Cheng, Davy*; Downey, Robert J.†; Kernstine, Kemp‡; Stanbridge, Rex§; Shennib, Hani¶; Wolf, Randall∥; Ohtsuka, Toshiya**; Schmid, Ralph††; Waller, David‡‡; Fernando, Hiran§§; Yim, Anthony¶¶; Martin, Janet*∥∥
Before the introduction of video-assisted thoracic surgery (VATS), lobectomy for lung cancer invariably required a thoracotomy, which was usually performed using division of chest wall musculature and rib spreading or cutting with resulting trauma to the patient. VATS for lung cancer lobectomy is held by proponents as being less traumatic to the patient, potentially allowing for less pain, shorter duration of chest tube drainage, and reduced length of hospital stay. These benefits have been shown in randomized trials of VATS for a variety of pulmonary and mediastinal indications other than lung lobectomy.1–4 In addition, improved postoperative cellular immunity has been observed for VATS compared with conventional thoracotomy.5–7 However, some evidence suggests that VATS may offer no significant improvement to clinical outcomes including postoperative pain, disease control, and length of stay in comparison with conventional treatment.4,8
To date, no comprehensive and methodologically rigorous meta-analysis of VATS lobectomy for lung cancer exists in the literature. One previous systematic review by Sedrakyan et al evaluated studies of a variety of indications for VATS including pneumothorax and lung resections9 and included only randomized trials published before 2003. Relevant randomized (RCT) and nonrandomized (non-RCT) controlled trials have been published since Sedrakyan et al's systematic review was conducted, and the totality of evidence remains to be addressed in an updated comprehensive systematic review with meta-analysis. Although individual trials of VATS considered in isolation will be underpowered to provide adequate estimates of clinical outcomes, aggregation of existing data through meta-analysis will improve the power and provide improved estimates of clinical and economic benefits and risks associated with VATS for lobectomy, as well as suggest possible baseline biases in the patient selection criteria.
This systematic review with meta-analysis sought to determine whether VATS lobectomy improves clinical and resource outcomes compared with open lobectomy (OPEN) in adults undergoing lung cancer resection.
This meta-analysis of randomized trials was performed in accordance with QUOROM Consensus and Cochrane Collaboration recommendations.10,11 and according to a protocol that prespecified outcomes, search strategies, inclusion criteria, and statistical analyses.
Definition of Endpoints
Clinical endpoints of interest included stage-specific survival, local or distal tumor recurrence, pain, need for analgesia, wound complications, respiratory failure, lung function, atelectasis, air leaks, need for repeat surgical intervention, re-exploration for bleeding, transfusions, severe adverse events, survival, patient satisfaction, functionality, quality of life, patient preference, duration of ventilation, duration of surgery, intensive care unit length of stay, total hospital length of stay, costs, and cost-effectiveness. Need for transfusions was defined as per the study authors, and typically included RBCs transfused intraoperatively and postoperatively until discharge. In some cases, only postoperative transfusions were reported. Perioperative complications were recorded individually, when available within the reported trials. When clinical trials also reported a composite of any perioperative complications, this composite was included in the overall complications graph.
A comprehensive literature search of MEDLINE, EMBASE, Cochrane CENTRAL, Current Contents and Science Citation Index, using keywords and variants of “video,” “thoracic,” “VATS,” and “thoracoscopic,” was performed from the earliest available date to April 2007. The most recent 6 months of relevant surgical and anesthesia journals were hand-searched, and databases of conference abstracts were reviewed electronically. Experts were contacted to solicit additional “in press” reports of clinical trials of VATS for lung cancer lobectomy.
To be eligible for inclusion in the systematic review and meta-analysis, trials had to be randomized or nonrandomized controlled trials comparing VATS (as defined by the study authors, with or without use of rib spreader) versus OPEN surgery in patients undergoing lobectomy for primary lung cancer. Published and unpublished trials were included, with no restrictions on language. Trials of patients undergoing primarily segmentectomy or wedge resection for diagnosis rather than treatment were excluded. For trials presenting the results of a mixed population of patients undergoing lung cancer resection through segmentectomy, wedge resection, or lobectomy, the trial was eligible for inclusion only if extractable data was presented separately for patients undergoing lobectomy, or if >80% of included patients had undergone lobectomy. Trials focused on patients undergoing lobectomy for metastases were excluded. Understanding that some trials will inadvertently (sometimes, even purposefully) include patients with benign pulmonary lesions, we excluded trials if greater than 20% of the patient population had benign lesions. Also excluded were trials where >20% of patients had undergone only segmentectomy or wedge resections for lung cancer treatment. All trials of VATS for the purpose of diagnosis rather than treatment of lung cancer were excluded.
Two authors independently extracted the following data points: baseline demographics including number of patients, inclusion/exclusion criteria for patient entry to study, age, sex, type of lung cancer, clinical stage, and pathologic stage. Details on the VATS procedure and the comparative conventional treatment provided were recorded, including the use of rib spreader and size of incision. Two authors extracted outcomes data, and verified the extraction with each other. Discrepancies were resolved by consensus.
Odds ratios (OR) and their 95% confidence intervals (95% CI) were calculated for discrete data. Weighted mean differences (WMD, 95% CI) were calculated for continuous data when similar metrics were used across the reported outcomes, and standardized mean differences (SMD, 95% CI) were calculated when different scales were reported across continuous outcomes. Heterogeneity was explored through the Q-statistic, and by calculating the I2.12 Summary OR and WMD were calculated using the fixed effects model when statistical heterogeneity was not found (ie, Q-test P value >0.10 and I2 < 50%). The random effects model was used when statistical heterogeneity was found (ie, Q-test P value <0.10 or I2 > 50%). When possible, data was analyzed by intention-to-treat (ie, if patients were crossed over from VATS to OPEN, they were analyzed in the VATS group as per their original intended allocation); however, in some cases the data was reported only in aggregate and was not separable for patients who had crossed over. Statistical significance for overall effect was defined as P < 0.05 or a confidence that excluded the value 1.00 for OR and 0.00 for WMD.
Subgroup analysis was planned a priori for randomized versus nonrandomized studies, and for trials in which a rib spreader was used versus trials in which no rib spreader was used during VATS. Publication bias was explored though visual inspection of funnel plots.
