The Risk of Varicella-Zoster Infections in Different Patient
Populations: A Critical Review
James J. Rusthoven
ARICELLA-ZOSTER (V-Z) is a ubiquitous
V virus among human populations. Gershon
and Steinberg reported a prevalence of seroposi-
tivity approaching 100% by the age of 60 years
among native-bom, urban-dwelling persons in the
United States. 1 Primary infection occurs almost
exclusively before adulthood; Preblud et al re-
ported that 90% of chickenpox cases occur be-
tween 1 and 14 years of age,2 with secondary at-
tack rates reported about 90% among household
members. 3 This artiele critically reviews the stud-
ies that identify the groups most at risk for V-Z
infection (Table 1). Uniess otherwise indicated,
the term immunocompromised connotes patients
with cancer or acquired immunodeficiency syn-
. drome (AIDS); those requiring chemotherapy or
steroids, those with immunodeficiency diseases,
and those receiving organ transplantation. Preg-
nant women and neonates are considered sepa-
rately.
VARICELLA INFECTION
Risk in the Generał Popułation
Primary infection, or chickenpox, is a common
disease and generally mild in nature; perhaps be-
cause of this, only 150,000 to 200,000 cases of an
estimated 3 million cases are reported to the Center
for Disease Controi (CDC) in the United States
each year. 2 Whereas year-to-year variations occur,
the only predictable pattem in incidence is a strik-
ing seasonal peak during March through May. Pre-
blud studied the age-specific risks of varicella
complications by comparing data collected be-
tween 1972 and 1977 from epidemiological
records from Illinois, New York, and New York
From the Department oj Medical Oncology, Ontario Cancer
Treatment and Research FountkJtion, Hamilton Regional Can-
cer Centre, Hamilton, Ontario, Canada.
Address reprint requests to James J. Rusthoven, MD, De-
partment ojMedical Oncology, Ontario Cancer Treatment and
Research Foundation, Hamilton Regional Cancer Center, 699
Concession St, Hamilton, Ontario, Canada.
Copyright © 1994 by W.B. Saunders Company
0887-7963/94/0802-0002$3.00/0
96
City. Data from the National Center for Health
Statistics Health Interview Survey (HIS) were also
obtained for the period 1972 through 1978, and
data from 13 state epidemiologists were solicited
for the year 1978. 4 These data were compared with
that of 389 cases of varicella encephalitis and 735
deaths reported from 1972 through 1978 to the
CDC from various reporting areas. Ninety-one
percent of varicella cases occurred between ages 1
and 14, whereas 2.4% and 1.6% occurred at ages
under 1 year and over 20 years, respectively (Table
2). Death associated with varicella occurred pro-
portionally more frequently in the older age group,
the rates being 63.2% in the age 1 to 14 group and
27.6% in the age 20 or older group. Encephalitis
and Reye's syndrome occurred predominantly in
the age 1 to 14 group; their incidence being 83.8%
and 98.3%, respectively.
The incidence of complications among suppos-
edly normal, nonimmunocompromised children is
unclear. Data from a study in 1935 showed com-
plications in 5.2% of 2,534 hospitalized children
with varicella. 5 Common complications included
secondary superficial bacterial infections (57%),
otitis media (28%), and pneumonia (16%),
whereas encephalitis occurred in only 4%. Eight
percent of those with complications died, for an
overall death rate of 0.4% in this hospital-based
population. In 1986, Fleischer et al reviewed the
cases of 96 children hospitalized for varicella. 6 Of
these, 84% were said to be "normal," with ages
ranging from 11 days to 19 years. Whereas sec-
ondary superficial bacterial infections were again
the most common complication (27%), encephali-
tis was next (25%), followed by Reye's syndrome
(21%). Pneumonia was reported in only 7%. In-
terestingly, 95% of encephalitis cases and all cases
of Reye' s syndrome occurred in nonimmunocom-
promised children, whereas pneumonia occurred
with the same frequency in compromised and non-
compromised patients. With information from
both of these studies, Preblud estimated the risk
per 100,000 varicella cases of varicella-associated
complications in normal children aged 1 to 14
years as follows: encephalitis 1.7, Reye' s syn-
drome 3.2, hospitalization 170, death 2.0. 2
Transfusion Medicine Reviews, Vol VIII, No 2 (April), 1994: pp 96-116
RISK OF v-z INFECTIONS IN DIFFERENT PATIENTS
Table 1. Population Subgroups at Increased Risk for
V-Z Infection
Pregnant women
Neonates
Cancer patients (particularly HO, NHL. and acute leukemia)
Organ transplantation patients
Acquired immunodeficiency disease patients
Perso ns with other immunodeficiency states
Certain closed populations and ethnic groups
seem to have a greater than normal risk for vari-
cella. Wallace et al have prospectively described
the natural history of varicella in adults at a naval
hospital. 7 The study was conducted in the context
af a randomized study of acyclovir versus placebo
as an intervention to alter varicella-related morbid-
ity. Patients were selected in whom rash had oc-
curred less than 72 hours before hospitalization.
Patients with classical varicella were selected; per-
haps because of this, no case of atypical lesions
were noted. Of 205 patients assessed, 148 were
evaluable for the study. As mentioned earlier, over
25% of deaths from varicella are reported in per-
sons age 20 years or older. Four patients had ra-
diographic evidence of pneumonia at hospitaliza-
tion (2.7%). All developed sepsis or symptoms of
pneumonia within 3 days of onset of rash, a find-
ing consistent with a 70% occurrence of this prob-
lem within this time period in a previous study. 8
This 2.7% incidence is on the low end of an inci-
dence spectrum that goes as high as 50%.9 In a
recent study of 10,687 United States Army and
Navy recruits admitted to the hospital for chicken-
pax, in a multivariate analysis, factors associated
with this varicella hospitalization included race
and climate. Black and Filipino naval recruits, for
example, have a 1.9-fold and 8.8-fold increase in
the likelihood of developing varicella, respec-
tively, compared with whites. lO The marked, in-
dependent association of high incidence in
younger-aged recruits suggested an effect of over-
crowding, with the more susceptible age group liv-
mg in crowded conditions compared with older
army personnel. Evidence of asymptomatic hepa-
titis was present in half of the patients and no
deaths occurred, likely because of the careful mon-
itoring at a naval hospital during a clinical trial.
!hus, in a re1atively closed population of naval
recruits complications of adult varicella are infre-
quent, presenting primarily as pneumonia, and the
mortality rate is very low.
97
Risk oj Varicella and lts Complications During
Pregnancy and in Neonates
Several studies have examined the complication
cate of varicella in pregnant women. An early es-
timate of the incidence of varicella in pregnancy
from collaborative prospective studies was 117 ,500
pregnancies, based on 60,000 pregnancies. ll In a
prospective cohort study, Siegel et al examined the
incidence and complication of various viral dis-
eases associated with pregnancy.12 Whereas the
fatality rate was only 1 in 125 pregnant women
with varicella (0.7%), a more recent study by
Paryani and Arvin reported pneumonia in 4 of 43
pregnant women with varicella, of whom 2 re-
quired respiratory support and 1 died (2.3%).13
The risk to the fetus also seems to be increased
with matemaI varicella infection. The incidence of
congenital varicella syndrome among pregnant
women with varicella is unclear, but transmission
af the virus to the fetus in the first 16 weeks of
gestation can lead to limb hypoplasia, skin scar-
ring, encephalitis, mental retardation, seizures, oc-
ular abnormalities, and low birth weight. The most
comprehensive study of this problem prospectively
examined 43 pregnant women with varicella. 13
The diagnosis was made on clinical grounds, all
patients were exposed within 21 days of diagnosis,
and none had a history of previous chickenpox.
Seventy-nine percent of the patients had only cu-
taneous involvement, whereas the remainder (9/
44, 21 %) had a medical or obstetrical complication
associated with varicella. One of the four women
with pneumonia died at 6 months gestation, with
loss of the fetus. Premature labor occurred in 4 of
the surviving 42 women, with premature delivery
resulting in two of these. Two of 13 women with
varicella in the first trimester had an elective abor-
tion. Forty-one infants were prospectively fol-
lowed up for consequences of their exposure to
varicella. Three infants exposed less than 10 days
before delivery may have acquired infection during
labor and delivery, and were considered sepa-
rately. Of the remaining 38 infants with intrauter-
ine exposure, 10.5% had clinical evidence of this
exposure; one of 11 patients (9%) with first trimes-
ter exposure developed congenital varicella syn-
drome, 1 of 11 infants with second trimester ex-
posure developed zoster infection at age 7 months,
and 2 infants had skin involvement with primary
varicella at birth among 16 infants with third tri-
98 JAMES J. RUSTHOVEN

mester exposure. One of these latter infants devel-
oped a second case of chickenpox at age 12 months
without known reexposure as well as zoster infec-
tion at age 30 months. Immunological evidence of
varicella infection using immunoglobulin M (IgM)
and IgG serum levels and lymphocyte transforma-
tion studies was present in 0.7 first tńmester, 2/11
second trimester, and 5/15 third tńmester-exposed
infants, or 21 % of tested infants. None of 12 other
infants bom to mothers with zoster infection dur-
ing pregnancy had evidence of an immune re-
sponse. Overall, 24% of infants had clinicalor
immunological evidence of congenital varicella in-
fection, but immunological evidence was not con-
sistent1y apparent in patients with clinical evidence
of infection. Siegel found congenital defects asso-

Table 2. Frequency and Oistribution ot Varicella Infection

Incidence Rates Proportion of Total Cases

Group Varicella Complication Complications Age Group

(reference no.) (reference no.) (reference no. I Varicella (reference no.) (yearsl

General population or hospitalized 75% (seropositivityl' 2.4% <1

nonimmunocompromised
91.0% 1-14
1.6% 20 or older
Encephalitis 2 83.8%4 1-14
1.71105 cases
Reyes syndrome 2 98.3%4 1-14
3.2/10 5 cases
63.2%4 1-14
2.0/10 5 cases 27.6%4 20 or older
Pregnancy 1n,500 Pneumonia 13: 10%
pregnancies'1
Death: 0.7%12,2.3%13
Obstetrical and
medical 13 : 21 %
Neonatal zoster: 4%
Infants of maternal exposure Clinical: Congenital defects
(first trimester exposure):
Clinical and 7.4%'4
seroconversion: 0%'5
24%'3
9%13

