|Home | About | Journals | Submit | Contact Us | Français|
Most hepatitis delta virus(HDV) prevalence estimates from the United States are over 10 years old, and HDV has shown significant temporal variation in other populations. HDV/hepatitis B(HBV) dual infection progresses rapidly, has more complications, and a different treatment regimen than HBV infection alone. Accurate estimates of prevalence and risk factors are important to help clinicians decide who to screen.
Injection drug users(IDUs) in Baltimore, MD positive for HBV serologic markers were tested for hepatitis delta antibody(HDAb) at two time periods: 1988-1989(n= 194) and 2005-2006(n=258). Those HDAb+ in 2005-2006 plus a random sample of HDAb-, HBV+ participants were tested for HDV-RNA, HBV-DNA, and HCV-RNA. Characteristics associated with HDV exposure and viremia were identified.
HDV prevalence declined from 15% in 1988-1989 to 11% in 2005-2006. Among those with chronic HBV infection, prevalence increased from 29%(n=15/48) to 50%(n=19/38), p = 0.05. Visiting a shooting gallery was a strong correlate of HDAb positivity (relative risk=3.08, p=0.01). 8(32%) of those HDAb+ were HDV viremic. Viremic participants had elevated liver enzymes and more ER visits.
The temporal increase in HDV prevalence among those with chronic HBV infection is concerning; understanding this change should be a priority to prevent the burden from increasing.
Hepatitis delta virus (HDV) is a single-stranded, circular RNA virus unusual in its dependence on the presence of another virus, hepatitis B virus (HBV), for replication in an infected individual [1-3]. A dual HBV/HDV infection is associated with a more rapid progression and a higher rate of complications such as cirrhosis, end-stage liver disease, and hepatocellular carcinoma (HCC) than HBV infection alone [4-10]. A recent multi-center study found that patients dually infected with HDV and HBV had three times the risk of HCC and twice the rate of death due to cirrhosis compared to patients with HBV alone; a natural history study of viral hepatitis-related cirrhosis found that the median time to cirrhosis was 2 years [5, 6], in contrast hepatitis C (HCV)-related cirrhosis has a median onset time of approximately 20 years 
HDV is found throughout the world, but its prevalence, incidence, clinical features, and epidemiological characteristics vary by geographic region . HDV is endemic in many populations with a high prevalence of HBV, ranging from 70% among chronic HBV carriers in the Amazon basin to 20% in Africa to <1% in North America [12-17].
While the overall HDV prevalence in the United States is very low, it is elevated in populations at higher risk of HBV and other blood born infections [12, 13, 16-18]. However, even in sub-populations with high levels of HBV, considerable variation in HDV prevalence has been reported. Nath et al. found that the prevalence of HDV among blood donors in the United States differed significantly by region, ranging from 1.4% among chronic HBV carriers in the Southeast to 12.1% among chronic carriers in the Pacific region . Other studies have found rates varying from as low as 2% in hepatitis B surface antigen positive (HBsAg+) homosexual men to as high as 21% in female prostitutes [17, 18]. It has been over ten years since HDV prevalence was systematically measured in the United States. Furthermore, there is limited information available from populations in the United States to elucidate what places an individual at risk for contracting HDV.
Injection drug users (IDUs) have high rates of HBV compared to the general population, making them particularly susceptible to HDV infection . Chaotic lifestyles, limited healthcare access, and fear of retribution for drug use make IDUs difficult for researchers to capture, and few studies have measured the burden of HDV among IDUs in the United States. The goals of our study were to (1) characterize current HDV prevalence and changes in prevalence over time in the Baltimore IDU population, (2) identify correlates of HDV exposure and viremia, and (3) to characterize HBV, HDV, and HCV viral dynamics among those with HDV exposure.
