- Email: [email protected]
Long-Term Survival of VATS Versus Open Lobectomy
408
CORRESPONDENCE
issue has not been specifically addressed in the accompanying papers. In addition, it has been shown that the subcellular distribution of Prox-1 is subject to change throughout development with cytoplasmic, as well as with nuclear staining [3]. The question, whether or not the subcellular distribution of Prox-1 in a mature cell is permanent, or may change under certain conditions (ie, a certain disease state) has not yet been investigated and is therefore unknown. In our article we have not raised any of these issues, as this would have been well beyond the focus of the paper. In addition, as pointed out correctly in the letter, the package insert of Acris antibodies does not specify the antibody’s staining pattern. In their letter Drs Castro and Galambos have referred to Figure 1D in our article. It is important to note, that the Figure 1D represents the immunohistostaining for vascular endothelial growth factor receptor-2 positive blood capillaries and not for Prox-1 positive lymphatics, which are shown in Figure 1, parts E and F. However, with regard to Figure 1 E and F we still disagree with the assumption that the nucleus which is seen in the figure, depicting the Prox-1 positive vessels, belongs to the lymphatic endothelial cell. But we accept the criticism of Dr Castro and Dr Galambos that the figure we chose for publication may give reason for discussion. In summary, we disagree with the notion that for Prox-1 only nuclear staining is relevant. However, we consider the ensuing controversy as very valuable, as it draws attention to the fact that not only the quantity of a specific marker, but that its subcellular distribution may be an important factor to look into, as well. Alexey Dashkevich, MD Department of Cardiovascular Surgery, University Freiburg Medical Center Hugstetter Strasse 55 79106 Freiburg, Germany e-mail: [email protected] Hans Joachim Geissler, MD, PhD Department of Cardiovascular Surgery, University Freiburg Medical Centre, Freiburg, Germany and Michael E. DeBakey Institute Texas A&M University College Station, TX 77843
References 1. Castro E, Galambos C. Prox-1: a specific and sensitive marker for lymphatic endothelium in normal and diseased human tissues (letter). Ann Thorac Surg 2011;92:407. 2. Dashkevich A, Heilmann C, Kayser G, et al. Lymph angiogenesis after lung transplantation and relation to acute organ rejection in humans. Ann Thorac Surg 2010;90:406 –11. 3. Duncan MK, Cui W, Oh DJ, Tomarev SI. Prox1 is differentially localized during lens development. Mech Dev 2002;112:195– 8.
Ann Thorac Surg 2011;92:403–10
Whitson and colleagues [2], patients who underwent VATS lobectomy improved survival versus patients with open lobectomy at 4 years (88.4% versus 71%; p ⫽ 0.003). Yan and colleagues [3] also reported in their systematic review that 5-year survival was significantly improved for patients who undergo VATS lobectomy for early-stage NSCLC (VATS relative risk, 0.72; p ⫽ 0.04). These systematic reviews [2, 3] included only one small, randomized controlled trial [4] and several observational comparative studies, and the authors’ abstracted and combined crude (unadjusted for confounding) data from observational studies. To reduce the effect of treatment-selection bias and potential confounding in observational studies, however, rigorous adjustment for significant differences in the baseline characteristics of patients should be conducted. Furthermore, adjusted estimates, and not unadjusted estimates, also ought to be pooled in a meta-analysis including observational studies. Herein, we would like to perform a meta-analysis of randomized controlled trials and only observational comparative studies providing adjusted estimates of VATS versus open lobectomy for prevention of long-term (5-year) all-cause mortality in early stage NSCLC. Our comprehensive search, which was current through September 2010, identified four observational comparative studies [5– 8] providing adjusted hazard ratios and only one randomized controlled trial [4]. All the observational studies