Combined Serum (1,3)-β-D-Glucan and Oral Wash PCR as a Noninvasive Diagnostic Strategy for Early Detection of Pneumocystis jirovecii Pneumonia: An Observational Retrospective Study

Abstract

Background

The diagnosis of P. jirovecii pneumonia (PJP) in immunosuppressed, HIV-negative patients is often limited by the inability to perform a bronchoalveolar lavage (BAL) due to respiratory insufficiency. In this study, we evaluated the diagnostic accuracy of combining serum beta-D-glucan (BDG) testing with P. jirovecii polymerase chain reaction (PCR) in oral wash (OW) samples as noninvasive tools for PJP diagnosis.

Methods

This was a retrospective, single-center study including all immunosuppressed non-HIV adult patients with suspected PJP in whom a Fujifilm Wako β-glucan test BDG and BAL PJ in-house PCR determination were performed during a 5-year period (2019–2024). The diagnostic performance (negative predictive value [NPV], positive predictive value [PPV], and accuracy) of BDG alone, OW PJ PCR alone, and combined with OW PJ PCR, when available, was assessed in patients with PJP. PJP diagnosis was established according to EORTC/MSGERC 2021 criteria.

Results

One hundred fourteen patients were included. PJP was confirmed in 15 patients (13.2%), of whom 14 had a positive BDG determination (median [interquartile range], 55.32 [19.98–137.90] pg/mL). BDG PPV was low (51.7%), but the overall accuracy (87.7%) for PJP diagnosis was good. Forty-seven (41%) patients had an OW performed, 11 of whom were diagnosed with PJP. The combined use of BDG and OW PJ PCR showed a high PPV (100%) and NPV (97.3%) and an excellent diagnostic accuracy (97.9%) for PJP diagnosis in this patient subset.

Conclusions

The additional use of OW PJ PCR to BDG should be considered as an accurate, noninvasive combined diagnostic strategy for PJP diagnosis.

With the universal use of highly active antiretroviral therapy in HIV-infected patients, the diagnosis of Pneumocystis jirovecii pneumonia (PJP) is currently increasing in immunosuppressed patients with malignancies, solid organ transplant recipients, and those with autoimmune diseases [1, 2]. In this population, PJP often presents with rapidly progressive pulmonary infiltrates, leading to respiratory insufficiency and intensive care unit (ICU) admission. The mortality rates in this critical scenario can reach 40%–60% [3]. Additionally, PJP can mimic other causes of pulmonary infiltrates in patients with underlying malignancies, such as tumor dissemination, drug-related toxicity, or other opportunistic infections [4–6]. While prophylaxis against P. jirovecii (PJ) is well established in patients with hematologic malignancies and solid organ transplant recipients [7–9], standardized recommendations are lacking in patients with solid malignancies or autoimmune diseases. Therefore, a prompt and accurate diagnosis of PJP in the heterogeneous non-HIV population is particularly important to initiate early and appropriate empirical treatment and, ultimately, improve outcomes.

Direct fluorescent microscopy examination (DFME) has classically been the standard laboratory method for the diagnosis of PJP [10]. However, its diagnostic yield is further decreased in the non-HIV population due to low fungal loads. To overcome this limitation, the use of PJ PCR in respiratory tract specimens has become increasingly relevant due to its higher sensitivity when compared with DFME [11, 12]. Nevertheless, the use of PJ PCR may be associated with overdiagnosis since the optimal threshold to differentiate colonization from true disease has not been established as it is dependent on each assay and not yet standardized technical factors.

The diagnostic performance of PJ PCR has been predominantly evaluated in low respiratory tract samples such as bronchoalveolar lavage (BAL) [13]. However, undergoing a flexible bronchoscopy (FBS) may be unfeasible in an urgent care situation or in cases of severe thrombocytopenia, a frequent complication in cancer patients. In this setting, the use of PJ PCR in upper respiratory tract samples, such as oral wash (OW), has been reported to be a less invasive option for PJP diagnosis. Several meta-analyses show that, while the specificity of OW PJ PCR remains similar to BAL PJ PCR, its sensitivity is generally lower [14, 15]. Several studies have reported good diagnostic performance of PJ PCR in OW samples, supporting its role as an attractive option to avoid invasive procedures [16, 17].

