Echinococcus spp. and genotypes infecting humans in Tibet Autonomous Region of China: a molecular investigation with near-complete/complete mitochondrial sequences

Background Molecular markers are essential to identify Echinococcus species and genotypes in areas with multiple Echinococcus species to understand their epidemiology and pathology. Tibet Autonomous Region (TAR) is one of the areas worst hit by echinococcosis. However, molecular epidemiology is still missing among echinococcosis patients in TAR. This research explored the Echinococcus species and genotypes infecting humans in TAR and the population diversity and the possible origin of G1 in TAR. Methods Cyst samples were collected in one echinococcosis-designated hospital in TAR. Echinococcus species and genotypes were identified through a maximum-likelihood approach with near-complete/complete mtDNA using IQ-TREE. Phylogenetic networks were built with PopART, and the phylogeographical diffusion pattern was identified using a Bayesian discrete phylogeographic method. Results Using phylogenetic trees made with near-complete/complete mtDNA obtained from 92 cysts from TAR patients, the Echinococcus species and genotypes infecting humans in TAR were identified as Echinococcus granulosus (s.s.) G1 (81, 88.04%), accounting for the majority, followed by G6 of the E. canadensis cluster (6, 6.52%), E. granulosus (s.s.) G3 (3, 3.26%), and E. multilocularis (2, 2.17%). An expansion trend and a possible recent bottleneck event were confirmed among the G1 samples in TAR. Adding the other near-complete mtDNA of G1 samples globally from the literature, we identified the possible phylogeographic origin of the G1 samples in TAR as Turkey. Conclusions Using near-complete/complete mtDNA sequences of Echinococcus spp. obtained from echinococcosis patients, a variety of Echinococcus species and genotypes infecting humans throughout TAR were identified. As far as we know, this is the first comprehensive molecular investigation of Echinococcus species and genotypes infecting humans throughout TAR. We identified, for the first time to our knowledge, the possible origin of the G1 in TAR. We also enriched the long mtDNA database of Echinococcus spp. and added two complete E. multilocularis mtDNA sequences from human patients. These findings will improve our knowledge of echinococcosis, help to refine the targeted echinococcosis control measures, and serve as a valuable baseline for monitoring the Echinococcus species and genotypes mutations and trends of the Echinococcus spp. population in TAR. Graphical Abstract Supplementary Information The online version contains supplementary material available at 10.1186/s13071-022-05199-6.


