Specific Cell-Mediated Immunity in Children with Congenital and Neonatal Cytomegalovirus Infection and Their Mothers

Transcription

1 THE JOURNAL OF NFECTOUS DSEASES. VOL. 140, NO.4. OCTOBER by The University of Chicago /79/ $00.75 Specific Cell-Mediated mmunity in with Congenital and Neonatal Cytomegalovirus nfection and Their David W. Reynolds, Paula H. Dean, Robert F. Pass, and Charles A. Alford From the Department of Pediatrics, School of Medicine, University of Alabama in Birmingham, Birmingham, Alabama Cell-mediated immunity (CM) to cytomegalovirus (CMV) and herpes simplex virus type 1 (HSV-1) was examined in 35 mothers and 30 of their offspring with congenital or neonatal CMV infection by means of the lymphocyte transformation assay. Eleven offspring did not respond to CMV antigen, and 15 of the 19 positive children displayed lower responses than those of normal immune adults. Productive infection in the younger children at the time of assay and the presence of disease correlated strongly with the absence of responses. The mothers as a group also demonstrated impaired CM to CMV while reacting normally to HSV-l antigen. Neither time after transmission of infection nor viral excretion was significantly associated with impaired CM, although a trend toward diminished responses was evident among women who were shedding virus. Mitogen stimulation was normal in all test subjects. These findings imply that deficient specific CM may play a role in the genesis of persistent CMV infection and fetal pathology. Fetal and neonatal cytomegalovirus (CMV) infections are characterized by long-term persistence of viral excretion despite the continued production of specific antibody [l, 2]. Humoral immunity is also incapable of controlling recurrent and occasionally persistent infections in the mothers of these children [3]. ndeed, transmission of CMV from mother to fetus [4] and neonate [5] has been shown to occur in the face of substantial levels of maternal serum neutralizing antibody. A similar situation pertains to immunosuppressed organ transplant patients. Among individuals who are seropositive for CMV prior to immunosuppression, endogenous infection is frequently reactivated [6, 7]. These observations plus data derived from experiments in animals [8, 9] clearly suggest that impaired cell-mediated immunity (CM) may Received for publication December 27, 1978, and in revised form March 13, This study was supported by grants no. 5-PO-HD10699 and 5-TO-HD00413 from the National nstitute of Child Health and Human Development; no. T32-A07041 from the National nstitute of Allergy and nfectious Diseases; no. 5-R01 CA6424 and 3-P30-CA13148 from the National Cancer nstitute; and no. 5-MO-RR32 from the National nstitutes of Health Clinical Research Center; and by the Robert E. Meyer Foundation. Please address requests for reprints to Dr. David Reynolds, University of Alabama Medical Center, Department of Pediatrics, Room 609 COLD Building, University Station, Birmingham, Alabama be important in the genesis of persistent or recurrent CMV infections. t is therefore important to assess the adequacy of CM to CMV in infected mothers and their offspring in order to ascertain whether a virus-induced or innate deficiency of cellular immunity may be associated with persistent productive infection, perinatal transmission of virus, or disease in the offspring. This report presents data on specific CM of children with three types of perinatal infection and their mothers in relation to age and disease state of the children and presence or absence of productive infection. Materials and Methods Subjects. The study population comprised 35 mothers and 30 of their children with either fetal or neonatal CMV infection. Participants were identified prospectively by methods previously described [4, 5] or by referral. All subjects were assessed semiannually for the serologic, virologic, and clinical expression of their CMV infection. Nine immune and seven nonimmune adult laboratory workers (seven male and nine female) between 20 and 48 years of age served as controls. Each time the lymphocyte transformation assay was performed, lymphocytes from one of the immune and one of the nonimmune controls were included. 493

