SEARCH

SEARCH BY CITATION

Keywords:

  • β-d-Glucan;
  • Pneumocystis jirovecii pneumonia;
  • prognosis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

Clin Microbiol Infect 2011; 17: 595–602

Abstract

Pneumocystis jirovecii (carinii) pneumonia (PJP) is a major cause of disease in immunocompromised individuals. However, until recently no reliable and specific serological parameters for the diagnosis of PJP have been available. (1[RIGHTWARDS ARROW]3)-β-d-Glucan (BG) is a cell wall component of P. jirovecii and of various other fungi. Data from the past few years have pointed to serum measurement of BG as a promising new tool for the diagnosis of PJP. We therefore conducted a retrospective study on 50 patients with PJP and 50 immunocompromised control patients to evaluate the diagnostic performance of serum BG measurement. Our results show an excellent diagnostic performance with a sensitivity of 98.0% and a specificity of 94%. While the positive predictive value was only 64.7%, the negative predictive value was 99.8% and therefore a negative BG result almost rules out PJP. BG levels were already strongly elevated in an average of 5 days and up to 21 days before microbiological diagnosis demonstrating that the diagnosis could have been confirmed earlier. BG levels at diagnosis and maximum BG levels during follow-up did not correlate with the outcome of patients or with the P. jirovecii burden in the lung as detected by Real-Time PCR. Therefore, absolute BG levels seem to be of no prognostic value. Altogether, BG is a reliable parameter for the diagnosis of PJP and could be used as a preliminary test for patients at risk before a bronchoalveolar lavage is performed.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

Pneumocystis jirovecii (carinii) pneumonia (PJP) is a major cause of mortality in immunocompromised individuals. Therefore, early diagnosis and therapy is vital.

Currently, the laboratory reference standard for the detection of P. jirovecii is staining of the organism with monoclonal antibodies, followed by immunofluorescence microscopy in bronchoalveolar lavage (BAL) fluid. Together with a compatible clinical presentation and confirmation by computed tomography, this allows the diagnosis to be made. However, diagnosis of PJP is hampered by the fact that the performance of BAL is not always possible, owing to the limited respiratory function of some patients. Furthermore, detection of P. jirovecii is not necessarily equivalent to infection, because of the existence of clinically insignificant P. jirovecii colonization [1,2]. Until recently, no reliable serological parameters for the diagnosis of PJP were available.

(1[RIGHTWARDS ARROW]3)-β-d-Glucan (BG) is a cell wall component of P. jirovecii and of various other fungi. During the course of an invasive infection, BG is released into the serum. Data collected over recent years have shown that serum BG measurement might be a powerful new tool for the diagnosis of PJP (Table 1). However, in these studies various test systems with different characteristics have been used, and direct comparison of the reported observations is not possible. The Fungitell assay is a Limulus amebocyte lysate-based test for the measurement of BG. So far, there have been only two retrospective studies with sizeable numbers of patients (20 and 28, respectively) and a relevant control group that have used the Fungitell assay for the diagnosis of PJP [3,4]. Furthermore, information on the prognostic benefit of BG measurement is limited, and results are inconclusive [5–8].

Table 1.   Published studies investigating (1[RIGHTWARDS ARROW]3)-β-d-glucan (BG) levels of patients with Pneumocystis jirovecii pneumonia
Author, year [Reference]Number of PJP patientsImmunosuppression of the PJP groupMedian BG level (pg/mL)Number of control patientsImmunosuppression of the control groupSensitivitySpecificityReference methodTest systemCut-off (pg/mL)
  1. CTD, connective tissue disease; HIV, human immunodeficiency virus; IFI, invasive fungal infection; IFS, specific immunofluorescence staining; PJP, P. jirovecii pneumonia; NS, not specified.

  2. Only studies with ten or more patients are shown. Three of these studies did not include a negative control group [5,19,23], and another four studies had a control group without immunosuppression or without specifications [7,8,22,24]. One study used clinical presentation only as diagnostic method, without confirming the diagnosis by IFS or PCR [22]. Both Fungitell and Fungitec G are Limulus amebocyte lysate-based tests. However, they use extracts from different horseshoe crab species (Fungitell, Limulus polyphemus; Fungitec G, Tachypleus tridentatus), and the results of the two tests cannot be compared directly, owing to the different reactivities of their factor G proteins.