Zoster: 1/11 exposed {9%)'3 First trimester

ISecond trimester) 9%'3 1/11 exposed (9%)'3 Second trimester
(ali trimestersl 2.6% 2/16 exposed (13%)13 Third trimester
Neonatal (exposure within 5 days 17% Death:
of delivery)'6
31% of cases
5% of infants of mothers
with predelivery varicella
Children with cancer or immuno- Visceral dissemination:
compromised 20 32% (on anticancer therapy)
0% (oft chemotherapy,
in remission)
Death:
7% (on anticancer therapy)
0% (oft chemotherapy,
in remissioni
Adults with cancer23 0.9%
Bone marrOW transplants 24 2.6% Visceral dissemination:
(allogeneic and syngeneic) 44%
Death:
28%
(autologoUS)2S 6.5% Visceral dissemination:
10%
Death:
0%
AISK OF v-z INFECTIONS IN DIFFERENT PATIENTS
ciated with varicella exposure in 2 of 27 infants
with ftrst trimester exposure, 14 but none occurred
aroong 23 such infants in another study.15 Clearly,
the risk of intrauterine infection is increased with
roatemal infection but, perhaps more importantly,
the consequences are often devastating for both
roother and infant.
Neonatal risk of varicella infection seems to be
also increased following matemaI infection as does
the risk of complications. When defined as new-
bom varicella associated with matemal varicella
occurring within 5 days of delivery, the risk of
neonatal varicella is about 17%, but the case-
fatality ratio is about 31 % overall; about 5% of
infants with mothers who have predelivery vari-
cella die of varicella. 16 Whereas exposure from
birth to the tenth day of life has been associated
generally with mild disease,17 one study reported
severe cases of varicella in 20 of 53 (38%) infec-
tions in infants less than l year of age. 18 However,
no deaths occurred and any reported excess of
deaths from varicella in this age group must be
tempered by the known under-reporting of vari-
cella cases in this groUp.2
Risk oj Varicella in
lmmunocompromised lndividuals
Although the occurrence of primary infection as
chickenpox generally confers life-long immunity
from a second attack of chickenpox, second pri-
mary infections have been documented. In a pro-
vocative study, Gershon et al serologically tested
patients for evidence of immunity to varicella at
the time of reexposure to varicella. 19 One hundred
fifteen patients from various institutions were
tested between 1975 and 1982. The median age
and range were not given but by inference the pa-
tients were children. Adequate follow-up informa-
tion was another criterion for inelusion into the
study. One third of patients were considered im-
munocompromised, although specific criteria for
classifying patients as such were not given. In ad-
dition, three patients with leukemia and no previ-
ous history of chickenpox and two children with
leukemia in remission who had received varicella
vaceine were ineluded. Therefore, this cohort was
highly selected, and interesting seroelinical corre-
lations were attempted. The fluorescent antibody
titer to membrane antigen (FAMA) of varicella
was used as an indicator of immunity. As ex-
pected, all patients with no demonstrated titers
99
« l :2) later developed clinical varicella. Of the
remaining 62 patients with exposure, titers were
~ l :8, and 6% developed clinical varicella within 3
weeks. One of these four patients had varicella
infection in infancy, whereas the others had no
history. Another patient had systemie lupus erythe-
matosus requiring high doses of corticosteroids at
the time of exposure. The other three patients were
considered immunologically norma!. Three chil-
dren in remission from leukemia with no history of
varicella, recent transfusions, or reeent immuno-
globulin therapy had been tested for FAMA titers
just before ehickenpox developed. All had re-
eeived immunoglobulin more than 4 months ear-
lier; all patients had positive FAMA titers but no
known history of recent exposure to zoster or
chickenpox. One child in remission had been vae-
einated for varicella 10 months before the devel-
opment of a varicelliform eruption. No recent ex-
posure was known, her FAMA titer 2 weeks before
illness was 1:8, she had been off chemotherapy for
several months, her disease was mild (70 vesieular
lesions), and v-z virus was isolated from the rash.
DNA analysis showed a wild-type, nonvaccine
strain. Another vaeeinee had amilder varicelliform
rash that developed 4 weeks after immigration.
Again, no exposure history was evident and the
vaecine strain was isolated from the rash. Whereas
no epidemiologieal data are available, this study
shows that a ehickenpoxlike rash without elinieal
evidenee of zoster (shingles) can oecur in seropos-
itive patients whether elinical ehickenpox had oc-
curred previously. Ił also shows that immunocom-
promised patients who received varicella vaccine
may not be protected eompletely against a second
ehickenpoxlike illness eaused by either a wild-type
or the vaceine strain. Whether the incidenee of this
second elinical bout of chickenpox in either vac-
cinated ar nonvaccinated patients who are seropos-
itive is higher in immunocompromised patients is
not yet elear.
Several studies of the risk of varieella compli-
cations in immunocompromised children have fo-
cused on children with eancer. In the study of
Fleiseher et al, 7 of the 15 patients considered
immunocompromised experienced pneumonia. 6
These figures contrast sharply with the 5.2% eom-
plication rate and 0.4% death rate in the early hos-
pital-based study of Bullowa and Wishik of pri-
marily nonimmunocompromised children. 5
Feldman et al characterized varicella infection
100
among 77 cases of varicella in children with cancer
at St Jude Children' s Research Hospital between
1962 and 1973. 20 The ages of the infected children
were not given. Varicella was diagnosed clinically
and no atypical generalized cases as described by
Schimpff et ae 1 were noted. Cases of disseminated
zoster were excluded. Although the incidence
among all cancer patients seen at that institution
during the study period was not given, the age
distribution and seasonal occurrence was similar to
studies of children in the general population. Va-
ricella recurred during the study period in 8%. Pa-
tients who were in remission from their cancer and
had been off anticancer treatment for 3 or more
months at the time of varicella infection experi-
enced no visceral dissemination or deaths (17 pa-
tients). In the remaining 60 patients receiving an-
ticancer therapy, 32% had visceral dissemination
and 7% (4/60) died of disseminated varicella. Va-
ricella pneumonia was evident in all deaths. Al-
though a multivariate analysis of potential risk
factors for dissemination was not performed, dis-
semination seemed to be linked with absolute lym-
phopenia less than 0.5 x 109/L. The type and sta-
tus of the tumor could not be correlated with the
risk of dissemination; similarly, dissemination risk
was not correlated with duration or type of anti-
cancer therapy. However, the cessation of cancer
therapy before the onset of varicella was associated
with a reduced risk of dissemination. Thus, vari-
cella in this cancer population is clearly more se-
vere than in the general population and the severity
is possibly related to whether the patient is receiv-
ing anticancer therapy or has lymphopenia.
Schimpff et al recorded all cases of varicella or
zoster that occurred from July 1969 through June
1971 among cancer patients at a single cancer re-
search center. 21 One child with acute leukemia and
a second with Burkitt's lymphoma developed va-
ricella among 419 patients; neither child had pre-
vious varicella and both were exposed. Six other
cases were described as atypical generalized
zoster. Whereas no antecedent localized zoster was
evident to allow the diagnosis of disseminated
zoster, the lesions were atypical for classical vari-
cella; lesions occurred individually rather than in
crops, and began on the limbs as opposed to the
usual truncal origins of varicella. Five of these
patients had Hodgkin's disease and their ages
ranged from 18 to 46 years. All had a history of
varicella and all had relatively mild symptoms
JAMES J. RUSTHOVEN
compared with those of patients with generalized
zoster. Whereas the investigators concluded that
these cases represented cases of generalized zoster
with better host defences than those with dissem-
inated zoster, they could also represent reinfection
with varicella as described by Gershon et al. 19
However, the latter did not describe the clinical
features of their three adult cases so clinical com·
parisons cannot be made.
Whereas Schimpff et al had reported several
small clusters of v-z infection during this 2-year
study,2l Morens et al published a more compre-
hensive case-control study characterizing 22 cases
during a 4-month period at the National Cancer
Institute. 22 These cases occurred from July 23 to
November 29, 1976; this was significantly more
frequent than the 12 cases reported in the previous
year (P < .Ol). Controi subjects were chosen for
the 22 cases and were matched for admission date,
date of the last outpatient visit, sex, and tumor
type. Four of the 22 patients died of V-Z infection,
a rate similar to the controls (5/22). In 8 of the 22
cases the initial derrnatomal zoster was absent or
doubtful (doubtful zoster). No difference in the
severity of symptoms or distribution of lesions was
evident compared with the fourteen cases of
zoster. However, patients with doubtful zoster
were significant1y more often in relapse of their
tumor OT were receiving chemotherapy. The inves-
tigators argue on the basis of this finding as well as
a statistically significant association with close ex-
posure in other patients with V-Z infection (P <
.005) that these two presentations are epidemio-
logically distinct diseases. Drawing paralleis with
the atypical generalized zoster cases in the study of
Schimpff et al they suggest that these atypicalor
doubtful forms of zoster are actually varicella in-
fections that resulted from exogenous reinfection.
They further argue that these cases may result from
low or absent levels of protective antibody, citing
the absence of antibody in seven of the eight cases.
However, the study did not clearly correlate the
presence of antibody with the type of V·Z infection
and this point remains inconclusive. Furtherrnore,
the investigators failed to highlight the differences
between these cases and those in the Schimpff et al
study, in which the lesions were individual rather
than clustered as in case of typical varicella. This
was likely because of the retrospective nature of
this study, whereas data were prospectively stud-
ied by Schimpff et al. However, combining the
RISK OF v-z INFECTIONS IN DIFFERENT PATIENTS
data of these studies and that of Feldman et al it is
probabie that varicella can recur in children ar
adults with eancer, particularly if the disease oe-
eurs during tumor relapse ar antieaneer therapy. In
a large retrospective study of 740 adult eaneer pa-
tients with V-Z infeetion, Rusthoven et al identi-
fied disseminated V-Z infection without a refer-
ence to anteeedent loealized zoster in only 0.9% of
cases. 23 Though this may underestimate the true
incidenee of varieelIa or atypieal or doubtful zoster
in a general adult cancer population with V-Z in-
fection, the true incidence in a nonepidemic setting
is likely closer to this figure than that in the other
studies.
Two risk groups that involve ehildren and adults
are patients undergoing organ transplantation and
those with AIDS. Locksley et al studied alI pa-
tients undergoing alIogeneic or syngeneie bone
marrow transplantation (BMT) in Seattle, WA
from 1969 through 1982. 24 Of these 1,394 pa-
tients, 231 patients with V-Z infection were iden-
tified; those with infeetion occurring within the
fust 100 days after transplant were prospectively
followed up, whereas retrospective chart reviews
were performed for patients with V-Z infection be-
yond that time. Although the underlying diagnosis
of these 1,394 patients was not given, 91 % of the
patients with V-Z infection received transplanta-
tion as part of therapy for acute leukemia of aplas-
tic anemia. Although the median folIow-up was
not given, the range varied from 2 months to 70
months after transplantation. Standard clinical def-
initions of primary varicelIa infection and herpes
zoster infection were used; cases in whieh herpes
simplex "could not be reliably discounted" were
not included and virological and serological meth-
ods were used to clarify this distinction. Some
cases of varicella may have included cases de-
scribed as atypieal generalized zoster, described by
Schimpff et al as sporadicalIy disseminated le-
sions, not typical of varicella but without anteced-
ent 10calized zoster. 21 V-Z infeetion could also be
based on viral isolation or antigen identification by
indirect immunofluorescence, but dermatomai 10-
calization was necessary to distinguish varicella
from zoster. Thirty-six patients with V-Z infection
presented with varicella (36/231, 16%), which
yielded an overalI incidence of varicelIa in this
BMT population of 2.6% (Table 2). The median
age af these 36 patients was 22 years, with a range
from 2 to 40 years. A previous history of varicella
101
was evident in 39% of these patients, two thirds of
infections occurred within 6 months of transplant,
and the yearly incidenee was constant over the
study period. Visceral dissemination was docu-
mented in 44% and 28% of these patients died of
V-Z infeetion. AlI deaths occurred in patients with
disseminated disease, whereas the frequeney of
visceral dissemination and death were signifieantly
greater in patients with varicella compared with alI
of those with zoster (P < .05): however, the fre-
quency of these problems were similar between
patients with varicella and those with zoster and
skin dissemination. In a logistic regression model
using multiple suspected risk factors in 1,109 pa-
tients with leukemia or aplastic anemia, patients
with alIogenic transplants and acute or chronic
graft-versus-host disease were associated with
greater risk for V-Z infection, whereas only acute
graft-versus-host disease was signifieant1y associ-
ated with dissemination or death among alI patients
with V-Z infection. Patients developing varicella
in the first 9 months after transplant were signifi-
cant1y older (25 v 15.5 years; P = .02), more
likely had an underlying hematologie malignancy
(P = .01), and were at greater risk of developing
viseeral dissemination (16126 v 2/10) compared
with those developing infeetion later. Serologi-
calIy, 21122 varicelIa patients with pretransplant
sera had detectable antibodies to V-Z virus before
transplants, whereas 23125 patients with sera taken
before infection and within the first 9 months after
transplant had antibodies. Varicella occurred ex-
clusively in young children when evident more
than 2 years after transplant; these four patients
had no history of previous chickenpox but twa
children had pretransplant sera evaluable and both
were negative for V-Z antibody. All four children
had a history of V-Z exposure.
This extensive study demonstrates that V-Z in-
fection, and particularly varicella, is associated
with a high risk of dissemination and death in the
early period following allogeneic bone marrow
transplantation. Among patients receiving autolo-
gous BMT (ABMT), Sehuchter et al reported a
23% incidence of varicella among patients with
V-Z infection; the overalI ineidenee among ABMT
recipients was 6.5%.25 Patients with renal trans-
plants also seem to be at increased risk. Naraqi et
al prospectively followed up thirty renal allograft
recipients over a 3-year period and found a 7%
yearly incidence of V-Z infection, whereas the
102
prevalence V-Z infection before hemodialysis or
transplant was similar to the general population. 26
Varicella Infection Associated With Human
Immunodeficiency Virus Infection
It is unc1ear whether the incidence of varicella
infection is significantly increased in children with
human immunodeficiency virus (HIV) infection.
This is caused in part by the relatively high inei-
dence of chickenpox in nonimmunocompromised
children. However, children with HIV infection
who develop varicella may seem to be at increased
risk for developing zoster infection within months
of experiencing chickenpox. 27 Distinctly uncom-
mon in healthy children,28 zoster occurs only after
years of latency following chickenpox; its devel-
opment with a latency period of several months
highlights the profound immunoincompetence of
these children.
HERPES ZOSTER INFECTIONS
The Incidence of Zoster and Its Complications in
the Generał Popułation and the
Immunocompetent Host
The most comprehensive general population
data on herpes zoster have been obtained from
Rochester, MI through the studies of Ragozzino et
ae9 and Guess et al. 28 In the former study, index
cases were identified from a database inc1uding
outpatients in c1inics, emergency rooms, house
calls, and nursing homes, as well as hospital pa-
tients. All health providers in this population for-
warded data for this study. A central index was
searched for all Rochester residents who had a di-
agnosis of herpes zoster between 1945 and 1959;
synonymous and complementary diagnoses in-
cluded herpes, shingles, postherpetic neuralgia, or
Ramsay Hunt syndrorne. A full description of le-
sions and symptoms was available in 92% of
cases. The entire medical history of each individ-
ual was reviewed during his/her residency in the
community. Risk factors that were reviewed in-
c1uded previous malignancy, trauma, systemic
chemotherapy, or corticosteroid therapy, radio-
therapy, or surgery. The latter four factors were
only considered if given within 1 year of zoster.
Incidence rates were reported in age- and sex-
speeific person-years (p-y) at risk.
The crude incidence of zoster in this predomi-
nantly white population was 1311100,000 p-y ad-
JAMES J. RUSTHOVEN
justed to the 1970 population structure in the
United States (Table 3). This incidence rate is
lower than those reported from England (340/10 5
p_y)30 and Scotland (480/105 p_y)31 during similar
and earlier time periods. However, age adjustment
for 1970 whites in the United States, the inclusion
of second episodes of zoster in these studies, and
the knowledge that those data come from medical
practices rather than a general population database
bring these figures closer to this study. After age
adjustment, the male-to-female ratio was 0.94:1.
The incidence increased with age for both sexes
with incidence increasing from about 100/105 p-y
at ages 25 to 34 to over 400/105 p-y by age 75 or
older. Unlike the incidence of varicella, no sea-
sonal trends were found. Ineidence rates also in-
creased over time, from 112/105 during 1945 to
1949 to 150/105 p-y during 1955 to 1959. This
increase approached linearity and was greater for
men than women. Whether this increase repre-
sented an increase in risk groups within the popu-
lation or an increase in medical care access is not
elear; the latter is unlikely because studies of other
diseases in this population have not shown system-
atic increases in incidence over this time. Ten per-
cent of patients had one or more of the previously
mentioned risk factors for zoster, with a history of
cancer leading these factors and involving 6% of
the entire cohort. Five percent had more than one
episode of zoster. Assuming that these recurrent
events were independent from similar events in
any given individual, the investigators calculated
an expected recurrence rate of 3.9% for this co-
hort; this was not significantly different from the
observed rate.19
Over one half of the cases involved unilateral
thoracic dermatomes, with involvement of cranial,
cervical, and lumbar nearly equally distributed
among the remaining cases (about 10% each). Less
than 1% of episodes were bilaterai and 2% were
episodes of disseminated zoster. The areas of in-
volvement and risk of dissemination did not vary
significantly with age, except for ophthalmic in-
volvement in which the age was significantly
greater than the cohort at large (mean age 56 v 46;
p < .005). Postherpetic neuralgia was the most
common complication, involving 9.3% of the co-
hort. This sequela was of relatively short duration
«8 weeks) in 45% of patients but lasted longer
than 1 year in 22%. The average age of patients
with this problem was 67 years, significantly
RISK OF v-z INFECTIONS IN DIFFERENT PATlENTS 103

