Performance of Cryptococcal Antigen Lateral Flow Assay Using Saliva in Ugandans with CD4 <100

Abstract

BackgroundCryptococcal meningitis can best be diagnosed by cerebrospinal fluid India ink microscopy, cryptococcal antigen detection, or culture. These require invasive lumbar punctures. The utility of cryptococcal antigen detection in saliva is unknown. We evaluated the diagnostic performance of the point-of-care cryptococcal antigen lateral flow assay (CrAg LFA) in saliva.MethodsWe screened HIV-infected, antiretroviral therapy naïve persons with symptomatic meningitis (n = 130) and asymptomatic persons with CD4+<100 cells/µL entering into HIV care (n = 399) in Kampala, Uganda. The diagnostic performance of testing saliva was compared to serum/plasma cryptococcal antigen as the reference standard.ResultsThe saliva lateral flow assay performance was overall more sensitive in symptomatic patients (88%) than in asymptomatic patients (27%). The specificity of saliva lateral flow assay was excellent at 97.8% in the symptomatic patients and 100% in asymptomatic patients. The degree of accuracy of saliva in diagnosing cryptococcosis and the level of agreement between the two sample types was better in symptomatic patients (C-statistic 92.9, κ-0.82) than in asymptomatic patients (C-statistic 63.5, κ-0.41). Persons with false negative salvia CrAg tests had lower levels of peripheral blood CrAg titers (P<0.001).ConclusionThere was poor diagnostic performance in testing saliva for cryptococcal antigen, particularly among asymptomatic persons screened for preemptive treatment of cryptococcosis.

The Point-of-Care Laboratory in Clinical Microbiology

“…In terms of the direct detection of pathogens, we created a POC meningitis menu that was systematically aimed at testing for enterovirus (NAAT), Streptococcus pneumoniae (NAAT), Neisseria meningitidis (NAAT), and herpesviruses, including herpes simplex and varicella-zoster viruses (NAAT), in cerebrospinal fluid collected from symptomatic patients clinically suspected of having meningitis (66). As for HIV-infected patients, an LFA detecting Cryptococcus neoformans polysaccharide capsule glucuronoxylomannan antigen performed adequately on CSF (67), plasma, serum, and urine samples (68), but tests on saliva were disappointing and should not be recommended (69). A few commercially available multiplexed NAATs currently under development would detect six bacteria, eight viruses, and two Cryptococcus species within 1 h.…”

Preventing Cryptococcosis—Shifting the Paradigm in the Era of Highly Active Antiretroviral Therapy

“…Attempts to improve the specificity of the urine assay by altering the diluent have been unsuccessful, leading to a reduction in test sensitivity compared to serum LFA as reference and continued sub-optimal specificity (sensitivity decreasing from 100 % to 80 % and specificity increasing from 73.8 % to 91.5 %) [ 46 ]. Using samples of saliva, a recent Ugandan study found excellent specificity but poor sensitivity (88 % in symptomatic patients and only 27 % in asymptomatic patients) of the LFA in detecting cryptococcal antigenemia using serum/plasma samples for reference [ 73 ]. In a review of the published literature addressing performance of the LFA assay, median CSF sensitivity was 100 %, and median specificity was 97.7 %.…”

Cryptococcal Meningitis: Diagnosis and Management Update

“…In a recent study in Uganda, there was perfect agreement between fingerstick whole blood, serum, and plasma CrAg LFA suggesting that testing from fingerstick whole blood is a viable option for detecting antigen CrAg, particularly in settings where phlebotomy is not available, or in patients with difficult venous access (15). The CrAg LFA has also been evaluated in both urine and saliva, but agreement with serum LFA was not sufficient to recommend routing screening using these fluids (16, 17). …”