Definitive diagnosis of Pneumocystis jirovecii pneumonia (PCP) can be difficult due to the inability to culture the organism and the need to visualize the organism on microscopic stains of respiratory secretions [1,2]. Measurement of blood levels of (1→3)-β-D-glucan (beta-glucan), which is a component of the cell wall of P. jirovecii and other fungi, has emerged as a potentially useful diagnostic tool in immunocompromised patients, including those with HIV [3–8].
We previously reported the performance characteristics of plasma beta-glucan in HIV-positive participants with a broad range of opportunistic infections . In that study of patients with advanced immunosuppression and diverse opportunistic infections, plasma beta-glucan was strongly associated with the diagnosis of PCP. However, in clinical practice, beta-glucan testing would likely have the greatest utility in patients with respiratory symptoms and suspected PCP, not as a screening test in all patients with acute opportunistic infections. As a result, here, we describe the test characteristics of beta-glucan testing only in study participants who presented with respiratory symptoms.
Materials and methods
The study population was participants in ACTG A5164, a strategy study of early versus deferred antiretroviral therapy (ART) in patients with acute AIDS-related opportunistic infections. Eligible HIV-related opportunistic infections for study enrolment specifically excluded tuberculosis (due to potential drug–drug interactions with ART) and opportunistic infections requiring ART for treatment (e.g. progressive multifocal leukoencephalopathy, AIDS-related dementia and cryptosporidiosis). Oral and oesophageal candidiasis were not eligible opportunistic infections for inclusion on their own, but investigators were required to note the diagnosis if present on entry. Study participants were required to be able to take oral medications and to provide written informed consent for participation.
Two study investigators independently adjudicated the diagnosis of PCP after reviewing reports from study sites. Participants could be enrolled with either ‘confirmed’ or ‘probable’ PCP. The study defined confirmed PCP as a history (within 3 months) of shortness of breath, dyspnoea on exertion, cough or fever, along with histological or cytological evidence of P. jirovecii in a bronchoalveolar lavage (BAL), lung biopsy or sputum specimen. Probable PCP was a history (within 3 months) of shortness of breath, dyspnoea on exertion, cough or fever, along with abnormal chest radiograph or computed tomographic (CT) scan or hypoxia based on arterial blood gas partial pressure of oxygen less than 80 mmHg or alveolar-arterial oxygen difference greater than 15 mmHg on room air, and specific antipneumocystis therapy initiation. The primary analysis showed no significant difference or trend in difference in beta-glucan results between confirmed versus probable PCP ; so, these two categories were considered together for the present analysis.
Respiratory symptoms were identified by investigators from a list of all signs and symptoms with an onset or resolution between 21 days prior to study entry and 14 days following study entry. Beta-glucan was categorized as positive if at least 80 pg/ml or negative if less than 80 pg/ml. Values of less than 31 pg/ml were censored at 31 pg/ml and values of more than 500 pg/ml were censored at 500 pg/ml. Sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) of beta-glucan for diagnosis of PCP among patients with respiratory symptoms were calculated on the basis of the results.
Among the 282 participants in A5164, 252 (89%) had a baseline sample yielding an analysable beta-glucan result. The participants had advanced HIV-related immunosuppression, with a median CD4 lymphocyte count of 26 cells/μl [interquartile range (IQR), 10–53 cells/μl] and a plasma HIV RNA level of 5.02 log copies/ml (IQR, 4.69–5.60 log copies/ml). The most common qualifying opportunistic infection was PCP (69%), followed by cryptococcal meningitis (14%) and bacterial pneumonia (9%). One hundred and twelve participants (44%) had concomitant oral or oroesophageal candidiasis with a qualifying opportunistic infection. Full baseline characteristics of the study population have been previously described .
Of 252 study participants with beta-glucan assay results, 159 (63%) reported at least one baseline respiratory symptom; the most common were dyspnoea or shortness of breath (52.1%), cough or sputum production (44.4%) and pleuritic chest pain (3.6%). One hundred and thirty-nine of 173 PCP cases (80%) had at least one baseline respiratory symptom.
