Immunogenicity of Inactivated Varicella Zoster Vaccine in Autologous Hematopoietic Stem Cell Transplant Recipients and Patients With Solid or Hematologic Cancer

Abstract Background In phase 3 trials, inactivated varicella zoster virus (VZV) vaccine (ZVIN) was well tolerated and efficacious against herpes zoster (HZ) in autologous hematopoietic stem cell transplant (auto-HSCT) recipients and patients with solid tumor malignancies receiving chemotherapy (STMc) but did not reduce HZ incidence in patients with hematologic malignancies (HMs). Here, we describe ZVIN immunogenicity from these studies. Methods Patients were randomized to ZVIN or placebo (4 doses). Immunogenicity was assessed by glycoprotein enzyme-linked immunosorbent assay (gpELISA) and VZV interferon (IFN)-γ enzyme-linked immunospot (ELISPOT) assay in patients receiving all 4 doses without developing HZ at the time of blood sampling. Results Estimated geometric mean fold rise ratios (ZVIN/placebo) by gpELISA and IFN-y ELISPOT ~28 days post–dose 4 were 2.02 (95% confidence interval [CI], 1.53–2.67) and 5.41 (95% CI, 3.60–8.12) in auto-HSCT recipients; 1.88 (95% CI, 1.79–1.98) and 2.10 (95% CI, 1.69–2.62) in patients with STMc; and not assessed and 2.35 (95% CI, 1.81–3.05) in patients with HM. Conclusions ZVIN immunogenicity was directionally consistent with clinical efficacy in auto-HSCT recipients and patients with STMc even though HZ protection and VZV immunity were not statistically correlated. Despite a lack of clinical efficacy in patients with HM, ZVIN immunogenicity was observed in this population. Immunological results did not predict vaccine efficacy in these 3 populations. Clinical trial registration NCT01229267, NCT01254630.

Cell-mediated immunity plays a critical role in the containment of varicella zoster virus (VZV), preventing the reactivation of VZV and subsequent onset of herpes zoster (HZ) [1]. Immunocompromised individuals, such as patients who have undergone autologous hematopoietic stem cell transplant (auto-HSCT) or patients with malignancies, are at ~3-18-fold increased risk of HZ infection compared with immunocompetent patients, depending on the nature of the underlying condition [2][3][4][5][6]. The reported incidence of HZ in auto-HSCT recipients, despite antiviral prophylaxis, ranges from 62 of 1000 person-years (PYs), based on a large retrospective analysis [4], to 92 of 1000 PYs, based on a recent phase 3 randomized clinical trial [7]. For patients with solid tumor malignancies receiving chemotherapy (STMc), reports of HZ incidence range from 15 of 1000 PYs [3] to 19 of 1000 PYs [8]. For patients with hematologic malignancies (HMs), HZ incidence is reported to be 31 of 1000 PYs [2,6,8]. In comparison, HZ incidence in the the general adult population is 5 of 1000 PYs [9].
The live attenuated VZV vaccine is contraindicated in immunocompromised patients [15]; therefore, an inactivated VZV vaccine (ZV IN ) was investigated as a preventive option for immunocompromised patients. Proof-of-concept studies and a phase 1 trial using a heat-treated ZV IN demonstrated immunogenicity and safety in auto-HSCT recipients and patients with STMc or HM through 28 days post-dose 4 following a 4-dose regimen administered ~30 days apart [16][17][18]. Subsequently, phase 1 and 2 trials using ZV IN inactivated by gamma irradiation confirmed immunogenicity and safety in patients with HM receiving anti-CD20 monoclonal antibodies and in adults with autoimmune disease receiving immunosuppressive therapy, respectively [19,20].
Primary safety and efficacy results from 2 phase 3 trials (V212-001 and V212-011) demonstrated that ZV IN was well tolerated, with the incidence of HZ and HZ-related complications significantly reduced in auto-HSCT recipients and patients with STMc but not in patients with HM [7,8]. In auto-HSCT recipients, the estimated vaccine efficacy of ZV IN against HZ (VE HZ ) was 63.8% (95% confidence interval [CI], 48.4%-74.6%) [7]; in patients with STMc, the estimated VE HZ was 63.6% (97.5% CI, 36.4%-79.1%) [8]. Immunogenicity was assessed as an exploratory end point in these 2 phase 3 trials, with the results presented here.

