Local Recurrence After Surgery for Early Stage Lung Cancer

Original Article Local Recurrence After Surgery for Early Stage Lung Cancer An 11-Year Experience With 975 Patients Chris R. Kelsey, MD 1 ; Lawrence B. Marks, MD 2 ; Donna Hollis, MS 3 ; Jessica L. Hubbs, MS 2 ; Neal E. Ready, MD, PhD 4 ; Thomas A. D Amico, MD 5 ; and Jessamy A. Boyd, MD 1 BACKGROUND: The objective of the current study was to evaluate the actuarial risk of local failure (LF) after surgery for stage I to II nonsmall cell lung cancer (NSCLC) and assess surgical and pathologic factors affecting this risk. METHODS: The records, including pertinent radiologic studies, of all patients who underwent surgery for T1 to T2, N0 to N1 NSCLC at Duke University between 1995 and 2005 were reviewed. Risks of disease recurrence were estimated using the Kaplan-Meier method. A multivariate Cox regression analysis assessed factors associated with LF in the entire cohort and a subgroup undergoing optimal surgery for stage IB to II disease. RESULTS: For all 975 consecutive patients, the 5-year actuarial risk of local and/or distant disease recurrence was 36%. First sites of failure were local only (25%), local and distant (29%), and distant only (46%). The 5-year actuarial risk of LF was 23%. On multivariate analysis, squamous/ large cell histology (hazards ratio [HR], 1.98), stage > IA (HR, 2.02), and sublobar resections (HR, 1.99) were found to be independently associated with a higher risk of LF. For the subset of patients (n ¼ 445) undergoing at least a lobectomy with negative surgical margins and currently considered for adjuvant chemotherapy (stage IB-II disease), the 5-year actuarial risk of LF was 27%. Within this subgroup, squamous/large cell histology (HR, 2.5) and lymphovascular space invasion (HR, 1.74) were associated with a higher risk of LF. The 5-year rate of LF was 13%, 32%, and 47%, respectively, with 0, 1, or 2 risk factors. CON- CLUSIONS: Greater than half of disease recurrences after surgery for early stage NSCLC involved local sites. Pathologic factors may help to distinguish those patients at highest risk. Cancer 2009;115:5218 27. VC 2009 American Cancer Society. KEY WORDS: nonsmall cell lung cancer, local recurrence, local/regional recurrence, patterns of failure, radiotherapy. Surgery is the preferred initial treatment for patients with early stage nonsmall cell lung cancer (NSCLC). Due to the relatively high risk of disease recurrence after surgery, adjuvant radiotherapy (RT) Corresponding author: Chris R. Kelsey, MD, Department of Radiation Oncology, Duke University Medical Center 3085, Durham, NC 27710; Fax: (919) 668-7345; kelse003@mc.duke.edu 1 Department of Radiation Oncology, Duke University Medical Center, Durham, rth Carolina; 2 Department of Radiation Oncology, University of rth Carolina, Chapel Hill, rth Carolina; 3 Duke Cancer Center Biostatistics, Duke University Medical Center, Durham, rth Carolina; 4 Department of Medicine, Division of Medical Oncology, Duke University Medical Center, Durham, rth Carolina; 5 Department of Surgery, Duke University Medical Center, Durham, rth Carolina Presented in part at the 44th Annual Meeting of the American Society of Clinical Oncology, Chicago, Illinois, May 30-June 3, 2008; and the 50th Annual Meeting of the American Society for Therapeutic Radiology and Oncology, Boston, Massachusetts, September 21-25, 2008. We thank Thomas A. Sporn, MD, from the Department of Pathology at Duke University for reviewing the text and providing invaluable suggestions. Received: December 1, 2008; Revised: May 14, 2009; Accepted: May 15, 2009 Published online August 11, 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/cncr.24625, www.interscience.wiley.com 5218 Cancer vember 15, 2009

Local Recurrence of Early Stage Lung Cancer/Kelsey et al and chemotherapy have each been investigated in randomized clinical trials. Several trials demonstrated that postoperative RT reduced the risk of local/regional (ie, local) disease recurrence. 1-3 However, RT was found to have either no effect, 1 or a detrimental effect, 3,4 on survival, due to competing risks of distant metastases and treatment-related mortality. Early adjuvant chemotherapy studies demonstrated a survival detriment with alkylating agents. 5 More recently, cisplatin-based regimens have demonstrated survival benefits in certain subsets of patients with early stage NSCLC. 