One hundred and twenty-nine of the 139 participants with respiratory symptoms and PCP had beta-glucan at least 80 pg/ml, for a sensitivity of 92.8% [95% confidence interval (CI) 87.2–96.5] (Table 1). Among the 20 participants with respiratory symptoms without PCP, 15 had a negative beta-glucan, for a specificity of 75.0% (95% CI 50.9–91.3). The 20 participants without PCP had the following diagnoses: 11 had bacterial pneumonia, including one with pneumonia secondary to Nocardia; seven had cryptococcal meningitis, including one with concomitant pulmonary cryptococcosis; one had toxoplasmosis; one had ‘bacterial infection of deep tissue;’ and one had disseminated histoplasmosis, including pulmonary disease. The five participants with positive beta-glucan, respiratory symptoms and no PCP included the participant with disseminated histoplasmosis and four with bacterial pneumonia (not Nocardia).
Among the 134 study participants with a baseline respiratory symptom and beta-glucan at least 80 pg/ml, 129 had PCP, for a PPV of 96.3% (95% CI 91.5–98.8). Conversely, among the 25 participants with a beta-glucan result less than 80 pg/ml, 15 did not have PCP, yielding an NPV of 60% (95% CI 38.7–78.9).
The median beta-glucan value among all participants with respiratory symptoms and PCP was 405 pg/ml (IQR, 209–500 pg/ml), and among those without PCP was 40.5 pg/ml (IQR, 31–84.5 pg/ml) (Fig. 1). Sixty of 62 participants (97%) with beta-glucan more than 500 pg/ml and respiratory symptoms had PCP. Thirty-four participants had PCP with no respiratory symptoms; among these participants, beta-glucan was positive in 31 (91.2%) and negative in three.
In this study, we demonstrate that a blood test for beta-glucan has a high PPV (96.3%) for PCP in AIDS patients with respiratory symptoms when 80 pg/ml is used as the cutoff for positive. This is higher than the PPV we previously reported (85%) when examining the performance of beta-glucan in patients presenting with a wider range of opportunistic infections . The PPV in the current analysis is higher than in the previously published study because of the higher pretest probability of PCP in AIDS patients presenting with respiratory symptoms. In addition, certain fungal infections that may yield a positive beta-glucan (such as histoplasmosis or, less commonly, cryptococcosis) are not primarily respiratory pathogens in patients with AIDS.
An inevitable limitation of the beta-glucan test is that it is not specific to P. jirovecii, and hence will have a lower PPV than the gold standard for diagnosis, cytologic staining of respiratory secretions. Furthermore, false-positive results can occur with the administration of certain agents that are filtered through cellulose filtres, such as albumin or immunoglobulin. False-positive results may also be caused by haemodialysis if cellulose membranes are used, by serosal exposure to cellulose gauze or possibly by certain antimicrobials [9,10]. Therefore, careful consideration of potential confounding factors is important when ordering the test.
Improved tests for PCP are needed because the diagnosis has many inherent challenges; most importantly, the organism cannot be cultured, so the diagnosis depends on examination of respiratory secretions [1,2]. The sensitivity of induced sputum examinations for PCP is highly variable and depends on patient and laboratory factors . Although bronchoscopy and lung biopsy are more sensitive, they are invasive. Other noninvasive markers, such as lactate dehydrogenase, have even lower specificity, and PCR is not readily available [11,12]. As a result, beta-glucan can provide a useful, noninvasive adjunctive marker.
In this study, 97% of participants with beta-glucan more than 500 pg/ml and respiratory symptoms had PCP. There were two participants with respiratory symptoms, beta-glucan more than 500 pg/ml and no diagnosis of PCP. On review, one of these patients was treated with trimethoprim-sulfamethoxazole and prednisone, yet the diagnosis entered in the case report form was bacterial pneumonia. It is possible that this patient actually had PCP and was miscoded. The other patient received piperacillin-tazobactam as part of treatment for bacterial pneumonia; data are conflicting as to whether antimicrobials such as piperacillin-tazobactam cause false elevations of beta-glucan, so it is unclear whether antibiotic administration led to the elevated result in this case [9,13,14]. There was also a patient who had respiratory symptoms and beta-glucan of 411 pg/ml but no diagnosis of PCP; this patient was diagnosed with disseminated histoplasmosis.
