Effectiveness of varicella vaccine in children infected with human immunodeficiency virus (HIV)

Introduction

HIV-infected children may develop severe varicella, and are >15 times more likely than the general population to develop herpes zoster (HZ), which is often severe [1]. Live attenuated varicella vaccine is safe, immunogenic, and protective against varicella-zoster (VZV) infection in both healthy and certain immunocompromised children [2]. Based on clinical trials, 2 doses of varicella vaccine 2 months apart were recommended in 1999 for relatively healthy HIV-infected children with CD4 cell counts ≥ 25% [3]. Later children with CD4 counts of ≥ 15% were safely vaccinated [4]. Although the vaccine is widely used, there is no published information on its effectiveness in these children [5, 6]. Therefore we assessed the effectiveness of varicella vaccine in preventing varicella and HZ in this population.

Methods

We conducted a longitudinal cohort study with clinical data collected from 2 units of the Pediatric AIDS Clinical Trials Group (PACTG): Columbia University-New York Presbyterian Hospital and St. Jude Children’s Research Hospital. This study was approved by both IRBs.

The medical records of 164 perinatally HIV-infected children between 1989 and 2007, determining whether and when they had received varicella vaccine, developed varicella, and/or developed HZ. All perinatally HIV-infected children were included except for those who developed varicella before one year of age, because varicella vaccine is not given before then. Vaccine was offered to children based on the recommendations of the Centers for Disease Control at the time.

Varicella was defined as an illness with a generalized pruritic vesicular rash with fever, diagnosed by a clinician. HZ was an illness with a unilateral vesicular rash in a dermatomal distribution without another identifiable cause. Laboratory confirmation of VZV usually was not obtained. Immunized children had at least 1 dose of live attenuated varicella vaccine. Highly active antiretroviral therapy (HAART) was defined as receipt of ≥ 3 antiretroviral agents, two non-nucleoside reverse transcriptase inhibitors, and either a protease inhibitor or a nucleoside reverse transcriptase inhibitor. HZ that occurred within 6–8 weeks after initiation of HAART was considered to be a possible instance of the immune reconstitution inflammatory syndrome (IRIS).

The statistical significance of differences of baseline characteristics of groups being compared was assessed with chi-squared tests for categorical variables and with t-tests for continuous variables. All follow-up times were calculated in person-years (P-Y). Rate ratios, their associated 95% confidence intervals and their statistical significance (two-tailed p≤0.05) were calculated using PEPI [7]. The effectiveness of the vaccine was calculated as 1 minus the rate-ratio.

Incidence rates of varicella were calculated for two periods: the pre-vaccine era for HIV-infected children, 1989–1998 (when none of the children received varicella vaccine) and the vaccine era 1999–2007. For unvaccinated subjects, time 0 was defined as one year of age, and the endpoint for time at risk (when they were censored from the denominator) was either the occurrence of varicella, receipt of varicella vaccine, or the last clinic visit. For vaccinated patients time 0 was the date they received varicella vaccine. Vaccinated subjects were censored at either onset of “breakthrough” varicella or the date of the last clinic visit.

Since HZ is a reactivation of a latent infection, only individuals previously infected with VZV through either natural disease or varicella vaccine were analyzed to assess the vaccine’s effectiveness against HZ [8]. Time 0 was either the date of immunization or the date of onset of varicella; subjects were censored at onset of HZ or the last clinic visit.

In the patients who developed natural varicella, HZ incidence rates were calculated for two periods: the pre-vaccine era (1989–1998) and the vaccine era (1999–2007). Rate-ratios for development of HZ among children who had been immunized versus those who had varicella were calculated.

The incidence rates and incidence-rate-ratio for HZ were also calculated for patients at risk of developing HZ among only those who were receiving HAART (to control for the possibility that HAART alone might decrease the risk of HZ). Time 0 was the date when the patients both had had varicella (or had received varicella vaccine) and were also receiving HAART. Censoring occurred at the date either of onset of HZ or of the last clinic visit.

