T cancer (NSCLC) has been the subject of a number of

Transcription

1 Randomized Trial of Neoadjuvant Therapy for Lung Cancer: Interim Analysis Harvey I. Pass, MD, Helen W. Pogrebniak, MD, Seth M. Steinberg, PhD, James Mulshine, MD, and John Minna, MD Thoracic Oncology Section, Surgery Branch, Biostatistics and Data Management Section, and Bethesda Naval Medical Center Oncology Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland The role of neoadjuvant chemotherapy in stage IIIa non-small cell lung cancer remains undefined. Since 1987, 27 patients with non-small cell lung cancer, all with histologically confirmed metastases to the ipsilatera1 mediastinal lymph nodes, have been enrolled in an ongoing prospective, randomized trial at our institution. Thirteen patients have been randomized to preoperative etoposide-platinum (EP) chemotherapy-surgery-postoperative EP, and 14 other patients have been randomized to surgery-postoperative mediastinal irradiation (SRT). Both groups are similar in sex, age, weight loss, tumor location, preoperative pulmonary function, physiologic grade, and tumor histology. Eight of the 1 EP patients have responded as evidenced by a 50% or greater radiographic tumor shrinkage after two cycles. Complete tumor and nodal resection rates were similar: 11/1 EP patients versus 12/14 SRT patients. There was no operative mortality for the 27 patients. Median potential follow-up is 29.9 months for the EP group and 4.9 months for the SRT group. Preliminary results suggest a trend toward increased survival time for the EP group (median, 28.7 months) versus the SRT group (median, 15.6 months) (p2 = 0.095). Eleven of 12 resected SRT patients have had recurrence versus 8 of 11 resected EP patients. Time to recurrence reveals no significant differences between the two groups but a trend toward increased disease-free interval in the EP group (12.7 months versus 5.8 months, EP versus SRT). This interim analysis demonstrates the feasibility of such a trial; however, despite the trends, definitive conclusions await further accrual and study maturation. (Ann Thorac Surg 1992;5:992-8) he management of higher stage non-small cell lung T cancer (NSCLC) has been the subject of a number of phase I1 trials since the early 1980s, although the first single-institution trial of preoperative chemotherapy for lung cancer was begun in the late 1950s at the Chiba University Lung Cancer Institute [l]. Numerous regimens have been evaluated including etoposide-cisplatinum [2]; mitomycin-c, vinblastine, and cisplatinum []; and 5-fluorouracil and cisplatinum [4]. Moreover, radiation was added in some phase I1 trials to provide greater control in the mediastinum, as well as to take advantage of the For editorial comment, see page 955. radiation sensitizing properties of certain chemotherapeutic drugs. The inability to compare studies due to heterogeneity of the stage IIIa population (ie, T NO, T1-2 N2) has been recently decreased due to the compulsive use of invasive staging to document the status of mediastinal nodal disease. Despite the multitude of phase I1 trials, a phase I11 trial that attempts to evaluate the role of neoadjuvant chemotherapy in a relatively uniform population of patients, all of whom receive appropriate resection and concomitant lymph node dissection, has been lacking. Presented at the Thirty-eighth Annual Meeting of the Southern Thoracic Surgical Association, Orlando, FL, Nov 7-9, Address reprint requests to Dr Pass, Thoracic Oncology Section, Surgery Branch, NCVNIH, Building 10, Room 2807, Bethesda, MD This report summarizes an ongoing phase 111 trial conducted at the National Cancer Institute, National Institutes of Health, since May 1987 comparing preoperative and postoperative chemotherapy to surgery plus mediastinal radiotherapy in stage IIIa patients with histologically confirmed N2 disease. Methods and Patient Population The phase I11 protocol was approved by the Clinical Research Subpanel of the National Cancer Institute and the Clinical Center of the National Institutes of Health in October of 1986 (Clinical Center #86-C-205). Patients were recruited for the study in the intramural program of the Surgery Branch, Clinical Oncology Program, Division of Cancer Treatment in Bethesda, Maryland. Eligibility requirements included the diagnosis of NSCLC, with histologic documentation of ipsilateral mediastinal nodal involvement. This nodal verification could be obtained by mediastinotomy, mediastinoscopy, or Wang needle biopsy. Patients who had undergone diagnostic thoracotomy without resection but with nodal biopsy within 0 days of referral were eligible. All pathologic specimens from referring institutions were reviewed at the NCI for verification of diagnosis of NSCLC. Patients with metastatic disease beyond the regional lobar and mediastinal nodes (ie, N, stage IIIb), prior malignancy, superior vena caval syndrome, malignant pleural effusion, performance status greater than ECOG, prior radiotherapy or chemo by The Society of Thoracic Surgeons /92/$5.00

