Dried Blood Spot Real-time Polymerase Chain Reaction Assays to Screen Newborns for Congenital Cytomegalovirus Infection FREE

Cytomegalovirus (CMV) is an important cause of congenital infection and a leading cause of sensorineural hearing loss (SNHL) in children.^1- 5 Of the estimated 20 000 to 40 000 infants born each year with congenital CMV infection in the United States, most (90% to 95%) have no detectable clinical abnormalities at birth and thus will not be identified by routine clinical examination.^2,6- 7 Furthermore, SNHL occurs in approximately 10% to 15% of infants with clinically inapparent congenital CMV infection and the majority of children with CMV-related SNHL will have late-onset losses, progressive losses, or both.^1,8- 9 Therefore, both routine physical examination and newborn hearing screening will miss potential diagnosis in many children who develop SNHL secondary to congenital CMV infection. To identify these at-risk infants early in life, rapid, reliable, and relatively inexpensive methods to screen newborns for congenital CMV infection are needed.¹⁰ Identification of children at increased risk of CMV-associated SNHL early in life will allow targeted monitoring of these children in order to intervene at critical stages of acquiring speech and language skills.¹¹

Although traditional virus isolation from saliva or urine specimens in tissue culture is considered the standard method for identification of infants with congenital CMV infection, it is not amenable to mass screening (even when modified to produce rapid results) because it is labor- and resource-intensive and requires tissue culture facilities. Real-time polymerase chain reaction (PCR) technology, in contrast, is well-suited for mass screening because it does not require tissue culture facilities and is amenable to automation with the screening of large numbers of specimens at low cost. A variety of newborn specimens including saliva, urine, and dried blood spots (DBS) can be tested with PCR-based methods for the diagnosis of congenital CMV infection.^12- 19 Since DBS are collected routinely for newborn metabolic screenings from all infants born in the United States, there has been considerable interest in using PCR assays for detecting CMV in newborn DBS specimens. Despite the benefits of DBS PCR-based methods, the sensitivity and specificity of these assays for universal newborn CMV screening have not been determined. Most reports have studied selected infant populations and none have prospectively compared the results of a DBS PCR assay with a standard (ie, tissue culture) method for identifying CMV infection in an unselected newborn population.^{13- 16,20- 22} This study examined the diagnostic accuracy of real-time PCR analysis of DBS as an approach for mass screening of newborns for congenital CMV infection.

Between March 2007 and May 2008, infants born at 7 US medical centers (University of Alabama at Birmingham Hospital; The University of Mississippi Medical Center, Jackson; Carolinas Medical Center, Charlotte, North Carolina; Saint Peter's University Hospital, New Brunswick, New Jersey; Good Samaritan Hospital, Cincinnati, Ohio; Magee Women's Hospital, Pittsburgh, Pennsylvania; and Parkland Memorial Hospital, Dallas, Texas) were enrolled prospectively in the National Institute on Deafness and Other Communication Disorders CMV and Hearing Multicenter Screening (CHIMES) study. Institutional review board approval was obtained at each site. Mothers were approached postpartum to obtain written informed consent for their newborn's enrollment in the study. Upon enrollment, saliva specimens were collected from participating infants along with an additional blood spot obtained at the time of newborn metabolic screening. The DBS specimen for the study was collected only after the completion of metabolic screening and infants were not subjected to additional heel sticks for the CHIMES study. Infants with positive saliva or DBS screening specimens were enrolled in the follow-up component of the study to confirm congenital CMV infection, as well as evaluate hearing outcomes during the first 4 years of life (ongoing). Race and ethnicity data were collected as self-reported by parents because the prevalence of congenital CMV infection has been shown to vary according to racial and ethnic composition of the delivery population.^23- 24

Saliva specimens were collected from the enrolled newborns at a mean (SD) age of 0.9 (0.6) days and before nursery discharge. Collection was made by swabbing inside the infant's mouth using a sterile polyester fiber-tipped applicator (PurFybr Inc, Munster, Indiana) and placed in 1.0 mL of transport medium containing sucrose phosphate.²⁵ The specimens were stored at 4°C until they were transported, on ice, within 1 week of collection. A temperature-monitoring device was included in shipments to monitor for temperature variation during transport (TL20, 3M, St Paul, Minnesota).

DBS specimens were collected at the time of newborn metabolic screening and the mean (SD) age at collection was 1.9 (1.8) days. The additional blood spots were collected on a separate filter paper (Whatman 903, Florham Park, New Jersey), placed in individual envelopes, and stored in plastic resealable bags containing desiccant. DBS specimens were maintained at room temperature and shipped once weekly. Saliva and DBS specimens were transported to the University of Alabama at Birmingham central laboratory.

