Salvage Lung Resection After Definitive Radiation (>59 Gy) for Non-Small Cell Lung Cancer: Surgical and Oncologic Outcomes

Transcription

1 Salvage Lung Resection After Definitive Radiation (>59 Gy) for Non-Small Cell Lung Cancer: Surgical and Oncologic Outcomes Julie E. Bauman, MD, Michael S. Mulligan, MD, Renato G. Martins, MD, Brenda F. Kurland, PhD, Keith D. Eaton, MD, and Douglas E. Wood, MD Departments of Medical Oncology and Surgery, University of Washington, and Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, Washington Background. Isolated local relapse occurs in 24% to 35% of patients after definitive chemoradiation for locally advanced non-small cell lung cancer. Although originally considered inoperable, select patients are referred for surgical salvage. We describe a series of salvage lung resection after curative-intent radiation. Methods. Twenty-four consecutive patients from 1997 to 2005 were identified retrospectively. Medical records reviewed. Patients were grouped by surgical indication: A, obvious relapse by computed tomography (CT), 7 patients; B, abnormal fluorodeoxyglucose-positron emission tomography (FDG-PET), 12; C, delayed conversion to trimodality, 4; and D, chronic bronchopleural fistula, 1. Results. All patients received definitive radiation (median, 63.9 Gray), 22 with concurrent chemotherapy. Original staging included cardiothoracic surgical consultation in 4. Median time from radiation to resection was 21 weeks. Twenty-four patients underwent 25 resections: one wedge, 10 lobectomies, 4 bilobectomies, and 10 pneumonectomies. Nineteen flaps were performed, 16 omental. Fourteen had complications, including one death from adult respiratory distress syndrome. Viable tumor was found in 19 patients. Median overall survival was 30 months (12 months, group A; 43 months, group B). Estimated 3-year survival was 47%. The Kaplan-Meier survival curve for group B was superior to that for group A(p 0.019). Conclusions. Salvage lung resection after definitive chemoradiation is feasible, with encouraging survival. Surgical indication is predictive, with higher survival among patients undergoing resection for abnormal FDG- PET than for obvious relapse by CT. FDG-PET should be studied prospectively in selecting patients for salvage lung resection. Systematic staging may have increased primary incorporation of surgery, minimizing the need for late salvage. (Ann Thorac Surg 2008;86:1632 9) 2008 by The Society of Thoracic Surgeons Lung cancer is the leading cause of cancer mortality in the United States [1]. Non-small cell lung cancer (NSCLC) is the predominant histology, comprising 80% to 85% of new cases. Approximately 30% of patients with NSCLC present with locally advanced disease (stage IIIA or IIIB) [2]. The optimal treatment of patients with resectable stage IIIA disease is controversial. Recommended standards include definitive, concurrent chemoradiation, induction chemotherapy followed by operation, or induction chemoradiation followed by operation [3]. Select patients with T4 N0-1 (stage IIIB) cancer may be appropriate for primary operation or multimodality therapy [4]. Results from standard management strategies are discouraging, however, with 3-year survival rates of 20% to 43% [5 7]. After treatment with definitive chemoradiation ( 59 Gy), 24% to 35% of patients with locally advanced NSCLC Accepted for publication July 16, Presented at the Forty-fourth Annual Meeting of The Society of Thoracic Surgeons, Fort Lauderdale, FL, Jan 28 30, Address correspondence to Dr Bauman, University of New Mexico, Cancer Research and Treatment Center, 900 Camino de Salud NE MSC , Albuquerque, NM 87131; jebauman@salud.unm.edu. experience isolated local relapse [6, 8]. The safety and efficacy of surgical salvage in this setting are not well characterized. Although planned trimodality therapy is feasible, such programs typically incorporate induction doses of radiation (30 to 50 Gy) and are followed 4 to 8 weeks later by lung resection [9]. In contrast, salvage lung resection follows definitive doses of radiation, where radiation is delivered as primary therapy with curative intent. Salvage lung resection is considered with evidence of treatment failure, usually more than 12 weeks after radiation. Salvage lung resection poses a substantial theoretic risk above that of conventional trimodality therapy due to the higher dose of radiation delivered and the inherent delay between radiation and operation. Fowler and colleagues [10] reported high rates of postoperative morbidity and mortality when lung resection followed 60 Gy of radiation. Moreover, the fibrotic response to radiationinduced injury matures over time, with interstitial deposition of fibrin and depletion of the microvasculature [11]. A longer interval between radiation and resection increases the likelihood of operating in a field of radiation fibrosis, where obliterated planes may compromise dissection and hypovascularity may impair wound healing by The Society of Thoracic Surgeons /08/$34.00 Published by Elsevier Inc doi: /j.athoracsur

