Varicella and Zoster in Children With Human Immunodeficiency Virus Infection

Varicella-zoster virus (VZV) is a herpesvirus that causes chickenpox as a manifestation of primary infection, during which time latent infection with VZV is usually established. This virus may later reactivate and cause zoster (shingles) in the partially immune host. Children infected with HIV are at extremely high risk to develop zoster within a few months or years after primary infection.1, 2 Zoster develops when the cell-mediated immune (CMI) response to VZV is low, a phenomenon that occurs usually as a result either of normal aging or of disease; however, in addition zoster may be the result of development of varicella in individuals with an impaired CMI response to VZV at the time of the primary infection.3 Children with HIV infection could develop zoster for both reasons. We previously reported in a collaborative study a 70% rate of zoster in a group of children with HIV infection who developed varicella when their CD4 cell counts were <15%.2 Even children with <25% CD4 cell counts at onset of varicella were significantly more likely to develop zoster than those whose CD4 values were >25% at that time.2 Other investigators, however, have not noted an association between low CD4 values at the time of varicella and development of zoster.4, 5 We examined the courses of all of the HIV-infected children currently being cared for at Babies and Children's Hospital, which included 5 previously reported patients and an additional 28 HIV-infected patients with varicella, to determine whether more recent information would confirm or refute our previous observations. Patients and methods. We performed a retrospective chart review of VZV infections in 63 HIV-infected children receiving medical care at Babies and Children's Hospital in New York City. The diagnosis of HIV infection was made by antibody titer (enzyme-linked immunosorbent assay and Western blot) and culture of HIV from peripheral blood or both. For the most part these children were enrolled in other studies on HIV-infected children, including the Columbia University AIDS Clinical Trials Unit and the Women and Infants Transmission Study. None of these children had received varicella vaccine. The medical records were reviewed for the documentation of whether varicella and zoster had occurred, the nature of the illness, whether passive immunization and/or specific antiviral therapy was given and the outcome of the illnesses. The CD4 cell counts at the onset of varicella were noted. Laboratory methods. CD4 and CD8 determinations were performed in AIDS Clinical Trials Group-certified laboratories by standard methods. Viral HIV-1 RNA was measured by quantitative RNA PCR (LabCorp). VZV antibodies were measured by indirect immunofluorescence, fluorescent antibody to membrane antigen6 and latex agglutination.7 VZV in skin lesions was detected by viral culture in human embryonic lung fibroblasts and/or PCR8 and/or direct immunofluorescence.9 Statistics. Continuous variables were assessed by two-tailed t test, and the chi square test was used to make group comparisons, using Stat View software on a Macintosh computer. This research is in conformity with human experimentation guidelines of the US Department of Health and Human Services and that of the authors' institution. Results. Varicella occurred in 33 of 63 (52%) HIV-infected children. The mean age of children with no previous varicella was 6.8 years; the mean age of children who had had varicella was 8.9 years. Twelve of the 33 (36%) also had a history of zoster. With regard to children with previous varicella, there were 17 males (51%) and 16 females (49%); 15 (45%) were black, 14 (42%) Hispanic, 1 (3%) white and 3 (10%) biracial. Twenty-eight (84%) were infected perinatally, 3 by transfusion (10%) and 2 (6%) from sexual abuse. The diagnosis of VZV infection was made clinically and substantiated by laboratory methods in 20%. Clinical information on these children is presented in Table 1. Twenty-eight of these children have not been previously reported; data from 5 were included in an early collaborative report.2 Zoster developed between 3 months and 6 years after varicella, with a mean of 2.7 years. Children who did not develop zoster had been followed for between 3 months and 10 years with a mean of 2.6 years.TABLE 1: Clinical information on 33 HIV-infected children with varicella and zoster Of the 33 children with varicella, 18 were given acyclovir orally and 1 intravenously. Three were hospitalized during varicella: one with right middle lobe pneumonia; one with a bacterial skin infection; and one with complicating infection with Mycobacterium avium. Only the last patient received intravenous acyclovir; the other 2 were given acyclovir orally. In general most of these children had mild to moderately severe cases of varicella. The numbers of skin lesions, when noted, varied from only a few to up to 300 in 1 child. No child had a disseminated infection. Eight children were not given either passive immunization with varicella-zoster immunoglobulin or antiviral therapy. Sequential viral RNA assays relating to the time of varicella were available for 2 patients. One patient had 432 HIV-1 RNA copies/ml 5 days after varicella and <400 RNA copies/ml 2 months later. Another had <400 RNA copies/ml 3 months before varicella and 1009 RNA copies/ml 1 month later. Neither of these children developed zoster after 1 year of follow-up. Their CD4 cell counts at onset of varicella were 25 and 28%, respectively. CD4 values, expressed as percentages, were