Patterns and risk factors of early recurrence after radical resection for intrahepatic cholangiocarcinoma

Abstract

BACKGROUND

Intrahepatic cholangiocarcinoma (ICC) represents the second-most prevalent primary hepatic malignancy, demonstrating increasing worldwide occurrence. Although surgical methods have progressed, outcomes following curative resection remain suboptimal due to substantial recurrence frequencies. Recurrence occurring early, usually during the initial postoperative year, indicates aggressive tumor characteristics and predicts inferior outcomes compared with delayed recurrence. Comprehending patterns and predictive indicators of early relapse is essential for refining postoperative monitoring approaches and directing adjuvant treatment choices. The present investigation concentrated on determining risk indicators and recurrence characteristics to enhance personalized patient management strategies for ICC.

AIM

To determine the characteristics and predictive indicators of early disease recurrence following curative surgical resection of ICC.

METHODS

We conducted a retrospective evaluation of 386 consecutive individuals who received curative surgical resection for ICC at our institution during the period spanning January 2017 through December 2021. Early disease recurrence was operationally defined as tumor relapse occurring within the first 12 postoperative months. Predictive indicators were examined through univariate and multivariate Cox proportional hazards regression analyses.

RESULTS

Within our cohort of 386 individuals, 237 patients (61.4%) demonstrated disease recurrence throughout the observation period, with 178 cases (75.1%) manifesting early recurrence. The predominant anatomical locations of recurrent disease included hepatic tissue (66.7%), regional lymph nodes (18.1%), and peritoneal surfaces (8.0%). Independent predictive indicators of early recurrence encompassed: Neoplasm diameter exceeding 5 cm [hazard ratio (HR) = 2.14, 95% confidence interval (CI): 1.52-3.01, P < 0.001], presence of multiple tumor foci (HR = 1.89, 95%CI: 1.34-2.67, P < 0.001), metastatic lymph node involvement (HR = 2.43, 95%CI: 1.71-3.45, P < 0.001), microscopic vascular invasion (HR = 1.76, 95%CI: 1.25-2.48, P = 0.001), carbohydrate antigen 19-9 concentrations surpassing 200 U/mL (HR = 1.92, 95%CI: 1.37-2.69, P < 0.001), and incomplete surgical margins (HR = 2.01, 95%CI: 1.38-2.93, P < 0.001). Individuals experiencing early recurrence demonstrated markedly reduced overall survival relative to those with delayed recurrence (median: 18.5 months vs 42.3 months, P < 0.001).

CONCLUSION

Early disease recurrence following ICC resection occurs frequently and correlates with unfavorable clinical outcomes. Several neoplasm-associated and treatment-associated characteristics predict early relapse. These indicators can facilitate identification of patients at elevated risk who might benefit from intensive monitoring protocols or adjuvant therapeutic interventions.

Key Words: Intrahepatic cholangiocarcinoma; Early recurrence; Radical resection; Risk factors; Prognosis; Survival

Core Tip: Intrahepatic cholangiocarcinoma demonstrates substantial susceptibility to early postoperative recurrence, which profoundly compromises clinical prognosis. Within this extensive retrospective investigation involving 386 participants, 46.1% manifested early disease recurrence during the initial 12 postoperative months, exhibiting a median overall survival duration of merely 18.5 months. Six autonomous predictive indicators were determined: Neoplasm diameter exceeding 5 cm, multiple lesions, metastatic lymph node involvement, microscopic vascular invasion, carbohydrate antigen 19-9 exceeding 200 U/mL, and R1 resection status. An innovative risk assessment framework successfully stratified individuals into low-, intermediate-, and high-risk categories, facilitating personalized surveillance protocols and adjuvant treatment strategies to optimize postoperative outcomes in intrahepatic cholangiocarcinoma management.

INTRODUCTION

Intrahepatic cholangiocarcinoma (ICC) constitutes the second-most common primary hepatic neoplasm, representing approximately 10%-15% of primary liver malignancies, with documented escalation in global incidence throughout recent decades[1,2]. Notwithstanding advancements in surgical methodology and perioperative care protocols, ICC prognosis remains discouraging, with 5-year overall survival (OS) proportions varying from 20% to 40% subsequent to curative-intent surgical intervention[3,4]. This unfavorable outcome predominantly derives from substantial recurrence frequencies, with 50%-70% of individuals experiencing disease relapse during the initial 2 postoperative years[5,6].

