A Systematic Review and Meta-analysis of Diagnostic Accuracy of Serum (1-3)-Beta-D-glucan for Pneumocystis Jirovecii Pneumonia Using the Fungitell Assay: Focus on Cutoff Levels

Purpose We conducted a meta-analysis to evaluate the diagnostic performance of various levels of serum (1-3)-Beta-D-glucan(BDG) for Pneumocystis jirovecii(PJ) infection using Fungitell assay. Methods EMBASE, the with no Meta-analysis was performed using random-effects models for bivariate analysis. Subgroup analyses were implemented in HIV-positive Pneumocystis jirovecii pneumonia (PJP), HIV-negative PJP, and PJP versus colonized patients. Results Nineteen individual studies that included a total of 2,310 participants met our inclusion criteria. The overall sensitivity, specicity, positive likelihood ratio(LR+) and negative likelihood ratio (LR−),and 95 % condence interval CI of serum-BDG were 0.94(95 % CI: 0.89–0.96),0.76 (95 % CI: 0.65–0.85), 3.99 (95 % CI: 2.59–6.13),0.08 (95 % CI: 0.05–0.15), respectively. Futher stratied analysis of diagnostic values showed that various levels of serum BDG differed in sensitivity, specicity, LR+ and LR− in the diagnosis of PJP. Subgroup analyses also indicated that the cutoff value of 200 pg/mL had sucient diagnostic accuracy in HIV-positive PJP patients versus controls. Moreover, the 80 pg/mL cutoff value had satisfactory diagnostic accuracy in PJP versus colonized patients, a overall sensitivity of 0.86 (95% CI, 0.73-0.93), a overall specicity of 0.82(95% CI,0.73-0.88), a overall LR+ 4.70(95 % CI: 3.11–7.08), and a overall LR−0.17(95 % CI: 0.09–0.34),individually. This meta-analysis the positive for BDG, the Fungitell and clinical validation rather the recommended cutoff of 80pg/mL by the manufacturer in the diagnosis of PJP. These could be further rened additional patients without HIV. When positive BDG is applied to patients, the results should be interpreted cautiously and in conjunction with clinical and radiological ndings. BAL is the preferred specimen for the diagnosis of PJP, in which the PJ-PCR result is highly sensitive and specicity. However, there are some limitations in differentiating patients with infection or colonization. Our meta-analysis showed that the positive results of serum BDG combined with PJ-PCR have high sensitivity and specicity in differentiating infection from colonization. This meta-analysis suggests that the optimal positive threshold for serum BDG requires better denition and clinical validation rather than the recommended cutoff of 80pg/mL in the diagnosis of PJP as well. And these cutoff values could be further rened in additional studies that focus on populations that are as homogeneous as possible. More studies are also needed, particularly large prospective cohort studies, to conrm the diagnosis value of serum BDG detection.


Study selection
Inclusion criteria were: 1) cross-sectional study for assessing the diagnostic value or cohort study for assessing the predictive value; 2) PJ infections were de ned as either de nite, probable, possible, or not PJP, as described for other invasive fungal diseases,with some modi cations (Table 1) [20][21], and the de nite and probable patients were our target population , namely PJP patients; 3) serum BDG measured by the Fungitell assay in all subjects and used recommended or different cutoff values than in current PJP criteria as an index test; 4) used the same reference standard for all participants to avoid differential veri cation; and 5) allowed reproduction of the diagnostic or predictive 2 x 2 contingency table (i.e., number of true-positive, false-negative, truenegative, and false-positive cases) for BDG tests. Disagreements were discussed among the group until a consensus was reached.

Data extraction
From each study two investigators (W. Long and K. Xiao) independently extracted the following information: rst author, publication year, BDG cutoff level, geographical region,language, mean or median age of subjects,population characteristics, study design, prophylactic antifungal therapy, HIV status of patient population and diagnostic data for two-by-two tables using a standardized protocol and reporting document to ensure consistency. If the criteria for the diagnosis of PJP were not directly provided, the article would be excluded in the present study. When analyzing the same population in several publications, the results were calculated only once. When evaluating HIV-related and -unrelated PJP patients in the same report, we emailed authors for further information and extracted data separately if available.
