Eur J Clin Microbiol Infect Dis (2010) 29:1147–1152 DOI 10.1007/s10096-010-0980-5

ARTICLE

A new method for the detection of Pneumocystis jirovecii using flow cytometry

J. Barbosa & C. Bragada & S. Costa-de-Oliveira & E. Ricardo & A. G. Rodrigues & C. Pina-Vaz

Received: 7 August 2009 /Accepted: 22 May 2010 /Published online: 16 June 2010 # Springer-Verlag 2010

Abstract Pneumocystis jirovecii is an opportunistic patho- reactions to bacteria or fungi. All positive cases detected by gen responsible for severe in immunocompro- IFS were positive by FC; however, FC classified eight mised patients. Its diagnosis has been based upon direct samples to be positive which were classified as negative by microscopy either by classic staining (Gomori Grocott) or routine technique. These samples were obtained from by epifluorescence microscopy (immunofluorescence stain- patients with respiratory symptoms who responded favour- ing, IFS), both of which are time-consuming and low on ably to Pneumocystis-specific therapy and were subsequently sensitivity. Our aim was to develop a flow cytometric (FC) considered to be true-positives. Using clinical diagnosis as a protocol for the detection of P. jirovecii on respiratory reference method, FC showed 100% sensitivity and speci- samples. In our study, 420 respiratory samples were analysed ficity, whereas IFS showed 90.9% sensitivity and 100% in parallel by IFS and FC, and compared from clinical specificity. According to our results, a new diagnostic diagnosis to its resolution upon specific anti-Pneumocystis approach is now available to detect P. jirovecii in respiratory therapy. The optimum specific antibody concentration for FC samples. analysis was determined to be 10 µg/ml, without any cross-

