Automated Nucleic Acid Isolation Methods for HDV viral Load Quantification can Lead to viral Load Underestimation

Introduction

HDV infection is a severe and not infrequent cause of liver disease with more than 15 million people infected worldwide [1]. Hepatitis D represents a particular public health problem in some Asian, African and South American regions [2].

The current treatment of HDV infection is based on pegylated interferon-α2a (PEG-IFN-α) administration, which can lead to sustained biochemical response in 40% and prolonged HDV RNA suppression in 25–30% of the patients [3]. Development of alternative treatments is imperative considering the severity of HDV-induced chronic liver disease and the limited efficacy of pegylated interferon therapy, which is also associated with frequent side effects [4]. Indeed, novel treatments including lonafarnib, a prenylation inhibitor, the nucleic acid polymer REP 2139 and myrcludex-B, an entry inhibitor, have been evaluated in small clinical trials [4–7].

HDV RNA levels may influence the natural course of liver disease, increasing the likelihood of progression to liver cirrhosis and the development of hepatocellular carcinoma [8]. Monitoring of viral load is critical to the evaluation of treatment response. Personalized treatment duration for PEG-IFN-α therapy has been suggested based on HDV RNA kinetics [9] and prolonged treatment may be needed for patients with low viral load [10]. In addition, to determine the efficacy of new treatments, reliable quantification of HDV RNA is essential.

For many years there was no international standard for HDV RNA quantification. Since 2012 a WHO standard for HDV RNA quantification has become available [11]. However, many in-house tests show poor performance and few commercial HDV RNA assays are available [12]. Recently CE-certified commercial kits have been introduced and have been shown to accurately quantify HDV genotype-1 in samples from European and Asian patients [13]. Even though primer selection and amplification methods may have been improved and optimized, the distinct role of RNA extraction methods has never been evaluated to date.

The aim of the present study was to compare four automated nucleic acid (NA) extraction methods commonly used in laboratories (AmpliPrep, MagNA Pure, QIAcube QBK and QIAcube VRK) versus a manual RNA extraction method and evaluate the possible effect of each method on HDV RNA yield with subsequent amplification with the Robogene HDV assay.

Methods

Serial serum samples from 6 HDV (genotype-1) positive patients, enrolled in the HIDIT-2 study (ClinicalTrials.gov Identifier: NCT00932971), were studied. The samples were taken from each patient at baseline before therapy start with PEG-IFN-α and 12 and 48 weeks after treatment start. The samples were stored at −80°C until extraction. The following methods were performed for nucleic acid (NA) extraction according to the manufacturer's instructions: four automated methods, COBAS AmpliPrep (Roche Molecular Systems, Manheim, Germany), MagNA Pure 96 (Roche Molecular Systems), QIAcube QBK and QIAcube VRK (Qiagen GmbH, Hilden, Germany) versus a manual RNA extraction method (Instant Virus RNA/DNA kit, Analytik Jena AG, Jena, Germany). In detail, the extraction volume was 400 μl for the manual method, 500 μl for AmpliPrep, 200 μl for MagNA Pure, 200 μl for QIAcube QBK and 140 μl for QIAcube VRK. The elution volume was 60 μl for the manual method, 75 μl for AmpliPrep, 100 μl for MagNA Pure, 100 μl for QIAcube QBK and 70 μl for QIAcube VRK. Hepatitis D viral load (HDVL) quantification was performed using the Robogene HDV RNA quantification kit 2.0 (Analytik Jena AG) according to the manufacturer's instructions.

Statistical analyses were performed using SPSS 21.0 (IBM SPSS, Inc., Chicago, IL, USA) and GraphPad Prism 5.0 (Graph Pad, San Diego, CA, USA). For the purpose of viral load statistical analyses undetectable samples were considered equal to zero. Samples that were detectable but not quantifiable were recorded as 1 IU/ml. The viral loads from samples extracted using each automated NA method were compared to those extracted using the manual extraction method using the Wilcoxon matched-pair signed-rank test and plotted using the Bland–Altman method [14]. Two-tailed P-values less than 0.05 were considered as statistically significant.

