Pneumocystis jirovecii Disease: Basis for the Revised EORTC/MSGERC Invasive Fungal Disease Definitions in Individuals Without Human Immunodeficiency Virus

Abstract

Background

Pneumocystis jirovecii pneumonia (PCP) causes substantive morbidity in immunocompromised patients. The EORTC/MSGERC convened an expert group to elaborate consensus definitions for Pneumocystis disease for the purpose of interventional clinical trials and epidemiological studies and evaluation of diagnostic tests.

Methods

Definitions were based on the triad of host factors, clinical-radiologic features, and mycologic tests with categorization into probable and proven Pneumocystis disease, and to be applicable to immunocompromised adults and children without human immunodeficiency virus (HIV). Definitions were formulated and their criteria debated and adjusted after public consultation. The definitions were published within the 2019 update of the EORTC/MSGERC Consensus Definitions of Invasive Fungal Disease. Here we detail the scientific rationale behind the disease definitions.

Results

The diagnosis of proven PCP is based on clinical and radiologic criteria plus demonstration of P. jirovecii by microscopy using conventional or immunofluorescence staining in tissue or respiratory tract specimens. Probable PCP is defined by the presence of appropriate host factors and clinical-radiologic criteria, plus amplification of P. jirovecii DNA by quantitative real-time polymerase chain reaction (PCR) in respiratory specimens and/or detection of β-d-glucan in serum provided that another invasive fungal disease and a false-positive result can be ruled out. Extrapulmonary Pneumocystis disease requires demonstration of the organism in affected tissue by microscopy and, preferably, PCR.

Conclusions

These updated definitions of Pneumocystis diseases should prove applicable in clinical, diagnostic, and epidemiologic research in a broad range of immunocompromised patients without HIV.

The EORTC/MSGERC consensus definitions of invasive fungal diseases (IFDs) published in 2002 [1] and updated in 2008 [2] have evolved into essential documents for research in clinical mycology. The definitions have fostered comparison of clinical research in patients with cancer and solid-organ and hematopoietic stem cell transplantation (HSCT) [3, 4]; they have been adopted by regulatory agencies for antifungal agents [4–7] and used to evaluate diagnostic tests [8] and for epidemiologic studies [9–12]. As such, they are specifically intended for research only, and not to direct patient care.

The 2008 revised definitions had their limitations, including poor applicability to patients treated in intensive care units (ICUs), lack of thresholds of positivity, and validation of, fungal biomarkers, and a focus on opportunistic mold infections [2]. Notably, no definitions were provided for diseases caused by Pneumocystis jirovecii including life-threatening pneumonia (PCP). Pneumocystis jirovecii pneumonia is particularly relevant in patients with profound impairment of T-cell–mediated immunity.

To overcome this limitation in a time of evolving anticancer immunotherapies, change in composition of immunocompromised patient populations, and new diagnostic tools, the EORTC/MSGERC established definitions for P. jirovecii disease in their second revision of IFD definitions [13]. The definitions were based on the established triad of host factors, clinical features, and mycologic tests with categorization into probable and proven disease, and were applicable to immunocompromised adults and children without human immunodeficiency virus (HIV). Here we present the scientific rationale behind these disease classifications.

METHODS

The Pneumocystis subcommittee of the EORTC/MSGERC was asked to restrict their purview to immunocompromised individuals without HIV [13]. This restriction was based on important differences in the biology and presentation of Pneumocystis disease between patients with advanced HIV infection and immunocompromised patients without HIV with attendant impact on applicability of diagnostic criteria [14–19].

Under the senior author (A. H. G.), host factors and diagnostic clinical and microbiologic criteria of Pneumocystis disease were evaluated through systematic literature review. Medical subject heading (MESH) terms were used as keywords to search articles published in English in PubMed. Host factors and clinical and microbiologic criteria were formulated and adapted after discussion within the group. The process for review and formulation of consensus is detailed in the 2019 revised IFD definitions [13].