Background
Echinococcosis is 1 of the 17 neglected tropical diseases prioritized to be controlled by the World Health Organization (WHO) [1]. Two main echinococcoses affect humans, namely cystic echinococcosis (CE), caused by Echinococcus granulosus (s.l.), and alveolar echinococcosis (AE), caused by Echinococcus multilocularis (E. multilocularis) [2]. These parasites rank second and third in the global food-borne parasite list of the Food and Agriculture Organization (FAO) [3].
As the treatment and prevention vary between AE and CE, accurate identification between them is very important. Even within CE, which was thought to be due to mostly Echinococcus granulosus (s.s.), it was later found that 11.07% of CE cases were due to E. canadensis [4,5]. A study in Mongolia identified all 18 children were infected by E. canadensis (94.4% G6/G7 of the E. canadensis cluster) [6]. Without molecular identification, our knowledge of CE was derived from a mixture of Echinococcus species and genotypes. Molecular markers are essential to detect cases in areas with multiple Echinococcus species and genotypes to understand their epidemiology, pathology, and infectivity to humans to establish corresponding control strategies [4,7,8]. Using short fragments, mainly through PCR, molecular epidemiology has been applied to identify the main circulating Echinococcus spp. infecting humans [5,[9][10][11]. However, short fragments of mitochondrial DNA (mtDNA) often could not distinguish closely related genotypes including G1/G3 and G6/G7 of the E. canadensis cluster [12,13]. These genotypes were classified based on mtDNA and named following the hosts identified. G1 was known as the sheep strain and was found worldwide, but it was reported not only from sheep, but also humans, kangaroos, dingo, cattle camel, pigs, and goats [14]. G3 was known as the buffalo strain and was first found in India [14]. Later, G3 was identified as the main genotype for human infection in Pakistan and North India [11,15]. G6 was known as the camel strain and was identified from camels in Somalia and Sudan, goats in Turkana, Kenya, and cattle and humans in China [14,16,17]. G7 was known as the pig strain and was first reported in Poland but was also reported from pigs in Bolivia and humans in Heilongjiang, China [14,18,19]. These strains/genotypes were mainly classified by sequencing partial mtDNA [20].
Western China is heavily affected by echinococcosis, accounting for 40% CE DALYs and 91% of the annual AE incidence globally [21][22][23]. Of the affected provinces, Tibet Autonomous Region (TAR) ranked top in echinococcosis prevalence [24]. Although there was a prevalence study in TAR using ultrasound diagnostics, molecular markers were seldom used to identify the Echinococcus species and genotypes infecting the echinococcosis patients in TAR. Previous molecular epidemiological studies in this region often included animal samples with very few samples from humans; even when samples from patients were involved, they were mainly patients from Qinghai or Sichuan Provinces, but rarely patients from TAR; besides, previous studies were using short fragments of the mtDNA such as cox1, a fragment of cox1, nad1, and atp6 [5,9,[25][26][27], which could not confidently allocate the samples into specific genotypes and could only generate networks with lower resolution; therefore, the exact Echinococcus species and genotypes infecting humans in TAR remain unknown. The accurate allocation of the samples into different species and genotypes is the prerequisite for the exploration of the epidemiology, pathology, and infectivity to humans of the different species and genotypes. After the exact species and genotypes infecting humans are identified, it is possible to explore the association between the genotypes and ultrasound presentation. The accumulated knowledge would lead to more accurate diagnosis and control measures [8]. In this study, we identified the Echinococcus species and genotypes infecting humans in TAR through near-complete/complete mitochondrial sequences and explored the corresponding ultrasound presentation and the genetic variability as well as the possible origin of the most common Echinococcus genotype in TAR.

Study area
TAR is a western province with seven prefectures and 3 million people; its average altitude is > 4000 m [28].
as we know, this is the first comprehensive molecular investigation of Echinococcus species and genotypes infecting humans throughout TAR. We identified, for the first time to our knowledge, the possible origin of the G1 in TAR. We also enriched the long mtDNA database of Echinococcus spp. and added two complete E. multilocularis mtDNA sequences from human patients. These findings will improve our knowledge of echinococcosis, help to refine the targeted echinococcosis control measures, and serve as a valuable baseline for monitoring the Echinococcus species and genotypes mutations and trends of the Echinococcus spp. population in TAR.
Keywords: Cystic echinococcosis, Alveolar echinococcosis, Tibet Autonomous Region, Echinococcus granulosus, Echinococcus multilocularis, Echinococcus canadensis, Mitochondrial genome, Next-generation sequencing TAR borders Xinjiang Uygur Autonomous Region to the north, Qinghai and Sichuan Provinces to the east, and Yunnan Province to the south (only a short section).

Sample collection
Patients who underwent surgery to remove hydatid cysts in the Second People's Hospital of TAR from 2018 to 2019 were invited to participate. After surgery, the cysts were checked for the presence of protoscoleces under a microscope and stored in 95% ethanol. Patient demographic information and ultrasound classification of the cyst were recorded and analyzed with R version 4.1.1. Maps were also drawn with R version 4.1.1. CE and AE differential diagnoses were made with ultrasound following the guidelines [29,30]. Species identification was carried out with qPCR, which was based on the primers designed by Boufana et al., which could identify three kinds of Echinococcus species, namely E. granulosus (G1), E. multilocularis, and E. shiquicus [31]. For each sample, three replicates were made, and the mean of the three was used as the final result.