(1,3)-ß-d-glucan (BDG) is a highly sensitive panfungal biomarker that has been proved to be a useful tool for PJP diagnosis [18]. The high negative predictive value of BDG allows ruling out PJP even when pretest probability is high [19]. In addition, a positive BDG determination enables differentiating colonization from invasive disease when the PJ PCR value in BAL fluid is inconclusive [20]. Nevertheless, one of its main drawbacks is its low positive predictive value [21], driven by several factors such as its panfungal nature, iatrogenic contamination, intestinal translocation, or bacterial infection, among others, causing false-positive results [22]. Therefore, a positive BDG test should trigger additional tests to achieve a more accurate diagnosis.

In light of the previous data and considering the clinical limitations of diagnosing PJP in non-HIV patients using conventional invasive techniques, optimizing diagnostic algorithms through the combined use of noninvasive tests is of utmost importance to guide empirical therapy and, ultimately, improve outcomes.

In the present study, we aimed to test the hypothesis that the combined use of BDG and OW PJ PCR could provide rapid and reliable results for the diagnosis of Pneumocystis jirovecii pneumonia, particularly when FBS performance is infeasible.

METHODS

Study Design and Participants

This was a retrospective, unicentric, observational study conducted at Vall d’Hebron University Hospital in Barcelona, Spain, from June 2019 to December 2024. All adult (≥18 years) non-HIV immunocompromised patients with suspected PJP in whom a BAL PJ PCR and serum BDG determination were tested <72 hours apart were included in the study. When available, PJ PCR values in OW were also recorded. The exclusion criteria were (1) prior HIV infection, (2) other fungal infection diagnosis, (3) clinical or radiological findings incompatible with PJP, and (4) active PJ treatment for >48 hours before BAL realization.

Data Collection

Patients were identified through microbiology records. Data were obtained from electronic hospital records and collected in a dedicated password-protected database hosted in Research Electronic Data Capture (RedCap; hosted at Vall d’Hebron Institute of Research). Demographics and clinical data were analyzed, including age, sex, underlying disease, hematopoietic stem cell transplantation, previous lymphopenia, previous immunosuppressive treatment (including corticosteroids), previous PJ prophylaxis or other PJ-active drugs, clinical and radiological features, and outcomes. Microbiological tests performed, including BDG value, BAL PJ PCR cycle threshold (ct), and OW PJ PCR ct, were also analyzed. Patients' follow-up extended from inclusion to hospital discharge.

Sample Collection and Microbiological Studies

BAL specimens were collected according to standard bronchoscopy procedures (3 separate 50-mL aliquots were instillated and later re-aspirated), and OW specimens were obtained by vigorous gargling of 10 mL of sterile saline for 30 seconds. Real-time PJ PCR targeting the single-copy nuclear dihydropteroate synthase (DHPS) gene and the mitochondrial small-subunit (mtSSU) rRNA were performed (DNA extraction, primers, probes, and thermal cycling protocol) and interpreted as previously described by our group [16]. Serum BDG was determined by Wako ß-glucan assay (Wako-BDG; Fujifilm Wako Chemicals) and, according to manufacturer's instructions, interpreted as positive if >7.0 pg/mL [23].

