Detection of Hepatitis B Virus–Host Junction Sequences in Urine of Infected Patients

Abstract

Integrated hepatitis B virus (HBV) DNA, found in more than 85% of HBV‐associated hepatocellular carcinomas (HBV‐HCCs), can play a significant role in HBV‐related liver disease progression. HBV‐host junction sequences (HBV‐JSs), created through integration events, have been used to determine HBV‐HCC clonality. Here, we investigate the feasibility of analyzing HBV integration in a noninvasive urine liquid biopsy. Using an HBV‐targeted next‐generation sequencing (NGS) assay, we first identified HBV‐JSs in eight HBV‐HCC tissues and designed short‐amplicon junction‐specific polymerase chain reaction assays to detect HBV‐JSs in matched urine. We detected and validated tissue‐derived junctions in five of eight matched urine samples. Next, we screened 32 urine samples collected from 25 patients infected with HBV (5 with hepatitis, 10 with cirrhosis, 4 with HCC, and 6 post‐HCC). Encouragingly, all 32 urine samples contained HBV‐JSs detectable by HBV‐targeted NGS. Of the 712 total HBV‐JSs detected in urine, 351 were in gene‐coding regions, 11 of which, including TERT (telomerase reverse transcriptase), had previously been reported as recurrent integration sites in HCC tissue and were found only in the urine patients with cirrhosis or HCC. The integration breakpoints of HBV DNA detected in urine were found predominantly (~70%) at a previously identified integration hotspot, HBV DR1‐2 (down‐regulator of transcription 1‐2). Conclusion: HBV viral–host junction DNA can be detected in urine of patients infected with HBV. This study demonstrates the potential for a noninvasive urine liquid biopsy of integrated HBV DNA to monitor patients infected with HBV for HBV‐associated liver diseases and the efficacy of antiviral therapy.

Abbreviations

bp

base pair

cfDNA

cell‐free DNA

DR1‐2

down‐regulator of transcription 1‐2

dsDNA

double‐stranded DNA

FOCAD

focadhesin

HBV

hepatitis B virus

HBV‐HCC

HBV‐infected hepatocellular carcinoma

HBV‐JS

HBV–host junction sequence

HCC

hepatocellular carcinoma

NGS

next‐generation sequencing

PCR

polymerase chain reaction

PPP2R2C

protein phosphatase 2 regulatory subunit Bgamma

RE

restriction endonuclease

SR

supporting read

TERT

telomerase reverse transcriptase

UMT

unique molecular tag

Hepatitis B virus (HBV) integration can occur early in the course of infection.^{( 1}, ², ^{3 )} Integration events are known to cause insertional mutagenesis, chromosomal instability, activation of oncogenes, and inflammation, and thus can play a critical role in liver disease progression and the development of hepatocellular carcinoma (HCC).^{( 4 )} Integrated viral DNA has been detected in more than 85% of HBV‐associated HCC (HBV‐HCC) cases.^{( 5}, ⁶, ⁷, ^{8 )} Although HBV integration sites have been observed throughout the human genome, integration into known HCC driver genes (such as TERT [telomerase reverse transcriptase], MLL4 [mixed‐lineage leukemia‐like protein], and CCNE1 [cyclin E1]) has been reported repeatedly in patients with HCC.^{( 9 )}

One outcome of HBV‐DNA integration is the creation of a unique HBV–host junction sequence (HBV‐JS), representing a specific molecular signature of the infected cell. HBV‐DNA integrations detected in tissue have been used to determine the clonality of recurrent HCC tumors.^{( 10}, ^{11 )} Recent studies have shown that integrated HBV DNA can be detected in the circulation as well.^{( 12}, ¹³, ¹⁴, ^{15 )} We have previously demonstrated that urine of patients with HCC contains cell‐free DNA (cfDNA) with HCC‐associated modifications,^{( 16}, ¹⁷, ¹⁸, ^{19 )} whereas urine of patients infected with HBV contains fragmented (but not intact) HBV DNA.^{( 20 )} We hypothesize that urine can serve as a noninvasive liquid biopsy for the investigation of HBV integration in patients infected with HBV.