2 Ann Thorac Surg 1992; PASSETAL 99 NEOADJWANT THERAPY FOR LUNG CANCER Randomization tt Surgery Pre-op Etoposide-Platinum I (2 cycles) t I v.it- Surgery Mediastinal RT Post-op Etoposide-Platinum (4 cycles) Fig 1. Schema for randomization of patients with stage IIIa non-small cell lung cancer. CRT = radiation therapy.) therapy, or serious medical conditions such as severe pulmonary or renal disease that would render the patients unsuitable for chemotherapy or surgery were excluded. The preliminary evaluation, in addition to invasive mediastinal staging, included computed tomographic analysis of the chest, abdomen, and brain as well as radionuclide bone scan. Functional criteria measured in all patients included routine pulmonary functions, baseline audiogram, and creatinine clearance. Complete blood count, measurement of serum electrolytes, and hepatic panel were performed at intervals, as dictated by the protocol. Upon verification of acceptability for protocol entry, patients were randomized to one of two treatment groups, as described below, and were stratified by histologic type, performance status (ECOG), and weight loss (Fig 1). Neoadjuvant Chemotherapy-Surgery-Chemotherapy (EP group) Patients randomized to the neoadjuvant arm received two 21-day cycles of etoposide and cisplatinum chemotherapy (Fig 2). Two 21-day cycles were administered in the outpatient clinic, consisting of a bolus of 120 mg/m' etoposide on consecutive days and cisplatinum (80 mg/m2) every 21 days after sufficient hydration and emetic therapy (metaclopromide, benadryl, and dexamethasone). Dose modification for neutropenia and renal or hepatic dysfunction was strictly defined in the protocol. At the completion of the second cycle, just before the operation, a repeat chest computed tomogram was performed to assess response to chemotherapy. Tumor response was defined as follows: complete response = complete disappearance of all tumor-bearing areas; partial response = a 50% or greater reduction in the sum of the products of two perpendicular diameters of all tumor sites; no response = less than a 50% reduction in the sum of the products of the two perpendicular diameters of all tumor sites. Patients were then taken to the operating room for resection of the primary tumor with concomitant lymph node dissection. Postoperatively, patients who responded to chemotherapy received four additional cycles of the same regimen. Patients who did not fulfill criteria for a partial or complete response preoperatively were offered postoperative mediastinal radiation (see below). Surge y and Postoperative Radiation Therapy (SRT group) Patients randomized to receive immediate surgery were operated on within 2 weeks of randomization. Briefly, the appropriate resection to establish negative margins was performed in all cases, including, if necessary, sacrifice of contiguous structures (chest wall, pericardium, and vena cava wedge in a solitary case). Both the interlobar nodes as well as the mediastinal nodes were resected as recommended by the North American Lung Cancer Study Group [5]. Postoperatively, 54 to 60 Gy of mediastinal radiation was administered over a 6.5-week period in the outpatient radiation oncology clinic. With extensive treatment planning, spinal cord doses were limited to a total dose of less than 45 Gy using multiple fields of 10 MEv x-rays. All patients were restaged every months with physical examination, laboratory evaluation, chest roentgenograms, and head, chest, and abdominal computed tomograms. Bone scans were repeated at 6-month intervals. Statistical Methods All patients randomized to date have been included in the analysis. Durations of survival or time to recurrence (= disease-free interval) were calculated from date of randomization to date of operation as indicated, until date of death, relapse, or last follow-up. The probability of survival or relapse-free survival was calculated using the Kaplan-Meier method [6]. The significance of the differences between pairs of Kaplan-Meier curves was computed by the Mantel-Haenszel technique [7]. Fisher's exact test was used to evaluate the significance of the difference between two proportions. All p values are two-sided and denoted by p2. 4 cycler Fig 2. Schema for the etoposide-platinum (Cis-DDP) -DDP*Etoposidc treatment arm of the randomized protocol Etoporidc 120 mglm Etoposide 120 mgln? 45

