Diagnosis of Progressive Disseminated Histoplasmosis in Advanced HIV: A Meta-Analysis of Assay Analytical Performance

Histoplasmosis is an important cause of mortality in people with advanced HIV, especially in countries with limited access to diagnostic assays. Histoplasmosis can be diagnosed using culture, histopathology, and antibody, antigen, and molecular assays. Several factors may affect the analytical performance of these laboratory assays, including sample type, clinical stage of the disease, and previous use of antifungal treatment, among others. Here we describe the results of a systematic literature review, followed by a meta-analysis of the analytical performances of the diagnostic laboratory assays employed. Our initial search identified 1631 references, of which 1559 references were excluded after title and abstract screening, leaving 72 references identified as studies relevant to the validation of histoplasmosis diagnostic assays. After evaluating the full text, 30 studies were selected for final review, including one paper not identified in the initial search. The meta-analysis for assay analytical performance shows the following results for the overall sensitivity (Sen) and specificity (Spe) of the various methods evaluated: Culture, Sen 77% (no data for specificity calculation); antibody detection assays, Sen 58%/Spe 100%; antigen detection assays, Sen 95%/Spe 97%; and DNA detection assays (molecular), Sen 95%/Spe 99%. Of the 30 studies reviewed, nearly half (n = 13) evaluated Histoplasma antigen assays, which were determined to be the most accurate methodology for diagnosis of progressive disseminated histoplasmosis in advanced HIV (inverse of the negative likelihood ratio was 13.2). Molecular assays appear promising for accurate diagnosis of histoplasmosis, but consensus on exact techniques is needed. Cultures showed variable sensitivity related to sample type and laboratory handling. Finally, antibody assays presented high specificity but low sensitivity. This poor sensitivity is most likely due the highly immunosuppressed state of this patient population. Diagnostic assays are crucial for accurate diagnosis of progressive disseminated histoplasmosis (PDH) with advanced HIV disease.


Introduction
Histoplasmosis is a disease caused by the thermally dimorphic fungus Histoplasma capsulatum. This disease has been reported worldwide, but is most frequently diagnosed in the Americas [1]. H. capsulatum is frequently found in soil, especially where it is contaminated with bird excreta and bat guano [2]. H. capsulatum primarily causes pulmonary infection when the human host inhales infectious propagules (microconidia and mycelial fragments) after soil disturbance. It can spread secondarily to other organs, especially those of the reticuloendothelial system [2]. In persons with advanced Human Immunodeficiency

Meta-Analysis Forest Plot
The analytical performance of the assays and meta-analysis results were graphically displayed using forest plots. Graphics represent summary measures and confidence intervals, and squares are proportional to the weights (study sample size) used in the meta-analysis [14].

Plot of Fitted Model
The resulting graph shows the following summaries, along with circles showing the individual study estimates: (i) a summary curve from the hierarchical summary receiver operating characteristic (HSROC) model; (ii) a summary of operating points (summary values for sensitivity and specificity); (iii) a 95% confidence region for the summary operating point; (iv) and a 95% prediction region (the confidence region for a forecast of the true sensitivity and specificity in a future study) [14]. Pairs of sensitivity and specificity extracted from each primary study are plotted on a ROC space, showing between-study heterogeneity as well as a relationship between sensitivity and specificity. The HSROC curve is plotted as a curvilinear line passing through summary point [14].

Hierarchical Summary Receiver Operating Characteristic (HSROC) Curves
HSROCs were used to evaluate the assays' accuracy. The positive and negative likelihood ratios (LR+ and LR-) were estimated. Larger values of the LR+ and of 1/LR-both indicate higher accuracy of the assay and can be compared to evaluate whether a positive or negative test result has a greater impact on the odds of disease. HSROC confidence regions and prediction regions were used to evaluate heterogeneity within and between studies [14].

Literature Search
We identified 1259 references using the narrow search and 2228 references with the broad search. After the removal of duplicate references, 1631 references were evaluated. A total of 1559 references were excluded after a manual review of titles and abstracts; the reasons for exclusion are summarized in Figure 1. Seventy-two studies were related to the validation of diagnostic assays for histoplasmosis in PLHIV. Of the 72 publications, 29 were selected; seven related to the validation of culture methodologies, five were related to assays for detection of antibodies, twelve were related to detection of circulating antigen, and five were related to molecular testing. By expert opinion, one study related to antigen testing validation that was not identified in the systematic review was included, increasing the number of studies related to antigen assay validation to thirteen and the total number of publications selected to 30, Figure 1  . There were no identified studies related to histopathology validation.

Evaluation of Culture Assays
Seven studies were identified [16][17][18][19][20][21][22]. Of that, two had enough available information to calculate the analytical performance of the culture according to the type of sample [21,22]. The overall sensitivity for culture was 77% (95%CI 72%-81%), Figure 2. Studies that evaluated blood culture processed by lysis methodology and those using bone marrow, Figure 2 (studies 1, 2 and 4) showed the highest analytical performance; sensitivity ranged between 60% and 90%. Respiratory samples presented poor sensitivity (from 0% to 60%). None of these studies have a non-histoplasmosis control group; for that reason, it was not possible to calculate assay specificity or HSROC curve.

Evaluation of Culture Assays
Seven studies were identified [16][17][18][19][20][21][22]. Of that, two had enough available information to calculate the analytical performance of the culture according to the type of sample [21,22]. The overall sensitivity for culture was 77% (95%CI 72%-81%), Figure 2. Studies that evaluated blood culture processed by lysis methodology and those using bone marrow, Figure 2 (studies 1, 2 and 4) showed the highest analytical performance; sensitivity ranged between 60% and 90%. Respiratory samples presented poor sensitivity (from 0% to 60%). None of these studies have a non-histoplasmosis control group; for that reason, it was not possible to calculate assay specificity or HSROC curve.

