Combined bronchoalveolar lavage and transbronchial lung biopsy: safety and yield in ventilated patients

Study population

All mechanically ventilated patients with pulmonary infiltrate(s) who underwent combined BAL/TBLB between June 1996 and the end of February 2000 were evaluated. The authors proceeded to use BAL/TBLB when a diagnosis could not be obtained by usual noninvasive methods, including thoracic imaging and cultures (e.g. sputum, endotracheal aspirate), and when lung tissue samples could optimise therapy. Exclusion criteria were: 1) severe coagulopathy (platelet count <50,000 thrombocytes·mm⁻³ despite platelet transfusion, fibrinogen ≤150 mg·dL, activated partial thromboplastin time (aPTT) ≥50 s, or prothrombin time ≤50%); and 2) haemodynamic instability (mean arterial pressure (MAP) <60 mmHg despite a continuous infusion of norepinephrine).

Procedure

Combined BAL/TBLB was performed at the bedside by the same physician (P.A. Bulpa). Before the procedure, all patients were managed as previously described 7. The bronchoscope (BF XT 20 or BF P 40; Olympus Optical Co., Tokyo, Japan) was introduced into the endotracheal tube through a special adaptor (Double Swivel Connector; Medisize Co., Hillegom, the Netherlands) to reduce gas leakage. Loss of tidal volume was compensated by increasing the peak pressure up to 60 cm H₂O. The bronchoscope was wedged in the affected lung segment identified by thoracic imaging. BAL was performed as reported by Marquette et al. 11, the first 20 mL aliquot being discarded to avoid bacterial contamination. TBLB immediately followed BAL. The forceps (FB 35 C or FB-19C‐1; Olympus Optical Co.) were advanced into the same lung segment as for BAL. The number of biopsies retrieved was determined by the operator. TBLB was executed, under fluoroscopic guidance, as described by O'Brien et al. 7, as was surveillance for pneumothorax. A pneumothorax occupying >20% of the hemithorax was drained, otherwise it was monitored by repeated chest radiographs.

Laboratory processing

Whenever possible, one biopsy specimen was placed in sterile water for culture. BAL and TBLB samples were immediately transported to the microbiology laboratory. BAL specimens were processed for bacterial, fungal, viral and mycobacterial culture as the authors previously described 12. TBLB samples were processed for bacterial and fungal culture. Total cell count was performed as reported elsewhere 13. The other biopsies were placed in Bouin's fluid. All biopsy and autopsy specimens were prepared following standard practices 12.

Data collection

Data were collected from the medical records and from the bronchoscopy database, a prospectively maintained record of all fibreoscopies conducted in the department. The following variables were evaluated: age, sex, intensive care unit (ICU) admitting diagnosis, reason for the procedure, patient category, pre-BAL/TBLB physiological parameters (blood gases, haemodynamic status), coagulation status and use of prophylactic anticoagulants, ventilator settings and subsequent airway pressures (peak and plateau).

During the procedure, haemodynamic parameters and oxygen saturation were recorded every minute (Monitoring HP OmniCare M1165/66A; Hewlett Packard Company, Andover, MA, USA) in 32 patients. Blood gases at fractional concentration of oxygen in inspired gas (F_I,O2) 1 were obtained immediately before the introduction and at the removal of the fibrescope in 33 patients.

The following post BAL/TBLB data were analysed: 1) physiological parameters and ventilator settings (after 2, 6, 24 and 48 h); 2) procedure complications; 3) cultures (from BAL in all patients and from TBLB in 33 patients); and 4) modifications in patient management as a result of the BAL/TBLB. Histological correlations between lung tissue obtained with TBLB or open-lung biopsy and/or autopsy were performed when available.

The following were considered procedure-related complications: significant bleeding (≥30 mL), pneumothorax, new pulmonary infection (according to the definitions used by O'Brien et al. 7), death, increases in body temperature of 1°C and >38°C within 24 h, severe hypotension (MAP ≤60 mmHg), bradycardia (heart rate (HR) ≤50 beats·min⁻¹), tachycardia (HR≥150 beats·min⁻¹) or an arterial oxygen saturation of <90% during the technique.

The patient's physician was free to modify treatment according to the BAL/TBLB results. The addition, withdrawal and lack of introduction (when a pre-procedure decision was cancelled due to the BAL/TBLB results) of a therapy were considered therapeutic modifications. Management changes were attributed to the BAL/TBLB procedure when the lung tissue biopsies yielded clinical information unobtainable without histological specimens (e.g. rejection in transplanted lung) and the results of the cultures provided new information that altered antibiotic therapy. The procedure was judged contributive when the results confirmed the suspected diagnosis even though treatment was not modified. Two authors (P.A. Bulpa and L. Mertens) reviewed the patients' data.

Statistical analysis

Numerical variables are expressed as mean±sd or median with interquartile ranges (IQR). A one-way analysis of variance for repeated measures was used to compare the various parameters at different time points and to detect any significant changes from baseline values. Blood gases before and immediately after bronchoscopy were compared by a paired t‐test. Correlation of particular complications with categorical or continuous variables was assessed by Chi-squared or Wilcoxon rank sum tests, respectively. All statistical tests were two-tailed. A p‐value of <0.05 was considered as statistically significant.

Patient and procedure characteristics

During the study period, 41 patients underwent combined BAL/TBLB, among whom 38 presented sufficient data to allow analysis. To facilitate the interpretation of results, patients were divided into two groups: immunocompetent (n=22, group 1) and immunocompromised (n=16, group 2), with the latter including those receiving long-term treatment with >10 mg equivalent prednisone. In addition, within group 1, patients with and without late-phase acute respiratory distress syndrome (ARDS, according to the definitions of the consensus conference criteria 14) were analysed separately: groups 1a and 1b respectively. All patients undergoing BAL/TBLB had radiographical evidence of pulmonary infiltrate(s), which occurred before or during invasive mechanical ventilation. BAL/TBLB was performed at a median time of 10 days (range 0–35) after tracheal intubation.

