Diagnosis and monitoring of HIV in infants: investigating the first fourth generation rapid test and two viral load technologies for use in the South African setting
Date
2014-11-17
Authors
Bhowan, Kapila
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Human immune deficiency virus (HIV) infection contributes to child mortality rates in South Africa.
Investigations of newer technologies for improving early infant diagnosis of HIV in the South
African setting could reduce child mortality as life saving treatment can be accessed early in life.
This study investigated three technologies: a fourth generation rapid HIV test and two viral load
(VL) platforms.
Determine Combo (DC) is a qualitative fourth generation rapid test that is able to detect HIV
antibodies and p24 antigen simultaneously. The performance of DC was evaluated in the field on
samples from pregnant and postpartum women; in the laboratory, on stored samples from children
and with the addition of heat denaturation.
In the maternal DC study 90 (8 .8%) of 1019 women tested HIV positive of whom 59 (17.1%
prevalence) were pregnant and 31 % (4.6% prevalence) were postpartum. The sensitivity and
specificity of the antibody component of DC on plasma was 100%(Confidence Interval (CI): 95.9-
100%) and 99.8%(CI: 99.2-99.9%) respectively. Three postpartum patients tested false positive for
HIV antibodies (n=2) and p24 antigen (n=1). No true positive p24 antigen was detected
DC was performed on stored samples from 182 (90%) HIV-exposed and 20 (10%) HIV-unexposed
children aged from birth to six years. The DC HIV antibody component returned false negative
results in 2 HIV-infected children; one clinically symptomatic and one asymptomatic aged 7 and 23
months respectively. The sensitivity of DC HIV antibody was 100% (CI :94.3-100%) in infants aged
6 months and younger with a specificity of 100% (CI:81.6-100%) for all ages. Of the 61 HIV infected
infants tested , the DC p24 antigen was reactive in only one clinically symptomatic infant
resulting in a sensitivity for detection of HIV infection of 1.7% (CI 0.3-8.9%).
A heat denaturation technique designed to improve p24 antigen detection was applied to HIVinfected
samples but failed to enhance p24 antigen detection on DC.
HIV viral load (VL) molecular assays are used to confirm an HIV-infected diagnosis and for VL
monitoring. In South Africa, plasma is the gold standard sample for VL monitoring in infants even
though dried blood spots (OBS) are the preferred specimen type in resource-constrained settings
and for early infant diagnosis. The use of OBS specimens for HIV VL monitoring would
convenience resource limited settings. The OBS matrix therefore requires validation to determine
accuracy (for establishing diagnosis) and precision (for VL monitoring) compared to plasma VL.
This study investigated the accuracy and precision limits of OBS VL on the Roche Cobas
AmpliPrep-Cobas TaqMan HIV-1 v2.0 assay (CAP/CTM) and the Abbott RealTime HIV-1 assay
(m2000) platforms on samples from HIV-infected adults and children. The CAP/CTM was
investigated on OBS containing 751J1 blood and the m2000 was investigated using one (50IJI) and
two (2x501J1) OBS.
Compared to plasma VL, OBS VL from adults and children were higher in the lower range
«310g,<1000copies/ml) and lower values in the higher range (>510g, >185,000copies/ml) on the
CAP/CTM in the study of OBS VL accuracy. Additionally, OBS VL values were >log1.0 higher in
42/100 (42%) of adult and 16/49 (33%) of measurements from children, which will have clinical
significance. On the m2000 platform, the differences between plasma and OBS VL were lower in
the range >5 log and higher in the range 2 log copies/ml (100 copies/ml) to 4 log copies/ml (10000
copies/ml). Compared to plasma VL, OBS VL values were >log1.0 higher in 20/82 (24%) adult and
7/43 (16%) of measurements from children.
Both platforms demonstrated 100% specificity in testing stored OBS from HIV-uninfected infants
who were diagnosed negative on HIV DNA PCR.
Acceptable limits for plasma VL precision is a coefficient of variation (CV) <35% and standard
deviation (SO) :50.19 log. Where plasma VL :5510g, OBS VL demonstrated poor precision with
CV>40% in 8/10 patients and total SO>0.30 log in 4/10 patients on the CAP/CTM. The m2000
total SO was >210g between adult plasma and OBS VLs under the 4 log copies/ml cut-off,
irrespective of the number of DBS used. DBS VLs were unreliable when using precision limits
used on plasma VLs on both platforms.
In conclusion the DC test does not offer any advantage over currently available rapid tests in
diagnosing new infection in women and children. The two VL platforms can be used to establish
an HIV status in treatment naive patients in view of the 100% specificity. HIV-infected patients on
treatment with undetectable plasma VL will always have detectable DBS VL on CAP/CTM, but
equally undetectable DBS VL on the m2000. With DBS, the CAP/CTM assay generates higher VL
values in the lower VL range than on plasma likely due to amplification of proviral DNA. Both
platforms display poor intra- and inter- assay precision, using plasma VL based criteria and the
variances would potentially affect clinical decision making. The acceptable limits for plasma VL
precision cannot be applied to DBS VL on either platform.
Description
Thesis (M.Sc. (Med.))--University of the Witwatersrand, Faculty of Health Sciences, 2013.