Introduction

Background

Gastric cancer is the second leading cause of cancer death and the fourth most common type of cancer. Globally, 989,600 new cases and 738,000 deaths per year can be encountered [1].

Only 3.1 % of patients with advanced gastric cancer survives up to 5 years, and the role of surgery is limited to 23 % of patients, which often depends on the delayed diagnosis because of nonspecific symptoms [2].

The goal of chemotherapy in metastatic gastric cancer (MGC) is to improve quality of life.

In the 1990s, three randomized trials [3–5] evaluated the benefit of chemotherapy versus supportive care alone, showing clearly its beneficial effect. A significant improvement in overall survival (OS) in favor of chemotherapy was also confirmed in a subsequent metanalysis of phase II–III clinical trials [6].

Very few studies have compared single-agent chemotherapy (mainly represented by 5-FU) with combination regimens, showing better response rates but minimal clinically significant survival benefits for polychemotherapy [7–9].

Fluoropyrimidines plus platinum derivatives [10], epirubicin [11], and, more recently, taxanes [12, 13] and irinotecan [14] represent the most active drugs in the treatment of advanced disease.

Despite the availability of numerous effective drugs and different combinations, it is still not possible to define a recommended chemotherapy regimen in patients with gastric cancer that is HER2 negative or with unknown receptor status. From the data available so far, the use, in patients with a good performance status, of a three-drug combination chemotherapy seems feasible. Randomized phase II [15] and phase III studies [16], on triplets as ECF (epirubicin, cisplatin, and 5-FU) and TCF (docetaxel), have shown an increase in response rate from 35 % to 40 % with a median survival estimated between 8 and 11 months. However, data on toxicity despite the clinical benefit [17] and an improved quality of life [18] highlight the need to move toward new schedules of administration.

We previously tested at our center, in a pilot study conducted in a different and separate group of patients (enrolled from November 2004 until June 2007), the feasibility and effectiveness of a TCF regimen modified according to a dose-dense schedule [19].

Some randomized clinical trials have already shown the potential benefits of dose-dense chemotherapy in neoplasms such as breast [20], non-Hodgkin lymphoma [21], and bladder cancer [22]. The availability of granulocyte growth factors has permitted reducing the incidence of neutropenia, allowing us to take advantage of the Norton–Simon hypothesis [23], according to which efficacy of chemotherapy increases by reducing the interval between treatment cycles.

The first studies on dose-density in gastric cancer date back to the late 1980s when regimens containing metrothexate, 5-FU, and doxorubicin (FAMTX) and then 5-FU, cisplatin, and epirubicin (PELF) were tested [24].

TCF q3w certainly represents one of the most effective regimes and is a reference three-drug regimen worldwide used in MGC, which is generally a chemosensitive disease.

To further improve its performance in terms of response rate maintaining an acceptable toxicity profile, we designed an innovative polychemotherapy scheme. Here we report the final efficacy and toxicity results of a phase II study with this modified intensified dose-dense TCF regimen.

Methods

Trial design

This was a nonrandomized, open-label, single-center phase II study of dose-dense chemotherapy with modified TCF regimen in locally advanced or MGC.

Participants

Main entry criteria of the study included histologically or cytologically confirmed gastric cancer, locally advanced nonresectable primary tumor, presence of measurable or evaluable tumor lesions, age ≥18 years, and adequate hepatic, renal, bone marrow, and cardiac function. Prior adjuvant chemotherapy and radiotherapy were allowed provided that these interventions had been completed at least 6 months before enrollment in the study. Major exclusion criteria were an ECOG PS ≥2, prior palliative chemotherapy, pregnancy, breast-feeding, child-bearing potentiality without use of any contraception, any other current or prior malignancy (with the exception of excised cervical carcinoma in situ or squamous cell skin carcinoma), and psychiatric disorders potentially affecting compliance to the therapeutic program. All patients provided written informed consent. The trial protocol was approved by the local ethics committee.

