Oral capecitabine for the treatment of hepatocellular carcinoma, cholangiocarcinoma, and gallbladder carcinoma

Abstract

BACKGROUND

The goal of the current study was to evaluate the efficacy and toxicity of capecitabine in patients with nonresectable hepatobiliary carcinoma.

METHODS

The authors performed a retrospective analysis of all patients with hepatocellular carcinoma (HCC), cholangiocarcinoma (CCA), or gallbladder carcinoma (GBC) who were ever treated with oral capecitabine. The medical records of 116 patients with hepatobiliary carcinoma who were treated at The University of Texas M. D. Anderson Cancer Center (Houston, TX) between July 1998 and March 1999 were reviewed.

RESULTS

A total of 63 patients were treated with capecitabine (37 with HCC, 18 with CCA, 8 with GBC). Capecitabine 1000 mg/m² was administered twice daily for 14 days. Treatment was repeated every 21 days. Each patient received 1–15 treatment cycles. Nine patients (14%)—11% of patients with HCC, 6% of patients with CCA, and 50% of patients with GBC—had either a complete response (CR) or a partial response. A CR was radiologically confirmed in one patient with HCC and in two patients with GBC. The median survival times were 10.1 months (95% confidence interval [CI], 4.5–15.7 months) for patients with HCC, 8.1 months (95% CI, 7.4–8.9 months) for patients with CCA, and 9.9 months (95% CI, 4.4–15.4 months) for patients with GBC. The most common toxicity was hand-foot syndrome (37%). Grade 3 thrombocytopenia occurred in 8% of patients with HCC. No other significant toxicities were observed. For all patients, response to treatment was positively correlated with survival and decline in tumor markers.

CONCLUSIONS

Capecitabine was found to be safe for patients with hepatobiliary carcinoma, including those with cirrhosis. The antitumor activity of single-agent capecitabine was most pronounced in patients with GBC, was modest in patients with HCC, and was poor in patients with CCA. Cancer 2004. © 2004 American Cancer Society.

Malignant tumors of the hepatobiliary tree are not infrequent worldwide, but they are less common in the United States. The prognosis for patients with these tumors is usually poor. The only curative treatment is surgical resection, which is feasible only for patients with early-stage disease. The role of radiotherapy and chemotherapy is mainly a palliative one.

The incidence of hepatocellular carcinoma (HCC) has been increasing in the United States.1 This increase has been attributed primarily to chronic hepatitis C virus (HCV) infection.2 HCC often develops in patients with cirrhosis,3 and the impact of cirrhosis on the outcome of HCC treatment is very significant. The hepatocellular damage associated with cirrhosis may decrease the ability of the liver to metabolize and excrete chemotherapeutic agents, thereby increasing the toxic side effects of chemotherapeutic drugs, further worsening cirrhosis-associated thrombocytopenia and leukopenia.

Many agents have been studied for their anti-HCC activity. The pyrimidine antimetabolite 5-fluorouracil (5-FU) was the first reported chemotherapeutic agent tested in the treatment of HCC.4, 5 Among the other agents tested is doxorubicin,6, 7 which has a reported single-agent activity of 25% and yields a survival advantage when compared with no treatment. Recombinant interferon-alpha (rIFN-α) also has been added to treatment with systemic doxorubicin, 5-FU, and cisplatin (PIAF), and this regimen has demonstrated encouraging results in the treatment of HCC. However, such treatment is unlikely to be tolerated by patients with advanced cirrhosis.8, 9

Bile duct tumors are rare in the United States. Fewer than 4000 new cases are diagnosed annually.10 However, between 1973 and 1997, the incidence of intrahepatic cholangiocarcinoma (CCA) increased by an estimated 9.11% per year, and the corresponding mortality rate increased by an estimated 9.44% per year. Between 1989 and 1996, the average 1-year and 2-year survival rates were 24.5% and 12.8%, respectively.11 Radiotherapy frequently has been used to treat patients with CCA, with response rates of only 20%. Systemic chemotherapy with 5-FU, carboplatin plus 5-FU, paclitaxel, and gemcitabine also has resulted in response rates of only 20%.12 More than 70% of patients with CCA and approximately 80% of patients with HCC have locally advanced disease and lymph node involvement or distant spread of disease at the time of diagnosis, and these features make surgical resection unfeasible.13–15

