Analysis of hepatitis B virus drug-resistant mutant haplotypes by ultra-deep pyrosequencing

Authors


Corresponding author: H.-B. Oh, Department of Laboratory Medicine, University of Ulsan College of Medicine and Asan Medical Centre, 88, Olympic-ro 43-gil, Songpa-gu, Seoul 138-736, Korea
E-mail: hboh@amc.seoul.kr

Abstract

Clin Microbiol Infect 2012; 18: E404–E411

Abstract

Direct sequencing and reverse hybridization are currently the main methods for detecting drug-resistance mutations of hepatitis B virus (HBV). However, these methods do not enable haplotype analysis so they cannot be used to determine whether the mutations are co-located on the same viral genome. This limits the accurate identification of viral mutants that are resistant to drugs with a high genetic barrier. In our current study, ultra-deep pyrosequencing (UDPS) was used to detect HBV drug-resistance mutations in 25 entecavir-treated and five treatment-naive patients. Of the 25 entecavir-treated patients, 18 had experienced virological breakthrough and two exhibited reduced susceptibility to entecavir. The results obtained by UDPS were compared with those of direct sequencing, and the haplotypes of the drug-resistant HBV mutants were analysed. The average number of reads per patient covering the region in which drug-resistance mutations are located was 1735 (range 451–4526). UDPS detected additional drug-resistance mutations not detected by direct sequencing in 19 patients (mutation frequency range 1.1–23.8%). Entecavir-resistance mutations were found to be co-located on the same viral genome in all 20 patients displaying virological breakthrough or reduced susceptibility to entecavir. In conclusion, UDPS was not only sensitive and accurate in identifying drug-resistance mutations of HBV but also enabled haplotype analysis of the mutants. This method may offer significant advantages in explaining and predicting the responses of patients with HBV to antiviral therapy.

Introduction

Nucleoside/nucleotide analogues such as lamivudine, adefovir and entecavir are widely used for the treatment of chronic hepatitis B [1–3]. However, the development of drug resistance is a common problem during long-term therapy [4]. Antiviral therapy can cause selection of drug-resistant mutants of the hepatitis B virus (HBV), leading to treatment failure and disease progression [4]. Hence, the accurate identification of drug-resistant HBV is vital for the management of patients with chronic hepatitis B.

Amino acid substitutions conferring drug resistance are located in the reverse transcriptase region of the HBV polymerase gene [1–4]. Resistance to entecavir appears to occur through a two-hit mechanism, with initial selection of a lamivudine-resistance mutation followed by substitutions at additional residues. Therefore, when identifying HBV mutants resistant to entecavir, it is important to perform haplotype analysis to determine whether mutations at multiple residues are co-located on the same viral genome [5].

At present, the most commonly used methods for detecting HBV drug-resistance mutations are direct sequencing and reverse hybridization [4]. Direct sequencing can identify all existing and emerging mutations but only those present in >20% of the circulating virus population. Reverse hybridization can detect mutations with frequencies of 5%, but only a limited repertoire of well-established mutations. Furthermore, neither of these methods enables haplotype analysis, hence precluding the determination of whether multiple mutations are co-located on the same viral genome.

Recently, several next-generation sequencing technologies have become available that generate more data than the conventional direct sequencing method [6,7]. The Roche GS FLX platform achieves this throughput by using emulsion PCR and simultaneous pyrosequencing on a picolitre scale [8]. The use of this technology to sequence a given nucleotide multiple times is referred to as ultra-deep pyrosequencing (UDPS) [9,10]. The read length of the GS FLX is 400–500 bp, making it suitable for analysing haplotypes of entecavir-resistant HBV mutants in which mutations are dispersed over a 250-bp range.

In our present study, UDPS was used to detect HBV drug-resistant mutations in entecavir-treated and treatment-naive patients. The results were compared with those of direct sequencing and an exclusive program was developed to analyse the haplotypes of the HBV mutants.