DNA extraction and sequencing
Genomic DNA was extracted from protoscoleces (in fertile cysts) and/or germinal layers (in infertile cysts) from cysts taken from echinococcosis patients using the phenol-chloroform method [32]. The extracted DNA was used for sequencing on BGISEQ-500/DIPSEQ-T1 platforms (MGI, Shenzhen, China) if the DNA amount was ≥ 0.7 µg [33]. Different from the Sanger method, BGISEQ-500/DIPSEQ-T1 sequencing platform uses DNA nanoball sequencing technology and uses Rolling circle replication to amplify DNA linearly. The sequence information was later obtained through sequencing-bysynthesis technology. The minimum DNA amount for samples to be handled by BGISEQ-500/DIPSEQ-T1 was 0.7 µg; thus, if the DNA amount was below this value, we used the PCR method first to increase the concentration.

PCR amplification and mitochondrial genome sequencing
For the rest of the samples with DNA < 0.7 µg, 13 published primer pairs were used to amplify mitochondrial genes through PCR, and the amplification products were sequenced [12]. As most of the PCR primers were designed for G1/G3 and tested with G6/G7 of the E. canadensis cluster [12,[34][35][36], these primers were expected to work.

Phylogenetic analysis and comparison tests
We used IQ-TREE (version 1.6.12) to construct a maximum-likelihood tree with the near-complete/complete mtDNA sequences (dataset A) to identify the genotypes and species of the Echinococcus spp., and ModelFinder was used to identify the best-fit nucleotide substitution models [40,41]. A bootstrap value of 1000 replicates was used to test the robustness of the phylogenetic tree. A clade was supported if there were three or more samples (including the reference sequence) and the approximate likelihood-ratio test SH-aLRT ≥ 80% and UFboot ≥ 95% [42,43]. The mitochondrial sequence data of Versteria mustelae (AB732957) were used as an outgroup of the genus Echinococcus [44]. The other reference sequences included E. granulosus G1 (AB786664), E. granulosus G3 (KJ559023); E. equinus G4 (AB786665); E. ortleppi G5 (AB235846); G6 of the E. canadensis cluster (AB208063), G7 of the E. canadensis cluster (AB235847), E. canadensis G8 (AB235848), E. canadensis G10 (AB745463); E. felidis (AB732958); E. multilocularis (AB018440.2); E. shiquicus (AB208064); E. oligarthra (AB208545); and E. vogeli (AB208546). The analysis was tested at least twice with the same dataset to verify the consistency of the results. After the species and genotypes were identified with near-complete/complete mtDNA, the same procedures were carried out with the major mitochondrial genes used in the literature, including cox1, nad1, nad2, nad5, and atp6 (dataset B), to test whether, using the key genes, the species and genotypes could be identified successfully.
After that, non-parametric two-tailed Wilcoxon-Mann-Whitney tests were performed to check whether the average cyst size of the different species/genotypes of CE infection was significantly different (P < 0.05 was considered significant). Only those genotypes with over five samples were included. Spearman's correlation (r) was used to assess the correlations between age and the average cyst size (P < 0.05 was considered significant).

Bayesian phylogeographic analysis
To understand the phylogeographical diffusion pattern, especially how G1 spread to TAR in China, we analyzed our G1 samples together with the 212 G1 samples from Kinkar et al. through a Bayesian discrete phylogeographic method [13,50]. Using the Bayesian stochastic search variable selection (BSSVS), we reconstructed the phylogeographic diffusion process through a Bayes factor (BF) test [50]. The procedures followed the Beast Tutorial titled "Phylogeographic diffusion in discrete space" (https:// beast. commu nity/ works hop_ discr ete_ diffu sion-accessed in March 2021). The sampling dates of our samples used the month of the surgery when the cysts were taken out. As no indication of the other samples' collection dates was available, we used the date when the manuscript with the sample was first submitted. The analysis was performed with BEAST v1.10.4 using the BEAGLE library [51,52]. The best fit model was chosen with ModelFinder [40]. Under the corrected Akaike information criterion (AIC), the best model was GTR + F + I + G4. MCMC chains were run for 100 million states, sampling every 10,000 steps with 10% burn-in, after which, Tracer v1.7.1 was used to check the effective sampling size (ESS) of estimates [51,53]. The MCC tree was produced with TreeAnnotator v1.10.4 and visualized with SpreaD3 v0.9.7.1 [51,54]. The combination of the runs was carried out with LogCombiner v1.10.4 [51]. BF and posterior probabilities were generated with SpreaD3 v0.9.7.1 [54]. Only routes which could meet BF ≥ 3 and the node posterior probabilities ≥ 0.5 were interpreted as supported [50,55]. The map was drawn with R version 4.1.1.