The mean (SD) interval between the collection of initial saliva specimens and testing at the University of Alabama at Birmingham central laboratory was 7.4 (4.0) days. The presence of CMV in saliva specimens was detected by a rapid culture method for detecting early antigen fluorescent foci using a monoclonal antibody against the CMV major immediate early antigen in duplicate wells of a 96-well microtiter plate.^25- 26 Each run included 2 positive control wells inoculated with the AD169 strain of CMV at a titer producing approximately 100 infectious foci per well. A specimen was considered positive if at least 1 focus of distinct nuclear fluorescence was detected in at least 1 well. Individuals ascertaining the results of the saliva rapid culture assay or the DBS PCR were blinded to the results of the other test.

From each DBS, two 3-mm disks were punched into 1.5-mL sample tubes using the BSD600 automated filter paper puncher (BSD Robotics, Acacia Ridge, Queensland, Australia). The punched filter paper disks were processed to extract DNA using the Qiagen M48 robotic system with MagAttract technology according to the manufacturer's instructions (Qiagen Inc, Valencia, California). The extracted DNA specimens were stored at −20^○C. A blank filter card was punched and included in each extraction run to serve as a negative control for DNA extraction and to monitor for cross contamination. In addition, a filter paper spotted with 10 000 copies of AD169 strain of CMV was punched and included in the extraction run to serve as a positive control and to monitor for consistency and reliability of the extraction protocol.

The mean (SD) interval between DBS specimen collection and PCR analysis was 14.6 (9.6) days. The detection of CMV DNA was performed using the ABI 7500 Real-time PCR System (Applied Biosystems Inc, Foster City, California) and ABsolute QPCR Low ROX Mix (ABgene USA, Rockford, Illinois). The reaction mixture contained primers at a concentration of 900 nM and the probe at 250 nM. Each specimen was run in duplicate using 25 μL of reaction mixture containing 20 μL of master mix and 5 μL of test specimen. To generate standard curves, each plate contained plasmid standards incorporating the target sequences in 10-fold dilutions ranging between 100 000 and 10 genomic equivalents per reaction. The real-time PCR amplification conditions have been previously described.^27- 28 During the first 10 months of the study, the real-time PCR assay included primers to detect the highly conserved AD-1 region of the major envelope glycoprotein B.^27- 29 During the final 5 months of the study, the PCR method was modified to include a second primer set from the highly conserved immediate early 2 exon 5 region (forward primer, GAG CCC GAC TTT ACC ATC CA; reverse primer, CAG CCG GCG GTA TCG A; and probe, VIC-ACC GCA ACA AGA TT-MGBNFQ) in an effort to improve the sensitivity of the assay (GeneBank accession numbers GU179001, AY446871, AY446870, FJ616285, AY446868). The real-time PCR was repeated on all specimens with a positive signal in either well and a specimen was considered positive if 1 or more genomic equivalents per reaction were detected on both PCR runs. In addition, real-time PCR was repeated on DBS specimens from infants with saliva specimens positive by rapid culture assay that were negative on the first PCR run. The detection limit of our real-time PCR assay, as determined by the sensitivity titration analysis, was 250 genomic equivalents per milliliter for the single-primer assay and 50 genomic equivalents per milliliter for the 2-primer assay (eAppendix).

To determine whether the sensitivity of DBS real-time PCR for CMV DNA detection was influenced by the extraction method, detection of CMV DNA by the 2-primer real-time PCR protocol was compared between a commercial column-extraction method (Qiagen Inc, Valencia, California) and the robot-extraction protocol used in this study (eAppendix). In addition, the amount of genomic DNA as determined by real-time PCR amplification of RNase P (TaqMan RNase P control reagents kit, Applied Biosystems Inc, Foster City, California) in 185 randomly selected DBS specimens from CMV-negative infants was compared between the robot- and column-extraction methods (eAppendix). A comparison of the 2-primer real-time PCR assay and a previously described nested PCR protocol was undertaken to assess whether our real-time PCR method would be as sensitive or more sensitive for detecting CMV DNA than a standard nested-PCR method (eAppendix).¹³

To account for the possibility that saliva rapid culture assay may be less than 100% sensitive in identifying CMV-infected newborns, infants with positive saliva specimens or DBS screening specimens were enrolled in the follow-up component of the CHIMES study to confirm congenital CMV infection.²⁵ Urine and repeat saliva specimens were obtained from these infants at the enrollment visit for the follow-up study and were tested for CMV with the rapid culture assay (previously described). The rapid culture assay on the follow-up saliva or urine specimen was considered the reference standard for this study and therefore, a confirmed congenital CMV infection was defined as identification of CMV in either saliva or urine obtained at enrollment into the follow-up study. Infants were considered to be uninfected if both the saliva and the urine specimens tested negative by rapid culture assay. Newborns who were negative for CMV by both screening assays (saliva rapid culture and DBS PCR) were not enrolled in follow-up and not retested with the reference standard assay.