Results

The median viral load in the 18 patient samples (three time points of 6 patients) differed significantly when using the different NA extraction methods. All automated methods yielded significantly lower HDVL compared to the manual method. The median viral load for the manual method performed with the instant Virus RNA/DNA kit was 10,665 IU/ml with an interquartile range of 1,474,795 IU/ml, 445 IU/ml with an interquartile range of 136,991 IU/ml for AmpliPrep, 3,209 IU/ml with an interquartile range of 191,742 IU/ml for MagNA Pure, 2,060 IU/ml with an interquartile range of 576,489 IU/ml for QIAcube QBK and 3,568 IU/ml with an interquartile range of 263,497 IU/ml for QIAcube VRK. HDVL quantification results of automatedly extracted NA significantly underestimated the viral load, when compared to the results of the manual extraction as shown in Figure 1.

OPEN IN VIEWER

Figure 1

HDVL quantification results of automatedly extracted NA significantly underestimated the viral load, when compared with the results of the manual extraction

We performed Bland–Altman analysis to compare the performance of the different extraction methods. The analysis showed that MagNA Pure, Ampliprep and QIAcube VRK extractions lead to the detection of about 10-fold lower HDVL compared to the manual extraction (ranging from −0.5 to 2.11 log₁₀ fold), while the difference with QIAcube QBK was smaller (5.86-fold) as shown in Figure 2. In more detail the median ratio of HDVL for the manual to the Ampliprep method was 10.3 with a minimum of 1.11 and a maximum of 51.68. The median ratio of HDVL was 11.85 (ranging from 1.08 to 123.62) when comparing the manual to the MagNA pure method, 5.86 and ranging from 0.32 to 36.35 for the manual to the QIAcube QBK method and 10.24 and ranging from 0.79 to 127.6 for the manual to the QIAcube VRK method.

OPEN IN VIEWER

Figure 2

Bland–Altman analysis

Furthermore, HDV RNA PCR tested negative in two samples, when performing the NA extraction with Magna Pure, but positive when using any other method as shown in Figure 3, which presents the virological course of all six patients.

OPEN IN VIEWER

Figure 3

The viral load of six HDV-positive patients treated with PEG-IFN-α was measured at baseline and 12 and 48 weeks after therapy start

Discussion

In this study we demonstrated that the NA extraction method can have a significant impact on HDVL quantification. The HDVL quantification results using automatedly extracted NA significantly underestimated the viral load determined with the Robogene HDV RNA quantification kit 2.0, when compared with RNA acquired through manual extraction. Use of the automated methods, MagNA Pure, Ampliprep and QIAcube VRK extraction led to the detection of about 10-fold lower HDVL compared with the manual method of RNA extraction, while the difference when using QIAcube QBK was smaller (about 6-fold lower). Furthermore, use of MagNA Pure led to the misclassification of two samples with low viral loads taken during pegylated interferon therapy as false negative.

Different extraction methods use different starting and elution volumes. For our experiments we used the starting and eluting volumes recommended by the manufacturer of each kit for each extraction to ensure reproducibility. Amongst the evaluated methods, higher starting volumes were not always associated with greater sensitivity: for example, the MagNA pure method, which demonstrated reduced sensitivity compared to the other methods, had a lower starting volume and greater elution volume than the manual method but the same starting and elution volume as the QIAcube QBK method, by which it was outperformed in terms of sensitivity. The difference in starting and elution volume alone could not account for the disparities in the measured viral load in the evaluated NA extraction methods.