available for 19 new children at the time they developed varicella. These values were assessed for children who developed varicella only and those who had varicella followed by zoster. Children with CD4 cell counts <25% at varicella onset were at very high risk to develop zoster within 3 years after experiencing varicella. The rate of zoster was 5 of 6 (83%) in those with <25% CD4 lymphocytes at onset of varicella and 1 to 18 (6%) in those with >25% CD4 lymphocytes. The CD4 values were significantly lower at the time of varicella in children who developed zoster than in those who did not (P = 0.009 for 19 new patients, and P = 0.0001 for a total of 24, 19 new patients and 5 patients reported previously, by 2-tailed t test). CD8 values, expressed as percentages, were available for 19 children at the time they developed varicella. For those who developed varicella only the average CD8 cell count was 46% (range, 20 to 56%), and for those who subsequently developed zoster it was 52% (range, 29 to 70%). Of the 12 patients who developed zoster 3 (25%) have had recurrent episodes. One child has had 4 recurrences, and 2 children have had 2 recurrences. Discussion. Despite early reports to the contrary,10 it is becoming clear that varicella is not usually a very severe disease in children with HIV infection, even if their CD4 cell count is low at the time of onset of illness. Varicella, however, does tend to be more severe in HIV-infected than in otherwise healthy children, and most physicians elect to treat these patient who develop chickenpox with either oral or intravenous acyclovir. In our current study eight children did not receive passive immunization or antiviral therapy yet they did not develop severe varicella. No child cared for at our institution had disseminated or fatal varicella infection. At present small numbers of HIV-infected children with normal CD4 cell counts are being immunized with live attenuated varicella vaccine to determine whether it is safe, in ongoing studies in the Pediatric AIDS Clinical Trials Group. There is little information concerning immunization of HIV-infected children with low CD4 values; however, caution must be advised. One recorded child with undiagnosed HIV infection and a very low CD4 count developed pneumonia caused by vaccine type VZV after he was inadvertently immunized (A Blackwood, et al., manuscript in preparation). HIV-infected children who have been infected with the wild-type VZV are potentially at risk to develop zoster. The risk of zoster is greatly increased in those who develop varicella in the setting of immune compromise, reflected by a low CD4 value at the time of onset of varicella. We reported this phenomenon in a previous collaborative study, and the new data reported here from 28 additional HIV-infected children are in agreement with our previous findings. In other studies of VZV infections in HIV-infected children, CD4 cell counts were often not available at onset of varicella, so that an analysis similar to ours could not be carried out.4, 5 Our data are in agreement with data from Japanese investigators concerning children who were not infected with HIV. They reported low primary CMI responses after varicella in children <1 year of age; these children are also at increased risk to develop zoster.3, 11 Thus the strength of the primary immune response to VZV also plays a role in protection from development of zoster. Although the data are scant there was no evidence of an increase in RNA viral load in two HIV-infected children after varicella. This would be in keeping with our previous reports in which we found that varicella did not seem to increase the progression to AIDS.2, 12 Compared with the incidence of zoster in leukemic children, HIV-infected children are at 7 to 20 times greater risk.2 Risk for development of zoster does not seem to be related to strategies to modify varicella; in the current study there was no apparent relationship between development of zoster and passive immunization or oral administration of antiviral therapy. Although zoster is not life-threatening for HIV-infected children, it clearly affects their quality of life, and it would be useful to devise means to prevent zoster in these children. HIV-infected children who have not had natural varicella should be considered for immunization, especially when their CD4 cell counts are within the normal range and response to vaccination is likely to be optimal. Moreover because the incidence of zoster is high in children with low CD4 cell counts at varicella onset within a few years of developing chickenpox (83% in the current study and 70% in our previous collaborative analysis), this might be an ideal group in which to test whether immunization with VZV vaccine can prevent or modify zoster in those who have already been infected with the wild-type virus. An inactivated form of varicella vaccine, given as three doses, substantially decreased the severity of zoster in patients who had undergone bone marrow transplantation.13 Similar vaccine trials might be worthwhile to try to modify zoster in HIV-infected children who already have latent infection with wildtype VZV and are at high risk for VZV reactivation. Acknowledgments. Supported by grants from the National Institutes of Health (AI 27562 and AI24021). AD was supported as a 1997 summer fellow by the Society for Pediatric Research. Amanda Derryck, A.B. Philip LaRussa, M.D. Sharon Steinberg Michelle Capasso Jane Pitt, M.D. Anne A. Gershon, M.D. Department of Pediatrics; Columbia University College of Physicians and Surgeons; New York, NY