In this proof‐of‐concept study, we assessed the feasibility of detecting HBV‐JSs in urine from patients infected with HBV. We first identified HBV‐JSs in HCC tissue, followed by detection in matched urine by sensitive junction‐specific polymerase chain reaction (PCR) assays. Next, we detected HBV‐JSs in urine from patients infected with HBV with hepatitis, cirrhosis, and HCC, including those undergoing recurrence monitoring (post‐HCC), using an HBV‐targeted next‐generation sequencing (NGS) assay. These findings support the potential of urine as a noninvasive liquid biopsy for integrated HBV DNA, which can be used to monitor the dynamics of HBV integration in the infected liver and to assess HBV treatment efficacy and disease progression.

Patients and Methods

Study Subjects and Specimens

Two studies, A and B, were performed. Study A specimens included archived DNA isolated from eight pairs of matched HBV‐HCC formalin‐fixed paraffin‐embedded tumor tissue and urine samples. The specimens were obtained from patients in stages I‐II undergoing surgery (Table 1) at the National Cheng‐Kung University Hospital, Taiwan, as described previously.^{( 17}, ^{19 )} Study B was performed on urine samples collected from 25 patients infected with HBV, including 5 with hepatitis, 10 with cirrhosis, 4 with HCC, and 6 post‐HCC, at the Thomas Jefferson University Hospital (Table 2). Five patients (B7, B15, B17, B19, and B20) provided more than one urine collection over the course of the study, resulting in a total of 32 urine samples. Additionally, urine samples from 8 HBV‐negative donors (4 male, 4 female), aged 23‐60 years, were collected and served as controls. All samples were collected with written, informed consent and in accordance with the guidelines of the institutional review board.

TABLE 1.

Characteristics of Archived DNA From Matched HBV‐HCC Tissue and Urine for Study A

Patient ID	Age (Years)	Sex (M/F)	Cirrhosis	Tumor stage*	Tumor Size (cm)
A1	57	M	‐	1	4.0
A2	29	M	+	2	7.0
A3	42	M	+	1	2.0
A4	49	M	+	2	3.4
A5	61	M	‐	2	2.3
A6	75	F	‐	1	3.0
A7	47	M	‐	1	4.5
A8	39	F	‐	2	10.0

^*HCC tumors were staged using the tumor‐node‐metastasis staging system.

Abbreviations: F, female; M, male.

TABLE 2.

Clinical Characteristics of Patients in Study B

Diagnosis	Patient ID	Age (Years)	Sex (M/F)	ALT (IU/L)	Serum AFP (ng/mL)	HBsAg+	HBV Serum Viral Load (IU/mL)	HBeAg	HBeAb	Tumor Stage#	Tumor Size (cm)
Hepatitis (n = 5)	B1	59	M	8	1.2	+	<20	‐	‐	N/A
	B2	56	M	58	3.3	+	105	‐	+
	B3	58	M	24	1.5	+	<20	‐	+
	B4	56	M	27	1.0	+	<20	‐	+
	B5	53	M	77	3.9	+	<20	‐	+
Cirrhosis (n = 10)	B6	60	M	40	2.8	+	<20	‐	+	N/A
	B7*	60	M	22	2.8	+	<20	‐	‐
	B8	74	M	30	1.4	+	<20	‐	‐
	B9	55	M	31	1.8	+	<20	‐	+
	B10	61	M	18	1.2	+	<20	‐	‐
	B11	47	M	22	1,8	+	<20	‐	+
	B12	61	M	16	3.1	+	<20	‐	+
	B13	61	M	16	1.1	+	<20	‐	‐
	B14	57	M	50	256.7	+	90	+	‐
	B15*	57	M	83	1.8	+	<20	+	‐
HCC (n = 4)	B16	73	M	29	88.9	+	<20	‐	+	I	1.4
	B17*	78	F	NA	1.2	+	<20	‐	+	I	0.7
	B18†	57	M	38	3045.0	+	50	+	‐	I	4.8
	B19*	70	M	23	756.4	+	<20	‐	+	I	3.0
Post‐HCC (n = 6)	B20*	51	M	28	3.5	+	<20	+	‐	N/A
	B21	66	M	46	1.2	+	<20	‐	‐
	B22	58	M	28	2.5	N/A	<20	‐	‐
	B23	74	M	28	35.3	+	N/A	‐	+
	B24	56	M	NA	47.1	+	<20	‐	+
	B25	56	M	50	5.5	+	<20	‐	+

^*More than one urine collection taken in the course of the study.

