Lymph Node Evaluation by Open or Video- Assisted Approaches in 11,500 Anatomic Lung Cancer Resections

Transcription

1 RICHARD E. CLARK AWARD Lymph Node Evaluation by Open or Video- Assisted Approaches in 11,500 Anatomic Lung Cancer Resections Daniel J. Boffa, MD, Andrzej S. Kosinski, PhD, Subroto Paul, MD, John D. Mitchell, MD, and Mark Onaitis, MD Department of Thoracic Surgery, Yale University School of Medicine, New Haven, Connecticut; Departments of Biostatistics & Bioinformatics and Thoracic Surgery, Duke University, Durham, North Carolina; Department of General Thoracic Surgery, New York-Presbyterian/Weill Cornell, New York, New York; and Division of Cardiothoracic Surgery, University of Colorado Denver School of Medicine, Aurora, Colorado Background. Unsuspected lymph node metastases are found in the surgical specimens of 10% to 25% clinical stage I lung cancers. Video-assisted thoracic surgery (VATS) is a minimally invasive alternative to thoracotomy. Because detection of clinically occult metastases is dependent on the completeness of surgical lymph node dissection, the influence of surgical approach on nodal evaluation is of interest. We determined the frequency of nodal metastases identified in clinically node-negative tumors by thoracotomy ( open ) and VATS approaches to approximate the completeness of surgical nodal dissections. Methods. The Society of Thoracic Surgery database was queried for lobectomies and segmentectomies from 2001 to Results. A total of 11,531 (7,137 open and 4,394 VATS) clinical stage I primary lung cancers were resected. Nodal upstaging was seen in 14.3% (1,024) in the open group and 11.6% (508) in the VATS group (p < 0.001). Upstaging from N0 to N1 was more common in the open group (9.3% versus 6.7%; p < 0.001); however, upstaging from N0 to N2 was similar (5.0% open and 4.9% VATS; p 0.52). Among 2,745 propensity-matched pairs, N0 to N1 upstaging remained less common with VATS (6.8% versus 9%; p 0.002). Conclusions. During lobectomy or segmentectomy for clinical N0 lung cancer, mediastinal nodal evaluation by VATS and thoracotomy results in equivalent upstaging. In contrast, lower rates of N1 upstaging in the VATS group may indicate variability in the completeness of the peribronchial and hilar lymph node evaluation. Systematic hilar dissection is encouraged, particularly as more surgeons adopt the VATS approach. (Ann Thorac Surg 2012;94:347 53) 2012 by The Society of Thoracic Surgeons The completeness of a cancer staging evaluation affects the accuracy of prognosis as well as the ability to optimize treatment. Nodal status in surgically managed lung cancer is particularly important, as the use of adjuvant chemotherapy in the setting of nodal metastases can improve survival [1, 2]. Noninvasive staging modalities, such as computed tomography and positron emission tomography scanning, are insufficient as they fail to identify nodal metastases in more than 20% of primary lung cancer patients [3]. Therefore a comprehensive surgical lymph node evaluation is a critical component of the lung cancer resection. Video-assisted thoracic surgery (VATS) is a safe, minimally invasive alternative to thoracotomy ( open approach) [4]. In fact, 80% of graduating dedicated thoracic surgeons consider themselves to be proficient in VATS Accepted for publication April 16, Presented at the Forty-eighth Annual Meeting of The Society of Thoracic Surgeons, Fort Lauderdale, FL, Jan 28 Feb 1, Winner of a Richard E. Clark Award. Address correspondence to Dr Boffa, Department of Thoracic Surgery, Yale University School of Medicine, 330 Cedar St BB205, PO Box , New Haven, CT ; daniel.boffa@yale.edu. lobectomy [5]. Differences in the surgical perspective, range of motion, and instrumentation associated with the VATS approach pose potential challenges to the identification and removal of the regional lymph nodes. Therefore, the efficacy of surgical lymph node evaluation during a VATS procedure for primary lung cancer is of interest. Nodal upstaging, or the identification of lymph node metastases in the surgical specimens of patients clinically staged to be node negative, is known to occur in a significant proportion of patients [6]. For example, 28% of clinical stage I lung cancer patients were upstaged (14% stage II and 14% stage III) in a Cancer and Leukemia Group B prospective clinical trial (CALGB 9761) [7]. Because imaging did not identify the nodal metastases, nodal upstaging is completely dependent on the efficacy of the surgical lymph node evaluation. Therefore, the prevalence of nodal upstaging may be used as surrogate for the completeness of nodal evaluation. To compare the completeness of the surgical lymph node evaluation during anatomic resection of primary lung cancer by open and VATS approaches, the rates of nodal upstaging were studied in a large prospective multiinstitutional database by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc

2 348 CLARK AWARD BOFFA ET AL Ann Thorac Surg LYMPH NODE EVALUATION BY OPEN OR VATS APPROACHES 2012;94: Patients and Methods Patients The Society of Thoracic Surgeons General Thoracic Database (STS-GTD) was created in 1999 as an expansion of The Society of Thoracic Surgeons prospective Cardiac Surgery database. Data collection templates have evolved during that time period, with the three earliest versions being folded into version 2.07 (56% of current study population) and in December of 2009 the adoption of version (44% of current study population) [8]. The STS-GTD was queried for anatomic pulmonary resections between 2001 and 2010 for treatment-naïve, clinically staged T1N0M0 or T2N0M0 primary lung cancer. Histologic diagnosis is not captured by these versions of the database; therefore, primary lung cancer cannot be further separated into small cell or non small cell lung cancer subtypes. Patients in whom pathologic N stage was missing were pathologically staged to be N3 (8 patients), and those who received preoperative treatment with chemotherapy or radiation were excluded from this study. Pneumonectomies (VATS 17 patients), bilobectomies (VATS 70), and sleeve lobectomies (VATS 10) were underrepresented in the VATS group and therefore were excluded from both cohorts. Segmentectomies and lobectomies were analyzed as a combined group because the latter version of the database (2.081) did not distinguish VATS lobectomies from VATS segmentectomies. More specifically, the coding nomenclature in version clusters VATS lobectomies and segmentectomies under a single designation (CPT code of 32663). For reference, the rates of segmentectomy in the previous version of the database (STS-GTD 2.07) were similar (open, 6.5%; VATS, 8%). A second analysis was performed in the STS-GTD of treatment-naïve, clinical N1 patients undergoing lobectomy or segmentectomy for primary lung cancer to assess pathologic confirmation of clinical N1 based on surgical approach as a second surrogate for the completeness of the surgical lymph node evaluation. Statistics All statistical analysis were approved through the STS- GTD committee and performed as a part of a contracted agreement with the Duke Clinical Research Institute. Comparisons of subject characteristics between groups were made using the two-sample Student s t test for continuous variables and the 2 test for categorical variables. Probability values were not adjusted for multiplicity of comparisons, and a probability value of less than 0.05 is considered statistically significant. To identify correlates of upstaging and adjust for patient characteristics, a logistic regression analysis was performed with the upstaging of cn0 to pn1 as a dichotomous variable as the outcome (dependent variable) and independent variables consisting of preoperative subject characteristics and the approach (VATS versus open) as independent variables. When developing the multivariable model we first considered univariable logistic regressions to evaluate univariable associations of each variable with outcome. A total of 9,893 patients had sufficiently complete data to be included in the multivariable analysis. The multivariable analysis initially considered variables with a univariable probability value of less than 0.30; subsequently, a parsimonious model was developed from various combinations of such candidate correlates by retaining predictors with a probability value of less than 0.05, and the approach (VATS versus open) was added to the developed model. The results of logistic regression modeling are summarized with odds ratio (OR), 95% confidence interval (CI), and probability value. To account for differences in the open and VATS groups in an alternative way, a propensity-matched analysis was performed. First, a logistic regression model was considered with approach (VATS versus open) as the dependent variable and preoperative subject characteristics as independent variables. Independent variables included use of