Performance evaluation of Xpert HBV viral load (VL) assay

Preparing to load PDF file. please wait...

0 of 0
100%
Performance evaluation of Xpert HBV viral load (VL) assay

Transcript Of Performance evaluation of Xpert HBV viral load (VL) assay

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .

1 Title- Performance evaluation of Xpert HBV viral load (VL) assay: Point-of-care molecular test 2 to strengthen and decentralize management of chronic hepatitis B (CHB) infection

3 Authors- Khodare Arvind1*, Gupta Ekta2*$, Nitiksha3*, Singh Gaurav4*, Aggarwal Kavita5*, 4 Sharma Manoj6#, Sarin SK7# 5 * Department of Clinical Virology, Institute of Liver and Biliary Sciences (ILBS), New Delhi 6 # Department of Hepatology, Institute of Liver and Biliary Sciences (ILBS), New Delhi 7 $ Corresponding author: [email protected]

8

9

1. [email protected]

10

2. [email protected]

11

3. [email protected]

12

4. [email protected]

13

5. [email protected]

14

6. [email protected]

15

7. [email protected]

16

17 Abstract

18 Introduction: Estimation of hepatitis B (HBV) viral load (VL) is critical in hepatitis-B cascade19 of-care and currently there is no point of care (POC) molecular assay available for that. This 20 study evaluated the performance of a new near point of care molecular assay Xpert HBV-VL 21 assay against FDA approved Real time PCR assays.

22 Materials & methods: In this retrospective study 119 archived plasma samples from HBV 23 infected patients, and 53 hepatitis B surface antigen (HBsAg) patients were simultaneously 24 tested for HBV DNA quantification on 2 real time PCR conventional assays and Xpert assay. 25 The routine method for reporting to patient was Abbott Real Time PCR.

26 Results: The range of HBV DNA load in samples was 1 to 8.76 log10IU/ml with a median load 27 of 4.46 (IQR: 1-8.76) log10IU/ml as detected by routine assay (Abbott Real-Time HBV VL 28 assay). Genotyping could be done in 95 (79.8%) samples and genotype D (83; 87.37%) was 29 found commonest. The Xpert assay demonstrated good correlation with Abbott (R2= 0.944) and

NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .
30 Roche (R2= 0.963). On comparison the mean difference (95% Confidence Interval) in average 31 viral load was -0.018 log10 IU/ml and -0.043 log10 IU/ml when Xpert was compared with the 32 Abbott and Roche assay, respectively. The overall sensitivity, specificity, negative predictive 33 value and positive predictive value of the Xpert assay was found 97.5%, 100%, 94.65 & 100% 34 respectively.
35 Conclusion: Xpert HBV-VL assay which has a potential for near point of care molecular testing 36 has shown excellent performance and found to be a reliable method for HBV DNA 37 quantification.
38 Introduction 39 Infection with the hepatitis B virus (HBV) remains an important global public health problem 40 with significant morbidity and mortality.[1] Globally the estimated prevalence of CHB infection 41 is 3.5% with 257 million cases which accounts for 80% cases of hepatocellular carcinoma (HCC) 42 along with chronic hepatitis C infection.[2] There is a large regional variation in hepatitis B 43 seroprevalence between low (<2%) and high (>8%) endemicity levels. India falls in the 44 intermediate endemicity zone (2%-8%) with an average prevalence of 4% ( 50 million cases) 45 which is responsible for 43% of HCC cases.[3] 46 The primary diagnosis of hepatitis B is based on the detection of HBsAg but HBV DNA 47 measurement is essential for the evaluation of patients with CHB for prognostication, to make 48 decision to treat and subsequent monitoring of patients for the efficacy of antiviral treatment. For 49 these indications serial monitoring of HBV-DNA levels is more important than any single 50 arbitrary cut-off value. Interpretation of HBV-DNA levels is important in the context of other 51 host factors including age, duration of infection, ALT elevation, and stage of disease when 52 making treatment decisions.[4] The risk of disease progression is reduced when a sustained