Reported 5-year survival rates of patients with gallbladder carcinoma (GBC) have ranged from 5% to 13%, with overall median survival times of < 6 months.16 Only 35% of patients with GBC have resectable disease, and results of treatment with chemotherapy have been unfavorable. Nonetheless, chemotherapy regimens containing intravenous (i.v.) 5-FU have been the mainstay of systemic treatment for patients with advanced hepatobiliary malignancies.8, 17

Capecitabine (5′-deoxy-5-fluoro-N-[(pentyloxy)carbonyl]-cytidine) is an orally administered systemic 5-FU prodrug that is absorbed as an intact molecule via the gastrointestinal tract.18–21 It is metabolized by carboxylesterase and by cytidine deaminase to 5′-deoxy-5-fluorocytidine, which is in turn converted to 5-FU by the tumor-associated angiogenic factor thymidine phosphorylase (dThdPase).20 Because dThdPase is more abundant in tumor cells than in surrounding tissues, treatment with capecitabine has led to 5-FU concentrations in tumor samples that were up to 127 times greater than the corresponding concentrations in plasma samples in mouse models22 and to 5-FU concentrations in human colorectal carcinoma tissue samples that were up to 20 times greater than the corresponding concentrations in plasma samples.19 A recent study indicated that mild-to-moderate liver dysfunction in patients with colorectal carcinoma liver metastases did not significantly affect capecitabine pharmacokinetics.23 Therefore, patients with such liver dysfunction should be monitored closely during treatment with capecitabine, but no dose adjustment solely on the basis of this condition is required.

Capecitabine was approved by the U.S. Food and Drug Administration to treat patients with colorectal carcinoma. The antitumor activity of capecitabine as a single agent for the treatment of patients with colorectal carcinoma24, 25 has been assessed in two large randomized studies. These studies revealed higher tumor response rates and an overall survival rate similar to the rate achieved using an i.v. bolus of 5-FU plus leucovorin. A recent Phase I–II trial of capecitabine combined with oxaliplatin has demonstrated promising results in patients with colorectal carcinoma.25

These observations prompted us to test capecitabine in patients with impaired liver function and hepatobiliary carcinoma who were not eligible for aggressive chemotherapy. These patients included those with advanced cirrhosis and low platelet counts who could not tolerate myelosuppressive therapy, those with uncontrolled ascites who could not tolerate the i.v. hydration required for treatment with platinum analogs, and those whose disease failed to respond to previously administered regimens. We performed a retrospective analysis of the efficacy and toxicity of capecitabine treatment in patients with HCC, CCA, or GBC, and the results of this analysis are reported herein.

MATERIALS AND METHODS

Between July 1998 and March 1999, 116 patients with hepatobiliary carcinomas (63 patients with HCC and 53 patients with carcinomas of the biliary tree) were treated at The University of Texas M. D. Anderson Cancer Center (MDACC; Houston, TX). During that period, no salvage protocols were available for patients whose disease did not respond to prior chemotherapy or for those who were not eligible for more aggressive chemotherapy.

A manual retrospective review of pharmacy records was performed, and 63 patients with HCC, CCA, or GBC who received ‘off-protocol’ treatment with capecitabine were identified. It was further determined that all of these patients were ineligible for higher-priority treatments and thus received off-protocol capecitabine. All 63 patients were evaluated and treated uniformly by a medical oncologist who specialized in hepatobiliary carcinoma management (Y.Z.P.). Patients were informed that treatment with capecitabine constituted ‘off-label’ use of the agent. All patients received information regarding the potential benefits and side effects associated with the agent, and all verbally consented to the treatment. Capecitabine was administered as frontline treatment to patients with advanced-stage disease, severe cirrhosis, or a Zubrod performance status score that did not satisfy the eligibility criteria for higher-priority protocols.

