The immunogenicity of standard intramuscular (IM) influenza vaccine is suboptimal in transplant recipients. Also, recent studies suggest that alloantibody may be upregulated due to vaccination. We evaluated a novel high-dose intradermal (ID) vaccine strategy. In conjunction, we assessed alloimmunity. Transplant recipients were randomized to receive IM or high-dose ID vaccine. Strain-specific serology and HLA alloantibody production was determined pre- and postimmunization. In 212 evaluable patients (105 IM, 107 ID), seroprotection to H1N1, H3N2 and B strains was 70.5%, 63.8% and 52.4% in the IM group, and 71.0%, 70.1%, 63.6% in the ID group (p = ns). Seroconversion to ≥1 antigen was 46.7% and 51.4% in the IM and ID groups respectively (p = 0.49). Response was more likely in those ≥6 months posttransplant (53.2% vs. 19.2%; p = 0.001). Use of mycophenolate mofetil was inversely associated with vaccine response in a dose-dependent manner (p < 0.001). Certain organ subgroups had higher response rates for influenza B in the ID vaccine group. Differences in anti-HLA antibody production were detected in only 3/212(1.4%) patients with no clinical consequences. High-dose intradermal vaccine is an alternative to standard vaccine and has potential enhanced immunogenicity in certain subgroups. In this large cohort, we also show that seasonal influenza does not result in significant alloantibody production.
Solid organ transplant (SOT) recipients have a high risk of morbidity and mortality from influenza infection [1, 2]. In addition to increased rates of lower respiratory infection with influenza, studies have also reported increased acute and chronic allograft rejection after infection, specifically in lung and kidney transplant recipients [1, 3-5]. Therefore, SOT patients are recommended to receive the annual trivalent inactivated influenza vaccine [6, 7]. However, immunogenicity is suboptimal as compared to immunocompetent persons . Responses may depend on several factors including time from transplantation, use of certain immunosuppressives such as mycophenolate mofetil, as well as type of transplant.
Various methods to increase immunogenicity of influenza vaccine in immunocompromised populations have been proposed such as the use of higher doses of vaccine, booster doses within the same season, and the addition of adjuvants. The intradermal (ID) route of delivery has also been proposed to increase immunogenicity since the dermis contains larger number of dendritic cells that in turn, act as potent antigen presenting cells. However, the amount of antigen that can be used is limited by the volume of vaccine that can be delivered intradermally. Low-dose intradermal immunization (3–6 μg antigen) has previously been compared to standard-dose intramuscular (IM) vaccination in lung transplant recipients and showed equivalent, albeit low immunogenicity . Recently, a more concentrated trivalent vaccine specifically for intradermal use has become available (Intanza, Sanofi-Pasteur, Canada) as 9 μg and 15 μg formulations. However, there are no published studies of the immunogenicity of high-dose ID vaccine in the transplant population or whether this vaccine has superior humoral responses compared to intramuscular vaccine. Also, more recently, the safety of influenza vaccine in transplant patients has been questioned due to concerns of vaccine induced alloreactivity leading to increased antibodies directed against donor human leukocyte antigen (HLA) .
Therefore, we conducted a prospective, randomized controlled trial designed to compare the safety and immunogenicity of IM versus high-dose ID influenza vaccine in transplant recipients. We hypothesized that due to its delivery directly into the dermis, high-dose ID vaccine would have improved immunogenicity in organ transplant patients. We also determined whether new HLA antibodies were formed after immunization with either vaccine.
Patient population and study design
The study was conducted at the University of Alberta, Canada after receiving institutional research ethics board approval. All patients provided written informed consent. Adult solid organ transplant recipients attending outpatient clinics were eligible if they were at least 3 months post transplant and had not yet received the 2010–11 influenza vaccine. In order to include stable outpatients, patients were excluded from the study if they had a history of biopsy proven or clinically treated rejection in the 30 days prior to transplant. Participants were screened for eligibility in August and September 2010 and randomized between October and December 2010. The study was registered at www.clinicaltrials.gov number NCT01180699.
