Early Release – Human Circovirus in Patients with Hepatitis, Hong Kong – Volume 30, Number 12—December 2024 – Emerging Infectious Diseases journal

Study Setting, Patient Specimens, and Animal Samples

This retrospective observational study was conducted at the Queen Mary Hospital (QMH) Department of Microbiology in Hong Kong, China. The study was approved by the Institutional Review Board of the University of Hong Kong, China/Hospital Authority Hong Kong, China West Cluster (UW 23-330). The QMH microbiology laboratory provides viral hepatitis diagnostic services to acute hospitals and transplant services across Hong Kong. We retrieved plasma or serum samples from patients sent to the laboratory for diagnostic evaluation of viral hepatitis (hepatitis cases) during October 1, 2022–November 15, 2023. We assigned patients to the hepatitis cases group if alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), or bilirubin were elevated above respective reference ranges (Appendix Table 1). We excluded persons with isolated elevated ALP if their gamma-glutamyl transpeptidase levels were within reference ranges (indicating extrahepatic source). Similarly, we excluded cases with isolated unconjugated hyperbilirubinemia. Archived plasma or serum samples from persons with unremarkable LFTs formed a control group; blood samples from those persons were sent to the QMH microbiology laboratory for organ donation evaluation or cytomegalovirus (CMV) reactivation monitoring. We retrieved patient clinical notes, imaging findings, and diagnostic test results from electronic patient records. As described in our previous study (5), we defined patients as immunosuppressed if they met any of the following criteria at time of blood taking: active hematological malignancy, history of solid organ or hematopoietic stem cell transplantation, receiving disease modifying antirheumatic drugs/marrow-suppressive chemotherapy/steroids (>0.5 mg/kg/d prednisolone-equivalent for >1 mo), or living with HIV infection with CD4 T-lymphocyte counts <200 cells/mm³.

HCirV PCR screening was conducted by staff blinded to group assignment. Patients were defined as infected with HCirV if they fulfilled both of the following criteria: HCirV PCR positive in serum or plasma with cycle threshold (Ct) <36 and at least a partial HCirV genome fragment sequenced from serum or plasma. The stringent requirement for sequenced isolates eliminated spurious cases of nonspecific PCR positivity because we did not have positive sample controls for PCR reactions at the outset of this study. Clinical records of HCirV-infected persons were assessed for hepatitis at time of blood taking. If hepatitis was present, clinical, imaging, and diagnostic investigation data were independently assessed by 2 experts blinded to study goals to determine the cause of hepatitis.

Real-Time PCR, Sequencing, and Phylogenetic Analysis

The HCirV real-time PCR targets the rep gene and was designed based on published genomes of HCirV strains from France (Paris strain) and China (YN09/J030 and YN09/347 strains) (GenBank accession nos. ON677309, ON226770, and OP744467) (Appendix Table 2). We determined performance characteristics including limit-of-detection, linearity, and analytical specificity of the assay as previously described (10,11) and obtained partial HCirV genomes by using Sanger sequencing. If the viral load was sufficiently high by Sanger sequencing, we attempted complete genome sequencing (Appendix Table 3). For phylogenetic analysis, we obtained complete genomes of the 65 exemplar circovirus species as listed by the International Committee on Taxonomy of Viruses as of September 9, 2024, from the National Center for Biotechnology Information (NCBI) GenBank database. We also included an additional 3 complete HCirV genomes (GenBank accession nos. OP744467.1, ON226770.2, and OR905605.1) and those of 4 strains of Circovirus porcine 3 (PCV3; accession nos. MK095623, MK496297, MH277112, and MK496292). We downloaded 3 complete Cyclovirus genomes (accession nos. KT732787, KF031466, GQ404857) for use as an outgroup to root the phylogeny.