Study quality was performed using the Quality Assessment of Diagnostic Accuracy Studies 2 score (QUADAS-2) tool [23], which is a quality assessment tool comprised of 4 domains: patient selection, index test, reference standard, and ow and timing. The risk of bias is assessed in each of the domains by indicating a 'low', 'high' or'unclear' rating. Of the 4 domains, the following 10 items were focused on as being most relevant for our purpose [23]. Domain 1 (Patient Selection): 1) Was a consecutive or random sample of patients enrolled? 2) Was a case-control design avoided? 3)Did the study avoid inappropriate exclusions?; Domain 2 (Index Test): 4) Were the index test results interpreted without knowledge of the results of the reference standard? 5) If a threshold was used, as it presigni ed? ; Domain 3 (Reference Standard): 6) Is the reference standard likely to correctly classify the target condition?; 7) were the reference standard results interpreted without knowledge of the results of the index test?; Domain 4 (Flow and Timing):8) Was there an appropriate interval between the index test and reference standard?; 9)Did all patients receive the same reference standard?; 10) Were all patients included in the analysis?.

Data synthesis
We carried out a meta-analysis of the point estimate of sensitivity, speci city, LR+ calculated as sensitivity/(1-speci city) , LR-calculated as (1sensitivity)/speci city , and diagnostic odds ratios (DOR) calculated as sensitivity/(1-speci city) ×speci city /(1-sensitivity) and their corresponding 95% CI by using a bivariate random-effects. On the other hand, we constructed a hierarchical summary receiver operating characteristics (HSROC) curve by using the two independent parameters that have normal distribution (accuracy, cutoff point) and scale parameters that allow asymmetry of the ROC curve [24]. Both models show different aspects of ROC data, but both are very closely related and are usually the same [25]. LR+ and LR-values would be used to assess diagnostic or predictive ability in PJP screening. We judged results with LR+ >10, 5 to ≤10, 2 to ≤5, and <2 as providing conclusive, strong, weak, and negligible evidence, respectively, for con rming PJP (i.e.,diagnosing or predicting PJP with a high level of con dence).
The results for heterogeneity were assessed by calculating a Q statistic, which we compared with a χ² distribution, and the I-squared index [26]. The Q test shows the statistical signi cance of homogeneity hypothesis and the I 2 index measures the degree of heterogeneity. The results of heterogeneity would be considered to be signi cant at P<0.10 (both sided). Publication bias was ascertained by visually examining for funnel plot asymmetry and quanti ed by using the Egger's test to calculate two tailed P values [27].

Diagnostic accuracy of serum BDG for HIV-negative PJP patients versus controls subgroup
Five studies [32][33]42,[44][45] reported the diagnostic accuracy of serum BDG at various cutoffs of 80,200,300,400,and 500 pg/mL in HIV-negative PJP patients versus controls subgroup, individually. The detailed results of strati ed analysis of the point estimates of sensitivity, speci city, LR+, LR-, and DOR are shown in Table 5. Except for 80mg/mL stratum (P=0.397), the heterogeneity studies in the other strata were consistently signi cant (P<0.001, for any other strata).
Except for the threshold of serum BDG 80mg/mL stratum (P=0.087),no signi cant publication bias was performed by Egger's test (P>0.10, for any other strata).

