Phase II study of capecitabine combined with gemcitabine in the treatment of androgen-independent prostate cancer previously treated with taxanes




The primary objective of the current study was to evaluate the effectiveness of capecitabine and gemcitabine in the treatment of patients with androgen-independent prostate cancer (AIPCa) who experienced disease progression after taxane therapy. The secondary objective was to evaluate the safety and tolerability of the combination of capecitabine and gemcitabine in these patients.


Patients with AIPCa, either metastatic or unresectable disease, and prior taxane therapy were eligible. Patients were treated with 800 mg/m2 of capecitabine orally twice daily (1600 mg/m2 per day) for 14 days, and 800 mg/m2 of gemcitabine intravenously on Days 1 and 8. This regimen was repeated every 21 days. Response to therapy was determined by measuring prostate-specific antigen concentration.


Sixteen patients participated in this study from June 2003 to January 2004. There were no responses as defined by a 50% decline in prostate-specific antigen. The study was terminated early because the response rate was not projected to exceed 30% (rejection error of 10%). Toxicities were notable: 3 patients had Grade 3 thrombocytopenia, 4 patients had Grade 3 neutropenia, and 3 patients had Grade 3 infections (according to the National Cancer Institute Common Toxicity Criteria [version 2.0]). Eight patients (50%) required dose reduction or treatment interruption.


The combination of capecitabine and gemcitabine for the salvage treatment of patients with AIPCa was associated with significant toxicities and was ineffective for induction of disease regression. Cancer 2006. © 2006 American Cancer Society.

In 2004, approximately 230,000 new diagnoses of prostate cancer were projected in the U.S. population, along with 30,000 deaths.1 Significant morbidity was attributed to features of either local progression or bone metastases. Furthermore, the evolution of prostate adenocarcinoma to androgen independence marks the onset of inexorable terminal progression. In published clinical studies, docetaxel and paclitaxel have demonstrated activity as components of frontline chemotherapy for androgen-independent prostate cancer (AIPCa). Treatment responses are typically brief in duration, with a median survival of 18 months reported in randomized trials.2, 3 Effective chemotherapy regimens are urgently needed for salvage treatment of patients whose disease progresses after taxane-based therapy.

Gemcitabine is an antimetabolite that has demonstrated activity in pancreatic, lung, breast, and bladder carcinomas. Gemcitabine also appears to modulate the activity of 5-fluorouracil (5-FU) in renal and gastrointestinal malignancies, presumably by inhibiting ribonucleoside reductase and other enzymes critical to DNA synthesis.4–6 Capecitabine is an oral prodrug of 5-FU and is metabolized to 5-FU in 3 enzymatic steps.5 The final step in the conversion is mediated by thymidine phosphorylase, an enzyme that is abundant in prostate tumor tissue.7 The response rates of AIPCa to single-agent 5-FU, capecitabine, or gemcitabine chemotherapy are only reported to be 7% to 12%,7–12 but to our knowledge, the effect of combining these agents is not known. We evaluated the efficacy, safety, and tolerability of gemcitabine and capecitabine as salvage therapy for AIPCa patients who were previously treated with taxanes in a single-center, Phase II clinical trial.


