A relatively simple, rapid, economic and practical genotyping system with high sensitivity and specificity was established based on the methods developed by Naito et al., (2001)  and Sugauchi et al., (2003) . Compared with the classical type-specific PCR method established by Naito et al., (2001) , low-cost common Taq polymerase was used in our method. The cost of the common Taq polymerase is roughly 40% of AmpliTaq Gold DNA polymerase used in Naito’s method. When common Taq polymerase was employed instead of AmpliTaq Gold DNA polymerase in Naito’s method, the specificity of the method was reduced (unpublished data). Only a single nucleotide difference between type-specific primers (e.g. antisense primer BB1R) and reference sequences of different genotypes may allow extension of mismatched primer-template. It was reported that, in some instances, a single terminal 3'-mismatched base did allow PCR amplification to proceed [31, 32], which reduced the specificity of type-specific PCR method. In our method, primers specific for each HBV genotype or subgenotype differed by ≥3 nucleotides (≥2 nucleotides at the 3’ end of the sequences). Double mismatches at 3’ end decreased the extension of mismatched primer-template. The sequencing results of 130 serum samples selected randomly from the 639 samples, confirmed the specificity of our nPCR method.
The sensitivity of the nPCR method was evaluated using serially diluted plasmids (101-109 IU/mL) and 127 clinical samples, in which HBV DNA was quantitated using the Abbott RealTime HBV DNA Assay System. The limit of detection was <101.18 IU/mL (Table 3) and 100% of the samples could be genotyped when the viral load was ≥102.3 IU/mL, which was sufficient for routine genotyping and subgenotyping. The sensitivity of Chen’s method  using single-round PCR for detection of genotype B and C was >104 IU/mL as we repeated this method in our lab. Therefore, we developed a nested PCR method for HBV genotyping and subgenotyping to improve the lower limit of single-round PCR method.
On the other hand, there were two fragments generated after the first round PCR reaction in the nPCR method. One of the fragments, for the specific detection of genotype B, was located mainly in the preC/C region of the HBV genome, and another one, for detecting genotypes C and D, was mainly in the preS/S region (see Supplementary data 1). This design increased the sensitivity and specificity of the detection of genotypes B, C and D, while mixed strains could be detected more sensitively as well. In this study, 13.1% (84/642) of the samples assayed by the nPCR method were mixed genotypes. The sequencing analysis of HBV DNA from the 21 cases with mixed HBV infection confirmed the results of the nPCR method (random sampling cases were 1/5 of total number of mixed infection samples).
Only three of the 642 samples could not be genotyped or subgenotyped by the nPCR method. Two of the three samples could not be genotyped and another one was identified as genotype B but could not be subgenotyped further by the nPCR method. The sequencing results of the three samples revealed that the first 2 samples were C/D recombinants. It was reported that this kind of C/D hybrid emerged in the preS/S region between genotype D and subgenotype C2, and belonged to genotype C. The recombination fragment of genotype D was located at the nt 10–799 or nt 10–1 499 region . As primers DF and C2F were designed at nt2 853–2 870 and nt147-166 respectively, the nPCR method could not detect this kind of C/D recombinant. Until now, whole genome sequencing was the only effective method to detect C/D recombinants. The third sample, which was identified as subgenotype B2 by sequencing, had a mutation in the region where primer BA was designed, and the mutation was exactly at the nucleotide position matching the 3’ end of the primer, which led to failure of detection by the nPCR method. However, the frequency of this kind of mutation was relatively low.
The present study demonstrated that genotype B, C and D were circulating in China. The results revealed marked geographical differences in the distribution of HBV genotypes. In the north and central China, genotype C was the predominant while the percentage of other genotypes was low. In the south and the east of China, the dominant genotypes were C and B. In the west, the dominant genotype was D. It was obvious that the percentage of mixed genotype infection was high in the south and the west of China. As previously reported, HBV infection rate with multiple genotypes was found to be 31% in Shenzhen  (in the south of China), and 22%- 43.8% in Xinjiang [35, 36] (in the west of China). In China, studies on the subgenotype of HBV were seldom reported. The predominant subgenotype was B2 and C2 in all parts of China.