Objective. Varicella-zoster virus (VZV) can cause severe disease in premature neonates. The fetus receives protective maternal VZV-immunoglobulin G (IgG) mainly in the third trimester of pregnancy. Therefore, premature neonates are considered at risk for VZV infection. Administration of varicella-zoster immunoglobulin (VZIG) within 96 hours after exposure effectively prevents severe illness in susceptible patients. The objectives of this study were to define the major determinants of the neonatal VZV-IgG titer and to determine the half-life of transplacentally acquired VZV-IgG. Guidelines provided by the Centers for Disease Control and Prevention for the use of VZIG in (premature) neonates were evaluated.
Methods. VZV-IgG titers were measured in sera of 221 neonates and 43 mothers using a quantitative enzyme-linked immunosorbent assay. In 27 neonates, VZV-IgG titers were followed for up to 14 weeks.
Results. In a linear regression model, the maternal antibody titer was the major determinant of the neonatal titer (β = 0.89); gestational age was only of minor importance (β = 0.18). The median half-life of VZV-IgG in neonates was 25.5 days (range: 14.6–76.0 days). In the first weeks of life, major fluctuations of the VZV-IgG titer occurred in >50% of the neonates. The predictive value of Centers for Disease Control and Prevention guidelines for identification of neonates who should receive VZIG in case of exposure to VZV was poor: positive and negative predictive values were 0.80 and 0.43, respectively.
Conclusions. The neonatal VZV-IgG titer is predominantly predicted by the maternal VZV-IgG titer, whereas birth weight and gestational age are much less predictive than previously reported.
Varicella (chickenpox) is a highly contagious disease caused by varicella-zoster virus (VZV), a ubiquitous human α-herpesvirus.1,2 VZV is transmitted by direct contact, droplet, or airborne spread of vesicle fluid or secretions of the respiratory tract of patients with chickenpox or the vesicle fluid of patients with herpes zoster.3–6 The attack rate among susceptible household contacts is approximately 90%,1,2 but the risk of transmission in the hospital is probably lower.7 This usually self-limiting disease can cause serious morbidity and even mortality in premature and critically ill neonates.8–11 During pregnancy, protective maternal antibodies are transferred to the fetus. It is generally accepted that this transfer occurs mainly in the third trimester of pregnancy and therefore that premature neonates may not have acquired enough maternal antibodies to be protected.12–16
Infected individuals start shedding the virus via the respiratory tract up to 5 days before the onset of symptoms.6 In addition, VZV-immunoglobulin G (IgG)-positive individuals may play a role in spreading the virus.17 Therefore, unintentional and inadvertent exposure of neonates cannot be completely prevented. For preventing such exposure, contact between neonates and siblings is avoided as much as possible, and nonimmune hospital staff are vaccinated.
Introduction of VZV on pediatric and neonatal wards occurs regularly.3,9,11,18–24 Administration of varicella-zoster immunoglobulin (VZIG) within 96 hours after exposure has proved to be effective in preventing severe illness in seronegative patients.25 In our hospital, all neonates in the neonatal intensive care unit (NICU) are tested for VZV-IgG within 96 hours after exposure, regardless of maternal history, because the reliability of the maternal history for VZV is still uncertain.26–28 If neonates test negative, then VZIG is administered.
The Centers for Disease Control and Prevention (CDC) recommends administering VZIG to exposed premature neonates with a gestational age <28 weeks and/or a birth weight of ≤1000 g, regardless of maternal history.16 Premature neonates of ≥28 weeks of gestational age who are born to immune mothers are considered to have acquired enough maternal antibodies to be protected from severe disease and complications.
To identify determinants of neonatal VZV-IgG levels, we measured VZV-IgG titers in 221 neonates and in 43 mothers. The utility of CDC guidelines was evaluated. In addition, we determined the half-life of VZV-IgG in 30 neonates by repeated measurements during the first 3 months after birth.
All neonates who were admitted to the NICU from April 1995 until February 2000 and were tested for VZV-IgG because of possible exposure to the virus were identified retrospectively by review of laboratory records (n = 157). To study the half-life of transplacentally acquired VZV-IgG in the neonate and the relation between the neonatal and maternal VZV-IgG titer, we studied 64 neonates and 43 mothers prospectively. After informed consent was obtained from the mother, serum samples from mother and neonate were obtained within the first week after delivery. If the serum of the child tested positive for VZV-IgG, then permission was asked to obtain follow-up samples. Samples were obtained as close as possible to the following schedule: every week during the first month of life, every second week during the second month, and once a month during the following 4 months. The research protocol was approved by the local ethical committee.
