- Email: [email protected]
HIV-1 viral load measurement in venous blood and fingerprick blood using Abbott RealTime HIV-1 DBS assay
Journal of Clinical Virology 92 (2017) 56–61
Contents lists available at ScienceDirect
Journal of Clinical Virology journal homepage: www.elsevier.com/locate/jcv
HIV-1 viral load measurement in venous blood and fingerprick blood using Abbott RealTime HIV-1 DBS assay
MARK
⁎
Ning Tanga, , Vihanga Pahalawattaa, Andrea Franka, Zowie Bagleyb, Raquel Vianab, John Lampinena, Gregor Leckiea, Shihai Huanga, Klara Abravayaa, Carole L. Wallisb a b
Abbott Molecular, 1300 East Touhy Avenue, Des Plaines, IL 60018-3315, USA BARC-SA and Lancet Laboratories, Johannesburg, South Africa
A R T I C L E I N F O
A B S T R A C T
Keywords: Human immunodeficiency virus Dried blood spot Viral load testing
Background: HIV RNA suppression is a key indicator for monitoring success of antiretroviral therapy. From a logistical perspective, viral load (VL) testing using Dried Blood Spots (DBS) is a promising alternative to plasma based VL testing in resource-limited settings. Objectives: To evaluate the analytical and clinical performance of the Abbott RealTime HIV-1 assay using a fully automated one-spot DBS sample protocol. Study design: Limit of detection (LOD), linearity, lower limit of quantitation (LLQ), upper limit of quantitation (ULQ), and precision were determined using serial dilutions of HIV-1 Virology Quality Assurance stock (VQA Rush University), or HIV-1-containing armored RNA, made in venous blood. To evaluate correlation, bias, and agreement, 497 HIV-1 positive adult clinical samples were collected from Ivory Coast, Uganda and South Africa. For each HIV-1 participant, DBS-fingerprick, DBS-venous and plasma sample results were compared. Correlation and bias values were obtained. The sensitivity and specificity were analyzed at a threshold of 1000 HIV-1 copies/ mL generated using the standard plasma protocol. Results: The Abbott HIV-1 DBS protocol had an LOD of 839 copies/mL, a linear range from 500 to 1 × 107 copies/mL, an LLQ of 839 copies/mL, a ULQ of 1 × 107 copies/mL, and an inter-assay SD of ≤0.30 log copies/ mL for all tested levels within this range. With clinical samples, the correlation coefficient (r value) was 0.896 between DBS-fingerprick and plasma and 0.901 between DBS-venous and plasma, and the bias was −0.07 log copies/mL between DBS-fingerprick and plasma and −0.02 log copies/mL between DBS-venous and plasma. The sensitivity of DBS-fingerprick and DBS-venous was 93%, while the specificity of both DBS methods was 95%. Conclusion: The results demonstrated that the Abbott RealTime HIV-1 assay with DBS sample protocol is highly sensitive, specific and precise across a wide dynamic range and correlates well with plasma values. The Abbott RealTime HIV-1 assay with DBS sample protocol provides an alternative sample collection and transfer option in resource-limited settings and expands the utility of a viral load test to monitor HIV-1 ART treatment for infected patients.
1. Background The 2013 WHO Consolidated Care and Treatment Guidelines for HIV indicate that viral load (VL) testing is the preferred option for monitoring treatment outcome in HIV infected individuals on antiretroviral therapy (ART) [1]. Furthermore, the UNAIDS 90-90-90 initiative aims to achieve the diagnosis of 90% of individuals infected with HIV; to get 90% of the HIV infected individuals on ART and ensure that 90% of these infected individuals on ART achieve viral loads (VL) suppression [2]. In order to achieve the third 90-90-90 objective
countries will need to roll-out country wide VL testing. There are barriers to wide-scale roll-out of VL testing in resource limited settings, such as, cost; logistics of getting plasma samples into central testing facilities and scarce technical skills. Using dried blood spots (DBS) instead of plasma as a sample matrix simplifies the VL logistic barrier, by simplifying sample collection and transport to a central laboratory [3–7]. The simplification of the logistic steps by using DBS reduces the costs by removing the cold chain as DBS are stable for several weeks prior to VL testing [3]. By contrast with the transport and cost advantages, DBS have two
Abbreviations: HIV, human immunodeficiency virus; DBS, dried blood spot; VL, viral load; LOD, limit of detection ⁎ Corresponding author. E-mail address: [email protected] (N. Tang). http://dx.doi.org/10.1016/j.jcv.2017.05.002 Received 17 January 2017; Received in revised form 17 April 2017; Accepted 3 May 2017 1386-6532/ © 2017 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
samples prepared from Ivory Coast and Uganda were stored at −20 °C or colder and shipped to testing lab in dry ice. After samples were received by the testing labs (Lancet and Abbott Molecular), both plasma and DBS samples were stored at −20 °C or colder. All DBS samples used in these studies were prepared using Munktell TFN paper (LabMate, South Africa) with 12 mm perforated circles. Additional studies were performed (data not shown) to demonstrate equivalent performance between DBS made using different papers including Munktell TFN, Whatman 903, and Ahlström 226.
challenges relative to plasma for HIV-1 VL testing. The first challenge is the need for high sensitivity using only 50–100ul of blood [4]. Some of the DBS protocols achieve the 1000 copies/mL VL threshold recommended by WHO [4,8]. The second challenge is potentially lower specificity when using the 1000 copies/mL VL threshold caused by proviral HIV-1 DNA found in whole blood but not in plasma [4]. To counteract this, DBS protocols must be designed to purify RNA selectively during sample preparation, or, to amplify only RNA targets during amplification. The Abbott RealTime HIV-1 assay sample preparation (iron oxide microparticle binding combined with phosphate buffer elution) was optimized to effectively capture RNA. This resulted in DBS assay specificity of ≥90% when using the 1000 copies/mL threshold [4]. The fully automated Abbott RealTime HIV-1 assay was developed using plasma as its sample type [9]. Subsequently, a research use only two-spot semi-automated DBS protocol was developed for this assay and was used in a number of studies [6,10–14]. A meta-analysis of over 1500 published data was summarized in 2014 where the assays with sensitivity and specificity above 90% were shown. [8]. To increase ease-of-use and user access, one-spot DBS protocol for the Abbott RealTime HIV-1 assay was developed and CE marked 2016. This manuscript describes the studies performed to evaluate the fully automated one-spot DBS protocol developed for use with the Abbott RealTime HIV-1 assay.
3.2. Abbott RealTime HIV-1 DBS and plasma protocols Perforated DBS were pushed out of the cards and submerged in 1.3 mL of DBS buffer in Abbott master mix tube. The tubes were mixed by swirling and incubated for 30 min in a heating block set at 55 °C. Following heating, the tubes were gently mixed again before being processed using an m2000 system. The Abbott HIV-1 RealTime assay uses automated extraction of viral nucleic acid on the m2000sp using purification reagents and purification process specific for RNA [16]. Plasma sample processing and testing was performed following the instructions in the package insert for the Abbott RealTime HIV-1 RNA assay with 0.6 mL protocol. Quantitative assay results were reported as copies/mL or log copies/mL [9].
2. Objectives 3.3. Analytical performance evaluation The objectives of this study were to evaluate the analytical (LOD, linearity, and precision) and clinical (correlation, bias, sensitivity and specificity) performance of the new one-spot DBS protocol when tested using the Abbott RealTime HIV-1 assay.