used Cox proportional hazards model to reduce the effect of potential confounding. Pooled analysis (representing 13,070 patients with early stage NSCLC) demonstrated no significant difference in mortality between VATS and open lobectomy in fixed-effects models (hazard ratio, 0.98; 95% confidence interval, 0.89 to 1.07; p ⫽ 0.61) (Fig 1). There was minimal between the studies heterogeneity (p ⫽ 0.79 by standard 2 tests) and no difference in the pooled result from random-effects modeling, accordingly. Exclusion of any single trial from the analysis did not substantively alter the overall result of our analysis. There was no evidence of significant publication bias (p ⫽ 0.81 by an adjusted rank-correlation test). The results of our analysis suggest no significant difference in long-term, all-cause mortality between VATS and open lobectomy in patients with early stage NSCLC. As Rueth and Andrade [1] stated in their review, a large, prospective, randomized, multi-institutional trial of VATS versus open lobectomy will likely never take place. The present meta-analysis must be better evidence than the previous meta-analyses [2, 3], because the former combined the adjusted results, and not the unadjusted results, from the observational studies. Hisato Takagi, MD, PhD Masafumi Matsui, MD Takuya Umemoto, MD, PhD Department of Cardiothoracic Surgery Shizuoka Medical Center 762-1 Nagasawa, Shimizu-cho, Sunto-gun Shizuoka, 411-8611 Japan e-mail: [email protected]
MISCELLANEOUS
Long-Term Survival of VATS Versus Open Lobectomy To the Editor:
References
We read with great interest a review by Rueth and Andrade [1] of the most current available information comparing videoassisted thoracoscopic surgical (VATS) lobectomy with open lobectomy for clinical stage I nonsmall cell lung cancer (NSCLC) from perioperative, biologic, and oncologic perspectives. The authors cited two systematic reviews by Whitson and colleagues [2] and Yan and colleagues [3]. In the systematic review by
1. Rueth NM, Andrade RS. Is VATS lobectomy better: perioperatively, biologically and oncologically? Ann Thorac Surg 2010;89:S2107–11. 2. Whitson BA, Groth SS, Duval SJ, Swanson SJ, Maddaus MA. Surgery for early-stage non-small cell lung cancer: a systematic review of the video-assisted thoracoscopic surgery versus thoracotomy approaches to lobectomy. Ann Thorac Surg 2008;86:2008 –16.
© 2011 by The Society of Thoracic Surgeons Published by Elsevier Inc
0003-4975/$36.00
Ann Thorac Surg 2011;92:403–10
CORRESPONDENCE
409
Fig 1. Long-term, all-cause mortality among patients with early stage, nonsmall cell lung cancer assigned to video-assisted thoracoscopic surgery (VATS) versus open lobectomy. (CI ⫽ confidence interval.)
Reply To the Editor: We thank Dr Takagi and colleagues [1] for challenging the evidence in the literature regarding the outcomes after videoassisted thoracoscopic (VATS) vs. open lobectomy for early stage non-small cell lung cancer (NSCLC) [2]. As with most topics, there are limitations to the available evidence. We encourage continued investigation to better understand these outcomes. Dr Takagi and coworkers criticized our systematic review and meta-analysis [3] (as well as a similar one conducted by Yan and associates [4]) for including only “1, small randomized trial and several observational comparative studies.” As a matter of fact, we included 2 randomized trials in our study. We recognize that any meta-analysis is inherently limited by the quality of the pooled data. Ideally, most of the data would come from well-conducted randomized trials. Unfortunately, most of the available data on VATS versus open lobectomies is based upon observational studies. We acknowledged this limitation in our paper. Despite their critique, the meta-analysis conducted by Takagi and colleagues also included “1, small randomized trial and several (4) observational comparative studies.” They also raised concerns that differences in baseline characteristics between VATS and thoracotomy patients could be confounding our results. However, the baseline characteristics in our study were similar (see Table 1 of our systematic review) [3].
