Evaluation of the ELITe InGenius PCR assay compared with immunofluorescence for Pneumocystis jirovecii detection in respiratory samples

Abstract

Purpose

Pneumocystis jirovecii (PJ) causes pneumonia primarily in immunocompromised individuals. Although direct immunofluorescence (IF) remains widely used in routine diagnostics, PCR assays are increasingly being adopted because they offer higher sensitivity and reduced observer dependency. This study evaluates the ELITe InGenius PJ PCR assay compared with IF for PJ detection in respiratory specimens.

Methods

Respiratory samples submitted for IF-based PJ testing at the University Hospital Zurich over a 19-month period were retrospectively analyzed using the ELITe InGenius PCR assay. Diagnostic accuracy was assessed using IF as the reference method, and performance was evaluated by receiver operating characteristic (ROC) analysis.

Results

A total of 222 samples from 213 patients were included (70 sputum, 152 bronchoalveolar lavage). PCR and IF results were concordant in 175 (78.8%) specimens, including 18 positives and 157 negatives. 47 (21.2%) samples were discrepant: six IF-positive/PCR-negative and 41 IF-negative/PCR-positive. IF reexamination was possible for three IF-positive/PCR-negative and eight IF-negative/PCR-high-positive samples (> 100’000 copies/ml (cp/ml) or Ct < 29), resulting in revised IF findings in 3/3 and 2/8 cases, respectively. Three IF-negative/PCR-positive cases were clinically confirmed as true infections. ROC analysis identified an optimal PCR threshold of 6’233 cp/ml, yielding 70.8% sensitivity and 88.3% specificity, improving to 82.6% and 89.5% after reexamination.

Conclusion

These findings support the higher sensitivity of PCR relative to IF for direct PJ detection. Clinical confirmation of IF-negative/PCR-positive cases and known limitations of IF underscore the diagnostic value of the ELITe InGenius assay. Its performance supports its use in routine diagnostics to improve early detection and guide timely treatment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10096-026-05474-9.

Introduction

Pneumocystis jirovecii (PJ) is an atypical, unicellular fungus with tropism for lung tissue causing pneumonia almost exclusively in immunocompromised patients [1–5]. Severe immunodeficiency remains the most essential host factor for Pneumocystis jirovecii pneumonia (PJP). Over the last two decades, the ratio of affected non-HIV- to HIV patients steadily increased in resource rich countries [6–9]. This shift is strongly linked to the widespread use of antiretroviral therapy (ART), which preserves cellular immunity and reduces opportunistic infections and mortality in HIV-positive patients [10–14]. Additionally, the increasing number of patients undergoing immunosuppressing or -modulating therapy for conditions such as solid organ transplantation, neoplastic, rheumatological and inflammatory diseases, has resulted in non-HIV patients representing the majority of PJP patients [15–18].

This changing epidemiology of PJP has introduced new challenges in its identification and management [9]. PJP typically manifests as a non-specific pneumonia with symptoms including fever, dyspnea and a non-productive cough, and is associated with a high mortality rate ranging from 27 to 55% [1–3, 19, 20]. Compared to people with HIV (PWH), non-HIV patients often experience a more acute onset and critical disease course, with higher rates of respiratory failure, ICU admission, and mortality [9, 19]. In contrast, PWH tend to exhibit a more prolonged disease progression and higher fungal loads. Radiological findings, including ground-glass opacities and, in more advanced stages, a denser mosaic pattern can support the clinical suspicion of PJP [21, 22].

Because PJ cannot grow in routine culture, it is not included in standard diagnostic algorithms [2 (White), 4 (Catherinot), 5 (Thomas), "24" (Bateman)]. Instead, microscopy or PCR analysis from respiratory specimens are used for rapid direct pathogen detection [2, 4, 23, 24]. Microscopy with conventional and immunofluorescent staining methods allows the visualization of cysts and trophozoites, with direct immunofluorescence (IF) demonstrating a more robust performance than conventional microscopy [2 (White), "24" (Bateman)]. In the last decades, PCR-based methods have gained importance due to their high sensitivity (97–99% in three meta-analyses), consistent performance regardless of HIV status, and reduced reliance on investigator experience [23, 25–28]. While it is essential to integrate pretest probability from clinical and radiological findings, particularly for the distinction between colonization and active infection, these promising PCR performance reports highlight its value as a time‑efficient and sensitive adjunct in PJP diagnosis.

