Herpes Zoster Recurrence: A Narrative Review of the Literature

In total, 11 of the studies in populations of immunocompetent or immunocompetent/immunosuppressed (mixed) individuals with an initial HZ episode reported on the percentage of individuals who experienced HZ recurrence, with estimates ranging from 1.2% to 9.6% (Table 1, Fig. 1) [10, 11, 14‐16, 19‐24]. Of these 11 studies, one provided only the percentage of study participants developing recurrent HZ, with no data on incidence rates per 1000 person-years (PY) and no time-to-event analyses [15]. The study was relatively large and long-term, surveying 14,343 inpatients and outpatients diagnosed with HZ between January 2005 and December 2015 at six hospitals in South Korea. The study reported that 169 patients experienced recurrence (a recurrence frequency of 1.2%); however, the number of recurrences per patient was not reported [15].

A total of eight of the studies (two each in Japan and the USA, and one each in Australia, China, South Korea, and Spain) in immunocompetent only or immunocompetent/immunosuppressed (mixed) populations with an initial HZ episode reported incidence rates per 1000 PY, with estimates ranging from 1.7 to 16.6 [11, 14, 16, 20‐22, 24, 25]. Most of these studies were retrospective reviews of electronic health records. The lowest recurrent incidence rate (1.7 per 1000 PY) was reported in one of the Japanese studies (a physician survey) [16]; the next lowest recurrent incidence rate (7.5 per 1000 PY) was reported in a US retrospective study with up to 4.3 years of follow-up [24]. The other Japanese study, a prospective cohort study of individuals aged ≥ 50 years with up to 3 years of follow-up, reported a recurrent incidence rate of 10.1 per 1000 PY [14]. A similar recurrent incidence rate of 11.1 per 1000 PY was reported in the only other prospective cohort study, an Australian study including all individuals with an initial HZ episode between 2006 and 2009, and with up to 8 years of follow-up [20].

A total of three studies (two of which are discussed above) in immunocompetent or mixed populations with an initial HZ episode included a time-to-event analysis in the overall population. All three were retrospective studies; one from South Korea [11] and two from the USA [10, 22]. A fourth study, from Germany, included a time-to-event analysis stratified by age, without reporting overall population results [19]; this study is listed in Table 1 but is discussed separately in a subsequent section on older age. The most recent of the three time-to-event studies was a US study that used state (California) electronic health records to analyze the first recurrence of HZ in immunocompetent, unvaccinated individuals aged ≥ 50 years who experienced an initial HZ episode between 2007 and 2008 [22]. Individuals were followed up until 2016, and recurrence was based on HZ International Classification of Diseases (ICD)-9 codes 053.xx (or ICD-10 codes B02.xx beginning from October 2015) and antiviral prescription (aciclovir, famciclovir, or valaciclovir). The cumulative risk of HZ recurrence was estimated as 2.5% at 2 years, 4.8% at 4 years, 6.6% at 6 years, 8.0% at 8 years, and 10.3% at 10 years (Fig. 2). These findings were consistent with an earlier US study that analyzed HZ recurrence among 1669 residents of Olmsted County, Minnesota, USA, of whom most (91.7%) were immunocompetent [10]. Following an initial HZ episode between 1996 and 2001, 95/1669 individuals (5.7%) had subsequent HZ episodes over a mean follow-up period of 7.3 years. Recurrence was based on detailed medical record review and was supported by laboratory testing in 24.8% of cases. Time-to-event analyses estimated the cumulative recurrence risk as 2.0% at 2 years, 3.6% at 4 years, 4.9% at 6 years, and 6.2% at 8 years (Fig. 2). Among immunocompetent individuals only, 80/1530 individuals (5.2%) had recurrence, with the 8-year cumulative recurrence risk estimated as 5.7% [10].