The following information was obtained from the medical records: gender, gestational age, birth weight, postnatal age (days) at which the serum sample was taken, multiple birth, and volume and type of transfusions administered before the serum sample was taken. Gestational age was calculated by menstrual history or, in case of irregular cycle, by ultrasound data obtained during the first trimester of pregnancy.
Serum samples (100 μL) were tested for VZV-IgG quantitatively with an automated enzyme-linked fluorescent immunoassay (Vitek Immuno Diagnostic Assay System [VIDAS], Biomerieux, France). In international literature, the sensitivity of VZV-IgG enzyme-linked immunosorbent assays has been reported to vary between 86% and 98%.16,29 Results of VZV-IgG enzyme-linked immunosorbent assays and the traditionally used fluorescent antibody to membrane assay are comparable.30 According to the manufacturer of the VIDAS assay, sensitivity and specificity of the test are 99.7% and 97.6%, respectively. Results are presented as the ratio (X) of the fluorescence value of the sample over that of a standard. Results are to be interpreted as follows: X < 0.60: negative; 0.60 < X < 0.90: equivocal; X ≥ 0.90: positive. Cutoff values are stored in the VIDAS testing machine and are deduced from test values of the standard. Because of the use of the standard as an internal control, the results of various runs are comparable. We considered equivocal results as negative.
Calibration of the VZV-IgG Assay
The VIDAS VZV-IgG assay was calibrated with the International Standard from the World Health Organization containing 50 International Units (IU) of VZV-IgG. The standard was dissolved in 1 mL of sterile water. A 2-fold dilution series was tested twice in 2 runs with 2 different kits with different lot numbers (total = 4 series).
Estimation of the Half-Life of VZV-IgG in Neonates
The half-life of VZV-IgG was estimated using the titers from neonates with positive VZV-IgG titers at birth. The half-life was calculated using the VZV-IgG titer of the first and the last serum sample from each child.
Study on Mother-Child Serum Pairs: Linear Regression Model of Neonatal Anti-VZV-IgG Titer
Data obtained from 51 neonates, for which a maternal titer was available, were included in a linear regression model. Among those 51 neonates were 8 twins; therefore, 43 mothers contributed to 51 serum pairs. The following variables were entered in the model: maternal VZV-IgG titer, gender, gestational age, age on day of serum sample, transfusions of packed cells, filtered erythrocytes, thrombocytes or fresh-frozen plasma, multiple birth, and the ethnicity of the mother (white or not). Birth weight was not entered because this variable was multicollinear with gestational age (r = 0.90). Two multivariate linear regression models are presented: a model in which all variables are entered and a stepwise model in which nonsignificant variables are withdrawn. Because the partial regression coefficients have different dimensions, we use the standardized coefficient β. This is a regression coefficient without dimension, which is calculated as follows: β = regression coefficient (X) × (standard deviation of X [SX]/standard deviation of outcome variable [neonatal titer]). Therefore, β depicts the relative influence of each separate variable.
Theoretical Influence of Transfusions of Blood Products Containing Various Amounts of VZV-IgG Antibodies and of Blood Withdrawal
Because blood products contain various amounts of plasma, neonatal VZV-IgG titers could be influenced by transfusions. The intensive monitoring of the clinical condition of the neonate requires frequent withdrawal of blood samples. Therefore, we calculated the theoretical effect of transfusions and blood withdrawal on VZV-IgG titers in 2 fictitious neonates of gestational age and birth weight of 26 weeks and 877 g, and 38 weeks and 3409 g, respectively. Their extracellular volume (ECV), in which VZV-IgG is distributed, was calculated as 500 mL and 1500 mL.31
Predictive Value of CDC Guidelines and Predictive Value Based on Maternal Titer
The predictive value of the CDC criteria was calculated for the total study population (N = 221).16 The predictive value of the maternal titer was calculated for the subset of 51 neonates for which a maternal titer was available. Positive predictive value (PPV) was defined as the percentage of neonates with a positive VZV-IgG titer, predicted correctly by CDC criteria and the maternal titer, respectively. Negative predictive value (NPV) was defined as the percentage of neonates with a negative VZV-IgG titer (at risk), predicted correctly.
