1. IntroductionSurgical resections for centrally located non-small cell lung cancer (NSCLC) are associated with massive parenchymal extirpation and poor survival outcomes due to the highly aggressive nature of the disease, along with interlobar and mediastinal lymph nodal diaspora [1]. Pneumonectomy (PN) represents the traditional surgical approach for patients with centrally located tumors, leading to a substantial decline in lung function and quality of life, thus precluding adjuvant treatment or the further resection recurrence of the disease [2]. Consequently, PN is associated with certain restrictions in the treatment pathway, along with significant postoperative morbidity. To face this challenge, several approaches have been described by departments that have implemented the bronchovascular sleeve resection as an aggressive lung-preserving alternative to PN [1,2,3,4]. The classic bronchoplastic procedure for NSCLC includes the reconstruction of the pulmonary artery (PA), which has superior short- and long-term outcomes compared to those of PN [1,2]. In this context, the implementation of bronchoplastic procedures has exceeded that of PNs in patients with centrally located NSCLC, thus leading to an increasing ratio of sleeve lobectomies compared with PNs [2,3]. The classic sleeve lobectomy involves the resection of one lobe with an end-to-end bronchial anastomosis. Nonetheless, the management of centrally located NSCLC tumors may require the resection of more than one lobe, along with airway anastomoses in segmental bronchi and pulmonary vascular reconstructions [4,5]. The extended sleeve lobectomy (ESL) represents an atypical bronchoplasty with resections of more than one lobe and, consequently, is a more technically demanding procedure. Nonetheless, ESL has certain theoretical advantages, leading to its proposal as an alternative approach to PNs. In fact, although there is a significant interest in ESL as a treatment for centrally located NSCLC, there is limited available evidence comparing its survival, perioperative, and oncologic outcomes with PN. Therefore, the purpose of the present study is to summarize the existing data in the literature by comparing the survival and perioperative outcomes of ESL and PN for centrally located tumors and to provide the best up-to-date and currently available level of evidence on the topic. 4. DiscussionThe current evidence provided by the literature on the benefits of ESL over PN for centrally located NSCLC remains limited, and there is no RCT available. In this context, the current meta-analysis represents the highest available level of evidence. In fact, there is no other meta-analysis available in the literature to the best of our knowledge. The present meta-analysis included nine articles comparing ESL and PN for central tumors using reconstructed time-to-event patient data. Given the great technical complexity of ESL compared with simple sleeve lobectomy and PN, adequate survival and oncologic outcomes should be demonstrated to counterbalance the perioperative risk. As a result, the survival outcomes were our primary endpoints. According to the outcomes of the present meta-analysis, ESL is associated with higher OS and DFS compared to PN. In the same context, ESL demonstrated a high rate of R0 resection, thus reaffirming the oncologic adequacy of the procedure. Moreover, the operative mortality was relatively low, with only three deaths (1.2%) reported in all the included articles. This outcome is similar to the mortality in patients undergoing sleeve lobectomy (SL) (1.3%) and lower than pneumonectomy (5.3%), as demonstrated in a large study including 1,230 patients [1]. Nonetheless, complications were not rare due to the high technical complexity of these procedures. Given the promising results of other treatment strategies, along with the induction treatment for advanced NSCLC, ESL has the potential to become a lung-sparing approach of choice for selected patients. Based on the promising outcomes of this strategy, more centers have tended to adopt it, a trend that explains the fact that most of the included studies were published during the last five years.Different ESL procedures were employed in the included studies. As was previously commented, these are divided according to the modified Okada classification to types A–E, and each one has its own characteristics. Type A procedures require a long bronchial resection from the level of the right main to the basal segment bronchus. Consequently, the management to reduce the anastomosis-related tension is important to prevent anastomotic complications. In this context, type A ESL procedures frequently require a combined angioplasty of the pulmonary artery. The same principles regarding the extent of bronchial resection exist also in type B procedures. On the other hand, type C ESL has different technical characteristics, given that a size discrepancy might occur between the proximal and distal bronchial stump, thus highlighting the need for a careful caliber adjustment in anastomosis. These characteristics are also similar to type D ESL procedures. Due to these special traits, the extent of lung-sparing in both C and D procedures might be less compared with other types. Consequently, the meticulous management of the residual pleural space is crucial to preventing space-related postoperative morbidities, such as empyema and BPF. Perhaps the meticulous drainage of the chest cavity or the artificial phrenic nerve palsy might reduce the incidence of these complications [18]. Finally, the characteristics of the type E ESL procedure are similar to type A.Due to their high technical complexity, ESL procedures are associated with a significant incidence of postoperative morbidity. However, the rate of reoperations was relatively low (1.5%) compared to SL (1.8%) and PN (23%), according to literature data [2]. Major complications are delayed air leakage, bronchial strictures, and BPFs. To reduce the tension of bronchial reconstruction in type A ESL procedures, a transposition of the inferior to the superior pulmonary vein or a PA reconstruction is often required. Owning to these operating maneuvers, vascular complications, such as PA thrombosis or the necrosis of the lung parenchymal remnant, are not rare [18]. According to our outcomes, the incidence of BPF and stricture was 4.6% and 3.1%, respectively. The incidence of BPF was higher compared with SL (1.8%) but significantly lower than pneumonectomy (14%), according to a previous study [2]. In addition, persistent air leakage was also relatively low (7.3%) compared to SL (12.7%) [2]. Furthermore, sputum retention was another relatively common complication according to our outcomes, which may require the endoscopic cleaning of airways to maintain good patency of the airways and lung expansion. An additional role of bronchoscopy is to evaluate the quality of the anastomosis prior to patient’s discharge [19]. Finally, PVT was another less frequent complication (1.5%) that can occur as a result of the overstretching of the pulmonary vein. The use of pericardial cutting has been proposed as a measure to prevent this complication in type A and B procedures [18].

Given the lack of randomized clinical trials comparing the feasibility of ESL over PN for centrally located NSCLC, the current work is the largest up-to-date comparative study, incorporating 431 patients. The present analysis supports the superiority of ESL in terms of long-term survival over PN for patients with central NSCLC. Nonetheless, due to the high complexity of these procedures, they should be performed by experienced thoracic surgeons in high-volume centers. Consequently, it is crucial to define the exact selection criteria for the best candidates to undergo ESL. In this context, the current meta-analysis provides the best currently available level of evidence that might help multidisciplinary decision-making on complex cases, given the lack of guidelines on the topic. The herein presented evidence should be taken into account during the composing of future guidelines on the management of central lung tumors. Nonetheless, our outcomes should be further validated by well-designed RCTs.

The limitations of the present study are mainly associated with the limitations of the included studies. Most of the studies were retrospective, and no RCT was identified through the literature search, thus posing a certain limitation in this study. Furthermore, the incorporated studies are related to biases related to participants’ selection and performance. In addition, the differences among institutions regarding the treatment protocols, selection criteria, and perioperative management pose several limitations. In the same context, the selection criteria were not homogenous and may have been based on the patients’ clinical attributes and status, thus posing a selection bias that could not be adjusted in the present study. Finally, patient data were gathered from Kaplan–Meier-derived data, and not from individual patient follow-ups, thus limiting the ability to perform further subgroup, multivariate, or propensity score matching analyses.

On the other hand, the strengths of the present meta-analysis include (i) the clear literature search and data extraction protocol, (ii) the well-specified inclusion/exclusion criteria, (iii) the literature search in three databases, (iv) the quality assessment of the included studies, (v) the detailed presentation of the outcomes, (vi) the extraction of survival data at the level of the independent patient, and (vii) the performance of sensitivity analyses.