Trial Designs
V212-001 (NCT01229267) and V212-011 (NCT01254630) were phase 3, randomized, double-blind, placebo-controlled multicenter trials that evaluated the safety, tolerability, efficacy, and immunogenicity of ZV IN for the prevention of HZ and HZ-related complications in auto-HSCT recipients (V212-001) and in patients with STMc or HM (V212-011). V212-001 was conducted between December 2010 and December 2015; V212-011 was conducted between June 2011 and April 2017. Ethical approval was obtained from the institutional review board at each trial site, and written informed consent was obtained from each participant before trial entry. The V212-001 and V212-011 protocols have been previously described [7,8]. The studies were conducted in accordance with the principles of Good Clinical Practice. Patients were monitored for clinical signs and symptoms of HZ and HZ-related complications through the entire trial period. HZ was diagnosed primarily by polymerase chain reaction [7,8].

Trial Population
Trials included males and females aged 18 years or older with a history of varicella infection or seropositivity for VZV antibody. V212-001 included participants scheduled to receive auto-HSCT for treatment of lymphoma, other malignancies, or any nonmalignant conditions within 60 days of enrollment. The exclusion criteria in V212-001 included underlying malignancies other than Hodgkin lymphoma associated with >2 disease relapses, planned tandem transplantation, and intended antiviral prophylaxis for more than 6 months after transplantation. V212-011 included patients with STMc or HM who were not likely to undergo HSCT and who were receiving a cytotoxic or immunosuppressive chemotherapy regimen. Patients with HM who were ≥50 years of age and not in remission were eligible, regardless of whether they were receiving chemotherapy. Common exclusion criteria in both trials were history of HZ within 1 year of enrollment and prior or expected receipt of any VZV vaccine. The exclusion criteria in V212-011 included current/expected receipt of long-term (>4 weeks) antiviral prophylaxis against HSV, VZV, or CMV.

Treatment Administration
In these trials, patients were randomly allocated to receive gamma-irradiated ZV IN or placebo, administered in a 4-dose regimen ~30 days apart [7,8]. For more details on the vaccine, see the Supplementary Methods. Auto-HSCT recipients received dose 1 ~30 days (60 to 5 days) before HSCT. Doses 2 through 4 were administered 30, 60, and 90 days after HSCT. Both patients with STMc and patients with HM received dose 1 of ZV IN or placebo at the time of enrollment (day 1). Doses 2 through 4 were administered ~30 days after each previous dose. Among patients receiving cyclic chemotherapy, dose 1 of ZV IN or placebo was administered ~5 days before any chemotherapy dose in the cycle. Doses 2 through 4 were administered ~20 to 40 days after the previous dose of vaccine or placebo; specifically, ZV IN or placebo had to be administered ~5 days before the upcoming chemotherapy dose. To complete the studies, patients had to have completed the studies' close-out questionnaires at the end of the safety follow-up period; these were administered over the phone. In V212-011, the HM group was discontinued due to statistical evidence of futility shown at a planned interim analysis.

Immunogenicity Measurements
Immunogenicity analyses were exploratory (no prespecified statistical hypotheses were tested) and were conducted in the per-protocol immunogenicity population. In both trials, the per-protocol immunogenicity population included patients who received all 4 doses and did not have HZ before blood sampling. For patients who received treatments interfering with measurements of VZV-specific antibody response (including those receiving immunoglobulin therapy) or who received medications interfering with B-cell function, the measurements at corresponding time points and thereafter were excluded from the immunogenicity analysis conducted by glycoprotein enzyme-linked immunosorbent assay (gpELISA). Patients who received immunoglobulin therapies were included in the immunogenicity analysis conducted by VZV interferon (IFN)-γ enzyme-linked immunospot (IFN-γ ELISPOT) assay because they did not interfere with T-cell function.
VZV-specific antibody responses were measured by gpELISA assay [21] in auto-HSCT recipients and in patients with STMc. This antibody assay was not conducted in patients with HM because the nature of the disease and treatments could have biased the test results, based on the results from previous phase 1 gpELISA data in patients with HM [17]. Cell-mediated immune responses were measured by VZV IFN-γ ELISPOT assay [22] in subsets of auto-HSCT recipients, patients with STMc, and patients with HM. The VZV ELISPOT assay detected IFN-γ-secreting cells from peripheral blood mononuclear cells stimulated with VZV before and after vaccination. For immune responses measured by gpELISA, end points were the GMT and geometric mean fold rise (GMFR). For immune responses measured by VZV IFN-γ ELISPOT assay, end points were geometric mean count (GMC) and GMFR.