Patients and Methods

Study subjects

Samples were obtained from 23 entecavir-treated patients with HBV mutations associated with resistance to lamivudine, adefovir or entecavir that had been detected by direct sequencing (nos. 1–23). Of these 23 patients, 18 (nos. 1–3, 5, 6, 8, 9, 11–18 and 20–22) had experienced virological breakthrough (>1 log10 IU/mL increase in serum HBV DNA from nadir) and two (nos. 7 and 10) exhibited reduced susceptibility to entecavir (responding at first by >1 log10 IU/mL decrease in serum HBV DNA but stopping to a fixed viral load later). Samples were also obtained from two entecavir-treated patients with no HBV drug-resistance mutation detected by direct sequencing (nos. 24 and 25) and from five treatment-naive patients (nos. 26–30).

Of the 25 entecavir-treated patients, 20 were treated with entecavir after lamivudine failure (no. 2–8, 10–21, and 23), four were switched from lamivudine to entecavir without lamivudine failure (nos. 1, 9, 22 and 24) and one (no. 25) had not been previously treated with lamivudine (Table 1). Virological breakthrough after entecavir treatment occurred in 15 of the 20 patients with a previous lamivudine failure (median time to virological breakthrough of 16 months), and in three of the remaining five patients (median time to virological breakthrough of 28 months).

Table 1.   Antiviral treatment history before ultra-deep pyrosequencing (UDPS) analysis, serum hepatitis B virus (HBV) DNA level at the time of UDPS, time to virological breakthrough after entecavir treatment, and historic resistance data by direct sequencing in the corresponding samples of 30 HBV-infected patients
Patient no. (sample)Antiviral treatment history (months)HBV DNA
(IU/mL)
Time to VB after ETV tx
(months)
Historic resistance data before UDPS by direct sequencingTime of direct sequencing taken (months after ETV tx)
  1. LMV, lamivudine; ADV, adefovir; ETV, entecavir; tx, treatment; VB, virological breakthrough.

  2. aDue to pregnancy.

  3. bDue to non-compliance.

  4. cIrregular intake.

1LMV (3)ETV (22)  6.3 × 10618Not done
2LMV (17)LMV + ADV (33)ETV (35) 6.8 × 10618L180M, T184L, M204V18
3LMV (57)LMV + ADV (42)ETV (22) >1.0 × 10910L180M, T184L, M204V13
4LMV (21)ADV (31)ETV (15)No tx (5)a7.2 × 106L180M, M204I15
5LMV (24)ADV (17)ETV (35) 2.2 × 10611L180M, S202G, 204V21
6LMV (57)ADV (33)ETV (23) 8.4 × 10516L180M, S202G, 204I/V16
7LMV (58)ADV (36)ETV (24) 4.7 × 104L180M, M204V13
8LMV (16)LMV + ADV (31)ETV (24) 2.8 × 10719Not done
9LMV (24)No tx (36)bETV (33) 7.4 × 10734Not done
10LMV (32)ADV (37)ADV + ETV (16) 5.4 × 104Not done
11LMV (43)ADV (41)ETV (27) 1.1 × 10715L180M, M204I13
12LMV (26)ADV (28)LMV + ADV (16)ETV (25)3.0 × 10516L180M, S202G, 204V16
13LMV (10)ADV (9)ETV (19) 2.0 × 10616Not done
14LMV (15)ADV (24)ETV (21) 7.2 × 10421Not done
15LMV (12)ADV (21)LMV + ADV (12)ETV (29)1.7 × 10512V173L, L180M, T184L, S202G, M204V15
16LMV (30)LMV + ADV (10)ETV (21) 1.5 × 10711Not done
17LMV (18)ADV (47)ETV (21) 3.3 × 10519Not done
18LMV (42)ETV (16)  3.6 × 10616Not done
19LMV (28)ETV (7)  6.0 × 105Not done
20LMV (9)ETV (29)  1.5 × 10712Not done
21LMV (20)ETV (13)  1.1 × 10713Not done
22LMV (24)No tx (9)bETV (28) 2.4 × 10728Not done
23LMV (9)ETV (4)  5.1 × 104M204I3
24LMV (36)No tx (27)bETV (3)No tx (6)b>1.0 × 109Not done
25ETV (12)c   3.0 × 108Not done
26    >1.0 × 109   
27    5.5 × 107   
28    >1.0 × 109   
29    5.9 × 108   
30    2.0 × 108   