^# HCC tumors were staged using the tumor‐node‐metastasis staging system.

^† Recurrence detected.

Abbreviations: +, HBsAg was taken at the time of HBV diagnosis; AFP, alfa‐fetoprotein; ALT, alanine aminotransferase; F, female; HBeAb, hepatitis B e antibody; HBeAg, hepatitis B e antigen; HBsAg, hepatitis B surface antigen; M, male; N/A, not available.

Urine Collection and DNA Isolation

Urine samples were collected and total urine DNA isolated, as previously described.^{( 21 )} For study A, cfDNA was obtained by isolating the <1‐kb fraction from total urine DNA using carboxylated magnetic beads, a method previously developed by our laboratory.^{( 22 )} For study B, the JBS urine cfDNA isolation kit (JBS Science Inc., Doylestown, PA) was used according to the manufacturer’s specifications.

HBV‐JS Detection by HBV‐Targeted NGS Assay and Validation by PCR‐Sanger Sequencing

Tissue DNA was fragmented by sonication and subjected to Illumina (San Diego, CA) NGS library preparation, as previously described.^{( 23 )} Ten cycles of library amplification with Herculase II Fusion polymerase were performed (Agilent Technologies, Santa Clara, CA). Urine cfDNA samples underwent NGS library preparation using the ThruPLEX Tag‐Seq kit (Cat# R400585; Takara Bio, Mountain View, CA), which contains unique molecular tags (UMTs) and uses double‐stranded DNA (dsDNA) as a substrate, according to manufacturer specifications. Of note, the TruPLEX Tag‐Seq kit converted dsDNA at least 300 times more efficiently than single‐stranded DNA (ssDNA) when control templates of 131‐nucleotide (nt) ssDNA and 140–base pair (bp) dsDNA were used to validate the performance of the kit. Library DNA was subjected to an HBV‐targeted NGS assay following the manufacturer’s instructions (JBS Science). The captured library DNA was pooled for NGS analysis on the Illumina MiSeq platform (Penn State Hershey Genomics Sciences Facility at Penn State College of Medicine, Hershey, PA) or on the NovaSeq platform (Psomagen Inc., Rockville, MD).

The ChimericSeq analysis software we developed previously^{( 24 )} was used to detect HBV‐JSs in NGS data sets. For NGS libraries containing UMTs, UMT family consolidation and consensus read generation were performed using the software package Connor.^{( 25 )} Briefly, UMT‐containing reads were aligned to human and HBV reference genomes, and paired reads were grouped into families based on the mapping coordinates and combined UMT sequence. For each UMT family, a consensus read pair was generated. The consensus reads were then processed by ChimericSeq for HBV‐JS identification.

To validate junctions identified by NGS, junction‐specific primers were designed for the host and viral sequences for tissue and urine (Supporting Table S1). Primers used for generating PCR products larger than 100 bp for Sanger sequencing were only used for tissue junction validation, whereas urine primers were designed to target the same junction sequence generating amplicons of 65 bp or less. The junction‐specific PCR was performed with Hotstart Plus Taq Polymerase (Qiagen, Valencia, CA). PCR products were visualized by agarose gel electrophoresis on a 2.2% FlashGel DNA Cassette (Lonza Group, Basel, Switzerland) and subsequently subjected to Sanger sequencing, a nested PCR reaction, or restriction endonuclease (RE) digestion using enzymes obtained from New England Biolabs (Ipswich, MA), according to the manufacturer’s specifications.

To control for artifacts generated by the HBV‐targeted NGS assay, which may be unavoidable with short PCR templates, an artifact library was generated using as input a mix of 30 ng of sonicated (predominantly 170‐200‐bp fragments) normal human genomic DNA and 9,000 copies of HBV plasmid DNA. The mix contained 10‐100 times more copies of HBV DNA than the >500 urine cfDNA samples examined (including previously described samples^{( 20 )}), in which HBV‐DNA content ranged from undetectable to hundreds of copies per milliliter of urine (as determined by quantitative PCR). The artifact library was used to remove HBV‐JSs from each test sample (Supporting Table S2).