Patient Selection

Men with pathologically documented metastatic or unresectable AIPCa who were previously treated with taxane-based therapy were eligible for this study. Patients were registered from June 20, 2003 through January 15, 2004. The following inclusion criteria were used: 1) adequate bone marrow function, defined as an absolute neutrophil count (ANC) ≥1500/μL and a platelet count ≥100,000/μL; 2) adequate hepatic function (bilirubin of ≤1.5 mg/dL and alanine aminotransferase ≤2× the upper limit of normal); 3) adequate renal function (serum creatinine clearance >50 mL/min or serum creatinine ≤1.6 mg/dL); 4) no strontium treatment within 12 weeks of study entry; 5) no samarium treatment within 8 weeks of study entry; 6) no antiandrogen treatment within 6 weeks of study entry; 7) no radiotherapy within 14 days of study entry; and 8) no chemotherapy within 21 days of study entry. Patients requiring continuous anticoagulation were required to switch from warfarin to subcutaneous low molecular weight heparin, because clinically significant prolongation of international normalized ratio has been reported with concomitant use of capecitabine and warfarin.13 A Zubrod performance score of ≤2 and a life expectancy of ≥12 weeks were required. Patients who had not undergone surgical castration were required to maintain luteinizing hormone–releasing hormone agonist therapy with leuprolide or goserelin. Patients provided written informed consent for treatment. Patients with small cell tumors of the prostate were excluded. Patients with other malignancies (except localized nonmelanoma skin cancer) within the past 5 years or gastrointestinal malabsorption syndromes were excluded. Patients were also excluded for underlying serious comorbid medical conditions such as uncontrolled diabetes mellitus or unstable cardiac conditions.

Study Design

The study protocol was approved by the institutional review board of the University of Texas M. D. Anderson Cancer and written informed consent was obtained from all participants. Patients were given 800 mg/m2 of capecitabine orally twice daily (1600 mg/m2 per day) for 14 days and 800 mg/m2 of gemcitabine by intravenous (i.v.) infusion over 30 minutes on Days 1 and 8. This treatment regimen was repeated every 21 days. Toxicities were graded according to the National Cancer Institute's Common Toxicity Criteria (version 2.0). Patients who remained free of tumor progression at the completion of a cycle were eligible for additional treatment cycles at the same dose level if no dose-limiting toxicities occurred. Dose reductions in capecitabine (600 or 400 mg/m2 orally twice daily) or gemcitabine (600 or 400 mg/m2 i.v.) were implemented as indicated for respective toxicities.

We performed complete blood counts every week, liver function tests with chemistry panel every third week, and serum prostate-specific antigen (PSA) every ninth week. Interval medical history and physical examination was performed every third week to assess for toxicity, including hand-foot syndrome and diarrhea. Treatment was interrupted for an ANC <1000/μL, a platelet count <100,000/μL, or a nonhematologic toxicity ≥Grade 3. Interruptions in treatment were also permitted for troublesome Grade 2 hand-foot syndrome or diarrhea, at the discretion of the treating physician.

Definitions of Response

The primary endpoint for the study was reduction in serum PSA concentration. A partial response (PR) was defined as a decline of ≥80% in the PSA level from the highest value recorded, sustained for 9 weeks with no increase in symptoms. A minor response (MR) was defined as a ≥50% decrease in the PSA level from the highest recorded value. Patients who completed 9 weeks (3 cycles) without PR, MR, or disease progression were evaluated as having stable disease and were followed for an additional 3 cycles to assess maximum response. Disease progression was defined as a ≥25% increase in the PSA level from the lowest level recorded, or the appearance of new disease sites or symptoms. The overall response rate (MR plus PR) defined by these parameters is consistent with the published PSA Working Group guidelines.14

Statistical Methods

The Simon 2-stage trial design was used.15 The sample size was chosen to minimize the accrual if the combination of capecitabine and gemcitabine was determined to be ineffective. The target response rate (MR plus PR) to the combination was 50%. Treatment would be considered ineffective if the response rate was ≤30%. Using these response rate targets, a total sample size of 46 patients was expected to provide 90% power with a significance level of 10%. In the first stage of the study, a maximum of 22 patients was planned for enrollment. If ≥ 8 of the first 22 patients responded to the therapy (PR plus MR), an additional 24 patients would be enrolled, and if there were ≥18 responses out of the 46 patients, the treatment would be considered effective. A 67% probability of early termination in the event of an unacceptable response rate was calculated.


