M expected to arise in 1.6% to 3.0% of all patients. Multiple Primary Lung Carcinomas: Prognosis and Treatment

Transcription

1 Multiple Primary Lung Carcinomas: Prognosis and Treatment Todd K. Rosengart, MD, Nael Martini, MD, Pierre Ghosn, MD, and Michael Burt, MD, PhD Thoracic Service, Department of Surgery, Memorial-Sloan Kettering Cancer Center, New York, New York From 955 to 99, patients have been treated for multiple primary lung carcinomas. Criteria for diagnosis were: () different histology (n = 44); or () same histology, but disease-free interval at least years (n = 9, origin from carcinoma in situ (n = 9), or metachronous disease in different lobe (n = 9) with no cancer in common lymphatics or extrapulmonary metastasis at the time of diagnosis. The second cancer was synchronous in patients (%) and metachronous in 78 (7%). disease developed at a median interval of 48 months. Five-year survival for patients with metachronous and synchronous disease from the time of initial diagnosis of cancer was 7% and 44%, and -year survival was 4% and %, respectively. Survival after the development of a metachronous lesion was % at 5 years. Survival from the time of initial diagnosis was significantly better for metachronous versus synchro- nous, late (4 month disease-free interval) versus early metachronous disease, and adenocarcinoma versus epidermoid carcinoma. The first cancer was completely resected in patients (9%), but complete resection of a metachronous tumor was possible in only 54 patients (9%). Complete resection of second primary cancers resulted in significantly (p <.) prolonged 5-year survival compared with incomplete resection (8% versus 9%). Excluding patients requiring pneumonectomy, initial resection limited subsequent resection in only 7 patients (9%) with metachronous disease. We conclude that patients surviving treatment of primary lung cancers require lifelong screening for multiple primary lung carcinoma, and complete resection is recommended whenever possible. (Ann Thoruc Surg 99;5:77-9) ultiple primary lung carcinomas (MPLC) may be M expected to arise in.% to.% of all patients with lung cancer, and in as many as % to 5% of those patients surviving more than years [l,. Although two separate primary lung cancers were first identified by Beyreuther in 94 [], reports of large series of patients with MPLC are sparse. Appropriate diagnosis and treatment for patients with MPLC therefore remain ill-defined. Establishing appropriate criteria for MPLC is critical in distinguishing between synchronous or metachronous MPLC versus metastatic or locally recurrent primary disease, respectively. This distinction can have important clinical and therapeutic implications. A synchronous second primary cancer judged to be metastatic disease would classify the patient as stage IV, probably contraindicating a potentially curative resection. The erroneous designation of a metachronous second primary cancer as a local recurrence might have similar therapeutic implications. The extent of resection is also an area of concern in treating MPLC. Some have suggested that first primary cancers be resected by a segmentectomy whenever possible, leaving sufficient lung parenchyma to permit subsequent resection of MPLC [4, 5. Recommended treatment for MPLC has ranged from local resection or nonoperative Presented at the Twenty-seventh Annual Meeting of The Society of Thoracic Surgeons, San Francisco, CA, Feb &, 99. Address reprint requests to Dr Burt, Memorial-Sloan Kettering Cancer Center, 75 York Ave, New York, NY. treatment for what has been labeled as "recurrent disease" [5, to standard lobar excisions when possible [7, 8. We have reviewed our experience with MPLC from 955 to 99. This review includes an initial 5 patients previously reported [8] and an additional patients seen and treated since. Material and Methods A review of the medical records of all patients treated for multiple lung carcinomas on the Thoracic Service of the Memorial-Sloan Kettering Cancer Center was carried out. All charts of patients with a probable diagnosis of MPLCs were available for analysis. Designation of multiple primary lung cancer was made based on the criteria established at our center and reported in 975 (Table ). Patients not meeting these criteria were excluded. Staging was according to the International TNM Classification for lung cancer [9] and was based on data obtained from imaging techniques, invasive diagnostic techniques (mediastinoscopy and percutaneous biopsy), bronchoscopy, and operative findings. Diagnosis of MPLC was made at autopsy in patients. Histologic typing was according to the World Health Organization classification [lo]. Cytologic diagnoses were not accepted for final histologic typing. Tumors were designated as "synchronous" when detected or resected simultaneously and "metachronous" when the second tumor was found at some time later. The larger of two synchronous tumors 99 by The Society of Thoracic Surgeons -4975/9/$.5