We performed multiple sequence alignment of the complete genomes using MAFFT version 7.526 software (–genfapair–maxiterate 1000—adjustdirectionaccurately) (12) and added partial genome sequences to the alignment with MAFFT (–localpair–addfragments–adjustdirectionaccurately) (13). We removed sites with >20% gaps with TrimAl version 1.4.22 (-gt 0.2) (14) and visualized the resulting alignment in JalView (15). We inferred a maximum-likelihood phylogenetic tree using IQ-TREE, using the best-fit substitution model automatically selected by ModelFinder and MixtureFinder (16,17; H. Ren et al., unpub. data, https://www.biorxiv.org/content/10.1101/2024.03.20.586035v2). We assessed branch support using the Shimodaira-Hasegawa–like approximate likelihood ratio test with 10,000 bootstrap replicates. We visualized and annotated the final phylogenetic tree using TreeViewer (18).

HCirV Peptide Expression, Characterization, and Immunoblots

The capsid (Cap) protein encoding the outer capsid of HCirV was expressed in Escherichia coli. Proteins were assessed using transmission electron microscopy and sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). Cap protein was used to raise hyperimmune polyvalent serum samples in mice and IgM/IgG immunoblots of human serum samples as previously described (19) (Appendix).

Cell-Binding Assay

We seeded human hepatoma cells (PLC/PRF/5, ATCC CRL-8024; Hep-G2, ATCC CRL-10741) and A549 lung epithelial cells (CCL-185) on glass coverslips in a 24-well plate. When confluence reached 80%, we washed cells with prechilled 1× phosphate-buffered saline (PBS). We incubated Purified Cap protein (10 µg/well) VLPs and control proteins with PLC/PRF/5, Hep-G2, and A549 cells for 1 hour at 4°C. After discarding supernatant, we extensively washed cells with cold PBS 3–4 times to remove unbound VLPs and fixed with 4% paraformaldehyde, then incubated with mouse anti-Cap protein polyvalent immune serum (1:200) diluted in 1% BSA. We used Alexa Fluor 488-conjugated anti-mouse IgG (ThermoFisher Scientific, https://www.thermofisher.com) as secondary antibody (1:2000). We counterstained nuclei at AntiFade mounting medium with DAPI (VectorLabs, https://www.vectorlabs.com). We examined slides using a fluorescence microscope and used ImageJ software (https://imagej.net/ij) to measure positive cells.

Testing for Other Causes of Viral Hepatitis

We screened samples positive for HCirV DNA for other viral causes of hepatitis including hepatitis A virus, hepatitis B virus, HCV, hepatitis E virus (HEV), adenovirus, enterovirus, human herpesvirus (HHV) 6, HHV-7, CMV, HIV, and Epstein-Barr virus (EBV) using commercial or in-house–developed serologic and molecular assays (Appendix Table 4). Protocols for molecular assays have been published previously (10,11,20–23). We measured Torque teno virus (TTV) DNA using a published protocol (24).

Statistical Analysis

We compared proportions of HCirV-positive cases and controls using χ² or Fisher exact test, depending on the number of persons. To differentiate cholestatic from hepatocellular liver injury, we calculated the R factor as follows: (ALT value/ALT upper limit of normal) ÷ (ALP value/ALP upper limit of normal). R factor <2 corresponds to cholestatic injury, >2 to <5 represents a mixed pattern, and >5 corresponds to hepatocellular injury (25).

Cases and Controls

We obtained serum or plasma samples from 278 hepatitis patients and 184 controls (Table 1). Immunosuppressed persons were well represented in both groups and accounted for more than half of all controls (Table 1). Degree of hepatitis was typically mild; most patients had anicteric hepatitis. A cause of hepatitis was identified by the clinical team in 123 (44.2%) cases; the remaining 155 (55.8%) were of unspecified etiology. For those with identified etiology, viral hepatitis (18.7%) and drug-induced liver injury (17.9%) were the most commonly diagnosed (Appendix Table 5).

Results of HCirV Screening in Human Samples

Figure 1

Figure 1. Results of screening in study of human circovirus in patients with hepatitis, Hong Kong. Ct, cycle threshold; HCirV, human circovirus.

