Biliary tract cancer (BTC) is a group of heterogeneous diseases that originate from the bile duct and gallbladder and includes intrahepatic cholangiocarcinoma (IHCC), extrahepatic cholangiocarcinoma (EHCC), and gallbladder cancer.¹ BTC is rare and accounts for less than 1% of global cancer cases; however, its incidence varies worldwide and has increased over the past four decades.^2-5 The prognosis of BTC is poor, with a 5-year survival rate of <20%.⁶ Surgical resection is the only potentially curative option for BTC; however, nearly two-thirds of patients are diagnosed at an advanced stage with inoperable disease; more than half of patients that undergo surgery experience recurrence.^7-10

The gemcitabine and cisplatin doublet chemotherapy has been a first-line treatment for BTC for over 10 years.¹¹ However, even with this potent doublet chemotherapy, the median overall survival (OS) remains <1 year.^12,13 Other chemotherapy combination regimens have been examined as candidate treatments that may improve survival outcomes in patients with advanced BTC.^14-17 Among them, triplet chemotherapy regimen that includes gemcitabine, cisplatin, and nanoparticle albumin-bound (nab)-paclitaxel has shown promising results in a phase II study with the median progression-free survival (PFS) of 11.8 months and the median OS of 19.2 months.¹⁸ Consequently, the triplet chemotherapy regimen is currently being used in clinical practice and a phase III trial of this regimen is ongoing (NCT03768414).

Nevertheless, real-world patient outcomes have not been compared between the doublet and triplet chemotherapy regimens to date. The present study aimed to show the efficacy and safety profiles of triplet chemotherapy in a real-world setting and compare them with those of the current standard doublet chemotherapy.

1. Patients

We identified 41 patients diagnosed with advanced BTC (histologically confirmed IHCC, EHCC, and gallbladder cancer) from September 1, 2019, to December 31, 2020, who received triplet chemotherapy at the Seoul National University Bundang Hospital, Seongnam, Korea. Among them, those who had undergone prior palliative chemotherapy or were lost to follow-up before the first response evaluation were excluded. Finally, 34 patients were included in the triplet chemotherapy group. In addition, we reviewed medical records of patients with advanced BTC who received doublet chemotherapy as first-line chemotherapy between January 1, 2011, and December 31, 2018. The total number of advanced BTC patients treated with doublet chemotherapy was 261.

We performed propensity-score matching analysis with the nearest neighbor matching method, using the ratio of one-to-one for two different chemotherapy groups matched on age, sex, tumor location, stage, Eastern Cooperative Oncology Group performance status, and carbohydrate antigen 19-9 (CA19-9) to overcome the heterogeneity of baseline characteristics. Finally, a matched cohort of 68 patients (n=34 patients per group) was obtained for further analyses. This study was approved by the Institutional Review Board of the Seoul National University Bundang Hospital (IRB number: L-2021-642) and informed consent was waived due to study design.

2. Treatment regimens

In the doublet chemotherapy group, patients received gemcitabine of 1,000 mg/m² and cisplatin of 25 mg/m² as a standard dose on days 1 and 8 every 21 days.¹¹ Patients in the triplet chemotherapy group were treated with reduced-dose regimen of 800 mg/m² gemcitabine, 25 mg/m² cisplatin, and 100 mg/m² nab-paclitaxel, as a standard dose on days 1 and 8 every 21 days.¹⁸ Dose modification or interruption were at the discretion of the attending physician. Granulocyte colony stimulating factors (GCSF) were administered therapeutically or prophylactically at the attending physician’s discretion.

3. Endpoints and assessment

The primary endpoints were PFS and OS of the triplet and doublet chemotherapy groups. Treatment responses were evaluated continuously at the intervals of 2 or 3 months with computed tomography or magnetic resonance imaging, according to the Response Evaluation Criteria in Solid Tumors, version 1.1. The median follow-up duration was evaluated in all patients including those who died; the median follow-up duration, PFS, and OS were calculated from the day of chemotherapy initiation. When a patient subsequently underwent surgery or radiotherapy, the median follow-up duration and PFS were measured until the time of surgery or radiotherapy.