mediastinal staging procedure, clinical T stage, age, sex, race, body mass index, forced expiratory volume in 1 second as percent predicted, year of surgery, clinical status at time of surgery, general thoracic reoperation, prior thoracic surgery, laterality, American Society of Anesthesiologists risk class, smoking status, Zubrod score, weight loss in 3 months before admission, hypertension, use of steroids, congestive heart failure, coronary artery disease, peripheral vascular disease, and renal insufficiency. Subsequently, propensity score (ie, estimated probability to receive VATS) was estimated based on this model. Second, pairs of VATS and open subjects were matched according to the value of the propensity score. A total of 7,896 patients contained sufficiently complete data to be included in the propensity score based matching (4,605 open and 3,291 VATS). Patients were matched using a greedy 5:1 digit algorithm [9], similar to the propensity strategy used in a previous report from the STS-GTD [4]. The majority of pairs had differences in propensity score less than , and matched pairs were not formed if the propensity score difference exceeded 0.1. Third, a univariable comparison between approach and the upstage outcome was performed using conditional logistic regression to account for matching. Balance between VATS and open groups was evaluated with conditional logistic regression as well. To determine whether VATS preferences influenced nodal upstaging, rates were determined across categories of STS-GTD participants, based on the proportion of study cases submitted by a participant that were performed using the VATS approach (0% to 20%, 20% to 40%, 40% to 60%, 60% to 80%, 80% to 100%). A test for trend in proportions of nodal upstaging along these categories was performed for the VATS patients. Results Patients Between 2001 and 2010, a total of 11,531 lobectomies or segmentectomies (7,137 open and 4,394 VATS) for clinical stage I primary lung cancer were recorded in the STS- GTD. The frequency of VATS increased in the study

3 Ann Thorac Surg CLARK AWARD BOFFA ET AL 2012;94: LYMPH NODE EVALUATION BY OPEN OR VATS APPROACHES Table 1. Demographics and Tumor Attributes by Approach Variable Open (n 7,137) VATS (n 4,394) p Value Clinical stage (%) T T Extent of resection, N (%) Lobectomy 6,667 (93.4) 1,948 (92) a Segmentectomy 470 (6.6) 170 (8) a Age at time of surgery (y) b (range, 18 96) (range, 21 95) Sex (% female) Race (%) Caucasian Black Hispanic Asian Body mass index 35 kg/m 2 (%) FEV 1 (% predicted) b Dlco (% predicted) b Year of surgery (%) Clinical status (%) Elective Urgent Emergent General thoracic reoperation (% yes) Laterality of procedure (% right) American Society of Anesthesiologists risk class (%) I II III IV IV Cigarette smoker (%) Never Ever Length of procedure (min) b Med staging procedure (%) c None Day of surgery or 45 days Robot assisted a These data are only available in the first edition of The Society of Thoracic Surgeons database (n 2,118 procedures), as the updated code does not distinguish VATS lobectomy from segmentectomy. b Results presented as mean standard deviation. c Med staging procedure includes mediastinoscopy, anterior mediastinoscopy, mediastinal lymph node biopsy, extended cervical mediastinoscopy, mediastinal lymph node dissection, mediastinal lymph node sampling, thoracoscopy diagnostic mediastinal space with biopsy, endobronchial ultrasound, thoracic lymphadenectomy. Dlco diffusing capacity of the lung for carbon monoxide; FEV 1 forced expiratory volume in 1 second; VATS video-assisted thoracoscopic surgery. cohort during this time period (in 2009, 44% of the current study population was approached by means of VATS). The patient demographics and tumor attributes are listed in Table 1. Several important differences were identified, including a greater frequency of T1 tumors in the VATS cohort than in the open cohort (76% versus 65%; p 0.001). Specific procedures to evaluate mediastinal lymph nodes were performed (on the day of resection or within the 45 days prior) in 69% of open resections and 80% of the VATS group (p 0.001). Mediastinoscopy

