Adjunctive steroid in HIV-negative patients with severe Pneumocystis pneumonia

Pneumocystis jiroveci pneumonia (PCP) is a major cause of acute respiratory failure in immunocompromised patients. Malignant disease, steroid treatment, and transplantation of solid organs or bone marrow are the leading causes of T-cell suppression, which is associated with a high risk of opportunistic infections, including PCP [1]. The number of patients with T-cell suppression has risen in recent years, resulting in an increased incidence of PCP [2, 3]. In recent studies, more than 8% of patients with hematological malignancies admitted to the ICU for acute respiratory failure had PCP [4]. Mortality rates of up to 30% have been reported in cancer patients with PCP [5].

Studies done in the 1990s showed that adjunctive treatment with high-dose steroids (HDS) was associated with a dramatic decrease in mortality during PCP episodes in HIV-positive patients [6, 7]. Corresponding proof of efficacy is not available for HIV-negative patients with PCP, and findings from the three available studies, all retrospective, are conflicting. In a 1998 study in 30 patients, the 16 patients given adjunctive HDS had no difference in mortality but spent less time on mechanical ventilation compared to the 14 patients managed without steroids [8]. The second study, reported in 1999, compared 15 patients with HDS and 8 without HDS and found no significant differences in mortality or ICU stay length [9]. The most recent study was published in 2011 and found no significant difference in mortality between the 59 patients given HDS and the 29 other patients [9]. More over two retrospective studies from our group could not conclude of outcome improvement with adjunctive steroid in that setting [5‐10]. The small sample sizes may have jeopardized the ability of these studies to detect significant differences between patients given HDS and other patients.

The pathophysiology of PCP may differ between patients with and without HIV infection. Studies have shown that HIV-negative patients with PCP were older and had a larger number of co-morbidities [2, 11], longer symptom duration at diagnosis [12], and higher neutrophil counts in bronchoalveolar lavage fluid [13], compared to HIV-positive patients [14]. Conceivably, these differences between the two populations might affect the ability of steroid therapy to provide therapeutic benefits.

Here, our objective was to determine whether HDS produced therapeutic benefits in HIV-negative patients with severe PCP requiring admission to the intensive care unit (ICU). To increase the sample size for our investigation, we included patients from four different sources, namely, three previously published studies and a teaching-hospital ICU database.

From 1988 to 2011, 139 HIV-negative patients were admitted to the ICU with PCP infection (Figure 1). Table 1 described underlying diseases and Table 2 reports main characteristics at ICU admission in survival and dead patients. Co-infection occurred at admission in 28 patients (20%) (Table 3). Mechanical ventilation was required in 84 patients (60%) and vasoactive agents in 41 (29.5%). Time on mechanical ventilation was 0 [0–7] days. Pneumothorax occurred in 11 (7.9%) patients and ICU-acquired infections in 31 (22.3%). Adjuvant steroids were given to 107 (76.9%) patients: 72 (51.8 %) patients received HDS and 35 (25.2%) LDS. The remaining 32 (23%) patients received no steroids in the ICU. More than half of patients received steroids before ICU admission (n=74, 53%), 35 patients in the LDS group and 39 patients in the HDS group. Of the 139 patients, 36 (25.8%) died in the ICU. Length of ICU stay was 8 [4‐13] days.

By univariate analysis (Table 2), ICU mortality was associated with age, non-hematological disease, ICU-acquired infection, higher SAPS II, lower P/F, pneumothorax, endotracheal mechanical ventilation, and shock. ICU mortality did not vary significantly across steroid-treatment groups: 29.2% in the HDS group, 20% in the LDS group, and 25% in the HS group (P=0.6).

By multivariate analysis, factors independently associated with ICU mortality were steroid use before ICU admission, SAPS II, non-hematological disease, HDS, and shock (Table 4).

In the group of 59 patients who required endotracheal mechanical ventilation, ICU mortality was not significantly different between the HDS patients (20/30 patients died, 66%) and the other patients (13/29 died, 45%) (P=0.09). Figure 2 described outcome according the severity of PCP and steroid group.

ICU acquired infection occurred in 31 (22.3%) patients (Table 4), respectively in 12.5%, 31.4% and 22.2% in no steroid group, LDS group and HDS group, respectively (p=0.19).

By multivariate analysis, independent risk factors for ICU-acquired infection were shock (OR= 6.80; [2.55-18.14], p<0.001) and pneumothorax (OR, 4.80; [1.09-21.1], p=0.04). The risk of ICU-acquired infection was not influenced by the SAPS II (p=0.36), steroid treatment before ICU admission (p=0.91), or steroid treatment in the ICU (p=0.46).

