Agreement Between Dried Blood Spot and Plasma PCR in Infants With Congenital Cytomegalovirus Infection

Authors: Masako Shimamura, Juhyeong Kim, Alexandra K. Medoro, Kaitlyn Flint, Irina Kaptsan, Huanyu Wang, Traci Pifer, Rachelle Harris, José Cuartas, Amy Leber, Pablo J. Sánchez

Abstract

Congenital cytomegalovirus (cCMV) infection is diagnosed by positive urine or saliva testing within 21 days after birth. Beyond this age, newborn dried blood spot (DBS) PCR can retrospectively diagnose cCMV infection but has lower sensitivity than urine or saliva PCR testing. The DBS PCR may be negative due to the absence of blood DNAemia at birth or to the technical limit of detection for DBS PCR. The objective of this study was to distinguish these two possibilities by determining agreement between DBS and plasma CMV PCR tests among cCMV‐infected infants. This single center retrospective cohort study evaluated 70 cCMV‐infected infants diagnosed by a positive urine CMV PCR, who had a CMV DBS at birth and a plasma PCR test within 31 days after birth. Clinical characteristics and viral loads were compared between groups according to paired DBS and plasma PCR results. Test agreement was calculated using Cohen's kappa coefficient. The DBS PCR sensitivity was 71% compared to urine PCR. Of the 70 subjects, 49 (70%) subjects were DBS+ /plasma+ , 1 (1.4%) were DBS+ /plasma−, 14 (20%) were DBS−/plasma+ , and 6 (9%) were DBS−/plasma−. Agreement between the tests was fair (κ = 0.348, 95% CI 0.115‐0.581). Of the 20 subjects with DBS− tests, 6 (30%) had undetectable plasma DNAemia. Of the infants with DBS−/plasma+ PCR, plasma viral loads were significantly lower than infants with DBS+ /plasma+ PCR testing. Nearly a third of cCMV infected infants may be missed by DBS testing due to both biological and technical limitations of this method.

Introduction

Congenital cytomegalovirus (cCMV) infection is the most common congenital viral infection in the U.S., affecting approximately 1 in 200 (0.5%) liveborn infants [1, 2]. Among cCMV infected newborns, 10–15% have symptomatic infection at birth, of whom approximately 40–58% will have permanent neurodevelopmental disabilities including sensorineural hearing loss (SNHL) [1, 2]. Additionally, among the 85–90% of infants who are asymptomatic at birth, 13.5% will have developmental delays and/or late onset SNHL [1]. The socioeconomic burden of cCMV infection is substantial [3, 4, 5]. Antiviral treatment can improve neurologic and hearing outcomes [6, 7], but asymptomatic and mildly symptomatic infants may not be identified at birth in the absence of universal cCMV screening programs.

The diagnosis of cCMV infection requires positive urine or saliva PCR testing or culture within the first 21 days after birth [8]. Beyond 21 days of age, a positive test may be due to postnatally acquired CMV infection, which is commonly transmitted through breastfeeding [9, 10]. Although universal newborn screening by urine or saliva PCR could identify cCMV infection within the necessary timeframe, universal screening is not routinely performed nationally in the US. Impediments to universal urine or saliva screening include cost, time, personnel training, potential difficulties obtaining bagged urine specimens, and necessity to establish protocols to follow‐up positive results for infants discharged before test reporting. As neonatal dried blood spots (DBS) are routinely obtained for state newborn screening programs, there is considerable interest in using DBS to screen for cCMV infection due to the feasibility of incorporating this sample type into existing state screening program workflows and reporting. Since February 2023, Minnesota became the first state in the U.S. to implement universal newborn screening for cCMV infection using DBS [11], followed by New York and several Canadian provinces. Additionally, neonatal DBS PCR can be used to identify cCMV infection retrospectively among children who are diagnosed with sensorineural hearing loss or neurologic sequelae beyond 21 days after birth.

