Postpneumonectomy bronchopleural f i stula: cohort of pat i ents i n a Pevur i an Nat i onal Hosp i tal

Background: Bronchopleural fistula (BPF) is a complication with a great impact on morbidity, mortality and survival in postoperative pneumonectomy patients. Our study aimed to determine the associated clinical-surgical characteristics of the BPF in a Peruvian cohort of patients with BPF following pneumonectomy Materials and Methods: Cross-sectional, and retrospective study during January 2015-June 2020 in a Peruvian National Hospital. Medical records and operative reports were reviewed, and all patients over age (>18 years) and with all-cause pneumonectomy were considered. The variables were grouped into pneumonectomies with and without BPF, and the characteristics were grouped as a baseline, clinical-surgical and postoperative. Results: Among fifty-seven patients who underwent pneumonectomy, 28% presented BPF. The mean age was 38.3 ± 15 years, and 62.5% were male. The main pathologies associated with BPF were post-tuberculosis fibrocavitary sequelae (62.5%), pulmonary hydatidosis (18.8%), and post-surgical cavitary sequelae (12.5%). 81.3% were left BPF (p = 0.342) and 100% were related to postoperative empyema (p &lt; 0.05). The mortality associated with BPF was 6.25%, and in the non-BPF group it was 4.87% (p = 0.098). Characteristics that were statistically associated with the presence of bronchopleural fistula were low predicted forced expiratory volume (FEV1) (56.3% vs. 31.7%; p = 0.046), preoperative empyema (37.5% vs. 2.4%; p = 0.001), postoperative empyema (100% vs. 9.8%; p = 0.000) and operative time (450.69 min vs. 367.73 min; p = 0.032). Conclusions: There are a series of factors associated with the presence of patients with BPF following pneumonectomy, many of them mainly related to operative and postoperative events, which is an important predictor of associated complications in our Peruvian population.


Introduction
Bronchopleural fistula (BPF) is a complication with a great impact on morbidity and mortality (6-71%) in postoperative pneumonectomy patients [1][2][3], and is defined as the communication of a bronchus into the pleural cavity.Many cases are more frequently associated with patients with BPF following pneumonectomy than with other pulmonary resections [4].It has been reported in 20% of postoperative patients with complete lung resection due to benign etiology (Mycobacterium tuberculosis) and 4.5% in lung cancer [5,6].The diagnosis of post-pneumonectomy BPF can be early or late and leads to a chronic situation that weakens the patient, leading to malnutrition problems, risk of reactivating tuberculous infections with Multidrug-Resistant germs (MDR), and even leading to death in 10% [1,4,7].Among the main factors widely described are a history of pulmonary tuberculosis (35%), malnutrition (21%), and right pneumonectomy (15%) [1][2][3][8][9][10][11].
Latin America has with the highest rate of indication of extensive tuberculous pulmonary resections in our environment.This has led to the need to implement operating rooms strictly for this type of surgery such as the one in our institution, with higher rates of BPF than in lung cancer surgeries.Although the surgical technique has progressed, which has generated a decrease in the incidence of BPF, it continues to generate great concern for the thoracic surgeon due to the multifactorial origin of this complication.Therefore, our study aimed to determine the associated clinical-surgical characteristics of the BPF in a Peruvian cohort of patients who underwent pneumonectomy.

Design, population, and sample size
A cross-sectional and retrospective study was conducted.Fifty-seven post-pneumonectomy patients previously evaluated by the thoracic surgery service of a Hospital Nacional from Peru during January 2015-June 2020 were studied.Inclusion criteria were the majority of age (>18 years), and pneumonectomy due to different etiologies (tuberculosis, pulmonary hydatidosis, malignant pulmonary tumor, benign pulmonary tumor, severe pulmonary sequelae, and others) and that ended with or without BPF.All patients who underwent pneumonectomy with or without BPF outside the study period and without complete information were excluded.