DIAGNOSTIC CRITERIA

Host Factors

Pneumocystis jirovecii is a transient fungal colonizer of human pulmonary alveoli [20–22]. Although not completely elucidated, the mode of acquisition of infection likely occurs by the airborne route and person-to-person spread [23]. Sero-epidemiologic studies suggest primary contact with the organism occurs in infancy [24, 25], with asymptomatic or mild upper respiratory tract infection [26]. There is no molecular evidence for a truly latent infection, and disease is believed to arise from prior colonization or by new infection [26, 27].

In immunocompromised patients, the organism may proliferate and cause lung disease through interaction with type I alveolar cells. Extrapulmonary manifestations are rare [28] and may be associated with atypical forms of P. jirovecii [29]. Pulmonary disease or PCP is typically diffuse with alveolar damage, an eosinophilic intra-alveolar foamy matrix, and interstitial inflammatory response, resulting in restrictive pulmonary disease, progressive hypoxemia, and death if untreated [26, 30]. Alveolar macrophages are the central effector cells in host defense against P. jirovecii, and PCP is exclusively associated with qualitative or quantitative impairment of T-cellular immunity [31–33].

Apart from institutionalized neonates with functional immaturity of cellular immunity, the principal populations at risk include children with primary T-cell immunodeficiencies, patients with very low CD4+ lymphocyte counts (eg, those undergoing intensive immunosuppressive therapy with glucocorticosteroids and other agents affecting CD4+ lymphocyte counts), and patients undergoing chemotherapy for cancer, or solid-organ transplantation or allogeneic HSCT prior to immune-reconstitution [26, 30, 34–36]. Historical incidence rates of PCP in immunocompromised patients without HIV not receiving prophylaxis are more than 20% for children with acute lymphoblastic leukemia [37], non-Hodgkin’s lymphoma [38], or soft tissue sarcoma [39]; and for adults, rates are 20–30% in non-Hodgkin’s lymphoma [40, 41], 5–15% in allogeneic HSCT recipients [42], 5–15% in solid-organ transplant recipients [43], and 6% in brain tumor patients with irradiation receiving glucocorticosteroids [44]. Compliance with trimethoprim/sulfamethoxazole renders Pneumocystis disease unlikely [39].

In case series, 80–90% of immunocompromised patients with PCP without HIV received corticosteroids [36, 45, 46] and, similar to patients with HIV [36, 47], a systematic review concluded that a CD4+ cell count of less than 200 was a sensitive biomarker of “high risk” in immunocompromised patients without HIV [48]. The strong association between Pneumocystis disease and immunosuppression suggests that risk should be focused on the net state of immunosuppression as opposed to underlying disease, including the use of glucocorticosteroids and therapies and conditions that specifically compromise T-cell–mediated immunity (Table 1).

Table 1.

Host Factors, Clinical Criteria, and Microbiologic Criteria Used for the Definition of Pneumocystis jirovecii Pneumonia

	Description
Host factors	• Use of therapeutic doses of ≥0.3 mg/kg prednisone equivalent for ≥2 weeks in the past 60 days • Low CD4+ lymphocyte counts (observed or expected; <200 cells/mm3) induced by a medical condition, anticancer, anti-inflammatory, and immunosuppressive treatment, including but not limited to:  - Primary immunodeficiencies with numeric/functional T-cell deficiency  - Acute leukemia, non-Hodgkin’s lymphoma, solid tumors, allogeneic HSCT  - Solid-organ transplantation  - Autoimmune- and hyperinflammatory disorders, including treatment with agents that lead to functional T-cell deficiencies
Clinical criteria	• Fever • Respiratory symptoms including cough, dyspnea, or hypoxemia • Bilateral or diffuse GGO on X-ray with interstitial infiltrates as the predominant feature; alveolar, alveolar-interstitial, and unilateral infiltrates are less frequent • Extensive, mostly diffuse GGO on CT scans, which typically has an upper lobe and perihilar predominance, sometimes with peripheral sparing or a mosaic pattern; consolidations, small nodules, and unilateral infiltrates are less frequent
Microbiologic criteria	• Visualization of P. jirovecii by microscopy using conventional staining methods (Gomori methenamine silver, Toluidine Blue O, Giemsa, Calcofluor White) or immunofluorescence staining in tissue, BAL fluid, induced sputum, expectorated sputum, or oral wash • Amplification of P. jirovecii DNA by quantitative real-time PCR in BAL fluid, induced sputum, or oral wash • Detection of β-d-glucan in serum if another invasive fungal infection and a false-positive result can be ruled out