2 494 Reynolds et al. Collection and processing ofperipheral blood mononuclear cells. Plasma for antibody testing and for use in the transformation assay was obtained by centrifugation of a heparinized venous blood sample (5-15 ml) at 200 g for 10 min; mononuclear cells were then separated from the remaining blood by centrifugation on a Ficoll-Hypaque gradient according to the technique of Boyum [10]. nterface cells were washed three times with cold medium 199 (Flow Laboratories, McLean, Va.) and adjusted to a final concentration of 5 X los cells/ml in RPM-1640 medium (Grand sland Biological Co., Grand sland, N.Y.) containing 100 -tg of gentamicin/ml and heat-inactivated, pooled, homologous CMV-immune sera. Preliminary testing had revealed that the cells of most control and test subjects responded better in this preparation than in either heat-inactivated homologous nonimmune sera or autologous plasma. Stimulants for the lymphocyte transformation assay. Two strains of CMV, herpes simplex virus type 1 (HSV-1), and phytohemagglutinin (PHA) were used as stimulants. The CMV strains employed were AD-169 (the standard laboratory virus) and W. C. (a cervical strain isolated and adapted in this laboratory). The strain of herpes simplex virus was isolated locally and typed by Dr. Andre Nahmias, Grady Memorial Hospital, Atlanta, Ga. Crude viral antigens were prepared by infection of human foreskin fibroblast monolayers at a multiplicity of input of 1.0 and by harvest of the infected cells by trypsinization 120 hr and 48 hr later for CMV and HSV-1, respectively. ntracellular viral antigens were released from the cells by resuspension of the cells in medium 199 and sonification at 30 W for 60 sec (Sonifer; Ultrasonics, Plainview, N.Y.). Following low-speed clarification, aliquots of the antigen were stored at - 70 C. Subsequent titration revealed titers of 1 X 10 6 and 2 x 10 6 pfu/ml for the CMV and HSV-l antigens, respectively. Control antigen was prepared in a similar manner by use of uninfected fibroblasts. Prior to use, antigens were heat-inactivated for 30 min and diluted 1:5 in medium 199-a dilution providing maximal stimulation for both strongly and weakly responding donors. For use as a mitogen, PHA-M (Sigma Chemical Co., St. Louis, Mo.) was reconstituted to a concentration of 1 mg/ml in distilled water, and aliquots of the suspension were stored at - 70 C. Lymphocyte transformation assay. For each culture, 1 X los mononuclear cells in 0.2 ml of RPM-1640 medium were dispensed into each of three wells of a 96-well tissue culture plate (Falcon, Oxnard, Calif.). Triplicate wells received 7 -tl ofeither antigen or mitogen. The cultures were incubated at 37 C in 5% CO 2 with humidity. On the sixth day of incubation, lci of [methyl- 14 C]thymidine in 7 -li of RPM-1640 medium was added to each well. Cells were harvested 24 hr later onto glass fiber filter paper strips with a semiautomatic precipitator (Otto Hiller, Madison, Wis.), and the radioactivity was measured in a Beckman model L52150T liquid scintillation spectrometer (Beckman, rvine, Calif.). The results of the transformation assay were expressed as stimulation indices (S), which were calculated by dividing the mean cpm of triplicate cultures containing viral antigen or mitogen by the mean counts of triplicate cultures containing uninfected fibroblast antigen. Serologic and virologic assays. Antibodies were measured by the microneutralization plaque reduction test as previously described for CMV [5] with use of steel pools of CMV strain AD-169 and HSV-l. Residual infectivity of HSV-l was assayed in BSC-l cells, a continuous cell line derived from African green monkey kidney tissue. Throat swabs, vaginal swabs, and urine specimens were processed for viral isolation as previously described [2]. Longitudinal evaluations allowed for an assessment of the pattern of excretion in each study subject. Statistical analyses. Student's t-test, Student's paired t-test, the X 2 test, Wilcoxon's two-sample ranking test, and linear regression analysis were used in the statistical evaluation of the data. Results Clinical and virologic characteristics of the study population. The study population was divided into three groups on the basis of time of acquisition of CMV by the children and whether or not the infection was symptomatic (table 1). Group 1 consisted of 19 mother-child pairs; these children were excreting CMV during their first week of life and were therefore identified as congenitally infected. All 19 of the children in group 1 were asymptomatic both at birth and at the time they were studied. Group 2 comprised seven

