Most patients with stage IIIA nonsmall-cell lung cancer (NSCLC) and with mediastinal node(s) involvement are usually not amenable to complete resection. If resected, the 5-year survival is < 10%. Because the majority of these patients ultimately die of distant metastases, recent efforts to improve their intermediate- and long-term survival have focused on neoadjuvant chemotherapy (with or without radiotherapy) as an induction regimen followed by surgical resection. The ideal trial would address the effect of chemotherapy (with or without radiotherapy) on the survival of patients with clinically visible and mediastinoscopically proven N2 disease.
The theoretical advantages of the neoadjuvant approach include systemic as well as local effects such as: (1) early control of distant micrometastatic disease; (2) prevention of viable tumor seeding at surgery; (3) an increase in the resectability of neoplastic lesions technically unresectable at diagnosis or, alternatively; (4) a reduction of tumor mass before definitive radiotherapy; (5) decreased incidence of positive margins at surgery; (6) possible use of less radical surgery with organ preservation; and (7) in vivo assessment of the effects of chemotherapy and radiotherapy. Moreover, early chemotherapy has been associated with greater efficacy and improved drug delivery to tumor cells via intact vasculature.
Disadvantages of the neoadjuvant approach include morbidity and mortality related to the side effects of induction chemotherapy (with radiotherapy), a reported increase in surgical morbidity and mortality, and a delay in time of surgery.
Over the last 15 years, more than 25 phase II trials and two small randomized phase III trials,[5,6] were terminated early because interim analyses confirmed the feasibility of the neoadjuvant approach and suggested a twofold increase in overall survival compared to surgery alone. More than 1,000 patients were included in the phase II studies. An average response rate of 62% (range 38% to 86%) was reported, with 55% of the patients submitted to thoracotomy and 49.1% resected. Median survival was 16.5 months, and the 5-year survival rate in responsive patients was 30% (50% in pathologic complete responders).
Beyond the inherent heterogeneity of stage IIIA disease, there were relevant differences among these trials, such as the inconsistent use of media-stinoscopy to confirm N2 disease, different surgical philosophies about what constitutes resectable disease, the use of various chemotherapy and radiotherapy regimens, and inconsistent use of postoperative radiotherapy. In some of these trials, good surrogate markers for 5-year survival were the resection rate, complete resection, and tumor downstaging by removal of mediastinal lymph node disease.
More recently, two phase II trials of concurrent chemoradiotherapy, which included patients with both stage IIIA and IIIB disease,[7,8] demonstrated comparable results in both subsets of patients in terms of response to induction therapy, median survival time, and long-term survival. If these results are confirmed, the neoadjuvant approach may be considered for selected patients with stage IIIB NSCLC.
This approach emphasizes both the importance of local as well as distant metastatic control. Concurrent chemoradiotherapy seems to produce higher downgrading rates (60% to 70% with neoadjuvant chemoradiotherapy vs 45% to 50% with chemotherapy alone), and a twofold increase in the proportion of pathologic complete responses, which does not translate to an increase in the complete resection rate.
Because of these results, most thoracic surgeons now accept the multimodality approach for the treatment of patients with stage III N2 NSCLC. Nevertheless, many aspects of the management of these patients need to be clarified. Is there a role for surgery? Is there a best induction strategy (ie, concurrent chemo- and radiotherapy)? What is the role of the new combinations? Is postsurgical chemotherapy warranted? Considering the high incidence of brain metastases observed in completely resected patients, is prophylactic cranial irradiation indicated? Ongoing phase II and III trials are addressing some of these questions.
Positron emission tomography (PET) for staging and monitoring chemotherapy response needs to be evaluated in large prospective trials. Studies on the use of molecular markers as prognostic tools to identify subsets of patients at higher risk of recurrence are also needed. Several studies have established that metabolic imaging with fluorine-18-fluorodeoxyglucose PET is more accurate than computerized tomography (CT) in staging mediastinal involvement in NSCLC and in identifying recurrence or restaging of patients after neoadjuvant chemotherapy.
One problem in assessing tumor response after neoadjuvant chemotherapy is the apparent discrepancy between degree of response on subsequent pathologic examination and response as evaluated by imaging techniques.
In a recently published study, Japanese investigators evaluated changes in tumor size, shape, and contrast enhancement on CT compared with results on pathologic examination of the surgical specimens of 24 patients with NSCLC who had undergone neoadjuvant chemotherapy followed by surgical resection. They found that CT size criteria tended to underestimate the therapeutic effects observed on pathologic examination.
