Association of time-to-treatment with outcomes of Pneumocystis pneumonia with respiratory failure in HIV-negative patients

All consecutive patients older than 20 years who were admitted to the medical ICU through the emergency department (ED) for acute respiratory failure were screened for inclusion (Fig. 1). Some of the clinical data from patients who were enrolled between 2005 and 2011 were also included in the previous study [18]. Patients were included if they had a microbiologically confirmed diagnosis of PCP and required respiratory support with mechanical ventilation including non-invasive ventilation or high-flow nasal cannula for respiratory failure. Patients were excluded if they had a positive HIV antibody test. Finally, patients who had neither respiratory symptom nor abnormal finding on chest radiographs at initial presentation were excluded from the study. In patients who had multiple admissions for acute respiratory failure due to PCP during the study period, only the first ICU admission was evaluated.

The diagnosis of PCP was based on clinical symptoms and the presence of a new infiltration on chest radiograph, along with morphological identification of the organism in bronchoalveolar lavage (BAL) fluid or lung tissue obtained by transbronchial lung biopsy (TBLB). The BAL fluid samples were stained using Gram and Ziehl-Neelsen methods and then cultured for bacteria, mycobacteria, and fungi. Multiplex nested polymerase chain reaction (PCR) assays were used to detect the following: influenza viruses A and B; parainfluenza viruses 1, 2, and 3; respiratory syncytial virus; and adenovirus [19]. Quantitative real-time PCR was used to measure the cytomegalovirus (CMV) deoxyribonucleic acid (DNA) in the BAL fluid [20]. Microbiological identification of P.jirovecii was confirmed by documenting the organism with Wright Giemsa or Gram-Weigert stain, or the cyst with the Gomori methenamine silver or calcofluor white stain [2].

The following data were extracted from the medical records: general demographic information, underlying diseases, medications used during the previous month, PCP prophylaxis, need for renal replacement therapy, need for vasopressor support, anti-PCP medication, and newly developed organ failure during the ICU stay. We also collected the following laboratory data from the medical records: the white blood cell count, absolute neutrophil count, absolute lymphocyte count, albumin, C-reactive protein, arterial partial pressure of oxygen (PaO₂)/fraction of inspired oxygen (FiO₂) ratio (PF ratio), and alveolar-arterial oxygen gradient [D(A-a)O₂]. The severity of illness was assessed using the Simplified Acute Physiology Score 3 (SAPS 3) and Sequential Organ Failure Assessment (SOFA) score [21, 22]. Finally, we studied the clinical outcomes, including the length of ICU stay, the length of hospital stay, and ICU and in-hospital mortality.

The time to anti-PCP treatment interval was defined as the number of hours from the time of emergency room arrival to the initial administration of anti-Pneumocystis antibiotics. We classified patients into two groups according to the timing of initial anti-PCP treatment before or after microbiologic diagnosis [23]: early empiric treatment group and definitive treatment group. The doses of corticosteroids used for immunosuppression and the adjunctive therapy for PCP were expressed as the prednisolone-equivalent doses [24]. The adjunctive corticosteroid therapy was defined as that started within 72 h of initiating the specific anti-PCP treatment (consisting of at least 40 mg prednisone twice daily for 5 days regardless of the subsequent tapering schedule), and the use of corticosteroids before the onset of PCP [25]. The failure of anti-PCP treatment was defined as follows: (1) progressive clinical deterioration as demonstrated by the inability to maintain a stable PaO₂ despite an increase in FiO₂; and (2) progressive deterioration of vital signs with an increased FiO₂ requirement despite 7 days of appropriate therapy [11, 26]. Breakthrough PCP infection was defined as that diagnosed in a patient receiving prophylactic agents with known activity against P. jirovecii for at least 7 days prior to the diagnosis [27].

All data are presented as medians and interquartile range (IQR) for continuous variables, and as numbers and percentages for categorical variables. Continuous variables were analyzed using the Mann-Whitney-Wilcoxon U test. Categorical variables were analyzed using the Pearson χ2 test or Fisher exact test. The Mantel–Haentzel trend test was used to examine the trends in the hospital mortality rate across the time quartiles to anti-PCP treatment [28]. To adjust for potential confouding factors in the association between the time to anti-PCP treatment and in-hospital mortality, multiple logistic regression analysis was used. Finally, we compared baseline characteristics and clinical outcomes between early empiric treatment group and definitive treatment group. Probabilities of survival after anti-PCP treatment for each group were estimated by Kaplan-Meier method and compared by the log-rank test. Variables with a P-value less than 0.2 on univariate analyses [29], as well as a priori variables of age and sex were entered into a multiple logistic regression model in which in-hospital mortality was the outcome variable of interest. To reduce the risk of multicollinearity, one closely correlated variable was a candidate for inclusion in the final model. Data are presented as adjusted odds ratios (OR) with 95% confidence intervals (CI). All of the tests were two-sided. P values < 0.05 were considered statistically significant. All analyses were performed using SPSS for Windows (ver. 23.0; IBM Corp., Armonk, NY, USA).