The serum HBV DNA levels at the time of UDPS analysis were quantified using a commercially available real-time PCR assay (Abbott Molecular Inc., Abbott Park, IL, USA) according to the manufacturer’s instructions (Table 1).

PCR and UDPS

The HBV DNA was extracted using the QIAamp MinElute Virus Spin Kit (Qiagen Inc., Valencia, CA, USA) and then amplified with eight primer pairs (Table 2). PCR was performed with 1 μL HBV DNA in a 25-μL reaction mixture containing 1 × PCR buffer (30 mM Tris–HCl; pH 9.0 30 mM salts consisting of K+ and NH4+; 2 mM Mg2+; and enhancer solution), 25 mM dNTPs, 12.5 μM of each primer, and 0.625 U i-Star Taq DNA polymerase (Intron Biotechnology, Sungnam, Korea). Amplification conditions consisted of 95°C for 10 min followed by 35 cycles of denaturation for 30 s at 95°C, annealing for 30 s at 55°C, and extension for 60 s at 72°C, with a final 10-min extension at 72°C.

Table 2.   Sequences and positions of the primers used for PCR amplification
Primer namePrimer sequence (5′→3′)DirectionPosition
  1. Numbering is based on the genotype C strain ayr genome, GenBank accession number NC_003977.

1 - FTCCAACTTGTCCTGGCTATForward352–370
1 - RGCCTTGTAAGTTGGCGAGAReverse1097–1115
2 - FCTACCAGCACGGGACCATForward498–515
2 - RGCCTTGTAAGTTGGCGAGAReverse1097–1115
3 - FAGAATTGTGGGTCTTTTGGForward994–1012
3 - RCAGAGGTGAAGCGAAGTGReverse1582–1599
4 - FGGTGGGAAGTAATTTGGAAForward2113–2131
4 - RCCAGCCTTCCACAGAGTATReverse2752–2770
5 - FGGCCTATATTTTCCTGCTGForward44–62
5 - RACATAGAGGTTCCTTGAGCAReverse535–554
6 - FAACCCTGTTCCGACTACTGForward86–104
6 - RGCCTTGTAAGTTGGCGAGAReverse1097–1115
7 - FTTGTTTAAAGACTGGGAGGAForward1719–1738
7 - RAGTTTCCGGAAGTGTTGATReverse2320–2338
8 - FCTTGGACAAAGGCATTAAACForward2678–2697
8 - RCCTCGAGAAGATTGACGATReverse115–133

The PCR amplicons were purified with AMPure beads (Beckman Coulter Inc., Brea, CA, USA) and subjected to UDPS. Library preparation, emulsion PCR and pyrosequencing were performed using the GS FLX (454 Life Sciences, Roche, Bandford, CT, USA), according to the manufacturer’s instructions. After library preparation, the DNA library was quantified using RiboGreen (Invitrogen, Eugene, OR, USA) and pooled at equimolar concentrations. Following emulsion PCR, the beads were counted on a Multisizer 3 Coulter Counter (Beckman Coulter Inc.). Pyrosequencing was performed with a GS FLX 454 Genome Sequencer on the 1/4 region of a 70 × 75 mm Picotiter Plate.

Direct sequencing

The reverse transcriptase region of the HBV polymerase gene in each sample was analysed by a direct sequencing method. The PCR was performed with an Absolute HBV DR (Entecavir) SBT kit (Biosewoom, Seoul, Korea) according to the manufacturer’s instructions. Direct sequencing was performed on an ABI 3130x1 genetic analyser (Applied Biosystems, Hitachi, Japan).