Desmet, 2009 [3]16 12HIV Non-HIV1496 377916 12HIV Non-HIV10096.4PCR + IFS/microscopyFungitell100
Watanabe, 2009 [8]111HIV175425NS96.487.8MicroscopyFungitec G23.2
Del Bono, 2009 [22]16Various42315 11Various NoneClinical diagnosisFungitell80
Nakamura, 2009 [7]19 16HIV Non-HIV300 85.424Bacterial pneumonia100 88NSMicroscopy + PCRFungitec G20
Persat, 2008 [4]16 4HIV Non-HIV94540 120None Risk of IFI100NSMicroscopyFungitell80
Marty, 2007 [23]16Various>500None93.8IFSFungitellNS
Tasaka, 2007 [24]57VariousNS222NS92.386.1Microscopyβ-Glucan test (WAKO)31.1
Fujii, 2007 [19]28HIV147None96.8NSNS5
Iikuni, 2006 [6]21CTD87.945Various77.876.9PCRNSNS
Shimizu, 2005 [5]15CTDNSNone86.7Microscopy and/or PCRFungitec G20

We therefore conducted a retrospective study on 50 patients with PJP and 50 control patients to evaluate the diagnostic performance of serum BG measurement, with a special focus on its prognostic relevance.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

PJP patients

During the study period (January 2002 to January 2010), 2287 BAL samples from 1488 patients were examined for P. jirovecii (2158 by immunostaining; 129 by nested PCR). Two hundred and sixty-five analyses on samples from 150 patients (10.1%) gave positive results. Seventy-eight of these P. jirovecii-positive patients had an archived serum sample around the day of the bronchoscopy (±7 days). Fifty of them (34 with positive immunostaining and positive PCR findings, 12 with positive immunostaining only and four with positive PCR findings only) showed a clinical picture typical of PJP, and were not subject to any of the exclusion criteria. Clinical presentation was considered to be typical if pulmonary infiltrates compatible with PJP were present and if at least four of the following criteria were met: existing immunosuppression, fever, dyspnoea, cough, elevated lactate dehydrogenase (LDH) level and hypoxia. Exclusion criteria were culture of any fungus from relevant materials, positive serum galactomannan assay (Platelia Aspergillus EIA; Bio-Rad Laboratories, GmbH, Munich, Germany) or positive serum Candida antigen assay (Cand-Tec; Ramco Laboratories, Inc., Stafford, TX, USA). Colonization with yeast or superficial candidiasis was not considered to be an exclusion criterion, because BG levels are not normally elevated in those patients [9,10].

Quantitative real-time PCR of the four patients who tested positive in PCR only showed very high copy numbers (two with >106 copies/mL; one with 105–106 copies/mL; and one with 104–105 copies/mL). Given the clinical findings, it seemed unlikely that PCR detected only colonization. These 50 patients were included in the PJP group (Table 2). Further analysis of the PJP group showed that 45 patients were admitted to the hospital with acute PJP. The remaining five patients had already been in hospital when they developed PJP. The reason for their initial admittance was intestinal graft-versus-host disease after haematopoietic stem cell transplantation, chemotherapy for B-cell non-Hodgkin lymphoma, methotrexate and corticosteroid therapy for pemphigus vulgaris, and heart transplantation (two patients). All 50 patients were hospitalized after diagnosis of PJP for intravenous antibiotic treatment.

Table 2.   Baseline characteristics of the study populations and main results
 PJP groupControl group
  1. BG, (1[RIGHTWARDS ARROW]3)-β-d-glucan; HIV, human immunodeficiency virus; HSCT, haematopoietic stem cell transplantation; HTX, heart transplantation; IQR, interquartile range; KTX, kidney transplantation; LTX, lung transplantation; OTX, organ transplantation; PJP, Pneumocystis jirovecii pneumonia.

  2. The PJP group contains a considerably higher number of HIV-positive patients than the control group. This is because the suspected diagnosis of PJP proved to be true in most cases, and therefore HIV patients with a negative result for P. jirovecii-specific immunofluorescence staining were rare.

No. of patients5050
Mean age (years) (range)51 (4–74)56.5 (6–83)
Sex (male/female)33/1725/25
Underlying disease (no. of patients)
 HIV171
 Haematological malignancies1018
 OTX (KTX/HTX/LTX)11 (6/5/0)6 (2/0/4)
 Immunological disorder510
 Haematological malignancies + HSCT412
 Solid tumours33
Mean CD4+ T-cells of HIV patients (cells/μL) (range)37 (3–139)120
Median BGDIAGNOSIS (pg/mL) (IQR, range)823 (461–4870, 31–38 400)27 (8–48, 8–273)
Median BGMAX (pg/mL) (IQR, range)1153 (461–7444, 31–38 400)
Interpretation of BG test
 Positive (>85 pg/mL)493
 Negative (≤85 pg/mL)147
Sensitivity (%) (95% CI)98.0 (89.3–99.7)
Specificity (%) (95% CI)94.0 (83.4–98.7)
Positive predictive value (%) (95% CI)64.7 (36.6–86.8)
Negative predictive value (%) (95% CI)99.8 (95.2–99.0)
Likelihood ratio, positive (95% CI)16.33 (15.1–17.7)
Likelihood ratio, negative (95% CI)0.021 (0.002–0.2)

Endpoints of the analysis were death from all causes or discharge. The median length of hospitalization was 25 days (interquartile range (IQR) 16–34). The median duration from admission to microbiological diagnosis was 5 days (IQR 2–10). The median duration from microbiological diagnosis to discharge was 21 days (IQR 12–29). Survivors (n = 35) had a median follow-up period of 29 days (IQR 22–43).