Data collection and study variables
Data collection was retrospective, and medical records and operative reports were evaluated using a data collection form.Subsequently, the data were tabulated in an orderly manner and respected the chronological order of the patients.A checklist was used, and to avoid any tabulation bias, each of the data obtained was double-checked.The study variables were divided into 2 groups (Pneumonectomies with BPF vs. Pneumonectomies without BPF) and the characteristics evaluated were baseline (age, sex, comorbidities, smoking and alcoholism history), clinical-surgical (pulmonary tuberculosis, type of surgical indication, malnutrition, anemia, hypoalbuminemia, chronic obstructive pulmonary disease (COPD), diabetes mellitus, forced expiratory volume during the first second (FEV1 predicted), overall health (ASA >II), mechanical ventilation greater than 2 hours, pleural empyema pre and post-surgery, pleural empyema pre and post-surgery (pleural empyema pre and post-surgery), mechanical ventilation greater than 24 hours, pre-and postoperative pleural empyema, neoadjuvant therapy and postoperative pulmonary infection, location and type of pneumonectomy, type of suture, operative bronchial closure technique (SWEET, separate loose stitch plication of the cartilaginous bronchial portion over the membranous part, and MABIT, bloc closure of the bronchial stump with arterial and/or venous vessels from the pulmonary hilum), bronchial stump coverage, lymph node dissection, residual tumor in bronchial stump, operative time, intraoperative bleeding and the presence of BPF).The present investigation respected ethical principles, and patient anonymity, and since it was a retrospective study, it was not necessary to implement informed consent.The study protocol was approved by the ethics committee of the Universidad de San Martín de Porres (USMP-00456/2022) and the Hospital Nacional Hipólito Unánue (NIT2309-06/2022), Lima, Peru.