We evaluated the ELITe InGenius PJ PCR assay as a not yet assessed platform in published studies. This study aimed to compare its performance with IF, as the reference standard, to support its implementation in routine diagnostics. It also provides a comprehensive overview to guide interpretation and establish a reliable threshold for clinical decision-making in PJP diagnosis.

Materials and methods

Study design and population

This study includes all samples submitted for PJ testing by IF at the University Hospital Zurich (USZ) between 13.03.2023 and 22.10.2024. The patient specimens originated from multiple hospitals and laboratories across the German-speaking part of Switzerland. In this study, all samples were retrospectively analyzed using ELITe InGenius PJ PCR assay to evaluate its suitability for implementation in routine diagnostic. For this study, the reference standard was defined as the routine microscopic result communicated to the treating physician, with immunofluorescence serving as the primary diagnostic method and conventional stains performed in parallel as part of the standard workflow.

Sample processing and immunofluorescence assay

Sample preparation varied depending on the sample type following the internal protocol [29]. For bronchoalveolar lavage (BAL) samples, 0.2 to 0.5 ml were cytocentrifuged onto polytetrafluoroethylene (PTFE)-printed microscope slides using a Cytospin 4 cytocentrifuge. For viscous samples, 0.5 to 5 ml of sputum, tracheal or bronchial secretions, or 1 to 5 ml of mucoid BAL samples, were mixed with 6 to 7 ml of a solution containing 20mM dithiothreitol and 20mM ethylenediamine-tetraacetic acid (EDTA) and incubated at 37 °C for 5 to 10 min. No filtration was performed on the BAL fluid. After dilution with phosphate-buffered saline (PBS), the samples were centrifuged, the supernatant discarded, and the sediment resuspended in a small volume of PBS (depending on the pellet size). Two to three drops of the suspension were applied to PTFE-printed slides and allowed to air dry. Air‑dried smears were heat‑fixed by passing them three times through a Bunsen burner flame. Slides intended for Giemsa and IF staining were subsequently fixed in acetone for 5 min in a Coplin jar.

Three slides were prepared from each sample for microscopic analysis. Two of the slides were stained for conventional microscopy with Toluidine blue and Giemsa stain, respectively. Staining with Toluidine Blue O (Sigma-Aldrich 198161) was performed using the procedure of Gosey et al. [30]. Giemsa staining was conducted using Giemsa’s azur eosin methylene blue solution (Merck Millipore 1.09204) diluted 1:10 in buffer according to Weise, pH 7.2 (Sigma-Aldrich 1.09468). The last slide was processed by direct immunofluorescence using either MONOFLUO™ Pneumocystis jirovecii IFA Test Kit (Bio-Rad 32515) for 148 samples or Merifluor^® Pneumocystis (Meridian Bioscience 222030) for 74 samples, according to the instructions of the manufacturers. Discrepant samples with sufficient material were reexamined using Bio-Rad staining for a second confirmatory IF. All samples in which P. jirovecii was microscopically detected were classified as positive, regardless of the quantity or form of observed organisms.

Real-time PCR assay

Following microscopy, PCR-testing was performed on the ELITe InGenius instrument using the corresponding Pneumocystis ELITe MGB^® Kit. The kit includes the PJ PCR Mix (ref. RTS150ING), the CPE Internal Control (ref. CTRCPE), the PJ Q PCR Standards (ref. STD150ING-2, STD150ING-3, STD150ING-4, STD150ING-5) for absolute quantification, and the PJ Positive Control (ref. CTR150ING). Molecular biology grade water was used as the negative control. This fully automated system extracts nucleic acids and amplifies the mitochondrial large subunit (mtLSU) rRNA gene specific to PJ. Samples were classified as negative if no DNA was detected or if the concentration was below the assay’s limit of detection (LoD), defined as 97copies/ml (cp/ml) according to the manufacturer.