Introduction * : : : : J. Barbosa ( ) :C. Bragada S. Costa-de-Oliveira E. Ricardo A. G. Rodrigues C. Pina-Vaz Pneumocystis jirovecii (previously named Pneumocystis Department of Microbiology, Porto Faculty of Medicine, carinii University of Porto, ) is an atypical . For 80 years, it was Al. Prof. Hernâni Monteiro, considered to be a protozoan, since its diagnostic form is 4200-319 Porto, Portugal a cyst, but ribosomal RNA and DNA studies demonstrated e-mail: [email protected] similarities to fungus [1]. In 2001, this microorganism was J. Barbosa officially reclassified as a fungus belonging to the class Escola Superior de Saúde Jean Piaget, Archiascomycetes [1] and was renamed P. jirovecii [2]. In Vila Nova de Gaia, Portugal humans, this pathogen is responsible for an at the lower respiratory tract, but it can lead to A. G. Rodrigues Pneumocystis Burn Unit, Department of Plastic and Reconstructive Surgery, severe pneumonia (PcP) in immunocompro- Hospital S. João, mised patients, particularly in those with AIDS [3, 4]. Prior Porto, Portugal to the AIDS epidemic, P. jirovecii had been sporadically reported as a cause of death in malnourished infants and C. Pina-Vaz Department of Microbiology, Hospital S. João, responsible for epidemics in institutions like nursing homes Porto, Portugal and hospitals [5]. Currently, it represents the most frequent : : : AIDS opportunistic pathogen, although PcP incidence has S. Costa-de-Oliveira E. Ricardo A. G. Rodrigues C. Pina-Vaz decreased significantly due to the extensive use of highly Cardiovascular Research and Development Unit, Faculty of Medicine, University of Porto, active anti-retroviral therapy (HAART) and prophylactic Porto, Portugal treatment [4–6]. PcP may also represent a small fraction of 1148 Eur J Clin Microbiol Infect Dis (2010) 29:1147–1152 respiratory infections in patients with pulmonary diseases patients, 130 were HIV-positive. Samples were kindly such as chronic obstructive pulmonary disease (COPD) or provided by the Microbiology Laboratories of several displaying immunity conditions like malignancy [7, 8]. Portuguese hospitals, namely, Hospital São João (HSJ, Porto, Over the past 15 years, P. jirovecii carriage in healthy Portugal), Instituto Português de Oncologia Francisco Gentil subjects has been widely demonstrated through the pres- (IPOFG, Porto, Portugal), Hospital Joaquim Urbano (HJU, ence of anti-P. jirovecii antibodies in their serum, with Porto, Portugal) and Centro Hospitalar de Coimbra (CHC, serologic analysis being used simply for the assessment of Coimbra, Portugal). A maximum of 30 mL of BAL or BW, the infection epidemiology [9]. PcP, however, is considered and 2 to 4 mL of BS samples were placed in a sterile container to result from de novo infection rather than from the and frozen at −70°C for further processing for immunofluo- reactivation of a latent infection [6]. It is assumed that rescence staining (IFS) and FC analysis. human transmission of P. jirovecii occurs through the airborne route, as suggested by studies performed in rodent Sample preparation models [6]. Nonetheless, inter-individual transmission has been suggested due to the occurrence of PcP case clusters For IFS, clinical samples were mixed with mucolytic agent in hospitals, namely, in paediatrics, hematology–oncology, N-acetyl-L-cysteine (Merck®) and incubated in a water intensive care, transplantation unit and infectious diseases bath at 37°C for 20 min or until complete dissolution, units [10]. P. jirovecii may be exhaled to the environment followed by centrifugation at 3,000g for 15 min. Subse- from infected patients, as demonstrated from cyst DNA quently, the samples were washed twice with sterile water detection in air filters from the hospital rooms of patients (H2O), centrifuged as above and the sediment re-suspended who developed PcP [10]. in a small volume of water which was divided into two The detection of P. jirovecii in an infected patient is portions; one portion was placed on a glass slide, air dried based upon the direct visualisation of organisms (cysts, and fixed, while the other portion was used for FC analysis. sporocysts or trophozoites) in clinical specimens, such as bronchoalveolar lavage (BAL), induced sputum (IS) or lung Immunofluorescence staining biopsy, which is the gold-standard specimen [6, 11]. Although specific fluorescent staining techniques were Smears were stained according to the manufacturer’sinstruc- developed for assessment under fluorescent microscopy, tions with the Detect IF™ kit Pneumocystis carinii such methods are very time-consuming, too cumbersome (Axis-Shield Diagnostics Limited, United Kingdom) and and subject to human error, especially when samples yield a analysed under an epifluorescence microscope Leitz Labor- low number of P. jirovecii organisms [11]. Molecular lux K (Leica, NY, USA). Smears were considered to be studies such as polymerase chain reaction (PCR) presented positive for P. jirovecii whenever two or more green cysts a higher sensitivity and variable specificity when compared were visualised, whether isolated or in a group. to the microscopic detection of P. jirovecii in BAL [12, 13]. However, PCR remains unavailable in most clinical Optimisation of a flow cytometry protocol microbiology laboratories and is an expensive technique [14, 15]. More so, a variable specificity has been described For the optimisation of the FC protocol, a mix of four ranging from 62.5 to 100%, depending on the type of positive samples and a mix of four negative samples specimen [14, 15]. Distinct applications of flow cytometry (evaluated by IFS) were prepared. Samples were divided (FC) in microbiology have been developed by our group in into two equal aliquots and one of them was passed through order to increase the diagnostic sensitivity in clinical a filter with a 30-μm pore size (Partec CellTrics®). Twenty samples [16–18]. In the present study, we have developed microlitres of specific enzyme from Detect IF™ kit and optimised a specific FC protocol for the detection of P. Pneumocystis carinii (Axis-Shield Diagnostics Limited, jirovecii on respiratory samples. United Kingdom) were added to 100 µl of the respiratory sample and incubated at 37°C for 30 min; 500 µl of sterile

