It is important to define subsets of patients who are at high risk for recurrence after radiotherapy for head and neck cancer so that therapy may be modified in an attempt to improve the likelihood of cure. Various tumor-related parameters that correlate with the probability of local control after radiotherapy include primary tumor stage (T stage) [1,2], tumor volume [3-5] primary tumor site [2,6-10], tumor morphology (infiltrative vs exophytic), intrinsic radiosensitivity , and potential doubling time [12-15]. In addition, time-dose-fractionation parameters have been shown to influence tumor control rates in patients treated with radiotherapy [16-18]. Extent of tumor oxygenation [19,20] and hemoglobin level during  or before treatment also may be important.
In 1985, Wall et al  first suggested that neck stage (N stage) may be indicative of the local aggressiveness of the primary cancer, and therefore, may be yet another factor that influences local tumor control. Before that time, most practitioners believed that the likelihood of control in the neck and at the primary site were independent of each other and were related to various pathologic and clinical factors at each site, as well as the treatment administered to each site. The question is of obvious importance because if a relationship between N stage and local control does exist, it may signal a need for treatment modification (eg, altered fractionation, concurrent chemotherapy) or, in some instances, a change in therapeutic modality (eg, surgery instead of radiotherapy).
The notion that there might be a dependent relationship between local control and N stage was controversial in 1985, and this issue has not been resolved during the last 10 years. Data from various institutions on this topic are conflicting. This paper reviews the literature pertinent to this question.
M. D. Anderson Data
Wall et al  reported on a series of 248 patients with squamous cell carcinoma of the supraglottis treated with radiotherapy alone for curative intent at the M. D. Anderson Cancer Center (Houston, Texas) between 1960 and 1980. The authors grouped primary lesions according to T stage (T1 and T2 vs T3 and T4) and analyzed the influence of N stage, histologic grade, and hemoglobin levels before and during radiotherapy on the probability of local control. Neck stage was stratified into 10 groups according to the number, size, and laterality of clinically positive nodes.
Patients with T1-T2 tumors had a lower local control rate in the presence of any positive node(s), as compared with patients with a clinically negative neck (P = .06). For patients with T3-T4 cancers, a single positive node more than 3 cm in diameter did not adversely influence local control, in contrast to the presence of more advanced neck disease (P =. 03). The authors also analyzed the data using the Cox proportional hazards test and noted an inverse relationship between N stage and local control, which was pronounced for T3-T4 lesions compared to T1-T2 tumors. There was no significant relationship between local control and histologic grade or hemoglobin levels.
Patterns of Fractionation Study
Withers et al  analyzed 676 patients treated in nine institutions with external-beam radiotherapy alone for primary squamous cell carcinoma of the tonsillar fossa between 1976 and 1985. The nine institutions included in the patterns of fractionation study were the Princess Margaret Hospital (Toronto, Canada), Massachusetts General Hospital (Boston), M.D. Anderson Cancer Center (Houston), Christie Hospital (Manchester, United Kingdom), Clatterbridge Hospital (Liverpool, United Kingdom), University of Florida (Gainesville), Royal Marsden Hospital (London), Mount Vernon Hospital (London), and Portsmouth Hospital (Portsmouth, United Kingdom). A multivariate analysis using the Cox proportional hazards model found that T stage, N stage, total dose, and overall treatment duration significantly (P <.05) influenced local control.
University of Florida Studies
Freeman et al  analyzed 607 patients treated for squamous cell carcinoma of the oropharynx, hypopharynx, or supraglottic larynx with continuous-course radiotherapy alone or followed by planned neck dissection at the University of Florida between 1964 and 1988. Patients with simultaneous primary lesions were excluded. All patients were followed for at least 2 years, and no patients were lost to follow-up. Parameters tested in a multivariate analysis of local control included T stage (T1, T2, T3, and T4), N stage (N0, N1, N2A-N3A, and N2B-N3B), primary site (oropharynx, hypopharynx, and supraglottic larynx), and fractionation scheme (once- and twice-daily).
Parameters that significantly influenced local control were T stage (P < .01), fractionation (P < .01), and primary site (P = .01). Local control rates were improved after twice-daily fractionation compared with once-daily irradiation, and patients with hypopharyngeal primary lesions had lower local control rates than those with cancer in the oropharynx and supraglottic larynx. Neck stage had no apparent impact on local control (P = .97).
A separate multivariate analysis also was performed for each mucosal site, (oropharynx, hypopharynx, and supraglottic larynx). Once again, N stage did not appear to significantly influence local control in any of these subgroups.