2 Ann Thorac Surg BAUMAN ET AL 2008;86: SALVAGE LUNG RESECTION 1633 At our academic medical center, salvage lung resection after definitive radiation for NSCLC is increasingly performed, with 1 patient in 1997 and 6 in The three most common reasons for a salvage operation are obvious relapse by computed tomography (CT), persistently abnormal fluorodeoxyglucose-positron emission tomography (FDG-PET) after completion of radiotherapy, and delayed decision to convert to a trimodality approach. We describe surgical and oncologic outcomes for 24 patients who underwent salvage lung resection after treatment with curative-intent radiation ( 59 Gy). Patients and Methods Data Collection Institutional Review Board approval was granted by the University of Washington (UW) in December 2005, with waiver of individual consent. Potential cases were identified from the 1997 to 2005 case log of the Thoracic Surgical Division. The first consultation note by a thoracic surgeon was reviewed for all cases of pulmonary wedge resection, lobectomy, or pneumonectomy. Entry criteria included (1) histologic or cytologic diagnosis of NSCLC; (2) prior treatment of NSCLC with curativeintent radiation ( 59 Gy), with or without chemotherapy; (3) no a priori plan for trimodality treatment incorporating surgical resection; and (4) subsequent salvage wedge, lobectomy, or pneumonectomy. An initial search identified 32 potential patients. Medical records were reviewed, and missing information was requested from referring physicians. Twenty-four patients met the inclusion criteria and are presented here. Five patients were excluded because they received induction radiotherapy only (30 to 50.4 Gy). One patient was excluded due to histologic diagnosis of small cell lung cancer (SCLC). Two patients were excluded because radiotherapy was delivered for SCLC, and lung resection was for NSCLC within the prior radiation field. Data collected from the preoperative period included patient demographics, medical comorbidities, forced expiratory volume in 1 second (FEV 1 ), stage assigned at diagnosis, staging procedures, reason for exclusion of operation from primary treatment, radiation dose to primary tumor, chemotherapy regimen, and time from completion of radiation to lung resection. Perioperative data included primary indication for operation, apparent clinical stage at diagnosis assigned retrospectively by the UW thoracic surgeon, surgical procedures, complications, hospital and intensive care unit (ICU) length of stay, resection status, and findings at pathology. Oncologic outcomes data included progression-free survival, site of first recurrence, and overall survival. Progressionfree and overall survival were measured in months from the date of lung resection. Patients were grouped retrospectively by primary indication for salvage lung resection: Group A: Obvious local relapse, defined by interval growth of primary tumor on thoracic computed tomography (CT) imaging. Routine biopsy was not conducted; however, 1 patient had proven recurrence of a central tumor amenable to bronchoscopic biopsy. Group B: Postradiotherapy FDG-PET demonstrated a qualitative hypermetabolic abnormality within the primary tumor or locoregional lymph nodes and the surgeon accepted the FDG-PET results as a surrogate of risk for viable tumor. Preoperative tissue confirmation was not sought, due to the poor sensitivity and negative-predictive value of percutaneous biopsy specimens after definitive chemoradiation. Group C: Empiric conversion to trimodality therapy. Four patients were offered resection after curative-intent radiotherapy when the review of staging determined the cancer was resectable and an operation had been inappropriately excluded from primary treatment. Group D: Chronic bronchopleural fistula. One patient had salvage lung resection for chronic bronchopleural fistula and empyema 22 months after definitive chemoradiation. Table 1. Patient Characteristics Characteristic No. (%) Gender Male 12 (50) Female 12 (50) Age, median (range), y 60 (36 73) Histology Adenocarcinoma 8 (33) Squamous 7 (29) Large cell 1 (4) NSCLC, NOS 8 (33) Stage at diagnosis a I 1 (4) IIB 1 (4) IIIA 14 (58) T3 N1 1 (4) T1 N2 3 (12.5) T2 N2 6 (25) T3 N2 4 (17) IIIB 6 (25) T4 N1 1 (4) T4 N2 2 (8) T2 N3 1 (4) T3 N3 1 (4) IIIB NOS 1 (4) IV 2 (8) T2 N2 M1 2 (8) FEV 1, median range, L b 2.3 ( ) a Assessed by referring provider. b Preoperative FEV 1 missing for 4 of 24 patients. FEV 1 forced expiratory volume in 1 second; NOS not otherwise specified; NSCLC non-small cell lung cancer.

