Pooled fecal samples were collected from 97 pig farms located throughout the Netherlands (20–60 fresh stool specimens per farm) throughout 2005 (20). Individual fecal samples from 50 pigs (5 pigs from each of 10 farms) were collected at a slaughterhouse in the southern part of the Netherlands in November 2006. Fecal samples were collected from 150 muskrats (Ondatra zibethicus) that were caught in 2 regions in the Netherlands, in the northeast (Groningen) and in the southeast (Limburg) (21). Individual fecal samples were collected from 26 wild boars in National Park De Hoge Veluwe in the center region of the Netherlands (Gelderland) in 2005. Two fecal samples obtained from 2 pigs on 1 farm in 2002 were included in this study. Fecal samples were stored as 50% suspensions in 15 g/L of Trypton Soya broth (CM 129; Oxoid, Cambridge, UK) and 10% glycerol at –70°C until testing.

From September 2004 through July 2005, twelve large-volume water samples (230 L–260 L) were collected from the Meuse River monthly by using a conventional filter adsorption–elution method and concentrated by ultrafiltration by using a cellulose-acetate filter (nominal molecular weight limit of 10,000) under high pressure (3 bar) (22). Resulting concentrates were stored at –70°C.

RNA was extracted from 100 μL of 10% pooled fecal suspensions from pigs or 0.1% fecal suspensions from muskrats according to the method Boom et al. (23). The QIAamp viral RNA mini kit (QIAGEN, Venlo, the Netherlands) was used to extract RNA from 140 μL of 10% fecal suspensions from individually sampled slaughterhouse pigs and 1% fecal suspensions from wild boars as described by the manufacturer. RNA was isolated from 12.5 μL of ultrafiltered water concentrate as described (22).

Reverse transcription–PCR (RT-PCR) was performed on 10-fold serially diluted RNA samples with primers HEVORF2con-s1 and HEVORF2con-a1, which were specific for the conserved open reading frame (ORF) 2 region, as described (24,25). An internal control RNA was included in the RT step to monitor for inhibition of the RT-PCR (20). A nested RT-PCR was used to detect ORF1 sequences coding for nonstructural proteins by using primers HEVConsORF1-s1 and HEVConsORF1-a1 (24) for the first round of amplification and primers ConsORF1-s2 and ConsORF1-a2 (26) for the second round as described (25). Negative controls were used, and measures were taken to prevent contamination by complete separation of activities for first- and second-round PCRs in time and space. Animal and environmental samples were analyzed in the Laboratory for Zoonoses and Environmental Microbiology of the National Institute for Public Health and the Environment (RIVM) in Bilthoven. Human sequences were obtained from the Laboratory for Infectious Diseases and Perinatal Screening of RIVM. Use of 2 physically separated laboratories excluded possible cross-contamination of human, animal, and environmental samples. Numbers of viral genomes in samples by RT-PCR (PCR-detectable units [PDUs]) were estimated as most probable numbers as described (27).

HEV RT-PCR products positive by hybridization (ORF2) or of the correct size (ORF1) were subjected to electrophoresis on a 2% agarose gel, excised, purified by using a Qiaquick Gel Extraction Kit (QIAGEN, Hilden, Germany), and sequenced by using the BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Nieuwerkerk a/d IJssel, the Netherlands). HEV RT-PCR products for which no sequences were obtained by direct sequencing were cloned into a pCRII-TOPO vector (Invitrogen, Breda, the Netherlands); >5 clones per product were sequenced. Nucleotide sequences without primer sequences were aligned and clustered by maximum-parsimony analysis in BioNumerics version 4.6 (Applied Maths, Kortrijk, Belgium) and corresponded with analysis of a 242-nt fragment of ORF1 and a 148-nt fragment of ORF2. Human HEV sequences were obtained from human serum samples used for diagnosis of acute viral hepatitis (18,19).

Ten-fold serially diluted RNA samples extracted from animal samples and river water were analyzed for HEV ORF2 sequences by RT-PCR. The highest prevalence of HEV ORF2 RNA in 51 (53%) of 97 pooled fecal samples was detected in pig farms housing pigs 5–27 weeks of age (Table 1). A prevalence of HEV RNA of 14% was detected in feces of pigs ≈6 months of age and ready for slaughter, which corresponded with a prevalence of 30% on pig farms. HEV RNA was detected in 1 (4%) of 26 wild boar fecal samples. Muskrat fecal samples contained many inhibitory factors for RT-PCR. Therefore, samples were tested at several dilutions (10%–0.1% vol/vol feces). HEV RNA was not detected in any of the muskrat fecal samples, although the possibility that some of the samples had false-negative results cannot be excluded because in 24 fecal samples no internal control RNA was detected.