Discussion
The current meta-analysis indicated that the usefulness of serum BDG at a recommended cutoff value of 80 pg /mL,measured by the Fungitell assay,had high pooled overall sensitivity (0.94,95%CI 0.89-0.96) for the diagnostic ability of PJP individuals, similar to the previous three meta-analysis studies [52-54]. However, our strati ed analysis showed that the ability to detect PJP patients was further lowered, as shown in Table 5 with a serum BDG cut-off value of 200 pg /mL (0.78,95%CI 0.66-0.87),300 pg /mL(0.77,95%CI 0.64-0.87),400 pg/mL(0.66,95%CI 0.52-0.78),500pg/mL(0.59,95%CI 0.49-0.68), respectively.There are obviously differences in comparison with the previous studies Salerno et al. [34] reported that the sensitivity was 0.91(95%CI, 0.83-0.99) at a threshold of 300 pg /mL and Esteves et al. [39] reported that the sensitivity was 0.928 at a threshold of 400 pg /mL. Secondly, our meta-analysis also analyzed the diagnostic sensitivity of serum BDG to PJP at various cutoff values in different HIV states, and the results showed that more sensitive in patients with HIV than in those without ,cutoff value of 80 pg/mL (0.94 vs. 0.87), 200 pg/mL (0.93 vs. 0.66), 300 pg/mL (0.87 vs. 0.64), 400 pg/mL (0.74 vs. 0.54), 500 pg/mL (0.70 vs. 0.47),individually. To our knowledge,it was the rst meta-analysis study to stratify the diagnostic sensitivity of plasma BDG to PJP, and there was high diagnostic sensitivity at various thresholds in HIV-positive populations, but the diagnostic sensitivity of other subgroups was not ideal. It may be interpreted that the superior diagnostic sensitivity of BDG in HIV patients was related to the higher burden of PJ. Finally, we particularly analyzed a subgroup of the PJP patients versus possible/colonized patients, which also had good pooled overall diagnostic sensitivity (0.86,95% 0.73-0.93) at a threshold of 80 pg/mL, indicating that serum BDG detection could effectively distinguish PJP from colonization. When PJ-PCR was introduced, it signi cantly increased the sensitivity to the diagnostic sensitivity of PJP, resulting in an increase in the number of false positive cases of PJP [28,56-57]. Given the high sensitivity of BDG detection in serum, negative serum B-D-glucan assay can rule out PJP in patients at risk for disease, particularly those who's a BAL test is not feasible [55].
Although our meta-analysis also indicated that the speci city of BDG for the diagnosis of PJP was moderate at the recommended threshold of 80mg/mL, with a pooled overall speci city of 0.76 (95%CI, 0.65-0.85), 0.68 (95%CI, 0.35 -0.89) for HIV-negative PJP patients subgroup, and 0.71 (95%CI,0.49-0.86) for HIVpostive patients subgroup, similar to what was previously described [52][53][54]. Nevertheless,our meta-analysis estimated that the combination of the BALF analyses PJ-PCR and serum BDG yielded a pooled overall speci city of 0.82(95% CI,0.73-0.88) at the lowest cut-off level of 80 mg/mL. The present metaanalysis described that serum BDG with cut-off level ≥200 pg/ml yielded speci city above 94% as well, suggesting a very high diagnostic performance for PJP patients.As a result, we suggest that PJP could be excluded by raising BDG threshold or combining with BALF PJ-PCR results.In particular, a higher cutoff value is appropriate for screening in populations whose BALF samples are not readily available.However, further studies are needed to con rm the validity of this recommendation.
Nevertheless,false positive tests for serum BDG may occur up to 35% of cases [41]. In fact, β-glucan is not speci c to PCP, it is also included in the criteria for deep-seated fungal diseases,such as invasive candidiasis and invasive aspergillosis [29]. False-positive results from serum BDG assays are mostly due to several conditions, presenting in patients undergoing hemodialysis through cellulose membranes, receipt of albumin and immunoglobulin products, concurrent gram-negative endotoxemia, use of cotton swabs during surgery and with certain medications such as amoxicillin/clavulanic acid [22,[30][31]. Therefore, if only serum BDG positive results are obtained, the diagnosis of PJP should be interpreted with care to avoid misdiagnosis and delay the patient's condition,and additional clinical and laboratory information such as PCR, BAL, computed tomography (CT) scan or autopsy will be required to validate the diagnosis of PJP.
