Local Therapy Reduces the Risk of Liver Failure and Improves Survival in Patients with Intrahepatic Cholangiocarcinoma: a Comprehensive Analysis of 362 Consecutive Patients

Abstract

Purpose

Treatment methods for intrahepatic cholangiocarcinoma (ICC) have improved, but their impact on outcomes remains unclear. We evaluated the outcomes of patients definitively treated with resection, radiation, and chemotherapy for ICC, stratified by era.

Methods

Clinicopathologic characteristics, cause of death, disease-specific survival (DSS), and intrahepatic progression-free survival (IPFS) were compared among patients who underwent resection, radiation, or chemotherapy as definitive treatment strategies for ICC (without distant organ metastasis) between 1997 and 2015. Variables were also analyzed by era (1997-2006 [early] or 2007-2015 [late]) within each group.

Results

Among 355 patients in our cohort, 122 underwent resection (early, 38; late, 84), 85 underwent radiation (early, 17; late, 68), and 148 underwent systemic chemotherapy alone (early, 51; late, 97) as definitive treatment strategies. In resection group, 3-year DSS rate was 58% for the early era and 67% for the late era (P=0.036), and 1-year IPFS was 50% for the early era and 75% for the late era (P=0.048). In radiation group, 3-year DSS was 12% for the early era and 37% for the late era (P=0.048), and 1-year IPFS was 48% for the early era and 64% for the late era (P=0.030). In chemotherapy group, DSS and IPFS did not differ by era. Patients treated with chemotherapy significantly more frequently developed liver failure at the time of death than patients treated with resection (P<0.001) or radiation (P<0.001).

Multivariable analysis identified local therapy (resection or radiation) as a sole predictor of death without liver failure.

Conclusion

Survival outcomes have improved for local therapy-based definitive treatment strategies for ICC, which may be attributable to maintaining control of intrahepatic disease, thereby reducing the occurrence of death due to liver failure.

Graphical Abstract

Condensed Abstract

This study evaluated the prognosis and causes of death in patients definitively treated with resection, radiation, and chemotherapy for intrahepatic cholangiocarcinoma, stratified by era. Both of disease-specific survival and intrahepatic progression-free survival have improved for local therapy (resection or radiation)-based definitive treatment strategies, which may be attributable to maintaining control of intrahepatic disease, thereby reducing the occurrence of death due to liver failure.

INTRODUCTION

Intrahepatic cholangiocarcinoma (ICC) is the second most common primary liver malignancy, accounting for up to 30% of all primary liver tumors and affecting 5,000-8,000 people per year in the United States.¹ In Western countries, incidence of ICC has steadily increased over the past three decades, and unfortunately, most reports indicate no improvement in overall survival, with less than 5% of patients surviving up to 5 years from diagnosis.^2-5 To date, the only potentially curative treatment is complete resection; however, less than 30% of patients are eligible for surgical resection.⁶

Although survival rates in general have not changed, some investigators have demonstrated improvements in prognosis in the past few decades for patients undergoing resection for ICC, in association with improved patient selection, surgical techniques, and perioperative multimodal treatments.^7-9 Current studies indicate that 5-year overall survival rates after resection of ICC range from 33% to 50%.^{8, 9} Other studies have reported that delivery of increased doses of radiation improved local control and prognosis in patients with unresectable ICC.^10-12 However, the current standard systemic therapy for advanced ICC, combined cisplatin and gemcitabine, has increased median overall survival by less than 4 months compared with gemcitabine alone.¹³

These data suggest that the definitive treatment strategy used for ICC may affect prognosis. The treatment strategy could also be related to the cause of death. The most frequent site of postoperative relapse has been reported to be the liver, either alone or associated with extrahepatic recurrence (79%); isolated extrahepatic recurrence is less common (21%).¹⁴ However, to date there has been no detailed analysis of causes of death following each treatment modality and the effects of each modality on outcome. In the current study, we sought (1) to analyze the outcomes of patients with ICC who underwent resection, radiation, or chemotherapy as the definitive treatment, stratified by era at a single center; and (2) to compare the occurrence of tumor-related liver failure as a cause of death among these treatment groups.

