Factors predictive of local disease control after intra-arterial concomitant chemoradiation (RADPLAT).

Abstract

To determine the relative risk of prognostic factors for local disease control following RADPLAT. Prospective study, academic medical center. Analyses of nine categories of risk factors among 240 patients with Stage II-IV carcinoma consecutively treated with RADPLAT (cisplatin 150 mg/m IA and sodium thiosulfate 9 g/m IV, weekly x4; radiotherapy 2 Gy/fraction/d, 5x weekly, 68-74 Gy over 6 to 7 weeks). Median follow-up: 58 months (range, 12-96 mo). The percentage of patients who had local disease control was 87.5%. Univariant analysis showed T classification (P =.01), laterality of neck disease (P =.026), number of neck levels involved (P =.008), total dose of radiation greater versus less than 60 Gy (P =.027), and presence of pathologically positive lymph nodes following chemoradiation (P =.01) to be significant. Logistic regression analysis showed total dose of radiation (P =.03) and the presence of pathologically positive lymph nodes following chemoradiation (P =.05) to be significant. For Kaplan-Meier estimates of local disease control at 5 years, T classification (P =.038), number of levels with nodal disease (P =.006), and total dose of radiation therapy (P =.0001) were significant. The log-rank test identified as significant the total dose of radiation therapy (P <.0001), the presence of pathologically positive lymph nodes following chemoradiation (P =.005), and the number of neck levels with positive nodes (P =.006). The Cox regression model showed significance for the total dose of radiation (P =.001), the presence of pathologically positive lymph nodes following chemoradiation (P =.007), and the T classification (P =.029). Risk factors significantly associated with local disease control after RADPLAT appears to differ from more traditional therapy and is suggestive of a paradigm shift.