Phylogenetic identification of the species and genotypes
A total of 92 near-complete/complete mtDNA sequences were successfully obtained from 92 patients' samples, among which 50 samples were sequenced directly, and 42 samples were sequenced after PCR amplification. Of the 13 PCR primer pairs published by Laurimae et al., 12 worked (the E9 primer pair failed) [12]. Phylogenetic identification of the sequences is shown in Fig. 1, which used the near-complete/complete mtDNA from all the cysts samples (dataset A) and the key references of Echinococcus spp. and genotypes. The major mtDNA genes of all the samples (dataset B) were used to construct the maximum-likelihood trees together with the key reference sequences' genes, which are given in Additional file 1: Figure S1, Additional file 2: Figure S2, Additional file 3: Figure S3, Additional file 4: Figure S4, Additional file 5: Figure S5.

Participants' demographic characteristics and clinical presentation
A summary of the patients' demographics (age, sex, and residency), sample information (endocyst/cyst fluid), cyst information (number, location, average size, WHO Informal Working Group on Echinococcosis (WHO-IWGE) cyst classification [56]), and qPCR results is given in Table 1. Detailed information including genotypes and species results (obtained from the maximum-likelihood method) of all the participants is given in Additional file 6: Table S1.
Most patients were females (67.39%). All except three patients were undergoing the surgery for the first time. Nearly all the cysts were from hepatic lesions (95.65%). Five patients had lesions in the liver and one more region, including two in the pelvic cavity, two in the spleen, and one in the abdomen. Most of the hepatic lesions were in the right liver only (73.86%), while 14.77% affected the left liver only.
All G3 and E. multilocularis infected patients' cysts were found in the right liver only, and in 66.67% of the G6 patients were found in the right liver only. All the G3 cysts were classified as CE1, and 85.71% of the G6 cysts (six out of seven) were classified as CE3, including two cysts (both CE3) from one patient. In comparison, 28.40% of G1 patients had at least one CE1, 18.52% had at least one CE3, and 33.33% G1 patients had at least one CE2. Some G1 patients' cysts were CE4 and/or CE5.
Most of the cysts/pseudocysts were > 5 cm (82.42%). The average size was 8.4 cm (95% CI 7.7-9.1) for all the cysts and G1, 11.2 cm for G3, and 6.3 cm for G6 (95% CI 3.8-8.8) and 9.5 cm for the pseudocysts caused by E. multilocularis. As too few samples were found in G3, no further comparison was made for them. A non-parametric two-tailed Wilcoxon-Mann-Whitney test was carried out to test whether the average size of G6 would be the same as that of G1 with the null hypothesis. Comparing the cyst sizes of G6 of the E. canadensis cluster and E. granulosus G1 identified P = 0.084, meaning the null hypothesis was likely, with no difference between the cyst sizes caused by the two genotypes. The mean ages of G1 and G6 patients were 31.3 (95% CI 27.9-34.6) and 24.7 (95% CI 17.2-32.2) years, respectively (two-sided Wilcoxon-Mann-Whitney test P = 0.361), meaning the two groups of patients' ages were not statistically different. Spearman's correlation between age and the average CE cyst size showed a positive correlation (r s = 0.3195, P = 0.0039).
Using ultrasound, only one AE case (NQBQ3) was identified successfully, but no conclusive diagnosis was made for the other AE case (STR3). In contrast, the qPCR method only identified one AE (STR3) accurately, and the other AE case (NQBQ3) was identified as "E. multilocularis and G1" infection, which was the same result as for one G1 sample (NQBQ4). All G3 and G6 samples were classified as G1 according to the qPCR result.

