Introduction

Head and neck cancers (HNC) are among the most common types of cancer and represent a major health problem. Annually, there are approximately 540,000 new cases and 271,000 deaths worldwide with a mortality of approximately 50 % [1]. Cisplatin plus 5-fluorouracil (5-FU) has been used as induction chemotherapy [2, 3] for patients with locally advanced squamous cell carcinoma of head and neck (PF protocol). The addition of taxane to induction chemotherapy, in the form of a taxane, cisplatin, and 5-FU triplet, has improved the efficacy of chemotherapy during the induction phase [4–7]. Two phase III trials have confirmed the superiority of this regimen with docetaxel (DCF protocol) over PF, followed by radiotherapy or chemoradiotherapy, in terms of progression-free and overall survival [8, 9]. Hence, DCF has become the standard choice for induction chemotherapy [10]. However, the DCF regimen is associated with a high risk of severe neutropenia or febrile neutropenia (FN).

Guidelines provide specific recommendations for the use of granulocyte colony-stimulating factor (G-CSF) [11, 12]. Primary prophylaxis with G-CSF is recommended for the prevention of FN in patients who are at risk, because of their age, medical comorbidities, disease characteristics, and chemotherapy myelotoxicity (overall FN incidence estimated at greater than 20 %). Multiple-dose daily G-CSF and pegfilgrastim are approved for use in the prevention of neutropenia and its complications with a similar efficacy in reducing the incidence and duration of FN [13]. According to the guidelines, G-CSF should only be administered 24 h after chemotherapy. However, in many chemotherapy regimens, especially those with IV continuous 5-fluorouracil (5-FU), the 24-h waiting period to administer G-CSF after stopping chemotherapy is not appropriate as the absolute neutrophil count (ANC) nadir appears earlier than 24 h. The implementation of the 24-h waiting period may result in a potential risk of severe neutropenia or FN. In our clinical practice with DCF (docetaxel, cisplatin, and IV 120 h continuous 5-FU) regimen, ANC nadir is usually observed from day 7, either 24 h after stopping 5-FU perfusion, or on the same day that pegfilgrastim administration starts. Such situations are not rare with standard chemotherapies for solid tumors, especially in head and neck or gastroesophageal cancers, where IV continuous 5-fluorouracil administration or oral 5-FU prodrug (capecitabine) is usually used and given until at least day 6 (DCF protocol) [8] or beyond [14, 15]. The increased risk of severe neutropenia with these protocols indicates that we should prescribe G-CSF primary prophylaxis, but the optimal administration schedule remains unclear for physicians despite the existence of current guidelines.

The concomitant use of G-CSF and chemotherapy has been investigated in several studies, with controversial results; some reports have shown that early G-CSF stimulation was safe and effective, in comparison to its administration on the following day after chemotherapy, in the prevention of neutropenia and its complications [16–20], whereas others showed that same-day administration of G-CSF was less efficacious in reducing the duration and severity of neutropenia [21, 22] than next-day administration. Moreover, no trial evaluated G-CSF during a prolonged IV 5-FU perfusion and none described optimal G-CSF administration modalities.

We conducted this retrospective and observational study to evaluate the optimal schedule of administration of G-CSF as primary prophylaxis with DCF chemotherapy regimen: given either early (on D3) in the course of 5-FU, or at the end of 5-FU perfusion (D7), as recommended by the Learned Societies. The secondary objectives were to evaluate the safety of G-CSF at day 3 versus day 7 and to estimate the impact of both administration modalities on patients’ outcomes (overall survival (OS) and progression-free survival (PFS)).

Patients and methods

Study design and patient selection

This was a retrospective, observational, and descriptive study. Data of 70 patients with advanced larynx or hypopharynx squamous cell carcinoma treated by DCF chemotherapy (docetaxel, cisplatin, and 5-fluorouracil) between January 2003 and December 2010 in a single institution were collected using patients’ medical records. To be eligible, patients had to be ≥19 years old, present no distant metastases, and have a WHO PS lower than 3. Before initiation, the study protocol was reviewed and approved by a scientific committee.

