Development and Clinical Validation of a Pangenotypic PCR-Based Assay for the Detection and Quantification of Hepatitis E Virus (Orthohepevirus A Genus)

The objective of this study was to design a pangenotypic PCR-based assay for the detection and quantification of hepatitis E virus (HEV) RNA from across the entire spectrum of described genotypes belonging to the Orthohepevirus A genus. The optimal conditions and the performance of the assay were determined by testing the WHO standard strain (6219/10) and the WHO HEV panel (8578/13). Similarly, performance comparisons were made with two commercial assays (Real Star HEV RT-PCR 2.0 and ampliCube HEV 2.

ABSTRACT

The objective of this study was to design a pangenotypic PCR-based assay for the detection and quantification of hepatitis E virus (HEV) RNA from across the entire spectrum of described genotypes belonging to the Orthohepevirus A genus. The optimal conditions and the performance of the assay were determined by testing the WHO standard strain (6219/10) and the WHO HEV panel (8578/13). Similarly, performance comparisons were made with two commercial assays (Real Star HEV RT-PCR 2.0 and ampliCube HEV 2.0 Quant) to detect HEV RNA at concentrations below 1,000 IU/ml with viral strains from the WHO and to test samples from 54 patients with acute hepatitis. The assay presented in this study was able to detect the entire spectrum of described genotypes belonging to the Orthohepevirus A genus, demonstrating better performance than both commercial kits. This procedure may represent a significant improvement in the molecular diagnosis of HEV infection.

INTRODUCTION

Hepatitis E virus (HEV) is an RNA virus that belongs to the genus Orthohepevirus A. The species included in this genus are a major cause of acute hepatitis worldwide (1). According to epidemiological criteria, the major four genotypes (i.e., genotypes 1 to 4) can be grouped into two main groups: genotypes 1 and 2, for which humans appear to be the only hosts and which produce large outbreaks in developing countries (primarily Africa and Asia) (1), and genotypes 3 and 4, which can infect humans and a large variety of animals and for which such infections are usually reported as sporadic cases (2). To these four primary genotypes, which have already been widely studied, other genotypes (i.e., genotypes 3ra and 5 to 8) with zoonotic capacity must be added (3–7).

HEV has shown considerable variability among subtypes (8–11). However, despite the emerging nature of some HEV genotypes and subtypes, at present, there is no PCR screening assay that has been designed to detect the broad spectrum of the species in the Orthohepevirus A genus. Thus, the implementation of an accurate quantitative assay for the detection of all currently described HEV genotypes, as well as for the potential detection of unidentified subtypes, is necessary in both human and animal public health, constituting the basis for a One-Health model for addressing HEV. In this sense, the diagnosis of HEV infection requires precise and sensitive tools. Serological tests may lack sensitivity or yield biased epidemiological data regarding seroprevalence (12, 13). Meanwhile, the molecular diagnosis of HEV infection by PCR exhibits high sensitivity and specificity with respect to such infections and has been crucial from both the epidemiological (blood and organ donor screening) and clinical (management of acute and chronic infection) perspectives.

For these reasons, our study consisted of the development of the concept of this assay, the evaluation of the performance of the pangenotypic reverse transcriptase PCR (RT-PCR) assay, and the clinical validation of the detection and quantification of HEV RNA from the Orthohepevirus A genus.

METHODS AND RESULTS

In silico development of primers and probes.

The identification of sequences to be used as pangenotypic primers was performed using an in silico procedure. The resources used were MEGA software (Version 10.0) and the MAFFT online service (14). The primers and probes used in the study were obtained by aligning all whole-genome sequences of 134 strains from the Orthohepevirus A genus (genotypes 1 to 8) with infective capacity in humans and/or animals that were available in GenBank as of December 2018 (see Table S1 in the supplemental material). Due to the high homology between sequences, the ORF3 region was selected as a target for the development of primers and probes. Specifically, the design of these primers and probe was based on the conserved ORF3 region used in the study carried out by Jothikumar et al. (15). Thus, a 70-bp sequence was selected for the development of primers and probes (nucleotides [nt] 5304 to 5373 in strain 3a; GenBank accession no. AB630970). The primers and probe selected were as follows: forward primer, 5′-RGTRGTTTCTGGGGTGAC-3′; reverse primer, 5′-AKGGRTTGGTTGGRTGA-3′, probe, 5′-FAM-TGAYTCYCARCCCTTCGC-TAMRA-3′. Subsequently, the specificity for HEV of the region amplified by these sequences was verified using BLAST (Basic Local Alignment Search Tool) without obtaining any homologous sequence or a sequence with a similarity level greater than 70%. Figure S2 shows show an alignment of primers and probe with the sequences described by Smith et al. (8).