Data Availability Statement

The sequence data for the subjects studied in this work who consented to data archiving have been deposited in the NCBI Sequencing Read Archive (accession no. PRJNA720423).

Statistical Analysis

The chi‐square test was performed for categorical variables. For all statistical tests, the level of significance was defined as P < 0.05 to determine whether differences were significant.

Results

Detection of HBV Integration in Urine DNA by Junction‐Specific PCR Assays

To investigate whether HBV‐JSs can be detected in urine, we identified eight pairs of archived DNA samples from matched tissue and urine from patients with HBV‐HCC (study A; Table 1), with each urine DNA isolate extracted from at least 1 mL of urine. We first detected HBV‐JSs in tissue DNA using an HBV‐targeted NGS assay, as described previously. HBV integrations were detected in all eight HCC tumor DNA samples (Table 3). The affected regions included the TERT gene, the most frequently reported integrant‐containing host gene (patient A5).^{( 9 )} The junctions were validated by Sanger sequencing (Supporting Fig. S1) using primers listed in Supporting Table S1.

TABLE 3.

HBV Integrations Identified in Archived Tumor‐Tissue DNA of Study A Patients

Patient ID	HBV–Host Junction Sequences	HBV–Host Junction Breakpoint Nucleotide Position*		Gene
		HBV	Human
A1	ccttgaggcgtacttcaaaa CCCAGACCCAGCT	1712	Chr.10: 31192695	LOC101929352
A2	aggagtgttatggtgtctca TAAATGGTGATATAA	247	Chr.1: 165248017	None
A3	gttgggggagggactag CTCATTAATCATTGT	1755	Chr.11: 92048629	LINE2
A4	gcaccatgcaacttttt TCCGAACCTGTGTACT	1825	Chr.16: 30207023	None
A5	ttgtactaggaggctg CATGGCCGGAAGTCT	1783	Chr.5: 1292170	TERT
A6	ctgtaggcataaattgCCTACAGCAATGTATA	1796	Chr.14: 67004392	GPHN
A7	gatctttgtactaggagg AACATGCCCAAGAA	1780	Chr.5: 165559760	None
A8	agactgtgtgtttaa CTCATCTGTCCAAACCC	1725	Chr.14: 103176895	None

Lowercase sequences represent HBV DNA. Underlined and capitalized sequences represent human DNA. Underlined, lowercase, bold bases represent overlapping human/HBV sequences. “None” indicates that no known gene was present within 100 kb of the integration site.

^*HBV accession #NC_003977.2; human genome GRCh38.

To detect HCC tissue–derived HBV junction DNA in urine, we designed a junction‐specific PCR assay for each junction (Supporting Table S1). The amplicons were designed to be short, as urine cfDNA is mostly fragmented to less than 250 bp.^{( 21 )} Matched tissue and urine DNA samples were subjected to junction‐specific PCR amplification. PCR products of expected size were generated from five of the eight matched urine samples tested. To determine whether the PCR products generated from the urine and tissue in each of the five tissue‐urine pairs were identical, we used either two‐step nested PCR followed by RE digestion of PCR products (patients A5, A6, and A8; Fig. 1A) or Sanger sequencing when no suitable RE site was available in PCR products (patients A1 and A2; Fig. 1B‐D). Four (patients A2, A5, A6, and A8) of the five PCR product pairs contained sequences indistinguishable by these methods, indicating that the HCC tissue–derived HBV‐JS was present in the matched urine.

FIG. 1.

Detection of HBV–host junction sequences in study A, matched archived tissue, and urine DNA. (A) RE digestion approach. Tissue and urine DNA was amplified using two‐step nested junction‐specific PCR. The products generated by HBV and human primers were incubated in the absence (‐) or presence (+) of the respective RE and analyzed by gel electrophoresis. The expected RE‐digested PCR product sizes are patient A5 (25 bp and 20 bp of the 45‐bp PCR product), A6 (24 bp and 26 bp of the 53‐bp PCR product), and A8 (22 bp and 36 bp of the 57‐bp PCR product, but the 22‐bp band was not included in the picture). (B) Sanger sequencing approach. The junction sequence of patient A2 was generated by PCR amplification of tissue and urine DNA using HBV and human primers. (C) Sequence of the HBV–host junction at chromosome 10 (Chr10) in the urine of patient A1. Sanger sequencing of the urine‐derived junction PCR product revealed a 23‐nt insertion compared with the tissue product, as shown by gel electrophoresis. (D) Detection of the 23‐nt inserted HBV–host junction in the corresponding tissue. Using a hybrid Chr5‐Chr10 primer, a specific tissue junction PCR product was obtained and found to be identical by sequencing to the urine‐derived 23‐nt insertion. Tissue DNA and HepG2 DNA served as positive and negative controls, respectively. Abbreviations: MW, molecular weight; Pos, positive; T, tissue; U, urine.

