Patterns of failure after postoperative radiotherapy for incompletely resected (R1) non-small cell lung cancer: Implications for radiation target volume design

Abstract Objective Overall survival (OS) and pattern of failure in R1-resected non-small cell lung cancer (NSCLC) patients treated with 3D-planned postoperative radiotherapy (PORT) was retrospectively evaluated. The outcomes and patterns of failure in patients with (+) and without (−) extracapsular nodal extension (ECE) were compared and analyzed with respect to the radiation target volume design. Materials and methods Eighty R1-resected (37 ECE+ and 43 ECE−) patients received PORT (60 Gy, 2 Gy daily) between 2002 and 2011. Patients with N2 disease received limited elective nodal irradiation (ENI); for pN0-1 disease the use of ENI was optional. Among ECE− (extranodal-R1) patients there were 35 pN0-1 and eight pN2 cases; in pN0-1 patients, patterns of failure and outcomes were analyzed with respect to the use of ENI. Loco-regional failure (LRF) was defined as in-field relapse; isolated nodal failure (INF) was defined as out-of-field regional nodal recurrence occurring without LRF, irrespective of distant metastases. Results The actuarial 3-year OS rate was 36.3% (median: 30 months). Three-year OS rates in the ECE− and ECE+ group were 40.4% and 31.4%, with median OS of 31 and 24 months, respectively ( p = 0.43). In multivariate analysis, the presence of ECE was correlated with OS (HR = 3.02; 95% CI: 1.00–9.16; p = 0.05). Three-year cumulative incidence of LRF (CILRF) was 14.5% and 15.5% in the ECE− and ECE+ groups, respectively ( p = 0.98). Three-year cumulative incidence of INF (CIINF) was 14.1% in the ECE− group and 11.1% in the ECE+ group ( p = 0.76). For pN0-1 patients treated with and without ENI (13 and 22 patients) 3-year CILRF rates were 7.7% and 20.8%, respectively ( p = 0.20); 3-year CIINF rates were 9.1% and 16.3%, respectively ( p = 0.65). Conclusion PORT resulted in a relatively good survival of R1-resected NSCLC patients. Relatively high incidence of INF was found in both ECE+ and ECE− patients. For ECE+ patients, treated with limited ENI, distant failure remains a major concern, so the design of ENI fields seems of lesser importance. Omission of ENI in pN0-1 (extranodal-R1) patients resulted in an unacceptably high incidence of INF. We postulate the use of some form of ENI in this setting.