Chemotherapy regimen consisted of docetaxel 75 mg/m2 on day 1, cisplatin 75 mg/m2 on day 1 (or carboplatin if contraindication), and 5-fluorouracil 750 mg/m2/day from day 1 to day 5 as 120 h of intravenous infusion. Chemotherapy was administered as per standard practice and the summary of product characteristics.

Chemotherapy was repeated every 3 weeks until disease progression or unacceptable toxicity. Patients were followed up as per the standard care of practice for the institution. A complete blood count and biochemistry tests were performed at the beginning of the chemotherapy regimen, then every 3 weeks thereafter, or as deemed necessary. Patients were evaluated for tumor response after 3 cycles according to WHO criteria. Patients who responded (response ≥50 %) underwent larynx remobilization and radiotherapy, performed by conformational techniques or by intensity modulation (2 Gy per fraction, one fraction per day, 5 days per week for 7 weeks). Patients who did not respond underwent a total pharyngolaryngectomy.

During the courses of chemotherapy, three distinct strategies to prevent infectious risk were performed. The patients received primary prophylaxis by G-CSF, either as a single injection of pegfilgrastim 6 mg on day 7, or at least 24 h after the end of chemotherapy, or as a single injection of pegfilgrastim 6 mg on day 3 or filgrastim 30 MU from day 3 to day 11. The third group received primary antibioprophylaxis by ciprofloxacine 500 mg twice a day for 10 days (from D5 to D15) at each cycle.

Statistical considerations

Descriptive statistics were used to describe patient demographics, disease characteristics, and the prophylactic strategies with G-CSF (mean and standard deviations for the continuous data, frequency and percentages for categorical data; 95 % confidence interval was calculated when relevant). The descriptive analysis of the population was performed using Pearson’s chi-squared test, Fisher’s exact test, or the Wilcoxon signed-rank test.

Primary study endpoint was the rate of grade 3–4 neutropenia and FN (defined as a high-risk infectious situation) assessed at each cycle of DCF chemotherapy (defined as the time elapsing between day 1 of two consecutives cycles). Safety evaluation was evaluated using the NCI-CTC V3.0 grading system of adverse events. The following variables were assessed: incidence of FN and neutropenic events (NE), median day and incidence of severe neutropenia (grades 3–4), percentage of patients requiring hospitalization due to FN, cycle delay, dose reduction, and impact of chemotherapy discontinuation due to NE on patient’s efficacy outcomes.

Patients were stratified according to the chemotherapy-induced risk of infectious complications. Patients at “high infectious risk” were defined as those who experienced severe (grade 3–4) neutropenia, or FN, and those who postponed chemotherapy courses due to the absence of hematopoietic recovery or severe infectious complications in the intercure period.

Overall response rates (ORR) after 3 cycles of DCF and larynx preservation rates at 3 months after chemoradiotherapy were collected. PFS or overall survival OS was respectively defined as the period between the date of diagnosis and the date of progression (clinical, biological, or radiological progression) or the date of death regardless of the cause.

Tumor response and toxicity were compared for the two schedules of G-CSF administration (day 3 or day 7) and the third with antibioprophylaxis, by chi-squared test or Fisher’s exact test. PFS and OS were compared between patients at low or high infectious risk and concerned only 61 patients with laryngeal preservation and similar treatment (excluding early death and laryngectomy to limit selection bias). Efficacy and safety predictive factors were analyzed by logistic regression. The type 1 error risk was always set at 5 %. Survival curves were added (Kaplan Meier method). Statistical analysis was performed on SAS 9.1.3© (USA). This study was approved by an ethics committee.

Results

A total of 70 patients who fulfilled the inclusion criteria were included. There were 63 male (90 %) and 7 female (10 %) patients. Median age of the study cohort was 56 years old, ranging from 45 to 77 years. The primary tumor was predominantly located in the hypopharynx (62.9 % of the patients), with 52.9 % at stage T3 and 25.7 % at stage T4. The majority of patients (90.0 %) had a good performance status (<2). The clinical characteristics of the patients are reported in Table 1. No significant differences were observed between the populations with “high or low infectious risk.”