Statistical analysis

To evaluate the diagnostic performance of the InGenius PCR assay relative to the IF assay, receiver operating characteristic (ROC) analysis was performed. Sensitivity and specificity were calculated across a range of thresholds to identify clinically relevant cut-offs. The optimal threshold was defined as the value maximizing Youden’s index (Specificity + Sensitivity − 1). Data analysis was conducted using R (version 4.4.3) in RStudio (version 2024.12.1) including the package pROC [31, 32]. To further contextualize the findings, a sensitivity analysis was performed incorporating detailed patient histories and additional data processing to account for potentially false-positive initial IF results.

Ethical considerations

The ethics committee of the Canton Zurich reviewed the study protocol (Kantonale Ethikkommission Zurich BASEC ID Req-2025-01064) and concluded that the study did not meet the criteria for requiring ethical approval within its remit. All research was carried out in accordance with Good Clinical Practice standards.

Results

Patient and sample characteristics

A total of 222 respiratory samples from 213 patients were retrospectively analyzed using quantitative real-time ELITe InGenius PJ-PCR and results compared to prior PJ detection by IF and conventional microscopy (Table 1).

Table 1.

Patient and sample characteristics (n = 213 patients, 222 samples)

Number of patients	213 (208 adults + 5 children)
Age*	Median: 62 years (range: 13 days – 89 years)
Sex: male/female	135/78
Year of sample collection: 2023/2024	190/32

*The age calculation refers to the year in which the sample was taken. For patients with multiple samples, age was recorded only once, as all samples were collected within the same year

Primary diagnostic performance

Of the 24 (10.8%) microscopy positive samples, 23 were IF positive, 9 Toluidine blue positive, and 6 Giemsa positive. In one case, only Toluidine blue staining was positive and IF negative. It was classified as microscopy-positive and included in the subsequent analysis. For clarity, results incorporating this sample are referred to as “microscopy” rather than “IF”. In all other cases, IF showed higher sensitivity and was positive whenever conventional microscopy detected PJ, as shown in supplementary Table S1 containing the raw data.

Quantitative PCR detected PJ DNA over the assay’s LoD of 97 cp/ml in 59 (26.6%) samples with a mean concentration of 449’108 cp/ml (range: 102–31’622’777 cp/ml). The initial, unaltered results from both diagnostic methods (microscopy and PCR) were referred to as “Primary data”. Within this dataset, 18 (8.1%) samples tested positive by both PCR and microscopy, 41 (18.4%) were only PCR positive and six (2.7%) samples were only IF positive. The latter were labelled as potentially false-negative PCR using microscopy as the reference standard. In one sample from this group, a single cyst was observed during the initial IF examination but could not be confirmed by a second observer a few days later, highlighting the variability and observer-dependency of IF as a diagnostic standard.

IF reexamination of discrepant samples

Among the six IF-positive/PCR-negative samples and 13 of the IF-negative/PCR-positive samples with high PCR positivity (defined as > 100’000 cp/ml or Ct value < 29), reexamination by IF was performed when sufficient material was available (Fig. 1). This was possible in 3/6 IF-positive/PCR-negative and 8/13 IF-negative/PCR-positive samples. The reexamination was conducted using the Bio-Rad IF staining reagent, selected for its superior diagnostic performance in our study and reported lower specificity of Merifluor in comparison to other staining agents [33–35].

Fig. 1.

(a)Flowchart showing data processing of discrepant results in microscopy and PCR analysis using the assay’s limit of detection (LoD; 97 cp/ml). The unadjusted primary data (orange) includes six IF+/PCR- and 41 IF-/PCR+ samples, 13 of which had high PCR positivity and thus were selected for further reexamination. IF reexamination (green) covered three false-negative and eight false-positive samples, with altered results in all three IF+/PCR- and two IF-/PCR+ cases. (b)Contingency tables comparing PCR and microscopy results before and after reexamination, including diagnostic performance values. PPV positive predictive value, NPV negative predictive value

In all three reexamined IF-positive/PCR-negative samples, the initial positive IF result could not be reproduced. This suggests a true-negative result further supported by a negative result in a second confirmatory PCR. In 2/8 reexamined IF-negative/PCR-positive samples, cysts were found confirming them as true positives. Based on these findings, the dataset was modified accordingly and referred to as the “Reexamination” in the following analyses.