Table 3. Frequencv of Zoster and Predictive Factors in Subgroups at Risk

Incidence Rates Risk Factors for
Group
(reference no.) Zosler Complicalion Zoster Oisseminalion

General populalion 20/10 5 p-y (age 0-4) Dissemination History of chickenpox

8 63/10 5 p-y (age 15--19) 2%% (first year of life only!
Age <20 onl1
Postherpelic neuralgia
0% Acule leukemia
Visceral involvement
0%
29 100/10 5 p-y (age 25--30) Oissemination
Ali ages
400/10 5 p-y (age "'751 2%
Postherpetic neuralgia
131/10 5 p-y (ali I 9.3%
Ocular complicalions
1.9%
Pneumonia
0.2%
Menlngoencephalitis
0.2%
13 Dissemination
Pregnancy and neonalal
7%
Neonatal ar infant 20sler
0%
Congenital defecls

Cancer Palients, 625/10 5 p-y Oissemination HO HO

Ganeral cancer 12% Non-Hodgkin's disease NHL
Population 23.40 (adults) Pain during zoster leukemia Head and neck cancer
66% {in multivariate analysis!
Postherpetic pain
23%
Death
0.3%
H038 Overall cumulative Chemotherapy plus
(ali agesl risk 9.5% (1 yearl radiotherapy
20.6% 13 years! Young age
Risk by treatment: B symptoms (in
Chemolherapy + irradia- mullivariate analysis)
tion 27.3%
Radiotherapy only 11.5%
Chemotherapy only 13.2%
Cancer patienls, Induction and mainlenance Maintenance chemotherapy
smali celi lung cancer39 chemotherapy 8. l % (particularly procarbazine
(adults) or nitrosoureasl
Induction only 1.6%
General cancer population 9% Cutaneous dissemination HO Lymphoma
(children!33 50% ALl
Encephal itis
4%
Poslherpetic pain
9%
Oeath
3%
Pneumonia
3% (ali died)
Meningoencephalitis
3%
Bone marrow transplants 14% Oissemination Allogeneic transplanls Acute graft-v-host
(allogeneic and 36.4% Acute graft-v-hosl diseage
syngeneic}24 Visceral disease
13.3%

(Continued on following page)

104 JAMES J. RUSTHOVEN

Table 3. Frequency o, Zoster and Predictive Factors in Subgroups at Risk (Cont'd)

Incidence Rates Risk Factors for
Group
Irelerence no.l Zoster Complication Zoster Dissemination

Oeath
6.7%
Postherpetic pain
25%
Multiple episodes
8.3%
Bone marrow transplants 22% Dissemination Underlying disease
(autologous}25 21% Lymphomas
Visceral
3%
Postherpetic neuralgia
15%
Cardiac transplants48 .50 6.9-9.1 per 100
patients per year
13% at 6 months
posttransplant
Renal transplant26 7% over lirst 3 years
postransplant
AIDS49.51,54 2,000/105 p-y
IUnited Statesl Cumulative risk age <18:
6 yr: 5%
10yr: 12%
Cumulative risk age "'18:
6 yr: 6%
10yr: 12%
(Alrica}53 15% HIV' Oissemination Severe weight loss
0%
Multiple episodes
29%