The third study using time-to-event analyses was conducted in South Korea and based on a national health insurance records database [11]. The study population was made up of 39,441 individuals with an initial HZ episode between 2002 and 2013 who were followed-up for recurrence, defined by diagnostic coding and antiviral prescription. Time-to-event curves indicated a linear increase in the cumulative rate of HZ recurrence in the decade following the initial episode, with an estimated cumulative rate exceeding 10% at 10 years [11]. Overall, the three time-to-event studies yielded broadly consistent findings, with no evidence of plateauing of HZ recurrence in the time periods covered (up to 10 years after the initial HZ episode). These studies suggest that individuals may be exposed to a uniformly increasing cumulative risk of HZ recurrence after the first episode. More studies of this type are required to confirm this hypothesis, both in more varied populations and over longer time periods, given that, when reported, mean time between the first HZ episode and first recurrent HZ episode ranged between 2.0 and 13.7 years [11, 15, 16, 20].

Another measure of the burden of HZ recurrence is the number of individuals who experience multiple recurrences; however, few studies provided this information. In the US study from Olmsted County, 6.3% of individuals had two recurrences, and 2.1% had three recurrences during follow-up [10]. In the South Korean study, 11.0% of individuals with recurrences had two recurrences and 1.2% had three recurrences [11]. Our review also identified a Canadian retrospective analysis of data for 194,584 individuals with HZ between 1994 and 2010, in which 15,909 individuals (8.2%) had one recurrence and 3964 (2.0%) had multiple recurrences [23]. A German analysis of health insurance records reported that 25% of patients (1030/4141) who experienced a first recurrence also experienced at least one further recurrence during the 10-year analysis; the authors also noted that the risk of a second recurrence was higher than the risk of a first recurrence [19].

Immunocompromised status is an important risk factor for HZ recurrence, with most [10, 11, 15, 20, 25], although not all [14, 21], studies that directly compare immunocompetent and immunocompromised individuals describing significantly higher recurrence rates in the latter subgroup. Two of the time-to-event studies compared HZ recurrence rates between immunocompetent and immunocompromised individuals. The South Korean study [11] and the US study from Olmsted County [10] reported significantly higher recurrence rates in immunocompromised individuals (p < 0.0001 and p = 0.006, respectively). However, only the US study from Olmsted County reported the specific recurrence rate, estimating an 8-year cumulative risk of HZ recurrence for immunocompromised individuals of 12.0%, which was more than twice that for immunocompetent individuals (5.7%) [10]. Moreover, only two studies reported incidence rates. A prospective Australian cohort study reported a recurrence rate of 17.0 per 1000 PY compared with 10.5 per 1000 PY in individuals without recent immunosuppression [20]. A Spanish retrospective study that included the electronic health records of 4.4 million individuals reported a recurrent HZ incidence rate of 22.8 per 1000 PY in immunocompromised individuals versus 15.6 per 1000 PY in immunocompetent individuals, which corresponds to a 25% increase in risk after adjustment for age, sex, comorbidities, reporting health department, and reporting year [25]. This study also provided rates for specific immunocompromising conditions, as summarized in Table 2. The highest rate reported was in hematopoietic stem cell transplant recipients (55 per 1000 PY), whereas the lowest rate was in patients with autoimmune thyroiditis or psoriasis (18 per 1000 PY) [25].

In total, 11 studies reported the percentage of immunocompromised individuals with observed HZ recurrence during follow-up, with estimates of 0–18.2%; the wide range is in part attributable to the fact that some studies focused on specific immunocompromising conditions, such as rheumatoid arthritis [26, 27], human immunodeficiency virus (HIV) infection [28], Wegener granulomatosis [29], systemic lupus erythematosus [30], or lung transplant [31, 32], while other studies included multiple conditions [11, 15, 21, 33]. The wide range is also likely a consequence of small population sizes in some studies.