4 350 CLARK AWARD BOFFA ET AL Ann Thorac Surg LYMPH NODE EVALUATION BY OPEN OR VATS APPROACHES 2012;94: Table 2. Distribution of Pathologic Nodal Staging (pn) of Clinical N0 (cn0) Lung Cancer Patients a Pathologic Nodal Stage Open (n 7,137) VATS (n 4,394) Total (n 11,531) pn0 6,113 (85.7%) 3,886 (88.4%) 9,999 (86.7%) pn1 666 (9.3%) 293 (6.7%) 959 (8.3%) pn2 358 (5%) 215 (4.9%) 573 (5%) a Results are number (%). VATS video-assisted thoracoscopic surgery. (2.081) of the database. The distribution of nodal status by tumor location was evaluated. Upstaging from cn0 to pn1 remained more common in open than VATS for both upper (9.4% versus 6.2%; p 0.002) and lower lobe tumors (9.9% versus 6.0%; p 0.003); however, the difference was not significant among the 307 middle lobe lesions (6.1% versus 6.3%; p 0.97). When tumor location and upstaging were examined using a logistic regression, the interaction between location and pathologic nodal status (pn0 versus pn1) was not significant (p 0.52). was performed in 10% (744) of the open procedures and 13% (562) of the VATS procedures. The use of robot assistance was minimal. Nodal Upstaging by Approach Overall nodal upstaging was more common in the open procedures, 14.3% (1,024), compared with 11.6% (508) in the VATS group (p 0.001; Table 2). Mediastinal upstaging (cn0 to pn2) or the removal of occult mediastinal lymph node metastases was similar, occurring in 5% of open resections and 4.9% of VATS (p 0.52). Hilar or peribronchial upstaging (cn0 to pn1) was more common with the open approach than with VATS (9.3% versus 6.7%; p 0.001). In an attempt to assess for differences by type of resection (as segmentectomies and lobectomies were analyzed together for this study), the data from version 2.07 were evaluated because VATS lobectomies and segmentectomies were coded separately in this version. During this period, 170 VATS segmentectomies were recorded, and upstaging from cn0 to pn1 was seen in 4% of patients compared with 5.3% among 280 open segmentectomies. Similarly, the lobectomy-only data from the same time frame and database version demonstrated the prevalence of hilar upstaging to be 9.6% in the open cohort and 7.5% in the VATS cohort (p 0.009). To further clarify the importance of tumor status (T stage) on the rates of occult hilar and peribronchial nodal metastases, nodal upstaging from cn0 to pn1 was further analyzed by clinical T stage. Upstaging was more common in open resections in clinical T1 (7.8% open versus 6% VATS; p 0.002) and in T2 tumors (13.6% open versus 10.4% VATS; p 0.013). In the revised version of the STS-GTD (version 2.081, representing 44% of study population), T stage was further stratified by size to comply with the 7th edition of the American Joint Committee on Cancer staging manual [10]. In clinical T1a (1 2cm) upstaging from cn0 to pn1 occurred in 7.4% of open procedures and 5.2% of VATS procedures (p 0.02); in T1b ( 2 3cm), 8.8% of open procedures and 7.1% of VATS procedures (p 0.33); in T2a ( 3 5cm), 11.5% of open procedures and 5.7 of VATS procedures (p ); and in T2b ( 5 7cm), 14.8% of open procedures and 15.2% of VATS procedures (not significant). It is worth noting that the number of patients was low in T2b (open, 189; VATS, 79). Tumor location (upper lobe, middle lobe, or lower lobe) was likewise only available in the revised version Multivariable Analysis of Hilar and Peribronchial Upstaging (cn0 to pn1) Relationships between the demographics and tumor variables and the rate of hilar and peribronchial nodal upstaging were evaluated by multivariable analysis (Fig 1). Resections by VATS were significantly less likely to result in upstaging to N1 than open resections (OR, 0.73; 95% CI, 0.62 to 0.85; p 0.001). Several other variables were identified as significant predictors of upstaging, including left-sided resections (OR, 1.33; 95% CI, 1.16 to 1.53; p 0.001). Propensity Matching of cn0 to pn1 In another attempt to eliminate confounding between VATS and open groups, a propensity-matched analysis was performed. Matched pairs were based on the probability of VATS approach estimated with a model built from 23 preoperative variables. A total of 2,745 propensity-matched pairs were studied (5,490 total subjects). Using the matched data, the hilar and peribronchial lymph node upstaging remained less common in the VATS than in the open cases (6.8% versus 9%; OR, 0.74; 95% CI, 0.61 to 0.90; p 0.002). There was no difference in mediastinal upstaging. VATS vs Open Med Staging done ct2 vs ct1 Female vs Male BMI > 35 Le vs Right 95% CI OR Fig 1. Multivariable analysis of upstaging from cn0 to pn1. For definition of med staging, see Table 1 footnote. (BMI body mass index; CI confidence interval; OR odds ratio; VATS videoassisted thoracoscopic surgery.)

