Lung cancer is the leading cause of cancer mortality in the United States. Over 172,000 people were diagnosed with lung cancer in 2005, and most will die of their disease. Non-small-cell lung cancer (NSCLC) accounts for over 80% of cases. Unfortunately, only a minority of patients present with early-stage disease, and the long-term survival of patients with more advanced disease is poor.
Stage III NSCLC encompasses a heterogeneous group of patients, which is reflected in the array of accepted treatment alternatives. These include definitive chemoradiotherapy, preoperative chemotherapy (with or without postoperative RT), and preoperative chemoradiotherapy. The literature does not clearly support one approach over the others. The choice of therapy is often dictated by the extent of local/regional disease, patient performance status, medical comorbidities, and patient/physician preference. In general, patients with bulky, unresectable disease or major comorbidities are managed with definitive chemoradiotherapy, whereas treatment involving surgical resection is offered to patients with more favorable disease.
The pattern of care for patients with stage III NSCLC is evolving. Prior to the 1990s, patients with mediastinal disease were generally considered unresectable and were treated with radiation therapy (RT) alone. This strategy yielded a median survival of ~10 months and a disappointing 2-year survival of ~15%, with few long-term survivors. The addition of systemic chemotherapy—initially sequential (Table 1)[2-4] and more recently concurrent (Table 2)[5-8]—was shown in multiple randomized trials to modestly improve survival over RT alone. Even with this advancement, 5-year survival remains less than 20% and local failure is common. This has rekindled interest in surgical resection of select patients with stage III (primarily N2) disease. However, 5-year survival after resection alone for patients with radiographic evidence of N2 disease is < 10%. Therefore, most patients with stage III disease are treated with a multimodality approach of chemoradiotherapy or surgery in combination with neoadjuvant/adjuvant chemotherapy and/or radiotherapy.
Several recent studies, largely involving patients with stage IB/II disease, have shown that adjuvant chemotherapy after resection confers a survival benefit over surgery alone (Table 3).[10-13] The only modern positive trial that enrolled stage III patients is the International Adjuvant Lung Cancer Trial (IALT). Among 1,867 patients enrolled in the trial, 479 had N2 disease. Subgroup analysis demonstrated that the hazard ratio for death (death rate with chemotherapy vs without chemotherapy) was lower for patients with stage III disease compared with stage I/II disease, suggesting that the value of chemotherapy is greater for more advanced stages.
However, other large randomized trials utilizing cisplatin(Drug information on cisplatin)-based regimens, also enrolling stage III patients, have failed to demonstrate a benefit for adjuvant chemotherapy.[14-16] Thus, while it seems clear that adjuvant chemotherapy improves survival in stage IB/II disease, its value in stage III disease is less certain but probably does confer a 5% to 10% survival advantage.
The administration of chemotherapy prior to surgery (induction therapy) is being used more frequently in patients with stage III lung cancer. This strategy is attractive for several reasons. In patients with advanced local/regional disease, initial chemotherapy may provide tumor regression that facilitates resection. For patients with potentially resectable disease, induction therapy is felt to be better tolerated with higher compliance than adjuvant chemotherapy after thoracotomy. Furthermore, surgical evaluation of the mediastinum after chemotherapy facilitates pathologic assessment of the response to chemotherapy, which may be prognostic and guide future systemic therapies. The local/regional response to induction therapy may also be a surrogate marker for the control of systemic micrometastases. Finally, induction chemotherapy allows for the early treatment of systemic disease.
Three small studies reported in the early 1990s,[17-19] each requiring pathologic confirmation of N2 disease, demonstrated that induction chemotherapy followed by resection yielded superior outcomes to surgery alone (Table 4). These studies, as well as the positive findings in the adjuvant setting for patients with stage I/II disease, are often cited to support chemotherapy plus surgery for stage III patients.
However, more recently reported, larger randomized studies have failed to demonstrate a benefit to induction chemotherapy for stage III NSCLC.[20,21] In the largest such study, conducted by the French Thoracic Cooperative Group, 355 patients (167 with IIIA disease) were randomized to induction and postoperative cisplatin-based chemotherapy vs surgery alone. No difference in overall survival was noted in the subset of patients with stage III disease (relative risk = 1.04, P = .85). Pathologic confirmation of mediastinal involvement was not required in that study. The Japan Clinical Oncology Group conducted a small trial in which 62 patients with pathologic evidence of N2 disease were randomized to surgery alone vs induction cisplatin-based chemotherapy. These investigators found no difference in median survival (16 vs 17 months) or 5-year survival (22% vs 10%, P = .53) between the two arms (Table 4).
Several points about these trials are worth mentioning. First, although the improvements in survival noted in the Roth and Rosell studies are provocative, only 60 patients were enrolled in each study. This increases the probability that prognostic factors were not equally balanced between the two groups. This may explain the unusually poor survival seen after surgical resection alone in the Rosell trial. Second, all of the trials required patients to have tumors deemed resectable prior to the initiation of chemotherapy. Thus, patients with more advanced lesions requiring induction therapy to facilitate resection would not have been eligible. Finally, patterns of failure were not consistently reported, but local/regional recurrence appears to have been a major obstacle.