Aim: To investigate the relation between cytomegalovirus (CMV) viral load on dried blood spots (DBS) from newborn biochemical screening (“Guthrie”) cards, and sensorineural hearing loss (SNHL) in congenital CMV.
Design: Cross-sectional study with retrospective case-note review.
Setting: Seven paediatric audiology departments in the United Kingdom.
Patients: 84 children, median age 7 years: 43 with known congenital CMV, 41 with unexplained SNHL.
Interventions: Half a DBS was tested for CMV DNA viral load by quantitative real-time polymerase chain reaction (PCR).
Main outcome measures: Pure tone average hearing thresholds (0.5–4 kHz).
Results: DBS CMV DNA viral load significantly correlated with hearing thresholds for the worse and better hearing ears (Spearman’s rank correlations: r = 0.445, p = 0.008 and r = 0.481, p = 0.004 respectively). Multivariable logistic regression showed that the effect of DBS viral load on the risk of SNHL remained important, when age and central nervous system involvement had been taken into account (odds ratio (OR) 2.76, 95% confidence interval (CI) 1.14 to 6.63, p = 0.024). The mean log DBS viral load was significantly higher in children with SNHL than in those with normal hearing (2.69 versus 1.64, 95% CI −1.84 to −0.27, p = 0.01). 8/35 (23%) children with unexplained SNHL tested positive for CMV DNA on DBS. One false positive result was obtained.
Conclusion: The risk of SNHL increased with DBS viral load. Further studies should investigate whether DBS CMV testing has a role in identifying asymptomatic congenitally infected neonates at risk of SNHL, and whether antiviral treatment can reduce this risk.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Congenital cytomegalovirus (CMV) affects approximately 3–12 per 1000 births and causes 10–30% of childhood sensorineural hearing loss.1 2 Only 10% of children are symptomatic at birth, and 22–65% of the survivors of this group have sensorineural hearing loss (SNHL).3 The remaining 90%, who are asymptomatic at birth, are also at risk of SNHL, which affects 6–23%.3 The SNHL usually presents too late to identify CMV as the causative agent because congenital CMV is confirmed by testing samples taken before day 21 of life.4 (CMV-positive samples taken after day 21 may reflect perinatal or acquired CMV, which have not been associated with childhood SNHL).5
CMV-related SNHL may be progressive and/or of late onset in approximately half of cases,6 7 but the pathogenesis is not fully understood. Virus-induced cell damage, virus reactivation, the host immune response and genetic factors have been implicated.8–10 Enlarged CMV-infected cells, inflammation and inner ear damage have been demonstrated during histopathological studies of temporal bone,8 and CMV DNA has been detected in perilymph collected during cochlear implantation.11
Early intervention is crucial in minimising the impact of SNHL on language development.12 Longitudinal studies predict that up to one half of all CMV-related SNHLs occur after the newborn hearing screen.3 The majority of late-onset CMV-related SNHL occurs before the age of 3 years, coinciding with the critical period for speech and language development.7 It is recommended that children with congenital CMV receive audiological follow-up after their newborn hearing screen.13 However, congenital CMV is rarely diagnosed in asymptomatic infants, making this largely unfeasible.3 14 A combined approach of screening for CMV and for SNHL has been suggested to prevent delay in detection, and facilitate early intervention.14 15
A randomised controlled trial showed that intravenous ganciclovir, which reduces CMV replication, also reduces hearing deterioration in infants with symptomatic congenital CMV and central nervous system (CNS) involvement.16 However, its side effects, such as neutropenia, can be severe.
Poorer outcomes in congenital CMV have been associated with increased viral loads in amniotic fluid and neonatal blood.17 18 Infants infected earlier in pregnancy may have greater neurological damage.19 20 CMV-related SNHL has been associated with signs of disseminated infection, such as intrauterine growth retardation (IUGR) and petechiae, the presence of CMV DNA in serum and increased urinary and blood viral load.21–23
CMV DNA has been extracted from dried blood spots (DBS) and detected by qualitative polymerase chain reaction (PCR).24 25 The DBS is collected during the first week of life, so this test has two important uses; firstly, to identify congenitally infected newborns for neonatal CMV screening,26 27 and, secondly, to retrospectively confirm congenital CMV in children who present with unexplained SNHL,27 28 neurological damage29 30 or autism.31 The technique is reported to be 71–100% sensitive and 98.5–100% specific.25 However, it remains a theoretical possibility that the CMV DNA could deteriorate with time of storage and/or that minute transfer of material from one card could cross-contaminate an adjacent card, sufficient to be detected by a sensitive PCR method.
