For patients with early-stage NSCLC (clinical stage I/II), lobectomy or pneumonectomy with mediastinal sampling or lymphadenectomy is the optimal treatment modality for the potential cure of patients and may result in a 5-year survival ranging from 67% in patients with pathologic T1, N0 disease to 38% in patients with pathologic T3, N0 disease. Clinical staging often understages NSCLC, particularly if newer modalities such as positron-emission tomography (PET) scanning are not utilized. As a result, patients with disease at a given clinical stage fare much worse than those at the corresponding pathologic stages (Table 1). The majority of patients with stage III NSCLC evaluated for resection have a disease extent that precludes immediate curative surgery. However, even in patients who are immediately resectable, those with pathologic T1-3, N2 disease (stage IIIA) have a 5-year survival of only 23%. Moreover, when bulky mediastinal involvement is noted on chest radiography, 3-year survival is dismal-less than 10%. Although in all of these patients disease is confined to the chest at onset and can be removed surgically, the vast majority suffer from recurrent NSCLC and ultimately die of metastatic disease. Adjuvant Approaches
In order to improve on the disappointing results of surgery alone, investigators have focused on chemotherapy and radiotherapy in addition to surgery. The majority of historical clinical trials examining adjuvant (ie, postoperative) therapy have not found unequivocal, reproducible evidence of efficacy for adjuvant chemotherapy.[ 5-7] However, a meta-analysis of 14 trials including over 4,000 patients and comparing postoperative chemotherapy with no further therapy demonstrated a survival advantage for chemotherapy regimens containing cisplatin(Drug information on cisplatin). More recently, a large randomized study showed that cisplatin- based adjuvant chemotherapy improves survival by 4% at 5 years. More recently, two trials in early-stage homogeneous populations demonstrated significant survival benefits in patients treated with platin-based adjuvant chemotherapy. [10, 11] Induction Therapy
Given this uncertainty, chemotherapy prior to surgery, or induction ther- apy,has been the subject of ongoing investigations. Induction therapy has been attractive to clinicians because regression of the primary tumor to chemotherapy serves as a surrogate marker for the control of clinically undetectable metastases. In fact, consistently higher response rates and better tolerability have been noted in locally advanced disease than in metastatic NSCLC.[12-14] Notably, a complete pathologic response rate of 13% has been reported in early clinical studies of induction chemotherapy. Thus, chemotherapy given before surgery has the potential to allow for assessment of tumor response to chemotherapy, to treat micrometastatic disease early, to facilitate the complete eradication of lesions, and to provide better dose delivery with fewer treatment interruptions. Additionally, decreasing tumor size may improve resectability and decrease the morbidity of resection. Resectable NSCLC Patients with stage I/II NSCLC can generally undergo immediate surgery with curative intent. Although patients with stage III disease have no evidence of distant metastases and are viewed as potentially curable, their prognosis is generally poor, and the overwhelming majority die of their disease. Stage IIIA encompasses patients for whom extended surgery can remove all known sites of disease. Patients with stage IIIB disease are frequently considered unresectable, because of contralateral mediastinal lymph node involvement (N3) or extensive tumor invasion (T4) that requires specialized surgical techniques to accomplish complete resections. A carefully selected subset of patients with T4 disease may be suitable for surgical resection. The basic division between stage IIIA and IIIB NSCLC was established in the 1980s. However, in the 1997 revision of the International Staging System, patients with T3, N0 disease were reclassified as having stage IIB because their survival was almost identical to patients with T2, N1 tumors. This grouping complicates interpretation of stage III trials before 1997, because treatment groups with T3, N0 patients have improved outcomes. Futhermore, within stage IIIA disease there is significant variability that can have an impact on prognosis. Larger tumors and multiple involved lymph node stations result in worse survival.[16-19] Staging NSCLC Computed Tomography
Patients with potentially resectable NSCLC, especially those with stage III disease, benefit from rigorous staging evaluation prior to therapy. Computed tomography (CT) scanning is critical in evaluating the location and extent of the primary tumor. CT scans, however, may under- or overestimate mediastinal lymph node involvement. Normal-sized lymph nodes (generally considered to be < 1 cm in the short axis) can contain tumor cells, and enlarged lymph nodes do not always contain metastatic tumor. Lymph node metastases become more likely with increasing nodal size. In one meta-analysis of 29 studies involving 2,226 patients utilizing CT scans for the detection of mediastinal nodal metastases, the mean sensitivity and specificity were 60% and 77%, with a positive predictive value of 50% and a negative predictive value of 85%. An enlarged mediastinal lymph node detected by CT scanning, therefore, only contains cancer 50% of the time, and 15% of normal-sized (< 1 cm) lymph nodes contain tumor cells. Positron-Emission Tomography
