Immune Response of Hepatitis B Vaccine Among Persons With Diabetes: A systematic review of the literature

Electronic searches of MEDLINE (via PubMed), EMBASE (via Ovid), Cochrane Library, and Web of Knowledge databases were performed. The search terms consisted of (hepatitis b vaccin* OR hbv vaccin* OR hepatitis b immuni* OR hbv immuni*) AND (immunogeni* OR immune response OR antibody) AND (diabetes). The search terms also were used as Medical Subject Heading terms (MEDLINE) or key words (EMBASE), as applicable. Where possible, limits included publication date from 1986 through 30 April 2012, English language, and study of humans. The MEDLINE and EMBASE searches were limited to items containing abstracts. Studies conducted in the U.S. and other countries were eligible for inclusion.

Peer-reviewed, published randomized clinical trials or observational studies (all types) assessing response to hepatitis B vaccine among persons with type 1 or type 2 diabetes were included. Studies among both children and adults were included to ascertain the effect of age upon vaccine response among persons with diabetes. Reports had to specify the proportion (numerator and denominator) of subjects seroprotected by diabetes status (or data available that allowed for calculation of the proportions), or available odds ratios from a multivariate analysis, which included diabetes as a predictor variable. Studies of specific subject populations of which persons with diabetes represented a subset were included.

Studies were excluded when immune response using an antibody to hepatitis B surface antigen (anti-HBs) threshold of 10 mIU/mL was not reported; immune response was not measured between 0 and 6 months after the last vaccine dose (because anti-HBs titers wane over time); 10 or fewer subjects with diabetes were included; vaccine was administered intradermally; or subjects were not naïve to the vaccine or had positive serology for HBsAg, anti-HBs, or antibody to hepatitis B core antigen, indicating past or current hepatitis B infection. When two studies reported results for duplicate subjects, one study was excluded. Studies were not excluded when subjects received additional vaccine dose(s) because they did not achieve seroprotective concentrations or when inclusion of a fraction of subjects meeting an exclusion criterion was deemed unlikely to affect the findings. The determination as to inclusion or exclusion was based on the above criteria and was made irrespective of study results.

Data were manually extracted and electronically recorded. Studies were classified into one of three subject categories: adults (mean age >18 years), children (mean age ≤18 years), and hemodialysis/chronic kidney disease patients. Elements for extraction included study characteristics (e.g., study design, sample size, publication year, country); subject characteristics (e.g., age, diabetes type); vaccine characteristics (e.g., vaccine, dosage, route, schedule); and the proportion of subjects attaining seroprotection, including the time interval (in months) from the last dose at which immune response was measured. When vaccine or route of administration differed by study arm, extraction elements were recorded for each arm when possible.

An anti-HBs threshold of 10 mIU/mL after series completion (an accepted marker of immune protection) defined seroprotection (4). When multiple immune response measurements were reported reflecting seroprotection at different intervals during the primary vaccine series, the result after the final dose was recorded. When immune response was measured more than once after vaccination, the result from the interval closest to 1 to 2 months after the last dose (corresponding to the recommended interval for postvaccination testing for serologic response in the U.S. [4]) was recorded. When immune response was measured after a booster dose(s) or novel adjuvant administration, results were recorded separately and noted.

Odds ratios for attainment of seroprotection were recorded for studies reporting results from a multiple logistic regression model. When odds ratios and 95% CIs for the diabetes predictor variable were reported for failure to achieve seroprotection, the probability of attaining seroprotection was calculated by taking the inverse of the reported values.

The systematic electronic search yielded 225 studies (53 from MEDLINE, 105 from EMBASE, 20 from Cochrane Library, and 47 from Web of Knowledge; a list of these studies is available upon request). Of these 225 studies, 85 duplicates were identified. Abstracts from the remaining 140 studies (62.2%) were reviewed, and 94 (67.1%) were deemed not relevant. Forty-six (32.9%) full-text studies were retrieved and reviewed, and one additional duplicate was identified (Fig. 1). Fifteen studies did not fulfill inclusion criteria, and 13 met at least one criterion for exclusion (Table 2 and Fig. 1), leaving 17 studies.

The 17 studies included 2 randomized clinical trials, 12 prospective, and 3 retrospective studies published between 1989 and 2012 (Table 3). The 17 studies included 16,310 unique subjects, of which approximately 9,286 had diabetes. The subject categories consisted of adults for 4 studies (285 subjects, of whom 152 had diabetes); children for 4 studies (472 subjects, of whom 206 had diabetes); and hemodialysis/chronic kidney disease patients for 9 studies (15,553 subjects, of whom approximately 8,928 had diabetes). Subjects in the studies among adults and children generally lacked comorbidities (including kidney disease) other than diabetes. One study among hemodialysis/chronic kidney disease patients included only subjects with chronic kidney disease (i.e., no hemodialysis patients) (32), and three studies included both hemodialysis and chronic kidney disease patients (36,42,44), one of which also included peritoneal dialysis patients (44). The remainder were comprised entirely of hemodialysis patients. Seroprotection against hepatitis B virus infection was assessed in subjects by diabetes status in 16 studies (1,764 subjects, 633 with diabetes). Five studies included diabetes in a multivariate model (31,32,37,42,44). No serious vaccine-related adverse event was reported in any study.