Results

Half of the subjects were female and 63% were non-Hispanic African Americans. Of 72 vaccinated children, 39 (54%) had received two doses of vaccine. Baseline characteristics of patients who developed varicella, received varicella vaccine and/or developed HZ, such as CD4 counts, were similar and not statistically significantly different (Table 1). In the pre-vaccine era, only 6% of unvaccinated patients were receiving HAART at the onset of varicella; in contrast, after 1998, 63% were receiving HAART when they contracted varicella. Of the 32 patients who developed HZ after varicella, 15 (47%) were receiving HAART (60% in the HAART era). Of the vaccinated patients, 85% were receiving HAART when immunized. None developed HZ.

Discussion

Before 1999, VZV infections, particularly HZ, were frequent causes of morbidity among HIV-infected children. Varicella vaccine was recommended for HIV-infected children based on safety and immunogenicity but its efficacy in these children was not evaluated. [3, 5, 6]. Our study indicates that varicella vaccine protects against both varicella and HZ. There was a 63% decrease in the incidence of varicella among unvaccinated children after 1999, likely due to herd immunity from widespread use of vaccine. Despite the lower attack rate of varicella in the vaccine era, however, we were still able to show that immunization was highly protective in the HIV-infected children.

Immunization also protected HIV-infected children against HZ. The Oka vaccine strain can establish latent infection and can reactivate; protection from HZ is probably not entirely related to the Oka strain itself. Possibly protection against HZ is due to the relatively normal immune function in vaccinees when they were immunized[1, 9] which may have resulted in only rare skin rashes after vaccination that in turn correlates with a lower risk of subsequently developing zoster. [10]

The incidence of HZ may increase transiently after the institution of HAART, [11, 12] often attributed to IRIS. To account for the possibility that some of the cases of HZ could be attributed to “immune reconstitution,” we excluded these cases in one of our analyses. Even when we excluded 3 cases of HZ that developed within 6–8 weeks after starting HAART, the vaccine was still statistically significantly effective in preventing HZ.

HAART was used more frequently after 1997, improving immune function and the quality of life and survival rates in HIV-infected children. In one report, introduction of HAART decreased HZ incidence by 50% in HIV-infected children [13]. In that study, however, vaccinees and those with natural varicella were analyzed together. Another study indicated that even in the HAART era, zoster remained a significant cause of morbidity for unvaccinated HIV-infected children [14]. In a different study in HIV-infected children, it was not possible to determine whether a decrease in HZ was attributable to immunization or to HAART [15]. In our study, by assessing the incidence rates of HZ only in children receiving HAART (vaccinated children and unvaccinated children who had had varicella), we found varicella vaccine to be highly effective in preventing HZ, even if children with possible IRIS were excluded from the analysis. Varicella vaccine, rather than HAART, therefore, appears to be the major factor in the reduction of HZ in immunized patients.

It also did not appear that the explanation for the vaccine’s effectiveness could be that vaccinees had better overall immunity than the unvaccinated children. The median CD4 cell percentage for the unvaccinated and unvaccinated groups were similar, suggesting minimal immunosuppression in both groups.

There are several possible limitations to this study. Our study was observational and not a randomized clinical trial; the data were obtained in a review of medical records. The diagnoses of VZV infection were not laboratory verified. The HIV-infected children included in our study, however, were intensively and prospectively followed under study conditions of the PACTG, and the data are therefore of high quality.

Regarding protection against HZ, the median age of the immunized group was lower than that of the unvaccinated group. Younger children were less likely to have already had varicella and more likely to have been immunized, in comparison with older children. However, there is evidence that the age at the time of the primary infection with VZV, except in children less than one year of age, is unrelated to the risk of zoster unlike that in adults [20].

In summary, our data show that varicella vaccine was highly effective in preventing varicella and HZ in a cohort of perinatally-HIV-infected children followed between 1989 to 2007. Additional follow-up over a longer period, with larger numbers of subjects may be useful to confirm our findings.