The chronological pattern of ICC recurrence following resection carries significant prognostic implications. Early disease recurrence, conventionally characterized as relapse occurring during the initial 6-24 postoperative months, possesses distinct prognostic significance compared with delayed recurrence[7,8]. A contemporary multi-institutional investigation revealed that individuals with very early recurrence (within 6 months) exhibited a median post-recurrence survival duration of only 8.4 months, contrasted with 24.3 months for those experiencing late recurrence[9]. Understanding patterns and predictive indicators of early recurrence proves essential for risk classification, patient consultation, and determination of adjuvant therapeutic strategies.

Numerous characteristics, encompassing neoplasm dimensions, quantity of lesions, vascular invasion, lymph node metastases, and surgical margin status, have demonstrated association with recurrence subsequent to ICC resection[10,11]. Furthermore, serum neoplasm biomarkers including carbohydrate antigen 19-9 (CA19-9) and carcinoembryonic antigen (CEA) have exhibited prognostic utility in ICC patients[12,13]. Contemporary investigations have additionally emphasized the significance of inflammatory biomarkers, incorporating the neutrophil-to-lymphocyte ratio and platelet-to-lymphocyte ratio, as predictors of recurrence and survival[14,15].

The function of adjuvant therapy in preventing recurrence following ICC resection has undergone substantial evolution subsequent to the BILCAP trial, which established the survival advantage of adjuvant capecitabine[16,17]. Nevertheless, optimal patient identification for adjuvant therapy remains challenging, particularly considering the heterogeneous characteristics of ICC and its variable recurrence patterns. Determining patients at maximum risk for early recurrence could facilitate treatment decision-making and enhance clinical outcomes. The purposes of this investigation were to: (1) Delineate the patterns and timing of recurrence subsequent to radical resection for ICC; (2) Determine independent predictive indicators for early recurrence; (3) Establish a risk stratification framework to predict early recurrence; and (4) Assess the impact of early recurrence on OS duration.

MATERIALS AND METHODS

Study design and patient selection

This retrospective investigation was implemented at Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology. The institutional Ethics Committee authorized the investigation protocol and exempted the requirement for informed consent considering the retrospective characteristics of the evaluation. All methodologies were executed in accordance with ethical principles of the Declaration of Helsinki.

We identified sequential individuals who received hepatic resection for histologically verified ICC during January 2017 through December 2021. The incorporation criteria comprised: (1) Histopathologically verified ICC based on World Health Organization standards; (2) Curative-intent resection characterized as complete macroscopic elimination of all neoplastic tissue; (3) Absence of extrahepatic disease manifestation at surgical intervention; (4) Availability of complete clinical and pathological information; and (5) Minimum observation period of 12 months or until mortality. The exclusion standards comprised: (1) Combined hepatocellular cholangiocarcinoma; (2) Perioperative mortality occurring during the initial 30 postoperative days; (3) R2 resection demonstrating macroscopic residual disease; (4) Previous hepatic resection for ICC; and (5) Utilization of neoadjuvant therapeutic intervention.

Preoperative evaluation and surgical approach

All individuals received comprehensive preoperative assessment, incorporating complete blood analysis, hepatic function evaluation, coagulation assessment, and tumor biomarkers (CA19-9, CEA, and alpha-fetoprotein concentrations). Cross-sectional imaging utilizing contrast-enhanced computed tomography or magnetic resonance imaging was executed in all cases to evaluate neoplasm extent, vascular participation, and presence of satellite nodules. Positron emission tomography-computed tomography was selectively implemented in individuals with suspected extrahepatic disease. Future liver remnant volume was determined utilizing three-dimensional volumetric evaluation when major hepatectomy was anticipated. Portal vein embolization was executed when the future liver remnant was considered insufficient.

The surgical methodology was determined based on neoplasm location, dimensions, vascular participation, and hepatic function reserve. Anatomic resection was prioritized when achievable, with the objective of securing negative surgical margins while maintaining adequate hepatic parenchyma. Hepatectomy extent was categorized according to the Brisbane 2000 Terminology for Liver Anatomy and Resection[18]. Systematic lymph node dissection of the hepatoduodenal ligament was routinely executed, incorporating stations 8 (common hepatic artery) and 12 (hepatoduodenal ligament) lymph nodes. Additional lymph node stations were dissected based on neoplasm location and intraoperative observations. Intraoperative ultrasonography was routinely utilized to direct resection and identify additional lesions undetected on preoperative imaging.