Interestingly, the sizes of the junction PCR products from patient A1 were different between tissue and urine (Fig. 1C). Sanger sequencing analysis identified a 23‐nt insert, indicating a re‐arrangement between chromosomes 5 and 10 in the urine DNA sample. To determine whether the urine‐identified integration site could be found in tumor‐tissue DNA, we designed a primer across the chimeric chromosome 5/10 sequence (Fig. 1D) to specifically amplify the 23‐nt insert–containing HBV‐JS species. Remarkably, the tripartite HBV‐JS was successfully amplified and validated in the matched tissue DNA sample (Fig. 1D ). In total, we were able to detect and validate five of the eight HCC‐tissue‐derived HBV‐JSs in matched urine samples.

Detection and Characterization of HBV‐JSs in the Urine of HBV‐Infected Patients by an HBV‐Targeted NGS Assay

After detecting liver‐derived HBV‐JSs in the urine of patients with HBV‐HCC by junction‐specific PCR assays, we next assessed whether it was feasible to analyze HBV integrations in urine without prior knowledge of junction sequences from tumor tissue and before disease progression to HCC. A total of 32 urine specimens were collected from 25 HBV‐infected individuals, including 5 patients with hepatitis, 10 with cirrhosis, 4 with HCC, and 6 post‐HCC (study B; summarized in Table 2 ). Five patients provided more than one urine specimen in the course of the study. For HBV‐negative controls, we included urine samples from 4 males and 4 females who had no known history of HBV infection. Urine DNA was isolated, enriched for HBV sequences, and subjected to NGS followed by HBV‐JS identification as described previously. We used a threshold of two unique (UMT‐consolidated) supporting reads (SRs) for junction detection for two reasons. First, this cutoff is consistent with other studies of HBV integration in HCC tissue.^{( 8}, ²⁶, ²⁷, ^{28 )} Second, the PCR‐Sanger sequencing validation rate we obtained in a previous study of 16 HBV‐HCC tumor tissue samples was 88%, which is in line with validation rates reported in the literature.^{( 23}, ²⁶, ^{28 )} No HBV‐JSs with two or more SRs were observed in the NGS data obtained from the 8 HBV‐negative controls. In contrast, 31 of 32 HBV‐infected urine samples contained HBV‐JSs at this cutoff.

Of the total of 712 HBV‐JSs identified, 84, 369, 41, and 218 junctions were detected in 4 hepatitis, 12 cirrhosis, 6 HCC, and 9 post‐HCC urine samples, respectively. Table 4 lists the major HBV‐JSs of each individual urine sample, defined as the two most abundant HBV‐JSs supported by at least 3% of total junction reads. As expected, we detected integrations in both gene‐coding and noncoding regions in all disease categories. The latter included repeat regions, such as LINEs, SINEs, and simple repeats,^{( 29 )} which have been previously reported as HBV integration sites.

TABLE 4.

Major HBV Integrations Identified in Urine Samples of Patients Infected With HBV (Study B)

