Poor Prognosis Head Research Articles

Interpretable survival network for progression risk analysis of multimodality imaging biomarkers in poor-prognosis head and neck cancers

This study explores the predictive utility of multi-time point, multi-modality quantitative imaging biomarkers (QIBs) and clinical factors in patients with poor-prognosis head and neck cancers (HNCs) using interpretable machine learning. We examined 93 patients with p16 + oropharyngeal squamous cell carcinoma or locally advanced p16- HNCs enrolled in a phase II adaptive radiation dose escalation trial. FDG-PET and multiparametric MRI scans were conducted before radiation therapy and at the 10th fraction (2 weeks). A survival network analyzed MRI and PET-derived biomarkers such as gross tumor volume (GTV), blood volume (BV), and metabolic tumor volume (MTV50), along with clinical factors to predict local (LF) and distant failures (DF). Feature attributions and interactions were assessed using Expected Gradients (EG) and Expected Hessian (EH). Through rigorous cross-validation, the model for predicting LF, incorporating biomarkers like p16 status and radiation boost, achieved a c-index of 0.758. Similarly, the DF prediction model showed a c-index of 0.695. The analysis of feature attributions and interactions enhanced understanding of important features and complex factor interplays, potentially guiding more personalized and intensified treatment approaches for HNC patients.

Prediction of Prognosis of Tongue Squamous Cell Carcinoma Based on Clinical MR Imaging Data Modeling.

Objective: Tongue squamous cell carcinoma (TSCC) is one of the most common and poor prognosis head and neck tumors. The purpose of this study is to establish a model for predicting TSCC prognosis based on clinical and MR radiomics data and to develop a nomogram. Methods: A retrospective analysis was performed on the clinical and imaging data of 211 patients with pathologically confirmed TSCC who underwent radical surgery at xx hospital from February 2011 to January 2020. Patients were divided into a study group (recurrence, metastasis, and death, n = 76) and a control group (normal survival, n = 135) according to 1 to 6 years of follow-up. A training set and a test set were established based on a ratio of 7:3 and a time point. In the training set, 3 prediction models (clinical data model, imaging model, and combined model) were established based on the MR radiomics score (Radscore) combined with clinical features. The predictive performance of these models was compared using the Delong curve, and the clinical net benefit of the model was tested using the decision curve. Then, the external validation of the model was performed in the test set, and a nomogram for predicting TSCC prognosis was developed. Results: Univariate analysis confirmed that betel nut consumption, spicy hot pot or pickled food, unclean oral sex, drug use, platelet/lymphocyte ratio (PLR), neutrophil/lymphocyte ratio (NLR), depth of invasion (DOI), low differentiation, clinical stage, and Radscore were factors that affected TSCC prognosis (P < .05). In the test set, the combined model based on these factors had the highest predictive performance for TSCC prognosis (area under curve (AUC) AUC: 0.870, 95% CI [0.761-0.942]), which was significantly higher than the clinical model (AUC: 0.730, 95% CI [0.602-0.835], P = .033) and imaging model (AUC: 0.765, 95% CI [0.640-0.863], P = .074). The decision curve also confirmed the higher clinical net benefit of the combined model, and these results were validated in the test set. The nomogram developed based on the combined model received good evaluation in clinical application. Conclusion: MR-LASSO extracted texture parameters can help improve the performance of TSCC prognosis models. The combined model and nomogram provide support for postoperative clinical treatment management of TSCC.

Randomized Phase II Study of DCE-MRI-Based Dose Escalation for Poor-Prognosis Head and Neck Cancer