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .
53 reduction of HBV DNA levels to undetectable is achieved, which in turn prevents the 54 progression of fibrosis to cirrhosis, HCC and death.[5] 55 International practice guidelines for hepatitis B recommend using sensitive nucleic acid 56 amplification test (NAAT) for quantifying HBV DNA in clinical practice.[1,4] There are several 57 commercial NAAT assays, mostly real-time polymerase chain reaction (real-time PCR) are 58 available for quantification of HBV DNA in clinical samples.[5–7] There are challenges in 59 doing testing by conventional PCR methods Specialized infrastructure, trained manpower and 60 longer turnaround time.[8,9]. A less complex, easy to use and inexpensive assay that has 61 potential for point-of-care (POC) molecular testing are needed for more widespread availability 62 of HBV management. 63 This study evaluated the performance of the new Xpert HBV VL assay (Cepheid, Inc. 64 Sunnyvale, CA, USA) on the GeneXpert system (Cepheid) which is almost a POC or near POC 65 test. 66 Study design67 Materials and methods 68 This was a retrospective study conducted in a tertiary care liver center of North India. Archived 69 samples within 3 months of testing were retrieved from -80 °C and were simultaneously tested 70 on all the three platforms from the same freeze-thaw cycle. Overall, 119 samples with confirmed 71 HBV DNA positivity of CHB patients and 53 samples of HBsAg negative patients were 72 included. Different technicians performed testing on the 3 platforms, and were blind to the 73 results. The study was approved by Institutional Ethical Review. Samples from patients co74 infected with HIV or HCV, children (<18 years old) and pregnant females were not included in 75 the study and with insufficient available volume. Hepatitis B virus genotyping results were

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .
76 available for samples with DNA load >3log10 IU/ml and was done by Sanger sequencing method 77 for the surface gene (S) of HBV genome. 78 Xpert® HBV VL assay 79 This is an automated, real-time PCR assay for the quantification of HBV DNA performed on the 80 GeneXpert System. The reported lower limit of quantification (LLOQ) is 1 log10 IU/ml and the 81 linear range of quantification is 1-9 log10 IU/mL. The test requires 600 µL plasma or serum. All 82 the steps like nucleic acid extraction, amplification, and detection of target done in a single 83 cartridge and the result available in 59 minutes. The assay cartridge contains internal controls to 84 ensure valid performance of the test and to quantify HBV DNA load by proprietary software. 85 The GeneXpert System is available in a 1, 2, 4 or 16-module configuration. A four modules 86 instrument was used for the present study. 87 Abbott Real-Time HBV VL assay (Abbott, Wiesbaden, Germany) 88 This is an automated real-time PCR for the quantification of HBV DNA in human plasma or 89 serum. The LLOQ is 1 log10 IU/mL and the linear range of quantification is 1-9 log10 IU/mL. The 90 sample input volume is 500 µL of serum or plasma. Samples were tested on automated 91 m2000sp-m2000rt Abbott real-time PCR. 92 93 COBAS® AmpliPrep/COBAS® TaqMan® HBV Test, v2.0 (Roche Diagnostics, GmbH, 94 Mannheim, Germany) 95 This is an automated HBV viral load quantitative assay. The LLOQ is 1.3 log10 IU/mL and the 96 linear range of quantification is 1.3-8.23 log10 IU/ml. Sample input volume is 650 µL of serum or 97 plasma. Test was performed on automated Roche COBAS Ampliprep/COBAS Taqman. 98 HBV genotyping