The impact of RNA or DNA extraction method has been evaluated in several studies quantifying HIV [15], HCV [16] or HBV viral load [17–19], with the manual extraction methods having better, equal or worse performance than automated methods in different settings. In a study evaluating HBV and HCV viral load quantification, the authors compared the performance of an automated extraction method (NucliSens easyMAG; BioMérieux, Craponne, France) and a manual extraction method (QIAmpMin Elute Kit; Qiagen, Hilden, Germany). For HBV, a higher amount of HBV DNA was measured in several samples when using the automated method compared with the manual method, while for HCV RNA the opposite was true, with more samples having a higher amount of HCV RNA when using the manual method [18]. Another study evaluating different methods of HBV nucleic acid extraction demonstrated that the QIAamp system, a manual extraction method, had comparable performance to the MagNA Pure 96 and a better performance than the Chemagic system [19], while DNA extraction to measure cytomegalovirus and Epstein–Barr virus viral load, with a manual and an automated (NucliSens easyMAG) method showed good sensitivity for both techniques [20,21]. A study comparing manual extraction (Abbott Sample Preparation System, RealTime HIV-1 m2000rt) versus automated (Abbott m2000sp platform) for HIV-positive samples found no clinically relevant differences in viral load between the two methods [22], while in another setting three HIV-positive samples with low levels of nucleic acids were detected as such when using MagNA Pure, but yielded negative results when using the manual method, demonstrating the reliability of the automatized MagNA Pure system for HIV RNA isolation [23]. Again, in yet another study using HIV-positive samples and comparing the automated RNA extraction methods, AmpliPrep and MagNA Pure, versus the manual method, High Pure Viral Nucleic Acid Isolation kit from Roche Diagnostics, demonstrated that the detected HIV-1 RNA concentration obtained by MagNA Pure system was higher than the other two methods [24]. In summary, the impact of the extraction method on viral quantification has been evaluated for different viruses, in different settings, with diverging performance results, with no clear universal advantage of any method over the others.

Very limited data exists about the impact of nucleic acid extraction on HDVL quantification. A recent study by Le Gal et al. [12], evaluating many available commercial and in-house assays for plasma HDV RNA quantification at an international level, underlined the high heterogeneity of assay characteristics in different laboratories. According to the authors, however, technical features, including RNA extraction methods, seemed to have no significant impact on the final quantitative results. The discrepancies in the measured HDVL were attributed to primer or probe mismatches related to the high genetic variability of HDV and, possibly, to the complex secondary structure of the target genomic RNA [12]. Our results demonstrated that different extraction methods of HDV RNA led to a significant difference in the quantified HDVL. In fact, for some samples the difference was hundred-fold. The highest viral load was measured in samples extracted manually, while one of the automated methods (MagNA pure) demonstrated reduced sensitivity, leading in two cases even to a false-negative result. A possible explanation for the discrepant conclusions in our study versus the study of Le Gal et al. [12] could be confounding factors such as the different PCR methods used in different laboratories in the latter study. In our study the RNA extracts were amplified using the same PCR assay in the same laboratory.

A limitation of our study is the relatively small sample number, but samples were well selected and all patients had HDV genotype-1 infection. We selected samples from the same patient at different time points in order to minimize interpatient variability and to evaluate the performance of each method with different viral loads. Still, the impact of NA extraction on different HDV genotypes needs to be determined in future studies.

Nucleic acid isolation is the first step to the quantification of viral load. Both automated and manual methods have advantages and disadvantages. The manual method takes a longer time, binds more human resources, is less standardized and is prone to human error and a risk of contamination. Use of automated methods, for all their other advantages, can be more expensive and, more importantly in the case of HDVL quantification, can significantly underestimate the viral load when using Robogene HDV RNA quantification kit 2.0.

The findings can have major implications for patient diagnosis, antiviral treatment and follow-up. It would be advisable to use a uniform method for RNA extraction, when measuring HDV RNA during follow-up. In case of a negative result, produced after RNA extraction with an automated method, it would advisable to retest the sample using manual RNA extraction, if a positive result would be of clinical relevance. Furthermore, due to the variability of HDVL depending on the RNA extraction method we believe that the RNA extraction method should be reported in studies including HDVL quantification.