Plasma virion reverse transcriptase activity and heat dissociation-boosted p24 assay for HIV load in Burkina Faso, West Africa

Background: In resource-limited settings, the requirement for inexpensive, easy-to-perform viral load monitoring has increased with greater antiretroviral drug availability.

Objectives: To evaluate feasibility, in Burkina Faso, of a simple assay for plasma HIV reverse transcriptase (RT) activity quantification compared to heat dissociation-boosted (HDB) p24 antigen and RNA-based quantifications in plasma samples from HIV-infected patients.

Methods: Plasma viraemia was quantified by RT activity, HDB-p24 and RNA copies in 84 samples from 70 HIV-1 group M-infected patients (82% non-B subtype, 93% treatment naive), including serial samples from nine patients.

Results: RT activity detected 86% of plasma samples containing measurable RNA copies; corresponding to 0, 93 and 100% of samples with 1.7-4.0 log₁₀, 4.1-4.8 log₁₀ and 4.9-6.7 log₁₀ RNA copies/ml, respectively. HDB-p24 detected 77% of plasma samples containing measurable RNA copies; corresponding to 27, 80 and 86% of samples with 1.7-4.0 log₁₀, 4.1-4.8 log₁₀ and 4.9-6.7 log₁₀ RNA copies/ml, respectively. Measurement error based on one-way analysis of variance between RT activity and HDB-p24 values with RNA copies showed good agreement with RT activity (ME, <10%), however poorer agreement was obtained with HDB-p24 values (ME, >10%). Patient follow up showed a similar pattern of viraemia with RNA and RT activity assays.

Conclusion: Field trials in Burkina Faso support the practical use of plasma RT activity assay as an affordable alternative for HIV viral load determination in regions where RNA detection remains difficult to perform. HDB-p24 use requires further evaluation before being considered as an alternative method in African HIV-infected patient follow up.

Introduction

RNA viral load quantification for the management of HIV-infected patients in industrialized countries relies on high maintenance technology and remains costly. Treatment with antiretroviral drugs is becoming more widely available in resource-limited countries. However, the cost of monitoring HIV infection remains an unaffordable part of treatment. Although central reference laboratories have become equipped with real-time PCR or conventional nucleic acids amplification-based assays, peripheral laboratories will require less demanding technologies.

Furthermore, viral load assays designed for use in Africa must also take into account the increasing genetic diversity of HIV as nucleic acid amplification is limited due to genome variability ^[1,2].

We previously demonstrated that plasma virion-associated RT activity was a reliable alternative for HIV load determination, whatever the HIV group and type ^[3]. In the present study, we evaluated a commercially available kit for quantification of RT activity in Burkina Faso, Africa. Viral load was measured by RNA copy-based, RT activity-based and heat dissociation-boosted (HDB) p24-antigen assays. We also investigated the reliability of RT activity assay for patient follow up.

Methods

Population

Eighty-four plasma samples were collected from 70 HIV-1 group M-infected patients in Ouagadougou. HIV-1 group and subtype were determined on plasma as previously described ^[4]. Seven samples (five patients) were infected by subtype B viruses while the remaining samples contained subtype non-B viruses. Most samples (93%) were obtained from treatment-naive patients and six samples were from six antiretroviral treated patients. Twelve samples were obtained from patients 2 months to 5 year old, 12 samples from 7-12-year-old patients, and 60 samples from patients aged 15 years or older. Plasma was aliquoted and stored at -20°C within 6 h of EDTA-preserved blood collection.

Plasma HIV RNA assay

HIV RNA load was determined with the Cobas Amplicor HIV-1 Monitor test v1.5 (Roche Diagnosis, Meylan, France) ^[5]. The detection limit was 1.7 Log₁₀ RNA copies/ml.

Isolation of virion-associated reverse transcriptase (RT) and RT activity assay

RT was recovered from plasma and quantified as previously described ^[3] using a commercially available RT activity kit (ExaVir Load version 1.0, Cavidi Tech AB, Uppsala, Sweden). After inactivation of cellular enzymes and removal of interfering factors, virion-associated RT was recovered with lysis buffer. RT activity was analysed by enzyme immunoassay (EIA) quantification of the DNA amount produced, using colorimetric or fluorimetric detection. The results calculated using the ExaVir Load Analyzer are expressed in RNA copies/ml equivalents and in fg HIV-1 RT activity/ml plasma, with a detection limit of 2 fg/ml.