greater than the average age of the cohort (P <
.0001). Other complications included ocular com-
plications (1.9%), motor deficit (1%), necrosis of
lesions (0.5%), pneumonia (0.2%), and meningo-
encephalitis (0.2%).
A second study by Guess et al examined 173
cases of zoster in persons aged under 20 years in
the same Rochester population from 1960 to
1981. 28 The same comprehensive database was
used. The data were gathered retrospectively and
therefore cases that were described as questionable
or possible were only included if V-Z virus was
cultured from a vesicle. One hundred seventy-
three residents experienced a first case of zoster in
the 22-year study period. The crude incidence was
42/105 p-y, compared with 131/105 p-y found by
Ragozzino et al over the entire Rochester popula-
tion in an earlier cohort. This is simUar to the crude
incidence found in that study in the same age co-
hort. 29 However, unlike the study of Ragozzino et
al wherein 5% of residents had a second episode of
zoster, none were found in this group. Statistically
significant increases in zoster occurred with in-
creasing age groups, from 20/105 p-Y at ages O to
4 years to 63/105 p-y at ages 15 to 19 (P < .01).
The establishment of a community pediatric ser-
vice at the Mayo Clinic was associated with a sig-
nificant increase in incidence: however, no tempo-
rai trends were noted within the time periods
before or after this event. This suggests that the
true incidence did not increase over this study pe-
riod. As in the study of Ragozzino et al, no sea-
sonal trend was noted and the distribution by gen-
der was equal.
One provocative finding was a highly significant
increase in a history of chickenpox in the first year
of life (9%) compared with the National Health
Interview Surveys of that age group in the general
population in the United States (3.3%, P < .002).
This was even more striking because in only 21 %
of the 173 residents was the age at onset of chick-
enpox documented and the comparison remains
significant even if none of the remaining 137 res-
idents had chickenpox. As a result, a sensitivity
analysis showed that the relative risk of childhood
zoster among children with chickenpox in the first
RISK OF v-z INFECTIONS IN DIFFERENT PAT1ENTS
year of life based on the 173 patients is 2.8 (95%
confidence intervals [C] 1.6 to 4.7), whereas that
derived from the 36 residents for whom the age of
chickenpox was recorded is 20.9 (95% CI 10.9 to
40.1; p < .0000001). The truth lies somewhere in
between, but the risk is clearly significant. This
finding supports the hypothesis of Weller that the
length of time from primary infection to reacti-
vated infection is shortened when primary infec-
tion occurs during pregnancy or in the neonatal
period because of an immature immune system. 32
No such increased risk was found in persons with
chickenpox onset in the second or any other year of
life. There was also no association between zoster
and childhood vaccination within the previous 30
days.
The dermatomai distribution was similar to that
in the Ragozzino study with thoracic dermatomes
involved in 60% of cases, as was the rate of dis-
semination (2%). However, no visceral dissemina-
tion was documented in this study. In general, the
disease was milder and complications fewer than
zoster described in adults 29 ; there were no cases of
postherpetic neuralgia but the incidence of pain
during zoster increased with age (P < .01). Unlike
the 6.1 % incidence of cancer found in adult zoster
patients, only 3% of these children with zoster had
a previously diagnosed malignant cancer. How-
ever, the 13% incidence of zoster among the 15
residents with acute lymphocytic leukemia (ALL)
was 122 times higher than that in the remainder of
the cohort without ALL (95% CI for relative risk;
15 to 181, P < .0003) and compares favorably
with the 10% incidence of zoster identified by
Feldman et al in a selective cancer population of
children seen at achildren's research hospital.33
As in the adult general population,29 there was no
association between zoster and an increased risk of
subsequent cancer.
Whereas such large population-based studies
provide us with valuable baseline information
about the natural history of this disease against
which data from high-risk groups can be com-
pared, they are highly dependent on clinical judg-
ment in establishing the diagnosis of zoster infec-
tion. A recent hospital-based study by Kalman and
Laskin suggests that clinical zoster may actually
represent herpes simplex disease in a surprisingly
high proportion of individuals. 34 This study re-
viewed the charts of patients presenting to hospital
dennatology service over a 3-year period who
105
were diagnosed with herpes zoster by a staff der-
matologist. Only patients without a history of
malignancy, collagen-vascular disease, immuno-
suppressive therapy, or AIDS (ie, who were im-
munocompetent) were included. One hundred
eleven cases were identified. Despite such clinical
selection for cases of true herpes zoster, 6 of 47
cases (13 %) with viral cultures of the lesions grew
herpes simplex. The mean age of these 6 patients
was somewhat younger than the other 105 patients
(35 v 50 years). Four patients had facial involve-
ment and the two others were women with breast
involvement. The 105 patients with no culture or
with lesions cultured for V-Z virus were younger
than the distribution seen in the general population
studies (32% were age 65 or older) but no season,
variation was seen and the dermatomaI distribution
was comparable to those studies. Despite the
younger patients, 91 % of them presented with
pain; this is likely another selection factor because
those without pain may not have sought medical
attention.
Risk oj Zoster During Pregnancy and
in Neonates
Until recently, the outcome of children whose
mothers experienced zoster during pregnancy had
not been prospectively studied. In fact, Enders re-
ported that 13 cases had been reported in the En-
glish literature before 1984 and that not alI of these
were welI documented. Similarly, 13 cases of
zoster in neonates were dubious as to whether ma-
temal zoster preceded them. 15 However, the risk
of developing neonatal zoster seems to be inexo-
rably linked to matemaI zoster during pregnancy;
BrunelI has reported that alI healthy children who
developed zoster during the first 2 years of life had
no history of varicella and alI were preceded by
matemai V-Z infection during pregnancy. 35 In ad-
dition, none of the infants had stigmata of congen-
ital V-Z infection.
A more recent prospective study by Paryani and
Arvin provides more reliable information on the
consequences of intrauterine zoster infection in the
infant. 13 Fourteen women with zoster during preg-
nancy were enrolled between 1978 and 1984
through a university hospital pediatric service. An
obstetrician referred patients on clinical suspicion,
and exposure to varicelIa was evident within 21
days of the onset of rash in alI women. AlI had a
history of varicelIa and herpes simplex was ruled
106
out by immunofluorescent staining of lesions if
they occurred in the lumbosacral area (unfortu-
nately, the dermatomai distribution was not given
and therefore it is not elear whether facial der-
matomes, where Kalman and Laskin found cases
of herpes simplex mimicking zoster, were in-
volved). All pregnancies resulted in live births,
most infants were evaluated at 1 and 2 years of age
but 5 infants were evaluated on1y under l year of
age.
Thirteen of the 14 women were free of eompli-
cations during pregnaney, whereas 1 woman had
mild cutaneous dissemination outside the der-
matomes involved. Nine eases occurred during the
third trimester (64%) and only one during the first.
Zoster infeetion oceurred in only one infant (4%)
at age 7 months who had been exposed to matemal
varicella in the seeond trimester. Of the infants
exposed to maternal zoster, none had physical
anomalies, nine were followed up for at least 1
year, and none developed zoster. This study dem-
onstrates the relatively benign course of matemai
zoster and the 10w risk of neonatal V-z infection
compared with maternał varieella.
Risk oj Zoster In Immunocompromised Patients
In the population study of Ragozzino et al 10%
of patients had one or more of six possible risk
factors for zoster infection. 29 Cancer topped the
list. Guess et al reported a marked inerease in risk
of developing zoster in ehildren with ehiekenpox
under the age of 2 but a relatively lower incidence
of eaneer (3%)28 eompared with the adults in the
same population (6%).29 Whereas the risk of zoster
among ehildren with aeute leukemia was estimated
to be 122 times higher than the rest of the generał
population, this was based on only two cases. The
ineidenee of other małignaneies was a1so low (one
ease each of Hodgkin's disease, [HD], non-
Hodgkin's lymphoma [NHL], and neuroblas-
toma). Therefore, studies of seleeted populations
are required to understand the ineidenee and be-
havior of zoster infeetions in these settings.
Whereas the earliest systematie study of zoster re-
ported that "the elassieal precipitants of zoster-
1ead, arsenie, syphiIis, spinał trauma, and neo-
plasm . . . are well-attested and indubitable, 30
these perceptions were based heavily on anecdotal
reports and personal experiences.
Mazur and Dolin published one of the first eom-
prehensive, albeit retrospeetive, reports on zoster
JAMES J. RUSTHOVEN
infections in a tertiary referral institution. 36 One
hundred seven cases of zoster were found in the
records of the elinieal center of the Nationa1 Insti-
tutes ofHealth between 1953 and 1974. The mean
age was 40 years with a range of 13 months to 94
years. Three patients experieneed two episodes.
Among a long list of underlying diagnoses that
refleeted the tertiary nature of the referrał pattem,
27 eases had HD, 10 had ALL, 7 had other lym-
phomas (NHL), 6 had ehronie lymphocytie leuke-
mia (CLL), and 5 had systemic lupus erythemato-
sus. Based on the number of patients with these
diagnoses seen during that 20-year period, zoster
oecurred signifieantly more frequently among pa-
tients with HD (4.7%) eompared with all other
diseases exeept CLL (2.9%) (P < .Ol). This dom-
inanee of HD as a risk of zoster in this seleet pop-
ulation was also seen in a children's cancer hospi-
tal setting by Feldman et al. 33 Among 1,132
children with malignaneies seen between 1962 and
1972,101 (9%) had a diagnosis of zoster, and 22%
of these had HD, whereas 10% had ALL. How-
ever, it should also be noted that the rates in dis-
ease other than HD and CLL were low (0.06% to
1.5%) and similar to those reported from other
medieal centers. Overal1, 60% of these 101 pa-
tients had an underlying małignaney. The associ-
ation of zoster with other risk faetors was more
inferential. A trend was found toward an assoeia-
tion with spleneetomy among HD patients (P =
.065). Other potential assoeiations with radiother-
apy or prior surgery were not put to statisticał sefU-
tiny. Zoster dissemination was reported in 50% of
patients; whereas the majority involved only skin
dissemination, 3% developed pneumonia and died,
3% had meningoeneephałitis, and 9% experieneed
postherpetie pain. However, no cases of dissemi-
nation to liver or gastrointestinal traet were found.
Whereas the overall association of malignaney
with dissemination was not greater than that with
loealized disease, lymphomas were assoeiated
more frequently with dissemination compared with
loealized disease (63 v 35%, P < .05). Interest·
ingly, this was not observed among HD patients
but the number of eases was small. Therefore, this
study strongly supported the contention that certain
lymphoproliferative malignancies were signifieant
risk factors for the development of zoster com-
pared with other diseases ineluding other malig-
naneies. However, whereas the risk of dissemina-
tion in this tertiary referral population was
RISK OF v-z INFECTIONS IN DIFFERENT PATIENTS
sevenfold higher than that reported in the Roches-
ter study,29 the morbidity was relatively low and
only one death (0.9%) was attributable to zoster.