Randomization and masking
At enrollment, patients were randomized to receive either standard dose IM or high-dose ID seasonal influenza vaccine in a 1:1 ratio. Randomization was done using a computer-generated schedule blinded to study investigators, in blocks of four to ensure equal numbers. Once the patient consent was obtained, the treatment assignment was provided by the study coordinating office. Since the study vaccines differed in appearance and were administered by different routes, both patients and the study team member administering the vaccine were unblinded to vaccine allocation. However, the outcomes assessor was masked to treatment allocation (observer blinded). Both vaccines contained the same three influenza antigens: influenza A/California/7/2009 (H1N1)-like virus, influenza A/Perth/16/2009 (H3N2)-like virus and influenza B/Brisbane/60/2008 (Victoria lineage). The IM vaccine was Vaxigrip® (Sanofi-Pasteur, Canada) containing 15μg antigen of each strain in 0.5 mL volume, and was injected into the deltoid muscle of the nondominant arm. The ID vaccine was Intanza® (Sanofi-Pasteur, Canada) and contained 9 μg antigen per strain in 0.1 mL volume as a prefilled syringe. Two doses of this vaccine (cumulative dose of 18 μg antigen per strain) were given in succession as an ID injection in the deltoid area of the nondominant arm. The vaccines did not contain an adjuvant.
Sera were collected pre- and 1 month postvaccination for strain-specific antibody testing and HLA antibody analysis and frozen at −80°C until the day of testing. Demographics, type of transplant and immunosuppression data were collected. Adverse events data were collected by those not aware of vaccine allocation. Safety data were collected at the following postimmunization times: 24 h, 48 h, 7 days, 1 month and 6 months. All adverse effects were graded as none, mild (no interference in daily activities), moderate (some interference in daily activities) and severe (patient unable to participate in activities of daily living). Patients were followed for a total of 6 months from enrollment for the development of influenza infection and biopsy-proven allograft rejection. For sample size calculation, previous studies have suggested response rates may range from 15% to 93% . Our study had sufficient power to detect a 20% increase in response rate with high-dose ID vaccine assuming a baseline response of 40% in the intramuscular arm.
Sera were stored at −80°C until the day of analysis. The samples underwent hemagglutination inhibition assay (HAI) for each of the three strains in the seasonal vaccine . The laboratory staff performing the assay was blinded to vaccine allocation. The test was performed at the Microbiology Services, Colindale, Health Protection Agency, UK using standard methods previously described . Briefly, serum was incubated with receptor-destroying enzyme (RDE II, Denka Seiken Ltd., Japan) to remove any nonspecific inhibitors of hemagglutination from the sera. Serially diluted sera were then incubated with influenza virus (containing 4 hemagglutination units of virus) followed by addition of a suspension of blood cells (0.5% turkey red blood cells for the H1N1 and B strains and 0.5% guinea pig red blood cells for the H3N2 strain). Titers were determined by doubling dilutions of antibody. Sera were tested in duplicate using an initial dilution of 1:10 and a final dilution of 1:2048. Antibody concentrations that were below the lower limit of detection (<10) were assigned a titer of 5 for the purposes of analysis.
All pre- and postvaccination samples were screened for HLA alloantibodies. Initial testing was performed using Lifecodes LifeScreen Deluxe (Gen-Probe, San Diego, CA, USA) as per manufacturer instructions. The screen was designed to detect IgG antibodies to HLA class I and II antigens using coated microparticle beads. To calculate whether the sample was positive for bound alloantibody, the median fluorescent intensity (MFI) of each bead was compared to the signal intensity of the negative control beads within the same preparation. Samples that were negative for alloantibody prior to vaccination and were positive postvaccination underwent specificity testing with the Luminex Single Antigen Bead Assay (OneLambda, Canoga Park, CA, USA). This assay allows for the detection of alloantibodies against specific HLA antigens. For purposes of the study, we defined a positive specimen (including de novo donor specific antibody (DSA)) as having a MFI > 1000 and a change in MFI > 20% from pre to postimmunization [12, 13]. Twenty percent was selected as there is significant assay variability from one batch to another [14, 15].
