Pulmonary complications in acute myelogenous leukemia in post allogenic hematopoietic stem cell transplant period and yield of bronchoalveolar lavage and biopsy- single center experience

Abstract Background Pulmonary complications occur in 40-80% of acute myeloid leukemia (AML) and hemopoietic stem cell transplant (HSCT) patients. Both infectious and noninfectious causes are associated with high mortality. Due to variable diagnostic yield and procedure related complications, the role of bronchoalveolar lavage (BAL) and transbronchial biopsy remains unclear. We evaluated the infectious and noninfectious diagnostic yield of BAL in this population. Methods Retrospective review of AML patients (>18 years) at Cleveland Clinic from 2012-2024 who underwent BAL for pulmonary complications within 3 years post HSCT. Detailed data was collected in redcap including demographics, HSCT, graft versus host diseases (GvHD), symptoms, BAL and biopsy details, anti-infective prophylaxis, empiric antimicrobial therapy, microbiological diagnosis, treatment, complications and outcomes. Diagnostic yield was defined as confirmed infectious or non-infectious diagnosis on BAL and/or biopsy when noninvasive testing was negative. Results From 2012-2024, a total of 435 AML pts underwent HSCT, 66 pts underwent BAL within 3 years post HSCT. Out of 66, 14 pts were also had transbronchial biopsy performed at the time of BAL. Males were 62%, females were 38% and median age was 57 (44-66) years. BAL was performed at a median of 191 (79-399) days post-HSCT. Type of HSCT included matched unrelated (61%), haploidentical (17%), and others. BAL was performed in 30% within 100 days and 50% within 6 months of HSCT. Concurrent GvHD was diagnosed in 51% of the pts (63% acute and 47% chronic), GI involvement in 13% followed by skin in 12% of the cases. Relapsed AML diagnosed in 18% of the pts at the time of BAL. The procedure was conducted in an inpatient setting in 86% of cases. At the time of BAL, 69% pts were hypoxic, 59% were admitted in ICU and 31% were required to have invasive mechanical ventilation and 40% developed septic shock. Almost half of the pts (48%) developed symptoms >2 weeks before BAL. Chest x-ray showed pulmonary infiltrates in 66% of the pts but most common CT scan finding was ground-glass opacities (60%), followed by consolidations (32%), multiple nodules (30%) and cavitary lesion (12%). Empiric antibiotics initiated 81% of the patients prior to BAL for a median of 8.5 days (3.0, 14.0) and 54% also received prophylactic/ empiric antifungal for median of 29 days. Piperacillin-tazobactam was the most used empiric antibiotic therapy (36%) and then meropenem (22%). Thirty percent of the patients were neutropenic with absolute neutrophilic count <1000 k/uL and 56% were thrombocytopenic with platelet count <50 k/uL at the time of BAL. Platelet transfusion was required in 22% of the pts prior to procedure. The most common infectious diagnosis was viral pneumonia (including adenovirus, covid, RSV, parainfluenza, CMV) followed by bacterial pneumonia (Pseudomonas aeruginosa, Stenotrophomonas maltophilia, Klebsiella pneumoniae, polymicrobial). Although less common, a number of fungal (aspergillus species, Mucorales species, pneumocystis jirovecii) and mycobacterial (mTB) infections were made. Noninfectious diagnosis included diffuse alveolar hemorrhage, drug pneumonitis, organizing pneumonia, mucus plug and engraftment syndrome. Overall diagnostic yield of BAL and biopsy was 32%. The overall yield of BAL in ICU cases was 28%, septic cases 22% and neutropenic cases 31%. Microbiological yield was low at 12%. BAL galactomannan was positive in 11 patients, but only 2 out of 11 cases have BAL culture confirmed Aspergillus species and 1 of them also have tissue histopathology positive for septate hyphae. BAL led to antimicrobial therapy changes in 33% of the pts: 18% it was stopped, 9% de-escalated and 6% escalated. No major procedure related complications occurred; only 2 pts developed mild pneumothorax. Conclusion Diagnostic yield of BAL was ~ 32% in AML, post-HSCT cases with pulmonary complications. Although microbiological yield was low at12%, BAL resulted in the diagnosis of multiple opportunistic infections including fungal, CMV and mTB. Prolonged use of preceding empiric antimicrobials may have contributed to low yield. Despite low diagnostic yield, BAL did impact changes in antimicrobial modification in 1/3 of the cases, supporting an early role prior to extensive antimicrobial exposure. The positive predictive value of BAL Aspergillus galactomannan was 18% so the results should be analyzed with caution and high clinical suspicion.