Herpes Zoster Caused by Vaccine-Strain Varicella Zoster Virus in an Immunocompetent Recipient of Zoster Vaccine

Abstract

We report the first laboratory-documented case of herpes zoster caused by the attenuated varicella zoster virus (VZV) contained in Zostavax in a 68-year-old immunocompetent adult with strong evidence of prior wild-type VZV infection. The complete genome sequence of the isolate revealed that the strain carried 15 of 42 (36%) recognized varicella vaccine–associated single-nucleotide polymorphisms, including all 5 of the fixed vaccine markers present in nearly all of the strains in the vaccine. The case of herpes zoster was relatively mild and resolved without complications.

Zoster vaccine (Zostavax) was licensed in the United States in 2006 and is currently recommended for adults aged 60 years and older. The vaccine comprises the same live-attenuated Oka seed stock of varicella zoster virus (VZV) used for the monovalent varicella vaccine used in the United States (Varivax), but the infectious dose is at least 14 times higher. Just as wild-type VZV can establish latent infection in dorsal root ganglia following primary varicella infection in susceptible hosts and subsequently reactivate decades later to cause herpes zoster (HZ), latency and HZ due to the Oka strain of VZV (Oka VZV) can occur following varicella vaccination, although the risks are considerably lower [1–3]. In contrast, almost all zoster vaccine recipients have had primary wild-type VZV infection earlier in life, with establishment of latency in dorsal root ganglia and a robust immune response that protects against a second episode of varicella. The ability of Oka VZV in zoster vaccine to establish latency and reactivate as HZ is unknown when zoster vaccine is administered to immunocompetent older adults with prior wild-type VZV infection. We report a case of HZ caused by Oka VZV in a 68-year-old woman vaccinated with zoster vaccine who had strong evidence of prior wild-type VZV infection.

CASE REPORT

The case patient was identified as part of a large prospective HZ study in progress at Kaiser Permanente Southern California (KPSC). She was a 68-year-old Hispanic woman who received zoster vaccine in her left deltoid region in March 2012. She presented to the outpatient clinic in December 2012 with a 3-day history of rash, mild but annoying pain, and burning sensation on her left middle chest and middle back and under the left upper arm the day before visit. The linear pattern was consistent with T1 dermatomal distribution. She reported no prodromal pain but had experienced more hot flashes lately. Her past medical history revealed no immunosuppressive conditions or medications, and she had an uncertain history of varicella. Her doctor diagnosed HZ and she was treated with acyclovir (800 mg 5 times a day for 10 days) and acetaminophen for pain. Both rash and pain resolved within 10 days, and the patient experienced no complications.

The National VZV Laboratory of the Centers for Disease Control and Prevention performs VZV strain identification for specimens submitted from patients with clinically diagnosed HZ at KPSC. Two swab specimens of lesion fluid/basal cells and 2 slide scrapings were collected and shipped dry at ambient temperature. Isolates were identified as strains originating from Oka vaccine via FRET real-time polymerase chain reaction, an approach that discriminates vaccine from wild-type VZV by melt curve analysis; 4 different assays targeting single-nucleotide polymorphisms (SNPs) in open reading frames (ORFs) 38, 54, and 62 each distinguish vaccine and wild-type VZV as described elsewhere [4–7]. The fixed ORF62 SNP at positions 106262 and 107252 definitively identified this isolate as Oka vaccine. In accordance with VZV nomenclature parameters established in 2010, this isolate has been named VZVs/Pasadena.USA/48/12/Z [2] (GenBank accession number KF811485 [8]).

The complete sequence data for this isolate revealed no evidence for mixed markers across the entire genome, indicating that the isolate is clonal—that is, a single vaccine-associated strain was responsible for this case (Figure 1).

Figure 1.

Profile of varicella vaccine–associated single-nucleotide polymorphisms (SNPs) for isolate varicella zoster vaccines/Pasadena.USA/48.12/Z [2]. Dumas strain is a clade 1 VZV isolate; the complete genome sequence of this virus serves as the primary reference strain for the numerical loci used in this publication (GenBank accession number XO4370). pOka is the parental strain from which the Biken, GlaxoSmithKline (Varilrix), and Merck (Varivax) vaccine preparations are derived; it is a clade 2 VZV strain. The complete genome sequence for VZVs/Pasadena.USA/48.12/Z [2] was determined using a long polymerase chain reaction method developed in-house. Sixteen amplicons overlapping by 500–1000 bp were amplified and sequenced using conventional Sanger sequencing. The genome termini were sequenced by ligating viral genomes to a circular form, amplifying the region across the junction and sequencing it. The detailed protocol is being prepared for publication. Abbreviations: NCR, noncoding region; ORF, open reading frame; POS, position; SNP, single-nucleotide polymorphism; WT, wild type.

The isolate carried 15 of 42 (36%) vaccine SNPs identified by Gomi et al and carried all 4 fixed vaccine SNPs in ORF62 (105705, 106262, 107252, and 108111) as well as the single fixed SNP located in ORF0 (noncoding region, position 560) [9]. Nine SNPs occurring as mixed markers (strains carrying either the vaccine or wild-type base are present in the vaccine) in Zostavax were present as the wild-type marker in VZVs/Pasadena.USA/48.12/Z [2], 5 of which were nonsynonymous, leading to an amino acid substitution in their respective encoded protein (5745 [ORF6], 31732 [ORF21], 101089 [ORF59], 105310 [ORF62], and 111650 [ORF64]) (Figure 1). Limited evaluation of the amino acid sequence changes revealed that none of the changes was predicted to alter protein function using the Provean software [10].