PET with 18F-fluorodeoxyglucose is a newer staging technique that uses the relatively increased metabolic activity of cancer cells to differentiate them from normal cells. In the aforementioned meta-analysis, data from 14 studies involving 514 patients were examined, and the authors found that the mean sensitivity and specificity for PET were 79% and 91%, respectively, with a positive predictive value of 90% and a negative predictive value of 93%. Pieterman et al reported their experience in 102 patients with resectable NSCLC with preoperative PET scanning and found sensitivity and specificity of 91% and 86%, respectively.[ 22] However, in this group of patients, the positive predictive value of PET was 74%. Therefore, 26% of patients with increased uptake on PET have no evidence of nodal disease on histopathology. Furthermore, even lymph nodes that are positive by CT and PET require histopathologic proof of involvement. The negative predictive value of both a negative PET and CT scan of 97% was excellent in this series, and many surgeons feel that PET scans can reliably rule out N2 disease but that surgical staging is necessary to confirm its presence. Mediastinoscopy
In many centers, mediastinoscopy is routinely performed to stage patients with lung cancer who are candidates for surgical resection. Mediastinoscopy is used to detect N3 disease for which surgical therapy would not be advised and to detect N2 disease for which induction therapy prior to definitive surgery would be warranted. Because not all lymph nodes can be sampled, cervical mediastinoscopy provides sensitivity between 72% and 89%.[23-25] Cervical mediastinoscopy allows access to the right and left paratracheal lymph nodes and subcarinal lymph nodes. The aorticopulmonary window and subaortic lymph nodes are not accessible via cervical mediastinoscopy and require anterior mediastinotomy. Combined Modalities
Appropriate staging procedures for patients with resectable NSCLC, therefore, include a CT scan not only to assess the primary tumor and mediastinum but also to evaluate for evidence of metastatic disease within the lung, liver, and adrenal glands. All patients with mediastinal lymph nodes that are greater than 1 cm by CT imaging or are positive by PET scan should have mediastinal disease confirmed histopathologically. In patients with normal mediastinal lymph nodes by CT scan, a negative PET scan may obviate the need for mediastinoscopy. Mediastinoscopy, however, may identify patients with micrometastatic lymph node involvement not apparent on imaging modalities who may benefit from induction therapy. Detecting Brain Metastases
The incidence of brain metastases in the initial staging of patients with primary lung cancer has been reported to between 12% and 18%.[26,27] Asymptomatic brain metastases occur more frequently in patients with more advanced stages of disease, with rates as high as 30% at 2 years in stage II/III patients. Higher stage and nonsquamous histology have been identified as risk factors for brain metastases.[ 28,29] In asymptomatic patients, gadolinium-enhanced magnetic resonance imaging (MRI) of the brain is superior for the detection of occult brain metastases. A recent study randomized 332 patients with potentially operable NSCLC, but without neurologic symptoms, to brain CT or MRI in order to detect occult brain metastasis before lung surgery. MRI showed a trend toward a higher preoperative detection rate than CT (P = .069), with an overall detection rate of approximately 7% from pretreatment to 12 months after surgery. In patients with stage I/II disease, the detection rate was 4% (8 of 200), whereas for individuals with stage III disease, the rate was 11.4% (15 of 132). Whether the improved detection rate of brain metastases by MRI translates into improved outcome remains unknown. Brain imaging is currently recommended as a part of the initial evaluation for all patients with potentially curable locally advanced NSCLC. Finally, reevaluation of disease after induction chemotherapy and prior to thoracotomy is recommended to exclude disease progression. The ability of noninvasive tests to predict pathologic response has thus far been suboptimal. Induction Chemotherapy Trials Initial Observations