Patient ID	Diagnosis	No. of HBV‐JS Species Identified	Major Junctions (% of Total Junction Reads)*
B1	Hepatitis (n = 5)	48	Chr4 (5.0%), SR (4.3%)
B2†		0	N/A
B3		2	Chr1‐a (50.0%), Chr1‐b (50.0%)
B4		20	CHCHD2P9 (20.7%), HSPE1P8 (17.2%)
B5		10	SINE (19.0%), SEMA6D (14.3%)
B6	Cirrhosis (n = 12)	21	MLLT4 (10.2%), FOCAD (8.2%), PAK3 (8.2%)
B7a‡		21	BAGE2 (6.8%), Chr12 (6.8%)
B7b		5	GCSHP2 (20.0%), PPP2R2C (20.0%)
B8		30	SR (6.3%), LINE (4.8%)
B9		10	SR (28.0%)
B10		40	XRCC3 (9.6%), MST1L (8.0%)
B11		18	TERT (48.8%), Chr16 (7.1%)
B12		53	TERT (30.5%), GCSHP2 (6.9%)
B13		54	SR (5.1%), LINC01481 (4.4%)
B14		13	AL954742.1 (13.8%), Chr5 (10.3%)
B15a‡		46	SINE (13.2%), LINE (3.9%)
B15b		58	SINE (8.0%), EXD3 (6.8%)
B16	HCC (n = 6)	4	KIF14 (25.0%), TET (25.0%)
B17a‡		2	VPS8 (50.0%); Chr9 (50.0%)
B17b		5	SR (20.0%), AC011193.1 (20.0%)
B18		11	SR (36.4%), TERT (9.1%)
B19a‡		17	CTD‐3229J4.1 (50.8%), SR (9.2%)
B19b		2	SR‐a (50.0%); SR‐b (50.0%)
B20a‡	Post‐HCC (n = 9)	16	SR (25.0%), LTR (12.5%)
B20b		63	LMF1‐AS1 (5.6%), hsd17b6 (5.1%)
B20c		14	GCSHP2 (60.8%), SR (10.1%)
B20d		13	SR (21.4%), LACTB2‐AS1 (7.1%)
B21		4	SR (75.0%)
B22		13	TRPM3 (37.0%), ANKRD9 (13.0%)
B23		2	PCDHGC5 (50.0%), HBB (50.0%)
B24		11	Chr12 (20.7%), GALNT6 (10.3%)
B25		82	LINE (12.4%), LTR (5.8%)

^*Defined as up to two most abundant junctions with at least two supporting reads.

^† No one junction sequence meets the major‐junction criteria.

^‡ Multiple collections were obtained from the patient.

Abbreviations: AC011193.1, ENSG00000225582.1; AL954742.1, microRNA ENSG00000251912; ANKRD9, ankyrin repeat domain 9; BAGE2, BAGE family member 2; CHCHD2P, coiled‐coil‐helix‐coiled‐coil‐helix domain containing 2 pseudogene 9; Chr, chromosome location; CTD‐3229J4.1, Long Intergenic Non‐Protein Coding RNA ENSG00000261555; EXD3, exonuclease 3'‐5' domain containing 3; FOCA, focadhesin; GALNT6, polypeptide N‐acetylgalactosaminyltransferase 6; GCSHP2, glycine cleavage system protein H pseudogene 2; HBB, hemoglobin subunit beta; HSD17B6, hydroxysteroid 17‐beta dehydrogenase 6; HSPE1P8, heat shock protein family E (Hsp10) member 1 pseudogene 8; KIF14, kinesin family member 14; LACTB2‐AS1, LACTB2 antisense RNA 1; LINC0148, long Intergenic Non‐Protein Coding RNA 1468; LINE, long interspersed nuclear elements; LMF1‐AS1, LMF1 antisense RNA 1; LTR, Long terminal repeat; MLLT4, Myeloid/Lymphoid or Mixed‐Lineage Leukemia; MST1L, macrophage stimulating 1 like; N/A, not available; PAK3, p21 (RAC1) activated kinase 3; PCDHGC5, protocadherin gamma subfamily C, 5; PPP2R2C, protein phosphatase 2 regulatory subunit Bgamma; SEMA6D, semaphorin 6; SINE, short interspersed nuclear elements; SR, simple repeat; TERT, telomerase reverse transcriptase; Translocated To, 4; TRPM3, transient receptor potential cation channel subfamily M member 3; VPS8, VPS8 subunit of CORVET complex; XRCC3, X‐ray repair cross complementing 3.