3 1634 BAUMAN ET AL Ann Thorac Surg SALVAGE LUNG RESECTION 2008;86: Statistical Methods Descriptive statistics were examined for patient characteristics, surgical outcomes, and oncologic outcomes. Kaplan-Meier survival curves for groups A and B were compared using log-rank tests. This planned subgroup comparison explored whether the imaging surrogate driving the decision for operation, which was measurable progression by CT vs persistent hypermetabolic abnormality by FDG-PET, predicted oncologic outcome. Because of the small sample size, a sensitivity analysis for the overall survival comparison examined the influence of removing one patient at a time from the analysis. Results Patient Characteristics The characteristics of the 24 patients are detailed in Table 1. In 23 of 24 patients, the initial staging and treatment occurred at another institution, and they were referred for consideration of salvage lung resection. Specifics of initial staging procedures were known in 23 of 24 patients. A surgeon participated in 9 of 23 (39%) staging evaluations, and in 4 patients the surgeon was certified by the American Board of Thoracic Surgery or its equivalent (44% of surgical evaluations and 17% of staging evaluations). Mediastinoscopy had been performed in 4 of 23 patients (17%), and resection was attempted in 1 patient. Primary rationale for exclusion of surgery from primary therapy included perceived tumor stage in 12 patients, medical comorbidities in 2, and physician preference for definitive chemoradiation in 10. At the first evaluation for salvage lung resection at the UW, the original clinical TNM stage was reviewed. In 14 of 24 patients (58%), results were concordant. In 5 patients (21%), the TNM stage at diagnosis was thought lower at UW review. In 5 patients, insufficient information was available to classify TNM at diagnosis. Radiotherapy All 24 patients received definitive chest radiation, with a median dose to the primary tumor of 63.9 Gy (range, 59.4 to 70.2 Gy), and 22 received concurrent chemotherapy. Regimens were nonstandardized and included carboplatin-paclitaxel in 9 patients, cisplatin-etoposide in 7, carboplatin-docetaxel in 2, cisplatin-docetaxel in 1, and an unknown regimen in 3. Median time between the last radiation treatment and salvage lung resection was 20.6 weeks (range, 5.4 to 93.7 weeks). Salvage Lung Resection The primary rationale for salvage lung resection was obvious relapse by CT in 7 (group A), persistently abnormal FDG-PET in 12 (group B), empiric conversion to trimodality therapy in 4 (group C), and chronic bronchopleural fistula in 1(group D). FDG-PET assessments for persistent disease were not performed at a standard interval; they occurred between 4 and 26 weeks after radiation. Table 2. Morbidity and Mortality Complication No. (%) Overall morbidity, patient rate a 14 (58) Major morbidity Adult respiratory distress syndrome b 1 (4) Chylothorax 1 (4) Major vessel injury 2 (8) Myocardial infarction, ventricular fibrillation 1 (4) arrest Pneumonia 4 (17) Prolonged intubation, requiring 2 (8) tracheostomy Renal failure 1 (4) Splenic injury, requiring splenectomy 1 (4) Minor morbidity Gastric ileus, prolonged 1 (4) Stump dehiscence (minor, spontaneously 1 (4) closed) Supraventricular tachycardia, transient 7 (29) Vocal cord palsy 4 (17) Mortality 1 (4) a At least one complication occurred in 14 of 24 patients. More than one complication occurred in 8; thus, itemized morbid events add up to more than 58%. b This complication led to one postoperative death. Twenty-four patients underwent 25 salvage lung resections: one wedge, 10 lobectomies, four bilobectomies, and 10 pneumonectomies (4 right). One patient had two salvage resections, with lobectomy followed 20 months later by completion pneumonectomy for a second local recurrence. Five patients underwent chest wall resection with reconstruction, and 1 patient required pulmonary arterioplasty. Mediastinoscopy was performed in 19 patients, including 1 repeat mediastinoscopy. Mediastinal lymph node dissection was performed in 14 patients. Nineteen of 25 lung resections were paired with a vascularized flap procedure to bolster the bronchial stump, including 16 omental, 2 intercostal muscle, and 1 pericardial fat flap. The median duration of resection surgery was 5.5 hours (range, 2.2 to 9.4 hours). Median estimated blood loss was 250 ml (range, 0 to 4400 ml, data available for 22 of 25 resections). Median hospital length of stay was 8 days (range, 4 to 46 days), and 12 of 25 procedures resulted in ICU time (median, 0; range 0 42 days). Perioperative complications were observed in 14 of 24 patients, including one death from postpneumonectomy adult respiratory distress syndrome (ARDS) and multiorgan failure. One patient sustained an aortic injury during mediastinoscopy that necessitated median sternotomy and aortic arch graft replacement. One patient had a pulmonary artery avulsion injury, followed by pulmonary arterioplasty. One patient had bronchial stump dehiscence without bronchopleural fistula; the omental flap securely covered the dehiscence and no surgical repair was indicated. A list of clinically significant complications is compiled in Table 2.