Discussion
BPF is a complication that represents 59% of morbidity and 5.4-71% of mortality in patients who underwent pneumonectomy [1][2][3].Our research reported an incidence of 28.1% of BPF, with a higher percentage reported in studies by Mammana (4.5%), Gursoy (10.8%), and La Serna (11%) [6,12,13].We described a history of sequelae of TB infection in 68.4%; however, several authors have reported higher rates exceeding 75% [10,12,14,15].This could be explained by the fact that chronic inflammation due to tuberculosis leads to the deposition of collagen at the hilar level and stiffer anatomical structures; consequently, bronchial tissue is predisposed to impaired healing and a greater risk of BPF [6].BPF was found to be mainly associated with the male gender (62.5%); being similar to that reported by La Serna, who described that 69% of the population was male [13].Mammana et al suggested that a possible explanation could be that the diameter of the male bronchus is larger and when closing the bronchial stump this would generate greater tension; on the other hand, Nachira et al reported that the female sex was an associated factor for post-pneumonectomy BPF (p = 0.03) [3,6].The mean age of patients with BPF reported in our study (38.31 ± 15 years) was somewhat different from that reported by Fuso and Ahmet et al as they described 66.1 ± 7.4 years and the latter would be due to the etiology associated with BPF such as neoplasms, unlike our case, mainly young population and with infectious etiologies [9,10,16].
We identified a statistically significant association (p = 0.015) between smoking and BPF, similar to that reported by Thomas et al and describing that active smoking was a significant predictor of BPF (OR 1.6; 95%CI, 1.34-1.95,p = 0.017) [11]; however, Yazgan and Mazzela et al did not find a clear significant association [8,10].Likewise, alcoholism was significantly associated with BPF (p= 0.008), similar to that described by Thomas et al (OR 2.2; 95%CI, 1.98-2.45,p = 0.017), describing alcohol abuse and dependence as a negative predictor for BPF [11].In Peru, the main indication for pulmonary surgery is for sequelae pulmonary TB, and we report more than 50% incidence of pulmonary TB for both groups; however, without statistically significant association with BPF [14].Mammana et al also reported no significant association in 20% of their cases [6]; but Gursoy and Ahmet et al reported that 33.3% of patients with pulmonary TB were associated with the presence of post-pneumonectomy BPF (p = 0.08) [12,15].Among the predictors of BPF development described by Pforr et al the presence of non-malignant pulmonary disease (p = 0.03) [10] and among which pulmonary sequelae disease (78%) and hydatidosis (34%), were also described by Birdas, and Ahmet et al [15,17].Factors such as malnutrition, underweight, anemia, and hypoalbuminemia have been widely related to BPF and were described by Thomas, Pforr, Haraguchi, and Mazzella et al.However, the latter highlighted that hypoalbuminemia was present in more than 20% of patients affected by BPF (p = 0.02) [8,10,11,18].COPD and low FEV1 predicted postpneumonectomy have been related to the incidence of BPF, because chronic inflammation of the bronchial mucosa in these patients may contribute to impair healing after lung resection [2].Algar et al described that low postoperative FEV1 is a predictive factor for postpneumonectomy BPF (p = 0.012); however, Yazgan and Thomas et al reported that there is no relationship and that more evidence is needed to accept this [2,11,19].Nachira (p < 0.019) and Mammana et al (p < 0.001), reported that prolonged mechanical ventilation causes a higher risk of suture tension and has been associated with the presence of BPF [3,6]; however, we do not report a significant association in this item but we do describe that preoperative empyema is strongly associated with this complication and was described by Haraguchi (p = 0.002), stating that preoperative pulmonary infections contribute significantly to BPF in postoperative pneumonectomy patients (p = 0.002) [19], stating that preoperative pulmonary infections contribute significantly to BPF in postoperative pneumonectomy patients [19].We did not find an association between BPF and postoperative pulmonary infection as did Fuso et al [9].However, Nachira et al concluded that postoperative pulmonary infections were strong predictors of BPF development (p = 0.008) [3].Despite this, it has been described that a mechanism of production of BPF is the infection of the pleural space and the increasing accumulation of infected fluid, resulting in the evacuation of empyema through the bronchial stump and subsequent complication into BPF [20].That is why we report that 100% of the patients who presented BPF had postoperative pleural empyema (p = 0.000), similar to what was described by Nachira et al, who reported that postoperative empyema increases the risk of having BPF by 8 times (p < 0.001) [3] and also Somocurcio et al concluded that postoperative complications in patients who underwent surgery for multidrug-resistant pulmonary TB were postoperative empyema followed by BPF [13].The greater frequency of BPF in right pneumonectomy is due to its blood vascularization which is nourished by one artery, compared to the left bronchus which is perfused by two bronchial arteries; likewise, another protective mechanism would be that in left pneumonectomy, the bronchial stump is covered by the aorta and surrounding tissues [2,18].All the above described has been reflected in publications by Mammana (OR 4.08; 95%CI, 3.67-4.67,p = 0.004), Gursoy (OR 5.90; 95%CI, 4.97-6.32,p = 0.001) and Mazzella et al (OR 2.53; 95%CI, 2.11-3.35,p = 0.023), concluding that right pneumonectomy is an associated factor for the development of BPF [6,8,12].One of the possible causes of BPF is the use of inadequate suture material.Non-absorbable thread (stainless steel, silk, polypropylene) gives better results, the other types are eliminated with the risk of BPF [21].In the present study; only suture thread was used for bronchial stump closure, no mechanical sutures were used.The most frequent thread used in patients with BPF was polypropylene (56.3%) but no significant association was found.Gursoy et al reported that of the cohort of patients with BPF, automatic suture (74.5%) and manual suture (polypropylene, 25.5%) were used but without significant differences [12]; however, Mammana et al described that the use of vicryl was one of the factors that showed a significant positive trend over time towards BPF (p < 0.001) [6].Ahmet et al reported that of the mechanical and manual suture group (polypropylene), 5% and 1.5% respectively, developed BPF, and it was concluded that mechanical suture was associated with BPF (p = 0.04) [15].The bronchus closure technique with continuous stitches (Mabit Technique) was performed in intraoperative complications such as bleeding, which required quick control of the pulmonary hilum, in sequelae lung, chronic empyema, and in case of previous pulmonary resection, where it was very difficult to separate vascular structures and bronchus.With this technique, 75% BPF was described, in contrast to the sweet technique with 32.5%.This factor was evaluated because, with continuous suturing, one of the basic principles of bronchial stump closure would be affected, that of conserving the blood supply and thus affecting the irrigation of the bronchial stump [1].In the literature, manual and mechanical closure techniques are compared, and their use in bronchial stump closure is still controversial [3,6,9].
Lymph node dissection in pneumonectomy, especially in right pneumonectomy, has been associated with greater devascularization of the bronchial stump, causing greater vascular damage of the nutrient vessel in surgery and being a risk factor for the presence of BPF [2,18], we did not report a significant association due to the few cases; neither was it described in other studies such as those of Pforr, and Thomas et al [10,11].Mammana et al reported that residual tumor in the bronchial stump is a risk factor for post-pneumonectomy BPF (p = 0.018); however, we did not find any association in the sample studied [6].La Serna described that the distorted anatomy of patients with sequelae of pulmonary tuberculosis would cause greater intraoperative complications since pulmonary dissections are difficult and increase the operative time [13].We described a longer operative time in patients who underwent pneumonectomy and BPF than those who did not present this complication, being a significant and related finding (p = 0.032); likewise, Thomas et al also concluded that increased operative time is an associated predictor factor for BPF [11].
Among the limitations of our study was the small sample size of our population; however, despite being one of the leading institutions in the surgical management of TB, we have a small caseload.We also believe that a longerterm follow-up study would be ideal for this group of patients with BPF who underwent pneumonectomy In conclusion, we have found a series of factors associated with the presence of BPF in patients who underwent pneumonectomy, many of them related to pulmonary tuberculosis infection, which is an important predictor of the associated complications; however, we believe that further multicenter studies with a larger number of patients are needed to identify more factors to be considered during the surgical management of this terrible complication such as BPF.

Figure 1 .
Figure 1.Thoracic radiographs showing various presentations of BPF, including postoperative empyema and collection (A-D), thoracic CT scans show BPF with the presence of postoperative collections and others with thoracic windows (E-H).

Table 1 .
Baseline characteristics of patients who underwent pneumonectomy with and without BF.

Table 2 .
Clinical-surgical characteristics of patients who underwent pneumonectomy with and without BF.