Among the targets of the 351 unique integrations in gene‐coding regions detected in urine, 11 genes had been previously reported in liver tissue from patients with severe liver diseases, such as cirrhosis and HCC. These genes are AC007880.1, ADAM12 [disintegrin and metalloproteinase domain‐containing protein 12], ARSA [arylsulfatase A], ATF6 [activating transcription factor 6], CLEC2L [C‐type lectin domain family 2 member L], FOCAD [focadhesin], PPP2R2C [protein phosphatase 2 regulatory subunit Bgamma], TNFRSF9 [TNF receptor superfamily member 9], TXNDC16 [thioredoxin domain containing 16], VEGFC [vascular endothelial growth factor C], and TERT. Most interestingly, all of the genes, except for AC007880.1 and ARSA, have been associated with the development of HCC (ADAM12,^{( 30 )} ATF6,^{( 31 )} CLEC2L,^{( 32 )} TERT, and VEGFC ^{( 33 )}) or other cancers (FOCAD,^{( 34 )} PPP2R2C,^{( 35 )} TNFRSF9,^{( 36 )} and TXNDC16 ^{( 37 )}).

Distribution of Integrated HBV Breakpoints Detected in Urine

We next examined the HBV genome breakpoints of the 712 junction sequences detected in urine. The HBV sequences of these junctions were plotted based on their locations in the HBV genome (Fig. 2A), and the SR statistics and HBV breakpoint regions are summarized in Fig. 2B. The number of SRs for a given junction varied widely within each disease type. Of the 712 junctions detected, 70% had two SRs. Approximately 70% of HBV‐DNA breakpoints detected in urine were clustered in the HBV DR1‐2 region, a known integration hotspot^{( 8 )} (Fig. 2B). The second‐most frequent of the previously reported HBV breakpoint regions, PreS,^{( 8 )} was also identified in urine, with higher frequencies found in hepatitis (7.1%), cirrhosis (10.5%), and post‐HCC (7.3%) as compared with HCC (4.9%). However, the association of integrated HBV breakpoint location (DR1‐2, PreS1/2, or other) with disease type was not statistically significant (χ² [6, 712 = 6.3]; P = 0.39) (Fig. 2B).

FIG. 2.

Analysis of HBV integrations in the urine of HBV‐infected patients with various liver diseases. (A) Distribution of HBV junction sequence breakpoints across the HBV genome. The HBV sequences indicated by black lines of unique junctions from each patient are plotted by disease category. Totals of 84, 366, 41, and 221 sequences from hepatitis, cirrhosis, HCC, and post‐HCC samples, respectively, are shown. (B) Comparison of HBV‐DNA breakpoints in HBV DR1‐2, PreS1/S2, and other HBV region and the supporting reads per junction detected among the HBV disease categories. Abbreviation: Avg, average.

Profiling of Detectable HBV Integration Sites in Serially Collected Urine Samples

Urine cfDNA containing HBS‐JSs may reflect the HBV integration profile of the infected liver. For a preliminary assessment of the possibility of using HBS‐JSs for disease monitoring, we collected serial urine samples from 5 patients (B7, B15, B17, B19, and B20) and compared the HBV‐JSs detected in different collections (Table 4). Patients B15 and B19, both with cirrhosis, had two collections 1 week apart and showed consistent integration in SINE and simple repeat regions, respectively, although the integration breakpoints were different. This finding indicates that the detection of these junctions does not reflect clonal expansion, but rather a high incidence of integration in the repeat regions. Similarly, patient B20 (post‐HCC) had four collections at 3‐month intervals, all showing integration in noncoding regions (e.g., simple repeats), although the specific sites varied. Patient B17 (HCC) had two urine collections 2 months apart, with different junction sequences detected. Patient B7, with cirrhosis, had two urine samples collected 7 months apart, with integrations initially detected in a noncoding region followed by integrations in gene‐coding regions. Interestingly, patient B7 developed HCC 4 years after the second collection. This initial comparison suggests that different HBV integrations can be captured in urine through serial monitoring. Collectively, our data demonstrate the feasibility of analyzing HBV integration using urine liquid biopsy.

Discussion

In this proof‐of‐concept study, we provide evidence for noninvasive detection of integrated HBV DNA in urine of patients infected with HBV with liver disease ranging from hepatitis to HCC. The importance of this study is threefold. First, we report the detection of integrated HBV DNA in urine. Second, we includes a comparative analysis of HBV integration in hepatitis, cirrhosis, HCC, and post‐HCC using solely ectopic samples. Third, we provide evidence that urine contains liver‐derived DNA, by taking advantage of the liver tropism of HBV and unique HBV‐JSs as molecular signatures of infected hepatocytes. Taken together, our results demonstrate the potential of urine liquid biopsy for integrated HBV burden and HBV‐related liver disease monitoring.