All analytical testing was performed at Abbott Molecular. The LOD of the DBS protocol was determined by testing five panels (125, 250, 500, 1000, and 3000 copies/mL). Each of the five panel members was tested seven times in four separate assay runs, resulting in 28 results per panel member. Each assay run used a different reagent lot. The LOD value was determined following the guidance in NCCLS standard EP17-A2 [17]. The linearity, lower limit of quantitation (LLQ), and upper limit of quantitation (ULQ) of the DBS protocol were determined by testing eight panels (500, 1000, 5000, 10,000, 100,000, 300,000, 1 × 106, and 1 × 107 copies/mL). Each of the panel members was tested three times in four separate assay runs, resulting in 12 results per panel member. One reagent lot was used for the testing. Linearity was determined per NCCLS guideline EP6A [18]. The lower limit of quantitation (LLQ) was defined as the lowest analyte concentration with a difference of < 0.5 log copies/mL as compared to the target concentration. The upper limit of quantitation (ULQ) was defined as the highest analyte concentration with a difference of < 0.5 log copies/mL as compared to the target concentration. The precision of the DBS protocol was determined by testing six panels (500, 1000, 5000, 10,000, 100,000, and 5 × 106 copies/mL). Each of the panel members was tested five times in 15 separate assay runs, resulting in 75 results per panel member. Three reagent lots were used for the testing. Within run, between run, and inter-assay (withinrun and between-run) standard deviations (SD) were determined following the guidance in NCCLS standard EP5-A3 [19].
3. Study design 3.1. Samples/DBS paper type Limit of detection (LOD) and precision were performed using samples prepared by diluting the Viral Quality Assurance (VQA) panel in seronegative EDTA whole blood. Linearity was performed using samples prepared by diluting an HIV1 armored RNA (Asuragen, Austin Texas) in seronegative EDTA whole blood. One hundred and twenty HIV seronegative whole blood donor samples were collected by ProMedx and used for assay analytical specificity evaluation. Clinical samples were obtained from 500 adults (≥18 years old) who visited HIV clinics in Ivory Coast, Uganda and South Africa. For each adult, DBS-venous, DBS-fingerprick and plasma were collected. Whole blood collected in EDTA tubes was used to prepare the DBS-venous samples (70 μL of blood per spot) and the plasma samples. The DBS-fingerprick samples were collected by either, 1) dropping the fingerprick blood into a small EDTA tube and manually pipetting 70 μL onto each spot [10] or, 2) by directly spotting the blood from the fingerprick onto the DBS paper card following a standard procedure for processing Dried Blood Spot [15]. Of the 500 collected samples, 3 samples from Ivory Coast were excluded because their plasma tubes were leaking leading to the exclusion of all samples for these 3 participants. For the remaining 497 participants, 244 were from participants under treatment at the time of collection, 227 were from participants that were treatmentnaïve, 21 were from participants that had previously received treatment but not receiving treatment at the time of collection, and 5 had an unknown treatment status. Blood samples stored at room temperature were used to prepare plasma and DBS-venous samples within 24 h of blood collection. Prepared plasma samples were stored at −20 °C or colder and shipped in dry ice from each site to the testing lab. The DBS samples prepared in South Africa were stored and shipped at room temperature. The DBS
3.4. Clinical performance evaluation Two hundred forty seven participant samples collected from Uganda and Ivory Coast were tested at Abbott Molecular (three of 250 samples were not tested because the plasma tubes had leaked during shipment) while the remaining 250 participant samples that were collected from South Africa were tested at Lancet Laboratories. For each participant, three sample types were tested: DBS-fingerprick, DBS-venous, and plasma. For the least squares correlation analysis between different sample types (DBS fingerprick vs. plasma, DBS venous vs. plasma, and DBS 57
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
fingerprick vs. DBS venous) each patient needed to have a quantitative value within the sample type linear range to be included in the analysis. A Bland-Altman bias analysis was performed using the same criterion. The rate of misclassification was calculated for DBS specimen results using the plasma result for each participant as the reference method. The rate of upward or downward misclassification was defined as the percentage of results that were incorrectly identified as greater or less than 1000 copies/mL, respectively, as compared to the plasma standard. Sensitivity was calculated as the percentage of virological failures as determined by DBS results in the total true failure cases as determined by plasma reference results. Specificity was calculated as the percentage of successful treatment cases as determined by DBS results in the total successful treatment cases as determined by plasma reference results. The threshold for defining successful treatment versus treatment failure was recommended by WHO to be 1000 copies/ml [8].
Fig 1. The assay was shown to be linear across the range tested (n = 92, r = 0.995, slope = 1.08, and intercept = −0.32) (500 to 1 × 107 copies/mL) made from HIV-1 Armored RNA. The claimed assay linear range, however, is from LOD (839 copies/mL) to 1 × 107 copies/mL.
4. Results
Table 2 Abbott RealTime HIV-1 DBS precision.
4.1. Analytical performance results The LOD of the DBS protocol was determined to be 839 copies/mL 95% confidence interval [CI] of 624 −1387 copies/mL (Table 1). The linearity of the DBS protocol extended from 500 copies/mL to 1 × 107 copies/mL (n = 92, r = 0.995, slope = 1.08, and intercept = −0.32) (Fig. 1). The LLQ is the same as the LOD of 839 copies/mL. The ULQ was 1 × 107 copies/mL. The inter-assay precision of the DBS protocol was determined to be ≤ 0.3 for all tested levels and ≤0.2 log copies/ mL for the tested levels that were above LOD (Table 2). 120 HIV-1 seronegative DBS-venous samples were tested for analytical specificity evaluation. None were positive, resulting in 100% analytical specificity (data not shown).
Panel Member
N
Mean Conc. (RNA copies/ mL)
Mean Conc. (log RNA copies/ mL)
Within-Run SD Component
BetweenRun SD Component
Inter-assaya SD (log RNAcopies/ mL)
1 2 3 4 5 6
54b 70b 74c 73d 75 75
417 692 4531 9034 108643 8130801
2.62 2.84 3.66 3.96 5.04 6.91
0.29 0.26 0.12 0.11 0.05 0.05
0.00 0.00 0.09 0.07 0.04 0.04
0.29 0.26 0.16 0.13 0.06 0.06
a
Inter-assay contains within-run and between-run components. Concentration means of Panel Members 1 and 2 are below the assay LOD (839 copies/mL). Precision estimates reflect only results that are detected and quantitated. c One replicate was invalid and was excluded from the data analysis. d Two replicates were invalid and were excluded from the data analysis.
4.2. Clinical performance results
b
Results were not generated for three of the 500 samples because the plasma sample tubes were damaged during transport. The following samples produced quantitative results within the assay linear range (between LLQ and ULQ): 65.2% (324/497) for plasma, 51.5% (256/ 497) for DBS-fingerprick, and 51.3% (255/497) for DBS-venous (Table 3). This difference in frequency of quantitation results reflects the greater sensitivity of the plasma protocol with an LOD/LLQ of 40 copies/mL as compared to the LOD/LLQ of 839 copies/mL for DBS. The least squares correlation was 0.896 between DBS-fingerprick and plasma, 0.901 between DBS-venous and plasma, and 0.951 between DBS-fingerprick and DBS-venous (Fig. 2A–C). Bland-Altman analyses of the same data sets show a bias −0.07 log copies/mL between DBS-fingerprick and plasma, −0.02 log copies/mL between DBS-venous and plasma, and −0.06 log copies/mL between DBSfingerprick and DBS-venous (Fig. 3A–C). Also shown in Fig. 3A–C are the individual sample biases. For DBSfingerprick vs. plasma, 83.79% of the results were within 0.5 log copies/mL, 15.02% were within one log copies/mL, and 1.19% of the
Table 3 Clinical sample distribution (total 497). Sample type
Results
Count
Percent (%)
DBS-fingerprick
Not detected Detected < LODa Detected (LOD −ULQ) Total Not detected Detected < LOD* Detected (LOD −ULQ) Total Not detected Detected < LODb Detected (LOD −ULQ) Total
170 71 256 497 185 57 255 497 137 36 324 497
34.2 14.3 51.5 100.0 37.2 11.5 51.3 100.0 27.6 7.2 65.2 100.0
DBS-venous
Plasma
a
DBS LOD: 839 copies/mL. b Plasma mL LOD: 40 copies/mL.