Diagnostic performance: IF vs. ELITe InGenius PJ PCR assay

The diagnostic performance of the ELITe InGenius PJ PCR assay was evaluated against IF as the reference standard (Figure 2). The area under the ROC curve (AUC) in the “Primary data” was 0.82, and when using the assay’s limit of detection (LoD; 97 cp/ml) as the cut-off, it demonstrated a sensitivity of 75.0% and specificity of 79.2%. Following reexamination, the AUC increased to 0.9, sensitivity to 87.0% and specificity to 80.4%.

Fig. 2.

ROC curves evaluating PCR performance against microscopy as the reference standard including the “Primary data” and “Reexamination” (RXM) data sets. (a) All samples; (b) IF assay stratified by supplier (Bio-Rad vs. Merifluor); (c) IF assay stratified by sample type (BAL vs. sputum)

ROC analysis identified an optimal threshold of 6’233 cp/ml in both datasets, maximizing the sum of sensitivity and specificity. At this cut-off, the “Primary data” showed a sensitivity of 70.8% and specificity of 88.3%. After reexamination, these values improved to 82.6% and 89.4%, respectively. In both datasets, microscopy-positive samples predominantly exceed a copy number of 6’233 cp/ml, while microscopy-negative samples more often show lower values (Figure 3). This optimal threshold remains unchanged when only one sample per patient is used for the analysis.

Fig. 3.

Distribution of PCR copy numbers (log10 scale) stratified by microscopy result with a dashed line indicating the optimal cut-off at 6’233 cp/ml based on the Youden index. (a) Primary data; (b) Reexamination (RXM)

Sensitivity and specificity vary depending on the selected cp/ml cut-off, as does the Youden index incorporating both parameters (Sensitivity + Specificity − 1). As shown in Fig. 4, specificity increases steadily with higher thresholds, while sensitivity declines more rapidly beyond a certain point. The Youden index peaks at a cut-off of 6’233 cp/ml, representing optimal balance between sensitivity and specificity in both the “Primary data” and “Reexamination” datasets.

Fig. 4.

Youden index, sensitivity, and specificity plotted against PCR thresholds with the optimal cut-off (6’233 cp/ml) indicated as dashed line by peak Youden index (Max Youden). (a) Primary data; (b) Reexamination (RXM)

Subgroup analysis of IF staining method and sample type

The calculated performance of the PCR assay compared to the reference standard varied depending on the IF staining method and sample type. IF assays from two different suppliers were used: Bio-Rad (n = 148) and Merifluor (n = 74). Of the 41 IF‑/PCR+ samples, 23 were stained with the Bio‑Rad assay and 18 with Merifluor, whereas in the IF+/PCR– group only one sample was processed with Bio‑Rad and five with Merifluor.

Samples stained by Bio-Rad demonstrated a better performance, with AUC values of 0.92 (Primary Data) and 0.98 (Reexamination), compared to Merifluor-stained samples (supplementary Figure S1 and Fig. 2). This was also reflected in the Bio-Rad group’s sensitivity and specificity at the optimal threshold of 6’233 cp/ml in both datasets (Table 2). Sensitivity and specificity reached 88.9% and 92.8%, respectively, and improved to 100% and 92.9% after reexamination. It is worth mentioning that at its optimal threshold only the Bio-Rad subgroup achieved zero false negatives and 100% sensitivity after the reexamination.

Table 2.

Diagnostic performance of PCR across datasets and subgroups. Optimal cut-offs were calculated using Youden’s index (Sensitivity + Specificity − 1), and the reported sensitivity and specificity values correspond to this threshold

	AUC (95% CI)	Cut-off in cp/ml	Sensitivity (95% CI)	Specificity (95% CI)
Primary Data
All samples	0.82 (0.72–0.93)	6233	0.71 (0.54–0.88)	0.88 (0.83–0.92)
BR	0.92 (0.80–1)	6233	0.89 (0.67–1)	0.93 (0.88–0.96)
Mf	0.73 (0.58–0.89)	181,178	0.53 (0.27–0.80)	0.92 (0.84–0.98)
BAL	0.87 (0.77–0.97)	5299	0.78 (0.56–0.94)	0.90 (0.84–0.94)
SP	0.68 (0.41–0.96)	4,198,338	0.50 (0.17–0.83)	0.98 (0.95–1)
Reexamination
All samples	0.90 (0.81–0.98)	6233	0.83 (0.65–0.96)	0.89 (0.85–0.93)
BR	0.98 (0.96–1)	6233	1 (1–1)	0.93 (0.89–0.96)
Mf	0.82 (0.69–0.95)	57,750	0.73 (0.53–0.93)	0.87 (0.80–0.95)
BAL	0.94 (0.87–1)	5299	0.89 (0.72–1)	0.92 (0.86–0.96)
SP	0.75 (0.45–1)	4,198,338	0.60 (0.20–1.20)	0.98 (0.95–1)
Clinical reevaluation*
All samples	0.92 (0.84–0.99)	31,377	0.81 (0.65–0.96)	0.95 (0.92–0.98)
No false negatives*
All samples	0.98 (0.96–1)	6233	0.96 (0.87–1)	0.91 (0.87–0.94)