5 Ann Thorac Surg CLARK AWARD BOFFA ET AL 2012;94: LYMPH NODE EVALUATION BY OPEN OR VATS APPROACHES Table 3. Distribution of Pathologic Nodal Staging (pn) of Clinical N1 (cn1) Lung Cancer Patients a Pathologic Nodal Stage Confirmation of Clinical N1 Pathologic confirmation of clinically detected lymph node metastases is similarly dependent on the completeness of the surgical resection. Hilar or peribronchial nodal metastases were identified in the surgical specimens of clinical stage N1 (cn1) patients more often by open approach than by VATS (54% versus 42%; p 0.002; Table 3). Conversely, downstaging of clinically node-positive tumors (cn1) to pathologically node-negative tumors (cn1 to pn0) was more common by VATS approach (30% open versus 43%VATS). Of note, upstaging from cn1 to pn2 was similar: 15.6% in open and 15.3% in VATS. Influence of Practice Patterns The study cases were submitted by 167 participants to the STS-GTD (of note, a participant may represent an individual, practice, section, department, and so forth). Participants were separated into five categories based on the fraction of study cases that were performed by VATS approach (0% to 20%, 20% to 40%, 40% to 60%, 60% to 80%, 80% to 100%). Rates of hilar or peribronchial upstaging (cn0 to pn1) were determined for each of the five participant categories (Fig 2). Upstaging from cn0 to pn1 occurred more frequently as the use of VATS by the submitting participant increased (p 0.004). In fact, when 989 VATS cases from 18 VATS-predominant participants (defined as 80% of submitted study cases by VATS) were compared with 3,668 open cases from open-predominant participants ( 80% open), the prevalence of upstaging from cn0 to pn1 is identical (8.7%). On the other hand, when a VATSpredominant participant submitted an open case (n 102), the prevalence of cn0 to pn1 was 17%. Comment Open (n 684) VATS (n 235) Total (n 919) pn0 206 (30.1) 100 (42.6) 306 (33.3) pn1 371 (54.3) 99 (42.1) 470 (51.1) pn2 107 (15.6) 36 (15.3) 143 (15.6) a Results are number (%). VATS video-assisted thoracoscopic surgery. The current study identified similar rates of mediastinal upstaging of clinical stage I lung cancers by open or VATS approaches. On the other hand, hilar and peribronchial upstaging occurred less often, and clinically detected N1 lymph node metastases were less likely to be confirmed by the VATS approach. Together these findings call into question the completeness of the surgical lymph node evaluation of the N1 nodal territories by VATS. This finding contradicts numerous studies that have reported that open and VATS approaches result in a similar number lymph nodes and lymph node stations being evaluated [11 14]. Interestingly, when the identification of occult lymph node metastases was compared in these same studies, the VATS approach was consistently less likely to result in nodal upstaging than the open approach. Upstaging is completely dependent on the accuracy of clinical staging. The STS-GTD does not capture clinical staging methodology, which may differ by approach. For example, a report from the Surveillance Epidemiology and End Results database suggests lung cancer patients approached by VATS are more likely to have a positron emission tomography scan than those approached by the open approach [15]. In fairness, positron emission tomography did not improve staging accuracy in the previously referenced CALGB 9761 of clinical stage I lung cancer (31% false negative rate, n 61) [7] or a retrospective positron emission tomography computed tomography study (nodal upstaging of 14.3%, which is the same as the STS-GTD open cohort) [16]. Nonetheless, without comparing the staging methodology, it is not possible to exclude important differences that could affect clinical staging accuracy. Perhaps the most critical assumption of this study is the ability to control for factors that may distort the relationship between approach and upstaging. At the crux of this matter is the motivation for a surgeon to select a particular approach (open versus VATS). Bias that is not accounted for (such as surgeons avoiding VATS for reasons that are not captured by STS-GTD, but correlate with an increased prevalence of occult nodal metastases) could exaggerate differences in upstaging. Tumor location within a lobe is one such example, as peripheral lesions appear to be less likely to be associated with occult mediastinal nodal metastases than central lesions [17, 18], and this variable is not captured by the STS-GTD. If a surgeon preferentially performed VATS for peripheral T1 lesions, this unaccounted for selection bias of tumors with fewer occult nodal metastases would result in less upstaging in the VATS cohort (and routing central lesions to open would increase upstaging in this cohort). In fairness, the association between central tumor location and radiographically occult lymph node metastases has primarily been made for occult mediastinal lymph node metastases, and not all studies made this association [16, 19]. Perhaps the best indication that some approach bias took place is seen in the open cases submitted by the % cn0 pn1 P = % VATS Fig 2. Prevalence of nodal upstaging (cn0 to pn1) by participant use of video-assisted thoracoscopic surgery (VATS). 351