Controls

The control group consisted of all patients, between December 2008 and January 2010, whose BAL fluids were negative for P. jirovecii in immunostaining (n = 312) and for whom an archived serum sample taken around the day of the bronchoscopy (±7 days) existed (n = 86). Only control patients who were immunocompromised and who had a suspected infectious respiratory disease (pulmonary infiltrates, clinical symptoms, elevated leukocyte count and/or elevated C-reactive protein (CRP)) were included. Exclusion criteria were identical to those for the study group. The collection was stopped when 50 patients had been included (Table 2).

Collection of clinical data

Patient demographics and clinical characteristics were collected, including age, sex, underlying disease, type of immunosuppression, CRP level, leukocyte count, CD4 cell count, LDH level, PaO2, albumin level, creatinine level, alanine transaminase level, aspartate transaminase level, γ-glutamyltransferase level, PJP prophylaxis and microbiological results. The therapy was reviewed for possible confounding factors for BG measurement, including intravenous immunoglobulin and albumin.

Serum collection and BG measurement

The sera retrospectively tested were drawn around the time of microbiological diagnosis and, if available, until discharge. One or more subsequent sera were available from 34 of the 50 P. jirovecii-positive patients. Serum samples were originally taken for various microbiological analyses other than BG, and were routinely frozen at −80°C. The use of these sera was approved by the local ethics committee (application number 105/09). Serum samples were examined for the presence of BG with the Fungitell assay (Associates of Cape Cod, Inc., East Falmouth, MA, USA). The test was performed at our own institution, according to the manufacturer’s recommendations. Each serum was tested in duplicate. The persons who tested the sera were not blinded. Samples with BG levels above 500 pg/mL were diluted and retested. BG levels below 31 pg/mL (lower validation limit) were calculated by extrapolation.

Indirect immunofluorescence staining

Monoclonal antibody staining for P. jirovecii was performed with the DETECT IF test (Axis Shield Diagnostics Limited, Dundee, UK) according to the manufacturer’s recommendations.

Touchdown PCR

DNA from BAL fluids was isolated by proteinase K digestion followed by phenol–chloroform extraction, and serum DNA was extracted with the QIAmp DNA Mini Kit (Qiagen GmbH, Hilden, Germany). Touchdown PCR was performed as previously described [11]. All amplification products were sequenced and confirmed to be part of the P. jirovecii mitochondrial large-subunit rRNA gene.

Real-time PCR

On the basis of a previously published protocol [12], quantitative real-time PCR was performed on DNA preparations from serum and BAL samples. Briefly, 5-μL aliquots were used as template DNA for subsequent PCR testing on a LightCycler (Roche Diagnostics, Mannheim, Germany). Samples positive for the specific amplicons were identified by the PCR instrument at the cycle number where the individual fluorescence value exceeded that measured for background. The quantitative interpretation of the results was assisted by a set of external standards that were tested in parallel.

Statistical methods

Statistical analysis was performed using SPSS, version 17.0. Unless otherwise stated, BG levels are expressed as median concentration with IQR. For comparison of variables, Pearson’s chi-square test, the Mann–Whitney U-test or the Kruskal–Wallis test was used. Differences were considered significant for p <0.05. Receiver operating characteristic (ROC) analysis was carried out using MedCalc, version 10.0. The optimal BG cut-off was determined with the maximum Youden index.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

BAL samples from 1488 patients were examined between January 2002 and January 2010 for the presence of P. jirovecii. Fifty P. jirovecii-positive patients met the clinical inclusion criteria and were enrolled in the PJP group. Between December 2008 and January 2010, examination for P. jirovecii in BAL fluid gave negative findings in 312 patients. Of these, 50 immunosuppressed patients with pneumonia were selected as controls (Table 2).

Serum BG measurement as an indicator for P. jirovecii pneumonia

Because the BG cut-off recommended by the manufacturer (80 pg/mL) was originally designed for the diagnosis of invasive fungal infections other than PJP, we first performed a ROC analysis on our data. According to the ROC analysis, the optimal BG cut-off for the diagnosis of PJP, determined with the maximum Youden index, would be >85 pg/mL. The area under the ROC curve was 0.987 (95% CI 0.941–0.998; p 0.0001). With this optimized cut-off, 49 of 50 patients in the PJP group tested positive for serum BG at the time of diagnosis. The negative patient had a serum BG level of 31 pg/mL. The median BG concentration in the PJP group at the time of diagnosis was 823 pg/mL (IQR 461–4870, range 31–38 400). The median highest BG level during hospitalization was 1153 pg/mL (IQR 461–7444, range 31–38 400). In the control group, three patients tested positive (Table 3) and 47 patients were negative for serum BG. The median BG concentration in the control group was 27 pg/mL (IQR 8–48), ranging from non-detectable to 273 pg/mL. The difference in BG levels between the PJP group and the control group was highly significant (p <0.001). On the basis of these results, the sensitivity of the BG assay for the diagnosis of PJP was 98.0% (95% CI 89.3–99.7%) and the specificity was 94.0% (95% CI 83.4–98.7%). The positive predictive value and negative predictive value were 64.7% (95% CI 36.6–86.8%) and 99.8% (95% CI 95.2–99.0%), respectively.