Two studies directly compared CMV DBS PCR to saliva testing in large cohorts of healthy newborns (n = 20448 and 12554, respectively) [12, 13]. The DBS PCR in these two studies had widely differing sensitivities of 34.4% and 85.7% compared to rapid saliva culture or PCR, respectively. The disparity in DBS sensitivity between these two studies was postulated to be due in part to differences in sensitivity of the DBS extraction and PCR methods [13]. Consequently, it has been proposed that additional technical improvements in DBS test sensitivity could enable its use as a cCMV screening method. However, neither study tested blood from venipuncture for PCR in parallel with the DBS to evaluate the possibility that the negative DBS PCR could represent a true negative caused by the absence of CMV DNAemia in some cCMV infected newborns. Supporting this premise, several studies of infants with confirmed cCMV infection reported that 5–17% had a negative blood PCR test [14, 15, 16, 17]. The objective of this study was to distinguish technical limitations in the DBS method from true negative DBS (absence of DNAemia) by determining agreement between DBS and plasma PCR tests among infants with proven cCMV infection diagnosed by positive urine PCR within the first 21 days of age.

Materials and Methods

Patient Population

This single center retrospective cohort study evaluated infants referred to the Nationwide Children's Hospital (NCH) Neonatal‐Infectious Diseases Clinic (“Neo‐ID Clinic”) due to diagnosis of congenital CMV infection from 2016 through 2021. Infants were tested for cCMV infection by clinicians due to clinical symptoms; failed newborn hearing screen; screening saliva PCR on admission to NCH affiliated neonatal intensive care units (NICUs) per institutional protocol; or through maternal enrollment in an interventional clinical trial to prevent cCMV transmission during pregnancy (NCT01376778). Infants with a positive saliva or urine PCR test at age ≤ 21 days were referred to Neo‐ID Clinic for clinical management. For all referred infants, confirmatory urine and plasma PCR were ordered through the NCH laboratory and the newborn DBS was requested from the Ohio Department of Health with parental consent for CMV PCR testing. None of the enrolled infants received antiviral therapy before performance of plasma PCR testing. Subjects with confirmed cCMV infection (positive urine PCR at age ≤ 21 days) who had a plasma CMV PCR test performed within 31 days after birth and a DBS PCR test were included in the study. Subjects were excluded if urine was tested beyond 21 days of age, blood PCR was performed beyond age 31 days, or if a DBS was not available. This study was approved with waiver of informed consent by the NCH IRB (IRB17‐00410).

Data Collection

Clinical data were collected in a REDCap form (Vanderbilt University, Nashville TN [18]), including patient demographics, maternal and birth history, clinical symptoms at birth, laboratory test results including qualitative urine PCR and quantitative plasma PCR tests, neuroimaging, ophthalmologic and hearing evaluations.

Clinical Plasma Testing

Viral loads in urine and plasma were ordered clinically as part of patient care. Testing was performed in the clinical microbiology laboratory at NCH using published primer/probe sequences and PCR conditions as described [19]. Briefly, total nucleic acid was obtained by extracting 200 μL of plasma or 400 μL of urine using the NucliSENS easyMag platform (bioMerieux, Durham, NC). 5 μL of the eluate was added to a 20 μL total volume reaction mixture (1X TaqMan Universal master mix, Life Technologies, Grand Island NY; 0.20 μM of each primer and probe). Real‐time PCR was performed using the ABI 7500 thermocycler (Life Technologies, Grand Island, NY) with the following running 50°C for 2 min, denaturation at 95°C for 10 min and 45 cycles of 95°C for 15 s and 60°C for 1 min.

DBS Processing

DBS were stored at the Ohio Department of Health at −80°C and obtained for testing at NCH after written parental consent. For each DBS, six 2‐mm circles were manually punched into a microcentrifuge tube and DNA was extracted into 100 μL sterile water per manufacturer protocol (Qiagen DNA Investigator Kit #56504, Qiagen, Germantown MD). Unused, clean filter papers (Whatman 903 Protein Saver Cards, Cytiva, Marlborough MA) were punched and extracted in identical fashion as controls for the DBS. Quantitative CMV PCR was performed in duplicate using 5 μL per reaction of the DBS or control filter paper extract using conditions as described for plasma PCR. DBS PCR results were reported qualitatively as positive or negative. To confirm that the DNA extracted from each DBS was of sufficient quantity and quality for PCR amplification, each DBS DNA extract was tested by PCR using primers and probe specific for human zinc finger gene [20]. All DBS extracts tested PCR positive for zinc finger targets.