6-8 An accurate assessment of patterns of failure after surgery is helpful to guide effective postoperative therapy. At the current time, adjuvant treatment is focused on chemotherapy and the risk of distant metastases. Postoperative RT is not currently being pursued out of concern for treatment-related toxicity. Furthermore, the risk of local recurrence is generally considered to be small in comparison with the risk of distant recurrence. However, the risk of local disease recurrence after surgery has not been well described and pathologic features placing patients at particularly high risk of local recurrence have been inadequately studied. This patterns-of-failure analysis was initiated to provide insight into these issues. MATERIALS AND METHODS This Institutional Review Board-approved retrospective study was performed by searching the Duke Comprehensive Cancer Center database for patients who underwent surgery for T1 to T2, N0 to N1 NSCLC at Duke University between 1995 and 2005. Patients who received preoperative chemotherapy and/or RT, presented with synchronous primary tumors, or had a prior history of lung cancer were excluded. Medical records and pertinent radiologic imaging were reviewed to characterize each patient s demographic information, obtain surgical and pathologic details, and score patterns of failure after surgery. Between 1995 and 1997, lung tumor specimens were interpreted by multiple pathologists at Duke University. After 1998, all specimens were interpreted by a single pathologist with a special interest in pulmonary pathology (T.A. Sporn). Hematoxylin and eosin stains were performed on all specimens. Elastin stains were used when necessary to evaluate involvement of the visceral pleura. Other pathologic studies, including immunostains, were performed at the discretion of the pathologist. Lymphovascular space invasion was reported when tumor emboli involved lymphatic, venous, and/or arterial channels. The primary objective of the current study was to assess the risk of local disease recurrence after surgery and to identify clinical, surgical, and pathologic features associated with a higher risk of local recurrence. Disease recurrence at the surgical resection margin, ipsilateral hilum, and/or mediastinum was considered a local (ie, local/regional) recurrence. All other sites of failure, including the supraclavicular fossa and contralateral hilum, were considered distant recurrences. Patterns of failure were assessed by means of followup imaging studies and data obtained from procedures such as bronchoscopy, mediastinoscopy, etc. Hilar and mediastinal lymph node failures were defined as a new or enlarging lymph node measuring 1 cm on the short axis on computed tomography (CT), and/or hypermetabolic on positron emission tomography (PET) imaging, which in the patient s subsequent clinical follow-up was consistent with disease progression. Thus, an enlarged lymph node or increased tissue thickening at the resection margin, or equivocal findings on PET, that failed to progress were not scored as a local recurrence. All cases of possible local and distant disease recurrence were reviewed by 2 authors to ensure accuracy (C.K. and J.B.). All patients had routine postsurgical surveillance with imaging studies, including chest CT, but testing was not standardized. Secondary objectives were to assess the risk of distant disease recurrence and the development of second primary lung cancers. A second primary tumor was scored when a patient presented with a different histology or the same histology but a clinical presentation that was most consistent with a new primary tumor. Distinguishing a second primary malignancy from a local and/or distant recurrence required clinical judgment, and such cases were reviewed by 2 authors (C.K. and J.B.). The Kaplan-Meier product-limit method 9,10 was used to estimate 5-year recurrence probabilities and 95% confidence intervals (95% CIs). Time to treatment failure was calculated from the date of surgery to the date of treatment failure (defined as local and/or distant recurrence or the development of a second primary lung cancer). Time to local failure and time to distant failure were calculated from the date of surgery to the date of local or distant Cancer vember 15, 2009 5219

Original Article disease recurrence, respectively. For these metrics, patients developing a second primary lung cancer were censored on the date the second primary malignancy was diagnosed, because it was often impractical to distinguish a local failure arising from the first versus the second lung cancer. Local and distant recurrences were scored independently (ie, patients developing a distant recurrence were not censored for local failure, but were assessed for local failure until the date of last follow-up or death). Presented median recurrence times were calculated using only those patients who had recurrent disease. A Fisher exact test was used to calculate all 2 2 table probabilities of independence. 