The first detailed study of zoster in cancer pa-
tients was reported by Schimpff et al. 21 Thirty-
seven patients with clinical evidence of V-Z infec-
tion were identified among 419 cancer patients
seen at acancer research center in a 24-month
period between July 1969 and June 1971. 21 The
mean age was not given but ages ranged from 8 to
61 years of age. However, two of the three chil-
dren under age 10 had varicella and the remaining
child developed two separate episodes of zoster.
Furthermore, as mentioned earlier, six patients
presented with disseminated lesions resembling
zoster or varicella but no antecedent dermatomaI
zoster. Therefore, 31 of the 39 patients presented
with classical zoster or disseminated zoster.
As in the studies of Feldman et al 33 and Mazur
and Dolin,36 zoster occurred with the highest fre-
quency among patients with HD (25%), folIowed
by other lymphomas (8.7%); again, only 1.8% or
less of patients with acute leukemias or solid tu-
mors seen during this period developed zoster (the
six cases of questionable zoster were included in
these figures). In this series, splenectomy was not
an added risk to the development of zoster. The
investigators reported a statistically significant as-
sociation between both radiation treatment in the
previous year and the presence of advanced stage
at diagnosis with the development of zoster among
the patients with HD. This association was not
significant with chemotherapy. However, the
methods used for the statistical analysis were not
presented and no multivariate analysis was re-
ported to test the independence of each putative
risk factor from others. Whereas 22 of the 31 pa-
tients who received radiation therapy developed
zoster in the radiation field (one half of them
within 3 months of receiving the therapy), it could
not be determined whether this could have been
also linked with the presence of tumor in the area.
Eight of 37 patients (22%) had more than one ep-
iSode of zoster identified, with one half of these
experiencing both episodes during the 2-year study
period.
In summary, this study supports the high sus-
ceptibility of patients with HD to develop zoster
and suggests that prior radiation therapy and ad-
vanced stage at diagnosis may predispose to the
development of zoster. However, no multivariate
107
analysis of risk factors for zoster or dissemination
was perfonned. This study also showed that the
clinical presentation of zoster may vary from the
classical description in these patients and care
should be taken to distinguish cases of atypical
zoster from those of herpes simplex by using im-
munofluorescent and culture methods.
Two subsequent studies focused exclusively on
zoster among patients with HD. Ruckdeschel et al
studied the immunoresponsiveness of 32 patients
with HD and 12 healthy donors to v-Z viruS. 37
Peripheral blood V-Z antibody titers were mea-
sured as was peripheral blood-derived lymphocyte
responsiveness to membrane-associated V-Z anti-
gen in vitro. Patients were selected according to
whether they had previously untreated HD (6 pa-
tients), received recent (within 6 months) radiation
therapy with or without chemotherapy (6 patients),
had recurrent HD (8 patients), or were in long-term
(over 1 year) remission. Patients with recent radio-
therapy, those with recurrent disease, and those in
remission but who received previous radiation and
chemotherapy, had significant1y impaired respon-
siveness compared with controls. Those previously
untreated had significant1y lower values as well but
the variability was greater in this subgroup. Two
patients with recurrence were tested when dissem-
inated zoster was evident. The patient whose lym-
phocytes failed to respond went on to progressive
dissemination over 10 days and recovered 3 weeks
later, only to die of recurrent HD l Y2 months later.
The patient with moderate responsiveness had a
mild episode with no new lesions after 2 days and
fuH recovery by 2 weeks. Treatment resulted in
complete remission.
This study supports the hypothesis that certain
patients with HD have impaired cellular immunity.
Two patients with recurrent HD and no lympho-
cyte responsiveness developed zoster several
months after testing. Whereas no patients with nor-
maI responsiveness were reported to have devel-
oped zoster after testing, the folIow-up time was
not reported for most patients. Therefore, it re-
mains unclear whether lymphocyte responsiveness
can be reliably predictive of the occurrence of
zoster.
More recent1y, Guinee et al retrospectively stud-
ied the incidence of zoster among 717 patients with
HD registered at six cancer centers between Janu-
ary 1978 and December 1981. 38 Most patients
were folIowed up for at least 18 months. The mean
108
or median age was not given but five centers in-
cluded patients under age 16. Data was obtained
by chart review; zoster was diagnosed by a physi-
cian at the cancer center in two thirds of cases and
by another physician in 22%. In 13%, the diagno-
sis was based on the patient's report to the physi-
cian. Univariate and multivariate analyses were
performed to determine the relationship between
patient characteristics and the incidence of zoster,
with the latter performed using Cox's proportional
hazards model. Patients folIowed up for 18 months
or more were examined for risk factors for zoster
occurrence using a stepwise logistic regression
procedure. Six dichotomous variables including
treatment, age, stage of HD disease, symptoms,
and laparotomy were considered for the regression
procedure, with stage entered in two different
ways.
One hundred sixteen cases of zoster were iden-
tified with the earliest of these occurring 2 months
after treatment for HD began. The largest number
of cases clustered around 8 months after therapy
began, with an incidence rate of 1.9% during that
month. The cumulative risk of zoster was 9.5%
after 1 year, 16.6% after the second year, and
20.6% after the third year. The incidence was
compared between patients receiving different
treatment modalities using a 36-month life-table
calculation and were as follows: chemotherapy-
radiation-chemotherapy (C + R + C), 27.3%;
radiotherapy only, 11.5%; chemotherapy, 13.2%.
Those receiving C + R + C had a significantly
higher risk compared with those receiving R (P =
.(01) or C (P = .018). Among those receiving
both modalities, the risk was significantly lower if
radiation folIowed chemotherapy (19.8%) com-
pared with the sandwich technique (that is, C + R
+ C) (P = .011). In the multivariate model, treat-
ment was classified as combined- versus single-
modality therapy and age groups were cases less
than age 16 or age 16 and older. When factors
found to be significantly associated with the oc-
currence were added sequentially into the regres-
sion model (treatment, folIowed by B symptoms
[fever, night sweats or weight loss], then age) all
three factors contributed independently to risk with
increased risk associated with C + R therapy,
younger age and the presence of B symptoms. To-
gether, this trio of factors predicted a 4.3-fold
greater risk of zoster compared with those with the
dichotomous counterparts. When the 81 patients
JAMES J. RUSTHQVEN
who were folIowed up for over 18 months were
entered into this model, age and treatment were
again found to affect independently the risk of
zoster, with younger age and C + R therapy in-
creasing the risk 3.5-fold compared with the other
combined-risk group. Stage of disease and laparot-
omy were not significant factors in any of these
analyses. Among the adults (age 16 and older), no
trend in incidence was noted when lO-year groups
were compared, although the 10west attack rate
occurred in the oldest (age 65 years or older) group
(6.2%). This is opposite to the pattem seen in the
general population but may reflect the more ag-
gressive therapy administered to younger patients.
This welI-executed, large study demonstrates
the impact of aggressive therapy, B symptoms,
and young age on the risk of developing zoster in
patients with HD. Unfortunately, the investigators
did not analyze this large database as to the risk of
dissemination, severity and duration of zoster-
related complication, and other clinical features.
The impact of combined-modality therapy on risk
was particularly impressive. The chemotherapy
drugs and doses were not given so the impact of
different regimens was not assessed. However,
similar agents are used for the treatment of smalI
celI lung cancer and Feld et al found a differential
effect of chemotherapy regimens on the incidence
of zoster in these patients. 39 Patients treated for
smalI celI lung cancer at a large provincial cancer
hospital between 1962 and 1975 were retrospec-
tively reviewed. Within this population, two co-
horts of patients were prospectively folIowed up
for zoster during chemotherapy clinical triais. Co-
horts of patients receiving an earlier regimen
(cyclophosphamide, doxorubicin, vineristine for
induction then lomustine, procarbazine, and meth-
otrexate for 1 year of maintenance therapy) and a
regimen used later in the study period (the induc-
tion regimen alone at higher doses of cyclophos-
phamide) were eompared for the risk of zoster.
Thoracic radiotherapy was similar in both eohorts
but prophylaetic eranial irradiation was adminis-
tered to responding patients in the later eohort. The
cumulative probability of developing zoster was
signifieant1y greater for the cohort reeeiving induc-
tion plus maintenance therapy compared with the
induetion-only eohort (P = .031) or eompared
with historieal controls with smalI celllung cancer
who received radiation alone or single-agent cy-
clophosphamide (P = .007). Zoster oecurred 5 to
RISK OF v-z INFECTION5 IN DIFFERENT PATIENTS
25 months after treatment began in the earlier
grOUP with 11/13 occurring during maintenanee
therapy, whereas zoster occurred from 2 to 7
months after treatment began in the induction-only
group. Survival was simi1ar in both cohorts. The
overall risk in the historical cohort was 1.6% (that
for non-smalI celIlung cancer was 0.8% during the
same period), whereas that for the induction and
maintenance versus induction onIy cohorts was
8.1% and 1.6%, respectively. This study provides
evidence that the risk of zoster mayaiso be related
to the type of combination chemotherapy; mainte-
nance therapy with nitrosoureas or procarbazine
may increase the risk. This mayaiso be important
in HD because patients received chemotherapy that
inc1uded procarbazine as part of their treatment
during the study era. Now that a regimen without
procarbazine has become standard therapy for HD
(doxorubicin, bleomycin, dacarbazine, vinblas-
tine) it will be interesting to determine if the risk in
patients receiving C + R decreases over the next 5
to 10 years.
Whereas the risk of zoster among HD patients
has been c1early illustrated, the risk for adults with
eancer of other types had until recent1y been based
on small numbers of patients over a short time
period, often under conditions of local in-hospital
epidemics. 21 ,22 Rusthoven et al reviewed the inei-
dence of zoster as recorded in the health records of
a large comprehensive cancer center between 1972
and 1980. 23 Whereas retrospective in nature, this
study provided data in a large general eancer pop-
ulation as to risk of zoster as well as the risk and
eonsequences of dissemination. Correlates with
faetors sueh as chemotherapy, type of cancer, ra-
diation therapy, etc, were also made using multi-
variate analyses. Seven hundred sixty-six cases of
zoster were identified among 740 adult patients
with cancer (ie, patients aged 16 years or older).
FolIow-up data were available for a minimum of 5
years for all patients. The cases involved both out-
patients and hospitalized patients. The cumulative
aetuarial incidence rate for the eancer population 5
years after diagnosis was 625/105 p-y, in contrast
to the overalI incidence rate of 1311105 p-y calcu-
lated for a 15-year period in the general popula-
. 22
tłon (Table 3). Whereas the proportion of pa-
tients with solid tumors exceeded that with