Statistical Analyses
A linear mixed longitudinal model was used on the natural logtransformed antibody titers for the comparison of GMTs between ZV IN and placebo recipients across the time points after vaccination. This longitudinal regression approach allowed for comparison of postvaccination antibody titers between the groups, adjusting for prevaccination antibody titer in the presence of incomplete data [23]. The model incorporated treatment group, visit, age (for V212-001: <50 vs ≥ 50 years; for V212-011: continuous variable), and treatment group-by-visit interaction. The fold-differences between the ZV IN and placebo recipients and the corresponding 95% CIs at the visits were obtained from this model. A Cox regression model was used for ZV IN and placebo recipients, with immune responses measured by gpELISA at prespecified time points as covariates, to evaluate the association between immune responses and risk of HZ. The GMT and GMFR were also summarized at these time points by treatment group and HZ outcome (patients who developed confirmed HZ during the trial vs patients who did not).
A time-varying Cox proportional hazards model was used to estimate the relationship between HZ occurrence and gpELISA titers among ZV IN and placebo recipients. The gpELISA titers were used as the time-dependent covariate to obtain a risk ratio for HZ per unit increase in the titer. Using the natural log-scale of GMC, analyses of the comparison of GMC between treatment groups and evaluation of the association between GMC and HZ risk were performed similarly to the analyses and evaluations for gpELISA.
The most common primary diagnoses were myeloma among auto-HSCT recipients, breast cancer among patients with STMc, and chronic lymphocytic leukemia among patients with HM (Supplementary Table 1) [7,8]. The most common concomitant medications included systemic antibacterial agents in auto-HSCT recipients, antineoplastic agents in patients with STMc, and analgesics in patients with HM (Supplementary Table 1) [8].

VZV-Specific Cell-Mediated Response by IFN-y Enzyme-Linked Immunospot Assay
The observed GMC values in the ZV IN and placebo groups at baseline and at time points examined in auto-HSCT recipients and patients with STMc and HM are shown in Figure 2 [7,8].

Association Between Immune Response and Risk of Herpes Zoster
Post-dose 4 gpELISA and VZV IFN-γ ELISPOT assay results were available for a small number of ZV IN recipients who subsequently developed HZ (Tables 3 and 4). Among auto-HSCT recipients who received ZV IN , gpELISA GMT values at ~28 days post-dose 4 were lower in the group that developed  HZ than in the group that did not, although CIs were broad and overlapping (Table 3). In patients with STMc, GMT values at ~28 days post-dose 4 were generally similar among those who did and did not develop HZ (Table 3). For both the auto-HSCT recipients and patients with HM who received ZV IN , VZV IFN-γ ELISPOT assay GMC values at ~28 days post-dose 4 were lower in the group that developed HZ than the group that did not, albeit with overlapping CIs (Table 4). Surprisingly, patients with STMc who developed HZ had high VZV IFN-γ ELISPOT assay GMC values at ~28 days post-dose 4, which were higher than GMC values among patients with STMc who did not develop HZ (Table 4). A statistical correlation was not  found between HZ protection and VZV gpELISA response in auto-HSCT recipients and patients with STMc (Table 5). In addition, no correlation was found by VZV IFN-γ ELISPOT assay response among patients with STMc (Table 5).

