Introduction

To date, the most widely used HCV RNA quantification assays are the Abbott RealTime HCV assay (RealTime) and the Roche COBAS AmpliPrep/COBAS TaqMan HCV test version 2 (CTM v2). They were calibrated with historical WHO standards (2nd) and may vary in sensitivity and precision depending on genotypes and particularly in the setting of low-level viremia, as previously reported [1–3]. A major role in DAA therapy (direct acting antivirals) was assigned to HCV RNA monitoring not only for investigating initial viral load but also for assessing patient adherence, sustained virological response (SVR; defined as HCV RNA < 25 IU/ml in DAA registration trials) and to follow stopping rules at 25 IU/ml for the compounds Simeprevir and Daclatasvir or even HCV RNA detectability or undetectability at certain points of time during therapy [4–6, 8, 9]. Viral loads below or above these thresholds are intended to guide therapy decision, in particular at the end of treatment. The clinical decision points were implemented using the manual Roche High Pure System/COBAS TaqMan assay (HPS) which, however, is rarely used in clinical routine.

The comparison of six commercially available HCV RNA assays (one laboratory per assay) revealed a substantial variety in assay precision at low viremic levels in particular at 25 and 100 IU/ml [7]. Additionally, substantial differences in quantification were also described between the HPS reference assay, which was used to establish the clinical decision points in clinical studies, and some assay comparators in this recent analysis [7].

In order to confirm the results of this initial HCV low-level viremia (LLV) study that used a one-laboratory-per-assay approach and in order to assess laboratory-specific biases, the present study focussed on the assessment of precision and detection rates of RealTime and CTM v2 results obtained by three independent laboratories each (multiple laboratories per assay) and using diluted 4th HCV WHO Standard samples as well as low viremic clinical samples near detection limits of chronically HCV-infected patients.

Materials and methods

Dilutions were prepared from the 4th WHO standard (NIBSC code: 06/102, genotype 1a) to achieve nominal concentrations of 100, 25 and 10 IU/ml using HCV-negative BaseMatrix (Seracare Life Sciences; HCV negativity was confirmed by using the Abbott RealTime HCV assay). For each participating laboratory, fifteen aliquots of each dilution level were prepared which were treated identically and underwent the same freeze-thawing cycles. Samples were transported on dry ice to the participating laboratories and successfully delivered in a frozen state.

Seven different laboratories participated in this study with 3 laboratories using RealTime (PZB Aachen, Germany/University of Essen, Germany/University of Basel, Switzerland), 3 laboratories using CTM v2 (University of Frankfurt, Germany/MHH Hannover, Germany/University Hospital Zurich, Switzerland) and 1 laboratory using HPS (University of Frankfurt, Germany). Assay characteristics as provided by the manufacturer are listed in Table 1.

Table 1 Assay characteristics as provided by manufacturers
Full size table

In each laboratory, three aliquots of each dilution were tested in five independent runs, respectively. Medians, coefficients of variation (CV) and their confidence intervals were determined for 100 IU/ml dilutions and detection rates for 25 and 10 IU/ml dilutions.

In a second approach, two clinical samples (one genotype 1a and one genotype 1b sample) were diluted in BaseMatrix to concentrations of 100 and 25 IU/ml and were transported to all participating centres. On each system, three replicates of each concentration were tested in five independent runs (15 aliquots in total), respectively, and CV values, confidence intervals as well as medians were determined. The initial HCV RNA concentrations of the clinical samples had been determined with the Abbott RealTime HCV system prior to dilution.

CVs were only calculated if at least 2 of 3 replicates in each of at least 4 of 5 runs per sample dilution were quantified.

Results

WHO standard dilutions

Assay result variation for WHO standard replicates

The overall testing results of WHO standard replicates at nominal concentrations of 100 and 25 IU/ml are shown in Fig. 1a, b. Results of the different laboratories are differentiated by blue, green and red colour. While the overall result range across all laboratories for 100 IU/ml replicates was similar for CTM v2 and RealTime in this analysis (Fig. 1: all colours), the CVs of the individual laboratories were lower across RealTime results with a range of 19.3–25.6 % compared to CVs of CTM v2 laboratories with a range of 26.1–47.3 % and the HPS laboratory with a CV of 34.9 % (Table 2). As shown in Table 2, overall median values across all replicates at 100 IU/ml and across all laboratories were similar for CTM v2 (56 IU/ml; range 22–72 IU/ml per laboratory) and RealTime (54 IU/ml; range 31–79 IU/ml per laboratory) and differed moderately from HPS median in this analysis (83 IU/ml) (Table 2). As shown in Fig. 1b, three-times more replicates at 25 IU/ml were quantified with RealTime (24) than with CTM v2 (8), with one CTM v2 centre (blue colour) and one RealTime centre (red colour) quantifying only one replicate each. In total, two replicates were quantified above 25 IU/ml in the HPS centre compared to two replicates by RealTime and none by CTM v2. Due to the low number of quantified replicates, a calculation of CV values at 25 IU/ml was not possible.