Forty-four percent of the study participants had concomitant oral and/or oroesophageal candidiasis (candidiasis alone was not an eligible study criterion, so all participants with candidiasis also had a concomitant opportunistic infection, the most common being PCP). As we reported in our original analysis of beta-glucan in A5164, the relationship between presence of candidiasis and elevated beta-glucan was not statistically significant . However, given the high rate of PCP among those with mucosal candidiasis, candidiasis is an important clinical marker of high PCP risk. High rates of PCP in the setting of candidiasis have been noted since early in the AIDS era .
We propose the following algorithm for use of blood beta-glucan as a diagnostic tool for PCP on the basis of the pretest probability in an AIDS patient with respiratory symptoms (Fig. 2): For a patient with low clinical suspicion for PCP, such as a patient with high CD4 cell count, acute onset of symptoms, focal or lobar infiltrate on radiograph or absence of fever, cough or dyspnoea, we recommend first treating more likely alternate diagnoses, and then checking the beta-glucan if there is a lack of clinical improvement. If positive, we would then initiate treatment and confirmatory testing for PCP [induced sputum and, if negative, BAL), particularly if the beta-glucan result was significantly elevated; the median result in our study for patients with positive beta-glucan, respiratory symptoms and PCP was 451 pg/ml.
For a patient with moderate to high pretest probability of PCP, such as a patient with CD4 cell count less than 200 cells/μl, subacute onset of fever, dyspnoea, cough or hypoxia, with diffuse interstitial infiltrates on chest radiograph or a combination of these factors, we recommend initiating empiric PCP treatment and checking beta-glucan as a way of all but confirming the diagnosis. In such patients, a beta-glucan result of more than 500 pg/ml would arguably obviate the need for cytologic confirmation of PCP. An intermediate result in this setting would necessitate confirmatory testing for PCP and a negative result should trigger additional work-up for alternate diagnoses.
An important caveat to any diagnostic strategy for patients with advanced HIV-related immunosuppression is that multiple infections may be present simultaneously. For example, in highly histoplasmosis-endemic areas, a high beta-glucan level could be triggered by PCP, histoplasmosis or concomitant infection with both pathogens. In this setting, supplementing beta-glucan with urine histoplasmosis antigen testing would help clarify the diagnosis. Furthermore, the beta-glucan test has not been thoroughly evaluated in less common fungal infections, such as coccidioidomycosis, blastomycosis, paracoccidioidomycosis and penicilliosis, so it should be used in conjunction with other diagnostic modalities in patients at risk for those infections. One study does describe a high rate of positivity in coccidioidomycosis, whereas another found no positivity in blastomycosis, though it was limited by a small sample size [16,17].
Limitations to our study include the lack of a consistent protocol for PCP diagnosis between study sites, leading to variable strategies (some sites were more likely to confirm the diagnosis than others) and the possible miscoding of one case that in hindsight probably did have PCP. In addition, we do not know how the test would perform in settings in which PCP is a less common opportunistic infection; this limits our ability to predict how the test would perform in a group with lower pretest probability. Another possible limitation is that A5164 excluded patients with tuberculosis; although tuberculosis is not known to raise the beta-glucan level, this may have excluded some patients with advanced immunosuppression and other concomitant infections, including fungal infections.
An important potential area of future research would be to conduct a similar analysis in resource-limited settings, wherein availability of definitive PCP diagnostics is limited and often relies on clinical parameters . Challenges to implementing the beta-glucan test in such settings will be the need for refrigeration of samples, equipment (such as a computer and an incubating plate reader) and a trained technician, as well as cost of the test. Currently, the Fungitell kit, as used in this study, permits testing of up to 42 patient samples in duplicate, per kit, and costs approximately $1229 (Associates of Cape Cod, personal communication).
In summary, we found that in patients with advanced HIV-related immunosuppression and respiratory symptoms, a positive beta-glucan test had a high predictive value for the diagnosis of PCP. In particular, values of more than 500 pg/ml were strongly suggestive of the diagnosis. We believe that, if available, blood beta-glucan testing can provide helpful adjunctive diagnostic information for work-up of respiratory symptoms in this population.
Beta-glucan assays were performed by Beacon Diagnostics Laboratory, a unit of Associates of Cape Cod, East Falmouth, Massachusetts, USA. The content in this article is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute of Allergy and Infectious Diseases or the National Institutes of Health.