Abbreviations: BAL, bronchoalveolar lavage; CT, computed tomography; GGO, ground-glass opacity; HSCT, hematopoietic stem cell transplantation; PCR, polymerase chain reaction.

Clinical Criteria

Symptoms and Signs

Unlike in patients with HIV where the onset of PCP is usually gradual and insidious, with few physical or radiologic findings, in immunocompromised patients without HIV, clinical presentation tends to be more acute with rapid onset of respiratory symptoms and faster progression to respiratory failure, higher ICU admission rates, and mortality exceeding 50% [16, 36, 49–51]. Clinical features include fever, progressive dyspnea, nonproductive cough, chest pain, circulatory failure, pneumothorax, and, very rarely, hemoptysis [36, 50–61]. The differences in clinical presentation between patients with and without HIV appear to be related to differences in severity of pneumonia and degree of lung inflammation. Patients with HIV have a higher organismal load and fewer granulocytes in the lung than do patients without HIV, with greater impairment of gas exchange [15] (Table 1). There is a paucity of findings at auscultation. Serum lactate dehydrogenase levels are not typically elevated in patients without HIV [49–51].

Very rarely, Pneumocystis disease may spread to other body sites and cause extrapulmonary manifestations whose signs and symptoms are nonspecific and will depend on the site involved [28, 29].

Radiographic Patterns

Between 10% and 15% of immunocompromised patients with PCP without HIV have normal chest radiographs and, among those with abnormalities, close to 30% have nonspecific findings. Typical findings are bilateral, diffuse ground-glass opacity (GGO) with interstitial infiltrates. Alveolar infiltrate patterns, unilateral involvement, lung nodules, or pleural effusions are less frequent. In mild or early presentations, opacities are usually perihilar. With advancing disease, opacities become diffuse and are in a butterfly pattern [19, 36].

Using high-resolution pulmonary computed tomography scans, extensive GGO is the main feature, representing exudate formation from alveolitis [15, 53]. Ground-glass opacity is usually symmetric, predominant in the perihilar regions and apices, with peripheral sparing (~20% of cases). A mosaic pattern has also been reported in 60% of cases, reflecting more severe disease [19, 62]. Pneumocystis jirovecii pneumonia treatment results in radiologic improvement, while ineffective therapy is associated with evolution to the mosaic pattern with architectural distortion and increasing pulmonary infiltrates [62] (Table 1). Because of host immune-mediated lung damage, GGO may be associated with rapid onset of lung consolidation [19]. Nodules and/or septal thickening are other findings [63]. Pulmonary cysts are rare and are attributed to longstanding, low-intensity inflammation, resulting in tissue destruction. Occasionally, pneumothorax or pneumomediastinum occurs [29, 64].

Microbiologic Criteria

Microbiologic diagnosis of PCP is hampered by the inability to cultivate the organism and little utility of serologic approaches using P. jirovecii–specific antibody tests. Although antibodies to P. jirovecii may be detected in up to 80% of individuals [25, 65], no commercial tests are available, results are variable, and the natural history of antibody persistence poorly understood. Many immunosuppressed patients are unable to produce antibodies.

The diagnosis of PCP hinges upon the visualization and/or detection of P. jirovecii in respiratory tract samples by (1) microscopy, (2) antigen detection, and (3) nucleic acid amplification tests (NAATs) (Table 1).