2 ~ ~ 774 ROSENGART ET AL Ann Thorac Surg 99;577>9 Table. Criteria for Diagnosis of Second Primary Lung TumoF Criteria No. ~~ ~~ ~ tumors Histology different 7 Histology the same,b if. Tumor-free interval at least years. Second cancer in different lobe or lung, and. Origin from carcinoma in situ Synchronous tumors Histology different 7 Histology the same,b if. Second tumor in different segment, lobe, or lung. Origin from carcinoma in situ a From Marhni et a [8] No extrapulmonary or common lymphatic carcinoma at hme of diagnosis 9 9 Number 4 4 F atients Median 9 4 years Time (yrs) Fig I. Time to occurrence of second tumor in 78 patients with metachronous disease. Median time interval was 48 months. was considered the first primary cancer for the purpose of some analyses. Follow-up was obtained to within months of this report in 95% of patients. In most patients disease status was assessed by serial chest roentgenograms, history, and physical examinations (n = ). Follow-up of several patients (n = 5) was obtained through contact with local physicians. Survival was calculated by the Kaplan-Meier method, and differences in survival were determined by log-rank analysis. Comparisons were also made using Student s t test and x analysis, as appropriate. Significance was defined as p less than or equal to.5. Results Pafienf Population and Demographics Multiple primary lung carcinomas were observed in patients. The criteria for diagnosing MPLC and the number of patients defined by each criterion are presented in Table. Synchronous tumors were noted in patients (%) and metachronous tumors in 78 patients (7%). The median age at time of initial diagnosis was years, with an age range of to 9 years. There were 87 male and 4 female patients. Patients with synchronous tumors were somewhat older (median age, years) than patients with metachronous tumors (median age, years). A third primary lung carcinoma was noted in 7 patients. Carcinoma The median time to diagnosis of metachronous disease was 48 months, with a range of to 8 months (Fig ). The disease-free interval was essentially the same whether the second tumor was of the same or different histology. primaries appeared greater than years after the first tumor in patients (5%). Squamous cell carcinoma was by far the most common histologic type, occurring as the first tumor in 45 (58%) of 78 patients and as the second tumor in 9 (5%) patients (Table ). The next most common histologic type was adenocarcinoma, occurring in (9%) and (%) of first and second cancers, respectively. First and second cancers shared the same histology in 5 (5%) patients (Table ). The same histologic type was more likely to occur if the first primary tumor was a squamous carcinoma as opposed to an adenocarcinoma ( of 45 patients (7%) and of (57%), respectively). tumors most commonly occurred in the upper lobes and occurred in the lung contralateral to the initial lesion in 5 cases (%) (Table 4). Stage I disease was noted in initial tumors (77%), and no tumor was more advanced than stage IIIA (Table 5). second cancers were somewhat more advanced than initial tumors. Nineteen patients (4%) had stage or IV disease when first seen. Synchronous Carcinoma Squamous cell cancer was also the predominant histologic type in patients with synchronous disease, occurring in 4 (%) of synchronous tumors (see Table ). Two synchronous tumors of the same histologic type were noted in (48%) patients (see Table ). Both squamous cell carcinoma and adenocarcinoma were equally likely to occur as synchronous tumors of the same of different histology (see Table ). Synchronous tumors were found in the same lung in 5 Table. Histologic Classification of Multiple Primary Lung Carcinomas* Classification First Tumor Second Tumor Synchronous Epidermoid 45 (58) (5) 4 () Adenocarcinoma (9) () (4) 9 () 9 () 5 (8) Large cell () () () Small cell () (4) () Total 78 () 78 () () a Numbers in parentheses are percentages.