6 352 CLARK AWARD BOFFA ET AL Ann Thorac Surg LYMPH NODE EVALUATION BY OPEN OR VATS APPROACHES 2012;94: VATS-predominant participants ( 80% of submitted study cases by VATS). When a VATS-predominant participant submitted an open procedure, the frequency of hilar and peribronchial upstaging (cn0 to pn1) was 17% (almost twice the average rate of the open cohort), implying these cases with a higher risk of occult nodal metastases were intentionally directed toward the open approach. Alternatively, the least bias would be expected when approach-predominant surgeons submitted cases by their preferred approach. This is in fact what was seen as both open cases by open-predominant participants ( 80% open) and VATS cases by VATS-predominant participants resulted in 8.7% cn0 to pn1 upstaging. The practice data can also be interpreted as experience with the VATS approach, as surgeons tend to increase the use of a technique as they gain familiarity. In this way, the observation that the VATS-predominant surgeons (and therefore the most experienced with the VATS approach) identified a similar number of occult nodal metastases as the open group suggests that equivalent nodal staging may be possible with increasing experience. It is worth noting that although 18 participants contributed cases to the highest group, a single participant was responsible for 46% of the cases in the VATS-predominant group ( 80% of submitted cases by VATS). With this high-volume participant removed, the trend remained significant. Rather than an intrinsic difference in the ability to assess the N1 nodal territories, the VATS surgical lymph node evaluation may in fact be a less recognized component to the learning curve of this approach. The STS-GTD does not contain information regarding conversion rates. This is potentially important as cases that were converted to open because of hilar or peribronchial disease, or other factors that were associated with occult nodal metastases, would be coded as open. Comparable rates of mediastinal upstaging suggest thoracotomy and VATS approaches achieve similarly complete nodal evaluations of the mediastinum during lobectomy for what is clinically determined to be early stage primary lung cancer. On the other hand, lower rates of hilar and peribronchial nodal upstaging and less pathologic confirmation of clinical N1 disease suggest variability in the completeness of the N1 nodal evaluation during VATS lobectomy. The extent to which this phenomenon is intrinsic to the approach is unclear, as differences in hilar and peribronchial upstaging appear to be exaggerated by approach bias and minimized by increasing experience with VATS approach. Surgeons are encouraged to apply a systematic approach to the hilar and peribronchial lymph node dissection during VATS lobectomy for lung cancer, particularly as they are adopting this approach. Addendum Additional tables for multivariable analysis and propensity matching are available on request to the corresponding author. References 1. 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: Butts CA, Ding K, Seymour L, et al. Randomized phase III trial of vinorelbine plus cisplatin compared with observation in completely resected stage IB and II non-small-cell lung cancer: updated survival analysis of JBR-10. J Clin Oncol 2010;28: Silvestri GA, Gould MK, Margolis ML, et al. Noninvasive staging of non-small cell lung cancer: ACCP evidencedbased clinical practice guidelines (2nd edition). Chest 2007; 132(3 Suppl):178S 201S. 4. Paul S, Altorki NK, Sheng S, et al. Thoracoscopic lobectomy is associated with lower morbidity than open lobectomy: a propensity-matched analysis from the STS database. J Thorac Cardiovasc Surg 2010;139: Boffa DJ, Gangadharan S, Kent M, et al. Self-perceived video-assisted thoracic surgery lobectomy proficiency by recent graduates of North American thoracic residencies. Interact Cardiovasc Thorac Surg 2012 Mar 1. [Epub ahead of print]. 