The investigators proposed three CT criteria for complete response based on radiologic-pathologic correlates: (1) more than a 50% reduction in tumor size; (2) change in tumor morphologic features from round or oval to irregular; and (3) disappearance of contrast enhancement after neoadjuvant chemotherapy.
Three randomized clinical trials compared the combination of gemcitabine(Drug information on gemcitabine) and cisplatin(Drug information on cisplatin) to a reference regimen (cisplatin/etoposide in the Spanish study, mitomycin(Drug information on mitomycin) C [Mutamycin]/ifosfamide [Ifex]/cisplatin in the Italian study, and cisplatin alone in the American study) in selected stage IIIB disease (N3 supraclavicular lymph node involvement or neoplastic pleural effusion) and in patients with metastatic disease. The regimen containing gemcitabine demonstrated a statistically significant superior response rate. Two trials also demonstrated superiority in time-to-disease progression.[12-14]
The main toxicity in the gemcitabine-containing combination was hematologic. Grade IV neutrophil toxicity ranged between 35% and 16% and grade 4 platelet toxicity between 38% and 16%. However, there were few episodes of neutropenic fever and only 10% to 15% of treated patients received platelet transfusions. Clinical manifestations of thrombocytopenia (ie, petechiae, mucosal bleeding) occurred seldomly and were usually short lived.
The relative dose intensity of gemcitabine seemed better with the 3-week schedule rather than the 4-week schedule with fewer gemcitabine dose reductions or omissions. Consequently, it seemed logical to investigate the activity of this new combination in locally advanced NSCLC. The preliminary results of five phase II trials have been published recently (Table 1).[15-19]
Phase II Trials
Nearly 5 years ago, the European Organization for Research and Treatment of Cancer (EORTC) Lung Cancer Cooperative Group designed a randomized trial comparing surgery to radiotherapy (EORTC 08941) in patients with stage IIIA NSCLC after they had achieved a major response to three cycles of induction chemotherapy. Any combination chemotherapy regimen containing cisplatin (>100 mg/m²) or carboplatin(Drug information on carboplatin) (Paraplatin) (> 400 mg/m²) was permitted.
A phase II trial of gemcitabine 1,000 mg/m² on days 1, 8, and 15 and cisplatin 100 mg/m² on day 2, administered every 4 weeks, was undertaken in 50 patients with pathologic N2 disease, who were registered in EORTC 08941. The response rate in 36 evaluable patients was 77.5%. Toxicity was comparable to that observed with the same schedule in advanced disease. Throm-bocytopenia was the most significant side effect.
In the other four studies,[16-19] a mixed population of stage IIIA and IIIB patients received the combination of gemcitabine/cisplatin. Almost all patients had a Zubrod performance status of 0 to 1.
In one study, squamous cell carcinoma was the prevalent histology. The cisplatin dose ranged from 90 to 100 mg/m² and was administered on day 1 (one study), day 2 (two studies), and day 15 (one study). The gemcitabine dose ranged between 1,250 mg/m² and 1,000 mg/m², and was administered weekly × 3 every 4 weeks in three studies and weekly × 2 every 3 weeks in one study. Globally, 120 patients entered these trials.
An average response rate of 63% has been reported. There were five complete responses and 13 (11%) patients with progressive disease. Of the 76 responsive patients, 63 (83%) underwent exploratory surgery and 51 (67%) were completely resected. Most of the resected patients were classified at diagnosis as stage IIIA, but complete resection was also feasible, to a lesser degree, in stage IIIB.
Clinical information about tumor downstaging is available from two of these studies. In the Taipei study, tumor downstaging was reported in 47% of evaluable patients and in the Italian study, in 9 of 10 surgically explored patients. A pathologic complete response (12% of the resected patients) was documented in six cases.
Hematologic toxicity, mainly thrombocytopenia, was more pronounced in patients on the 4-week schedule and with cisplatin administered on day 2. Day-15 administration of cisplatin was associated only with grade 3 hematologic (mainly neutropenia) and nonhematologic toxicity (Table 2).
In conclusion, the gemcitabine/cisplatin combination was well tolerated and was effective induction therapy for stage III NSCLC patients with an average response rate > 60%. The dose of gemcitabine was frequently omitted on day 15 of the 4-week schedule, mainly due to the development of thrombocytopenia.