RNA extraction and real-time RT-PCR.

All measurements for this study were performed by the use of a unique one-step real-time RT-PCR method with a Qiagen One-Step RT-PCR kit (Qiagen, Hilden, Germany) and using a CFX Connect instrument (Bio-Rad, CA, USA). For the RT-PCR assay employed in this study, the 1st World Health Organization (WHO) International Standard for HEV RNA Nucleic Acid Amplification Techniques (NAT)-Based Assays (PEI code 6219/10) and 1st WHO International Reference Panel for HEV Genotypes for NAT-Based Assays (PEI code 8578/13) supplied by the Paul-Ehrlich-Institut were used. The WHO standard HEV strain and all genotypes from the WHO HEV panel were reconstituted with 500 μl of diethyl pyrocarbonate (DEPC)-treated water (Thermo Fisher Scientific, Waltham, MA, USA). For the WHO strains, RNA was extracted from reconstituted virus solution (200 μl) using a commercial QIAamp MinElute virus spin kit (Qiagen, Hilden, Germany) by an automated procedure (QIAcube; Qiagen, Hilden, Germany). The primers and probe, supplied by IDT (Integrated DNA Technologies, IA, USA) and purified by the desalting technique, were reconstituted at a concentration of 20 μM. RT-PCR analyses were performed in final volumes of 50 μl containing 2 μl of deoxynucleoside triphosphates (dNTPs), 2 μl of enzyme mix, 10 μl of 5× buffer, 2.5 μl of each primer (20 μM), 1.12 μl of probe (20 μM), and differing amounts of RNase-free water depending on the amount of extracted RNA used. For quantification of results of all RT-PCRs carried out in this study, a standard curve with 16 points was designed using 1:2 dilutions of the WHO HEV standard strain.

Determination of optimal thermocycling conditions.

To establish an optimum annealing temperature, a temperature gradient of 8 points between 50°C and 56°C was performed using 45 cycles of amplification and the WHO HEV standard strain (see Fig. S1 in the supplemental material). The optimal annealing temperature obtained in this experiment was 51°C. The definitive thermocycling conditions are shown in Table S2 with the indications provided by the manufacturer of the One-Step kit.

Detection limit setup.

The WHO HEV standard strain was employed to establish the detection limit for 5 μl, 10 μl, 15 μl, 20 μl, and 25 μl of RNA extracted. To this end, 16 replicates were analyzed in decreasing 1:2 dilutions for each amount of RNA extracted. The starting dilution was 1/512, which corresponded to a concentration of 976.5 IU/ml. The detection limit was determined by probit analysis (at the 95% confidence level). Table S3 presents the number of samples tested, as well as the number of positive samples in each of the dilutions. Table 1 presents the detection limit for different concentrations, as determined using probit analysis. For the subsequent development experiments, we chose 25 μl of RNA extracted, which exhibit a detection limit of 21.86 IU/ml (17.38 to 34.30 IU/ml).

TABLE 1.

Limit of detection according to volume of RNA extracted

RNA extracted vol	Limit of detection in IU/ml (95% CI)a
5 μl	100.2 (77.1–161.5)
10 μl	74.1 (60.8–101.2)
15 μl	69.2 (56.7–94)
20 μl	36.39 (29.3–52.2)
25 μl	21.86 (17.38–34.30)

^aCI, confidence interval.

Performance analysis of the WHO HEV panel and emerging genotypes.

