1 Introduction
Varicella zoster virus (VZV) causes varicella (chickenpox), a highly contagious disease that affects most people in the absence of vaccination [
1]. After infection, VZV remains dormant in sensory nerve ganglia, from where it may reactivate to cause herpes zoster (HZ; shingles) [
2]. The incidence of HZ has been estimated to range between 5.2 and 10.9 cases per 1000 person-years in those aged ≥ 50 years [
3]. Generally, the incidence increases with age [
3‐
6], which is primarily attributed to an age-related decline in immunity (immune senescence) and a higher likelihood of the presence of immunosuppressive conditions [
5]. HZ is characterised by a painful, localised rash and can be associated with serious complications of the nervous system such as ophthalmic HZ and post-herpetic neuralgia (PHN) [
2,
7‐
9].
To prevent HZ and its complications, adults aged ≥ 50 years may be vaccinated with a live attenuated HZ vaccine or the adjuvanted recombinant zoster vaccine (RZV; Shingrix, GSK, Belgium) [
10,
11]. Additionally, RZV can be used in adults aged ≥ 18 years who are at increased risk of HZ [
12]. RZV consists of a truncated form of the VZV glycoprotein E (gE) antigen adjuvanted with the AS01
B system [
10]. In adults aged ≥ 50 years, two doses of RZV had a vaccine efficacy against HZ ranging between 89.8 and 97.2%, depending on the participants’ age, as demonstrated in two large, parallel, phase III, randomised, observer-blind, controlled trials of RZV (ZOE-50 [NCT01165177] and ZOE-70 [NCT01165229]) [
13,
14]. Pooled safety data from ZOE-50/70 showed that RZV was more reactogenic than placebo (i.e., individuals vaccinated with RZV reported more injection site reactions and common systemic symptoms such as fever, myalgia, fatigue, chills, headache, or gastrointestinal symptoms than individuals vaccinated with placebo). The occurrence of serious adverse events (SAEs), potential immune-mediated diseases (pIMDs [
15]), and deaths was similar between vaccine and placebo recipients during the median 4 years of follow-up. Skin and subcutaneous tissue disorders (system organ class) and various types of cutaneous eruptions by Medical Dictionary for Regulatory Activities (MedDRA
®) preferred term (PT) also occurred at the same rate between vaccine and placebo recipients [
16]. Overall, these data demonstrated no safety concern related to RZV vaccination in adults aged ≥ 50 years [
13,
14,
16,
17]. Furthermore, clinical trials in adults with a compromised immune system, who are at increased risk of HZ, also did not identify safety concerns of RZV vaccination in this population [
18‐
21].
With the approval of RZV for immunisation of adults aged ≥ 50 years from October 2017 onwards [
12,
22,
23], GSK established enhanced post-marketing safety surveillance measures to promptly identify safety signals [
24], for which data were continuously shared with regulatory authorities. A review of the first 1.5 years of post-marketing safety surveillance data showed that the post-marketing safety profile of RZV was consistent with that previously observed in pre-licensure clinical trials and reflected in the RZV patient leaflet [
25].
Following the marketing of RZV in Germany in 2018, the Drug Commission of the German Medical Association (DCGMA) and Paul-Ehrlich-Institut (PEI) received reports of vesicular and bullous cutaneous eruptions, including blistering eruptions clinically compatible with HZ rash, that occurred in close temporal association with RZV vaccination. The DCGMA and PEI initiated a study to further investigate such reports [
26]. Here, the occurrence of vesicular and bullous cutaneous eruptions following vaccination with RZV was evaluated, based on the available post-marketing data comprising 2.5 years of spontaneously reported data.
4 Discussion
In this analysis, spontaneous reports of AEs suggestive of HZ or HZ complications and other (non-HZ) vesicular and bullous cutaneous eruptions that occurred following vaccination with RZV were assessed (Fig.
1a). It is important to note that cutaneous eruptions can have multiple aetiologies, such as viral infections, autoimmune diseases, vaccination, and allergic-type reactions, even when occurring in close temporal relation with vaccination. In particular, the elderly, one of the target populations for RZV, can show a marked susceptibility to dermatological disorders manifesting with eruptions. This is because of structural and physiological changes in their cutaneous membranes that may occur as a consequence of ageing and a lifetime of exposure to environmental and lifestyle factors and that can result in xerosis and reduced strength and elasticity [
35]. The elderly also have an increased risk of cutaneous drug reactions because of such comorbidities as well as polypharmacy, which can complicate accurate diagnosis and safety assessment of dermatological disorders reported in this patient group [
35,
36].