H2O were then added and a subsequent centrifugation was Materials and methods performed at 3,000g for 5 min. One hundred microlitres of centrifuged sample were Clinical specimens stained with serial concentrations (0, 5.0, 10.0, 15.0 and 20.0 μl) of specific P. jirovecii mouse monoclonal antibody A total of 420 samples, including 380 BAL and bronchial (Axis-Shield Diagnostics Limited, United Kingdom), washing (BW) specimens, and 40 bronchial secretion (BS) followed by dark incubation at 37°C for 15 min. Subse- aspirates, were prospectively collected from patients with quently, 500 µl of sterile H2O were added and centrifuga- different immunological conditions. From those 420 tion at 3,000g for 5 min was performed; the supernatant Eur J Clin Microbiol Infect Dis (2010) 29:1147–1152 1149 was discarded and the pellet re-suspended in 100 μlof Data comparison sterile H2O, vortexed for 30 s, transferred to a propylene tube and analysed by FC. The results of IFS were compared to those obtained by FC. Clinical diagnosis was used as the standard. When results Flow cytometry analysis were discrepant, clinical symptoms and signs were consid-

ered; patients with oxygen level PaO2 <70 mmHg, an X-ray The optical characteristics of the cysts suspensions were depicting a diffuse bilateral infiltrate and with a favourable evaluated on a FACSCalibur flow cytometer (BD Bioscien- outcome upon specific therapy were considered as true- ces, Sydney , Australia) standard model equipped with three positive cases for PcP. photomultipliers (PMTs), standard filters (FL1: BP 530/30 nm; FL2: BP 585/42 nm; FL3: LP 650 nm), a 15-mW 488-nm argon laser and with CellQuest Pro Software Results (version 4.0.2, BD Biosciences, Sydney, Australia). Operating conditions included log scales on all detectors (forward scatter Respiratory samples were considered to be positive for P. [FSC], side scatter [SSC] and fluorescence detectors [FL1]). jirovecii when two or more green round formations, Acquisition settings were defined using a non-stained sample namely, cysts, were observed on epifluorescence micros- (autofluorescence) and adjusting the PMTs’ voltage to the first copy. By IFS, 340 samples were negative and 80 samples logarithmic (log) decade. Instrument controls followed the were positive for P. jirovecii. described standard procedures indicated by the manufacturer. Since there is not an available commercial type strain of P. jirovecii, we used a mix of four IFS-positive and four Assessment of the sensitivity and specificity of the flow IFS-negative samples (without cysts) as material for the cytometry staining protocol optimisation of the FC protocol. As we increased the specific antibody concentrations, an increase of the mean To assess the specificity, positive and negative IFS samples intensity of fluorescence (MIF) was also clear. Twenty for P. jirovecii were mixed with 1×106 cells/ml of different microlitres of specific antibody resulted in the highest value microorganisms, namely, Escherichia coli type strain of MIF at FL1, although a similar staining pattern was ATCC 35218 from the American Type Culture Collection obtained with 10.0 or 15.0 µl (Fig. 1). As such, all further (ATCC), Staphylococcus aureus ATCC 25923 and Candida experiments were carried out with the lowest concentration albicans ATCC 10231. The mix was stained with the sufficient to detect the organisms from the autofluorescence specific P. jirovecii antibody under optimised conditions background, i.e. 10.0 µl. and analysed by FC. To determine the specificity of the staining, micro- organisms usually present in respiratory samples were Evaluation of clinical respiratory samples mixed with negative and positive control samples, which were then incubated with the monoclonal antibody. Al- All clinical samples were analysed by both IFS and by FC though an increase in events was detectable in the presence according to the previously optimised conditions and the of high concentrations of bacteria, those events were not results were subsequently compared. selected in our settings due to their smaller size;