3 Cellular mmunity and CMV 495 Table 1. Mean age and prevalence of productive infection with cytomegalovirus at the time of study in mothers and their offspring grouped according to time of perinatal transmission and presence of symptoms in the children. No. of symptomatic Type of children/no. No. shedding No. shedding Group transmission examined" Age virus/total Age virus/total Congenital 0/ / /19 (20-34) (1/12-tO) 2 Congenital 6/6 23 3/ /6 (19-26) (1/12-7) 3 Natal 0/5 23 1/9 4 3/5 (18-27) (1-6) * mmediate or late-onset disease. t Mean age in years (range). mothers with a child with cytomegalic inclusion disease (CD); six of the symptomatic children were examined. Each child had some degree of mental or perceptual impairment at the time of testing. The nine mothers in group 3 were shedding CMV from the cervix late in pregnancy. None of their offspring was excreting virus at birth, but five began to do so during their second month of life and were thus considered to have natal infection. There was no statistically significant difference in either mean age or prevalence of viral shedding at the time of study for mothers or children of any of the three groups. The majority of children in each group were excreting CMV at the time of study. Lymphocyte transformation responses. Test characteristics. Background counts from control antigen-stimulated cultures ranged from a mean (± SE) of 197 ± 33 in the adult immune control group to 78 ± 11 in the children with asymptomatic (silent) congenital infection. As shown in figure 1, seven nonimmune adults had a mean S of 1.0 ± 0.5 with CMV antigen, and in only two of 28 assays did a seronegative individual achieve an index between 2.0 and 3.0; therefore, responses of ~2.0 were judged to be negative. n contrast, the lymphocytes from the normal immune adults proliferated in response to CMV in each of 28 assays; their group mean S was 42 ± 5. Forty-nine of 65 test subjects and each of the immune controls responded to a similar degree to the laboratory and wild CMV strains. Of the remaining 16 patients, four children and one mother reacted at low levels (mean S of 4 ± 0.07) to AD-169 only. One child's lymphocytes transformed only in response to the wild isolate (S of 7). Eight mothers and two children responded to both strains but had a higher S with one or the other antigen. Specifically, four mothers and one child reacted with a twofold or greater (group mean, 4.7) increase in S to AD-169; an identical number of mothers and children responded with a similar increment (group mean, 3.6) to the wild isolate. The results to follow are expressed as the higher S obtained with the two antigens.. Lymphocytes from five mothers, three in group 1 and one from each of the other two groups, failed to respond to CMV antigen. Moreover, the mean S of each of the maternal groups was significantly lower than that of the positive controls (P<0.05). The mothers of group 1 and group 3 had almost identical S to CMV-20 ± 6 and 21 ± 6, respectively. Group 2 mothers had a lower mean S than did group 1 or 3 mothers - 7 ± 2; however, the differences between the mean responses of women in group 2 and those of group 1 and 3 mothers were not statistically significant (P = 0.22 and P = 0.07, respectively). The specificity of this apparent defect in maternallymphocyte transformation response (LTR) to CMV was investigated by comparing responses to CMV, HSV-l, and PHA (table 2). All 16mothers with antibody to HSV-1 had lymphocyte responses to that viral antigen. Their mean S to HSV-l was significantly higher than to CMV (P = 0.05). The S to both HSV-1 and PHA were similar in mothers and normal adult controls. Three women who failed to respond to CMV reacted with S similar to those of their peers (mean S, 25 ± 14) when tested with HSV-1.. Eleven children had no lymphocyte proliferation in response to CMV antigen (figure 1). Responses were absent in five of six children in group 2, five of 19 children in group 1, and one of