3 994 PASS ET AL Ann Thorac Surg 1992;5:992-8 Table 1. Patient Characteristics Variable EP SRT Maleifemale Age (Y) Weight loss > 5% Location (lefthight) I FT FEV, (%) FVC (%) His tology Adenocarcinoma Epidermoid Large cell * 1/ * / f EP = etoposide-platinum/surgery/etoposide-platinum; FEV, = forced expiratory volume in 1 second; FVC = forced vital capacity; PFT = pulmonary function tests; SRT = surgerylradiation therapy. Results From May 1987 to July 1991,27 patients were randomized on the protocol. The full protocol accrual requirement was for 148 patients to detect a 20% improvement in 5-year survival as a result of chemotherapy. However, due to the selectivity of the patient population and single-institution construction of the trial, accrual was much slower than anticipated. There were 12 men and 15 women, of whom 1 randomized to chemotherapy and 14 randomized to immediate operation. As seen in Table 1, there were no differences with regard to functional status, age, location of the primary tumor, or histologic subtypes in either group. Histologic confirmation of N2 disease was obtained by mediastinoscopy in 1 patients, Wang needle biopsy in, mediastinotomy in 9, prior thoracotomy in 1, and percutaneous transthoracic needle biopsy in 1. Sixtythree percent (17) of the patients were symptomatic. Chemotherapy Group All 1 chemotherapy patients (100%) received full dose chemotherapy, and all of this treatment was delivered in the outpatient clinic. Alopecia (1/1) was the most common complication. Eight patients (62%) had a roentgenographic partial response; all other patients had a marginal (2) or no response (). Patients underwent resection 47 * 1 days after initiation of chemotherapy, as called for in the protocol. Chemotherapy was reinstituted a mean of 46 * 11 days after the operation, and 24 of 2 predicted cycles were delivered. Of the 8 preoperative responders, 6 received 100% of the preoperative and postoperative chemotherapy. One patient refused further chemotherapy, whereas chemotherapy was terminated after four cycles in another patient because of deterioration of renal function. In all 50 cycles of chemotherapy, 4 patients required admission: for dehydration due to chemotherapy-related nausea and vomiting, and 1 because of neutropenic fever. Complications and Mortality of Operation There were no postoperative deaths in either the EP or SRT groups. Additionally, we found no objective differ- 8 ence in the conduct of the operation whether or not the patients had received preoperative chemotherapy. Two EP patients had unresectable disease. Both were found to have stage 11% disease at exploration, one with diffuse pleural carcinomatosis and a second with a malignant pericardial effusion. Two patients in the SRT group had unresectable disease because of mediastinal invasion and were reclassified as 11%. There were no significant differences between the two groups in the magnitude of procedure required to establish negative margins (pneumonectomy: versus 5; lobectomy: 7 versus 6; lobectomy and chest wall resection: 1 versus 1; unresectable: 2 versus 2; EP versus SRT, respectively). Postoperative complications in the chemotherapy group included deep venous thrombosis (l), recurrent nerve transection (l), and arrhythmia (1). One patient each in the SRT group suffered pericarditis, hoarseness, and arrhythmia. Another SRT patient had dehiscence of his sleeve resection months after receiving radiation therapy, which necessitated completion pneumonectomy from which he recovered without incident. Pathologic Findings There was only one histologic complete response in a patient with adenocarcinoma who underwent chest wall resection and upper lobectomy. No tumor was found in the surgical specimen, including the lymph nodes. Nevertheless, the patient had recurrence 7 months later in the ipsilateral supraclavicular nodes. All other patients were found to have residual tumor of the