11 Univariate Cox proportional hazards models 9,10 were used to assess the strength of the association between time to disease recurrence and surgical and pathologic risk factors (covariates), and to estimate hazards ratios (HR) and 95% CIs using the Wald chi-square statistic. The assumption of proportional hazards was assessed by adding each covariate by log time interaction to the model and assessing the significance of the product term using the partial likelihood ratio test. Stages and histologies were combined according to similar HRs. In the multivariate Cox proportional hazards regression model, all variables with P values <.10 in the univariate analysis were included and a stepwise variable selection approach was used, with entrance and exit significance levels of.05. Kaplan-Meier survival plots were used to assess subgroup survival similarities. This univariate/multivariate approach was performed on the entire patient cohort, and repeated in a subgroup of patients undergoing optimal surgery (at least lobectomy with negative margins) and currently considered potential candidates for adjuvant chemotherapy (those with stage IB-II disease per the American Joint Committee on Cancer staging system, 6th edition). All statistical tests were 2-sided and a P value <.05 was considered statistically significant. SAS 9.1 statistical software (SAS Institute, Inc, Cary, NC) was used for all statistical analyses. RESULTS We identified 975 patients. The median follow-up was 33 months (range, 1-149 months). Patient characteristics and surgical/pathologic details are shown in Table 1. The majority of patients had stage I disease (45% with stage IA and 39% with stage IB). Hilar lymph nodes (N1, levels 10-14) were sampled in 81% of patients. The median number of N1 lymph nodes sampled was 4 (range, 1-31 N1 lymph nodes). Mediastinal lymph nodes (N2, levels 1-9) were sampled or dissected in 85% of patients. The median number of mediastinal lymph node stations examined was 4 (range, 1-9 mediastinal lymph node stations). Surgical margins were positive in 4% of patients. Adjuvant chemotherapy was administered to 7% of patients. Carboplatin and paclitaxel were used most frequently (49%), although several regimens were used. Adjuvant RT was given to 3% of patients. Patients with positive surgical margins were more likely to receive adjuvant RT than patients with negative surgical margins (31% vs 2%; P <.0001). Considering all 975 patients, the 5-year actuarial risk of treatment failure (local and/or distant disease recurrence or diagnosis of a second primary lung cancer) was 42% (95% CI, 38-46%) (Fig. 1). A local and/or distant recurrence developed in 250 patients. First sites of failure were local only (25%), local and distant (29%), and distant only (46%). The 5-year actuarial risk of local and/or distant recurrence was 36%. A second primary lung cancer was diagnosed in 45 patients. Local disease recurrence was identified in 140 patients. Diseaserecurrencewasconfirmedbymeansofbiopsyin 47% of patients, PET in 33% of patients, and CT in 20% of patients. Distribution of local failures were as follows: mediastinum (36%), ipsilateral hilum (11%), wedge resection line (10%), bronchial stump (9%), and multiple subsites (34%). The median time from surgery to local recurrence was 14.1 months (range, 1-78 months) for those patients with local recurrence. The 5-year actuarial risk of local disease recurrence was 23% (95% CI, 19-26%) (Fig. 2). Distant recurrence was identified in 207 patients and was pathologically confirmed in 103. The median time from surgery to distant disease recurrence was 12.5 months (range, 1-79 months) for those developing distant metastases. For all patients, the 5-year actuarial rate of distant disease recurrence was 34% (95% CI, 30-39%). The risk of local recurrence was higher for patients withstageib,iia,andiibdiseasecomparedwiththose with stage IA disease (HRs of 2.0, 2.6, and 1.9, respectively; P ¼.0001). Variables associated with a higher rate of local failure on univariate analysis were sublobar resections (HR, 1.64), lack of mediastinal lymph node sampling (HR, 1.65), increasing size (HR, 1.13 per cm), 5220 Cancer vember 15, 2009