The AIDS Linked to the Intravenous Experience (ALIVE) cohort is a longitudinal cohort study of 3,360 current and former IDUs (both HIV positive and HIV negative) in Baltimore, Maryland that has been operational for over 20 years. Participants are seen at semiannual visits which include HIV testing and collection of behavioral data, clinical data, and biological samples. Details of the ALIVE cohort have been provided elsewhere . This study was approved by the Johns Hopkins IRB and all participants gave informed consent.
We utilized a cross-sectional design and stratified random sampling scheme to estimate the prevalence of exposure to HDV among current and former IDUs in Baltimore at two time periods: between 1988 and 1989 (baseline), and between 2005 and 2006 (recent). Because the majority of the participants in the 2005-2006 cohort were recruited after 1994, this study is best described as a cross-sectional prevalence study of IDUs in Baltimore at two time points rather than a longitudinal study of the same patients over time. ALIVE participants were stratified into three categories according to their HBV serology: (1) hepatitis B surface antigen (HBsAg) positive > 6 months (chronically infected, ) (2) hepatitis B core antibody (anti-HBc) and hepatitis B surface antibody (anti-HBs) positive, and HBsAg negative (prior infection and immunity), and (3) anti-HBc positive only (likely previously exposed ). Those who were anti-HBs positive only were assumed to have been vaccinated for HBV and excluded from analysis (n = 76, 2.9% of 1988-1989 cohort and n = 62, 5.0% of 2005-2006 cohort). A random sample of patients was chosen from each HBV category at each time point by using a random number generator. Stored serum samples collected during the same visits at which HBV testing was performed were evaluated for hepatitis delta antibody (HDAb).
Testing for anti-HBc and anti-HBs was performed using the Abbott Corzyme immunoassay (Abbott Laboratories, Abbott Park, IL). The Abbott Auszyme immunoassay was used to detect HBsAg in the baseline samples (Abbott Laboratories, Abbott Park, IL), and the DiaSorin HBsAg assay was used for the recent samples (Dia-Sorin, Saluggia, Italy). A sensitivity study of commercial HBsAg assays found DiaSorin and Abbott to be comparable . All assays were performed according to the manufacturer's protocols. Reactive samples were confirmed through repeat testing; there were no discordant results. The baseline samples were tested for presence of HDAb using radioactive immunoassay (RIA), the standard assay available in 1994 when the testing was performed. The recent samples were tested using an enzyme-linked immunoassay (ELISA) for the determination of antibodies to hepatitis delta virus in sera (Diagnostic Bioprobes Srl, Milano, Italy). Validation studies of RIA and ELISA for detection of HDAb have shown that the ELISA has 100% specificity, but only 92% sensitivity when compared to RIA. The reduced sensitivity generally occurs in patients with no or low-level viral replication who have lower titers of antibody, as the RIA has a lower minimum limit of detection than the ELISA [23, 24]. To account for the use of different testing methods, we report prevalence results both unadjusted and adjusted for the difference in sensitivity between the tests. The adjusted results reflect what would be expected if RIA had been used at both time points.
All HDAb+ serum samples (n=25) from the recent (2005-2006) cohort were tested for HDV RNA (qualitative), HBV DNA (quantitative, HBV Roche COBAS Taqman HBV Anylate specific reagent), and HCV RNA (quantitative, Roche COBASE ampliprep/Taqman). A random sample of 25 HDAb- participants, frequency matched on HBsAg status (in other words, the proportion of HBsAg+ participants is the same among cases and matched controls), were also tested for HBV DNA and HCV RNA using quantitative methods. All tests were run in duplicate. Results are reported in international units.