same histologic cell type as when they were first entered on protocol. Although a higher proportion of the patients were found to have nodal positivity in the immediate surgery group (Nl, 7/11 versus 10112, EP versus SRT; N2, 6/11 versus 12/12, EP versus SRT), no significant differences were seen at this time. Table 2 details the preoperative clinical staging of the two groups of patients before resection as well as the final pathologic staging based on complete nodal sampling and examination of margins (all of which including bronchial, vascular, and chest wall were negative for tumor in the 2 resected patients). Because complete nodal sampling was performed after therapy in the EP group, it is difficult to relate levels and location with prognostic significance for these patients. Survival and Recurrences The median potential follow-up for the EP group is 29.9 months, compared with 4.9 months for the SRT group. The 18-month survival probability estimate for all patients (n = 27) is 59.7%, with a 95% confidence interval from 40.1% to 76.7%. This did not differ much from the 18-month survival estimate of 6.7% for the 2 resected patients (median survival, 21. months) (Fig ). At this interim analysis, there is no significant difference in survival since randomization of the EP versus the SRT group, with the former having a median survival of 28.7 months versus 15.6 months for the latter (p, = 0.095) (Fig 4). For this interim analysis, a p value of no more than 0.01 would have been needed to consider declaring a significant result before reaching the planned accrual, and thus

4 Ann Thorac Surg 1992;5:992-8 PASSETAL 995 NEOADJWANT THERAPY FOR LUNG CANCER Table 2. Preoperative and Postoperative Pathologic Staging for the 2 Patients With Total Resection Chemotherapy-Surgery-Chemotherapy Surgery and Postoperative Radiation Therapy Preresection Postresection Number of N2 Levels Preresection Postresection Number of N2 Levels Stage Stage Positive for Tumor Stage Stage Positive for Tumor T2 N2 T2 N2 2 T2 N2 T2 N2 1 T2 N2 T2 NO 0 T2 N2 T2 N2 1 T N2 TO NO 0 T2 N2 T2 N2 2 T2 N2 T1 N2 2 T2 N2 T2 N2 T2 N2 T2 N2 2 T N2 T N2 2 T2 N2 T4 NO" 0 T2 N2 T2 N2 1 T2 N2 T2 NO 0 T2 N2 T2 N2 1 T2 N2 T2 N2 2 T2 N2 T2 N2 2 T2 N2 T1 NO 0 T2 N2 T2 N2 2 T2 N2 T2 N2 T2 N2 T2 N2 1 T2 N2 T2 N2 2 T2 N2 T2 N2 1 T2 N2 T2 N2 2 a Partial resection of superior vena cava. this merely indicates a trend. A selected analysis of the survival since randomization for the 8 responders versus the 14 SRT patients reveals a median survival of 28.7 months and 15.6 months, respectively (p2 = 0.07) (Fig 5). However, this analysis is best viewed descriptively because this is a selected comparison and responders needed to survive long enough to be classified as responders. Possibly, a more legitimate method would be to compare EP responders with SRT patients from the date of operation as all patients are known to be either chemotherapy responders or treated with surgery plus radiation therapy at the operation date. Nevertheless, the difference in survival from the date of operation was also not yet statistically significant despite a trend toward increased survival in the chemotherapy responders (p2 = 0.076). Of course, this is also a selected comparison which needs to be interpreted with caution. Of the 2 resected patients, 19 have had recurrence (8%). Eleven of 12 resected SRT patients (92%) and 8 of 11 resected EP patients (7%) have had recurrence (p2 = _ (u 50-0 Resected patients (n = 2) I 20 - I I I I I I I Survival Since Randomization (months) Fig. Survival of the 2 patients in the protocol who had complete resection of disease: median survival was 21. months porn the time of randomization. 0.52). The initial site of treatment failure in the SRT group was systemic in 10 (91%); 1 additional patient had locoregional recurrence (9%). In contrast, 6% (5/8) of the EP patients had locoregional recurrence, and 7% (/8) failed initially systemically (p2 = 0.048). Overall, the brain was the most frequent site of recurrence (7/19; 7%), followed by the locoregional nodes (6/19; 1%) and bone (5119; 26%). Time to recurrence (or disease-free interval) since randomization revealed no significant differences between the two groups (p2 = 0.08) despite a trend toward longer disease-free intervals in the EP group. The 4 patients with unresectable disease were considered to have failed at their date of operation. Analysis of the responders to EP revealed an even greater trend toward increased diseasefree interval compared with the SRT group from the time of randomization (p2 = 0.047) or from the time of operation (p2 = 0.065). These comparisons were both selected ones, and have potential bias because of the lead time for response as well as the several week difference in time of operation after randomization between the two groups. No differences in survival or relapse were found be I I I I I I 1