Analysis of sequence data generated by UDPS

More than 99% of the HBV isolates in Korea are genotype C [11]. Hence, sequences of each read obtained by UDPS were aligned to the NC_003977 (http://www.ncbi.nlm.nih.gov/nuccore/NC_003977) sequence using Amplicon Variant Analyzer software (Roche). However, because this software is designed for single-nucleotide polymorphism analysis, we developed an exclusive program to analyse HBV mutant haplotypes as well as estimate mutation frequencies.

The program first loads the sequences of each read obtained by UDPS, and counts the total number of reads covering the region in which HBV drug-resistance mutations are located. Sequences with insertions and deletions are then considered as follows throughout the correction process: when insertions or deletions are detected at positions of other than homopolymer of more than three bases, the number of reads with a specific insertion or deletion for which a quality value (phred-equivalent values for each UDPS base call) is >20 is counted by each nucleotide position. If these reads constitute >2% of the total number of reads, the insertion or deletion is regarded as authentic and the reads are excluded from calculation of HBV drug-resistance mutation frequency. There were four samples with insertions (nos. 6, 9, 10 and 29) and they constituted approximately 2.3–3.3% of the reads.

The program then converts the nucleotide sequences to the amino acid sequences of the HBV polymerase, which is compared with the NC_003977 sequence to identify mutations. Mutations conferring resistance to antiviral agents included rtM204I/V, with or without rtL180M, for lamivudine resistance; rtA181T/V or rtN236T for adefovir resistance; and a combination of lamivudine-resistance mutations plus rtI169T, rtV173L, rtT184S/A/I/L/F/G, rtS202G/I or rtM250V for entecavir resistance. The program counts the number of reads for each drug-resistant mutant, and the frequency of each drug-resistance mutation is calculated as the number of reads with that mutation relative to the total number of reads covering the rt169–rt250 region.

The cut-off for the mutation frequency was set at 1%. The reliability of drug-resistance mutations detected at frequencies of 1–2% by UDPS was confirmed by counting the number of reads with a quality value >20 at all three nucleotides of the corresponding mutation.

Results

Drug-resistance mutations of HBV detected by direct sequencing and UDPS

The average lengths of the reads obtained by UDPS from samples of 30 HBV-infected patients ranged from 402 to 442 bp, depending on the segment. The average number of reads covering the rt169–rt250 region was 1735 (range 451–4526).

The drug-resistance mutations detected by direct sequencing among the 23 entecavir-treated patients (nos. 1–23) were associated with the antiviral treatment history of the patients. Of these 23 patients, 19 (all but nos. 4, 10, 19 and 23) had entecavir-resistance mutations. No drug-resistance mutation was detected by direct sequencing in any of the five treatment-naive patients (nos. 26–30).

All drug-resistance mutations detected by direct sequencing were detected by UDPS. The UDPS also detected additional drug-resistance mutations not detected by direct sequencing in 19 of the 30 patients (Table 3). A total of 38 drug-resistance mutations were detected only by UDPS, including 37 at frequencies <20% and one (V173L from no. 10) at a frequency >20% (23.8%). Of the seven patients in whom no drug-resistance mutations were detected by direct sequencing (nos. 24–30), two (nos. 25 and 30) had drug-resistance mutations detected only by UDPS. The treatment response of patient no. 25 could not be evaluated because this patient died 2 months later due to septic shock. Patient no. 30 continued entecavir treatment and the serum HBV was undetectable after 15 months without any virological breakthrough.

Table 3.   Drug-resistance mutations of hepatitis B virus detected by direct sequencing and additional drug-resistance mutations detected only by ultra-deep pyrosequencing (UDPS), with their frequencies in each patient
Patient no. (sample)Drug-resistant mutations detected by direct sequencing (%a)Drug-resistant mutations detected only by UDPS (%)
  1. a% Analysed by UDPS.