HDB-p24 antigen ELISA

HDB-p24 antigen quantification was performed according to Nadal et al. ^[6]. One-hundred microlitres plasma, diluted 1: 6 with 0.5% Triton X-100 in phosphate-buffered saline (PBS), was heat denatured for 5 min at 100°C. P24-antigen was quantified with the Innotest HIV antigen ELISA kit (Innogenetics NV, InGen, Rungis, France) boosted by the ELAST ELISA amplification system (PerkinElmer Life Sciences, Courtaboeuf, France). Samples with an end-point absorbance above the mean absorbance of five HIV-negative controls run on the same plate plus five standard deviations were considered positive ^[6] and expressed as fg HIV-1 p24/ml plasma.

Within- and between-run variations

Identical plasma aliquots were quantified for RT activity in two to seven replicates. Within-run variation was measured by processing five samples (4.3-5.6 log₁₀ RNA copies/ml, one subtype B and four subtype non-B) and between-run variation was measured on six samples (4.3-5.6 log₁₀ RNA copies/ml, one subtype B and five subtype non-B).

Serial dilution variation

Seven plasma samples were serially diluted in healthy blood donor plasma and processed for RT quantification. Samples with 4.3-5.6 log₁₀ RNA copies/ml, one subtype B and six subtype non-B, were diluted up to 3- or 20-fold, for RNA copies smaller or higher than 5 log₁₀, respectively.

Statistical analysis

All the values above cut-off were log₁₀ transformed and used for analyses. Variability between RT activity replicates was calculated using SD. The association between RT activity and HDB-p24 values with RNA copies was determined using the Spearman correlation coefficient. To determine agreement between two methods, we plotted difference against average of the two methods as recommended by Altman and Bland ^[7,8]. Due to the lack of relationship between mean level and difference of the methods, we calculated the so-called 'measurement error' (ME, equivalent to the coefficient of variation), based on one-way analysis of variance as: se/m, where se is the within-subject error and m is the mean of total values. ME was considered good when <0.10.

Results

RT activity assay shows low within-run, between-run and serial dilution variations

RT activity processed in Burkina Faso produced within-run SD of 0.06-0.09 log₁₀ and between-run SD of 0.06-0.19 log₁₀. Compared with undiluted plasma, RT activity differences obtained with serial plasma dilutions up to 20-fold were not considered significant, with SD in the range 0.03-0.40 log₁₀.

Sensitivity of RT activity assay relative to HIV RNA assay

Eighty-four identical plasma samples from 70 HIV-1 group M-infected patients were assayed for RT activity both in Ouagadougou (colorimetric detection) and in Rouen (colorimetric and fluorimetric detections). Plasma from HIV-seronegative individuals, routinely included in each experiment, were repeatedly negative for RT activity. RT activity values were compared with HIV RNA values in HIV-1 group M plasma samples (Table 1). In Ouagadougou, 72 samples (86%) were positive for both RT activity and HIV RNA. RT activity was measurable in 93% of samples containing 4.1-4.8 log₁₀ RNA copies/ml and in 100% of samples containing ≥4.9 log₁₀ copies/ml. Eleven samples containing <4.0 log₁₀ RNA copies/ml were positive for RNA alone, five of these samples were from patients receiving antiretroviral therapy (ART). In Rouen, RT activity was measurable in 36 and 73% of samples containing 1.7-4.0 log₁₀ RNA copies/ml for colorimetric and fluorimetric detection, respectively, and 100% of samples containing ≥4.1 log₁₀ RNA copies/ml with both detection methods (Table 1). Two patients (subtype non-B) had RNA loads >4.0 log₁₀ copies/ml and non-measurable or low RT activity. One 5-year-old child had a 4.8 log₁₀ RNA load smaller than those quantified for the age group 11 months to 5 years (mean ± SD, 5.9 ± 0.3 log₁₀ RNA copies/ml) and the RT activity value was 5 fg/ml in Rouen and undetectable in Ouagadougou. The second patient had recently been diagnosed HIV-1 seropositive; the RNA value was 5.3 log₁₀ and the RT activity value was 18-61 fg/ml by colorimetry-fluorimetry.

Table 1

Sensitivity of HDB-p24 antigen assay relative to HIV RNA assay

Sixty-five plasma samples (77%) were positive for both HDB-p24 antigen and HIV RNA. HDB-p24 antigen was positive in 27, 80 and 86% of samples with 1.7-4.0, 4.1-4.8 and 4.9-6.7 log₁₀ RNA copies/ml, respectively (Table 1). The 19 plasma with HDB-p24 antigen values below the cut-off were from 18 adult patients and one 10-year-old child. The child, with 3 log₁₀ RNA copies/ml, and three adults with 2.2-3.2 log₁₀ RNA copies/ml were treated with ART. The other adults with 3.6-5.6 log₁₀ RNA copies/ml were ART naive.

