Analysis of risk factors for the development of a post-bronchoscopy respiratory infection in lung cancer patients

Background: The development of pneumonia following bronchoscopy is very important as a post-bronchoscopy complication. Most patients with post-bronchoscopy respiratory infections show typical pneumonia, and lung abscesses are rare. However, bronchoscopic techniques have advanced, and recently, we have observed patients with lung abscess after bronchoscopy. Therefore, the risk factors might vary from those in past reports. This study aims to identify the incidence of and risk factors for post-bronchoscopy respiratory infections. Methods: We retrospectively studied adult patients diagnosed with lung cancer by bronchoscopy at Fukujuji Hospital between January 2017 and June 2019. Patients in the infection and noninfection groups were compared. The incidence of lung abscess was compared between recent periods and 2013, when endobronchial ultrasonography with a guide sheath (EBUS-GS) was not yet used in our hospital. Results: We reviewed 327 patients, including 20 patients (6.1%) in the infection group. The risk factors for infection were necrosis and/or a cavity in the tumor (p<0.001), large tumor diameter ( ≥ 30 mm) (p=0.003), and low serum albumin (<4.0 g/dL) (p=0.012). We developed a predictive score that included these risk factors, and the area under the curve of the score was 0.737 (95% Cl: 0.610-0.864). Conversely, no signicant differences in age, current smoking status, or abnormal bronchoscopic ndings were observed, even though these factors were reported as risk factors in past reports. Other risk factors for infection were a high white blood cell count (p=0.007), high C-reactive protein level (p=0.014), and expression of programmed death-ligand 1 expression in the tumor cells (p=0.033). In total, 12 patients had lung abscesses (3.7%), which represents a higher incidence than that in 2013 (0.8%). Other types of infection were post-obstructive pneumonia in four patients (1.2%) and typical pneumonia in four patients (1.2%). Conclusions: The risk factors for developing post-bronchoscopy respiratory infection in our study varied from those in past reports, possibly because of the advancement of bronchoscopic techniques such as EBUS-GS. We retrospectively studied patients who were diagnosed with lung cancer by bronchoscopic examination at the Respiratory Disease Center of Fukujuji Hospital from January 2017 to June 2019. Patients with post-bronchoscopy respiratory infections were compared to patients who did not develop an infection after bronchoscopy. We selected adult patients (age ≥ 18 years) with lung cancer who were diagnosed by a bronchoscopic examination. The main outcome of our study was the diagnosis of post-bronchoscopy respiratory infection. Data were collected about the symptoms, laboratory data, radiological ndings, bronchoscopic ndings, histopathologic ndings, and other relevant ndings. Patients who were diagnosed by EBUS-TBNA without transbronchial lung biopsy (TBLB) and those who could not be diagnosed with lung cancer by bronchoscopic examination were excluded. The types of post-bronchoscopy respiratory infection in 2017-2019 were compared to those in 2013, when EBUS-GS was not yet used in our hospital. The study was approved by the Institutional Review Board of Fukujuji Hospital. Patient consent was not required. The decisions made by this board are based on and in accordance with the Declaration of Helsinki.


Introduction
A exible beroptic bronchoscope can safely perform assessments, aid in diagnosis, and provide treatment to patients with respiratory diseases [1], and complications associated with bronchoscopy occur in 0.2-5.6% of patients with a mortality rate of 0.004-0.02% [2][3][4][5]. The development of pneumonia following bronchoscopy is very important in patients suffering from lung cancer since pneumonia causes a delay in treating the malignancy [6]. Most of the patients with post-bronchoscopy respiratory infections show a new in ltrate or consolidation on chest radiographic ndings similar to the signs of typical pneumonia [3,6,7]. Additionally, lung abscesses rarely develop in post-bronchoscopy respiratory infections and occurred in 0.22-1.06% of patients in a past report [7,8]. However, recently, we have observed patients with lung abscesses after bronchoscopy. Bronchoscopic techniques have been further developed to include methods such as endobronchial ultrasonography (EBUS), EBUS with a guide sheath (EBUS-GS) and EBUS-guided transbronchial needle aspiration (EBUS-TBNA); therefore, an updated review should be considered [8]. Some past reports have demonstrated the risk factors for post-bronchoscopy respiratory infection, such as an elderly age, current smoking status, and central location of the tumor [6,9]. However, the risk factors for post-bronchoscopy respiratory infection might differ from those in past reports. This study aims to identify the incidence of and risk factors for post-bronchoscopy respiratory infections.

