This is a retrospective study, and the study subjects consisted of 146 patients diagnosed with iCCA who were initially treated with HAIC or first-line SC between March 2016 and March 2022 at Sun Yat-sen University Cancer Center, China. Participants were included if they conformed to the following criteria: (Bridgewater et al., 2014) age 18 years old or elder; (Sirica et al., 2019) histopathological evidence confirmation of iCCA; (Valle et al., 2021) confirmed records of primary HAIC or first-line SC; (Oh et al., 2022) an Eastern Cooperative Oncology Group (ECOG) score of 2 or below; and (Banales et al., 2016) complete medical follow-up data. Patients were excluded based on the following exclusion criteria: (Bridgewater et al., 2014) patients with any other malignant tumor and (Sirica et al., 2019) patients who had contraindications to HAIC and SC.

HAIC was performed according to our previously reported protocol (Li et al., 2022). Femoral artery puncture and catheterization were performed on day 1 of the HAIC cycle, and the patient was transferred to the inpatient ward for drug infusion through the hepatic artery. Oxaliplatin was administered at 130 mg/m² from 0 to 2 h on day 1; leucovorin was administered at 400 mg/m² from 2 to 3 h on day 1; fluorouracil was administered at 400 mg/m² from hour 3 on day 1. Infusional fluorouracil was given at 2400 mg/m² over 23 h or 46 h. HAIC cycles were performed every 3 weeks. In the GEMCIS group, each cycle comprised cisplatin (25 mg per square meter of body-surface area), followed by gemcitabine (1,000 mg per square meter), which was administered on days 1 and 8 every 3 weeks. In the GEMOX group, each cycle comprised oxaliplatin (85 mg/m2) on day 1 and gemcitabine (1,000 mg per square meter) between days 1 and 8 every 3 weeks. HAIC or SC was suspended at 24 weeks or because of disease progression, unacceptable toxic effects, or patient’s own choice. As a part of treatment, HAIC or SC may be combined with the PD-1 inhibitor or tyrosine kinase inhibitor according to the needs of the condition and patient’s own choice.

All clinical data were obtained from the medical records of the Sun Yat-sen University Cancer Center. Demographic and clinical characteristics included age, sex, hepatitis infection status, ECOG, aspartate aminotransferase (AST), alanine transaminase (ALT), albumin (ALB), total bilirubin (TBIL), carcinoembryonic antigen (CEA), carbohydrate antigen 19–9 (CA19–9), white blood cell count (WBC), platelet count (PLT), creatinine (CRE), largest tumor size, tumor number, macroscopic vascular invasion, lymph node metastasis, extra-hepatic metastasis, and tumor–node–metastasis (TNM) stages. A summary of demographic and clinical characteristics is presented in Table 1. The blood tests and tumor burdens were measured within 5 days before the treatment. After treatment had been initiated, the radiological response was evaluated by magnetic resonance imaging (MRI) or computed tomography (CT) performed at baseline and every 6 weeks. Response Evaluation Criteria in Solid Tumors (RECIST)1.1 and modified RECIST (mRECIST) were used for evaluating the tumor response (Eisenhauer et al., 2009; Llovet and Lencioni, 2020).

Overall survival (OS) was defined as the time interval from first-line treatment to cancer-related death. Progression-free survival (PFS) was defined as the interval from first-line treatment to disease progression, iCCA relapse, or the date of death from iCCA or the date of the last follow-up. Intrahepatic progression-free survival (IPFS) was defined as the interval from the first-line treatment to intrahepatic tumor progression, iCCA relapse, or the date of death from iCCA or the date of the last follow-up, regardless of extrahepatic metastasis.

Non-normally distributed data were expressed as medians and ranges. Continuous parametric variables were analyzed by the unpaired Student’s t-test, and continuous non-parametric variables were analyzed by the Mann–Whitney U test. Categorical data were analyzed by Pearson’s correlation coefficient, chi-squared test with continuity corrections, or Fisher’s exact probability method. Forward LR-based univariate and multivariate Cox regression analyses were conducted to identify independent predictive variables. The OS and PFS were shown by Kaplan–Meier curves, and differences between the groups were compared using the results of the log-rank test. The p-value <0.05 was considered statistically significant. All the analyses were performed using SPSS 25.0 software (SPSS Inc., Chicago, IL) and R version 4.0.1.

Between March 2016 and March 2022, 146 patients diagnosed with iCCA who initially received HAIC or first-line SC were selected at Sun Yat-sen University Cancer Center, China. There were 75 patients in the HAIC group and 71 patients in the SC group (Figure 1). Detailed characteristics of each group are shown in Table 1. No significant baseline differences existed between the HAIC and SC groups.

