Paclitaxel, carboplatin, and gemcitabine in metastatic nasopharyngeal carcinoma

A Phase II trial using a triplet combination

Abstract

BACKGROUND

Patients with nasopharyngeal carcinoma (NPC) are treated primarily with radiotherapy. In the disseminated state, platinum-based, 2-drug combination regimens yielded response rates of 55–75%, achieving a median survival of 10–12 months. With the proven efficacy of second-generation cytotoxics like paclitaxel and gemcitabine in patients with metastatic NPC, the authors hypothesized that a triplet combination incorporating these newer cytotoxics may improve treatment results.

METHODS

Thirty-two patients with metastatic NPC were treated with combination chemotherapy that included paclitaxel 70 mg/m² on Days 1 and 8, carboplatin dosed to area under curve of 5 on Day 1, and gemcitabine 1000 mg/m² on Days 1 and 8 every 21 days for a maximum of 8 cycles.

RESULTS

Two patients achieved a complete response, and 23 patients achieved a partial response, for an overall response rate of 78%. The main toxicities were hematologic, with 41% of patients experiencing Grade 3 or 4 anemia, 41% of patients experiencing Grade 3 or 4 thrombocytopenia, and 78% of patients experiencing Grade 3 or 4 neutropenia. The median time to disease progression was 8.1 months, and the median overall survival was 18.6 months.

CONCLUSIONS

The combination of paclitaxel, carboplatin, and gemcitabine showed promising efficacy against metastatic NPC but at the expense of considerable toxicity. Cancer 2005. © 2004 American Cancer Society.

Nasopharyngeal carcinoma (NPC) is a tumor that is highly sensitive to both radiotherapy and chemotherapy. Unlike other malignancies of the head and neck region, in which surgery remains the primary mode of treatment, radiotherapy is the treatment of choice in patients with early-stage NPC, whereas chemotherapy is important in the metastatic setting as well as in locally advanced disease when used in combination with radiotherapy.1, 2

Like most squamous cell head and neck carcinomas,3–5 the most widely used chemotherapy regimen in the treatment of NPC also has been the combination of 5-fluorouracil and cisplatin.1, 6, 7 The response rate to this regimen among patients with metastatic NPC is ≈ 55–65%, with a median time to disease progression of 6–8 months.8, 9 In the last few years, other agents have demonstrated significant single-agent activity, including carboplatin,10 paclitaxel,11 and gemcitabine,12 with response rates of 44%, 22%, and 48%, respectively. When paclitaxel was given together with carboplatin, the response rate was as high as 75%.13 This combination has gained popularity over time. However, regardless of the treatment regimen used, the duration of response remained short, and the median survival has not exceeded 12 months.11–13 It would seem prudent to capitalize on the chemotherapy-responsive nature of NPC and to continue the search for a more effective regimen that potentially may have an impact on survival.

Based on the good activity of two-drug combinations, we anticipated that a three-drug combination would result in further improvements in response rates and response duration. The favorable toxicity profile and encouraging single-agent activity of gemcitabine12 made it a suitable addition to the combination of paclitaxel and carboplatin. This three-agent combination already has shown feasibility and promising activity in patients with nonsmall cell lung carcinoma14 and ovarian carcinoma.15

We initiated a Phase II trial using the combination of paclitaxel, carboplatin, and gemcitabine as first-line treatment in patients with metastatic NPC. The primary objective of this study was to assess the response rate of the regimen. The secondary objectives were to evaluate treatment toxicities, to assess the quality of life (QoL) of patients, and to determine the time to disease progression when patients were treated with this regimen.

MATERIALS AND METHODS

Eligibility Criteria

Chemotherapy-naïve patients who had a histologically or cytologically proven diagnosis of NPC with bidimensionally measurable disease and distant metastasis were eligible for the study. Prior radiotherapy was permitted, but the sites of measurable disease had to be outside the radiation portal. Patients were required to have adequate bone marrow function (absolute neutrophil count ≥ 1 × 10⁹/L, platelet count > 100 × 10⁹/L, hemoglobin level ≥ 9 g/L) and renal function (serum creatinine ≤ 140 μmol/L). In patients with no liver metastasis, a normal bilirubin level and transaminases ≤ 2 × the normal upper limit (NUL) were required; whereas, for patients with liver metastasis, they could be enrolled if their bilirubin level was ≤ 2 × the NUL and transaminases were ≤ 4 × the NUL. Under such circumstances, the starting dose of paclitaxel was lower (see Treatment Regimen, below). A good performance status of ≤ 1 on the Eastern Cooperative Oncology Group (ECOG) scale was required. Other malignancies, severe comorbid conditions, and peripheral neuropathy ≥ Grade 1 were considered exclusion criteria.