*Two additional datasets were used for sensitivity analysis: “Clinical reevaluation” reclassified patients with confirmed PJP as true positives, while “No false negatives” further excluded all potential false-negative cases (see section “Sensitivity analysis”)

AUC area under the curve BR bio-rad Mf merifluor, SP sputum

In contrast, samples stained by Merifluor required higher thresholds to optimize diagnostic performance (181’178 before and 57’750 after reexamination). Even in the “Reexamination” dataset, sensitivity and specificity reached only 73.3% and 88.1%, respectively, compared to 53.3% and 91.5% in the “Primary Data”.

Sample type also influenced assay performance. Among the 222 samples, 152 were BAL and 70 were sputa. BAL samples showed higher AUC values (0.87 and 0.94 after the reexamination) than sputum samples (0.68 and 0.75 after the reexamination) (Fig. 2; Table 2).

Evaluation of patient characteristics from discrepant results

We reviewed patient histories for all six IF-positive/PCR-negative cases and 10 of the 13 IF-negative/PCR-positive samples with high PCR positivity (Table 3 and detailed clinical descriptions in supplementary Table S2). The remaining three cases were excluded due to unavailable clinical information.

Table 3.

Clinical characteristics of patients with discrepant primary IF/PCR results. Left: IF+/PCR− cases; Right: IF−/PCR+ cases with high PCR positivity

Number of patients	IF+/PCR-	IF-/PCR+ (*)
	6 (4 adults + 2 children)	10 (10 adults)
Age	49.67	57
Sex: male/female	3/3	4/6
HIV positive	0	1
Transplantation	1	6
Kidney	0	3
Heart	1	1
Stem cell	0	1
> 1 organ	0	1
Neoplastic disease	2	1
Solid tumor	1	0
Myelodysplastic syndrome	1	0
Anaplastic Large Cell Lymphoma	0	1
Autoimmune disease	0	2
Autoimmune hepatitis	0	1
Temporal arteritis	0	1
Other causes	5	1
SCID	1	0
Glucocorticoids > 15 mg/d before sampling	1	1
Glucocorticoids ≤ 15 mg/d as only identifiable host factor	3	0
Other opportunistic infections or conditions associated with immunosuppression	3	7
Preceding PJP prophylaxis	3	1
PJ therapy	6	9
Therapeutic dose: adequate length or until death	6	3
Therapeutic dose: shorter than recommended	0	3
Prophylactic dosage	0	3
Death	3	1
Pneumocystis diagnosis made	6	3
Therapy initiation before sampling	0	2
Positive PJ testing result during hospital stay	6	3
Sample type: BAL/sputum	3/3	5/5
Immunofluorescence positive	6	0
Bio-Rad/Merifluor	1/5	4/6
Different IF result after Reexam/all reexamined samples	3/3	2/8
Mean of PCR results (cp/ml)	0	1’545’449

*(> 100’000 cp/ml or Ct value < 29; data not available for three patients) Patient data were collected during hospitalization only, with no follow-up after discharge

Both subgroups included a similar proportion of BAL and sputum samples. However, in the IF-positive/PCR-negative group, five of six samples were stained with Merifluor rather than Bio-Rad, including all that could not be reexamined by IF (n = 3/3). In three of the six cases, low-dose glucocorticoid use (≤ 15 mg/d) was the only identifiable host factor, which is generally not considered clinically relevant [36]. All six IF-positive/PCR-negative patients received therapeutic doses of trimethoprim-sulfamethoxazole (TMP-SMX), with three already on prophylaxis. The impact of TMP-SMX remains unclear in 3/6 cases due to their death within the first 10 days following PJP diagnosis.