The secondary endpoints were overall response rate (ORR) and disease control rate (DCR) in both groups. In addition, the median number of treatment cycles and the mean relative dose intensity were calculated to confirm that each chemotherapy agent was suitably administered. Lastly, safety profiles including the rates of hematologic adverse events, non-hematologic adverse events, and use of GCSF prophylaxis were reviewed in both groups. Adverse events were monitored at every hospital visit and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.0.

4. Statistical analysis

Continuous variables were analyzed using the t-test, paired t-test, or Mann-Whitney U test, depending on whether the assumption of normal distribution was satisfied or not. Categorical variables were compared with the McNemar test, chi-square test, or Fisher exact test. The Kaplan-Meier curves and Cox proportional hazard models were used to perform survival analysis. All statistical analyses were performed using SPSS version 25.0 (IBM Corp, Armonk, NY, USA).

1. Baseline characteristics

A total of 34 patients in each of the triplet and doublet chemotherapy groups were matched (Fig. 1). Before propensity-score matching, there were differences between the groups in several baseline variables; the baseline characteristics became balanced after matching (Table 1). In the matched cohort, the mean age (standard deviation) in the triplet and doublet groups was 62.8 (±10.8) and 63.6 (±10.1) years, respectively. There were more female than male patients in both groups (19/34 [55.9%] vs 18/34 [52.9%], respectively). The most common tumor location was EHCC (17/34 per group, 50.0%), and metastatic cancer was more common than locally advanced cancer in both groups (19/34 [55.9%] vs 20/34 [58.8%], respectively). The median CA19-9 level (interquartile range) was 350 (27–1,625) U/mL in the triplet chemotherapy group and 221 (33–2,125) U/mL in the doublet chemotherapy group.

Table 1 . Baseline Characteristics before and after Propensity-Score Matching.

Characteristics	Unmatched cohort		p-value	Matched cohort		p-value
	Triplet (n=34)	Doublet (n=261)		Triplet (n=34)	Doublet (n=34)
Age, yr	62.8±10.8	64.4±10.5	0.421	62.8±10.8	63.6±10.1	0.768
Sex			0.009			1.000
Male	15 (44.1)	174 (66.7)		15 (44.1)	16 (47.1)
Female	19 (55.9)	87 (33.3)		19 (55.9)	18 (52.9)
Site			0.003			1.000
IHCC	11 (32.4)	169 (64.8)		11 (32.4)	11 (32.4)
EHCC	17 (50.0)	69 (26.4)		17 (50.0)	17 (50.0)
GBC	6 (17.6)	23 (8.8)		6 (17.6)	6 (17.6)
Stage			0.002			1.000
Localized	15 (44.1)	54 (20.7)		15 (44.1)	14 (41.2)
Metastatic	19 (55.9)	207 (79.3)		19 (55.9)	20 (58.8)
ECOG			0.824			1.000
0	15 (44.1)	101 (38.7)		15 (44.1)	15 (44.1)
1	19 (55.9)	151 (57.9)		19 (55.9)	19 (55.9)
2	0	9 (3.4)		0	0
CA19-9, U/mL	350 (27–1,625)	510 (52–2,700)	0.210	350 (27–1,625)	221 (33–2,125)	0.949

Data are presented as mean±SD, number (%), or median (interquartile range)..

IHCC, intrahepatic cholangiocarcinoma; EHCC, extrahepatic cholangiocarcinoma; GBC, gallbladder cancer; ECOG, Eastern Cooperative Oncology Group performance status; CA19-9, carbohydrate antigen 19-9..

Figure 1. Flowchart of matched cohort creation.
BTC, biliary tract cancer.

2. Efficacy

During the median follow-up duration (standard deviation) of 8.7 months (95% confidence interval [CI], 6.3 to 11.2) in the triplet chemotherapy group, 21 of 34 patients (61.8%) experienced disease progression, three patients (8.8%) underwent surgery, and two patients (5.9%) received radiotherapy. In the doublet chemotherapy group, 33 of 34 patients (97.1%) experienced disease progression or death, except for one patient who underwent surgery.

