The diagnostic accuracy of serologic and molecular methods for detecting visceral leishmaniasis in HIV infected patients: meta-analysis

Abstract

Background: Human visceral leishmaniasis (VL), a potentially fatal disease, has emerged as an important opportunistic condition in HIV infected patients. In immunocompromised patients, serological investigation is considered not an accurate diagnostic method for VL diagnosis and molecular techniques seem especially promising.

Objective: This work is a comprehensive systematic review and meta-analysis to evaluate the accuracy of serologic and molecular tests for VL diagnosis specifically in HIV-infected patients.

Methods: Two independent reviewers searched PubMed and LILACS databases. The quality of studies was assessed by QUADAS score. Sensitivity and specificity were pooled separately and compared with overall accuracy measures: diagnostic odds ratio (DOR) and symmetric summary receiver operating characteristic (sROC).

Results: Thirty three studies recruiting 1,489 patients were included. The following tests were evaluated: Immunofluorescence Antibody Test (IFAT), Enzyme linked immunosorbent assay (ELISA), immunoblotting (Blot), direct agglutination test (DAT) and polimerase chain reaction (PCR) in whole blood and bone marrow. Most studies were carried out in Europe. Serological tests varied widely in performance, but with overall limited sensitivity. IFAT had poor sensitivity ranging from 11% to 82%. DOR (95% confidence interval) was higher for DAT 36.01 (9.95-130.29) and Blot 27.51 (9.27-81.66) than for IFAT 7.43 (3.08-1791) and ELISA 3.06 (0.71-13.10). PCR in whole blood had the highest DOR: 400.35 (58.47-2741.42). The accuracy of PCR based on Q-point was 0.95; 95%CI 0.92-0.97, which means good overall performance.

Conclusion: Based mainly on evidence gained by infection with Leishmania infantum chagasi, serological tests should not be used to rule out a diagnosis of VL among the HIV-infected, but a positive test at even low titers has diagnostic value when combined with the clinical case definition. Considering the available evidence, tests based on DNA detection are highly sensitive and may contribute to a diagnostic workup.

Figure 1. Terms used in PubMed search.

Figure 2. Study selection process.

Figure 3. Tests performance summary.

Footnote: Immunofluorescence antibody test (IFAT), Enzyme linked immunosorbent assay (ELISA), Immunoblotting (BLOT), direct agglutination test (DAT) and polymerase chain reaction (PCR) in whole blood, standard deviation (SD), 95% confidence interval (95% CI), *standard error (SE). * SE is a measure of precision and it is not a measure of confidence interval, which is shown in sROC plot, except for ELISA.

Figure 4. Estimates of sensitivity and specificity (95% confidence interval) of tests.

Footnote: Combined results are shown using both options: fixed and random effects model. When both results are similar with low heterogeneity, both can be used. When they are different, we prefer results from random effects model, which gives wide and conservative confidence interval for heterogeneous results.

Figure 5. sROC curve for PCR in peripheral whole blood.