hematologie malignancies (56.7% v 43.3%), the
eumulative incidence at 5 years was highest among
patients with lymphomas and leukemias (HD,
109
14%, leukemia, 10%; NHL, 5%; breast, lung, and
gynecologic malignancies, 2% each). As in the
study of Ragozzino et al, the proportion offemales
with zoster was higher than males, but this could
reflect the number of patients with underlying fe-
male malignancies. Three percent of patients had
multiple episodes; multiple episodes were more
frequent among patients with lymphomas, leuke-
mia, and myeloma (14% of episodes seen in these
patients) than patients with solid tumors (2%) and
the most common underlying tumors among these
patients with HD and NHL (63% of patients with
multiple episodes). Dissemination occurred in
12% of patients; dissemination developed more
frequent1y in patients with extensive tumors at di-
agnosis (P = .002) and more often among patients
with active tumor known at the time of the diag-
nosis of V-z infection compared with those in
complete remission (P = .011). Death attributable
to zoster was low (0.3%). No epidemics or sea-
sonal variations were noted during the study pe-
riod. The involvement of thoracic dermatomes
documented the sites of involvement (54%). In an
attempt to correlate the site of malignancy and the
site of subsequent zoster, a chi-squared analysis
was performed on patients with no regional over-
lap of zoster (ie, purely thoracic, lumbar, or sacral
involvement). A significant correlation was found
(P = .0001) with significantly more patients with
lung and breast cancer developing thoracic zoster
and those with gynecologic tumors developing
lumbar and sacral zoster. By dermatomal regional,
slightly more than two thirds of patients who re-
ceived previous irradiation developed zoster within
the radiation field, except for trigeminal zoster in
which only 29% developed zoster within the field.
Pain was reeorded in 80% of cases where the pres-
ence or absence of pain was recorded and in 66%
of alI patients studied. Twenty-three percent of all
patients had postherpetic pain with 20% of these
experieneing pain more than 6 months after the
episode of zoster began (postherpetic neuralgia
was diagnosed in 9% of patients in the study of
Ragozzino et al and none in the pediatrie study of
Guess et al. 28) •
The risk of disseminated zoster was also as-
sessed. 40 After univariate analysis of selected van-
ables for zoster risk (including site of original tu-
mor, tumor status at diagnosis, tumor status at time
of zoster, site of zoster, age, gender, and previous
treatment), a diagnosis of HD, decreasing age,
110
chemotherapy with 6 months of zoster, and exten-
sive tumor at tumor diagnosis were considered sig-
nificant ńsks; this was consistent with the findings
in the previously noted study of HD patients. 38
When added into a multivańate analysis, only the
diagnosis of HD (P < .001), NHL (P = .016),
and head and neck cancer (P = .043) predicted an
increased ńsk, whereas complete remission at di-
agnosis of zoster and previous radiation therapy
predicted a reduced ńsk.
This study and that of Guinee et a1 38 provide the
largest and most thoroughly analyzed database of
cancer patients as to their ńsk of zoster and dis-
semination of zoster. At this time, no recommen-
dations can be made for the use of prophylactic
antiviral agents, although these data can be useful
to identify patients in whom early diagnosis may
lead to reduced morbidity and perhaps reduced ńsk
or dissemination by instituting early antiviral ther-
apy.
Some investigators have suggested that zoster
may predict for the occurrence of cancer41 ,42 and
some have recommended work-up for occult ma-
lignancy in certain situations. 43 ,44 Ragozzino et al
followed up 590 residents of their Rochester pop-
ulation for 9,389 p-y after the diagnosis of
zoster. 45 The relative ńsk of developing cancer
was 1.1 with 95% C of 0.9 to 1.3. Subgroups
including those with dissemination and those with
postherpetic neuralgia had no increased ńsk of
cancer. The relative ńsk of developing colon can-
cer regardless of gender and of developing bladder
cancer among women was increased (colon cancer:
2.2, 95% CI 1.3 to 3.3, P < .05; bladder cancer:
2.0, 95% CI 0.8 to 4.2), reaching statistical sig-
nificance only for colon cancer and only at 6 or
more years after zoster. However, this finding was
appropńately tempered by the investigators be-
cause of the reduced power of this post hoc sub-
group analysis. The finding required confirmation
before cHnical significance can be claimed.
Organ transplantation defines a growing group
of immunocompromised patients at ńsk for zoster.
In the study of Locksley et al, 195 cases of zoster
developed among 1,394 patients (14%) receiving
allogeneic (n = 1,270) or syngeneic (n = 124)
BMT from 1969 to 1982.24 Childhood vańcella
was evident in 41 %. Dissemination occurred in
36.4%; on1y skin was involved in 23.1 % (64% of
disseminated cases). Whereas the incidence of vis-
ceral dissemination and death (13.3% and 6.7%,
JAMES J. RUSTHOVEN
respectively) were relatively high compared with
the general population ( < l % and approaching 0%,
respectively),22 these figures were significantly
lower than those for vańcella cases in the same
population (visceral dissemination, 44.4%; death,
27.8%; P < .05). The median age was 20 years
(range 2 to 52 years) with acute leukemia as the
diagnosis in two thirds of cases. The median time
to onset of zoster was 5 months with over 85%
occurńng within l year. More patients developing
zoster less than 9 months after BMT had leukemia,
whereas the later development of zoster was more
likely associated with aplastic anemia.
The clinical presentation showed a high propor-
tion of patients with cranial nerve involvement and
less dominance of thoracic involvement compared
with the general population and cancer patient
studies. The morbidity among those with cranial
nerve involvement was particularly stńking; 19/32
patients developed comeal or facial scarring, cra-
nial nerve palsy or heańng loss, or death. Five
patients had more than one episode after transplant
(2.6%) and 11 patients had a pretransplant history
of zoster (5.7%). In all eight patients in whom the
previous site of zoster was known recurrence de-
veloped in the same site. Postherpetic pain oc-
curred in 25%, much higher than the general pop-
ulation (9%)29 but similar to that reported in a
general cancer population (23%).23 Factors that in-
dependently predicted for zoster or vańcella in-
cluded allogeneic transplants and acute graft-
versus-host disease, whereas a diagnosis of
chronic myloid leukemia was associated with less
ńsk. The ńsk of dissemination or death was only
associated with acute graft-versus-host disease (P
= .006 and P = .0002, respectively).
ABMT has become an altemative to allogeneic
BMT for some patients with lymphomas, leuke-
mias, and solid tumors. Schuchter et al recently
reviewed the ńsk of zoster among 209 patients
receiving ABMT at a single tertiary referral cancer
center between January 1983 and December
1987. 25 Unlike the study of Locksley et al, only
patients surviving longer than 100 days after trans-
plant were included. The data were collected ret-
rospectively and a minimum follow-up of 6
months was achieved. One hundred fifty-three pa-
tients received ABMT and survived long enough to
be included, whereas 56 patients were excluded on
this basis. Zoster developed in 33 evaluable pa-
tients (22%), a figure higher than the previouS
RISK OF v-z INFECTIONS IN DIFFERENT PATIENTS
study in allogeneic BMT patients (14%). AlI pa-
tients had pretransplant serology positive for V-Z
antibody, indicating reaetivation as the likely
mechanism of zoster infeetion in most eases. With
a median follow-up of l year, the actuarial inci-
dene e of V-Z infection (>75% of eases were
zoster) at l year was 25% (95% CI 22% to 41 %)
and the overall actuarial probability was 40% (95%
CI 27% to 55%). These figures are higher than the
cumulative incidence for HO 3 years after diagno-
sis in the study of Guinee et al (20.6%?8 and for
HO 5 years after diagnosis in the study of
Rusthoven et al (14%),23 wherein HO was assoei-
ated with the highest cumulative risk of zoster
compared with other cancers. However, the latter
study was exclusively in adults, whereas the me-
dian age of patients with zoster in this study was 23
years with a range from 7 years to 58 years. Lym-
phoma was the underlying diagnosis in 17/33
cases, whereas acute leukemia was present in 14
cases. The development of a second episode of
zoster was comparable to that in the Loeksley et al
study (3.0%) and the proportion of patients with
cranial dermatomes involved was even higher than
that study and was similar to those with thoracic
involvement (33%). However, unlike the sugges-
tion in the Rusthoven et al study that the site of
malignancy and previous radiotherapy may influ-
ence the location of zoster,23 no such assoeiation
was found in this study. Cutaneous dissemination
occurred in only 6/33 (18%), whieh was compara-
ble to the allogeneic BMT reeipients (23.1 %), but
a lower proportion of these patients developed vis-
cerał dissemination as well (14% v 36% for allo-
genek BMT). As in the Locksley et al study, com-
plications of V-Z infection were frequent (32%)
with postherpetk neuralgia and scarring leading
the list of problems. However, cranial involvement
was not as severe with no ophthalmic eompliea-
dons.
Of multiple potential risk faetors that were en-
tered into a proportional hazards model (including
age, gender, V-Z titer, disease status, preparative
BMT regimen, and previous radiation therapy),
only underlying disease was a significant risk fae-
tor for zoster. Not surprisingly, patients with lym-
phomas were more likely to develop V-Z infection
compared with those with leukemia or solid tumors
(P < .002).
These two studies show the similarities of zoster
in these two populations including high ineidenee,
111
relatively early time to onset, and severe morbid-
ity. However, the true ineidence of early zoster is
likely higher than reported by Sehuchter et al be-
eause they did not inelude patients whose survival
was less than 100 days posttransplant. The eauses
of death of those exeluded by this criterion were
not mentioned. The risk factors for developing
zoster seem to be different between these popula-
tions. Whereas acute graft-versus-host disease was
the major prediction in allogeneic BMT patients,
the underlying malignancy was the only indepen-
dent risk factor among those receiving ABMT.
These comparisons suggest that immune factors
associated with acute graft-versus-host disease
may override underlying malignancy as the main
risk factor for zoster: however, far fewer patients
with alIogeneic BMT had underlying lymphoma
and this may partly account for the difference in
risk factors. What is elear is that intensive antican-
cer therapy using BMT puts patients at increased
risk for zoster infection as well as significant risk
of zoster-associated morbidity and mortality. As a
result, acyelovir has been used in an attempt to
prevent zoster infection in allogeneic BMT recip-
ients. Perren et al conducted a randomized, pla-
cebo-controlled comparison of patients receiving
intravenous followed by oral acyelovir compared
with placebo after allogeneic BMT for leukemia. 46
Although the study was small (82 patients), no
patient receiving acyelovir intravenously for 23
days followed by oral acyelovir for 6 months de-
veloped zoster versus six patients receiving pla-
cebo for the same time period (P = .006). This
was similar to an earlier, small randomized
study.47 However, unlike the earlier study, Perren
et al continued to follow up patients for 6 months
after therapy stopped. Ouring this time, two more
patients previously receiving placebo developed
zoster (total 8 eases), whereas six cases developed
in the acyelovir-treated group. Therefore, the in-
cidence of zoster after l-year posttransplant was
the same; furthermore, no cases of disseminated
zoster occurred.
This study demonstrates that prophylactic acy-