To determine the limit of detection for the DBS extracts, whole blood from a CMV seronegative adult was spiked with CMV strain TR [21] in 10‐fold serial dilutions with a range of 10^1^–10^5^ copies/ml. For each dilution, 50 μL of blood was spotted onto filter papers in duplicate, dried overnight, then extracted and processed as described to determine the DBS PCR detection limit.

Outcomes and Statistical Analysis

Descriptive analyses were used to summarize clinical characteristics among groups with concordant and discordant DBS and plasma PCR tests using medians with interquartile ranges and frequency distributions as appropriate. DBS sensitivity relative to urine PCR was calculated. Agreement between the plasma and DBS PCR tests was calculated using Cohen's kappa coefficient. Plasma viral loads were compared between groups with DBS+ or DBS− tests using the Mann‐Whitney test and further stratified according to postnatal age at testing and analyzed using the Kruskal‐Wallis test.

Results

Clinical Characteristics

Of 86 infants with DBS testing, eight infants were excluded due to urine PCR performed > 21 days after birth, 5 did not have plasma PCR testing, and three infants had plasma PCR performed > 31 days after birth. After exclusions, 70 cCMV infected infants were included in the study. Among this cohort (Table 1), 30 (43%) were female, 51 (73%) were white, 13 (19%) were black, 3 (4%) were Asian, 3 (4%) were multiracial, and 3 (4%) were Hispanic. Infants had a median gestational age of 38 weeks (interquartile range [IQR] 37–39 weeks), median birthweight 2531 g [IQR 2095–2935 g], median birth length 47.0 cm [IQR 44.5–49.5 cm], and median head circumference 32.0 cm [IQR 31.0–33.5 cm]. The asymptomatic group consisted of 16 infants (23%) who had no symptoms of congenital CMV infection at birth. Reasons for CMV testing of asymptomatic infants, including reasons for NICU admission, are shown in Table S1. Of the 54 infants (77%) in the symptomatic group, clinical symptoms at birth are shown in Table 2. Within the symptomatic group, 23 infants had non‐CNS manifestations, 27 had CNS abnormalities on physical examination or neuroimaging, and 18 had SNHL at birth.

DBS PCR Performance Characteristics

Based on the study definition for cCMV infection, 100% of infants in this cohort had a positive urine PCR test result. Of the 70 subjects, 50 had a positive DBS PCR, resulting in 71% sensitivity compared to urine PCR. All 50 of the positive DBS PCR tests detected amplification in both PCR well replicates for each sample, and all 20 of the negative DBS PCR tests had no amplification in either PCR well replicate. Comparing DBS and plasma PCR results, 49 (70%) had DBS+ /plasma+ PCR; 1 (1.4%) had DBS+ /plasma− PCR; 14 (20%) had DBS‐/plasma+ PCR; and 6 (9%) had DBS−/plasma− PCR (Table 3). Agreement between the tests was fair (κ = 0.348, 95% confidence interval, 0.115–0.581). Table 1 shows the distribution of sex, race, ethnicity, gestational age, birthweight, birth length, birth head circumference, and symptomatic infection for groups according to DBS/plasma PCR results.

The majority of subjects (n = 55, 79%) had concordant DBS and plasma PCR testing (+/+ or −/−), supporting that the DBS test reflected the presence or absence of CMV DNAemia in these subjects. One infant (1.4%) had a positive DBS PCR and negative blood PCR. This subject's plasma sample was obtained at 9 days after birth, which was below the median age at testing for the DBS+ /plasma+ group (median [IQR], 13 days [7.5–20.5]), indicating that the negative plasma test was not due to the subject's postnatal age at plasma testing. The DBS PCR was negative in 6 subjects (9%) due to undetectable plasma DNAemia (plasma PCR‐). Fourteen infants (20%) had a negative DBS PCR and positive plasma PCR. The plasma viral loads were compared between the DBS+ /plasma+ and DBS‐/plasma+ groups (Figure 1A). The DBS−/plasma+ group had significantly lower plasma viral loads (321 IU/ml [IQR 165–464]) than the DBS+ /plasma+ group (2245 IU/ml [IQR 761–5773]) (p < 0.0001). The median postnatal age at plasma testing was similar between the DBS+ /plasma+ and DBS−/plasma+ groups, indicating that the difference in viral loads between these groups was not due to a difference in the postnatal age at plasma testing (Figure 1B). Viral loads within each group (DBS+ /plasma+ or DBS−/plasma+) were further stratified according to postnatal age at time of plasma testing (Table S2), which confirmed that viral loads were statistically similar across postnatal ages within each group. Within the DBS+ /plasma+ and DBS−/plasma+ groups, 10 of 49 subjects and 5 of 14 subjects, respectively, were asymptomatic (p = 0.27, not significant). The DBS PCR limit of detection for exogenously spiked dried blood spots was 10^2^–10^3^ copies/ml of whole blood. All “control” extracts from blank filter papers tested negative by CMV PCR.