Interventions

Upon study entry, a complete medical history was taken, and all the patients underwent a physical examination, evaluation of ECOG PS, blood chemistry tests, computed tomography scan of the abdomen, of the chest, and of all measurable and assessable sites. Bone scan, magnetic resonance imaging scan, and ultrasound endoscopy were carried out only if clinically indicated. Patients subsequently underwent a physical examination and laboratory tests (blood cell count, serum creatinine, bilirubin, AST, ALT) before each cycle of treatment. Tumor evaluations were carried out every 2 months until disease progression or withdrawal from study medication, on the basis of the response evaluation criteria in solid tumors (RECIST) criteria version 1.0. In addition, survival was monitored every 2 months in each patient leaving the study. Adverse events were classified according to National Cancer Institute (NCI) common toxicity criteria (CTC), version 3.0.

The TCF-dd regimen consisted of docetaxel (Taxotere; Sanofi-Aventis, Paris, France), 70 mg/m2 over a 1-h intravenous (i.v.) infusion on day 1; cisplatin, 60 mg/m2 on day 1 (1- to 3-h i.v. infusion); l-folinic acid, 100 mg/m2 administered in 5 % glucose over 2 h i.v. on days 1 and 2; followed by 5-FU, 400 mg/m2 bolus i.v. on days 1 and 2, and then 5-FU, 600 mg/m2 as a continuous i.v. infusion over 22 h on days 1 and 2. Pegfilgrastim (Neulasta; Amgen, Thousand Oaks, CA, USA), 6 mg, was administered subcutaneously on day 3. Patients aged ≥65 years received the same schedule with a dose reduction of all agents by 30 %.

Anti-emesis treatment (5-hydroxytryptamine receptor antagonists and dexamethasone), appropriate hydration, and corticosteroid premedications were always administered before chemotherapy infusion. Treatment was repeated every 14 days and was continued for up to six cycles (one cycle = 14 days) in the absence of disease progression, unacceptable toxicity, patient’s refusal, or physician’s decision. Treatment was delayed in case of insufficient hematological function (neutrophil count <1,500/mm3 and/or platelet count <100,000/mm3), and/or nonhematological toxicity grade >1 on day 15 of any cycle. If toxicity lasted longer than 2 weeks, the treatment was continued, after recovery, with a dose reduction by 20 %, but always maintaining the 2-week schedule. In the event of febrile neutropenia, grade 4 nonfebrile neutropenia lasting longer than 5 days, or grade 4 or grade 3 with bleeding thrombocytopenia, the dose of each drug was reduced by 25 %. The same dose reduction was indicated for grade 3 and 4 nonhematological toxicity.

Objectives of the study

The primary objective was the activity evaluated as ORR (complete + partial responses). Secondary endpoints were PFS, toxicity, and OS.

Sample size and statistical methods

To measure the real advantage of the biweekly scheme, we assumed a 20 % improvement in response rate compared to reported data in the literature (classical TCF: ORR 37 %). According to Simon’s two-stage design, setting a 5 % alpha significance level and an 80 % power of the study, the estimated total number of patients to be enrolled was 46. The first stage involved the recruitment of 15 patients. If the number of observed responses would have been <5, the recruitment would be stopped and the experimental treatment considered equal to the reference treatment in terms of response. If the number of responses observed would have been >5, then 31 additional patients would have been enrolled. The experimental treatment had to be considered effective if the total number of responses observed were >18.

The efficacy analyses were based on the intent-to-treat population. Descriptive statistics were reported as proportions and medians. Kaplan–Meier estimates were used in the analysis of time-to-event variables, and the 95 % confidence interval (CI) for the median time to event was computed.

Patients who have not received at least one dose of the drug were excluded from the analysis of safety.

Results

Patients’ characteristics

Forty-six patients were enrolled from December 2007 to November 2010. Patient characteristics are reported in Table 1. Median age was 66 years (range, 38–76); 78 % of patients were male and 22 % female. Twenty-two patients had an ECOG PS of 0 (48 %), 21 had a PS of 1 (46 %), and 3 had a PS of 2 (6 %). Most of the patients had histologically confirmed adenocarcinoma (98 %); the most frequent histological grade was G3 (53 %). Only 2 patients had a gastroesophageal junction cancer. Four patients (9 %) had received prior adjuvant chemotherapy. The most common disease sites were lymph nodes (78 %), liver (47 %), stomach (45 %), peritoneum (17 %), lung (6 %), and bone (2 %).