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .
99 HBV DNA extraction from plasma samples was done by a spin column-based method using the 100 commercially available high pure viral nucleic acid kit (Roche Diagnostics, GmbH, Mannheim, 101 Germany). The DNA extract was subjected to PCR using in-house designed primers for surface 102 gene (S) genomic region (forward primer 5’-CATCAGGATTCCTAGGACCCCT-3’, reverse 103 primer 5’-AGGACAAACGGGCAACATAC-3’) on Phusion® high fidelity DNA polymerase 104 (Thermo Scientific Inc., Waltham, MA). Amplified products were purified by gel-excision using 105 the QIAquick gel extraction kit (QIAGEN, GmbH, Mannheim, Germany) to remove 106 unincorporated dNTPs and primers. This was followed by bidirectional sequencing using ABI 107 Big Dye chemistry on the ABI 3500Dx series genetic analyzer (Life Technologies, Waltham, 108 MA). Forward and reverse sequence reads were aligned and assembled using DNA Baser v3.5.1 109 software (Heracle BioSoft SRL, Romania). Genotype assignment was done by comparing the 110 obtained sequences with the consensus sequences on the Basic Local Alignment Search Tool 111 (BLAST) database of NCBI. 112 Statistical analysis 113 The HBV DNA values were expressed in log10 format. Linear regression analysis was done and 114 correlation coefficients were calculated using SPSS version 22. Agreement between the two 115 assays was determined using Bland-Altman plots and estimated the overall bias. Sensitivity, 116 specificity, positive predictive value (PPV), negative predictive value (NPV) was calculated by 117 taking Abbott and Roche, both assay as reference. 118 Results119 Of the 172 subjects included in the study, male preponderance was seen with M: F ratio of 3.3:1 120 and most were adults with a mean age of 43 (±14.7) years. The range of HBV DNA load in 121 samples was 1 to 8.76 log10IU/ml with a median load of 4.46 (IQR: 3.12-6.39) log10IU/ml as

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .

122 detected by reference method (Abbott) which was used as a routine assay for reporting of results

123 to the patients as per our institution’s practice. Genotype D was the commonest, seen in 83

124 (87.37%) of the cases.(Table 1) Median values of viral loads detected by all three assays are

125 given in table 1.

126

Table 1. Baseline characteristics of the study population, n=172

Variable

Values

Age (Mean ± SD) years

43 ±14.7

Male gender, n (%)

132 (76.7%)

Female gender, n (%)

40 (23.3%)

Male: Female

3.3:1

ALT (median, IQR) IU/ml

45 (77-29)

AST (median, IQR) IU/ml

41 (68-29)

Genotype, n (%)

A

12 (12.63%)

D

83 (87.37%)

Median (IQR) HBV DNA (log10 IU/ml)

Abbott

4.46 (6.39-3.12)

Roche

4.47 (6.45-3.00)

Xpert

4.35 (6.53-3.18)

Xpert assay performance

Sensitivity

97.5% (95% CI; 92.8-99.5)

Specificity

100% (95% CI; 93.3 -100)

PPV

100% (95% CI; 96.9 -100)

NPV

94.6% ( 95% CI; 85.1-98.9)

DNA: Deoxyribonucleic acid; IU: international unit, ALT: Alanine

transaminase; AST: aspartate aminotransferase; SD: standard

deviation; ml: millilitre. PPV & NPV: Positive & negative predictive

value, CI: Confidence interval

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .

Table 2. Level of agreement between the assays

Abbott vs Xpert
Roche vs Xpert

Overall D A Overall D A

Mean (log10 IU/ml) paired
difference -0.018 -0.096 0.038 -0.043 -0.155 0.074

95% limit of agreement Upper Lower 1.10 -1.13 1.04 -1.23 1.10 -1.02 0.88 -0.97 0.65 -0.96 0.39 -0.24

