Phase II Trial of Cisplatin, Interferon α-2b, Doxorubicin, and 5-Fluorouracil for Biliary Tract Cancer¹

The annual incidence of all of the hepatobiliary cancers has been steadily increasing in the United States from 15,000 in 1993 (1) to 22,200 in 2000; (2) and that of HCC³ has increased over the past 2 decades by 79% when comparing the periods 1976–1980 and 1991–1995 (3). Data from the Veterans Administration hospitals throughout the Unites States (4) and from M.D. Anderson Cancer Center⁴ suggested that hepatitis C virus is a major factor contributing to the increase in the incidence of HCC in the United States. However, there are no adequate United States epidemiological studies that assessed the change in the incidence of biliary tree cancers (gallbladder cancers and cholangiocarcinoma). It is not impossible that hepatitis C virus-related cirrhosis is a risk factor for American patients with cholangiocarcinoma, because it has been reported for Japanese patients (5).

Both diagnosis and treatment of biliary tree cancers pose several challenges. These tumors may be difficult to visualize by CT or MRI, because the tumor may be present inside the ductal system and all that can be seen on CT or MRI are dilated bile ducts. Indeed, many patients with biliary tree cancer present with obstructive jaundice (6, 7) which must be treated, often by ERCP or PTC, to drain and decompress the biliary tree before the patient can begin chemotherapy. Modern technology such as magnetic resonance cholangiopancreatography may provide better visualization of the biliary tree but cannot offer bile duct decompression.

It is well accepted that surgical resection is the treatment of choice for patients with biliary tree cancers. However, a high percentage of these tumors are unresectable and are usually associated with a very poor prognosis because of the lack of active chemotherapy regimens. A review article of the systemic chemotherapy for bile duct cancers reported that mitomycin C, doxorubicin, and 5-FU are the most active agents against cholangiocarcinoma (8). The overall collective PR of 97 reviewed patients was 29%, and the median survival ranged between 6 and 11 months. However, no CR was reported.

Our previous large-scale study of i.v. 5-FU and s.c. IFNα-2b produced a PR rate of 34% and no CR in 32 evaluable patients (6). Additionally, the combination of 5-FU, carboplatin, and leucovorin was tested in 14 patients with adenocarcinoma of the biliary tree showing an overall response rate of 21.4% and a median survival of 5 months (9). Ducreux et al. (10) reported recently that the combination of 5-FU and cisplatin produced an overall response rate of 24% in 25 previously untreated patients with inoperable, locally advanced, or metastatic biliary tree carcinoma. Similar results were observed in unresectable or metastatic gallbladder carcinoma. A study of 53 patients with gallbladder carcinoma treated by 5-FU or 5-FU with other agents reported similar antitumor response rate of 12% in each treatment arm (11). Moreover, systemic therapy with fluoropyrimidines combined with doxorubicin generated an objective response rate of 30–40% (12, 13).

Indeed, the achievement of CR in patients with unresectable biliary tree cancers is very rare. The inconsistent response rates reported in different studies may have resulted from the addition of other agents to 5-FU and study sample sizes.

Facing tumors so difficult to diagnose and to treat, and based on these observations, we embarked on this Phase II trial of cisplatin, rIFNα-2b, doxorubicin, Adriamycin, and 5-FU (PIAF) in patients with biliary tree cancers.

Between September 1995 and July 1998, 41 patients with unresectable, histologically confirmed, and radiologically measurable adenocarcinoma of the intrahepatic or extrahepatic biliary tree or the gallbladder were registered on a protocol approved by The University of Texas M. D. Anderson Cancer Center Institutional Review Board. For the purpose of this analysis, patients with intra- or extrahepatic biliary tree cancer were referred to as having cholangiocarcinoma and the others as having gallbladder carcinoma. The study sample included 22 cholangiocarcinoma patients (2 extra- and 20 intrahepatic) and 19 gallbladder carcinoma patients. Disease was staged according to the American Joint Committee Cancer Tumor-Node-Metastasis staging criteria (14) and is summarized in Table 2. Thirty two patients (12 gallbladder carcinoma and 20 cholangiocarcinoma) had stage IV A, and 8 patients (6 gallbladder carcinoma and 2 cholangiocarcinoma) had stage IV B disease. None of the patients seemed a surgical candidate because of multifocal lesions and multilobe disease. Even the patient with stage III gallbladder carcinoma was only a marginal candidate and, therefore, was treated with primary PIAF chemotherapy. However, the patient developed metastasis to segment IV of the liver by the second month of treatment, and the consideration of resection was abandoned. All of the patients provided written informed consent to participate. Eligibility criteria included histological confirmation of the disease and the presence of lesions that could be measured by CT or MRI. Other eligibility criteria were life expectancy of ≥16 weeks, performance status of ≤2 on the Zubrod scale, absolute granulocyte count ≥1500/μl, platelet count ≥100,000/μl, serum creatinine level ≤1.5 mg/dl, serum bilirubin level ≤3 mg/dl, serum albumin level ≥2.8 g/liter, and normal coagulation profile. Patients with jaundice and evidence of bile duct obstruction of which the biliary tree could be decompressed by means of ERCP or PTC with a subsequent reduction in bilirubin level to ≤3 mg/dl, were eligible for the study. Previous exposure to any of the four trial drugs was not allowed but prior exposure to nontrial drugs or to radiation therapy was allowed.