Baseline demographics of the participants were compared between the participants from 1988-1989 and the 2005-2006. HDAb prevalence was calculated within each HBV serologic category and compared between the two time periods. For the 2005-2006 sample, difference in demographic and clinical factors including age (shown as mean and also categorized as 18-45, 46-60, and >60), gender, race (categorized as White Non-Hispanic, Black, Native American, or Other), HIV serostatus, hepatitis C antibody (HCV Ab) status, HCV RNA status (detectable HCV RNA versus no detectable HCV RNA), serum aspartate aminotransferase (AST) level, alanine aminotransferase (ALT) level, serum albumin level, and visit to an emergency room in the 6 months prior to study date (yes or no) were compared between those who had no evidence of exposure to HDV (HDAb-) and those who were exposed (HDAb+). The analysis was repeated comparing those who were HDV exposed but not viremic (HDAb+, HDV RNA-) to those who were viremic (HDV RNA+). HBV and HCV viral load values were log transformed and compared between (1) HDV RNA+ (viremic) and HDV RNA-, Ab+ (exposed, not viremic) and (2) Exposed, not viremic and HDAb- (not exposed). Categorical variables were compared using two-sided chi-square tests; continuous variables using two-sided t-tests.
To examine associations between drug use and sexual behaviors and HDV, we calculated the relative risk of HDV exposure (HDAb + test result), in those who reported each behavior in the 6 months prior to visit date compared to those who did not. We report 95% confidence intervals and p-values for each relative risk. These were performed on the 2005-2006 samples only, as more complete behavioral data was available for this cohort. Analyses examining sexual behavior were stratified by current injection drug use (yes or no). All data were analyzed using STATA statistical software version 11.0 (Stata Corp., College Station, TX).
Figure 1 depicts the number of participants eligible and sampled at each time point. A total of 243 participants who were HBV-positive at baseline (between 1988 and 1989) were randomly selected. Of these, 194 (79.8%) had serum available in the repository at the appropriate visit; which was tested for HDV. In the second cohort 258 of 259 randomly selected HBV+ participants had serum available for HDV testing. To evaluate whether the samples not available in the repository were missing at random, the demographic characteristics of participants with serum available were compared to those whose serum was unavailable. Within HBV strata, there were no statistically significant differences in the demographic or clinical characteristics between participants tested and participants missing serum, indicating the sera were missing at random (data not shown).
Table 1 compares the demographic characteristics of the two cohorts from each time-period. Participants from the baseline cohort were more likely to be younger, male, married, have an income, and be HIV-negative than participants from the current cohorts. The proportion of African-Americans and HCV Ab+ was similar between the two time periods.
HDV prevalence in 1988-1989 was 11% in previously HBV-exposed participants, and 29% in those chronically HBV-infected (Table 2). HDV prevalence decreased to 3% among HBV-exposed participants in 2005-2006, but rose to 50% in the chronically HBV-infected. When we calculated the 2005-2006 prevalence expected by RIA, we estimated an HDV prevalence of 55% among those with chronic HBV infection and 4% among previously exposed participants. The increase in prevalence among the chronically HBV-infected between the two time-points was statistically significant in both the uncorrected (p = 0.048), and corrected (p = 0.01) analyses.
In an analysis of drug use behaviors and HDV infection, visiting a shooting gallery was the only statistically significant correlate of HDV infection (Table 3a). No sexual behaviors were independently associated with HDV infection in an un-stratified analysis; however, in analyses stratified by active injection drug use, sexual behavior tended to be a stronger correlate in those reporting no injection drug use, and those who reported sex with an anonymous partner were significantly more likely to be HDAb + (RR = 4.48, 95 % CI: 1.23-16.35, p = 0.02).
Compared to those with no exposure to HDV (HDAb-), a lower proportion of HDV exposed (HDAb+) participants were Black (75% compared to 89.6%, p =0.03), and a higher proportion identified as White, Non-Hispanic or Other. Those with HDV exposure were more likely to be HIV+ (56% compared to 35.6%, p = 0.046). There were no statistically significant differences in age, gender, HCV Ab status, serum AST, ALT, or albumin level, or likelihood of an ER visit between HDV exposed and non-exposed.