Microscopy

Definitive diagnosis of PCP has traditionally relied on microscopic visualization of P. jirovecii in respiratory specimens using optical brighteners, silver stains, and toluidine blue [66, 67]. Bronchoalveolar lavage (BAL) fluid or washings, with/without transbronchial biopsy, induced sputum (IS), and expectorated sputum are most often submitted for examination, but other upper tract specimens (eg, oral rinses, to avoid invasive sampling procedures) also have utility. Immunofluorescent staining for all the above specimen types exhibits superior sensitivity to conventional microscopy [66, 68, 69]. Today, conventional stains may be used (1) in laboratories that do not offer NAAT or immunofluorescent staining and (2) to visualize the cyst/trophic forms in histologic or cytologic specimens.

The use of mouse anti–P. jirovecii monoclonal antibodies to detect cysts and trophic forms in an immunofluorescent assay (IFA) format is the preferred method of microscopic diagnosis [70]. Direct antigen-detection formats identify both morphotypes while indirect IFAs detect only cysts. Although it may be an advantage to detect both forms, direct IFAs suffer from more artefact than those that detect cysts only. Rath and colleagues [66] recommend an IFA that detects only cysts as the most useful assay in contemporary routine diagnostics. The main limitations of IFAs are cost and need for a fluorescent microscope. Notably, detection of Pneumocystis microscopically in tissue, BAL fluid, or expectorated sputum remains the criterion for proven PCP. Due to suboptimal sensitivity, negative microscopy does not rule out infection.

Nucleic Acid Amplification Test Approaches

Polymerase chain reaction (PCR) and other NAAT methods are more sensitive than microscopic examination for the detection of P. jirovecii; however, their high sensitivity does not allow for easy distinction between PCP and colonization with P. jirovecii. Hence, quantification of the fungal load is essential to interpret PCR results. Purely qualitative endpoint PCR tests (single round or nested) are not recommended for PCP diagnosis.

Instead, real-time PCR is now preferred, as this approach provides quantitative results, is rapid, and allows inclusion of a PCR inhibition control. The Pneumocystis multicopy mitochondrial large-subunit ribosomal RNA (mtLSU) gene is most commonly targeted, but assays targeting the mitochondrial small-subunit ribosomal RNA (mtSSU) gene, multi-copy major surface antigen (MSG) gene, 18S ribosomal RNA (rRNA), internal transcribed spacer (ITS), 5S rRNA, DHPS, B-tubulin, and HSP70 genes have been developed. Overall, the performance of in-house PCR tests is similar to commercial tests [66].

Attempts have been made to define a quantitative PCR (qPCR) threshold. Based on literature data and including the results of a prospective multicenter laboratory evaluation over 4 years [71], consideration was given to defining 2 types of PCR “thresholds”: a “high” threshold that would diagnose PCP with 100% specificity and a “low” threshold that would exclude PCR with a high degree of certainty (eg, where PCR is performed on BAL fluid). Inevitably, there will be patients with results in the gray zone in-between the 2 thresholds. Therefore, these results should be interpreted in the context of the patient’s underlying disease, immunosuppressive therapies, and other treatments to inform decisions of whether or not to institute anti-Pneumocystis therapy. Indeed, all Pneumocystis qPCR assays (in-house or commercial) should be validated in the appropriate clinical context (eg, non–HIV-positive patients) to define the aforementioned thresholds.

Perret and colleagues [72] have suggested a single cutoff of 5 × 10³ copies/mL to discriminate PCP from colonization while assessing an in-house qPCR on BAL samples from 71 patients with positive PCR results including 62 patients without HIV. However, test variability observed due to master mix and thermocycler parameters prevented the application of a consensual threshold and test standardization is essential [73]. The Pneumocystis Working Party of the Fungal PCR Initiative (www.fpcri.eu) has been working towards such a consensus method. A 16-laboratory international study confirmed the large (10 000-fold) variation between qPCR assays for a given sample. Assays targeting whole nucleic acid and the mtSSU gene were the most sensitive and have been put forward as a basis for standardizing P. jirovecii loads [74].