Only infants enrolled in the follow-up component of the study for confirmation of congenital CMV infection status were included in determining the diagnostic ability of the DBS real-time PCR assays. Sensitivity, specificity, and predictive values for both the single-primer and the 2-primer DBS real-time PCR assays were calculated using standard methods for proportions and exact 95% confidence limits. The positive predictive value was the ratio of true positives to all positive DBS PCR results and the negative predictive value was the ratio of true negatives to all negative DBS test results. Likelihood ratios (LRs) were calculated to summarize the diagnostic accuracy of the DBS PCR assays. Positive LR was sensitivity/(1−specificity) and the negative LR was (1−sensitivity)/specificity. Confidence intervals (CIs) for LRs were determined using the method described by Simel et al.³⁰ Statistical differences between nested and real-time PCR methods were calculated using the χ² test. All statistical analyses were performed using SAS software version 9.2 (SAS Institute Inc, Cary, North Carolina).

Of the 36 130 eligible infants, 22 758 (63%) infants were enrolled in the study. Although all live-born infants were eligible for participation, some of the infants born over holidays or weekends and those discharged prior to obtaining consent for participation in the study (n = 10 876) were not enrolled. Additional reasons for nonenrollment included refusal to participate (n = 1359); unable to obtain consent due to maternal factors such as illness, mental capacity, age, or language (n = 677); and infant death or illness (n = 460).

Both saliva and DBS specimens were collected from 20 613 (91%) infants, only saliva specimens were collected from 1837 infants, only DBS specimens were collected from 262 infants, and 46 infants had neither specimen collected (Figure). The reasons that both specimens were not available from these newborns included (1) the infants were unavailable or discharged from the nursery prior to collection (n = 1214); (2) the newborn metabolic screening was completed before infants were enrolled in the study or there was insufficient blood left for the study DBS specimen (n = 731); or (3) the specimens were mislabeled or misplaced (n = 200). The infants (n = 2145) who did not have both specimens collected were more likely to be in the neonatal intensive care unit than infants who had both specimens collected (14.7% vs 2.9%; χ² test = 707.2; P < .001). Of the 20 613 infants who had both specimens collected, saliva specimens from 165 infants could not be tested due to leakage or temperature variations during shipment (Figure). Thus, the study population comprises the 20 448 infants who had both saliva and DBS specimens collected and tested.

Most of the study infants (19 858 [97.1%]) were from the well-baby nurseries (Table 1). The infants were evenly distributed by sex (male, 51.0% vs female, 49.0%). Mean (SD) maternal age was 27.3 (6.1) years. The mean (SD) age at enrollment into the follow-up study for confirmation of congenital CMV infection in infants positive by screening saliva rapid culture or DBS PCR was 6.4 (6.1) weeks of age. Overall, 92 of the 20 448 (0.45%; 95% CI, 0.36%-0.55%) infants had confirmed congenital CMV infection.

Between March 2007 and December 2007, 11 422 newborns were screened for congenital CMV infection using saliva rapid culture and the single-primer DBS PCR assay (Figure). Eighty-one newborns tested positive for CMV infection by either saliva rapid culture assay (n = 71), the DBS PCR assay (n = 26), or both methods (n = 16). Sixty-six of the 81 infants (81%) who tested positive by either screening method were enrolled in the follow-up study and of those, 60 children were confirmed to have congenital CMV infection based on the positive reference standard assay. Congenital CMV infection status could not be determined in 15 infants because they were not enrolled in the follow-up study. Reasons for not enrolling in the follow-up study included refusing participation (n = 8), loss to follow-up (n = 6), and relocation (n = 1).

Screening saliva rapid culture correctly identified 59 of the 60 infants (98%) with confirmed congenital CMV infection, whereas the single-primer DBS PCR only identified 17 of the 60 infants (28%) confirmed to have congenital CMV infection (Table 2). Congenital CMV infection was not confirmed in 2 of 61 infants (3%) with saliva specimens positive by rapid culture assay and in 4 of 21 infants (19%) who were DBS PCR-positive because of the negative reference standard assay. The sensitivity and specificity of the single-primer DBS PCR assay in identifying infants with confirmed congenital CMV infection were 28.3% (95% CI, 17.4%-41.4%) and 99.9% (95% CI, 99.9%-100%), respectively. The positive LR for the single-primer DBS PCR assay was 803.7 (95% CI, 278.7-2317.9) and the negative LR was 0.7 (95% CI, 0.6-0.8). The positive predictive value of the single-primer PCR assay was 80.9% (95% CI, 58.1%-94.5%) and the negative predictive value was 99.6% (95% CI, 99.5%-99.7%).