4 Ann Thorac Surg BAUMAN ET AL 2008;86: SALVAGE LUNG RESECTION 1635 Fig 1. (A) Progression-free survival (solid line) and 95% confidence interval (dashed lines) in 23 patients. (B) Progression-free survival by indication for surgery, abnormal fluorodeoxyglucose-positron emission tomography (FDG-PET; dashed line) versus obvious relapse by computed tomography (CT; solid line). (C) Overall survival (solid line) and 95% confidence interval (dashed lines) in 24 patients. (D) Overall survival by indication for surgery, abnormal FDG-PET (dashed line) versus obvious relapse by CT (solid line). Surgical Pathology Pathology specimens were reviewed for all patients. Resection status is defined for all 25 resections: 23 were R0 and two were R1. Viable tumor cells were identified in 20 of 25 specimens, including both the lobectomy and pneumonectomy specimens from the patient who underwent two salvage resections. Thus, 19 (79%) of 24 patients undergoing salvage lung resection had pathologic proof of residual or recurrent disease. The proportion of patients in each group with viable tumor present in the resection specimen was as follows: group A, 7 of 7; group B, 10 of 12; group C, 2 of 4; and group D, 0 of 1. In 1 patient with FDG-PET abnormality suggesting persistent disease, Candida infection was found in the fibrotic tumor bed without residual neoplasm. Nodal involvement in the 24 patients was N0 in 13 (54%), N1 in 6 (25%), and N2 in 5 (21%). Oncologic Outcomes Ten of 24 patients (42%) were still alive at last contact, and 9 were without relapse. The median follow-up for survivors was 29 months (range, 15.6 to 91 months). Among the 10 survivors, 5 had viable tumor within the surgical specimen. Progression-free survival information was missing for 1 patient known to be living 60.9 months after resection. For 2 deceased patients, time to progression was assumed to be the same as overall survival (7.5 months and 11.7 months). Median progression-free survival was 12 months for the entire cohort, 9 months for group A, had not been reached for group B, and was 9 months for group C. Comparison of estimated Kaplan-Meier curves showed a trend toward increased progression-free survival for group B compared with group A (p 0.07, see Fig 1). Median overall survival for the entire cohort was 30 months, and estimated 3-year survival was 47%. Median overall survival was 12 months for group A, 43 months for group B, and 17 month for group C. Surgical indication was predictive of survival. By log-rank comparison, the survival curve of group B (persistently abnormal FDG- PET) was significantly superior to that of group A (obvious relapse by CT; p 0.019, see Fig 1). Sensitivity analysis did not reveal any observations with excessive influence on the group comparison: the overall survival curves were still found to be different (p 0.04) when individual patients were removed from the analysis. To address possible lead-time bias (ie, that groups A and B differed only in timing of the operation but not prognosis), a post hoc analysis compared time from radiation to operation between groups A and B. Average time to the operation was significantly different: The mean time to operation was 45.7 weeks for patients with obvious relapse by CT and 21.3 weeks for those with persistently abnormal FDG-PET (p ). To eliminate apparent lead-time bias favoring group B, an additional analysis recalculated survival curves from the last

5 1636 BAUMAN ET AL Ann Thorac Surg SALVAGE LUNG RESECTION 2008;86: day of radiation. A significant survival difference persisted (log-rank test, p 0.048). The first site of relapse was known for 11 of 13 relapsing patients: 3 experienced simultaneous local and distant relapse; 2 had local relapse only, and 1 of these underwent a second salvage lung resection; 6 had distant relapse only, and 3 of these experienced first failure in the brain. Of note, two patients relapsed with multiple brain metastases less than 2 months after surgery. The result of brain imaging in 1 of these patients was negative at the initial diagnosis, and the brain was not restaged before salvage resection 9 months later. In 1 patient (T3 N0), no brain imaging was performed at the initial diagnosis or before salvage. Comment Intergroup 0139, a randomized phase III comparison of definitive chemoradiation vs induction chemoradiation followed by operation, did not demonstrate a survival benefit to trimodality therapy in patients with resectable stage IIIA NSCLC [9]. Although planned trimodality therapy remains an accepted standard in certain cases, the optimal therapy for patients with N2 disease remains controversial. National guidelines recommend that patients with resectable, locally advanced NSCLC be treated with multimodality therapy, including either definitive, concurrent chemoradiation, or induction therapy followed by surgical resection [3]. In the United States, the dominant, curative-intent strategy for stage III disease appears to be chemoradiation. A recent national survey, conducted by the National Cancer Data Base for the year 2001, identified patterns of care for 40,909 