Analysis was performed with the SPSS Statistical Package version 9.0 (SPSS Inc, Chicago, IL), Excel 4.0 (Microsoft, Redmond, WA), and Lotus-123 version 2.01 (IBM Software Group, Cambridge, MA). P < .05 was considered significant.
We included 221 neonates in the study. General features of the study population are presented in Table 1. Twenty-five percent of neonates had a negative VZV-IgG titer. In the prospectively recruited cohort of 64 neonates, a maternal serum sample was not available in 13 cases.
Calibration of the VIDAS
Serial measurements of VZV standard dilution samples showed an excellent reproducibility of the test (fitted calibration curve: R2 = 0.993).
Estimation of the Half-Life of Maternally Acquired VZV-IgG Antibodies
Results are shown in Fig 1. For all neonates taken together (n = 27), the median half-life of VZV-IgG was 25.5 days (range: 14.6–76.0 days). In 15 neonates with continuous and regular decreasing VZV-IgG titers, the median half-life was 21.4 days. In 12 neonates, at some time points, the titer was higher than at previous time points. The increases occurred mainly during the first weeks of life. During the follow-up period, transfusions of blood products were administered to neonates with regularly as well as irregularly decreasing titers. In 3 cases, the last VZV-IgG titer (on days 5, 16, and 16 of life, respectively) was higher than the first; these 3 cases were excluded from the half-life calculations.
Linear Regression Model of Factors That Influence Neonatal VZV-IgG Titer
Results are shown in Figs 2 and 3. In a crude analysis of correlation (Fig 2), gestational age was a poor predictor of neonatal VZV-IgG antibody titer (R2 = 0.07). This was also the case for birth weight (R2 = 0.06). In contrast, maternal VZV-IgG titer (Fig 3) correlated well with neonatal VZV-IgG titer (R2 = 0.72). This was also the case for neonates who were at risk according to the CDC guidelines. The linear regression model, in which all variables were entered, explained 85% of variance of neonatal VZV-IgG titer, and the maternal titer was the major predictor of the neonatal VZV-IgG titer (β= 0.90; P < .01). Less important predictors were gestational age (β = 0.17; P = .02) and the ethnic background of the mother (β = 0.12; P = .05). In the stepwise linear regression model, explaining 84% of variance in neonatal VZV-IgG titer, gestational age (β = 0.18; P < .01), day of sampling serum (β= −0.17; P < .01), and the ethnic background of the mother (β = 0.15; P = .01) were included, but the maternal titer remained the single most important predictor of the neonatal titer (β = 0.89; P < .01).
Three neonates had a negative VZV-IgG titer, despite a positive titer of the mother (Table 2, Fig 3). The first was a male neonate of gestational age 31 weeks and birth weight 1855 g; the other 2 were male twins with a twin-to-twin transfusion syndrome, of gestational age 28 weeks and a birth weight of 766 and 1040 g, respectively. We could not identify any particular features that could explain the discrepancy between maternal and neonatal VZV-IgG titer.
Theoretical Influence of Transfusions and Blood Sampling
In the simulation model, withdrawal of blood samples varying between 0 and 100 mL and administration of various volumes (0–50 mL) of transfusions with different VZV-IgG titers (0.5, 0.16, 0.12, and 0.05 IU/mL, respectively) were evaluated. In general, withdrawal of blood and blood transfusions had little effect on neonatal VZV-IgG titers. Only in the simulation of a very premature neonate (gestational age of 26 weeks) with a VZV-IgG titer approximately at the cutoff point did transfusions with a volume of >20 mL of blood products containing a high titer of VZV-IgG (0.5 IU/mL) or withdrawal of >50 mL of blood affect the serostatus, changing it from negative to positive and positive to negative, respectively.
Predictive Value of CDC Criteria and Predictive Value Based on the Maternal Titer
The predictive value of the CDC guidelines is shown in Table 3. Because 57% of the neonates who were at risk according to CDC criteria tested positive for VZV-IgG, the NPV of these criteria was very low (0.43). The PPV was higher (0.80). The PPV and NPV of the maternal titer were significantly better: 0.93 and 1.0, respectively (Table 2).