<h3>Purpose/Objective(s)</h3> Loco-regional failure accounts for the majority of failures in loco-regionally advanced head and neck cancer (HNC). We have previously shown that poor perfusion on DCE-MRI in HNC was associated with worse LRC and OS. A prospective randomized multi-center clinical trial was designed to evaluate the role of DCE-MRI based RT dose escalation in poor prognosis HNC. <h3>Materials/Methods</h3> Eligibility included 1) AJCC 8 stage III p16+ OPSCC or 2) T3-4 or N3 p16- oropharynx/non-oropharynx HNSCC planned for definitive chemoradiation, ECOG 0-2, with concurrent weekly cisplatin 40mg/m2 or carboplatin (AUC = 2) for cisplatin ineligibility. FDG-PET and DCE-MRI were acquired at baseline and at fraction 10. BV was derived from DCE-MRI < 0.076, and ADC was evaluated per previous report; low BV/ADC maps were created within the GTV. 79 patients with > 1cc low BV/ADC persistent tumor subvolume (GTVboost) at fraction 10 were randomized to total dose of 70Gy (arm A) or 80Gy (86Gy EQD2) total dose to GTVboost subvolume (arm B). The primary endpoint of DFS and secondary endpoints including LRC and OS were compared between treatment arms using stratified log-rank tests. Acute and late toxicities were compared with Chi-squared test. EORTC QLQ30/HN35 PROs, VFSS endpoints, HPV DNA kinetics were also analyzed. <h3>Results</h3> All patients included in this analysis (N = 91) completed definitive therapy with min 12 mo and mean 32 mo follow-up. Treatment arms were balanced with respect to sex, p16+ status, smoking status, ECOG status, and chemotherapy. Mean age was higher in arm B (<i>P</i> = 0.01). Disease sites included 62% p16(+) AJCC 8 stage 3 oropharynx cancer. Randomization was stratified by GTVtotal (≥ 56cc) and GTV boost (> 10cc). Mean GTVtotal was 91cc in p16+ OPSCC and 67 cc in p16- OPSCC/non-OPSCC. Mean GTVboost was 15cc. 12 pts (7 p16+ OPSCC) were not randomized due to < 1cc persistent boostable tumor subvolume and were treated on an observation arm with LRC 83% and OS of 83% at 2 years. Treatment outcomes for randomized patients listed in Table 1. In a subgroup analysis of p16+ OPSCC patients, LRC was improved in the boost arm 93% vs 73% (2-sided <i>P</i> = 0.088). All local failures occurred within the 70Gy isodose line. Grade 3+ toxicities occurred in 53% (arm A) vs 64% (arm B) (<i>P</i> = 0.33). Feeding tube (FT) rate at 3 months was 15% (arm A) vs 27% (arm B) (<i>P</i> = .17) with no FT at 12 months post RT. There were 2 early deaths in arm B related to lingual arterial bleeding, 1 at 3 months and 1 at 12 mo post RT. <h3>Conclusion</h3> LRC improvement was numerically but not significantly improved with DCE-MRI directed boost for all randomized patients and p16+ oropharynx subgroup; DFS was not improved. Boost volume < 1cc was associated with good prognosis. Grade 3+ toxicities and 3 mo FT use were non-significantly higher in Arm B. Table 1. Outcomes according to treatment arm for randomized patients.

Adaptive Chemoradiation Therapy for Head and Neck Cancer Based on Multiparametric MRI: Interim Results of a Prospective Randomized Trial

Tumor regions characterized by hypoperfusion (HP) and restricted diffusion (RD) by multiparametric MRI have been reported to harbor aggressive disease. We report interim results of a randomized trial assessing boost to these subvolumes in poor-prognosis head and neck cancer (HNC). Patients with HPV-positive, T4/N3 oropharyngeal cancer (OPC); HPV (-), T3-4/N2-N3 OPC; EBV (-), T3-4/N3 nasopharyngeal cancer; locally-advanced, bulky, or unresectable T3-4 laryngeal or hypopharyngeal cancer; or stage III-IV oral cavity or paranasal sinus cancer who refuse or are unfit for surgery undergo MRI at baseline and 2-weeks midtreatment. Patients with stable HP or RD subvolumes are randomized to boost to these areas to 80 Gy versus standard dose (70 Gy). Cox regression and Kaplan-Meier were used to compare rates of failure. CTCAE v4, and EORTC QLQ-C30 and H&N35 are used to assess toxicity and quality of life (QOL). At time of analysis, 31 of the planned 80 patients have completed treatment: 15 randomized to boost and 16 to standard therapy. At a median follow-up of 13.3 months, there were 7 (22.6%) locoregional failures (LRFs) in all patients: 5 (31.2%) in the 70-Gy arm and 2 (13.3%) in the 80-Gy arm (HR 0.36, P = .218 ). There were 6 (19.4%) distant failures (DFs): 4 (25%) in the 70-Gy arm and 2 (13.3%) in the 80-Gy arm (HR 0.46, P = .365). Rates of HPV positivity were 56.2% and 66.7% in the 70-Gy and 80-Gy arms, respectively (P = .567). In HPV (+) patients, there were 2 (10.5%) LRFs: 2 (22.2%) in the 70-Gy arm and 0 in the 80-Gy arm (HR 0.19, P = .211). There were 3 (15.8%) DFs: 2 (22.2%) in the 70-Gy arm and 1 (10%) in the 80-Gy arm (HR 0.41, P = .463). In HPV (-) patients, there were 5 (41.7%) LRFs: 3 (42.8%) in the 70-Gy arm and 2 (40%) in the 80-Gy arm (HR 0.60, P = .586). There were 3 (15.8%) DFs: 2 (22.2%) in the 70-Gy arm and 1 (20%) in the 80-Gy arm (HR 0.59, P = .670). On univariate analysis, HPV positivity (HR 0.22, P = .073), as well as pretreatment hypermetabolic (HFDG) (HR 1.57, P = .079) and union of HP, RD, and HFDG (HR 2.15, P =.046) volumes were significant or marginally significant correlates of LRF. There were no significant differences in acute toxicity between arms. “Lost-Weight” QOL scores were worse in the boost arm at baseline and at 4 weeks posttreatment, although the difference at 4 weeks was of lesser magnitude than at baseline. At 72 weeks, “Cough” scores were worse in the boost arm at 2 versus 1.2 on a 4-point scale (P = .010). In this interim analysis of MRI-guided boost in poor-prognosis HNC, there were not yet significant differences in outcomes between arms. However, the HR for LRF was encouraging as it was consistent with the value used to calculate power for the study. Boost was associated with no increased acute toxicity and minimal difference in QOL.

