Purpose: We previously demonstrated high locoregional control, in patients with poor-prognosis head and neck cancer (HNC), using paclitaxel, 5-fluorouracil, hydroxyurea, and concomitant hyperfractionated radiotherapy. In the present phase I trial, gemcitabine, a novel antimetabolite with strong radiation-enhancing activity, replaces hydroxyurea. We sought to determine the recommended phase II dose and clinical efficacy in poor-prognosis HNC patients.
Experimental Design: Seventy-two patients enrolled. Eligibility criteria included recurrent or second primary HNC, metastases or expected 2-year survival <20%. Chemoradiotherapy consisted of 5-fluorouracil, 600 mg/m2/d, for 5 days; paclitaxel, 100 mg/m2 on Day 1; and concurrent 1.5 Gy twice-daily radiation for 5 days. Gemcitabine was dose escalated, 50–300 mg/m2 on day 1. Cycles repeated every 14 days until the completion of chemoradiation. Dose-limiting toxicities (DLTs) included: neutropenic fever; grade ≥4 neutropenia or thrombocytopenia for >4 days; grade ≥4 mucositis or dermatitis for >7 days; or grade 3 toxicity necessitating chemotherapy dose reductions. Non-DLT dose reductions in 5-fluorouracil and/or paclitaxel were allowed.
Results: Seventy-nine percent of assessable patients experienced a clinical response. Five-year actuarial survival is 33.0%, and locoregional control is 61.4%. The recommended phase II dose of gemcitabine in this regimen is 100 mg/m2 during cycles 1–5 (1 of 7 patients with DLT) or 200 mg/m2 delivered only during cycles 3–5 (3 of 19 with DLT). Grades 3 and 4 mucositis (56 and 21%, respectively) and dermatitis (25 and 21%, respectively) were common.
Conclusions: Gemcitabine, 5-fluorouracil, paclitaxel, and twice-daily radiation, delivered on alternating weeks, is active in patients with poor-prognosis HNC, although severe mucositis limits the clinical applicability of this regimen. Refinements in radiotherapy, including intensity-modulated radiation therapy, may improve the tolerance for this regimen.
Surgery and/or radiation therapy have been the standard treatment for locoregionally advanced head and neck cancer. Definitive radiation therapy offers an organ-sparing alternative to surgery, with altered fractionation schemes resulting in improved outcomes (1) . In multiple randomized trials, definitive chemoradiotherapy has proved superior to radiation therapy alone in terms of locoregional control, progression-free survival, and overall survival (2, 3, 4, 5, 6, 7, 8, 9) . Some of these studies used twice-daily radiation therapy (2 , 3 , 6) .
The University of Chicago has investigated chemoradiotherapy as an organ-sparing definitive treatment of advanced head and neck malignancies, with concurrent radiation therapy, 5-fluorouracil and hydroxyurea administered on an alternating-week schedule (FHX regimens; refs. 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 ). Both agents are radiosensitizers; in addition, hydroxyurea enhances the intracellular activity of 5-fluorouracil via the inhibition of ribonucleotide reduction. Hydroxyurea depletes dUMP, which enhances 5-fluorouracil binding to thymidylate synthetase.
Twice-daily radiation therapy (1.5 Gy twice a day) was introduced into our protocols to accelerate treatment delivery (13 , 15) . In recent studies, concurrent paclitaxel was added to 5-fluorouracil and hydroxyurea with twice-daily radiation therapy (TFHX regimens; refs. 20, 21, 22, 23 ). The TFHX regimens yielded locoregional control on the order of 65–86%. TFHX was relatively well tolerated, with most patients able to receive full-dose chemotherapy.