Table 1 Patient characteristics
Full size table

Efficacy and safety

A median of four cycles (range, one to six) per patient was administered. Twenty-eight patients (61 %) received four cycles of treatment and not six as planned because of toxicity. Moreover, 5 patients received fewer than four cycles (3, early disease progression; 1, early death from toxicity; 1, treated with radiotherapy). Twelve patients received six well-tolerated cycles of therapy and all showed a disease response. Twenty-two patients (46 %) were treated at full doses without any delay, thus respecting the dose-dense schedule. The main cause of noncompliance with the dose-dense schedule was hematological toxicity.

Forty-three (93 %) patients were evaluable for response. Three complete responses (CR) (7 %), 25 partial responses (PR) (54 %), 10 stable disease (SD) (21 %), and 5 progression of disease (PD) (11 %) were observed, for an ORR by intention to treat (ITT) analysis of 61 % (95 % CI, 47–75) (Table 2).

Table 2 Efficacy: intention-to-treat analysis
Full size table

As far as age is concerned (≥65 years), 24 patients were evaluable for response. In this population we registered 2 CR (8 %), 14 PR (58 %), 5 SD (21 %), and 3 PD (13 %); similarly, in younger patients, 1 CR (5 %), 11 PR (58 %), 5 SD (26 %), and 2 PD (11 %) were observed.

After a median follow-up of 23 months (95 % CI, 13–35), median OS was 17.63 months (95 % CI, 13.67–20.67) and median PFS was 8.9 months (95 % CI, 6.5–13.4) (Figs. 1, 2).

Fig. 1
figure 1

Kaplan–Meier estimates for overall survival

Full size image
Fig. 2
figure 2

Kaplan–Meier estimates for progression-free survival

Full size image

Thirty-eight patients (83 %) received a second-line therapy after disease progression (37 were treated with chemotherapy and 1 with radiotherapy). Most of the patients received a scheme containing three drugs (oxaliplatin, irinotecan, and 5-FU), in the context of a subsequent clinical trial.

It is noteworthy that two patients initially treated with TCF-dd underwent gastrectomy.

Toxicities observed during treatment are listed in Table 3. All patients were evaluated for toxicity. The most frequent grade 3–4 toxicities were neutropenia (20 %), leukopenia (4 %), thrombocytopenia (2 %), anemia (2 %), febrile neutropenia (6 %), asthenia (22 %), diarrhea (4 %), nausea/vomiting (11 %), and hypokalemia (6 %).

Table 3 Toxicity according to NCI CTC version 3.0 criteria
Full size table

In patients aged ≥65 years, the most frequent grade 3–4 toxicities were neutropenia (8 %), asthenia (16 %), diarrhea (4 %), nausea/vomiting (8 %), and hypokalemia (4 %); in younger patients, neutropenia (33 %), asthenia (29 %), diarrhea (5 %), nausea/vomiting (14 %), and hypokalemia (9 %) were seen. No febrile neutropenia events were encountered in older subjects.

Discussion

The response rate we obtained with TCF-dd (61 %) is much higher than that reported in previous studies with this combination [15, 16].

To increase the percentage of responses it was not necessary to increase drug dosage. In our protocol the following doses were used: docetaxel 70 mg/m2, cisplatin 60 mg/m2, bolus 5-FU 400 mg/m2, and 5-FU by continuous infusion for 44 h at 600 mg/m2. Nearly half the patients (46 %) complied with the dose-dense criterion without receiving any dose reduction. In the remaining 54 % of patients, it was necessary to reduce the dosage or delay the cycle of chemotherapy because of ≥G2 hematological or gastrointestinal toxicity.

In the study by Van Cutsem and colleagues [16], docetaxel and cisplatin were administered at higher doses, both 75 mg/m2 on day 1 and 5-FU at a dose of 750 mg/m2 from day 1 to day 5, every 3 weeks. In 64 % of cases, cycles were postponed, and in 41 % of cases, dose was reduced.