Among the 25 participants who were HDAb+, 8 (32%) had detectable HDV RNA. Those with detectable HDV viremia were on average younger (mean age 41.1 years versus 49.9 years, p =0.004), with the majority between 18 and 45 years old. HDV viremic participants had significantly higher AST levels (125.7 U/L compared to 49.6 U/L for those without viremia, p = 0.03), ALT levels (132.3 U/L compared to 31.1 U/L, p = 0.009), and lower albumin levels (3.6 g/DL compared to 4.2 g/DL, p = 0.03). They were also significantly more likely to report at least 1 visit to the ER in the 6 months prior to study date (62.5% of HDV viremic versus 18.8% of HDV non-viremic, p= 0.03). All HDV viremic participants (100%) were HIV+, compared to only 35.3% of non-viremic participants (p = 0.002). Only 37.5% of those viremic for HDV had detectable HCV-RNA, compared to 76.5% of those who were non-viremic; however this difference was not statistically significant. None of the HDV viremic participants were female, compared to 31.3% of those non-viremic (p = 0.1).
Those viremic for HDV had a mean HBV-DNA level of 232,229,000 IU/mL compared to 700 IU/mL for those who were not viremic (p <0.001, table 5a). When the analysis was restricted to those who were HBsAg+ only, a significant difference between the two groups remained (p = 0.02). HCV viral load was slightly higher in those viremic for HDV (150,288,000 IU/mL compared to 3,246,000 IU/mL), but that difference was not statistically significant.
Participants who were exposed to HDV (HDAb+) but not viremic had significantly lower levels of HBV DNA than those with no exposure (700 IU/mL compared to 245,887,000 IU/mL, p = 0.007, table 5b). This difference persisted even when the analysis was restricted to those who were HBsAg+ (1,300 IU/mL compared to 312,087,000 IU/mL, p = 0.01). HCV viral load was slightly lower in the HDV exposed, non-viremic group (3,246,000 IU/mL compared to 38,869,000 IU/mL among those not exposed to HDV), but the difference was not statistically significant.
Our study found a higher prevalence of HDV among IDUs with chronic HBV infection than has previously been reported in the United States, with 50% infected in 2005-2006. The majority of prior studies were performed in health-care settings, which may lead to an underestimation of HDV prevalence by excluding individuals with limited access to healthcare, given that such individuals may be at increased risk of acquiring HDV. The most recent estimate of HDV infection among HBsAg positive individuals comes from NHANES 2003-2004, in which participants who were HBsAg positive and anti-HBc positive were tested for HDAb. Only 1 of 28 (3.6%) were HDAb positive , suggesting that the higher HDV prevalence may be concentrated in the IDU population.
Visiting a shooting gallery (a location where people gather to inject illegal drugs) emerged as the most important behavioral risk factor for HDV infection. Previous studies suggest that HDV is more efficiently spread parenterally than sexually . Our study corroborates these findings. However, there was evidence to suggest that high-risk sexual behavior might be important among infrequent or non-IDUs. More studies are needed to better understand the role of sexual transmission in HDV infection. However, our data are concerning as sexual transmission could facilitate the spread of HDV from IDUs to their sexual partners and subsequently to the general population.
Participants viremic for HDV had significantly higher levels of AST, and ALT, lower levels of albumin, and increased likelihood of a recent ER visit. These differences may be partially explained by HIV, which was present in all HDV viremic participants compared to only 31.1% of non-viremic participants. Progression to chronic HBV after acute infection is largely related to the immune status of the host. HIV infection is associated with greater likelihood of progression to HBV chronicity , and this may be true for HDV as well. Unfortunately the cross-sectional nature of our study does not allow for determination of whether or not HIV infection predated HBV/HDV infection.
We found that participants who were exposed to HDV but were not viremic had significantly lower HBV viral load levels compared to both HDV- patients and HDV viremic patients. It is possible this group is mostly comprised of participants who were coinfected with HDV (they acquired both HDV and HBV at the same time, an event associated with clearance of both viruses the vast majority of the time) . The lowered HCV viral load among HDV viremic participants corroborates previous findings that HCV replication is inhibited in the presence of HDV [30, 31]. However, HCV viral load was not uniformly lowered among HDV viremic participants and in fact some had very high viral loads, as such careful monitoring of hepatitis viral levels is necessary in HDV patients.