The average age of subjects by study ranged from 8.4 to 79.5 years. Diabetes type was type 1 in five studies, type 2 in one study, and type 1 and type 2 in two studies. Diabetes type was not specified in eight of the nine studies among hemodialysis/chronic kidney disease patients; persons with type 2 diabetes constitute more than half of persons with diabetes starting dialysis in the U.S. (45). The vaccine type was exclusively recombinant in 15 studies, one of which used a combination hepatitis A/hepatitis B vaccine (43), and was exclusively plasma derived in one study; both recombinant and plasma-derived vaccines were used in one study. A novel adjuvant was administered to subjects in two studies (40,41). Vaccine dosage ranged from 3 to 40 µg. Route of administration was exclusively intramuscular in 15 studies and exclusively subcutaneous in one study; one study used intramuscular or intradermal administration routes in separate study arms. The standard 0-, 1-, and 6-month schedule was used exclusively or partially in 11 studies. Others were a 0-, 1-, 2-, and 12-month schedule and a 0-, 1-, 2-, and 6-month schedule (both U.S.-approved alternatives), as well as a 0-, 1-, and 2-month schedule (not approved in the U.S.). The sex distribution was generally equivalent for most studies, although men predominated in the studies among hemodialysis/chronic kidney disease patients. The type of vaccine, dosage, route of administration, schedule, and the interval after the last dose when seroprotection was assessed were equivalent for the diabetes and comparison groups except as noted in Table 4.

Seroprotection proportions ranged from 31.3–100.0% (median, 73.4%) among persons with diabetes and 35.2–100.0% (median, 87.1%) for those without diabetes. The proportion protected was generally greatest among children, ranging from 54.2–100.0% (median, 93.9%) among children with diabetes and 98.0–100.0% (median 100.0%) among those without diabetes. Among adults, seroprotection proportions ranged from 31.3–94.4% (median, 88.2%) for those with diabetes compared with 35.2–96.9% (median, 93.6%) for those without diabetes. Seroprotection proportions were lowest for hemodialysis/chronic kidney disease patients, ranging from 41.8–85.3% (median, 60.1%) for those with diabetes and 61.8–87.5% (median, 75.1%) for those without diabetes.

Seroprotection proportions among subjects with diabetes were generally lower when vaccine was administered on a 0-, 1-, and 2-month schedule (not approved in the U.S.) and greater when vaccine was administered on a 0-, 1-, 2-, and 12-month schedule (U.S.-approved alternative schedule) compared with the standard 0-, 1-, and 6-month schedule. Two studies using a 0-, 1-, and 2-month schedule reported relatively low seroprotection proportions for persons with diabetes in their corresponding subject categories (54.2% among children [35] and 45.8% among hemodialysis/chronic kidney disease patients [34]), and two studies using a 0-, 1-, 2-, and 12-month schedule reported relatively high seroprotection proportions among subjects with diabetes in their corresponding subject categories (88.6% and 94.4% [13] among adults and 66.7% among hemodialysis/chronic kidney disease patients [33]). When a fourth dose was administered to children with diabetes after a primary series on a 0-, 1-, and 2-month schedule, seroprotection increased from 54.2 to 100.0% in one study (35). Another study, however, achieved an 88.2% seroprotection proportion when vaccinating adults with diabetes subcutaneously using a 0-, 1-, and 2-month schedule with thymopentin as an adjuvant (41). Thymopentin is a synthetic pentapeptide with activity characteristic of the thymic hormone thymopoietin and is not approved for use in the U.S. It has been used to enhance interleukin 2 production and function of macrophages (46).

Administration of additional hepatitis B vaccine doses may improve immune response among adults with diabetes. Douvin et al. (13) reported seroprotective anti-HBs levels in 91.5% of adult subjects with diabetes compared with 75.0% as reported by Bouter et al. (14). Subjects in both studies received 20-μg dosages of recombinant hepatitis B vaccine intramuscularly. Subjects in the study by Douvin et al. were, on average, older than those in the study by Bouter et al. (mean age, 49.2 vs. 34.3 years). It is therefore reasonable that subjects from the study by Douvin et al. would have responded more poorly than those from the study by Bouter et al. because older age (6–9,31,47) is associated with a decline in response. However, differences in the number of doses administered likely contributed to higher seroprotective proportions in the Douvin et al. study, in which 30 subjects (42.3%) receiving vaccine on a 0-, 1-, 2-, and 12-month schedule also received a booster dose at month 4 (for an anti-HBs titer <10 mIU/mL, according to the study protocol). Therefore these subjects received five total doses, compared with three total doses on a 0-, 1-, and 6-month schedule reported by Bouter et al.