In the Roth et al. study [15], the initial docetaxel dose was 85 mg/m2, but after the first 29 patients enrolled, a protocol amendment was made to reduce the dose to 75 mg/m2 because of the high incidence of febrile neutropenia.

In 2006 the randomized comparison between DCF (docetaxel, cisplatin, and 5-FU) and CF (cisplatin and 5-FU) showed an increase in time to progression (TTP) (5.6 vs. 3.7 months) and OS (9.2 vs. 8.6 months) [16].

In 2007, three different regimens were compared: ECF (epirubicin, cisplatin, and 5-FU) versus TC (docetaxel and cisplatin) versus TCF (docetaxel, cisplatin, and 5-FU), showing a median survival of 8.3, 11.0, and 10.4 months, respectively [15].

In this study, median OS was 17.63 months and TTP 10.67 months. Although these were not the primary endpoints and their value is only descriptive, they deserve to be highlighted.

The median survival in the present study was much longer than in our previous one [19], probably because we administered to all patients lower doses of docetaxel (70 vs. 85 mg/m2) and cisplatin (60 vs. 75 mg/m2) from the beginning; this modification resulted in a higher response rate (61 vs. 56 %) combined with lesser hematological toxicity (neutropenia, 20 vs. 53 %; febrile neutropenia, 6 vs. 22 %).

One could correctly argue that this is a phase II study, but efficacy results are among the highest ever reported with this combination.

In all studies containing the triplet docetaxel, cisplatin, and fluorouracil, major treatment-related concerns were represented by hematological and nonhematological toxicity. Taking again into consideration all the limits of this trial (phase II, selected population, lack of control group, etc.), if we make an indirect comparison with the V325 trial [16], we can easily notice that TCF-dd was likely associated with a better tolerability. One of the reasons for these results probably relies on the prophylactic use of pegfilgrastim, which resulted in a better compliance to treatment and allowed us to limit the febrile neutropenia cases to 6 % and G3–G4 neutropenia to 20 %, much less than reported in the literature (G3–G4 neutropenia >80 % and febrile neutropenia in 29 % of the patients).

Additionally, G3–G4 gastrointestinal toxicities (diarrhea and nausea/vomiting) were also much lower and well controlled compared with classical TCF (4 vs. 19 % and 11 vs. 14 %, respectively).

Among nonhematological side effects, asthenia was particularly frequent and severe in 22 % of patients. We recorded one toxic death from septic shock. Overall, TCF-dd was shown to be safe.

To reduce the significant toxicity related to TCF while maintaining its efficacy, several schedules with different modes of administration have been explored.

In a phase II study, including 60 patients, the administration of docetaxel and cisplatin on days 1, 15, and 29 every 8 weeks with weekly fluorouracil produced an overall response rate of 47 % with a TTP of 8.1 months and an OS of 15.1 months. Rate of febrile neutropenia was 5 % [25].

More recently, an Italian study [26] tested the administration of cisplatin and fluorouracil every 2 weeks for four cycles, sequentially followed by docetaxel every 3 weeks in case of response or stable disease. The aim of the study, which enrolled 34 patients, was to use the three most effective drugs for gastric cancer, but with a modified schedule, to reduce toxicity. Response rate was 38.2 % with PFS and OS of 4.8 and 10.6 months, respectively. The rate of febrile neutropenia reported was 11.8 %.

Despite the intensity of our therapeutic scheme, because of the association of three potential toxic drugs and the close interval between cycles, safety apparently was not compromised. This design translated into a relatively high compliance to treatment and into an impressive response rate that has never previously been registered in a phase II study with the same combination.

Possible explanations of these results probably rely on the appropriate selection of the population in study. In effect, the median age was not particularly high (66 years), and 94 % of patients had a PS = 0–1.

These findings confirm that in specific settings such as good clinical baseline conditions, young age, and low tumoral burden, an intensified chemotherapy regimen such as that used in this protocol may find its proper placing.

Nevertheless, more and more efforts are needed to identify new reliable molecular biomarkers of response to chemotherapy and OS.