Study Design and Setting
We retrospectively studied patients who were diagnosed with lung cancer by bronchoscopic examination at the Respiratory Disease Center of Fukujuji Hospital from January 2017 to June 2019. Patients with post-bronchoscopy respiratory infections were compared to patients who did not develop an infection after bronchoscopy. We selected adult patients (age≥18 years) with lung cancer who were diagnosed by a bronchoscopic examination. The main outcome of our study was the diagnosis of post-bronchoscopy respiratory infection. Data were collected about the symptoms, laboratory data, radiological ndings, bronchoscopic ndings, histopathologic ndings, and other relevant ndings. Patients who were diagnosed by EBUS-TBNA without transbronchial lung biopsy (TBLB) and those who could not be diagnosed with lung cancer by bronchoscopic examination were excluded. The types of postbronchoscopy respiratory infection in 2017-2019 were compared to those in 2013, when EBUS-GS was not yet used in our hospital. The study was approved by the Institutional Review Board of Fukujuji Hospital. Patient consent was not required. The decisions made by this board are based on and in accordance with the Declaration of Helsinki.

De nitions
Post-bronchoscopy respiratory infection was de ned as a respiratory infection that developed a new or progressive in ltrate on chest radiographs or computed tomography (CT) scans with an elevated white blood cell count (WBC) and/or C-reactive protein (CRP) during the period from the bronchoscopic examination to starting treatment for lung cancer or within a month. The types of post-bronchoscopy infection are classi ed as lung abscesses in the tumor, post-obstructive pneumonia, and typical pneumonia (Fig. 1). Lung abscesses were de ned as follows: (1) primary shadow developing enlargement, and (2) cavity formation, low-density area in a primary shadow appearing or enlargement in post-bronchoscopy respiratory infection. Post-obstructive pneumonia was defined as an infection of the lung parenchyma distal to a bronchial obstruction [10]. Typical pneumonia was de ned as a new in ltrate or consolidation around the target lesion during bronchoscopic examination. Necrosis in the tumor was classi ed as a low-density area in the tumor with a maximum of 30 Houns eld units nonenhanced CT scans [11,12]. Lung cancer staging referred to the 8 th edition of the TNM classi cation [13]. Abnormal bronchoscopic ndings were identi ed with intrabronchial tumor observations via bronchoscopy.

Statistical Methods
All data were analyzed and processed using EZR, version 1.35 [14]. Student's t test, Mann-Whitney U test, and Fisher's exact test were used to compare groups. Sensitivity, speci city, and odds ratios were calculated. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were calculated for each predictor model. ROC curves determined the cutoff value, and the AUC accurately measured the prediction model. The level of statistical significance was set at p=0.05 (2-tailed).