Treatment with capecitabine was only offered to patients with histologically confirmed HCC, CCA, or GBC. Additional criteria were radiologically measurable disease in 2 dimensions, life expectancy ≥ 16 weeks, and a Zubrod performance status score ≤ 2. An absolute peripheral neutrophil count ≥ 1.0 × 10⁹/L and a platelet count ≥ 40 × 10⁹/L were also required. Laboratory studies were performed to ensure that patients met the following additional criteria: serum creatinine levels ≤ 2 mg/dL, total bilirubin levels ≤ 3.5 mg/dL (for patients with HCC who had cirrhosis), and serum albumin levels ≥ 2.8 g/dL. Capecitabine treatment was offered regardless of whether patients had been treated previously with other agents or radiotherapy. Women who were pregnant or breastfeeding were ineligible for the treatment. Before the start of therapy, all patients received detailed information about the drug and its potential side effects and verbally consented to treatment. In view of the importance of information regarding potential treatments for patients with hepatobiliary carcinoma, the institutional review board at MDACC approved the retrospective identification of study patients. The analysis generated the data reported in the current article.

Therapy

Therapy was administered on an outpatient basis at MDACC. Capecitabine (Xeloda; Roche Laboratories, Nutley, NJ) was administered orally twice daily at a dosage of 1000 mg/m² for 14 consecutive days followed by 7 days of rest. This cycle was repeated every 21 days. Dose reduction to Level −1 (1000 mg/m² administered orally twice daily for 7 consecutive days followed by 7 days of rest, repeated every 14 days) or Level −2 (875 mg/m² administered orally twice daily for 7 days followed by 7 days of rest, repeated every 14 days) was allowed before the second treatment cycle. All patients were evaluated at either 9 weeks (if they received a 21-day treatment cycle) or at 8 weeks (if they received a 14-day treatment cycle).

Evaluation Criteria

Responses to treatment were measured using computed tomography before and after treatment and were evaluated according to World Health Organization criteria.26 A complete response (CR) was defined as the complete disappearance of disease as determined by all objective parameters on 2 observations made ≥ 4 weeks apart, with the absence of any new lesions. A partial response (PR) was defined as a reduction of > 50% in the sum of the products of the 2 largest perpendicular diameters of all measurable lesions as determined on 2 observations made ≥ 4 weeks apart, with the absence of any new lesions. A minor response was defined as a reduction of > 25% but < 50% in the sum of the products of the 2 largest perpendicular diameters of all measurable lesions as determined on 2 observations made ≥ 4 weeks apart. Stable disease (SD) was defined as a reduction of < 25 % or an increase of < 25% in the sum of the products of the 2 largest perpendicular diameters of 1 or more measurable lesions. Progressive disease (PD) was characterized by the appearance of new lesions or by an increase of > 25% in the product of the 2 largest perpendicular diameters of 1 or more measurable lesions. Imaging response was analyzed for correlations with serum carcinoembryonic antigen (CEA) and α-fetoprotein (AFP) levels.

Time to progression was defined as the time from the start of treatment to the date that PD was first observed. Survival time was defined as the interval between study entry and death. Patients were reported as being alive or as having achieved a CR at the time of the most recent follow-up visit. Toxic effects were also assessed.

Statistical Analysis

The objective of the current study was to collect all pertinent clinical information regarding patients with hepatobiliary carcinoma who were treated with capecitabine during the period 1998–1999. All available patients who had ever received any dose of capecitabine were included in the analysis.