Spatial distribution of the genotypes identified
All patients were Tibetans living in TAR throughout their lives. The spatial distribution of the species and genotypes can be found in Fig. 2.
As shown in Fig. 2, the patients were from all seven prefectures, with most from Naqu and Shigatse. The two AE cases were from Naqu and Shigatse. Most of the G6 cases (four out of six) were from Naqu, and all G3 cases were from Shigatse.

Phylogenetic networks of the E. granulosus (s.s.) samples and references
As shown in Fig. 3, the phylogenetic networks were made using near-complete mtDNA (11,007 bp) and cox1 (1674 bp) of the mtDNA sequences from E.  Fig. 1 Phylogenetic tree of Echinococcus spp. and genotypes with a maximum-likelihood approach using the near-complete/complete mtDNA sequences granulosus (s.s.) samples with reference G1 (AB786664) and G3 (KJ559023) [57,58]. Figure 3a shows that using the near-complete G1 and G3 sequences, 78 haplotypes were formed, and the pattern went radially from the center. There were two major G1 clusters of sequences, with a branch of G3 farthest from the root by 34 mutations away from the G1 cluster. Using G3 reference KJ559023 and our G3 samples as the outgroup, Hap_14/NQBR6 from Naqu was closest to the root. Equally far, Hap_46/ STN14 from Nanmulin in Shigatse was the center of the other cluster. The reference G1 sequence AB786664 or Hap_17 was three mutations from the center Hap_46/STN14. The location of STN14, Nanmulin, was bordering Naqu. The G1 samples from Shigatse and Naqu were interwoven, with no obvious geographical clustering. Figure 3b, which is the network of the E. granulosus (s.s.) samples using the cox1 gene, shows that our samples formed 47 haplotypes. Using G3 as the outgroup (Hap_45/KJ559023, Hap_46/STK1&STB1, and Hap_47/ STN2), an expanding trend was observed, though with reduced resolution compared with Fig. 3a. A major haplotype at the center Hap_2 included 17 samples (8 from Shigatse, 6 from Naqu, 1 from Shannan and Lhasa each, and G1 reference AB786664). G3 cox1 reference from KJ559023 was five mutations away from the root, clustered with our G3 samples.

Phylogenetic relationship of the G1 samples of TAR and neighboring provinces
To understand the relationship with other neighboring provinces, we added G1 samples from Qinghai, Xinjiang, and Sichuan downloaded from NCBI based on a previous paper [9] to draw the network, shown in Fig. 4. As mentioned, previous research mainly used shorter fragments such as partial cox1, cox1, or nad1; we could only use shorter fragments to draw the network with the samples from the neighboring provinces. We also added the reference sequence of G1 (AB786664).
A star pattern was identified, and the major haplotype Hap_2 included 34 samples from TAR, 1 sample from Qinghai, and the common haplotype G01 (reference AB786664-scg1). Many haplotypes were shared by the samples from TAR and Qinghai.
As many of our G1 haplotypes were shared with Qinghai samples, we calculated the Fst of the G1 population from the four provinces in China, shown in Table 2.
As Table 2 shows, Fst indicated that there was no differentiation in sub-populations between TAR and Qinghai. Moderate differentiation was found between TAR and Sichuan (Fst = 0.09) and Xinjiang (Fst = 0.05).

Population diversity and neutrality indices of the G1 samples
Population diversity and neutrality indices of the G1 samples were calculated based on the near-complete sequences and the key genes. The results are shown in Table 3.
As shown in Table 3, among the 81 G1 samples, the haplotype diversity (Hd) index was the highest with the near-complete sequences. The second highest Hd index and the number of haplotypes were with cox1. Both nearcomplete mtDNA and most of the genes (except atp6) showed significant negative values, which indicated the deviation from neutrality was significant.  [13,50]. Based on the selection criteria, diffusion routes with BF ≥ 3 and node posterior probabilities ≥ 0.5 were supported. The results are shown in Fig. 5.
As shown in Fig. 5, 17 well-supported spatial routes were identified. Ten routes had BF > 20 and seven BF > 300. Six out of the 17 routes were originated from Turkey, including the one spreading to China (BF > 53, posterior probabilities > 0.7). The six countries whose routes originated from Turkey included Iran, Greece, Mongolia, China, Moldova, and Romania (BF and posterior probability in decreasing order). Four routes originated from Tunisia, namely Algeria, Italy, Argentina, and Spain (BF and posterior probability in reducing order).