Definitions

According to EORTC/MSGERC 2021 criteria [10], a probable PJP case was established in patients with compatible clinical symptoms (fever, cough, dyspnea, or hypoxemia), new-onset or worsened interstitial pattern on radiological tests (chest x-ray or computed tomography scan), and a positive P. jirovecii BAL PCR (DHPS ct <36.8 and mtSSU ct <30.9) [16]. Colonization was considered in cases of an alternative diagnosis and clinical improvement without specific anti–P. jirovecii treatment, regardless of the PJ PCR value. Previous lymphopenia was defined as an absolute lymphocyte count (ALC) <1000/μL, maintained for at least the previous 30 days. Systemic corticosteroid treatment was defined as any dose of methylprednisolone (or equivalent) within the previous 30 days. Other immunosuppressive treatments (conventional chemotherapy, targeted treatments, methotrexate) received during the previous 6 months were also recorded. Active PJ prophylaxis included cotrimoxazole (trimethoprim/sulfamethoxazole [TMP/SMX]) 80/400 mg once a day or 160/800 mg 3 times a week, inhaled pentamidine every 28 days, atovaquone 1500 mg once a day, or dapsone 50 mg twice a day [7]. As per local protocol, cotrimoxazole (TMP 15–20 mg/kg/d) was considered the first-line treatment. If contraindicated, alternative second-line regimens included clindamycin plus primaquine, intravenous pentamidine, or atovaquone.

Diagnostic Performance

The diagnostic performance of serum BDG for the diagnosis of PJ pneumonia was analyzed in all included patients. The diagnostic accuracy of OW PJ PCR alone and combined with BDG for PJP diagnosis was analyzed only in patients who additionally had an available OW sample (OW cohort).

Statistical Analysis

To define cohort characteristics, categorical variables were presented as number and percentage and compared using the chi-square or Fisher exact test as appropriate. Continuous variables were presented as median and interquartile range (IQR) and compared using the Mann-Whitney U test. Sensitivity (Se), specificity (Sp), positive and negative predictive values (PPVs and NPVs, respectively), positive and negative likelihood ratios (PLRs and NLRs), and accuracy (overall and balanced) were calculated for BDG positivity alone or in combination with OW PJ PCR positivity. To estimate the 95% CIs for Se, Sp, PPV, NPV, and accuracy, bootstrapping resampling was used. Specifically, 1000 bootstrap samples were generated by resampling the paired true and predicted class labels with replacement. For each bootstrap replicate, the classification metrics were recalculated. The 95% CIs were defined by the 2.5th and 97.5th percentiles of the resulting bootstrap distributions. To avoid infinite or 0 values when calculating PLRs and NLRs, the Haldane-Anscombe correction was applied. Specifically, 0.5 was added to each 2 × 2 contingency table cell before computing PLR and NLR, providing more stable estimates and preventing extreme values. The statistical analyses were performed using IBM SPSS Statistics (Mac, version 27.0; IBM Corp., Armonk, NY, USA) and Python (Windows version 3.13.3; Python Software Foundation).

Ethics

The need for informed consent was waived due to the retrospective nature of the study. The study was approved by the ethical committee of Vall d’Hebron University Hospital (protocol number PR [AG] 108/025). It was conducted in accordance with the Declaration of Helsinki guidelines. The study results are reported following the STROBE recommendations (Supplementary Table 1).

RESULTS

Patients

Among the 176 eligible patients, 62 met at least 1 exclusion criterion, leaving 114 included patients, of whom 47 had an oral wash performed (Figure 1).

Figure 1.

Study cohort and patient selection. The flowchart shows patient selection: Of 176 eligible patients, 62 were excluded due to any exclusion criterion. The remaining 114 patients were included, of whom 15 had PJP and 99 had other diagnoses. Among these, 47 underwent oral wash, including 11 with PJP and 36 with other diagnoses. Abbreviations: BAL, bronchoalveolar lavage; BDG, 1,3-ß-D-glucan; PJP, Pneumocystis jirovecii pneumonia.

The respiratory episode was attributed to PJP in 15 patients (13.2%), while in 99 (86.8%) an alternative diagnosis was identified, mainly viral or bacterial pneumonia (61/99, 61.6%) or drug-induced lung injury (13/99, 13.1%). Among patients without PJP, 12/99 (12.1%) had a positive BAL PJ PCR (median DHPS ct [IQR], 40 [37.49–40]; median mtSSU ct [IQR], 34.83 [32.86–37.99]) and were later considered colonized.