Table 1 Abbott RealTime HIV-1 DBS limit of detection (LOD) evaluation. Target concentration (RNA copies/mL)
Number tested
Number detected
Percent detected (%)
3000 1000 500 250 125
27* 28 28 28 28
27 27 24 10 4
100 96 86 36 14
results were greater than one log copies/mL different. For DBS-venous vs. plasma, 87.30% of the results were within 0.5 log copies/mL, 11.86% were within one log copies/mL, and 0.79% of the results were greater than one log copies/mL different. For DBS-fingerprick vs. DBSvenous, 95.04% of the results were within 0.5 log copies/mL, and 4.96% were within one log copies/mL. One sample had a large quantitation difference (DBS venous < LOD/ LLQ, DBS-fingerprick 5.62 log copies/mL, and plasma 5.78 log copies/ mL). Repeat testing for all 3 sample types produced similar results. Sample mislabeling may explain the observed difference. This sample was still included in the correlation, bias and sensitivity/specificity evaluation.
*One replicate was invalid and was excluded from the analysis. Probit analysis of the data determined that the concentration of HIV-1 RNA detected with 95% probability was 839 copies/mL (95% CI 624 to1387copies/mL).
58
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
Fig. 2. A total 497 clinical samples was tested. Sample pairs (DBS-fingerprick vs. plasma, DBS-venous vs. plasma, DBS-fingerprick vs. DBS-venous) that fall within the range of LOD (839 copies/mL) to 1 × 107 copies/mL were plotted. The correlation coefficient for DBSfinger versus plasma (n = 253) was 0.896 (Fig. 2A), for DBS-venous versus plasma (n = 252) was 0.901 (Fig. 2B), for DBS-finger versus DBS-venous (n = 242) was 0.951 (Fig. 2C).
Fig. 3. Bland-Altman plots showing mean biases of −0.07, −0.02 and −0.06 for DBSfingerprick vs. plasma (a); DBS-venous vs. plasma (b) and DBS-fingerprick vs. DBS venous (c) respectively. The mean bias plus minus 2 standard deviations were shown in the plots. The percentage of individual discrepant samples between HIV-1 VL quantitated from plasma, from DBS-fingerprick and DBS venous were also summarized in the table under each plot.
59
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
Table 4 Classification with 1000 RNA copies/mL threshold: DBS-fingerprick vs. plasma.
Table 7 Agreement between DBS-venous vs. plasma for 244 currently under treatment participants.
Reference testing using plasma Reference testing using plasma
VL ≥ 1000 copies/ml VL < 1000 copies/ml
DBS-finger prick
Sensitivity Specificity Overall Percent Agreement
VL ≥ 1000 copies/ml 237
VL < 1000 copies/ml 11
248
19
230
249
256 Ratio 93% 95% 94%
241 Score 237/256 230/241 467/497
497 95% CI 88.70–95.20% 92.01−97.43% 91.51–95.74%
DBS-venous
Sensitivity Specificity Overall Percent Agreement
Reference testing using plasma
DBS-venous
VL ≥ 1000 copies/ml VL < 1000 copies/ml
Sensitivity Specificity Overall Percent Agreement
VL < 1000 copies/ml 13
252
17
228
245
256 Ratio 93% 95% 94%
241 Score 239/256 228/241 467/497
497 95% CI 89.62–95.81% 90.99−96.82% 91.51–95.74%
Using the 1000 copies/mL threshold as determined by the plasma protocol, the sensitivity and specificity were 93% and 95% respectively, for both DBS-fingerprick and DBS-venous protocols (Tables 4 and 5). When analyzing only patients under antiretroviral therapy, the sensitivity and specificity of the DBS-fingerprick protocol were 93% and 95% respectively, versus 95% and 95% for the DBS-venous protocol (Tables 6 and 7). 5. Discussion A new DBS protocol has been developed for routine use with the Abbott RealTime HIV-1 assay. This protocol uses one DBS spot per patient, uses heated elution (55 °C for 30 min) to improve assay sensitivity (data not shown), and is fully automated. The new protocol Table 6 Agreement between DBS-fingerprick vs. plasma for 244 currently under treatment participants. Reference testing using plasma
DBS-finger prick
VL ≥ 1000 copies/ml VL < 1000 copies/ml
Sensitivity Specificity Overall Percent Agreement
VL ≥ 1000 copies/ml 70
VL < 1,000 copies/ml 8
78
5
161
166
75 Ratio 93% 95% 95%
169 Score 70/75 161/169 231/244
244 95% CI 85.32–97.12% 90.94–97.58% 91.10–96.86%
VL < 1000 copies/ml 8
79
4
161
165
75 Ratio 95% 95% 95%
169 Score 71/75 161/169 232/244
244 95% CI 87.07−97.91% 90.94–97.58% 91.60–97.16%
replaces an earlier research version that used two DBS per patient, room temperature elution for 20 min, and was not fully automated. Analytical testing demonstrated that the new DBS protocol has an LOD of 839 copies/mL. This meets the 1000 copies/mL threshold required by WHO for HIV-1 VL testing for virological failure determination [4,8]. The LOD was established using VQA panels diluted into HIV seronegative blood and therefore proviral DNA could not falsely contribute to the estimate of assay sensitivity. Other analytical characteristics of the new DBS protocol are compatible with routine HIV-1 VL testing. This includes a linearity ranging from 500 copies/mL to 1 × 107 copies/mL, an LLQ of 839 copies/mL (which is the same as the DBS LOD), a ULQ of 1 × 107 copies/mL, and an inter-assay precision of ≤0.3 log copies/mL for all tested panels. In clinical sample testing, the correlation values (0.896 for DBSfingerprick vs. plasma and 0.902 for DBS-venous vs. plasma) and BlandAltman bias values (-0.07 log copies/mL for DBS-fingerprick vs. plasma and −0.02 log copies/mL for DBS venous vs. plasma) using the DBS protocol correlated well with their counterparts generated using the standard plasma protocol. Using 1000 copies/mL as a threshold for virological failure, the DBS protocol had a sensitivity of 93% and a specificity of 95% for both DBSfingerprick and DBS-venous evaluated in all tested population (n = 497). Similar sensitivity and specificity values were obtained from the subset of samples collected from participants under antiretroviral treatment (n = 244). The high specificity confirms that RNA is selectively purified from the samples during Abbott RealTime HIV-1 assay DBS sample preparation [16]. Results from additional studies also indicated that the genomic DNA recovery rate using the Abbott RealTime HIV-1 nucleic acid extraction chemistry and protocol is approximately 3% (data not shown), compared with approximately 60% or more in RNA recovery. Therefore, any potential contribution of genomic DNA to final HIV-1 viral load results is minimal, as supported by the high specificity and correlation between DBS and plasma HIV-1 viral load results. Most DBS versus plasma VL studies are performed using DBS from EDTA-blood collected by venipuncture (venous blood) [4,8]. However, DBS are most likely to be prepared by fingerprick (adults) or heel-stick (infants) in remote settings [4]. In this study, a side by side evaluation of DBS-fingerprick and DBS-venous versus the plasma standard was performed to generate the DBS-venous and DBS-fingerprick comparative data. The results (correlation of 0.951 and bias of −0.06 log copies/mL) demonstrate that both the DBS-fingerprick and DBS-venous protocols have very similar performance, when compared to standard plasma protocol.