In the clinical setting, a rapid laboratory BDG assay is critical to diagnose or rule out PJP, which may allow physicians to postpone or even skip the BAL.Although most of the articles included in our meta-analysis were retrospective studies, diagnostic ability of serum BDG LR+ to diagnose PJP in HIVpositive, HIV-negative, or PJ-PCR-positive people was judged to be 'weak' at the recommended threshold of 80pg/mL. A positive serum BDG in these populations should prompt a comprehensive microbiological study to allow the diagnosis of bacterial and fungal diagnosis other than PJP. And it appears more and more clearly that serum BDG detection of PJ appears to be a more appropriate strategy for the diagnosis of PJP in HIV patients.Our ndings, if con rmed, may indicate that the optimal cut-off level for diagnosing PCP may be different from serum BDG for invasive fungal infections. However, the above ndings should be explained in terms of the high statistical heterogeneity noted in all the studies we analyzed. The presentation of the patient and the presence or absence of specimens as described above will in uence the approach to the interpretation of the test.In such cases, choosing a higher threshold value can help identify those patients with clinically signi cant disease. Raw patient data were unavailable, so the optimum cut-offs were unable to determine.Future studies are needed to clarify this topic.
Some limitations should be addressed. Firstly, we could not access unpublished reports, which may allow publication bias. Secondly, most of the studies included in our meta-analysis were retrospective case-control studies, and there were signi cant issues of selection bias when evaluating the quality of the included studies. We found signi cant heterogeneity in the pooled sensitivity and speci city estimates. This may owe to the differences in the target population and the biases associated with patient selection in these case-control studies. Thirdly, several studies do not evaluate serum BDG as the sole purpose of diagnosing PJP, which may lead to an overestimate of sensitivity and speci city. Forthly, some studies did not perform hierarchical assessment of the diagnostic ability of serum BDG for PJP or the strati cation strategies were different among studies, resulting in data loss or inaccuracy. Fifthly, the time of specimen collection is critical. The time interval between the serum BDG test and the patient's diagnosis of PJP is different between studies, or some specimens have been collected before the use of prophylactic therapies, which may affect the results of serum BDG test. Finally, the included trials were not large enough for us to obtain su cient data to explore the potential interference between the various factors contributing to signi cant heterogeneity. These factors may have potential effects on the ndings in the current study. As a result,the ndings of this meta-analysis need to be con rmed by more welldesigned and large-scale investigations.
In conclusion, the main strength of this study is that the strati ed systematic review and meta-analysis of existing studies, although most of the studies included were retrospective controlled studies, were su cient to assure us that serum BDG tested by the Fungitell assay at various thresholds differed signi cantly in the ability to diagnose PJP patients with and without HIV. The current meta-analysis suggests that a negative serum-BG result can be used as a laboratory indicator to "exclude" PJP, especially in patients without HIV. When positive BDG is applied to patients, the results should be interpreted cautiously and in conjunction with clinical and radiological ndings. BAL is the preferred specimen for the diagnosis of PJP, in which the PJ-PCR result is highly sensitive and speci city. However, there are some limitations in differentiating patients with infection or colonization. Our meta-analysis showed that the positive results of serum BDG combined with PJ-PCR have high sensitivity and speci city in differentiating infection from colonization. This meta-analysis suggests that the optimal positive threshold for serum BDG requires better de nition and clinical validation rather than the recommended cutoff of 80pg/mL in the diagnosis of PJP as well. And these cutoff values could be further re ned in additional studies that focus on populations that are as homogeneous as possible. More studies are also needed, particularly large prospective cohort studies, to con rm the diagnosis value of serum BDG detection.
a Shortness of breath, dyspnea on exertion, increased O 2 requirements, dry or productive cough, and hemoptysis.