Patients' Clinical Data

Clinical data and baseline characteristics are detailed in Table 1. No differences regarding smoking habit, lung structural disease, previous lymphopenia, previous treatment with corticosteroids, or other immunosuppressive drugs were identified between PJP and non-PJP patients. None of the patients diagnosed with PJP had received active PJ prophylaxis, except 1 who was under inhaled pentamidine prophylaxis. Thirty-four patients (29.8%) received empirical treatment for P. jirovecii (32 patients TMP/SMX and 2 patients atovaquone due to previous cytopenia).

Table 1.

Clinical Characteristics of 114 Included Patients Based on Clinical, Radiological, and Microbiological Findings

	Total (114, 100)	P. jirovecii Pneumonia (15, 13.2)	No P. jirovecii Pneumonia (99, 86.8)	P Value
Male gender	73 (64)	8 (53.3)	65 (65.7)	.354
Age, median (IQR), y	57.5 (42.5–68)	58 (34–65)	57 (43–68)	.651
Smoker	16 (14)	1 (6.7)	15 (15.2)	.691
Former smoker	47 (41.2)	6 (40)	41 (41.4)	.917
Structural lung disease	54 (47.4)	5 (33.3)	49 (49.5)	.243
Solid organ malignancy	28 (24.6)	7 (46.7)	21 (21.2)	.033
Lung neoplasm	8 (7)	3 (20)	5 (5.1)	.069
Lung metastasis	7 (6.1)	3 (20)	4 (4)	.047
Hematological malignancy	78 (68.4)	7 (46.7)	71 (71.7)	.052
Solid organ transplantation	8 (7)	0 (0)	8 (8.1)	.594
Autoimmune disease	10 (8.1)	2 (13.3)	8 (8.1)	.618
ALC <1000/μL for >30 da	62 (60.2)	8 (57.1)	54 (60.7)	.802
Days since lymphopenia started, median (IQR)	70 (24–219)	114 (54–151)	69 (24–225.75)	.605
Systemic corticosteroids <30 d	54 (47.4)	6 (40)	48 (48.5)	.540
Alone	11 (9.6)	3 (20)	8 (8.1)	.158
In combination	43 (37.7)	3 (20)	40 (40.4)	.160
Systemic corticosteroids without PJ prophylaxis	26 (40)	6 (42.9)	20 (39.2)	.805
Any other immunosuppressive therapiesb	100 (87.7)	13 (86.7)	87 (87.9)	.894
Conventional chemotherapy <6 mo	55 (48.2)	3 (20)	52 (52.5)	.019
Targeted therapies <6 mo	79 (69.3)	10 (66.7)	69 (69.7)	.773
Previous PJ prophylaxis	49 (43)	1 (6.7)	48 (48.5)	.002

Data are expressed as No. (%) unless otherwise indicated.

Abbreviations: IQR, interquartile range; ALC, absolute lymphocyte count; PJ, Pneumocystis jirovecii.

^aAvailable in 103/114 patients.

^bOther immunosuppressive therapies include targeted therapies, methotrexate, conventional chemotherapy, and mycophenolate.

Regarding patients with PJP, solid organ and hematological malignancy accounted for the first cause of immunosuppression. Patients with solid cancer diagnosed with PJP more frequently had lung neoplasms or lung metastasis.

ICU admission and in-hospital mortality were high in the whole cohort, with no differences between patients with and without PJP (40% vs 47.5%; P = .589; and 26.7% vs 25.3%; P = .907; respectively). Four patients (4/15, 26.7%) with PJP died, with PJP being the main cause of death in 3, which accounted for an attributable mortality of 75%.