All patients who entered the study were fully informed of the nature and purpose of the study and provided written informed consent. The trial was approved by the Ethics Committee of the National Cancer Center, Singapore.

Pretreatment Evaluation

Pretreatment evaluation consisted of a detailed medical history and physical examination, laboratory investigations (complete blood cell counts, serum electrolytes and creatinine, liver function tests) and electrocardiogram. Computer tomography (CT) scans of the relevant regions (sites of measurable disease) were obtained within 3 weeks prior to the first treatment, and a bone radionuclide scan was obtained if there was clinical suspicion of bone metastasis. QoL information was recorded prior to first treatment using the core QoL questionnaire of the European Organization for Research and Treatment of Cancer (EORTC), QLQ-C30.16

Treatment Regimen

Paclitaxel at a dose of 70 mg/m² was given as a 1-hour intravenous (IV) infusion on Days 1 and 8 of the treatment cycle. For patients who had liver metastasis that resulted in compromised liver function, the dose was reduced by 20%. On Day 1 of each cycle, this was followed by carboplatin dosed to a target area under the time-concentration curve (AUC) of 5, as determined according to the formula reported by Calvert et al.17 Creatinine clearance was estimated for each patient using the serum creatinine level and the Cockroft–Gault formula.18 Gemcitabine at a dose of 1000 mg/m² was then given as a 30-minute infusion on Days 1 and 8 of the treatment cycle. All patients received dexamethasone at a dose of 20 mg IV, cimetidine at a dose of 300 mg IV, and diphenhydramine at a dose of 25 mg IV as premedication 30 minutes before each paclitaxel infusion, and ondansetron was given as antiemetic prophylaxis. The treatment was repeated every 21 days until 2 cycles beyond the maximum response, or up to a maximum of 8 cycles.

Cycles were delayed if a minimum neutrophil count of 1.0 × 10⁹/L and a minimum platelet count of 100 × 10⁹/L was not maintained on Day 1. Treatment on Day 8 was reduced by 20% if the neutrophil count was 0.5–1.0 × 10⁹/L and/or the platelet count was 50–100 × 10⁹/L, and treatment was omitted if the cell counts fell below these levels. This was made more stringent (neutrophil count = 0.8–1.0 × 10⁹/L and/or platelet count = 75–100 × 10⁹/L) after the first 19 patients due to the high incidence of hematologic toxicities and the concerns of physicians involved with the study. In addition, all subsequent treatment doses were reduced by 20% if any of the following occurred: 1) Grade 4 neutropenia that lasted ≥ 7 days or associated with fever, Grade 3 thrombocytopenia associated with bleeding, or any Grade 4 thrombocytopenia.

Response and Toxicity Evaluation

Patients were reviewed every 3 weeks, and toxicity from the previous treatment was assessed with direct questioning and physical examination. All toxicities were documented according to the World Health Organization (WHO) criteria.19

Response to treatment was assessed after every two cycles of treatment according to WHO response criteria.19 A response was confirmed if it persisted beyond 4 weeks. An independent evaluation of radiologic responses also was made by a radiologist who was not involved directly with the study.

Study Design and Statistical Analysis

In total, 32 patients were to be enrolled for the trial, which assumed an overall target response rate (complete responses plus partial responses) of 80% and no further interest in the regimen if the response was as low as 60%. Using the Bayesian dual-threshold design20 with a Stage 1 threshold probability of 0.65, a final threshold probability of 0.7, and a prior expected response rate of 80%, 19 patients were to be recruited in Stage 1. If at least 11 responses were observed (corresponding to a probability < 0.65 that the true response rate would be < 60%), then we would proceed to recruit an additional 13 patients in Stage 2. Efficacy ≥ 80% would be demonstrated if ≥ 28 responses were observed among all 32 patients (corresponding to a probability of ≥ 0.7 that the true response rate would be > 80%).

Sample estimates, 95% confidence intervals (95% CIs), and Bayesian posterior probabilities were to be computed for the response rate. The 95% CIs would be calculated using the formula described by Newcombe and Altman.21 The time to disease progression and overall survival would be estimated using the Kaplan–Meier method. The time to progression was defined from the time of registration to the first documentation of disease progression. Similarly, survival was calculated from the time of registration to the time of death or, in patients who survived or were lost to follow-up, the date of last contact.

QoL was assessed prior to the start of treatment (baseline) and on Day 15 of every cycle. Summary scores for each QoL domain were computed and transformed to take a value between 0 and 100 in accordance with the procedure outlined in the EORTC QLQ-C30 scoring manual.22 The QoL domain involving global QoL (Questions 29 and 30) was the main outcome of interest. Each patient's global QoL score was plotted against the time since baseline. In addition, the mean change in global QoL score relative to the baseline also was plotted against the treatment cycle (i.e., the difference between the score at each cycle and at baseline).

