The National Emphysema Treatment Trial (NETT) randomized patients to bilateral lung volume reduction surgery (LVRS) or medical management for severe emphysema, comparing the primary endpoints of mortality and maximal exercise capacity.1 After excluding the highest risk patients,2 LVRS resulted in the significant functional improvement in three of four subgroups of patients and a survival benefit for the subset of patients with upper lobe emphysema and low exercise capacity. A secondary analysis of the NETT was designed to compare mortality, morbidity, and functional outcomes between the two approaches to surgical resection: median sternotomy (MS) and video-assisted thoracoscopic surgery (VATS).3 Of 17 centers in the NETT, six patients were randomized to VATS versus MS, whereas the others were not randomized according to technique. Therefore, both randomized and nonrandomized data were available on this topic. Before the NETT, there had only been individual institutional reports comparing MS with VATS in a nonrandomized fashion.4–6
In the NETT report addressing outcomes after the MS versus VATS approaches, there was no significant difference found between the groups in functional results, and the morbidity incurred was described as “comparable.”3 VATS patients did, however, demonstrate a shorter median hospital length of stay (LOS) and earlier return to independent living in both the nonrandomized and randomized data sets. Furthermore, in both data sets, VATS patients incurred significantly lower costs in the 6 months after surgery. Finally, in the randomized data set, the MS group showed a higher rate of failure to wean. A recent study measuring serum cytokine responses following these two approaches to LVRS demonstrated reduced postoperative release of proinflammatory cytokines for the VATS approach.7 The authors suggested that this might be a factor either reflecting, or etiologic in, the slightly improved outcomes seen after VATS in the NETT.
In the current report, we present a single surgeon's series of bilateral LVRS performed by either MS or VATS. We felt that the consistency in surgical technique and perioperative management that would be found in a series of patients cared for by the same surgeon might provide information additive to that which was obtained in the NETT and single-institution studies that used predominately one approach or the other. The surgeon was comfortable performing the procedure by either approach, believing that each approach has certain advantages. Patients were selected for one approach or the other on the basis of individual patient characteristics and desires—strongest indication for the VATS approach being an overall sense of an elevated risk profile in a particularly patient. Our hypothesis was that if we selected patients who are estimated to have a higher risk profile—by a variety of criteria—for the VATS approach, that these patients who might otherwise be at greater risk for postoperative complications might be offered the operation without a substantial increase in morbidity and mortality. The objectives of this study, therefore, were to determine (1) whether we could establish quantitatively that patients selected for VATS in fact had a higher risk profile, and (2) whether, despite this increased risk profile, the VATS patients had equivalent or reduced perioperative morbidity and mortality compared with lower risk MS patients.
Over a 9-year period, 67 patients underwent LVRS by a single surgeon outside of the NETT. We performed a retrospective chart review of these patients with distribution into two groups: patients who underwent bilateral video-assisted thoracoscopic resection (VATS) and patients who underwent MS for bilateral resection. The choice of surgical approach (VATS vs. MS) was determined in a nonrandomized fashion. Key factors in this determination were the surgeon's assessment of specific patient preoperative characteristics (Table 1), which are discussed below. Generally, patients with a higher risk profile based on a subjective assessment of the several criteria listed in Table 1, plus several less easily quantified criteria, were selected to undergo VATS. Those less easily quantified criteria included degree of infirmity and work of breathing observed on physical examination, and degree of chronic bronchitis as suggested by reported amount of chronic sputum production. It was not a specific combination of preoperative characteristics that identified a higher risk patient but ultimately the surgeon's general impression of the overall profile. The selection criteria for LVRS itself were based initially on the recommendations from the early experience of Cooper et al8 and subsequently by the results of the NETT.1,2
Both groups underwent routine flexible bronchoscopy at the start of the operation with sputum aspiration and culture to guide any early postoperative need for antibiotic therapy. In the MS group, operation was performed in the supine position through a standard MS incision. A bougie retractor was used to alternately elevate each hemisternum to the minimal degree that allowed sufficient exposure for adhesiolysis and resection. A single 28F, straight intercostal chest tube was placed into each hemithorax at the completion of the procedure. In the VATS group, operation was performed in the full lateral decubitus thoracotomy position. After completion of one side, the patient was repositioned to the opposite lateral decubitus position for contralateral resection. The VATS technique evolved to a three-port, nontriangulated technique (Fig. 1). We term this “nontriangulated” because the anterior and inferior ports were placed spatially very close to one another, in contrast to typical VATS approaches where each of the three ports is approximately equidistant from the two other ports. The two closely spaced ports were used primarily for the camera and linear stapler, and we feel this arrangement provided better visualization of the junction of the previous staple line than some other port placements might, allowing accurate placement of each additional staple line directly at this established junction. The third port was placed more cephalad, just posterior to the tip of the scapula. More posterior placement than this was avoided to reduce the risk of intercostal nerve injury that can occur when working through the narrower, far posterior intercostal spaces. A single, 28F, straight chest tube was tunneled on each side to the apex of the chest posteriorly. All patients were extubated in the operating room. Immediate waterseal was routinely used to minimize air leaks. Low-level suction (−10 cm) was applied to the tubes only if large pneumothoraces or progressive subcutaneous emphysema developed.
The method of parenchymal excision in both groups was by serial nonanatomic continuous resection lines using a linear stapling device (GIA by MS; endoGIAII by VATS; Covidien Autosuture, Norwalk, CT). In all patients, the most diseased areas as assessed by preoperative imaging and intraoperative inspection were resected, with a goal of removing at least 25% of the lung volume on each side. The visceral pleura within the intended area of resection was frequently incised to deflate the lung before resection. Care was taken to place all subsequent staple lines precisely at the junction of the previous staple lines, so as not to have crisscrossing staple lines. The resection on each side was most often achieved with a single, long, u-shaped resection line in the MS group, and with two or three nonintersecting, continuous staple lines in the VATS group (Figs. 2A and B). In the MS group, all of the staple lines were reinforced with polyterafluoroethylene strips (W.L. Gore & Associates, Inc, Flagstaff, AZ). In the VATS group, the first 9 patients had no staple line reinforcement because the surgeon was not satisfied with the buttresses available for VATS staplers at that time. The subsequent VATS patients had reinforcement with bioabsorbable Seamguard (W.L. Gore & Associates, Inc).
Patient-controlled thoracic epidural anesthesia with fentanyl and bupivacaine was maintained until at least postoperative day 4, and at least until removal of one of the chest tubes, before conversion to oral narcotics. Ketorolac was used as needed during the first 48 hours postoperatively if not contraindicted. Ibuprofen was used on an as-needed basis after 48 hours to supplement the epidural or oral narcotics. Patients who were otherwise clinically stable but with a continued, unilateral air leak on waterseal at postoperative day 7 were discharged home with a Heimlich valve.
Several preoperative and postoperative variables were collected for each patient. Preoperative factors felt important to the risk profile included age, weight, oxygen requirement, DLCO, FEV1, total lung capacity, residual volume, PaO2, PaCO2, and evidence of pulmonary hypertension. Although the test was consistently done, we found it difficult to find the six-minute walk test results on many of the patients, so we do not have reliable data on this potentially important variable. Postoperative outcomes analyzed included LOS, total chest tube days, total intensive care unit (ICU) days, total operating time, estimated blood loss, discharge with a Heimlich valve, major complications (bleeding requiring transfusion, perforated ulcer, pneumonia, bacteremia, and respiratory failure), and atrial fibrillation. We used the independent samples t test to examine differences between VATS and MS groups for continuous outcome variables and Fisher exact test tests for dichotomous variables.
Of the 67 patients undergoing LVRS procedures, only the 53 who underwent single-stage, bilateral LVRS were included in this study. The remaining 14 patients underwent either unilateral LVRS procedures or unilateral, combined LVRS/lung cancer operations.
Eighteen patients underwent bilateral VATS LVRS. Fifteen of these had full records and were therefore included in the final analysis. All 35 patients who underwent MS for LVRS had full records and were thus included. The male to female ratio was similar with 47% males and 46% males in the VATS and MS groups, respectively (P = 1.0). The average perfusion to the upper lung zones was 7.3% and 6.9% for the VATS and MS groups, respectively (P = 0.66).
There was only one (1.9%) 90-day mortality in the entire study group of 53 patients (due to respiratory failure in a single patient in the MS group). There were no mortalities for the VATS group.
With regard to what we have called “risk profile,” there were no differences that reached statistical significance between the two groups (Table 1). There were, however, nonsignificant, but notable differences, which we believe, confirm our hypothesis that our subjective selection process created a VATS group that had a slightly higher risk profile than the MS group. Age was the closest parameter to reaching significance but fell short (P = 0.08) with a slightly older population in the VATS group. The VATS group also contained a substantially greater fraction of patients with more than mild pulmonary hypertension (P = 0.11). Variables that were farther from reaching statistical significance included the findings that mean body mass was slightly lower in the VATS group (P = 0.35), and that the VATS group seemed to be more hyperexpanded, with a slightly higher mean residual volume (P = 0.32).
In terms of outcomes, only one parameter demonstrated a statistically significant difference between the groups (Table 2). The operative time was significantly longer for the VATS group (P = 0.01), which is not surprising given the need for repositioning in the midst of the operation by our technique. The VATS group also had a lower mean blood loss, although the difference did not reach significance (P = 0.13). The total major complication rate for both groups combined was 15% and was similar between the two groups, with a slightly lower rate in the VATS group (P = 0.39). The atrial fibrillation rate was lower within the VATS group, but this difference also did not reach statistical significance (P = 0.33). Although both LOS and ICU days seems higher (P = 0.10 and 0.69, respectively) in the VATS group, this resulted from a single outlier in the VATS group who spent 61 days in the hospital and 53 days in the ICU. If this patient is excluded from the analysis, then the mean LOS and ICU days become very similar in the VATS and MS groups, with the differences far from approaching significance (Table 2 footnotes).
The mean number of chest tubes days was statistically similar between the two groups but greater within the VATS group. This coincides with a greater number of VATS patients, requiring a Heimlich valve on discharge (which were then monitored for the possibility of removal once weekly), but again this difference did not reach significance. Heimlich valve management of air leaks was highly effective, as there were no reoperations for air leaks in either group.
A secondary analysis of the NETT compared VATS versus MS in both small, randomized and larger, unrandomized cohorts.3 Although the conclusions of this analysis were that outcomes were “comparable” between the groups, a closer look reveals that the VATS group had shorter LOS, earlier return to independent living, lower costs, and likely a lower rate of failure to wean.
The current nonrandomized study selected patients for VATS or MS based on a single surgeon's estimation of an individual patient's risk profile. Because there was only one attending surgeon involved in this study, no surgeon variability (in patient selection, intraoperative, or postoperative care) was introduced. Patients who were felt to be at greater risk for postoperative complications, yet who remained reasonable surgical candidates, were offered a minimally invasive approach in hopes of minimizing morbidity and mortality. A prior study involving the senior author reporting a reduced serum level of proinflammatory cytokines with the bilateral thoracoscopic approach encouraged him to view the VATS approach as potentially less morbid.7
Our data confirm, though perhaps only weakly, that a higher risk profile was present within the VATS group. Of the 10 data points listed in Table 1 that we measured as a reflection of risk profile, all but two suggested that the VATS patients were at greater risk. We found, however, only trends in this regard, with no measure meeting statistical significance. This may be due to the small sample size within the VATS group: a clear limitation of the study. The VATS patients, did, however, tend to be older, more hyperexpanded, with a greater incidence of pulmonary hypertension, and with slightly lower mean body mass.
With regard to early outcomes, the only finding reaching significance that we identified was the longer operative time for VATS. This was felt to be due to the need for repositioning between sides by our technique of VATS. We feel, however, that this “repositioning time” has potential benefit, as it allows time to ventilate both lungs to reduce pCO2 before operating on the second side. This is important because it is critical to minimize inspiratory pressures on the lung that was operated on first and subsequently becomes the only lung available for ventilation, and there is therefore a tendency for respiratory acidosis to develop while resecting the second side. The surgeon's recollection (no data collected on this issue) is that one generally needed to wait longer in the operating room after surgery for MS patients to breathe off CO2 before extubation than was necessary for the VATS patients.
VATS patients did have a greater (though not significantly greater) number of patients requiring Heimlich valves for management of prolonged air leaks. It is possible that this resulted in part from the VATS patients having more severe emphysema or from the fact that the earliest VATS patients had no staple line buttressing used. We believe it is most likely, though, that this finding resulted from the VATS patients, apart from a unilateral persistent air leak, being otherwise ready for discharge at postoperative day 7 more frequently than were the MS patients. Because air leaks in Heimlich valves were checked as outpatients only once weekly, patients with Heimlich valves (more VATS than MS patients) would have therefore had falsely elevated chest tube [CT] durations in contrast to those who remained in the hospital. This likely explains the greater (but not significantly greater) CT duration in the VATS group. The randomized data for the NETT showed no difference between the groups for the presence of prolonged air leaks (>7 days), the total number of air leak days, or the rate of reoperation for air leaks.3
Our results for LOS and number of ICU days seem at first glance to be higher for the VATS group, though not significantly greater. When the single VATS outlier who spent 61 days in the hospital and 53 days in the ICU was excluded from the analysis, the differences between VATS and MS groups did not even approach significance (VATS=10.8 day mean LOS, P = 0.56, VATS = 0 ICU days, P = 0.19; Table 2 footnote). We therefore believe that the divergent results on these outcomes can be attributed to our low number and to chance.
Perhaps most importantly, the rates of major complications were no different between the VATS and MS groups, and there were no deaths in the VATS group and only one in the MS group. We believe the finding that our risk profile suggested higher preoperative risk in the VATS group, but that the VATS group nevertheless had no increase (in fact less) major complications, suggests that the VATS approach is an optimal approach for the highest risk LVRS patients. It can be argued that from a purely technical point of view the MS approach is ideal. Perhaps, the worst mistake one can make in performing LVRS is to resect too little lung tissue, as doing this incurs the morbidity of the incision(s) without providing much physiological benefit. The authors believe that this is a common mistake, particularly early in one's LVRS experience, and it is certainly possible that the magnification obtained by the thoracoscopic approach, and the less wide jaw-opening of available endoscopic staplers, encourages one to make this mistake. However, the randomized data set for the NETT showed no difference in actual grams of lung parenchyma resected, so this potential downside of VATS LVRS can clearly be overcome.3 The benefits of the reduced initial surgical insult to patients incurred by the VATS approach over MS7 may only be important in the most compromised surgical candidates.
A brief comment on technical lessons learned from this experience is appropriate. Our three-port, nontriangulated approach for VATS evolved over time. The initial approach was more of a triangulated approach as is generally optimal for other types of thoracoscopic procedures. We found, though, that for VATS LVRS, placing the camera and stapling ports in close proximity to one another allows for improved visualization of the junction of the previous staple line, helping to allow placement of each subsequent staple firing directly at the junction created by the previous staple fire. We believe that the ability to do this may minimize air leaks. Because of the different perspective our VATS approach provided, we elected to resect slightly different areas of the lung in our VATS than in our MS cases. As shown in Figures 2A and B, we found that the “inverted U” of the MS approach was not ideal by VATS. We therefore evolved to resecting a nearly equivalent region of lung during VATS by creating two separate staple lines in the right upper lobe: one anterior and one posterior. We kept these staple lines nonintersecting. Because the VATS approach provides better exposure than MS to the superior segment of the lower lobe, and because this area is often severely involved with emphysema, we more often excised a portion of the superior segment by the VATS approach.
Based on the results of this study, our personal experience with LVRS, and other studies including the NETT, if we were to propose a set of criteria to choose patents for the VATS approach going forward, it would include some combination of the following: FEV1 <25% predicted, DLCO <25% predicted, homogeneous disease, pCO2 >45, room air pO2 <60, BMI >10% below ideal, 6-minute walk distance <800 feet after pulmonary rehabilitation, more than mild pulmonary hypertension, subjective component of chronic bronchitis (ie, sputum production), and older than 70 years. An algorithm including these criteria could be studied in prospective fashion.
In conclusion, in this nonrandomized, single-surgeon series comparing bilateral VATS with bilateral MS for LVRS, patients who underwent LVRS by the VATS approach tended to have a higher risk profile by a variety of criteria. Despite this, there were no substantial differences in outcomes. We believe that this suggests that the VATS approach creates a slightly lesser initial setback to LVRS patients than does the MS approach, and that therefore the VATS approach is optimal in the most compromised patients. Only a comparison of longer term, physiological outcomes would allow us to answer the question of whether there is any reason to continue to use the MS approach in lower risk profile patients. Given the low postoperative morbidity and mortality rates we have achieved, we continue to use both approaches according to the individualized selection processes described here.