Sample Collection

BDG determination and BAL specimens were obtained within a median interval (IQR) of 1 (0–2) day. Among patients with OW, BDG determination and OW specimens were collected on the same day (median [IQR], 0 [0–1]) and 1 day before BAL collection (median [IQR], 1 [1–2]). Anti-PJ empirical treatment was mostly initiated on the same day as OW collection (median [IQR], 0 [0–0]) and 1 day before BAL collection (median [IQR], 1 [1–2]).

End Points

BDG Diagnostic Yield in Patients With Suspected PJP

Overall, BDG was positive in 27 out of 114 (23.7%) patients; 14/27 were diagnosed with PJP and 13/27 with an alternative diagnosis. In this latter group, a possible cause of a false positive was identified in 9 of the 13 (69.2%) patients, 8 were receiving treatment with beta-lactam antibiotics, and 1 was under extracorporeal membrane oxygenation. All colonized patients had a negative BDG determination. Median BDG values were significantly higher in patients with PJP than in patients with an alternative diagnosis (median [IQR], 55.32 [19.98–137.90] pg/mL; vs median [IQR], 3.52 [3.13–3.96] pg/mL; P < .001) (Supplementary Figure 1).

The Se, Sp, PPV, and NPV of a positive BDG alone in patients with suspected PJP were 93.3%, 86.9%, 51.9%, and 98.8%, respectively. BDG positivity achieved an overall and balanced accuracy of 87.7% and 90.1%, respectively, for identifying patients with PJP. The PLR and NLR were 7.11 and 0.077, respectively (Table 2A).

Table 2.

Summary of (A) Serum 1,3-ß-D-Glucan Alone and (B) Combined Postive and Negative Serum 1,3-ß-D-Glucan and Oral Wash PCR Performance for Noninvasive P. jirovecii Pneumonia Diagnosis

A
Whole Cohort (n = 104)	P. jirovecii Pneumonia (n = 15)	No P. jirovecii Pneumonia (n = 99)	Se, %	Sp, %	PPV, %	NPV, %	Accuracy, %	PLR	NLR
BDG >7 pg/mL	14	13	93.3	86.9	51.9	98.8	87.7	7.11	0.077
BDG <7 pg/mL	1	86	…	…	…	…	90.1	…	…

B
OW Cohort (n = 47)	P. jirovecii Pneumonia (n = 11)	No P. jirovecii Pneumonia (n = 36)	Se, %	Sp, %	PPV, %	NPV, %	Accuracy, %	PLR	NLR
Positive OW	11	4	100	88.9	73.3	100	91.5a	7.88	0.047
Negative OW	0	32	…	…	…	…	94.4b	…	…
OW and BDG both positive	10	0	90.9	100	100	97.3	97.9a	64.8	0.127
Either OW or BDG negative	1	36	…	…	…	…	95.5b	…	…
Either OW or BDG positive	11	8	100	77.8	57.9	100	83a	4.17	0.054
OW and BDG both negative	0	28	…	…	…	…	88.9b	…	…

In the OW cohort, PLR and NLR were calculated using the Haldane-Anscombe correction to account for 0 counts in contingency tables and to provide stable estimates in small samples.

Abbreviations: BDG, 1,3-ß-D-glucan; LR+, positive likelihood ratio; LR−, negative likelihood ratio; NPV, negative predictive value; OW, oral wash; PPV, positive predictive value; Se, sensitivity; Sp, Specificity.

^aOverall accuracy.

^bBalanced accuracy.

OW Cohort

Among 47 (41.2%) patients with an OW PJ PCR performed, 11 were diagnosed with PJP, while 36 had an alternative diagnosis.

OW Diagnostic Yield in Patients With Suspected PJP

All 11 (100%) PJP cases had a positive OW PJ PCR, while 4 patients with an alternative diagnosis also had a positive OW PJ PCR and were considered colonized. In this latter group, all had a metastatic solid organ tumor on active chemotherapy, only 1 was on cotrimoxazole prophylaxis, and all had a negative BDG determination. BAL results ruled out PJP in all 4 cases (1 negative PJ PCR and 3 with ct values compatible with colonization).