MATERIAL AND METHODS

Patients Selection and Review of Patient Records

All patients with a diagnosis of ICC evaluated at The University of Texas MD Anderson Cancer Center between January 1, 1997, and December 31, 2015 (with follow-up through April 30, 2016) were identified from the prospectively maintained databases of the Department of Surgical Oncology, the Department of Radiation Oncology, and the MD Anderson Tumor Registry. This study was approved by the Institutional Review Board of MD Anderson (PA15-0202). Our analysis included patients with histologically confirmed ICC and no distant organ metastases who underwent resection (with or without perioperative chemotherapy), radiation (with or without chemotherapy), or systemic chemotherapy alone as definitive treatment strategies at MD Anderson. In patients who did not undergo resection, the diagnosis was made on the basis of percutaneous biopsy results combined with cross-sectional imaging appearance, as well as the absence of an extrahepatic primary tumor.

Variables extracted from the database or updated by review of electronic medical records for each patient included sex, age, body mass index, number of tumors, largest diameter of the tumor, lymph node status, presence of distant organ metastasis, clinical stage (based on the seventh edition of the American Joint Committee on Cancer [AJCC] staging manual), chemotherapy regimen, dose and modality of radiation, and cause of death.^{11, 15-17} Liver failure was considered the cause of death when clinical findings such as portal hypertension (esophageal or gastric varices, splenomegaly, or low platelet count), refractory ascites, or hepatic encephalopathy were reported.¹⁸ Causes of death unrelated to liver failure included biliary complication (cholangitis and biloma), vascular complication (portal vein occlusion and hepatic vein or inferior vena cava occlusion), extrahepatic-peritoneal carcinomatosis, extrahepatic-lymphangitic disease in the lung, and others.¹¹ In the patients who underwent resection as the definitive treatment strategy, the following data were also retrieved: reasons for preoperative chemotherapy, perioperative therapy and outcomes, and pathologic findings. Imaging and medical chart assessments were performed by two readers (S.Y., a hepatobiliary surgeon with 8 years of experience; and E.J.K., a radiation oncologist with 6 years of experience), who reached agreement about each assessment.

Treatments

Decisions about the treatment strategy for each patient, including whether to use perioperative chemotherapy, were made at a weekly meeting of physicians with various specialties. Patients were considered candidates for upfront resection when they were medically fit and able to undergo macroscopically curative resection in which more than 20-30% of the total estimated liver volume could be preserved with or without preoperative portal vein embolization, two continuous hepatic segments could be spared, and vascular inflow and outflow and biliary drainage could be maintained with or without reconstruction.¹⁹ For borderline resectable cases—i.e., patients with vascular or biliary invasion—preoperative chemotherapy was administered.^{7, 20} Preoperative chemotherapy primarily consisted of regimens containing a combination of gemcitabine and platinum administered for 4 to 6 months, and postoperative chemotherapy was primarily gemcitabine- or capecitabine-based regimens.^{7, 21, 22} During preoperative chemotherapy, the disease was restaged on the basis of Response Evaluation Criteria in Solid Tumors, and ICC was deemed resectable when a hepatectomy could achieve a negative margin. Concomitant lymphadenectomy was performed selectively in patients with known or suspected nodal involvement.¹⁴

Patients with unresectable disease, identified either at initial presentation or during preoperative chemotherapy and intraoperatively, were treated with radiation (with or without chemotherapy), systemic chemotherapy alone, or best supportive care, depending on performance status. Generally, multidisciplinary liver tumor conference considered locally advanced unresectable ICC or medically inoperable without multifocal bilobar distribution as being suitable to undergo radiation. Radiation consisted of three-dimensional conformal radiation therapy with high-energy photons (6-18 MV), intensity-modulated radiation therapy with high-energy photons (6MV), or passive scatter proton therapy.¹¹ Since 2010, radiation was more frequently delivered using a simultaneous integrated boost (doses of 60-75 Gy in 15 fractions or 75-100 Gy in 25 fractions) with simultaneous integrated protection of the adjacent organs, achieving biologic equivalent doses above 80.5 Gy.²³ Systemic chemotherapy continues to vary; most regimens prior to 2000 contained 5-fluorouracil and leucovorin, and newer regimens often contained gemcitabine and other non-5-fluorouracil-based compounds.^{13, 14, 20, 24} Patients who were medically unfit for or refused the treatments described above received best supportive care.