The median PFS was 7.5 months (95% CI, 4.1 to 10.9) in the triplet chemotherapy group and 7.2 months (95% CI, 5.6 to 8.9) in the doublet chemotherapy group (hazard ratio, 0.93; 95% CI, 0.53 to 1.62; p=0.793) (Fig. 2A). The median OS estimates in the triplet and doublet groups were 13.7 months (95% CI, 8.8 to 18.7) and 12.2 months (95% CI, 8.4 to 16.0), respectively (hazard ratio, 0.73; 95% CI, 0.38 to 1.41; p=0.354) (Fig. 2B). The median PFS and OS in the triplet chemotherapy group did not show any significant improvement, as compared with those in the doublet chemotherapy group. Adjustment for other baseline variables did not alter these outcomes (Table 2).

Table 2 . Multivariate Analysis of Progression-Free Survival and Overall Survival.

Prognostic factors	Progression-free survival			Overall survival
	HR (95% CI)	p-value		HR (95% CI)	p-value
Triplet vs doublet	1.06 (0.55–2.04)	0.857		0.70 (0.33–1.48)	0.351
Age	1.01 (0.97–1.06)	0.519		1.00 (0.95–1.05)	0.963
Sex
Male	Reference			Reference
Female	1.58 (0.76–3.32)	0.224		1.27 (0.58–2.78)	0.548
Tumor site
IHCC	Reference			Reference
EHCC	1.50 (0.70–3.21)	0.294		0.87 (0.38–2.00)	0.743
GBC	0.94 (0.39–2.41)	0.955		1.13 (0.44–2.92)	0.803
Stage
Localized	Reference			Reference
Metastatic	3.29 (1.63–6.64)	0.001		2.70 (1.27–5.76)	0.010
ECOG
0	Reference			Reference
1	1.10 (0.50–2.43)	0.819		1.06 (0.44–2.56)	0.905
CA19-9
<300 U/mL	Reference			Reference
≥300 U/mL	1.29 (0.66–2.54)	0.458		1.24 (0.63–2.43)	0.540

HR, hazard ratio; CI, confidence interval; IHCC, intrahepatic cholangiocarcinoma; EHCC, extrahepatic cholangiocarcinoma; GBC, gallbladder cancer; ECOG, Eastern Cooperative Oncology Group performance status; CA19-9, carbohydrate antigen 19-9..

Figure 2. Kaplan-Meier curves for median progression-free survival and overall survival. (A) The median progression-free survival was 7.5 months in the triplet chemotherapy group and 7.2 months in the doublet chemotherapy group (hazard ratio, 0.93; 95% confidence interval, 0.53 to 1.62; p=0.793). (B) The median overall survival was 13.7 months in the triplet chemotherapy group and 12.2 months in the doublet chemotherapy group (hazard ratio, 0.73; 95% confidence interval, 0.38 to 1.41; p=0.354).

In the triplet chemotherapy group, one patient showed complete response and nine patients showed partial response, yielding the ORR of 29.4% (Table 3). The DCR for the triplet chemotherapy group was 79.4%, including 16 patients with stable disease. In the doublet chemotherapy group, nine patients showed partial response, accounting for the ORR of 26.5%. The DCR was 85.3% with additional 20 patients who had stable disease. There was no significant difference in ORR and DCR between two groups.

Table 3 . Response to Treatment in the Matched Cohort.

Response	Triplet (n=34)	Doublet (n=34)	p-value
Complete response	1 (2.9)	0	-
Partial response	9 (26.5)	9 (26.5)	-
Stable disease	17 (50.0)	20 (58.8)	-
Progressive disease	7 (20.6)	5 (14.7)	-
Overall response	10 (29.4)	9 (26.5)	0.595
Disease control	27 (79.4)	29 (85.3)	0.525

Data are presented as number (%)..

3. Dose administration and adverse events

The median number of treatment cycles in the triplet group and doublet group was 6 (interquartile range, 3 to 12) and 8 (interquartile range 4 to 13), respectively (Table 4). The mean relative dose intensity of each chemotherapy agent in triplet chemotherapy group was comparable to doublet group (gemcitabine of 85.2%, cisplatin of 81.1%, and nab-paclitaxel of 73.4% in triplet group vs gemcitabine of 84.1% and cisplatin of 79.1% in doublet group). Seven patients (20.6%) in triplet group had to discontinue at least one chemotherapy agent due to adverse events (nab-paclitaxel of six patients and cisplatin of one patient), but three patients (8.8%) in doublet group (cisplatin of three patients).