Lee et al  evaluated 243 patients treated between 1964 and 1990 with external-beam irradiation alone or combined with an interstitial implant for tonsillar cancer at the University of Florida. Patients treated before 1989 were included in the analysis of Freeman et al . Local control rates were examined using multivariate analysis; parameters included in the analysis were anterior extension, T stage, anterior tonsillar pillar primary, nasopharynx extension, fractionation schedule, palate extension, tongue extension, mid-line extension, brachytherapy boost, and neck stage. Anterior extension (P = .0001) and T stage (P = .0144) significantly influenced local control, whereas N stage had the least impact (P = .9415) on this end point.
Mendenhall et al  updated the University of Florida experience with radiotherapy alone for squamous cell carcinoma of the supraglottic larynx between 1964 and 1992. Patients treated before 1989 were included in the analysis by Freeman et al . Multivariate analysis revealed the following effects on local control: T stage (P = .0001), vocal cord mobility (P = .0373), N stage (P = .0768), fractionation schedule (P = .3532), suitability for conservation surgery (P = .3532), sex (P = .5405), primary site (P = .6340), and pretreatment CT scan (P = .7299). There was a trend toward decreasing local control with increasing N stage.
Pernot et al  described a series of 361 patients treated with brachytherapy alone (18 patients) or combined with external-beam radiotherapy (343 patients) for squamous cell carcinoma of the tonsillar region, soft palate, and glossotonsillar sulcus at the Centre Alexis Vautrin (Nancy, France) between 1977 and 1991. Parameters included in a multivariate analysis of local control were T stage, N stage, primary site, safety margin of the interstitial implant, overall treatment time, and interval between external-beam radiotherapy and brachytherapy.
Primary tumor stage (P = .0001), overall treatment time (P less than .0001), and primary site (P < .0001) significantly influenced local control. Patients completing treatment in less than 55 days had improved local control, as did those who had primary lesions of the tonsillar fossa, soft palate, and posterior tonsillar pillar, when compared with those with tumors of the anterior tonsillar pillar and glossotonsillar sulcus. Neck stage did not significantly affect local control.
Jaulerry et al  analyzed 166 patients treated with external-beam radiotherapy for squamous cell carcinoma of the base of the tongue at the Institut Curie (Paris) between 1960 and 1980. Stage of the primary tumor, regression of the primary tumor during radiotherapy, N stage (N0 vs N1 vs N2-N3), and histologic grade (well-differentiated vs moderately or poorly differentiated) were evaluated in a multivariate analysis. Primary tumor stage and regression of the primary lesion significantly influenced local control, whereas N stage and histologic grade did not.
Jaulerry et al  also reported on 228 patients with squamous cell carcinoma of the head and neck (oral cavity, 59 patients; oropharynx, 65 patients; hypopharynx, 37 patients; larynx, 67 patients) treated with radiotherapy alone at the Institut Curie between 1986 and 1990. Univariate analysis revealed that local control was significantly influenced by primary site (P < .009) and T stage (P< .0001) but not by N stage (P = .26). In a multivariate analysis, local control was significantly related to primary site (P = .002), T stage (P < .0001), and tumor regression at a radiation dose of 55 Gy (P = .0009).
Data from Hong Kong
Teo et al  evaluated 659 patients with nasopharyngeal carcinoma treated with radiotherapy at the Prince of Wales Hospital (Hong Kong) between 1984 and 1987. Multivariate analysis revealed that tumor confined to the nasopharynx was associated with improved local control, whereas cranial nerve II-VIII palsies and male gender were related to an increased risk of local recurrence. Extent of neck disease did not significantly influence local control.
Lee et al reported on 1,008 patients treated with radiotherapy alone for nasopharyngeal squamous cell carcinoma between 1976 and 1985 at the Queen Elizabeth Hospital (Hong Kong) . Multivariate analysis of local control revealed the following relationships: age (P = .33), sex (P = .60), N stage (P = .5), total dose (P < .01), dose per fraction (P = .21), and overall treatment time (P = .17). Neck stage did not affect the likelihood of local failure.
Medical College of Virginia-Tufts Study
Johnson et al analyzed 76 patients treated with "concomitant boost accelerated superfractionated irradiation" for advanced squamous cell carcinoma of the head and neck at the Medical College of Virginia (Richmond) and Tufts-New England Medical Center (Boston) . Tumor volume at the primary site (TV) and in the neck nodes (NV) was estimated for each patient based on pretreatment CT. Total tumor volume was defined as the sum of TV and NV. Univariate analysis of local control revealed the following: total tumor volume (P = .0001), T stage (P = .03), N stage (P = .009), sex (P = .10), age (P = .50), and primary site (P = .0004). Multivariate analysis demonstrated that total tumor volume (P = .0001), primary site ( P= .008), and sex (P = .03) significantly influenced local control, whereas T stage, N stage, stage group, and age had no statistically significant impact on this end point.