Quantitative analysis of RT activity and HDB-p24 antigen assays relative to HIV RNA assay

The Spearman correlation coefficients were considered statistically significant with values equal to 0.85 (P < 0.001) for RT activity and 0.39 (P = 0.005) for HDB-p24, indicating a positive relationship between each of the two methods and RNA quantification. Altman and Bland plots are shown in Fig. 1a and b. The mean difference (95% limit of agreement) versus RNA was equal to 0.3 (-0.5-1.1) for RT activity and equal to 0.7 (-0.87-2.27) for HDB-p24. These plots demonstrate that mean differences between new and reference methods are not related to the average level of the values. ME, calculated when differences between two methods vary over the range of measurement, was 7% for RT activity and 15% for HDB-p24. The agreement between RT activity by fluorimetry and colorimetry was assessed with ME equal to 2%.

Fig. 1

Serial samples follow up

RT activity and HDB-p24 and RNA copies were quantified in serial plasma collected from one HIV-1 subtype non-B-infected adult patient naive of treatment (Fig. 1c). Changes in RT activity and RNA values were similar. Curves were less matched for HDB-p24 and RNA values.

Serial viral load was evaluated in a further eight patients (two subtype B and six non-B). RT activity and RNA curves again showed similar changes, but HDB-p24 and RNA showed discrepant curves (data not shown).

Discussion

In Burkina Faso, we compared RT activity, using an early version of ExaVir Load EIA and HDB-p24 with RNA copies. Correlation coefficient showed an association between measurements. The difference between two methods of measurements demonstrated significant agreement between RT activity and RNA methods. In contrast, a limited agreement was found between HDB-p24 and RNA methods.

In Burkina-Faso, CRF06-cpx together with CRF02-AG and subtypes A and G predominate ^[9]. RT is crucial for viral replication and its activity is not affected by the broad sequence diversity of HIV subtypes, groups and types ^[3,10,11]. RT-activity was routinely quantified in plasma >1800 copies/ml with fluorimetric detection. In Ouagadougou, as in other developing countries, only colorimetric detection can be performed with a routine detection limit of 10 000 copies/ml. Fluorimetric and colorimetric detection were equivalent (ME = 2%). A newer generation RT assay recently launched with an increased sensitivity, according to the manufacturer, now exists with the aim of improving monitoring.

The discrepancy observed for two plasma whose RNA-based quantifications were >10 000 copies/ml and the RT activity low or undetectable could reflect a proportion of defective virions. This hypothesis is supported by previous reported studies showing that plasma HIV RNA level exceeded, by an average of 60 000-fold, virus titres measured by endpoint dilution culture ^[12] and that total virions exceeded culturable infectious units by factors of 10⁴-10^{7 [13]}. One-tenth RT-defective virions, in the viral population, could result in RT quantification in agreement with the results we obtained, i.e., below the limit of RT detection for the child and in the range of RT activity (12-83 fg/ml) quantified for 4.3 log₁₀ RNA copies/ml for the adult. Attempts to isolate the virus starting from the adult plasma through spinoculation ^[14] was negative (data not shown).

Eleven plasma samples with RNA viral load >4.1 log₁₀, all from adult patients, were not measurable for HDB-p24 antigen. The p24 antigen kit we used is unlikely to be type or subtype dependent ^[15]. Previous studies reported that HDB-p24 could reliably quantify non-B HIV plasma in Belgium ^[16], Switzerland ^[10], Thailand ^[17], Malawi ^[18], that the correlation with RNA load was fairly strong in South Africa ^[19] but weak in Ivory Coast ^[20]. The detection failure tends to show that heat dissociation of immune complex remains difficult in plasma from adult African patients containing high antibody levels ^[20,21]. These results strengthen the importance of field evaluations with clinical samples. HIV RNA, RT activity or HDB-p24 loads could be equivalent for early diagnosis of African infants born to seropositive mothers as immune complex should be easier to dissociate. HDB-p24 has been used successfully for early diagnosis in Tanzanian infants ^[22] but follow up of ART-treated African adults by RNA load and HDB-p24 quantification showed discrepant patterns ^[16,20]. Our data show that RNA load curves matched RT activity curves but not HDB-p24 curves.

In summary, robust, simple and easy-maintenance technologies are required as an affordable alternative to viral RNA quantification in peripheral laboratories. Our field data support the practical routine use of RT activity assay which could have a major impact on the monitoring of ART.

Acknowledgements

The authors thank the technicians of the Virology Laboratory CHU Charles Nicolle, Rouen and Richard Medeiros, Medical Editor, Rouen University Hospital, for his valuable advice in editing the manuscript. This work was supported by ANRS, the French agency for AIDS and viral hepatitis research, France.