All of the i.v. drugs were administered through a central venous catheter inserted into a cephalic or subclavian vein. Patients were first given 2 liters of normal saline, i.v., followed by either 1 mg of granisetron or 20 mg of ondansetron and then 10 mg of dexamethasone, all given i.v. Next, rIFNα-2b (Intron A; Schering-Plough Inc., Kerrilworth, NJ) was administered s.c. at a dose of 5 million units/m². This same dose was repeated every 24 h for a total of four rIFNα-2b doses per 4-day treatment cycle. Then, cisplatin (80 mg/m²) was administered i.v. in 200 ml of normal saline over 2 h followed by doxorubicin (40 mg/m²) administered in 200 ml of normal saline i.v. over 2 h, followed by 5-FU (500 mg/m²) by continuous infusion for 3 consecutive days (total 5-FU dose, 1500 mg/m² over 3 days). The treatment plan and dose adjustments for toxic reactions are shown in Table 1. Dose adjustment to −1 and −2 dose levels were mandated if patients developed grade 3 or grade 4 regimen related toxicity. From day 6 to day 13, patients were given prophylactic s.c. doses of granulocyte colony-stimulating factor (Filgrastim) at a dose of 300 μg/day for patients with body surface area ≤2 m² or 480 μg/day for patients with body surface >2 m². The treatment regimen dose, dose adjustment, and prophylactic use of Filgrastim were implemented after our previous experience in using this regimen in patients with HCC (15, 16). Hydration and antiemetic medications were continued if needed for 2–3 days after the administration of cisplatin. The 4-day treatment cycles were repeated every 28 days or as soon as they could be tolerated thereafter.

In general, patients were treated as outpatients and admitted to the hospital only for management of toxicities and treatment-related complications.

Treatment response was assessed at the end of 8 weeks after 2 cycles of chemotherapy had been completed. Liver tumors were evaluated by CT or MRI, and the sum of the product of the largest two perpendicular diameters was used for baseline and follow-up tumor measurements. All of the responses were reviewed and confirmed by one of the authors (K. A. W.). Dimensions of lung metastases were assessed by chest radiography. CR was defined as disappearance of all of the lesions measurable previously and no new tumor lesions appearing for at least 8 weeks. PR was defined as a decrease of >50% in the product of two perpendicular diameters of each measurable lesion (17). Minor response was defined as a 25–49% decrease in such measurements. Imaging response was correlated with serum CEA levels. Time to disease progression and survival were determined from the first day of treatment.

Simon’s two-stage design (18) was used to calculate the sample size for this Phase II trial, using two levels of response rate, P⁰ (20%) and P¹ (40%). Accordingly, 37 patients were required for this study; 17 patients were accrued for the first stage followed by 20 more patients when four or more responses were observed during the first stage.

The χ² test (19) was used to determine the significance of differences in responses between patients with cholangiocarcinoma and patients with gallbladder carcinoma. Survival curves were generated by the Kaplan-Meier method (20), and the statistical significance of differences was determined according to Gehan’s modification of the Wilcoxon signed-rank test (21). A P of <0.05 was considered statistically significant.

The characteristics of the 41 patients with biliary tract cancer registered on the protocol (19 men and 22 women) are listed in Table 2. The patient median age was 56 years (range, 36–80 years); 22 had cholangiocarcinoma and 19 had gallbladder carcinoma. Thirteen patients (32%) required biliary-tract decompression by ERCP or PTC because of obstructive jaundice; 34 (83%) had disease outside the biliary tree or the gallbladder fossa. Metastatic liver involvement was more common in patients with gallbladder carcinoma (n = 17) than in those with cholangiocarcinoma (n = 8; P = 0.03). Of the 41 patients enrolled, 3 patients could not be evaluated for response, 1 withdrew consent to participate, and 2 were lost to follow-up after the first cycle. None of the patients had been exposed to radiotherapy or chemotherapy before enrollment in the study. However, 9 patients with gallbladder carcinoma (47%) underwent a cholecystectomy before PIAF treatment. The overall response rate was 21% (one-sided 95% CI, 11–32%). Of the 21 evaluable patients with cholangiocarcinoma 2 had a PR and a response rate of 9%; and of the 17 patients with gallbladder carcinoma, 1 had a CR and 5 had a PR, for a response rate of 35% (one-sided 95% CI, 16–54%). In addition, 1 patient with cholangiocarcinoma and 1 patient with gallbladder carcinoma had minor responses (Table 3). Disease did not become resectable in any of these patients despite radiological responses. Before- and after-treatment CT scans of 2 patients with responding disease are shown in Figs. 1 and 2. Radiological results were correlated with changes in CEA levels. A significant decrease in CEA serum level as compared with the baseline values was observed among all of the patients with CR and PR, with a median decrease of 83% (range of decrease, 60–99%). Conversely, 10 of 18 patients with disease progression showed an increase in their follow-up serum CEA levels. The estimated median increase was 228% compared with baseline values (range of increase, 6.6–1366%). The remaining 8 patients with progressive disease showed normal levels of CEA at baseline, and that value remained stable throughout the study.