The South Korean time-to-event study reported that 17.8% of individuals with HZ recurrence were immunocompromised, compared with 13.5% of individuals without HZ recurrence (p < 0.001). Notably, individuals with HZ recurrence had an increased prevalence of solid cancers, hematologic malignancies, autoimmune diseases, and chronic renal disease (all p < 0.05 versus individuals without HZ recurrence; see Table 2 for percentages) [11]. No differences were observed between individuals with HIV infection or chronic hepatic disease, although these subgroups were small [11]. For individuals with HIV infection, this may be an encouraging reflection of the efficacy of current antiretroviral therapies. For patients with autoimmune diseases receiving immunosuppressive therapy, it can be difficult to separate the effects of disease and treatment, but a retrospective South Korean study of individuals receiving biologic disease-modifying drugs found that tofacitinib increased the risk of recurrence compared with abatacept [adjusted hazard ratio 2.46; 95% confidence interval (CI) 1.61–3.76; p < 0.001], indicating that the risk may be differentially modulated by specific immunosuppressive medications [26].

HZ recurrence is more frequently reported in women than men. One of the time-to-event studies assessed cumulative risk of HZ recurrence by sex. The US study from Olmsted County reported a significantly (p = 0.04) higher cumulative risk of recurrence in women than men, with 8-year recurrence rates estimated as 7.2% and 4.5%, respectively [10]. Two studies reported the relative risk of HZ recurrence by sex, with both studies finding a significantly greater risk for women than men. In a Spanish study, women had a 19% greater risk of HZ recurrence (95% CI 1.14–1.24) [25], while in a South Korean study, women had a 48% greater risk (95% CI 1.35–1.62) [11].

Most studies reporting the incidence rate of HZ recurrence by sex show higher rates in women. In a prospective Australian study, the incidence rate was 12.8 recurrences per 1000 PY for women versus 8.3 recurrences per 1000 PY for men [20]. Similarly, a Japanese prospective study reported an incidence per year for recurrent HZ of 1.07% for women versus 0.91% for men [14]. A US retrospective study in individuals aged ≥ 60 years reported incidence rates for recurrent HZ of 9.5 cases per 1000 PY for women versus 4.5 cases per 1000 PY for men [24]. However, a retrospective Chinese study in individuals aged ≥ 50 years reported incidence rates of 14.3 recurrences per 1000 PY for women versus 19.5 recurrences per 1000 PY for men [21]. Such variation could be explained by different approaches to surveillance for HZ in different countries; differences in healthcare-seeking behavior between women and men; and cultural factors, such as the use of traditional Chinese medicine or patient stoicism, leading to underreporting of symptoms and underuse of antivirals.

Older age might be expected to increase the risk of HZ recurrence, given that it is an established risk factor for the first episode of HZ, but this was not seen consistently across the studies identified in our review. The Korean national health-insurance-based study reported that the risk of HZ recurrence was 45% higher for individuals aged 51–70 years versus 21–50 years (95% CI 1.31–1.60) [11]. A large Spanish cohort study found that, relative to individuals aged 18–29 years, the risk of HZ recurrence increased for individuals aged 40–49 years (relative risk 1.14; 95% CI 1.01–1.28), and then increased further in each decade of life up to ≥ 80 years of age (relative risk 2.14; 95% CI 1.91–2.40) [25]. Conversely, in the California time-to-event study, the 10-year cumulative risk of recurrence was 11.1% for individuals aged 50–59 years, decreasing consistently in older cohorts to 9.0% for individuals aged ≥ 80 years [22]. The US study from Olmsted County did not find a significant increase with age (p = 0.65), although there was an increase in the estimated cumulative 6-year (the longest timepoint for the analysis) recurrence risk for men only (3.3% versus 5.5% in those aged < 50 years versus ≥ 70 years) [10]. The German health insurance study noted above, which used a matched-cohort design, conducted a time-to-event analysis stratified by age, and reported that the highest risk, of > 15% at 10 years, was observed in individuals aged < 50 years and with ≥ 1 underlying condition, whereas the lowest risk was observed in individuals aged ≥ 60 years, who had a cumulative 10-year risk of recurrence of between 5 and 10% [19]. People who are older at the time of an initial HZ diagnosis may be more likely to die before experiencing a recurrence than those who are younger on initial HZ diagnosis, which may influence recorded data on risk of HZ recurrence by age.