Patient features are summarised in table 1⇓. On admission, the Acute Physiology, Age, and Chronic Health Evaluation (APACHE) II score was 25±9 and the Simplified Acute Physiology Score (SAPS) II was 48±16. Because diagnosis was essential for appropriate therapy to be instituted, the procedure was performed in one patient with an aPTT of 58 s (limited bleeding (35 mL) was observed).

According to the protocol, to allow a good return of BAL fluid, positive end-expiratory pressure (PEEP) was maintained at ≤5 cmH₂O, except in one case where oxygen desaturation was observed when PEEP was <10 cmH₂O. Before the technique, F_I,O2 ranged from 0.26–1 (median 0.49). Twelve patients required an F_I,O2 of ≥0.6, seven of whom were receiving nitric oxide.

In 33 patients, blood gases at F_I,O2 100% were analysed immediately before and after bronchoscopy. Arterial oxygen tension (P_a,O2) decreased from 262±121 to 244±120 mmHg (p=0.09). Carbon dioxide arterial tension (P_a,CO2) increased significantly from 40±7 to 51±9 mmHg (p<0.001). As a consequence, pH decreased from 7.41±0.06 to 7.32±0.07 (p<0.001). Interestingly, P_a,CO2 and pH returned to their pre-procedure levels within 2 h in the 23 patients in whom blood gases were obtainable after 2, 6, 24 and 48 h.

Respiratory frequency, tidal and minute ventilations, and peak and plateau pressures were stable between the pre-procedure period and 2 h after the bronchoscopy. As per protocol, F_I,O2 was increased to 1 before and during the procedure, and then adjusted to maintain an oxygen saturation of ≥90%; it took 6 h for levels to return to baseline values. The median duration of the procedure was 23 min (range 11–53).

BAL and TBLB results and contribution to therapy

A median of five (range 1–8; IQR 5–6) biopsies was performed per patient. With experience, the number of biopsies taken increased to a minimum of five. On two occasions at the beginning of the study, just one biopsy was taken. Based on the results of histopathology and/or TBLB culture, aetiological diagnosis was obtained in 24 of 38 cases (63%, table 2⇓), and, as the only positive test, contributed to specific diagnosis in 17 patients, i.e. nine out of 11, three out of 11 and five out of 16 in groups 1a, 1b, and 2, respectively. In noncontributive TBLB, histological diagnosis included diffuse alveolar damage, bronchial or pleural tissue, normal parenchyma and cryptogenic organising pneumonia. The number of TBLB specimens retrieved in patients with contributive histopathological diagnoses was not significantly different from the number obtained in patients without specific diagnoses.

The average amount of BAL fluid retrieved from the patients was 55±24 mL (range 15–95 mL). The median number of cells in the BAL fluid was 265 mm⁻³ (range 15–1,000 mm⁻³). There was no correlation between the differential cell count in the BAL fluid and a specific diagnosis. However, BAL yielded 11 (29%) specific diagnoses and, as the only positive test, contributed to specific diagnosis in four cases (table 2⇓).The therapeutic contributions of BAL/TBLB are listed in table 2⇓. In groups 1a, 1b and 2, diagnosis was obtained in 11 of 11, seven of 11 and 10 of 16 patients, respectively, and therapy changed in 11 of 11, six of 11 and seven of 16.

The performance of the procedure (combined BAL/TBLB or each separately), calculated as the ratio of the number of patients in whom the procedure contributed to therapy to the total number of patients studied, was 74% (28 out of 38) for BAL/TBLB, 29% (11 out of 38) for BAL, and 63% (24 out of 38) for TBLB (fig. 1⇓).

Three patients underwent OLB because the bronchoscopic procedure was inconclusive and autopsy examination was obtained in 10 cases. There was a good correlation between BAL/TBLB and tissue obtained at necropsy in four patients (confirmed diagnoses: invasive pulmonary aspergillosis (IPA) in two patients and emphysematous lesions in two). In three patients, no correlation could be found because BAL/TBLB yielded no diagnostic information (final diagnoses: lipoid pneumonia, usual interstitial pneumonia and IPA associated with some foci of bronchiolitis obliterans organising pneumonia). In the remaining patients, therapy was modified and subsequent treatment could have altered the autopsy findings since the delay between the two procedures exceeded 10 days.

Complications

No fatalities were attributable to the procedure. Some complications were procedure-related (table 3⇓). Pneumothorax was always discovered during fluoroscopic guidance. The air leak was transient (median 1 day, range 1–12) and no persistent bronchopleural fistulas were observed. There was no significant correlation between the number of TBLB samples, the level of PEEP or peak and plateau pressures and the appearance of a pneumothorax (p>0.7).

Median platelet counts were significantly lower in patients with haemorrhage (72,000; IQR 35,250–159,000 thrombocytes·mm⁻³) than in patients without bleeding (187,500; IQR 100,500–301,750) (p=0.023). Although 27 patients received low molecular weight heparin for venous thromboembolism prophylaxis, there was no association between its use and the occurrence of haemorrhage.

MAP always remained >60 mmHg, except in two patients in whom transient decreases were promptly corrected. No patient developed bradycardia or tachycardia during bronchoscopy.

During the procedure, three patients experienced temporary oxygen desaturation requiring the withdrawal of the fibrescope. No new pulmonary infection was procedure-related. Thirteen transient episodes of fever occurred after the procedure.