This work was supported by the National Institute of Allergy and Infectious Diseases, National Institutes of Health (5U01AI069472-05, for the ACTG and 1 U01 AI068634 for the Statistical and Data Management Center for the ACTG); beta-glucan testing was provided by Associates of Cape Cod, Inc.
Conflicts of interest
There are no conflicts of interest.
1. Baughman RP, Liming JD. Diagnostic strategies in Pneumocystis carinii pneumonia
. Front Biosci
2. Kovacs JA, Gill VJ, Meshnick S, Masur H. New insights into transmission, diagnosis, and drug treatment of Pneumocystis carinii pneumonia
3. Fuji T, Nakamura T, Iwamoto A. Pneumocystis pneumonia in patients with HIV infection: clinical manifestations, laboratory findings, and radiological features
. J Infect Chemother
4. Nakamura H, Tateyama M, Tasato D, Haranaga S, Yara S, Higa F, et al. Clinical utility of serum beta-D-glucan and KL-6 levels in Pneumocystis jirovecii pneumonia
. Intern Med
5. Onishi A, Sugiyama D, Kogata Y, Saegusa J, Sugimoto T, Kawano S, et al. Diagnostic accuracy of serum 1,3-(-D-glucan for pneumocystis jiroveci pneumonia, invasive candidiasis, and invasive aspergillosis: systematic review and meta-analysis
. J Clin Microbiol
6. Sax PE, Komarow L, Finkelman MA, Grant PM, Andersen J, Scully E, et al. AIDS Clinical Trials Group Study A5164 TeamBlood (1–>3)-beta-D-glucan as a diagnostic test for HIV-related Pneumocystis jirovecii pneumonia
. Clin Infect Dis
7. Tasaka S, Hasegawa N, Kobayashi S, Yamada W, Nishimura T, Takeuchi T, et al. Serum indicators for the diagnosis of pneumocystis pneumonia
8. Watanabe T, Yasuoka A, Tanuma J, Yazaki H, Honda H, Tsukada K, et al. Serum (1(3) beta-D-glucan as a noninvasive adjunct marker for the diagnosis of Pneumocystis pneumonia in patients with AIDS
. Clin Infect Dis
9. Marty FM, Koo S. Role of (1(3)-beta-D-glucan in the diagnosis of invasive aspergillosis
. Med Mycol
2009; 47 (Suppl 1):S233–S240.
10. Mennink-Kersten MA, Warris A, Verweij PE. 1,3-Beta-d-glucan in patients receiving intravenous amoxicillin-clavulanic acid [letter]
. N Engl J Med
11. Cruciani M, Marcati P, Malena M, Bosco O, Serpelloni G, Mengoli C. Meta-analysis of diagnostic procedures for Pneumocystis carinii pneumonia in HIV-1 infected patients
. Eur Respir J
12. Vogel MN, Weissgerber P, Goeppert B, Hetzel J, Vatlach M, Claussen CD, Horger M. Accuracy of serum LDH elevation for the diagnosis of Pneumocystis jiroveci pneumonia
. Swiss Med Wkly
13. Marty FM, Lowry CM, Lempitski SJ, Kubiak DW, Finkelman MA, Baden LR. Reactivity of (1→3)-beta-d-glucan assay with commonly used intravenous antimicrobials
. Antimicrob Agents Chemother
14. Metan G, Agkus C, Nedret Koc A, Elmali F, Finkelman MA. Does ampicillin-sulbactam cause false-positivity of (1,3)-beta-D-glucan assay? A prospective evaluation of 15 patients without invasive fungal infections
15. Phair J, Munoz A, Detels R, Kaslow R, Rinaldo C, Saah A. The risk of Pneumocystis carinii pneumonia among men infected with human immunodeficiency virus type 1. Multicenter AIDS Cohort Study Group
. N Engl J Med
16. Thompson GR, Bays DJ, Johnson SM, Cohen SH, Pappagianis D, Finkelman MA. Serum (1→3)-β-D-glucan measurement in coccidioidomycosis
. J Clinc Microbiol
17. Girouard G, Lachance C, Pelletier R. Observations on (1–3)-beta-D-glucan detection as a diagnostic tool in endemic mycosis caused by Histoplasma or Blastomyces
. J Med Microbiol
18. de Armas RY, Wissman G, Müller AL, Pederiva MA, Brum MC, Brackman RL, et al. Pneumocystis jirovecii pneumonia in developing countries