Accumulating data have revealed the frequent existence of mosaic HBV genomes, which are generally considered to be the result of recombination between two different genotype strains. In western China, two types of genotype C/D recombinant have been identified 
. In the present study, a large-scale survey on the geographical and ethnic distribution of the HBV C/D recombinant in western China reinforce the results of our previous reports based on smaller samples 
. By analyzing a large cohort of 624 patients with chronic HBV infection from five provinces of western China, we found that the two types of C/D recombinant account for 51.9% of the patients, suggesting that the HBV C/D recombinants prevail in this region.
Our results show that the ‘CD1’ recombinant has a higher prevalence than the ‘CD2’ strain in western China, but that its distribution follows a gradient from east to west, such that the ‘CD1’ strain has a remarkably higher prevalence than the ‘CD2’ in the eastern part of western China. In Mongolia and Inner Mongolia, which are located in the north of Gansu and Ningxia provinces, genotype D is the most prevalent HBV strain, and ‘CD1’ recombinant was observed with a very low prevalence 
. In contrast, we found that ‘CD2’ exists at a slight higher prevalence than ‘CD1’ in Tibet, contradicting a previous prediction that all C/D hybrids were ‘CD2’ recombinant in this region 
. Interestingly, the further west, the more ‘CD2’ recombinant can be observed. Across the Himalayas, genotype D is the most prevalent HBV strain, followed by genotypes A and C in India and Nepal 
. Our results demonstrate that the HBV C/D recombinant is restricted to a specific region in western China, mainly on the Qinghai-Tibet Plateau, which rises about 4000 meters above sea level. Around this region, genotype D is the predominant HBV strain in the west, north and south, whereas genotype C is the most prevalent HBV strain in the east 
The geographic origin of the HBV C/D recombinant remains largely unknown because of the shortage of evidence. In the light of the findings from this study we speculate on the possible geographic origin of HBV C/D recombinant. Of the five provinces in this study, Gansu and Xinjiang were on the Silk Road, a well-traveled transcontinental trade route that linked Europe (mainly genotype D) in the West with China (mainly genotype C) in the East. If the HBV C/D recombinant is the results of co-infection by genotype C and D followed by recombination, Gansu and Xinjiang should have a much higher prevalence of the C/D recombinant than Tibet and Qinghai. Contrary to this, our investigation shows that the HBV C/D recombinant is the predominant HBV strain in Tibet and Qinghai. The Qinghai-Tibet Plateau is a mountainous area with a series of huge mountain ranges, the Qilian, Kunlun, Tanggula, Gangdisi, and Himalayas, that run through from east to west. Because of the particularly high altitude, extreme environmental conditions, and the special religious traditions of this area, it is difficult for people from surrounding areas to enter the region. This physical and cultural isolation also prevents interbreeding and social contact of the Tibetan ethnic population with outside people. This may explain why the C/D recombinants once introduced and flourished there.
It is well known that HBV genotypes correlate well with ethnicity and geography, but the mechanism is still unclear. Study from hepatitis C virus has shown that adaptation to multiple host Human leukocyte antigen (HLA) alleles is an important cause of viral mutation and evolution, and thus divergence 
. Previous investigations have shown the great polymorphism of HLA alleles among the Tibetan, Uygur, and Han populations in western China 
. In particular, although the Tibetan and northern Han Chinese populations shared many similar HLA alleles, they also had distinct frequencies of many HLA alleles and haplotypes. However, some allelic distributions in the Uygur population were more similar to those among Caucasians. These observations are coincidentally in agreement with the present investigation on HBV genotype distributions among these three ethnic populations, with genotype D having the highest prevalence among Uygur, the C/D recombinant being the highest among Tibetans, and genotype C being the highest among Han Chinese. It is still uncertain whether the high prevalence of the C/D recombinant in Tibetan population is the viral adaptation to the specific HLA alleles of Tibetan.
Traditionally, HBV recombination is presumed to be the consequence of genetic material exchange of two HBV strains after one patient has been co-infected with two different HBV genotypes. Different HBV genotype strains co-infection has been widely described 
, but intergenotype recombination events have been rarely reported in co-infected patients 
. A more recent study indicated that recombination in HBV is not as extensive as previously assumed by using an alternative software STRUCTURE 
. For intergenotype recombinant, the mosaic sequences should be derived from the parental strains. When we phylogenetically analyzed the backbone sequences of ‘CD1’ and ‘CD2’ recombinants with the corresponding sequences of local subgenotype C2, we found that ‘CD1’ recombinants were grouped separately, instead of clustering among local C2 isolates within subgenotype C2 (). When we phylogenetically analyzed the mosaic fragments of ‘CD1’ and ‘CD2’ with the corresponding fragment of genotype D, we also observed that the C/D recombinants did not cluster with subtypes D1–D4, but rather formed separate clusters of its own (). Likewise, these findings suggest that the C/D recombinants (‘CD1’ and ‘CD2’) could have evolved after the recombination events and this has taken place over a long period of time. The ‘CD1’ and ‘CD2’ may have evolved over 500 years and 800 years, respectively.
Many studies have shown the association of HBV genotype/subgenotype and specific mutations with clinical outcome of HBV, though the role of HBV genotypes in response to antiviral therapy is still uncertain 
. Studies from Asia populations suggested that genotype C was associated with an increased risk of hepatocellular carcinoma, lower rate of spontaneous HBeAg seroconversion and higher rate of core promoter double mutations (T1762/A1764) compared with genotype B 
. A recent study from Alaska showed that HBeAg seroconversion occurred decades later in patients infected with HBV genotype C than in those infected with genotypes A, B, D and F, suggesting that genotype C may be responsible for most perinatal transmission 
. In the present study, the prevalence of HBeAg in the carriers of HBV/CD1 was comparable with that in the carriers of HBV/C2, suggesting that patients with HBV/CD1 had the similar HBeAg duration to patients with HBV/C2. The T1762/A1764 and A1896 mutations are the most common HBeAg-negative variants that reduce or abolish HBeAg production. These two types of mutation pattern may be preferentially selected by different genotype HBV strain in developing HBeAg-negative infection 
. Here we observed a significant lower tendency to develop A1896 (4.6% vs. 18.8%, P
0.004), but a slight higher tendency to develop T1762/A1764 (32.1% vs. 22.9%, P
0.244) mutations in patients with HBV/CD1 than in patients with HBV/C2, suggesting that the T1762/A1764 mutations were more selected than the A1896 mutation by the HBV C/D recombinant. The lower level of TBIL in patients with HBV/CD1 than in patients with HBV/C2 suggested that patients with HBV/CD1 may have a lower risk of liver damage. But it remains hard to conclude that HBV/CD1 has a lower capacity of disease-inducing than HBV/C2.
In summary, our present investigation shows that the HBV C/D recombinant is specifically restricted to the Qinghai-Tibet Plateau. This region is geographically located between East and West Asia where genotype C and D are the most prevalent HBV strains, respectively. HBV/CD1 had a lower tendency to develop A1896 mutation than HBV/C2, but the prevalence of HBeAg was comparable. Further studies are still needed to investigate the clinical significance of the HBV C/D recombinant and its association with hepatocarcinogenesis.