Takagi and coworkers claim they can reduce selection bias in a meta-analysis by “rigorous adjustment for significant differences in the baseline characteristics of patients.” Their methods for performing such an analysis are unclear [1]. Selection bias is an inherent limitation of observational studies; consequently, it is a limitation of meta-analyses that include such data. Dr Takagi seems to imply that a regression analysis will reduce selection bias. On the contrary, regression is simply one method to reduce confounding; it will not reduce bias since selection bias is a systematic error in the way patients are chosen for inclusion in a study (not an error in data analysis). All meta-analyses should include a rigorous systematic review of the literature. Their methods for performing a “comprehensive search” for studies to include in their analysis and their methods for extracting data are ill-defined [1]. It’s peculiar that they only identified 5 studies for inclusion in their analysis yet 2 similar, independent studies published in high impact factor, peer-reviewed journals identified over 20 studies each [3, 4]. We do not understand how Takagi and associates identified 13,070 patients for inclusion. Our review of those same papers identified 10,750 patients with early stage disease. Furthermore, most of the patients (about 90%) included in Takagi’s analysis were based on data from one study [5]. That study used pooled data on patients who underwent either lobectomy or segmentectomy for any stage of disease (not just stage I or II); we are unable to extract individual data on lobectomies for early stage NSCLC. As such, we suspect Dr Takagi’s analysis included patients with more advanced disease as well as patients who underwent segmentectomies. Often, the best indicator of evidence is consistency of results between studies. Contrary to Takagi’s findings, previously published meta-analyses (including our study), randomized trials, and observational studies suggest that a VATS approach is associated with more favorable outcomes. We appreciate Dr Takagi and colleagues’ critical review and encourage them to submit their study in greater detail for publication in a peerreviewed journal. Shawn S. Groth, MD, MS Bryan A. Whitson, MD, PhD Natasha M. Rueth, MD, MS Michael A. Maddaus, MD Rafael S. Andrade, MD Division of General Thoracic and Foregut Surgery Department of Surgery, University of Minnesota MMC 207, 420 Delaware St SE Minneapolis, MN 55455 e-mail: [email protected]
MISCELLANEOUS
3. Yan TD, Black D, Bannon PG, McCaughan BC. Systematic review and meta-analysis of randomized and nonrandomized trials on safety and efficacy of video-assisted thoracic surgery lobectomy for early-stage non-small-cell lung cancer. J Clin Oncol 2009;27:2553– 62. 4. Sugi K, Kaneda Y, Esato K. Video-assisted thoracoscopic lobectomy achieves a satisfactory long-term prognosis in patients with clinical stage IA lung cancer. World J Surg 2000;24:27–31. 5. Flores RM, Park BJ, Dycoco J, et al. Lobectomy by videoassisted thoracic surgery (VATS) versus thoracotomy for lung cancer. J Thorac Cardiovasc Surg 2009;138:11– 8. 6. 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;137:1415–21. 7. Sakuraba M, Miyamoto H, Oh S, et al. Video-assisted thoracoscopic lobectomy vs. conventional lobectomy via open thoracotomy in patients with clinical stage IA non-small cell lung carcinoma. Interact Cardiovasc Thorac Surg 2007;6: 614 –7. 8. Sawada S, Komori E, Yamashita M, et al. Comparison in prognosis after VATS lobectomy and open lobectomy for stage I lung cancer: retrospective analysis focused on a histological subgroup. Surg Endosc 2007;21:1607–11.
CORRESPONDENCE
issue has not been specifically addressed in the accompanying papers. In addition, it has been shown that the subcellular distribution of Prox-1 is subject to change throughout development with cytoplasmic, as well as with nuclear staining [3]. The question, whether or not the subcellular distribution of Prox-1 in a mature cell is permanent, or may change under certain conditions (ie, a certain disease state) has not yet been investigated and is therefore unknown. In our article we have not raised any of these issues, as this would have been well beyond the focus of the paper. In addition, as pointed out correctly in the letter, the package insert of Acris antibodies does not specify the antibody’s staining pattern. In their letter Drs Castro and Galambos have referred to Figure 1D in our article. It is important to note, that the Figure 1D represents the immunohistostaining for vascular endothelial growth factor receptor-2 positive blood capillaries and not for Prox-1 positive lymphatics, which are shown in Figure 1, parts E and F. However, with regard to Figure 1 E and F we still disagree with the assumption that the nucleus which is seen in the figure, depicting the Prox-1 positive vessels, belongs to the lymphatic endothelial cell. But we accept the criticism of Dr Castro and Dr Galambos that the figure we chose for publication may give reason for discussion. In summary, we disagree with the notion that for Prox-1 only nuclear staining is relevant. However, we consider the ensuing controversy as very valuable, as it draws attention to the fact that not only the quantity of a specific marker, but that its subcellular distribution may be an important factor to look into, as well. Alexey Dashkevich, MD Department of Cardiovascular Surgery, University Freiburg Medical Center Hugstetter Strasse 55 79106 Freiburg, Germany e-mail: [email protected] Hans Joachim Geissler, MD, PhD Department of Cardiovascular Surgery, University Freiburg Medical Centre, Freiburg, Germany and Michael E. DeBakey Institute Texas A&M University College Station, TX 77843
References 1. Castro E, Galambos C. Prox-1: a specific and sensitive marker for lymphatic endothelium in normal and diseased human tissues (letter). Ann Thorac Surg 2011;92:407. 2. Dashkevich A, Heilmann C, Kayser G, et al. Lymph angiogenesis after lung transplantation and relation to acute organ rejection in humans. Ann Thorac Surg 2010;90:406 –11. 3. Duncan MK, Cui W, Oh DJ, Tomarev SI. Prox1 is differentially localized during lens development. Mech Dev 2002;112:195– 8.