During the study period between January 2008 and May 2008, there were 9026 newborns screened for congenital CMV infection using saliva rapid culture and the 2-primer DBS PCR assay (Figure). Forty-three newborns tested positive for CMV infection by either saliva rapid culture assay (n = 43) or the DBS PCR assay (n = 14). Thirty-five of the 43 infants (81%) who screened positive were enrolled in the follow-up study and of those, 32 children were confirmed to have congenital CMV infection based on a positive reference standard assay (Figure). Congenital infection status could not be determined in 8 infants since they did not enroll in the follow-up study. Reasons for not enrolling in the follow-up study included refusing participation (n = 4), loss to follow-up (n = 2), death (n = 1), and relocation (n = 1).

Screening saliva rapid culture correctly identified all 32 infants (100%) who were confirmed to have congenital CMV infection, whereas the 2-primer DBS PCR identified only 11 of the 32 infants (34%) confirmed to have congenital CMV infection (Table 2). Congenital CMV infection was not confirmed in 3 of 35 infants with saliva rapid culture (8%) and 1 of 12 screening DBS PCR-positive infants (8%) because the reference standard assay was negative. The sensitivity and specificity of the 2-primer DBS PCR assay for detecting infants with confirmed congenital CMV infection were 34.4% (95% CI, 18.6%-53.2%) and 99.9% (95% CI, 99.9%-100%), respectively. The positive LR for the 2-primer DBS PCR assay was 3088.9 (95% CI, 410.8-23 226.7) and the negative LR was 0.7 (95% CI, 0.5-0.8). The positive predictive value of the 2-primer assay was calculated to be 91.7% (95% CI, 61.5%-99.8%) and the negative predictive value was 99.8% (95% CI, 99.6%-99.9%).

Of the 71 DBS specimens from infants with positive saliva specimens, 29 robot-extracted specimens (41%) were positive for CMV DNA, whereas only 19 column-extracted specimens (29%) were positive (χ² test, 3.14; P = .08) (eTable 1). In addition, in 185 randomly selected DBS specimens from infants testing negative for CMV, the mean (SD) amount of genomic DNA obtained using robotic extraction (0.86 [0.46] μg/mL) and using a commercial column kit (0.78 [0.44] μg/mL) was similar (t [368] = −1.58; P = .11) as measured by amplifying the RNase P gene (TaqMan Gene Expression Assays Protocol, PN 4333458) (eAppendix).

In 86 infants with confirmed congenital CMV infection, 40 (47%) were positive on the 2-primer PCR and 30 (35%) were positive by the nested PCR assay (χ² test = 2.41; P = .12). Both methods failed to identify 48% (41/86) who were confirmed CMV-positive (eTable 2).

This study demonstrates that real-time PCR analysis of DBS has low sensitivity for correctly identifying infants with congenital CMV infection. These results have major public health implications because they indicate that such methods, as currently performed, will not be suitable for the mass screening of newborns for congenital CMV infection—the most common nongenetic cause of deafness in the United States. Our data indicate that as many as 80% of infants with congenital CMV infections could be missed, even when using 2-primer DBS real-time PCR assays. The high positive LRs for the single-primer and the 2-primer PCR assays provide strong evidence that a positive DBS PCR result using these assays will identify infants with congenital CMV infection. However, the negative LRs for both PCR assays are not sufficiently small enough to rule out congenital CMV infection in newborns with a negative DBS PCR result.

PCR testing of peripheral blood has been widely used as a standard diagnostic method to detect invasive CMV infections in immunocompromised individuals including allograft recipients and patients with AIDS.^31- 32 These results, together with those of several studies that reported successful identification of infants with congenital CMV infection by DBS PCR, has led to anticipation that DBS PCR methods would become valuable tools in newborn CMV screening.^{13- 16,20- 22} However, the pathogenesis of congenital CMV infection is likely to be different from that in immunocompromised hosts. Immunocompromised patients usually experience acute CMV infection or symptomatic reactivation shortly before blood CMV PCR testing, whereas congenitally infected infants may have acquired CMV infection months before birth and thus are no longer viremic when tested as newborns.

