False Positive Ultrasensitive HIV bDNA Viral Load Results in Diagnosis of Perinatal HIV-Infection in the Era of Low Transmission

Use of an Indeterminate Range in HIV Early Infant Diagnosis: A Systematic Review and Meta-Analysis

BACKGROUND

Expanded access to HIV antiretrovirals has dramatically reduced mother-to-child transmission of HIV. However, there is increasing concern around false-positive HIV test results in perinatally HIV-exposed infants but few insights into the use of indeterminate range to improve infant HIV diagnosis.

METHODS

A systematic review and meta-analysis was conducted to evaluate the use of an indeterminate range for HIV early infant diagnosis. Published and unpublished studies from 2000 to 2018 were included. Study quality was evaluated using GRADE and QUADAS-2 criteria. A random-effects model compared various indeterminate ranges for identifying true and false positives.

RESULTS

The review identified 32 studies with data from over 1.3 million infants across 14 countries published from 2000 to 2018. Indeterminate results accounted for 16.5% of initial non-negative test results, and 76% of indeterminate results were negative on repeat testing. Most results were from Roche tests. In the random-effects model, an indeterminate range using a polymerase chain reaction cycle threshold value of ≥33 captured over 93% of false positives while classifying fewer than 9% of true positives as indeterminate.

CONCLUSIONS

Without the use of an indeterminate range, over 10% of infants could be incorrectly diagnosed as HIV positive if their initial test results are not confirmed. Use of an indeterminate range appears to lead to substantial improvements in the accuracy of early infant diagnosis testing and supports current recommendations to confirm all initial positive tests.

Early infant diagnosis HIV-1 PCR cycle-threshold predicts infant viral load at birth

BACKGROUND

HIV-1 viral load (VL) has been found to be an independent predictor for disease progression among untreated HIV-infected children. However, qualitative polymerase chain reaction (PCR) assays are routinely used for early infant diagnosis (EID).

OBJECTIVES

To predict HIV-1 VL at birth using qualitative EID real-time PCR cycle-threshold (Ct) values.

STUDY DESIGN

This study was a secondary analysis of results from a cohort of intrauterine HIV-1 infected neonates. Neonates were enrolled at Rahima Moosa Mother & Child Hospital in Johannesburg, South Africa between June 2014 and November 2017. Laboratory EID HIV-1 PCR testing was performed at birth using COBAS AmpliPrep/COBAS TaqMan HIV-1 Qualitative Test v2.0 (EID CAP/CTM). Some infants had simultaneous EID point-of-care (POC) testing using Xpert HIV-1 Qualitative assay (EID Xpert). Neonates with a confirmed HIV-1 detected EID result and plasma HIV-1 RNA VL test were included in this analysis. Bland-Altman analysis was used to determine extent of agreement between Ct values of both EID assays. Multivariable linear regression models adjusting for time between EID and VL testing were used to describe the association between EID Ct values and VL and to predict VL at given EID Ct values.

RESULTS

Among 107 HIV-1 infected neonates included in the study, 59 had POC EID testing. Median VL was 28 400 copies per millilitre (cps/ml) (IQR: 1 918-218 358) - two neonates had VL < 100 cps/ml prior to antiretroviral therapy initiation. There was good correlation between Ct values of both EID assays (Spearman correlation coefficient 0.9, 95% CI: 0.8-1.0). The limits of agreement between EID CAP/CTM and Xpert Ct values were 4-11 cycles. For every one cycle increase in Ct value there was 0.3 log₁₀ RNA decrease (95% CI: -0.3 to -0.2) for both EID assays. An EID CAP/CTM Ct value ≤ 23 and an EID Xpert Ct value ≤ 31 predicted a VL of > 5.0 log₁₀ cps/ml in 82.2% (95% CI: 73.9-88.3) and 84.7% (95% CI: 73.7-91.8) of cases, respectively.

CONCLUSION

EID Ct values at birth predict VL and accurately identify infants with VL > 5.0 log₁₀ cps/ml.