 1L180M(99.1), S202G(84.4), M204V(97.2)T184A(1.3), M250V(13.8)
 2L180M(97.3), T184L(97.1), M204V(95.8)A181T(1.7), M204I(1.2)
 3L180M(94.8), T184L(94.8), M204V(94.7)
 4L180M(29.3), M204I(96.4)
 5L180M(99.2), T184I(69.9), S202G(97.2), M204V(97.8)T184L(2.0)
 6L180M(97.9), S202G(98.5), M204V(98.0)M204I(1.1)
 7V173L(98.7), L180M(98.9), M204V(99.1)
 8L180M(98.3), T184L(93.2), M204V(98.1)A181T(1.4), T184F(2.7)
 9V173L(99.4), L180M(99.2), M204V(98.1)M250V(3.9)
10L180M(42.6), A181V(15.3), M204V(41.6), N236T(21.6)V173L(23.8), M204I(12.3)
11I169T(68.9), L180M(97.7), T184A(97.3), M204V(97.1)
12L180M(98.5), S202G(96.9), M204V(95.1)M204I(1.5)
13L180M(97.2), S202G(94.3), M204V(93.3)A181T(2.8), T184I(2.6), M204I(1.8)
14L180M(93.7), T184L(38.5), S202G(51.8), M204V(90.6)A181V(2.5), A181T(1.7), M204I(3.5), N236T(2.1)
15V173L(36.9), L180M(97.1), T184L(63.5), S202G(23.4), M204V(95.0)A181T(2.4), T184A(4.2), M204I(3.2)
16L180M(99.3), S202G(98.4), M204V(97.5)T184I(1.7)
17L180M(98.9), T184A(95.8), M204V(99.2)T184S(3.9)
18L180M(99.1), S202G(99.6), M204V(99.1)
19L180M(39.9), M204I(59.0), M204V(39.3)A181T(1.4)
20L180M(98.1), T184L(13.4), S202G(77.0), M204V(98.2)I169T(3.6), T184I(1.9)
21L180M(98.6), S202G(86.7), M204V(97.7)A181T(1.8), T184L(7.1)
22L180M(99.4), S202G(99.8), M204V(97.5)
23M204I(97.7)I169T(1.4), L180M(3.6), A181T(5.0), T184I(1.2), M204V(1.8)
24
25A181T(10.4), M204I(3.2)
26
27
28
29
30A181T(3.9), M204I(6.1)

Reliability of HBV drug-resistance mutations detected at frequencies of 1–2% by UDPS

The UDPS detected 14 drug-resistance mutations at frequencies of 1–2% in 11 patients. Of the 508 reads with these 14 mutations, 452 (89%) had quality values >20 at all three nucleotides of the corresponding mutations. Drug-resistance mutations in these reads were considered to have a high reliability.

Haplotypes of the drug-resistant HBV mutants

Entecavir-resistance mutations were found to be co-located on the same viral genome (entecavir-resistant variants) in the 18 patients (nos. 1–3, 5, 6, 8, 9, 11–18, and 20–22) who had been treated with lamivudine before being started on entecavir and had experienced virological breakthrough (Table 4). Entecavir-resistant variants constituted >75% of the circulating virus population in 17 of these 18 patients (all but patient no. 9). The entecavir-resistant variants observed were L180M-T184A-M204V, L180M-T184L-M204V, L180M-T184S-M204V, L180M-T184F-M204V, L180M-T184G-M204V, L180M-S202G-M204V, L180M-M204V-M250V, L180M-A184I-M202G-M204V, L180M-M202G-M204V-250V, I169T-L180M-T184A-M204V, V173L-L180M-T184A-M204V, V173L-L180M-T184L-M204V, and V173L-L180M-M204V-M250V.

Table 4.   Haplotypes of drug-resistant hepatitis B virus mutants with frequencies >1% with their estimated load in each patient
Patient no. (sample)Clinical statusHaplotypeFrequency (%)Loada (IU/mL)
  1. VB, virological breakthrough.

