AN ANALYSIS OF FACTORS INFLUENCING THE OUTCOME OF POSTOPERATIVE IRRADIATION FOR SQUAMOUS-CELL CARCINOMA OF THE ORAL CAVITY

Purpose: To analyze factors influencing outcome in patients who receiv ed postoperative irradiation for advanced squamous cell carcinoma of t he oral cavity. Methods and Materials: Between October 1964 and Novemb er 1993, 134 patients with 135 previously untreated primary invasive s quamous cell carcinomas of the oral cavity (excluding the lip) were tr eated postoperatively with continuous courses of external-beam irradia tion at the University of Florida. All patients had a minimum follow-u p of 2 years (analysis, December 1995). No patient was lost to follow- up. Results: The 10-year actuarial rates of primary site, neck, and lo cal-regional control were 79%, 88%, and 71%, respectively. Recurrence of cancer above the clavicles developed in 35 patients. Ninety-four pe rcent of the recurrences were within the primary field of irradiation (anterior to the plane of the spinal cord); there were 24 recurrences at the primary site and nine in the upper neck alone. There were no fa ilures in the neck area behind the plane of the spinal cord (i.e., the ''posterior strip''). Two failures occurred in the low neck below the level of the thyroid notch. In univariate analyses, factors that affe cted local-regional control included pathologic stage (I-II vs. III-IV , p = 0.04), margin status (invasive cancer at the margin vs. other,p = 0.0007),multifocal tumor(p = 0.05), perineural invasion (p = 0.04), and number of indications for postoperative irradiation (p = 0.05). Ex tracapsular nodal extension was marginally significant (p = 0.07). In multivariate analysis, positive margins and number of indications rema ined significant. These factors were used to define relatively favorab le (< 4 indications, margins not positive) and unfavorable (greater th an or equal to 4 indications and/or margins positive for invasive canc er) groups. For both favorable and unfavorable groups, there were nons ignificant trends toward improved local-regional control for patients who began irradiation within 45-50 days, compared with those whose irr adiation began later. There were also nonsignificant trends toward imp roved control for patients treated with shorter overall irradiation tr eatment courses. An analysis was also performed on the effects of dura tion of the overall ''treatment package'' (from the date of surgery un til the lest day of irradiation). For patients with unfavorable tumors , there was a significantly higher probability of local-regional contr ol for patients whose overall ''treatment package'' was less than or e qual to 100 days (60% vs. 14%,p = 0.04). The 5-year rate of distant me tastasis as the sole site of failure was 8% and was predicted by patho logic N stage (NO-NI, 3%; N2-N3, 16%,p = 0.02), as well as the presenc e (20%) or absence (6%) of extracapsular nodal extension (p = 0.06). T he 5-year freedom-from-relapse rate was 63%. The 5-year survival and c ause-specific survival rates were 50% and 67%, respectively. Four seve re radiation injuries occurred (3%). Conclusion: This paper provides d ata that define relatively favorable and unfavorable groups of patient s in the postoperative setting. Patients with four or more indications for irradiation and/or invasive cancer at the surgical margins have a worse outcome than patients who do not have these negative factors; t his is true in spite of the fact that the unfavorable group received h igher doses of radiation. Attention should be focused on not only the interval between surgery and irradiation, but also time-dose parameter s and the overall duration of the treatment ''package.'' (C) 1997 Else vier Science Inc.