Introduction

Pneumocystis jirovecii pneumonia (PJP) is an opportunistic fungal infection that mostly afflicts immunosuppressed patients with deficits in cell-mediated immunity [1]. PJP was first described in malnourished infants in Europe following World War II [2]. In the 1960s and 1970s, it was diagnosed primarily among patients with hematologic malignancies [3]. From the 1980s until the present, PJP is most often found in patients with advanced AIDS. Other known risk-factors include corticosteroid treatment for various indications [4], solid tumors [1, 5], inflammatory diseases [6], and immunosuppressive drug use following bone marrow and solid organ transplantation [7]. The introduction of active antiretroviral therapy and the use of cotrimoxazole prophylaxis resulted in a dramatic drop in the incidence of PJP among patients with HIV infection. Several studies have reported increased rates of PJP among patients treated with new immunomodulatory drugs (monoclonal antibodies, anti-TNF therapy, tyrosine kinase inhibitors and proteosome inhibitors) [8–12]. Patients in most of these studies were treated with combinations of immunosuppressive drugs; thus, it was impossible to separate the role of each one.

The prognosis of PJP among patients without HIV is typically worse than in those with HIV, with high rates of in-hospital mortality (50% to 86%, according to some reports [4, 13]), respiratory failure and ICU admission [14]. Due to a tendency for rapid and severe progression, early diagnosis and treatment of PJP is vital. Furthermore, since the diagnosis of PJP is usually established by bronchoalveolar lavage (BAL), a high level of suspicion is needed for early detection and initiation of appropriate treatment.

The aim of this study was to assess risk-factors for PJP among non-HIV patients.

Methods

Study design: This retrospective, test-negative, case-control study was performed at 6 tertiary-level medical centers in Israel. All participating hospitals had a dedicated haemato-oncology department, 5 had stem cell transplantation units, and 5 had solid organ transplantation services. The 6 general hospitals had a total of 5,516 beds, among 15,871 in the entire country (35%). The study was conducted from January 2006 through December 2016.

Case definition: A case was defined as a hospitalized patient with clinical and radiological evidence of pneumonia and a positive direct microscopic examination for PJP (May-Grunwald-Giemsa stain, Gomori-Grocott staining or immunofluorescence) from BAL fluid. All PJP diagnosed by pathology were available, and electronic medical records were reviewed for evidence of pneumonia and HIV status. All cases diagnosed during the study period were followed until discharge.

Controls were defined as hospitalized patients with clinical and radiological evidence of pneumonia, for which PJP was excluded by BAL. Controls were matched 2:1 by age and sex to PJP cases. Exclusion criteria were younger than 18 years-of-age and HIV-positive status. HIV negativity was determined based on documentation in the medical records. To minimize false positive results representing PJP colonization, we did not include patients with PJP detected by PCR alone [15, 16].

Data extracted from the electronic medical records of cases and controls included demographics, past medical history, presence of an immunosuppressive condition considered a risk-factor for PJP (including hematological malignancies, solid tumors, solid organ transplant, bone marrow transplant and inflammatory disorders), use of immunosuppressant drugs (including corticosteroids with a dosage equivalent to ≥20 mg prednisone for ≥2 weeks, cytotoxic chemotherapy and immunomodulatory drugs) and use of trimethoprim/sulfamethoxazole as prophylactic agents. Laboratory results and clinical outcomes were collected, as well. Due to the large variety of risk-factors in the medical history and the large number of immunosuppressive drugs in use, we grouped drugs and diseases by mechanisms. The complete list is presented in S1 and S2 Tables.

Meir MC review board approved the study approval number 268–16. Tel-Aviv MC review board approved the study, Sheba MC review board approved the study, Hadassah MC review board approved the study, Assaf Harofeh MC review board approved the study, Rabin MC review board approved the study.

No consent was required, data were analyzed anonymously.

Results

Seventy-six patients met the case definition for non-HIV related PJP and were compared to 159 matched control patients. The average age of both groups was 63 years. Most patients were men, 64% and 67% in the case and control groups, respectively. Demographic, clinical, laboratory and hospitalization outcome data for both groups are presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographics, medical history, drug, clinical, laboratory, imaging and hospitalization outcomes for case and control groups.

https://doi.org/10.1371/journal.pone.0239042.t001

Univariate analysis (Table 1), demonstrated that, compared to controls, patients with PJP had a higher incidence of hematologic malignancies (44.7% vs. 28.9%, p = 0.017), and inflammatory diseases (18.4% vs. 6.9%, p = 0.007), and a lower frequency of organ transplantation (9.2% vs. 20.8%, p = 0.028). They were treated more frequently with corticosteroids, (61.8% vs. 22.6%, p < 0.001) chemotherapy (36.8% vs. 19.5%, p = 0.006) and antilymphocyte monoclonal antibodies, mainly rituximab (15.8% vs. 5%, p = 0.008). Fewer were treated with sulfamethoxazole and trimethoprim prophylaxis (6.6% vs. 16.4%, p = 0.038). Patients with PJP had lower body mass index (21 vs. 25, p < 0.001), lower albumin levels (2.7 g/dL vs. 3.2 g/dL, p < 0.001), and greater incidence of lymphopenia (<1000/mcl) (80.3% vs. 46.5%, p < 0.001).