The FHX (10, 11, 12) and TFHX (20 , 21) regimens were initially studied in phase I trials in which patients with recurrent or second primary head and neck cancer (some previously irradiated), metastatic disease, poor prognosis, and/or poor anticipated survival were eligible. Eligibility criteria were the same in the present phase I trial, in which concurrent gemcitabine replaces hydroxyurea in TFHX regimen (TFGX regimen). Reirradiation (with concurrent chemotherapy) of unresectable locoregionally recurrent (or second primary) head and neck cancer, offers the possibility of long-term survival on the order of 20% (24 , 25) ; in contrast, palliative chemotherapy alone prolongs median survival but does not offer the potential for long-term disease control.
Gemcitabine is a deoxycytidine analogue activated intracellularly to its 5-triphosphate product. It acts as an inhibitor or substrate for replicative DNA polymerases resulting in the inhibition of DNA synthesis. It also exerts cytotoxic effects through a self-potentiation mechanism, whereby both the diphosphate form (which inhibits ribonucleotide reductase) and triphosphate form (which acts as a competitive inhibitor) contribute to the reduction of dCTP levels, facilitating increased phosphorylation and decreased elimination of the drug (26) . Its inhibition of ribonucleotide reduction may enhance the intracellular activity of 5-fluorouracil, hence justifying the replacement of hydroxyurea with gemcitabine in the present study.
In both preclinical and clinical studies, gemcitabine has demonstrated single-agent activity in head and neck cancer (27 , 28) . In preclinical studies, gemcitabine has also proved to be a potent radiosensitizer in human cancer (including head and neck) cell lines (29, 30, 31, 32, 33) . Gemcitabine-based radioenhancement is associated with a depletion of endogenous deoxynucleotide pools and redistribution of cells into the S phase and likely results in the lowering of the threshold for radiation-induced apoptosis and the inhibition radiation-induced DNA repair (30 , 34) .
Several studies have investigated gemcitabine-based chemoradiotherapy regimens in resected pancreatic cancer (35) , unresectable pancreatic cancer (36, 37, 38, 39, 40, 41, 42) , and lung and pleural malignancies (43, 44, 45, 46) . With pancreatic malignancies, dose-limiting toxicities (DLTs) of gemcitabine-based chemoradiotherapy include cytopenias, anorexia, nausea, and vomiting, whereas with lung malignancies, DLTs include esophagitis, pneumonitis, and cytopenias. These toxicities suggest strong radiosensitization.
Here we report the DLT, recommended phase II dose, and efficacy of the TFGX regimen in locoregionally advanced head and neck cancer. Quality of life and late toxicity will be evaluated in a separate analysis. When this trial began, no toxicity or efficacy data were available for gemcitabine-based chemoradiotherapy in patients with advanced head and neck cancer.
PATIENTS AND METHODS
Patients were discussed in a multidisciplinary conference (at each individual institution), with a pathologist and medical, radiation, and surgical oncologists, to determine stage, optimal management and protocol eligibility. Patients were eligible if they had histological or cytological documentation of a malignant neoplasm located in the head and neck that required locoregional radiotherapy. Measurable disease was not required. Prior radiation therapy, chemotherapy (including 5-fluorouracil and/or paclitaxel), and/or surgery did not preclude enrollment. Patients presenting with new primary head and neck cancer were eligible if they had unresectable disease or an expected 2-year survival of <20%. Patients with metastatic disease or coexisting malignancies were eligible if the predominant site of head and neck cancer and associated symptoms required local treatment with radiation therapy. A Cancer and Leukemia Group B (CALGB) performance status ≤2 was required. Patients were excluded if they had a white blood cell (WBC) count ≤3,000/μl, absolute neutrophil count (ANC) ≤1,500/μl, platelet count ≤100,000/μl, creatinine >2.0 mg/dl, total bilirubin >1.5 mg/dl, or transaminase concentrations >2.5 times the upper limit of normal. Pregnancy was an absolute contraindication.
All of the patients underwent computed tomography scans of the head, neck and chest, bone scan, barium swallow, panendoscopy, tumor mapping, complete dental evaluation (unless edentulous), and routine blood work (serum chemistries and blood counts).
Written informed consent was obtained before the initiation of treatment. The protocol was approved by the institutional review boards of all of the participating institutions.
Each cycle of chemotherapy lasted 2 weeks, with treatment given on days 0 to 5 (week-on/week-off). Patients were admitted to the hospital on a Sunday (day 0) night, at which time, 120 h of continuous infusion 5-fluorouracil at 600 mg/m2/d was started. Paclitaxel was administered on day 1, between the first and second radiation therapy dose, at 100 mg/m2 over 1 h. Gemcitabine was administered after paclitaxel, at the appropriate phase 1 dose level, over 30 min. The doses of 5-fluorouracil and paclitaxel were derived from prior phase I studies (10, 11, 12 , 20 , 21) . 5-Fluorouracil was prepared in 1 liter of dextrose 5% water (D5W). Paclitaxel was prepared by diluting the total dose in 500 ml of D5W containing 0.9% NaCl. Gemcitabine was prepared by making a solution containing 10 or 40 mg/ml final concentration, and then diluted with 200–500 ml of normal saline, to allow for appropriate dose administration (50–300 mg/m2) over 30 min. Gemcitabine was started in the first (C1), third (C3), or fourth (C4) cycle (see below). Chemotherapy was infused through indwelling double-lumen tunneled catheters.
Patients with no measurable disease were prescribed four cycles (C1–4) of therapy, whereas those with measurable disease were prescribed five cycles (C1–5).
All of the patients were recommended to undergo central-line and feeding-tube placement. Oral hygiene was required, with normal saline, antifungal mouth rinse, or a combination of lidocaine, diphenhydramine, and sodium bicarbonate. A water-based aloe containing gel was applied to the skin within the radiation field after the afternoon radiation therapy and was removed before the next morning’s radiation therapy. Daily hydration with 1 to 2 liters of normal saline was given after chemotherapy. Electrolytes were replenished as needed. Antiemetics were used per physician discretion. In patients experiencing ≥grade 3 neutropenia, granulocyte colony-stimulating factor (G-CSF) was used at a dose of 5 μg/kg during days 6 through 12 during all remaining cycles.
Radiation Therapy Guidelines.
Each cycle of chemoradiotherapy comprised 5 consecutive days of radiation therapy, followed by a 9-day break. Radiation therapy was administered in twice-daily 1.5-Gy fractions. A daily 2.0-Gy fraction was allowed if the patient could only receive one radiation treatment on a given day. The prescribed doses to uninvolved areas or areas of microscopic disease were as follows: 39 to 50 Gy to the supraclavicular fossa, 45 to 60 Gy to the posterior neck, and 50 to 60 Gy to the anterior neck. The prescribed dose to gross disease was as follows: 66 to 74 Gy to the supraclavicular fossa, 66 to 74 Gy to the posterior neck, 70 to 74 Gy to small volume disease (≤4 cm), and 72 to 76 Gy to bulkier disease (>4 cm). Patients with M1 and Mx disease were treated in the same manner as those with M0 disease.
All of the patients were simulated before the start of radiation therapy with an appropriate immobilization device. Initial opposed lateral fields were used to treat the primary tumor and draining cervical lymphatics as deemed appropriate. A wedged pair or opposed oblique fields were allowed as well. If a supraclavicular field was used, it was administered by a separate, matched anterior-posterior field. A posterior spinal cord block or midline block on the lateral field was allowed to minimize overlap on the spinal cord. Custom blocks were used to shape the fields to shield critical structures and other areas not at risk for disease. Electrons were used to boost the posterior neck if the prescribed dose to the posterior neck was >40 Gy. Appropriate field reductions were used to administer the boost volumes.
In those patients with prior radiation therapy to the head and neck, the radiation therapy volume, techniques, and doses were left to the discretion of the radiation oncologist. At a minimum, the radiation therapy fields included all gross disease and areas of known microscopic disease. Generally, uninvolved lymphatics were included only if the cumulative spinal cord dose (including prior radiation therapy, although neglecting internal scatter) could be kept below threshold (50 Gy). Most were treated at least to the ipsilateral regional nodes. As with primary cancer, the prescribed protocol dose depended on whether or not measurable disease was present.
Patients were treated in cohorts of at least three individuals. 5-Fluorouracil and paclitaxel were administered at fixed doses. If a total of one patient experienced a DLT (see below), the gemcitabine dose was increased in the subsequent cohort. If three patients experienced a DLT, treatment was to be stopped. If two patients developed a DLT, an additional three patients were enrolled. Enrollment of additional patients at a given dose level was allowed to ascertain the safety of further dose escalation. Gemcitabine doses were to be escalated from 50 mg/m2 to 300 mg/m2 in 50-mg/m2 increments.
Initially, gemcitabine was to be given during all five cycles (C1–5). However, a significant number of patients developed grade 3–4 mucositis and/or dermatitis at the 150-mg/m2 dose level. Thus the protocol was amended such that gemcitabine was to be given only during cycles 3 to 5 (C3–5), starting at a dose of 150 mg/m2. After a dose escalation to a maximum 300 mg/m2 (at which level, all patients experienced a DLT), additional patients were enrolled at the 250-mg/m2, 200-mg/m2, and 150-mg/m2 dose levels to better ascertain the recommended phase II dose. The expanded dose cohorts are combined in the analyses.
Toxicity was graded by the Cancer and Leukemia Group B common toxicity criteria. Toxicities experienced during any cycle in which gemcitabine was delivered were used to determine DLTs. DLT was defined as follows: (a) grade ≥4 neutropenia for >4 days (on day 1 of next cycle) or a neutropenic fever despite therapy with G-CSF, grade ≥4 thrombocytopenia for >4 days on day 1 of next cycle; (b) grade ≥4 mucositis and/or dermatitis exceeding 7 days duration or not resolving to ≤grade 3 on day 1 of next cycle; and (c) any other grade 3 toxicity (except nausea and/or vomiting, or alopecia) resulting in a failure to receive the full dose of chemotherapy within 24 h of the scheduled time during the first three cycles. The administration of G-CSF in and of itself was not considered a DLT.
Chemotherapy dose reductions were allowed as follows. For grade 4 WBC count, grade 4 or grade 3–4 platelet count on day 1 of a cycle, the following modifications were allowed: no paclitaxel and no gemcitabine. For grade 3 WBC count, grade 3 ANC or grade 2 platelet count on day 1 of a cycle (which were not DLTs), the following reductions were allowed: reduce paclitaxel by 50% and gemcitabine by one dose level. For grade 4 mucositis and/or dermatitis exceeding 7 days’ duration or persisting on day 1 of a cycle, the following reductions were allowed: 5-fluorouracil to 500 mg/m2/d, paclitaxel to 80 mg/m2, and gemcitabine reduced by one dose level. No treatment delays were allowed for mucositis, dermatitis, or diarrhea. A cycle could be postponed in the presence of fever >38°C, for any clinically apparent infection, or at the discretion of the treating physician. Some patients underwent dose reductions for toxicities not considered DLTs (i.e., grade 4 WBC count, ANC, mucositis, and/or dermatitis resolved by day 1 of the next cycle, or grade 3 toxicities resulting in dose reductions during cycles 4–5).
Study End Points.
The primary end point was to describe the DLT of the regimen and to define the recommended phase II dose of gemcitabine combined with paclitaxel, 5-fluorouracil, and twice-daily radiation therapy (TFX). Response rate and survival were evaluated as secondary end points. Patients underwent re-evaluation, including computed tomography scans of the head, neck, and chest 4–6 weeks posttherapy and every 3 months thereafter. The initial re-evaluation included panendoscopy, tumor mapping, and biopsy when feasible. Complete response was defined as the disappearance of all detectable disease as assessed 4 to 8 weeks after completion of therapy. Partial response was defined as a ≥50% reduction in the sum of the products of the longest perpendicular diameters of all measurable tumor. Progressive disease was defined as a ≥25% increase. Patients with stable disease did not meet the criteria for partial response or progressive disease.
All of the patients were followed until recurrence or death. No patient was lost to follow-up. Survival and disease control parameters were calculated using Kaplan-Meier actuarial analyses, with survival and failure times defined from the first day of treatment until an event or date of last follow-up. Progression-free survival events included locoregional failure, distant failure and treatment related death. Treatment related deaths were also incorporated into cause-specific survival. Second cancers were not considered failures. Patients dying without evidence of disease recurrence were censored at the time of death in the analyses of progression-free survival, time to locoregional failure, and time to distant failure. Multivariate survival analyses were performed by Cox’s proportional hazards regression.
Patient accrual began in December of 1997 and was completed in January of 2001. One patient who enrolled at the 100-mg/m2 dose level refused any treatment; he is excluded from all analyses. Follow-up data are available through June of 2003. Follow-up ranged from 0.4 to 63.1 months (median, 19.7 months). For living patients, the median follow-up was 43.1 months. Patient and tumor characteristics of the 72 patients treated on the study are listed in Table 1⇓ . Table 2⇓ outlines the basis for patient eligibility in the study and the details of prior therapy. Thirty-five patients received prior radiation therapy to the head and neck; of these, 1 was treated with mantle radiation therapy (with an unknown dose) for Hodgkin’s disease. Of the 34 patients previously irradiated for head and neck cancer, the prior radiation therapy doses ranged from 45 to 111 Gy with a mean and median of 65.6 Gy and 64.9 Gy, respectively, from 4.7 to 133 months before protocol enrollment (mean, 30.6 months; median, 16.8 months). Forty-one patients had recurrent disease, whereas 31 presented with new primary cancers. Table 3⇓ summarizes the treatment delivered to the 72 patients during the study.
Phase I Dose Levels and Toxicity.
Table 4⇓ outlines the dose levels and the number of patients treated per cohort. Fifteen patients failed to complete treatment; of these, 4 experienced a DLT. The reasons for treatment discontinuation are also summarized in Table 4⇓ . Of the 15 patients who discontinued treatment, 4 received no gemcitabine (and are not evaluable for acute toxicity or for DLT), and 11 received at least one cycle with gemcitabine (and are included in the acute toxicity and DLT analyses).
DLTs were considered only in cycles in which gemcitabine was delivered, although no toxicity equivalent to a DLT was experienced in those cycles delivered without gemcitabine. Table 5⇓ outlines the DLTs at each dose level. Fifteen experienced a DLT; of these, 10 underwent dose reductions (see below), 4 discontinued treatment (Table 4)⇓ , and 1 continued treatment without a dose reduction. One patient, enrolled at the C3–5 150-mg/m2 dose level, died unexpectedly in her sleep; no autopsy was performed per family wishes. Although she is considered to have had a treatment-related death in the survival analysis, she is not considered a DLT because she experienced no discernible grade ≥3 toxicity. There was one grade 5 infection (fungemia) occurring after the fifth cycle at the C3–5 200-mg/m2 dose level. This patient experienced grade 3 WBC count and ANC toxicity; hers was considered a treatment-related death but not a DLT.
Table 6⇓ summarizes the worst grade 3–5 toxicities experienced at each dose level. The predominant toxicities were mucositis and dermatitis. Overall, grade 3 mucositis occurred in 56% of patients, whereas grade 3 dermatitis occurred in 25%. Grade 4 mucositis occurred in 21%, and grade 4 dermatitis occurred in 21%. No grade 4 mucositis occurred at the two lowest gemcitabine dose levels. Twenty (35%) of the 57 patients completing treatment required a reduction in one or more chemotherapeutic agents; 10 of these had a DLT. Fifteen underwent a gemcitabine dose reduction (all but 1 patient enrolled at ≥150 mg/m2 dose levels); 9 also had dose reductions in 5-fluorouracil and/or paclitaxel; 5 underwent reductions in 5-fluorouracil and/or paclitaxel without reductions in gemcitabine.