To analyze the performance of the pangenotypic primers against different genotypes, the WHO HEV panel was used. This sample panel consists of 11 different members, including genotypes 1a (2 strains), 1e, 2a, 3b, 3c, 3e, 3f, 4c, and 4g, as well as a human isolate related to rabbit HEV (3ra). For each of these genotypes, five replicates were performed. Table 2 shows that the assay was able to detect all genotypes present in the WHO HEV panel and was able to measure the viral load of each of the genotypes. An in silico-designed, single-stranded DNA fragment belonging to the ORF3 region of each genotype, supplied by IDT, was tested to verify the capacity of the pangenotypic assay to detect emerging viral genotypes (genotypes 5, 6, 7, and 8) not included in the HEV genotype panel. A 339-bp fragment (nt 5183 to 5522; GenBank accession no. AB573435) was designed for genotype 5, a 339-bp fragment (nt 5186 to 5525; accession no. AB602441) for genotype 6, a 342-bp fragment (nt 5161 to 5503; accession no. KJ496143) for genotype 7, and a 342-bp fragment (nt 5154 to 5496; accession no. KX387867) for genotype 8. The fragments were reconstituted at the concentration recommended by the manufacturer (10 ng/μl). As a control, a fragment of strain 1a (8567/13) from the HEV genotype panel was used. In screening these genotypes, the pangenotypic assay was able to detect genotypes 5, 6, 7, and 8 using these synthetic strains.

TABLE 2.

Viral load obtained by the pangenotypic procedure using the WHO HEV panel^a

Panel member	Genotype (source)	Cq mean (SD)	Mean viral load (log10 IU/ml)
			Pangenotypic	WHOb
8567/13	1a	38.50 (±0.38)	2.64	2.57
8568/13s	1a (stool)	31.01 (±0.86)	5.21	4.28
8569/13	1e	35.69 (±1.07)	3.61	3.25
8570/13	3b	32.74 (±0.73)	4.62	4.24
8571/13	3c	35.88 (±0.90)	3.54	3.32
8572/13	3e	35.80 (±0.41)	3.57	3.47
8573/13	3f	34.29 (±0.68)	4.09	3.69
8574/13s	3ra (stool)	29.69 (±0.37)	5.67	4.73
8575/13	4c	32.70 (±0.84)	4.63	3.96
8576/13	4g	34.34 (±0.59)	4.07	3.68
8577/13s	2a (stool)	28.08 (±1.07)	6.22	5.22

^aC_q, quantification cycle; SD, standard deviation.

^bMean data are based upon results obtained from 17 quantitative assays (16).

Performance comparison between pangenotypic and commercial kits: HEV panel.

The performance of the pangenotypic assay was compared with those of the commercial kits ampliCube 2.0 HEV (Mikrogen Diagnostik, Neuried, Germany), referred to here as kit no. 1, and RealStar HEV RT-PCR kit 2.0 (Altona Diagnostic, Hamburg, Germany), referred to here as kit no. 2. The first comparison consisted of the detection at concentrations below 1,000 IU/ml of genotypes 1e, 3c, 3f, and 3ra, which are included in the HEV genotype panel. The mean viral load obtained in the WHO report (16) was employed as a reference for the dilutions. For each of the dilutions, 5 replicates were made following the protocol described above for the pangenotypic assay and following the manufacturer's instructions for the two commercial kits. In the comparison with the two commercial kits, the pangenotypic assay showed better performance in the 4 genotypes tested at low concentrations. The percentage and number of positive replicates detected are shown in Table 3. Except for genotype 3c, the performance of both commercial kits did not exceed 25% of positive replicates at low concentrations.

TABLE 3.