Table 3.   Characteristics of false-negative and false-positive patients
 False-negative patientFalse-positive patient 1False-positive patient 2False-positive patient 3
  1. ALL, acute lymphoblastic leukaemia; ALT, alanine transaminase; AML, acute myeloid leukaemia; AST, aspartate transaminase; BAL, bronchoalveolar lavage; BG, (1[RIGHTWARDS ARROW]3)-β-d-glucan; CMV, cytomegalovirus; GGT, γ-glutamyltransferase; HSCT, haematopoietic stem cell transplantation; IFS, specific immunofluorescence staining; RSV, respiratory syncytial virus.

  2. The false-negative patient had a proven PJP by standard criteria. Despite high-dose trimethoprim–sulphamethoxazole therapy, the condition of the patient deteriorated, and she died 6 days after admission to the intensive-care unit. In the post-mortem lung biopsy specimen, P. jirovecii could still be detected by immunostaining. The reason for the negative BG result is unclear. False-positive patient 2 had a positive blood culture for Pseudomonas aeruginosa 11 days after the serum for BG testing was drawn. However, blood cultures taken on the day of serum sampling and 1 day and 2 days after serum sampling were negative for Pseudomonas aeruginosa. Although Pseudomonas aeruginosa is described as a source of false-positive BG results [25], it seems unlikely that, in this patient, the delayed bacteraemia was responsible for the elevated BG level. Another possible confounding factor in this patient was dialysis. However, the dialysis membranes used at our hospital were tested, and dialysis had no significant effect on BG levels. False-positive patient 3 had a mucositis, and it is possible that BG from the gastrointestinal tract was entering the bloodstream.

Age (years)37707311
SexFemaleFemaleMaleFemale
Underlying diseaseArthritis under corticosteroid therapyAML with HSCTAML with HSCTALL
Detection of Pneumocystis jiroveciiIFS + PCRNegativeNegativeNegative
BGDIAGNOSIS31273177145
Confounding factorsNoneNoneDialysisNone
Other pathogen in BAL fluidNoNoPseudomonas aeruginosaRSV
Bacteraemia/viraemiaCMVNoPseudomonas aeruginosaNo
MucositisNoNoNoYes
Creatinine (mg/L)0.232.702.920.85
ASTDIAGNOSIS (U/L)Not tested4735554
ALTDIAGNOSIS (U/l)Not tested5820199
GGTDIAGNOSIS (U/L)Not tested218Not tested124

BG measurement for earlier confirmation of diagnosis

Sera from 36 patients were available in the 3 weeks prior to diagnosis. The median time of serum sampling before diagnosis was 7 days. Thirty-four of these patients (94%) already had highly elevated BG levels (792 pg/mL, IQR 316–3557, range 122–14 220) at an average of 5 days (IQR 1.8–9.3, range 1–21) before microbiological diagnosis. As expected, the shorter the time interval between sample date and diagnosis date, the higher the serum BG levels. Within the last 10 days before diagnosis, the median BG level was 831 pg/mL (n = 26, IQR 541–4202, range 122–14 220). In the period between day 10 and day 21 before diagnosis, the median BG concentration was 269 pg/mL (n = 8, IQR 154–656, range 127–6460). For the two patients who had negative results, serum samples were drawn at day 19 and day 20 before diagnosis.

Correlation of BG levels with clinical outcome

To determine the prognostic value of BG levels at diagnosis (BGDIAGNOSIS) and of maximum BG levels during hospitalization (BGMAX), the patients were divided into groups based on the median, the tertile and the quartile BG levels. The cumulative mortality of the respective groups was then compared. However, no statistically significant differences in mortality were found between the groups (Table 4).

Table 4.   Mortality of Pneumocystis jirovecii (carinii) pneumonia patients stratified by median, tertile and quartile (1[RIGHTWARDS ARROW]3)-β-d-glucan (BG) levels
 1st2nd3rd4thp-value
  1. For determination of the prognostic value of BG levels, the patients were divided into groups based on the median, tertile and quartile BG levels. This was performed for the BG levels at diagnosis and for the maximum BG levels during hospitalization. The mortality of the respective groups was then compared with Pearson’s chi-square test. There was no significant difference in mortality, and therefore absolute BG levels do not predict the outcome.

BG measurement at microbiological diagnosis
 Mortality, median (%)28.028.01.000
 Mortality, tertiles (%)31.311.841.20.125
 Mortality, quartiles (%)41.715.433.323.10.482
Maximum BG levels during hospitalization
 Mortality, median (%)28.028.01.000
 Mortality, tertiles (%)31.323.529.40.874
 Mortality, quartiles (%)41.715.430.825.00.524

Because the survivor group was very diverse with respect to clinical course, we subdivided them into patients with rapid improvement after the initiation of therapy (n = 18) and patients with a severe disease course (n = 17). Patients were considered to have a severe disease course if there was no clinical improvement within the first 7 days of therapy, if treatment in an intensive-care unit or mechanical ventilation was necessary, if therapy needed to be continued after 21 days, or if treatment failure was suspected and the antifungal regimen had to be changed. Again, the patients were divided on the basis of the median, tertile and quartile BG levels, and the percentages of patients in the groups with a different clinical course were compared. Subdivision of the study population on the basis of clinical course also did not result in significant differences (data not shown).