Five studies among hemodialysis/chronic kidney disease patients included diabetes in a multivariate model (31,32,37,42,44) (Table 5). All studies controlled for age, and three controlled for obesity. Four studies, with a total of 587 subjects (186 with diabetes), reported that diabetes was independently associated with failure to achieve seroprotection (odds ratio [OR] 0.23–0.50 [95% CI 0.09–0.96]) (31,32,42,44). In their study of 14,546 predominantly hemodialysis patients, Lacson et al. (37) reported no decrease in seroprotection associated with diabetes (OR 1.02 [95% CI 0.93–1.12]), although a subanalysis of 814 pairs (including subjects with and without diabetes) matched for age, sex, race, dialysis vintage, and dialysis modality reported a decrease in seroprotection associated with diabetes (OR 0.71 [95% CI 0.52–0.97]).

A lower seroprotection proportion was achieved among children with diabetes who received the vaccine using intradermal administration, which is not a U.S.-approved route of administration. Li Volti et al. (38) reported 77.8% seroprotection with intradermal administration compared with 100.0% seroprotection with intramuscular administration. Among children without diabetes, 100.0% in both intradermal and intramuscular arms achieved seroprotection (38). Other studies have found improved results with intradermal administration of the hepatitis B vaccine in dialysis patients (48).

Thymopentin or priming with tetanus toxoid (TT) was used in two studies to enhance the immune response to the hepatitis B vaccine. Thymopentin was administered to adults with diabetes three times per week for a week before hepatitis B vaccination and for 3 weeks afterward. The results were compared with hepatitis B vaccination among normal control subjects (41). TT was administered 2 days before an additional dose of hepatitis B vaccine to hemodialysis/chronic kidney disease patients who were nonresponders after 6 or 7 doses of hepatitis B vaccine that had been administered during the previous 18 months (40). Although differences in route of administration and number of booster doses administered prevent comparison with other studies, seroprotection was reported in 88.2% of adults with diabetes who received the thymopentin adjuvant and in 100.0% of previously nonresponding hemodialysis/chronic kidney disease patients with diabetes who had received TT before vaccination. Subsequent meta-analysis of studies using thymopentin adjuvant with hepatitis B vaccine in end-stage renal disease patients (with and without diabetes) found limited benefit from the addition of thymopentin to hepatitis B schedules (46).

When hepatitis B vaccination is performed in accordance with standard administration procedures, children and young adults with diabetes generally have responses similar to persons of comparable age without diabetes. Older adults have a reduced response, and older adults with diabetes seem to have further impairment in vaccine response, particularly those with coexisting kidney conditions.

There are several hypotheses about the biological basis for potentially impaired response to vaccination among persons with diabetes. Although persons with diabetes have appropriate humoral immune responses to vaccination (49), impaired cellular response may account for less robust antibody production after hepatitis B vaccination (14). Proposed explanations include a reduction in the number of circulating helper T cells, the CD4-to-CD8 lymphocyte ratio, and lymphocyte blastogenesis (23) and defects with antigen presentation (39). Impaired vaccine response also has been linked to the presence of DR3, DR7, and DQ2 human leukocyte antigen alleles among persons with diabetes (13). However, no association between seroprotection and glycemic control (13,14,30,39), duration of diabetes (13,30,39), insulin requirement (30,39), or microangiopathic complications (39) was demonstrated in studies included in this review.

Obesity is a major risk factor for type 2 diabetes; 53% of adults with diabetes are obese (50). Studies have demonstrated an association between obesity and a reduced response to hepatitis B vaccine in adults (51). A needle length that is inadequate to penetrate the deltoid fat pad and reach the muscle mass may account for the reduced immune response among persons with obesity (4,52). The less abundant blood supply in adipose tissue may delay antigen presentation to the B and T cells responsible for the immune response (52). However, the needle should not be so long that it involves the underlying bone (4).

In addition to obesity (52), older age (6–9,47), comorbid conditions (11), and medication use (12) have been associated with impaired vaccine response and may confound the relationship between diabetes and immune response. Four multivariate analyses among hemodialysis/chronic kidney disease patients included in this review, all of which controlled for age and three of which controlled for obesity, found diabetes to be an independent factor for impaired vaccine response (31,32,42,44). Lacson et al. (37), however, reported a null association with diabetes and seroprotection among hemodialysis/chronic kidney disease patients (N = 14,546; OR 1.02 [95% CI 0.93–1.12]) in their principal multivariate analysis.