This study, in which the 2 DBS real-time PCR assays were directly and prospectively compared with a reference standard for identification of infants with congenital CMV infection, provides important test measures of the use of DBS PCR. Several previous reports have demonstrated that newborns with congenital CMV infection can be identified with varying degrees of success by testing DBS using different PCR methods.^{13,16,33- 34} However, the prospective studies that confirmed CMV infection after identifying CMV DNA in DBS did not determine the number of false negatives (infants with congenital CMV infection who tested negative on DBS PCR). Having the complete denominator, as provided by this study, is essential to determine the use of DBS PCR for newborn CMV screening.

The low sensitivity of the DBS PCR method could possibly be explained by several factors: (1) the method used for DNA extraction; (2) the real-time PCR techniques; or (3) the possibility that not all infants with congenital CMV infection have detectable CMV DNA in their blood at birth. To evaluate extraction methods, we compared the ability of the 2-primer DBS real-time PCR to detect CMV DNA in DBS specimens processed with the robot-extraction protocol used in this study and the column-extraction method and found no difference.

A number of amplification methods including qualitative, quantitative, and real-time PCR protocols with different primers, probes, and cycling parameters have been reported with varying performance characteristics.^{12- 15,21- 22,34- 35} The single-primer real-time PCR assay used in this study was developed in the University of Alabama at Birmingham central laboratory.^27- 28 A number of newborn CMV screening studies in which DBS specimens were tested using a nested PCR protocol report a sensitivity of the DBS PCR assay approaching 100% in some populations.^13,20- 21 However, these studies did not include a direct comparison of the DBS PCR results with a standard culture-based assay. Further, in a more recent study in which different laboratories were given similar sample panels, the sensitivity of the CMV PCR method has been shown to vary from laboratory to laboratory.³⁶ A comparison of our 2-primer real-time PCR assay with a nested PCR protocol demonstrated that the 2-primer PCR had a higher sensitivity than the nested PCR but neither method identified most of the infants with congenital CMV infection.

Previous studies observed that some infants with clinically apparent or symptomatic congenital CMV infection had no detectable CMV DNA in whole-blood specimens obtained during the neonatal period.^27- 28,37 These findings argue strongly that the low sensitivity of our DBS PCR methods is most likely not due to our assay performance, but to the absence of detectable CMV DNA in the peripheral blood of some newborns with congenital CMV infection. Since about 10% to 15% of infants with asymptomatic or clinically inapparent congenital CMV infection develop hearing loss, it is critical that an ideal CMV screening method identify most newborns with asymptomatic congenital CMV infection.

A limitation of our study is that the 20 324 infants who had negative results on both screening assays, saliva rapid culture and DBS PCR, did not have urine and repeat saliva specimens collected and tested with the rapid culture, resulting in the possibility that some CMV-infected newborns may have been missed by the saliva rapid culture. However, it is unlikely that the screening saliva rapid culture missed significant numbers of infants with congenital CMV infections. The saliva rapid culture assay used in our study was adapted from the shell vial assay, which has been shown to be as sensitive and specific as the conventional tube culture method and, thus, considered a standard method for the diagnosis of CMV infections in a variety of clinical settings.^38- 39 In addition, the saliva rapid culture assay we used has been demonstrated to be at least 98% sensitive in identifying infants with congenital CMV infection.²⁵ Finally, the results of our study showed that 99% (91 of 92) of infants with confirmed congenital CMV infection were identified on screening saliva rapid culture assay.

Another possible limitation is the relative overrepresentation of African Americans in our study population, which could make the findings of this study less generalizable to other populations. Although African American infants have a greater risk of infection, there is no scientific evidence that the clinical course or the sensitivity of diagnostic assays differs by race or ethnicity.^23- 24 However, the overrepresentation of African Americans may have influenced the prevalence of congenital CMV infection in our study. For populations with differing prevalences of congenital CMV infection than we found in this study, the predictive values calculated for the DBS PCR assays would not be appropriate since predictive values are dependent on the underlying prevalence of disease in the population.

In summary, the results of this large, prospective newborn CMV screening study that included a direct comparison of the DBS real-time PCR assays with the culture-based method on saliva specimens demonstrated that real-time DBS PCR assays are not suitable for screening newborns for congenital CMV infection since they miss approximately two-thirds of the infections. As the disease burden from congenital CMV infection remains a significant public health problem, there continues to be a need to identify the large number of infants with clinically inapparent congenital CMV infection early in life. The results of our study underscore the need for further evaluation of high-throughput methods performed on saliva or other specimens that can be adapted to large-scale newborn CMV screening.