Ann Thorac Surg 2011;92:403–10
Whitson and colleagues [2], patients who underwent VATS lobectomy improved survival versus patients with open lobectomy at 4 years (88.4% versus 71%; p ⫽ 0.003). Yan and colleagues [3] also reported in their systematic review that 5-year survival was significantly improved for patients who undergo VATS lobectomy for early-stage NSCLC (VATS relative risk, 0.72; p ⫽ 0.04). These systematic reviews [2, 3] included only one small, randomized controlled trial [4] and several observational comparative studies, and the authors’ abstracted and combined crude (unadjusted for confounding) data from observational studies. To reduce the effect of treatment-selection bias and potential confounding in observational studies, however, rigorous adjustment for significant differences in the baseline characteristics of patients should be conducted. Furthermore, adjusted estimates, and not unadjusted estimates, also ought to be pooled in a meta-analysis including observational studies. Herein, we would like to perform a meta-analysis of randomized controlled trials and only observational comparative studies providing adjusted estimates of VATS versus open lobectomy for prevention of long-term (5-year) all-cause mortality in early stage NSCLC. Our comprehensive search, which was current through September 2010, identified four observational comparative studies [5– 8] providing adjusted hazard ratios and only one randomized controlled trial [4]. All the observational studies used Cox proportional hazards model to reduce the effect of potential confounding. Pooled analysis (representing 13,070 patients with early stage NSCLC) demonstrated no significant difference in mortality between VATS and open lobectomy in fixed-effects models (hazard ratio, 0.98; 95% confidence interval, 0.89 to 1.07; p ⫽ 0.61) (Fig 1). There was minimal between the studies heterogeneity (p ⫽ 0.79 by standard 2 tests) and no difference in the pooled result from random-effects modeling, accordingly. Exclusion of any single trial from the analysis did not substantively alter the overall result of our analysis. There was no evidence of significant publication bias (p ⫽ 0.81 by an adjusted rank-correlation test). The results of our analysis suggest no significant difference in long-term, all-cause mortality between VATS and open lobectomy in patients with early stage NSCLC. As Rueth and Andrade [1] stated in their review, a large, prospective, randomized, multi-institutional trial of VATS versus open lobectomy will likely never take place. The present meta-analysis must be better evidence than the previous meta-analyses [2, 3], because the former combined the adjusted results, and not the unadjusted results, from the observational studies. Hisato Takagi, MD, PhD Masafumi Matsui, MD Takuya Umemoto, MD, PhD Department of Cardiothoracic Surgery Shizuoka Medical Center 762-1 Nagasawa, Shimizu-cho, Sunto-gun Shizuoka, 411-8611 Japan e-mail: [email protected]
MISCELLANEOUS
Long-Term Survival of VATS Versus Open Lobectomy To the Editor:
References
We read with great interest a review by Rueth and Andrade [1] of the most current available information comparing videoassisted thoracoscopic surgical (VATS) lobectomy with open lobectomy for clinical stage I nonsmall cell lung cancer (NSCLC) from perioperative, biologic, and oncologic perspectives. The authors cited two systematic reviews by Whitson and colleagues [2] and Yan and colleagues [3]. In the systematic review by
1. Rueth NM, Andrade RS. Is VATS lobectomy better: perioperatively, biologically and oncologically? Ann Thorac Surg 2010;89:S2107–11. 2. Whitson BA, Groth SS, Duval SJ, Swanson SJ, Maddaus MA. Surgery for early-stage non-small cell lung cancer: a systematic review of the video-assisted thoracoscopic surgery versus thoracotomy approaches to lobectomy. Ann Thorac Surg 2008;86:2008 –16.
© 2011 by The Society of Thoracic Surgeons Published by Elsevier Inc
0003-4975/$36.00
Ann Thorac Surg 2011;92:403–10
CORRESPONDENCE
409
Fig 1. Long-term, all-cause mortality among patients with early stage, nonsmall cell lung cancer assigned to video-assisted thoracoscopic surgery (VATS) versus open lobectomy. (CI ⫽ confidence interval.)