Our search in the GSK worldwide safety database using the MedDRA
® PTs indicative of LoE (ESM 2) identified 2423 reports of AEs suggestive of HZ or HZ complications. Among these, only two reports were laboratory-confirmed cases, and 643 reports were suspected cases based on the presence of HZ clinical symptoms and TTO post-vaccination. Conservatively, suspected cases were included in the analysis despite limited diagnostic certainty. Most confirmed and suspected vaccination failures (49%) were reported from Canada via a Canadian RZV social media page. These cases were not medically confirmed, and the provided clinical details and possibility for follow-up were very limited. The reporting rate for vaccination failures was low (2.0 cases per 100,000 RZV doses distributed), in line with the high efficacy of RZV in adults aged ≥50 years demonstrated in clinical trials [
13,
14,
16,
17]. A reporting rate of 1.37 cases per 100,000 RZV doses distributed for AEs suggestive of HZ or HZ complications that occurred in individuals who previously had an episode of HZ was observed. HZ recurrence incidence rates of 1–10 cases per 1000 person-years have been reported in immunocompetent populations [
7,
37,
38]. Together, the reporting rates observed in our analysis did not raise any safety concerns.
Our search using the MedDRA
® PTs suggestive of other (non-HZ) vesicular and bullous cutaneous eruptions (ESM 3) identified a total of 810 non-HZ reports (i.e., that were not co-reported with an LoE event). Most of the non-HZ reports that included sufficient information for full assessment were non-allergic and non-injection site localised eruptions. This may be explained by RZV being first licensed for immunisation of older adults, who have a higher prevalence of and susceptibility for dermatological disorders in general [
35,
36] and for hypersensitivity rashes including vesicular, blistering, or pustular rashes as clinical manifestations. Hypersensitivity rashes are rarely reported with vaccines in general and are also recognised and reflected in the patient leaflet for RZV [
10,
12]. Local injection site reactions secondary to the vaccine’s reactogenicity occurred in 74 reports and are also reflected in the patient leaflet for RZV [
10,
12]. Among the 19 AEs suggestive of other (non-HZ) vesicular and bullous cutaneous eruptions that were reported in the context of a pIMD diagnosis, four reports (one autoimmune bullous skin disease, one SJS, one psoriasis, and one systemic lupus erythematosus report) had a TTO of < 5 days. Studies have shown that pro-inflammatory cytokine production induced by the adjuvant system AS01 is transient, with a peak on day 1 and return to baseline by day 2–3, and localised at the injection site and draining lymph node [
39]. Based on this, it would likely take more than 5–7 days for symptoms of a new pIMD to manifest [
15], indicating that those four pIMDs were unlikely a consequence of the vaccination. In the remaining reports, the TTO was unknown, and/or no medical assessment was possible because clinical information was limited or because assessment of a causal relationship with vaccination was confounded and alternative aetiologies may be considered. Overall, the review of spontaneous reports of AEs suggestive of other (non-HZ) vesicular and bullous cutaneous eruptions following vaccination with RZV did not raise safety concerns.
There were 1928 reports among those identified using the MedDRA
® PTs indicative of LoE (ESM 2) that met criteria for possible VZV reactivations (TTO < 30 days or unknown after any vaccine dose). An O/E analysis demonstrated that, generally, the observed incidence of HZ cases following RZV vaccination was below the background incidence in the general population [
31‐
34], and this may be explained by the protection conferred by the vaccine or underreporting of cases. Worldwide and in the USA, the observed number of cases was higher than expected only for very low levels of RF (< 10%). In Germany and Canada, the observed number of cases was higher than the expected number for RFs < 56%. Estimates have indicated that between 50 and 81% of AEFIs are generally reported [
29], meaning that the RFs estimated for Germany and Canada are at the lower end of this range. Consequently, it is likely that the actual RFs for these countries are higher than the RF limits estimated in our analysis. There is some indication for a higher likelihood to be vaccinated with RZV within 30 days after an HZ episode (data on file; publication in preparation). Therefore, the risk of recurrent HZ may be higher than the background incidence rates identified from the literature, which decreases the O/E ratio, and the RF limits reported here may therefore be an overestimation. Overall, RF limits below which the observed number of cases was higher than the expected number were lower in the sensitivity analysis than in the main analysis. The risk period of the sensitivity analysis (starting 7 days post-vaccination) seems more reflective of an actual HZ episode from a pathophysiological perspective. More specifically, when VZV is reactivated, it is transported along microtubules within sensory axons to infect epithelial cells. At these sites, VZV replicates and causes the typical vesicular rash usually about 10–21 days after infection [
40]. Although very unlikely, if postulating that inflammatory cytokines, of which levels peaked 1 day following administration of a vaccine containing the adjuvant system AS01 in animal models [
39], could theoretically induce VZV reactivation, a period of more than 7 days would likely be needed for the apparition of HZ vesicular rash.
