Colon cancer is still one of the leading causes of

Long-Term Survival After Repeated Resection of Pulmonary Metastases From Colorectal Cancer Stefan Welter, MD, Jan Jacobs, MD, Thomas Krbek, MD, Bettina Krebs, MD, and Georgios Stamatis, MD Department of Thoracic Surgery, Ruhrlandklinik, Essen, Germany Background. The purpose of this study was to evaluate the outcome of patients undergoing repeated resection of pulmonary metastases from colorectal cancer and specify factors promising long-term survival. Methods. From January 1993 to December 2003, 175 patients were diagnosed and resected for pulmonary metastases of colorectal cancer. Follow-up information was collected for 169 patients, and 33 (19.5%) had had recurrent metastasectomies up to three times. Their follow-up information was updated in August 2006. The first repeated resection was performed for up to six bilateral metastases, the second and third metastasectomies were each unilateral and for a single metastasis only. Lymph node involvement was present in 5 patients who underwent repeat resections. Results. The overall (n 169) median survival was 47.2 months after the first metastasectomy. The 33 patients with repeated resections had a median survival of 72.6 months, with survival of 53.8% at 5 years and 20.6% at 10 years. After reoperation, age, sex, primary tumor stage, preoperative carcinoembryonic antigen, disease-free interval, prior resection of liver metastases, and lymph node involvement were not found to be of prognostic importance. The only factor that significantly influenced survival was the number of metastases (hazard risk, 1.299). Perioperative mortality even for repeated resections was 0%. Conclusions. Repeated resection of pulmonary metastases secondary to colorectal cancer is safe and can provide long-term survival for highly selected patients. (Ann Thorac Surg 2007;84:203 10) 2007 by The Society of Thoracic Surgeons Colon cancer is still one of the leading causes of cancer death in Western countries. After the initial diagnosis, more than two thirds of these patients undergo primary curative resection. Nonetheless, more than half of the resected patients eventually die from the underlying disease [1]. Tumor recurrence is usually due to distant metastases [2]. Patients with untreated metastatic disease have a 5-year survival of less than 5% [3]. But despite stage IV disease, some highly selected patients with only a few pulmonary metastases can undergo surgical metastasectomy and still have a curative option with a 5-year survival rate up to 60%. The results of 5206 cases assessed in the International Registry of Lung Metastases showed that lung metastasectomy in general is a safe and potentially curative procedure [4]. To date, very limited, but encouraging information about repeated pulmonary metastasectomy is available [5 9]. Depending on tumor size and location, number of metastases and performance status, repeated surgical resection can be offered, and even long-term survival is possible. In this retrospective study we reviewed our experience with recurrent pulmonary metastases from colorectal cancer and focused on the prognostic impact of repeated surgical interventions. Accepted for publication March 12, 2007. Address correspondence to Dr Welter, Department of Thoracic Surgery, Ruhrlandklinik, Tüschener Weg 40, Essen, 45239, Germany; e-mail: stefan.we@t-online.de. Material and Methods Our hospital Ethics Committee approved this study and waived the need for individual consent. From January 1993 to December 2003, 175 patients with colorectal lung metastases were resected in our hospital, and 169 patients were available for further examination (Table 1). Six were lost to follow-up after discharge and were excluded from this study; none had a recurrent procedure. As of August 2006, 33 (19.5%) of the 169 former identified patients had at least one repeated thoracotomy for removal of recurrent metastases. Data Collection Preoperative, operative, and pathologic data were collected by review of medical records. Sequential operations on both lungs were counted as one resection with the date of first resection, and the number of metastases from each side was added. Follow-up information included date of death, tumor recurrence, and location and was gathered by written inquiries to the general practitioner, the oncologists, or (if otherwise impossible) the registry offices. The 33 patients with repeated operative procedures were followed up until August 2006. Three were still alive in June 2004, but could not be contacted in August 2006. Twenty-four patients died: 2 from comorbid disease, and the remaining 22 died from tumor recurrence. All had some sort of palliative chemotherapy before they died, but no patient had adjuvant chemotherapy or radiation directly after 2007 by The Society of Thoracic Surgeons 0003-4975/07/$32.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2007.03.028