It has been reported that a number of confounding factors exist that are capable of causing false-positive BG results. The major candidates are administration of fractionated blood products (e.g. immunoglobulins and albumin) and haemodialysis with certain cellulose membranes [13–18]. We reviewed the treatment protocols of every patient to identify a possible role of such confounding factors that might distort the analysis. None of our patients had been exposed to any of these confounding factors prior to the day on which the first serum sample was taken, but during the time of their treatment, 12 patients were given immunoglobulins and/or albumin. Surprisingly, median BGMAX did not change when patients with possible confounding factors were excluded (n = 38; BGMAX 1153 pg/mL, IQR 461–7040, range 31–38 400), and there was no significant difference in the cumulative mortality between patients of the different groups.

Correlation of BG levels with clinical characteristics (Table 5)

Table 5.   Clinical characteristics of patients with Pneumocystis jirovecii pneumonia
 HIVHaematological malignanciesOrgan transplantationImmunological disordersHaematological malignancies + HSCTSolid tumours
  1. ALT, alanine transaminase; AST, aspartate transaminase; BG, (1[RIGHTWARDS ARROW]3)-β-d-glucan; CRP, C-reactive protein; GGT, γ-glutamyltransferase; HIV, human immunodeficiency virus; HSCT, haematopoietic stem cell transplantation; LDH, lactate dehydrogenase; time periodBG-DIAGNOSIS, time period from first positive BG measurements to microbiological confirmation of diagnosis

  2. Organ transplant recipients had significantly higher BG levels and creatinine levels than the other patients.

Number of patients171011543
Mean age (years) (range)41 (27–59)58 (4–74)62 (39–68)71 (38–76)65 (26–67)55 (33–58)
Sex (male/female)11/67/36/54/12/23/0
Median length of hospitalization (days)282532201316
BGDIAGNOSIS (pg/mL) (range)778 (177–15 430)554 (142–14 220)2928 (481–38 400)507 (31–4820)662 (199–3300)628 (527–7300)
BGMAX (pg/mL) (range)929 (180–31 360)554 (146–25 600)9730 (481–38 400)507 (31–4820)1349 (199–300)628 (527–9000)
Median time periodBG-DIAGNOSIS (days) (range)6 (1–19)8 (1–13)2 (1–9)210 (3–21)6.5 (1–12)
CRP (mg/L) (range)17 (2–227)71 (6–331)32 (6–228)41 (24–425)106 (73–188)17 (9–36)
Leukocyte count (cells/μL) (range)5.60 (1.7–12.9)6.50 (3–98.9)9.55 (3.6–18.3)10.90 (2.2–14.4)4.35 (1.5–12.5)8.00 (4.3–12.0)
CD4+ cells (cells/μL) (range)37 (3–139)151478 (348–677)456 (129–782)
LDH (U/L) (range)374 (173–1186)499 (248–1493)555 (354–970)285 (162–633)240 (177–714)582 (213–687)
PaO2 (mmHg) (range)49.70 (40–86)59.10 (27–72)69.85 (21–88)71.40 (24–76)54.45 (50–59)58.00
CreatinineDIAGNOSIS (mg/dL) (range)0.70 (0.46–1.10)0.80 (0.25–1.20)1.70 (0.50–3.00)1.20 (0.23–2.95)1.13 (0.57–1.70)0.79 (0.75–0.90)
CreatinineMAX (mg/dL) (range)0.90 (0.66–3.00)1.02 (0.25–2.1)2.22 (0.7–4.3)1.27 (0.23–2.95)1.39 (0.60–2.91)0.90 (0.79–1.20)
AST (U/L) (range)44 (22–368)61 (18–3002)58 (25–406)33 (16–66)40 (18–61)36 –
ALT (U/L) (range)31 (12–295)59 (23–964)46 (7–274)32 (22–41)26 (23–100)43 (32–54)
GGT (U/L) (range)64 (14–421)164 (25–347)46 (21–530)66 (58–162)112 (25–279)81 (65–138)