The in-hospital mortality rate (47.2% vs. 37%, respectively, p = 0.15) was similar in both groups. However, the rate of invasive ventilator support was higher among the PJP cohort (61.3% vs. 44.3%, p = 0.015).

In multivariate logistic regression analysis for risk factors (Table 2), PJP was independently associated with Rituximab use (OR 11.47 (95%CI 1.5–87.7), p = 0.019), high-dose steroid treatment (OR 4.39 (95%CI 1.52–12.66), p = 0.006), lymphopenia (OR 8.13 (95%CI 2.48–26.61), p = 0.001), low albumin (OR 0.15 (95%CI 0.04–0.54), p = 0.004) and low BMI (OR 0.80 (95%CI 0.68–0.93), p = 0.018).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Risk factors for PJP based on multivariate logistic regression analysis^a.

https://doi.org/10.1371/journal.pone.0239042.t002

Following the identification of rituximab as a major risk factor for PJP, we present the demographic and clinical data of this cohort of patients, including the time frame from last rituximab treatment to PJP onset (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Demographic and clinical data of PJP patients treated with rituximab.

https://doi.org/10.1371/journal.pone.0239042.t003

Discussion

In this multicenter study of risk factors for PJP in non-HIV patients, antilymphocyte antibodies, mainly rituximab, were found to be the leading risk factor for PJP, with an OR >11. Rituximab, an anti-CD20 antibody was approved in 1997 for B-cell hematological malignancies and autoimmune disorders. In 2007, a black box warning regarding excessive risk for progressive multifocal leukoencephalopathy was issued. Several publications implied a role of rituximab in causing CD4 lymphopenia, although most patients treated with rituximab did not have serial CD4 measurements [17]. Another study on the immunity required for PJP clearance identified a critical role for B cells. In addition to production of PJP-specific antibodies, B-cells are required to prime CD-4 cells for normal expansion and memory generation [18]. The link between rituximab and PJP was reported previously [19]; although, most patients received it as part of combination therapy. Martin-Garrido et al. followed patients treated with rituximab: 30 subsequently developed PJP, but only 3 had been treated with rituximab alone, while the others had received it in combination with either steroids or chemotherapy [19]. In a group of 7,554 rituximab-treated patients in Taiwan, the prevalence of PJP was found to be 2.95%, as compared to 1.3% in controls [20].

In our cohort of 12 PJP patients treated with rituximab, all developed the disease within 90 days of last treatment. This is in accordance with a short report by Alexandre et al. of 11 similar patients who developed PJP within 11 weeks [21]. Although the numbers are small, this might suggest that long term preventive therapy might not be required.

Steroid use (OR 4.98) is a well-known risk-actor for PJP and an indication for preventive therapy [4]. Of note, the odds for developing PJP were twice as high in patients treated with antilymphocyte antibodies than in those treated with steroids. Low BMI and low albumin are proxies for inadequate nutrition, and thus, are expected risk-factors [2].

Mansharamani et al. showed that CD4 <300 denoted a higher risk for PJP infection among immunocompromised patients and recommended they receive prophylaxis [22]. We identified lymphopenia (<1000/mcl) as a risk-factor (OR 7.96), but we did not have CD4 lymphocyte counts for our non-HIV patients.

This study had several limitations. It included patients whose PJP was diagnosed by microscopy only. Although this diagnostic method decreased the number of false positive results and instances of colonization in the case group, it might have increased the incidence of false negative results in the control group. Due to the retrospective nature of the study, data on BMI were incomplete, and HIV status was not ascertained in the HIV-negative individuals. However, information regarding drug therapy, the primary focus of the study, was retrieved from pharmacy records and thus, was complete.

In conclusion, rituximab, which is prescribed for a wide variety of malignant and inflammatory disorders, was found to be significant risk-factor for PJP. Increased awareness of possible PJP infection in this patient population is warranted.

Supporting information

Acknowledgments

Faye Schreiber, MS edited the manuscript. She is an employee of Meir Medical Center.