Our primary endpoint was defined as seroconversion to at least one of three influenza vaccine antigens between IM and high dose ID vaccine. All definitions were based on EMEA/CPMP criteria for use in influenza trials . Seroconversion was defined as a greater than or equal to fourfold rise in titer from baseline for any of the three influenza vaccine antigens. All analyses were performed using IBM SPSS version 19.0 (Chicago, IL, USA) and GraphPad Prism version 4.0 (La Jolla, CA, USA).
Secondary outcomes included seroconversion factor (SCF) [derived by dividing the postimmunization titer by the prevaccine titer]; seroprotection to a strain [defined as a strain-specific postvaccination titer ≥ 1:40]; and, geometric mean fold rise (GMFR) [calculated as the geometric mean of the seroconversion factor]. In addition, we evaluated rates of rates of microbiologically confirmed influenza infection. Significant rejection was defined as biopsy-proven and/or clinically treated. Protocol biopsies are not routine at our center; therefore, asymptomatic rejection was not evaluated. To further examine factors associated with seroconversion between the two vaccine groups, certain a priori defined secondary analyses were performed to determine the most significant factors affecting vaccine response. Factors analyzed were patient age, time from transplant (less than or greater than 6 months from transplant), type of organ transplanted and immunosuppression. For the analysis of immunosuppression, medications at the time of vaccination were used. Each immunosuppressive was analyzed as a yes/no variable. For cyclosporine, tacrolimus, sirolimus, drug levels at the time of vaccination were correlated with response. With MMF, dosing was correlated with vaccine response. In addition, a multivariate model was constructed using variables that had a p-value < 0.2 on univariate analysis. Multivariate analysis was performed using logistic regression. Statistical significance was defined as a p < 0.05.
Role of the funding source
The funding source (CIHR) had no role in design, conduct, analysis, writing of the manuscript, or decision to submit.
During October 2010–December 2010, we randomized 229 organ transplant recipients to ID (n = 114) and IM (n = 115) vaccines. Baseline characteristics of the cohort were similar between groups and are detailed in Table 1. The most common types of transplant were kidney (n = 94, 41.0%), lung (n = 74, 32.3%), liver (n = 26, 11.4%) and heart (n = 18, 7.9%). The overall median time from transplant to vaccination was 4.9 (0.3–32.4) years. Maintenance immunosuppression as well as the number of patients that received antithymocyte globulin in the 6 months prior to enrolment were similar between the IM and ID groups. The majority of patients reported receiving influenza vaccine in prior years although documentation of having received vaccine in the prior year was available for 86 patients (of these 52.3% and 47.7% received IM and ID vaccines respectively); however, since patients were randomly assigned to each group, prior receipt of vaccine should be equally distributed.
Table 1. Baseline characteristics of the study population
Of the 229 randomized patients, 17 patients did not have either a pre- or postvaccination sera (Figure 1). For the intention to treat analysis, these patients were assumed not to have reached seroconversion. Seroconversion to at least one antigen was 55/114 (48.2%) versus 49/115 (42.6%) in the ID and IM groups respectively (p = 0.47). There was no significant difference in the distribution of patients not having sera measurements between the ID and IM groups (p = 0.46). Therefore for the remaining analysis, only patients that had a measurable endpoint (n = 212) were included (107 ID, 105 IM). The seroconversion rate to influenza antigens was low regardless of vaccine type. Seroconversion to at least one antigen was 51.4% and 46.7% in the ID and IM groups respectively (OR 95%CI 1.20 (0.70–2.08) p = 0.49). There were no significant differences between groups for seroconversion to two or three vaccine antigens (Figure 2). Overall, no significant differences in immunogenicity between the two cohorts were seen (Table 2). However, there was a trend toward greater geometric mean titers (GMTs) in the ID group for influenza B (p = 0.074). Postimmunization seroprotection to H1N1, H3N2 and B strains was 71.0%, 70.1%, 63.6% and 70.5%, 63.8% and 52.4% in the ID and IM group respectively (p = 0.93, 0.33 and 0.10).