204 WELTER ET AL Ann Thorac Surg REPEATED RESECTION OF METASTASES 2007;84:203 10 Table 1. Patient Characteristics (n 169) Characteristic Patients, n (%) Age, mean (range) 63.5 (34.1 81.5) Sex Male 96 (56.8) Female 73 (43.2) Primary tumor site Colon 73 (43.2) Rectum 96 (56.8) Prior resection of liver metastases Yes 37 (21.9) Number of metastases 1 81 (48) 2 4 61 (36.1) 5 27 (15.9) Maximum tumor size 5 cm 150 (88.8) 5 cm 19 (11.2) Tumor recurrence Pulmonary only 39 (23.1) None 38 (22.5) Pulmonary others 34 (20.1) Others 50 (29.6) Unknown 8 (4.7) Major complications (268 operations) 41 (15.3) Arrhythmias 8 Pneumonia 9 Pulmonary embolism 1 Prolonged ventilation 2 Bronchopleural fistula 1 Empyema 4 Gastric ulcer bleeding 2 Others 14 the removal of pulmonary metastases. Nine patients were alive, and 6 (18.2%) were free of tumor. Staging The preoperative staging in our hospital consisted of abdominal ultrasound, computed tomography (CT) scan of the thorax and upper abdomen, fiberoptic bronchoscopy, and clinical examination. Brain and bone scans were only done if metastases were clinically suspected. Regular brain and bone scans were completed before repeated surgical interventions. Barium enema or colonoscopy to rule out local recurrence was accepted if not older than 6 months. As far as possible, wedge resection or segmentectomy was the procedure of choice; when necessary, pneumonectomy or completion pneumonectomy was performed. Lymph nodes suggestive of disease were sampled. After the first resection of metastases, the routine follow-up consisted of a chest roentgenogram, an ultrasound scan of the abdomen, and blood tests done by the oncologist every 3 months for 2 years and every 6 months thereafter. A yearly CT scan of the chest was advised. Definitions Initial TNM stage was extracted from external medical reports. All histologic reports of pulmonary metastases were reviewed and confirmed adenocarcinoma (G2) of intestinal origin. Preoperative selection criteria were completely resectable lesions in the lung, not more than 10 seen on CT scan, no evidence of metastases other than in the lung, resected primary colorectal tumor, no suspicion of mediastinal lymph node metastases, good performance score (World Health Organization, 0 to 2), sufficient lung function, and no contraindication due to concomitant disease. Patients with excessive pleural scarring or more than five metastases on one side were not accepted for repeated resection. If new nodules had evolved a minimum of 6 months after the last procedure, than reoperation was defined as repeated operation. An exception was a patient included from 1993 who had bilateral metastasectomy and one new metastasis only 2 months later, which was resected immediately. The pathologist s statement about microscopic resection margins were extracted for patients who had repeated operations, and the number of local recurrences was counted. The maximum resection (wedge, segment, lobe, lung) in one operation was counted as a type of resection, even if further wedge or segmental resections were performed. In case of multiple metastases in 1 patient, only the largest diameter was counted. Locoregional recurrence was defined as recurrent tumor in the same lobe. Table 2. Characteristics of 33 Patients Undergoing Repeated Resections Characteristic Patient, n (%) Age, mean (range) 62.3 (39.5 79.6) Sex Male 20 (60.6) Female 13 (39.4) Primary tumor site Colon 15 (45.5) Rectum 18 (54.5) Primary tumor stage a I 4 (12.9) II 7 (22.5) III 10 (32.3) IV 10 (32.3) DFI, median months (range) DFI 1 27.0 (0 106.3) DFI 2 16.8 (2.0 38.2) DFI 3 15.2 (6.9 20.8) Prior resection of liver metastases Yes 11 No 22 a Data for 31 patients, 2 were unknown. DFI disease-free interval.