For data management and statistical analysis, SPSS software (SPSS Inc., Chicago, IL) was used. Baseline laboratory markers were expressed as median values with ranges. Differences among patients with the three different types of hepatobiliary carcinoma were calculated using Kruskal–Wallis one-way analysis of variance (ANOVA). The chi-square test was used to determine the significance of differences in response among patients with HCC, patients with CCA, and patients with GBC. Survival curves were generated using the Kaplan–Meier method,27 and the statistical significance of each difference was determined using the Gehan modification of the Wilcoxon signed-rank test.28 Because we were dealing with three types of malignant disease and with different factors that may have had an impact on overall survival and type-specific survival, we performed multivariate analysis. The Cox proportional hazards model29 and multivariate analyses were used to test potential prognostic factors for survival in all patients. Cox analyses were performed such that only variables with significant survival prognosis on univariate analysis (P < 0.05) were included in the overall Cox survival model. The variables tested included age, gender, race, type of disease, stage of disease, performance status, presence of cirrhosis, and presence of metastases. The assumptions of Cox regression analysis were verified, and all reported P values were two sided. To assess differences in tumor markers (AFP and CEA) among patients, Friedman ANOVA was applied to all repeated measurement parameters, and the Wilcoxon signed-rank test was used to compare baseline values and subsequent measurements during the follow-up period.

RESULTS

Patient Characteristics

Sixty-three patients treated at MDACC during the 9-month study period were included in the current analysis. Histologic studies confirmed that 37 patients had HCC, 18 had CCA, and 8 had GBC. Table 1 summarizes patient characteristics grouped according to disease type. The median age of the entire group was 62 years (range, 28–85 years); age distributions were similar in all 3 groups (P = 0.5). Of the patients with HCC, 73% had cirrhosis; 40.6% had Child–Pugh Class A disease, whereas 32.4% had Child–Pugh Class B disease. Twenty-five of 32 (68%) patients with HCC had disease confined to the liver that was not amenable to local therapy measures. Ten patients (seven with GCC and three with GBC) had jaundice and evidence of bile duct obstruction and required biliary tree decompression. This was accomplished using endoscopic retrograde cholangiopancreatography or percutaneous transhepatic cholangiography, resulting in a subsequent reduction in bilirubin levels to ≤ 3 mg/dL before the start of treatment. Compared with patients with CCA and patients with GBC, patients with HCC had increased AFP levels (P = 0.0001) and a higher incidence of hepatitis B (HBV) infection only, HCV infection only, and infection with both HBV and HCV. Before the start of treatment with capecitabine, 19 patients had received chemotherapy—11 patients with HCC (30%), 5 patients with CCA (28%), and 3 patients with GBC (38%; P = 0.6). Of the 37 patients with HCC, 3 (8%) had previously received chemotherapy with irinotecan, 8 (22%) had received a combination of 5-FU and rIFN-α or PIAF, 22 (59%) were chemotherapy naive, and 4 (11%) had previously undergone surgery alone. For patients with CCA or GBC, previous treatment included PIAF and gemcitabine; surgery had been performed for one patient with CCA and four patients with GBC (Table 1). Seven patients (88%) with GBC had undergone cholecystectomy. None of the patients had been exposed to radiotherapy. There was no statistically significant difference among the three groups in terms of performance status or the presence of extrahepatic metastases (P = 0.4). Surgical resection had been performed for 8 patients (22%) with HCC and 2 patients (11%) with CCA.