There was no correlation between age, CRP level, leukocyte count, LDH level, Po2, albumin, alanine transaminase level, aspartate transaminase level, γ-glutamyltransferase level and BG level. However, high BG levels correlated with high creatinine levels (p 0.012). In contrast to a previous study [7], our data showed no significant difference in BG levels between human immunodeficiency virus (HIV)-positive patients and non-HIV patients. The group of organ transplant recipients, however, showed significantly higher BGMAX levels than the remainder of the patients (pDIAGNOSIS 0.063; pMAX 0.022). We further stratified this group according to the organ that had been transplanted. The group consisted of six kidney and five heart transplant recipients. There was a tendency for there to be higher BG levels in kidney transplant recipients than in heart transplant recipients (BGDIAGNOSIS 2928 pg/mL, IQR 736–13 208 for kidney transplant recipients vs. 897 pg/mL, IQR 669–21 710 for heart transplant recipients; BGMAX 13 415 pg/mL, IQR 1543–19 455 for kidney transplant recipients vs. 5020 pg/mL, IQR 689–22 430 for heart transplant recipients), but the difference was not statistically significant (pDIAGNOSIS 0.93; pMAX 0.54). Inclusion of other clinical parameters in the analysis showed that organ transplant recipients had higher creatinine levels (p 0.003) and LDH levels (p 0.021) than the other patients.

Correlation of BG levels with DNA levels in BAL fluid and sera as detected by PCR

To evaluate whether the P. jirovecii burden in the lung correlated with the BG levels measured in the serum, available BAL fluids (n = 38) were tested for P. jirovecii DNA by real-time PCR. P. jirovecii DNA could be detected in all BAL samples (>106 copies/mL in 21 patients, 105–106 copies/mL in six patients, 104–105 copies/mL in five patients, 103–104 copies/mL in four patients, and 102–103 copies/mL in two patients). However, there was no significant correlation between the quantitative PCR results and serum BG levels (pRealTimePCR 0.766).

Furthermore, sera of all patients (n = 50) at the time of diagnosis as well as at the time of maximum BG levels were tested for P. jirovecii DNA by touchdown PCR, and selected sera were additionally tested by quantitative real-time PCR. Each PCR system could detect P. jirovecii DNA in a total of three sera (sensitivity of 6%). Two of these sera were positive in both PCR systems, whereas the third serum sample tested positive in one system but not in the other. Neither the levels of BG nor the outcome correlated with serum PCR positivity.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

We have conducted a retrospective study with the Fungitell assay to evaluate the usefulness of serum BG measurement for the diagnosis of PJP. Our results show an excellent diagnostic performance, with a sensitivity of 98% and a specificity of 94%, and thereby confirm the data of previous studies. The positive predictive value was only 64.7%, possibly because BG is influenced by a number of as yet unknown confounding factors, and because BG is elevated not only in PJP but also in most other invasive fungal infections. As a consequence, BG levels alone cannot prove the existence of PJP, and they must not be interpreted without considering the clinical findings. However, with a negative predictive value of 99.8%, a BG level ≤85 pg/mL almost rules out PJP, and this could be extremely helpful in patients who cannot undergo bronchoscopy or in patients where the clinical suspicion of PJP is low.

Usually, there is a significant delay between the onset of symptoms and the diagnosis of PJP. Median intervals are reported to be 21–28 days in HIV-positive patients [19]. In our study, BG levels were already strongly elevated at an average of 5 days before microbiological diagnosis. For the majority of patients, there was only one pre-existing serum available, and the median of serum sampling was at day 7. Therefore, it is unclear exactly when BG levels exceeded the cut-off. The average time-span of 5 days between BG positivity and microbiological diagnosis is most likely an underestimation caused by a lack of samples dating further back. So far, our data suggest that BG levels are elevated up to 21 days before microbiological diagnosis, and timely BG measurement could allow for the earlier initiation of effective treatment (Fig. 1).

image

Figure 1.  (1[RIGHTWARDS ARROW]3)-β-d-Glucan (BG) kinetics of a patient with Pneumocystis jirovecii (carinii) pneumonia in whom timely measurement of BG could have prevented a significant delay in initiation of effective therapy. A 26-year-old female with acute lymphoblastic leukaemia underwent haematopoietic stem cell transplantation in December 2005. On 29 May 2006 (day −5), she presented with fever, dyspnoea and non-productive cough. The patient was on trimethoprim–sulphamethoxazole (SXT) prophylaxis. Chest X-ray showed bilateral infiltrates suggestive of atypical pneumonia. Serum Aspergillus galactomannan testing gave negative findings. Therapy with clarithromycin was initiated but showed no effect. The patient was re-admitted to the hospital on day −3. A computed tomography scan showed progression of the infiltrates, and therapy was changed to meropenem, voriconazole and foscavir. Bronchoscopy was performed on day 0, and P. jirovecii cysts were detected by specific immunofluorescence and PCR. Intravenous trimethoprim–sulphamethoxazole was started, with good treatment response. The subsequent clinical course was uneventful, and the patient was discharged on day 23. Retrospectively, BG levels were already highly elevated (836 pg/mL) on day −5, and a steady decrease in BG levels reflected a good clinical response. The red line indicates the cut-off (85 pg/mL). BAL, bronchoalveolar lavage; CL, clarithromycin; F, foscavir; M, meropenem; Pj, P. jirovecii; V, voriconazole.