Pathological evaluation

All resected specimens received processing according to standardized protocols. Neoplasm dimensions were characterized as the maximum diameter of the largest lesion. Multiple neoplasms were characterized as the presence of two or more distinct tumor nodules, excluding satellite lesions. Histological differentiation was categorized as well, moderate, or poor according to World Health Organization standards. Microvascular invasion was characterized as neoplastic cells within vascular spaces lined by endothelium visible exclusively on microscopic examination. Perineural invasion was evaluated as present or absent. Surgical margin status was categorized as R0 (negative margins with > 1 mm clearance), R1 (microscopic positive margins or margins ≤ 1 mm), or R2 (macroscopic residual disease). The American Joint Committee on Cancer (AJCC) 8^th edition staging framework was utilized for pathological staging.

Postoperative management and follow-up

Postoperative complications were categorized according to the Clavien-Dindo classification. Major complications were characterized as grade III or higher. Adjuvant chemotherapy utilizing capecitabine was recommended for all individuals based on BILCAP trial findings, particularly for individuals with high-risk characteristics, incorporating lymph node metastases, R1 resections, or multiple neoplasms. The standard regimen comprised oral capecitabine at 1250 mg/m² twice daily on days 1-14 of a 21-day cycle for 8 cycles.

The observation protocol incorporated clinical evaluation, serum tumor biomarkers, and cross-sectional imaging every 3 months during the initial 2 years and every 6 months subsequently. Recurrence was characterized as radiological or histological evidence of disease following complete resection. The anatomical location of initial recurrence was categorized as intrahepatic exclusively, extrahepatic exclusively, or combination of intrahepatic and extrahepatic. Extrahepatic recurrence sites incorporated lymph nodes, peritoneum, lungs, bone, and other distant organs. Recurrence treatment was personalized based on disease extent, performance status, and hepatic function, incorporating repeat resection, locoregional therapy, systemic chemotherapy, or optimal supportive care.

Statistical analysis

Continuous variables are presented as medians with interquartile ranges (IQRs) or mean ± SD as appropriate. Categorical variables are displayed as n (%). The optimal threshold value for characterizing early recurrence was established utilizing the minimum P value methodology and confirmed utilizing piecewise regression evaluation. Comparisons between groups were executed utilizing the Mann-Whitney U test for continuous variables and the χ² or Fisher’s exact tests for categorical variables.

Recurrence-free survival was determined from the surgical date to the date of initial recurrence or final observation. OS was determined from the surgical date to mortality or final observation. Post-recurrence survival was determined from the recurrence diagnosis date to mortality or final observation. Survival curves were constructed utilizing the Kaplan-Meier methodology and compared utilizing the log-rank test.

Predictive indicators for early recurrence were determined utilizing Cox proportional hazards regression frameworks. Variables demonstrating P < 0.10 in univariate evaluation were incorporated in multivariate evaluation utilizing backward stepwise selection. The proportional hazard assumption was evaluated utilizing Schoenfeld residuals. A risk assessment for early recurrence was established based on regression coefficients from the final multivariate framework, with points allocated proportionally to the hazard ratios (HRs). The discriminatory capability of the risk assessment was evaluated utilizing the concordance index (C-index) and time-dependent receiver operating characteristic curves. Internal validation was executed utilizing 1000 bootstrap samples. All statistical evaluations were executed utilizing R software version 4.2.0 (R Foundation for Statistical Computing, Vienna, Austria). Statistical significance was characterized as a two-sided P value of < 0.05.

RESULTS

Patient characteristics

A total of 386 individuals who received radical resection for ICC were incorporated in the evaluation. The median age was 62 years (IQR: 54-69), and 218 individuals (56.5%) were male. Hepatitis B infection was present in 98 individuals (25.4%), hepatitis C infection in 31 (8.0%), and cirrhosis in 67 (17.4%) individuals. The median neoplasm dimensions were 5.8 cm (IQR: 3.5-8.2), with 142 individuals (36.8%) possessing neoplasms > 7 cm. Multiple neoplasms were present in 89 individuals (23.1%). Lymph node metastases and microvascular invasion were documented in 124 (32.1%) and 168 (43.5%) individuals, respectively. R0 resection was accomplished in 298 individuals (77.2%), while 88 (22.8%) received R1 resection. Major hepatectomy (≥ 3 segments) was executed in 245 individuals (63.5%). Adjuvant chemotherapy was provided to 268 individuals (69.4%; Table 1).

Table 1 Baseline characteristics of patients, median (interquartile range)/n (%).

Characteristic	Total (n = 386)	Early recurrence (n = 178)	No early recurrence (n = 208)	P value
Demographics
Age, years	62 (54-69)	61 (53-68)	63 (55-70)	0.082
Male sex	218 (56.5)	102 (57.3)	116 (55.8)	0.764
BMI, kg/m2 (mean ± SD)	23.8 ± 3.4	23.6 ± 3.5	24.0 ± 3.3	0.248
Comorbidities
Diabetes mellitus	87 (22.5)	42 (23.6)	45 (21.6)	0.645
Hypertension	134 (34.7)	58 (32.6)	76 (36.5)	0.415
Hepatitis B	98 (25.4)	51 (28.7)	47 (22.6)	0.171
Hepatitis C	31 (8.0)	13 (7.3)	18 (8.7)	0.625
Cirrhosis	67 (17.4)	35 (19.7)	32 (15.4)	0.267
Laboratory values
CA19-9 > 200 U/mL	156 (40.4)	98 (55.1)	58 (27.9)	< 0.001
CEA > 5 ng/mL	112 (29.0)	67 (37.6)	45 (21.6)	< 0.001
NLR > 3.0	143 (37.0)	89 (50.0)	54 (26.0)	< 0.001
PLR > 150	128 (33.2)	75 (42.1)	53 (25.5)	< 0.001
Albumin < 3.5 g/dL	76 (19.7)	42 (23.6)	34 (16.3)	0.074
Total bilirubin > 1.2 mg/dL	89 (23.1)	45 (25.3)	44 (21.2)	0.339
Tumor characteristics
Tumor size, cm	5.8 (3.5-8.2)	7.2 (5.0-9.5)	4.5 (3.0-6.8)	< 0.001
Tumor size > 5 cm	213 (55.2)	132 (74.2)	81 (38.9)	< 0.001
Multiple tumors	89 (23.1)	62 (34.8)	27 (13.0)	< 0.001
Bilobar distribution	54 (14.0)	38 (21.3)	16 (7.7)	< 0.001
Pathological features
Poor differentiation	98 (25.4)	67 (37.6)	31 (14.9)	< 0.001
Lymph node metastasis	124 (32.1)	89 (50.0)	35 (16.8)	< 0.001
Microvascular invasion	168 (43.5)	112 (62.9)	56 (26.9)	< 0.001
Perineural invasion	145 (37.6)	91 (51.1)	54 (26.0)	< 0.001
Surgical factors
Major hepatectomy	245 (63.5)	125 (70.2)	120 (57.7)	0.011
R1 resection	88 (22.8)	67 (37.6)	21 (10.1)	< 0.001
Blood loss > 500 mL	167 (43.3)	89 (50.0)	78 (37.5)	0.013
Transfusion required	98 (25.4)	56 (31.5)	42 (20.2)	0.011
Postoperative course
Major complications	67 (17.4)	35 (19.7)	32 (15.4)	0.267
Adjuvant chemotherapy	268 (69.4)	112 (62.9)	156 (75.0)	0.010

BMI: Body mass index; CA19-9: Carbohydrate antigen 19-9; CEA: Carcinoembryonic antigen; NLR: Neutrophil-to-lymphocyte ratio; PLR: Platelet-to-lymphocyte ratio.

Recurrence patterns and timing

Throughout a median observation of 28.5 months (IQR: 18.2-42.3), 237 individuals (61.4%) demonstrated recurrence. The median duration to recurrence was 8.7 months (IQR: 5.2-15.8). Utilizing the minimum P value methodology, 12 months was determined as the optimal threshold for characterizing early recurrence. Among individuals who experienced recurrence, 178 (75.1%) experienced early recurrence during the initial 12 months, while 59 (24.9%) experienced delayed recurrence.

The predominant anatomical location of initial recurrence was hepatic tissue (158 individuals, 66.7%), followed by lymph nodes (43 individuals, 18.1%), peritoneum (19 individuals, 8.0%), lungs (11 individuals, 4.6%), and bones (6 individuals, 2.5%). Among individuals with intrahepatic recurrence, 87 (55.1%) possessed solitary lesions amenable to repeat resection or locoregional therapy, while 71 (44.9%) possessed multifocal disease. Individuals with early recurrence demonstrated increased likelihood of possessing multiple recurrence sites (42.1% vs 23.7%, P = 0.012) and extrahepatic disease (45.5% vs 28.8%, P = 0.025) compared with those experiencing delayed recurrence (Table 2).

Table 2 Patterns of recurrence according to timing, n (%).

Recurrence pattern	Total recurrence (n = 237)	Early recurrence (n = 178)	Late recurrence (n = 59)	P value
Site of first recurrence
Intrahepatic only	134 (56.5)	89 (50.0)	45 (76.3)	< 0.001
Solitary	74 (55.2)	42 (47.2)	32 (71.1)	0.007
Multiple	60 (44.8)	47 (52.8)	13 (28.9)	0.007
Extrahepatic only	79 (33.3)	64 (36.0)	15 (25.4)	0.135
Lymph nodes	43 (54.4)	35 (54.7)	8 (53.3)	0.923
Peritoneum	19 (24.1)	16 (25.0)	3 (20.0)	0.681
Lung	11 (13.9)	8 (12.5)	3 (20.0)	0.449
Bone	6 (7.6)	5 (7.8)	1 (6.7)	0.878
Combined	24 (10.1)	25 (14.0)	0 (0)	0.002
Number of recurrence sites
Single site	162 (68.4)	103 (57.9)	45 (76.3)	0.012
Multiple sites	75 (31.6)	75 (42.1)	14 (23.7)	0.012
Treatment of recurrence
Repeat resection	34 (14.3)	15 (8.4)	19 (32.2)	< 0.001
Locoregional therapy	67 (28.3)	38 (21.3)	29 (49.2)	< 0.001
Systemic chemotherapy	112 (47.3)	98 (55.1)	14 (23.7)	< 0.001
Best supportive care	24 (10.1)	27 (15.2)	0 (0)	0.001

Risk factors for early recurrence

In univariate evaluation, characteristics demonstrating significant association with early recurrence incorporated neoplasm dimensions > 5 cm, multiple neoplasms, bilobar distribution, poor differentiation, lymph node metastases, microvascular invasion, perineural invasion, R1 resection, CA19-9 concentrations > 200 U/mL, CEA concentrations > 5 ng/mL, neutrophil-to-lymphocyte ratio > 3.0, and platelet-to-lymphocyte ratio > 150 (all P < 0.05).

In multivariate Cox regression evaluation, independent predictive indicators of early recurrence comprised neoplasm dimensions > 5 cm (HR = 2.14, 95%CI: 1.52-3.01, P < 0.001), multiple neoplasms (HR = 1.89, 95%CI: 1.34-2.67, P < 0.001), lymph node metastasis (HR = 2.43, 95%CI: 1.71-3.45, P < 0.001), microvascular invasion (HR = 1.76, 95%CI: 1.25-2.48, P = 0.001), CA19-9 concentrations > 200 U/mL (HR = 1.92, 95%CI: 1.37-2.69, P < 0.001), and R1 resection (HR = 2.01, 95%CI: 1.38-2.93, P < 0.001; Table 3).

Table 3 Univariate and multivariate analysis of risk factors for early recurrence.

Variable	Univariate analysis, HR (95%CI)	P value	Multivariate analysis, HR (95%CI)	P value
Age > 65 years	0.89 (0.65-1.22)	0.467
Male sex	1.04 (0.76-1.42)	0.812
Hepatitis B/C	1.23 (0.88-1.72)	0.227
Cirrhosis	1.31 (0.89-1.93)	0.172
Tumor size > 5 cm	2.87 (2.06-3.99)	< 0.001	2.14 (1.52-3.01)	< 0.001
Multiple tumors	2.65 (1.91-3.68)	< 0.001	1.89 (1.34-2.67)	< 0.001
Bilobar distribution	2.34 (1.61-3.40)	< 0.001		NS
Poor differentiation	2.18 (1.57-3.03)	< 0.001		NS
Lymph node metastasis	3.42 (2.48-4.72)	< 0.001	2.43 (1.71-3.45)	< 0.001
Microvascular invasion	2.89 (2.09-4.00)	< 0.001	1.76 (1.25-2.48)	0.001
Perineural invasion	2.13 (1.55-2.93)	< 0.001		NS
R1 resection	3.21 (2.29-4.50)	< 0.001	2.01 (1.38-2.93)	< 0.001
CA19-9 > 200 U/mL	2.76 (2.00-3.81)	< 0.001	1.92 (1.37-2.69)	< 0.001
CEA > 5 ng/mL	1.89 (1.37-2.61)	< 0.001		NS
NLR > 3.0	2.03 (1.48-2.79)	< 0.001		NS
PLR > 150	1.78 (1.29-2.45)	< 0.001		NS
Major hepatectomy	1.42 (1.02-1.97)	0.038		NS
Blood loss > 500 mL	1.51 (1.10-2.07)	0.011		NS
Major complications	1.34 (0.91-1.97)	0.139
No adjuvant chemotherapy	1.67 (1.21-2.31)	0.002		NS

HR: Hazard ratio; CI: Confidence interval; CA19-9: Carbohydrate antigen 19-9; CEA: Carcinoembryonic antigen; NLR: Neutrophil-to-lymphocyte ratio; PLR: Platelet-to-lymphocyte ratio; NS: Not significant in multivariate analysis.

Risk stratification model

Based on multivariate evaluation, a risk scoring framework was established with the following point allocations: Neoplasm dimensions > 5 cm (2 points), multiple neoplasms (2 points), lymph node metastases (3 points), microvascular invasion (2 points), CA19-9 concentrations > 200 U/mL (2 points), and R1 resection (2 points). The total score ranged from 0-13 points. The individuals were stratified into three risk categories: Low risk (0-3 points, n = 98), intermediate risk (4-7 points, n = 176), and high risk (8-13 points, n = 112). The 12-month recurrence proportions in the low-, intermediate-, and high-risk categories were 18.4%, 48.3%, and 82.1%, respectively (P < 0.001). The C-index of the risk assessment was 0.78 (95%CI: 0.74-0.82), indicating satisfactory discriminatory capability. The area under the receiver operating characteristic curve for predicting 12-month recurrence was 0.81 (95%CI: 0.77-0.85; Table 4).

Table 4 Risk stratification model for early recurrence.

Risk group	Score range	n (%)	12-month recurrence rate, %	RFS, median (95%CI), months	P value
Low risk	0-3 points	98 (25.4)	18.4	38.5 (32.1-44.9)	Reference
Intermediate risk	4-7 points	176 (45.6)	48.3	14.2 (11.8-16.6)	< 0.001
High risk	8-13 points	112 (29.0)	82.1	5.8 (4.7-6.9)	< 0.001

RFS: Recurrence-free survival; CI: Confidence interval.

Survival outcomes

The median OS for the complete cohort was 31.2 months (95%CI: 27.8-34.6), with 1-, 3-, and 5-year survival proportions of 82.4%, 42.7%, and 28.3%, respectively. Individuals with early recurrence demonstrated significantly inferior OS compared with those experiencing delayed recurrence (median OS: 18.5 months vs 42.3 months, P < 0.001) or no recurrence (median OS: 18.5 months vs 58.7 months, P < 0.001). The median post-recurrence survival was 9.8 months (95%CI: 8.4-11.2) for individuals with early recurrence vs 18.6 months (95%CI: 14.9-22.3) for those experiencing delayed recurrence (P < 0.001). Among individuals with early recurrence, those who received repeat resection or locoregional therapy possessed superior post-recurrence survival compared with those receiving systemic chemotherapy exclusively or optimal supportive care (median 15.2 months vs 8.3 months vs 3.1 months, P < 0.001; Table 5).

Table 5 Survival outcomes according to recurrence status, %.

Group	n	OS, median (95%CI), months	1-year OS	3-year OS	5-year OS	P value
Overall cohort	386	31.2 (27.8-34.6)	82.4	42.7	28.3
By recurrence status
No recurrence	149	58.7 (52.3-65.1)	96.6	71.8	54.2	Reference
Early recurrence	178	18.5 (16.2-20.8)	71.3	18.0	5.1	< 0.001
Late recurrence	59	42.3 (36.7-47.9)	100	52.5	23.7	< 0.001
Post-recurrence survival
Early recurrence	178	9.8 (8.4-11.2)	38.2	8.4	2.8	Reference
Late recurrence	59	18.6 (14.9-22.3)	71.2	25.4	10.2	< 0.001
By treatment of recurrence
Repeat resection/RFA	34	28.3 (21.7-34.9)	82.4	35.3	17.6	Reference
Locoregional therapy	67	19.2 (15.8-22.6)	71.6	19.4	7.5	0.012
Systemic chemotherapy	112	12.4 (10.1-14.7)	51.8	8.9	0	< 0.001
Best supportive care	24	3.1 (2.2-4.0)	4.2	0	0	< 0.001

RFA: Radiofrequency ablation; CI: Confidence interval; OS: Overall survival.

DISCUSSION

This extensive single-institution investigation of 386 individuals with resected ICC established that early recurrence during the initial 12 months was prevalent, occurring in 46.1% of all individuals and 75.1% of those who demonstrated recurrence. Our observations emphasize the aggressive characteristics of ICC and the essential significance of determining individuals at elevated risk who might benefit from intensive surveillance and adjuvant therapies. The comprehensive risk stratification framework established in this investigation provides a practical instrument for clinical decision-making and patient consultation.

The substantial proportion of early recurrence documented in our cohort aligns with previous reports, although the characterization of early recurrence has varied across investigations. Spolverato et al[19] documented that 59.6% of 588 individuals with ICC demonstrated recurrence, with the majority occurring during the initial 24 months. More recently, Tsilimigras et al[9] discovered that 45% of recurrences occurred during the initial 6 months, which they characterized as “very early recurrence”. Our utilization of 12 months as the threshold for early recurrence was data-driven and corresponded with the clinical observation that the most aggressive neoplasms recur during the initial year subsequent to surgical intervention.

The predominance of intrahepatic recurrence (66.7%) in our series highlights the challenges of accomplishing local control for ICC. This observation resembles the 66% hepatic recurrence proportion documented in a contemporary multi-institutional investigation[20]. Interestingly, individuals with early recurrence demonstrated increased likelihood of possessing multiple recurrence sites and extrahepatic disease, suggesting that early recurrence might reflect systemic dissemination present at surgical intervention but below the detection threshold of current imaging modalities. This observation substantiates the rationale for utilization of systemic adjuvant therapy in individuals at elevated risk.

Multivariate evaluation determined six independent predictive indicators of early recurrence: Neoplasm dimensions > 5 cm, multiple neoplasms, lymph node metastases, microvascular invasion, CA19-9 concentrations > 200 U/mL, and R1 resection. These characteristics reflect both neoplasm biology (dimensions, multiplicity, and vascular invasion) and the adequacy of surgical treatment (margin status). The strong association between lymph node metastases and early recurrence (HR = 2.43) is particularly significant and aligns with multiple previous investigations establishing the prognostic significance of nodal status in ICC[21,22]. A contemporary evaluation of 449 individuals from an international database discovered that lymph node metastasis was associated with a median survival of only 20 months, compared with 40 months for node-negative individuals[23].

The prognostic utility of CA19-9 concentrations in ICC is progressively acknowledged. Our observation that CA19-9 concentrations > 200 U/mL independently predicted early recurrence (HR = 1.92) is substantiated by contemporary investigations. He et al[24] established that preoperative CA19-9 concentrations > 200 U/mL were associated with inferior OS and recurrence-free survival in 341 individuals with ICC. Nevertheless, it should be acknowledged that approximately 10% of individuals do not express the Lewis antigen and consequently cannot produce CA19-9, potentially constraining its universal applicability[25].

The influence of surgical margin status on outcomes remains an area of active investigation. Our investigation discovered that R1 resection was a strong predictive indicator of early recurrence (HR = 2.01), which aligns with the observations of Hewitt et al[26], who documented significantly inferior survival for R1 vs R0 resection. Nevertheless, optimal margin width remains controversial. A Japanese nationwide survey discovered that margins ≥ 5 mm were associated with enhanced survival exclusively in node-negative individuals[27]. More recently, Endo et al[28] suggested that the influence of margin status might depend on overall neoplasm burden, with reduced benefit from wide margins in individuals with elevated neoplasm burden.

The risk-stratification framework established in this investigation established satisfactory discriminatory capability with a C-index of 0.78. This performance was comparable to other prognostic frameworks for ICC, incorporating the recently published post-recurrence survival score by Tsilimigras et al[29] (C-index: 0.69). The capability to stratify individuals into distinct risk categories, with 12-month recurrence proportions ranging from 18.4% to 82.1%, possesses significant clinical implications. Individuals at elevated risk with scores ≥ 8 points might be candidates for more intensive adjuvant therapy or clinical trials of innovative agents.