Fig. 1 Dot plot (side scatter [SSC] versus fluorescence intensity [FL1]) of staining (autofluorescence); (b) stained with 5.0 µg/mL of specific a positive sample containing Pneumocystis jirovecii cysts labelled with monoclonal antibody; (c) stained with 10.0 µg/mL of specific monoclo- serial concentrations of specific antibody Detect IF™ kit Pneumocystis nal antibody. Region R corresponds to the acquisition gate of P. jirovecii carinii (Axis-Shield Diagnostics Limited, United Kingdom): (a) without cysts stained with specific antibody 1150 Eur J Clin Microbiol Infect Dis (2010) 29:1147–1152

Fig. 2 Histogram correlating counts versus fluorescence intensity of antibody and spiked with ; (c) positive sample Pneumocystis marker (FL1: green fluorescence 530 nm) of respiratory stained with 10.0 µg/mL of specific monoclonal antibody. Regions M1 samples for P. jirovecii:(a) without staining (autofluorescence); (b) and M2 correspond to autofluorescence and to specific antibody negative sample stained with 10.0 µg/mL of specific monoclonal fluorescence, respectively appeared near the acquisition gate corresponding to P. occurred on samples without filtration, leading to difficulties jirovecii cysts but displayed much lower fluorescence in its interpretation. A better separation of stained cysts from intensity than that of the stained cysts (Fig. 2). debris was obtained upon filtration without signal lost Upon optimisation of the cytometric protocol, clinical (Fig. 3). For FC analysis, we considered positive all the samples were analysed. Sample preparation included muco- respiratory samples whose events appeared on the acquisi- lytic agent, with and without filtration (30-µm pore size). A tion gate R corresponding to P. jirovecii cysts stained with greater number of non-fluorescent cytometric events specific antibody (typical example shown in Fig. 3).

Fig. 3 Two-dimensional dot plot correlating a side scatter (SSC) (c) negative sample upon filtration; (d) positive sample upon filtration versus fluorescence intensity of Pneumocystis marker (FL1: green stained with 10.0 µg/mL of specific monoclonal antibody. Region R fluorescence 530 nm) of respiratory samples for P. jirovecii:(a) corresponds to the acquisition gate of P. jirovecii cysts stained with negative sample without filtration; (b) positive sample without specific antibody filtration stained with 10.0 µg/mL of specific monoclonal antibody; Eur J Clin Microbiol Infect Dis (2010) 29:1147–1152 1151