HEV RNA concentrations in positive fecal samples varied from 10³ to 10⁶ PDU/g. In 2 (17%) of 12 river water samples analyzed, HEV RNA was detected at concentrations of 2 PDU/L to 100 PDU/L.

Thirty-six newly generated nucleotide sequences of 148 nt of HEV ORF2 were obtained from pooled fecal samples from pig farms that were previously shown to be positive for HEV ORF1 sequences (20). Nine nucleotide sequences of HEV ORF2 were obtained from individually sampled pigs either on a pig farm or during slaughter. The latter samples represented the infectious status of pigs at the time of consumption. One HEV ORF2 sequence was obtained from an HEV-positive wild boar fecal sample and 2 from HEV-positive water samples (Table 1). These sequences, together with 3 previously published HEV sequences detected in Dutch pig livers (28), were compared with available genotype 3 HEV strains from animals and humans (www.ncbi.nlm.nih.gov) by phylogenetic analysis. This comparison showed that all belonged to HEV genotype 3.

Sequences grouped within 4 previously proposed genotype 3 clusters (3a, 3c, 3e, and 3f) (3) (Figure 1), of which cluster 3c is unique to the Netherlands. Comparison of newly generated HEV ORF2 sequences obtained from pooled fecal pig samples with the previously published HEV ORF1 sequences from the same samples showed identical clustering in 4 HEV genotype 3 clusters (20). HEV genotype 3e sequences were not detected in the Netherlands until 2005. At 1 pig farm sampled in 2002, an unrelated HEV genotype 3 variant was detected. Most HEV sequences obtained from animals and water detected during 2004–2006 grouped within cluster 3f (43%): 18 HEV sequences isolated from pig farms, 1 from pig liver, and 1 from water, with identities of 87.2% to 97.3% (Table 2). The second largest cluster (3c) included 35% of environmental HEV sequences and contained 14 HEV sequences obtained from pig farms, 1 HEV sequence from pig liver, and 1 HEV sequence from wild boar. The identities ranged from 88.5% to 100%. Six sequences obtained from pig farms grouped within clusters 3a, with similarities ranging from 96.0% to 99.3%. Three HEV sequences obtained from pig liver, feces, and water grouped within cluster 3e. Analysis of the geographic distribution of all Dutch environmental HEV sequences by postal code showed that HEV sequences belonging to the 2 major clusters (3c and 3f) were distributed randomly. Too few HEV strains within clusters 3a and 3e were available to be informative for geographic distribution.

Two sequences in cluster 3f obtained from different pig farms were 100% identical within the 148-nt fragment of ORF2. A total of 242 nt of ORF1 were sequenced and showed that the 2 sequences were 98.8% identical within this region of the HEV genome. This result indicated that the strains were similar but not identical.

Dutch environmental and animal HEV sequences obtained during 2004–2006 were compared with HEV ORF2 sequences from a study on Dutch hepatitis E patients without a travel history who received a diagnosis during 2004–2006 (Figure 2) (19). Nucleotide identities between environmental isolates and between human and environmental isolates were similar, as shown in Table 2. These results suggest that variability in HEV strains circulating in Dutch pigs, wild boar, and surface water is similar to variability in strains circulating in humans.

Human sequences grouped within the same 4 HEV genotype 3 clusters, with most of the sequences clustering within Dutch cluster 3c. A relatively high percentage of the sequences in cluster 3c is of human origin (43%). In this cluster, 1 human sequence was 100% identical with a porcine HEV sequence. For confirmation of homology, ORF1 RT-PCRs were performed, but only the pig sample yielded an ORF1 sequence. No direct geographic evidence could be established for an association between the patient and the pig farm; the patient who contracted the HEV infection in 2005 lived in the northern region of the Netherlands, but the pig farm was located in the eastern region. Comparison of postal codes of other Dutch patients without a recent travel history with postal codes of environmental sample locations did not show any geographic clustering.

In cluster 3f, only 1 (5%) of 21 sequences identified during 2004–2006 was of human origin. Comparison with HEV sequences derived from human serum samples showed 2 close homologies, 1 between sequences from a hepatitis E patient in 2005 and a pig (96.6%) and 1 between sequences from a patient in 2006 and a surface water sample in 2005 (97.3%).