Table 2. Vaccine immunogenicity by hemagglutination inhibition assay after 2010–2011 seasonal influenza vaccination
Intradermal group (n = 107)
Intramuscular group (n = 105)
GMT = geometric mean titer.
GMT (95% CI)
Seroprotection rate (%)
Seroconversion rate (%)
Geometric mean seroconversion factor
Baseline seroprotection (prior to vaccination) to A/H1N1, A/H3N2 and B was present in 42.9%, 42.9% and 36.3% respectively, and was not significantly different in the two groups. If those seroprotected at baseline were excluded, after ID and IM vaccination seroconversion rates to A/H1N1 were 48.4% and 45.8% (p = 0.77), to A/H3N2 40.4% and 32.8% (p = 0.39), and to influenza B 31.8% and 21.7% (p = 0.19), respectively.
We also analyzed factors that influenced seroprotection. Seroprotection to A/H1N1 was significantly lower in patients receiving ≥ 2 g daily of MMF (37.3% vs. 62.7%; p = 0.019), lung transplant recipients (28% vs. 72%; p = 0.016) and those less than 6 months from transplant (8.7% vs. 91.3%; p = 0.013). Seroprotection to B strain was significantly lower in those receiving ≥2 g daily of MMF (36.6% vs. 63.4%; p = 0.042), and those less than 6 months from transplant (6.5% vs. 93.5%; p = 0.003). No factors were found to significantly influence seroprotection to A/H3N2.
Three patients developed microbiologically documented influenza A infections (two A/H3N2 and one unknown subtype) between 1 and 6 months after immunization (2 intradermal group and 1 intramuscular group). All three participants were double lung transplant recipients who had been transplanted more than 1 year previously. Two patients had achieved seroprotective titers to A/H3N2 after immunization. No patient was seroprotected against A/H1N1.
Response and type of organ transplant
Seroconversion to at least one vaccine antigen was seen in 46.5%, 75%, 41.7% and 58.8% of kidney, liver, lung and heart recipients respectively. Lung transplant recipients (70/212) had the lowest response to vaccine (p = 0.12 vs. other transplant types). In a subgroup analysis of nonlung transplants (n = 142, comprising primarily of liver and kidney patients), recipients of the ID vaccine had significantly greater geometric mean titers (49.1 vs. 30.0, p = 0.036), seroprotection rates (70.4% vs. 52.1%, p = 0.025) and seroconversion factors (2.65 vs. 1.71, p = 0.011) to influenza B.
Other factors affecting vaccine response
Using the whole cohort (n = 212), further analysis was performed to determine factors that might affect vaccine response (Table 3). On univariate analysis, vaccine response (seroconversion to at least one vaccine antigen) was lower for either group if the immunization was performed when patients were less than 6 months posttransplant (19.2% vs. 53.2%, p = 0.001). Also GMTs for all three vaccine strains were significantly lower in those vaccinated at less than 6 months posttransplant compared to those vaccinated more than 6 months posttransplant (A/H1N1, p = 0.044; A/H3N2 p<0.001; B, p = 0.015). Upon further analysis of time from transplant, timepoints after 6 months (i.e. 6–12 months, 12–24 months, 24 months–5 years and > 5 years) had no significant effect on vaccine immunogenicity. In addition, vaccine response had a strong inverse correlation with total daily dose of mycophenolate mofetil (MMF) (Figure 3). Multivariate analysis showed that MMF dosing had a significant effect on vaccine response (OR 0.77 (0.66–0.91) for each 500 mg increment in MMF dose, p = 0.002).
Table 3. Factors associated with vaccine responsea
We screened 212 pre- and postvaccination sera for HLA antibody. Of these, 24 (11.3%) patients with a negative screening test prior to vaccination developed a newly positive screen (Table 4). The screen was positive for class I antibody only (9/24), class II antibody only (14/24) and both class I and II (1/24). One patient of 24 had biopsy-proven rejection at the 6-month follow-up. Further specificity testing in this cohort using the single antigen bead method confirmed the presence of anti-HLA antibody in 3/24 (12.5%) patients (1 kidney, 1 kidney-pancreas and 1 liver; 2 in IM and 1 in ID group). Two of these patients had preexisting nondonor specific antibody that increased and one had preexisting DSA that increased (Table 5). The remaining 21 patients did not show any defined antibody specificities. Graft function was stable in all patients at 6 months postvaccination.