Table 1 Results obtained from 420 respiratory samples for IFS FC microorganisms in animal models. The flow cytometer is an Pneumocystis jirovecii diagnosis expensive piece of equipment present in most immunology by immunofluorescence staining Positive 80 88 laboratories that could probably be easily shared by micro- (IFS) and flow cytometry (FC) Negative 340 332 biologists. Since the amount of specific monoclonal antibody is lower than that used for IFS, it has the potential to be more economic. Through FC, we found 88 positive samples, 35 of Our study aimed to develop and optimise the cytometric which belonged to HIV patients (27%). Eighty positive analytical protocol regarding clinical samples. An important cases (19%) were detected by both IFS and FC methods. condition should be fulfilled: the concentration of the probe However, eight clinical samples previously considered to should be at its minimum to allow a clear separation of the be negative by IFS were found to be positive by FC assay stained population from all other events; while insufficient (Table 1). All eight cases were from HIV-positive patients staining would not allow a clear separation from debris, with clinical diagnostic criteria of PcP such as fever, excessive staining could result in unspecific fluorescence. cough, shortness of breath, weight loss associated with Our study faced a number of limitations such as the absence pulmonary infiltrates on chest X-ray and an arterial of a commercially available type strain, as well as the oxygen low level. In addition, the patients had improved relatively low number of positive samples, often containing upon specific treatment with trimethoprim, despite nega- very few organisms, which made it difficult to assess the tive IFS. Considering the clinical criteria as the diagnostic sensitivity limit. Cytometric analysis could be difficult to standard, such cases should be considered as true- perform in the presence of organic debris and, as such, the positives. According to our study, FC analysis displays filtration step prior to specific antibody staining resulted in 100% sensitivity and specificity versus 90.9% sensitivity an improved discrimination of debris from P. jirovecii and 100% specificity for the IFS assay. Consequently, IFS organisms. In order to confirm the specificity of the presented a positive predictive value of 100% and a monoclonal antibody used in this study, cross-reactions negative predictive value of 97.6%. were investigated using both prokaryotic (E. coli, S. aureus) and eukaryotic microorganisms (C. albicans) mixed in a negative sample for P. jirovecii. An increased number of Discussion events was clear, although without staining with the specific antibody (only fungi showed a very slight increase in green Specific and prompt diagnosis of P. jirovecii infection has fluorescence), demonstrating that the test is quite specific. significant clinical relevance, since this opportunistic The results from the 420 clinical samples collected from organism can cause severe pneumonia in immunocompro- putative PcP patients revealed that, in accordance to the mised patients [6]. As this organism cannot be cultured in literature reports, only 19% of the samples were positive by vitro, the standard diagnostic procedure for the detection of both methods (IFS and CF), possibly due to the extensive P. jirovecii relies upon the direct microscopic examination use of prophylaxis among the risk population [21, 22]. of the clinical specimen. Using commercialised fluorescein However, regarding the HIV-positive population, the or enzyme-labelled monoclonal or polyclonal Pneumocystis percentage which we found is in agreement with the antibodies in respiratory samples, a higher detection literature [21]. Comparing the results from FC to those of sensitivity can be achieved [11]. Considering the increasing IFS, several cases would have to be considered as false- complexity of diagnostic laboratory techniques, resulting in positives. However, taking into account medical criteria as higher costs and the need for greater human expertise, the the final diagnosis, those should be considered as true- development of more sensitive and automated methods positives. which allow a rapid assessment of clinical specimens is Thus, a new cytometric detection method is now highly recommended. FC is a method based upon the available for a more reliable detection of P. jirovecii evaluation of cell fluorescence, which has frequently shown organisms in respiratory clinical samples. to be specific and yielding great sensitivity. Moreover, it allows the detection of low threshold limits, which are Acknowledgements J. Barbosa was supported by a PhD grant (SFRH/ difficult to obtain by standard methods [16–19]. The first BD/44346/2008)fromFundaçãoparaaCiênciaeaTecnologia(FCT). cytometric study regarding the formerly P. carinii was developed by Libertin et al. [19] in lung homogenates samples from infected mice and humans, however, without References practical application. Later assays were performed by Lapinsky et al. [20] in order to observe the pharmacological 1. Murray PR, Rosenthal KS, Pfaller MA (2005) Fungal classifica- action of drugs upon the growth and viability of these tion, structure, and replication. In: Murray PR, Rosenthal KS, 1152 Eur J Clin Microbiol Infect Dis (2010) 29:1147–1152