Oral wash PJ PCR in patients with suspected PJP had an Se, Sp, PPV, and NPV of 100%, 88.9%, 73.3%, and 100%, respectively. Overall and balanced accuracy were 91.5% and 94.4%, respectively, for PJP diagnosis. The PLR and NLR were 7.88 and 0.047, respectively (Table 2B).

Diagnostic Performance of BDG and OW PJ PCR in Combination

Among PJP cases, 10 (90.9%) tested positive for both BDG and OW PJ PCR. The remaining patient showed discordant results: Clinical and radiological criteria were met, the OW tested positive, but the BDG was negative. Therefore, as no alternative diagnosis was identified, it was considered a BDG false negative.

Regarding the 36 patients without PJP, 28 (77.8%) tested negative in both BDG and OW PJ PCR. Eight patients (22.2%) had discordant results (4 patients BDG+/OW–, 4 patients BDG−/OW+). All patients with a positive BDG and negative OW had a negative BAL PCR and discontinued PJP treatment; therefore, they were classified as false-positive BDG. Among the remaining 4 patients with a negative BDG and a positive OW PJ PCR, BAL PJ PCR was negative in 1 patient and positive with ct values compatible with colonization in 3 patients.

When evaluating the performance of the combination of both BDG and OW PJ PCR positivity for PJP diagnosis, Se was 90.9% (95% CI, 70.0%–100.0%) and Sp 100% (95% CI, 100.0%–100.0%). The PPV reached 100% (95% CI, 100.0%–100.0%), whereas the NPV was 97.3% (95% CI, 91.4%–100.0%). PLR and NLR were 64.8 and 0.127, respectively. Positivity of both biomarkers achieved an overall and balanced accuracy of 97.9% (95% CI, 93.6%–100.0%) and 95.5% (95% CI, 85.0%–100.0%), respectively, for PJP diagnosis, whereas NPV and NLR of both negative BDG and OW PJ PCR were 100% and 0.054, respectively. Table 2B presents the performance of BDG and OW PJ PCR combined for PJP diagnosis and exclusion.

Compared with BDG alone, the addition of OW PJ PCR increased Sp, PPV, and diagnostic accuracy to 15.1%, 92.7% and 11.6%, respectively. Likewise, the addition of BDG to OW PJ PCR enhanced Sp by 12.5%, PPV by 36.4%, and diagnostic accuracy by 7%. Comparative values and estimated CIs of Se, Sp, PPV, NPV, and accuracy for BDG alone, OW alone, and BDG combined with OW PJ PCR are represented in Figure 2 and Supplementary Table 2.

Figure 2.

Performance of BDG alone, OW PJ PCR, and both oral wash PJ PCR and BDG positivity performance for noninvasive P. jirovecii pneumonia diagnosis. For BDG alone: Se 93.33% (95% CI, 76.92%–100%), Sp 86.87% (95% CI, 80.0%–92.93%), PPV 51.85% (95% CI, 33.33%–70.97%), NPV 98.85% (95% CI, 96.30%–100.0%), overall accuracy 87.72% (95% CI, 81.58%–92.98%), balanced accuracy 90.1% (95% CI, 81.53%–95.76%). For OW PJ PCR alone: Se 100% (95% CI, 100.0%–100.0%), Sp 88.9% (95% CI, 77.8%–97.4%), PPV 73.3 (95% CI, 50.0%–93.3%), NPV 97.3% (95% CI, 100.0%–100.0%), overall accuracy 91.5% (95% CI, 83.0%–97.9%), balanced accuracy 94.4% (95% CI, 88.9%–98.7%). For BDG and OW PJ PCR positivity: Se 90.9% (95% CI, 70.0%–100.0%), Sp 100% (95% CI, 100.0%–100.0%), PPV 100% (95% CI, 100.0%–100.0%), NPV 97.3% (95% CI, 91.4%–100.0%), overall accuracy 97.9% (95% CI, 93.6%–100.0%), balanced accuracy 95.5% (95% CI, 85.0%–100.0%). Abbreviations: BDG, 1,3-ß-D-glucan; NPV, negative predictive value; OW PJ PCR, oral wash polymerase chain reaction; PPV, positive predictive value; Se, sensitivity; Sp, specificity.