  2. aEstimated mutant loads calculated as per cent of each mutant detected × sample viral load.

1VB180M-202G-204V78.24.9 × 106
180M-204V-250V10.36.5 × 105
180M-204V2.41.5 × 105
180M-202G-204V-250V2.31.4 × 105
2VB180M-184L-204V92.26.3 × 106
3VB180M-184L-204V91.89.2 × 108
Wild2.82.8 × 107
4 204I65.24.7 × 106
180M-204I28.92.1 × 106
Wild1.51.1 × 105
5VB180M-184I-202G-204V66.51.5 × 106
180M-202G-204V25.65.6 × 105
180M-184L-204V1.43.1 × 104
6VB180M-202G-204V94.78.0 × 105
204I1.19.2 × 103
7Reduced susceptibility173L-180M-204V95.14.5 × 104
8VB180M-184L-204V90.22.5 × 107
180M-184F-204V2.98.1 × 105
180M-204V1.13.1 × 105
9VB173L-180M-204V92.36.8 × 107
173L-180M-204V-250V3.82.8 × 106
10Reduced susceptibility173L-180M-204V23.11.2 × 104
180M-204V20.01.1 × 104
236T17.99.7 × 103
181V14.17.6 × 103
204I10.35.6 × 103
Wild2.01.1 × 103
180M-204I1.26.5 × 102
11VB169T-180M-184A-204V67.57.4 × 106
180M-184A-204V27.33.0 × 106
12VB180M-202G-204V90.72.7 × 105
180M-202G-204V-250V1.13.3 × 103
13VB180M-202G-204V88.81.8 × 106
180M-184I-202G-204V1.73.4 × 104
14VB180M-202G-204V51.53.7 × 104
180M-184L-204V34.22.5 × 104
181V-236T2.11.5 × 103
180M-204V2.11.5 × 103
204I2.11.5 × 103
15VB180M-184L-204V31.75.4 × 104
173L-180M-184L-204V22.43.8 × 104
180M-184G-204V17.63.0 × 104
173L-180M-204V4.27.1 × 103
173L-180M-184A-204V3.45.8 × 103
180M-204V2.84.8 × 103
16VB180M-202G-204V95.11.4 × 107
180M-184I-202G-204V1.32.0 × 105
180M-204V1.32.0 × 105
17VB180M-184A-204V92.53.1 × 105
180M-184S-204V3.91.3 × 104
18VB180M-202G-204V96.73.5 × 106
19 204I69.14.1 × 105
180M-204V25.51.5 × 105
20VB180M-202G-204V56.68.5 × 106
180M-184L-204V12.01.8 × 106
180M-202G-204V10.81.6 × 106
180M-184I-202G-204V1.52.3 × 105
21VB180M-202G-204V79.78.8 × 106
180M-184L-204V6.57.1 × 105
180M-204V4.34.7 × 105
22VB180M-202G-204V94.12.3 × 107
180M-184I-202G-204V1.12.6 × 105
23 204I87.94.5 × 104
181T-204I4.22.1 × 103
180M-204I3.41.7 × 103
169T-204I1.47.1 × 102
184I-204I1.26.2 × 102
24 Wild95.69.6 × 108
25 Wild80.12.4 × 108
181T9.32.8 × 107
204I1.75.1 × 106
26Treatment-naiveWild97.29.7 × 108
27Treatment-naiveWild96.85.3 × 107
28Treatment-naiveWild95.79.6 × 108
29Treatment-naiveWild95.85.7 × 108
30Treatment-naiveWild89.91.8 × 108
204I3.16.2 × 106
181T1.22.4 × 106
181T-204I1.02.0 × 106