In the Netherlands, 98% of young adults are positive for VZV-IgG.32 Vaccination against VZV is not routinely performed. Because of this high seroprevalence, the majority of neonates acquire anti-VZV-IgG antibodies during gestation. In our study population, the maternal titer was the single most important predictive factor of neonatal VZV-IgG titers. The generally accepted hypothesis that very premature neonates have not acquired sufficient maternal antibodies to be protected from VZV could not be confirmed: 57% of premature neonates who were at risk according to the CDC guidelines had sufficient levels of VZV-IgG. The subanalysis of mother-child serum pairs showed that 2 very premature neonates (at risk according to CDC) had the highest VZV-IgG titer. In our opinion, CDC recommendations as to which neonates should receive VZIG in case of exposure to VZV are not universally applicable. From our data, it seems impossible to predict the serostatus of a premature neonate on the basis of gestational age or birth weight alone. Possibly, the utility of the CDC definitions of which neonates are at risk for VZV should be defined for different populations, according to the prevalence of VZV-IgG in women of childbearing age. Our finding that almost 25% of neonates tested negative for VZV-IgG is most likely attributable to the fact that 52 of 221 neonates were tested at a postnatal age of >1 week and that there was overrepresentation in our study population of neonates from immigrant mothers. Seroprevalence of VZV-IgG positivity in (sub)tropical regions is known to be much lower than in regions with more temperate climates.33,34 In fact, both factors had a small but significant predictive value in the linear regression model.
Mendez et al35 found that 10 of 12 neonates with a gestational age <30 weeks had a positive VZV-IgG titer. In another study, 23 premature neonates (gestational age: 22–31 weeks), born from VZV-IgG-positive mothers, were tested for VZV-IgG with 2 serologic assays; 17 of the 23 tested positive with at least 1 of the 2 VZV-IgG assays.23 Also, for IgG antibodies directed against other viruses, it was shown that the maternal titer is the most important predictor of the neonatal titer, although there is a small but significant difference between mature and premature neonates.36,37
In healthy adults, the half-life of IgG antibodies is 21 days.38 The median half-life in our premature neonates was 5 days longer, although the subgroup with regular decreasing titers had a mean half-life almost equal to that of adults. In the first weeks of life, when the neonates are intensively treated, major fluctuations of the VZV-IgG titer occurred in 44% of the neonates. We could not explain these fluctuations by administration of blood products, because blood products were also given to neonates who showed a steadily decreasing VZV-IgG titer.
In (premature) neonates, there are many factors (with opposite effects) that, in theory, can affect the neonatal VZV-IgG titer. Some of those factors cannot be measured and therefore could not be entered in the linear regression model. The neonatal VZV-IgG titer at birth is determined by the amount of maternal antibodies received during gestation as well as by the ECV in which these maternal antibodies are distributed. The ECV of neonates varies from 59% of total body weight at 24 weeks of gestation to approximately 44% at term and is approximately 30% 6 months after birth.31 However, because of growth, the absolute ECV increases, which results in a steady increase of distribution volume. Second, at birth, a physiologic hemoconcentration occurs but resolves in a few days.39 Third, compared with the circulating blood volume, relatively large amounts of blood may be withdrawn for investigation, especially in the first weeks of life. In our NICU, in critically ill neonates, sometimes >50% of blood is withdrawn in the first weeks of life; this is approximately 12% of ECV.25 Finally, administration of multiple transfusions of blood products is common.40 Through transfusions, seronegative individuals can become temporarily seropositive for VZV-IgG.41 Because of the high seroprevalence of VZV-IgG in the Netherlands, it is reasonable to assume that most transfusions will contain a certain amount of VZV-IgG. We showed in a theoretical model that only large volumes of transfused blood products or of blood withdrawals influence the VZV-IgG titer and only in very low birth weight neonates.
In our hospital, measurement of the VZV-IgG titer can be accomplished within a few hours. Therefore, it is always possible to identify neonates who are at risk for VZV and should receive VZIG within the 96-hour period after exposure to VZV. VZIG cannot prevent the development of varicella, but it prevents serious life-threatening disease. Neonates lack optimal cellular responses to VZV. Therefore, it is possible that very premature neonates may benefit from boosting VZV antibody titers, even if they have measurable VZV-IgG titers at birth. So far, however, the effect of such boosting has not been studied. Therefore, until additional research is conducted, it may still be justifiable to provide passive antibodies to the very low birth weight infant.