patients with NSCLC in the United States [12]. Among patients with stage III disease, only 10.7% were treated with a multimodality program incorporating surgical intervention, and 36.2% underwent combined radiation and chemotherapy. Because isolated local relapse occurs in approximately one-third of patients who have undergone definitive chemoradiation for locally advanced NSCLC, a legitimate population may benefit from surgical salvage [6, 8]. In principle, successful identification of this population would permit selective application of trimodality therapy to the subgroup that would benefit most. Given that the dominant practice pattern for stage III disease is definitive chemoradiation, approximately 5000 patients could be candidates for salvage lung resection annually in the United States (based on 2001 data). However, the feasibility and efficacy of salvage lung resection have not been defined, nor has the optimal method for identifying appropriate candidates. Salvage lung resection poses two risks above conventional trimodality therapy: patients have been exposed to definitive doses of radiation, and the operation is often performed later than the optimal window of 4 to 8 weeks after radiation. Induction doses of radiation conventionally have been limited to 45 to 50 Gy because of an early report of unacceptable morbidity and mortality when lung resection followed 60 Gy of radiation [10]. In 1993 Fowler and colleagues [10] described surgical outcomes in 13 patients who underwent lobectomy or pneumonectomy after an induction regimen of concurrent 5-fluorouracil, cisplatin, and etoposide, and thoracic radiation to 60 Gy. Although 6 patients underwent lobectomy without mortality, 3 of 7 patients died after pneumonectomy, resulting in an overall 23% mortality rate for the series. Major complications were chiefly characterized by ARDS and bronchopleural fistula. The conventional limit in induction radiation dose has been challenged by three institutional series that demonstrated feasibility of planned trimodality therapy with induction doses exceeding 59.4 Gy [13 15]. In 2004 Sonett and colleagues [14] reported 40 consecutive patients undergoing lung resection after curative-intent radiotherapy ( 59 Gy) and concurrent, platinum-based chemotherapy. This series had no operative deaths and only 1 patient each with ARDS and bronchopleural fistula, despite 11 pneumonectomies. The lower complication rate may be attributable to radiation technique, use of vascular flaps, minimization of intraoperative fluid, and ventilatory strategies. Notably, patients in these planned, high-dose trimodality series underwent lung resection in the optimal window after radiation: The median time from radiation to surgery was less than 2 months in all three series. In contrast, surgical resection in the salvage setting is typically delayed beyond 2 months. Thus, the fibrotic response to radiation is more mature. In a rat model of radiation-induced lung injury, the early postradiation phase (6 to 12 weeks) is characterized by parenchymal and vascular inflammation, and the late phase (34 to 38 weeks) is characterized by fibroblast hypercellularity and collagen deposition [16]. In practice, the surgeon operating on fibrotic lung encounters brittle, devascularized tissue and obliterated planes, resulting in greater difficulty dissecting important structures such as major vessels. The delay inherent to salvage lung resection may increase intraoperative risk to major vessels, lengthen resection time, and further increase the risk of bronchopleural fistula due to impaired wound healing. In this cohort of 24 patients undergoing salvage lung resection, all patients had definitive doses of radiation. The median time from radiation to operation was 5 months, with resection occurring as late as 22 months after radiation. Ten patients underwent pneumonectomy. Despite these high-risk features, salvage lung resection was feasible, with acceptable toxicity. As expected, median operating time was high, at 5.5 hours. However, the hospital and ICU median stay did not appear substantially different from Sonett and colleague s [14] planned trimodality series. The complication rate was high: 14 of 24 patients experienced perioperative morbidity, with 8 patients (33%) experiencing at least one major complication. Yet, only one postoperative death occurred, secondary to ARDS after left pneumonectomy. Intraoperative risk to major vessels was borne out in 2 patients, with one aortic and one pulmonary arterial injury; dense adhesions and fibrosis were encountered