Administration of additional doses of hepatitis B vaccine improves the proportion of persons responding to it. Additional doses have not caused unusual adverse reactions (16). It is recommended that health care personnel at continuing risk of hepatitis B exposure and infants with perinatal hepatitis B exposure repeat a primary series of hepatitis B vaccine if they are found to be nonresponsive to the initial series (4,5). Among health care personnel who did not respond after a primary series, 67.6% mounted a protective response up to three additional doses (11). Data exclusively on revaccination of nonresponding persons with diabetes are sparse. Although postvaccination serology to determine responses among adults with diabetes is not cost effective (Centers for Disease Control and Prevention, 2011, unpublished data) and not recommended, using schedules requiring four rather than three doses or administering additional doses of hepatitis B vaccine remains an option.

A longer interval between the final two doses of the primary hepatitis B vaccine series has been associated with higher final anti-HBs levels in general (4). Findings from this review suggest that a longer interval between the final two doses also is associated with increased seroprotection proportions among persons with diabetes. Excessively long intervals have the drawback of increasing the risk for acquisition of HBV infection among those with an incomplete series (4).

Other strategies that may augment the response to hepatitis B vaccine in persons with diabetes include the use of novel adjuvants, higher dosages, or combination vaccines. No hepatitis B vaccine with a novel adjuvant is currently licensed for use in the U.S. Recent prelicensure trials using a two-dose schedule with a hepatitis B vaccine containing a toll-like receptor 9 (HBsAg-1018 ISS) agonist as the adjuvant seems promising (53). Other strategies that could be explored in clinical trials include the use of the combined hepatitis A virus and HBV vaccine Twinrix (GlaxoSmithKline Biologicals) (54,55) or hemodialysis dosage of the hepatitis B vaccine (56). Because response declines with advancing age and comorbidities (11), vaccination soon after diabetes diagnosis likely would constitute the most practical way to optimize seroprotection.

This review has several limitations. It is possible that all relevant studies may not have been identified in the literature search, especially those studies in which persons with diabetes represented a subset of the population. There is the potential for publication bias. Studies with significant results are more likely to be published than those with null results, and this review consisted only of published studies. A null result for some studies in this review would consist of an equivalent vaccine response in persons with and without diabetes. It is therefore unlikely that publication bias would lead to a reported overestimate of vaccine response among persons with diabetes. Only 17 of the identified studies were included in this review (and only 5 multivariate analyses), and data pertaining to adults with diabetes but without hemodialysis/chronic kidney disease were sparse. Observational studies were included, and the results of these studies may not be as robust as randomized clinical trials. The matched comparison groups and objective outcome measure (i.e., anti-HBs levels) strengthen the data quality. Outcomes consisted of serological correlates of protection (anti-HBs levels), which may not correlate with maintaining full vaccine effectiveness over time. Immunocompetent persons found to have anti-HBs levels of ≥10 IU/L after the primary series, consistent with the threshold used by studies included in this review, are considered to be protected (4,5,57). Duration of protection was not assessed in this review, although other evidence suggests protection lasts for two decades in healthy primary vaccine responders. However, data on the duration of immune memory in immunocompromised persons are limited (4). Although general conclusions were drawn from a synthesis of the results, no attempt was made to test for statistically significant differences. A meta-regression was not conducted because colinearity likely exists among sources of heterogeneity (e.g., children most likely have type 1 diabetes and are treated with insulin). Heterogeneity between studies by diabetes type, management, glycemic control, and vaccine type, dosage, administration route, and schedule limit synthesis of results.

High levels of seroprotection from recombinant hepatitis B vaccine are achieved safely in children with diabetes. Data from studies reviewed here suggest that adults with diabetes have a reduced response to vaccination. Seroprotection is decreased among older adults and persons on hemodialysis in general, including those with diabetes. Administration of schedules using an extended interval to the final dose, four vs. three doses, or additional vaccine doses may achieve seroprotection in a greater proportion of adults with diabetes, including those who do not respond to an initial series. Other strategies to increase response among older adults with diabetes, including those with other comorbid conditions, should be explored.

No potential conflicts of interest relevant to this article were reported.

S.F.S. performed the literature review, abstracted data, and wrote the manuscript. P.R.S. critically reviewed and edited the manuscript. T.V.M. suggested the topic and critically reviewed and edited the manuscript. S.F.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

A portion of the data from five included studies was presented at a Meeting of the Advisory Committee on Immunization Practices, Atlanta, Georgia, 25–26 October 2011.

The authors thank Jack (Rick) Colbert, MLIS, Reference Librarian, CDC Public Health Library and Information Center, Atlanta Georgia, for his assistance with the literature search.