During the initial phase of the trial, in which patients received gemcitabine during all five cycles, there were two DLTs and one treatment discontinuation among seven patients treated at the C1–5 150-mg/m2 dose level. The C1–5 100-mg/m2 dose level, with one DLT and one treatment discontinuation is, therefore, the recommended phase II dose for C1–5 gemcitabine dosing.
During the initial dose escalation phase, none of four patients at the C3–5 200-mg/m2 dose level experienced a DLT, and one of four patients discontinued treatment (progressive disease). At the C3–5 250 mg/m2 dose level, one of seven patients experienced a DLT (grade 3 anemia and diarrhea) resulting in a treatment discontinuation. At the C3–5 300-mg/m2 dose level, five patients were enrolled, with one discontinuing treatment after 1 cycle because of personal reasons and the other four experiencing DLTs (see Table 5⇓ ). The C3–5 300-mg/m2 dose level was not tolerated, and, hence, additional patients were enrolled at lower dose levels to better ascertain the recommended phase II dose. An additional four patients were enrolled at the 250-mg/m2 dose level. Although only one DLT was observed in the 11 patients treated at the C3–5 250-mg/m2 dose level, these patients experienced significant grade 3–4 toxicity (see Table 6⇓ ). An additional 18 patients were enrolled at the C3–5 200 mg/m2 dose level. Of the 22 patients enrolled at this level, 3 discontinued treatment before receiving gemcitabine (see Table 6⇓ ), leaving 19 assessable for toxicity. Three of 19 experienced a DLT, and an additional 3 discontinued treatment. The 200-mg/m2 dose is the recommended phase II dose for C3–5 gemcitabine dosing
Eleven patients were enrolled at the C4–5 150 mg/m2 dose level, which demonstrated similar toxicity to the C3–5 150–250 mg/m2 regimens (see Table 6⇓ ), with 3 DLTs, 1 of which resulted in a treatment discontinuation.
Response to Treatment.
Twenty-one patients were treated with no measurable disease, whereas 51 patients had measurable disease, amenable to quantification of response to treatment. Five were not evaluable for the following reasons: two early deaths, one entered hospice, and two refused further treatment. A complete response was seen in 27 (59%), with pathological confirmation of a complete response in 16 of 27. Two (4%) patients with pathological partial responses were surgically salvaged after therapy. A clinical partial response was noted in seven (15%) and clinically stable disease in two (4%). Only 3 (7%) of 46 experienced local progression, and five experienced early distant failure.
For all 72 patients, the 2-year and 5-year actuarial values are as follows, respectively: overall survival of 45.7 and 33.0%; cause-specific survival of 49.1 and 42.5%; progression-free survival of 41.5 and 39.6%; locoregional control of 64.1 and 61.4%; and distant control of 58.4 and 58.4%. The median survival is 21.9 months, and median progression-free survival is 14.3 months. The actuarial survival and disease control curves are shown in Figs. 1⇓ and 2⇓ . The 5-year overall survival and progression-free survival of the 12 patients with non-squamous cell cancer is 33 and 40%, respectively, essentially equivalent to that of the 60 patients with squamous cell cancer. There was better locoregional control with non-squamous cell versus squamous cell histologies (5-year locoregional control of 88.8 versus 56.8%; P = 0.10), and worse distant control (5-year distant control of 40.0 versus 62.3%; P = 0.15).
In the 29 patients with previously irradiated head and neck cancer and no distant metastases (no M1 or Mx) at presentation, the 5-year overall survival is 37.6% (shown in Fig. 3⇓ ). Survival is more than doubled in the cohort of 12 patients with disease amenable to debulking before chemoradiotherapy (57.1 versus 23.5%; P = 0.068). These patients will be discussed in more detail in a future analysis.