RESULTS

Patient Characteristics

In total, 35 patients were enrolled between July, 2001, and April, 2003. Three patients were ineligible and were excluded from the analysis: One patient had an ECOG performance status of 3, 1 patient had elevated transaminases beyond the range required by the trial, and 1 patient had a baseline radiologic evaluation > 21 days prior to the start of treatment.

Among the remaining 32 patients, 27 patients (84%) had an ECOG performance status of 1. All patients had WHO Type III (undifferentiated) NPC. Twenty-six patients (81%) were male, and the majority of patients (94%) were Chinese. The median age was 52 years, with a range of 22–71 years. The most common sites of metastasis were the bones/bone marrow and distant lymph nodes (69% for each site), followed by the liver (66%) and the lungs (41%). Twenty-six patients (81%) had multiple sites of metastasis (Table 1).

Table 1. Demographic and Clinical Characteristics of Patients

Characteristic	No. of patients (%)
ECOG: Eastern Cooperative Oncology Group; RT: radiotherapy.
Total patients	32 (100)
Median age (yrs) (range)	52 (22–71)
Gender
Male	26 (81)
Female	6 (19)
Race
Chinese	30 (94)
Malay	1 (3)
Indian	1 (3)
ECOG performance status
0	5 (16)
1	27 (84)
Previous RT
Yes	21 (66)
No	11 (34)
Site of metastasis
Bone or bone marrow	22 (69)
Liver	21 (66)
Lung	13 (41)
Distant lymph node	22 (69)
Multiple sites	26 (81)

Treatment Received

The median number of treatment cycles received was six (Table 2). Twenty-four of 32 eligible patients (75%) completed at least 6 cycles of chemotherapy. Two patients developed reactivation of hepatitis B virus (HBV) after one cycle of chemotherapy, and a third patient developed HBV reactivation after the sixth cycle. Those three patients had to be discontinued from the trial due to elevated transaminase levels. Two patients requested to stop treatment after the first cycle. Of the remaining four patients who received less than six cycles of treatment, three patients had intolerable side effects or deteriorating performance status (after receiving three, four, and five cycles of treatment, respectively), and one patient had disease progression at first reevaluation after two cycles of chemotherapy. In total, 179 cycles of chemotherapy were administered. The median relative dose intensities for paclitaxel, carboplatin and gemcitabine were 0.75, 1.00, 0.75, respectively.

Table 2. Summary of Treatment Received

Treatment received	No.
Chemotherapy (no. of cycles)
Total (all patients)	179
Median (per patient)	6
Treatment modification (no. of doses)
Before amendment
Dose reduction	27
Dose omission	33
Dose delay	6
After amendment
Dose reduction	1
Dose omission	41
Dose delay	4
Total
Dose reduction	28
Dose omission	74
Dose delay	10

Response

The response rates of the first 19 patients satisfied the Stage 1 criteria of the Bayesian dual-threshold design, allowing the trial to complete accrual. Of the 32 eligible patients, 2 patients achieved a complete response, and 23 patients had a partial response, for an overall response rate of 78% (95% CI, 61–89%). Disease progressed in one patient and was stable in two patients, and the response was considered nonevaluable in four patients. The final posterior probability that the true response rate exceeded the pretrial desired rate of 80% was 0.37. This was below the target of 0.7 that had been set based on the trial design, although the point estimate of 78% only marginally was below the pretrial desired rate of 80%. Assessment made by the independent reviewer concurred closely with the investigators, with 1 complete response and 25 partial responses, for an overall response rate of 81%.

Survival

The median follow-up was 14.5 months (range, 4.4–26.5 months). At the time of this analysis, disease had progressed in 20 patients (63%). The median time to disease progression was 8.1 months (range, 3.6–9.7 months) (Fig. 1). At the time of this analysis, 15 deaths (47%) were reported. The median follow-up for the patients who remained alive was 15.6 months (range, 6.3–18.4 months). The Kaplan–Meier estimate of median survival was 18.6 months (range, 4.4–26.5 months). The 1-year survival rate was 83.5% (95% CI, 70.1–96.8%) (Fig. 2).

Figure 1. Time to disease progression.

Figure 2. Overall survival.

Toxicity

The main toxicity was bone marrow suppression: Thirteen patients (41%) had Grade 3 or 4 anemia, 13 patients (41%) had Grade 3 or 4 thrombocytopenia, and 25 patients (78%) had Grade 3 or 4 neutropenia (Table 3). The mean nadir neutrophil count decreased with increasing numbers of treatment cycles, suggesting that the bone marrow toxicity may have been cumulative (Fig. 3). Despite the high incidence of neutropenia, only two patients had neutropenic fever. Other Grade 3 and 4 toxicities were not common. Three patients developed reactivation of HBV, requiring interruption of chemotherapy after the first cycle of treatment in two patients and after the sixth cycle in the third patient. All three patients recovered from the episodes. The two patients who developed HBV reactivation after the first cycle were able to receive further chemotherapy (out of trial) after they were started on lamivudine. Three patients had deterioration in performance status and had to drop out of the study after receiving between three and five cycles of chemotherapy. Only 2 patients (6%) had significant alopecia, and 10 patients (31%) had Grade 1 peripheral neuropathy. Mild rash observed in 5 patients (16%) was attributed to the use of gemcitabine.