Discussion
Western China is an endemic region of both CE and AE, and TAR ranks top in human echinococcosis prevalence among the western provinces of China [24,28,59]. Although the prevalence of echinococcosis was explored [28], the causative agents of human infection have not yet been fully understood in TAR. Previous molecular studies on the Tibetan Plateau often included samples from animals; even when samples from humans were involved, they were mainly patients from Qinghai or Sichuan Fig. 4 Median-joining network using partial cox1 gene (789 bp) from E. granulosus (s.s.) G1 samples and other G1 samples (n = 43 haplotypes) from three neighboring provinces [9] and also the G1 reference sequence AB786664 (SCG1) Provinces, with few patients from TAR; moreover, previous studies were using short fragments of the mtDNA with limited phylogenetic resolution [5,9,[25][26][27].
In this research, using the near-complete/complete mitochondrial sequences obtained from the cysts of 92 echinococcosis patients in a designated hospital, the exact Echinococcus species and genotypes infecting humans all over TAR were clarified to be three species and four genotypes: E. granulosus (s.s.) G1 accounting for the majority, followed by G6 of the E. canadensis cluster, E. granulosus (s.s.) G3, and E. multilocularis. As far as we know, this is the first comprehensive molecular investigation of Echinococcus species and genotypes infecting humans based on near-complete/complete mtDNA sequences. The genetic variability was explored, and the population expansion trend of G1 was identified. For the first time, we also identified the possible origin of the G1 in TAR, which was probably from Turkey, in Table 3 Diversity and neutrality values of the G1 samples included in this study using different lengths/genes of mtDNA sequences n number of samples examined, Hn number of haplotypes, Hd haplotype diversity, pi (π) nucleotide diversity, D Tajima [13,60]. However, the analysis, as highlighted by Kinkar et al., could be influenced by the unbalanced samples obtained [13]. The complete mtDNA database was also enriched through this study. Previously, in NCBI, there was only one complete genome of E. multilocularis (AB018440.2, 13,738 bp), which was from a naturally infected vole from Hokkaido and maintained through Mongolian gerbils [61]. Two complete E. multilocularis mtDNA sequences (13,738 bp) obtained from Tibetan patients were added through this study. Similarly, there was only one complete mtDNA of E. granulosus G3 stored in NCBI (KJ559023.1 13,702 bp), which was from a Sichuan patient [58]. Three mtDNAs of E. granulosus G3 sequences (13,702 bp) were added through this study. Using near-complete/complete mitochondrial sequences to identify the exact species and genotypes of the echinococcosis infection was feasible, providing a better phylogenetic resolution and accuracy than commonly used short genes. Understanding of the species and genotypes is essential for targeted diagnostic tool development and control strategies [8,62]. For example, Yang et al. reported only G1 was identified in the Tibetan communities in Sichuan Province; thus, the key control method would be the Eg95 vaccine together with dog anthelmintic treatment [63]. However, this method may not be sufficient in areas with more Echinococcus species and genotypes, where more monitoring of the respective hosts, more hygiene education, and dog treatment are needed [63]. The Echinococcus species or genotypes identified in human patients could all be due to eggs produced by adult Echinococcus spp. parasites in the definite hosts, which indicates the relationships between the pathogens of humans and animals. More Echinococcus species and genotypes identified in an area mean more animals involved in the transmission cycle, and the targeted control in this region should be more than targeting sheep and dogs. Though human pathogens are not involved in the transmission of echinococcosis at all, given the sources of human pathogens could all be animals, it is possible to make phylogenetic trees with samples from humans and animal sources. With near-complete/complete mtDNA, the closely related genotypes such as G1/ G3 and G6/G7 were differentiated, which could not be achieved using key genes including cox1, nad1, nad2, and atp6, as demonstrated in the phylogenetic trees, but it was possible to differentiate G1/G3 with nad5, which confirmed the previous study finding of Kinkar et al. [64].