In the IF-negative/PCR-positive subgroup (n = 13), despite negative IF results, three patients received prophylactic TMP-SMX and six received full treatment. In three of these treated patients, therapy was stopped within six of the recommended 21 days [37]. The other three completed ≥ 21 days of therapy and were formally diagnosed with PJP based on a later IF-positive BAL after an initial negative sputum in one patient; a highly positive external PCR result in another patient; and a prior IF-positive result (13 days earlier) with treatment initiation in the third patient. Two other IF-negative/PCR-positive patients had recent or ongoing PJ prophylaxis.

Based on these findings, additional data processing for a subsequent sensitivity analysis was conducted. First, a dataset named “Clinical reevaluation” was created, in which the three IF-negative/PCR-positive patients with confirmed PJP diagnosis during hospitalization were reclassified as true-positive (supplementary Figure S2). Using the assay’s limit of detection (LoD; 97 cp/ml) as the cut-off, sensitivity increased to 88.5%, specificity to 81.6% and the AUC reached 0.92 in this dataset. The optimal threshold was identified at 31’377 cp/ml with a sensitivity of 80.8% and specificity of 94.9% (Table 2).

As the three IF-positive/PCR-negative samples that we could reexamine proved to be true-negatives, we reclassified the three remaining IF-positive/PCR-negative samples as true-negative in a final sensitivity analysis (supplementary Figure S3). This dataset referred to as “No false negatives”, reached a sensitivity of 100% and a specificity of 81.9% (supplementary Figure S2). The optimal threshold remained at 6’233 cp/ml consistent with the “Primary data” and “Reexamination” datasets (Table 2) achieving a sensitivity of 95.7% and specificity of 91%. To reach a sensitivity of 100%, the threshold would have to be lowered to 128 cp/ml with a corresponding drop in specificity to 82.4%.

Discussion

In our study, IF detected PJ in 24 samples whereas PCR identified 59 positive samples. Based on these findings, we performed ROC analyses and determined diagnostic thresholds using Youden’s index. We also included the results from conventional microscopic staining methods (Toluidine Blue and Giemsa), which overall demonstrated lower sensitivity than IF. Although IF was used as the reference standard, it is important to note that no universally accepted gold standard exists for identifying an active PJ infection [38–40]. Thus, we conducted reexaminations of discrepant samples using IF, reviewed patient histories of discrepant samples, and performed a sensitivity analysis excluding false-negative results to contextualize our results. Our findings are in line with previous reports showing that quantitative PCR is more sensitive than IF for detecting PJ, supporting its implementation in the diagnostic process of PJP [20, 38–53].

Direct pathogen detection is a crucial part in the diagnosis of PJP and should be interpreted alongside the pretest probability from clinical and radiological findings [2, 4]. In recent years, several PCR assays have been developed to replace microscopic detection in routine diagnostics [19]. Our findings reflect the reported advantages of PCR, including higher sensitivity, reduced observer-dependency, and improved time-efficiency [9, 19, 23].

Among the 47 discrepant cases, 41 were IF-negative but PCR-positive. 13 showed a highly positive PCR result (> 100’000 cp/ml or Ct value < 29) and eight of them had enough material for IF reexamination with PJ detected in two. Clinical reevaluation showed that three additional cases were diagnosed with PJP during hospitalization, including two with negative reexaminations and one lacking material, further supporting the higher sensitivity of PCR.

In the first case, the original IF report noted “questionable trophozoites found; re-sampling suggested” and a follow-up BAL two days later tested IF-positive. The second patient, previously diagnosed and treated for PJP, had a negative IF result 13 days later but remained PCR‑positive (318’229 cp/ml), with an even higher pathogen load than in the earlier specimen (262’303 cp/ml). This likely reflects residual DNA fragments rather than active infection with viable PJ organisms, illustrating a limitation of PCR for short‑term treatment monitoring. A third HIV-positive patient showed IF-negative results one day after treatment initiation but tested strongly positive by both external PCR performed at that time (2’000’000 cp/ml) and our retrospective PCR analysis (5’644’021 cp/ml).