Partially resistant variant V173L-L180M-M204V was observed in three patients with a history of previous lamivudine treatment before receiving entecavir. They were patient no. 9, who had experienced virological breakthrough, and patients nos. 7 and 10 with reduced susceptibility to entecavir. In patient no. 23, variants of unknown significance were observed. I169T and T184I, which were detected at very low frequencies by UDPS, were co-located with M204I (I169T-M204I and T184I-M204I). This patient continued entecavir treatment and the serum HBV level decreased to 1 log after 6 months. However, the HBV level started to increase afterwards and reached 7 log after 14 months. L180M, M204I, M204V, T184I and T184L were detected by direct sequencing at this point. Mutations conferring resistance to lamivudine or adefovir were detected by UDPS in four patients (nos. 4, 19, 25 and 30) and no drug-resistance mutations were detected by UDPS in five patients (nos. 24, 26–29).

Discussion

Entecavir is widely used for the treatment of chronic hepatitis B. As the development of resistance to entecavir requires multiple mutations in the HBV reverse transcriptase, haplotype analysis is essential for the accurate prediction of resistance to this drug. As previously applied, however, haplotype analysis involves in vitro cloning and sequencing methods that are too labour-intensive for routine laboratory analysis. On the other hand, UDPS not only allows the detection of minor variants but also permits haplotype analysis of drug-resistant HBV mutants and may therefore be a good alternative approach.

To date, only three studies have applied UDPS to the detection of drug-resistance mutations of HBV. Two studies focused on the detection sensitivity and error rate of UDPS by using HBV plasmid clones, and both selected 1% as the cut-off for distinguishing sequencing errors from authentic minor variants [9,10]. We also demonstrated that UDPS is highly sensitive in detecting HBV mutations conferring drug-resistance. Several drug-resistance mutations were detected at very low frequencies of 1–2%. Approximately 90% of the reads with these mutations had quality values >20 at all three nucleotides of the mutations. Hence, a cut-off >1.0% for the detection of drug-resistance mutations by UDPS seems appropriate. The other study found that drug-resistant HBV mutants are commonly present in the liver and serum of treatment-naive patients in various abundances [12].

However, for the correct interpretation of minor variants, G-to-A hypermutation induced by host cytidine deaminase APOBEC3G (apolipoprotein B messenger RNA editing enzyme, catalytic polypeptide-like 3G) should be considered [13]. This causes deamination of cytidine bases to uridine in negative-stranded viral genomes, resulting in G-to-A hypermutation of the virus. G-to-A hypermutated viruses are non-functional because their replication is inhibited. This type of mutation was first reported in HIV [14] and has recently been reported for HBV also [15]. The actual incidence of hypermutated genomes is very low so they are rarely detected by conventional sequencing analyses. Among the drug-resistance mutations of HBV, A181T and M204I are caused by a G-to-A mutation. Hence, the possibility that some proportion of the A181T and M204I detected only by UDPS are actually the result of G-to-A hypermutated viruses cannot be excluded at this stage.

Mutations conferring resistance to entecavir were found to be co-located on the same HBV genome in all 20 patients who had experienced virological breakthrough or who exhibited reduced susceptibility to entecavir. In patient no. 10, the L180M, A181V, M204V and N236T mutations were detected by direct sequencing, but the presence of these mutations could not explain this patient’s reduced susceptibility to entecavir. However, UDPS detected an additional V173L mutation at a frequency of 23.8% and haplotype analysis showed that this mutation was co-located with L180M–M204V, explaining the patient’s reduced susceptibility to entecavir. Patient no. 23 responded to entecavir treatment with HBV DNA level decreasing. Only the M204I mutation was detected by direct sequencing and additional I169T and T184I mutations were detected at very low frequencies by UDPS in this patient. Haplotype analysis showed that each of these mutations was co-located with M204I. To investigate whether the virological breakthrough that eventually occurred in this patient was caused by M204I and the minority variants or by de novo M204V, haplotype analysis has to be performed on the HBV strain at the time of virological breakthrough. Unfortunately, however, no sample at virological breakthrough was available for further UDPS. Based on the results of direct sequencing carried out at the time of virological breakthrough, L180M was dominant over wild-type, M204I and M204V were present in similar proportions, as were T184I and T184L. Considering this finding, the previously detected M204I as well as de novo M204V seem to have played a role in the emergence of virological breakthrough in this patient.