Local Recurrence of Early Stage Lung Cancer/Kelsey et al Table 1. Patient and Treatment Characteristics (N¼975) Characteristic. % Age, y Mean 66.4 Median 67 Range 20-93 Gender Male 533 55 Female 442 45 Race White 824 85 Black 124 13 Other 27 2 Surgical procedure Wedge 161 17 Segmentectomy 41 4 Lobectomy 668 68 Sleeve resection 36 4 Pneumonectomy 69 7 Surgical approach Open 600 62 VATS 375 38 Hilar lymph node sampling Yes 790 81 185 19 Mediastinal lymph node sampling Yes 824 85 151 15 Size, cm Mean 3.23 Median 2.6 Range 0.3-13 Location Right upper lobe 349 36 Right middle lobe 51 5 Right lower lobe 152 16 Right lung, NOS 2 <1 Left upper lobe 280 29 Left lower lobe 141 14 Histology Adenocarcinoma 441 45 Squamous cell* 364 38 Large cell 48 5 Bronchioloalveolar 28 3 NSCLC, NOS 94 10 Histologic differentiation Well 77 8 Moderate 424 43 Poor 298 31 NS 176 18 Lymphovascular space invasion Yes 210 22 /NS 765 78 (Continued) Table 1. (Continued) Characteristic. % Visceral pleural invasion Yes 201 21 /NS 774 79 Pathologic stagey IA 435 45 IB 385 39 IIA 43 4 IIB 112 11 Surgical margins Negative 939 96 Positive 36 4 Adjuvant chemotherapy Yes 65 7 910 93 Adjuvant radiotherapy Yes 33 3 942 97 VATS indicates video-assisted thoracoscopic surgery; NOS, not otherwise specified; NSCLC, nonsmall cell lung cancer; NS, not stated. * Including adenosquamous (n¼6). y Grading determined according to the American Joint Committee on Cancer staging system, 6th edition. FIGURE 1. Kaplan-Meier curve of time to treatment failure (defined as local and/or distant disease recurrence or the development of a second primary lung cancer) is shown. squamous or large cell histology (HR, 2.15), moderate or poorly differentiated grade (HR, 2.88), lymphovascular space invasion (HR, 1.54), visceral pleural invasion (HR, 1.93), stage IB to IIA disease (HR, 2.00), and positive surgical margins (HR, 1.85). A stepwise multivariate proportional hazards analysis resulted in squamous or large cell histology (HR, 1.98), stage > IA (HR, 2.02), and sublobar resection (HR, 1.99) being independently Cancer vember 15, 2009 5221

Original Article FIGURE 2. Kaplan-Meier curve of time to local failure is shown. associated with a higher risk of local disease recurrence (Table 2). For the subset of patients (n ¼ 445) undergoing optimal surgery (at least a lobectomy with negative surgical margins) and currently considered for adjuvant chemotherapy (stage IB-II disease), the 5-year actuarial risk of local disease recurrence was 27% (95% CI, 21-33%) and that for distant recurrence was 42% (95% CI, 36-48%). On multivariate analysis within this subgroup of patients, lymphovascular space invasion (HR, 1.74; 95% CI, 1.1-2.8) and squamous/large cell histology (HR, 2.5; 95% CI, 1.5-4) were found to be associated with a higher risk of local recurrence (Table 3). The 5-year actuarial risk of local recurrence was 13% (95% CI, 8-22%), 32% (95% CI, 25-41%), and 47% (95% CI, 30-67%), respectively, with 0, 1, or 2 risk factors (Fig. 3). DISCUSSION NSCLC remains the leading cause of cancer mortality in the United States. 12 Even patients with early disease are at relatively high risk of disease recurrence. Cisplatin-based chemotherapy regimens have decreased the risk of recurrence after surgery and provided modest survival gains. An updated meta-analysis from the NSCLC Collaborative Group demonstrated an average survival benefit of 5.4% at 5 years for patients with resected stage I to III disease. 13 Rates of local failure are infrequently reported in adjuvant lung cancer trials. It is generally believed that the risk of local disease recurrence after surgery for patients with early stage NSCLC is low. However, there is substantial variation in the literature, with rates ranging from 6% to 45% for stage I disease 2,3,14-21 and 7% to 55% for stage II disease. 1,18,21-24 There are multiple reasons why the risk of local recurrence is poorly defined. To begin with, the definition of local failure varies widely among publications, even prospective randomized trials. In the Adjuvant Navelbine International Trialist Association (ANITA) trial of chemotherapy versus observation for patients with stage IB to IIIA NSCLC, the crude rate of local failure was 12% in the arm randomized to receive chemotherapy. 7 However, only ipsilateral mediastinal recurrences were scored as local failures. In other words, a contralateral mediastinal recurrence was not considered a local recurrence. We believe the most appropriate definition of local (ie, local/ regional) failure is disease recurrence at the surgical resection margin, ipsilateral hilum, and mediastinum. Practically speaking, these are the sites that are encompassed by typical radiation fields in the postoperative setting. Furthermore, many studies only report first sites of failure. Distant metastases commonly develop after surgery for lung cancer and are typically easier to assess radiologically than local failures. Unless there is a thorough evaluation performed at the time of disease recurrence (to assess for a concurrent local failure), a local recurrence may be missed. It is interesting to note that studies with local control as a primary endpoint 2,3,15 have generally reported higher rates of local failure than those with other primary endpoints, such as disease-free survival. 