In this study, quantitative real-time PCR was used to measure the number of genomic copies of CMV DNA per semicircle of DBS to obtain a DBS CMV viral load. We aimed to investigate the relation between DBS CMV viral load and SNHL, and how DBS viral load related to viral load in whole blood. We also tested for decrease in DBS viral load over time, and for cross-contamination of CMV DNA between neighbouring DBS.
SUBJECTS AND METHODS
There were two groups of eligible children, aged 0–18 years: children known to have congenital CMV, with or without SNHL; and children with unexplained SNHL.
Children with known congenital CMV already had a diagnosis of congenital CMV confirmed by positive CMV culture, antigen detection or DNA PCR in body fluid or tissue samples taken before day 21, and were deemed “symptomatic” at birth if they had any of the following clinical features: petechiae, hepatosplenomegaly, jaundice with conjugated hyperbilirubinaemia, microcephaly, seizures and chorioretinitis.32
SNHL was defined (for one or both ears) as air conduction thresholds >20 dBHL for pure tones, or >30 dBnHL for clicks or tone pips or bursts on auditory brainstem response testing (ABR) when middle ear function was normal, and as bone conduction thresholds >20 dBHL for pure tones, or >30 dBnHL for clicks or tone pips or bursts on ABR, when middle ear function was abnormal.
Exclusion criteria consisted of having a known cause of SNHL (other than congenital CMV), being born in a country without newborn blood spot screening and, for the unexplained SNHL group, having another cause of SNHL strongly suspected from their history or examination.
From April 2005 to June 2006, audiological physicians, paediatricians, otolaryngologists and virologists participating in the CMV in Hearing Impaired Children (CHIC) Study notified the researchers of eligible participants. All known eligible children were invited to participate. Signed informed consent allowed retrieval and testing of the child’s stored DBS for CMV DNA, data collection from the child’s notes and voluntary donation of blood and urine samples for CMV DNA PCR testing. The results (either “CMV DNA detected” or “CMV DNA not detected”) were sent to the participant’s consultant and general practitioner.
The virologist tested the DBS while blind to the child’s CMV status and level of hearing. The audiovestibular physician collected the audiological and medical data while blind to the child’s viral load results.
Measurement of DBS CMV viral load
The commercial QiaAmp DNA blood mini kit (Qiagen) extraction system was used to extract nucleic acid from half a DBS. To prevent contamination, disposable gloves were worn and changed after handling each card. Scissors were cleaned with 0.1 M hydrochloric acid. Three samples of water were processed at the same time as the DBS to test for contamination during the extraction procedure.
(5′-GAGGACAACGAAATCCTGTTGGGCA-3′ [gB1] and 5′-TCGACGGTGGAGATACTGCTGAGG-3′ [gB2])
Positive and negative controls were included on each run. Serial dilutions of cloned CMV glycoprotein B plasmid (from 200 copies/ml to 2×106 copies/ml) were included on each run, for accurate quantification of the CMV viral load. DBS viral loads were calculated using the sequence detection system software on the ABI 7000 platform. All clinical samples were analysed in triplicate. Positive samples were re-extracted and tested again.
Study Guthrie cards
Donated CMV-positive and CMV-negative blood samples were used to make “study” Guthrie cards (50 μl whole blood per circle). After one week, the study Guthrie card DBS extract, and the whole blood used to make that card, were tested. CMV-positive and CMV-negative cards were stored at room temperature, with one spot abutting the back of another, to mimic real-life storage conditions. They were tested for CMV DNA viral load at 1–2 monthly intervals for 12 months.
Demographic data, antenatal history, method of diagnosis of congenital CMV and sequelae of congenital CMV infection were recorded. Children with known congenital CMV were recorded as having CNS involvement if they had any of the following: abnormal brain imaging, cerebral palsy, seizures, global developmental delay or greater than mild cognitive impairment. The most recent pure tone hearing thresholds from 0.5–4 KHz were averaged for each ear. Risk factors for SNHL (for example, aminoglycoside treatment) and the results of aetiological investigations were recorded.