3 Ann Thorac Surg 99;5:77-9 ROSENGART ET AL 775 Table. Association of Histologic Types Type Synchronous Total Same Histology Epidermoid Adenocarconima Total Different Histology E p i d e r m o i d a Adenocarcinomab Oat cell Large cell Adenocarcinoma Epidermoid Large cell Oat cell Epidermoid Adenocarcinoma Large cell Adenocarcinoma Total a First tumor. Second tumor Table 4. Tumor Location First Tumor Second Tumor LLL LUL L Hilum RLL RML RUL R Hilum Synchronous (n = ) LLL(n = 4) LUL (n = ) RLL(n = 5) RML(n = ) RUL (n = ) (n = 78) LLL(n = 8) LUL (n = 4) RLL (n = ) ML(n = 5) RUL 8... (n = ) RHilum (n = ) Table 5. Staging of Patients With Multiple Primary Lung CarcinomasR Synchronous First Second First Second Stage Tumor Tumor Tumor Tumor Total 78 () 78 () () () a Numbers in parentheses are percentages. patients (7%) (see Table 4). Twelve synchronous tumors (%) were found in the same lobe. The staging of patients with synchronous tumors was similar to that of patients with metachronous disease (see Table 5). Treatment Resection was performed in 7 of the patients (9%) at initial presentation (Table ). Lobectomy was the procedure most commonly performed (7 patients). Lobectomy was combined with wedge resection in patients with synchronous disease, and sleeve resection was performed in patients with first primary tumors. Complete resection was possible in patients (9%) at initial presentation, but in only 54 patients (9%) with metachronous disease. Significantly more patients underwent wedge or segmental resection at reoperation for metachronous disease compared with initial operation (p < Table. Treatment First Second Treatment Tumor Tumor Synchronous Operative Pneumonectomy Bilobectomy Lobectomy Lobectomyiwedge Wedgeisegment Sleeve resection Total operative Nonopera tivea Chemotherapy Radiation Implant External beam Laser ablation Supportive care Total nonoperative a Number in parentheses indicates additional patients receiving nonoperative treatment as adjuvant therapy to operation.

4 ~~ 77 ROSENGART ET AL Ann Thorac Surg 99;5:77%9 Table 7. Limited Resections for Disease Reason Inadequate pulmonary reserve COPD Required pneumonectomy Previous pneumonectomy Presumed metastatic disease Advanced disease Age Unknown Total a Numbers in parentheses are percentages COPD = chronic obstructive pulmonary disease. No. of Patients".). There were four postoperative deaths, yielding an operative mortality of.4%. The reason for limited or incomplete resection was analyzed in 45 patients with metachronous disease (Table 7). The most common reason given was concern regarding pulmonary reserve, noted in patients (%). Four of these patients had previous pneumonectomy, and 5 would have required pneumonectomy for the new lesion. Excluding these patients, initial resection limited subsequent resection in only 7 patients (9%) with metachronous disease. Fourteen patients (%) underwent limited resections because the surgeon at the time of operation believed that the new tumor represented recurrent or metastatic disease. Survival Overall follow-up ranged from to 5 months, with a median of 7 months. Overall 5-year survival for all patients with MPLC from the time of initial diagnosis was % (median survival, 7 months). Ten-year survival was 7%. Survival from the time of initial diagnosis of lung cancer was significantly better (p =.9) for patients in whom metachronous compared with synchronous disease developed (Fig ). Survival for patients with metachronous disease, measured from time of diagnosis of the second tumor, was not significantly different than the survival for patients with synchronous disease (see Fig ). The 5-year survival from the time of initial diagnosis for patients with metachronous and synchronous lesions was 7% and 44%, respectively; -year survival was 4% and %, respectively. Median survival was 84 and 4 months, respectively. Survival after the development of a second, metachronous tumor was % and 9% at 5 and years, respectively, with a median survival of 4 months. When the interval between two metachronous lesions was equal to or greater than 4 months, survival was significantly longer (p <.) than in patients with a shorter disease-free interval (Fig ). Five-year survival from the time of initial diagnosis for late and early metachronous disease was 85% versus %, and -year survival was 5% versus %, respectively. The median survival was months for patients with early metachro- Proportion Surviving Time (mos) Fig. Actuarial survival curves of patients in whom synchronous and metachronous multiple primary lung carcinomas developed. Top and middle curves indicate the survival of patients with metachronous and synchronous disease, respectively, from the time of diagnosis of the first primary tumor. Bottom curve indicates survival for patients with metachronous disease from the time of diagnosis of the second tumor. nous disease, and had not been reached for those with late disease. The stage of either primary cancer did not correlate with overall survival, possibly because of the influence on survival imposed by the second lesion or because of such other factors as disease-free interval. Therefore, the 79 patients with stage I disease were analyzed. The 5-year and -year survival from the time of second diagnosis of lung cancer in 5 patients with either a metachronous or a second synchronous stage I lesion was 8% and %, respectively, with a median survival of 7 months (Fig 4). This was significantly longer (p =.9) than the survival of patients with a stage I or IV second tumor. Median survival in these patients was only 9 months, with 5% 5-year survival. :::IF] ;:"!+/: Time (mod Fig. Actuarial survival curves of patients in whom metachronous disease developed at an interval equal to or greater than 4 months (late) or less than 4 months (early) after the first primary lung tumor. Survival is from the time of diagnosis of initial primary tumor