Reply To the Editor: We thank Dr Takagi and colleagues [1] for challenging the evidence in the literature regarding the outcomes after videoassisted thoracoscopic (VATS) vs. open lobectomy for early stage non-small cell lung cancer (NSCLC) [2]. As with most topics, there are limitations to the available evidence. We encourage continued investigation to better understand these outcomes. Dr Takagi and coworkers criticized our systematic review and meta-analysis [3] (as well as a similar one conducted by Yan and associates [4]) for including only “1, small randomized trial and several observational comparative studies.” As a matter of fact, we included 2 randomized trials in our study. We recognize that any meta-analysis is inherently limited by the quality of the pooled data. Ideally, most of the data would come from well-conducted randomized trials. Unfortunately, most of the available data on VATS versus open lobectomies is based upon observational studies. We acknowledged this limitation in our paper. Despite their critique, the meta-analysis conducted by Takagi and colleagues also included “1, small randomized trial and several (4) observational comparative studies.” They also raised concerns that differences in baseline characteristics between VATS and thoracotomy patients could be confounding our results. However, the baseline characteristics in our study were similar (see Table 1 of our systematic review) [3].
Takagi and coworkers claim they can reduce selection bias in a meta-analysis by “rigorous adjustment for significant differences in the baseline characteristics of patients.” Their methods for performing such an analysis are unclear [1]. Selection bias is an inherent limitation of observational studies; consequently, it is a limitation of meta-analyses that include such data. Dr Takagi seems to imply that a regression analysis will reduce selection bias. On the contrary, regression is simply one method to reduce confounding; it will not reduce bias since selection bias is a systematic error in the way patients are chosen for inclusion in a study (not an error in data analysis). All meta-analyses should include a rigorous systematic review of the literature. Their methods for performing a “comprehensive search” for studies to include in their analysis and their methods for extracting data are ill-defined [1]. It’s peculiar that they only identified 5 studies for inclusion in their analysis yet 2 similar, independent studies published in high impact factor, peer-reviewed journals identified over 20 studies each [3, 4]. We do not understand how Takagi and associates identified 13,070 patients for inclusion. Our review of those same papers identified 10,750 patients with early stage disease. Furthermore, most of the patients (about 90%) included in Takagi’s analysis were based on data from one study [5]. That study used pooled data on patients who underwent either lobectomy or segmentectomy for any stage of disease (not just stage I or II); we are unable to extract individual data on lobectomies for early stage NSCLC. As such, we suspect Dr Takagi’s analysis included patients with more advanced disease as well as patients who underwent segmentectomies. Often, the best indicator of evidence is consistency of results between studies. Contrary to Takagi’s findings, previously published meta-analyses (including our study), randomized trials, and observational studies suggest that a VATS approach is associated with more favorable outcomes. We appreciate Dr Takagi and colleagues’ critical review and encourage them to submit their study in greater detail for publication in a peerreviewed journal. Shawn S. Groth, MD, MS Bryan A. Whitson, MD, PhD Natasha M. Rueth, MD, MS Michael A. Maddaus, MD Rafael S. Andrade, MD Division of General Thoracic and Foregut Surgery Department of Surgery, University of Minnesota MMC 207, 420 Delaware St SE Minneapolis, MN 55455 e-mail: [email protected]
MISCELLANEOUS
3. Yan TD, Black D, Bannon PG, McCaughan BC. Systematic review and meta-analysis of randomized and nonrandomized trials on safety and efficacy of video-assisted thoracic surgery lobectomy for early-stage non-small-cell lung cancer. J Clin Oncol 2009;27:2553– 62. 4. Sugi K, Kaneda Y, Esato K. Video-assisted thoracoscopic lobectomy achieves a satisfactory long-term prognosis in patients with clinical stage IA lung cancer. World J Surg 2000;24:27–31. 5. Flores RM, Park BJ, Dycoco J, et al. Lobectomy by videoassisted thoracic surgery (VATS) versus thoracotomy for lung cancer. J Thorac Cardiovasc Surg 2009;138:11– 8. 6. 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;137:1415–21. 7. Sakuraba M, Miyamoto H, Oh S, et al. Video-assisted thoracoscopic lobectomy vs. conventional lobectomy via open thoracotomy in patients with clinical stage IA non-small cell lung carcinoma. Interact Cardiovasc Thorac Surg 2007;6: 614 –7. 8. Sawada S, Komori E, Yamashita M, et al. Comparison in prognosis after VATS lobectomy and open lobectomy for stage I lung cancer: retrospective analysis focused on a histological subgroup. Surg Endosc 2007;21:1607–11.