When using qPCR, IS and BAL fluid are equally appropriate as samples to diagnose PCP, and potentially the same interpretive cutoff values can be used; however, the number of patients with PCP in whom both IS and BAL fluid have been tested remains small [75]. Sensitivity of qPCR on upper respiratory tract samples is lower than on BAL fluid. Such positive results can be used as a microbiologic criterion to diagnose PCP, but negative results cannot exclude PCP.

Antigen Detection

Because of the invasiveness of BAL sampling and imperfect specificity of PCR, the utility of the fungal biomarker, 1,3-β-d-glucan (BDG), while not specific for P. jirovecii, has been studied in PCP diagnosis [66, 76–78].

BDG is concentrated in the cell wall of the cyst (but not the trophic) form of P. jirovecii [76] Several BDG commercial assays can be used—these differ in their cutoff value to call a “positive” test and hence affect study comparability. The most commonly studied are the Fungitell (Associates of Cape Cod, Inc, East Falmouth, MA) test and the Wako β-D-glucan Test (FUJIFILM Wako Chemicals, Osaka, Japan).

A systematic review and meta-analysis of the utility of serum BDG testing provided data studying 997 patients with PCP and 3062 controls [79]. Pooled sensitivity and specificity for PCP were 91% (95% confidence interval [CI], 87–94%) and 79% (95% CI, 72–84%), respectively. The sensitivity in patients with HIV was higher than in those without HIV (94% vs 86%; P = .02) with similar specificity. The authors concluded that a negative BDG test is only associated with a low post-test probability of PCP (≤5%) when the pre-test probability was low (≤20%) in patients without HIV. The moderate specificity can be explained by the positivity due to other fungal infections (eg, candidiasis) and the false positivity seen in patients with hemodialysis, receipt of immunoglobulins, and certain medications. A positive BDG result should hence trigger tests to exclude other IFDs (Table 1).

The BDG assay may also be able to distinguish P. jirovecii colonization from infection [80]. In 166 immunocompromised patients with pulmonary infiltrates, the results of BAL fluid PCR and serum BDG (Wako; FUJIFILM) were compared. BDG levels in patients with definite PCP were significantly higher than those in patients with probable infection, colonization, and patients without PCP (all P < .001). BDG levels in patients with definite/probable PCP (173.1 ± 18.8 pg/mL) were also higher than those in colonized patients who had PCR-positive results (P < .002). The cutoff level for discrimination was estimated at 33.5 pg/mL (positive-predictive value, 0.925).

The combination of qPCR and serum BDG testing may result in greater diagnostic performance. In 1 study, patients considered to have PCP (by qPCR on BAL fluid) had BDG levels of 100 pg/mL or higher (Fungitell; Associates of Cape Cod) compared with colonized patients (BDG <100 pg/mL), suggesting that qPCR on BAL fluid plus serum BDG testing can differentiate between PCP and colonization [81]. In patients with unexplained lung infiltrates who underwent evaluation for suspected PCP with bronchoscopy, higher BDG values (>200 pg/mL; Fungitell; Associates of Cape Cod) were associated with clinical PCP among PCR-positive patients [82]. If BAL sampling is not feasible, combined BDG measurement with qPCR on nasopharyngeal aspirates has been an alternative [83]. It remains uncertain whether serum BDG can inform treatment response or prediction of the outcome [84]. There are no supporting data for BDG detection in BAL fluid.

BDG detection for PCP diagnosis has adequate sensitivity. Requiring 2 consecutive positive results improves specificity [85]. More experience is needed with commercial assays other than the Fungitell assay, with assignment of an optimal cutoff value. One study evaluating 116 PCP cases revealed the performance of the Wako β-glucan assay (FUJIFILM; cutoff 11 pg/mL) to be similar to the Fungitell (Associates of Cape Cod) assay with lower inter- and intra-run variability [86].

DISEASE DEFINITIONS

The criteria for the diagnosis of PCP by the 2019 Update of the EORTC/MSGERC Consensus Definitions of Invasive Fungal Disease are summarized in Table 1, and the disease definitions based on this triad of criteria are in Table 2.