16,17 This most likely reflects the diligence in which local failure is assessed and recorded. In addition, some studies only score local failures if they occur in the absence of a distant failure. For example, in the Lung Cancer Study Group, crude rates of local failure for stage IA disease were 6% after lobectomy and 17% after sublobar resections. Local failure was defined as disease recurrence in the ipsilateral lung and mediastinum. However, local failure was only scored if it occurred in the absence of a concomitant distant failure. In the current study, 72 of 140 (51%) local failures occurred with a concomitant distant failure. In those patients who developed concurrent local and distant failures, the true temporal relation between these 2 events is uncertain. Although they were detected concurrently, it remains possible that the local failure occurred before, and was a cause of, the distant failure. 5222 Cancer vember 15, 2009

Local Recurrence of Early Stage Lung Cancer/Kelsey et al Table 2. Univariate and Multivariate Analyses for Local Recurrence in All Patients (n¼975) Factors* Hazards Ratio Univariate Analysis 95% CI Wald Chi-Square P Hazards Ratio Multivariate Analysis 95% CI Wald Chi-Square P Age 1 0.98-1.02.9784 Gender Female 1.40 1.00-1.96.0504 Male Surgical procedure Lobectomy 1.64 1.12-2.39.0109 1.99 1.35-2.94.0005 Wedge/segmentectomy Surgical approach Open 1.06 0.75-1.49.7482 VATS Hilar lymph node sampling Yes 1.36 0.91-2.05.1374 Mediastinal lymph node sampling Yes 1.65 1.09-2.49.0184 Size, cm 1.13 1.05-1.22.0011 Histology nsquamous 2.15 1.54-3.01 <.0001 1.98 1.14-2.78 <.0001 Squamous or large cell Histologic differentiation Well 2.858 1.18-7.04.0199 Moderate or poorly Lymphovascular space invasion /NS 1.54 1.05-2.25.0260 Yes Visceral pleural invasion /NS 1.93 1.34-2.77.0004 Yes Pathologic stage IA 2.00 1.41-2.87.0001 2.02 1.40-2.92.0002 IB-IIB Surgical margins Negative 1.85 0.90-3.76.0931 Positive Adjuvant chemotherapy Yes 1.04 0.56-1.92.9126 Adjuvant radiotherapy Yes 1.09 0.48-2.48.8326 95% CI indicates 95% confidence interval; VATS, video-assisted thoracoscopic surgery; NS, not stated. * The first factor level listed is the reference set. Cancer vember 15, 2009 5223

Original Article Table 3. Subgroup (n¼445) Undergoing Lobectomy for Stage IB-II Disease: Univariate and Multivariate Analyses for Local Recurrence Univariate Analysis Multivariate Analysis Factors* Hazards Ratio 95% CI Wald Chi-Square P Hazards Ratio 95% CI Wald Chi-Square P Age 1.02 0.996-1.05.10 Gender Female 1.84 1.11-3.05.0177 Male Surgical approach Open 1.34 0.83-2.17.2386 VATS Hilar lymph node sampling Yes 3.55 0.49-25.53.2085 Mediastinal lymph node sampling Yes 1.17 0.51-2.70.7163 Size, cm 1.05 0.94-1.17.4171 Histology nsquamous 2.48 1.53-4.00.0002 2.45 1.51-3.96.0003 Squamous or large cell Histologic differentiation Well 4.12 0.57-29.41.1597 Moderate or poorly Lymphovascular space invasion /NS 1.77 1.10-2.85.0191 1.74 1.08-2.80.0236 Yes Visceral pleural invasion /NS 1.26 0.78-2.01.3438 Yes Adjuvant chemotherapy Yes 1.62 0.78-3.37.1970 Adjuvant radiotherapy Yes 2.67 0.66-10.89.1720 95% CI indicates 95% confidence interval; VATS, video-assisted thoracoscopic surgery; NS, not stated. * The first factor level listed is the reference set. Finally, local failure rates are often reported as crude percentages, rather than actuarial probabilities. There are statistical limitations with Kaplan-Meier statistics when assessing nonsurvival endpoints, such as time to local failure, due to competing risks. 25 There is no ideal method to correct for competing risks. We did not censor patients for local failure after a distant disease recurrence was identified, but rather assessed all patients until the date of last follow-up or death for patterns of failure. However, only 6 of 140 local recurrences developed >30 days after a distant recurrence was identified. The majority of patients who failed at distant sites initiated systemic chemotherapy, which could potentially have prevented residual local disease from becoming clinically apparent. Although the majority of patients had chest imaging after developing distant metastases, some patients did not. These, and other issues, will cause actuarial local failure rates to be underestimated. In our study of 975 patients with stage I to II NSCLC, the 5-year rate of local failure was 23%. By 5224 Cancer vember 15, 2009