Pfaller MA (eds) Medical microbiology, 5th edn. Mosby, 12. Durand-Joly I, Chabé M, Soula F, Delhaes L, Camus D, Dei-Cas Philadelphia, Pennsylvania, pp 67–73 E (2005) Molecular diagnosis of . FEMS 2. Stringer JR, Beard CB, Miller RF, Wakefield AE (2002) A new Immunol Med Microbiol 45:405–410 name (Pneumocystis jiroveci) for Pneumocystis from humans. 13. Jiancheng W, Minjun H, Yi-jun A, Lan S, Zengzhu G, Jianrong S, Emerg Infect Dis 8:891–896 Xixiong K (2009) Screening Pneumocystis carinii pneumonia in 3. Kelley CF, Checkley W, Mannino DM, Franco-Paredes C, Del non-HIV-infected immunocompromised patients using polymer- Rio C, Holguin F (2009) Trends in hospitalizations for AIDS- ase chain reaction. Diagn Microbiol Infect Dis 64:396–401 associated Pneumocystis jirovecii pneumonia in the United States 14. Alvarez-Martínez MJ, Miró JM, Valls ME, Moreno A, Rivas PV, (1986 to 2005). Chest 136:190–197 Solé M, Benito N, Domingo P, Muñoz C, Rivera E, Zar HJ, 4. Su YS, Lu JJ, Perng CL, Chang FY (2008) Pneumocystis jirovecii Wissmann G, Diehl AR, Prolla JC, de Anta MT, Gatell JM, Wilson pneumonia in patients with and without human PE, Meshnick SR (2006) Sensitivity and specificity of nested and virus infection. J Microbiol Immunol Infect 41:478–482 real-time PCR for the detection of Pneumocystis jiroveci in clinical 5. Wazir JF, Ansari NA (2004) Pneumocystis carinii infection. specimens. Diagn Microbiol Infect Dis 56:153–160 Update and review. Arch Pathol Lab Med 128:1023–1027 15. Robberts FJL, Liebowitz LD, Chalkley LJ (2007) Polymerase 6. Krajicek BJ, Thomas CF Jr, Limper AH (2009) Pneumocystis chain reaction detection of Pneumocystis jiroveci: evaluation of 9 pneumonia: current concepts in pathogenesis, diagnosis, and assays. Diagn Microbiol Infect Dis 58:385–392 treatment. Clin Chest Med 30:265–278 16. Barbosa JM, Costa-de-Oliveira S, Rodrigues AG, Hanscheid T, 7. Morris A, Sciurba FC, Norris KA (2008) Pneumocystis: a novel Shapiro H, Pina-Vaz C (2008) A flow cytometric protocol for pathogen in chronic obstructive pulmonary disease? COPD detection of Cryptosporidium spp. Cytometry A 73:44–47 5:43–51 17. Barbosa J, Costa-de-Oliveira S, Rodrigues AG, Pina-Vaz C (2008) 8. Maschmeyer G, Beinert T, Buchheidt D, Cornely OA, Einsele H, Optimization of a flow cytometry protocol for detection and Heinz W, Heussel CP, Kahl C, Kiehl M, Lorenz J, Hof H, viability assessment of Giardia lamblia. Travel Med Infect Dis Mattiuzzi G (2009) Diagnosis and antimicrobial therapy of lung 6:234–239 infiltrates in febrile neutropenic patients: guidelines of the 18. Barbosa J, Rodrigues AG, Pina-Vaz C (2009) Cytometric infectious diseases working party of the German Society of approach for detection of Encephalitozoon intestinalis,an Haematology and Oncology. Eur J Cancer 45:2462–2472 emergent agent. Clin Vaccine Immunol 16:1021–1024 9. Morris A, Netravali M, Kling HM, Shipley T, Ross T, Sciurba FC, 19. Libertin CR, Woloschak GE, Wilson WR, Smith TF (1984) Norris KA (2008) Relationship of pneumocystis antibody Analysis of Pneumocystis carinii cysts with a fluorescence- response to severity of chronic obstructive pulmonary disease. activated cell sorter. J Clin Microbiol 20:877–880 Clin Infect Dis 47:e64–e68 20. Lapinsky SE, Glencross D, Car NG, Kallenbach JM, Zwi S (1991) 10. Nevez G, Chabé M, Rabodonirina M, Virmaux M, Dei-Cas E, Quantification and assessment of viability of Pneumocystis carinii Hauser PM, Totet A (2008) Nosocomial Pneumocystis jirovecii organisms by flow cytometry. J Clin Microbiol 29:911–915 infections. Parasite 15:359–365 21. D’Avignon LC, Schofield CM, Hospenthal DR (2008) Pneumo- 11. Procop GW, Haddad S, Quinn J, Wilson ML, Henshaw NG, cystis pneumonia. Semin Respir Crit Care Med 29:132–140 Reller LB, Artymyshyn RL, Katanik MT, Weinstein MP (2004) 22. Morris A, Wei K, Afshar K, Huang L (2008) Epidemiology and Detection of Pneumocystis jiroveci in respiratory specimens by clinical significance of Pneumocystis colonization. J Infect Dis four staining methods. J Clin Microbiol 42:3333–3335 197:10–17