4 496 Reynolds et al. 100 ~ 80 ~ :..J.. J. x Q) -0 c c... o o ::> E ~ r: ~ [ O.6 '-~_.._. i---l..-_+_---l l..-_+_---l l.-+_---l l- Non-mmune Adults mmune Adults Silent Congenital nfection.. 2r ~ 0.8 ~ Symptomatic Congenital nfection Natal nfection Group 1 Group 2 Group 3 Figure 1. Lymphocyte transformation by cytomegalovirus (CMV) of mononuclear cells from nonimmune adults, immune adults, and three groups of mothers and their offspring with different types of perinatal CMV infection. Results reflect the highest reponse obtained in each assay when more than one CMV strain was used as antigen. t indicates mean ± SE. five children in group 3. The differences in frequency of absent responses were statistically significant between groups 2 and 1 (P = 0.02) and groups 2 and 3 (P = 0.04) children. Among those children with positive 81 to CMV, the magnitude of the mean response was substantially attenuated in comparison with group responses of their mothers or the adult controls, with the exception of group 2 women. Because of the low incidence of H8V-1 infection in the children, it was more difficult to test for the specificity of the defect in LTR to this antigen than to CMV. Only four children had antibody to H8V-; all of them had a positive response to H8V-, but they had a low mean 81-4 ± 0.8 (table 2). Three of these four also had blastogenic responses to CMV with a mean 81 identical to that for H8V-1. LTR to PHA were normal in the 19 children tested, including seven who had. no LTR to CMV. nfluence of viral excretion, age of the offspring, and serum antibody levels. Viral excretion and age of the offspring relative to LTR are shown in figure 2. Of the 20 children with persistent viruria, 19 were three years of age or younger, and 11 failed to respond to CMV antigen. n con-

5 Cellular mmunity and CMV 497 Table 2. Comparative lymphocyte transformation responses to cytomegalovirus (CMV), herpes simplex virus type 1 (HSV-), and phytohemagglutinin (PHA) in CMV-immune controls, children with congenital or neonatal infection with CMV, and their mothers. mmune controls Stimulant HSV-l CMV PHA No. with No. with No. with Mean S positive S/ Mean S positive S/ Mean S positive S/ ± SE no. tested ± SE no. tested ± SE no. tested 42 ± 14 3/3 34 ± 10 16/16 4 ± 0.8 4/4 40 ± 10 3/3 14 ± 3 13/16 4 ± ± 15 9/9 104 ± 20 20/20 81 ± 15 19/19 NOTE. All subjects seropositive to HSV-l were tested concomitantly with that antigen and CMV. S = stimulation index. trast, the eight with nonproductive infection responded, although most had low S. Seven of these children were older than three years of age. The difference in positive and negative S between those with and without viral excretion is significant (P < 0.(05). Likewise, a significantly greater percentage of children three years of age or younger failed to respond (P = 0.(05). Among the eight children who were not excreting virus, virus shedding had ceased from six to 48 months prior to examination. n these children, there was no correlation between the magnitude of the LTR and the duration of the nonproductive infection. The two nonexcreting children with high responses had fetal infection. The older child was symptomatic at birth and had severe neurological sequelae, while the younger patient was normal at birth and was still so at the time of this study. Neither time after delivery nor viral excretion significantly altered the CMV-specific responses of the mothers, although a trend toward lower stimulation was apparent in excreting women as compared to nonexcreting women (P = 0.10). Of interest in this regard is the observation that cells from two of the three women with persistent viral excretion of greater than 18 months' duration failed to respond to CMV but proliferated normally in response to HSV-1 antigen. All subjects had serum neutralizing antibody to CMV at the time of testing. There was no correla- ~ CHLDREN MOTHERS Excreting X Q,) Not excreting -0 c 26 0 c 22.Q "0 ::> 18 E 14 V> eo 6 0 i OJ ~2 oe ~ ( ~85 Age of the Child in Months Figure 2. Lymphocyte transformation responses to cytomegalovirus among mothers and children in relation to virus excretion and age of the child at the time of assay.