The function of adjuvant therapy in preventing recurrence has undergone substantial evolution subsequent to the BILCAP trial, which established the survival advantage of adjuvant capecitabine[16]. In our cohort, 69.4% of individuals received adjuvant chemotherapy, primarily capecitabine-based regimens. Although adjuvant therapy was associated with reduced risk of early recurrence in univariate evaluation, this effect was not significant in multivariate evaluation, possibly due to selection bias, as higher-risk individuals were more likely to receive adjuvant therapy. The ongoing ACTICCA-1 trial comparing gemcitabine-cisplatin with capecitabine might provide additional insights into optimal adjuvant regimens[30].

Contemporary advances in molecular profiling have determined targetable alterations in ICC, incorporating FGFR2 fusions and IDH1/2 mutations, which are present in approximately 15%-20% and 10%-20% of cases, respectively[31]. The integration of molecular biomarkers with clinical risk characteristics might further enhance risk stratification and direct the selection of targeted therapies. The ongoing PROOF trial is evaluating adjuvant pemigatinib in individuals with FGFR2-altered ICC, representing a precision-medicine methodology to prevent recurrence.

The unfavorable outcomes associated with early recurrence emphasize the necessity for enhanced surveillance strategies in individuals at elevated risk. Although current guidelines recommend imaging every 3-6 months subsequent to resection, our data suggest that more frequent monitoring throughout the initial year might be warranted for individuals at elevated risk. Furthermore, emerging biomarkers, including circulating tumor DNA, might enable earlier detection of recurrence and direct treatment decisions, although prospective validation is necessary.

This investigation possesses several limitations that should be acknowledged. First, the retrospective design introduces potential selection bias and constrains causal inference; specifically, the decision to administer adjuvant therapy was not randomized, which might confound the observed associations between treatment and outcomes. Second, as a single-center investigation from a high-volume tertiary center, the findings might not be generalizable to community practice settings where surgical volume and expertise differ, potentially affecting recurrence patterns and survival. Third, molecular profiling data were not available for most individuals, preventing evaluation of the prognostic influence of genetic alterations including FGFR2 fusions and IDH1/2 mutations; this constrains our capability to establish a comprehensive precision-medicine methodology and might explain some of the observed variability in recurrence timing among individuals with similar clinical characteristics.

Fourth, the heterogeneity in adjuvant therapy regimens and the absence of standardized criteria for treatment selection constrain conclusions about the effectiveness of specific adjuvant methodologies; this variability makes it challenging to isolate the genuine influence of adjuvant therapy on early recurrence prevention. Fifth, circulating tumor DNA and other emerging biomarkers were not evaluated, which could have enhanced early detection of subclinical recurrence. Finally, although our risk stratification framework demonstrated satisfactory discriminatory performance (C-index: 0.78), external validation in independent cohorts from different geographic regions and practice settings is essential before widespread clinical implementation, as the framework’s performance might vary in populations with different baseline characteristics and treatment methodologies.

CONCLUSION

Early recurrence (during the initial 12 months) subsequent to radical resection for ICC is prevalent and associated with significantly inferior survival outcomes. Multiple neoplasm-related characteristics, incorporating neoplasm dimensions > 5 cm, multiplicity, lymph node metastases, microvascular invasion, elevated CA19-9 concentrations, and incomplete margins, independently predict early recurrence. The risk stratification framework established in this investigation can determine individuals at elevated risk for early recurrence who might benefit from intensified adjuvant therapy and surveillance protocols. Future investigations should concentrate on integrating molecular biomarkers with clinical characteristics to further refine risk stratification and evaluate innovative adjuvant strategies to prevent early recurrence in individuals at elevated risk.

ACKNOWLEDGEMENTS

The authors would like to express their sincere gratitude to all the physicians, sonographers, and nursing staff of the Department of Medical Ultrasound, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, for their invaluable assistance in patient care and data collection. We also appreciate the contributions of the Department of Hepatobiliary Surgery and the Department of Pathology for providing essential clinical and pathological data. The authors thank the institutional biostatistics team for their support with statistical analyses and methodological review. Finally, we acknowledge all the patients and their families who participated in this study for their trust and cooperation, which made this research possible.