Statistical analysis was performed using the Statistical Package for Social Sciences version 13.0. Relations between hearing thresholds, log DBS viral load, demographic factors and risk factors for SNHL were explored for potential confounding effects before multivariable logistic regression analysis. A four-parameter logistic regression model was used for modelling the relation between viral load and hearing loss. The equation for the model was as follows:
y = A*C/A+(C−A)*exp(−B*x))+D
The optimum fit was achieved by repeatedly estimating the curves with one parameter held constant and then progressing to the other parameters in the sequence A then C and B then D and was carried out to minimise the sum of the squares and provide the best fit. The parameter values which satisfied this criterion were A = 1.06×10−6, B = 7.95, C = 91.1 and D = 16.9 with an associated r2 = 0.51. Computations were performed using Regress+ (downloadable from http://www.causascientia.org/software/Regress_plus.html). The Bland and Altman method was used to analyse the relation between DBS and whole blood viral load data.35
In all, 92 children were invited and 84 (91%) took part; 52 participants were female. The median age was 7 years and the mean duration of audiological follow-up was 6 years. All retrieved DBS had been taken within 21 days of birth. DBS could not be retrieved for 10/84 (12%) children (fig 1).
Children with known congenital CMV
A total of 39 children known to have congenital CMV had their DBS tested. Five children had received ganciclovir: three had normal hearing at ages 3.5 years, 3.3 years and 1.7 years respectively, one had bilateral severe-profound SNHL from birth and one died in infancy. Of the 34 children not treated with ganciclovir, nine had normal hearing and 25 had SNHL (four unilateral, 21 bilateral). Progression of SNHL was observed in 19/25 (76%) and onset after normal newborn ABR was documented in 3/25 (12%). By the end of the study period, all hearing losses had been confirmed by age 3 years in the symptomatic group and by age 5 years in the asymptomatic group. Nineteen of the 34 untreated children (56%) had CNS involvement.
For the 34 children with congenital CMV not treated with ganciclovir, there were significant positive Spearman’s rank correlations between pure tone average hearing thresholds and log DBS viral load for the worse and better hearing ears (r = 0.445, p = 0.008 and r = 0.482, p = 0.004 respectively). The best-fit curve was sigmoid, suggesting a rapid increase in hearing thresholds (indicating more severe SNHL) with increasing DBS viral loads (fig 2).
The mean log DBS viral load was significantly higher in the 25 children with SNHL than in the nine children with normal hearing, (2.69 versus 1.64; p = 0.01, 95% CI −1.84 to −0.27). The mean log DBS viral load was also significantly higher in the 18 children with bilateral severe SNHL (>70 dBHL) than in the 16 children with better hearing, (2.84 versus 1.93; p = 0.01, 95% CI −1.60 to −0.2). As the DBS viral load increased, so did the percentage of children with SNHL (fig 3).
Better ear pure tone average hearing thresholds did not significantly correlate with sex, age, ethnic group, known risk factors for SNHL or for the symptomatic group, the presence of petechiae or IUGR, but they did significantly correlate with log DBS viral load and CNS involvement (r = 0.481, p = 0.004 and r = 0.487, p = 0.004 respectively). On multivariable logistic regression analysis log DBS viral load remained independently associated with SNHL >70 dBHL in the better hearing ear (OR 2.76, 95% CI 1.14 to 6.63, p = 0.024), while CNS involvement was of borderline significance for being an independent predictor (OR 5.21, 95% CI 1.0 to 27.18, p = 0.050).
Children with unexplained SNHL
Of the 35 children with unexplained SNHL, eight (23%) tested positive for CMV DNA on DBS, consistent with a diagnosis of congenital CMV. One child was found to have bilateral widened vestibular aqueducts on computed tomography scan and so had dual pathology for progressive SNHL.
A 5-month-old girl with a low-level positive DBS CMV viral load was CMV seronegative and was not excreting CMV in her urine. Her mother was also CMV seronegative. Forensic testing confirmed that the blood on the DBS sample with her name matched her subsequent blood sample, so the initial PCR result was falsely positive.
Study Guthrie card results
Five study Guthrie cards were made using CMV-positive blood. Seven were made using CMV-negative blood. Over 12 months the viral loads of the positive study Guthrie cards were not seen to decrease, and the negative cards continued to test below the lower limit of detection. No study cards made with whole blood with viral load <1700 genomic copies/ml tested positive, which is in keeping with the previously reported level of detection of 1600 copies/ml.36 Although there was a significant correlation between blood CMV load and DBS load (p<0.001) the Bland and Altman analysis illustrates that the two results cannot be treated synonymously (fig 4).