Statistical analyses

Continuous variables were compared using the Wilcoxon rank-sum test, and categorical variables were compared using the χ2 test. Intrahepatic progression was defined as progression at the primary tumor and/or intrahepatic metastases. Intrahepatic progression-free survival (IPFS) was measured from the date of definitive treatment until the date of radiographic detection of intrahepatic progression or last follow-up. Disease-specific survival (DSS) was measured from the date of the start of definitive treatment to the date of death caused by ICC or to the date of last follow-up. In patients who underwent resection as the definitive treatment strategy, recurrence-free survival was also measured from the date of the start of definitive treatment to the date of radiographic detection of recurrence or to the date of last follow-up. Survival curves were generated using the Kaplan-Meier method, and differences between curves were evaluated using the log-rank test. To evaluate changes over time, we bisected the study period (1997-2006 [early era] and 2007-2015 [late era]) and compared outcomes in each era within each definitive treatment group. This separated patients close to evenly and it also represented a time at which the technology for radiation therapy had improved, incorporating CT-based image guidance to deliver higher doses.²³ Univariable and multivariable analyses were performed using logistic regression to identify predictors of death without liver failure. In the resection group, univariable and multivariable analyses were performed to identify predictors of poor recurrence-free survival and DSS using Cox proportional hazards regression models. Variables with P<0.1 in the univariate analysis were selected for the multivariable analysis. P<0.05 was considered statistically significant in all analyses. Statistical analyses were performed using IBM SPSS software (version 23.0; SPSS Inc., Chicago, IL, USA).

RESULTS

Study Population

A total of 583 patients who underwent primary therapy for ICC at MD Anderson were identified (Figure 1). Among them, 221 patients with distant organ metastases were excluded, and the remaining 362 comprised the current study cohort.

Fig 1.

Flowchart showing treatments administered in the study population.

Potentially Resectable or Borderline Resectable Disease at Diagnosis

Among the 160 patients who were initially considered to have potentially or borderline resectable disease, 38 patients were later shown to have unresectable disease on the basis of intraoperative findings or image findings during preoperative chemotherapy. These patients received systemic chemotherapy (n = 32), radiation (n = 4), or best supportive care (n = 2). The remaining 122 patients with potentially or borderline resectable disease underwent upfront resection (n = 79) or preoperative chemotherapy followed by resection (n = 43). Among the 43 patients who underwent preoperative chemotherapy followed by resection, the reasons for selecting preoperative chemotherapy were locally advanced disease (n = 28), extensive lymph node metastases (n = 8), and transient poor performance status caused by medical comorbidity (n = 7). Preoperative radiation was not performed.

Unresectable Disease at Diagnosis

The 202 patients initially diagnosed with unresectable disease were treated with systemic chemotherapy (n = 116), radiation (n = 81), or best supportive care (n = 5; Figure 1).

Definitive Treatment Groups

Altogether, in our study cohort, 122 patients underwent resection (79 underwent upfront resection and 43 underwent preoperative chemotherapy followed by resection), 85 underwent radiation (21 underwent radiation alone and 64 underwent chemoradiation), and 148 underwent systemic chemotherapy alone as definitive treatment strategies, and 7 received best supportive care (Figure 1). Ablative treatments (radiofrequency, microwave, and cryoablation) were not used as definitive treatment.

Patient Characteristics According to Definitive Treatment Group

Table 1 lists the clinicopathologic characteristics and posttreatment outcomes for each definitive treatment group. The number of patients increased over time in each group. Compared with patients in the resection and chemotherapy groups, patients in the radiation group were significantly older. During the period studied, the proportion of patients in the radiation group significantly increased over time (early era, 17/106 [16%]; late era, 68/249 [27%]; P=0.023). Patients in the chemotherapy group had significantly more advanced disease status in terms of number of tumors, T stage category, regional lymph node metastases, and AJCC stage compared with patients in the other treatment groups. In patients who received systemic chemotherapy alone or resection following preoperative chemotherapy, a gemcitabine-based regimen was most frequently selected. A similar proportion of patients received gemcitabine-based and capecitabine-based regimens in the radiation group (i.e., among those who received chemoradiation). In patients who received radiation, the median biologic equivalent dose was 72 Gy (range 44-180 Gy) and intensity-modulated radiation therapy was most frequently used (Table 1).