6 Ann Thorac Surg BAUMAN ET AL 2008;86: SALVAGE LUNG RESECTION 1637 intraoperatively in both patients. The absence of postoperative bronchopleural fistula is conspicuous. A higher rate of fistulas would have been expected because of impaired wound healing within fibrotic tissue. The low rate is likely attributable to high utilization of vascularized flaps to buttress the bronchial stump. Omentum was used frequently because it can reach anywhere within the chest, is pliable to cover any shape, has excellent angiogenic properties, is universally outside of the radiation field that may include chest wall muscles, and does not result in any musculoskeletal disability once the laparotomy is healed. Oncologic outcomes were likewise encouraging for this highly selected population. The median survival of 30 months and the median 3-year survival of 47% are within the ranges reported by prospective, trimodality series for stage IIIA and IIIB patients [13, 14, 17]. A planned subgroup comparison noted significantly higher survival among patients undergoing resection for persistently abnormal FDG-PET than for measurable relapse by CT. A similar trend for progression-free survival suggests the observed survival difference was related to cancer progression. These findings are of clear clinical relevance. If the imaging surrogate driving the decision for operation predicts outcome, then algorithms for repeat imaging and surgical referral after definitive chemoradiation should be developed accordingly. An obvious criticism in this study is lead-time bias, because patients with abnormal FDG-PET underwent resection earlier after radiation than did patients with obvious relapse by CT. Yet, a significant difference persisted when survival curves were recalculated according to the date of last radiation. Two explanations are plausible. First, the optimal window for salvage lung resection is before measurable local progression. Captured before obvious relapse, patients in group A may have benefited from surgical salvage. Alternately, patients with measurable local progression may have innately aggressive tumor biology and would be destined to do poorly even with earlier surgical salvage. In this series, 5 of 10 surviving patients had viable tumor present at operation and would presumably have developed obvious clinical relapse without surgical salvage. FDG-PET appeared to be a useful surrogate for local disease persistence. Ten of 12 patients brought to salvage for abnormal FDG-PET had viable tumor, representing 83% accuracy in preoperative diagnosis of persistent disease. Nonetheless, our result should be approached with caution. FDG-PET abnormality was defined retrospectively and broadly included any qualitative hypermetabolic abnormality within tumor or ipsilateral lymph nodes. Moreover, FDG-PET was performed at varying time points after radiation and at multiple facilities with variable expertise in image acquisition and interpretation. Finally, the absence of a comparison population with normalized tumor metabolism on postradiation FDG-PET precludes estimates of sensitivity or specificity. Nonetheless, FDG-PET may represent the best available surrogate for residual disease after definitive chemoradiation. In a series of 26 patients with resectable stage IIIA or IIIB NSCLC treated with neoadjuvant chemoradiotherapy, FDG-PET was performed for the initial staging as well as for restaging 2 weeks after the completion of induction [18]. When prespecified levels of FDG uptake were exceeded relative to noncardiac mediastinal structures, the sensitivity and specificity of FDG-PET for distinguishing viable tumor from a complete response were 67% and 63%, respectively. Sensitivity and accuracy may be augmented by concomitant CT evaluation [19]. FDG-PET after definitive chemoradiation, at a set time point and with predetermined criteria for abnormality, should be studied prospectively in the selection of candidates for surgical salvage. Relapse patterns in this study were similar to prospective trials in locally advanced NSCLC [20]. Notably, 2 patients relapsed with multiple brain metastases in less than 2 months after salvage surgery, resulting in death. Neither patient s brain was imaged before resection, although 1 patient had negative result of brain imaging at diagnosis 9 months before resection. Brain imaging during preoperative assessment may have avoided futile thoracotomy and lung resection for both patients. This observation indicates the brain should be screened for metastases before salvage lung resection, even if the result of the preradiation evaluation was negative. This study has several important limitations. This is a retrospective, single-institution series where the intervention salvage lung resection occurred in a nonstandardized population during an 8-year period. Patients were referred at heterogeneous time points for consideration of surgical salvage, and the indications for operation varied. Surgical and oncologic outcomes could not be determined in an intent-to-treat manner and are difficult to generalize to the multi-institutional setting. The study reflects a practical reality, likely unique to the academic cancer center, of the application of a surgical salvage procedure after nonstandardized staging and a decision for nonoperative therapy in the community. Experience with these select patients raises an important question of how frequently patients are overstaged or deemed unresectable in the absence of appropriate thoracic surgical consultation. Here, 23 of 24 patients were referred for surgical salvage from the community, including 12 patients originally labeled as being