Performance of the three RT-PCR assays against genotypes 1e, 3c, 3f, and 3ra^a

Genotype and dilutionb	No. of replicates detected (Cq mean ± SD) or total no. positive/no. tested (%)c
	Pangenotypic	ampliCube HEV 2.0 quant	RealStar HEV RT-PCR kit 2.0
1e
1/20 (88.9 IU/ml)	4/5 (37.6 ± 0.3)	1/5 (37.2)	1/5 (42.3)
1/80 (22.2 IU/ml)	4/5 (37.7 ± 1.3)	0/5 (NC)	2/5 (39.1 ± 0.4)
1/160 (11.1 IU/ml)	1/5 (38)	0/5 (NC)	0/5 (NC)
1/400 (4.4 IU/ml)	2/5 (39.7 ± 4.2)	0/5 (NC)	0/5 (NC)
Total no. positive/no. tested (%)	11/20 (55)	1/20 (5)	3/20 (15)
3c
1/5 (625 IU/ml)	5/5 (37.3 ± 0.5)	5/5 (37.9 ± 0.5)	5/5 (37.9 ± 0.4)
1/20 (125 IU/ml)	5/5 (37.7 ± 3.5)	4/5 (38.8 ± 0.4)	5/5 (39.8 ± 2.8)
1/40 (62 IU/ml)	5/5 (39.2 ± 1.1)	3/5 (39.9 ± 3.1)	4/5 (37.8 ± 0.8)
1/100 (25 IU/ml)	5/5 (40.6 ± 1.1)	2/5 (38.7 ± 0.8)	4/5 (39.1 ± 0.6)
Total no. positive/no. tested (%)	20/20 (100)	14/20 (70)	18/20 (90)
3f
1/60 (115 IU/ml)	4/5 (37.9 ± 3.4)	1/5 (37.5 ± 3.2)	0/5 (NC)
1/240 (28 IU/ml)	3/5 (39.4 ± 0.9)	3/5 (37.9 ± 0.2)	0/5 (NC)
1/480 (14.41 IU/ml)	0/5 (NC)	0/5 (NC)	0/5 (NC)
1/1,200 (5.76 IU/ml)	2/5 (40.7 ± 1.8)	0/5 (NC)	1/5 (40.3 ± 1.4)
Total no. positive/no. tested (%)	9/20 (45)	4/20 (20)	1/20 (5)
3ra
1/1,250 (76 IU/ml)	5/5 (38.3 ± 4.1)	2/5 (37.2 ± 0.1)	5/5 (39.5 ± 2.8)
1/5,000 (19 IU/ml)	3/5 (39.3 ± 0.8)	1/5 (39)	0/5 (NC)
1/10,000 (9.5 IU/ml)	1/5 (38.7)	0/5 (NC)	0/5 (NC)
1/25,000 (3.81 IU/ml)	1/5 (43.1)	0/5 (NC)	0/5 (NC)
Total no. positive/no. tested (%)	10/20 (50)	3/20 (15)	5/20 (25)

^aSD, standard deviation; NC, not calculable.

^bAll dilutions were based on the viral load estimated from the analyses performed by the WHO.

^cData represent the number of replicates detected (C_q mean ± SD) unless otherwise indicated.

Performance comparison between pangenotypic and commercial kits: testing in a real setting.

This comparison consisted of testing clinical specimens of patients with acute hepatitis in care at the Reina Sofía Hospital in Córdoba between 2017 and 2019. Acute hepatitis was defined as acute illness with (i) onset of such symptoms as nausea, anorexia, fever, or malaise and (ii) elevated serum aminotransferase levels. Sera were tested by enzyme immunoassay for anti-hepatitis A IgM, anti-hepatitis B core, and hepatitis B surface antibody; total hepatitis C antibodies; total hepatitis D antibodies; anti-Epstein Barr IgM; and anti-cytomegalovirus IgM. In addition, sera were tested by means of PCR for the detection of hepatitis B virus DNA, hepatitis C virus RNA, cytomegalovirus RNA, and Epstein-Barr DNA. All samples were screened for HEV infection by enzyme immunoassay for anti-hepatitis E IgM (Wantai; Beijing Wantai Biological Pharmacy Enterprise Ltd., Beijing, China) and HEV RNA in triplicate using the pangenotypic assay, kit no. 1, and kit no. 2. RNA extraction from human serum samples was performed following the same procedure as that described above in the “RNA extraction and real-time RT-PCR” section. Samples with detectable HEV used for any assay were genotyped by targeting the ORF2 region using primers HEV_5920S (5′-CAAGGHTGGCGYTCKGTTGAGAC-3′) and HEV_6425A (5′-CAAGGHTGGCGYTCKGTTGAGAC-3′) in the first round and primers HEV_5930S (5′-GYTCKGTTGAGACCWCBGGBGT-3′) and HEV_6334A (5′-TTMACWGTRGCTCGCCATTGGC-3′) in the second round. The second amplification product of 467 bp was sequenced using a BigDye Terminator cycle sequencing ready reaction kit on an ABI Prism 3100 genetic analyzer (Applied Biosystems, Foster City, CA, USA). SnapGene software (Version 3.1; GSL Biotech) was used for sequence analysis. The consensus sequence was obtained using SeqMan NGen software Version 12.0 (DNASTAR. Madison, WI). Subtype assignment and phylogenetic analyses were performed using the HEVnet genotyping tool (https://www.rivm.nl/mpf/typingtool/hev/) (17), and the results were confirmed by BLAST.