Table 4 . Administration of Chemotherapy Agents in Both Groups.

Administration	Triplet	Doublet	p-value
Treatment cycle	6 (3–12)	8 (4–13)	0.416
Relative dose intensity, %
Gemcitabine	85.2±15.0	84.1±11.7	0.733
Cisplatin	81.1±16.3	79.1±17.5	0.636
Nab-paclitaxel	73.4±21.5	-	-

Data are presented as median (interquartile range) or mean±SD..

The most frequent treatment-related adverse events in both groups were hematologic adverse events. Grade 3 or higher hematologic adverse events were more frequent in the triplet chemotherapy group than in the doublet chemotherapy group (neutropenia rates of 29.4% vs 20.6%; febrile neutropenia rates of 14.7% vs 8.8%; anemia rates of 26.5% vs 20.6%; thrombocytopenia rates of 20.6% vs 17.6%, respectively) (Table 5). In particular, neutropenia was relatively common in the triplet chemotherapy group, despite six patients (17.6%) in this group having received prophylactic GCSF and no one in doublet chemotherapy group.

Table 5 . Safety Profiles in the Matched Cohort.

Safety profiles	Triplet (n=34)	Doublet (n=34)	p-value
Any grade ≥3 AEs	20 (58.8)	14 (41.2)	0.146
Grade ≥3 hematologic AEs
Neutropenia	10 (29.4)	7 (20.6)	0.401
Anemia	9 (26.5)	7 (20.6)	0.567
Thrombocytopenia	7 (20.6)	6 (17.6)	0.758
Febrile neutropenia	5 (14.7)	3 (8.8)	0.452
Prophylactic GCSF use	6 (17.6)	0	0.025
Grade ≥3 non-hematologic AEs
Infection	2 (5.9)	2 (5.9)	1.000
Diarrhea	2 (5.9)	0	0.493
Constipation	1 (2.9)	1 (2.9)	1.000
Nausea	1 (2.9)	3 (8.8)	0.614
Vomiting	1 (2.9)	1 (2.9)	1.000
Rash	1 (2.9)	0	1.000
Liver dysfunction	1 (2.9)	0	1.000
Renal dysfunction	0	0	-
Neuropathy	2 (5.9)	1 (2.9)	1.000
Thromboembolic event	1 (2.9)	0	1.000

Data are presented as number (%)..

AE, adverse event; GCSF, granulocyte colony stimulating factors..

The present study compared the efficacy of a triplet chemotherapy regimen (gemcitabine, cisplatin, and nab-paclitaxel) with that of a doublet chemotherapy regimen (gemcitabine and cisplatin) in the treatment of advanced BTC. Contrary to expectations, in the present study, survival outcomes associated with the triplet chemotherapy (median PFS of 7.5 months, median OS of 13.4 months, and ORR of 29.4%) were not superior to those associated with the doublet chemotherapy.

In the present doublet chemotherapy group, the median PFS was 7.2 months, median OS was 12.2 months, and ORR was 26.5%. These findings were comparable to those of the ABC-02 study, in which the corresponding outcomes for the doublet chemotherapy group were 8.0 months, 11.7 months, and 26.1%, respectively.¹¹

However, the efficacy of the triplet chemotherapy regimen in this study was inferior to that of previous phase II study with triplet chemotherapy regimen (median PFS of 11.8 months, median OS of 19.2 months, and ORR of 45.1%).¹⁸ This discrepancy may be due to the differences in the participants’ baseline characteristics. In this study, patients were older (mean age 62.8 years vs 58.4 years in the previous study), the most common tumor location was EHCC (17/34, 50.0%) in contrast to IHCC (38/60, 63%) in the previous study, and the median CA19-9 level was higher (350 U/mL vs 99 U/mL in the previous study).¹⁸ Age and CA19-9 levels are well-documents prognostic factors;^19,20 nevertheless, differences in prognoses according to tumor location remain controversial.^13,21,22 In a large retrospective study with 740 advanced BTC patients treated with the doublet chemotherapy, tumor location did not affect survival outcomes.¹³ However, another collective study with individual data from ABC trials showed that IHCC was associated with relatively better OS.²¹ Therefore, considering the effect of potential prognostic factors, we performed a propensity-score matching study showing no difference in survival outcomes between the triplet and doublet chemotherapy groups, even though the doublet chemotherapy group showed outcomes comparable to those reported in the ABC-02 study.¹¹