Table 1 The clinical characteristics of patients
Full size table

Treatment administration and efficacy outcomes

A total of 59 (84.3 %) patients received at least 3 cycles of chemotherapy. Mean (±SD) number of cycles received was 2.9 (±0.76). The majority of patients (n = 60; 85.7 %) received the full dose of chemotherapy. Dose reduction occurred in only 10 patients (14.3 %), of whom 8 (11.4 %) had cycle delay (Table 2). A total of 36 patients (51.4 %) received pegfilgrastim on D7, and 28 patients (40 %) started G-CSF prophylaxis during the course of chemotherapy, of whom 12 patients (17.1 %) had daily administration of filgrastim and 16 patients (22.9 %) received pegfilgrastim on D3. A total of six patients (8.6 %) had ciprofloxacin as primary antibioprophylaxis during the course of chemotherapy.

Table 2 Chemotherapy administered during the induction
Full size table

Of the 70 treated patients, 67 (95.7 %) were evaluable for response. The main reason for nonevaluability was early death (n = 3 patients), due to infectious complications arising during induction chemotherapy which only occurred in the late G-CSF group (day 7). A complete response (CR) was reported in 45 of the 67 evaluable patients (67.2 %) and a partial response (PR) was reported in 15 patients (22.4 %), yielding an ORR of 89.6 %.

All responder patients underwent radiotherapy with (n = 29) or without (n = 31) concomitant chemotherapy. The rate of larynx preservation was 89.6 % (60 patients) with a functional larynx at 3 months. Of the seven patients who did not respond, six underwent radical surgery (pharyngolaryngectomy), and one patient was treated by conventional radiotherapy (refused surgery). Median follow-up was in 24.8 months; PFS was 74.2 % at 1 year, 64.6 % at 2 years, and 62.7 % at 3 years; and OS was 86.5 % at 1 year, 77.2 % at 2 years, and 72.8 % at 3 years (16 events and 28 patients were followed up).

Incidences of grade 3–4 neutropenia, FN, and cycle delay due to infectious complications and their impact on patient outcomes

Table 3 summarizes the incidence of severe neutropenia (grades 3–4), FN, and cycle delays due to infectious complications in all patients and according to the prophylactic strategies developed to prevent these events. Overall, severe neutropenia was reported in 22.9 % of patients. Day 7 pegfilgrastim was correlated with a statistically significant increase in severe grade 3–4 neutropenia (p = 0.0235), FN (p = 0.0290), and chemotherapy discontinuation (p = 0.0057) in comparison with daily G-CSF or D3 pegfilgrastim administration. No significant difference was observed between antibioprophylaxis and use of prophylactic G-CSF during the course of chemotherapy.

Table 3 Incidences of grade 3–4 neutropenia, FN, and cycle delay due to infectious complications in all patients and according to the prophylactic strategies
Full size table

No significant difference in PFS (p = 0.28) was reported between patients at “high infectious risk” (12 evaluated patients out of 61) in comparison with the group of “low-risk” patients (49 patients). PFS rates were 71.4 % at 1 year and 57.1 % at 2 years in the group with “higher risk of infectious complications” versus 79.8 % and 72.8 % at 1 and 2 years, respectively, in the low-risk group (Fig. 1). A significant difference in OS (p = 0.035) in favor of the group of patients at a lower risk of infectious complications during the induction chemotherapy was observed (Fig. 2).