Discussion

In this study, there was fair agreement (κ) between DBS and plasma CMV PCR. Our DBS sensitivity of 71% is lower than the rate of 86% reported by Dollard et al. [13] but higher than that reported in the CHIMES study (34%) [12]. Some of these differences could be due to differences in the “gold standard” used to calculate sensitivity. Our study compared the DBS to urine PCR, whereas the Dollard group compared DBS to saliva PCR, which found than saliva testing had both false‐positive and false‐negative results compared to urine PCR (13.3% and 6.7%, respectively). Differences between our study and CHIMES included DBS comparison to urine PCR versus rapid saliva culture, our use of a slightly larger DBS surface area per extraction (18.84 vs. 14.13 mm^2^), manual versus automated extraction, and different PCR primers and cycling conditions [12, 22, 23]. An additional key difference is that the CHIMES study evaluated test performance by screening a large, unselected population of healthy newborns. Table 4, section A, summarizes six studies that used DBS PCR to screen newborn populations for cCMV infection and found that the DBS was positive for 0.12–0.48% of newborns [11, 12, 13, 24, 25, 26]. In contrast, 6 studies testing DBS in children with known cCMV infection (Table 2, section C) found that 6.7–86% of subjects had a positive DBS [12, 13, 15, 27, 29, 30]. Our study more closely resembled the latter group by evaluating DBS among a cohort who underwent targeted testing indicated by specific clinical characteristics or NICU admission. In our cohort, 77% had symptomatic infection and 39% had CNS involvement. Studies have shown that blood viral loads are generally higher in infants with symptomatic than asymptomatic infection [42, 43, 44] and those with CNS involvement [28], so it is possible that the high DBS sensitivity in our study may simply be due to the high viral loads present in this symptomatic cohort.

Our results indicate that DBS PCR is more often positive among children with higher viral loads but may be negative due to low or absent DNAemia. A small study of four cCMV infected infants found that all four had positive plasma PCR, but only two had positive DBS PCR [27]. More recently, in a multicenter study from Spain, DBS PCR was positive in 58 of 103 cCMV infected infants (56%), blood PCR was positive in 82 of 95 infants (86%), and a negative DBS was associated with low blood viral load [15]. Inclusion criteria for this study differed from ours in that cCMV infection was diagnosed via positive CMV PCR or culture from any body fluid (urine, blood, cerebrospinal fluid) within the first 2 weeks after birth, and the blood PCR tests were performed in various clinical laboratories that may have differed from the PCR assay used for DBS testing.

Other studies have reported blood viral loads without concomitant DBS testing (Table 4, Section B). In one study of 20 cCMV infected neonates with positive CMV urine culture, 19 (95%) had a positive serum PCR, with 1 (5%) negative [14]. In another study of 18 infants with cCMV infection, 15 had positive blood PCR but three (17%) were negative [16]. In a larger cohort of 256 infants, 147 (92.5%) were positive at diagnosis but 12 (7.5%) had negative blood viral loads [17]. These studies show that up to17% of cCMV infected infants lack CMV DNAemia, with our results (10%) in agreement within this range. Table 4, sections B and C, summarizes studies of blood or DBS PCR testing for cCMV infected infants, with blood PCR being positive for 83–100% of cCMV infected infants, and DBS PCR having lower positivity rates ranging from 6.7% to 86%. Together, these data indicate that the DBS method may fail to identify all cCMV infected infants, as some may lack DNAemia at birth.