Table 4. Characteristics of study patients that showed newly positive HLA screening test after immunization
N = 24
Median age (years; range)
Time from transplant to vaccination (years; range)
Class I only
Class II only
Class I and II
Table 5. Details of patients that showed significant increases in alloantibody postvaccination
Vaccine safety was assessed at regular intervals up to 6 months postvaccination in 228/229 patients. Within the first 7 days of immunization, there was no significant difference in rates of systemic adverse effects between ID and IM vaccine except gastrointestinal symptoms such as nausea and diarrhea which were more common with the ID vaccine. A significantly higher rate of local adverse events was seen with ID vaccine. These included erythema (p < 0.001), induration (p < 0.001), tenderness (p < 0.001) and pruritus (p = 0.005). No hospitalizations were a direct result of immunization. By 6 months postvaccination 12/228 (5.3%) patients had been diagnosed with mild rejection on biopsy (5 in the IM group and 7 in the ID group) and of these, 5 were treated with augmented immunosuppression. At 6 months posttransplant, At 6 months postvaccination, 3 patients had died (2 of cardiovascular events and one of recurrence of primary liver disease), all in the ID group.
Since traditional influenza vaccines have suboptimal immunogenicity in transplant patients, it is important to assess novel approaches to vaccination in this group of patients. We report the first randomized trial comparing high doses of intradermally administered vaccine compared to standard IM vaccine. Our trial was powered to detect a 20% difference in seroconversion which we felt would be clinically important. We found that in the overall cohort, seroconversion rates to both vaccines were low, and there were no significant differences in vaccine response (defined as seroconversion to at least one vaccine antigen) between the two vaccine groups (54.1% in ID and 46.7% in IM). Both vaccines were well tolerated although an increased frequency of local adverse events was observed with ID vaccine. Significant anti-HLA antibody production secondary to vaccination was not observed with either vaccine.
Higher doses of intradermal vaccine have only been evaluated in immunocompetent persons. In transplant patients, low dose intradermal vaccines (3–6 μg) have not demonstrated significant benefit when compared to standard vaccination, or when administered as a booster dose following IM vaccination [8, 17]. Studies in immunocompetent people ages 18–59 have demonstrated that seroconversion rates with 15 μg of intradermal antigen range from 56–79% for each specific vaccine strain [18-20]. In contrast, in our study, seroconversion to each specific vaccine strain following ID vaccination ranged from 21.5% to 37.4%. In a study of immunocompetent patients ≥ 60 years of age, 15 μg of intradermal vaccine resulted in higher antibody titer and seroprotection rates compared with intramuscular vaccine . Two patients in the study developed influenza infection despite achieving seroprotection. We hypothesize this may be due to lack of development of cellular immunity which also likely plays an important role in influenza infection.
We analyzed factors associated with a poor vaccine response. Time from transplant < 6 months and daily MMF and prednisone doses were significant factors on univariate analysis for poor vaccine responses. However, in multivariate analysis, only MMF dose remained significant and demonstrated a strong dose–response relationship. The mechanism of action of MMF includes potent inhibition of B-cell proliferation . The association of vaccine response with time from transplant likely reflects the greater intensity of immunosuppression in the first 6 months. These findings support the current recommendation to administer influenza vaccine starting after 3–6 months posttransplant when patients are often on lower immunosuppressive doses . Previous studies have also suggested consistently low immunogenicity of influenza vaccine in lung transplant recipients [8, 17]. We observed this in our study as well, and also noted that all three vaccinated patients who developed influenza infection during the follow-up were lung transplant recipients. In an ‘a priori’ planned analysis by organ type we observed that intradermal vaccine may have greater immunogenicity in the remaining (nonlung) transplant recipients. This was statistically significant for the influenza B strain with higher seroprotection, seroconversion factor and antibody titers postvaccination. Overall, the B strain in influenza vaccine is known to be less immunogenic than influenza A vaccine strains, a finding also observed in our study . These findings may have more importance to future quadrivalent influenza vaccines that will contain two B strains.