6. Rocha AT, McCormack M, Montana G, Schreiber G. Association between lower lobe location and upstaging for earlystage non-small cell lung cancer. Chest 2004;125: D Cunha J, Herndon JE 2nd, Herzan DL, et al. Poor correspondence between clinical and pathologic staging in stage 1 non-small cell lung cancer: results from CALGB 9761, a prospective trial. Lung Cancer 2005;48: Boffa DJ, Allen MS, Grab JD, Gaissert HA, Harpole DH, Wright CD. Data from The Society of Thoracic Surgeons General Thoracic Surgery database: the surgical management of primary lung tumors. J Thorac Cardiovasc Surg 2008;135: Parsons LS. Reducing bias in a propensity score matchedpair sample using greedy matching techniques. Proceedings of the Twenty-Sixth Annual SAS Users Group International Conference, Cary, NC: SAS Institute Inc; Available at Detterbeck FC, Boffa DJ, Tanoue LT. The new lung cancer staging system. Chest 2009;136: Watanabe A, Koyanagi T, Ohsawa H, et al. Systematic node dissection by VATS is not inferior to that through an open thoracotomy: a comparative clinicopathologic retrospective study. Surgery 2005;138: Denlinger CE, Fernandez F, Meyers BF, et al. Lymph node evaluation in video-assisted thoracoscopic lobectomy versus lobectomy by thoracotomy. Ann Thorac Surg 2010;89: D Amico TA, Niland J, Mamet R, Zornosa C, Dexter EU, Onaitis MW. Efficacy of mediastinal lymph node dissection during lobectomy for lung cancer by thoracoscopy and thoracotomy. Ann Thorac Surg 2011;92: Shiraishi T, Hiratsuka M, Yoshinaga Y, Yamamoto S, Iwasaki A, Shirakusa T. Thoracoscopic lobectomy with systemic lymph node dissection for lymph node positive non-small cell lung cancer is thoracoscopic lymph node dissection feasible? Thorac Cardiovasc Surg 2008;56: Farjah F, Wood DE, Mulligan MS, et al. Safety and efficacy of video-assisted versus conventional lung resection for lung cancer. J Thorac Cardiovasc Surg 2009;137: Park HK, Jeon K, Koh WJ, et al. Occult nodal metastasis in patients with non-small cell lung cancer at clinical stage IA by PET/CT. Respirology 2010;15: Lee PC, Port JL, Korst RJ, Liss Y, Meherally DN, Altorki NK. Risk factors for occult mediastinal metastases in clinical stage I non-small cell lung cancer. Ann Thorac Surg 2007;84:

7 Ann Thorac Surg CLARK AWARD BOFFA ET AL 2012;94: LYMPH NODE EVALUATION BY OPEN OR VATS APPROACHES 18. Al-Sarraf N, Aziz R, Gately K, et al. Pattern and predictors of occult mediastinal lymph node involvement in non-small cell lung cancer patients with negative mediastinal uptake on positron emission tomography. Eur J Cardiothorac Surg 2008;33: Kanzaki R, Higashiyama M, Fujiwara A, et al. Occult mediastinal lymph node metastasis in NSCLC patients diagnosed as clinical N0 1 by preoperative integrated FDG-PET/CT and CT: risk factors, pattern, and histopathological study. Lung Cancer 2011;71: DISCUSSION DR GAIL E. DARLING (Toronto, Ontario, Canada): Thank you for the opportunity to discuss your paper, Dr Boffa. It is excellent work. I have no disclosures. Doctor Mack, Dr Fullerton, members, and guests. Using the STS (Society of Thoracic Surgeons) General Thoracic Database, Dr Boffa has presented a comparison of lymph node assessment during pulmonary resection by open or VATS (video-assisted thoracoscopic surgery) techniques. At the STS meeting in 2005, Alex Little analyzed data from the Commission on Cancer Patient Care Evaluations study and reported that only 27% of patients having surgical treatment for non small cell lung cancer had a preoperative mediastinoscopy, but less than half of those patients actually had any lymph nodes biopsied, and at the time of lung resection, only 58% of patients had lymph nodes sampled or removed from the mediastinum. Today we have heard that the surgeons who voluntarily submit their data to the STS General Thoracic Database are doing a good job of assessing mediastinal lymph nodes. We have clearly come a long way. Recognizing that this is not a randomized comparison and that data submission is voluntary, this is, nevertheless, a real world snapshot of what is being done by those surgeons who submit their data. Eighty percent of patients having VATS resections and 59% of those having open resections had a mediastinal lymph node staging procedure within 45 days of their resection. Instead of counting or weighing lymph nodes, which has its hazards, Dr Boffa has used nodal upstaging as a surrogate for completeness of nodal evaluation in comparing the two approaches. Eighty-six percent of open and 88% of VATS resection patients were pathological N0. Only 5% of patients in both groups were upstaged at resection to N2. This number is almost identical to the number obtained in the ACOSOG (American College of Surgeons Oncology Group) Z0030 randomized trial wherein rigorous