6 498 Reynolds et al. tion by linear regression analysis between antibody levels and LTR in mothers, children, or adult controls. Discussion Gehrz et al. [11] and Starr et al. [12] have previously demonstrated decreased specific LTR in a small number of mothers of children with either symptomatic or subclinical congenital CMV infection. As noted here, this lymphocyte defect is relatively common, not only in women who transmit CMV in utero but also in those with productive gestational infection who do not transmit the infection to the fetus. Certainly, then, intrauterine transmissibility is not solely dependent on this particular cellular immune defect. The finding that CM was reduced in mothers of diseased infants as compared to women who transmitted silent infection to their offspring before or at delivery may imply that the virulence of the fetal infection is causally related to this anomaly. More women in each category will have to be examined to confirm the observation. This maternal lymphocyte defect appears to be specific for CMV since LTR to HSV-l were normal in 16 mothers with previous HSV infection when tested concomitantly with HSV and CMV antigens. Three of these women had no response to CMV at all, and the LTR to CMV were attenuated in the remainder as compared to controls. Whether this specific deficiency in CM is genetically determined or secondary to the viral infection remains speculative. There was a definite trend toward lowered LTR to CMV in women with viral excretion at the time of the study, and persistent productive infection observed in three women was associated with absent responses in two of them. Although this observation, confirmed by the study of Starr et al. [12], establishes a relationship between impaired specific CM and productive infection, it does not define whether the lymphocyte defect is the cause or result of persistent active infection. Resolution of this issue as well as definition of the role of CM in the genesis of vertical transmission of CMV will require a prospective and long-term study of specific CM among mothers with documented primary and recurrent infection who do and do not transmit virus in utero. rrespective of whether CMV acquisition occurs in utero, at delivery, or shortly after delivery, virtually all of the infected offspring have grossly impaired specific CM in comparison with the mothers or adult controls, with the possible exception of mothers who delivered diseased infants. The defect is directly related to viral excretion in children, even more so than in the mothers. Younger children with the greatest level of virus excretion have a significantly higher prevalence of negative responses than do their older nonexcreting cohorts. This defect does not appear to be genetically based since there is qualitative restoration of LTR with advancing age and concomitant reduction of virus shedding. n children, however, functional immaturity of the lymphocyte in virus recognition may also play a role in depression of CM. This possibility is suggested by the following. First, as mentioned, the youngest patients have the most profound defect in specific CM. Second, unlike the adults, the lymphocyte defect was not confined exclusively to CMV but also involved HSV antigen recognition in four children with antibody to HSV. Consequently, the lymphocytes of infants and young children may have a broad inability to recognize viral.antigens, at least those of the herpesvirus family, even though the cells can respond normally to nonspecific mitogen, such as PHA, as noted here and by others [11, 13-15]. Studies of specific CM of children with documented postnatally acquired CMV infection would help answer this question but are logistically very difficult to perform. The pathologic contribution of the in vitro cellular immune defect observed in children is not clearly defined by these or other available data [11, 12]. n our study, specific cellular immunity was more profoundly decreased in symptomatic than in subclinically infected infants with intrauterine acquisition of virus. A significantly higher frequency of negative responses was noted among subjects with CD (five of six) than among children with inapparent fetal infection (five of 19) (P = 0.04). This relationship held true even when adjustments were made for age and virus excretion. Of the 10 children with disease in previous reports [11, 12], all but two had no LTR to CMV, and the two with positive responses were older and no longer excreting virus. Even though these data suggest an important etiologic role for deficient CM in the virulence of fetal.infection, this defect cannot be the exclusive factor, since eight silently