On univariate analysis, poor prognostic factors were as follows: metastatic involvement was significant for progression-free survival (P = 0.005) and distant control (P = 0.001), whereas measurable disease was significant for overall survival (60.4 versus 21.2% at 5 years; P = 0.002), cause-specific survival (65.9 versus 31.8% at 5 years; P = 0.01), and progression-free survival (60.0 versus 29.7% at 5 years; P = 0.04). A multivariate analysis was performed with the following variables: gender, age, performance status, extent of tobacco use, extent of alcohol use, metastatic involvement, nodal involvement, measurable disease, recurrent versus primary disease, and gemcitabine dose. Gender, extent of tobacco use, extent of alcohol use, recurrent versus primary disease, and gemcitabine dose were not significant. The multivariate analysis was re-run without these five variables. Poor prognostic factors were as follows: measurable disease was significant for overall survival (P = 0.018) and cause-specific survival (P = 0.043), nodal involvement was significant for locoregional control (P = 0.048) and distant control (P = 0.016), and both age and metastatic involvement were significant for progression-free survival (P = 0.041 and P = 0.051, respectively, and distant control (P = 0.003 and P < 0.0001, respectively).
This study demonstrates the clinical radioenhancement of gemcitabine given in conjunction with 5-fluorouracil and paclitaxel, concurrently with twice-daily radiation therapy on an alternating week schedule, in patients with locally advanced head and neck cancer. In the TFGX regimen, the recommended phase II dose of gemcitabine is 100 mg/m2 weekly, delivered in cycles 1 to 5, or 200 mg/m2 weekly, delivered in cycles 3–5, each dose given with concurrent continuous infusion 5-fluorouracil (600 mg/m2/d) and weekly paclitaxel (100 mg/m2). At the C1–5 100-mg/m2 weekly dosage, 1 of 7 patients experienced a DLT; at the C3–5 200-mg/m2 weekly dosage, 3 of 19 experienced a DLT.
In patients with locoregionally advanced head and neck cancer, TFGX yields good clinical outcome considering the advanced stage of disease, with complete response of 59%, 5-year locoregional control >60%, and 5-year cause-specific survival >40%. From prior experience, 60–70% of clinical and/or radiographic partial responses proved to be pathological complete responses (22) . This promising clinical outcome is at the expense of a large percentage of patients experiencing grade 3–4 toxicity. Only 79% completed their course of chemoradiotherapy, and, of these, 35% underwent chemotherapy dose reductions because of dose-related events. DLT criteria were liberal, reflecting the anticipated need for aggressive treatment in these patients. Acute mucositis and/or dermatitis accounted for most of the toxicity of this regimen. Grade 3 mucositis is common with our aggressive chemoradiotherapy regimens and was, therefore, expected. However, the relatively high rate of grade 4 dermatitis (21%) and grade 4 mucositis (21%) was disappointing. Certainly, these toxicities reflect the potent radiosensitizing activity of gemcitabine.
The TFGX regimen followed from our experience with TFHX (21 , 22) . TFGX differs from TFHX in that gemcitabine was given instead of hydroxyurea. TFHX was relatively well tolerated in the initial phase I study, in which no grade 3–4 mucositis or dermatitis was seen in eight patients treated at the recommended phase II dose of hydroxyurea, 500 mg twice daily; 5-fluorouracil, 600 mg/m2/d; and paclitaxel, 20 mg/m2/d (20) . However, in the phase II study of TFHX at that dose, 84% experienced grade 3–4 mucositis (∼30% grade 4) and 38% experienced grade 3–4 dermatitis (22) . Although the rates of acute mucositis and dermatitis are similar to those seen in the TFGX regimen, TFHX was better tolerated in that fewer patients underwent dose reductions, and 95% received all of the planned weeks of radiation therapy (versus 79% in the present trial).