A comparison of three time periods (1991–1996, 1997–2004, and 2005–2011) showed a trend for declined in mortality over time (respectively 36%, 26%, 18%; p=0.12). The proportion of patients managed with endotracheal mechanical ventilation was higher in the most recent period (respectively 36%, 40%, 44%; p=0.73), whereas the proportion of patients given HDS was higher in the earliest period (respectively 83%, 48%, 33%; p<0.001).

Data on BAL count were available for 47 patients: 13 patients had neutrophil alveolitis (>10%), 7 patients had lymphocyts alveolitis (>5%), 4 patients had normal BAL count and 21 had alveolitis with neutrophils and lymphocyts.

In this study, HDS was associated with an increase in ICU mortality in HIV-negative patients with PCP. This increase was not related to a higher rate of ICU-acquired infection.

In recent non-ICU studies of patients with PCP, up to 30% of patients died [15]. However, the 25.8% overall ICU mortality rate in our study compares favorably with the 39% to 66% rates in other ICU studies done between 1995 and 2011, with the highest rates being reported in mechanically ventilated patients [2, 11, 15]. During the same period, survival improved overall among immunocompromised patients admitted to the ICU [16, 17].

The pathophysiology of PCP may differ between HIV-positive and HIV-negative patients. For instance, differences in lymphocyte and neutrophil counts in bronchoalveolar lavage fluid have been reported [13, 14]. In our study, in agreement with Limper et al. most BAL included neutrophil alveolitis. [13]. Diagnosis remained easier in HIV infected patients with higher organism burden. [15]. In ICU studies, PCP was more often fatal in non-HIV-infected patients than in HIV-positive patients [2, 11]. The HIV-positive patients were younger and had less co-morbidity [11]. Similarly, our patients were older than those in studies of HIV-positive patients. We did not collect co-morbidities. Co-infections, most notably with viruses, were considerably more prevalent among HIV-negative patients than HIV-positive patients in previous studies [11, 18].

In HIV-negative patients with PCP, there is no proof that adjunctive steroid therapy is beneficial [5, 8‐10]. The most recent study compared 29 patients treated with adjunctive steroid therapy to 59 patients treated without steroids [16]. Adjunctive steroid therapy was defined as at least 40 mg of prednisone bid, and no effort was made to separate patients on HDS and LDS. The proportions of patients with hematological malignancies and solid organ transplants were similar to those in our population. ICU mortality was 35.8% overall, with no significant difference between patients managed with and without steroids. In our study, mortality was lower (25.8%), despite greater acute illness severity at ICU admission for acute respiratory failure. Adjunctive HDS was independently associated with increased mortality in our study. Steroid therapy might be expected to increase the rate of viral reactivation and bacterial infection in immunocompromised patients [18]. However, the adverse effect of HDS on mortality was not related to an increase in the rate of ICU-acquired infections. ICU-acquired infections occurred in only 22% of our patients, a low rate given their immunocompromised status.

Our study has several limitations. First, only 139 patients were included over a 23-year period. Outcome of severe PCP tends to improve throughout this study period and more patients in the earlier period received HDS. PCP in immunocompromised patients has become uncommon since the use of routine prophylactic treatment [19, 20]. In addition, we confined our study to patients with severe illness requiring ICU admission. Second, we used a retrospective design. We had no information on the factors that governed the use and timing of steroid therapy. Moreover, steroid therapy seemed less beneficial in the patients with greater disease severity (Figure 2). We cannot exclude that HDS was started in some patients as rescue therapy after endotracheal mechanical ventilation became necessary. Nevertheless, our study is the largest in this field [8, 16] and showed that HDS was independently associated with mortality. Third, we did not have complete data collected on the day of steroid therapy initiation. Fourth, some patients were included many years ago, at a time when the proportion of patients given HDS was higher than in recent years. In this early period, few tests were available for detecting viral infections. Viral co-infection may increase ICU mortality and the risk of ICU-acquired infections [2, 18]. Moreover, ICU mortality of immunocompromised patient improved over time [17, 21]. Fifth, we separate patients in three groups according the steroids dose, resulting in a limited number of patients in each group. Particularly, in LDS group mortality was the lowest but not significantly different from other group due to the small number of patients.

In this study, adjunctive HDS in HIV-negative patients admitted to the ICU with mild-to-severe PCP was associated with an increased risk of ICU mortality. This increase in mortality was not related to a higher rate of ICU-acquired infection in this retrospective study. A prospective is warranted to confirm these results.

VL, AD, EA, designed the study, collected and interpreted the patient data and wrote the manuscript, AD, CA, analyzed the data. All authors read and approved the final manuscript.