Local Recurrence of Early Stage Lung Cancer/Kelsey et al FIGURE 3. Kaplan-Meier curve of time to local failure by multivariate risk factors (squamous/large cell histology and lymphovascular space invasion) in the subgroup of patients undergoing at least a lobectomy for stage IB to II nonsmall cell lung cancer is shown. comparison, the 5-year rate of distant failure was 34%. After excluding those patients who underwent less than optimal surgery (sublobar resections and/or positive surgical margins), and excluding patients with stage IA disease (in whom adjuvant therapy is not typically recommended), the risk of local disease recurrence was 27% at 5 years. This appears sufficiently high to warrant further examination into adjuvant local therapy. Furthermore, certain pathologic features may place patients at particularly high risk of local disease recurrence, such as lymphovascular space invasion and/or squamous or large cell histology. Lymphovascular space invasion predisposes to regional (and distant) spread and may be a marker of occult lymph node involvement. Squamous cell carcinomas often arise from a background of premalignant changes induced by tobacco smoke, often referred to as a field defect. This may predispose to local disease recurrence. Such changes are not routinely and methodically described in pathology reports, making it difficult to evaluate this possibility. Our findings that lymphovascular space invasion and squamous/large cell histology are independent predictors of local disease recurrence will need to be validated in independent datasets. Several surgical and pathologic factors have been associated with a higher risk of local recurrence, the most consistent being limited surgery (wedge or segmentectomy), 14,19,20,26,27 positive surgical margins, 27-30 and lack of or limited mediastinal sampling. 18,20 In the current study, sublobar resections were associated with a higher risk of local disease recurrence in multivariate analysis but lack of mediastinal sampling was not. Patients with positive surgical margins were more likely to receive RT than patients with negative margins, but the limited number of events made it difficult to evaluate the effectiveness of RT in the presence of positive surgical margins. An accurate assessment of patterns of failure after surgery is helpful to guide postoperative therapy. Given the recognized risk of local recurrence, even in patients with early stage disease, a reanalysis of postoperative RT appears prudent. There is significant concern in proceeding with such a study, and rightly so, given the results of the PORT Meta-analysis Trialists Group indicating a survival detriment in patients with early stage NSCLC. 4 Nevertheless, there are several reasons to support a reassessment of postoperative RT. First, the current analysis suggests that local failure occurs relatively often after lobectomy for early stage disease and appears to be higher than is widely appreciated. Second, just as chemotherapy approaches have improved over the past 20 years (alkylating agents replaced by cisplatin-based regimens), so have RT techniques. Advances in 3-dimensional (3D) treatment planning and radiation delivery systems appear to have improved the therapeutic ratio of RT. Two small randomized trials in which modern 3D treatment planning was used both demonstrated improved local control and trends toward improved survival with postoperative RT in patients with lung cancer. 2,31 Furthermore, prior studies of postoperative RT used generous RT fields, which were directed at the entire mediastinum and often the supraclavicular fossa. This was logical because these studies were largely conducted in the pre-ct era, and there was much uncertainty regarding lobe-specific lymphatic spread 32,33 and patterns of failure after surgery. 34 A better understanding of these issues provides an opportunity to more carefully design RT fields that target the intrathoracic sites at highest risk, and by so doing, enhance the therapeutic ratio. A recent randomized trial of postoperative RT in patients with stage I NSCLC using very small customized RT fields 2 demonstrated a significant improvement in local control without excess toxicity using this approach. Third, the ability of local measures to enhance survival is limited by the risk of distant failure. Local disease recurrence may not influence survival if the risk of distant recurrence is high. The increased use of adjuvant Cancer vember 15, 2009 5225

Original Article chemotherapy, with the associated reduction in systemic risk, may allow improvements in local control to be more readily translated into improvements in overall survival. In the recently published ANITA trial, postoperative RT appeared to improve survival for patients with stage III, but not stage II, NSCLC after resection and adjuvant chemotherapy. 35 However, because RT was not a randomized variable, it is not possible to accurately evaluate the effect on either group. Chemotherapy also has the potential to decrease the risk of local failure. The ANITA trial, which to our knowledge is the only recent randomized trial to report patterns of failure, did demonstrate a reduction in the risk of local failure with cisplatin-based chemotherapy. 