Dosimetric parameters predictive for esophageal injury

Purpose/Objective: To evaluate the 3D dosimetric parameters predicting for clinically significant esophageal injury after conformal radiotherapy of non-small cell lung cancer (NSCLC). Materials/Methods: The records of 254 patients treated with 3D conformal radiotherapy at Duke University for NSCLC between 1992 and 2001 were reviewed. Age range 26–87 years, median 64; men 143, women 111. Stage distribution; 36 stage I, 24 II, 102 IIIA, 86 IIIB, and 6 unknown. Radiation delivered dose to the isocenter was ranged from 30 Gy to 86.4 Gy with the median 66 Gy. The 3D dose distribution was reviewed and a variety of metrics describing the esophageal dose were extracted, including dose volume histogram (DVH), dose circumferential histogram (DCH) on the esophagus. All doses reflect tissue density heterogeneity. Chemotherapy was given in 143 patients (56%), concurrently in 32 and sequentially 111 patients. The RTOG toxicity criteria for grading of esophageal injury were used. Grade >2 and >3 acute and late esophageal toxicities were identified. Seventeen patients who died within three months after radiotherapy were excluded from the analysis of late esophageal injury. Logistic regression, contingency table analyses and Fisher’s exact tests were used to assess the relationship between various parameters (dosimetric, treatment, patient characteristics) and the incidence of esophageal toxicity. The median follow-up time of all patients was 43 months with the range of 0.5–120 months. Results: Acute esophageal toxicity occurred in 78% patients (199/254); 138 grade 1, 38 grade 2, 22 grade 3, and 1 grade 4. On univariate analysis, BID-RT, pre-RT dysphagia, nodal stage (dichotomous ≤1 vs ≥2), maximal esophageal point dose, the maximum percent of the circumference receiving >50–70 Gy, and the length of esophagus where ≥75% of the circumference received >50–70 Gy, and the percent esophageal volume receiving >50–70 Gy were predictive for the development of ≥grade 3 acute esophagitis. On multivariate analysis, the clinical factors of BID (vs. QD RT), and N-stage were dominant predictors. Of 238 patients evaluable for late esophageal toxicity, it occurred in 7% patients (17/238); 5 grade 1, 4 grade 2, 5 grade 3, and 3 grade 4. Late toxicity occurred in 2%, 3%, 16%, 23%, and 100% of patients with acute grade 0, 1, 2, 3 and 4 toxicity, respectively. The severity of acute esophagitis is significantly related to the occurrence of late esophageal injury (p<0.0001) with a greater than 3 fold increase in the probability of development with each increase in grade. Late toxicity appeared 3–40 months after completion of radiotherapy with the median 5 months. On univariate analysis, the development of severe late symptoms, the percent esophageal volume receiving >70–80 Gy, the length of esophagus where 75% of the circumference received >55–70 Gy, the length of esophagus where the full circumference received >50 Gy, and the maximum percent of the circumference receiving >70–80 Gy were predictive. On multivariavte analysis, the severity of acute symptoms appeared to be a more powerful predictor than the dosimetric parameters. Conclusions: Dosimetric and clinical parameters are predictive for the development of acute and late esophageal injury. For both, the clinical parameters are more predictive than are the dosimetric parameters. The development of acute esophagitis is most related to the use of BID (vs QD) RT, and the nodal stage; but BID treatment typically resulted in higher doses. The latter is likely a surrogate for target proximity to the esophagus. For late injury, the severity of the acute injury was most predictive. Additional studies to better define predictors of RT-induced esophageal injury are needed. We gratefully acknowledge the University of North Carolina for the PLUNC treatment planning software.