DISCUSSION

In this study, the accuracy of combining 2 noninvasive tests—BDG and OW PJ PCR—for the diagnosis of P. jirovecii pneumonia was evaluated. Wako BDG assay showed good overall accuracy, with significantly higher values in patients with PJP compared with those with an alternative diagnosis. However, its PPV and PLR were insufficient to reliably confirm PJP diagnosis. Consequently, the combination of BDG with OW PJ PCR enhanced its diagnostic performance, achieving an excellent diagnostic accuracy (97.9%) for PJP diagnosis.

Currently, most of the studies analyzing BDG as an adjunctive tool for PJP diagnosis have been focused on the Fungitell assay in mixed HIV and non-HIV populations. Although some limitations have been described using the Wako assay, such as reduced stability at room temperature and lower sensibility for the diagnosis of other invasive fungal infection, several diagnostic advantages should be noted: easy performance, high specificity, and low intra- and interassay variability (1.8%–13.2% and 3.3%–7.6%, respectively), allowing individual sample testing and sequential processing [24, 25]. However, data regarding its performance for PJP diagnosis in non-HIV patients are limited.

In our study, the use of the Wako assay showed good accuracy for the diagnosis of PJP in immunosuppressed non-HIV patients, with a high negative predictive value, as similarly reported with Fungitell [25]. Considering the advantages of this assay, our results encourage its use in the diagnostic algorithm of PJ in this high-risk population.

In our study, BDG values were significantly higher in patients with PJP compared with patients with an alternative diagnosis [23–25]. The manufacturer-recommended cutoff was initially set at 11 pg/mL, with reported sensitivities ranging from 78% to 88.9% and specificities from 98% to 100%. To improve its diagnostic yield, Carolis et al. lowered the cutoff for positivity to 7 pg/mL and proved an increase in Se from 88.2% to 94.1% with a slight decrease in Sp (from 99.5% to 97.3%) in HIV-negative patients with PJP [23]. In line with these results, we observed similar BDG diagnostic accuracy using the aforementioned cutoff, thus supporting the routine use of this lowered threshold.

Similarly to previous literature, solid organ malignancy accounted for the most common underlying disease in our cohort. In this group, structural lung disease (as primary or metastatic) was more frequently encountered [1, 3, 26–28]. Importantly, none of the patients with PJP were receiving active prophylaxis except 1. As reported in a recent retrospective study [29], the use of prophylaxis in this population remains low, probably as a direct consequence of the lack of standard recommendations.

In this vulnerable population, in whom FBS is often infeasible, noninvasive techniques such as BDG or PCR in upper respiratory tract samples have emerged as key diagnostic tools for PJP diagnosis. Most studies have analyzed the accuracy of BDG or PCR in nasal swabs or induced sputum alone, whereas only a few have studied the efficacy of the combined use of both tests. Price et al. [30] developed an algorithm for PJP diagnosis, based on combining PCR results (assessed with OLM PneumID in BAL and upper respiratory tract samples) and Fungitell BDG assay (based on Del Corpo [31] meta-analysis results). A prospective study in France [32] aimed to assess the diagnostic performance of a quantitative PCR (qPCR) in nasopharyngeal aspirate combined with the Fungitell BDG assay. The combination of a qPCR ct determination <35 and BDG value >143 pg/mL achieved an Se, Sp, PPV, and NPV of 93.75%, 97.87%, 97.83%, and 93.88%, respectively. Similarly, combining both tests significantly enhanced the Sp, PPV, and overall diagnostic accuracy of BDG positivity in our study, compared with using either test alone.

