Dried blood spot sampling for hepatitis B virus quantification, sequencing and mutation detection

Hepatitis B virus (HBV) diagnosis is performed on serum samples, but the access to this diagnosis is difficult in low-income regions. The use of dried blood spot (DBS) samples does not require special structure for collection, storage or transport. This study evaluates the use of DBS for detection, quantification and sequencing of HBV DNA using in-house techniques. Two study groups were included: 92 HBsAg + individuals and 49 negative controls. Serum and DBS samples were submitted to quantitative and qualitative in-house PCR for S/pol genes, sequencing and phylogenetic analyses. Total of 84 serum samples were successfully amplified. Of them, 63 paired DBS were also positive in qualitative PCR. Qualitative PCR in DBS presented a sensitivity of 75% and specificity of 100% (Kappa = 0.689). Quantitative PCR in DBS presented a detection limit of 852.5 copies/mL (250 IU/mL), sensitivity of 77.63% and specificity of 100% (Kappa = 0.731). A total of 63 serum samples and 36 DBS samples were submitted to sequencing, revealing the circulation of genotypes A (65.08%), D (4.8%), E (3.2%) and F (27%) with 100% of correspondence between serum and DBS. All sequenced samples displayed polymorphisms in HBsAg gene. An HIV-coinfected patient presented the rtM204V/I-rtL180M double resistance mutation in serum and DBS. In conclusion, DBS is an alternative to detect, quantify and characterize HBV DNA, being a possibility of increasing diagnosis in low-income settings, closing gaps in HBV control.

www.nature.com/scientificreports/ Thus, the advent of a quantitative in-house real time PCR that can be adapted for DBS samples could improve the access to HBV molecular diagnosis in areas of limited infrastructure 16,17 . Moreover, the use of DBS in molecular epidemiology studies, to better understand the role of genotypes and viral mutations in the course of hepatitis B is crucial to promote ways to access HBV diversity in settings where the gold standard methods are not available.
Once most HBV chronically infected individuals live in remote regions, with limited laboratory infrastructure to perform the virus diagnosis, the access to economical and accurate diagnostic methods is an essential step for the eradication of HBV. At our knowledge, scarce studies have tracked HBV genotypes, resistance and vaccine-escape mutations in DBS samples and most of them did not evaluate the reliability of DBS compared to their paired serum samples 12,[18][19][20][21] .
The present study aimed to evaluate the usefulness of DBS for HBV quantification and sequencing, using in-house techniques.

Results
Characteristics of participants. A total of 141 individuals were enrolled in this study. Group 1 was composed by 92 HBV infected individuals, 53 (57.6%) were men; mean age was 44.26 ± 14.97 years. Group 2 had 49 healthy individuals, 35 (71.4%) were women; mean age was 37.12 ± 11.27 years.
Eighteen patients (19.6%) related previous antiviral treatment (lamivudine, entecavir or tenofovir). According to HBV risk behaviours, both groups reported frequently going to the manicure and sharing razors. However, 39/49 (79.6%) people from group 2 mentioned using their own pliers during the procedure. Another point is the use of condoms. Almost 40% of individuals from group I reported never using a condom during sexual intercourse in contrast to 12.24% from group 2 (p = 0.33). Regarding coinfections, 6 patients from group I presented HBV-HIV coinfection.
From group 1, 24 (26.1%) individuals were HBeAg positive, and 69 (75%) were anti-HBe positive. No patient was positive for HCV. Biochemical and serological profiles of the subjects are shown in Table 1.
Analytical sensitivity: linear dynamic range and detection limit. A serial dilution panel was used to evaluate the linear dynamic range and detection limit of DBS samples. The lowest HBV dilution detected in qPCR was 3 log10 (3,410 copies/mL or 1,000 IU/mL). Then, this sample was twofold diluted, from 1/2 to 1/64, and these dilutions were subjected to another run of the same qPCR in duplicate. HBV DNA was detected until the dilution of 1/4, resulting in an estimated detection limit of 852.5 copies of HBV/mL or 250 IU/mL (Table 2).
To assess the precision of qPCR, a DBS sample with HBV viral load of 4 log IU/mL (~ 4 log copies/mL) was quantified 20 times in the same run and the coefficient of variation (CV) was 0.73.

Reproducibility and repeatability.