DISCUSSION
Defects in T-cell immunity increase the risk for HZ [1]. The gpELISA [21], which measures T-cell-dependent antibody responses, was shown in clinical studies of zoster vaccine to correlate with protection against HZ in healthy adults aged 50 years and older [24,25]. At the time the phase 3 studies of ZV IN were conducted, it was unknown if the same relationship between gpELISA and VE HZ would be seen in immunocompromised patients. Therefore, gpELISA and VZV IFN-γ ELISPOT assay-a direct measure of T-cell immunity [22] [17]. With respect to VZV-specific cell-mediated responses measured by the IFN-γ ELISPOT assay, ZV IN elicited a ~2-5-fold higher estimated GMFR ratio between ZV IN and placebo at ~28 days post-dose 4 across different immunocompromised populations. Among auto-HSCT recipients, the estimated GMFR ratio between treatment groups remained high at 1 and 2 years post-dose 4 (4.12 and 3.32, respectively). These results were similar to those observed in auto-HSCT recipients, patients with STMc (n = 56), and patients with HM (n = 60) enrolled in V212-002 [17].
The immunogenicity findings of ZV IN described here are consistent with those observed with zoster vaccine live in a nonimmunocompromised patient population. In the zoster vaccine live shingles prevention study, conducted in 38 546 patients aged ≥60 years, cell-mediated immunity, assessed by IFN-γ ELISPOT assay, and humoral immunity, assessed by gpELISA, were significantly increased in patients receiving live attenuated VZV vaccine measured at 6 weeks after vaccination compared with those receiving placebo (IFN-γ ELISPOT: 70.1 vs 31.7 GMC; gpELISA: 471.3 vs 292.3 GMT), and the increase in cell-mediated immunity persisted for 3 years of follow-up [26]. Similarly, in the zoster vaccine live efficacy and safety trial, conducted in 22 439 patients 50-59 years of age, humoral immunity, assessed by gpELISA, was significantly increased, with a GMFR of 2.3 [27]. In both studies, a specific level for any immune response that was predictive of protection against HZ was not identified. The efficacy results from trials V212-001 and V212-011 demonstrated ZV IN vaccine efficacy in the prevention of HZ and HZ-related complications in auto-HSCT recipients and patients with STMc but not in patients with HM [7,8]. The immunogenicity data presented here support an immune mechanism for ZV IN vaccine efficacy in auto-HSCT recipients and patients with STMc, although a statistical correlation between immunogenicity and protection against HZ was not demonstrated. In fact, the IFN-γ ELISPOT assay GMCs observed ~28 days post-dose 4 in patients with STMc who received ZV IN and subsequently developed HZ were higher than those observed in patients with STMc in the ZV IN group who did not develop HZ. However, in this population that may have received chemotherapy after receiving 4 doses of ZV IN , measurement of immunity at ~28 days post-dose 4 may not accurately reflect immunity before the development of HZ. Waning of cell-mediated immunity due to chemotherapy could subsequently increase HZ risk, which is thought to depend on cell-mediated immunity at the time that latent VZV reactivates [1]. Interestingly, in patients with HM, positive VZV-specific cell-mediated responses measured by IFN-γ ELISPOT assay were observed in the ZV IN group but did not translate into ZV IN vaccine efficacy. These unexpected findings point to the importance of assessing vaccine clinical efficacy, as well as immunogenicity, to understand the protective potential of investigational HZ vaccines or established HZ vaccines administered to new patient populations. It is possible that immunogenicity measured by the assays reported here may not predict vaccine efficacy in certain immunocompromised patient populations or for all vaccines.
Long-term immunogenicity data assessed at 1 and 2 years post-dose 4 in auto-HSCT recipients revealed a decline in humoral immunity in the ZV IN group, while cell-mediated immunity remained sustained. VZV-specific antibody responses by gpELISA in placebo-treated auto-HSCT recipients at 1 to 3 years post-dose 4 were lower than baseline values, a phenomenon previously seen in phase 1 testing [17], due to the highly immunosuppressive nature of the transplantation procedure performed after the baseline time point. As expected, VZVspecific cell-mediated immunity in the placebo group improved over the 2-year follow-up period of this study, as immunity has been shown to be restored following auto-HSCT procedures [28]. ZV IN   A limitation of the V212-011 trial is that humoral immunity and cell-mediated immunity were not measured at 1 year postdose 4; therefore, long-term immunogenicity data in patients with STMc and HM are not available. Although these are the first large vaccine trials conducted in these patient groups, the number of patients enrolled in both the V212-001 and V212-011 trials who were eligible to be included in the immunogenicity analysis was small. This precluded a robust analysis of the association between immune response and the risk of HZ. In trial V212-001, most exclusions from the immunogenicity analyses were due to use of prohibited concomitant medications, as expected in this patient population. Another limitation is that the Cox proportional hazards model used for the immunogenicity analyses did not include postrandomization use of immunosuppression. One key difference of the transplant population studied in the current trial vs the trials in immunocompetent patients is the relatively high proportion of patients with "nonvalid assays" in the ELISPOT assay (37 of 188 ZV IN recipients [20%] and 39 of 182 placebo recipients [21%]) (Supplementary Figure 1). The observed effect may be due to the high level of immunosuppression, which is a well-known phenomenon among transplant recipients [29][30][31].
In summary, these 2 phase 3 trials demonstrated that ZV IN elicits cellular immune responses when measured by IFN-γ ELISPOT assay in the 3 immunocompromised populations examined: auto-HSCT recipients, patients with STMc, and patients with HM. Although our immunogenicity data are consistent with the ZV IN clinical efficacy in HZ prevention in auto-HSCT recipients and patients with STMc, demonstrated in the same studies, the immunogenicity data we report in patients with HM did not translate into clinical efficacy in HZ prevention.

Supplementary Data
Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.