Several clinical trials have tested combination chemotherapy prior to surgery to improve the outcome of patients with resectable NSCLC. Initial studies using older chemotherapy regimens suggested that preoperative chemotherapy may be beneficial.[31,32] Martini demonstrated that otherwise resectable patients with ipsilateral mediastinal lymphadenopathy as their sole site of distant spread can have 3-year survivals of 43%, and 5-year survivals of 24%, if both the primary tumor and ipsilateral mediastinal nodes were completely resected and followed by mediastinal irradiation.[ 33,34] These same studies revealed that individuals with bulky ipsilateral mediastinal lymphadenopathy had only an 8% 3-year survival. Based on these observations, a preoperative combination chemotherapy program with MVP (mitomycin [Mutamycin], vinca alkaloids, and highdose cisplatin [Platinol] at 120 mg/m2) was developed at Memorial Sloan- Kettering Cancer Center for use in stage IIIA patients with clinical N2 disease. In a group of 136 patients, the objective major response rate to MVP was 77%, with a 10% complete response rate. Overall, 65% of patients underwent complete resections; 14% had achieved a pathologic complete response at surgery. The median survival was 19 months for all patients. The 3-year survival for completely resected patients was 41%-a significant improvement over the prior surgery-only experience, where the 3-year survival for this group was 8% (P = .001). Particularly noteworthy was that a pathologic complete response was observed in approximately 12% of advanced NSCLC patients who received preoperative MVP chemotherapy, with survival estimates in this population of 54% at 5 years. The Cancer and Leukemia Group B (CALBG) conducted a second similarly designed trial. A total of 74 patients with surgically staged IIIA (N2) NSCLC were treated with two cycles of preoperative cisplatin and vinblastine(Drug information on vinblastine).[ 36] Patients who underwent resection were then treated with two cycles of adjuvant cisplatin and vinblastine followed by thoracic radiation. Approximately 64% of patients achieved a radiographic response or stable disease, and 62% underwent a complete resection. Operative mortality was 3% (2 deaths). Median survival was 15 months. Elias and investigators at the Dana- Farber Cancer Institute treated 34 patients with pathologically confirmed N2 disease, using infusional cisplatin, fluorouracil(Drug information on fluorouracil) (5-FU), and leucovorin plus postoperative radiotherapy to 54- 60 Gy. The radiographic response rate to this regimen was 65%. Thoracotomy was performed in 82% of patients. No operative mortality was noted, and 18% of patients achieved a pathologic complete response. Median survival was 18 months. The authors noted that few local recurrences occurred and that 15% of first relapses were solely in the brain. Randomized Studies
Following these initial observations that suggested promising results with induction therapy, a number of randomized studies aimed at establishing the role of induction therapy were conducted.
- Spanish Study-The first randomized trial that compared cisplatinbased chemotherapy plus surgery with surgery alone was conducted by Rosell et al in patients with stage IIIA disease.[ 38] Pathologic confirmation of N2 was not mandatory, and approximately 27% of patients had clinical T3, N0/1 disease. Chemotherapy consisted of the MIP regimen (mitomycin, ifosfamide(Drug information on ifosfamide) [Ifex], cisplatin) given at 3-week intervals before surgery for three cycles. All patients received mediastinal irradiation (50 Gy) after surgery. A total of 60 patients were randomized, 30 in each arm. After 2 years, the trial was prematurely terminated because of the significant benefits seen in the chemotherapy arm. No toxic deaths occurred during preoperative therapy. The partial response rate was 53%, with a 7% clinical complete response rate. Postoperative mortality was similar in both groups (two deaths each). Overall median survival was 26 months in the chemotherapy arm vs 8 months in the surgery arm (P < .001). Survival rates at 2 years were 27% and 0%. Survival in the control arm of surgery alone was much lower than expected from historic controls. An analysis performed 7 years later confirmed the survival advantage in favor of preoperative chemotherapy, with median survival of 22 months in the chemotherapy arm vs 10 months in the surgery-alone arm (P = .005). The survival rate at 5 years was 17% and 0%, respectively.
- M. D. Anderson Study-Roth and investigators at M. D. Anderson Cancer Center conducted another randomized study of preoperative chemotherapy in stage IIIA NSCLC. Chemotherapy consisted of three cycles of cyclophosphamide(Drug information on cyclophosphamide) (Cytoxan, Neosar), etoposide(Drug information on etoposide), and cisplatin. Patients who responded to preoperative chemotherapy were given an additional three cycles of adjuvant therapy after complete resection. Patients who were found to have unresectable disease with incomplete resection could undergo postoperative radiation therapy. A total of 28 patients were randomized to receive preoperative chemotherapy, and 32 were randomized to receive primary surgery. Almost one-quarter of patients had T3, N0/1 disease. The overall response rate was 35%, including one pathologic complete response. Disease progression developed in four patients during chemotherapy; however, the resectability rate and the rate of complete responses were the same in both arms. An interim analysis showed a statistically significant survival advantage for the preoperative chemotherapy arm (P < .008). The trial was therefore halted after 60 patients were enrolled. The estimated 2-year survival rate in the preoperative arm was 60% and in the surgery arm was 25%. An analysis 4 years later revealed that median survival in the chemotherapy arm was 21 months and in the surgery arm was 14 months (P = .056).