Table 3. Grade 3 and 4 Toxicity

Toxicity	No. of patients (%)
	Grade 3	Grade 4
a Two patients had neutropenic fever. b Transaminitis was due to re-activation of hepatitis B in all 3 patients.
Anemia	12 (38)	1 (3)
Thrombocytopenia	7 (22)	6 (19)
Neutropenia	13 (41)a	12 (38)a
Transaminitis	2 (6)b	1 (3)b
Alopecia	1 (3)	0 (0)
Syncope	1 (3)	0 (0)
Lethargy	1 (3)	0 (0)
Giddiness	1 (3)	0 (0)

Figure 3. Nadir neutrophil count by cycle.

QoL

Thirty-one patients had adequate QoL assessment data. One patient did not have a baseline assessment and was excluded from analysis. Figure 4 shows the variation in the mean change in global QoL scores from baseline. There was a trend suggesting improvement of QoL in patients who received increasing numbers of treatment cycles.

Figure 4. Mean change in global quality of life (QoL) on Day 15 of each cycle.

DISCUSSION

NPC is a disease that is very sensitive to chemotherapy. Many chemotherapeutic agents have shown significant activity against NPC, including bleomycin, methotrexate, 5-fluorouracil, the platinums, paclitaxel, and (recently) gemcitabine and docetaxel.1, 6–13, 23, 24 Despite the good response rates of various regimens, the survival of patients with metastatic NPC remained poor, and the median survival had not exceeded 12 months in various studies. In the current study of the combined 3-drug regimen with paclitaxel, carboplatin, and gemcitabine, encouraging efficacy was shown, with 78% of patients achieving confirmed responses. More important, the 1-year survival rate was 83.5%, and the median survival was 18.6 months.

The high incidence of Grade 3 and 4 bone marrow toxicity encountered in the current study was not unexpected. When a 2-drug combination of paclitaxel and carboplatin was used, the Grade 3 or 4 neutropenia rate was 31% in the study reported by Tan et al.13 In a series that involved other types of carcinoma, the neutropenia rate ranged from 47% to 59%.25–27 With the addition of gemcitabine, Hainsworth et al. reported 49% Grade 3 or 4 leukopenia among his patients with nonsmall cell lung carcinoma,14 whereas Hussain et al. documented 73% Grade 3/4 neutropenia and 43% Grade 3/4 thrombocytopenia among patients with urothelial carcinoma.28 In our current study, although the Grade 3 or 4 neutropenia and thrombocytopenia rates were 78% and 41%, respectively (very similar to the rates reported by Hussain et al.), there were only 2 patients who required hospitalization for neutropenic fever, and there was no treatment-related mortality. There also were no significant episodes of bleeding, and no platelet transfusions were required. The other toxicity commonly associated with any regimen using paclitaxel was peripheral neuropathy. In the literature, significant (Grade 2 or 3) peripheral neuropathy occurred in 24–57% of patients who were given paclitaxel as a 3-hour infusion every 3 weeks.29–31 By dosing the paclitaxel weekly in our study, this toxicity was negligible, with only 31% of patients experiencing Grade 1 neuropathy and none with toxicity ≥ Grade 2. Three patients (9%) developed HBV reactivation during the course of their treatment, requiring discontinuation from the trial. This complication has been reported with the use of gemcitabine and other chemotherapy regimens32–35 and seems to be a peculiarity with Asian patients. The prevalence of HBV carriers is ≈ 4–10% in this region36 (much higher than in other parts of the world), and HBV flares reportedly have occurred in 10–50% of HBV carriers during aggressive cytotoxic treatment.33 Zhong et al. showed that prechemotherapy viral load was an important risk factor,33 and data also suggested that preemptive treatment with lamivudine may reduce the incidence HBV reactivation during chemotherapy.34

Although the efficacy of the regimen fell short of the target response rate of 80%, nonetheless, it resulted in an improved median overall survival of 18.6 months, which is encouraging compared with our historic data of ≈ 12 months with 2-drug regimens.8, 13 The toxicities were significant but tolerable. A routine pretreatment screening for HBV carrier status is warranted to reduce the risk of hepatotoxicity.

Acknowledgements

The authors thank Eli Lilly for providing the gemcitabine.