unresectable. The rate of upfront thoracic surgical consultation was only 17%, indicative of a lost opportunity for systematic staging and appropriate incorporation of surgical intervention into primary therapy. National lung cancer treatment guidelines recommend evaluation by a multidisciplinary team that includes an experienced thoracic surgical oncologist [3, 21]. Upfront multidisciplinary evaluation with a thoracic surgeon could have minimized the need for surgical salvage and potentially improved surgical and oncologic outcomes. Despite clear limitations, this series demonstrates that salvage lung resection after definitive radiation for NSCLC is technically feasible, with acceptable toxicity even when performed at a delayed interval. Although oncologic outcomes are encouraging, with a subset of

7 1638 BAUMAN ET AL Ann Thorac Surg SALVAGE LUNG RESECTION 2008;86: long-term survivors, determination of efficacy requires prospective validation in a rigorously defined population. The superior survival of patients undergoing resection for persistently abnormal FDG-PET vs for obvious relapse by CT is hypothesis generating. Prospective trials incorporating response assessment by FDG-PET after definitive chemoradiation for resectable, locally advanced NSCLC, with predetermined criteria for salvage lung resection, are justified. This study was conducted without grant support. Nominal costs were covered by the research budget of the Thoracic Oncology Research Group at the University of Washington. References 1. Ries LAG, Eisner MP, Kosary CL, et al. SEER Cancer Statistics Review Bethesda, MD, National Cancer Institute, Mountain CF. Revisions in the international system for staging lung cancer. Chest 1997;111: NCCN Clinical Practice Guidelines in Oncology. Nonsmallcell Lung Cancer v Available at: org/professionals/physician_gls/pdf/nscl.pdf. Accessed Jan 11, DiPerna CA, Wood DE. Surgical management of T3/T4 lung cancer. Clin Cancer Res 2005;11(13 suppl):5038s 44S. 5. DeCamp MM Jr, Ashiku S, Thurer R. The role of surgery in N2 nonsmall-cell lung cancer. Clin Cancer Res 2005;11(13 suppl):5033s 37S. 6. Fournel P, Robinet G, Thomas P, et al. Randomized phase III trial of sequential chemoradiotherapy compared with concurrent chemoradiotherapy in locally advanced nonsmallcell lung cancer: Groupe Lyon-Saint-Etienne d Oncologie Thoracique-Groupe Français de Pneumo-Cancérologie NPC study. J Clin Oncol 2005;23: Roth J, Atkinson EN, Fossella F, et al. Long-term follow-up of patients enrolled in a randomized trial comparing perioperative chemotherapy and surgery with surgery alone in respectable stage IIIA non-small-cell lung cancer. Lung Cancer 1998;21: Furuse K, Fukuoka M, Kawahara M, et al. Phase III study of concurrent versus sequential thoracic radiotherapy in combination with mitomycin, vindesine and cisplatin in unresectable stage III nonsmall-cell lung cancer. J Clin Oncol 1999;17: Albain KS, Scott CB, Rusch VR, et al. Phase III comparison of concurrent chemotherapy plus radiotherapy (CT/RT) and CT/RT followed by surgical resection for stage IIIA (pn2) non-small-cell lung cancer. Initial results from intergroup trial 0139 (RTOG 93 09). Proc Am Soc Clin Oncol 2003:621; Abstract Fowler W, Langer CJ, Curran WJ, et al. Postoperative complications after combined neoadjuvant treatment of lung cancer. Ann Thorac Surg 1993;55: Kong FM, Ten Haken R, Eisbruch A, et al. Nonsmall-cell lung cancer therapy-related toxicity: an update on radiation pneumonitis and fibrosis. Semin Oncol 2005;32(suppl 3): S Little AG, Gay EG, Gaspar LE, Stewart AK. National survey of nonsmall-cell lung cancer in the United States: epidemiology, pathology and patterns of care. Lung Cancer 2007;57: Vora SA, Daly BD, Blaszkowsky L, et al. High dose radiation therapy and chemotherapy as induction treatment for stage III non-small cell lung carcinoma. Cancer 2000;89: Sonett JR, Suntharalingam M, Edelman MJ, et al. Pulmonary resection after curative intent radiotherapy ( 59 Gy) and concurrent chemotherapy in nonsmall-cell lung cancer. Ann Thorac Surg 2004;78: Suntharalingam M, Sonett JR, Haas ML, et al. The use of concurrent chemotherapy with high-dose radiation before surgical resection in patients presenting with apical sulcus tumors. Cancer J 2000;6: Novakova-Jiresova A, van Luijk P, van Goor H, Kampinga HH, Coppes RP. Changes in expression of injury after irradiation of increasing volumes in rat lung. Int J Radiation Oncology Biol Phys 2007;67: Albain KS, Rusch VW, Crowley JJ, et al. Concurrent cisplatin/etoposide plus chest radiotherapy followed by surgery for stages IIIA(N2) and IIIB non-small-cell lung cancer: mature results of Southwest Oncology Group phase II study J Clin Oncol 1995;13; Ryu JS, Choi NC, Fischman AJ, Lynch TJ, Mathisen DJ. FDG-PET in staging and restaging nonsmall-cell lung cancer after neoadjuvant chemoradiotherapy: correlation with histopathology. Lung Cancer 2002;35: Wong CY, Schmidt J, Bong JS, et al. Correlating metabolic and anatomic responses of primary lung cancers to radiotherapy by combined F-18 FDG PET-CT imaging. Radiat Oncol 2007;2: Mamon HJ, Yeap BY, Jänne