Of over 984 studies screened, 304 were identified as potentially relevant, and were retrieved for review. Of these, 36 met the inclusion criteria for the primary analysis of VATS versus OPEN (three unique randomized trials reported in four articles including a total 205 patients13–16 and 33 unique nonrandomized trials of 3384 patients reported in 41 articles,8,17–54 for an overall total of 3589 patients in 36 trials (Fig. 1, QUOROM Flowchart). For trials that reported more than once on an identical or overlapping population, only the most recent or most complete results were included. For the secondary analysis of C-VATS versus A-VATS, 1 randomized trial55 and selected arms of 2 nonrandomized trials27,50 met the inclusion criteria. Most trials were published after the year 2000; however, one trial was performed in 199516 (range: 1995–2007). Four trials were in languages other than English. The median quality score for the randomized studies was 2 (range: 1–2) out of a maximum score of 5. One randomized trial used a pseudo-randomized technique, because randomization was performed by patient identification number.13 For the purposes of this meta-analysis, this pseudo-randomized trial was classified as a randomized trial, but with the caveat that the randomization code may have been predictable and potentially introduced bias. Some trials excluded converted patients, and some trials excluded patients with complications. For nonrandomized studies, 12 indicated that they collected data prospectively, and the remainder were retrospective. Publication bias was not evident after visual inspection of funnel plots; however, the test was underpowered for most clinical outcomes.
Table 1 lists all the included trials and baseline characteristics are compared in Table 2. At baseline, the mean age was 65 years in both groups. Baseline pulmonary function including FEV1 and vital capacity (absolute volume and percent predicted) did not differ at baseline for VATS versus OPEN. Approximately 55% and 64% were male in the VATS and OPEN group, respectively. Although baseline characteristics were generally similar for randomized trials, the small size of the randomized trials precluded adequate power to rule out potential differences. In nonrandomized trials, baseline characteristics differed significantly for tumor size and for pathologic stage, favoring VATS because of the smaller tumor size (0.48 cm) and greater number of patients with pathologic stage I tumor versus stage II or III relative to patients selected for OPEN surgery. In addition, significantly fewer patients in the VATS group had squamous cell carcinoma or “other” lung lesions and more adenocarcinoma compared with OPEN surgery patients. Therefore, in the nonrandomized trials, baseline prognosis was more favorable for VATS (smaller tumor size, earlier stage, tumor more likely to be peripheral) than for OPEN surgery patients (Figs. 2A–G).
Significant heterogeneity across studies was observed for length of stay, duration of surgery, tumor size, duration of chest tube drainage, visual analog scale (VAS) scores, time to return to full activity, arrhythmias, analgesic dose, analgesic duration, analgesic administrations, number of patients transfused, blood loss, nodes dissected, number of stations sampled. Nevertheless, much of the heterogeneity was less concerning because it was driven by uncertainty about the size of effect and not the direction of effect, and therefore did not generally put into question the presence of significant benefit (when detected).
Crossovers and Drop-Outs
The overall number of patients who were crossed over from planned VATS to OPEN surgery was 6%, and was similar for randomized and nonrandomized trials; however, a number of trials did not explicitly report crossovers.
Postoperative complications, considered as a composite outcome, were significantly reduced in the VATS group compared with OPEN surgery when both randomized and nonrandomized trials were considered in aggregate (OR 0.48, 95% CI 0.32–0.70) (Fig. 3A). When randomized trials were considered alone, the reduction was similar (OR 0.30, 95% CI 0.11–0.81; two trials). Nonrandomized trials suggested a 48% reduction in risk of postoperative complications (OR 0.52, 95% CI 0.34–0.80; six trials) (Table 3).
Pulmonary complications were significantly reduced (OR 0.39, 95% CI 0.21–0.73; five non-RCTs) (Fig. 3B). Other complications, including respiratory dysfunction (OR 0.53, 95% CI 0.16–1.75), pneumonia (OR 0.56, 95% CI 0.26–1.21; one RCT, 10 non-RCTs), chylothorax (OR 1.17, 95% CI 0.42–3.27; four non-RCTs), pyothorax (2.10, 95% CI 0.18–25.01; one non-RCT), atelectasis (OR 0.36, 95% CI 0.01–9.47; one non-RCT), prolonged air leaks (>7 days) (OR 1.67, 95% CI 0.92–3.03; one RCT; 11 non-RCTs), arrhythmias (OR 0.93, 95% CI 0.49–1.76; one RCT, six non-RCTs), and cardiac complications (OR 0.68, 95% CI 0.32–1.48; one non-RCT) were not significantly reduced with VATS versus OPEN surgery. In two nonrandomized trials that included wound infections as an outcome, no incidents of wound infection were found.
Although blood loss was significantly reduced with VATS versus OPEN (−80 mL, 95% CI −100 to −50 mL; 18 non-RCTs) (Fig. 3C), the incidence of excessive blood loss (generally defined as >500 mL) was not significantly reduced, (OR 0.45, 95% CI 0.13–1.60; one RCT, three non-RCTs) and incidence of re-exploration for bleeding was not significantly reduced (OR 0.10, 95% CI 0.01–2.00; one non-RCT). The number of patients transfused was not significantly reduced (OR 0.59, 95% CI 0.14–2.49; one RCT, six non-RCTs) (Table 3).
Chest Tube Drainage
Mean volume of chest tube drainage was significantly reduced (−106 mL, 95% CI 206 to −7 mL; three non-RCTs), and the number of days of drainage was also significantly reduced (WMD −1.0 days; 95% CI −1.6 to −0.3; one RCT, 16 non-RCTs) (Fig. 3D).
Although the overall incidence of postoperative pain (any severity) was not reduced (OR 0.47, 95% CI 0.17–1.29; two non-RCTs), the incidence of severe postoperative pain was significantly reduced (OR 0.03, 95% CI 0.00–0.30; one non-RCT). Similarly, postoperative pain measured after discharge and up to 1 year was significantly reduced (OR 0.25, 95% CI 0.10–0.60; two non-RCTs), and postoperative pain measured after 1 year was significantly reduced with VATS versus OPEN (OR 0.39, 95% CI 0.16–0.93; one RCT, three non-RCTs). Recurrent nerve palsy/injury was not significantly reduced (OR 1.07, 95% CI 0.32–3.57; two non-RCTs). Pain measured via 10-point VAS was significantly reduced by <1 point on day 1 (WMD −0.65 points, 95% CI −1.26 to −0.04; two non-RCTs), by >2 points at 1 week (WMD −2.38, 95% CI −3.40 to −1.36; five non-RCTs), and by <1 point at weeks 2 to week 4 (WMD −0.28 points, 95% CI −0.38 to −0.19 points; three non-RCTs) in the VATS group (Fig. 4A). At 3 months to 3 years' follow-up, the reduction in VAS score did not reach significance (WMD −0.10 points, 95% CI −0.20 to 0.01 points; three non-RCTs).
Analgesic requirements were significantly reduced, whether measured by total dose requirements (SMD −4 units, 95% CI −7 to −1 units; four non-RCTs), number of times analgesics were administered (WMD −4 administrations, 95% CI −56 to −1 administrations; six non-RCTs), or duration of analgesic requirements (WMD −3 days, 95% CI −5 to −1 day; five non-RCTs) (Fig. 4B).
Percent change in postoperative vital capacity postoperatively was significantly improved for VATS versus OPEN (WMD 20%, 95% CI 14%–25%), and percent predicted forced vital capacity at 1 year was significantly greater for VATS versus OPEN surgery (WMD 7%, 95% CI 2%–12%). However, other measures of pulmonary function were not significantly improved with VATS versus OPEN surgery; including postoperative percent predicted FEV1, vital capacity, Pao2, Paco2.
Functionality, Satisfaction, and Quality of Life
The incidence of patients who were dependent at discharge (ie, discharged to home care) was significantly reduced (OR 0.15, 95% CI 0.04–0.52; one non-RCT), but was reported in only one trial. The percent change in 6-minute walk distance was significantly improved for VATS versus OPEN (WMD 17%, 95% CI 12%–22%; one non-RCT).
Most satisfaction and quality of life scores did not differ significantly, including overall quality of life scores (WMD 8.8 points, 95% CI −2.4 to 20.0; one non-RCT), overall impression of operation (SMD 0.32, 95% CI −0.12 to 0.75; two non-RCTs) and satisfaction related to wound numbness (WMD −0.20, 95% CI −0.63 to 0.23; one non-RCT). Significant improvement was found for satisfaction with limitations of arm or shoulder (WMD −0.60, 95% CI −0.93 to −0.27; one non-RCT), but due to the risks of multiple testing across multiple scores, the validity of this statistical significance is uncertain. Shoulder strength was greater at 1 week postoperatively (WMD 0.60, 95% CI 0.11–1.09; one non-RCT), but did not differ at 3 months.
The incidence of patients reporting limited activity at 3 months was reduced (OR 0.04, 95% CI 0.00–0.82; one non-RCT), and time to full activity was significantly reduced in the VATS group versus OPEN surgery (WMD −1.48, 95% CI −2.11 to −0.85; two non-RCTs). Overall patient-reported physical function scores did not differ between groups at 3 years follow-up (WMD 1.40 points, 95% CI −5.23 to 8.03 points; one non-RCT).
The incidence of cancer recurrence (local or distal) was not significantly different between VATS versus OPEN (OR 0.78, 95% CI 0.58–1.04). Similarly, distal recurrences considered alone were not significantly different (OR 0.77, 95% CI 0.55–1.07), and local recurrences considered alone were not significantly different between group (OR 0.59, 95% CI 0.34–1.03) (Fig. 5A).
Delay in delivery of planned adjuvant chemotherapy treatment was significantly reduced for VATS versus OPEN (OR 0.15, 95% CI 0.06–0.38), but was reported in only 1 trial. Similarly, the need for chemotherapy reduction was reduced (OR 0.37, 95% CI 0.16–0.87), and the number of patients who did not receive at least 75% of their planned chemotherapy without delay or reduction was reduced (OR 0.41, 95% CI 0.18–0.93) (Fig. 5B).
Nodes Dissected or Biopsied
The mean number of lymph nodes dissected or biopsied was not significantly different (WMD −0.2, 95% CI −0.8 to +0.5; two RCTs, 12 non-RCTs) (Fig. 6). Two non-RCTs reported data on the number of lymph node stations dissected/sampled. Combining these two trials together suggests that the number of lymph node stations sampled did not significantly differ overall between VATS and OPEN (WMD 0.6 station, 95% CI −0.3 to 1.5; two non-RCTs); however, there was significant heterogeneity (I2 = 59%) between the two trials, with 1 trial suggesting no difference43 and the other trial suggesting significant difference.32
The incidence of death during hospitalization or up to 30 days was not significantly different for VATS versus OPEN (OR 0.79, 95% CI 0.38–1.64; two RCTs, 19 non-RCTs). Similarly, the incidence of death did not differ between groups at 1 year (OR 0.78, 95% CI 0.34–1.77; five non-RCTs), 3 years (OR 1.09, 95% CI 0.60–1.97; one RCT, five non-RCTs) (Figs. 7A–C). The odds for death were similarly nonsignificant when the randomized (OR 1.40, 0.35–5.53) and five non-RCTs (OR 1.03, 95% CI 0.53–1.98) were subanalyzed separately.
Death at 5 years was reported in eight trials (one RCT, seven non-RCTs) (Fig. 7D). Considered in aggregate, the eight trials showed a significant reduction in mortality at 5 years (OR 0.67, 95% CI 0.47–0.97). The randomized trial showed no significant reduction in 5-year death rate (OR 0.64, 95% CI 0.19–2.11), but was a small trial (n = 100 patients).13 Because there was negligible statistical heterogeneity between randomized and nonrandomized studies, the weight of the evidence in all trials suggests a similar overall reduction in 5-year mortality; however, the majority of the weight for these estimates comes from the nonrandomized data because of the paucity of randomized data reporting this outcome.
Death at Maximum Follow-up
A total of 20 trials reported death at any time point. We therefore combined all 20 trials reporting survival at the maximum time of follow-up reported by the trial (ie, up to 5 years in some trials). The aggregate analysis for all trials reporting overall death suggests a significant 29% reduction in the odds of death (OR 0.71, 95% CI 0.54–0.93; two RCTs, 18 non-RCTs) at maximum follow-up. The overall incidence of death was 14.9% in the VATS group versus 23.8% in the OPEN group at maximum follow-up. The risk of death was not significantly reduced when randomized trials were considered alone (OR 0.64, 95% CI 0.19–2.11; P = 0.46); however, there were only two randomized trials reporting this outcome. When non-RCTs were considered alone (18 non-RCTs), the risk of death at maximum follow-up was significantly reduced (OR 0.72, 95% CI 0.55–0.94; P = 0.02) (Fig. 7E).
A subset of the above trials (nine non-RCTs) also reported death rates by pathologic stage at approximately 5 years follow-up. Four non-RCTs reported death rates for stage IA (OR 0.96, 95% CI 0.58–10.60; P = 0.89), three non-RCTs reported death rates for stage IB (OR 0.69, 95% CI 0.43–1.12, P = 0.13), and no significant differences in 5 year survival were found for stage IA and stage IB considered separately. Similarly, when stage IA and stage IB were considered together, the overall risk of death for VATS versus OPEN was not significantly different (OR 0.85, 95% CI 0.61–1.19) (Fig. 8A). One non-RCT reported death rate for stage II (OR 3.15, 95% CI 0.34–29.53; P = 0.26) (Fig. 8B) and stage III (OR 0.96, 95% CI 0.19–4.82, P = 0.96), and found no significant difference (Fig. 8C).
Duration of Surgery and Length of Stay
Hospital length of stay was significantly reduced by 2.6 days for VATS versus OPEN surgery (WMD −3 days, 95% CI −5 to −1; two RCTs, 17 non-RCTs); however, there was significant heterogeneity (I2 = 95%) due to variation across the trials in the magnitude of reduction of hospital stay and also due to differing directionality of effect (Table 3) (Fig. 9A). Operation time was significantly increased by 16 minutes with VATS versus OPEN (WMD 16 minutes; 95% CI 3–30 minutes) (Fig. 9B). Although there was significant heterogeneity between trials for differences in OR times, the heterogeneity was due to variation in magnitude of differences, rather than due to differences in direction of effect. Total costs associated with surgical procedure plus hospital stay was significantly increased for VATS versus OPEN surgery patients in the aggregate analysis of two non-RCTs reporting costs (SMD 0.72, 95% CI 0.21–1.23; P = 0.0005). However, this is based on a standardized mean difference of two small studies that evaluated differences in costs retrospectively in two different settings, using differing definitions of relevant costs (Table 4).
Complete (C-VATS) Versus Assisted (A-VATS)
In a secondary meta-analysis, trials comparing patients undergoing lobectomy specified as being a completely endoscopic VATS (C-VATS) were compared with trials specifying that VATS was performed using a rib spreader (assisted VATS or A-VATS) (Table 5). Pain scores or duration of analgesia were significantly reduced for C-VATS versus A-VATS (SMD −0.68, 95% CI −0.95 to −0.41; one RCT; one non-RCT). Length of hospital stay was also reduced with C-VATS compared with A-VATS (WMD −4 days, 95% CI −5 to −2; one RCT, one non-RCT). Duration of chest tube drainage was reduced (WMD −2 days, 95% CI −2 to −1), and blood loss was significantly reduced (WMD −50 mL, 95% CI −60 to −30 mL; one RCT; two non-RCTs). However, OR time was significantly increased (WMD 27 minutes; 95% CI 17–37).
To further explore whether VATS definition was related to differences in outcome, a subgroup analysis was conducted to include only trials of VATS without use of rib spreader versus OPEN. Table 6 outlines the results of key outcomes of this subgroup analysis. In general, differences between VATS versus OPEN were similar when only studies using VATS without rib spreader were included, with the exception that the reduction in death in hospital reached statistical significance for VATS versus OPEN. This subanalysis together with the results above suggest that patients undergoing less invasive VATS without a rib spreader may have more dramatic improvements in results than those with a rib spreader.
This meta-analysis suggests that, when feasible and within experienced surgeon hands, there maybe short-term, and possibly long-term advantages to performing lung resections with VATS techniques rather than through conventional open thoracotomy for patients with clinical stage I non-small cell lung cancer (NSCLC). Overall, synthesis of limited current evidence suggests that VATS for lobectomy may reduce acute and chronic pain and functionality, and with the possibility of improved survival at 5-year follow-up. However, the results are largely dependent on non-RCTs, and cannot be considered definitively conclusive. Existing randomized trials comparing VATS versus OPEN lobectomy are few, and underpowered to find clinically-important benefits and risks.
Evidence From Randomized Trials
When evidence from randomized trials was considered in aggregate apart from the non-RCTs, the three small trials together showed significant reduction in overall postoperative complications. For all other reported outcomes, no significant difference was found in the meta-analysis of randomized data, including OR time, length of stay, air leaks, pneumonia, arrhythmia, excess bleeding, pain after 1 year, and death in hospital, death at 5 years. Unfortunately, a number of important outcomes were not measured in the randomized trials including longer term survival, and cancer related outcomes. These studies were underpowered (n = 205) to show important differences for most clinical outcomes.
Evidence From Nonrandomized Trials
When data from non-RCTs and RCTs was combined and analyzed, significant differences were found for a number of additional clinical outcomes, including a 52% reduction in risk of postoperative complications and 61% reduction in the risk of pulmonary complications. Also, severe postoperative pain and longer term incidence of pain was significantly reduced with VATS versus OPEN surgery. The mean volume (106 mL reduction) of chest tube drainage was also reduced with VATS, but the clinical relevance of this small reduction is questionable. The mean duration of chest tube drainage was also reduced by about 1 day for VATS versus OPEN surgery patients, and length of hospital stay was reduced by over 2.5 days. There is an additional 16 minutes of operation time required, on average, and preliminary evidence suggests that VATS may increase overall costs. Whether this potential incremental cost and effort is worthy of the benefits it affords remains an important future research question.
The overall impact on quality of life, functionality, and satisfaction was inconsistent, but there was some indication that patients undergoing VATS have faster return to full activity, better performance on shoulder function tests and walking tests, and may have lesser dependent status at discharge.
Although blood loss was also significantly reduced by an average of 80 mL, this is considered clinically negligible, and none of the more clinically relevant indicators of important blood loss were reduced for VATS versus OPEN lobectomy, including incidence of massive blood loss (>500 mL), need for transfusion, and need for surgical re-exploration for bleeding.
Cancer outcomes appeared to be similar, or even improved, with VATS versus OPEN lobectomy. The number of nodes biopsied or dissected and stations sampled was similar across trials; however, the latter result should be interpreted cautiously because it was reported in only two nonrandomized trials and was associated with significant heterogeneity (one trial showed significant differences in station sampling, whereas the other did not). Overall, there were no differences in local or distal cancer recurrences in longer term follow-up studies, and overall risk of death was reduced by 29% at 5 years' follow-up in the aggregate analysis. However, because there were concerning imbalances in baseline prognostic factors (larger tumor size and more advanced stages in the OPEN group), the stage-specific survival analyses may represent a more balanced estimate of survival differences. None of the stage specific survival analysis showed statistically significant differences for overall survival at 5 years. Because the trend strongly favored VATS for all stage-specific survival comparisons, it maybe safe to conclude that, at the very least, VATS does not increase the risk of 5-year mortality.
Whether the potential for survival advantage associated with VATS could be eventually attributed to a reduced need to delay or reduce adjuvant chemotherapy,42 or whether it maybe related to reduced impairment of immunologic function as found in a number of included studies14,22–24,30,47,55 remain important hypotheses to be tested in future randomized trials.
The finding of reduced inflammatory markers and cytokine production with VATS versus OPEN is congruent with other research in the field of minimally invasive surgery, where minimal access technique has led to decreased pro-inflammatory and anti-inflammatory cytokines not only for lobectomy, but also in minimally invasive off-pump coronary artery bypass surgery56 and gastrointestinal surgery.57
Strengths and Limitations
This meta-analysis provides a systematic review of all available comparative studies of VATS versus OPEN lobectomy. To date, no meta-analyses have been done to address this question and surgeons and their patients have had to make decisions about the use of VATS without a clear understanding of what the balance of the evidence suggests. Now that the evidence base is clearly summarized in this analysis, decisions should be maximally informed by the knowledge of potential benefits and risks, and in light of the remaining uncertainties.
A number of limitations should be highlighted to facilitate appropriate interpretation of this meta-analysis. First, and foremost, the majority of this data is reliant on nonrandomized studies. Nonrandomized studies are subject to selection bias that may result in unbalanced selection of patients for VATS versus OPEN surgery. Examination of the baseline patient characteristics (Table 1) reveals clear evidence of selection biases in these studies, whereby patients had more favorable prognostic characteristics (more adenocarcinoma rather than squamous cell carcinoma, smaller tumor size, less advanced stage, smaller tumor size and more likely to be female) than patients selected for OPEN surgery (more likely to have squamous cell carcinoma, with larger tumor size, more advanced stage, and more likely to be male). The combined effect of these baseline differences on differences in outcomes is unknown, but is likely to have biased the included studies individually and in combination in this meta-analysis toward more favorable outcomes with VATS than with OPEN surgery. In particular, patients with smaller, more peripheral, and less advanced tumors may be comparatively less likely to die in the 5-year follow-up. In addition, other unbalanced factors may influence analyses of outcomes in unclear ways, and these uncertainties cannot be corrected for within this meta-analysis.
Because the majority of studies were nonrandomized, the selection of patients for VATS was at the discretion of the surgeons. In most trials, the criteria for selecting patients for VATS rather than OPEN surgery were not provided. In some trials where criteria for selecting patients suitable for VATS were described, it was clear that the criteria were systematically different for VATS versus OPEN surgery.
Examination of the baseline pulmonary function tests for these studies suggests that most patients had relatively good pulmonary function (VC > 3 L, FEV1 > 70% predicted). None of the studies reported data specifically for patients considered to be at high risk (ie, VC < 2.5 L, FEV1 < 50%), and a number of studies specifically excluded these types of patients. Therefore, the representativeness of the patient populations to the patients who may be selected for VATS in the real world setting is uncertain, and the generalizability of these results to patients with poor lung function (or other known high risk groups not included in studies to date) is uncertain.
In addition, the majority of patients included in this analysis were clinical stage I NSCLC patients, and few patients with advanced stages of disease were included in the trials. In some trials, patients with clinical stage I disease were upstaged to pathologic stage II or III upon surgery; however, advanced stages were proportionately under represented. Caution is warranted before presuming these results would apply to patients with advanced stages of lung cancer.
Because the definition and surgical techniques for VATS have not been standardized, there was considerable heterogeneity in the definitions and techniques for VATS represented in the included trials. Although some trials performed VATS completely endoscopically through a small incision (<8 cm) without a rib spreader, there were some trials that used larger incisions and rib spreaders to allow for direct visualization. However, in many trials the technique was not described clearly enough to determine whether rib spreaders were used. Table 1 outlines this information, as much as it was available. At least seven trials used rib spreaders,18,21,32,34,35,41,52 and 13 trials did not.16,22,24–26,30,31,38,43,47,49,51,54 In three trials, the use of rib spreader was compared with no rib spreader.27,50,55 The remaining included trials did not specify. Although subanalyses showed some evidence of greater benefit in trials without rib spreader use, caution should be exercised in interpreting these subanalyses because most of the outcomes were reported in too few trials to allow for robust subanalyses and the statistical tests for interactions or trends across subgroups were severely underpowered. In the analysis of the three trials with direct comparisons of less invasive versus more invasive VATS (C-VATS versus A-VATS), there was also some preliminary evidence of reduced pain (although variously reported definitions were combined), reduced chest tube duration, reduced length of stay, reduced bleeding (although, again the clinical relevance of the 45 mL reduction is questionable). Together these results suggest a strong hypothesis for reduced invasiveness being related to improvement in outcomes, even across the spectrum of definitions of VATS.
This meta-analysis was also limited by the quality of information reported in the available clinical trials. These limitations are testament to the surprising scarcity of valid comparative trials that have been conducted in this area. Even when trials were conducted, very few of them (only three) were randomized, and most nonrandomized trials failed to adequately control for baseline differences in patient characteristics of prognostic significance. Furthermore, many trials failed to report clinically-important outcomes such as survival beyond discharge from hospital, tumor recurrence, acute or chronic pain, analgesic requirements. Examination of the figures highlights the scarcity of trials reporting on each outcome of interest. Furthermore, some trials that reported on these clinically-relevant outcomes failed to provide data that was sufficiently explicit to include in the meta-analysis (ie, in some cases, means were provided without standard deviations, and in other cases results were reported qualitatively without providing any numeric data).
Also of concern, most trials excluded patients who were crossed over from VATS to OPEN surgery (Table 1). In particular, the retrospective trials typically failed to account for patients who were originally intended to undergo VATS but eventually underwent OPEN surgery because of anatomic issues or complications encountered intraoperatively. Whether these converted patients were included in the OPEN arm of retrospective trials was sometimes unclear. If they were included in the OPEN arm, this may add bias against OPEN surgery as converted patients may be at risk for blood loss, prolonged length of stay, mortality, or other morbidities. In the randomized trials, patients who were randomized to VATS, but converted to OPEN, were often excluded from the analysis or were crossed over and accounted for in the OPEN arm. In some nonrandomized trials, patients were excluded for reasons other than conversion (ie, in some retrospective trials, complicated patients were explicitly excluded,8,20,38 whereas in other retrospective trials, complicated patients were implicitly excluded through survivor bias). This overall lack of intention-to-treat analysis jeopardizes the validity of the results of each trial individually as well as for the overall aggregate meta-analysis.
In some retrospective trials, patients in the VATS group were followed for a shorter period of time than patients in the OPEN group (who were sometimes drawn from historical cohorts with a longer retrospective time line of observation). This raises the risk of death as an event in the OPEN groups if their available chart record provided a longer history than the more contemporary VATS patients.32,52
Additional limitations include the wide range of dates of the included studies, and the range of surgeon expertise represented. The included trials were published in the years ranging from 1995 to 2007. The surgeons' level of expertise was rarely specified in the included trials, and leaves the reader with uncertainty whether the results will apply equally to contemporary situations, with newer technologies, and an overall increasing expertise in the techniques associated with VATS and with OPEN surgery for lobectomy. Furthermore, in some studies, one surgeon was assigned to perform only VATS whereas another surgeon performed only OPEN surgeries. This may bias the results if the surgeons' skills or criteria for patient selection or treatment algorithm are significantly different from one another (ie, were trainees more likely to be assigned to OPEN lobectomy patients, and more experienced surgeons assigned to VATS?), and any observed differences could be due to difference in surgeons rather than differences inherent in the VATS versus OPEN approach.
Finally, treatment for lung cancer has changed significantly over the last decade, and the impact of these changing trends (primarily in the administration of adjuvant therapy and the utilization of PET imaging for staging) is unknown.
Further Research Required
Examination of the limitations of this meta-analysis highlights the urgent need to address the deficiencies inherent in the evidence base for VATS versus OPEN surgery. This meta-analysis aggregates the best available evidence, but it does not obviate the need for further randomized controlled trials; rather, this meta-analysis highlights the need for future randomized trials of adequate power to measure important differences in survival, tumor recurrence, short-term pain, long-term neuralgias, and quality of life. Also, there is a dearth of information available regarding the incremental resource requirements for VATS versus OPEN surgery, and whether these incremental requirements are worthy of the incremental benefits and potential risks. Cost-effectiveness analyses will be required to inform future policies in this area.
Lobectomies using VATS are feasible, but require novel skill sets for surgeons trained in open techniques.58,59 Concerns about the potential for intraoperative accidents and the uncertain long-term oncologic validity have limited the widespread acceptance of this technique.60,61This meta-analysis suggests that intraoperative accidents have not been commonly reported, and that long-term oncologic success maybe comparable or, perhaps, even better than conventional OPEN surgery. Caution is warranted in interpreting these results due to the paucity of prospective randomized trials to provide definitive evidence of these benefits. Furthermore, the success of VATS will depend largely on the skill of the individual surgeon, and adoption of the technique, even if it is associated with inherent advantages, does not guarantee superior results if surgical technique is suboptimal. VATS performed poorly will undoubtedly be worse than conventional OPEN performed well.
The findings of this systematic review differ somewhat from that of an earlier systematic review which contained fewer studies than the present systematic review.9 Sedrakyan et al systematically reviewed VATS for a variety of indications, including lobectomy for cancer. Because their systematic review was limited to randomized trials, they included three trials.13,14,16 Our systematic review excluded one of these because wedge resections were performed, and we identified further relevant randomized and nonrandomized trials. Therefore, this current systematic review represents an update of the earlier analysis, and allows for comparison of randomized and nonrandomized data. In addition, this current analysis allowed for statistical aggregation through meta-analysis, which was not performed in the earlier publication.
This meta-analysis suggests that there maybe some short-term, and possibly even long-term, advantages to performing lung resections with VATS techniques rather than through conventional thoracotomy. Overall, VATS for lobectomy may reduce acute and chronic pain, perioperative morbidity, and improve delivery of adjuvant therapies, without a decrease in stage specific long-term survival. However, the results are largely dependent on non-RCTs, and future adequately powered randomized trials with long-term follow-up are encouraged.
Cost-effectiveness of VATS versus OPEN for lobectomy remains unknown. These results should be interpreted conservatively because they rely primarily on nonrandomized trials and selection biases may have influenced the results favorably for VATS. Clearly, these results are also highly dependent on the appropriate selection of the patient, and the skill set of the surgeon performing VATS.
The authors acknowledge Dr. Aya Saito for her services in translating Japanese language articles, and Kathleen Broad for her services in facilitating the literature searches and retrieval.
1. Ayed AK, Al-Din HJ. Video-assisted thoracoscopy versus thoracotomy for primary spontaneous pneumothorax: a randomized controlled trial. Med Princ Pract. 2000;9:113–118.
2. Waller DA, Forty J, Morritt GN. Video-assisted thoracoscopic surgery versus thoracotomy for spontaneous pneumothorax. Ann Thorac Surg. 1994;58:372–377.
3. Ayed AK, Raghunathan R. Thoracoscopy versus open lung biopsy in the diagnosis of interstitial lung disease: a randomised controlled trial. J R Coll Surg Edinb. 2000;45:159–163.
4. Miller JD, Urschel JD, Cox G, et al. A randomized, controlled trial comparing thoracoscopy and limited thoracotomy for lung biopsy in interstitial lung disease. Ann Thorac Surg. 2000;70:1647–1650.
5. Luh S, Liu H. Video-assisted thoracic surgery—the past, present status and the future. J Zhejiang Univ Sci B. 2006;7:118–128.
6. Nakamura H. Controversies in thoracoscopic lobectomy for lung cancer. Ann Thorac Cardiovasc Surg. 2007;13:225–226.
7. Koizumi K. The revolution of thoracotomy for lung cancer surgery. Ann Thorac Cardiovasc Surg. 2007;13:228–235.
8. Demmy TL, Plante AJ, Nwogu CE, et al. Discharge independence with minimally invasive lobectomy. Am J Surg. 2004;188:698–702.
9. Sedrakyan A, van der Meulen Lewsey J, Treasure T. Video assisted thoracic surgery for treatment of pneumothorax and lung resections: systematic review of randomized clinical trials. BMJ. 2004;329:1008–1011.
10. Alderson P, Green S, HJiggins J. Cochrane Reviewers' Handbook. 4.2.1 ed. Chichester, UK: Jonhn Wiley & Sons Ltd.; 2004.
11. Moher D, Pham B, Jones A, et al. Does quality of reports of randomized trials affect estimates of intervention efficacy reported in meta-analyses? Lancet. 1998;22:609–613.
12. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 21:1539–1558.
13. 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.
14. Craig SR, Leaver HA, Yap PL, et al. Acute phase responses following minimal access and conventional thoracic surgery. Eur J Cardiothorac Surg. 2001;20:455–463.
15. Leaver HA, Craig SR, Yap PL, et al. Lymphocyte responses following open and minimally invasive thoracic surgery. Eur J Clin Invest. 2000;30:230–238.
16. Kirby TJ, Mack MJ, Landreneau RJ, Rice TW; Review Group(s): Cochrane Anaesthesia Group, Cochrane Lung Cancer Group. Lobectomy–video-assisted thoracic surgery versus muscle-sparing thoracotomy. A randomized trial. J Thorac Cardiovasc Surg. 1995;109:997–1001.
17. Giudicelli R, Thomas P, Lonjon T, et al. Video-assisted minithoracotomy versus muscle-sparing thoracotomy for performing lobectomy. Ann Thorac Surg. 1994;58:712–717.
18. Thomas P, Doddoli C, Yena S, et al. VATS is an adequate oncological operation for stage I non-small cell lung cancer. Eur J Cardiothorac Surg. 2002;21:1094–1099.
19. Kawai H, Tayasu Y, Saitoh A, et al. Nocturnal hypoxemia after lobectomy for lung cancer. Ann Thorac Surg. 2005;79:1162–1166.
20. Li WWL, Lee RLM, Lee TW, et al. The impact of thoracic surgical access on early shoulder function: video-assisted thoracic surgery versus posterolateral thoracotomy. Eur J Cardiothorac Surg. 2003;23:390–396.
21. Nakata M, Saeki H, Yokoyama N, et al. Pulmonary function after lobectomy: video-assisted thoracic surgery versus thoracotomy. Ann Thorac Surg. 2000;70:938–941.
22. Nagahiro I, Andou A, Aoe M, et al. Pulmonary function, postoperative pain, and serum cytokine level after lobectomy: a comparison of VATS and conventional procedure. Ann Thorac Surg. 2001;72:362–365.
23. Ng CS, Lee TW, Wan S, et al. Thoracotomy is associated with significantly more profound suppression in lymphocytes and natural killer cells than video-assisted thoracic surgery following major lung resections for cancer. J Invest Surg. 2005;18:81–88.
24. Ng CS, Wan S, Hui CW, et al. Video-assisted thoracic surgery lobectomy for lung cancer is associated with less immunochemokine disturbances than thoracotomy. Eur J Cardiothorac Surg. 2007;31:83–87.
25. Ohbuchi T, Morikawa T, Takeuchi E, Kato H. Lobectomy: video-assisted thoracic surgery versus posterolateral thoracotomy. Jpn J Thorac Cardiovasc Surg. 1998;46:519–522.
26. Stanbridge RDL, Hon JKF, Bateman E, Roberts S. Minimally invasive anterior thoracotomy for routine lung cancer resection. Innovations. 2007;2:76–83.
27. Tajiri M, Maehara T, Nakayama H, Sakamoto K. Decreased invasiveness via two methods of thoracoscopic lobectomy for lung cancer, compared with open thoracotomy. Respirology. 2007;12:207–211.
28. Tajiri M, Maehara T, Nakayama H. [Thoracoscopic lobectomy (non rib-spreading method): versus a video assisted lobectomy and a conventional thoracotomy]. Kyobu Geka. 2000;53:13–17.
29. Yamagishi S, Koizumi K, Shimizu K. Assessment of the perioperative hemodynamics and right ventricular performance of lung cancer patients using a continuous cardiac output monitoring system: comparison between video-assisted thoracic surgery and muscle-sparing thoracotomy. Ann Thorac Cardiovasc Surg. 2006;12:166–173.
30. Yim APC, Wan S, Lee TW, Arifi AA. VATS lobectomy reduces cytokine responses compared with conventional surgery. Ann Thorac Surg. 2000;70:243–247.
31. Cattaneo SM, Park BJ, Wilton AS, et al. Use of video-assisted thoracic surgery (VATS) for lobectomy in the elderly results in fewer complications. Ann thorac Surg. 2008;85:231–235.
32. Demmy TL, Curtis JJ. Minimally invasive lobectomy directed toward frail and high-risk patients: a case-control study. Ann Thorac Surg. 1999;68:194–200.
33. Inada K, Shirakusa T, Yoshinaga Y, et al. The role of video-assisted thoracic surgery for the treatment of lung cancer: lung lobectomy by thoracoscopy versus the standard thoracotomy approach. Int Surg. 2000;85:6–12.
34. Inaoka M, Obama T, Kawaharada N. [Video-assisted minithoracotomy versus conventional posterolateral thoracotomy for performing lobectomy of lung carcinomas]. Kyobu Geka. 2000;53:18–21.
35. Iwasaki A, Shirakusa T, Kawahara K, et al. Is video-assisted thoracoscopic surgery suitable for resection of primary lung cancer? Thorac Cardiovasc Surg. 1997;45:13–15.
36. Kaseda S, Aoki T, Hangai N, Shimizu K. Better pulmonary function and prognosis with video-assisted thoracic surgery than with thoracotomy. Ann Thorac Surg. 2000;70:1644–1646.
37. Koizumi K, Tanaka S, Haraguchi S, et al. Lobectomy by video-assisted thoracic surgery for primary lung cancer: experiences based on provisional indications. Surg Today. 1998;28:36–40.
38. Li WWL, Lee TW, Lam SSY, et al. Quality of life following lung cancer resection: video-assisted thoracic surgery vs thoracotomy. Chest. 2002;122:584–589.
39. Muraoka M, Oka T, Akamine S, et al. Video-assisted thoracic surgery lobectomy reduces the morbidity after surgery for stage I non-small cell lung cancer. Jpn J Thorac Cardiovasc Surg. 2006;54:49–55.
40. Nomori H, Horio H, Naruke T, Suemasu K. What is the advantage of a thoracoscopic lobectomy over a limited thoracotomy procedure for lung cancer surgery? Ann Thorac Surg. 2001;72:879–884.
41. Nomori H, Ohtsuka T, Horio H, et al. Difference in the impairment of vital capacity and 6-minute walking after a lobectomy performed by thoracoscopic surgery, an anterior limited thoracotomy, an anteroaxillary thoracotomy, and a posterolateral thoracotomy. Surg Today. 2003;33:7–12.
42. Petersen RP, Pham D, Burfeind WR, et al. Thoracoscopic lobectomy facilitates the delivery of chemotherapy after resection for lung cancer. Ann Thorac Surg. 2007;83:1245–1250.
43. Petersen RP, Pham D, Toloza EM, et al. Thoracoscopic lobectomy: a safe and effective strategy for patients receiving induction therapy for non-small cell lung cancer. Ann Thorac Surg. 2006;82:214–218.
44. Shiraishi T, Yoshinaga Y, Yoneda S, et al. [Clinical evaluation of VATS lobectomy for lung cancer]. Kyobu Geka. 2000;53:4–7.
45. Shiraishi T, Shirakusa T, Hiratsuka M, et al. Video-assisted thoracoscopic surgery lobectomy for c-T1N0M0 primary lung cancer: its impact on locoregional control. Ann Thorac Surg. 2006;82:1021–1026.
46. Shiraishi T, Shirakusa T, Miyoshi T, et al. A completely thoracoscopic lobectomy/segmentectomy for primary lung cancer – technique, feasibility, and advantages. Thorac Cardiov Surg. 2006;54:202–207.
47. Sugi K, Kaneda Y, Esato K. Video-assisted thoracoscopic lobectomy reduces cytokine production more than conventional open lobectomy. Jpn J Thorac Cardiovasc Surg. 2000;48:161–165.
48. Sugi K, Kaneda Y, Nawata k, et al. Cost analysis for thoracoscopy: thoracoscopic wedge resection and lobectomy. Jpn J Surg. 1998;28:41–45.
49. Sugiura H, Morikawa T, Kaji M, et al. Long-term benefits for the quality of life after video-assisted thoracoscopic lobectomy in patients with lung cancer. Surg Laparosc Endosc Percutan Tech. 1999;9:403–408.
50. Shigemura N, Akashi A, Funaki S, et al. Long-term outcomes after a variety of video-assisted thoracoscopic lobectomy approaches for clinical stage IA lung cancer: a multi-institutional study. J Thorac Cardiovasc Surg. 2006;132:507–512.
51. Tashima T, Yamashita JI, Nakano S, et al. Comparison of video-assisted minithoracotomy and standard open thoracotomy for the treatment of non-small cell lung cancer. Minim Invasive Ther Allied Tech. 2005;14:203–208.
52. Tatsumi A, Ueda Y. Video-assisted thoracic surgery for lung cancer: is it a feasible operation for stage I lung cancer? Jpn J Thorac Cardiovasc Surg. 2003;51:646–650.
53. 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;138:510–517.
54. Watanabe A, Koyanagi T, Obama T, et al. Assessment of node dissection for clinical stage I primary lung cancer by VATS. Eur J Cardiothorac Surg. 2005;27:745–752.
55. Shigemura N, Akashi A, Nakagiri T, et al. Complete versus assisted thoracoscopic approach: a prospective randomized trial comparing a variety of video-assisted thoracoscopic lobectomy techniques. Surg Endosc. 2004;18:1492–1497.
56. Wan S, Izzat MB, Lee TW, et al. Avoiding cardiopulmonary bypass in multivessel CABG reduces cytokine response and myocardial injury. Ann Thorac Surg. 1999;68:52–57.
57. Ng CS, Whelan RL, Lacy AM, Yim AP. Is minimal access surgery for cancer associated with immunologic benefits? World J Surg. 2005;29:975–981.
58. Roviaro G, Varoli F, Vergani C, et al. Long-term survival after videothroacoscopic lobectomy for stage I lung cancer. Chest. 2004;126:725–732.
59. Swanson SJ, Herndon JE, D'Amico TA, et al. Video-assisted thoracic surgery lobectomy: report of CALGB 39802—a prospective, multi-institution feasibility study. J Clin Oncol. 2007;25:4993–4997.
60. McCulloch P. Half full or half empty VATS? BMJ. 2004;329:1012.
61. Sedrakyan A, van der Meulen J, Lewsey J, et al. Variation in use of video assisted thoracic surgery in the United Kingdom. BMJ. 2004;329:1011–1012.
© 2007 Lippincott Williams & Wilkins, Inc.