In the present study, the triplet chemotherapy group was initially treated with a reduced-dose regimen, which was recommended in the previous phase II trial. Given that only 28 of 60 patients (46.7%) in the previous study received the reduced-dose regimen and the ORR in reduced-dose group was lower than that in the high-dose group in the previous study (39.1% vs 50.0%),¹⁸ a reduced-dose may be associated with poor outcomes in the triplet chemotherapy group of this study. However, the reduced-dose group in the previous study included only 10 (35.7%) progression or mortality events, obscuring efficacy outcomes of the reduced-dose regimen. Further large-scale studies are required to confirm the efficacy and safety profiles of the reduced-dose regimen.

In addition, the differences in efficacy outcomes may result from genomic diversity of BTC patients. Several genomic studies in BTC have shown that the molecular profile of BTC varies depending on its anatomical location and underlying etiology.²³ Moreover, the prognosis may differ according to specific genetic alterations such as KRAS or TP53, or clusters by mutational signature.^23-25 These genomic findings may help account for the differences in BTC outcomes, which seem to depend on multiple factors such as tumor location, risk factor or etiology, and genetic predisposition.^23-26 Consequently, further studies on the clinical outcomes of BTC require models that account for diverse tumor biology.

In the present study, hematologic adverse events were the most common type of treatment-related adverse events in both groups. Meanwhile, grade 3 or higher hematologic adverse events, in particular, neutropenia, were more frequent in the triplet chemotherapy group than in the doublet chemotherapy group. The safety profile of the triplet chemotherapy regimen, including hematologic and non-hematologic adverse events, in this study was similar to that in the previous phase II study, except that severe neutropenia was less common in this study (29.4% vs 40.4%).¹⁸ However, it should be noted that 30% of the patients experienced grade 3 or higher neutropenia, even though all patients in the present triplet chemotherapy group received a reduced-dose regimen and 18% of the patients received prophylactic GCSF.

Our study has several limitations. First, the number of patients treated with triplet chemotherapy was small because this regimen is not yet recommended as standard treatment for advanced BTC. Second, this was a retrospective study, so there was no standardized protocol in the treatment process such as dose modification or GCSF administration. However, chemotherapy administration in both groups, as expressed by the median number of treatment cycles and the mean relative dose intensity, were similar enough to compare the efficacy of two different chemotherapy regimens. Lastly, follow-up duration of patients in triplet chemotherapy group was relatively short. However, the triplet chemotherapy group (n=34) included 21 (61.8%) patients with disease progression event, five (14.7%) patients with another subsequent treatment such as surgery or radiotherapy who could not provide additional survival information associated with triplet chemotherapy, and five (14.7%) patients with a longer follow-up duration than the median PFS. Consequently, these data were considered sufficient for the assessment of survival outcomes in the triplet chemotherapy group.

Our study has some strengths. First, to the best of our knowledge, this is the first study on the new triplet chemotherapy regimen in a real-world setting; in addition, this is the first study to compare patient outcomes between the triplet and doublet regimens. Although this was a retrospective study, the potential confounding factors were well balanced in both groups by propensity-score matching.

In conclusion, the use of gemcitabine, cisplatin, and nab-paclitaxel in patients with advanced BTC did not show survival outcomes better than those associated with the use of gemcitabine and cisplatin, which is the current standard regimen. Moreover, treatment-related severe adverse events, such as neutropenia, were more common in the triplet chemotherapy group than in the doublet chemotherapy group. The use of the triplet chemotherapy in clinical practice requires further evidence from a phase III trial.

No potential conflict of interest relevant to this article was reported.

Conception and design: K.J., J.H.H. Acquisition of data: K.J., J.P., J.H.J., J.C.L., J.K. Analysis: K.J., J.H.H. Interpretation of the data and drafting of the article: K.J., J.H.H. Critical revision of the article for important intellectual content: J.P., J.H.J., J.C.L., J.K., J.H.H. Final approval of the article: all authors.