Table 5 Classification with 1000 RNA copies/mL threshold: DBS-venous vs. plasma.
VL ≥ 1000 copies/ml 239
VL ≥ 1000 copies/ml VL < 1000 copies/ml
VL ≥ 1000 copies/ml 71
6. Conclusion The new one-spot Abbott RealTime HIV-1 DBS protocol has an 60
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
[4] N.T. Parkin, Measurement of HIV-1 viral load for drug resistance surveillance using dried blood spots: literature review and modeling of contribution of DNA and RNA, AIDS Rev. 16 (2014) 160–171. [5] A. Johannessen, Dried blood spots in HIV monitoring: applications in resourcelimited settings, Bioanalysis 2 (2010) 1893–1908. [6] U. Neogi, S. Gupta, R. Rodridges, P. Sahoo, S. Rao, B. Rewari, S. Shastri, A. Costa, A. Shet, Dried blood spot HIV-1 RNA quantification: a useful tool for viral load monitoring among HIV-infected individuals in India, Indian J. Med. Res. 136 (December (6)) (2012) 956–962. [7] R. Hamers, P. Smit, W. Stevens, R. Schuurman, T. Rinke de Wit, Dried fluid spots for HIV type-1 viral load and resistance genotyping: a systematic review, Antivir Ther. 14 (5) (2009) 619–629. [8] Interim Technical Update. Technical and Operational Considerations for Implementing HIV Viral Load Testing, World Health Organization, 2014, pp. 1–26. [9] N. Tang, S. Huang, J. Salituro, W.-B. Mak, G. Cloherty, J. Johanson, Y.H. Li, G. Schneider, J. Robinson, J. Hackett Jr., P. Swanson, K. Abravaya, Abbott RealTime HIV-1 viral load assay for automated quantitation of HIV-1 RNA in genetically diverse group M subtypes A-H, group O and group N samples, J. Virol. Methods 146 (2007) 236–245. [10] A. Sousa, J. Houston, M. Assane, J. Chang, E. Koumans, I. Jani, J. Sabatier, P. Vaz, C. Yang, E. Rivadeneira, Performance of HIV Viral Load with Dried Blood Spots in Children on ART in Mozambique, CROI, 2015. [11] C. Garrido, N. Zahonero, A. Corral, M. Arredondo, V. Soriano, C. de Mendoza C, Correlation between human immunodeficiency virus type 1 (HIV-1) RNA measurements obtained with dried blood spots and those obtained with plasma by use of Nuclisens EasyQ HIV-1 and Abbott RealTime HIV load tests, J. Clin. Microbiol. 47 (2009) 1031–1036. [12] M. Arredondo, C. Garrido, N. Parkin, N. Zahonero N, S. Bertagnolio, V. Soriano, C. de Mendoza, Comparison of HIV-1 RNA measurements obtained by using plasma and dried blood spots in the automated Abbott real-time viral load assay, J. Clin. Microbiol. 50 (2012) 569–572. [13] S. Mtapuri-Zinyowera, F. Taziwa, C. Metcalf, E. Mbofana, S. De Weerdt, L. Flevaud, S. Simons, J.-F. Saint-Saveur, E. Fajardo, Field Evaluation of Performance of Dried Blood Spots (DBS) from Fingerprick for the Determination of Performance of Viral Load in a Resource-constrained Setting in Urban and Rural Zimbabwe, Presented at The 7th International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention, Kuala Lumpur, Malaysia, June 30–July 3, 2013, 2017 Abstract WEPE610. [14] S. Rutstein, D. Kamwendo, L. Lugali, I. Thengolose, G. Tegha, S.A. Fiscus, J.A. Nelson, M.C. Hosseinipour, A. Sarr, S. Gupta, F. Chimbwandira, R. Mwenda, R. Mataya, Measures of viral load using Abbott RealTime HIV-1 Assay on venous and fingerstick dried blood spots from provider-collected specimens in Malawian District Hospitals, J. Clin. Virol. 60 (2014) 392–398. [15] Processing of Dried Blood Spots Standard Operating Procedure, ACTG/IMPAACT Lab Tech committee, 2012. [16] S. Fernandes, K. Morosyuk, M. Ramanathan, L. Rainen, Evaluation of effect of specimen-handling parameters for plasma preparation tubes on viral load measurements obtained by using the abbott RealTime HIV-1 load assay, J. Clin. Microbiol. 48 (2010) 2464–2468. [17] EP17-A2, Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures. Approved Guideline, by Clinical Laboratory Standards Institute, second edition, June, (2012). [18] EP6-A, Evaluation of the Linearity of Quantitative Measurement Procedures: A Statistical Approach; Approved Guideline, by Clinical Laboratory Standards Institute, April, (2003). [19] Evaluation of Precision of Quantitative Measurement Procedures: Approved Guideline −Third edition, by Clinical Laboratory Standards Institute, October, (2014).
analytical sensitivity below 1000 copies/mL, tested linear range from 500 copies/mL to 107 copies/mL (claimed linear range from an LLQ of 839 copies/mL to a ULQ of 107 copies/mL), and ≤0.3 log copies/mL inter-assay SD. The clinical performance evaluation showed high sensitivity of 93% and specificity of 95% for both DBS-fingerprick and DBS-venous when compared to the plasma assay results at 1000 copies/mL threshold for virologocal failure. The data supports the use of the Abbott RealTime HIV-1 assay with the DBS protocol as a convenient sample collection/transport method for HIV-1 viral load monitoring in resource limited settings and thus aiding the 90:90:90 initiatives Funding This work was funded by Abbott Molecular Inc. Ethical approval All samples were collected under ethical approval. The protocols were approved by country/local ethical committees. Conflict of interest The following authors: Ning Tang, Vihanga Pahalawatta, Andrea Frank, John Lampinen, Gregor Leckie, Shihai Huang, and Klara Abravaya are employees of Abbott Molecular (Ning Tang, Gregor Leckie, Shihai Huang, and Klara Abravaya are also shareholders of Abbott Laboratories). Acknowledgements We would like to thank all study participants in South Africa and blood donors in Uganda and Ivory Coast; the submitting doctors (Dr Osman Ebrahim; Dr David Johnson; Dr Lydia K Masemola); the staff at BARC-SA and Lancet, the staff at Abbott Molecular. References [1] World Health Organization, Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV Infection: Recommendations for a Public Health Approach, World Health Organization, 2013, pp. 1–269. [2] 90-90-90 An Ambitious Treatment Target to Help End the AIDS Epidemic, (2017) http://www.unaids.org/sites/default/files/media_asset/90-90-90_en_0.pdf. [3] G. Wu, M. Zaman, Low-cost Tools for Diagnosing and Monitoring HIV Infection in Low-resource Settings 90 Bulletin of the World Health Organization, 2012, pp. 914–920, http://dx.doi.org/10.2471/BLT.12.102780.
