March 19-20, 2002
Report and Recommendations
Congenital cytomegalovirus (CMV) infection occurs in around 0.5 to 1.5% of live births in the U.S. The public health impact of congenital CMV infection is largely due to its ability to damage the central nervous system, including the auditory system. Although few population-based studies of the etiology of hearing loss in infants have been performed, when such studies have included assays for congenital CMV infection, they have strongly suggested that congenital CMV infection is a leading cause of sensorineural hearing loss in children.
Based on these significant but limited studies, the National Institute on Deafness and Other Communication Disorders (NIDCD) convened a workshop attended by experts in the related fields of congenital CMV infection and hearing loss, intrauterine CMV transmission, treatment of CMV-positive neonates, neonatal hearing and metabolic screening, and CMV diagnostics. The purpose of the workshop was: 1) to present the current research in these related areas; 2) to better determine the degree to which congenital CMV infection contributes to hearing loss in children; and 3) to facilitate a discussion among these experts in order to develop a set of recommendations for future research in the area of congenital CMV infection and hearing loss.
Dr. Robert Pass opened the workshop with a general overview of our current understanding of congenital CMV infection and hearing loss. Congenital CMV infection occurs in around 0.5 to 1.5% of live births in the U.S. Cytomegalovirus causes lytic infection of cells and autopsy and experimental animal studies have localized CMV infection to specific components of the auditory system. In limited population-based studies of hearing loss in infants, which have included virologic ascertainment of congenital infection, the results have suggested that CMV infection is a leading cause of sensorineural hearing loss and perhaps the leading cause in children. Follow-up studies have shown that a substantial portion of hearing loss due to congenital CMV infection either has onset after the newborn period or shows progressive decline in auditory thresholds. Therefore, universal hearing screening based on hospital nursery screening of newborns would fail to identify late onset hearing loss due to CMV.
Although observational studies of women with naturally acquired infection suggest that immunity to CMV infection can significantly reduce the rate of congenital infection, a vaccine is years in the future. Preliminary results from a randomized trial of antiviral treatment of newborns with severe congenital CMV infection show that such treatment might be able to reduce ongoing damage to hearing. Studies of the pathogenesis of hearing loss in infants with congenital CMV infection provide a biologic rationale for the effectiveness of antiviral treatment.
Studies at the University of Alabama, Birmingham
Dr. Karen Fowler described a cohort of 860 children with congenital CMV infection who were born between 1966 and 1999, and have been followed in a longitudinal study at the University of Alabama at Birmingham (UAB). Children were identified either by referral or as part of routine newborn CMV screening at two hospitals in Birmingham. The population consisted predominantly of children of young black mothers. All infections were confirmed by isolation of virus in urine or saliva. Infected children received medical, psychometric, vision and hearing evaluations. 651 children had no clinically apparent infection at birth and were classified as having asymptomatic congenital CMV infection (ASX group). The remaining children (209) were documented as having symptomatic congenital CMV infection (SX group). Sensorineural hearing loss (SNHL) occurred in 7.4% of the ASX group and 40.7% of the SX group. Late onset SNHL occurred in both groups of children with the median age of hearing loss onset ranging from 33 to 44 months.
Variability in the severity of hearing loss ranging from unilateral high frequency loss only (4000-8000 Hz) to bilateral profound (>90 dB HL) loss followed congenital CMV infection. About 60% of all children with SNHL due to congenital CMV infection had bilateral loss. Also, progressive and fluctuating SNHL were observed in children with hearing loss due to CMV. In a subset of asymptomatic children identified by routine newborn CMV screening, there were no maternal or perinatal factors that indicated which children were at risk for developing late onset SNHL due to CMV.
In another subset of 388 children born from 1980 to 1996 at UAB University Hospital, SNHL was present in 5.2% of CMV infected infants at birth. Late onset SNHL occurred among the CMV infected children throughout the first 6 years of life. By 72 months of age, the cumulative incidence of SNHL was 15.4% in the group. If universal newborn hearing screening had been in place during this time less than half of all SNHL caused by CMV would have been identified by newborn auditory screening. Since most of the infants with congenital CMV infection are without symptoms at birth these children are unlikely to be recognized as being at risk for SNHL and would not receive further hearing evaluations to detect late-onset hearing loss.
Studies at the Baylor College of Medicine
Dr. Gail Demmler of the Baylor College of Medicine discussed work begun in 1982 involving a cohort of patients followed in a longitudinal study on the long term effects of congenital CMV infection on SNHL. Hearing loss has developed in 26 (66.7%) of children with symptomatic infection, 15 (15.3%) with asymptomatic congenital infection, and 1 (5%) uninfected control.
Frequency of hearing loss progression was at least 86% and 100% if mid-frequency SNHL was detected. Documented late onset hearing loss was seen in 12 children so far (4 asymptomatic and 8 symptomatic) at a mean age of 2.16 years (range 3 months to 5.1 years) for mid-frequency SNHL and 10.8 years (range 9.4 - 13.6 years ) for high-frequency SNHL. Distortion product otoacoustic emission (DPOAE) testing showed excellent correlation for detection of SNHL in 138 of 157 subjects. Dr. Demmler concluded that her results demonstrate that SNHL is a frequent complication in children born with congenital CMV infection. The loss is almost always progressive, leading to severe/profound loss in the affected ear(s) of both symptomatic and asymptomatic congenitally infected children.
Studies in Iowa and Identification of Multiple CMV Strains
Dr. James Bale, Jr. discussed the epidemiology of congenital CMV infection and disease in an Iowa study, which included multiple molecular approaches for the identification of CMV strains in infected infants. To determine factors that influence the occurrence of congenital CMV infection, 8,254 infants born in eastern Iowa between October 1989 and June 1994 were surveyed. A case-control study was implemented to identify maternal risk factors, matching each CMV-infected infant with three uninfected infants according to hospital and date of birth. CMV strains were compared using polymerase chain reaction (PCR)-based methods to identify common sources of infection. Of 7,229 infants cultured successfully for CMV, 35 (0.48%) were congenitally infected. PCR analysis detected 27 distinct CMV strains in the study population.
Recently, short tandem repeat (STR) length polymorphisms have been identified in the human CMV genome. Multiplexed polymerase chain reaction assays using primers based upon these STRs accurately mapped CMV strains. Using primers for ten microsatellite regions, the STR profiles of 44 wild-type and two laboratory strains of CMV were characterized. The results of STR analysis were compared with strain characterization using nucleotide sequencing and restriction fragment length polymorphism (RFLP) analysis. STR analysis accurately identified strains that were indistinguishable or distinct by conventional molecular analysis indicating that STR analysis allows rapid, precise molecular characterization of CMV.
Pre-existing Maternal Immunity against CMV
Dr. Suresh Boppana described his studies showing that preexisting maternal immunity against CMV provides substantial but incomplete protection against intrauterine transmission and severe fetal infection. CMV can be transmitted across the placenta in women who are CMV seropositive before pregnancy. Although mother-to-fetus transmission of CMV does occur in women with preexisting immunity, previous studies have suggested that the incidence of sequelae in children born to women with non-primary infection (women with pre-existing immunity) is significantly lower than those infants born following primary maternal CMV infection. These observations led to the strategy that an effective vaccine for the prevention of primary CMV infection during pregnancy could prevent or significantly reduce the neurologic damage from congenital CMV infection. However, more recent studies from Dr. Boppana’s laboratory and from Ahlfors, et al. in Sweden, demonstrated that symptomatic congenital CMV infection is also seen following a non-primary maternal infection. In addition, congenitally infected children born to immune mothers have been shown to be at significant risk for an adverse outcome.
The mechanisms and pathogenesis of intrauterine transmission of CMV in immune mothers has not been defined. The transmission of CMV to the fetus in the setting of a non-primary maternal infection could result either from reactivation of a previously acquired virus or from reinfection with a new virus strain. Recent studies addressed whether reinfection with a new virus strain could result in intrauterine transmission of CMV in immune mothers. Antibody responses against the aminoterminal region of envelope glycoprotein gH, which contains a strain specific neutralizing antibody epitope, were tested in serial sera from 46 women who were CMV seropositive at the time of their first pregnancy and delivered another infant at UAB during the study period. Sixteen of the mothers delivered infants with congenital CMV infection (transmitters) and the remaining 30 had uninfected infants (non-transmitters). Acquisition of new antibody specificities between pregnancies was observed in the majority of the transmitters (10/16) compared with only 4/30 non-transmitters. CMV virus isolates were available from seven of the congenitally infected infants and the maternal neutralizing antibodies were measured against a reference laboratory strain and the virus strain isolated from the infants. There was a significant boost in the neutralizing titers against the virus from 4 infants in the corresponding maternal sera obtained at 2nd delivery. In addition, three of the 4 infants had symptomatic infection. These results suggested that in women with preexisting immunity to CMV, acquisition of a new virus strain between pregnancies was associated with intrauterine transmission and disseminated fetal infection. However, the exact frequency of CMV infection in seropositive women and the significance of such reinfection to intrauterine transmission are not known. Furthermore, the factors that are associated with CMV infection in healthy seropositive individuals have not been defined. Developing a clear understanding of these issues is important for the formulation of effective strategies to prevent or reduce the incidence of hearing loss in children.
Treatment of Symptomatic Congenital CMV-Infected Newborns
Dr. David Kimberlin presented the study design and results of a phase III randomized clinical trial in which symptomatic neonates were treated with an anti-herpes virus drug. The study is part of the NIAID-supported Collaborative Antiviral Study Group (CASG). The study population consisted of congenitally infected newborns less than one month of age with significant CMV-induced CNS disease and abnormalities.
Patients received intravenous administration of ganciclovir at 12 mg/kg/day for six weeks. The primary endpoint of the study was improved hearing (measured as brain stem evoked response) at six months compared with the baseline measurement, or for those babies with normal hearing at baseline, maintenance of normal hearing at the six month assessment. Follow up assessments were performed at months 6, 12, 24, 36, 48, and 60.
At 6 months, none (0%) of 26 treated newborns had worsening in hearing compared to 44% in the untreated group whose hearing worsened during the first six months of life. When assessed at one year or older, only 20% of the treated group had progressive hearing loss, compared to 70% of the untreated group. At the 6 month hearing assessment, 85% of treated infants had either improved hearing or maintenance of normal hearing, compared to 56% in the untreated group. All of the above results were statistically significant.
Sixty-three percent of ganciclovir recipients developed Grade 3 or 4 neutropenia, compared with 21% of patients in the no treatment group. Of the 29 ganciclovir recipients experiencing such toxicity, 14 required dosage adjustments: 3 infants responded favorably to dose adjustments; treatment was stopped and successfully restarted in 7 infants; and treatment had to be permanently stopped in 4 cases. Of note, ganciclovir is gonadotoxic and carcinogenic in some animal models, although no correlates have been seen with use in humans.
The general conclusions of the study were the following. Ganciclovir therapy both improves hearing (or maintains normal hearing) and prevents hearing deterioration at 6 months. Therapy may have a functional effect on prevention of hearing deterioration at ≥ 1 year. Two thirds of treated patients developed significant neutropenia. Long-term therapy with orally bioavailable drugs should be evaluated in clinical trials. Drug therapy may be considered in patients with symptomatic congenital CMV disease involving the CNS.
Current Neonatal Hearing and Metabolic Screening Programs
Irene Forsman and Dr. Michele Lloyd-Puryear of the Maternal and Child Health Bureau of the Health Resources and Services Administration (HRSA) discussed the status of hearing screening and metabolic screening, respectively.
Currently there are 47 grants to states or jurisdictions for implementation of sustainable statewide systems of newborn hearing screening, audiologic testing by 3 months of age and enrollment in early intervention programs by the age of 6 months. An important component of all the programs is strong links to a medical home for infant/family and family-to-family support services.
The establishment of universal newborn hearing screening prior to hospital discharge was initially endorsed by an NIH Consensus Conference in 1993 and in 1999 congress passed into law a bill supporting a national program to develop and support universal newborn hearing screening. It has been estimated that approximately 67% of children born in the U.S. currently receive newborn hearing screening. Although the Academy of Pediatrics endorsed newborn hearing screening in 1999, many pediatricians remain uninformed, and therefore, skeptical about the benefits of newborn hearing screening.
Newborn screening for metabolic disorders leading to catastrophic health consequences has been a concern of public health departments since the development of screening algorithms in the early 1960s. Shortly after it was shown that newborn population screening for inborn errors of metabolism could reduce the frequency of or prevent mental retardation and reduce the economic burden to the family and to society, state-based screening programs spread rapidly across the country. These early studies and the health promotion that followed were funded by federal health dollars and were critical in establishing public health newborn screening systems that included education, screening, follow-up, diagnosis, management, and evaluation.
Within newborn screening systems, there are critical junctures between system components where responsibilities and coordinated activities must be seamless and non-duplicative. Lessons learned from newborn screening implementation and expansion over the years must be utilized as new systems are developed. Current dried blood spot screening programs evolved from fragmented systems of public and private laboratory services, with disjointed follow-up, to their current levels of integrated systems over a period of almost 40 years. Today’s improved systems resulted from valuable input from the public and professional communities such that virtually all newborns receive newborn metabolic screening and essentially all who need follow-up testing, diagnosis, and treatment, receive it. In addition, newborns now are receiving expanded dried blood spot screening for heritable disorders in some states (often uncoordinated through private screening laboratories), and almost all states have established universal newborn hearing screening programs.
This expanded testing must be embraced by public health systems in such a way that there is: (1) minimal duplication of data tracking between programs that serve the same population; (2) rapid follow-up coupled with efficient and effective delivery of ancillary medical services; (3) adequate privacy protection; and (4) appropriate outcome data for system evaluation and improvement. Where nationally recognized standards of care exist, public health systems must comply. Ultimately, state public health newborn screening systems must address nationally issues of uniformity in: (1) service availability (education, testing, follow-up and care), (2) public health decision-making processes related to testing, and (3) methods of financing.
Regarding the possibility of newborn screening for congenital CMV infection, Dr. Puryear outlined criteria that would have to be met for such screening to become routinely integrated into current screening programs. The right assay must be found/developed. Characteristics of such an assay include a low incidence of false negative calls and technology that blends with that already in place. For instance, a CMV assay from the dried blood spots used for most metabolic screening would be most desirable. Importantly, the data on congenital CMV infection and neonatal disease in the U.S. has primarily utilized urine testing and no states use urine as a source for any of their current metabolic tests.
Status and Efficacy of Assays for Congenital Cytomegalovirus Infection
Dr. Maria Paola Landini described the aggressive approach to congenital CMV detection that has been put into place in Italy and other European Countries. She also discussed results showing the clinical value of a number of assays for CMV utilizing the serum of mothers, amniotic fluid and blood of newborns.
Unlike the United States, in most European Countries pregnant women are tested for CMV during the first trimester of pregnancy. Within the last five years European laboratories have made significant progress in determining the optimal assays for determining the CMV disposition of the fetus and predicting the probability of significant clinical outcomes due to congenital CMV infection.
With advances in CMV serology, the presence of anti CMV-specific immunoglobulin-M (IgM) detected by a screening test such as the enzyme linked immunosorbent assay (ELISA), can be confirmed by Immunoblot, identifying pregnant women undergoing an active or recent infection. Furthermore, primary infections observed or suspected due to the presence of CMV IgM in mothers’ blood, can now be readily diagnosed by disclosing the presence of anti-CMV low avidity immunoglobulin-G (IgM) antibody in IgM-positive mothers, greatly reducing the number of women who should be considered at risk of transmitting the virus to the fetus. Of great significance is the finding that only 30% of primarily infected mothers transmit the virus. Identifying these women is a major diagnostic problem that has been solved by moving to in utero diagnosis.
In utero diagnosis carried out by quantitative PCR in amniotic fluid of primarily infected mothers taken at 21-22 weeks of gestation can identify fetuses that are at no risk of congenital infection and also fetuses with a low viral load that should not be considered at risk of developing a symptomatic congenital infection. These results can help physicians to counsel infected mothers, thus allowing a much higher number of mothers to continue their pregnancy with confidence. In addition, as treatment options arise, the pre-natal identification of a severely infected fetus could also lead to treatment of the mother with an anti-CMV drug that crosses the placenta for pre-emptive fetal therapy.
Most recently, significant progress has been made on neonatal diagnosis. The quantitation of the number of CMV genomes in neonatal blood is being used successfully to predict the severity of disease. Both symptomatic infections in newborns at birth and congenital infections in newborns that will develop delayed sequelae can be identified by an increased number of CMV genomes in the neonatal blood. In contrast, asymptomatic congenitally infected newborns clearly show a reduced number of CMV genomes. This is an important advance that allows for the identification of newborns requiring a more careful clinical follow up.
Recommendations of Areas for Future Research
At the conclusion of the meeting, participants reviewed the material presented during the workshop and discussed research needs, by considering the question, “What knowledge would be most valuable in developing a better understanding of the role of congenital CMV infection in hearing loss and in developing approaches to prevent or control damage to hearing from congenital CMV infection?” As a result of this discussion, a number of areas were recommended as top priority for research support. These priority areas for research could be addressed by focused studies in humans, molecular studies of virus or host cells in the laboratory, or in animal model systems. The following key questions/research priorities were identified:
- What is the role of congenital CMV infection as a cause of hearing loss in infants, and how can we optimize early detection of hearing loss in infants with congenital CMV infection? What is the epidemiology of hearing loss in children and the role that CMV plays in relation to other causes of hearing loss?
- Why do some infants with congenital CMV infection develop hearing loss while others maintain normal hearing? To what extent do maternal or infant host factors or virus-determined factors influence outcome with respect to development of hearing loss? Very little is known about the effect of viral strain differences as determinants of virulence of human CMV infection. Can viral genetic heterogeneity or specific viral genotypes be linked to dissemination of CMV, tropism of virus for cells of the inner ear or host response to infection? Similarly, it will be important to determine whether host genetic factors influence the development of hearing loss in infected infants, and if so, how these genetic factors operate.
- How does congenital CMV infection cause hearing loss? What cells of the auditory system are damaged and how does this damage occur? Is this damage the result of lytic viral infection of cells or does the host immune response cause or amplify damage to cells of the inner ear?
- What mechanisms account for delayed onset hearing loss or progressive hearing loss in infants and children with congenital CMV infection?
- Can antiviral treatment administered for a longer period of time provide additional improvement in hearing function or stop progressive deterioration of hearing in infants with hearing loss due to congenital CMV infection? Is there a role for antiviral therapy initiated beyond the neonatal period for the management of CMV-related hearing deterioration in older children and adolescents?
- Can we prevent hearing loss due to congenital CMV infection by preventing maternal or congenital CMV infection by vaccine or other means? This question could be addressed by studies aimed at developing a better understanding of the extent to which maternal immunity protects the fetus and infant from CMV infection and from damage to the auditory system. This question could also be addressed by proposals to develop or evaluate vaccines or other interventions aimed at prevention of maternal and congenital CMV infection.
- Can prenatal diagnosis of maternal and congenital CMV infection be used to determine damage to the fetal auditory system or risk for damage to hearing? Can prenatal antiviral treatment of selected pregnancies complicated by virologically proven, clinically evident fetal CMV infection be used to improve the outcome for the fetus?
Due to the compelling but limited data on congenital CMV infection and hearing loss in neonates, the NIDCD convened a panel of experts working on different aspects of congenital CMV infection as well as experts on the screening of newborns for hearing loss and metabolic disorders. The latest research findings were presented and there was lively and productive discussion throughout the workshop. The resulting recommendations will be extremely useful to NIDCD and scientific community regarding future research programs on congenital CMV infection and hearing loss.