Table 1. Characteristics of Patients Treated with Capecitabine

Characteristic	No. of patients (%)
	Hepatocellular carcinoma (n = 37)	Cholangiocarcinoma (n = 18)	Gallbladder carcinoma (n = 8)
SD: standard deviation; 5-FU: 5-fluorouracil; IFN-α: interferon-alpha; HCV: hepatitis C virus; HBV: hepatitis B virus; PIAF: recombinant interferon-α, doxorubicin, 5-FU, and cisplatin.
Gender
Male	23 (62)	11 (61)	3 (37)
Female	14 (38)	7 (39)	5 (63)
Race/ethnicity
White	32 (87)	15 (83)	6 (75)
Black	2 (5)	1 (6)	1 (12)
Hispanic	2 (5)	2 (11)	1 (13)
Asian	1 (3)	0 (0)	0 (0)
Mean age in yrs (± SD)	58.7 (+ 13.2)	59.7 (+ 12.6)	65.8 (+ 9.9)
Cirrhosis
Yes	27 (73)	5 (28)	1 (13)
No	10 (27)	13 (72)	7 (87)
Previous treatment
CPT-11	3 (8)	0 (0)	0 (0)
5-FU and IFN-α	3 (8)	2 (11)	0 (0)
PIAF	5 (14)	3 (17)	1 (12.5)
Gemcitabine	0 (0)	0 (0)	2 (25)
Surgery	4 (11)	1 (6)	4 (50)
None	22 (59)	12 (66)	1 (12.5)
CLIP score
1	8 (22)
2	12 (32)
3	7 (19)
4	10 (27)
> 5	0 (0)
Extrahepatic disease
Yes	12 (32)	7 (39)	4 (50)
No	25 (68)	11 (61)	4 (50)
Zubrod performance status score
0 or 1	26 (70)	14 (78)	7 (88)
2 or 3	11 (30)	4 (22)	1 (12)
Hepatitis virus infection
HCV only	12 (32)	2 (11)	0 (0)
HBV only	7 (19)	3 (17)	0 (0)
HCV and HBV	4 (10)	0 (0)	0 (0)
None	14 (38)	13 (72)	8 (100)

Disease was staged according to the American Joint Committee on Cancer TNM staging criteria. Pretreatment Stage IV disease was present in all 37 patients with HCC (Stage IV A in 28 patients and Stage IVB in 9 patients) and in 17 patients (94%) with CCA (Stage IVA in 13 patients and Stage IVB in 4 patients). In addition, Cancer of the Liver Italian Program (CLIP)32 scores were calculated for all patients with HCC.31, 32 Twenty patients with HCC (54%) were assigned a CLIP score of 1 or 2, whereas 17 (46%) were assigned a CLIP score of 3 or 4 (Table 1). Among patients with GBC, 4 (50%) had TNM Stage IVA disease, and 4 (50%) had Stage II/III disease. The four patients with Stage II/III disease were offered capecitabine treatment for the following reasons: disease stability and poor tolerance of the PIAF regimen (one patient); coronary artery disease, which excluded the use of anthracycline-containing regimens (one patient); infected central venous catheter access (one patient); and reluctance to receive the aggressive PIAF regimen (one patient).

Table 2 summarizes the results of pretreatment laboratory measurements. All laboratory values were similar across disease types. However, pretreatment platelet counts were significantly lower and AFP levels were significantly higher among patients with HCC compared with patients with CCA and patients with GBC (P < 0.001).

Table 2. Laboratory Values before Treatment with Capecitabine

Laboratory value	Median (range)
	Hepatocellular carcinoma (n = 37)	Cholangiocarcinoma (n = 18)	Gallbladder carcinoma (n = 8)
AGC: absolute granulocyte count; U: units; ALT: alanine aminotransferase; IU: international units; LDH: lactate dehydrogenase; CEA: carcinoembryonic antigen; AFP: α-fetoprotein; CA125: cancer antigen 125.
Platelet count (10^3 cells/μL)	133 (43–870)	214 (80–387)	259.5 (130–422)
AGC (10^9 cells/L)	3.4 (1.4–15.6)	4.3 (1.5–21.1)	5.2 (2.2–16.9)
Alkaline phosphatase (U/L)	149 (81–491)	180 (69–804)	95.5 (0.20–482)
Albumin (g/dL)	3.5 (2.8–4.7)	3.9 (3–4.30)	2.7 (2.9–4.4)
Bilirubin (mg/dL)	0.9 (0.3–3.20)	0.8 (0.50–2.6)	0.9 (0.20–1.20)
ALT (IU/L)	62 (26–237)	47 (20–855)	35.5 (21–212)
LDH (IU/L)	639 (49–1643)	670.5 (367–1462)	573 (48–7762)
CEA (ng/mL)	3 (0.75–13.10)	4 (0.75–316)	2 (0.75–62)
AFP (ng/mL)	438.1 (1–49,468.80)	5.6 (1.9–155)	3.3 (1.3–10.6)
CA125 (U/mL)	8.1 (7–119)	26.4 (1–171)	29.1 (10–56)