clovir in this setting may delay rather than prevent
zoster. Beeause no dissemination oceurred, acy-
clovir may be as effeetive in preventing this prob-
lem if used after localized zoster has developed.
Whereas more prolonged use of prophylactic acy-
elovir may reduce the cumulative incidence of
zoster, this praetice could also be associated with
112
the development of resistance strains as has been
reported among AIDS patients receiving oral acy-
clovir to suppress zoster infeetion. 48 However, this
risk of resistant strains may be even lower with
ehronic use in transplant patients in whom immu-
nocompetenee improves over time and treatment is
administered before clinical infeetion develops.
Zoster also occurs at increased frequeney among
patients with other organ transplants. Rand et al
observed the incidence of zoster among 51 cardiac
transplant patients. 49 Twenty-three patients were
followed up prospeetively for 3 months, whereas
the remaining patients were reviewed retrospee-
tively at between 6 months and 6 years posttrans-
plant. The ineidence was 9.1 per 100 patients per
year (9.1I100/yr) in the patients folIowed up pro-
speetively for 3 months eompared with 6.9/100/yr
for those studied retrospeetively. The 11 observed
cases were evenly distributed over the first 2 years
after transplant. The investigators found that re-
dueed lymphoeyte transfonnation responses and
reduced interferon produetion to herpes zoster an-
tigen eorrelated with inereased suseeptibility to
zoster infeetion, a finding similar to that reported
by Ruekdesehel et al in patients with HD.3 7 Preik-
saites et al also reported a high incidenee among
patients reeeiving cardiae transplant from 1976 to
1978 and from 1980 to 1981. 50 At 6 months post-
transplant, 8/64 patients (13%) developed zoster,
whieh was loealized in all but one case. This was
similar to the previous study (9170 at 3 or more
months after transplant)49 and the incidenee was
not affeeted by ehanges in immunosuppressive
therapy over the study period. Among renal trans-
plant reeipients, the ineidenee is relatively low
(7%) over the first 3 years posttransplant. 26 The
lower ineidenees in renal and eardiae transplants
compared with BMT likely reflects the contribu-
tion of the underlying malignaney to the oeeur-
renee of zoster in the latter population.
Occurrence and Manifestations in Patients With
HIV Infection
In addition to BMT, the relatively new clinical
setting within whieh zoster has emerged at greater-
than-nonnal frequeney is HIV infeetions. Beeause
HIV infeetion has been associated with eertain risk
groups, studies of zoster infeetion have generally
foeused on these groups. Eyster et al reeently fol-
lowed up prospeetively hemophiliaes from 16 he-
mophilia centers for the deve10pment HIV-1 anti-
JAMES J. RUSTHOVEN
bodies and eight HIV-related non-life threatening
eonditions, one of whieh was herpes zoster infee-
tions. 51 Patients were registered between 1982 and
1990 in eight centers, 1987 and 1990 in 12 eenters,
and 1989 and 1990 in 16 eenters. HIV positivity
was tested in samples going back as far as 1976.
The cumulative 6-year and lO-year ineidence of
HIV-related zoster (as a percentage ± standard er-
ror [SE]) in the subgroup of 130 ehildren and 193
adults in whom the time of seroeonversion was
known as 5 ± 2 and 14 ± 6, respeetively, in
patients less than 18 years of age, and 6 ± 2 and 12
± 3, respectively, for adults aged 18 years and
older. The eumulative 2-year incidence of AIDS
after the development of zoster (as a pereentage ±
SE) was 26 ± 9 for adults and 7 ± 6 for ehildren.
However, after adjustment of CD4 counts and du-
ration of HIV infeetion, zoster was not signifi-
cantly associated with the development of AIDS in
either age group. The cumulative risk rates for
AIDS in the adult groups were very similar to
those reported by Melbye et al among homosexual
men in New York City (22.8% at 2 years; 34.8%
at 3 years. 52 In this latter study, the investigators
reported a significant association between severe
pain and the severity of zoster in general and the
risk of AIDS, using a proportional hazard model.
However, whereas the investigators suggested that
such infonnation may help AIDS counselors in
predieting the development of AIDS in HIV-
positive homosexual men, a prospeetive cohort
study following the development of AIDS in HIV-
positive men with and without zoster would be
more convincing in showing whether zoster is truly
predictive or just another consequenee of profound
immunosuppression associated with the develop-
ment of AIDS.
HIV infection is also endemie in some areas of
Afriea. Colebunders et al examined the relation-
ship between zoster and HIV infeetion among hos-
pitalized and nonhospitalized patients treated in a
large hospital in Kinshasa, Zaire. 53 HIV-positive
patients hospitalized August through November
1983, inpatients admitted several months later re-
gardless of HIV status, and patients referred to
physicians at the hospital with a history of zoster or
who developed zoster between April 1985 and Jan-
uary 1987 were studied as distinct groups. Among
the hospitalized patients, the overall ineidence of
zoster was 11 %; zoster was signifieantly more fre-
quent among HIV-positive eompared with HIV-
RISK OF v-z INFECTIONS IN DIFFERENT PATIENTS
negative patients (15% v 1%, P = .0005). Among
146 patients referred with a history of zoster (mean
age 29 years), the positive predictive value of a
bistory of zoster for predicting HIV infection by
serology was 91 %. The presence of severe weight
10ss (undefined in the study) added significantly to
the predictive value (P = .(01) and a11 patients
with recurrent zoster (n = 27) were HIV seropos-
itive. Among seropositive patients, 40% devel-
oped zoster as the initial clinical manifestation of
HIV infection, 79% had at !east one episode of
zoster, whereas 29% of these had multiple epi-
sodes. Except for the exclusive occurrence of re-
current zoster among only HIV seropositive pa-
tients, clinical characteristics of zoster were
similar among seropositive and seronegative pa-
tients. No cases of dissemination were reported.
No gender distinctions were evident among the
107 men and 177 women who were seropositive.
The data from these studies demonstrate that
zoster occurs in frequencies similar to patients with
lymphomas among subgroups of patients at risk for
AIDS who are HIV seropositive. Furthermore, the
occurrence of zoster has predictive value for the
development of AIDS among these patients, par-
ticularly if accompanied by weight loss and severe
symptoms related to zoster infection. These fea-
tures are similar to geographically distant groups
as well. Friedman-Kien et al reported an incidence
of zoster of 2,000 per 105 p-y in a retrospective
study of 300 patients with AIDS-associated Kapo-
si's sarcoma,54 7- to l5-fold higher than that re-
ported in general population studies. 29 ,30 In the
American outpatient population at high risk for
AIDS, the incidence of zoster and the incidence of
AIDS among seropositive patients were strikingly
similar to the African study. Wbat is somewhat
surprising is the low incidence of dissemination
and relatively mild clinical features. However,
without clinical evidence of AIDS, the immune
status of HIV seropositive patients may allow for
containment of the infection.
When dissemination does occur in patients with
HIV infection, the consequences are often more
devastating than in less immunocompromised pa-
tients. These zoster infections can be associated
with atypical lesions55 with chronic and progres-
sive encephalitis,56 and with the development of
chronic cutaneous infection associated with acy-
clovir resistant fo11owing chronic or multiple treat-
lIlents with acyclovir.48
113
PREVENTION OF VARICELLA-ZOSTER
INFECTION: IMMUNOGLOBULIN AND VACCINES
First available in 1944 for the prevention and
treatment of measles infections, gamma globulin
has been tested as a treatment for V-Z infections.
In a nonrandomized, prospective cohort study,
Ross fo11owed the course of chickenpox in 322
primary cases, inoculated 242 secondary famUy
contacts with intravenous pooled gamma globulin
(PIG), and fo11owed another 209 secondary cases
as uninoculated controls between 1957 and 1960. 3
No reference was made as to the matching of
treated subjects and controls as to age, gender,
geography, etc. However, most ofthe 318 families
included one primary and one or more secondary
cases. Gamma globulin was administered within 3
days of exposure and patients were assigned
through an undisclosed method to one of five doses
ranging from 0.1 to 0.6 mL/lb. The secondary at-
tack rates were 87% for both treated and untreated
cases. There was no correlation between failure to
contract chickenpox and age or severity of the pri-
mary case. Similarly, the incubation period was
not affected by gamma globulin treatment. Prodro-
maI symptoms were milder or absent in a11 but the
lowest dose groups compared with controls but sta-
tistical correlations were not reported. At best, this
study suggested that gamma globulin may reduce
the morbidity associated with secondary attacks of
V-z infection, particularly at the higher doses but
no preventative effect in attack rates was seen.
A second immunoglobulin preparation, zoster
immune globulin (ZIG), became available in the
late 1960s. Produced from patients convalescing
from zoster infections, this product could prevent
chickenpox if administered within 3 days of expo-
sure in immunocompetent and immunocompro-
mised patients. Whereas complete protection from
primary and secondary family exposure could be
achieved in normal children given 2 mL within 72
hours of household exposure,57 varicella did occur
after this dose in children with leukemia. In a sub-
sequent study of 9 children with acute leukemia (7
patients) thymic aplasia (1 patient) or renal trans-
plant (l patient), patients received 5 mL of ZIG
within 48 hours after household exposure. 58 Two
children developed varicella after exposure to sec-
ondary cases at 4 weeks after ZIG therapy and one
child developed varice11a 21 days after ZIG, but
varicella infection was mild in a11 three cases. The
114
investigators presented these data as evidence that
ZłG can prevent or modify varicella, probably
based on the expected secondary household attack
rates of greater than 90% reported in the general
population. 3 Stevens and Merigan conducted a
randomized trial of PIG and ZIG to compare the
rates of dissemination when these agents were ad-
ministered within 9 days of the development of
localized zoster in patients with malignancy or on
cytotoxic drugs including corticosteroids. 59 The
power of the study was relatively small; 47 patients
receiving ZIG, whereas 50 received PIG. Dissem-
ination developed with similar frequency in both
groups (30% and 34%; P = NS), despite the much
higher antibody titer in ZIG. Similarly, morbidity
did not differ between the groups. Unfortunately
the small sample size results in a high risk that a
true difference in these therapies could have been
missed. Combining these data with the other non-
randomized trial it cannot yet be concluded that
immunoglobulin therapy with these preparations
can reliably prevent V-Z infection in immunocom-
promised patients.
A third preparation became available in the early
1980s, derived from outdated, pooled plasma se-
lected for complement-fixation titers greater than
1: 16. Zaia et al showed the effectiveness of this
V-Z immunoglobulin (VZIG) preparation in pre-
venting varicella in immunocompromised pa-
tients. 60 However, as with ZIG,59 VZIG has not
been shown to modify zoster infections. The most
recent preparation is intravenous serum immune
globulin (lVIG), which produces antibody levels
comparable to VZIG. 61 Produced from the Cohn
fraction 2 of pooled plasma, this preparation is
more practical and convenient to administer, is
readily available, and avoids the risk of hepatitis.
Whereas no comparator trial with other prepara-
tions has been performed, the equal effectiveness
of the earlier three preparations in preventing