Fig. 1
figure 1

Total WHO standard variation analysis was performed in triplicates in 5 independent runs (3 × 5 replicates) for each participating centre, as indicated by different marks. a WHO standard replicates with a nominal concentration of 100 IU/ml (dotted line). b WHO standard replicates with a nominal concentration of 25 IU/ml

Full size image
Table 2 Median and CV data for the WHO dilution at 100 IU/ml nominal concentration
Full size table

Analysis of assay detection rates for WHO standard replicates

Furthermore, the ability of each laboratory and each assay to detect HCV RNA in 15 WHO replicates at nominal concentrations of 25 and 10 IU/ml, respectively, was analysed (Table 3).

Table 3 Detection rates for 4th WHO replicates with nominal concentrations of 25 and 10 IU/ml
Full size table

Overall, in nearly all 25 IU/ml replicates, HCV RNA was detected by CTM v2, HPS and RealTime. However, the amount of quantified replicates differed between CTM v2 (18 %) and RealTime (55 %). HPS, which has a limit of quantification of 25 IU/ml, quantified 13 % of 15 replicates. For replicates at 10 IU/ml, 22 % were not detected by CTM v2 centres compared to 7 % by RealTime laboratories. The HPS reference centre did not detect 20 % of replicates. Overall detection rates were 97 % for RealTime, 88 % for CTM v2 and 90 % for HPS.

Analysis of assay precision and quantification of replicates of clinical samples with genotypes 1a and 1b

The overall assay variations in clinical genotype 1 samples at nominal concentration of 100 and 25 IU/ml are illustrated in Fig. 2. As already shown for WHO standard replicates, this analysis confirmed differences in precision across the evaluated assays independent of the participating centres (Fig. 2). Results from clinical samples revealed higher precision and lower CVs for RealTime as compared to CTM v2 (Table 4). The calculated CVs obtained across RealTime laboratories ranged from 13.1 to 21.0 % for replicates at 100 IU/ml and from 17.6 to 34.9 % for those at 25 IU/ml. In comparison, the calculated CVs obtained across CTM v2 laboratories ranged from 15.0 to 32.3 % at 100 IU/ml and were particularly higher for replicates at 25 IU/ml with CVs of 28.2–54.9 %. The CVs for HPS were 17.7–23.6 % at 100 IU/ml and 22.8–42.5 % at 25 IU/ml (Table 4).

Fig. 2
figure 2

Variability charts of 15 replicates for each of the clinical genotype 1a sample (a, b) and genotype 1b sample (c, d) at nominal concentrations of 100 and 25 IU/ml, respectively. Numbers on the x-axis represent the different laboratories (1–3 represent RealTime centres, 4–6 represent CTM v2 centres, 7 represents the HPS reference centre)

Full size image
Table 4 Median and CV data for clinical genotype 1a and 1b samples
Full size table

In general, results obtained with CTM v2 and HPS when analysing the genotype 1b replicates at 25 IU/ml had higher coefficients of variation than corresponding genotype 1a results.

Median values for genotype 1a and 1b replicates at 100 and 25 IU/ml nominal concentrations across the participating laboratories are listed in Table 4. Compared to genotype 1b replicates, median values of genotype 1a replicates measured by HPS or CTM v2 showed discrepant quantification results relative to RealTime results (Table 4).

Discussion

The results of this multicentre-per-assay study confirm and amend the conclusions of the previously reported initial investigation which was based on a one-laboratory-per-assay approach [7]. The intention of this study was to focus particularly on very low viremic specimens near the clinical cut-off of 25 IU/ml in a more comprehensive setting. RealTime demonstrated the highest detection rate (97 vs. 90 % for HPS and 88 % for CTM v2) in detecting WHO standard samples with nominal concentrations of 25 IU/ml and below. This corresponds to the ranking in the previous one-laboratory-per-assay study with standard dilutions at 25, 10 and 5 IU/ml. In the clinical sample analysis of genotypes 1a and 1b, intra-assay variation was consistently lower for RealTime (CVs at 100 and 25 IU/ml ranged from 13–21 % and 16–35 %, respectively) as compared to CTM v2 (15–32 and 28–55 %, respectively) and HPS (18–24 and 23–43 %, respectively), indicating a higher precision of this assay especially at a low nominal concentrations of 25 IU/ml. Despite the slightly different study design of the initial HCV LLV study with 10 replicates tested individually in 10 independent runs [7], results of this multicentric study are in line with the mean CV results obtained from the initial HCV LLV study (18 % at 100 IU/ml and 27 % at 25 IU/ml for RealTime, 32 and 53 % for CTM v2, and 32 and 48 % for HPS, respectively).

In registration trials for the new direct acting antivirals that were approved since 2014, SVR was consistently defined as HCV RNA < 25 IU/ml (either detected or undetected). Thus, a high precision at 25 IU/ml is of major importance to properly assess SVR. Furthermore, previously observed differences of quantification levels across assays were confirmed also in this multicentre-per-assay setting. This should be taken into consideration when switching assays during therapy and a comparison of quantification is suggested in this case.

Current guidelines of the European Association for the Study of the Liver (EASL) continue to recommend monitoring of viral load also in IFN-free regimens to assess patient adherence and therapy efficacy [8, 9]. Newer compounds and additional substances may lead to a simplification of virological management in HCV treatment in the future. However, it might also take several years for these compounds to become available and refundable especially in resource-limited countries.

Conclusions

Presumably, quantification assays with high accuracy and precision will have their clinical relevance for the upcoming years since only one single measurement is used to assess SVR or to decide if a patient is eligible to continue an expensive antiviral therapy.