Two possible biological mechanisms have been hypothesised for VZV reactivation to occur following vaccination with RZV [
26]. The first hypothesis would be that vaccination with RZV may cause immune exhaustion, which is the result of chronic T-cell stimulation and causes suboptimal control of infections [
26,
41]. However, to date, such chronic stimulation has not been described following HZ disease or vaccination with RZV. On the contrary, it has been demonstrated that VZV-specific T cells that functioned as effector T cells at the peak of an HZ episode reverted to a polyfunctional memory phenotype within 4 months of recovery [
42]. Additionally, considering that gE is not expressed during VZV latency [
43] and that the gE antigen in RZV rapidly degrades following vaccination [
44], it is unlikely that chronic antigenic stimulation of gE-specific T cells occurs following vaccination with RZV. Furthermore, telomere shortening, which has been linked with immune exhaustion for several viruses [
45], has not been described for VZV-specific T cells; in contrast, one report demonstrated increased telomere size in the VZV-specific T cells of one patient [
46]. Lastly, in line with what has been shown for the live attenuated HZ vaccine [
47], RZV may mobilise a naive pool of T cells in addition to re-stimulating memory cells. Such naive cells are thus not impacted by previous VZV exposure or previous episodes of HZ. However, it has also been speculated that RZV vaccination during subclinical VZV reactivation could lead to symptomatic HZ because the excess gE might saturate T-cell-mediated immunity that would normally control such subclinical reactivation [
26]. The second hypothesis would be that massive inflammation and cytokine production in response to vaccination with RZV may increase the risk of VZV reactivation [
26]. The authors of a case report of HZ ophthalmicus following RZV vaccination in a patient who had had HZ ophthalmicus 20 years prior to the reported episode speculated that, since histopathological studies had found VZV DNA in the human cornea up to 8 years after the onset of HZ ophthalmicus, an immune response to the vaccine might trigger an immune response to the VZV DNA in the eye, leading to disease [
48] (although detection of VZV DNA does not imply presence of VZV gE). Pre-clinical studies have demonstrated that specific cytokines (i.e., interferon [IFN]-γ, IFNα, interleukin [IL]-6, and tumour necrosis factor-α) inhibit VZV replication and promote latency [
49,
50] and that the use of anti-nerve growth factor antibodies reactivates VZV [
51,
52]. A study in macaques showed that administration of gE adjuvanted with AS01 resulted in transient increases in IL-6 and low levels of IFNγ with a peak on day 1 [
53]. Similar data following administration of RZV in humans are currently not available; however, immunisation of adults with hepatitis B surface antigen adjuvanted with AS01
B induced a rapid and transient response of five (IL-6, IFNγ, IFNγ-induced protein 10, monocyte chemotactic protein 2, and macrophage inflammatory protein 1β) of 24 measured cytokines, all peaking between 12 and 24 h post-vaccination [
54]. This indicates that no signs of massive inflammation and cytokine production could be observed following immunisation with RZV and that it is very unlikely that an increase in cytokines could be the trigger for VZV reactivation and HZ. Overall, the O/E analysis and discussion of potential underlying pathophysiological mechanisms supported the medical review of the reports retrieved and did not raise safety concerns.
This analysis included the use of a large data set based on 32,597,779 distributed RZV doses from three countries where RZV was first marketed. As such, the analysis was based on a data set that was representative of the population currently exposed to RZV. This analysis had some limitations that are inherent to analyses using spontaneously reported data as a passive reporting system from a population of unknown size, e.g., underreporting, missing information, and misclassification [
55]. These limitations may have biased the analyses towards specific regions or patient groups and may have hindered medical review and assessment of the risk period. Furthermore, most identified reports of AEs suggestive of HZ or HZ complications included only clinical diagnosis without any confirmation using laboratory testing. As such, case identification could be biased because of a close temporal relation with RZV vaccination. To formally test whether RZV vaccination increased the frequency of VZV reactivations, an O/E analysis was performed. However, this O/E analysis had uncertainties surrounding the number of doses distributed, the true risk period, and the relevance of the background incidence rates extrapolated from the literature.