pre-resection staging was performed. Therefore, I think that is a very reliable number. Where we see the difference, as has been pointed out by Dr Boffa, is in the N1 disease. Patients having open resections were more likely to have N1 disease identified than patients having VATS resections. This tells us that although we are doing a good job of staging the mediastinum, we do need to do a better job of sampling the hilar and interlobar lymph nodes, at least for the VATS resections. This has huge implications when doing VATS sublobar resections. If less than a lobectomy is being considered, we need to be sure that the N1 nodes are truly negative. Otherwise the patients may as well have SBRT (stereotactic body radiotherapy). And this is key as we embark on a randomized trial of SBRT versus sublobar resection. It is encouraging to note that as surgeons use VATS more frequently, the rate of upstaging from N0 to N1 goes up, and for those surgeons who do over 80% of their pulmonary resections by VATS, there is no difference in the rate of upstaging from N0 to N1. This data is encouraging, but we must remember that the STS database is voluntary and all surgeons who submit their data are board certified in cardiothoracic surgery. What would the results be if we queried the SEER (Surveillance, Epidemiology, and End Results) database? I have several questions, and I think you have actually answered one of them already. Were all patients staged with PET-CT (positron emission tomography computed tomography)? We know that this has made a huge difference in our clinical staging. Two, the use of mediastinoscopy you didn t give us this data in the presentation was low in both groups; I think 10% and 12%. What were the methods used to evaluate the mediastinal lymph nodes? If EBUS-TBNA (endobronchial ultrasound with real-time guided transbronchial needle aspiration) and EUS (endoscopic ultrasound) were used, was this done by the surgeon or by the pulmonologist or gastroenterologist? And do you have any data on nodal upstaging in open versus VATS segmentectomies? Thank you for the privilege of discussing your paper. DR BOFFA: Thank you, Dr Darling. With respect to the use of PET-CT, there was no data in the STS database concerning the use of CT scanning, PET scanning, or any imaging modalities. I do think it is worth pointing out that in the CALGB (Cancer and Leukemia Group B) 9761 study that I referenced at the beginning, there was a false-negative rate of PET-CT in that study that was 31%, and a subsequent retrospective study also showed there was a relatively high rate of false-negative studies in the hilum when looking at [what was] presumed to be early stage lung cancer. With respect to the methodology for evaluating the mediastinum, the nomenclature has changed around a little bit as the database template has evolved, but two thirds of the mediastinal evaluations were coded as either thoracic lymphadenectomy or lymph node dissections that were done at the time of the operation. You mentioned the low mediastinoscopy rate of 10% and 12%. It is challenging to capture procedures done prior to the day of surgery, as the STS database is a procedure-based database, and so it takes a bit of finagling to actually link patients to procedures that have happened on a day other than the day of their dominant procedure. With respect to the use of EBUS, this is a great question, because EBUS actually can evaluate the hilar lymph nodes. Less than 1% of the patients received EBUS in both cohorts, but, to be honest, I didn t think to evaluate EBUS as an independent variable for upstaging, because it is conceivable that in EBUSstaged patients there would be fewer surprise hilar lymph nodes because they would have likely been evaluated. With respect to who does what, the STS database codes participants, which may be surgeons, practices, hospitals, departments. That is about as focused as you can get. So we don t get information on the specific practice of the individual surgeon that did an individual case. With respect to segmentectomies, the database changed versions from 2.07 to in 2009, and in the earlier version, lobectomies and segmentectomies were actually captured separately. In 170 VATS segmentectomies, the hilar upstaging was 4%. In 280 open segmentectomies, it was higher; it was 5.3%. So even in segmentectomies, there does appear to be a difference in hilar upstaging. The other way to look at that is lobectomy-only data. So for open lobes that were not clumped with segmentectomies, the hilar upstaging rate was 9.6% in the open cohort and in the VATS cohort it was significantly lower at 7.5%. Thank you for your questions.