Sensitivity and Specificity of Histoplasma Antigen Detection by Enzyme Immunoassay

Introduction

Histoplasma capsulatum is a dimorphic soil-borne fungus found throughout most of the Southeastern and Midwestern United States.¹ Prevalence is high within the Mississippi, Missouri, and Ohio River valleys. In a study of mongrel dogs in Kentucky, 154 out of 386 (39.6%) had confirmed Histoplasma organisms on necropsy, with higher prevalence in rural locations.² Infection commonly occurs after inhalation of the microconidia. Within tissues, microconidia are converted to a yeast phase. If the organism is not eliminated, further multiplication and dissemination occurs within macrophages, causing multisystemic pyogranulomatous inflammation.¹

Clinical signs include fever, lymphadenopathy, dyspnea, cough, lethargy, and diarrhea.^1,3,4 Occasionally, lameness and central nervous system signs are also noted. Similar to humans without underlying immunosuppression, the majority of exposed dogs experience subclinical infection.² With clinical disease, the lungs, gastrointestinal tract, lymph nodes, spleen, liver, bone marrow, and eyes are most commonly affected.^1,3–5

Historically, diagnosis has relied on detection of the organism on either cytology or histopathology of fluid or tissue samples or by performing fungal culture of affected tissues. The organism is slow growing and specialized laboratories are required. As such, fungal culture is not routinely relied upon clinically. In addition, previous studies have documented false negative fungal culture results from samples, including tissue biopsy, the buffy coat, and lymph node aspirates.⁶ Several studies have defined cytological identification of the organisms, either within macrophages or freely in pyogranulomatous exudates, as the diagnostic gold standard.^3–6 Organisms can be missed cytologically with low organism burdens and nonrepresentative aspirates. Further, histopathological sections can be falsely negative without the addition of special stains, such as Gomori methenamine silver, which is specific for fungal organisms. Serologic tests, including complement fixation and immunodiffusion are available, but are associated with poor sensitivity.^1,4,6,7 It has been shown in humans that, in acute disease, serology can be negative for the first 2 mo after infection.⁵ Also, false positive results due to cross-reactivity with other fungal organisms may occur.⁸ Lastly, past versus active infection can be difficult to define with serology alone.^3,5,6

Currently, enzyme immunoassay (EIA)-based antigen detection is available as a noninvasive means of diagnosis of histoplasmosis in people.^5,9–11 Antigen quantification is based on the identification of the terminal (1-6)-α-D-galactofuranose, a major component of Histoplasma galactomannan, found in the fungal cell wall. Previous studies have identified cross-reactivity with other fungal organisms, including a high level of cross-reactivity with Blastomyces spp., Paracoccidioidomyces spp., and Penicillium marneffei. A lower level of cross-reactivity was found with Coccidioidomyces spp., and only rarely with Aspergillus spp. (<10%).^5,11–13 No cross-reactivity with Cryptococcus spp. was found.¹⁴

The purpose of this study was to define the sensitivity and specificity of the EIA-based third-generation Histoplasma antigen assay and the detection of antigenuria in the diagnosis of canine histoplasmosis.

Materials and Methods

The medical records of dogs from Oklahoma State University, Texas A&M University and Kansas State University small animal hospital that had a urine sample submitted to a diagnostic laboratory^a for EIA-based antigen assay were reviewed. Dogs were confirmed to have histoplasmosis only if Histoplasma capsulatum organisms were identified either by cytological examination of tissue aspirates or bronchoalveolar lavage (BAL) fluid or histological examination of tissue biopsies. Dogs were determined to have an alternative diagnosis only if the diagnosis was supported with cytological or histopathological findings of the organ system involved. Information retrieved from the medical records of all dogs that had urine antigen EIA testing included date of testing, the organ system(s) involved, and cytological or histological findings. The quantitative limit of detection of the EIA was 0.6 ng/mL. Values lower than that were still considered positive but were not quantifiable. A negative EIA test result required finding no antigen.

Patients with incomplete medical records, patients without a definitive diagnosis, and patients having been administered antifungal treatment within 1 mo prior to EIA testing were excluded from the study. In cases where more >1 urine sample was tested, only the initial test result was considered.