To assess the reproducibility of the qPCR in DBS samples, high and low positive controls (HPC and LPC) were tested by three operators in three different days, twice a day, using the same reagents. Results are presented in Table 3. For HPC, a CT variation from 16.62 to 20.61 (mean of 17.56 ± 0.93) was observed, while for the LPC the variation was from 31.03 to 36.54 (mean of 33.65 ± 1.5), presenting CV of 0.05 and 0.04, respectively. There was no statistical difference between the mean CT values for both HPC and LPC demonstrating good reproducibility of the method.
Regarding repeatability, viral loads of HPC in the same day conducted by the same operator were similar, while for LPC, variation of viral load was noted in the same reaction and in different days.    www.nature.com/scientificreports/ When HBV viral load was compared between serum and DBS, significantly higher viral load was observed among serum samples, in comparison to DBS (3.03 vs. 1.62 log copies/mL). However, qPCR results for serum and DBS showed good correlation, as demonstrated by the coefficient of Pearson (r = 0.844; 95% IC: 0.747-0.905, p < 0.0001). Figure 1A shows the comparison of log copies/mL values between DBS and serum samples in qPCR. In this case, the qPCR in serum had a median equal to 3.03 while the qPCR in DBS had a result equal to 1.62.
Finally, the pairwise comparison of HBV viral load between qPCR in serum and DBS was performed using a Bland-Altman Plot (Fig. 1B). A mean point of 1.66 log copies/mL was estimated, with a standard deviation of 1.96 log copies/mL. The upper limit of agreement was 4.15 (95% CI 3.58-4.73) and the lower limit was − 0.83 (95% CI − 1.40 to 0.26). Three samples were outside the limits of agreement, with an estimated value of − 0.92; − 1.23 and 6.6; which represent 5.08% of the samples, an index slightly above the tolerated limit (5%). Within the region of agreement, there were differences between the measurements obtained by the two samples.
Regarding DBS samples, there was 100% agreement with the genotype observed in the respective paired serum sample, as illustrated in the phylogenetic tree containing the 36 serum and DBS pairs (Fig. 2B). From them, 23 pairs belonged to genotype A (63.8%), 10 to genotype F (27.8%), 2 to genotype D (5.6%) and one to genotype E (2.8%). Only one pair was located in different branches in the phylogenetic tree, which may suggest the classification of these two samples in distinct subgenotypes.
In addition, variables as viral load, HBsAg titers, ALT and AST values, HBeAg, anti-HBe and anti-HIV positivity in relation to the different genotypes/subgenotypes and to the sequencing result were evaluated. Genotype A was statistically associated with low levels of AST and higher HBsAg titers when compared to genotype F (p < 0.005). Furthermore, sequenced samples had higher HBV viral load and HBeAg positivity than not sequenced samples (p < 0.05) while not sequenced samples had lower anti-HBe positivity than sequenced samples (p = 0.01).
Nucleotide sequences of serum and DBS were analyzed for HBsAg/polymerase mutations. All samples presented at least one polymorphism; most of them were neutral mutations. Overall, ~ 98% of polymorphisms observed in serum were also found in their paired DBS samples. The presence of HBV surface antigen-related mutations were identified in 10 serum samples (15.87%), with 144D → A (1), 133 M → I (1), 100Y → C (6), 109L → V (1) and 109L → R (1) being the most frequent mutations ( Table 7). The paired DBS was available for 6 sequences and presented 100% concordance with the HBsAg mutations described in serum.
One patient had the double resistance polymerase mutation rtM204V/I-rtL180M, in both serum and DBS. This is a 52-year-old man, heterosexual, from Northeast region, HIV co-infected, self-declared history of syphilis in the past, HBeAg + /anti-HBe-, subgenotype F4, HBV load in serum of 5.14 log UI/mL (5.67 copies/mL).

Discussion
Broad access to HBV molecular diagnosis is a basic condition for starting therapy and monitoring the emergence of drug-resistant strains. However, ensuring this access is still a challenge in remote places and low-income settings 1,11,16 . In these cases, when the gold-standard tests are not available, the use of DBS and in-house techniques play a role in providing molecular diagnosis to hard-to-reach populations. Here, in-house qPCR was employed to detect and quantify HBV DNA in DBS samples. In addition, the applicability of DBS for sequencing purposes, as HBV genotyping and tracking mutations, was evaluate.