The superiority of cisplatin/gemcitabine in terms of response rate has been clearly established in advanced disease compared with reference regimens such as cisplatin/etoposide and mitomycin C/ifosfamide/cisplatin. The ability of a combination to induce tumor shrinkage could be considered questionable in metastatic disease, but not in the neoadjuvant setting in which tumor shrinkage and downstaging are critical issues for a successful therapeutic approach.
In a phase III study, French investigators examined the administration of two courses of mitomycin/ifosfamide/cisplatin, followed by surgery (N = 187) compared to surgery alone (N = 180) in patients with resectable stages I (with the exclusion of T1 N0 lesions), II, and IIIA disease. A survival advantage, potentially delayed for high perioperative toxicity, was observed in the combined arm. Median survival time favored the combined approach (36 months vs 26 months; P = .11, log-rank test). After 150 days, the effect of perioperative chemotherapy on survival was significantly favorable (relative risk = 0.71; P = .03).
A quantitative interaction between N factor and treatment was also noted, with benefit from perioperative chemotherapy confined to N0-1 patients (P = .008). Disease-free survival was significantly longer in the perioperative chemotherapy arm (P = .02) with similar interaction with N0-1 patients (P = .002). Perioperative chemotherapy induced a pathologic complete response in 19 patients (11%) and a pathologic partial response in 95 patients (53%), for an overall response rate of 64%.
A phase II trial assessed the response, toxicity, resectability, pathologically confirmed complete response, and surgical mortality of two courses of paclitaxel(Drug information on paclitaxel) (Taxol) 225 mg/m²/3-hour infusion and carboplatin at an area under the curve (AUC) of 6 administered every 21 days in 94 patients with stages IB, II, and selected IIIA NSCLC. Completely resected patients received three additional courses of paclitaxel/carboplatin after surgery and 92 patients completed perioperative chemotherapy. Major responses occurred in 54% of enrolled patients.
Of the 92 patients potentially eligible for surgery, 80 (87%) were explored and 70 (76%) were completely resected. No increased or unexpected toxicity or surgical morbidity have been observed. Based on the results of this feasibility study, a randomized Intergroup trial comparing three cycles of induction paclitaxel/carboplatin and surgery to surgery alone in early-stage NSCLC has been planned in the United States.
Ongoing Italian Trial
Using a nearly identical study design, a randomized clinical trial (ChEST [Chemotherapy for Early Stages Trial]) of surgery alone vs surgery plus preoperative gemcitabine/cisplatin in early clinical stages (T2-3 N0, T1-2 N1, T3 N1) NSCLC was launched in January 2000 in Italy. In the experimental arm, gemcitabine 1,250 mg/m², days 1 and 8, and cisplatin 75 mg/m² on day 2, are administered every 3 weeks for a total of three cycles.
The primary end point of the ChEST study is to determine whether preoperative chemotherapy improves time to recurrence compared to surgery alone. Secondary end points include overall survival, patterns of relapse, operative mortality, surgery-related morbidities, response rates, and chemotherapy-related toxicities.
The sample size calculation is based on data of previous adjuvant chemotherapy in completely resected NSCLC (ALPI trial) in which progression-free survival at 3 years is expected to range from 45% to 55%, depending on distribution of the disease stage. Estimated efficacy of preoperative chemotherapy might be in the range of a 20% to 25% reduction in progression-free survival and death.
This assumption is based mainly on an overview of NSCLC, which demonstrates a 13% reduction in mortality in the adjuvant setting, and a 27% reduction in mortality in advanced disease, where compliance to treatment is usually higher. Estimate of the sample size is based on these additional assumptions: duration of accrual of 3 years; minimum follow-up of an additional 3 years; a error equal to 0.05 (two sides), and b error of 0.20 using the log-rank test. Based on these assumptions, the required number of events is about 390, corresponding to a number of patients ranging from 600 to 700.
In stage III NSCLC, response rates to induction chemotherapy are at least twofold higher than in metastatic disease. Tumor downstaging and increased surgical resectability of tumors that were considered at diagnosis to be marginally or totally unresectable have been shown in phase II-III studies.
Regimens, including new drugs such as gemcitabine and the taxanes, have been proven to be at least as active as previously reported with triple-drug combinations like mitomycin C/vinblastine/cisplatin, with the newer regimens likely to be more tolerable. These positive results led us to test prospectively the role of neoadjuvant chemotherapy in stage I and II NSCLC.