In other studies, test performance of Guthrie cards spiked with blood of known viral loads and various extraction/amplification methods yielded inconsistently positive DBS PCR results at blood viral loads between 2 and 4 log10 copies/ml [22, 45]. These findings align with our results showing that DBS is less often positive than blood samples, and that the DBS may be negative among infants with low‐level DNAemia.

Targeted CMV DBS PCR has been used to identify cCMV infection among neonates who fail the newborn hearing screening (NBHS), but this approach misses many cCMV infected infants who pass the NBHS at birth [46]. Neonatal urine and saliva samples are rarely available retrospectively to confirm cCMV infection among children who develop SNHL beyond the newborn period. Some investigators have sought to diagnose cCMV infection in children with late‐onset SNHL by PCR testing of dried blood spots that were collected at birth for state newborn screening programs, with positive results supporting cCMV infection [29, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41]. In these studies, 2.7–34% of subjects with SNHL were DBS positive, which confirmed cCMV infection in these children. These studies are summarized in Table 4, section D. DBS was also tested among a cohort of 1388 infants with symptomatic cCMV infection or at risk for cCMV infection, of whom 7.5% had a positive DBS PCR [30]. In summary, DBS PCR can be used to make a retrospective diagnosis of cCMV infection in children with SNHL or other symptoms of cCMV infection, but a negative test does not rule out cCMV infection.

One subject in our cohort had a positive DBS with negative blood PCR. It is unclear whether this was due to a false‐positive DBS PCR test, or if DNAemia was present at birth but cleared before plasma testing at age 9 days. A false‐positive DBS could occur via false‐positive PCR amplification, amplification due to laboratory contamination, or possibly from CMV cross‐contamination from an uninfected infant's DBS card being stored adjacent to the DBS card of a cCMV infected infant [47, 48]. The likelihood of false‐positive amplification or laboratory contamination of this subject's DBS specimen is low, since the “control” blank filter paper processed in parallel tested negative by PCR. Potential contamination from an adjacent dried blood spot during storage at the Ohio Department of Health cannot be excluded.

Limitations of this study include a relatively small sample size derived from targeted testing and consisting predominantly of symptomatic infants. Analysis was also limited to infants who had proven cCMV infection within the first 21 days of age, as well as those who had plasma PCR testing and DBS samples available for analysis. Results from this cohort could therefore differ from a general, unselected population being screened for cCMV infection. It is also possible that some infants with negative plasma PCR testing may have had DNAemia at the time of delivery but cleared within the first 31 days of age. However, the likelihood of this is low, given that other studies have shown that CMV DNAemia persists longitudinally for months to years after congenital infection [17, 49]. In our study, blood CMV PCR amplification might also differ between whole blood (DBS) and plasma compartments due to the presence of leukocytes containing CMV DNA in whole blood. However, a recent study showed correlation between whole blood and plasma CMV PCR testing from paired samples (p = 0.631) among cCMV infected infants [50].

In summary, this study showed fair agreement between DBS and plasma PCR in cCMV infected infants. DBS was negative for 29% of infants with low or absent CMV DNAemia, raising concern for use of DBS PCR as a large‐scale cCMV screening method. Efforts may be better directed toward operationalizing the use of urine or saliva PCR for universal cCMV screening [12, 51].

Author Contributions

Conceptualization and design of the Masako Shimamura, Juhyeong Kim. Data acquisition, analysis, Masako Shimamura, Juhyeong Kim, Alexandra K. Medoro, Kaitlyn Flint, Irina Kaptsan, Huanyu Wang, Traci Pifer, Rachelle Harris, José Cuartas, Amy Leber, Pablo J. Sánchez. Drafting and critically revising article for important intellectual Masako Shimamura, Juhyeong Kim, Alexandra K. Medoro, Kaitlyn Flint, Huanyu Wang, Pablo J. Sánchez. Final approval of the all authors.

Ethics Statement

This study was approved by the Nationwide Children's Hospital Institutional Review Board with waiver of patient consent for analysis of clinical data and remnant clinical specimens (NCH IRB17‐00410).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information