In this study, the detection limit of HBV in DBS by qPCR was 852.5 copies/mL (~ 250 IU/mL). Other studies using COBAS ® TaqMan ® HBV adapted for DBS have described detection limits of 914 IU/mL 19 and 1400 IU/ mL 24 . These discrepancies can be explained by differences between commercial and in-house methods, such as the target gene (S/Pol vs. Pre-Core/Core genes) and several automated steps.
The qPCR in DBS samples presented good agreement (86.40%), sensitivity (77.63%) and excellent specificity (100%) compared to serum samples. In-house qPCR was used as reference since this method was previously optimized and compared to commercial methods 25 Some positive samples by commercial method were excluded to compare the results obtained in DBS what could impact in the estimation of sensitivity and specificity of the assay in this study and it is considered a limitation of the present work. Some studies carried out in DBS showed higher sensitivity values [26][27][28][29] , however most of them were carried out using commercial methods and with a smaller sample size. Moreover, satisfactory values of viral load and detection limit were observed in the repeatability and reproducibility analysis, demonstrating the reliability of the method.
Although HBV viral load was higher in serum compared to DBS, we observed good agreement between results from qPCR in serum and DBS as demonstrated by kappa value. Bland-Altman analysis demonstrated only few discordant results between qPCR in serum and DBS probably due to small differences in the performance of the assay among those samples. Other studies also demonstrated good correlation between qPCR for HBV in serum and DBS 19,26,27 . These results reinforce that the qPCR using DBS samples may be an economic and efficient alternative to HBV quantification in settings where serum sampling is not accessible.
Some variables were associated with qualitative and quantitative detection of HBV DNA in DBS, such as higher viral loads detected in qPCR, higher HBsAg values, HBeAg positivity and the absence of anti-HBe in the respective serum samples. Other studies have reported a greater HBV DNA detection in DBS samples with higher viral loads detected in serum 19 www.nature.com/scientificreports/ related to viral replication, were also associated with success in HBV detection by qPCR. In agreement, it was observed that anti-HBe reactive samples were more likely to have negative results in DBS, being related to a lower viral replication and, consequently, lower viral genome detection 31 . Despite DBS is commonly used in mass serological screening, studies reporting its use for sequencing purpose are scarce and the robustness of DBS for molecular characterization is still an issue. Consequently, in low-income settings virus diversity is seldom determined, contributing to the lack of association of HBV DNA diversity and the hepatitis B phenotypes [32][33][34] .
In this study, we also investigated the genetic variability of HBV variants in 63 samples and evaluated the applicability of DBS for molecular epidemiology studies.
The comparison between paired serum and DBS sequences revealed great similarity, ranging from 98 to 100%. Phylogenetic analysis demonstrated that genotypes A and F were the most prevalent. Individuals infected with HBV/A showed higher values of HBsAg and lower AST titters compared to individuals infected with HBV/F (p < 0.005). Croagh et al. 35 reported that HBV/A was associated to chronicity, while infection promoted by HBV/F tends to have limited progression and beneficial evolution, which may explain the difference in HBsAg levels.
Regarding subgenotypes, in both serum and DBS, HBV/A1 (57.1%) was the most prevalent, followed by HBV/F2 (23.8%), A2 (7.9%), F4 (3.2%), E (3.2%), D2, D3 and D4 (1.6% each). As have been stated, genotypes A, D and F are the most prevalent in Brazil [36][37][38][39] . On the other hand, genotype E is rarely detected in Brazil and when found, is often linked to the recent waves of African migration 38,39 . Here, genotype E was identified in two individuals from Angola living in Brazil. In agreement, phylogenetic analyses indicate that these HBV/E isolates are closely related with strains from Angola and Guinea. Table 6. HBV DNA sequencing in DBS samples correlated to HBV viral load, biochemical and serological markers. *ALT alanine-aminotransferase, AST aspartate-aminotransferase, GGT gama-glutamyltransferase (GGT). www.nature.com/scientificreports/ It is noteworthy that HBVA1 was the most prevalent subgenotype in the Southeast, while F2 was the most frequent in Northeast. High proportion of HBV/F in Northeast was previously observed by Mello et al. 36 , and more recently, confirmed by Lampe et al. 38 . Phylogenetic analyses revealed that samples from the Northeast and Southeast had a heterogeneous distribution along the tree, suggesting that HBV strains circulating in these two regions don't present a monophyletic origin. Brazil is a continental country where differences in colonization influenced the HBV genotype distribution along the regions 39 . However, expressive migration flows between the Northeast and Southeast over time may play a role in HBV dispersal and mixing of viral variants. Although this study does not suggest a distinction between the isolates from the two regions, studying the HBV variability distribution by region is crucial for increasing the knowledge of HBV dispersal patterns and for surveillance of viral variants circulating in chronic carriers.