These results suggest that the reported increased risk of SNHL with increased viral load in urine and whole blood23 may also apply to the DBS sample, despite the variation in age and storage conditions for Guthrie cards. We anticipated that a cut-off value might be apparent, above which antiviral treatment could be given to prevent hearing loss. However, the steepness of the curve means that any neonate testing positive (irrespective of viral load) on DBS may be at risk of SNHL, consistent with the qualitative PCR results reported for serum.22
Viral load changes reflect the net result of virus replication and immune control. A “threshold effect” of viral load is seen in the immunocompromised, where antiviral therapy may be used pre-emptively to reduce CMV load and prevent CMV disease.37 The increased risk of SNHL with increased neonatal viral load provides a possible explanation for the reported beneficial effects of a 6-week course of ganciclovir on hearing.16 (Ganciclovir lowered plasma and urinary viral loads in treated infants.38 39) A sigmoid curve would be compatible with such transient lowering of viral load reducing the risk of SNHL to a greater extent than would be expected if there was a linear relation with viral load.
The proportion of children with unexplained SNHL testing positive for CMV DNA on DBS (23%) is consistent with previous reports.27 28 Diagnosing congenital CMV is important: it provides an explanation for the SNHL, it alerts us to the possible progression of SNHL, neurodevelopmental, visual and vestibular complications, and allows parents to act to prevent further CMV infections.40
The sensitivity and specificity of the PCR assay used here were similar to those reported in the literature for qualitative assays.25 Although we were unable to artificially create a false-positive DBS result through cross-contamination in the laboratory, serology results helped identify a false positive DBS result in a single patient. Addition of a second CMV DNA primer set to the diagnostic algorithm is planned to see if this increases test specificity. It is important to remember that the viral load in DBS cannot be used interchangeably with CMV load in blood. Nevertheless, we have shown that the two are related so that both can provide important diagnostic/prognostic information. Evidence of previous exposure to CMV is important to document before requesting DBS testing, or considering antiviral therapy (fig 5).
The study is limited by retrospective data collection. However, this was the first time the DBS test has been used quantitatively in this population (to our knowledge), so we could not predict CMV DBS viral loads for children before invitation, and the audiovestibular physician was blind to these results until the study had finished.
Until congenital CMV is preventable by vaccination, or treatable antenatally, it will continue to cause significant childhood SNHL. Further work is needed to understand the mechanisms underlying late-onset and progressive SNHL, to determine the benefit of neonatal DBS CMV screening followed by audiological follow-up, and to ascertain whether CMV-related SNHL may be prevented by treatments reducing viral load in infancy.
What is already known on this topic
Congenital CMV is a significant cause of sensorineural hearing loss (SNHL) in children and the hearing loss occurs after the newborn hearing screen in up to half of cases.
CMV-related SNHL has been associated with increased viral load in urine and blood.
What this study adds
CMV-related SNHL may also be associated with increased viral load on “Guthrie” card dried blood spots (DBS).
DBS CMV testing may identify asymptomatic, congenitally infected neonates at risk of SNHL to enable audiological follow-up and early intervention, and may assist the aetiological investigation of SNHL.
We wish to thank all the families who kindly took part, Mrs Carmen Burton and Mrs Keri Stephenson of the Congenital CMV Association, local principal investigators: Dr Deirdre Lucas, Dr Soo Yoong, Dr Yogi Thakker, Mr Kevin Gibbin and Dr Tracey Davis, and also Mrs Jayne Ramirez Inscoe, Ms Rebecca Reeves-Jones, Dr Philip Marsh, Dr Antony Hale, participating consultants and specialist registrars, and the Leeds Virology Laboratory for their help, Jan Poloniecki for his statistical advice and Regional Infant Screening Centre Staff: Ms Julia Rider, Ms Deborah Lawrence, Dr Jeffrey Barron, Dr Mick Henderson, Mrs Susan Carter, Dr Jim Bonham, Ms Melanie Downing, Mr Wyn Griffiths, Dr Helena Kemp, Ms Kate Hall, Mr Ian Smith, Mrs Joan Mackenzie, Dr John Sherwin and Ms Heidi Lerner for retrieving the newborn screening cards. This work was submitted by SW as fulfilment of an MSc in audiovestibular medicine.
Funding: This study was not externally funded.
Competing interests: None.
Ethics approval: The study was approved by Wandsworth main research ethics committee.