Systematic Review into Diagnostics for Post-Kala-Azar Dermal Leishmaniasis (PKDL)

Abstract

Identification of post-kala-azar dermal leishmaniasis (PKDL) is important due to the long and toxic treatment and the fact that PKDL patients may serve as a reservoir for visceral leishmaniasis (VL). We summarized the published literature about the accuracy of diagnostic tests for PKDL. We searched Medline for eligible studies investigating the diagnostic accuracy of any test for PKDL. Study quality was assessed using QUADAS-2. Data were extracted from 21 articles including 43 separate studies. Twenty-seven studies evaluated serological tests (rK39 dipstick, ELISA, DAT, and leishmanin tests), six studies molecular tests, eight microscopy, and two cultures. Only a few of these studies reported a valid estimate of diagnostic accuracy, as most were case-control designs or used a reference standard with low sensitivity. The included studies were very heterogeneous, for example, due to a large variety of reference standards used. Hence, no summary estimates of sensitivity or specificity could be made. We recommend well-designed diagnostic accuracy trials that evaluate, side-by-side, all currently available diagnostics, including clinical symptoms, serological, antigen, molecular, and parasitological tests and possible use of statistical modelling to evaluate diagnostics when there is no suitable gold standard.

Figure 1

Flow of included studies.

Figure 2

(a) Raw ROC plot of the serological tests. Every symbol refers to the sensitivity (y-axis) and specificity (x-axis) of a test in a study. The height of the symbol represents the number of diseased in the study, and the width of the symbol represents the number of nondiseased in the study. Circles = DAT; diamonds = rK39 LF; squares = rK39 ELISA; triangles = other serological tests. The majority of studies are clustered into the upper hand left corner, indicating a near perfect sensitivity and specificity; however, poor patient selection must be taken into account when drawing conclusions. (b) Raw ROC plot of PCR and IHC. Every symbol refers to the sensitivity (y-axis) and specificity (x-axis) of a test in a study. The height of the symbol represents the number of diseased in the study, and the width of the symbol represents the number of nondiseased in the study. Circles = PCR; diamonds = IHC. The majority of studies have a sensitivity of above 90%, but specificity varies widely.

Figure 3

Forest plots of included studies. Overview of all 2 × 2 tables with forest plot (TP = true positives; FP = false positives; FN = false negatives; TN = true negatives; DAT = direct agglutination test; PCR = polymerase chain reaction; IHC = immunohistochemistry). Numbers 1–4 refer to different set of data from the same paper.

Figure 4

QUADAS-2 results. (a) shows the risk of bias from each of the 41 2 × 2 contingency tables. Note that less than 20% of the included 2 × 2 tables had a low risk of bias for patient selection. (b) shows the risk of applicability of the patients, index, and reference tests to use in PKDL areas.