Results
A total of 344 patients were diagnosed with lung cancer by bronchoscopic examinations at our hospital.
Seventeen patients who were diagnosed with EBUS-TBNA without TBLB were excluded, so we reviewed 327 patients in our study. Among all patients, 20 patients (6.1%) developed post-bronchoscopy respiratory infections (the infection group), and 307 patients were in the noninfection group ( Table 1) .003) and more necrosis and/or cavities in the tumors (n=8 (40.0%) vs. n=29 (9.4%), p<0.001) than those in the noninfection group. The incidence of using a guide sheath (p=0.588), abnormal bronchoscopic ndings (p=0.175), and antibiotic prophylaxis (p=0.429) were not signi cantly different between the two groups.
The odds ratios of a tumor diameter of 30 mm or more, necrosis and/or a cavity in the tumor, WBC of 6500 cells/µL or more, CRP of 0.55 mg/dL or more, and serum albumin less than 4.0 g/dL are shown in Table 2. Necrosis and/or a cavity in the tumor showed the highest odds ratio at 6.33 (95% con dence level (Cl): 2.07-18.5, p<0.001). We developed a predictive score for the development of post-bronchoscopy respiratory infections that included necrosis and/or cavities in the tumors, tumor diameter ≥30 mm, and serum albumin <4.0 g/dL (Fig. 2). Each variable was assigned a value of 1 point, thus totaling 3 points.
The ROC curve of the score demonstrated a high AUC of 0.737 (95% Cl: 0.610-0.864). A score of 2 or more was regarded as the cutoff value, and the sensitivity, speci city, and odds ratio were 70.0%, 69.2%, and 5.21 (95% Cl: 1.81-17.1), respectively. Table 3 shows the characteristics of the infection group. Of all the patients, 12 patients had lung abscesses (3.7%), 4 patients had post-obstructive pneumonia (1.2%), and 4 patients had typical pneumonia (1.2%). The median duration from the bronchoscopic examination to treatments for postbronchoscopy respiratory infection was 12.5 days (interquartile range (IQR): 7.0-18.3). Eight out of the 12 patients with lung abscesses showed necrosis and/or a cavity in the tumor; however, no patients with post-obstructive pneumonia and typical pneumonia showed those ndings. All patients with postobstructive pneumonia showed abnormal bronchoscopic ndings and underwent biopsy without EBUS-GS. Conversely, many patients with lung abscesses and typical pneumonia underwent EBUS-GS and did not show abnormal bronchoscopic ndings. One patient with a lung abscess required surgical treatment, and another patient died. In 2013, when EBUS-GS was not yet used in our hospital, 131 patients were diagnosed with lung cancer by bronchoscopic examination, and seven of these patients had postbronchoscopy respiratory infection (5.3%), including only one patient with a lung abscess (0.8%) and six patients with other types of pneumonia (4.6%). The incidence of developing lung abscesses after bronchoscopy has increased in recent years (Fig. 3). In addition, the number of biopsies performed in 2019 was higher than that in 2017 (p=0.018).