Ann Thorac Surg WELTER ET AL 2007;84:203 10 REPEATED RESECTION OF METASTASES 205 Operative Results Of the 33 patients with repeated surgical treatment, 20 were men (60.6%) and 13 were women (39.4%), with a median age of 63.8 years (range, 39.5 to 79.6) years. The initial tumor site was the colon in 15 (45.5%) and the rectum in 18 (54.5%). The initial tumor stage was stage I in 4 patients, II in 7, III in 10, and IV in 10, and 2 were unknown, according to International Union Against Cancer Classification (1997). The median disease-free interval (DFI) in the 33 patients from colorectal resection to pulmonary metastasectomy (DFI 1) was 27.0 months (range, 0 to 106 months). The median DFI from first to first repeated (DFI 2) and from first repeated to second repeated (DFI 3) metastasectomy was 16.8 months (range, 2.0 to 38.2 months) and 15.2 months (range, 6.9 to 20.8 months). Of the 33 patients, 11 (33.3%) had a complete resection of liver metastases before repeated pulmonary metastasectomy (Table 2). All together, 155 metastases had been removed with up to seven in a sequential bilateral thoracotomy. At the time of first lung resection, 15 patients had a single metastasis and 18 had multiple. At the time of first repeated lung resection, 15 patients had a single metastasis, and 18 had multiple (2 to 6). Six patients with a second and 1 patient with a third repeated metastasectomy had one lesion each. Locoregional recurrence was found in 22 of 33 patients during first repeated resection, 4 of 7 during second, and 1 of 1 during the third repeated resection. Table 3. Treatment and Pathologic Results of 33 Patients With Repeated Interventions Repeated Metastasectomy Fig 1. Kaplan-Meier probability of survival of 33 patients after the first metastasectomy. Median survival time was 72.6 months, and the survival rate was 53.8% at 5 years and 20.6% at 10 years. The pathologist described microscopically tumorfree resection margins concerning all specimens in 22 of 33 patients after the first metastasectomy, in 29 of 33 after the first repeated metastasectomy, in 4 of 7 after the second, and 1 out of 1 after the third repeated metastasectomy. During the first repeated resection, eight local recurrences were found where microscopically incomplete resection was found before. No patient who had microscopically incomplete resection Variable First Second Third (n 7) Fourth (n 1) Main resection a Wedge 29 14 5 Segment 13 7 1 Lobe 8 15 1 Pneumonectomy 1 1 Total 50 37 7 1 Number of metastases 1 15 15 7 1 2 4 12 16 5 10 6 2 Maximum tumor size 5 cm 31 32 7 1 5 cm 2 1 Lymph node involvement Pulmonary/hilar 2 4 1 Mediastinal 1 Resection margins R0 (all sites) 22 29 4 1 R1 ( 1 site) 11 4 3 Locoregional recurrence 22 4 1 Recurrence R1 site 8 0 a The main procedure of bilateral resections were counted separately. Fig 2. Kaplan-Meier probability of survival in 33 patients after the first repeated metastasectomy. Median survival time was 49.0 months, and the survival rate was 37.1% at 5 years and 13.8% at 10 years.

206 WELTER ET AL Ann Thorac Surg REPEATED RESECTION OF METASTASES 2007;84:203 10 Fig 3. Kaplan-Meier probability of survival in 7 patients after the second repeated metastasectomy. Median survival time was 54.6 months. after the second repeated metastasectomy was operated on a third time. The largest diameter of the first resected metastases was 0.7 to 8.0 cm according to the pathologist s measurement and was 0.7 to 5.5 cm in the repeated resections. At the first procedures, 50 unilateral and bilateral operations were performed in 33 patients, consisting of 29 wedge resections, 13 segmentectomies, and eight lobar resections as the main procedure. The next 37 first repeated operations consisted of 14 wedge, seven segmental, 15 lobar resections, and one pneumonectomy, including four bilateral interventions. These were again followed by seven second repeated resections on one side consisting of five wedge resections, one segmental, and one lobar resection, and again were followed by one completion pneumonectomy as the third repeated metastasectomy (Table 3). The surgical access was a muscle-saving anterioraxillary thoracotomy that allowed palpation of the whole lung in all patients, even those undergoing repeated operations. Two patients had incomplete removal of lung metastases after the first repeated procedure, and one after the second repeated thoracotomy; the latter had intrapulmonary lymph node involvement as