Abstract PR05: Adaptive chemo-radiotherapy for head and neck cancer based on multiparametric MRI: Interim results of a prospective randomized trial

Abstract Purpose/Objective(s): Tumor regions characterized by hypo-perfusion (HP) with dynamic contrast-enhanced (DCE)-MRI and restricted diffusion (RD) with diffusion-weighted (DW)-MRI have been reported to harbor aggressive disease. We report interim results of a randomized trial assessing boosting the doses to HP and RD sub-volumes in poor-prognosis head and neck cancer (HNC). Materials/Methods: Prospective randomized trial: Patients with either T4/N3 HPV(+) and heavy smoking history or T3-4/N3 HPV(-) HNC undergo PET-FDG and multiparametric MRI at baseline and at 2 weeks mid-treatment. HP is defined as DCE-MRI-measured fractional plasma volume &amp;lt;0.042. RD is defined as DW-MRI-measured apparent diffusion coefficient &amp;lt;1.2 x 10-3 mm2/sec. We characterize the HP and RD regions within the tumor pretreatment and after 2 weeks of chemoradiation. Patients with spatially stable HP or RD sub-volumes at both baseline and 2 weeks are randomized to boost to these sub-volumes to biologically equivalent 86 Gy, or to standard dose (70 Gy). Paired two-tailed t-tests were used to compare changes in sub-volumes, and univariate Cox-regression to correlate clinical and imaging variables to locoregional failure (LRF). Results: Of the planned 80, 36 patients were accrued to date: 15 randomized to receive boost and 15 to standard therapy. In addition, 6 patients did not retain high- risk sub-volumes at 2 weeks and were treated like the standard dose arm. HP sub-volume comprised a mean of 23% of the primary tumor at baseline and 15% at 2 weeks (p=0.042). The RD sub-volume comprised a mean of 41% of the primary tumor at baseline and 21% at 2 weeks (p&amp;lt;0.001). The average volume of overlap between the HP and RD sub-volumes comprised 8% of the primary tumor at baseline and 3% at 2 weeks (p=0.036). The sub-volumes planned for boost, comprising both stable HP and RD, averaged 22% (SD 11%) of the gross tumor volumes. At median follow-up 8.9 months, 31 patients were evaluable for clinical outcome: 13 on the boost arm and 18 on the standard dose arm. Of these, 2 (15%) and 4 (22%) patients experienced LRF in the boost and standard arms, respectively. All failures in the boost arm were HPV (-), while in the standard arm, failures occurred in both HPV (+) and (-) patients. No difference in grade &amp;gt; 3 acute or late toxicities was observed between arms. Mean RD subvolume at 2 weeks emerged as a near-statistically significant factor for LRF. updated clinical and radiological results will be presented. Conclusion: HP and RD sub-volumes are spatially distinct and only slightly overlap in advanced HNC, suggesting they represent different biological features of aggressive cancer and provide rationale to using both for radiation boost, as has been done in the current study. As this study continues, we will assess whether LRF arise preferentially in regions of HP or RD, and whether boosting these regions improves locoregional control. This abstract is also being presented as Poster 01. Citation Format: Avraham Eisbruch, Yue Cao, Peter Hawkins, Jae Lee, Choonik Lee, Madhava Aryal, Michelle Mierzwa, Matthew Spector, Frank Worden, Paul Swiecicki, Gregory Wolf. Adaptive chemo-radiotherapy for head and neck cancer based on multiparametric MRI: Interim results of a prospective randomized trial [abstract]. In: Proceedings of the AACR-AHNS Head and Neck Cancer Conference: Optimizing Survival and Quality of Life through Basic, Clinical, and Translational Research; April 23-25, 2017; San Diego, CA. Philadelphia (PA): AACR; Clin Cancer Res 2017;23(23_Suppl):Abstract nr PR05.