5 Ann Thorac Surg 99;5:77>9 ROSENGART ET AL 777 Proportion Surviving.9.8 was 4 versus months, respectively. Differences in survival for these two histologic types may in part be related to differences in staging Time (mod Fig 4. Actuarial survival curves of 79 patients with an initial stage tumor as a function of the stage of their second tumor. Patients with stage or stage IV second tumors had significantly shorter survival compared with patients with two stage I tumors. Survival is from the time of diagnosis of the second tumor. Patients undergoing complete resection of the second tumor had a significantly longer survival (p <.) compared with patients with tumors that were not completely resected (Fig 5). Five-year survival for these two groups was 8% and %, respectively; -year survival was % and %, and median survival was months and 9 months. Survival was not significantly different between patients undergoing lobectomy or pneumonectomy compared with those undergoing wedge or segmental resection. Patients with adenocarcinoma as a first tumor, including the large cell and bronchioalveolar subtypes, appeared to survive longer ( p =.) than patients with an initial squamous carcinoma. Five-year survival was 7% and % and median survival was 99 versus 7 months, respectively. Survival differences for patients with adenocarcinoma versus squamous carcinoma as a second primary approached statistical significance (p =.89). Fiveyear survival was 44% and %, and median survival after detection of the second tumor for the two histologic types Proportion Surviving Time (mos) Fig 5. Actuarial survival curves of patients having complete as opposed to incomplete resections of their second tumors, calculated from the time of second diagnosis Comment Survival after appropriate treatment for synchronous or metachronous MPLC is significantly better than that for metastatic or locally recurrent disease. Locally recurrent disease has been reported to develop at a median time interval of months, and results in a % to % -year and a 5% 4-year survival [ll], far lower than the % 5-year survival we now report. Pairolero and associates [] using similar criteria to our own, reported a 5% -year survival for metachronous primary tumors, compared with % and 9% -year survival for local and distant recurrent disease, respectively. A recent report by Deschamps and co-workers [7] of 8 patients treated at the Mayo Clinic also using criteria similar to our own demonstrated equally encouraging survival statistics for patients with second primary lung cancers. These data suggest that patients with MPLC have a more favorable prognosis than patients with locally recurrent or metastatic disease. Therefore, appropriate identification of MPLC is critical. primary cancers appeared greater than years after the first tumor in (5%) of our patients. Survival for patients with early-stage first and second primary lung tumors was significantly better than those with an advanced second lesion, as previously noted [l]. However, advanced disease resulted in limited resection of metachronous disease in % of our patients (see Table 7). These data highlight the need for careful, lifelong screening for MPLC to improve overall survival. Diagnosis Our previously proposed criteria (see Table ) for second primary lung cancer include tumors of the same or different histologic types [8]. Histologic heterogeneity of lung cancers may theoretically make the diagnosis of MPLC based on different histology unreliable [l], but a truly "mixed" histologic pattern of this sort was not encountered in our series. Partly in light of this consideration, however, we have accepted only pathologic tissue specimens, as opposed to cytologic evidence of tumor type, in making final designations. The overall survival and the interval to ocurrence of metachronous tumors was similar whether the metachronous lesion was of the same or different histologic type in our series, and in each case was significantly longer than that expected for recurrent disease. These findings suggest that both these histologic patterns represent multiple primary, rather than recurrent disease. Although the majority of locally recurrent disease will become manifest within the designated -year exclusion interval, extending this interval might be expected to improve the specificity of this criterion [l,. This modification, however, would exclude some patients with metachronous MPLC. Twelve (5%) of our patients with metachronous disease had development of metachronous