In this study, 10 serum sequences displayed mutations in HBsAg gene. All mutations found in serum were also detected in the respective DBS sequences (when available), showing the potential of this sampling as an alternative for molecular analysis. All the amino acid changes were detected within the major hydrophilic region (MHR), where the determinant "a", the main target of B and T cells, is located. Mutations in this region can affect the antigenicity of HBsAg and may be related to occult HBV infection and escape from vaccine-induced immunity 40 .
Overall, Y100C was the most frequent substitution, being present in 6/10 sequenced samples. Despite this mutation has been associated with occult HBV infection [41][42][43] , in this study all carriers were HBsAg-positive. As demonstrated by Mello et al. 44 , Y100C alone may not affect HBsAg production, secretion or HBsAg affinity by commercial serological assays, as for our samples. Thus, the presence of other potential reasons that may influence on HBsAg detection should be further investigated. In addition, L109R/V mutations were present in 2/10 samples and have been related to HBV vaccine escape and virus evasion to the host immune system 7,45 .
Regarding resistance mutations, the double rtM204V/I-rtL180M mutation in polymerase gene (to lamivudine, telbivudine and entecavir-partial) was observed in both serum and DBS samples from an HIV-treated patient. As observed by Komas et al. 18 , this finding reinforces the robustness of DBS to detect clinically relevant mutations. The accurate detection of mutations with clinical significance, as vaccine-escape and resistance mutations, is essential for better design immunization strategies and managing chronic carriers under antiviral therapy.
Our study has some limitations such as the selection of samples with viral loads above 1000 copies/mL. In addition, the assay described herein target only one region S/Pol of the HBV genome. Moreover, DBS is a less sensitive method providing a lower HBV quantification when compared to the gold-standard serum sampling, which can lead to an underestimation of viral loads. Due to this limitation, optimizations are still necessary to provide a larger number of sequenced DBS samples. DBS samples could be useful for initial screening followed by second test in serum samples before to initiate the treatment. In addition, DBS could be used to evaluate mutations and for molecular epidemiology studies when serum sampling is not available. At our knowledge, this study analysed the largest number of HBV sequences using DBS available in literature, thus contributing to increase the knowledge in this setting.
In conclusion, our results reinforce the applicability of DBS for molecular analyses, such as HBV quantification and sequencing, using in-house techniques. In low-income settings and remote locations, where the gold-standard tests are not available, the use of DBS for molecular purpose may be a convenient and economical alternative, increasing the access to HBV diagnosis. Providing alternative diagnosis methods to low-income and hard-to-reach population, who are frequently unaware of their carrier status, may close important gaps in HBV control.

Material and methods
Participants. A total of 141 individuals were included in this study: Group 1 was composed by 92 HBsAg and HBV DNA reactive individuals in serum with HBV viral load greater than 2.5 IU/mL by COBAS ® TaqMan ® HBV (Roche Diagnostics, Branchburg, NJ, USA) or Abbott RealTime HBV (Abbott Diagnostics, Des Moines, USA). These subjects were recruited from ambulatories and laboratories related to the diagnosis and monitoring of viral hepatitis located in the states of Ceará and Rio de Janeiro of Brazil from 2011 to 2015. Group 2 (negative control) was composed by 49 healthy volunteers with no HBV or HCV serological markers (HBsAg, anti-HBc and anti-HCV) and with serum and DBS samples. These samples were collected in 2010 in hepatitis educational events and were part of the biorepository of the Laboratory of Viral Hepatitis of the Oswaldo Cruz Institute, with operation reviewed and approved by the institutional ethical committee on December 16, 2013. All individuals agreed to participate to the study by signing an informed consent. Sociodemographic and risk behaviour data were obtained through the application of a questionnaire. The study protocol was approved by the Research Ethics Committee of the Oswaldo Cruz Institute (number CAAE: 34055514.9.0000.5248), in accordance with the Declaration of Helsinki. Inclusion criteria were patients with more than 18 years old of any gender, race or ethnicity. Individuals with advanced cirrhosis (Child-Pugh B and C) and hepatocellular carcinoma were excluded from the study. All methods were performed in accordance with the relevant guidelines and regulations. HBV DNA extraction in serum and DBS. HBV DNA was extracted from serum samples using a commercial kit (High Pure Viral Nucleic Acid Kit, Roche Diagnostics, Mannhein, Germany) following manufacturer's instructions. The same kit was used for HBV DNA extraction from DBS, according to Bezerra et al. 22 . Thus, three circles of 3 mm of DBS were directly used in DNA extraction and all the steps recommended by manufacturer's were carried out.