Download figure to PowerPoint

Data on the correlation between BG levels and the outcomes of patients are limited, and the results are inconclusive. Two studies reported that high levels of BG are associated with poor prognosis [5,6], whereas two other studies could not find a significant difference in BG levels between survivors and non-survivors [7,8]. The analysis of our data showed no significant difference in the mortality or severity of PJP between patients with high and those with low BG levels. These findings are supported by our observation that there was no correlation between pulmonary P. jirovecii burden, as detected by quantitative real-time PCR, and serum BG levels. However, the results of real-time PCR on BAL fluids are strongly dependent on the quality of the material, and we cannot rule out the possibility that suboptimal sampling affected the data quality.

Although the absolute BG level at diagnosis and the maximum BG level during hospitalization do not seem to have prognostic relevance, this might be different for BG kinetics. It has been shown in individual cases that a favourable outcome can be paralleled by decreasing BG levels [20], an observation that is supported by our own experience. However, further work is clearly necessary to confirm these reports in larger series.

The severity and mortality of PJP, as well as BG levels, have been reported to be lower in HIV-positive patients than in non-HIV patients [19]. We did not observe a significant difference in the BG levels between those two groups of patients. However, subgroup analysis of patients with different underlying diseases showed significantly higher BG levels in organ transplant recipients. At the same time, organ transplant recipients and patients with high BG levels in general showed higher creatinine levels, whereas liver enzyme levels were not elevated. The route of BG elimination is still unknown. In rabbit models, 80% of the organ-associated BG was found in the liver and 10% in the kidney. BG was detected in organs, blood and urine at 23%, 18% and 10%, respectively [21]. Together with the observation that elimination of BG in dialysis patients takes much longer than in healthy subjects, this indicates that renal clearance seems to play some role [15]. However, another study found that the median plasma half-life of BG in dialysis patients was approximately 20 h (range 3.1–181.3 h) and was not affected by renal or hepatic impairment [14]. Therefore, a direct correlation between high BG levels and impaired renal function is hypothetical, and the reason for our observation remains unclear.

In accordance with the findings of previous studies, the diagnostic potential of serum PCR—irrespective of whether touchdown PCR or real-time PCR was used—is poor, and it cannot be recommended for the diagnosis of PJP.

In summary, serum BG levels in patients with PJP are strongly elevated and the negative predictive value is high. Therefore, measurement of BG could be used as a preliminary test for patients with suspected PJP before BAL is performed. Neither the outcome nor the pulmonary P. jirovecii burden seems to correlate with the serum BG level at diagnosis or with the peak BG level. Accordingly, the prognostic value of single BG measurements in PJP is low. BG levels start to increase up to 21 days before microbiological diagnosis, and this could be helpful in reducing the time until adequate treatment is initiated.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

We would like to thank M. Olschewski and D. Huzly for assistance with the statistical analysis, the Department of Virology, University of Freiburg for providing some of the serum samples, E. Rappolt for extensively searching our serum archive, and G. Häcker, V. Bui and F. von Loewenich for critically reviewing the manuscript. Some of the results were presented at the 61st Annual Meeting of the German Society for Hygiene and Microbiology (DGHM), 22 September 2009, in Göttingen, Germany.