Phase I trial of pemetrexed in combination with cetuximab and concurrent radiotherapy in patients with head and neck cancer

BackgroundWe studied the combination of pemetrexed, a multi-targeted antifolate, and cetuximab, an mAb against the epidermal growth factor receptor, with radiotherapy in poor prognosis head and neck cancer. Patients and methodsPatients received pemetrexed on days 1, 22, and 43 on a dose-escalation scheme with starting level (0) 350 mg/m2 (level -1, 200 mg/m2; level +1, 500 mg/m2) with concurrent radiotherapy (2 Gy/day) and cetuximab in two separate cohorts, not previously irradiated (A) and previously irradiated (B), who received 70 and 60–66 Gy, respectively. Genetic polymorphisms of thymidylate synthase and methylenetetrahydrofolate reductase were evaluated. ResultsThirty-two patients were enrolled. The maximum tolerated dose of pemetrexed was 500 mg/m2 in cohort A and 350 mg/m2 in cohort B. Prophylactic antibiotics were required. In cohort A, two dose-limiting toxicities (DLTs) occurred (febrile neutropenia), one each at levels 0 and +1. In cohort B, two DLTs occurred at level +1 (febrile neutropenia; death from perforated duodenal ulcer and sepsis). Grade 3 mucositis was common. No association of gene polymorphisms with toxicity or efficacy was evident. ConclusionThe addition of pemetrexed 500 mg/m2 to cetuximab and radiotherapy is recommended for further study in not previously irradiated patients.

Phase I Study of Bevacizumab Added to Fluorouracil- and Hydroxyurea-Based Concomitant Chemoradiotherapy for Poor-Prognosis Head and Neck Cancer

We conducted a phase I dose escalation study to determine the maximum-tolerated dose (MTD) and dose-limiting toxicity (DLT) of bevacizumab, when added to the standard FHX (fluorouracil [FU], hydroxyurea [HU], radiation) chemoradiotherapy platform in poor-prognosis head and neck cancer (HNC) patients. Patients with recurrent, previously radiated or poor-prognosis, treatment-naive HNC were eligible. Treatment was repeated every 14 days for seven cycles: Bevacizumab was escalated 2.5 to 10 mg/kg, FU 600 to 800 mg/m(2) (120 hours continuous infusion), and hydroxyurea from 500 to 1,000 mg (twice daily for 5 days), starting day 1. At the MTD, the cohort was expanded. Forty-three patients were treated. DLT was reached at level 3 (bevacizumab 5 mg/kg, FU 800 mg/m(2), HU 1,000 mg) with two grade 3 transaminase elevations and one grade 4 neutropenia, attributed to the combination of chemotherapy with bevacizumab. For level 4, chemotherapy doses were reduced (FU 600 mg/(2), HU 500 mg), and bevacizumab escalation continued to 10 mg/kg. Treatment of six assessable patients resulted in one venous thrombosis; this dose level was expanded to 26 patients. Late complications included five patients with fistula formation (11.6%) and four with ulceration/tissue necrosis (9.3%). Serious toxicities (hemorrhage/thrombosis/death) were comparable to prior reirradiation reports. Median overall survival for reirradiated patients with recurrent, nonmetastatic disease was 10.3 months [95% CI, 5.6 to 13.5]; 2-year cumulative incidence of death resulting from disease was 51.7% (95% CI, 31.7 to 68.5). Bevacizumab can be integrated with FHX chemoradiotherapy at a dose of 10 mg/m(2) every 2 weeks with decreased chemotherapy doses because of neutropenia. The regimen shows antitumor activity. Observed fistula formation/tissue necrosis may be bevacizumab related, and further investigation should proceed with careful monitoring.