Almost 9% of patients with PJP suspicion showed false-positive BDG values, a higher rate than previously reported [18, 23, 25]. All these cases had a negative OW determination and a probable alternative diagnosis, supporting the exclusion of PJP.

The use of PCR in OW alone has certain drawbacks. The detection of colonized patients decreases the specificity of the test [16]. In our study, 11% of patients who had an OW performed and who did not have PJP showed a positive OW PJ PCR result and were considered colonized, as BDG was negative in all of them. This rate of OW PJ PCR false positivity is higher than the rates reported by Hviid et al. [33], which may be related to the specific fungal target [16]. These authors detected PJ in 9 out of 250 (3.6%) OW samples collected from HIV-negative patients. In situations where a false-positive PCR PJ is suspected, interpretation of the positive results may be facilitated by a concomitant BDG determination, which increases both Sp and PPV and contributes to a more accurate diagnosis.

All PJ cases in our study had a positive OW PJ PCR leading to an NPV of 100%, in line with our previously published data. However, several authors have detailed a false-negative OW PJ PCR rate ranging from 9% to 52% [17, 34, 35]. Consequently, improper OW sample collection and low fungal burden need to be considered in case of a negative OW PJ PCR in a high-risk patient.

Our data support the implementation of a stepwise diagnostic algorithm (Figure 3) that includes the combined use of BDG and OW PJ PCR in immunosuppressed non-HIV patients at high risk of PJP, particularly if FBS is infeasible. In this setting, an early empirical treatment should be initiated, followed by a tailored therapy according to PCR and BDG results. When noninvasive tests show inconclusive results, an FBS performance is of paramount importance, along with a thorough evaluation of host and clinical factors. The use of this strategy, together with prompt diagnostic suspicion and a 24-hour turnaround of microbiological results, may have contributed to the low all-cause mortality observed in our cohort in comparison to previous studies [3].

Figure 3.

Proposed algorithm for Pneumocystis jirovecii pneumonia diagnosis. For non-HIV immunocompromised patients with suspected PJP, BDG and OW PJ PCR should be performed simultaneously. If both tests are negative, PJP can be ruled out, while if both are positive, PJP is confirmed. When results are discordant, the patient's risk should be carefully assessed, and a fibrobronchoscopy (if feasible) should be performed to establish a definitive diagnosis. Abbreviations: BDG, 1,3-ß-D-glucan; OW, oral wash; PJP, Pneumocystis jirovecii pneumonia.

This study has several limitations. First, its retrospective nature could have led to missing information in some patients or introduced bias in patient selection. Second, it was carried out in a single center using specifically the Wako assay and an in-house PJ PCR, limiting the generalizability of our results, particularly to centers using other techniques. However, our positivity cutoffs for PJP PCR had been previously validated, supporting the robustness of our results. Third, despite the definition of PJP being based on the EORT/MSGERC criteria, the definition of colonized patients still includes subjective criteria, which can lead to misclassification. Fourth, allowing a 72-hour gap between samples could have facilitated variability in disease progression. Finally, a BAL was required to ensure PJP diagnosis, which led to a small number of included patients and resulted in wide confidence intervals for some diagnostic metrics, particularly in the OW cohort. Nevertheless, the statistical methods used are rigorous for small samples, so in conjunction, this approach allowed for robust results in a highly targeted population. Additionally, patients whose BAL was performed >48 hours after treatment initiation were also excluded to minimize the risk of false-negative BAL PJ PCR results.

In summary, the combined use of BDG and OW P. jirovecii PCR showed good diagnostic accuracy for PJP in susceptible hosts. Therefore, the integration of both noninvasive assays should be considered in a diagnostic strategy for patients with suspected PJP, particularly when FBS is not feasible. Further randomized controlled trials are warranted to assess the impact of the exclusive use of both tests for PJ diagnosis in these high-risk patients.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.