61
Contents lists available at ScienceDirect
Journal of Clinical Virology journal homepage: www.elsevier.com/locate/jcv
HIV-1 viral load measurement in venous blood and fingerprick blood using Abbott RealTime HIV-1 DBS assay
MARK
⁎
Ning Tanga, , Vihanga Pahalawattaa, Andrea Franka, Zowie Bagleyb, Raquel Vianab, John Lampinena, Gregor Leckiea, Shihai Huanga, Klara Abravayaa, Carole L. Wallisb a b
Abbott Molecular, 1300 East Touhy Avenue, Des Plaines, IL 60018-3315, USA BARC-SA and Lancet Laboratories, Johannesburg, South Africa
A R T I C L E I N F O
A B S T R A C T
Keywords: Human immunodeficiency virus Dried blood spot Viral load testing
Background: HIV RNA suppression is a key indicator for monitoring success of antiretroviral therapy. From a logistical perspective, viral load (VL) testing using Dried Blood Spots (DBS) is a promising alternative to plasma based VL testing in resource-limited settings. Objectives: To evaluate the analytical and clinical performance of the Abbott RealTime HIV-1 assay using a fully automated one-spot DBS sample protocol. Study design: Limit of detection (LOD), linearity, lower limit of quantitation (LLQ), upper limit of quantitation (ULQ), and precision were determined using serial dilutions of HIV-1 Virology Quality Assurance stock (VQA Rush University), or HIV-1-containing armored RNA, made in venous blood. To evaluate correlation, bias, and agreement, 497 HIV-1 positive adult clinical samples were collected from Ivory Coast, Uganda and South Africa. For each HIV-1 participant, DBS-fingerprick, DBS-venous and plasma sample results were compared. Correlation and bias values were obtained. The sensitivity and specificity were analyzed at a threshold of 1000 HIV-1 copies/ mL generated using the standard plasma protocol. Results: The Abbott HIV-1 DBS protocol had an LOD of 839 copies/mL, a linear range from 500 to 1 × 107 copies/mL, an LLQ of 839 copies/mL, a ULQ of 1 × 107 copies/mL, and an inter-assay SD of ≤0.30 log copies/ mL for all tested levels within this range. With clinical samples, the correlation coefficient (r value) was 0.896 between DBS-fingerprick and plasma and 0.901 between DBS-venous and plasma, and the bias was −0.07 log copies/mL between DBS-fingerprick and plasma and −0.02 log copies/mL between DBS-venous and plasma. The sensitivity of DBS-fingerprick and DBS-venous was 93%, while the specificity of both DBS methods was 95%. Conclusion: The results demonstrated that the Abbott RealTime HIV-1 assay with DBS sample protocol is highly sensitive, specific and precise across a wide dynamic range and correlates well with plasma values. The Abbott RealTime HIV-1 assay with DBS sample protocol provides an alternative sample collection and transfer option in resource-limited settings and expands the utility of a viral load test to monitor HIV-1 ART treatment for infected patients.
1. Background The 2013 WHO Consolidated Care and Treatment Guidelines for HIV indicate that viral load (VL) testing is the preferred option for monitoring treatment outcome in HIV infected individuals on antiretroviral therapy (ART) [1]. Furthermore, the UNAIDS 90-90-90 initiative aims to achieve the diagnosis of 90% of individuals infected with HIV; to get 90% of the HIV infected individuals on ART and ensure that 90% of these infected individuals on ART achieve viral loads (VL) suppression [2]. In order to achieve the third 90-90-90 objective
countries will need to roll-out country wide VL testing. There are barriers to wide-scale roll-out of VL testing in resource limited settings, such as, cost; logistics of getting plasma samples into central testing facilities and scarce technical skills. Using dried blood spots (DBS) instead of plasma as a sample matrix simplifies the VL logistic barrier, by simplifying sample collection and transport to a central laboratory [3–7]. The simplification of the logistic steps by using DBS reduces the costs by removing the cold chain as DBS are stable for several weeks prior to VL testing [3]. By contrast with the transport and cost advantages, DBS have two
Abbreviations: HIV, human immunodeficiency virus; DBS, dried blood spot; VL, viral load; LOD, limit of detection ⁎ Corresponding author. E-mail address: [email protected] (N. Tang). http://dx.doi.org/10.1016/j.jcv.2017.05.002 Received 17 January 2017; Received in revised form 17 April 2017; Accepted 3 May 2017 1386-6532/ © 2017 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
samples prepared from Ivory Coast and Uganda were stored at −20 °C or colder and shipped to testing lab in dry ice. After samples were received by the testing labs (Lancet and Abbott Molecular), both plasma and DBS samples were stored at −20 °C or colder. All DBS samples used in these studies were prepared using Munktell TFN paper (LabMate, South Africa) with 12 mm perforated circles. Additional studies were performed (data not shown) to demonstrate equivalent performance between DBS made using different papers including Munktell TFN, Whatman 903, and Ahlström 226.
challenges relative to plasma for HIV-1 VL testing. The first challenge is the need for high sensitivity using only 50–100ul of blood [4]. Some of the DBS protocols achieve the 1000 copies/mL VL threshold recommended by WHO [4,8]. The second challenge is potentially lower specificity when using the 1000 copies/mL VL threshold caused by proviral HIV-1 DNA found in whole blood but not in plasma [4]. To counteract this, DBS protocols must be designed to purify RNA selectively during sample preparation, or, to amplify only RNA targets during amplification. The Abbott RealTime HIV-1 assay sample preparation (iron oxide microparticle binding combined with phosphate buffer elution) was optimized to effectively capture RNA. This resulted in DBS assay specificity of ≥90% when using the 1000 copies/mL threshold [4]. The fully automated Abbott RealTime HIV-1 assay was developed using plasma as its sample type [9]. Subsequently, a research use only two-spot semi-automated DBS protocol was developed for this assay and was used in a number of studies [6,10–14]. A meta-analysis of over 1500 published data was summarized in 2014 where the assays with sensitivity and specificity above 90% were shown. [8]. To increase ease-of-use and user access, one-spot DBS protocol for the Abbott RealTime HIV-1 assay was developed and CE marked 2016. This manuscript describes the studies performed to evaluate the fully automated one-spot DBS protocol developed for use with the Abbott RealTime HIV-1 assay.
3.2. Abbott RealTime HIV-1 DBS and plasma protocols Perforated DBS were pushed out of the cards and submerged in 1.3 mL of DBS buffer in Abbott master mix tube. The tubes were mixed by swirling and incubated for 30 min in a heating block set at 55 °C. Following heating, the tubes were gently mixed again before being processed using an m2000 system. The Abbott HIV-1 RealTime assay uses automated extraction of viral nucleic acid on the m2000sp using purification reagents and purification process specific for RNA [16]. Plasma sample processing and testing was performed following the instructions in the package insert for the Abbott RealTime HIV-1 RNA assay with 0.6 mL protocol. Quantitative assay results were reported as copies/mL or log copies/mL [9].
2. Objectives 3.3. Analytical performance evaluation The objectives of this study were to evaluate the analytical (LOD, linearity, and precision) and clinical (correlation, bias, sensitivity and specificity) performance of the new one-spot DBS protocol when tested using the Abbott RealTime HIV-1 assay.