Similar findings were seen in a second phase-I-II study of TFHX, in which paclitaxel was administered on the first day of each cycle (as was done in the present trial) with dose escalation from 60–150 mg/m2 (21) . The toxicity and dose intensity in the second TFHX regimen is similar to that seen in TGHX. However, TFHX has remained well tolerated in patients definitively treated with primary cancers (22 , 23 , 47 , 48) .
The University of Michigan recently published a phase I study in which 29 patients with unresectable head and neck cancer were treated with weekly gemcitabine concurrent with daily radiation therapy to a dose of 70 Gy (49) . Unlike the present study, patients previously treated with chemotherapy or radiation therapy were not eligible, nor were patients presenting with metastatic disease. Gemcitabine was de-escalated because of toxicity, initially from 300 mg/m2 down to 10 mg/m2. No patients treated at the 10 mg/m2 level experienced grade 3–4 acute mucosal or skin toxicity; in contrast, at the 50–300 mg/m2 dose levels, 77% experienced grade 3–4 mucosal toxicity and 65% experienced grade 3–4 skin toxicity. The complete response rate was 78% (21 of 27); locoregional failure developed in 13 (44%) of 29 and distant failure in 5 (17%) of 29. Interestingly, gemcitabine acts as a potent radiosensitizer at doses of 5 to 30% of the maximum tolerated doses of gemcitabine delivered alone (49) . DLTs occur at lower doses of gemcitabine than used in pancreatic or lung malignancies, which may be attributable to lower threshold doses of oropharyngeal mucosa as compared with duodenal or esophageal mucosa (49) .
The TFGX regimen differs from the University of Michigan regimen in several respects: paclitaxel and 5-fluorouracil were used in conjunction with gemcitabine; radiation therapy was delivered twice daily; and treatment was delivered via an alternating week schedule. With TFGX, the rate of grade 3–4 acute mucositis was 77% and dermatitis was 46%, similar to the University of Michigan regimen. However, TFGX resulted in a higher percentage of grade 4 mucositis (21, versus 4% in the University of Michigan regimen) and grade 4 dermatitis (21, versus 8% in the University of Michigan regimen). A comparison of response rate, disease control, and survival rate with TFGX versus the University of Michigan regimen would not be meaningful because the patient populations were substantially different.
In summary, gemcitabine is a potent radiosensitizer, yielding excellent disease response and disease control at relatively low doses in the TFGX regimen. Toxicity, namely severe mucositis and dermatitis, is prohibitive. The long-term survival in those patients previously irradiated to the head and neck is particularly promising and warrants further study. In recent studies, we and others have devoted efforts to minimizing skin and mucosal toxicity while maintaining excellent clinical outcomes (22 , 23 , 47 , 48) . Most recently, we returned to a TFHX regimen, in which three cycles of induction carboplatin and paclitaxel precede chemoradiotherapy. With this regimen, much less skin and mucosal toxicity is seen as compared with the present study, and with impressive clinical response rates and disease control (47 , 48) . Future directions include attempts at reducing radiation dose to clinically uninvolved areas (48) , using intensity-modulated radiation therapy (50) , and studying novel chemotherapeutics.
We would like to thank Allison Dekker for nursing care and Rosalind Williams for data management. We would like to thank Dr. Merrill Kies for patient care.
Grant support: This study was supported in part by the University of Chicago/Northwestern University Oral Cancer Research Center (P50 DE11921-04); University of Chicago Cancer Research Center (P30 CA14599); The Francis Lederer Foundation; The Geraldi Norton Memorial Corporation; The Robert and Valda Svendsen Memorial; and Eli Lilly Pharmaceuticals, Indianapolis, IN.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Everett Vokes, University of Chicago, 5841 South Maryland Avenue, MC 2115, Chicago, IL 60637. E-mail:
- Received November 24, 2003.
- Revision received April 30, 2004.
- Accepted May 3, 2004.