Poor Reproducibility of HIV­1 Low-level Viraemia Results with 3 Commercial Real-time PCR Assays

1. Meeting 2011 September 30, Leuven • 1. Background and Objectives • Ultra-sensitive real-time PCR-based HIV-1 viral load assays theoretically allow the detection of any RNA copy present in the sample, allowing the detection of residual viral replication in patients previously defined as virologically suppressed. • Recently, studies on the clinical outcome of HIV‐1 patients experiencing low-level replication (<50 cop/ml) in comparison to fully suppressed patients were published. • We hypothesised that this exceeds the technical limit of PCR. Hence, we tested the analytical variability below 50 cop/ml of three commercial real‐time assays. Intra-assay variability Each participating laboratory tested 3 times40 samples with detectable VL below 50 cop/ml, in separate runs. Variability was assessed qualitatively with a discrepancy score: a score of +1 was attributed if any of the replicates crossed the thresholds of <1 or >50 cop/ml. Poor Reproducibility of HIV­1 Low-level Viraemia Results with 3 Commercial Real-time PCR Assays • 2. Methods • Assays tested: • - Versant HIV-1 RNA 1.0 kPCR (Siemens), limit of quantification (LOQ) 37 cop/ml. • - Realtime HIV-1 (Abbott), LOQ 40 cop/ml. • Cobas Ampliprep/Cobas Taqman HIV-1 v2 (Roche Diagnostics), LOQ 20 cop/ml. • Analysis were performed in the clinical routine settings of Belgian AIDS reference laboratories, following the assay manufacturers’recommandations (one assay per site). • Samples: • Prospective collection in each participating laboratory from HIV-1 positive patients. • 3 aliquots of 1.1 ml of plasma from fresh EDTA-blood samples were stored at -80°C, no supplementary thaw/freeze cycle was applied. • The selection was made at random: the only criteria applied was a detectable plasma viral load (VL) under 50 cop/ml with the first aliquot. • The same assay lot was used throughout the study, except for the re-evaluation of blips. When a detectable sample with VL under 50 cop/ml was tested in triplicate, the result is not reproduced in more than 50% of cases where at least one replicate was not detected. As a consequence, when the VL is not detected in clinical routine settings, we should not consider it as a proof of full viral suppression. Inter-assay variability 61 clinical samples with detectable VL below 50 cop/ml were tested in triplicate, one time on each platform. Mean score was 0.7, 52% of the samples were not detected by at least one assay, 11.5% detected by one assay only. 3. Results Dilutions of a reference sample A shared sample used as a quality control tested on each run was used as a reference. Its absolute value was defined as the mean of all the results obtained in routine settings with the present control lot in the 3 labs. Ten replicates at 4 target dilutions of 100, 50, 25 and 12.5 cop/ml were tested on each platform. In the boxplot, the bottom and the top of the box represent the lower and upper quartiles, respectively. The triangle in the box is the median. The ends of the whiskers represent the minimum and the maximum of all data. No assay gave signal for 10 replicates of the HIV negative plasma used for dilutions. Si: Versant HIV-1 RNA 1.0 kPCR (Siemens) ; Ab: Realtime HIV-1 (Abbott); Ro: Cobas Ampliprep/Cobas Taqman HIV-1 v2 (Roche) Jean Ruelle1, Laurent Debaisieux2, Ellen Vancutsem3, Annelies De Bel3 and Patrick Goubau1. 1:  UCLouvain, AIDS Reference Laboratory, Avenue Hippocrate 54 - B1.54.05, 1200 Brussels, Belgium. 2:  Hôpital Universitaire Erasme, AIDS Reference Laboratory, Route de Lennik 808, 1070 Brussels, Belgium.   3:  Vrije Universiteit Brussel, AIDS Reference Laboratory, site Universitair Ziekenhuis Brussel, Laarbeeklaan 101, 1090 Brussels,  Belgium.     Contact author: jean.ruelle@uclouvain.be Blips As variability is high round the clinically relevant limit of 50 cop/ml, we reanalysed 26 blips, defined as a VL between 50 and 100 cop/ml experienced by patients whose previous and next sample was <50 cop/ml. A low but detectable VL under therapy may be considered as treatment failure, or a parameter used in clinical studies to consider a patient as non-responder. The VL blips could only be reproduced in 19.2% of cases. In the remaining cases, the sample had a VL below 50 or completely undetectable. In 1 case (4 %), the VL was above 100 cop/ml. Therefore, the variability measured here is in agreement with the 2011 revision of DHHS guidelines which define the virological failure threshold as 200 cop/ml. All assays showed important variability; it was the highest for the Roche assay. The Abbott assay was here less sensitive for very low target concentrations, and the Siemens assay quantified under 50 cop/ml some aliquots of the 100 cop/ml target dilution. 4. Conclusions The most recent versions of widespread commercial real‐time assays show an important variability at VL below 50 cop/ml. Patient outcome studies based on very low‐level viraemia are biased. Blips under 100 cop/ml are mostly due to analytical assay variability. Acknowledgements: Companies provided free assay kits for the evaluation. Abbott and Roche diagnostics offered support to calculate quantitative results from the PCR Cq values under the LOQ.