well. Operative mortality included patients who died within 30 days after thoracotomy or during the same hospital stay and was 0%. The rate of major complications for all repeated operative treatments in the 33 patients was 22.0% (9/41). The latter included 4 patients with prolonged air leak, and 1 patient each with relevant arrhythmias, pneumonia, empyema, recurrent nerve palsy, and others. Lymph nodes suggestive for disease were sampled. Interlobar, hilar, or mediastinal lymph nodes were removed in 23 of 33 primary lung interventions and in 25 of 33, two of seven, and one of one first, second, and third repeated metastasectomies. The pathologist described up to 10 lymph nodes in 1 patient in the first, and up to 10, eight, and four lymph nodes in the first, second, and third repeated resection specimen. Lymph node metastases were found in 3 of 33 patients during the first intervention, and in 5 of 33 during one of the recurrent thoracotomies. Statistical Analysis The data were analyzed using the program SPSS 11.5 (SPSS Inc, Chicago, IL) for Windows (Microsoft, Redmond, WA). Probability of survival was calculated according to the method of Kaplan-Meier [10] using the date of the first pulmonary resection and the date of last follow-up or death of the patient as start and end points. The prognostic influence of variables on survival was analyzed using the log-rank test [11]. Cox proportional hazards model [12] was used to perform a multivariate analysis of prognostic factors. Variables with a value of p 0.15 were included into the stepwise regression model. A p 0.05 was considered significant. Stepwise regressions of parameters due to highest value of p were achieved with the likelihood quotient method (LR method). Results All pulmonary resections were performed in the Department of Thoracic Surgery at Ruhrlandklinik Essen. The initial colonic resections were all done in other hospitals. Fig 4. Kaplan-Meier probability of survival in 33 patients after the first metastasectomy, depending on the number of metastases. The survival difference between patients with 1 (thick black line), 2 to 4 (grey line) and 5 to 10 (thin black line) metastases, removed at the first metastasectomy, was statistically significant (p 0.0249 and p 0.0096).

Ann Thorac Surg WELTER ET AL 2007;84:203 10 REPEATED RESECTION OF METASTASES 207 Table 4. Univariate Analysis of Prognostic Factors (Log-Rank Test) First Metastasectomy First Repeated Metastasectomy Variable MST 95% CI Log-Rank p Value MST 95% CI Log-Rank p Value Age 63 49.6 39.8 59.4 33.5 8.1 59.0 63 75.2 53.0 97.3 2.23 0.135 55.5 30.0 81.1 2.96 0.085 Sex M 71.8 45.7 97.9 48.8 22.3 75.2 F 77.5 35.6 119.4 0.10 0.752 50.3 30.7 69.9 0.13 0.716 PTS I 95.3 59.0 131.7 65.5 38.7 92.2 II 71.8 38.3 105.4 2.44 0.118 55.5 42.1 68.9 2.44 0.118 III 46.6 0.0 105.3 0.23 0.630 21.5 0.0 67.6 0.21 0.650 IV 51.9 46.9 57.0 0.05 0.815 38.3 32.0 44.7 0.00 0.966 DFI a 24 51.9 40.9 63.0 24 79.3 68.1 90.6 1.54 0.215 NOM 1 95.3 49.0 141.6 59.6 40.3 79.0 2 4 51.4 47.4 55.4 5.03 0.025 b 33.5 17.6 49.5 2.63 0.105 5 10 33.7 14.3 53.0 6.71 0.009 b 43.1 0.18 0.668 LM No 72.6 51.9 93.4 55.5 42.4 68.7 Yes 51.9 31.2 72.6 0.24 0.624 38.3 25.3 51.3 0.10 0.756 LNI No 72.6 27.0 118.3 49.0 9.0 89.0 Yes 71.8 47.9 95.7 0.46 0.497 55.5 33.5 77.5 0.81 0.368 a DFI 24 was not calculated for repeated resections. b Significant at p 0.05. CI confidence interval; DFI disease-free interval; LM liver metastases; LNI lymph node involvement; MST median survival time; NOM number of metastases; PTS primary tumor stage. Survival The actuarial overall median survival (n 169) was 47.2 months (95% confidence interval, 39.8 to 54.5 months). Overall survival was 39.1% at 5 years and 20.0% at 10. After the first pulmonary metastasectomy, only 38 patients were free of tumor in June 2004, and 8 patients were unknown. A total of 123 (75.7%) patients faced tumor recurrence, of which 39 had intrapulmonary recurrence only. Six of them proved to be inoperable, and 33 had recurrent operations. Patients With Repeated Resections The median survival of the 33 patients who had repeated resections was 72.6 months, with a corresponding overall survival rate of 53.8% after 5 years and 20.6% after 10. The median survival after first and second repeated metastasectomy was 49.0 and 54.6 months, respectively (Figs 1, 2, and 3). The 2 patients with completion pneumonectomy survived 30 and 74 months but both died of tumor recurrence in the remaining lung. No significant