The median survival time for all of the 41 patients was 14 months (95% CI, 9.5–18.5 months; Fig. 3). The median survival time of the 22 patients with cholangiocarcinoma was 18.1 months (95% CI, 11.6–24.6 months) and that of the 19 patients with gallbladder carcinoma was 11.5 months (95% CI, 5.4–17.6 months), a statistically insignificant difference. The median survival times for responding patients were also not different between groups (14 months for the gallbladder carcinoma group and 15.9 months for the cholangiocarcinoma group). However, median survival time for patients with progressive gallbladder cancer (7.5 months; 95% CI, 4.9–10 months) was shorter than that for patients with progressive cholangiocarcinoma (18.1 months; 95% CI, 7.0–29.2 months; P = 0.02). Post-PIAF salvage treatments were given to 10 patients with gallbladder carcinoma (53%) and to 13 patients with cholangiocarcinoma (59%). However, such salvage therapy had no effect on gallbladder carcinoma patient (P = 0.3) or cholangiocarcinoma patient (P = 0.07) survival. Surgical intervention did not became a treatment option in any of the study patients.

Treatment-related toxic effects are summarized in Table 4. The most common grade III-IV effects were neutropenia (17 patients, 41%), and nausea and vomiting (14 patients, 34%); grade III-IV thrombocytopenia was present in 8 patients (20%) and grade III-IV anemia in 6 (15%). Lesser degrees of fatigue, nausea and vomiting, mucositis, diarrhea, and anemia were also common (Table 4). All of the reported incidents of vomiting were attributable to treatment related toxicity. Because of grade III and IV hematological toxicities, drug dosages were lowered in 23 patients (56%).

The results of this study indicate that the PIAF regimen appeared to be more toxic but no more effective than 5-FU plus rIFNα-2b for the treatment of biliary tract cancer (5). Moreover, the response rate of patients with cholangiocarcinoma given PIAF (9%), was less than that of patients with cholangiocarcinoma who were given 5-FU plus rIFNα-2b (37.5%) in the previous study (6). The response rate for patients with gallbladder cancer was 35%, with 1 CR. The overall median survival time after PIAF (14 months) was no different from that after 5-FU and rIFNα-2b (12 months; Ref. 6).

In the current study, response rates to PIAF were not different statistically for patients with cholangiocarcinoma versus those with gallbladder carcinoma, perhaps because of the relatively small number of patients tested. Despite a seemingly higher response rate, survival time for patients with gallbladder carcinoma was shorter than that for patients with cholangiocarcinoma, particularly among those with progressive disease; however, survival times of patients with responding disease were similar between the two groups. These findings may suggest that the natural history of gallbladder carcinoma, at least in this study population, may be more aggressive than cholangiocarcinoma. It also could be related to the characteristics of the patients in our series, because patients with gallbladder carcinoma had more extensive disease and a higher rate of liver involvement than did patients with cholangiocarcinoma.

The rationale for selecting this four drug combination (PIAF) was based on the activity reported previously of 5-FU and rIFNα-2b (6) and the activity of doxorubicin (12, 13) and cisplatin (7) in biliary tree cancers. To improve tolerance to cisplatin, dexamethasone was added once as antiemetic, with the assumption that a single administration of steroids was unlikely to abrogate the effects of all four daily doses of rIFNα-2b.

One possible explanation for the inferior activity of PIAF relative to that of 5-FU and rIFNα-2b for cholangiocarcinoma (6) may be the smaller doses and shorter exposure times to 5-FU in this study (500 mg/m² for 3 days every 4 weeks) relative to our earlier one (750 mg/m² for 5 days every 14 days). For gallbladder carcinoma patients, the shorter survival time despite the seemingly higher antitumor activity of PIAF may reflect large tumor burdens in these patients. Therefore, an argument could be made for conducting separate Phase II trials for gallbladder carcinoma and for cholangiocarcinoma.

In summary, the aggressive, toxic, and expensive PIAF regimen produced some dramatic antitumor responses but is probably not indicated for the treatment of cholangiocarcinoma. Continuous exposure to a fluoropyrimidine-based regimen or an oral fluoropyrimidine analogue may well be less toxic and more effective for patients with cholangiocarcinoma. Given the apparently high response rate, which included one CR in gallbladder carcinoma, PIAF chemotherapy could be used as a salvage treatment for gallbladder cancer. Nevertheless, newer agents and combinations should continue to be explored in Phase II trials.