"With further follow-up, the significant difference in progression-free survival is still very solid. There is no weakening of the strength of the data at this point," said Dr. Albain, of Loyola University of Chicago’s Cardinal Bernardin Cancer Center.
Surgical resection has yielded encouraging results in patients with stage IIIA NSCLC and pathologically confirmed N2 nodes (pN2), but due to increased mortality and morbidity associated with surgery, CT/RT is the standard of care. Intergroup Trial 1039 was designed to test the value of resection after induction CT/RT vs full-course CT/RT therapy, in 429 patients with T1-3, pN2, M0 tumors.
All patients had induction chemoradiation with two cycles of cisplatin(Drug information on cisplatin) (Platinol) and etoposide(Drug information on etoposide) and daily radiation to 45 Gy, then were randomized to either surgical resection (surgery arm) if disease progression had not occurred or uninterrupted radiotherapy to 61 Gy (CT/RT arm), each followed by two more courses of chemotherapy.
The median follow-up in this report was 71 months, which represented an additional 8 months beyond the 63-month data reported at ASCO. The analysis yielded updated disease-free and overall survival, a new Cox multivariate model, and revised patterns of failure in 392 evaluable patients (n = 201 surgery arm; n = 191 CT/RT arm).
Induction chemotherapy was delivered per protocol in 95% to 96% of patients in both arms. Significantly more patients in the surgery arm did not receive consolidation chemotherapy (42% vs 21%), but RT was delivered per protocol in significantly more patients in the surgery arm (97% vs 81%).
Toxicities were similar, although more patients died on the surgery arm. The only difference in grade 3-4 toxicity was more esophagitis in the CT/RT arm (20% vs 9%). Three deaths occurred on the CT/RT arm during or after consolidation. In the surgery arm, 14 patients died (10 of postoperative complications). Twelve deaths followed pneumonectomy, 8 of which were from acute respiratory distress syndrome.
Updated Survival Analysis
Progression-free survival (PFS) was significantly better for patients in the surgical arm (P = .03). Median PFS was 13.4 vs 11.8 months, and 3-year PFS was estimated at 28% vs 19% for surgery vs CT/RT, respectively. Median overall survival was 22 months vs 22.3 months, and 3-year survival was 37% and 34%, respectively, Dr. Albain reported. "The hazard ratios crossed due to treatment-related deaths for both PFS and overall survival, and the log-rank P value was not significant," she said.
While early survival favors the CT/RT arm, the curves superimpose, then cross, and begin to separate at the median survival point of 22 months, she said. By year 3, there is an absolute survival benefit of 3% favoring the surgery arm, for a 10% proportional risk reduction. However, confidence intervals are wide and overlap at this point of follow-up. In the surgical arm, more patients are alive without progression (P = .004), and more died without progression (P = .007).
"Very little change has occurred in the outcome endpoints since the ASCO 2003 analysis," she noted. "Progression-free survival remains significant in favor of surgery, with no difference in overall survival. The absolute survival improvement with surgery at 3 years is still small. There is no change in the greater number of patients alive without progression with surgery, and in the fewer number who died without progression after CT/RT."
Patterns of Failure
Overall patterns of failure were still similar between arms. There was a trend for greater local relapse with CT/RT in the primary site, nodes, and both (17% with surgery vs 26% with CT/RT; P = .08). The updated Cox multivariate model for overall survival found the significant independent predictors of poor outcome to be higher T stage, weight loss less than 5%, and older age. Performance status and tumor size did not prove predictive.
In the analysis of overall survival according to interim
pathologic response data, N0 status continued to predict improved outcome
whether or not there was residual primary tumor. Patients who were T0, N0
(pathological complete response) had a median survival of 36.7 months and a
3-year survival of 52%. Patients who were T0-1, N0 had a median survival of
36.7 months and a 53%
"Progression-free survival is superior for the trimodality arm with an apparent plateau, but overall survival does not appear to be statistically different. It is uncertain whether the plateau in median survival will translate into improved overall survival with longer follow-up," Dr. Albain concluded.
Discussant Harvey Pass, MD, professor of surgery and oncology, Karmanos Cancer Institute, Detroit, called Intergroup 0139 "the most anticipated randomized trial in lung cancer for a long time. It is a standard for randomized trials in induction therapy and serves as a springboard for trials we need to plan for the future."
He said that the study confirmed phase II trials showing that full doses of cisplatin/etoposide could be delivered with radiotherapy and that pathologic complete responses could be achieved in 18%. "The most intriguing part of the trial is that downstaging of N2 disease to N0 will occur in 46% to 48% of patients, and 3-year survival of this group is a phenomenal 53%, which is even better than the 44% we achieved in SWOG 8805," he said. [In SWOG 8805, 126 patients with pathologically staged IIIA/IIIB disease received cisplatin plus etoposide concurrent with radiation therapy followed by surgical resection.]
Dr. Pass said that results in the surgery arm "re-amplified the importance of nodal clearance on survival." The findings do not demonstrate that "everybody should get surgery after induction," but point to the need to determine which patients will have nodal downstaging by any induction method. To accomplish this, he said, pre- and postinduction PET scanning should be used in future trials.
The study still leaves unanswered several questions, such as whether the N2 downstaging is due to the radiotherapy, the chemotherapy, or the combination, and what contribution surgery really is making to the outcome. These questions are difficult because of the heterogeneity of the N2 population, and future studies should stratify these patients, he said.