True positives were defined as a positive antigen test and either cytological or histological confirmation of disease. True negatives were defined as a negative antigen test and a final diagnosis other than histoplasmosis. False positives had a positive antigen test and a final diagnosis other than histoplasmosis. False negatives had a negative antigen test and Histoplasma organisms found on either cytology or histopathology. A sensitivity, specificity, negative predictive value, positive predictive value, and kappa coefficient with 95% confidence interval were calculated using commercial software^b.

Results

A total of 133 cases had at least one urine sample submitted for Histoplasma antigen testing between January 2007 and December 2011. Six cases were excluded from analysis due to incomplete case information in the medical record (n = 3) or previous treatment with an antifungal agent (n = 3). An additional 67 cases were excluded from analysis due to lack of a definitive diagnosis. Sixty cases met the inclusion criteria and remained for further analysis.

Seventeen cases had EIA results classified as true positives. Organisms were found on cytological (n = 16) or histopathological (n = 1) examination of affected tissues. Cytology samples included rectal scraping (n = 6); peripheral blood smear (n = 3); BAL fluid (n = 2); bone marrow aspirate (n = 2); and fine-needle aspirate of liver (n = 4), spleen (n = 3), lung (n = 2), and lymph nodes (n = 2). The single case diagnosed on histopathology was based on identification of Histoplasma organisms within lung tissue on postmortem examination.

Urine antigen levels for the 16 dogs with cytological evidence of Histoplasma organisms ranged from <0.6 to 24.04 ng/mL with a median of 11.9 ng/mL (Figure 1). Two of those cases had positive values less than the quantifiable limit of 0.6 ng/mL.

View Figure

• Download as Powerpoint
• Download Figure

Forty-one cases had EIA results classified as true negatives. Definitive diagnoses included neoplasia (n = 12); inflammatory bowel disease (n = 7); meningitis (n = 3); heartworm disease (n = 2); bacterial diskospondylitis/osteomyelitis (n = 3); immune-mediated polyarthritis (n = 2); hepatic disease (n = 2); bacterial pneumonia (n = 2); and one case each of immune-mediated neutropenia, coccidioidomycosis, pancreatitis, eosinophilic bronchopneumopathy, pythiosis, pulmonary fibrosis, bacterial pyelonephritis, and bacterial sepsis. All of the above cases were classified as having a definitive alternative diagnosis by either the failure to identify Histoplasma organisms on cytological/histopathological examination of affected tissue (n = 34) or the absence of Histoplasma on postmortem examination (n = 7). Additionally, two of the cases with cytologically negative samples had long-term response to steroids, further decreasing the likelihood of histoplasmosis. Several cases where bacterial disease was diagnosed were also supported with a positive bacterial culture (n = 7), including bone (n = 3), urine (n = 1), abdominal effusion (n = 1), and BAL fluid (n = 2). No cases had EIA results that were classified as false positives.

Two cases had EIA results classified as false negatives. Both dogs had negative urine EIA, but Histoplasma organisms were found on histopathological examination. Both of those cases had clinical signs confined to the gastrointestinal tract. Organisms were identified in the colon of both patients in addition to the duodenum and ileum of one dog.

Based on the above findings, the Histoplasma urine antigen EIA has an overall specificity and positive predictive value of 100% because no false positives were observed. The sensitivity was 89.47%, and the negative predictive value was 95.35%. Given that no false positives were observed, the kappa coefficient measuring the agreement between traditional means of diagnosis and the Histoplasma urine antigen EIA was 0.9207 (95% confidence interval, 0.8131–1), indicating a very strong agreement.

Discussion

To the authors' knowledge, this is the first study to investigate the utility of urine antigen EIA to support the diagnosis of canine histoplasmosis. The clinical utility of this assay is well established in the human literature for both respiratory and disseminated histoplasmosis.^5,9–11 The results of the current study are consistent with a human study reporting an overall sensitivity of 95.4% and specificity of 99% and one feline study reporting a sensitivity of 94.4%.^11,15

One limitation of this study is that in the areas in which the participating hospitals are found, systemic mycoses other than histoplasmosis are relatively uncommon. Considering the cross-reactivity with other fungal organisms, the specificity of this assay is likely to be lower in a region with a higher prevalence of other mycoses. Clinically, cross-reactivity may not be a significant issue. Although definitive diagnosis is ideal, treatment of histoplasmosis is similar to that of blastomycosis. As such, a misdiagnosis of histoplasmosis as blastomycosis, or vice versa, is expected to alter the treatment plan minimally. Given the limitations of this retrospective study, no conclusions can be drawn regarding the cross-reactivity of this assay with other fungal organisms; however, it is important to note that the single case of Pythium in this study was Histoplasma antigen negative.