Evaluation of Culture Assays
Seven studies were identified [16][17][18][19][20][21][22]. Of that, two had enough available information to calculate the analytical performance of the culture according to the type of sample [21,22]. The overall sensitivity for culture was 77% (95%CI 72%-81%), Figure 2. Studies that evaluated blood culture processed by lysis methodology and those using bone marrow, Figure 2 (studies 1, 2 and 4) showed the highest analytical performance; sensitivity ranged between 60% and 90%. Respiratory samples presented poor sensitivity (from 0% to 60%). None of these studies have a non-histoplasmosis control group; for that reason, it was not possible to calculate assay specificity or HSROC curve.

Discussion
Using a literature search and meta-analysis, we evaluated the analytical performance of multiple laboratory assays used to diagnose progressive disseminated histoplasmosis in PLHIV with advanced HIV disease. The results of the meta-analysis showed that laboratory assays based on the detection of circulating Histoplasma antigen demonstrated the best analytical performance. However, these results may be partially related to the number of antigen assay studies that were available for evaluation (n = 13) compared to those available to evaluate the other methodologies (n = 5-7).
Because validated Histoplasma antigen assays are commercially available, incorporating this testing into clinical laboratories is often easier, reducing the technical problems that can be related with performance of in-house assays. In locations where these assays have been implemented, the performance of these more sensitive and specific assays have resulted in a significantly increased number of patients diagnosed compared with conventional diagnostic tests (culture and histopathology) [28,30,31]. Likewise, the relative ease and speed in which these antigen assays can be performed has resulted in a shortened time of diagnosis and a reduction of mortality associated with PDH in PLHIV [44][45][46][47]. Antigen assays can also be performed in laboratories that do not have the higher levels of biocontainment needed for handling Histoplasma cultures for conventional identification or DNA extractions for molecular identification. An advantage of ELISA-based antigen assays is that they are designed to easily evaluate a larger number of patient specimens. However, a limitation is that some of these assays require multiple wells (up to 8-10 wells) for quality control and standard curve formation, reducing the cost effectiveness of the assaying when only a small number of samples are tested. Development of simple and rapid diagnostic technology, like laboratory assays based on immunochromatographic assays are being explored [29].
The results from this study also demonstrated that molecular testing may be another alternative for accurate diagnosis of histoplasmosis in PLHIV. However, the lack of commercially available kits, the lack of standardized methods, and the limited number of validation studies are current limitations in performing these assays in laboratories. It has yet to be determined which DNA extraction method, gene target and primers, or amplification methodology is optimal, which will most likely vary between the use in culture confirmation and the direct detection in patient specimens. Furthermore, the definitions of invasive fungal diseases published by the EORTC/MSG Consensus Group in 2008 do not include molecular assay as a methodology for the diagnosis of endemic mycoses due to a lack of consensus on gene targets and laboratory protocols for molecular assays. Further investigations, laboratory consensus and development of commercial kits are needed [8,48].
Culture is still considered the gold standard for the diagnosis of histoplasmosis, however the results of this literature review and meta-analysis show variable sensitivity when the cultures are not properly processed (i.e., blood culture processed by lysis centrifugation versus not processes by lysis centrifugation). Selection of the proper specimen is crucial; blood and bone marrow cultures provided the highest assay sensitivity, while the other specimen types demonstrated lower sensitivity. Additionally, there are several limitations in performing the culture for the diagnosis of histoplasmosis. Primarily, the need for laboratory infrastructure for handling isolates (biosafety level 3), the need for staff who have appropriate laboratory training and experience, and a prolonged turn-around time for results. This final variable is directly associated with an increased risk of patient mortality [49]. However, conventional assays like culture or special histopathologic stains may be the only option for laboratory diagnosis of certain forms of histoplasmosis (e.g., muco-cutaneous, central nervous system and pulmonary localizations) [3].
Results from this meta-analysis demonstrated that antibody detection assays have high specificity, but the sensitivity of this type of diagnostic assay in PLHIV with advanced diseases was poor. These findings can be explained by the highly immunosuppressed state of this patient population. However, the detection of specific anti-Histoplasma antibodies in patient's sera or cerebrospinal fluid could be a complementary diagnostic tool, particularly in PLHIV with subacute and chronic forms of PDH where patients have a progressive long-term infection characterized by a lower fungal burden [3,11]. Additionally, it is important to mention that antibody detection assays are useful for the diagnosis of histoplasmosis in non-immunosuppressed patients (subacute and chronic histoplasmosis) [11].

Conclusions
Diagnostic assays are crucial to improving the care of patients with advanced HIV disease who are most at risk of developing progressive disseminated histoplasmosis (PDH). Results from this meta-analysis demonstrated that antigen and molecular diagnostic assays had greater sensitivity and specificity in identifying PDH in PLHIV compared with culture and antibody assays. In contrast, multiple diagnostic assays may be helpful in identifying non-disseminated histoplasmosis. Further investigation, diagnostics developments, and guidelines for diagnosis are needed to improve capacity to rapidly detect PDH in PLHIV. Finally, this systematic review is limited to the analysis assays performance on PDH in PLHIV, due the variety of histoplasmosis clinical presentations, further systematic reviews and meta-analysis are needed.