A phase I study of bevacizumab (B) with fluorouracil (F) and hydroxyurea (H) with concomitant radiotherapy (X) (B-FHX) for poor prognosis head and neck cancer (HNC)

5530 Background: Increased VEGF levels are found in HNC. Preclinical data suggest synergistic antitumor activity of B with radiation and chemotherapy. We conducted a Phase I dose escalation study to determine the maximum tolerated dose and dose limiting toxicity (DLT) of B, when added to infusional 5-FU, Hydroxyurea (HU), and daily radiotherapy administered every other week in patients (pts) with poor prognosis HNC. Methods: Eligible pts had recurrent, or newly diagnosed HNC with high risk of recurrence, ± metastatic disease requiring local control, ECOG PS ≤2, and life expectancy &gt;12 weeks. Two week cycles were repeated 6–7 times (see table ). Results: 43 pts were treated (34 completed). DLT was reached at level 3 with 2 pts having gr 3 transaminase elevations and one pt gr 4 neutropenia. Treatment of 7 (6 evaluable) pts on level 4 resulted in one DLT (SMV thrombosis) and this dose level was chosen for expanded evaluation. In all level 4 pts (N = 27) gr 3 mucositis occured in 73.1% and gr 3 hand-foot syndrome in 15.4%. Additional gr 3 or worse toxicities in the expanded level 4 included: esophageal bleed (tumor bed, grade 5), stroke (grade 4), carotid rupture (3 wks post RT, grade 5), and neck ulceration with need for carotid stent (3 months post RT, grade 4). One sudden death of unclear etiology occurred. Median overall survival is 389 days. One/two year survival is 52.1/26%. Median survival for patients treated with re-irradiation for recurrent, non-metastatic HNC is 314 days; one/two year survival is 45.9/17.2%. With a median follow-up of 317 days 13 patients are still alive (12 are cancer free). Conclusions: B can be integrated with FHX chemoradiotherapy at a dose of 10 mg/m2 every 2 weeks. While B related toxicities are seen, there appears to be no major synergistic toxicity. Long term activity is observed in this very high risk patient population. A randomized phase II trial of FHX with or without B in a lower risk population is ongoing. [Table: see text] No significant financial relationships to disclose.

A phase I–II study of concomitant chemoradiotherapy with paclitaxel (one-hour infusion), 5-fluorouracil and hydroxyurea with granulocyte colony stimulating factor support for patients with poor prognosis head and neck cancer

Concomitant chemoradiotherapy is an effective treatment modality for advanced head and neck cancer, but improved regimens are needed. We sought to define the toxicities, recommended phase II dose, and outcome of a combination chemotherapy regimen with concomitant hyperfractionated radiotherapy in patients with poor prognosis cancers of the head and neck, including those having received prior curative intent radiotherapy. From 1995 until 1997, 54 patients were treated, 25 of whom had received a prior full course of radiotherapy to the head and neck. Patients were treated with 5-fluorouracil (5-FU) 600 mg/m2/day continuous infusion x 5 days (days 1-5), hydroxyurea, 500 mg p.o. bid x 11 doses (days 1-6) and paclitaxel (60-150 mg/m2) by one-hour infusion on day 2 using a dose escalation strategy. Radiotherapy was given concomitantly on days 2-6, 150 cGy bid. Each of 4-5 cycles was delivered every other week. The MTD of paclitaxel was 100 mg/m2. The regimen was feasible; radiotherapy was delivered at a median of 7300 cGy and 83% of patients received > or = 80% planned dose intensity. Hematological toxicity, with granulocyte colony stimulating factor, was very mild. Dose limiting toxicities were mucositis and dermatitis. Despite poor prognosis, two-year survival was 45%. The recommended phase II dose of this regimen is 5-FU 600 mg/m2/day x 120 hours (days 1-5), hydroxyurea 500 mg p.o. b.i.d. x 11 doses (days 1-6), paclitaxel 100 mg/m2 over one hour on day 2, and radiotherapy 150 cGy b.i.d. days 2-6. Concomitant chemotherapy and re-irradiation was feasible on this protocol and resulted in long-term survival in patients without other curative intent options.