All analytical testing was performed at Abbott Molecular. The LOD of the DBS protocol was determined by testing five panels (125, 250, 500, 1000, and 3000 copies/mL). Each of the five panel members was tested seven times in four separate assay runs, resulting in 28 results per panel member. Each assay run used a different reagent lot. The LOD value was determined following the guidance in NCCLS standard EP17-A2 [17]. The linearity, lower limit of quantitation (LLQ), and upper limit of quantitation (ULQ) of the DBS protocol were determined by testing eight panels (500, 1000, 5000, 10,000, 100,000, 300,000, 1 × 106, and 1 × 107 copies/mL). Each of the panel members was tested three times in four separate assay runs, resulting in 12 results per panel member. One reagent lot was used for the testing. Linearity was determined per NCCLS guideline EP6A [18]. The lower limit of quantitation (LLQ) was defined as the lowest analyte concentration with a difference of < 0.5 log copies/mL as compared to the target concentration. The upper limit of quantitation (ULQ) was defined as the highest analyte concentration with a difference of < 0.5 log copies/mL as compared to the target concentration. The precision of the DBS protocol was determined by testing six panels (500, 1000, 5000, 10,000, 100,000, and 5 × 106 copies/mL). Each of the panel members was tested five times in 15 separate assay runs, resulting in 75 results per panel member. Three reagent lots were used for the testing. Within run, between run, and inter-assay (withinrun and between-run) standard deviations (SD) were determined following the guidance in NCCLS standard EP5-A3 [19].
3. Study design 3.1. Samples/DBS paper type Limit of detection (LOD) and precision were performed using samples prepared by diluting the Viral Quality Assurance (VQA) panel in seronegative EDTA whole blood. Linearity was performed using samples prepared by diluting an HIV1 armored RNA (Asuragen, Austin Texas) in seronegative EDTA whole blood. One hundred and twenty HIV seronegative whole blood donor samples were collected by ProMedx and used for assay analytical specificity evaluation. Clinical samples were obtained from 500 adults (≥18 years old) who visited HIV clinics in Ivory Coast, Uganda and South Africa. For each adult, DBS-venous, DBS-fingerprick and plasma were collected. Whole blood collected in EDTA tubes was used to prepare the DBS-venous samples (70 μL of blood per spot) and the plasma samples. The DBS-fingerprick samples were collected by either, 1) dropping the fingerprick blood into a small EDTA tube and manually pipetting 70 μL onto each spot [10] or, 2) by directly spotting the blood from the fingerprick onto the DBS paper card following a standard procedure for processing Dried Blood Spot [15]. Of the 500 collected samples, 3 samples from Ivory Coast were excluded because their plasma tubes were leaking leading to the exclusion of all samples for these 3 participants. For the remaining 497 participants, 244 were from participants under treatment at the time of collection, 227 were from participants that were treatmentnaïve, 21 were from participants that had previously received treatment but not receiving treatment at the time of collection, and 5 had an unknown treatment status. Blood samples stored at room temperature were used to prepare plasma and DBS-venous samples within 24 h of blood collection. Prepared plasma samples were stored at −20 °C or colder and shipped in dry ice from each site to the testing lab. The DBS samples prepared in South Africa were stored and shipped at room temperature. The DBS
3.4. Clinical performance evaluation Two hundred forty seven participant samples collected from Uganda and Ivory Coast were tested at Abbott Molecular (three of 250 samples were not tested because the plasma tubes had leaked during shipment) while the remaining 250 participant samples that were collected from South Africa were tested at Lancet Laboratories. For each participant, three sample types were tested: DBS-fingerprick, DBS-venous, and plasma. For the least squares correlation analysis between different sample types (DBS fingerprick vs. plasma, DBS venous vs. plasma, and DBS 57
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
fingerprick vs. DBS venous) each patient needed to have a quantitative value within the sample type linear range to be included in the analysis. A Bland-Altman bias analysis was performed using the same criterion. The rate of misclassification was calculated for DBS specimen results using the plasma result for each participant as the reference method. The rate of upward or downward misclassification was defined as the percentage of results that were incorrectly identified as greater or less than 1000 copies/mL, respectively, as compared to the plasma standard. Sensitivity was calculated as the percentage of virological failures as determined by DBS results in the total true failure cases as determined by plasma reference results. Specificity was calculated as the percentage of successful treatment cases as determined by DBS results in the total successful treatment cases as determined by plasma reference results. The threshold for defining successful treatment versus treatment failure was recommended by WHO to be 1000 copies/ml [8].
Fig 1. The assay was shown to be linear across the range tested (n = 92, r = 0.995, slope = 1.08, and intercept = −0.32) (500 to 1 × 107 copies/mL) made from HIV-1 Armored RNA. The claimed assay linear range, however, is from LOD (839 copies/mL) to 1 × 107 copies/mL.
4. Results
Table 2 Abbott RealTime HIV-1 DBS precision.
4.1. Analytical performance results The LOD of the DBS protocol was determined to be 839 copies/mL 95% confidence interval [CI] of 624 −1387 copies/mL (Table 1). The linearity of the DBS protocol extended from 500 copies/mL to 1 × 107 copies/mL (n = 92, r = 0.995, slope = 1.08, and intercept = −0.32) (Fig. 1). The LLQ is the same as the LOD of 839 copies/mL. The ULQ was 1 × 107 copies/mL. The inter-assay precision of the DBS protocol was determined to be ≤ 0.3 for all tested levels and ≤0.2 log copies/ mL for the tested levels that were above LOD (Table 2). 120 HIV-1 seronegative DBS-venous samples were tested for analytical specificity evaluation. None were positive, resulting in 100% analytical specificity (data not shown).
Panel Member
N
Mean Conc. (RNA copies/ mL)
Mean Conc. (log RNA copies/ mL)
Within-Run SD Component
BetweenRun SD Component
Inter-assaya SD (log RNAcopies/ mL)
1 2 3 4 5 6
54b 70b 74c 73d 75 75
417 692 4531 9034 108643 8130801
2.62 2.84 3.66 3.96 5.04 6.91
0.29 0.26 0.12 0.11 0.05 0.05
0.00 0.00 0.09 0.07 0.04 0.04
0.29 0.26 0.16 0.13 0.06 0.06
a
Inter-assay contains within-run and between-run components. Concentration means of Panel Members 1 and 2 are below the assay LOD (839 copies/mL). Precision estimates reflect only results that are detected and quantitated. c One replicate was invalid and was excluded from the data analysis. d Two replicates were invalid and were excluded from the data analysis.
4.2. Clinical performance results
b
Results were not generated for three of the 500 samples because the plasma sample tubes were damaged during transport. The following samples produced quantitative results within the assay linear range (between LLQ and ULQ): 65.2% (324/497) for plasma, 51.5% (256/ 497) for DBS-fingerprick, and 51.3% (255/497) for DBS-venous (Table 3). This difference in frequency of quantitation results reflects the greater sensitivity of the plasma protocol with an LOD/LLQ of 40 copies/mL as compared to the LOD/LLQ of 839 copies/mL for DBS. The least squares correlation was 0.896 between DBS-fingerprick and plasma, 0.901 between DBS-venous and plasma, and 0.951 between DBS-fingerprick and DBS-venous (Fig. 2A–C). Bland-Altman analyses of the same data sets show a bias −0.07 log copies/mL between DBS-fingerprick and plasma, −0.02 log copies/mL between DBS-venous and plasma, and −0.06 log copies/mL between DBSfingerprick and DBS-venous (Fig. 3A–C). Also shown in Fig. 3A–C are the individual sample biases. For DBSfingerprick vs. plasma, 83.79% of the results were within 0.5 log copies/mL, 15.02% were within one log copies/mL, and 1.19% of the
Table 3 Clinical sample distribution (total 497). Sample type
Results
Count
Percent (%)
DBS-fingerprick
Not detected Detected < LODa Detected (LOD −ULQ) Total Not detected Detected < LOD* Detected (LOD −ULQ) Total Not detected Detected < LODb Detected (LOD −ULQ) Total
170 71 256 497 185 57 255 497 137 36 324 497
34.2 14.3 51.5 100.0 37.2 11.5 51.3 100.0 27.6 7.2 65.2 100.0
DBS-venous
Plasma
a
DBS LOD: 839 copies/mL. b Plasma mL LOD: 40 copies/mL.