7 Finally, most patients with early stage NSCLC will not develop local disease recurrence. Identifying those that will, with a high degree of certainty, is not currently possible. Recent studies have shown that gene expression profiles can distinguish patients at particularly high risk of disease recurrence who are more likely to benefit from adjuvant treatment. 36,37 To the best of our knowledge, these genomic signatures have not distinguished between local and distant disease recurrence. If patients at high risk of local recurrence could be reliably identified, this would further improve the therapeutic ratio and may facilitate clinical trials examining the role of postoperative RT in early stage NSCLC. As with any retrospective analysis, the current study has limitations. Although nearly 50% of local recurrences were confirmed with biopsy, the remaining were scored using imaging studies that can overestimate, as well as underestimate, the extent of disease. Furthermore, clinical judgment was necessary when evaluating patterns of failure, in particular distinguishing recurrent disease from second primary lung cancers. Every case was reviewed by 2 physicians to ascertain, to the best of our ability, an accurate description of patterns of failure. We acknowledge that there may have been errors in this process. Furthermore, although pathologists with an expertise in lung cancer rendered an initial interpretation of the pathologic findings in the majority of cases, we did not re-review the pathologic specimens. The primary strength of this analysis is the number of patients analyzed. Indeed, to the best of our knowledge, the current study represents 1 of the largest patterns of failure analyses of patients undergoing surgery for early stage NSCLC. Conclusions Greater than 50% of disease recurrences occurring after surgery for early stage NSCLC involve local sites. The risk of local disease recurrence after surgery for patients with stage I to II NSCLC in the current series was 23%. Despite optimal surgery, current candidates for adjuvant chemotherapy are at relatively high risk of local recurrence. Pathologic factors, in particular squamous or large cell histology and lymphovascular space invasion, may help distinguish those patients at highest risk. Conflict of Interest Disclosures The authors made no disclosures. References 1. Feng QF, Wang M, Wang LJ, et al. A study of postoperative radiotherapy in patients with non-small-cell lung cancer: a randomized trial. Int J Radiat Oncol Biol Phys. 2000;47:925-929. 2. Trodella L, Granone P, Valente S, et al. Adjuvant radiotherapy in non-small cell lung cancer with pathological stage I: definitive results of a phase III randomized trial. Radiother Oncol. 2002;62:11-19. 3. Van Houtte P, Rocmans P, Smets P, et al. Postoperative radiation therapy in lung cancer: a controlled trial after resection of curative design. Int J Radiat Oncol Biol Phys. 1980;6:983-986. 4. PORT Meta-analysis Trialists Group. Postoperative radiotherapy in non-small-cell lung cancer: systematic review and meta-analysis of individual patient data from 9 randomised controlled trials. Lancet. 1998;352:257-263. 5. n-small Cell Lung Cancer Collaborative Group. Chemotherapy in non-small cell lung cancer: a meta-analysis using updated data on individual patients from 52 randomised clinical trials. BMJ. 1995;311:899-909. 6. Arriagada R, Bergman B, Dunant A, Le Chevalier T, Pignon JP, Vansteenkiste J. Cisplatin-based adjuvant chemotherapy in patients with completely resected non-small-cell lung cancer. N Engl J Med. 2004;350:351-360. 7. Douillard JY, Rosell R, De Lena M, et al. Adjuvant vinorelbine plus cisplatin versus observation in patients with completely resected stage IB-IIIA non-small-cell lung cancer (Adjuvant Navelbine International Trialist Association [ANITA]): a randomised controlled trial. Lancet Oncol. 2006;7:719-727. 8. Winton T, Livingston R, Johnson D, et al. Vinorelbine plus cisplatin vs. observation in resected non-small-cell lung cancer. N Engl J Med. 2005;352:2589-2597. 9. Collett D. Modeling Survival Data in Medical Research. 2nd ed. London: Chapman & Hall; 2003. 5226 Cancer vember 15, 2009