- NCI Study-A third randomized trial in histologically confirmed N2 NSCLC was conducted by Pass et al at the National Cancer Institute (NCI). Chemotherapy consisted of cisplatin and etoposide. At interim analysis, 27 patients had been randomized. The response rate was 61%. Preliminary results suggested a trend toward increased survival for the preoperative chemotherapy arm (median survival = 29 vs 16 months, P = .095).
- Japanese Study-A fourth randomized study, conducted in Japan, compared induction chemotherapy with cisplatin and vindesine(Drug information on vindesine) (Eldisine) vs surgery alone in patients with stage IIIA (N2) NSCLC. After 62 patients were enrolled, this study was terminated due to slow accrual. There were no statistically significant differences in survival in this trial (median overall survival was 17 months for patients receiving induction and 16 months for patients undergoing surgery alone).
Taken together, the four small, randomized studies described above demonstrated a potential benefit with the addition of induction chemotherapy in stage IIIA NSCLC. Studies have since been designed to evaluate the role of induction chemotherapy before surgery in even earlier-stage NSCLC.
- French Study-The French Cooperative Oncology Group conducted a large randomized study in stage I (except T1, N0), II, and IIIA resectable NSCLC.[44,45] The 355 eligible patients were randomized to induction therapy with two cycles of mitomycin(Drug information on mitomycin), ifosfamide, and cisplatin or to primary surgery. Patients who responded to chemotherapy received an additional two cycles postoperatively. In both arms, patients with pT3 or pN2 disease or incomplete surgery received radiotherapy after either surgery or completion of chemotherapy. This regimen yielded a response rate of 64% with a pathologic complete response rate of 11%. The complete resection rate was similar between the two arms: 92% with induction vs 86% with surgery alone. A 10-month improvement in median survival in the chemotherapy arm was documented (26 vs 36 months), although this was not statistically significant (P = .11). Operative mortality was similar in both arms, but there was a trend toward increased postoperative mortality (7% vs 4.5%) and increased postoperative complications in the chemotherapy arm. Although a statistically significant improvement in survival could not be demonstrated, subset analysis showed that patients with stage N0/1 disease derived a survival benefit from induction chemotherapy (hazard ratio = 0.68, 95% CI = 0.49-0.96, P = .027). This survival benefit was not observed in patients with N2 disease (hazard ratio = 1.04, 95% CI = 0.68-1.60, P = 0.85). The study by Depierre et al is the largest trial of induction chemotherapy conducted to date. However, it remains nonconclusive as it included a heterogeneous group of patients with various stages of disease and its implications for the treatment of patients with stage IIIA NSCLC are only based upon subset analysis.
- US Studies-The role of induction chemotherapy for early-stage NSCLC has also been under investigation in the United States. Pisters et al reported promising results from the multicenter phase II study, the Bimodality Lung Oncology Team (BLOT) trial. Patients with stage IB-IIIA (T3, N1, M0) NSCLC received induction therapy with two cycles of carboplatin(Drug information on carboplatin) (Paraplatin) and paclitaxel(Drug information on paclitaxel) followed by three additional cycles postoperatively. Evaluable patients included 94 evaluated for survival and toxicity and 90 for response. The response rate was 56%, the pathologic complete response rate was 6%, and 3% of patients had disease progression during induction. The 1-year survival rate was 85%. Results from a second group of patients who received three rather than two cycles of induction chemotherapy with carboplatin and paclitaxel have also been reported. In a total of 124 patients, the response rate was 51% and 3-year survival was 61%, which was superior to historical controls. The surgical mortality rate was only 1%. The North Central Cancer Treatment Group (NCCTG) conducted a similar phase II study with induction carboplatin and paclitaxel in patients with stage T1-3, N0/1, M0 disease. Of 52 patients, 3 died postoperatively. The 2-year survival rate was 73%. An ongoing phase III randomized study (S9900) is randomizing patients to three cycles of induction carboplatin and paclitaxel followed by two additional cycles of postoperative carboplatin and paclitaxel or surgery alone. Over 325 patients have thus far been entered into this trial, which has an accrual goal of 600.