Response

All 63 patients were alive at the end of treatment and were assessable for response. The overall response rate was 14% (9 of 63 patients). Tumors became resectable after capecitabine treatment in 2 (5%) of 37 patients with HCC and in 1 (6%) of 18 patients with CCA (Table 3). The high response rate observed among patients with GBC (50%) was not attributable to disease stage. In fact, one patient with Stage III disease had PD, and one patient with Stage II disease had SD. In contrast, two patients with Stage IVA disease at baseline evaluation went on to experience a PR.

Table 3. Response to Treatment with Capecitabine According to Type of Malignancy

Outcome	No. of patients (%)
	Hepatocellular carcinoma (n = 37)	Cholangiocarcinoma (n = 18)	Gallbladder carcinoma (n = 8)
OLT: orthotopic liver transplantation; CR: complete response; PR: partial response; MR: minor response; SD: stable disease; PD: progressive disease. a 95% Confidence interval, 4–23%. b 95% Confidence interval, 0–15%. c 95% Confidence interval, 21–79%.
CR	1 (3)	0 (0)	2 (25)
PR	3 (8)	1 (6)	2 (25)
CR or PR	4 (11a)	1 (6b)	4 (50c)
MR	0 (0)	3 (17)	1 (13)
SD	4 (11)	2 (11)	1 (13)
PD	23 (62)	10 (56)	2 (25)
Resection or OLT	2 (5)	1 (6)	0 (0)
Median no. of treatment cycles (range)	3 (1–15)	3 (1–9)	5 (3–8)

Analysis of response to treatment according to previous exposure to chemotherapy indicated that two patients with HCC had received PIAF and that one patient with GBC had received 5-FU and rIFN-α, whereas the responder in the CCA group had not previously been exposed to any chemotherapy. The median time to progression was 2.9 months (range, 1.2–23.9 months) for HCC, 3.1 months (range, 0.7–56.80 months) for CCA, and 6.5 months (range, 2.8–16.2 months) for GBC.

A significant decrease in follow-up AFP serum levels compared with the baseline values was observed among patients with HCC who achieved CR or PR (P = 0.04). The estimated median decrease was 86% (range, 86–99%). In contrast, there was no change in AFP values among the 23 patients with HCC who had PD (median, 396.7 ng/mL; range, 1–49,648.8 mL) or among the 4 patients with HCC who had SD (median, 2309.7 ng/mL; range, 9.1–38,714.7). In the GBC group, among patients with CR or PR who had elevated baseline values of CEA (n = 2), significant decreases in serum CEA levels (61% and 73%, respectively) were detected. In addition, the patient with CCA who achieved a PR experienced a decrease in baseline serum CEA levels, from 387.9 ng/mL to 2.30 ng/mL. The remaining patients with CCA or GBC who had PD exhibited elevated or stable CEA levels throughout follow-up.

Survival

The overall median survival time for the entire group was 9.9 months (95% confidence interval [CI], 5.8–14.0 months). Stratification of patients according to type of disease revealed no statistically significant difference in survival times (P = 0.6) (Fig. 1). The median survival duration was 10.1 months (95% CI, 4.5–15.7 months) for patients with HCC, 8.1 months (95% CI, 7.4–8.9 months) for patients with CCA, and 9.9 months (95% CI, 4.4–15.4 months) for patients with GBC. Multivariate analysis demonstrated that radiologic response to treatment was significantly associated with longer survival duration. A hazard ratio of 3.7 (95% CI, 1.4–9.6) was calculated using the Cox proportional hazards regression model, indicating that patients with PD had a fourfold increased risk of death due to disease compared with patients who had a PR or CR. Overall survival was not significantly affected by disease stage, type of malignancy, presence of extrahepatic disease, cirrhosis, gender, age, or performance status.