6 778 ROSENGART ET AL Ann Thorac Surg 99:577-9 tumors in the -year to -year interval. The median survival of this group was equivalent to that for the patients with disease occurring in the -year to 4-year interval. Because there is no significant dropoff in the survival of patients with development of metachronous disease shortly after years, it seems appropriate to treat all these patients similarly. Patients with synchronous and early metachronous disease did experience significantly reduced survival compared with those with late metachronous disease, similar to results in other published reports [4, 7. The decreased survival noted in these patients may be related to a more aggressive nature of the disease, an increased susceptibility to malignancy, or the actuarial effects imposed by a shortened disease-free interval. The possibility that some patients with recurrent or metastatic disease have been included in these groups cannot be excluded. Treatment The incidence of MPLC can be as high as % to 5% in survivors of or more years after potentially curative resection of primary lung carcinomas. However, the overall incidence of MPLC is only % to %, emphasizing the fact that most patients do not survive long enough to have development of a second primary. To compromise their survival risk by a 5% to 5% incidence of local recurrence in performing wedge resections or segmentectomies, as advocated by some authors [4, 5, therefore seems inappropriate to us. Further, only about one third of patients were unable to tolerate lobectomy for metachronous disease on the basis of pulmonary function (see Table 7). Pulmonary reserve was even less of a consideration if patients with previous pneumonectomy were excluded. Re-resection can almost uniformly be performed safely, as indicated by our and others operative mortality rates [l, 4. The only patient dying of postoperative respiratory failure in our series was also the only patient who had an initial pneumonectomy and a subsequent lobectomy. A significantly greater number of metachronous tumors were treated with limited (wedge, segmental) resections, radiation therapy, chemotherapy, or supportive care compared with the treatment of first primaries. Survival after treatment of the second primary tumors was influenced by a number of factors. In the older patient population (7 years or older) with second primary lung tumors, wedge resections thus seem a reasonable alternative, especially when pulmonary function or medical condition contrain- dicates larger resections. Survival was, however, significantly better in patients undergoing complete resection of their second carcinoma as opposed to an incomplete resection or a nonoperative treatment, and complete resection should be performed whenever possible. Conclusions Follow-up of patients surviving treatment of lung primaries must be indefinite. Complete resection of MPLC is possible and results in favorable survival for these patients. Patients with multiple or recurrent lung masses should not, therefore, be relegated to any treatment other than complete resection because of presumed recurrent or metastatic disease without fully considering the diagnosis of MPLC. References. Van Bodegam PC, Wagenaar SS, Corrin B, et al. Second primary lung cancer: importance of long term follow up. Thorax 989;44:78%9.. Shields T, Humphrey E, Higgins G, et al. Long-term survivors after resection of lung carcinoma. J Thorac Cardiovasc Surg 978;7:494.. Struve-Christensen E. Diagnosis and treatment of bilateral primary bronchogenic carcinoma. J Thorac Cardiovasc Surg 97;: Mathisen DJ, Jensik RJ, Faber P, Kittle CF. Survival following resection for second and third primary lung cancers. J Thorac Cardiovasc Surg 984;88: Jensik RJ, Faber LP, Kittle CF, et al. Survival following resection for second primary bronchogenic carcinoma. J Thorac Cardiovasc Surg 98;8: Smith RA, Nigam BK, Thompson JM. Second primary lung carcinoma. Thorax 97;: Deschamps C, Pairolero PC, Trastek VF, et al. Multiple primary lung cancers: results of surgical treatment. J Thorac Cardiovasc Surg 99;99: Martini N, Melamed MR. Multiple primary lung cancers. J Thorac Cardiovasc Surg 975;:-. 9. Mountain CF. A new international staging system for lung cancer. Chest 98;89:5%S.. World Health Organization. World Health Organization histologic typing of lung tumours. nd ed. Am J Clin Pathol 98; Green N, Kern W. The clinical course and treatment results of patients with postresection locally recurrent lung cancer. Cancer 978;4: Pairolero PC, Williams DE, Bergstralh EJ, et al. Postsurgical stage I bronchogenic carcinoma: morbid implications of recurrent disease. Ann Thorac Surg 9&;8:4. DISCUSSION DR PETER C. PAIROLERO (Rochester, MN): I would like to thank Rosengart and associates for an excellent presentation and for the opportunity to review and discuss the article. No uniform agreement exists as to what constitutes a second primary lung cancer. Distinguishing multicentric lung cancers with absolute certainty either from a lung cancer with pulmonary metastases or from an extrapulmonary primary cancer with lung metastases can be exceedingly difficult and, at times, impossible. Nonetheless, clarification among these three distinct entities is of more than academic interest, inasmuch as each has different therapeutic and survival implications. Clearly, second lung cancers do occur. The criteria set forth by Rosengart and associates for metachronous cancers certainly can be accepted with minimal reservation. The problem arises with

7 Ann Thorac Surg 99;5:77-9 ROSENGART ET AL 779 the criteria for synchronous cancers; different histology seems reasonable, but half of the synchronous cancers in the present study had the same histology. Are these metastases or multiple primary tumors? This latter point, however, may be moot because synchronous neoplasms, whatever their origins, have a very ominous prognosis. In the present series, only % of patients survived years. This is similar to the 5% reported by us previously. Clearly, these patients with synchronous lesions should be classified as stage IV. In view of these grim long-term results for synchronous cancers, I would like to ask three questions. Should these patients be better staged before thoracotomy? What then are the indications for both mediastinoscopy and transthoracic needle aspiration? Finally, should patients with known synchronous lesions undergo pulmonary resection? In contrast to synchronous cancer, -year survival for patients with metachronous neoplasms was 44% and, not unexpectedly, was better for those patients with stage I disease and a long tumor-free interval. We agree with Rosengart and associates conclusion that aggressive follow-up of all patients with lung cancer is mandatory. When recurrent lung cancers do develop, strong consideration should be given to further surgical intervention if there is no evidence of extrathoracic cancer and if the patient is otherwise in good health. Pulmonary resection, conserving as much lung parenchyma as possible, is safe and offers the greatest possibility of long-term survival. DR ROSENGART: In answer to your first question, we agree with you and have looked at your report extensively. We noted that the patients whose synchronous tumors had different histology performed exactly the same in terms of survival, at least in our series, as patients with the same histology. Therefore, although obviously it is more difficult to classify two tumors of the same histology as multiple primary tumors, the fact that their survival characteristics were the same as those of patients with different histology leads us to believe that our criteria are acceptable in terms of distinguishing a distinct population. In terms of how we evaluate these patients, we proceed as we would for any primary branch gene carcinoma, which would include a computed tomographic scan of the chest. With a negative mediastinum, we would proceed as if we were dealing with primary tumors until proven otherwise. A caveat is that we do not significantly use cytologic diagnosis because of some concern for histologic variability, and we would really make the diagnosis based on pathology as opposed to cytologic diagnosis. DR TOM R. DEMEESTER (Los Angeles, CA): The management of a synchronous tumor has occupied our interest as well [l]. I wish you would have limited your synchronous tumors to those without lymph node metastases, because the presence of lymph node metastases is a strong indication that the lesion is metastatic. Have you made any effort to look at your survival of synchronous tumors when there was no lymph node involvement? DR ROSENGART: Yes, we did look at those data, and essentially all of our patients with synchronous tumors had no lymphatic involvement, as this was one of our criteria. We have not specifically looked at staging, I do not have that information for you, sir. DR PAUL A. KIRSCHNER (New York, NY): I enjoyed this paper very much. I have one question. Were there any synchronous tumors found in the resected specimen that you did not detect preoperatively? A recent paper from British Colombia [] reported that 9% of lobectomy specimens of adenocarcinoma were found to contain additional tumors on serial gross sections of the inflated fixed lobes, some as small as to mm. DR ROSENGART. The pathologic determination was not done with that in mind specifically, and we had no study protocol designed when our pathologists were looking at the specimens; we did not find any unrpcognized or unsuspected second nonsynchronous tumors. References. Ferguson MK, DeMeester TR, DesLauriers J, Little AG, Piraux M, Golomb H. Diagnosis and management of synchronous lung cancers. J Thorac Cardiovasc Surg 985;89: Miller RB, Nelems 8, Evans KG, et al. Glandular neoplasia of the lung. Cancer 988;:9-4. INVITED COMMENTARY The discussants have made the following points: () patients with synchronous lesions should be classified as having stage IV cancers, and pulmonary resections that may be undertaken should conserve as much lung tissue as possible; () the definition of synchronous cancers should require that there be no lymph node metastases; and () multicentric foci of bronchogenic cancers may occur in about % of patients. Although Rosengart and associates reported a remarkable 7% 5-year survival rate for metachronous multiple cancers and a 44% 5-year survival rate among patients with synchronous cancers, these findings may not be representative of experience elsewhere. Until there are more data, the points that were made by the discussants speak in favor of continuing relatively conservative, tempered, and selective indications for resections of multiplesite lung cancers. John R. Benfield, MD University of California, Davis Professional Building 4 X St Sacramento, CA 9587