age or sex related mortality was found. Only 17 of 33 patients had a carcinoembryonic antigen investigation before resection, and 5 were elevated with no significant impact on prognosis (p 0.49). Diseasefree interval from colon resection to first, second, and third metastasectomy, primary tumor stage, and lymph node involvement was not found to be of prognostic significance in the univariate analysis as well. The 5-year and 10-year survival of 11 patients with liver metastases was 42.4% and 21.2% compared with 59.1% and 21.5% for those without liver involvement; the difference was not significant (p 0.62). The number of metastases was the only strong predictor of survival in the univariate analysis (Fig 4). The survival difference between patients with a single and two to four metastases was significant (p 0.025) after the first metastasectomy and not significant (p 0.10) after the first repeated resection (Table 4). The 5-year survival of patients with one metastasis at the primary intervention was 79.0% compared with 33.3% for those with more than one lesion. No patient with more than one lesion at the initial pulmonary operation survived 10 years, despite repeated resections. A multivariate analysis of independent prognostic factors was done after the first and the first repeated metastasectomy using the Cox regression method. Covariables without significant values were stepwise excluded. Beginning with age, sex, primary tumor stage,

208 WELTER ET AL Ann Thorac Surg REPEATED RESECTION OF METASTASES 2007;84:203 10 Table 5. Multivariate Analysis of Prognostic Factors Covariable Estimator (B) Exp (B) (95% CI) disease free interval, number of metastases, size of metastases, preoperative liver metastases, and lymph node involvement, six iteration steps left three independent variables as statistically significant prognostic factors: disease free interval (hazard risk, 0.981), number of metastases (hazard risk, 1.299) and age (hazard risk, 0.933; Table 5). The same iteration procedure was performed after the first repeated removal of metastases, which identified the number of metastases (hazard risk, 1.276) as the only independent prognostic factor (Table 5). These multivariate results indicate that assuming that all other covariables are constant, the prolongation of the DFI by 1 month reduces the hazard by 1.9%, which means 20.6% per year. The patient with one more metastasis (n 1) has a 1.299 (29.9%) increased risk to die. And every additional year of life before metastasectomy reduces the risk to die by 6.7%. Comment p Value First pulmonary metastasectomy DFI 0.019 0.981 (0.964 0.998) 0.032 NOM 0.262 1.299 (1.028 1.642) 0.028 Age 0.07 0.933 (0.869 1.001) 0.053 First repeated pulmonary metastasectomy NOM 0.244 1.276 (1.057 1.541) 0.011 CI confidence interval; DFI disease-free interval; NOM number of metastases. This investigation was especially focused on the question of whether repeated removal of lung metastases from colorectal cancer has a beneficial effect on survival. To date, resection of pulmonary metastases from colorectal cancer becomes an increasingly accepted method in the treatment of this cancer. Very limited but encouraging information is available about repeated surgical interventions for recurrent metastases from various primaries, yielding 5-year survival rates from 30.6% to 57% [5 9, 13, 14] and 53.8% in our study. Survival curves were found to be equal between first and first repeated metastasectomy for osteosarcomas [8], soft-tissue sarcoma [14], colorectal cancer [9, 15], and various tumor types [6, 7] in other series. We treated 33 patients for recurrent colorectal lung metastases with a repeated resection and found a 53.8% 5-year survival rate, which is almost equal to the results (52.1%) of Saito and colleagues [15]. Selection Criteria Because regular cancer follow-up is in the hands of oncologists, pneumologists, and general practitioners, only a highly selected group of repeated resection candidates were seen. A real weakness in this series is the lack of a formalized follow-up. We do not know how many patients might have been resectable a second or even third time if a thoracic surgeon had seen them routinely. From 169 patients, 33 (19.5%) had repeated procedures. This is even more than the 15% from the International Registry data [4] and the 14.6% described by Jaklitsch and colleagues [7], where a prospective follow-up after primary metastasectomy was established in the same department. Accepted inclusion criteria are the surgical and functional resectability and the exclusion of other distant organ metastases. The number of expected metastases did not influence the decision to perform a repeated metastasectomy in the reports of Kandioler and colleagues [6] and Jaklitsch and colleagues [7]. We did not perform a second or third repeated metastasectomy in case of more than one single lesion. Taking into account the results of this study that median survival after the second repeated metastasectomy was 54 months, this restriction does not seem to be necessary. Eleven initial microscopically incomplete resections caused eight local recurrences in the same lobe, whereas 14 were local recurrences after initially clear margins. Only three R1 resections did not lead to a local recurrence. This fact underlines the necessity for wide resection margins. Age should not be a selection criterion: operative mortality is low regardless of age, 0% in our series and up to 3% [5]. Repeated resections had a slightly elevated rate of major complications (22.0%) compared with the rate (15.3%) of first metastasectomy in 169 patients. Operative Access From the beginning, our policy was to perform open procedures, thus allowing the palpation of the whole lung for every metastasectomy. Other authors [4, 16 18] reported that up to 30% of metastases are missed on CT scan, and the success rate to detect all nodules by video assisted thoracoscopic surgery was only 22% [18]. Others even performed repeated resections with video assistance [7, 13] and could show excellent survival rates. Prognostic Factors AGE. In the 33 patients who had repeated resections, we found a survival advantage for the elderly in the multivariate analysis. We have not found an explanation for this in the literature on pulmonary metastasectomy and we cannot provide an explanation from our data. A similar age-related survival difference in stage I non-small cell lung cancer was found by Cerfolio and Bryant [19], who compared elderly patients with a younger control group. The explanation was given in the greater percentage of less aggressive, well-differentiated tumors in the older age group and the detection of earlier tumor stages. DISEASE-FREE INTERVAL. The DFI is a prognostic factor of low importance in our study. In the group with repeated resections, we found the DFI ( 24 months) from colorec-

Ann Thorac Surg WELTER ET AL 2007;84:203 10 REPEATED RESECTION OF METASTASES 209 tal resection to first metastasectomy (DFI 1) to be significantly (p 0.032) correlated with a better prognosis, associated with a hazard risk of 0.981 for every additional month without tumor recurrence. There is little doubt in other series that synchronous metastases have a minor prognosis and a long DFI from primary tumor resection to first metastasectomy is superior [4, 8, 13, 14, 20]. The International Registry [4] showed the best prognostic group being those with a single resectable metastasis and a DFI of more than 3 years with no other risk factor. Furthermore, resectability is most important. As Jaklitsch and colleagues [7] could show, prognosis is good (5-year survival, 33% to 59%) as long as repeated resections can provide local control. Once local control had been lost, further survival fell dramatically to a median of 8 months, regardless of the number of previous procedures. LIVER METASTASES. In our series, as well as in others [15, 20, 21, 22], resected liver metastases were not associated with a decreased survival. Patients in our series with repeated resections and removed liver metastases had a 5-year survival of 42.4% compared with 59.1% of those without liver involvement (p 0.624). NUMBER OF METASTASES. The number of metastases is the most important predictor of survival [5, 14, 20, 22 24]. The results of 5206 cases of the International Registry [4], including various histologies, showed a survival of 43% at 5 years for single metastasis and 27% for more than four metastases. Even in repeated resections, the number of metastases is most important. We found a median survival of 95.3, 51.4, and 33.7 months in the groups with one, two to four, and more than 4 metastases. The difference was statistically significant (p 0.025 and p 0.009). LYMPH NODE INVOLVEMENT. We did not find a prognostic influence of lymph node metastases in the 33 patients with repeated pulmonary resections. The number of cases is probably too small to make further statements. Lymph node involvement was not investigated in the other series with repeated carcinoma resections [6, 7, 13]. Completion pneumonectomy (CP) for recurrent pulmonary metastases is still controversial. We performed two CPs. Survival was 74 months and 30 months after CP, and both patients died of tumor recurrence. Much worse results for CP were found by Grunenwald and colleagues [25] in a 1997 series with a median survival of 6 months in of 12 patients with pulmonary metastases and by Jungraithmayr and colleagues [26] in a series with a 3-year survival rate of 34% in 9 patients. Both authors did not see a survival benefit for CP. Better results were found by Koong and colleagues [27] in 1999, in which 38 patients had a CP with an operative mortality of 3% and a 5-year survival of 30%. CHEMOTHERAPY. Finally, chemotherapy has to be discussed. A comparison is very difficult because we have not found a trial that differentiates resectable from irresectable lung involvement before chemotherapy, and nearly all series contain different sites of metastases, not only restricted to the lung [28 30]. Even the best prognostic groups (one tumor site, good performance status) had a median survival of only 15 to 19 months [28, 31, 32]. The addition of bevacizumab to fluorouracil and irinotecan (in a randomized controlled study with 923 patients) can prolong median survival up to 20 months in metastatic disease [30]. The objective response rate was 44.8% including only 3.7% complete responses, 2.8% of patients died from toxicity, and about 50% of patients had further second-line and third-line chemotherapy. In this context, the operative results appear to be excellent, but a direct comparison is not possible. All surgical candidates were highly selected and had some form of chemotherapy in the course of their disease. Conclusion Repeated resection of pulmonary metastases in colorectal cancer is indicated in some highly selected patients. If other organ involvement can be ruled out, repeated thoracotomy can be offered with a low mortality and slightly elevated morbidity. In the patient with one or few recurrent operable metastases, chemotherapy does not offer better results. Some patients can be cured, and long-term survival is possible. References 1. Rosen M, Chan L, Beart RW, Vukasin P, Anthone G. Follow-up of colorectal cancer: a meta-analysis. Dis Colon Rectum 1998;41:1116 26. 2. Hermanek P, Wiebelt H, Riedel S, Staimmer D, Hermanek P. Long term results of surgical therapy of colon cancer. Results of the Colorectal Cancer Study Group. Chirurg 1994;65:287 97. 3. Poon MA, O Connell MJ, Moertel CG, et al. Biochemical modulation of fluorouracil: evidence of significant improvement of survival and quality of life in patients with advanced colorectal carcinoma. J Clin Oncol 1989;7:1407 18. 4. Pastorino U, Buyse M, Friedel G, et al. Long-term results of lung metastasectomy: prognostic analysis based on 5206 cases. J Thorac Cardiovasc Surg 1997;113:37 49. 5. McAfee MK, Allen MS, Trastek VF, Ilstrup DM, Deschamps C, Pairolero PC. Colorectal lung metastases: results of surgical excision. Ann Thorac Surg 1992;53:780 6. 6. Kandioler D, Krömer E, Tüchler H, et al. Long-term results after repeated surgical removal of pulmonary metastases. Ann Thorac Surg 1998;65:909 12. 7. Jaklitsch MT, Mery CM, Lukanich JM, et al. Sequential thoracic metastasectomy prolongs survival by reestablishing local control within the chest. J Thorac Cardiovasc Surg 2001;121:657 67. 8. Briccoli A, Rocca M, Salone M, et al. Resection of recurrent pulmonary metastases in patients with osteosarcoma. Cancer 2005;104:1721 5. 9. Ogata Y, Matono K, Hayashi A, et al. Repeat pulmonary resection for isolated recurrent lung metastases yields results comparable to those after first pulmonary resection in colorectal cancer. World J Surg 2005;29:363 8. 10. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Soc 1958;53:457 81. 11. Peto R, Peto J. Asymptotically efficient rank and invariant procedures. J R Stat Soc 1972;135:185 207. 12. Cox DR. Regression models and life-tables. J R Stat Soc (Ser B Stat Methodol) 1972;34:187 220. 13. Groeger AM, Kandioler D, Mueller MR, End A, Eckersberger F, Wolner E. Survival after surgical treatment of recurrent pulmonary metastases. Eur J Cardiothorac Surg 1997;12:703 5. 14. Weiser MR, Downey RJ, Leung DH, Brennan MF. Repeat resection of pulmonary metastases in patients with softtissue sarcoma. J Am Coll Surg 2000;191:184 90.

210 WELTER ET AL Ann Thorac Surg REPEATED RESECTION OF METASTASES 2007;84:203 10 15. Saito Y, Omiya H, Kohno K, et al. Pulmonary metastasectomy for 165 patients with colorectal carcinoma: a prognostic assessment. J Thorac Cardiovasc Surg 2002;124:1007 13. 16. Loehe F, Kobinger S, Hatz RA, Helmberger T, Loehrs U, Fuerst H. Value of systematic mediastinal lymph node dissection during pulmonary metastasectomy. Ann Thorac Surg 2001;72:225 9. 17. McCormack PM, Ginsberg KB, Bains MS, et al. Accuracy of lung imaging in metastases with implications for the role of thoracoscopy. Ann Thorc Surg 1993;56:863 6. 18. McCormack PM, Bains MS, Begg CB, et al. Role of videoassisted thoracic surgery in the treatment of pulmonary metastases: results of a prospective trial. Ann Thorac Surg 1996;62:213 17. 19. Cerfolio RJ, Bryant AS. Survival and outcomes of pulmonary resection for non-small cell lung cancer in the elderly: a nested case-control study. Ann Thorac Surg 2006;82:424 30. 20. Kobayashi K, Kawamura M, Ishihara T. Surgical treatment for both pulmonary and hepatic metastases from colorectal cancer. J Thorac Cardiovasc Surg 1999;118:1090 6. 21. Ike H, Shimada H, Togo S, Yamaguchi S, Ichikawa Y, Tanaka K. Sequential resection of lung metastasis following partial hepatectomy for colorectal cancer. Br J Surg 2002;89:1164 8. 22. Regnard JF, Grunenwald D, Spaggiari L, et al. Surgical treatment of hepatic and pulmonary metastases from colorectal cancers. Ann Thorac Surg 1998;66:214 9. 23. Okumura S, Kondo H, Tsuboi M, et al. Pulmonary resection for metastatic colorectal cancer: experiences with 159 patients. J Thorac Cardiovasc Surg 1996;112:867 74. 24. Pfannschmidt J, Muley T, Hoffmann H, Dienemann H. Prognostic factors and survival after complete resection of pulmonary metastases from colorectal carcinoma: experiences in 167 patients. J Thorac Cardiovasc Surg 2003;126:732 9. 25. Grunenwald D, Spaggiari L, Girard P, Baldeyrou P, Filaire M, Dennewald G. Completion pneumonectomy for lung metastases: is it justified? Eur J Cardiothorac Surg 1997;12: 694 7. 26. Jungraithmayr W, Hasse J, Stoelben E. Completion pneumonectomy for lung metastases. Eur J Surg Oncol 2004;30: 1113 7. 27. Koong HN, Pastorino U, Ginsberg RJ. Is there a role for pneumonectomy in pulmonary metastases? Ann Thorac Surg 1999;68:2039 43. 28. Kohne CH, Cunningham D, Di CF, et al. Clinical determinants of survival in patients with 5-fluorouracil-based treatment for metastatic colorectal cancer: results of a multivariate analysis of 3825 patients. Ann Oncol 2002;13:308 17. 29. Saltz LB, Cox JV, Blanke C, et al. Irinotecan plus fluorouracil and leucovorin for metastatic colorectal cancer. N Engl J Med 2000;343:905 14. 30. Hurwitz H, Fehrenbacher L, Novotny W, et al. Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med 2004;350:2335 42. 31. De Gramont A, Figer A, Seymour M, et al. Leucovorin and fluorouracil with or without oxaliplatin as first-line treatment in advanced colorectal cancer. J Clin Oncol 2000;18: 2938 47. 32. Giacchetti S, Perpoint B, Zidani R, et al. Phase III multicenter randomized trial of Oxaliplatin added to chronomodulated fluorouracil-leucovorin as first-line treatment of metastatic colorectal cancer. J Clin Oncol 2000;18:136 47.