Patient Characteristics

Sixteen men with a median age of 67 years (range, 39-84 years) were included in the study (Table 1). Four patients (25%) had a Zubrod performance status of 2, and the other 12 men (75%) had a performance status of 1. All patients had progression of disease after a taxane-based treatment for androgen-independent metastatic disease. Seven patients had previously responded to a paclitaxel-based regimen, and 4 had a previous response to docetaxel-based chemotherapy. Seven patients (44%) had undergone ≥3 prior chemotherapy regimens, whereas 5 patients (31%) had 2 prior chemotherapy regimens, and 4 patients (25%) had only 1 prior treatment. All the patients were receiving androgen deprivation treatment. All patients were assessable for toxicity; 15 patients received at least 1 course of therapy on the study and were assessable for response.

Table 1. Patient Characteristics
CharacteristicMedian (Range)No. of Patients (%), n = 16
Age, y67 (39-84) 
Zubrod performance status  
 1 12 (75)
 2 4 (25)
Testosterone (ng/dL)19.5 (<10.0-45.0) 
Hemoglobin (g/dL)12.3 (8.1-13.5) 
Granulocyte count (×103/μL)3.61 (1.64-8.62) 
Platelet count (×103/μL)234 (106-513) 
Alkaline phosphatase (IU/L)147 (57-896) 
Prostate-specific antigen (ng/mL)114 (8.1-915) 
Sites of metastases  
 Lymph nodes 11 (69)
 Bone only 5 (31)
Prior treatment  
 Androgen deprivation 16 (100)
 Prior taxane 16 (100)
  Docetaxel 7 (44)
  Paclitaxel 9 (56)


There were no treatment-related deaths. Three patients had Grade 3 thrombocytopenia, 4 patients had Grade 3 neutropenia, and 3 patients had Grade 3 infections (Table 2). Nonhematologic toxicities included Grade 2 diarrhea in 2 patients and Grade 2 hand-foot syndrome observed in 1 patient. Eight patients (50%) had either an interruption in treatment or required dose reduction during treatment. Six of these patients had dose interruption or a reduction in both capecitabine and gemcitabine, whereas 2 patients had only a capecitabine dose reduction. Four patients were unable to proceed beyond the first course due to Grade 3 infection (2 patients), Grade 3 hematologic toxicity (1 patient), or clinical progression of malignancy (1 patient). Two patients required treatment interruption in the second course due to hematologic toxicity. One patient required dose reduction of capecitabine in the third course of treatment due to troublesome Grade 2 diarrhea, mucositis, and hand-foot syndrome.

Table 2. Treatment Toxicity
Toxicity*No. of Patients by Grade (n = 16)
Grades 1 or 2Grade 3Grade 4
  • *

    All toxicities were graded according to the National Cancer Institute Common Toxicity Criteria (NCI/CTC) (version 2).

  • One case of Grade 3 thrombocytopenia would be classified as Grade 4 under version 3 of the NCI/CTC criteria.

Hand–and–foot syndrome100

Duration of Treatment

The median duration of treatment was 2 courses. Twelve patients remained on therapy for at least 2 courses, and 2 patients completed 4 courses (12 weeks).


None of the 15 assessable patients achieved a PR or MR based on the PSA level. The trial was terminated early according to the criteria established in the statistical design, which required at least 8 of the initial 22 enrolled patients to achieve response in the first stage of the trial. Two patients achieved disease stabilization for 9 weeks. A median overall survival of 11 months was observed from study entry.

There were 3 patients removed during the first course because of toxicity, and the remaining 13 patients all demonstrated evidence of tumor progression upon removal from the study. Eleven patients had a PSA increase of ≥25%. One patient had a minor increase in PSA with worsening bone pain and alkaline phosphatase increase from 580 to 1003 IU/L after 2 courses. Another patient had a minor increase in PSA with palpable progression of supraclavicular lymphadenopathy after 1 course. Alkaline phosphatase measurements were available at baseline and at completion of the study for 14 patients, and showed an increase of ≥25% in 7 patients (50%). We did not routinely obtain imaging studies at the time of progression. One patient did undergo repeat imaging that demonstrated a stable bone scan and slight progression of retroperitoneal adenopathy after 3 courses.