Although the remaining IF-negative/PCR-positive cases were not formally diagnosed with PJP, many received TMP-SMX either therapeutically or prophylactically. Active PJ infection or colonization could not be definitively excluded.

The IF-positive/PCR-negative group included six non-HIV patients. Three samples were reexamined, and none confirmed the initial IF result, suggesting true negatives. One had a prior note indicating the observed cyst could not be verified by a second examiner. The remaining three samples, all stained with Merifluor, were associated with glucocorticoid use, though two of them received only a low dosage (≤ 15 mg/d) as the only identifiable host factor.

Clinical context supports the possibility of false-positive initial IF results. One patient on 15 mg/day prednisolone for lymphocytic pneumonia and ongoing prophylaxis, died shortly after IF positivity and TMP-SMX initiation, preventing assessment of treatment response. Residual colonization from a prior infection even after successful treatment may explain the positive IF [54]. Another patient with COPD GOLD3, using intermittent inhaled steroids, improved on TMP-SMX, though alternative pathogens could not be excluded. The third patient started glucocorticoid treatment (60 mg/d) three days before sample collection, while suffering from multi-organ failure and E. coli sepsis. This patient died 10 days later and the impact of TMP-SMX remains unclear.

In the literature, 13 of 17 studies comparing IF and PCR reported no IF-positive/PCR-negative cases [20, 38, 41–45, 47, 49–53]. Of the remaining four studies, one did not perform IF on PCR-negative samples, and three reported IF-positive/PCR-negative results [39, 40, 46, 48]. In one study, four such cases were observed among 156 IF-positive samples (305 total samples total, 204 PCR-positive) attributed to two PCR inhibitors, one false-positive IF result and one unidentified technical error [40]. Another study reported five IF-positive/PCR-negative results out of seven IF-positives (222 total samples, 93 PCR-positive) with one linked to radiation pneumonitis [39]. The third study documented one IF-positive/PCR-negative case among 33 IF-positive samples (411 total samples, 61 PCR-positive), without specifying the cause [46]. Definitions of IF positivity also vary across studies and while our study considered any visible PJ form (trophozoites, cysts) as positive, others required ≥ 2 or ≥ 5 cysts for a positive result [39, 40, 47, 48]. Given these factors, the cause of IF-positive/PCR-negative discrepancies remains uncertain. Potential explanations include Merifluor-related false positive results, residual organisms post-treatment, observer variability or sample degradation. PCR inhibition can be ruled out due to the assay including an internal control, making this explanation unlikely.

The PCR assay’s limit of detection (97 cp/ml) is not appropriate as a diagnostic threshold in clinical practice. Using it as a cut-off may lead to unnecessary treatment, particularly in patient populations with a potentially higher susceptibility to colonization, such as children, immunosuppressed adults, individuals with chronic lung diseases, and healthcare workers [55]. While specificity is important, ensuring high sensitivity is crucial given the high mortality associated with PJP. Therefore, identifying a clinically meaningful cut-off that balances both sensitivity and specificity is essential for accurate diagnosis and appropriate patient management. In our study, the optimal threshold, based on the highest Youden index, was consistently identified as 6’233 cp/ml across the “Primary data”, “Reexamination” and “No false negatives” datasets.

An optimal cut-off value ideally discriminates between an acute infection and colonization. One study aiming to make this distinction and targeting the mtLSU rRNA gene proposed similar thresholds, suggesting < 10³ cp/ml may indicate colonization and > 5 × 10³ cp/ml an acute infection [56]. Another study using the same PCR target reported 100% sensitivity < 1.6 × 10³ cp/ml and 100% specificity > 2 × 10⁴ cp/ml for active PJ infection [57]. Other authors with PCR assays targeting other PJ-specific genes proposed cut-offs of > 1.3 × 10³ cp/ml and > 1.45 × 10³ cp/ml for a high probability of PJP [38, 58]. Although cut-off values are assay-specific, our findings align with other previously published thresholds and thus, our analysis suggests the following diagnostic algorithm for PJP identification using PCR results (Fig. 5). The cut-off values are derived from the validation of our specific PCR assay while integrating previous literature and its validity relies on accurate pre-analytical processing.

Fig. 5.