5 996 PASSETAL Ann Thorac Surg 1992;5: I I I I I I I tween the major histologic groups. Patients with large cell lung cancer or epidermoid histology who were randomized to EP (n = 7) had a trend toward increased survival since randomization over the same histologies randomized to immediate operation (n = 6; p2 = 0.09) (Fig 6). A similar trend was not seen comparing the 6 adenocarcinoma patients randomized to EP with the 8 SRT adenocarcinoma patients (pl = 0.57). Comment The concept of neoadjuvant therapy for higher stage NSCLC has become almost an obsession with thoracic and medical oncologists. The recognition of agents effective in NSCLC, especially in disease limited to the hemithorax, has resulted in a plethora of phase I1 trials trying to identify the population of these patients who would exhibit prolonged survival and freedom from recurrence [8]. Multiple regimens, with and without mediastinal radiation, have been investigated. As stated previously, the heterogeneity of populations studied, entrance criteria loo r-- 80 gol - k-7 Large Cell or (n = 7) 0 Surgery + RT (n = 6) pz = 0.09 I I I I L I I OO b 6 42 Fig 6. Survival of patients with large cell or epidermoid cancer who received or did not receive etoposide-cisplatinum neoadjuvant therapy. with regard to invasive mediastinal staging, and intraoperative conduct of lymphadenectomy have made these studies difficult to interpret. Moreover, the absence of a nonhistorical control group with which to compare the possible effect of the adjuvant therapy only prolongs the agony of trying to definitively answer the question of neoadjuvant utility. These difficulties not only persist, but also were present in 1987 when this study was initiated. Therefore, despite the possibility of poor accrual we attempted to design a single-institution study that could possibly elucidate the role of upfront therapy in NSCLC. There were certain criteria, however, that made the study unique at the time of inception including the insistence of a positive mediastinal nodal biopsy for entrance. From the International Staging System Survival curves, and as corroborated by the Japanese [9], we realized that T NO disease was distinctly different in survival from T1- N2 disease; hence, entrance into the program was restricted to patients with N2 node involvement. We chose EP as the neoadjuvant routine because it is better tolerated and can be administered in an outpatient setting [lo]. A truly effective neoadjuvant program, in our opinion, is one that not only gives consistent response rates but also is well tolerated by the patients. Etoposidecisplatinum, compared with vinblastine and 5-fluorouracil-cisplatinum, had comparable response rates for limited disease. Furthermore, EP did not require extremely high doses of drugs and could be delivered on an outpatient basis (important issues in preserving quality of life and minimizing hospital costs). Moreover, the design of the study was time-limited in that only two cycles would be delivered preoperatively. It was important to deliver these cycles efficaciously to allow timely graduation to the operating room so that the patients disease would not become unresectable if it progressed. Probably the most ambitious portion of the trial design was the insistence of a surgery control arm. Thus, all patients had to have a resection option, albeit aggressive, at study entry. This mandated careful study of the computed tomograms of the chest, and correlation with referral notes when the invasive staging was not performed at the NCI. By definition only a select group of patients were eligible as all patients with presumed IIIb disease were excluded. One patient in the chemotherapy arm who had had a response to chemotherapy was found to have disease requiring segmental resection of the vena cava (T4) after a true partial response. He completed the protocol and was therefore included in the analysis. One could argue that the addition of radiation to the standard treatment group was not in the best interests of the study. In light of the studies by the Lung Cancer Study Group [ll], however, which document decrease in the local recurrence rates by postoperative radiation therapy after resection of higher stage NSCLC without influencing long-term survival, we felt justified in the addition of the mediastinal radiation. One could argue that this arm could have been just as easily treated without radiation; however, as described in the results, the disparity of