Figure 1. Median survival according to hepatobiliary carcinoma diagnosis (Kaplan–Meier method). CCA: cholangiocarcinoma; GBC: gallbladder carcinoma; HCC: hepatocellular carcinoma.

Toxicity Effects

All patients were evaluated for toxic effects (Table 4). Grade 1 or 2 hand-foot syndrome was the most common side effect (37%), followed by Grade 1 or 2 nausea and emesis (19%) and fatigue (14%). Grade 3 thrombocytopenia occurred in 8% of patients with HCC. Other toxic effects were mild. All toxic effects were reversible. Dose adjustments to Level −1 were required for patients with HCC (n = 5), patients with CCA (n = 2), and patients with GBC (n = 2). All toxicities were related to treatment. Only one patient with HCC required dose reduction to Level −2.

Table 4. Toxic Effects of Capecitabine Treatment

Toxic effect	No. of patients (%)
	Hepatocellular carcinoma (n = 37)	Cholangiocarcinoma (n = 18)	Gallbladder carcinoma (n = 8)
AGC: absolute granulocyte count; PLT: platelet count; Hgb: hemoglobin level.
Hand-foot syndrome	14 (38)	5 (28)	4 (50)
Nausea and gemesis	3 (8)	5 (28)	4 (50)
Fatigue	4 (11)	2 (11)	3 (38)
Mucositis	5 (13)	0 (0)	1 (13)
Diarrhea	7 (19)	2 (11)	2 (25)
Skin rash	3 (8)	3 (0)	0 (0)
Neutropenia
Grade 1 or 2 (AGC 1.1–1.9 × 10^3 μ/L)	9 (24)	3 (16)	9 (0)
Grade 3 or 4 (AGC < 1.0 × 10^3 μ/L)	0 (0)	0 (0)	0 (0)
Thrombocytopenia
Grade 1 or 2 (PLT, 50–75 × 10^3 μ/L)	5 (14)	3 (16)	1 (13)
Grade 3 or 4 (PLT < 50 × 10^3 μ/L)	3 (8)	0 (0)	0 (0)
Anemia
Grade 1 or 2 (Hgb, 8–9 g/dL)	3 (8)	3 (16)	1 (13)
Grade 3 or 4 (Hgb < 8 g/dL)	0 (0)	0 (0)	0 (0)

DISCUSSION

More than 70% of patients with CCA and approximately 80% of patients with HCC have locally advanced disease, lymph node involvement, or distant spread of disease at the time of diagnosis, making resection unfeasible.13–15 Because biliary tract malignancies are relatively rare in the United States, few clinical studies describing radiotherapy and chemotherapy (5-FU, carboplatin plus 5-FU, paclitaxel, and gemcitabine) for such malignancies have been reported. Response rates of approximately 20% have been reported in previous studies.12 The reported 5-year survival rate for patients with GBC ranges from 5% to 13%, with an overall median survival of < 6 months.16 Data from our laboratory on the combination of continuous-infusion 5-FU and subcutaneous rIFN-α33 and on the PIAF regimen17 have also indicated response rates of < 20%. These poor results continue to drive the quest for new therapies.