Table 2.

Diagnostic Criteria for Definition of Proven and Probable Pneumocystis jirovecii Pneumonia

	Description
Proven PCP	• Clinical and radiologic criteria, plus:  - Demonstration of P. jirovecii by microscopy using conventional or immunofluorescence staining in tissue or  - Demonstration of P. jirovecii by microscopy using conventional or immunofluorescence staining in respiratory specimens
Probable PCP	• Appropriate host factors and clinical and radiologic criteria, plus:  - Amplification of P. jirovecii DNA by quantitative real-time PCR in respiratory specimen or  - Detection of β-d-glucan in serum (alternative method; another IFD and a false-positive result should be ruled out)

Abbreviations: IFD, invasive fungal diseases; PCP, Pneumocystis jirovecii pneumonia; PCR, polymerase chain reaction.

The diagnosis of proven PCP is based on clinical and radiologic criteria plus demonstration of P. jirovecii by microscopy using conventional or immunofluorescence staining in tissue or respiratory tract specimens. The diagnosis of proven PCP does not require a host factor; however, in the absence of a host factor at the time of diagnosis, investigations for a predisposing host factor should be initiated. Quantitative PCR is not accepted as a microbiologic criterion for proven PCR because of the lack of standardized methodology and clear interpretation rules to distinguish colonization from infection.

Probable PCP is defined by the presence of appropriate host factors and clinical and radiologic criteria, plus detection of P. jirovecii by qPCR in respiratory tract specimens (BAL fluid, induced sputum, or oral wash) and/or detection of BDG in serum, provided that another IFD and a false-positive result can be ruled out. Two types of PCR “thresholds” for distinguishing colonization by P. jirovecii from disease have been proposed: a “high” threshold that would diagnose PCP with 100% specificity and a “low” threshold that would exclude PCR with a high degree of certainty, at least on BAL fluid; however, thresholds have not been defined by consensus. The inclusion of the serum BDG test is based on high sensitivity and excellent negative-predictive value; uniformly accepted thresholds, however, have not been defined.

Whereas the definitions of proven and probable IFDs are reliable for research purposes, a diagnosis of possible IFD per se is inconclusive due to lack of a microbiologic criterion but may be upgraded during the diagnostic workup if an appropriate microbiologic test result becomes positive. For PCP, possible disease is defined by appropriate host factors and clinical and radiologic criteria but absence of microbiologic confirmation by microscopy and PCR in tissue or respiratory specimens and BDG in serum, respectively (not done or negative result). Pneumocystis jirovecii pneumonia is highly unlikely in cases of failure to demonstrate P. jirovecii by microscopy in lung or to demonstrate P. jirovecii by PCR in BAL material and a negative BDG in serum immediately prior to or within 3–5 days after the start of appropriate treatment [82].

Finally, the diagnosis of extrapulmonary Pneumocystis disease requires demonstration of the organism in involved tissue by microscopy and, preferentially, by NAAT.

SUMMARY

These definitions represent consensus expert opinion based on current evidence. They will need regular review for relevance, particularly regarding the role of qPCR in supporting the definition of proven disease. However, this is more a consideration of scientific accuracy than of practical relevance, as probable and proven disease is usually grouped as 1 entity in clinical research.

Notes

Financial support. No funding was received in the context of the development of the disease definitions and for the writing of this document.

Supplement sponsorship. This supplement is sponsored by MSGERC and EORTC.

Potential conflicts of interest. K. L. reports personal fees from SMB Laboratoires, Gilead, FUJIFILM Wako, and ThermoFisher Scientific, and personal fees and nonfinancial support from Pfizer and MSD, all outside the submitted work; S. C.: untied educational grants from MSD Australia and F2G Ltd UK, outside the submitted work. C. V. reports personal fees from Merck, Sharp & Dohme, personal fees from Gilead Sciences, personal fees from Angelini, personal fees from Forrest Italia, personal fees from Pfizer, personal fees from Basilea, and personal fees from Astellas, outside the submitted work. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.