P value
0.74 0.14 0.81 0.32 0.001 0.62

127

Table 3. Repeatability of Xpert assay

Number High positive Low positive

of tests

standard

standard

(4log10IU/ml) (2log10IU/ml

1

4.04

2.04

2

4.00

2.03

3

4.00

2.02

4

4.01

2.01

5

3.99

1.99

6

4.00

1.99

7

3.99

1.99

8

3.95

1.98

9

4.02

2.02

10

4.01

2.01

Mean

4.001

2.08

SD

0.023

0.020

CV%

0.56

0.99

128

129

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .
130 Comparison between Abbott and Xpert assay 131 On linear regression analysis of the quantifiable viral loads a very high positive correlation was 132 seen (R2= 0.944) (Fig.1) and on Bland-Altman analysis 95% samples showed results within the 133 limit of agreement of the assay ranging from -1.13 to 1.1 log10 IU/ml with a bias of -0.018 log10 134 IU/ml (Fig.2A). 135 136 Comparison between Abbott and Roche assay 137 On linear regression analysis of the quantifiable viral loads a very high positive correlation was 138 seen (R2= 0.963) (Fig.1) and on Bland-Altman analysis 95% samples showed results within the 139 limit of agreement of the assay ranging from -0.96 to 0.65 log10 IU/ml with bias of -0.043 log10 140 IU/ml (Fig.2B). 141 142 Clinical performance of Xpert 143 Overall concordance was seen in 169 (98.25%) of 172 samples when Xpert assay was compared 144 with Abbott and Roche assay. The sensitivity of Xpert assay was found 97.5% (95% CI; 92.8145 99.5). All the 53 HBsAg negative samples tested negative by all three assays, so the specificity 146 was found 100% (95% CI; 93.3 -100). There were 3 (2.52%) out of 119 samples had discrepant 147 results and the viral load detected by Abbott assay in these three was 14, 12 and 12 IU/ml, but 148 were not detected by Xpert. In these 3 samples, HBV DNA was also detected by Roche and it 149 was less than the LLOQ i.e. 20 IU/ml. Due to insufficient sample volume the assays were not 150 repeated in these 3 cases.
151 152 Analytical performance of Xpert assay

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .
153 To check analytical performance of Xpert assay 3rd WHO International Standard for HBV 154 (NIBSC code 10/264) was used.[10] The reconstituted standard was diluted in human plasma 155 tested negative for all markers of HBV, hepatitis C virus (HCV) and HIV to make high positive 156 (4 log10IU/ml) and low positive (2 log10 IU/ml) standard. For precision testing, high and low 157 positive standards were tested on Xpert in duplicates on 5 consecutive days (to obtain 10 data 158 points for each dilution) to evaluate the precision of the assay. The high positive standard was 159 also used to make four serial 10-fold dilutions having a concentration of 4, 3, 2 and 1 log10IU/ml 160 and tested to evaluate the linearity of Xpert assay for quantification. 161 The Xpert assay demonstrated good precision with a coefficient of variation (CV) for the ten 162 HBV DNA load measurements which was 0.56% for the high positive standard and 0.99% for 163 the low positive standard. (Table 3) On linear regression analysis of four serial 10 fold dilutions 164 of WHO standard, an excellent correlation was seen between the measured HBV DNA 165 concentrations and the expected concentrations (R2 = 0.998). (Figure 3) On analytical 166 performance, the Xpert assay was found to be linear and reproducible. 167 168 Figure 1 Linear regression analysis for correlation of Xpert assay with Abbott and Roche assay 169 for quantification of HBV DNA.
170 Figure 1 has shown the correlation of measurements between Abbott, Roche and Xpert assay. An 171 excellent correlation between the Abbott and Xpert assay was observed (R2= 0.944, P < 0.001). 172 An excellent correlation was also observed between the Roche and Xpert assay (R2 =0.963, P < 173 0.001).

medRxiv preprint doi: https://doi.org/10.1101/2020.05.31.20104760; this version posted June 2, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC 4.0 International license .

174 175 Figure 2. Level of agreement for quantification of HBV DNA between the assays by Bland176 Altman plot; between Abbott and Xpert assay (A), between Roche and Xpert assay (B)

177 Figure 2A has shown the mean difference (Abbott − Xpert) of -0.018 log10IU/mL (limits of

178 agreement: −1.13 to 1.1 log10IU/mL) and 2B has shown the mean difference (Roche − Xpert) of

179 -0.043 log10IU/mL (limits of agreement: −0.97 to 0.88 log10IU/mL).

180

181

A

B

182 183 184 Figure 3. Xpert assay HBV DNA quantification linearity
LicensePreprintXpert AssayAbbottPeer Review