DBS and serum samples.
HBV quantification. Serum samples were submitted to the commercial assay COBAS ® TaqMan HBV ® Test (Roche Diagnostics, Branchburg, NJ, USA) or to the commercial test Abbott Real Time HBV (Abbott Diagnostics, Des Moines, USA) for HBV-DNA quantification, according to the availability of the test and following the manufacturer's instructions. All results were presented in copies/ml for comparative purposes. Therefore, the quantification values obtained by COBAS ® TaqMan ® HBV and Abbott RealTime HBV needed to be converted for copies/mL. For COBAS ® TaqMan ® HBV, the value was multiplied by 5.82 (conversion factor) to obtain the result in copies/mL, while for Abbott RealTime HBV the value in UI/mL was multiplied by 3.41.
Serum and DBS samples were also quantified using in-house real-time PCR (qPCR) as previously described 6 . For analysis of serum and DBS samples, the DNA was added to a PCR mix in a concentration of 25 ng/μL. All samples were tested in duplicate.
A serial dilution panel was used to assess analytical sensitivity and to determine the limit of detection of the qPCR in DBS samples. Dilution panel was made as previously described 22 . DBS samples ranging from 10 0 to 10 7 copies/mL, as estimated in commercial quantitative method, were tested in duplicate.
The repeatability of qPCR was evaluated by testing eight times in the same reaction a HBV DBS sample with high viral load (high positive control, HPC, 3.41 × 10 10 copies/mL) and a HBV DBS sample with low viral load (low positive control, LPC, 3.41 × 10 4 copies/mL). The reproducibility was obtained by testing HPC and LPC eight times in the same reaction, twice a day, for three days by three different operators 46 .
To demonstrate the precision of qPCR, a DBS sample containing 4log10copies of HBV/mL, previously determined by commercial quantitative method, was tested 20 times in the same reaction 46 .
HBV DNA qualitative detection and S/Pol genes sequencing. HBV DNA from both serum and DBS samples were used for PCR amplification and sequencing of the overlapped small envelope and polymerase genes (S/Pol regions), producing a DNA fragment of ~ 900 bp, as previously described 23 .
The identity of the sequences was checked through the BLAST tool algorithm. Sequences representative of known genotypes and sub-genotypes were retrieved from GenBank and included in phylogenetic analysis. Phylogenetic trees were constructed with MEGA 7.0 software 47 using the Maximum Likelihood method and General Time Reversible substitution model with gamma distribution, as the best-fit model.
To assess the presence of mutations in both viral envelope and in the reverse transcriptase domain of viral polymerase, the sequences obtained were submitted to Geno2pheno HBV algorithm (Max-Planck-Institut für Informatik, Germany, at http:// hbv. geno2 pheno. org/ index. php). This algorithm also reports a trend in the sample's phenotypic resistance to five antiretroviral drugs. Data analysis. Descriptive statistical analysis was performed with calculation of means and standard deviation, with a preliminary assessment using contingency tables and respective statistics. Categorical variables were compared between groups using the chi-square test or Fisher's exact test, and continuous variables were analyzed using the Mann-Whitney U test. A p-value of < 0.05 was considered significant.
To evaluate the effectiveness of the in-house qPCR for DBS samples, we measured precision, reproducibility, repeatability according to each reaction, per day, by operator and total. In addition, sensitivity and specificity were also determined.
Concordance between the results obtained for the DBS and sera samples was assessed using the Kappa index (k). According to international standards, findings should be interpreted as follows: < 0.20 corresponds to poor agreement; 0.21-0.40 as fair agreement; 0.41-0.60 as moderate agreement; 0.61-0.80 as good agreement, and 0.81-1.00 corresponds to very good agreement.
HBV DNA viral load correlation was calculated using Pearson correlation test. P value < 0.05 was considered significant.
Ethical approval. The study protocol was approved by the Research Ethics Committee of the Oswaldo Cruz Institute (number CAAE: 34055514.9.0000.5248).