Table 1.

Patient Characteristics in Each Definitive Treatment Group (whole cohort: n = 355)

	Definitive treatment strategy, no. (% )			P value*
Characteristic	Resection	Radiation	Chemotherapy	Resection vs radiation	Resection vs chemotherapy	Radiation vs chemotherapy
No. of patients	122	85	148	-	-	-
Median age (range)	62 years (26-83 years)	66 years (30-88 years)	60 years (30-88 years)	0.008 †	0.480†	<0.001 †
Male:female sex ratio	53:69	38:47	73:75	0.857	0.335	0.497
Median body mass index (range)	26 kg/m2 (15-43 kg/m2)	25 kg/m2 (18-49 kg/m2)	26 kg/m2 (13-54 kg/m2)	0.304†	0.315†	0.950†
Era
Early: 1997-2006	38 (31)	17 (20)	51 (34)	0.074	0.565	0.019
Late: 2007-2015	84 (69)	68 (80)	97 (66)
Ratio of solitary liver tumors to multiple liver tumors	97:25	63:22	66:82	0.364	<0.001	<0.001
Median size of largest tumor (range)	7.3 cm (2.6-13 cm)	6.0 cm (2.0-17 cm)	7.0 cm (1.2-24 cm)	0.319†	0.231†	0.156†
Extent of tumor
T1/T2:T3/T4	103:19	38:47	44:104	<0.001	<0.001	0.021
N0:N1	104:18	48:37	52:96	<0.001	<0.001	0.002
American Joint Committee on Cancer stage
I	47 (39)	4 (5)	2 (1)	<0.001	<0.001	0.001
II	44 (36)	15 (18)	23 (16)
III	9 (7)	7 (8)	4 (3)
IVA	22 (18)	59 (69)	119 (80)
Chemotherapy regimen‡
Gemcitabine-based	39 (32)	32 (38)	112 (76)	<0.001	<0.001	<0.001
Fluorouracil-based	0 (0)	2 (2)	11 (7)
Capecitabine–based	2 (2)	30 (35)	15 (10)
Other	2 (2)	0 (0)	10 (7)
Median biologic equivalent dose (range)	-	72 Gy (44-180 Gy)	-	-	-	-
Radiation modality
Intensity-modulated	-	53 (62)	-	-	-	-
Proton	-	19 (22)	-	-	-	-
3D-plan	-	13 (15)	-	-	-	-
Death	58 (48)	53 (62)	121 (82)	0.035	<0.001	0.001
Cause of death known	47 (81)	44 (83)	99 (82)	0.786	0.899	0.849
Related to liver failure§	14 (30)	18 (41)	71 (72)	0.267	<0.001	0.001
Unrelated to liver failure§	33 (70)	26 (59)	28 (28)

^*χ² test, unless indicated otherwise. Italics indicate statistical significance.

^†Wilcoxon rank-sum test.

^‡Preoperative chemotherapy regimen in the resection group, chemotherapy regimen used along with radiation in the radiation group, and initial regimen prescribed in the chemotherapy group.

^§Liver failure as the cause of death was determined by clinical findings such as portal hypertension (esophageal or gastric varices, splenomegaly, or low platelet count), refractory ascites, and hepatic encephalopathy. Causes of death unrelated to liver failure included biliary complication, vascular complication, extrahepatic-peritoneal carcinomatosis, extrahepatic-lymphangitic disease in the lung, and others.

Causes of Death, and Independent Predictors of Death without Liver Failure

A total of 232 patients died during the period studied. The cause of death was known in 47 patients in the resection group, 44 patients in the radiation group, and 99 patients in the chemotherapy group. A total of 42 patients (18% of all patients who died) died from unknown causes (11 in the resection group, 9 in the radiation group, and 22 in the chemotherapy group). Patients in the chemotherapy group more frequently died from liver failure (71/99, 72%) compared with patients in the resection group (14/47, 30%; P<0.001) and the radiation group (18/44, 41%; P<0.001; Table 1 and Figure 2). There was no difference in the rate of liver failure as a cause of death between the resection and radiation groups (P=0.267).