Fifty-four patients were included in the validation procedure (Fig. 1). The etiology of acute hepatitis is shown in Table S4. Of the 54 patients, 33 (61.1%) were found to be HEV RNA negative by the pangenotypic assay. None of these patients gave positive results by any of the commercial kits. Of these 33 patients, 7 were diagnosed with HEV infection by anti-HEV IgM antibody testing. These 7 patients gave negative results for HEV RNA by both commercial kits.

FIG 1.

Flow chart of validation procedure.

Twenty-one patient samples (38.9%) were determined to be HEV RNA positive using the pangenotypic assay, 20 of which could be sequenced. Nineteen strains were consistent with genotype 3f, and 1 strain was genotype 3. Another RT-PCR determination was repeated in duplicate for the sample that could not be sequenced, and positivity was confirmed. Of these 21 patients, 15 were determined to be HEV RNA positive by kit no. 1, and 17 were determined to be HEV RNA positive by kit no. 2 (Table 4). Four of the 6 patients who were negative with kit no. 1 were also negative with kit no. 2.

TABLE 4.

Performance details of the three RT-PCR assays for clinical specimens with detectable HEV RNA

Patient	HEV RNA (log10 IU/ml)			HEV genotype	Accession no.	IgM
	Pangenotypic	ampliCube 2.0	RealStar HEV 2.0
01	4.11	Not detected	Not detected	3f	MT776551	Negative
02	5.22	Not detected	Not detected	3f	MN628557	Negative
03	5.60	3.82	3.53	3f	MT776550	Positive
04	5.20	2.00	1.54	3f	MN628562	Positive
05	5.05	Not detected	Not detected	NAa	NA	Negative
06	5.00	4.18	4.61	3f	MN628561	Positive
07	4.97	3.17	4.39	3f	MN628565	Positive
08	5.62	3.81	4.20	3f	MN628563	Positive
09	4.39	Not detected	2.00	3f	MN914126	Negative
10	4.02	3.80	4.35	3f	MN914127	Negative
11	5.34	3.23	4.09	3f	MN537838	Negative
12	4.73	Not detected	Not detected	3	MW143072b	Negative
13	4.30	5.16	1.95	3f	MT776552	Positive
14	5.77	5.11	5.82	3f	MT776553	Positive
15	5.59	3.29	3.96	3f	MN628564	Positive
16	7.22	6.94	6.81	3f	MN628566	Positive
17	6.70	6.24	6.56	3f	MN628567	Positive
18	5.00	4.76	5.24	3f	MN688224	Positive
19	3.76	Not detected	5.90	3f	MT776554	Positive
20	3.41	3.92	4.78	3f	MT250082	Positive
21	6.41	4.64	5.46	3f	MT854329	Positive

^aNA, not available.

^bSample genotyped by means of protocol described by Inoue et al. (33).