chickenpox in immunocompromised patients pre-
dicts that IVIG can substitute for VZIG as therapy
for those patients unable to receive IVIG.
Live attenuated varicella vaccine is highly effec-
tive in reducing the risk of secondary V-Z infection
in susceptible (V-Z antibody-negative) individuals
REFERENCES
l. Gershon AA, Steinberg SP: Antibody responses to vari-
cella-zoster viros and the role of antibody in host defence. Am
J Med Sci 282:12-17,1981
JAMES J. RUSTHOVEN
exposed to V-Z virus. Weibel et al compared 468
vaccinated, susceptible, nonimmunocompromised
children with 446 placebo-treated children in a
double-blind, randomized trial. 62 The vaccine was
100% effective of preventing varicella, with 39
cases in the placebo group and none in the vacci-
nated group during a 9-month surveillance period.
Among children with leukemia who have an in-
creased risk of varicella infection and a greater risk
of severe consequences compared with healthy
children,2,20 the vaccine confers protection against
varicella, with children benefiting more than
adults. 63 Furthermore, Hardy et al have shown that
the incidence of zoster is significantly lower in
vaccinated children with leukemia compared with
their unvaccinated counterparts who develop nat-
ural varicella infection (vaccinated: 0.80 cases/100
p-y; natural: 2.46 cases/100 p-y; P = .01).64 How-
ever, these children also tend to have the most
serious side effects of vaccination, including
rashes resembling natural varicella that may re-
quire antiviral therapy and hospitalization.
Whereas live vaccines may have more advan-
tages than immunoglobulin therapy, including a
reduction in the risk of both chickenpox and zoster
in exposed, immunocompromised patients, it is
not yet available in North America. Guidelines for
the use of VZIG or IVIG include its administration
to newboms of mothers with varicella occurring
within the 5 days before or 48 hours after delivery
and to susceptible immunocompromised patients
exposed to a household contact, an infectious hos-
pital roommate, or a hospital staff worker. 65 A
major problem is the failure to treat susceptible
immunocompromised patients because of a failure
to identify exposure. Paryani et al reported that
48% of their patients who developed varicella were
in this group. 61 Whereas treating patients with
IVIG produces protective levels of antibodies for 3
to 4 weeks, the full scale use of IVIG or VZIG for
all such patients during peak times of V-Z infec-
tion to protect such patients may not be cost-
effective, and a cost-effectiveness study of this ap-
proach compared with vaccination may be required
in the future when a vaccine is approved in the
United States or Canada.
2. Preblud SR, Orenstein WA, Bart KJ: Varicella: Clinical
manifestations, epidemiology, and health impact in children.
Pediatr Infect Dis 3:505-509, 1984
RISK OF v-z INFECTIONS IN DIFFERENT PATIENTS
3. Ross AH: Modification of chicken pox in farnily contacts
by administration of gammaglobulin. N Engl J Med 267:369-
376, 1962
4. Preblud SR: Age-specific risks of varicella complications.
Pe(\iatrics 68:14-17, 1981
5. Bullowa JGM, Wishik SM: Complications of varicella. I.
1beir occurrence among 2,534 patients. Am J Dis Child 49:
923-926, 1935
6. F1eisher G, Henry W, McSorley M, et al: Life-
tbreatening complications of varicella. Am J Dis Child 135:
896-899, 1981
7. Wallace MR, Bowler WA, Murray NB, et al: Treatrnent
of adult varicella with oral acyclovir. A randomized, placebo-
controlled trial. Ann Intern Med 117:358-363, 1992
8. Triebwasser JH, Harris RE, Bryant RE, et al. Varicella
pneumonia in adults. Medicine 46:409-421, 1967
9. Hockberger RS, Rothstein RJ: Varicella pneumonia in
adults: A spectrum of disease. Ann Emerg Med 15:931-934,
1986
10. Gray GC, Palinkas LA, Kelly PW: lncreasing incidence
ofvaricella hospitalizations in the United States army and navy
personnel. Pediatrics 86:867-873, 1990
11. Sever JA, White LR: Intrauterine viral infections. Annu
Rev Med 19:471-486, 1968
12. Siegel M, Fuerst HT, Peress NS: Comparative fetal mor-
tality in matemai virus disease: A prospective study in rubella ,
measles, mumps, chicken pox, and hepatitis. N Engl J Med
274:768-771, 1966
13. Paryani SG, Arvin AM: lntrauterine infection with va-
ricella-zoster virus after matemai varicella. N Engl J Med 314:
1542-1546, 1986
14. Siegel M: Congenital malformations following chicken-
pax, measles, mumps, and hepatitis: Results of a cohort study.
JAMA 226:1521-1524, 1973
15. Enders G: Varicella-zoster infection in pregnancy. Prog
Med Virol 29:166-196, 1984
16. Meyers JD: Congenital varicella in term infants: Risk
reconsidered. J Infect Dis 129:215-127, 1974
17. Brunell PA: Fetal and neonatal varicella-zoster infec-
tions. Semin Perinatol 7:47-56, 1983
18. Baba K, Yabuuchi H, Takahashi M, et al: Immunologic
and epidemiologie aspects of varicella infection acquired during
infancyand early childhoOO. J Pediatr 100:881-885, 1982
19. Gershon AA, Steinberg SP, Gelb L, et al: Clinical re-
infection with varicella-zoster virus. J lnfect Dis 149:137-142,
1984
20. Feldrnan SF, Hughes WI, Daniel CB: Varicella in chil-
dren with cancer: Seventy-seven cases. Pediatrics 56:388-397,
1975
21. Schimpff S, Serpick A, Stoler B, et al: Varicella·zoster
infection in patients with cancer. Ann Intern Med 76:241-254,
1972
22. Morens DM, Bregman DJ, West M, et al: An outbreak
of varicella-zoster virus infection among cancer patients. Ann
Intern Med 93:414-419, 1980
23. Rusthoven n, Ahlgren P, Elhakim T, et al: Varicella-
ZOSter infection in adult cancer patients: A population study.
Arch Intern Med 148:1561-1566, 1988
24. Locksley RM, Flournoy N, Sullivan KM, et al: Infection
With varicella-zoster virus after marrow transplantation. J Infect
Dis 152:1172-1181, 1985
25. Schuchter LM, Wingard JR, Piantadosi S, et al: Herpes
115
zoster infection after autologus bone marrow transplantation.
BloOO 74:1424-1427,1989
26. Naraqi S, Jackson 00, Jonasson 0, et al: Prospective
study of prevalence, incidence, and source of herpes virus in-
fections in patients with renal allografts. J Infect Dis 136:531-
540, 1977
27. Patterson LER, Buder KM, Edwards MS, et al: Clinical
herpes zoster shortly following primary varicella in two HIV-
infected children. Clin Pediatr 28:354, 1989
28. Guess HA, Broughton DD, Melton III U, et al: Epide-
miology of herpes zoster in children and adolescents: A popu-
lation-based study. Paediatrics 75:512-517, 1985
29. Ragozzino MW, Melton III U, Kurland LT, et al: Pop-
ulation-based study of herpes zoster and its sequelae. Medicine
61:310-316, 1982
30. Hope-Simpson RE: The nature ofherpes zoster: A long-
term study and a new hypothesis. Proc R Soc Med 58:9-20,
1965
31. McGregor RM: Herpes zoster, chickenpox, and cancer
in generał practice. Br Med J 1:84-87, 1957
32. Weller TH: Varicella and herpes zoster: Changing con-
cepts of the natural history, control, and importance of a not-
so-benign virus (fiest of two parts). N Engl l Med 309:1362-
1368, 1983
33. Feldman S, Hughes WT, Kim HY: Herpes zoster in
children with cancer. Am l Dis Child 126:178-184, 1973
34. Kalman CM, Laskin OL: Herpes zoster and zosteriform
herpes simplex virus infections in immunocompetent adults.
Am J Med 81:775-778, 1986
35. Brunell PA: Fetal and neonatal varicella-zoster infec-
tions. Semin Perinatol 7:47-56, 1983
36. Mazur MH, Dolin R: Herpes zoster at the NIH: A twenty
year experience. Am l Med 65:738-744, 1978
37. Ruckdeschel JC, Schimpff SC, Smyth AC, et al: Herpes
zoster in impaired cell-associated immunity to the varicella-
zoster virus in patients with Hodgkin's disease. Am l Med
62:77-85, 1977
38. Guinee VF, Gnido H, Pfalgraf KA, et al: The incidence
ofherpes zoster in patients with Hodgkin's disease. An analysis
of prognostic factors. Cancer 56:642-648, 1985
39. Feld R, Evans WK, DeBoer G: Herpes zoster in patients
with carcinoma of the lung. Am l Med 73:795-80 I, 1982
40. Rusthoven H, Ahlgren P, Elhakim T, et al: Risk factors
for varicella zoster disseminated infection among adult cancer
patients with localized zoster. Cancer 62:1641-1646, 1988
41. Merselis JG, Kaye D, Hook EW: Disseminated herpes
zoster: A report of seventeen cases. Arch Intern Med 113:679-
686, 1964
42. de Moragas JM, Kierland RR: The outcome of patients
with herpes zoster. Arch Dermatol 75:193-196, 1957
43. Fulginiti VA, John TJ, Sieber OF: Herpes zoster, in
Demis D, Dobson R, McGuire J (eds): Clinical Derrnatology,
vol3 (section 14-7). New York, NY, Harper and Row, 1977,
pl
44. Sibrack LA: Cutaneous signs of internal malignant dis-
ease. Primary Care 5:263-280, 1978
45. Ragozzino MW, Melton U III, Kurland L, et al: Risk of
cancer after herpes zoster. A population-based study. N Engl J
Med 307:393-397, 1982
46. Perren Tl, Powles RL, Easton D, et al: Prevention of
herpes zoster in patients by long-term oral acylovir after allo-
116
geneic bone marrow transplantation. Am J Med 85:99-101,
1988 (suppl 2A)
47. Lundgren G, Wilczek H, Uinnqvist B, et al: Acyclovir
prophylaxis in bone marrow transplant recipients. Scan J Infect
Dis 47:137-144, 1985 (suppl)
48. Jacobson MA, Berger TG, Fikrig S, et al: Acyclovir-
resistant varicella-zoster infection after chronic oral acyclovir
therapy in patients with acquired irnrnune deficiency (AIDS).
Ann Intern Med 112:187-191, 1990
49. Rand KH, Rasmussen LE, Pollard RB et al: Cellular
irnrnunity and herpes virus infections in cardiac-transplant pa-
tients. N Engl J Med 296:1372-2377, 1976
50. Preiksaitis JK, Rosno S, Grumet C, et al: Infections due
to herpes viruses in cardiac transplant recipients: Role of the
donor heart and irnrnunosuppressive therapy. J Infect Dis 147:
974-981, 1983
51. Eyster ME, Rabkin CS, Hilgartner MW, et al: Human
irnrnunodeficiency virus-related conditions in children and
adults with hemophilia: Rates, relationships to CD for counts
and predictive value. Blood 81:828-834, 1993
52. Melbye M, Goedert JJ, Grossman RJ, et al: Risks of
AIDS after herpes zoster. Lancet 28:728-731, 1987
53. Colebunders R, Mann JM, Francis H, et al: Herpes
zoster in African patients: A clinical predictor of human irnrnu-
nodeficiency virus infection. J Infect Dis 157:314-318, 1988
54. Friedman-Kien AE, Lafleur FL, Gendler E, et al:
Herpes zoster: A possible early clinical sign for development of
acquired irnrnunodeficiency syndrome in high-risk individuals.
JAm Acad Dermatol 14:1023-1028, 1986
55. Gilson IH, Barnett JH, Conant MA, et al: Disseminated
ecthymatous herpes varicella-zoster virus infection in patients
with acquired immunodeficiency syndrome. J Am Acad Der-
matoI20:637-642, 1989
JAMES J. RUSTHOVEN
56. Gilden DH, Murray RS, Wellish M, et al: Chronic pro-
gressive varicella-zoster virus encephalitis in an AIDS patient.
Neurology 38:1150-1153, 1988
57. Burnell PA, Ross A, Miller LH, et al: Prevention of
varicella by zoster irnrnune globulin. N Engl J Med 280:1191-
1194, 1969
58. Burnell PA, Gershon AA, Hughes WT, et al: Prevention
of varicella in high-risk children: A collaborative study. Pedi-
atrics 50:718-722, 1972
59. Stevens DA, Merigan TC: Zoster irnrnune globulin pro-
phylaxis of disseminated zoster in compromise hosts. A ran-
domized tria!. Arch Intern Med 140:52-54, 1980
60. Zaia JA, Levin MJ, Preblud SR, et al: Evaluation of
varicella-zoster immune globulin: Protection of irnmunosup-
pressed children after household exposure to varicella. J Infect
Dis 147;737-743, 1983
61. Paryani SG, Arvin AM, Koropchak CM, et al: Compar-
ison of varicella zoster antibody titers in patients given intra-
venous irnmune serum globulin or varicella zoster irnrnune
globulin. J Pediatr 105:200-205, 1984
62. Weibel RE, Neff BJ, Kuter BJ, et al: Live attenuated
varicella virus vaccine. Efficacy trial in healthy children. N
Engl J Med 310:1409-1415, 1984
63. Gershon AA, LaRussa P, Hardy I, et al: Varicella vac-
cine: The American experience. J infect Dis 166:S63-S68, 1992
(suppl I)
64. Hardy I, Gershon AA, Steinherg SP, et al: The inei-
dence of zoster after irnrnunization with live attenuated vari-
cella vaccine. A study in children with leukemia. N Engl J Med
325:1545-1550, 1991
65. MMWR 40(RR-12):69 + 88, 1991