Fig 2. Cause of death.

Presence and absence of liver failure at the time of death in patients with known causes of death who underwent resection, radiation, or chemotherapy as definitive treatment.

Among the patients who died of known causes (n=190), 103 (54%) died of liver failure secondary to local tumor progression. In the multivariable analysis, local therapy (resection or radiation) as the definitive treatment strategy was the sole independent predictor of death without liver failure (odds ratio 4.11, 95% confidence interval 2.12-8.20, P<0.001; Table 2).

Table 2.

Univariable and Multivariable Analysis of Predictors of Death without Liver Failure among Patients with a Known Cause of Death

		Death without liver failure		Odds ratio	95% confidence interval	Multivariable P
Variable	No.	No. (%)	Univariate P
All patients	190	87 (46)
Sex
Male	99	45 (45)	0.923	–	–	–
Female	91	42 (46)
Age
≥60 years	105	50 (48)	0.574	–	–	–
<60 years	85	37 (44)
Body mass index
≥25 kg/m2	104	48 (46)	0.912	–	–	–
<25 kg/m2	86	39 (45)
Era
Early: 1997-2006	73	33 (45)	0.898	–	–	–
Late: 2007-2015	117	54 (46)
No. of tumors
Solitary	102	52 (51)	0.122	–	–	–
Multiple	88	35 (40)
Size of largest tumor
>5 cm	152	71 (47)	0.610	–	–	–
≤5 cm	38	16 (42)
T category*
T1/T2	81	44 (54)	0.042	1.02	0.41-2.59	0.958
T3/T4	109	43 (39)
Lymph node metastasis*
No	89	51 (57)	0.003	1.58	0.68-3.65	0.282
Yes	101	36 (36)
American Joint Committee on Cancer stage*
I/II	55	32 (58)	0.029	1.10	0.34-3.51	0.877
III/IVA	135	55 (41)
Local therapy (resection or radiation)*
Yes	91	59 (65)	<0.001	4.11	2.12-8.20	<0.001
No	99	28 (28)

^*Variables entered into the multinomial logistic regression analyses.

Survival Outcomes

The 1-year IPFS rate after resection was 68%, which was superior to that after radiation (61%, P=0.020) and after chemotherapy (21%, P<0.001; Figure 3A). In the analysis of IPFS stratified by era, there was no significant improvement in IPFS over time in the chemotherapy group (Figure 3B). However, significant improvements in IPFS were observed over time in the resection group (1-year IPFS 50% in the early era, 75% in the late era, P=0.048; Figure 3C) and the radiation group (1-year IPFS 48% in the early era, 64% in the late era, P=0.030; Figure 3D).

Fig 3. Intrahepatic progression-free survival (IPFS) by treatment group and era.

A, IPFS by definitive treatment strategy. B, IPFS in the chemotherapy group by era. C, IPFS in the resection group by era. D, IPFS in the radiation group by era.

The 3-year DSS rate after resection was 64% and the median DSS was 54 months, which was superior to DSS after radiation (3-year DSS 30%, median DSS 23 months, P<0.001; Figure 4A). A significantly worse 3-year DSS rate and median DSS were observed after chemotherapy (3-year DSS rate 14%, median DSS 17 months) compared with those observed after radiation (P=0.004; Figure 4A). In the analysis of DSS stratified by era, there was no significant improvement in DSS over time in the chemotherapy group (Figure 4B). However, significant improvements in DSS over time were observed in the resection group (3-year DSS in the early era: 58%, in the late era: 67%, P=0.036; Figure 4C) and the radiation group (3-year DSS in the early era: 12%, in the late era: 37%, P=0.048; Figure 4D).

Fig 4. Disease-specific survival (DSS) by treatment group and era.

A, DSS by definitive treatment strategy. B, DSS in the chemotherapy group by era. C, DSS in the resection group by era. D, DSS in the radiation group by era.