6 Ann Thorac Surg 1992; PASSETAL 997 local versus systemic recurrences seen between the two groups (EP versus SRT) may not have been appreciated. The chemotherapy regimen was well tolerated. Few patients had complications and these were reversible. The necessity for in-hospital resources was minimal. Moreover, all of the chemotherapy was delivered under the auspices of the surgery branch of the NCI, in the Surgery Branch Outpatient Clinic. There was no involvement of other disciplines in the delivery of treatment, ie, medical oncology, and therefore routine follow-up and management of these patients was consistent and easy to perform. This emphasizes the ability of thoracic oncologic surgeons to play more of a role than just technicians in the management of these patients. The response rate of 62% is comparable with that of other regimens used (vinblastine, 5-fluorouracil-cisplatinum). The other regimens, however, seem to be associated with greater toxicity, specifically myocardial and pulmonary [ 121, and required greater inpatient involvement. The solitary histologic complete response was less than that seen with other regimens; however, the size of this study must be taken into account when defining this parameter. Moreover, the value of the histologic complete response in affording a survival advantage is not fully established in the literature for this population of patients. In our 1 case, the patient had local recurrence in the mediastinum despite the histologically negative tumor bed and nodal basin. The only thoracic surgeon involved in this study was the principal investigator (H.P.). The ability to perform complete resection in these patients probably reflects the homogeneity in the selection of the IIIa group. Only 15% of the patients explored (2 from each group) were found to have unresectable IIIb disease. The disease therefore could be completely resected to the surgeon s satisfaction after pulmonary resection and lymphadenectomy. Any opinions regarding the degree of difficulty of these operations, operative results, or patient management decisions remain uncontrolled by other observers. Nevertheless, in our series, there was no difference in the difficulty or magnitude of resection, conduct of operation, or estimated blood loss. Both the CT and the SRT groups had comparable lengths of operation and duration of hospital stay, and neither group suffered operative mortality. Moreover, in neither group was any patient left with a positive margin at the completion of the resection, and the extent of resection was comparable. Therefore, at least with EP, one cannot conclude that preoperative chemotherapy affects the ability to conserve pulmonary parenchyma. Despite the trend toward a decreased number of patients in the chemotherapy group with N2 or N1 involvement at resection, and the uniformity of lymph node dissection in these patients (one surgeon, one technique), one cannot say that this truly represents downstaging of disease due to the small number of patients in the series. It is interesting, however, that the majority of the recurrences in the chemotherapy group were locoregional (but extrathoracic), and the majority of the recurrences in the SRT group were systemic. This finding titillates the imagination in that perhaps the addition of preoperative or postoperative radiation therapy might have an impact on these local recurrences in future studies. This study is only at interim analysis with regard to survival. Nevertheless, the median survival of all resected patients (21. months) is significantly better than that in other studies of IIIa NSCLC patients with clinical N2 MO disease. We found no difference at this time in survival from time of randomization, yet the disease-free interval has a trend toward increased time to recurrence in the EP group. This is vaguely reminiscent of the findings of the LCSG in a number of studies that investigated postoperative adjuvants after NSCLC resection. What, then, is the take-home message from this study at interim analysis? Certainly, at this time this study does not justify the routine use of neoadjuvant chemotherapy in the management of N2 positive NSCLC patients. Moreover, it is foolhardy to say that the EP regimen is the only or best regimen for these patients, and without doubt, other regimens will give similar response rates [-5, 121. It is even possible that more platinum should be delivered for even greater response rates; however, a close scrutiny of Klastersky s work regarding EP would not support this claim, at least in higher stage patients [lo]. Nevertheless, there are no phase I11 trials that compare different regimens for neoadjuvant therapy to each other in a homogeneous group of patients with stage IIIa disease. Furthermore, this study does not imply that preoperative chemotherapy will significantly improve long-term survival in these patients. Despite high response rates, it is too early to define favorable populations that will enjoy the benefits of these responses. The study does reinforce the safety of operating on patients with higher stage lung cancer, and implies that a fixed, but small percentage of patients will be salvaged by operation alone. Although this study continues to accrue patients, it will be the responsibility of the cooperative groups to continue ongoing phase I11 trials such as this to definitively define the role of neoadjuvant chemotherapy. References 1. Katsuki H, Shimada K, Koyama A, et al. Long-term intermittent adjuvant chemotherapy for primary resected lung cancer. J Thorac Cardiovasc Surg 1975;70: Wozniak A, Kraut M, Herskovic A. Treatment of locally advanced non-small cell lung cancer with continuous infusion VP-16 and cisplatin and concurrent radiation therapy. Proc Annu Meet Am SOC Clin Oncol 1990;9:A975.. Martini N, Kris MG, Gralla RJ, et al. The effects of preoperative chemotherapy on the resectability of non-small cell lung carcinoma with mediastinal lymph node metastases (N2 MO). Ann Thorac Surg 1988;45: Faber LP, Kittle CF, Warren WH, et al. Preoperative chemotherapy and irradiation for stage I11 non-small cell lung cancer. Ann Thorac Surg 1989;46: Ginsberg RJ, Goldberg M, Waters PF. Surgery for non-small cell lung cancer. In: Roth JA, Ruckdeschel JC, Weisenburger TH, eds. Thoracic oncology. Philadelphia: W.B. Saunders, 1989: Kaplan E, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc 1958;5: Mantel N. Evaluation of survival data and two new rank

7 998 PASSETAL Ann Thorac Surg 1992;5:9924? order statistics arising in its consideration. Cancer Chem Rep ;50:16? Ruckdeschel JC, Holmes EC. Preoperative chemotherapy for 11. locally advanced non-small cell lung cancer. Chest Surg Clin N Am 1991;l:l Naruke T, Goya T, Tsuchiy R, et al. Prognosis and survival in resected lung carcinoma based on the new international staging system. J Thorac Cardiovasc Surg 1988;96: Klastersky J. Therapy with cisplatin and etoposide for nonsmall cell lung cancer. Semin Oncol 1986;1: Weisenburger TH. Postoperative radiotherapy for non-small cell lung cancer. Chest Surg Clin N Am 1991;1:7148. Spain RC. Neoadjuvant mitomycin c, cisplatin, and infusion vinblastine in locally and regionally advanced non-small cell lung cancer: problems and progress from the perspective of long term follow-up. Semin Oncol 1988;15:&15. DISCUSSION DR CAROLYN E. REED (Charleston, SC): I applaud your efforts and I think we all desperately need to join in a multiinstitutional trial. I have several questions for you. One concerns the selection process of your N2 patients. I know a number of the patients are referred to you. Are all these patients sent to you by surgeons who just routinely do mediastinoscopy? The reason I ask this is that one of your arms is surgery alone, and in many of the studies that are reported we are talking about clinical N2 disease, disease that we would not expect to be able to take to surgery and perform a complete resection. In other words, we are using the chemotherapy to make them resectable and you are using surgery alone on this. That is one question. The second question is, as you had an increased local failure rate in an arm that showed decreased systemic recurrence, what do you think the role of radiation therapy should be in that arm? Finally, would you give us some words of wisdom as to what you think would be the ideal multiinstitutional study that we should mount? DR PASS: Regarding your mediastinoscopy question, approximately 50% of these patients had their nodal disease diagnosed invasively by me. I think the most important thing is the screening of the computed tomographic scan. When you look at enough computed tomographic scans, you can say whether there is going to be gray area, IIIb-IIIa, or whether this is going to be a patient in whom you can perform resection. Personally I think there is not enough emphasis in the literature on that. I think that essentially you ought to be able to define bulky versus nonbulky disease ; also, by doing the mediastinoscopy yourself you get the idea whether you are going to be able to resect those nodes. DR REED: What is going to be the role of radiation therapy? DR PASS: I have no idea what the role of radiation therapy is going to be. That is why I wrote the program originally to not even involve that in the up-front treatment. The purpose of this trial was not to look at what radiation therapy does in non-small cell lung cancer up front. I wanted to see how good the chemotherapy was. Now, what the study is showing is that you may have to add something to the chemotherapy for local control. With regard to the ideal study, I am not so sure that the ideal control is a nonsurgical group. My own feeling is that if you have enough surgeons who operate on these higher stage lung cancer patients, they can sift out which are the real IIIa s with N2 disease and resect them and see if you salvage a fixed number, as opposed to just giving up on them. DR EDWARD F. PARKER (Charleston, SC): I enjoyed very much this report and your diligence in pursuing this study. I was impressed with the fact that among the survivors with recurrence, apparently the first recurrence was in the cervical lymph nodes. One would wonder if they could have been positive at the start. Brantigan in Baltimore some years ago showed that in patients with carcinoma of the lung, the ipsilateral nodes on routine biopsy were positive in about 20%, even though they were nonpalpable. We found some years ago the same thing; we also found that the contralateral nodes were positive for carcinoma in about 20% of the patients even though there was no abnormality on palpation of the neck. What is your opinion of the value of preoperative cervical lymph node biopsy, unilateral or bilateral, in the study of these patients to ascertain, more accurately than we do now, the preoperative staging and the prognosis after excision of the primary lesion? DR PASS: Dr Parker, I think that is a good question. If you look at the locoregional recurrences, what you see is that many patients fail in supraclavicular nodes. We also have had patients with recurrence in paratracheal and azygos nodes. I am not so sure that routine scalenus or supraclavicular nodal biopsy is going to salvage many more of these patients. The bigger question, though, is, does this recurrence affect long-term survival, and I do not really have an answer to that question. So if it does not affect long-term survival, I am not so sure whether it is necessary to have the patient submit to concomitant preoperative radiation therapy, which in itself could be toxic. DR JOE 8. PUTNAM, JR (Houston, TX): Dr Pass, I enjoyed your presentation and certainly share your concern for better treatment for lung cancer. Specifically I was interested in your standard criteria for inclusion, how these patients were staged, and your comments on the value of routine brain computed tomograms before operation and also in following up these patients long-term. DR PASS: I do everything to make sure that these patients are selected as a right homogeneous group. All patients have preoperative computed tomograms of the head, chest, and abdomen, and bone scans in the staging evaluation. You could argue that you are not going to pick up that many patients with silent brain metastases in this group. I personally feel, though, that for study purposes, to answer the question, you must do the preoperative computed tomography. These patients have follow-up computed tomograms every months to attempt to pick up a recurrence as early as possible. I do not think the studies need to be performed any more often than that. References 1. Cromartie RS 111, Parker EF, May JE, Metcalf JS, Bartles DM. Carcinoma of the lung: a clinical review. Ann Thorac Surg 1980;60: Brantigan JW, Brantigan CO, Brantigan OC. Biopsy of nonpalpable scalene lymph nodes in carcinoma of the lung. Am Rev Respir Dis 197;