The limit-of-detection of the Rep gene real-time PCR was 600 copies/mL (Appendix Table 6) with good linearity (Appendix Figure 1). This result was comparable to the limit-of-detection in a recent study (26). We screened samples from hepatitis case-patients and from controls for HCirV infection using the Rep gene assay, followed by sequencing of positive samples. Of 278 hepatitis patients, 10 (3.6%) tested positive by HCirV PCR, but a sequence could not be obtained from 2 of those. Of the 184 controls, 1 (0.5%) tested positive for HCirV DNA, but a sequence could not be obtained (Figure 1). The 3 samples that tested positive by PCR without an identifiable sequence had high Ct values at >33. We did not regard those as HCirV infections for the purpose of this study.

HCirV Case Characteristics

Figure 2

Figure 2. Liver function test kinetics in 8 patients with hepatitis and human circovirus infection, Hong Kong. A) Patient 1; B) patient 2; C) patient 3; D) patient 4; E) patient 5;…

The median age of 8 confirmed HCirV-infected persons (PCR-positive with identifiable sequences) was 47.5 years; 5 were male and 3 female (Table 2). A total of 4 patients had received an organ graft or blood transfusion within 3 months of hepatitis onset, and 4 persons were immunocompromised. None of the patients were known to be infected with HCV or HIV. R factor indicated cholestatic injury in 5 patients, mixed pattern in 2 persons, and hepatocellular injury in 1 person. Consensus on the cause of hepatitis was only available for 1 patient (patient 7); both experts (who were blinded to study objectives and HCirV screening results) agreed with the clinical team on a diagnosis of cholangitis (Appendix Table 7). Investigations conducted for this study identified HEV viremia in 1 patient (patient 3) who had tested negative for HEV IgM (Appendix Table 7). Consensus on the cause of hepatitis in the other 6 patients was lacking. All but 1 patient had acute hepatitis that resolved within a month. HCirV DNA was detectable in plasma or serum of all patients on >1 day during episodes of hepatitis (except for patient 3, who had only 1 blood sample available for testing) (Figure 2). A total of 3 patients died; cause of death was progression of underlying disease or unexplained cardiac arrest rather than hepatitis.

Figure 3

Figure 3. Hematoxylin and eosin stained liver sections of patients in study of human circovirus in patients with hepatitis, Hong Kong. A, B) Cholestasis with bile pigment in zone 3 canaliculi (arrows)…

Of the 4 persons with immunosuppressive conditions, 3 were solid organ transplant recipients and1 was receiving chemotherapy for neuroblastoma (patient 1). Patient 1 had relapsing hepatitis episodes over 3 years before eventually dying of the underlying disease (Figure 2). HCirV was detected consistently in all plasma samples over a period of 18 months. Virus loads ranged from 5.65 × 10³ copies/mL to 1.75 × 10⁶ copies/mL. Patient 4 (kidney graft recipient) and patient 8 (liver graft recipient) also had persistent HCirV viremia over several months despite resolution of the initial hepatitis episode; viremia was highest during hepatitis (10⁴–10⁵ copies/mL) and declined as LFTs normalized in subsequent months. Archived plasma samples from patient 8 obtained at 1-month and 12-months after transplant tested negative for HCirV DNA, indicating an infection in the second year. Patient 5 (liver graft recipient) had acute hepatitis with transient HCirV DNA detection. Although the hepatitis relapsed around 3 months later, HCirV DNA was no longer detectable in the patient’s plasma. The liver histology of patient 5 during the first hepatitis episode was consistent with cholestasis, whereas histology of patient 8 showed neutrophils around biliary ducts and acidophil bodies with no features of acute cellular rejection (Figure 3). Immunohistochemical staining for HCirV capsid protein on a small liver core biopsy from patients 5 and 8 did not show positive signals (Appendix Figure 2).

For some samples with sufficient volume, we also measured concomitant TTV viral load. HCirV was detected in 2 persons with undetectable TTV viremia. In patient 1, who had serial concomitant measures of HCirV and TTV viral loads, viral loads of HCirV tended to fluctuate much less than those of TTV, suggesting that HCirV is not merely an indirect measure of immunosuppression (Appendix Table 8).