Fig. 1
figure 1

Progression-free survival according to chemotherapy-induced risk of infectious complications

Full size image
Fig. 2
figure 2

Overall survival according to chemotherapy-induced risk of infectious complications

Full size image

Discussion

The purpose of our study was to evaluate the safety and efficacy of early administration of G-CSF (pegfilgrastim at day 3 or filgrastim from day 3 to day 11) versus pegfilgrastim day 7 (as per current guidelines), in patients treated for locally advanced HNC by DCF protocol with continuous 5-FU. After 3 cycles of DCF, ORR was 89.6 %, with 67.2 % of CR and 89.6 % larynx preservation at 1 year. PFS and OS were also higher than the reported data in the literature (74.2 % at 1 year and 62.7 % at 3 years for PFS, and 86.5 % and 72.8 % at 1 and 3 years, respectively, for OS) [4, 8]. It should be noted that our cohort comprised only laryngeal and hypopharyngeal carcinoma with 52.9 % of patients at stage T3 and 21.4 % at stage T1/T2, and the majority of the patients had a good PS (90 % < 2). Three (4.3 %) toxic deaths occurred during the course of chemotherapy, due to infectious complications, and all were in the late G-CSF group (day 7), which remains unacceptable for this curative intent. Only six patients (8.6 %) received primary antibioprophylaxis by fluoroquinolones. The use of primary antibioprophylaxis remains a subject of debate because of the increased risk of development of bacterial resistances. In spite of its effectiveness in reducing the risk of chemo-induced infectious complications, primary antibioprophylaxis is not recommended in routine practice by the EORTC [23]. In our study, we evaluated two modes of administration of primary G-CSF prophylaxis: an early administration during the course of chemotherapy (at day 3 or 4 in 40 % of the patients), covering in theory the neutrophil nadir, and an administration after the end of chemotherapy, as per the current recommendations (at day 7 in 51.4 % of the patients). Our results show that patients with a higher infectious risk (16 out of 62; 25.8 %) had a significantly lower OS than the others (57.1 % at 2 years versus 72.8 %, p = 0.0348). The only predictive factor that was found to be associated with an increased risk of infectious complications due to chemotherapy was the administration of G-CSF after the end of chemotherapy, i.e., at day 7. Indeed D7 administration of pegfilgrastim was significantly associated with an increased infectious risk or more precisely an inefficiency in the prevention of this risk, with 32 events (14 neutropenia events and 9 discontinuations of cures) in 36 patients against only 3 events (two grade 3–4 neutropenia and one FN) in 34 patients having received early G-CSF administration or antibioprophylaxis. No difference in OS was observed between the two groups of patients, probably because of a lack of statistical power, although a trend was observed in favor of the group of patients who received early G-CSF administration (84.7 % for early D3; G-CSF versus 77.2 % at 2 years for G-CSF stimulation at day 7). Our findings are consistent with those published by Bonnin et al. [7] who reported an incidence of 23 % for grade 3–4 neutropenia and 14 % for FN despite the use of primary G-CSF prophylaxis at the end of 5-FU perfusion. DCF protocol is associated with a high risk of infection, FN, and severe neutropenia. The use of G-CSF prophylaxis appears to be an alternative option to prevent this risk. However, current guidelines do not specify the optimal modality of administration of G-CSF when chemotherapy is given continuously for several days. The only recommendation is to administer G-CSF at the end of cytotoxic chemotherapy based on the theoretical risk of myelotoxicity due to an increase in the sensitivity of rapidly dividing myeloid cells. There is a lack of published data on the concomitant use of G-CSF and chemotherapy, for instance with continuous IV-based chemotherapy as the DCF regimen. Our study is the first to evaluate this modality of administration of G-CSF with the DCF regimen. In our study, concomitant administration of G-CSF with chemotherapy did not result in increased hematological toxicity and even appeared to be the only relevant mode of administration which could reduce the risk of infectious complications and maintain dose density of the chemotherapy. The analysis presented here shows that early administration of G-CSF, concomitant to chemotherapy, is safe and feasible and could be generalized to similar situations in clinical practice, such as advanced gastric cancer for example, where the use of intravenous continuous 5-fluorouracil or oral 5-FU prodrug (DCF, ECX, EOX, or ECF) is standard.

In conclusion, our results have shown that early administration of G-CSF, as primary prophylaxis for locally advanced HNC, appeared to be effective and safe in reducing the risk of severe neutropenic events during a DCF regimen.