PA, et al. High risk of brain metastases in surgically staged IIIA nonsmall-cell lung cancer patients treated with surgery, chemotherapy, and radiation. J Clin Oncol 2005;23: Alberts WM. Diagnosis and management of lung cancer executive summary: ACCP evidenced-based practice guidelines (2nd edition). Chest 2007;132:1 19. DISCUSSION DR DAVID H. HARPOLE (Durham, NC): I have to make a comment. That was extremely well presented, and I wish all of the thoracic medical oncologists in North America would have your grasp of surgical techniques. It was very well done. DR MARK J. KRASNA (Towson, MD): Welcome to the lion s den. First of all, I congratulate you on your willingness to look at this group and specifically, Dr Wood, your willingness to reconsider operating after high-dose radiation therapy. Welcome back to the club. I think that using Fowler as a historical perspective, with all respect to Dr Fowler, is just that it s historical. I think we have enough good retrospective large series, whether it s from Dr Cerfolio in Alabama, Dr Daly, Duke, Maryland, Brigham, et cetera, that really we should no longer limit our radiation dose because of one small series that was retrospective with bad outcomes and high morbidity and mortality. I congratulate you for doing this, and perhaps we just need to move on and no longer refer to that study. One of the other comments that I have will lead to a question to either you or Dr Wood. I was impressed that despite the fact that these patients were not staged surgically up front, you did successfully stage about 19 of them surgically, if I saw correctly. My question to you or Doug is why not surgically stage all of them? Were there technical reasons why those patients were not staged? My only other specific question to you is, when you look at the path complete response [CR] rate, it is surprising that it is so low.

8 Ann Thorac Surg BAUMAN ET AL 2008;86: SALVAGE LUNG RESECTION 1639 I agree with your hypothesis that likely you have already identified that group that has an innately bad tumor biology and that is why there was such a low path CR rate. This does lead to your conclusion, that these patients should be seen prospectively, and I want to support it. The NCI [National Cancer Institute] has established the NCCCP [NCI Community Cancer Centers Program], which is a community cancer center pilot program, to do exactly what we used to only do in academic centers throughout North America, which is do true multidisciplinary care. I applaud you for doing it, and hopefully, Doug, you will get your entire community in Seattle to send you those patients prospectively and sit in with you on your conferences. Thank you very much. DR BAUMAN: I would gladly defer the questions regarding surgical technique to Dr Wood. DR WOOD: Julie could probably answer them, but she is just being polite to me so that I can have a chance to say something. In terms of the dose of radiation, I think the point that Dr Krasna makes is very legitimate. However, that still doesn t get at the timing of radiation related to surgery, because the studies that you are referring to with a new standard dose of radiation that would be at the definitive level are being performed with the intent of subsequent surgical resection, which is in a narrow time window, as opposed to surgical salvage, which is a patient that is a late referral to the thoracic surgeon with a plea to consider surgery at a late stage, usually when surgery was felt to be contraindicated primarily. This is something that we are all familiar with and a problem that we face that results from the lack of thoracic surgical consultation at the outset of that patient s staging and initial therapy. In terms of staging, this is a very heterogeneous population who has had a variety of staging interventions before salvage surgical referral. Some of them had had mediastinoscopy, and we did not repeat mediastinoscopy in very many of them. So we were pretty thorough about staging, as you could see, but I think you don t see 100% mediastinoscopy for a variety of reasons relating to the previous treatment and mediastinoscopy. DR ARA VAPORCIYAN (Houston, TX): I enjoyed your paper very much. I noticed that a subgroup of the patients on final pathology had no viable cancer in the specimen. Did a significant number of these patients come from the persistent abnormal FDG-PET [fluorodeoxyglucose-positron emission tomography] group, and, if so, could this have contributed to the improved outcome seen in this group? Thanks again for a wonderful presentation. DR BAUMAN: That is a very reasonable question. The finding of viable tumor at pathology did relate to the group and indication for surgery. All 7 of 7 patients referred for salvage because of obvious local relapse had viable tumor at surgery. However, 10 of 12 patients in the PET-positive group also had viable tumor at surgery, thus the difference may not be explained wholly on that basis. The other 3 patients without viable tumor were the 1 treated for chronic bronchopleural fistula, and 2 of the 4 empirically converted to trimodality therapy.