The utility of taxane-based therapy has been well established and is a superior first-line therapy in the management of AIPCa.2, 3, 16–20

To our knowledge, there is currently no standard salvage chemotherapy regimen, however, for patients with AIPCa after taxane-based therapy has failed. We report herein that the combination of capecitabine and gemcitabine did not demonstrate activity in men who previously received taxane-based therapy for AIPCa. We observed no responses in terms of PSA decline in 15 assessable patients after salvage treatment with the capecitabine and gemcitabine regimen, indicating that the actual response rate is likely to be <30%. Other potential salvage regimens to be considered include mitoxantrone and prednisone2, 3; ketoconazole and hydrocortisone21; or cyclophosphamide, vincristine, and dexamethasone,22 all of which have response rates of approximately 30% (or less, depending on the extent of prior therapy).

Antimetabolite drugs are known to have only modest activity in prostate cancer. Gemcitabine and 5-FU each have single-agent response rates of ≤12% (Table 3).8–12 In a recent Phase II study of single-agent capecitabine for treatment of AIPCa, capecitabine at a dose of 1250 mg/m2 given twice daily on Days 1-14 was administered every 21 days.7 PSA responses were observed in 12% of participants in this study. Grade 3 hematologic toxicity was observed in only 4% of patients. However, 4 patients developed Grade 2 hand-foot syndrome and an additional 4 patients (16%) developed Grade 3 hand-foot syndrome at this treatment dose. Most notable, 3 patient deaths (12%) were observed during the study, which were attributable to treatment toxicity. Two deaths were due to sepsis after mucositis.

Table 3. Phase II Studies with 5-Fluorouracil or Gemcitabine as Single-Agent Therapy
StudyRegimenNo. of PatientsResponses (%)
  1. 5-FU: 5-fluorouracils.

Hansen et al., 199185-FU, 250-300 mg/m2/d continuous infusion, toxicity-adjusted duration202 (10)
Kuzel et al., 199395-FU, 1000 mg/m2/d bolus × 5 d, every 4 wks180
Daliani et al., 1995105-FU, 750 mg/m2/d bolus × 5 d, followed by additional bolus (Days 15 and 29), every 6 wks172 (12)
Berlin et al., 1998115-FU, 425 mg/m2/d bolus × 5 d, every 4 wks383 (8)
Morant et al., 200012Gemcitabine, 1000 mg/m2(Days 1, 8, and 15), every 4 wks433 (7)

We hypothesized that the addition of gemcitabine to a fluoropyrimidine could provide an antitumor effect superior to that observed with either drug alone. Experience with gemcitabine in patients with renal cell carcinoma suggested that it enhances the response to infusional 5-FU,4 and the combination of gemcitabine and capecitabine produced PRs in patients with colon cancer and pancreatic cancer who previously were treated with 5-FU.5, 6 The most commonly encountered toxicities with capecitabine and gemcitabine were neutropenia and mucositis, and the doses recommended for Phase II testing were: capecitabine at a dose of 830 mg/m2 given twice daily on Days 1-21 and gemcitabine at a dose of 1000 mg/m2 on Days 1, 8, and 15, repeated every 28 days.5 However, because of the risk of hematologic toxicity, patients with AIPCa and prior exposure to taxanes were not likely to tolerate these doses. In the current study, we chose a starting dose of 800 mg/m2 gemcitabine per week, which was tolerated in a Phase I/II study of gemcitabine and strontium-89 treatment for AIPCa.23 However, even at this dose and on a 21-day cycle, the combination of capecitabine and gemcitabine was difficult to tolerate for patients with AIPCa.

The regimen used in the current study did not yield an antitumor effect and appears to be less active than other options such as cyclophosphamide, vincristine, and dexamethasone or anthracycline-based regimens. In addition, substantial toxicities were observed in our study. Extensive prior therapy was a likely factor in both the poor tolerance and lack of efficacy, because the majority of the patients had ≥2 prior regimens. Better results might be obtained with untreated patients or in other selected subgroups. We conclude that further development of this regimen in the treatment of patients with AIPCa who have undergone prior taxane therapy is not warranted.