Transparency Declaration

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

This work was not specially funded. Some of the results were generated during routine diagnostic activities. No commercial relationships or potential conflicts of interest exist.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References
  • 1
    Ponce CA, Gallo M, Bustamante R, Vargas SL. Pneumocystis colonization is highly prevalent in the autopsied lungs of the general population. Clin Infect Dis 2010; 50: 347353.
  • 2
    Morris A, Wei K, Afshar K, Huang L. Epidemiology and clinical significance of pneumocystis colonization. J Infect Dis 2008; 197: 1017.
  • 3
    Desmet S, Van Wijngaerden E, Maertens J et al. Serum (1-3)-beta-d-glucan as a tool for diagnosis of Pneumocystis jirovecii pneumonia in patients with human immunodeficiency virus infection or hematological malignancy. J Clin Microbiol 2009; 47: 38713874.
  • 4
    Persat F, Ranque S, Derouin F, Michel-Nguyen A, Picot S, Sulahian A. Contribution of the (1[RIGHTWARDS ARROW]3)-beta-D-glucan assay for diagnosis of invasive fungal infections. J Clin Microbiol 2008; 46: 10091013.
  • 5
    Shimizu A, Oka H, Matsuda T, Ozaki S. (1[RIGHTWARDS ARROW]3)-beta-D glucan is a diagnostic and negative prognostic marker for Pneumocystis carinii pneumonia in patients with connective tissue disease. Clin Exp Rheumatol 2005; 23: 678680.
  • 6
    Iikuni N, Kitahama M, Ohta S, Okamoto H, Kamatani N, Nishinarita M. Evaluation of Pneumocystis pneumonia infection risk factors in patients with connective tissue disease. Mod Rheumatol 2006; 16: 282288.
  • 7
    Nakamura H, Tateyama M, Tasato D et al. Clinical utility of serum beta-D-glucan and KL-6 levels in Pneumocystis jirovecii pneumonia. Intern Med 2009; 48: 195202.
  • 8
    Watanabe T, Yasuoka A, Tanuma J et al. Serum (1[RIGHTWARDS ARROW]3) beta-D-glucan as a noninvasive adjunct marker for the diagnosis of Pneumocystis pneumonia in patients with AIDS. Clin Infect Dis 2009; 49: 11281131.
  • 9
    Yasuoka A, Tachikawa N, Shimada K, Kimura S, Oka S. (1[RIGHTWARDS ARROW]3) beta-D-glucan as a quantitative serological marker for Pneumocystis carinii pneumonia. Clin Diagn Lab Immunol 1996; 3: 197199.
  • 10
    Obayashi T, Yoshida M, Mori T et al. Plasma (1[RIGHTWARDS ARROW]3)-beta-D-glucan measurement in diagnosis of invasive deep mycosis and fungal febrile episodes. Lancet 1995; 345: 1720.
  • 11
    Probst M, Ries H, Schmidt-Wieland T, Serr A. Detection of Pneumocystis carinii DNA in patients with chronic lung diseases. Eur J Clin Microbiol Infect Dis 2000; 19: 644645.
  • 12
    Larsen HH, Masur H, Kovacs JA et al. Development and evaluation of a quantitative, touch-down, real-time PCR assay for diagnosing Pneumocystis carinii pneumonia. J Clin Microbiol 2002; 40: 490494.
  • 13
    Ikemura K, Ikegami K, Shimazu T, Yoshioka T, Sugimoto T. False-positive result in Limulus test caused by Limulus amebocyte lysate-reactive material in immunoglobulin products. J Clin Microbiol 1989; 27: 19651968.
  • 14
    Kanda H, Kubo K, Hamasaki K et al. Influence of various hemodialysis membranes on the plasma (1[RIGHTWARDS ARROW]3)-β-D-glucan level. Kidney Int 2001; 60: 319323.
  • 15
    Kato A, Takita T, Furuhashi M, Takahashi T, Maruyama Y, Hishida A. Elevation of blood (1[RIGHTWARDS ARROW]3)-beta-D-glucan concentrations in hemodialysis patients. Nephron 2001; 89: 1519.
  • 16
    Ogawa M, Hori H, Niiguchi S, Azuma E, Komada Y. False positive plasma (1[RIGHTWARDS ARROW]3)-β-D-glucan following immunoglobulin product replacement in adult bone marrow recipient. Int J Hematol 2004; 80: 9798.
  • 17
    Usami M, Ohata A, Horiuchi T, Nagasawa K, Wakabayashi T, Tanaka S. Positive (1[RIGHTWARDS ARROW]3)-beta-D-glucan in blood components and release of (1[RIGHTWARDS ARROW]3)-beta-D-glucan from depth-type membrane filters for blood processing. Transfusion 2002; 42: 11891195.
  • 18
    Ohata A, Usami M, Horiuchi T, Nagasawa K, Kinoshita K. Release of (1[RIGHTWARDS ARROW]3)-beta-D-glucan from depth-type membrane filters and their in vitro effects on proinflammatory cytokine production. Artif Organs 2003; 27: 728735.
  • 19
    Fujii T, Nakamura T, Iwamoto A. Pneumocystis pneumonia in patients with HIV infection: clinical manifestations, laboratory findings, and radiological features. J Infect Chemother 2007; 13: 17.
  • 20
    Cuétara MS, Alhambra A, Chaves F, Moragues MD, Pontón J, del Palacio A. Use of a serum (1[RIGHTWARDS ARROW]3)-beta-D-glucan assay for diagnosis and follow-up of Pneumocystis jiroveci pneumonia. Clin Infect Dis 2008; 47: 13641366.
  • 21
    Yoshida M, Roth RI, Grunfeld C, Feingold KR, Levin J. Soluble (1[RIGHTWARDS ARROW]3)-beta-D-glucan purified from Candida albicans: biologic effects and distribution in blood and organs in rabbits. J Lab Clin Med 1996; 128: 103114.
  • 22
    Del Bono V, Mularoni A, Furfaro E et al. Clinical evaluation of a (1[RIGHTWARDS ARROW]3)-beta-D-glucan assay for presumptive diagnosis of Pneumocystis jiroveci pneumonia in immunocompromised patients. Clin Vaccine Immunol 2009; 16: 15241526.
  • 23
    Marty FM, Koo S, Bryar J, Baden LR. (1[RIGHTWARDS ARROW]3)-beta-D-glucan assay positivity in patients with Pneumocystis (carinii) jiroveci pneumonia. Ann Intern Med 2007; 147: 7072.
  • 24
    Tasaka S, Hasegawa N, Kobayashi S et al. Serum indicators for the diagnosis of pneumocystis pneumonia. Chest 2007; 131: 11731180.
  • 25
    Mennink-Kersten MA, Ruegebrink D, Verweij PE. Pseudomonas aeruginosa as a cause of 1,3-beta-D-glucan assay reactivity. Clin Infect Dis 2008; 46: 19301931.