Ultrasound only diagnosed one AE patient accurately. QPCR only identified one AE patient accurately, and the other AE patient had "E. multilocularis and G1" infection, which was the same result as in one G1 patient.
The qPCR method could only identify the three species as indicated in the reference, namely E. granulosus (G1), E. multilocularis, and E. shiquicus [31]; closely related species/genotypes may be identified as one of the three. Near-complete/complete mtDNA could be considered in areas with multiple species and genotypes such as TAR to identify the accurate species and genotypes among patients who need surgery to remove the lesion when the differential diagnosis such as imaging or qPCR is not conclusive. No E. shiquicus infection was identified with either qPCR or mtDNA method, which confirmed that humans were probably insusceptible to E. shiquicus, a new species found in the Tibetan Plateau only, despite the concern of its potential, given dog infection of E. shiquicus was reported [65]. Besides, near-complete/complete mitochondrial sequences of patients could serve as baseline data to monitor the Echinococcus species and genotype prevalence changes, mtDNA mutations, and trends of Echinococcus spp. population.
Like many other areas in the world, E. granulosus (s.s.), especially G1, accounted for most of our samples [4,9,63,66,67]. Using the cox1 gene, Heath et al. reported the samples obtained from yaks from Tibetan Plateau as G1 genotype infection, but no protoscoleces were identified in their cysts [67]. Due to the short target genes used, most of the previous studies could not differentiate the genotypes within E. granulosus (s.s.) with confidence. With near-complete/complete mtDNA, three G3 samples were identified, all from Shigatse. Interestingly, G3 was reported as the dominant genotype infecting humans in Pakistan and North India [11,15], which bordered TAR, especially Shigatse directly or via Nepal. There were limited molecular studies on causative species and genotypes of echinococcosis patients in Nepal, but it was reported that 5% of the water buffalos examined were found to have hydatid cysts, and G3 was previously called the buffalo strain [14,68]. Though limited in the number of G3 samples, we found all were active CE1. Further studies in this region might help us to explore the characteristics and ultrasound presentation of G3 infection.
Our identified genotypes G1, G3, and G6 are in line with the genotypes identified in the study with animal samples (sheep and yaks) in TAR [27]. Using nad1 and nad5 genes through BLAST comparison, G1, G3, and G6 were found to be the causative agents of cysts in the animal samples from Shigatse and Lhasa, and the authors confirmed the two G6 samples from Zhongba county in the west of Shigatse with complete G6 mtDNA [27]. The two G6 samples in our study were from Nanmulin and Renbu in the east of Shigatse, and the other four G6 samples were from Naqu. The geographical distribution of G6 was wider than in the previous reports, and further molecular studies in TAR could improve our understanding of G6. The reason behind the wide distribution of G3 and G6 in TAR is to be explored.
Molecular identification of the causative agents is not only a prerequisite to understanding the epidemiology and pathology of Echinococcus spp. but also essential to formulate targeted control measures [4,7,8]. Our study, though preliminary, improved our understanding of the ultrasound characteristics of G1, G3, and G6. G3 infections were large active CE1, and G6 infections were mostly transitional CE3 with smaller cysts. The average age of patients infected by G6 seemed younger at 24.7 (95% CI 17.2-32.2) compared to G1 patients at 31.3 (95% CI 27.9-34.6) years, but no significant difference was found. Though it was reported in Mongolia that all pediatric cases were infected with E. canadensis, especially G6/G7 of the E. canadensis cluster [6], in our study, 90% of the children were infected with G1, with only two children infected with G6. It was hypothesized that the lack of adult cases infected with G6/G7 in the Mongolia study could be explained in two opposite directions: E. canadensis infection could be mild and spontaneously healed among children or lead to early death of the children infected with G6/G7 [4]. Our study supports the mild G6 hypothesis, given that (i) most of the G6 cysts were transitional CE3; (ii) most of the G6 cysts were small; (iii) all of the G6 patients were > 14 years old, with 66.7% > 27 years old; (iv) most of the G6 patients came to the hospital because of detection of the cyst by ultrasound in either regular body check-ups or echinococcosis screening programs instead of symptoms. As there were no symptoms, even after ultrasound positive results of echinococcosis, the patients waited 5-10 months before coming to the hospital to undergo surgery. This was in line with the findings of Schneider et al. among the patients in Europe [10]. Due to the limited sample size, this mild hypothesis of G6 needs further verification in larger scale studies with molecular markers.