We showed a large variability in neonatal VZV-IgG titers, which is predominantly determined by the maternal VZV-IgG titer. More than 50% of our premature neonates of <28 weeks’ gestation and/or birth weight of <1000 g showed a measurable VZV-IgG titer. With a few exceptions, the maternal titer of VZV-IgG predicted the neonatal titer very well. The only means of determining whether a neonate is at risk for VZV infection with certainty is by actually measuring his or her VZV-IgG titer.
Dr van der Zwet is supported by an AGIKO grant of the Netherlands Organization for Scientific Research.
We are indebted to Dr Erik F. Ree, a pediatrician, for assistance in initiating the study, as well as for entering many mothers/neonates into the study by obtaining informed consent from the mothers.
- ↵Whitley RJ. Varicella-zoster virus. In: Mandell GL, Bennett JE, Dolin RJ, eds. Principles and Practice of Infectious Diseases. 5th ed. Philadelphia, PA: Churchill Livingstone;2000:1580– 1586
- ↵Arvin AM. Varicella-zoster virus. Clin Microbiol Rev.1996;9 :361– 381
- Gustafson TL, Lavely GB, Brawner ER, Hutcheson RH, Wright PF, Schaffner W. An outbreak of airborne nosocomial varicella. Pediatrics.1980;70 :550– 556
- Sawyer MH, Chamberlin CJ, Wu YN, Aintablian N, Wallace MR. Detection of varicella-zoster virus DNA in air samples from hospital rooms. J Infect Dis.1994;169 :91– 94
- ↵Benenson AS, ed. Control of Communicable Diseases Manual. 16th ed. Washington, DC: American Public Health Association;1995 :88– 89
- Rubin L, Leggiadro R, Elie MT, Lipsitz P. Disseminated varicella in a neonate: implications for immunoprophylaxis of neonates postnatally exposed to varicella. Pediatr Infect Dis J.1986;5 :100– 102
- Conway SP, Dear PR, Smith I. Immunoglobulin profile of the preterm baby. Arch Dis Child.1985;60 :208– 221
- ↵Prevention of varicella: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Morb Mortal Wkly Rep.1996;45(RR-11) :5, 22– 23
- ↵Connely BL, Stanberry LR, Bernstein DI. Detection of varicella-zoster virus DNA in nasopharyngeal secretions of immune household contacts of varicella. J Infect Dis.1993;168 :1253– 1255
- ↵Gershon AA, Raker R, Steinberg S, Topf-Olstein B, Drusin LM. Antibody to varicella-zoster virus in parturient women and their offspring during the first year of life. Pediatrics..1976;58 :692– 696
- Myers G, Rasley DA, Hierholzer WJ. Hospital infection control for varicella zoster virus infection. Pediatrics.1982;70 :199– 202
- ↵Varicella zoster immune globulin for the prevention of chickenpox. MMWR Morb Mortal Wkly Rep. 1984;33:84–90, 95–29
- ↵Vandersmissen G, Moens G, Vrankcx R, De Schrijver A, Jacques P. Occupational risk of infection by varicella-zoster virus in Belgian healthcare workers: a seroprevalence study. Occup Environ Med.2000;57 :621– 626
- ↵Shanley J, Myers M, Edmond B, Steele R. Enzyme-linked immunosorbent assay for detection of antibody to varicella-zoster virus. J Clin Microbiol.1982;15 :208– 211
- ↵Bell EF, Oh W. Fluid and electrolyte management. In: Avery GB, Fletcher MA, MacDonald MG, eds. Neonatology: Pathophysiology and Management in the Newborn. 5th ed. Philadelphia, PA: JB Lippincott;1999:345– 361
- ↵Weers-Pothoff G, De Boo TM, Willemse MJ. Prevalence of Antibodies to Human Herpesviruses in a Part of the Dutch Population [PhD thesis]. Nijmegen, the Netherlands: University Medical Center Nijmegen; 1991
- ↵Mandal BK, Mukherjee PP, Murphy C, Mukherjee R, Naik T. Adult susceptibility to Varicella in the tropics is a rural phenomenon due to the lack of previous exposure. J Infect Dis.1998;178(suppl) :S52– S54
- Copyright © 2002 by the American Academy of Pediatrics