7 Cellular mmunity and CMV 499 infected children with negative LTR have been noted here and in the study by Starr et al. [12]. The fact that the responses among the great majority of test subjects to the standard laboratory and typical wild strain of CMV are comparableargues against the postulate that absent or low S could be due to lack of antigenic experience with the test antigens. Strain-specific responses, however, have also been demonstrated among children with CMV infection whose age at acquisition was unknown by Beutner et al. [16]. Definitive resolution of this question will require that the assays be performed with use of the patient's own strain of virus. Whatever the nature of the suppression of LTR may be, it is quite clear that the defect does not involve humoral immunity. All test subjects had antibody to CMV, and there was no correlation between the height of antibody titer and the presence, absence, or magnitude of the blastogenic response to CMV. Similar results were reported by Gehrz et al. [11]. ndeed, children with fetal and neonatal CMV infection have been shown to have normal antibody responses to heterologous viruses [17] as well as bacterial antigens [18]. Thus, since CM is the only immune function definitely shown to be impaired in these patients and is quite likely related to the persistence and pathology of perinatal CMV infections, it is imperative to define better the mechanism and extent of this deficiency. References 1. Melish, M. E., Hanshaw, J. B. Congenital cytomegalovirus infection. Am. J. Dis. Child. 126: , Stagno, S., Reynolds, D. W., Tsiantos, A., Fuccillo, D. A., Long, W., Alford, C. A., Jr. Comparative, serial virologic and serologic studies of symptomatic and subclinical congenital and natally acquired cytomegalovirus infections. J. nfect. Dis. 132: , Reynolds, D. W., Stagno, S., Alford, C. A. Recurrent cytomegalovirus (CMV) infection in adult females [abstract no. 231]. n Program and Abstracts of the Fifteenth nterscience Conference on Antimicrobial Agents and Chemotherapy, Washington, D.C., Stagno, S., Reynolds, D. W., Huang, E. S., Thames, S. D., Smith, R. J., Alford, C. A., Jr. Congenital cytomegalo- virus infection: occurrence in an immune population. N. Engl. J. Med. 196: , Reynolds, D. W., Stagno, S., Reynolds, R., Alford, C. A., Jr. Perinatal cytomegalovirus infection-the influence of placentally transferred maternal antibody. J. nfect. Dis. 137: , Pollard, R. B., Rand, K. H., Arvin, A. M., Merigan, T. C. Cell-mediated immunity to cytomegalovirus infection in normal subjects and cardiac transplant patients. J. nfect. Dis. 137: , Ho, M. Virus infections after transplantation in man. A brief review. Arch. Virol. 55:1-24, Gardner, M. B., Officer, J. E., Parker, J., Estes, J. D., Rongey, R. W. nduction of disseminated virulent cytomegalovirus infection by immunosuppression of naturally chronically infected wild mice. nfec. mmun. 10: , Starr, S. E., Allison, A. C. Role of T lymphocytes in recovery from murine cytomegalovirus infection. nfec. mmun. 17: , Boyum, A. solation of mononuclear cells and granulocytes from human blood. Scand. J. Clin. Lab nvest. 21(Suppl. 97):77-89, Gehrz, R. c., Marker, S. c., Knorr, S. 0., Kalis, J. M., Balfour, H. H. Specific cell-mediated immune defect in active cytomegalovirus infection of young children and their mothers. Lancet 2: , Starr, S., Tolpin, M., Friedman, H., Paucker, K., Plotkin, S. Cellular immunity to cytomegalovirus in congenitally infected infants and their mothers [abstract no. 515]. n Program and Abstracts of the Eighteenth nterscience Conference on Antimicrobial Agents and Chemotheraphy, Atlanta, Ga., American Society for Microbiology, Washington, D.C., Marshall, W. c.. Cope, W. A., Soothill, J. F., Dudgeon, J. A. n vitro lymphocyte response in some immunity deficiency diseases and intrauterine virus infections. Proc. R. Soc. Med. 63: , Gibas, H., Hayes, K. n vitro activity of lymphocytes from children with prenatal cytomegalovirus infection. Aust. Paediatr. J. 10:86-93, Montgomery, J. R., Desmond, M. M., Mason, E. 0., Yow, M. D., Hittner, H. A prospective study of congenital cytomegalovirus infection. Clinical Research 23: 67A, Beutner, K. R., Morag, A., Deibel, R., Morag, B., Raiken, D., Ogra, P. L. Strain-specific local and systemic cellmediated immune responses to cytomegalovirus in humans. nfec. mmun. 20:82-87, Reynolds, D. W., Stagno, S., Herrman, K. L., Alford, C. A. Antibody response to live virus vaccines in congenital and neonatal cytomegalovirus infections. J. Pediatr. 92: , Michaels, R. H. Suppression of antibody response in congenital rubella. J. Pediatr. 80: , 1972.