To our knowledge, the current report is the first to address the use of capecitabine to treat patients with hepatobiliary carcinomas. Unfortunately, the study was a retrospective analysis of off-protocol treatment, and its validity therefore is somewhat compromised. Another limitation of the study was the small cohort size, particularly for patients with GBC and CCA. Nonetheless, the current study provided important information regarding the potential use of capecitabine, which can be tested in subsequent larger-scale trials, to treat hepatobiliary carcinomas. The current retrospective analysis of capecitabine in this setting demonstrated a response rate of 11% among patients with HCC, 6% among patients with CCA, and 50% among patients with GBC. Many of these patients would have been unable to tolerate more toxic chemotherapeutic regimens. Capecitabine was well tolerated by patients with all three types of disease, including patients with HCC who had thrombocytopenia due to cirrhosis and patients with GBC or CCA who previously required bile duct decompression. Despite the finding that capecitabine was well tolerated by patients with HCC who had cirrhosis, it should be kept in mind that the degree of cirrhosis was limited to Child–Pugh Class A and Child–Pugh Class B; none of the treated patients had Class C cirrhosis. Thus, the use of capecitabine to treat patients with HCC and Child–Pugh Class C cirrhosis cannot be presumed to be safe on the basis of the current study.

Palmar-plantar erythrodysesthesia was the most common side effect of treatment, while hematologic toxicity was limited to Grades 1 and 2. There was no significant difference in grade of toxicity between patients who had previously received treatment and patients who had not. The mildness of the observed toxicity may be attributable to the less aggressive starting dose (1000 mg/m²) used for this group of patients. The use of 1000 mg/m² capecitabine has become common in the treatment of patients with other malignancies, such as colorectal carcinoma; this reduced dose was suggested by Borner et al.,25 who also recommended the use of 1000 mg/m² twice daily when capecitabine is used in combination with oxaliplatin.

In human tumor xenograft models, oral administration of capecitabine yielded substantially higher concentrations of 5-FU in tumor specimens than in specimens of plasma or normal tissue.21, 22 It is noteworthy that levels of 5-FU after administration of capecitabine were much higher than those achieved by i.v. administration of 5-FU at doses producing equal levels of toxicity. The susceptibility of the xenografts to capecitabine were correlated with levels of dThdPase in tumor tissue specimens. Therefore, the efficacy of capecitabine may be optimized by selecting candidates for treatment on the basis of dThdPase expression or by combining this agent with other agents (e.g., taxanes,34 cyclophosphamide,22 irinotecan,35 oxaliplatin,25 and rIFN-α36) that can up-regulate dThdPase expression within tumor tissue.

Although impaired hepatic function can exacerbate toxicity or inhibit the efficacy of many agents, the presence of mild-to-moderate hepatic dysfunction had no clinically significant effect on the pharmacokinetics of capecitabine and its metabolites.23 This finding suggests that capecitabine may be useful for patients with hepatobiliary carcinoma, including patients with mildly-to-moderately impaired hepatic function.

We recently conducted a Phase II trial in which a combination of continuous i.v. 5-FU and rIFN-α was used to treat patients with HCC.36 The overall response rate of 14% was similar to the rate achieved in the current study. However, the median survival duration observed in that study (15.5 months; 95% CI, 8.5–22.5 months) was greater than the median survival duration noted in the current study. The use of single-agent thalidomide in a Phase II trial involving patients with HCC was initially promising in terms of survival results.37 However, a more recent analysis found a median survival time of 6.8 months, (95% CI, 5.4–8.2 months [unpublished data]) in that trial.

Although comparisons of groups of patients in sequential nonrandomized trials are inconclusive, the difference in survival duration between patients treated with rIFN-α plus i.v. 5-FU and patients in the other two treatment groups suggests that a combination of an antimetabolite and IFN-α may result in improved survival. The combination of capecitabine and rIFN-α is particularly attractive in view of the potential effects of rIFN-α on dThdPase gene expression and the associated potentiation of capecitabine activity as a result of effects on fluoropyrimidine metabolism and interactions with tumor angiogenesis.38–40

The addition of other agents with the capacity to up-regulate dThdPase expression (and, thus, the antitumor activity of capecitabine) may be of interest and should be investigated in patients with hepatobiliary carcinomas. We recommend that the combination of capecitabine with low doses of interferon or pegylated interferon be explored in a Phase II trial involving patients with HCC. If the results of that study are promising, a Phase III randomized trial should be performed.