Vaccine safety was also evaluated in our study. Mild gastrointestinal side effects were slightly more common with ID vaccine as were local reactions. Local reactions are known be more common with ID vaccine with up 29% pruritus and 71.9% erythema reported for immunocompetent adults age ≥ 60 years receiving a 15 μg dose . Overall, our rates of local reactions were lower than reported in immunocompetent patients perhaps due to the blunting effect of immunosuppression. Several recent studies have questioned the safety of influenza vaccine in transplant recipients due to their potential to trigger alloreactivity [9, 24]. A proposed mechanism for this is molecular mimicry whereby vaccine antigens induce cross-reactive allo-immune responses. In a study of two cohorts of kidney transplant recipients (n = 92 and 59) two doses of ASO3-adjuvanted pandemic H1N1 vaccine were associated with a 17.3% and 11.9% rate of alloantibody development . Similarly, a case–control study of 60 heart transplant recipients demonstrated a higher rate of biopsy proven rejection after adjuvanted influenza vaccine . In our study the incidence of confirmed HLA alloantibody development was extremely low (3/212; 1.4%) and could not be attributed to any clinical consequences. Other patients that were newly screen positive displayed nonspecific reactivity but no clear specificity pattern. Nonspecific reactivity can be found commonly in the healthy population . In transplantation, one study showed that de novo DSA was seen in up to 15% of kidney transplant patients followed over a mean of 4.6 years . Given the large sample size, this study provides important reassurance that nonadjuvanted seasonal influenza vaccine does not appear to result in significant allo-reactivity regardless of the route of administration.
Our study has a number of limitations. First, in order to deliver a sufficient dose of intradermal vaccine, we used two successive doses of the 9 μg preparation of ID vaccine (the only available dose in Canada at that time). This may have increased the rate of local reactions observed in our study. The dose of vaccine also differs slightly between the two groups (18 μg ID vs. 15 μg IM); the reasoning for the dose selection has been discussed above. Given no significant differences in overall immunogenicity were observed, it is unlikely these limitations affected the primary outcome of the study. We also enrolled all types of organ transplants which may have created a heterogeneous population. Although transplant recipients at our center are on similar maintenance immunosuppression, groups such as lung transplant recipients are generally much more immunosuppressed compared to liver and kidney recipients. This was the rationale for the planned analysis of immunogenicity by organ group. We did not specifically collect data on rejection within the 6 months prior to enrolment; however, we included only patients that had been on stable immunosuppression and did not have ongoing treatment for or a diagnosis of rejection in the past 30 days. We did not conduct a study of clinical protection against influenza. However, immunogenicity is widely used as a surrogate marker for protection as well as for annual licensure of influenza vaccines. Our study results may not apply to patients with hematopoietic stem cell transplants and those on immunosuppressive agents other than those included in this study.
In summary, this is the first study to evaluate high-dose ID vaccine in organ transplant recipients. We show comparable immunogenicity to standard IM vaccine. We also show in the largest cohort to date that either vaccine did not lead to a significant rise in donor specific HLA-antibodies. These are important findings for the transplant community since intradermal vaccine is available and may be preferred by patients due to ease of administration. Our study shows that the intradermal vaccine has similar immunogenicity to standard IM vaccine. Based on this, we suggest that high-dose ID vaccine is an alternative to standard-dose IM vaccine. Potentially improved immunogenicity observed in certain subgroups merits further evaluation.
This study is funded by a grant from the Canadian Institutes of Health Research. Both vaccines used in this study were kindly provided by Sanofi-Pasteur, Canada. Sanofi-Pasteur had no role in study design, conduct, analysis, or publication of this study. A.E. is supported by the Swiss National Fund Grant PBBSP3-130963.
The authors of this manuscript have conflicts of interest to disclose as described by the American Journal of Transplantation: