Cytomegalovirus Infection of Extremely Low Birth Weight Infants via Breast Milk

Transcription

1 MAJOR ARTICLE Cytomegalovirus Infection of Extremely Low Birth Weight Infants via Breast Milk J. Maschmann, 1,4 K. Hamprecht, 2 K. Dietz, 3 G. Jahn, 2 and C. P. Speer 1,4 Departments of 1 Neonatology, 2 Medical Virology, and 3 Medical Biometry, University of Tübingen, Tübingen; and 4 Children s Hospital, University of Würzburg, Würzburg, Germany In addition to seroprevalence and transmission rate, the clinical symptoms of postnatal cytomegalovirus (CMV) infection in infants with a very low birth weight (VLBW;!1500 g;!32 weeks gestational age at birth) were assessed in a 3-year prospective study. CMV monitoring included serologic testing (of the mother and child) and virus culture and PCR (of samples of both breast milk and the infant s urine). Within 3 weeks of the initial virus detection in the infant, clinical and laboratory parameters were evaluated. Of 170 infants, no CMV transmission was found in the 80 infants of seronegative mothers and in the 3 infants of seropositive mothers who did not shed CMV DNA into breast milk. Transmission occurred in 33 of the 87 CMV-exposed infants, 16 of whom presented with such symptoms as hepatopathy, neutropenia, thrombocytopenia, and sepsis-like deterioration. Low birth weight and early postnatal virus transmission were risk factors for symptomatic infection. VLBW infants of CMV-seropositive mothers are at high risk of acquiring a symptomatic CMV infection postnatally via breast milk. Postnatal cytomegalovirus (CMV) transmission to preterm infants has been reported to have serious consequences for example, respiratory deterioration and hepatosplenomegaly [1, 2]. Of special concern is the donation of CMV-seropositive blood products, which may have a deleterious impact with fatal outcome [3 5]. Since the early 1970s, it has been known that human breast milk is a potential source of CMV infection [6 8], and different rates of virus transmission from mother to child have been reported. In a prospective study, we described the epidemiology and ki- Received 14 February 2001; revised 10 July 2001; electronically published 12 November Presented in part: 17th European Congress of Perinatal Medicine, Porto, Portugal, June, The study was approved by the Ethics Committee of the Medical Faculty of the University of Tübingen, and informed consent of the parents was obtained. Financial support: Deutsche Forschungsgemeinschaft (grant HA ). Reprints or correspondence: Jens Maschmann, University Children s Hospital, Josef-Schneider-Str. 2, D Würzburg, Germany (jens.maschmann@mail.uni -wuerzburg.de). Clinical Infectious Diseases 2001; 33: by the Infectious Diseases Society of America. All rights reserved /2001/ $03.00 netics of maternal CMV reactivation [9] in 176 preterm infants of 151 mothers; in that study, 6 mothers did not breast feed. Of the remaining subjects, 69 mothers (with 80 infants) were seronegative and 76 mothers (with 90 infants) were seropositive; 73 (96%) of these 76 mothers shed CMV DNA into breast milk. None of the infants of seronegative mothers or of the seropositive mothers who did not shed CMV into breast milk were CMV infected. A total of 57 recipients of CMV DNA contained in breast milk were without transmission, but 33 became infected. Preliminary results of the clinical symptoms possibly associated with CMV infection after 1 year of observation have been reported elsewhere [10]. In the present article, we present detailed clinical data on primary CMV infection of premature infants for the entire 3-year observation period. PATIENTS AND METHODS From 1 July 1995 through 30 June 1998, all motherinfant pairs of inborn premature babies who were!1500 g and/or born before the completed 32d week of gestation were included in the study and prospec CID 2001:33 (15 December) Maschmann et al.

2 Table 1. Anthropometric and obstetrical data from a study of the cytomegalovirus (CMV) status of 170 premature infants. Characteristic Seronegative group (n p 80) Without CMV infection (n p 57) Seropositive group With CMV infection (n p 33) Gestational age, median weeks (range) a 28 6 ([23 4] to [34 2]) 29 5 ([23 3] to [33 6]) 28 6 ([24 0] to [32 1]) Birth weight, median g (range) 1080 ( ) 1160 ( ) 1100 ( ) Cesarean section, n/n (%) 67/80 (84) 51/57 (89) 28/33 (85) Premature rupture of membranes b Occurrence, n/n (%) 6/76 (8) 11/54 (19) 10/33 (30) Duration, hours NOTE. There were no differences in birth weight, gestational age, and rate of cesarean section among the 3 groups. Significantly more premature ruptures of membranes were found in the seropositive group with CMV infection, compared with the seronegative group ( P p.006). There were no significant differences between the other groups. a For x y in the gestational age, x denotes completed weeks of gestation and y denotes the current day (1 6) of the subsequent week. b Defined as being longer than 12 h before birth. No information about membrane rupture was available for 7 cases. tively screened for CMV infection. Screening included initial serologic testing of the mother-infant pair and virus culture and PCR of breast milk and urine samples obtained at biweekly intervals until discharge. During a follow-up examination at the corrected age of 3 months, we obtained one last urine specimen from each child. Congenitally or perinatally acquired CMV infection was ruled out, and transfusion-associated virus transmission was prevented by exclusive use of CMV-seronegative blood products. Strict donation of the infant s mother s own breast milk was mandatory. Additional detailed information on study design and the performance of CMV screening by use of serologic testing, cell culture, and PCR has been provided elsewhere by Hamprecht et al. [9, 11]. We considered a 3-week interval starting 1 week before and ending 2 weeks after the first detection of CMV (by means of PCR or culture of a urine specimen) to be critical for the manifestation of CMV-associated symptoms. During this period, a careful evaluation of the clinical symptoms (apnea-bradycardia, hepatomegaly, petechiae, and seizures) was performed. In addition, throat swabs were obtained in all cases of virus transmission. The assessed laboratory parameters were WBC count, absolute number of neutrophils, thrombocyte count, and liver enzyme activity. For the period before we were aware of virus transmission, analysis was done by retrospective chart review. Neutropenia was defined as!1500 neutrophils/ml; thrombocytopenia, as!150 cells/nl; elevated aspartate aminotransferase, as 138U/L; elevated alanine aminotransferase, as 130U/L; and elevated g-glutamyl transferase (g-gt), as 1116U/L. In infants with clinical deterioration, infection, or cholestasis, a small blood sample (0.2 ml) was obtained for CMV PCR analysis of leukocytes and plasma. Any possible abnormal findings were monitored until there was a return to normal or the infant was discharged. The discharge criteria were respiratory stability and sufficient weight gain, with a body weight of g. On our ward, feeding was started routinely within the first 48 h of life by bolus instillation of freshly pumped (Egnell Elite) breast milk via a naso-gastric tube. Supplementation with breast milk fortifier (FM 85; Nestlé), trace elements, and vitamins was introduced after tolerance of 100 ml of breast milk per kg of body weight. Inadequate amounts of breast milk were supplemented with preterm formula milk (Beba FN; Nestlé). Statistical analysis was done using a logistic regression model with use of the logarithm of the day of life at initial virus detection and the birth weight of the infant as influencing variables on the outcome variable symptomatology of virus transmission. For the comparison of anthropometric and obstetrical characteristics between the groups, we used the Wilcoxon and Fisher s exact tests, where appropriate. Values were calculated using the JMP statistical software, version 4.04 (SAS Institute). RESULTS The anthropometric data regarding the infants are given in table 1. Infants were assigned to 1 of 3 different groups: preterm babies of seronegative mothers, preterm babies of seropositive mothers who did not have postnatal virus transmission, and preterm babies of seropositive mothers who had postnatal virus transmission. There were no statistical differences among the 3 groups with regard to birth weight, gestational age, and rate of cesarean section. However, premature rupture of membranes occurred more frequently in the group of seropositive infants who had virus transmission than it did in the seronegative group ( P p.006). The 2 subgroups of seropositive infants (with and without CMV transmission) did not differ with regard to the rate of premature rupture of membranes ( P p.293). Although none of the babies of seronegative mothers and none of the 3 infants of seropositive mothers who did not shed CMV DNA into breast milk acquired a CMV infection, 33 Cytomegalovirus Infection via Breast Milk CID 2001:33 (15 December) 1999

3 Figure 1. Clinical symptoms associated with cytomegalovirus (CMV) infection in a male infant of 24 weeks and 4 days of gestation and birth weight of 630 g. The initial findings were no viruria on day 15 and no DNAemia on day 32. The initial detection of CMV happened on day 47, with leukocytes and plasma DNAemia and viruria of 2 different CMV strains. There was subsequent need for reintubation on day 53 because of severe apnea-bradycardia and sepsis-like appearance. On day 55, a CMV isolate was recovered from specimens of tracheal secretions. Hepatomegaly occurred from day 56 onward. There was spontaneous recovery of all parameters and recovery of a CMV isolate from the throat on day 96., Platelet count;, neutrophil count; dashed lines, level for neutropenia (!1500 neutrophils/ml); dotted lines, level for thrombocytopenia (!150 cells/nl). (38%) of the 87 CMV-exposed babies of seropositive mothers did. In all cases, shedding of CMV or CMV DNA into breast milk preceded the transmission. The systematic screening of the infants with virus transmission with regard to clinical and laboratory findings showed that 16 (48%) of the 33 affected infants demonstrated at least 1 symptom. The most common finding was an absolute neutropenia (in 14 [42%] of 33 infants). Ten of these 14 infants had neutrophil counts of 335 neutrophils/ml to 966 neutrophils/ml. If one considers the symptomatic infants alone, 14 (88%) of 16 infants had low neutrophil counts. All episodes of neutropenia resolved spontaneously after 1 7 weeks, and there were no opportunistic bacterial infections during that period. Platelet counts were low in 4 infants (range, cells/nl). In 1 case, petechiae of the lower legs was present. After 1 7 weeks, the thrombocyte count returned to normal in all cases. The most severe clinical condition was present in 4 infants with sepsis-like symptoms; it consisted of apnea, bradycardia, and a gray pallor of the skin. Two of the affected premature babies had to undergo reintubation because of severe apnea attacks. There was no bacterial growth in blood cultures, but there was simultaneous initial CMV DNA detection in blood or urine samples by means of PCR, followed by urine cultures that yielded virus 2 weeks later. Three of these 4 infants with sepsis-like symptoms also had thrombocytopenia. Liver involvement was detectable in 5 infants with elevated g-gt activities ( U/L), 4 of whom had normal aspartate aminotransferase and alanine aminotransferase activities. However, 1 infant had hepatitis with an aspartate aminotransferase level of 145 U/L, alanine aminotransferase level of 222 U/L, and (along with 2 other infants) hepatomegaly. Both hepatopathy and petechiae were observed in 1 infant in the following sequence: hepatopathy on day 35, thrombocytopenia on day 41, and neutropenia and petechiae on day 54. In this infant, and in the 4 infants with sepsis-like symptoms, virus could be isolated from the additionally obtained throat swabs. In 2 of the infants, abnormal neurologic findings were present 6 and 9 days after the first virus detection with myoclonias of arms and legs, normal findings on an electroencephalograph, and spontaneous amelioration. Both of these infants also had low neutrophil counts during this episode. A representative time course of a symptomatic CMV transmission via breast milk of a 630-g birth weight male infant of 24 weeks and 4 days of gestation is shown in figure 1. Initial CMV detection on postpartum day 47 in leukocytes and samples of plasma and urine was paralleled by a decrease in the platelet count from normal values to 70 platelets/nl, followed by a period of thrombocytopenia of nearly 7 weeks in duration. Clinical deterioration occurred 6 days after the initial virus detection, with severe apneas and recurrent episodes of bradycardia, a distended bowel, and gray pallor of the skin that finally required reintubation to stabilize the infant. Because sepsis was suspected, antibiotic treatment was started, but the results of blood cultures remained negative. C-reactive protein concen CID 2001:33 (15 December) Maschmann et al.

4 trations increased to a maximum of 50 mg/l, and the results of PCR of plasma and peripheral blood leukocytes samples were positive for CMV. In addition, on day 55, CMV could be isolated from specimens of tracheal secretions. An enlargement of the liver (3 cm below the costal margin), together with an increase of g-gt levels to 279 U/L, was found 3 days after clinical deterioration. However, clinical signs of cholestasis were not detectable. The number of neutrophils gradually decreased to 624 neutrophils/ml on day 70 of life, when the clinical signs were already gone, and it reached the threshold of 1000 neutrophils/ml 14 days later. The serostatus of the infant on day 50 of life was CMV IgG negative and CMV IgM positive, whereas the inverse was found in the mother. Of interest, the viral genotype analysis of this case revealed that the mother reactivated sequentially 2 different strains of CMV that were shed into breast milk, and both of them could be detected in the infant [12]. Besides the maternal CMV strain 1 (detected on postpartum day 15), another genotypically distinct strain was found on postpartum day 26 in milk whey. In the infant s urine sample, both CMV strains were detectable initially on postpartum day 47, but only strain 2 could be detected on postpartum day 158. A logistic regression for possible risk factors for symptomatic virus transmission showed a significant association with birth weight and the logarithm of the age of the infant at the initial virus detection, as shown in figure 2. This means that a reduction of the birth weight by 100 g increases the risk for a symptomatic virus transmission by an OR of 1.39 (95% CI, ) and a 2-fold reduction of the time period until initial virus detection increases the risk for a symptomatic virus transmission by an OR of 3.58 (95% CI, ; figure 2 [symptomatic cases are represented by the x symbols in the graph]). Premature babies with a greater birth weight and a delayed virus transmission have a greater chance to cope with the primary CMV infection without symptoms (represented by the white triangles in the graph), just as full-term babies do. The linear score function for the logistic regression has the equation (birthweight) 4.24 log 10 (day of virus transmission). With respect to this score function, the area under the curve for the receiver operating characteristics curve equals 0.90 that is, one makes a correct prediction of symptoms in 90% of the cases if the prevalence is 50%. Only 1 baby with a 190% predicted risk of having a symptomatic virus transmission showed no clinical signs of infection. DISCUSSION During this 3-year evaluation period, there was a rate of CMV transmission to preterm infants via breast milk of 38% (33 of 87 infants), which is much lower than the rate of 59% reported Figure 2. Logistic regression model demonstrating the influence of birth weight and time until initial virus detection on the risk of symptomatic CMV infection. Of interest, 1 infant with a 190% risk for symptomatic CMV infection did not show clinical abnormalities. Open triangles, asymptomatic infants; blackened dots, infants with clinical symptoms; diagonal lines, calculated risk (%) for a symptomatic virus transmission. elsewhere after preliminary evaluation of the first year of the 3-year study [10]. However, all of the infected infants received CMV-positive breast milk from their CMV IgG positive mothers, whereas in the control group of CMV IgG negative motherchild pairs, there was neither viral detection in breast milk samples nor CMV infection of a premature baby. This finding further strengthens the evidence that breast milk is the major source of postnatal CMV infection, especially for preterm infants [9]. The molecular identity of maternal CMV strains isolated from breast milk with the strains from the infant s urine clearly demonstrates that breast milk was the only source of postnatal CMV infection in the preterm infant [12]. However, it is not clear why the majority of exposed infants (54 of 87) did not acquire CMV infection. Birth weight, gestational age, and rate of cesarean section did not differ among the 3 groups. Only the rate of premature rupture of membranes was significantly higher in the group of infants who acquired CMV via breast milk, as compared with the seronegative control group. We think that this difference is not related to postnatal CMV transmission, because there was no such difference between the 2 seropositive subgroups (subjects with and those without CMV transmission), and no single infant in the group with postnatal CMV transmission showed evidence of perinatal infection, as ruled out by PCR-negative surface swabs from the throat and ear. As shown in our study, the early onset of CMV reactivation, detected by DNA and virolactia in the milk whey fraction, and not the duration of CMV exposure, is a risk factor for transmission. Furthermore, we observed 2 pairs of twins who had discordant CMV transmission during lactation [9]. Therefore, it is very likely that the means to defend and protect from CMV infection must be sought in the premature infant. With regard to the group of infants with virus transmission, Cytomegalovirus Infection via Breast Milk CID 2001:33 (15 December) 2001

5 we could clearly demonstrate that the status of immaturity is positively correlated with the risk of a symptomatic virus transmission. One might speculate about the reasons for this finding for example, a greater permeability of the epithelial layer of the mouth and gastrointestinal tract and a lower level of circulating maternal antibodies. An attempt to prevent postnatal CMV infection of premature neonates by giving them iv CMV immunoglobulin failed. However, there was a trend that the rate of infants with CMV syndrome was reduced in the group who received immunoglobulin when compared with those who received placebo [13]. The most common laboratory finding of a postnatal CMV infection was an absolute neutropenia (14 of 16 infants with symptomatic CMV transmission). In addition to thrombocytopenia and clinical deterioration of the infant, one should always consider CMV primary infection. A longitudinal diagnostic workup should be started with a CMV PCR of leukocyte and plasma specimens to detect the period of DNAemia. The next test of CMV infection can be done by means of serologic testing (seroconversion with positive CMV IgM), whereas the virus detection in the infant s urine is always the last step in the transmission cascade, often lagging up to 2 weeks behind the virus detection in the blood. In all of our cases, the symptomatology was self limited, and there was no fatal outcome. The currently used definitions for CMV disease, CMV syndrome, and asymptomatic CMV infection of immunosuppressed recipients of bone marrow or organ transplants were not applicable for our study population of preterm infants [14]. Experience in the use of antiviral treatment of postnatally acquired CMV infection of preterm infants is very scarce [15] and must be considered experimental. Thus, none of the infants in this study received antiviral therapy. With regard to possible long-term sequelae of early postnatal CMV infection of preterm infants (e.g., sensorineural hearing loss), there are different findings reported in the literature [16 18]. Our own investigations are currently being performed. In our opinion, one should try to prevent postnatal CMV transmission to preterm infants by giving noninfectious breast milk, especially to babies of!30 weeks of gestation or!1000 g birth weight. To achieve this aim, breast milk must be treated in a way that inactivates CMV without destroying the nutritional and immunologic components. Holder treatment (63 C for 30 min) only does the first part effectively [19], whereas freezing does not totally eliminate viral activity [20, 21]. The gold standard of careful CMV inactivation in breast milk seems to be short-term pasteurization (72 C for 10 s) [22], but the means for this procedure are not commercially available. Thus, the dilemma of how to treat the milk of CMV seropositive mothers is still unsolved, but we think that, for more than psychological reasons, the use of the mother s own breast milk should not be refused [23 25]. The issue of breast milk treatment is currently being evaluated. Acknowledgments We thank E. Kisch and A. Baumeister for their excellent assistance in performing this study. References 1. Ballard RA, Drew WL, Hufnagle KG, et al. Acquired cytomegalovirus infection in preterm infants. Am J Dis Child 1979; 133: Dworsky M, Yow M, Stagno S, et al. Cytomegalovirus infection of breast milk and transmission in infancy. Pediatrics 1983; 72: de Cates CR, Roberton NRC, Walker JR. Fatal acquired cytomegalovirus infection in a neonate with maternal antibody. J Infect 1988; 17: Adler SP. Neonatal cytomegalovirus infections due to blood. Crit Rev Clin Lab Sci 1986; 23: Yeager AS, Grumet FC, Hafleigh EB, et al. Prevention of transfusionacquired cytomegalovirus infections in newborn infants. J Pediatr 1981; 98: Hayes K, Danks DM, Gibas H, et al. Cytomegalovirus in human milk. N Engl J Med 1972; 27: Peckham CS, Johnson C, Ades A, et al. Early acquisition of cytomegalovirus infection. Arch Dis Child 1987; 62: Stagno S, Reynolds DW, Pass RF, et al. Breast milk and the risk of cytomegalovirus infection. N Engl J Med 1980; 302: Hamprecht K, Maschmann J, Vochem M, et al. Epidemiology of cytomegalovirus transmission of mothers to preterm infants by breastfeeding. Lancet 2001; 357: Vochem M, Hamprecht K, Jahn G, et al. Transmission of cytomegalovirus to preterm infants through breast milk. Pediatr Infect Dis J 1998; 17: Hamprecht K, Vochem M, Baumeister A, et al. Detection of cytomegaloviral DNA in human milk cells and cell free milk whey by nested PCR. J Virol Methods 1998; 70: Prix L, Kuner R, Speer CP, et al. Evaluation of restriction fragment length polymorphism analysis of the UL10 UL13 genomic region for rapid identification of human cytomegalovirus strains. Eur J Clin Microbiol Infect Dis 1998; 17: Snydman D, Werner B, Meissner H, et al. Use of cytomegalovirus immunoglobulin in multiply transfused premature neonates. Pediatr Infect Dis J 1995; 14: Ljungman P, Griffiths P. Definitions of cytomegalovirus infection and disease. In: Michelson S, Plotkin SA, eds. Multidisciplinary approach to understanding cytomegalovirus disease. Amsterdam: Elsevier Science Publishers B.V., 1993: Yeager AS. Use of acyclovir in premature and term neonates. Am J Med 1982; 73: Kumar ML, Nankervis GA, Jacobs IB, et al. Congenital and postnatally acquired cytomegalovirus infections: long-term follow-up. J Pediatr 1984; 104: Paryani SG, Yeager AS, Hosford-Dunn H, et al. Sequelae of acquired cytomegalovirus infection in premature and sick term infants. J Pediatr 1985; 107: Yeager AS, Palumbo PE, Malachowski N, et al. Sequelae of maternally derived cytomegalovirus infections in premature infants. J Pediatr 1983; 102: Ford JE, Law BA, Marshall VME, et al. Influence of the heat treatment of human milk on some of its protective constituents. J Pediatr 1977; 90: Dworsky M, Stagno S, Pass RT, et al. Persistence of cytomegalovirus in human milk after storage. J Pediatr 1982; 101: CID 2001:33 (15 December) Maschmann et al.

6 21. Maschmann J, Speer CP, Jahn G, et al. Inactivation of cytomegalovirus (CMV) in breast milk [abstract 208]. European Society for Paediatric Research, Annual Meeting, June Pediatr Res 1999; 45: Goldblum RM, Dill CW, Albrecht TB, et al. Rapid high-temperature treatment of human milk. J Pediatr 1984; 104: American Academy of Pediatrics, Workgroup on Breastfeeding. Breastfeeding and the use of human milk (RE9729). Pediatrics 1997; 100: Goldman A. Modulation of the gastrointestinal tract of infants by human milk. Interfaces and interactions. An evolutionary perspective. J Nutr 2000; 130:426S 31S. 25. Bernt KM, Walker WA. Human milk as a carrier of biological messages. Acta Paediatr Suppl 1999; 88: Cytomegalovirus Infection via Breast Milk CID 2001:33 (15 December) 2003