One limitation of our study was the use of different HDAb testing methods at the two testing points. Based on the results of a previous study comparing HDAb RIA to ELISA in 1000 participants, we were able to adjust for the differences in sensitivity between the assays. The authors of the validation study noted that the discrepancy was predominately seen in patients with little or no active viral replication . This is most likely to affect patients who have already cleared both viruses because antibody titers wane over time and may eventually become undetectable. In this setting, antibody detection will depend on the sensitivity of the assay used. Patients with persistent HDV infection will maintain high levels of anti-HDV antibodies because of chronic antigenic stimulation . In this scenario, anti-HDV will be positive using both methods. Therefore we believe that the increase among chronic carriers is likely to be accurate. In previously exposed patients, some individuals negative by ELISA may have been positive by RIA. This would suggest that if incorrect, the observed increase HDAb prevalence in previously HBV-exposed individuals may actually be an underestimate. Furthermore, the difference in sensitivity of 8% observed between ELISA and RIA was seen in patients who were all HBsAg positive. This difference may be greater at lower antibody titers likely to be seen in HBsAg negative patients, meaning that even the adjusted ELISA estimate in the previously HBV-exposed participants may be an underestimate of the true HDV prevalence. A further limitation of our study is the use of behavioral data measured at time of testing rather than time of exposure. We cannot exclude the possibility that infection with HDV or HBV might change an individual's behavior, thus introducing a systematic bias into our results. Misclassification of this type would bias results towards the null; as such our behavioral risk factor analysis should be interpreted with this in mind.
Our findings underscore the importance of HBV vaccination, as it can prevent both HBV and HDV infections. Rates of HBV vaccination as measured by serology (HBsAb positive, cAb and sAg negative) were very low (2.9% in 1988-1989 and 5.0% in 2005-2006). As such efforts should be made to expand HBV vaccination coverage to at risk populations such as IDUs. It is possible that the low prevalence of HBsAb+ only serology is not entirely explained by decreased access to the vaccine in this population, but might also reflect decreased efficacy due to high rates of HIV . It is also possible that HBV infection predated vaccination for many in this cohort. Further research is needed to determine true levels of access and efficacy of the HBV vaccine in IDU populations.
Increasing prevalence of HDV among IDUs has potentially important clinical implications. HDV/HBV co-infection is generally more severe than HBV mono-infection in the acute setting, with a higher risk of fulminant hepatitis [29, 33]. In the setting of super-infection and chronicity, HDV has been shown to suppress HBV replication, potentially confounding clinicians because significant liver disease often occurs despite low or undetectable levels of HBV DNA . This can lead to delayed diagnosis, which is particularly concerning given the more aggressive nature of HBV/HDV chronic infection [4-6]. In addition, oral nucleoside/tide analogue therapy, standard treatment for HBV, is ineffective in the presence of HDV, leaving high-dose prolonged interferon therapy as the only option . Our data suggest that all patients infected with HBV with a history of past or present IDU should be screened for HDV. Careful screening combined with further studies to better understand this change in epidemiology should be a priority in order to reduce the significant morbidity associated with HDV infection and to prevent the prevalence from rising further.
Financial Support: HDAb ELISA kits were donated by International Immunodiagnostics (CA). Associated laboratory expenses were supported by the Johns Hopkins Bloomberg Infectious Disease Program Laboratory. This research was partially supported by the Intramural Research Program of the NIDDK/NIH. The ALIVE study was supported by NIDA grants R01DA04334 and R01DA12568
Conflicts of Interest: No authors have any conflicts of interest to report
Human Subjects: This study was approved by the Johns Hopkins IRB and all participants gave informed consent.