Intensified concomitant chemoradiotherapy for poor-prognosis head and neck cancer.

Head and neck cancer presents most commonly as a locoregionally advanced malignant process [1]. Distant metastases are clinically rare, but may be present more frequently at the microscopic level. Patterns of failure following standard surgery and/or radiotherapy indicate the inability of that therapy to successfully eradicate locoregional disease in the majority of patients. Investigational therapy, therefore, must focus on increasing the efficacy of locoregional therapy as its primary goal. Eradication of systemic microscopic disease remains a secondary goal that may gain greater importance in patients whose locoregional disease can be controlled [2]. Finally, the prevention of second malignancies represents an important long-term goal that can help to maintain the cancer-free status in patients cured of their primary disease [3].

Treatment of poor-prognosis head and neck cancer (HNC) with alpha-interferon (a-IFN) and concomitant radiotherapy

Treatment of poor-prognosis head and neck cancer (HNC) with alpha-interferon (a-IFN) and concomitant radiotherapy

Survival and Analysis of Failure Following Hydroxyurea, 5-Fluorouracil and Concomitant Radiation Therapy in Poor Prognosis Head and Neck Cancer

Thirty-nine patients with head and neck cancer were entered into Phase I-II study of simultaneous radiation therapy with continuous infusion fluorouracil at 800 mg/m2/day and escalating doses of hydroxyurea. Twenty of these patients had recurrent disease after previous surgery and/or radiation therapy (group 1). Nineteen patients had not received prior local therapy but had advanced-stage disease (group 2). Cycles were repeated every other week until the completion of radiation therapy. The median follow-up was 32 months. Patients with recurrent disease were generally treated with palliative doses of radiation (median dose 5,000 cGy) while previously untreated patients received radiation with curative intent (median dose 7,040 cGy). The response rate for 15 evaluable patients with recurrent disease was 93% with 40% of patients achieving a complete response. For 17 evaluable patients without recurrent disease the response rate was 100%, with a complete response rate of 71%. This regimen exhibited a high activity and significant palliative benefit in group 1 patients. However the local control rate was 25% (5/20) because the majority of patients in this group eventually developed a local recurrence. The local control rate for group 2 patients was 84% (16/19). The higher local failure rate in group 1 patients appeared to be attributable to the palliative doses of radiation delivered and the fewer cycles of treatment received. We conclude that this regimen has palliative benefit in patients who have failed prior local therapy and has the potential for cure in patients with poor prognosis advanced stage disease as well.

Hydroxyurea, fluorouracil, and concomitant radiotherapy in poor-prognosis head and neck cancer: a phase I-II study.

Hydroxyurea and fluorouracil (5-FU) are active cytotoxic drugs in head and neck cancer and have shown synergistic activity in vitro. Both drugs also act as radiosensitizers. Therefore, we administered radiotherapy at daily fractions of 180 to 200 cGy with simultaneous continuous infusion 5-FU at 800 mg/m2/d and escalating daily doses of hydroxyurea for five days. Cycles were repeated every other week until completion of radiotherapy. Thirty-nine inoperable patients were treated at six dose levels of hydroxyurea ranging from 500 mg to 3,000 mg orally daily. Little effect of hydroxyurea on the WBC or platelet count was noted in patients receiving less than 2,000 mg daily, whereas both parameters decreased progressively in patients receiving 2,000 mg daily or more. Mucositis occurred at all dose levels, requiring frequent dose reduction of 5-FU; however, in patients receiving a daily hydroxyurea dose of 2,000 mg or less, the median weekly 5-FU dose administered was 1,725 mg/m2 (86% of the intended 5-FU dose), whereas at daily hydroxyurea doses exceeding 2,000 mg, the median weekly 5-FU dose decreased to 1,133 mg/m2 (57%) (P = .001). Of 15 evaluable patients with recurrent disease after prior local therapy only one failed to respond; six had a complete response (CR), and eight a partial response (PR). Of 17 evaluable patients without prior local therapy, 12 had a CR, with no patient developing recurrence in the irradiated field to date; five patients had a PR. We conclude that the recommended dose of hydroxyurea in this regimen is 2,000 mg daily. That dose will cause mild to moderate myelosuppression and will allow for delivery of greater than 80% of the intended 5-FU dose. The activity of this regimen in poor-prognosis head and neck cancer exceeds 90%; its further investigation in previously untreated patients is warranted.