Although the rate of resistance to entecavir is known to be 1.2% after a 5-year treatment in treatment-naive patients [16], it is much higher in patients previously treated with lamivudine because the rate of resistance to lamivudine is quite high [17]. The cumulative incidence of genotypic resistance to entecavir in patients previously treated with lamivudine increases from 6% at year 1 to almost 60% by year 6 [4]. At present in areas endemic for chronic hepatitis B, entecavir is used more frequently in patients with a previous history of lamivudine treatment than in patients without any history of other antiviral treatments. Hence, it is highly probable that lamivudine-resistant HBV mutants are already present in patients commencing entecavir.

In treating lamivudine-resistant patients, the European Association for the Study of the Liver [1] recommends adding tenofovir, or adefovir if tenofovir is not available. The American Association for the Study of Liver Diseases had recommended adding adefovir or tenofovir, or stopping lamivudine and switching to tenofovir or entecavir [18], although switching to entecavir has been eliminated in the updated recommendations [2]. Moreover, a recent multicentre cohort study has indicated that entecavir should not be used in patients resistant to lamivudine [19]. On the other hand, the Asian-Pacific Association for the Study of the Liver still states that add-on adefovir therapy is indicated and switching to entecavir is an option for such cases [3]. In some countries endemic for chronic hepatitis B, this entecavir strategy is still being implemented for patients with lamivudine resistance. As entecavir-resistant HBV variants emerge easily in patients with lamivudine resistance, the Asian-Pacific Association for the Study of the Liver guidelines should be revised to discourage the use of entecavir in these patients and policy should be changed to avoid treatment failures. In Korea, however, medical insurance does not reimburse for combination therapy in patients who are resistant to lamivudine, so these patients are treated with entecavir monotherapy rather than the combination of lamivudine plus adefovir in many instances. For this reason, UDPS would be a useful method of accurately predicting resistance to entecavir in patients being switched to this drug after developing lamivudine resistance.

Another issue in the management of antiviral resistance is add-on versus switching therapy. The concept of superiority of add-on over switching therapy has been adopted for patients with resistance to lamivudine and adefovir [20]. However, a recent study of patients with chronic hepatitis B resistant to lamivudine and adefovir has found that entecavir monotherapy results in a more effective suppression of HBV DNA than lamivudine plus adefovir [21]. This finding suggests that most of the HBV mutations conferring resistance to each drug after sequential monotherapy may be co-located on the same viral genome. Although a previous in vitro clonal analysis has indeed found that most of the HBV drug-resistance mutations are co-located on the same viral genome, this study analysed only 215 clones from 11 samples [5]. UDPS can yield far more information because hundreds to thousands of read sequences per sample can be analysed.

In conclusion, UDPS was highly sensitive and accurate in identifying drug-resistance mutations of HBV compared with the conventional direct sequencing method. UDPS detected mutations at frequencies >1.0% and estimated mutation frequencies by counting an average of 1735 reads. Haplotype analysis of drug-resistant HBV mutants allowed us to determine whether the mutations are co-located on the same viral genome. UDPS identification of viral mutants resistant to drugs may play a significant role in explaining and predicting the responses of patients with chronic hepatitis B to antiviral therapy.

Acknowledgements

This study was presented orally at the 21st European Congress of Clinical Microbiology and Infectious Disease/27th International Congress of Chemotherapy in Milan, Italy on 7 May 2011.

Transparency Declaration

This research was supported by the Pioneer Research Centre Programme through the National Research Foundation of Korea funded by the Ministry of Education, Science and Technology (2012-0001090).

The authors declare no conflicts of interest with respect to this manuscript, other than the fact that Jong-Eun Lee is the CEO of DNA Link, Seoul, Korea.

Ancillary