In a multivariable analysis of recurrence-free survival in the resection group, the sole independent predictor of poor outcome was tumor size ≥5 cm (Supplementary Table 1). In a multivariable analysis of DSS in the resection group, two factors were independent predictors of poor outcome: tumor size ≥5 cm and lymph node metastasis (Supplementary Table 2). There was no significant difference in terms of both recurrence-free survival and DSS between patients with R0 (no macroscopic or microscopic tumor remaining) and R1 (microscopically positive surgical margins).

DISCUSSION

In the current study, we found that local therapies (resection and radiation) were associated with fewer deaths related to liver failure in patients with ICC (30% for resection and 41% for radiation) compared with systemic chemotherapy alone (72%); furthermore, survival rates with local therapies significantly improved over time. Compared with patients who underwent resection and radiation in the early era, patients who received these treatments in the late era had better DSS and IPFS rates. To the best of our knowledge, this is the first report to show an improvement in survival outcomes after local therapy in association with a decrease in deaths related to liver failure. The number of patients who underwent definitive local therapy increased over time, and local therapy was an effective treatment option in a substantial proportion of patients in our study cohort during the late era (61%).

It is reasonable to postulate that the improvement in IPFS and DSS over time in patients who underwent resection could have been the result of multiple factors: better candidate selection based on anatomical evaluation by precise preoperative imaging and biological evaluation during preoperative chemotherapy, advances in surgical techniques, and sophisticated perioperative management strategies.^{7, 19, 25, 26} The most recent case series showed a 5-year DSS rate of 50% after resection of ICC, with the median survival duration of 58 months; this is similar to the outcomes we observed in the late era in the present study.⁹ The IPFS curve trended toward a plateau in the resection group in our study, suggesting that resection provides the potential for long-term disease-free status in a subset of patients with ICC. Although a residual tumor status of microscopic or macroscopic positive has been reported to predict poor survival outcomes in patients who undergo resection of ICC, the impact of a local residual tumor on survival might be diminished with local therapy.²⁷ In fact, the current study, which included patients who received preoperative chemotherapy, demonstrated that a microscopically positive surgical margin did not predict poor prognosis, which is consistent with previous studies.^{28, 29}

The improvements in IPFS and DSS that we observed over time in patients treated with radiation may be due to increased radiation doses with enhanced safety since 2007. In a previous study, we demonstrated that patients with unresectable ICC treated with a biologic equivalent dose >80.5 Gy had significantly better local tumor control and overall survival than patients treated with a biologic equivalent dose ≤80.5 Gy.¹¹ As with surgery, the administration of ablative radiation doses for ICC prolongs life because the patients do not die of liver failure. This critical point emphasizes the importance of local control in this disease and warrants investigation of aggressive liver-directed therapies for all stages of disease.

The current study has a few limitations. First is the limitation associated with any retrospective study. Though we have prospectively collected these data over time, we acknowledge that there may be bias in the selection of treatment for patients and unmeasured imbalances in the patient characteristics, which could influence the interpretation of the results. Comparisons regarding outcomes among patients treated with 3 definitive therapies had to be viewed in light of this possible selection bias. We attempted to address this concern in the study design by excluding patients with extrahepatic metastases. Second, previous studies have illustrated the limitations of death certificates in ascribing the cause of death.¹⁸ To minimize this limitation in our study, we had two authors review detailed electronic records to validate the cause of death. These records were available for most of the deaths that occurred in our cohort (190/232, 82%), which was similar to the records available for previous reports that used this method.³⁰ Finally, there was some heterogeneity in the analyzed subsets because we included patients who underwent combined, sequential treatment strategies. However, such multimodal treatments reflect clinical practice.

In conclusion, the most common cause of death in patients with ICC is liver failure secondary to local tumor progression. The current study emphasizes the need for intrahepatic disease control with innovative and aggressive liver-directed strategies for all stages of disease. Improvement of systemic therapy with a more personalized approach will likely extend the benefit of local therapy.^31-33 We have demonstrated that survival outcomes improved over the 19-year period studied for both resection- and radiation-based definitive treatment strategies for ICC. This improved survival in patients undergoing local therapy at our institution may be attributable to establishing and maintaining control of the intrahepatic disease, thereby reducing the occurrence of death due to liver failure.