Table 1 Abbott RealTime HIV-1 DBS limit of detection (LOD) evaluation. Target concentration (RNA copies/mL)
Number tested
Number detected
Percent detected (%)
3000 1000 500 250 125
27* 28 28 28 28
27 27 24 10 4
100 96 86 36 14
results were greater than one log copies/mL different. For DBS-venous vs. plasma, 87.30% of the results were within 0.5 log copies/mL, 11.86% were within one log copies/mL, and 0.79% of the results were greater than one log copies/mL different. For DBS-fingerprick vs. DBSvenous, 95.04% of the results were within 0.5 log copies/mL, and 4.96% were within one log copies/mL. One sample had a large quantitation difference (DBS venous < LOD/ LLQ, DBS-fingerprick 5.62 log copies/mL, and plasma 5.78 log copies/ mL). Repeat testing for all 3 sample types produced similar results. Sample mislabeling may explain the observed difference. This sample was still included in the correlation, bias and sensitivity/specificity evaluation.
*One replicate was invalid and was excluded from the analysis. Probit analysis of the data determined that the concentration of HIV-1 RNA detected with 95% probability was 839 copies/mL (95% CI 624 to1387copies/mL).
58
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
Fig. 2. A total 497 clinical samples was tested. Sample pairs (DBS-fingerprick vs. plasma, DBS-venous vs. plasma, DBS-fingerprick vs. DBS-venous) that fall within the range of LOD (839 copies/mL) to 1 × 107 copies/mL were plotted. The correlation coefficient for DBSfinger versus plasma (n = 253) was 0.896 (Fig. 2A), for DBS-venous versus plasma (n = 252) was 0.901 (Fig. 2B), for DBS-finger versus DBS-venous (n = 242) was 0.951 (Fig. 2C).
Fig. 3. Bland-Altman plots showing mean biases of −0.07, −0.02 and −0.06 for DBSfingerprick vs. plasma (a); DBS-venous vs. plasma (b) and DBS-fingerprick vs. DBS venous (c) respectively. The mean bias plus minus 2 standard deviations were shown in the plots. The percentage of individual discrepant samples between HIV-1 VL quantitated from plasma, from DBS-fingerprick and DBS venous were also summarized in the table under each plot.
59
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
Table 4 Classification with 1000 RNA copies/mL threshold: DBS-fingerprick vs. plasma.
Table 7 Agreement between DBS-venous vs. plasma for 244 currently under treatment participants.
Reference testing using plasma Reference testing using plasma
VL ≥ 1000 copies/ml VL < 1000 copies/ml
DBS-finger prick
Sensitivity Specificity Overall Percent Agreement
VL ≥ 1000 copies/ml 237
VL < 1000 copies/ml 11
248
19
230
249
256 Ratio 93% 95% 94%
241 Score 237/256 230/241 467/497
497 95% CI 88.70–95.20% 92.01−97.43% 91.51–95.74%
DBS-venous
Sensitivity Specificity Overall Percent Agreement
Reference testing using plasma
DBS-venous
VL ≥ 1000 copies/ml VL < 1000 copies/ml
Sensitivity Specificity Overall Percent Agreement
VL < 1000 copies/ml 13
252
17
228
245
256 Ratio 93% 95% 94%
241 Score 239/256 228/241 467/497
497 95% CI 89.62–95.81% 90.99−96.82% 91.51–95.74%
Using the 1000 copies/mL threshold as determined by the plasma protocol, the sensitivity and specificity were 93% and 95% respectively, for both DBS-fingerprick and DBS-venous protocols (Tables 4 and 5). When analyzing only patients under antiretroviral therapy, the sensitivity and specificity of the DBS-fingerprick protocol were 93% and 95% respectively, versus 95% and 95% for the DBS-venous protocol (Tables 6 and 7). 5. Discussion A new DBS protocol has been developed for routine use with the Abbott RealTime HIV-1 assay. This protocol uses one DBS spot per patient, uses heated elution (55 °C for 30 min) to improve assay sensitivity (data not shown), and is fully automated. The new protocol Table 6 Agreement between DBS-fingerprick vs. plasma for 244 currently under treatment participants. Reference testing using plasma
DBS-finger prick
VL ≥ 1000 copies/ml VL < 1000 copies/ml
Sensitivity Specificity Overall Percent Agreement
VL ≥ 1000 copies/ml 70
VL < 1,000 copies/ml 8
78
5
161
166
75 Ratio 93% 95% 95%
169 Score 70/75 161/169 231/244
244 95% CI 85.32–97.12% 90.94–97.58% 91.10–96.86%
VL < 1000 copies/ml 8
79
4
161
165
75 Ratio 95% 95% 95%
169 Score 71/75 161/169 232/244
244 95% CI 87.07−97.91% 90.94–97.58% 91.60–97.16%
replaces an earlier research version that used two DBS per patient, room temperature elution for 20 min, and was not fully automated. Analytical testing demonstrated that the new DBS protocol has an LOD of 839 copies/mL. This meets the 1000 copies/mL threshold required by WHO for HIV-1 VL testing for virological failure determination [4,8]. The LOD was established using VQA panels diluted into HIV seronegative blood and therefore proviral DNA could not falsely contribute to the estimate of assay sensitivity. Other analytical characteristics of the new DBS protocol are compatible with routine HIV-1 VL testing. This includes a linearity ranging from 500 copies/mL to 1 × 107 copies/mL, an LLQ of 839 copies/mL (which is the same as the DBS LOD), a ULQ of 1 × 107 copies/mL, and an inter-assay precision of ≤0.3 log copies/mL for all tested panels. In clinical sample testing, the correlation values (0.896 for DBSfingerprick vs. plasma and 0.902 for DBS-venous vs. plasma) and BlandAltman bias values (-0.07 log copies/mL for DBS-fingerprick vs. plasma and −0.02 log copies/mL for DBS venous vs. plasma) using the DBS protocol correlated well with their counterparts generated using the standard plasma protocol. Using 1000 copies/mL as a threshold for virological failure, the DBS protocol had a sensitivity of 93% and a specificity of 95% for both DBSfingerprick and DBS-venous evaluated in all tested population (n = 497). Similar sensitivity and specificity values were obtained from the subset of samples collected from participants under antiretroviral treatment (n = 244). The high specificity confirms that RNA is selectively purified from the samples during Abbott RealTime HIV-1 assay DBS sample preparation [16]. Results from additional studies also indicated that the genomic DNA recovery rate using the Abbott RealTime HIV-1 nucleic acid extraction chemistry and protocol is approximately 3% (data not shown), compared with approximately 60% or more in RNA recovery. Therefore, any potential contribution of genomic DNA to final HIV-1 viral load results is minimal, as supported by the high specificity and correlation between DBS and plasma HIV-1 viral load results. Most DBS versus plasma VL studies are performed using DBS from EDTA-blood collected by venipuncture (venous blood) [4,8]. However, DBS are most likely to be prepared by fingerprick (adults) or heel-stick (infants) in remote settings [4]. In this study, a side by side evaluation of DBS-fingerprick and DBS-venous versus the plasma standard was performed to generate the DBS-venous and DBS-fingerprick comparative data. The results (correlation of 0.951 and bias of −0.06 log copies/mL) demonstrate that both the DBS-fingerprick and DBS-venous protocols have very similar performance, when compared to standard plasma protocol.
Table 5 Classification with 1000 RNA copies/mL threshold: DBS-venous vs. plasma.
VL ≥ 1000 copies/ml 239
VL ≥ 1000 copies/ml VL < 1000 copies/ml
VL ≥ 1000 copies/ml 71
6. Conclusion The new one-spot Abbott RealTime HIV-1 DBS protocol has an 60
Journal of Clinical Virology 92 (2017) 56–61
N. Tang et al.
[4] N.T. Parkin, Measurement of HIV-1 viral load for drug resistance surveillance using dried blood spots: literature review and modeling of contribution of DNA and RNA, AIDS Rev. 16 (2014) 160–171. [5] A. Johannessen, Dried blood spots in HIV monitoring: applications in resourcelimited settings, Bioanalysis 2 (2010) 1893–1908. [6] U. Neogi, S. Gupta, R. Rodridges, P. Sahoo, S. Rao, B. Rewari, S. Shastri, A. Costa, A. Shet, Dried blood spot HIV-1 RNA quantification: a useful tool for viral load monitoring among HIV-infected individuals in India, Indian J. Med. Res. 136 (December (6)) (2012) 956–962. [7] R. Hamers, P. Smit, W. Stevens, R. Schuurman, T. Rinke de Wit, Dried fluid spots for HIV type-1 viral load and resistance genotyping: a systematic review, Antivir Ther. 14 (5) (2009) 619–629. [8] Interim Technical Update. Technical and Operational Considerations for Implementing HIV Viral Load Testing, World Health Organization, 2014, pp. 1–26. [9] N. Tang, S. Huang, J. Salituro, W.-B. Mak, G. Cloherty, J. Johanson, Y.H. Li, G. Schneider, J. Robinson, J. Hackett Jr., P. Swanson, K. Abravaya, Abbott RealTime HIV-1 viral load assay for automated quantitation of HIV-1 RNA in genetically diverse group M subtypes A-H, group O and group N samples, J. Virol. Methods 146 (2007) 236–245. [10] A. Sousa, J. Houston, M. Assane, J. Chang, E. Koumans, I. Jani, J. Sabatier, P. Vaz, C. Yang, E. Rivadeneira, Performance of HIV Viral Load with Dried Blood Spots in Children on ART in Mozambique, CROI, 2015. [11] C. Garrido, N. Zahonero, A. Corral, M. Arredondo, V. Soriano, C. de Mendoza C, Correlation between human immunodeficiency virus type 1 (HIV-1) RNA measurements obtained with dried blood spots and those obtained with plasma by use of Nuclisens EasyQ HIV-1 and Abbott RealTime HIV load tests, J. Clin. Microbiol. 47 (2009) 1031–1036. [12] M. Arredondo, C. Garrido, N. Parkin, N. Zahonero N, S. Bertagnolio, V. Soriano, C. de Mendoza, Comparison of HIV-1 RNA measurements obtained by using plasma and dried blood spots in the automated Abbott real-time viral load assay, J. Clin. Microbiol. 50 (2012) 569–572. [13] S. Mtapuri-Zinyowera, F. Taziwa, C. Metcalf, E. Mbofana, S. De Weerdt, L. Flevaud, S. Simons, J.-F. Saint-Saveur, E. Fajardo, Field Evaluation of Performance of Dried Blood Spots (DBS) from Fingerprick for the Determination of Performance of Viral Load in a Resource-constrained Setting in Urban and Rural Zimbabwe, Presented at The 7th International AIDS Society Conference on HIV Pathogenesis, Treatment and Prevention, Kuala Lumpur, Malaysia, June 30–July 3, 2013, 2017 Abstract WEPE610. [14] S. Rutstein, D. Kamwendo, L. Lugali, I. Thengolose, G. Tegha, S.A. Fiscus, J.A. Nelson, M.C. Hosseinipour, A. Sarr, S. Gupta, F. Chimbwandira, R. Mwenda, R. Mataya, Measures of viral load using Abbott RealTime HIV-1 Assay on venous and fingerstick dried blood spots from provider-collected specimens in Malawian District Hospitals, J. Clin. Virol. 60 (2014) 392–398. [15] Processing of Dried Blood Spots Standard Operating Procedure, ACTG/IMPAACT Lab Tech committee, 2012. [16] S. Fernandes, K. Morosyuk, M. Ramanathan, L. Rainen, Evaluation of effect of specimen-handling parameters for plasma preparation tubes on viral load measurements obtained by using the abbott RealTime HIV-1 load assay, J. Clin. Microbiol. 48 (2010) 2464–2468. [17] EP17-A2, Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures. Approved Guideline, by Clinical Laboratory Standards Institute, second edition, June, (2012). [18] EP6-A, Evaluation of the Linearity of Quantitative Measurement Procedures: A Statistical Approach; Approved Guideline, by Clinical Laboratory Standards Institute, April, (2003). [19] Evaluation of Precision of Quantitative Measurement Procedures: Approved Guideline −Third edition, by Clinical Laboratory Standards Institute, October, (2014).
analytical sensitivity below 1000 copies/mL, tested linear range from 500 copies/mL to 107 copies/mL (claimed linear range from an LLQ of 839 copies/mL to a ULQ of 107 copies/mL), and ≤0.3 log copies/mL inter-assay SD. The clinical performance evaluation showed high sensitivity of 93% and specificity of 95% for both DBS-fingerprick and DBS-venous when compared to the plasma assay results at 1000 copies/mL threshold for virologocal failure. The data supports the use of the Abbott RealTime HIV-1 assay with the DBS protocol as a convenient sample collection/transport method for HIV-1 viral load monitoring in resource limited settings and thus aiding the 90:90:90 initiatives Funding This work was funded by Abbott Molecular Inc. Ethical approval All samples were collected under ethical approval. The protocols were approved by country/local ethical committees. Conflict of interest The following authors: Ning Tang, Vihanga Pahalawatta, Andrea Frank, John Lampinen, Gregor Leckie, Shihai Huang, and Klara Abravaya are employees of Abbott Molecular (Ning Tang, Gregor Leckie, Shihai Huang, and Klara Abravaya are also shareholders of Abbott Laboratories). Acknowledgements We would like to thank all study participants in South Africa and blood donors in Uganda and Ivory Coast; the submitting doctors (Dr Osman Ebrahim; Dr David Johnson; Dr Lydia K Masemola); the staff at BARC-SA and Lancet, the staff at Abbott Molecular. References [1] World Health Organization, Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV Infection: Recommendations for a Public Health Approach, World Health Organization, 2013, pp. 1–269. [2] 90-90-90 An Ambitious Treatment Target to Help End the AIDS Epidemic, (2017) http://www.unaids.org/sites/default/files/media_asset/90-90-90_en_0.pdf. [3] G. Wu, M. Zaman, Low-cost Tools for Diagnosing and Monitoring HIV Infection in Low-resource Settings 90 Bulletin of the World Health Organization, 2012, pp. 914–920, http://dx.doi.org/10.2471/BLT.12.102780.
61