During the study period, a total of 117 patients with PCP were admitted to the ICU for respiratory support. Nineteen of these patients were excluded from the study because they had HIV infections. After excluding patients who were initially admitted to general ward from outpatient care with other cause (n = 29), those who were transfer to our hospital (n = 1), and those with an initial chief complaint that was unrelated to respiratory symptoms (n = 17), 51 patients with acute hypoxemic respiratory failure and a new infiltration on chest radiograph were included.

The patients’ baseline characteristics are summarized in Table 1. The median patient age was 52.0 (IQR 40.5–66.0) years. Thirty-five (68.6%) patients were men. The main underlying conditions associated with the development of PCP included malignancies (n = 28, 54.9%) and history of solid organ transplant (n = 18, 35.3%). All of the patients were immunosuppressed before developing PCP. Twenty-two (43.1%) patients received systemic steroids. The median prednisolone-equivalent dose was 16.1 (6.2–34.7) mg/day in patients who received systemic steroids. Two (3.9%) patients had a history of receiving prophylactic TMP/SMZ.

At the time of ICU admission, all of the patients had hypoxemia requiring mechanical ventilation (n = 43, 84.3%) or high-flow nasal cannula support (n = 8, 15.7%). The median PF ratio was 137.2 (112.8–164.6) mmHg. The D(A-a)O₂ was 39.9 (30.8–52.3) mmHg. Twelve (23.5%) patients required vasopressor support and one (2.0%) needed renal replacement therapy. The median SAPS 3 and SOFA scores on ICU admission were 48.0 (37.0–57.5) and 6.0 (5.0–8.0), respectively.

All of the patients underwent bronchoscopy with BAL ± TBLB within 57.7 (30.2–91.7) hours of the ER visit (Table 2). The diagnosis of PCP was based on microbiological identification of P. jirovecii in the BAL fluid (n = 44, 86.3%) or lung biopsy specimens (n = 14, 27.5%). Seven (13.7%) patients were positive for P. jirovecii in both the BAL fluid and pathology specimens. In addition to P. jirovecii, other pathogens were simultaneously isolated from respiratory specimens at the time of diagnosis. The most common simultaneously isolated pathogen was CMV in 13 patients (25.5%), bacteria in 12 (23.5%), and viruses other than CMV in 7 (13.7%).

All of the patients were treated with appropriate anti-PCP treatment, primarily involving TMP (20 mg/kg/day)/SMZ (100 mg/kg/day). Thirty-one (60.8%) patients were treated with TMP/SMZ empirically prior to confirmation of the microbiological diagnosis. The median time to anti-PCP treatment was 58.0 (28.0–97.8) hours. Baseline characteristics and the hospital mortality rates according to the time to anti-PCP treatment in quartiles are presented in Additional file 1: Table S1 and Fig. 2, respectively. Interestingly, the hospital mortality rates were not associated with increasing quartiles of time to anti-PCP treatment (P = 0.818, test for trend).

All of the patients except one (98.0%) were treated with adjunctive corticosteroids. One patient had an adverse reaction to TMP/SMZ involving an electrolyte imbalance after 7 days of use. After excluding this patient, the initial treatment response was assessed on day 7. Of the 50 patients, 20 (40.0%) did not respond to the initial treatment with TMP/SMZ. These 20 patients were subsequently treated with clindamycin-primaquine (n = 14) or pentamidine (n = 6) as salvage therapy for PCP.

During their ICU stays, organ failure newly developed in 33 (64.7%) patients, which included shock requiring vasopressors in 30 patients and acute kidney injury requiring renal replacement therapy in 13 (Table 3). The ICU mortality was 37.3% with a median stay of 15.0 (6.0–29.0) days. Ultimately, 23 (45.1%) patients died while hospitalized. The median hospital stay was 24.0 (15.5–32.5) days.

The early empiric treatment group contained 31 (60.8%) patients and the definitive treatment group contained 20 (39.2%) patients. There were no significant differences in baseline characteristics, and diagnosis and treatment of PCP, except of the median time to anti-PCP treatment (34.2 h for early empiric treatment group vs. 91.7 h for definitive treatment group, P < 0.001) (Additional file 2: Table S2 and Additional file 3: Table S3). In comparison of clinical outcomes between the two groups, there was no significant difference in mortalities and length of stay (Table 3). In addition, early empiric treatment did not significantly affect overall survival (Fig. 3). Although the probability of survival for early empiric treatment group appears to increase in the first few weeks after anti-PCP treatment, this difference is not statistically significant (P = 0.8673, log-rank test).

Univariate comparisons of the clinical characteristics of hospital survivors and non-survivors are presented in Table 4. Compared to survivors, non-survivors were likely to be older (48.0 [34.5–59.0] vs. 64.0 [50.5–69.0], P = 0.007), require more mechanical ventilation (67.9% vs. 100%, P = 0.009), SAPS 3 score increase (43.0 [30.5–54.5] vs. 56.0 [46.5–58.0], P = 0.030), more CMV co-infection (10.7% vs. 43.5%, P = 0.011), and failed to initial anti-PCP treatment (21.4% vs. 69.6%, P = 0.002). After adjusting for potential confounding factors, age (adjusted OR 1.07, 95% CI 1.01–1.14, P = 0.010) and failure to initial treatment (adjusted OR 13.03, 95% CI 2.34–72.65, P = 0.005) were independently associated with increased mortality (Table 5). However, the time to anti-PCP treatment was not associated with increased mortality.