The remaining 2 patients (patients 2 and 6) did not have immunosuppressive conditions. Patient 2, a 59-year-old man with good past health, was admitted with fever for 1 week and a maculopapular skin rash. In addition to hepatitis, blood tests revealed a lymphocytosis (3.8 × 10⁹ cells/L) with occasional atypical lymphocytes. Serologic workup for known causes of infectious mononucleosis including EBV, CMV, and HIV yielded negative results (Appendix Table 7). The peripheral blood EBV viral load was detected at <100 IU/mL. Patient 6 was a 75-year-old man admitted for fever. Blood tests showed parenchymal liver dysfunction (Figure 2). Nasopharyngeal swab samples tested positive for SARS-CoV-2 by reverse transcription PCR with a Ct value of 30.4. Chest radiograph did not show lower respiratory tract involvement, and he did not require supplemental oxygen. Ultrasound showed a gallstone but no cholecystitis or cholangitis.

HCirV Genome and Phylogenetic Analysis

Figure 4

Figure 4. Phylogenetic tree of representative circoviruses and human circoviruses from study of human circovirus in patients with hepatitis, Hong Kong. Maximum-likelihood phylogenetic tree was inferred from a multiple sequence alignment of…

Complete HCirV genome was obtained for patient 1 (GenBank accession no. PP968832). Predicted genomic organization of this strain was like other circovirus strains: 2 major open reading frames (ORFs) encoding the Rep gene at nucleotide positions (nt) 247–1134 and capsid protein at nt 1368–2012. The Rep gene is encoded on the sense strand and the Cap gene on the antisense strand. The genome size is 2021 bp, and it is most closely related to the HCirV1 Paris strain (GenBank accession no. ON677309.1), sharing a nucleotide identity of 98.7%. Partial gene sequences were obtained from all the other patients (Appendix Table 9). A phylogenetic tree of complete HCirV genome of patient 1 (HK P1) and partial gene sequences of other isolates showed that strains related to both Paris and Yunnan strains circulate in Hong Kong (Figure 4).

HCirV Immunoassays

Figure 5

Figure 5. IgM and IgG immunoblots of 7 patients for whom sufficient serum was available from study of HCirV in patients with hepatitis, Hong Kong. P indicates positive control, mouse polyclonal serum…

HCirV Cap protein was expressed in E. coli. This peptide showed a band at 25 kDa on SDS-PAGE (Appendix Figure 3, panel A). The peptides spontaneously assembled into VLPs on transmission electron microscopy (Appendix Figure 3, panel B). Serum samples from 7 HCirV-infected persons were available for antibody immunoassays using HCirV VLP-based immunoblots (Figure 5). All patients were negative on IgM immunoblots. Four patients had discernible bands on IgG immunoblots (patients 1, 4, 7, and 8); of those, 3 patients were immunocompromised. As negative controls, we used 9 HCirV PCR-negative donor samples, as well as a serum pool of 900 healthy donors. None of those samples showed bands in the immunoblot, suggesting reasonable specificity (Appendix Figure 3, panel C).

HCirV Cell-Binding Assay

Figure 6

Figure 6. Cell-binding assay showing binding of HCirV VLPs to PLC/PRF/5 cells in study of HCirV in patients with hepatitis. HEV VLPs (bHEV p239) is included as a positive control;…

To further investigate hepatic tropism of HCirV, we tested whether HCirV VLPs could bind to hepatoma cells (PLC/PRF/5). HEV VLPs that are known to bind strongly to PLC/PRF/5 cells served as positive controls. SARS-CoV-2 nucleoprotein served as negative controls. HCirV VLPs were capable of binding to PLC/PRF/5 cells (Figure 6). We found that 27/90 (30%) of cells bound to HCirV VLPs, compared with 30/468 (6.4%) cells that bound to HEV VLPs. We confirmed that HCirV VLPs could bind with Hep-G2, another hepatoma cell line (Appendix Figure 4). Of note, no binding was observed with A549, a lung epithelial cell line (Appendix Figure 4).