Local Recurrence of Early Stage Lung Cancer/Kelsey et al 10. Therneau TM, Grambsch PM. Modeling Survival Data. New York: Springer; 2000. 11. Altman DG. Practical Statistics for Medical Research. London: Chapman & Hall; 1991. 12. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2008. CA Cancer J Clin. 2008;58:71-96. 13. Pignon JP, Tribodet H, Scagliotti GV, et al. Lung adjuvant cisplatin evaluation: a pooled analysis by the LACE Collaborative Group. J Clin Oncol. 2008;26:3552 3559. 14. Ginsberg RJ, Rubinstein LV. Randomized trial of lobectomy versus limited resection for T1 N0 non-small cell lung cancer. Lung Cancer Study Group. Ann Thorac Surg. 1995;60:615-622; discussion 622-623. 15. Lafitte JJ, Ribet ME, Prevost BM, Gosselin BH, Copin MC, Brichet AH. Postresection irradiation for T2 N0 M0 non-small cell carcinoma: a prospective, randomized study. Ann Thorac Surg. 1996;62:830-834. 16. Feld R, Rubinstein LV, Weisenberger TH. Sites of recurrence in resected stage I non-small-cell lung cancer: a guide for future studies. J Clin Oncol. 1984;2:1352-1358. 17. Kato H, Ichinose Y, Ohta M, et al. A randomized trial of adjuvant chemotherapy with uracil-tegafur for adenocarcinoma of the lung. N Engl J Med. 2004;350:1713-1721. 18. Lardinois D, Suter H, Hakki H, Rousson V, Betticher D, Ris HB. Morbidity, survival, and site of recurrence after mediastinal lymph-node dissection versus systematic sampling after complete resection for non-small cell lung cancer. Ann Thorac Surg. 2005;80:268-274; discussion 274-275. 19. Harpole DH Jr, Herndon JE Jr, Young WG Jr, Wolfe WG, Sabiston DC Jr. Stage I nonsmall cell lung cancer. A multivariate analysis of treatment methods and patterns of recurrence. Cancer. 1995;76:787-796. 20. Martini N, Bains MS, Burt ME, et al. Incidence of local recurrence and second primary tumors in resected stage I lung cancer. J Thorac Cardiovasc Surg. 1995;109:120-129. 21. Immerman SC, Vanecko RM, Fry WA, Head LR, Shields TW. Site of recurrence in patients with stages I and II carcinoma of the lung resected for cure. Ann Thorac Surg. 1981;32:23-27. 22. Sawyer TE, Bonner JA, Gould PM, et al. Factors predicting patterns of recurrence after resection of N1 non-small cell lung carcinoma. Ann Thorac Surg. 1999;68:1171-1176. 23. Luzzi L, Voltolini L, Campione A, et al. Pneumonectomy vs lobectomy in the treatment of pathologic N1 NSCLC: could the type of surgical resection dictate survival? J Cardiovasc Surg (Torino). 2003;44:119-123. 24. Kim YT, Kang CH, Sung SW, Kim JH. Local control of disease related to lymph node involvement in non-small cell lung cancer after sleeve lobectomy compared with pneumonectomy. Ann Thorac Surg 2005;79:1153-1161; discussion 1153-1161. 25. Gelman R, Gelber R, Henderson IC, Coleman CN, Harris JR. Improved methodology for analyzing local and distant recurrence. J Clin Oncol. 1990;8:548-555. 26. Warren WH, Faber LP. Segmentectomy versus lobectomy in patients with stage I pulmonary carcinoma. Five-year survival and patterns of intrathoracic recurrence. J Thorac Cardiovasc Surg 1994;107:1087-1093; discussion 1093-1094. 27. Lee JH, Machtay M, Kaiser LR, et al. n-small cell lung cancer: prognostic factors in patients treated with surgery and postoperative radiation therapy. Radiology. 1999;213: 845-852. 28. Hsu HC, Wang CJ, Huang EY, Sun LM. Post-operative adjuvant thoracic radiotherapy for patients with completely resected non-small cell lung cancer with nodal involvement: outcome and prognostic factors. Br J Radiol. 2004;77: 43-48. 29. Sawyer TE, Bonner JA, Gould PM, et al. Effectiveness of postoperative irradiation in stage IIIA non-small cell lung cancer according to regression tree analyses of recurrence risks. Ann Thorac Surg 1997;64:1402-1407; discussion 1407-1408. 30. Sawyer TE, Bonner JA, Gould PM, et al. The impact of surgical adjuvant thoracic radiation therapy for patients with nonsmall cell lung carcinoma with ipsilateral mediastinal lymph node involvement. Cancer. 1997;80:1399-1408. 31. Mayer R, Smolle-Juettner FM, Szolar D, et al. Postoperative radiotherapy in radically resected non-small cell lung cancer. Chest. 1997;112:954-959. 32. Hata E, Hayakawa K, Miyamoto H, et al. Rationale for extended lymphadenectomy for lung cancer. Theor Surg. 1990;5:19 25. 33. hl-oser HC. An investigation of the anatomy of the lymphatic drainage of the lungs as shown by the lymphatic spread of bronchial carcinoma. Ann R Coll Surg Engl. 1972;51:157 176. 34. [no authors listed]. Patterns of failure after resection of non-small-cell lung cancer: implications for postoperative radiation therapy volumes. Int J Rad Oncol Biol Phys. 2006;65:1097 1105. 35. Douillard JY, Rosell R, De Lena M, Riggi M, Hurteloup P, Mahe MA. Impact of postoperative radiation therapy on survival in patients with complete resection and stage I, II, or IIIA non-small-cell lung cancer treated with adjuvant chemotherapy: the adjuvant Navelbine International Trialist Association (ANITA) Randomized Trial. Int J Radiat Oncol Biol Phys. 2008;72:695-701. 36. Potti A, Mukherjee S, Petersen R, et al. A genomic strategy to refine prognosis in early-stage non-small-cell lung cancer. N Engl J Med. 2006;355:570-580. 37. Chen HY, Yu SL, Chen CH, et al. A 5-gene signature and clinical outcome in non-small-cell lung cancer. N Engl J Med. 2007;356:11-20. Cancer vember 15, 2009 5227