held (P ¼ *

Table 4 .
(52)rted studies of radioembolization in iCCA The following definitions of level of evidence were used(52): Level A: there exists a meta-analysis of high standard or several randomized trials with consistent results; Level B: if randomized studies (level B1), therapeutic trials, quasi-experimental trials, or comparisons of populations (level B2) provide consistent results when considered together; Level C: there exist studies, therapeutic trials, quasi-experimental trials, or comparisons of populations, of which the results are not consistent when considered together; Level D: if scientific data either do not exist or there is only a series of cases; Expert agreement: data does not exist but the experts are unanimous in their judgment.CI¼ confidence interval; iCCA ¼ intrahepatic cholangiocarcinoma; mo ¼ months; nr ¼ not reached; OS ¼ overall survival; PFS ¼ progression-free survival; TTP ¼ time-to-progression. †All patients were refractory to standard chemotherapy.‡Patients in total (23.2%; 45 patients were iCCA).§Consecutive patients.Downloaded from https://academic.oup.com/jnci/advance-article-abstract/doi/10.1093/jnci/djz071/5488118 by guest on 27 June 2019 addition, a homogeneous population of patients treated with the same chemotherapy schedule was presented (treatmentnaı ¨ve patients treated with first-line cisplatin and gemcitabine) and the same chemotherapy protocol for dose reductions and duration of chemotherapy. *