Except for atp6, all the genes of G1 studied inferred significant deviation from neutrality. The expansion trend and a possible recent bottleneck event were confirmed in the region as hypothesized by Nakao et al. [9]. The nad5 gene was not complete in four samples sequenced after PCR and the G1 samples were only near-complete, probably because the nad5 gene was located next to the tandem repeat region [69], which made it difficult to carry out PCR; thus, for G1 samples, it was hard to obtain the full mtDNA sequences. The other possible reason could be that these four samples were not as good as the other samples and the PCR primers for nad5 failed. Of the 13 PCR primer pairs published before, 12 pairs worked; the E9 primer pair was specially designed for G6/G7 [12], which might not be suitable for G1/G3 samples. For samples with low-quality and low-concentration DNA, these 12 primer pairs could be used to amplify DNA to obtain near-complete mtDNA, but the targeted region of the E9 primer pair should be improved by considering more genotype sequences such as G1/G3. Our team is exploring this direction.
Only one hospital specialized in hepatic cyst removal was selected as the study site, thus limiting the cysts/ pseudocysts from other organs and regions, which might have limited the generalizability of our study. It was hypothesized that different Echinococcus species and genotypes might have different organotropism, for example, G6 might exhibit more brain tropism than G1 [70]. In this research, the hospital only examined the brain if some symptoms/indicators possibly involved the brain. Even with this limitation, patients from all seven prefectures were included, and a variety of species and genotypes infecting humans were identified; thus, more cyst/ pseudocyst samples from more organs and hospitals would probably increase the diversity identified.
Another limitation was that no follow-up of the patients after the surgery was performed; thus it was not known if there was any difference between the prognoses of the patients infected by different Echinococcus species and genotypes. The lack of follow-up also limited the possible application of mtDNA sequences to differentiate between re-infection and relapse. It is hypothesized that if the mtDNA sequences obtained from the new cysts/ pseudocysts are identical to the corresponding patients' mtDNA identified this time, it would probably be due to relapse, but if the mtDNA sequences obtained from the new cysts/pseudocysts are different from the corresponding patients' mtDNA identified this time, it would probably be re-infection. It is believed that with the nearcomplete mitochondrial sequence records and mtDNA sequences obtained from the patients' new cyst, the reinfection or relapse could be differentiated. A larger scale study design with follow-up should be planned in the future in this direction.

Conclusions
We conducted a comprehensive molecular investigation of Echinococcus species and genotypes infecting humans throughout TAR based on near-complete/complete mtDNA sequences. We found patients infected by three Echinococcus species and four genotypes including E. granulosus (s.s.) G1 and G3, G6 of the E. canadensis cluster, and E. multilocularis. The genetic variability was explored, and the population expansion trend of G1 was identified. For the first time, we also identified the possible origin of the G1 in TAR in China. The findings also enriched the long mtDNA database of Echinococcus spp. and added two complete E. multilocularis mtDNA sequences from humans. The findings improved our knowledge of echinococcosis infecting humans in TAR, a region heavily affected by echinococcoses, which would help to refine the targeted echinococcosis control measures and serve as valuable baseline data for monitoring the Echinococcus species and genotype prevalence changes, mtDNA mutations, and trends of the Echinococcus population in the region.