Our previous study demonstrated that only fragmented HBV DNA was detected in the urine of patients with HBV infection even when the blood viral load was as high as 10⁸ copies/mL.^{( 20 )} Urine cfDNA is mostly monocleosomal and dinucleosomal in size and largely derived from apoptotic cells.^{( 21 )} Therefore, an HBV sequence containing urine cfDNA likely originates from degradation of host‐integrated HBV genomes protected by nucleosomes in apoptotic HBV–infected hepatocytes. Hepatocytes from the HBV‐infected liver have a higher turnover rate than hepatocytes from uninfected normal liver.^{( 38 )} Thus, urine provides a unique biopsy source to noninvasively detect integrated HBV DNA and monitor infected‐cell burden, neither of which is achievable with current clinical HBV tests.

As expected, HBV genome integration breakpoints detected in urine were found to be mostly (>70%) in the HBV DR1‐2 region, consistent with previous HBV‐HCC tissue studies.^{( 8 )} Additionally, 11 HBV‐JSs identified in urine and reported previously in HCC tissue were all found in patients with cirrhosis or HCC. We and others have previously shown that urine contains apoptotic DNA mostly from the circulation,^{( 21 )} which is expected to include DNA from infected hepatocytes undergoing apoptosis throughout the liver. Therefore, urine DNA is likely to reflect the heterogeneity of HBV‐JSs in the liver more accurately compared with a single tumor biopsy or resection. HBV‐JSs have been detected in plasma of patients with HCC in recent studies.^{( 12}, ¹³, ¹⁴, ^{15 )} However, because urine cfDNA consists of mostly 1‐2‐nucleosomal‐sized DNA from apoptotic cells,^{( 21 )} its analysis is less likely to be confounded by the possible presence of any non‐nucleosome‐associated replicating HBV DNA or virions.^{( 20 )} As a result, urine is likely to provide a more sensitive liquid biopsy than blood for assessing integrated HBV‐DNA sequences in infected patients. Although we did not directly compare matched urine and plasma for HBV‐JS detection due to plasma unavailability, the large number of HBV‐JSs we detected in urine samples is encouraging. Studies directly comparing urine with matched plasma will be needed to determine which body fluid is better for detection of HBV‐JSs in cfDNA.

The detection of HBV–host sequences and a chromosome rearrangement in HCC urine echoes our recent progress in developing a noninvasive, urine‐based HCC screening test.^{( 16}, ¹⁹, ^{39 )} Characterization of integrated HBV DNA may provide insight into clonal expansion and turnover of infected hepatocytes, HCC risk stratification, monitoring recurrence following treatment, and treatment efficacy. A urine‐based test allows routine at‐home collection; thus, frequent monitoring can be feasible for patients undergoing antiviral therapy for treatment efficacy evaluation, or for patients at risk of developing primary or recurrent HCC. Characterization of HBV‐JSs in urine can be highly applicable in the development of HBV‐directed T‐cell immunotherapies, as targeting HBV antigens derived from integrated HBV DNA has been shown to have antiviral and antitumor effects.^{( 40}, ^{41 )} Furthermore, detection of urinary integrated HBV sequences may also present a unique opportunity to noninvasively monitor the infected cell burden that may not be captured by current clinical HBV tests. Although more studies are required, the potential of monitoring dynamics of HBV‐JS quantity and quality can provide critical molecular profiling to aid in delineating liver pathogenesis in patients with HBV. Because integrations into known HCC drivers were only found in urine of patients with cirrhosis or HCC in this study cohort, characterization of HBV‐JSs may hold potential for application in assessing HBV‐related liver disease in infected patients. In the present proof‐of‐concept study, we demonstrated that detection of HBV‐JSs in urine is feasible. Further study with a much larger sample size and quantitative measurement in addition to qualitative characterization is needed to evaluate its potential clinical applications.

Supporting information

Supplementary Material

Table S2