Histoplasmosis in dogs is commonly multisystemic. Much like humans and cats, most canine gastrointestinal histoplasmosis is thought to be a part of disseminated disease. Although uncommon, reports have been made in humans of acute pulmonary manifestations without dissemination.⁵ Human studies have shown that the distribution of Histoplasma organisms has a direct effect on the sensitivity of the antigen EIA test. In patients with acute pulmonary histoplasmosis, the antigen EIA sensitivity is reported to be 77%. With progressive disseminated disease, it has a sensitivity of 92%.⁵ A separate human study showed a positive antigen EIA in 19 out of 20 cases of progressive disseminated disease, with only 8 out of 17 cases of subacute or chronic pulmonary disease being EIA positive.¹⁶ Based on the location of organisms found, the vast majority of dogs in the current study had disseminated disease.

Clinical signs of histoplasmosis are commonly confined to the gastrointestinal tract in dogs, found in approximately 78% of the cases in one study.³ Despite intestinal histoplasmosis being associated with disseminated disease in cats, there is one clinical report of isolated intestinal involvement. Cases of gastrointestinal histoplasmosis in humans are considered to have gastrointestinal involvement as a manifestation of progressive disseminated disease. In the current study, the two cases of histoplasmosis in which antigenuria was not detected were dogs with Histoplasma organisms only identified within the gastrointestinal tract. In those cases, clinical signs were also isolated to the gastrointestinal tract. It is possible that finding the organisms in the gastrointestinal tract may have limited further investigation that would have revealed Histoplasma organisms elsewhere. Interestingly, in a previous study in cats, the single cat with histoplasmosis and a negative EIA was in a cat in which organisms were only identified within the gastrointestinal tract.¹⁵

It is possible that testing for antigenemia in addition to antigenuria may have improved antigen assay sensitivity. Although sensitivity associated with various sample types was not an objective of this study, in previous human studies, antigenemia and antigenuria have been assessed concurrently.^9,11 Those studies showed that antigenuria was more sensitive and was detectable earlier compared with antigenemia.⁹ It remains unknown how assessment of antigenemia in addition to antigenuria would affect the ability to diagnose histoplasmosis in dogs.

An additional limitation of this study is inherent to its retrospective nature and includes a lack of consistency in case management and intensity of diagnostic work up. Strict inclusion criteria, with intention to minimize the consideration of dogs without a definitive diagnosis, were used. Diagnostic criteria were likely more stringent than those used clinically to warrant a specific treatment. It is possible that only including dogs with organisms found on cytology or histopathology selected for dogs with higher fungal burdens, thus falsely representing the sensitivity of the assay.

Future studies should be aimed at a prospective evaluation of confirmed histoplasmosis cases with a standardized diagnostic investigation (including detection of antigen in serum) and standardized treatment. In addition, investigation of the correlation of antigen concentration with clinical manifestations of histoplasmosis, including disease severity is warranted. Finally, further evaluation of testing techniques which, according to a conversation with J. Wheat, MD, Medical Director and Founder of MiraVista Diagnostics, in November 2012, would include ultrafiltration methods to detect weakly positive antigen samples, as well as tests used to differentiate between the similar fungal antigens in blastomycosis and histoplasmosis should be refined for our continued use of these tests in veterinary patients.¹⁹

Conclusion

The antigen EIA is a clinically useful test to support the diagnosis of histoplasmosis in dogs, specifically in regions where there is a high prevalence of disease. In regions where other fungal disease is more prevalent, the specificity will be lower. The utility of this assay in cases with clinical disease limited to the gastrointestinal tract requires further investigation. The Histoplasma antigen EIA should be considered in cases of suspected canine histoplasmosis when more invasive diagnostic investigation is not feasible.