Long-lasting pain following an initial HZ episode was only assessed in a few studies in our review, but was consistently identified as a significant predictor of subsequent recurrence. The US study from Olmsted County study reported that pain lasting ≥ 30 days was a significant (p < 0.001) predictor of recurrence compared with pain lasting < 30 days, and the effect was stronger in younger individuals (aged ≤ 50 years) and during the first 3–4 years after the initial episode [10]. In the South Korean national health insurance study, cumulative recurrence rates were 4.9%, 5.7%, and 10.2% in individuals with durations of HZ-related pain of < 31, 31–90, and > 90 days, respectively (p < 0.001) [11]. In another South Korean study, the reported cumulative risk of recurrence was significantly different between individuals with pain lasting < 30 days (0.9%), 30–90 days (2.2%), or ≥ 90 days (1.8%) (p < 0.001) [15].

In addition to immunocompromising medical conditions, other comorbidities were associated with an increased risk of recurrent HZ in the studies we reviewed. The list of comorbidities reported, and their effect sizes, varied between studies. A large German analysis of health insurance data that included patients with asthma, coronary heart disease, chronic heart failure, chronic obstructive pulmonary disease (COPD), depression, diabetes, and rheumatoid arthritis reported that the presence of any of these comorbidities increased the cumulative risk of recurrent HZ in all age groups by approximately 10–17% over 10 years. Overall, 9.6% of the study population had at least one HZ recurrence, and 25% of these patients had at least two HZ recurrences [19]. However, the authors did not report recurrence risks associated with the individual comorbidities [19]. A large Spanish epidemiologic study reported increases in recurrent HZ risk of 63% for heart failure, 26% for COPD, 20% for chronic kidney disease, and 5% for diabetes [25]. South Korean and Chinese analyses also described associations with dyslipidemia, hypertension, and/or diabetes [11, 21]. Further studies are required to understand the impact of specific non-immunocompromising medical conditions on HZ recurrence.

The severity of HZ recurrences may be inferred from factors such as the proportion of study participants with complications (e.g., PHN) or who require hospitalization, and by patient-reported outcome measures (PROMs; e.g., pain severity). Evidence regarding the severity of recurrence versus initial HZ episodes is limited and mixed. In a prospective Japanese study, HZ recurrence was characterized by a significantly lower severity of skin lesions and pain (p < 0.001) and a reduced incidence of PHN (p = 0.03) in individuals aged 50–79 years when compared with an initial HZ episode [14]. In an Australian study, although the hospitalization rate was similar between initial HZ episodes and recurrence (2.5% versus 2.7%; p = 0.8), fewer individuals with a recurrent HZ episode received pain medication (9.3% versus 14.8%; p < 0.001) [20]. Further, in a pooled analysis of tofacitinib trials in rheumatoid arthritis, second versus first HZ episodes were less likely to be classified as severe (1.6% versus 4.3%) or serious (4.8% versus > 7.0%), or lead to PHN (3.2% versus 6.9%) [27]. Taken together, these data suggest generally milder events at the time of clinical presentation with recurrence. A Chinese study reported a longer duration of hospitalization for recurrent than initial HZ episodes (mean 24.0 versus 21.6 days; p = 0.573), but this finding was not statistically significant [21].

HZ typically erupts in one or, less frequently, two contiguous dermatomes, commonly affecting the thoracic, cervical, and trigeminal dermatomes [34]. Evidence indicates that the dermatome(s) affected, and the anatomic distribution of recurrent HZ episodes, frequently differ from those of the initial HZ episode. In the US study from Olmsted County, 44.2% of individuals had a recurrence contralateral to the first episode [10]. Notably, ipsilateral recurrence was observed in all ten individuals with multiple recurrences (p = 0.008) [10]. In two studies recording dermatomal distribution, dermatomes supplied by the thoracic, trigeminal, and/or cervical nerves were frequently involved in recurrent HZ episodes [15, 16]. Recurrence usually involved a single dermatome (76.3% of individuals) [16], and more commonly affected different dermatome(s) compared with the initial HZ episode [15, 16].