DISCUSSION

HEV is an emerging disease with high phylogenetic diversity, especially genotype 3, which has a global distribution and a large number of recently described subtypes (9, 18, 19). On the other hand, new species are gradually emerging that may serve as reservoirs of new genotypes and that can be potentially zoonotic (4, 6). Therefore, we designed a pangenotypic detection assay that could minimize the number of false negatives and thereby substantially increase the number of isolated cases of the virus. This assay could clinically facilitate the management of patients infected by HEV, both in acute and chronic infections. On the other hand, this assay could increase the number of cases and therefore may provide new strains not previously described.

To the best of our knowledge, there are no studies comparing the performances of HEV diagnostic assays in patients with acute hepatitis. In this study, our assay and two other commercial assays were compared by two analyses. In the first analysis, it was observed that the pangenotypic assay exhibited better performance than commercial kits at very low concentrations of four different genotypes. It should be noted that there was a greater difference in performance seen in genotypes 1e, 3f, and 3ra. Some studies have shown performance problems in detecting genotype 3f at low concentrations using both in-house and commercial assays. In this vein, Baylis et al., using an in-house RT-PCR procedure, showed that at concentrations below 12,000 IU/ml, the efficiency for the detection of HEV 3f was considerably reduced (20). Similarly, a study conducted by Abravanel et al. observed that RealStar HEV RT-PCR kit 1.0 had poor performance below a concentration of 100 IU/ml (21). Using the assay proposed in this paper, genotype 3f was detected with a viral load greater than the mean viral load estimated by the WHO (16). On the other hand, to the best of our knowledge, there are no other studies that have conducted a performance analysis of the assays for genotypes 1e and 3ra at low virus concentrations. In the second analysis, which consisted of a clinical validation procedure, despite the relatively small sampling performed, the two commercial assays were unable to detect the genotype for between four (kit no. 2) and six (kit no. 1) samples, four of which belonged to the 3f genotype. Furthermore, this lower level of performance observed with this genotype was independent of the viral load of the sample (5 of 6 samples had a viral load greater than 10,000 IU/ml). Taken together, the data show that both commercial kits exhibited lower performance for genotype 3f, because both failed to detect this genotype at concentrations below 100 IU/ml and because of the lack of detection of two or four (depending on the assay) clinical species infected by this genotype. This result is notable, since genotype 3f is one of the most prevalent genotypes in Europe and shows a worldwide distribution (9). Therefore, the use of our assay could minimize the number of false negatives.

HEV is the main etiological agent of acute viral hepatitis (1). Suboptimal performance in the detection of HEV can lead to an erroneous diagnosis of acute HEV infection in patients with hepatitis of unknown etiology or to a nonnegligible percentage of false negatives in the detection of the virus in blood donations in countries where the screening is implemented (22, 23). From an epidemiological point of view, these false-negative results could lead to underestimation of the percentage of infections. In this sense, there is evidence of discrepancies between positive HEV-Ag and negative HEV RNA; therefore, the possibility of biased HEV RNA determinations on the basis of false negatives cannot be ruled out. This bias has been observed in studies of human cohorts (24–29). Some of those studies used broad-spectrum primers included in the ORF1 or ORF2 region. In this work, we designed an assay that focuses on the ORF3 region, since it has been identified as the ideal area for the development of molecular diagnostic assays due to its low profile of mutations between viral genotypes/subtypes (9, 30).

This study had several limitations. First, genotypes not included in the panel provided by the WHO (i.e., genotypes 5, 6, 7, and 8) were tested with synthetic strains, which means that the performance of this assay has not been tested on real samples for these genotypes. Second, the study was evaluated only with the Qiagen One-Step PCR kit; therefore, the performance demonstrated in this study may not be maintained if other PCR kits are used. Third, the assay has been designed to detect only species of Orthohepevirus A, which means that other strains of other species with observed zoonotic potential that was demonstrated after the design of this study, such as those belonging to the Orthohepevirus C (31, 32), have not been considered. Finally, this assay would need to be validated in other real-world settings with a wider distribution of genotypes and subtypes.

In conclusion, our assay was able to detect all described genotypes of the Orthohepevirus A genus and showed stronger performance than the two commercial kits. Our assay may represent a significant improvement in the molecular diagnosis of infection with important clinical and epidemiological implications.

Data availability.

Accession numbers for sequence data determined in this work are listed in Table 4.