Flowchart illustrating the PJP diagnostic algorithm based on PCR results, highlighting the clinical consequence of different cp/ml cut-off values

A sensitivity of 100% was achieved only after reexamination in the Bio-Rad group at the 6’233 cp/ml threshold, and in the “No false negatives” dataset at a lower threshold of 128 cp/ml. Importantly, the used IF assay affects only the reference standard, not the PCR itself. The potentially lower specificity of Merifluor likely influenced the apparent performance of PCR when all samples are considered. 83.3% of IF-positive/PCR-negative results were analyzed using Merifluor, including all three that could not be reexamined. The reexamination was conducted using Bio-Rad. Literature suggests Merifluor may have higher sensitivity than conventional microscopic stains but also lower specificity with a significantly higher false-positive rate [33–35]. One author describes a positive predictive value (PPV) of 81.9% for Merifluor, compared to > 96% for the other conventional microscopic stains (Diff-Quik, Grocott-Gomori methenamine silver (GMS), and Calcofluor white stains) [33]. Another study reports one IF-positive/PCR-negative result from a total of 83 tested samples [35].

The broad gap between thresholds achieving 100% sensitivity and those yielding 100% specificity illustrates the sensitivity-specificity trade-off, introducing diagnostic uncertainty in interpreting intermediate PCR results. This reinforces the importance of integrating pretest probability from clinical and radiological findings into diagnostic decision-making. In some cases, treatment may be warranted even without any direct pathogen detection. While PCR’s high sensitivity may lead to positive results in colonized individuals due to its inability to differentiate colonization from active infection, the risk of missing a true PJP diagnosis cannot be ignored.

A major issue in evaluating PJ PCR assays is the absence of a universally accepted gold standard for PJP diagnosis even though the visualization of PJ through direct pathogen staining methods remains widely regarded as the current standard method [38–40]. Both PCR and microscopy are influenced by sample quality and type, but IF is further affected by the utilized assay. Our findings suggest that Merifluor may reduce IF specificity and representing a potential bias in PCR performance analysis. Additionally, prior treatment may degrade fungal structures and disproportionately affecting IF sensitivity. The predominant presence of trophozoites during active infection may also negatively impact microscopic detection [23]. These factors highlight the limitations of using IF as a reference standard and justify our sensitivity analyses, which demonstrated excellent performance of the ELITe InGenius PJ PCR assay. Because sputum samples showed suboptimal performance compared with BAL specimens, their results should be interpreted with caution. Additionally, all specimens except non-mucoid BAL were diluted during IF preparation, and all samples had been stored frozen for at least eight months prior to analysis. Of the 222 samples analyzed, only 24 were IF‑positive, meaning that sensitivity and specificity estimates were driven by a relatively small number of positive cases. Consequently, a follow‑up validation in a larger cohort is warranted to confirm the clinical utility and impact of these findings. Finally, the absence of residual material in certain cases limited the ability to perform IF reexaminations and the reliance on pre-existing data inherent to the retrospective design represented additional limitations.

In conclusion, our data suggests a higher sensitivity of the ELITe InGenius PJ PCR assay compared to IF for direct PJ detection. Confirmed PJP diagnoses in IF-negative/PCR-positive cases support this finding. Although the origin of discrepant results cannot always be definitively determined, the 100% negative reexamination among IF-positive/PCR-negative samples and the known limitations of IF support the reliable performance of PCR. Overall, given its higher sensitivity and robust performance, the ELITe InGenius PJ PCR assay represents a valuable addition to routine diagnostics for PJP with the potential to improve early detection and guide timely treatment decisions.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization: ID, ASZ, RDK, HFG, PK, SDB. Funding acquisition: ASZ, HFG, SDB. Analysis: HK, BK. Clinical analysis, case analysis, and materials: MAG, PMMS, UWK, FT, CB, HFG, SDB. Statistical analysis: ID, RDK, SDB. Drafting of the manuscript: ID, RDK, SDB. Review and editing: all authors. All authors read and approved the final manuscript.

Funding

This work was supported by the Swiss National Science Foundation (grant no. 211422) to SDB. The authors have no relevant financial or non-financial interests to disclose.

Data availability

All data supporting the findings of this study are available within the paper and its Supplementary Information.

Declarations

Competing interests

The authors declare no competing interests.