In the HAIC group, the median age was 54 years old, 52 patients were male subjects, the largest tumor size of 25 (33.3%) patients was longer than 10 cm, a majority of patients had multiple tumors (66.7%), a total of 23 (30.7%) patients had macrovascular invasion, 51 (68%) patients had lymph node metastasis, and 17 (22.7%) patients had extra-hepatic metastasis. In the SC group, the median age was 57 years old, and 40 patients were male subjects, the largest tumor size of 14 (19.7%) patients was longer than 10 cm, a majority of patients had multiple tumors (69%), a total of 18 (25.4%) patients had macrovascular invasion, 47 (66.2%) patients had lymph node metastasis, and 24 (33.8%) patients had extra-hepatic metastasis. According to characteristics of a tumor, most patients in this study had large tumor burden and advanced iCCA.

Prognostic factors of all clinical variables were analyzed in univariate analysis. Univariate analyses showed that ECOG, tumor number, extra-hepatic metastasis, and TNM stages were significant risk factors for patients’ OS. Univariate analysis for PFS showed that ECOG, CA19–9, and extra-hepatic metastasis were significant risk factors. More details are described in Table 2. The multivariate Cox proportional analysis revealed that ECOG (p < 0.001) and extra-hepatic metastasis (p = 0.026) were significant and independent prognostic factors of OS (Table 2). The multivariate Cox proportional analysis revealed that ECOG (p < 0.001), CA19–9 (p = 0.02), macrovascular invasion (p = 0.02), and extra-hepatic metastasis (p = 0.001) were significant and independent prognostic factors of PFS (Table 2).

The median OS in the HAIC and SC groups was 18.0 and 17.8 months, respectively. Meanwhile, the median PFS times in the HAIC and SC groups were 10.8 and 11.4 months, respectively. There was no significant difference between the two groups in OS (p = 0.84; Figure 2A) and PFS (p = 0.59; Figure 2B). However, patients in the HAIC group had significantly longer IPFS than patients in the SC group (p = 0.035; Figure 2C). The median IPFS in the HAIC and SC groups was 13.7 and 11.4 months, respectively. The median follow-up in the HAIC and SC group was 16.8 and 17.7 months, respectively (Supplementary Figure S1). Patients in the SC group were divided into two subgroups (GEMCIS and GEMOX). GEMCIS and GEMOX were compared with HAIC in OS and PFS (Supplementary Figure S2).

The subgroup analyses of OS and PFS are shown in Figure 3. HAIC provided a clinical benefit for OS and PFS in tumor number subgroups. Single-tumor patients appeared to benefit more from it in terms of OS (p = 0.047; Supplementary Figure S3A) and PFS (p = 0.009; Supplementary Figure S3B). The intrahepatic tumor responses of the patients are shown in Table 3. On the basis of RECIST1.1 and mRECIST criteria, HAIC showed an ORR two times higher than SC (40% vs. 16.9%, p = 0.002, RECIST1.1; 45.3% vs. 21.2%, p = 0.002, mRECIST). The optimal response for intrahepatic target lesions by patients according to RECIST1.1 criteria is shown in the waterfall plot in Figure 4.

In general, the SC resulted in more AEs than those in HAIC (Table 4). The frequencies of rash (3 [4%] vs. 20 [28.2%]; p < 0.001), vomiting (27 [36%] vs. 51 [71.8%]; p < 0.001), fatigue (19 [25.3%] vs. 35 [49.3%]; p < 0.001), leukopenia (9 [12%] vs. 20 [28.2%]; p = 0.014), anemia (13 [17.3%] vs. 33 [46.5%]; p < 0.001), and sensory neuropathy (9 [12%] vs. 18 [25.4%]; p = 0.038) were lower in the HAIC group. Meanwhile, the overall incidence of serious AEs was higher in the SC group than that in the HAIC group. The frequencies of grades 3–4 vomiting (1 [1.3%] vs. 8 [11.2%]; p = 0.032), leukopenia (0 [0] vs. 5 [7%]; p = 0.025), and anemia (0 [0] vs. 6 [8.5%]; p = 0.012) were significantly higher in the SC group than those in the HAIC group. There were no significant differences in the frequencies of fever (15 [20%] vs. 10 [14.1%]; p = 0.343), abdominal pain (19 [25.3%] vs. 13 [18.3%]; p = 0.305), diarrhea (2 [2.7%] vs. 2 [2.8%]; p = 1.000), neutropenia (6 [8%] vs. 9 [12.7%]; p = 0.352), thrombocytopenia (8 [10.7%] vs. 16 [22.5%]; p = 0.053), elevated ALT (20 [26.7%] vs. 16 [22.5%]; p = 0.563), elevated AST (30 [40%] vs. 24 [33.8%]; p = 0.438), hyperbilirubinemia (12 [16%] vs. 10 [14.1%]; p = 0.746), hypoalbuminemia (37 [49.3%] vs. 34 [47.9%]; p = 0.861), and elevated creatinine (8 [10.7%] vs. 6 [8.5%]; p = 0.649). In the HAIC group, three (4%) patients delayed and discontinued treatment because of AEs. In the SC group, seven (9.86%) patients delayed and discontinued the treatment because of AEs.