Three vaccine-associated SNPs that appeared to be fixed in Merck Oka carried the wild-type nucleotide in VZVs/Pasadena.USA/49.12/Z [2] (703 [ORF1], 105 356, and 107136 [ORF62]). All of these were synonymous changes resulting in no differences in the respective proteins. Finally, there were 9 SNPs that were unique to isolate VZVs/Pasadena.USA/48.12/Z [2], located in ORF 11 (15865), ORF 17 (24651), ORF28 (50382), ORF33/33.5 (61029), ORF34 (63723), ORF44 (80966), ORF62 (106650 and 108505), and a noncoding region between ORF61 and ORF62 (104971); 4 of these led to an amino acid substitution (Figure 1: 15865, 24651, 63723, and 108505) These changes were also not predicted to alter protein function [10].

DISCUSSION

Similar to wild-type VZV, vaccine Oka (vOka) is capable of establishing latency in neurons following varicella vaccination and reactivating at a later time to cause HZ. This phenomenon has been well documented in both healthy and immunocompromised individuals, although HZ due to vOka appears to be uncommon [1, 11]. We provide evidence indicating that vOka found in zoster vaccine can establish latency and reactivate to cause clinical zoster.

One limitation of our report is that we were unable to confirm a past primary infection with varicella either by history or presence of VZV-specific antibodies prior to receipt of zoster vaccine. Our patient was 68 years old, was born in California, and had spent her entire life in the United States, and was likely exposed to VZV for many decades while varicella was circulating widely prior to introduction of routine varicella vaccination for children. Rates of VZV infection in this cohort of US residents are virtually 100% [12]. In addition, the patient lived with a large number of siblings during childhood, and recalled that several of them had experienced varicella, further reducing the likelihood that she was never infected with wild-type VZV. Therefore, it is likely that this case represents Oka strain superinfection in the face of prior wild-type immunity. We cannot rule out the unlikely possibility that our case represented a rare VZV-naive 68-year-old person, with zoster vaccine serving as her primary VZV infection.

Our findings also have implications in terms of zoster vaccine safety. To date, >16 million doses of zoster vaccine have been distributed (Merck and Company, Inc, unpublished data), yet this is the only reported case of Oka strain HZ reported in a zoster vaccinee. Although in general clinical practice, few cases of HZ are tested for VZV strain type, >19 000 zoster vaccine recipients were actively followed for episodes of HZ in the Shingles Prevention Study (SPS), yielding about 58 200 person-years of observation time, and 294 HZ cases were VZV strain tested [13]. No vOka strain HZ was detected, suggesting that the incidence of vOka HZ following zoster vaccination in VZV-immune individuals is not more than 0.06 per 1000 person-years (upper 95% confidence interval [CI]), despite the fact that zoster vaccine contains a 14-fold higher titer of Oka VZV than the varicella vaccine.

Viewed differently, between our study and the SPS, there have been 634 zoster vaccine recipients who developed HZ and were tested for VZV and for VZV strain type, with just 1 case (0.16% [95% CI, .0%–.47%]) of vOka VZV detected. For the physician seeing a patient with HZ following a history of zoster vaccination, the likelihood that the HZ is due to vOka VZV is extremely low. The overwhelming majority of HZ cases occurring among vaccine recipients are due to the limited ability of zoster vaccine to prevent reactivation of wild-type virus, which is 50%–55% [13–15]. In addition, our HZ case resolved within 10 days with no complications.

In a previous study, we identified 7 SNPs present in small numbers of Varivax isolates (in different combinations) that appear to be associated with enhanced pathogenicity [16]. These markers were previously undetected due to the limited sensitivity of Sanger sequencing, which fails to detect SNPs present in <10% of strains in a mixed population such as Oka vaccine [9]. We did not find any of these SNPs in our patient's isolate.

Altogether, 21 individual SNPs were identified from the complete genomic sequence of VZVs/Pasadena.USA/48.12/Z [2] that distinguished it from the composition of most of the VZV strains in Zostavax vaccine, including 9 that resulted in amino acid changes to their respective proteins. It is conceivable that 1 or more of these changes could reflect a strain in the vaccine with enhanced pathogenicity, although this is merely speculation. The functions (where known) of the 9 proteins that display amino acid changes were ORFs 6 (helicase-primase subunit), 11 (RNA binding protein), 17 (messenger RNA–specific RNAse), 21 (virion morphogenesis), 34 (DNA encapsidation), 44 (virion morphogenesis), 59 (DNA repair), 60 (gL, membrane glycoprotein), 62 (2 changes, major transactivating protein), and 64 (function unknown). The affected proteins are variously involved in virus entry, virion maturation, replication, transcription, nucleic acid metabolism, and gene expression, and as such they play important roles in every phase of VZV infection and reproduction. However, it is not possible to ascribe any of these changes to an increased potential for developing HZ, and we speculate that most of the individual virus strains in the vaccine have roughly equivalent capacities for symptomatic reactivation.

In conclusion, we report the first case of vOka HZ following zoster vaccination in an immunocompetent adult who very likely had prior immunity to varicella. Although clinicians should be aware of the possibility of HZ due to reactivation of vOka from zoster vaccine, they can be reassured that the risk appears to be extremely low. Zoster vaccine offers an important opportunity for preventing HZ and its often devastating complications.