Discussion
This study identi ed the characteristics and risk factors for developing a post-bronchoscopy respiratory infection. A total of 6.1% of all patients developed post-bronchoscopy respiratory infections, and the risk factors were high WBC counts or CRP levels, low serum albumin levels, necrosis and/or a cavity in the tumor, and large tumor diameter. However, these risk factors differ from those in past reports [6,9]. We developed a predictive score that included necrosis and/or a cavity in the tumor, tumor diameter ≥30 mm, and serum albumin <4.0 g/dL. A score of a least 2 was a good predictor for developing a postbronchoscopy respiratory infection. In total, 60% of the patients with post-bronchoscopy respiratory infections showed lung abscesses. The incidence of developing lung abscesses after bronchoscopy has increased in recent years. We believe that these different risk factors are caused by the advancement of bronchoscopic techniques.
A past report from 2012 to 2013 suggested risk factors for post-bronchoscopy respiratory infection, such as an elderly age, current smoking status, and central location of the tumor [6]. However, these factors did not differ signi cantly between the infection group and the noninfection group in our study. These differences might be caused by the increasing incidence of lung abscesses after bronchoscopy since the bronchoscopic techniques have been changed. Generally, lung abscesses are rare in post-bronchoscopy respiratory infections and occur in 0.22-1.06% of the patients [7,8]. Similar to the percentage in past reports, 0.8% of the patients in our study had lung abscesses after bronchoscopy in 2013, when EBUS-GS was not yet used in our hospital. However, the incidence of lung abscesses after bronchoscopy increased to as high as 5.3% in 2019. Recently, the EBUS-GS technique has been used and has signi cantly increased the diagnosis rate of peripheral pulmonary lesions [15,16]. Moreover, the number of TBLB procedures performed is higher because the specimens from EBUS-GS are smaller in size than those from routine TBLB [15]. The British Thoracic Society guidelines for bronchoscopy recommend taking at least ve biopsy samples [1], and an additional five bronchial forceps biopsies should be considered for phenotyping and genotyping exams [17]. In our hospital, the number of biopsy by bronchoscopy procedures performed increased from October 2018 because of phenotyping and genotyping, and the number of biopsies performed by bronchoscopy was signi cantly greater in 2019 than in 2017. Therefore, we thought that post-bronchoscopy lung abscesses can easily occur due to the type of approach, infection of the central tumor using EBUS-GS and increases in the number biopsies performed. Other risk factors for post-bronchoscopy respiratory infection were low serum albumin before the bronchoscopic examination and PD-L1 expression in tumor cells. Low serum albumin might indicate that the immune systems of the patients in the infection group were worse than those of the patients in the noninfection group. Generally, the nutritional status as assessed by the serum albumin level relates to the immune response, and malnutrition is a common cause of immunode ciency [19,20]. In addition, PD-L1 expression in tumor cells is related to the immune system. The combination of programmed cell death 1 (PD-1) and PD-L1 leads to tumor immune evasion through the suppression of T cells [21], and the role of PD-1 is to regulate infections due to the limited function of macrophages as well as the T cellindependent B cell response [22]. A case report of a patient with a lung abscess after bronchoscopy showing high levels of PD-L1 was previously published [23]. Thus, immune disorder is an important risk factor for the development of post-bronchoscopy respiratory infections.
The British Thoracic Society guidelines for bronchoscopy do not recommend antibiotic prophylaxis before bronchoscopy for endocarditis, fever, or pneumonia [1]. In our study, antibiotic prophylaxis did not effectively prevent post-bronchoscopy respiratory infections. However, very few patients received antibiotic prophylaxis, so whether antibiotic prophylaxis was useful cannot be determined. Kanazawa H et al. reported that azithromycin was effective as prophylactic antibiotherapy in preventing infections post bronchoscopy compared to the no-treatment group (3.0% vs 14.8%, p=0.02) [7]. Generally, the common causative organisms of lung infections after bronchoscopy are oro-or nasopharyngeal microbes, such as Streptococcus, Staphylococcus, Moraxella, Neisseria, and anaerobic bacterial species [5]. Therefore, prophylactic antibiotherapy might be better adapted to anaerobic bacteria. Many patients with post-bronchoscopy respiratory infections canceled their planned oncologic treatments [6], and some post-bronchoscopy respiratory infections could not improve with antibiotics and required surgical invention [8]. Therefore, prophylactic antibiotherapy might be necessary for high-risk patients. A prospective study of effective antibiotic prophylaxis is required. This investigation had several limitations. The study was conducted retrospectively in a single center. The target disease was lung cancer; therefore, the incidence in this study does not represent the overall incidence of bronchoscopy. Some medical data were not recorded. Some patients received prophylaxis antibiotherapy. We reviewed all types of infections together. Some patients did not have PD-L1 tumor cell expression data because the analysis was not requested or sample size from TBLB was too small.

Conclusion
The study demonstrated the characteristics of post-bronchoscopy respiratory infections, which was mainly lung abscesses. The risk factors were high in ammation, low serum albumin, necrosis and/or a cavity in the tumor, and large tumor diameter. In addition, we developed a good predictive score, which included necrosis and/or a cavity in the tumor, tumor diameter ≥30 mm, and serum albumin <4.0 g/dL.
We believe that these risk factors vary from those in past reports because of the advancement of bronchoscopic techniques. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.   The receiver operating characteristic curve of the score for predicting post-bronchoscopy respiratory infections The area under the curve was 0.737 (95% Cl: 0.610-0.864). The predictive accuracy for postbronchoscopy respiratory infection at a cutoff value of 2 points or more included a sensitivity of 70%, speci city of 69.2%, and odds ratio of 5.21 (95% Cl: 1.81-17.1). Cl: con dence interval