Evaluation of Cytomegalovirus (CMV) DNA Quantification in Dried Blood Spots: Retrospective Study of CMV Congenital Infection

Abstract

We compared two protocols for extracting DNA from dried blood spots for cytomegalovirus (CMV) DNA detection and quantification by real-time PCR. Both extraction methods were reliable for the retrospective diagnosis of CMV congenital infection. Quantification of CMV DNA was valuable after normalization of viral loads with albumin gene PCR amplification results.

Human cytomegalovirus (CMV) is the main organism responsible for congenital infection and permanent deafness in young children in developing countries (13). Policies for screening during pregnancy and at birth have not been implemented in European countries or in the United States, essentially because there is no well-established treatment for pregnant women or for newborns with CMV infection (7). Retrospective diagnosis of congenital infection has been achieved by PCR detection of the CMV DNA in dried blood spots (DBS) stored on perinatal Guthrie cards (2, 5, 6, 8, 9, 14, 16, 17). Only one protocol (heat DNA extraction, followed by nested PCR) has been extensively evaluated in a clinical setting, with excellent sensitivity and specificity compared to that of viral isolation in the urine (1, 2). However, lower sensitivities (63 and 71%) were reported using the same protocol (5, 16). Alternative methods based on either phenol-chloroform or silica extraction protocols have been proposed, but their sensitivities (81% and 100%, respectively) were studied only with small numbers of patients (9, 14).

In the current study, we compared two DNA extraction protocols (phenol-chloroform versus silica-based technology) followed by quantitative in-house real-time CMV DNA-specific PCR amplification (12). Indeed, knowing the CMV DNA load in DBS could provide unique insights regarding the pathogenesis and outcome of CMV congenital infection, especially the relationship between viral load and the risk of hearing loss (4, 10).

DNA extraction using a whole DBS cut into thin strips with single-use scissors was performed using two protocols. In protocol one, the strips were submerged twice in 1.5 ml of washing solution (10 mM NaCl, 10 mM EDTA) for 30 min at room temperature. Then, 150 μl of lysis buffer (0.32% NaOH) was added onto the strips, and the lysate was recovered after centrifugation (at 10,000 × g for 2 min) and supplemented with 30 μl of neutralization solution (1 M Tris, pH 7.5) before DNA extraction with a QIAamp DNA blood mini-kit (QIAGEN, Courtaboeuf, France). Protocol two was performed as described previously (15). Briefly, the strips were submerged in 400 μl of extraction buffer and incubated at 56°C for 1 h. The supernatant was recovered after centrifugation and purified by phenol-chloroform extraction, followed by ethanol precipitation.

CMV DNA-specific PCR amplification and human albumin PCR amplification were carried out with in-house real-time PCR assays in duplicate (11, 12). The normalized value of the CMV DNA load was expressed as the number of CMV genome copies per 10⁵ cells.

The 45% and 95% sensitivity values of the assays were calculated with a nonlinear regression sigmoidal model (Graph Pad). Nonparametric Spearman correlation coefficients were used to assess the association between continuous variables. Median copy numbers obtained from DBS from different groups of patients were compared by the Mann-Whitney U test. A P value of <0.05 was accepted as statistically significant.

For sensitivity assessment, a prequantified CMV DNA-positive whole-blood sample obtained from an infected patient who had undergone transplantation was diluted in CMV-negative whole blood at concentrations of 10⁵, 10⁴, 5 × 10³, 10³, 5 × 10², 10², and 10 genome copies/ml. Fifty μl (10 spots for each of the three highest concentrations and 20 spots for each of the other four dilutions) were applied to Guthrie cards (903 specimen collection paper; Whatman, Maidstone, England). With extraction protocol one, the CMV DNA-specific PCR was positive for 100% (15/15) of the spots loaded with the three highest concentrations of genome copies/ml; positive for 80% (8/10), 50% (5/10), and 10% (1/10) of the spots loaded with, respectively, 10³, 5 × 10², and 10² genome copies/ml; and positive for 0% (0/10) of the spots loaded with 10 genome copies/ml. The 45% sensitivity of the assay was 421 (2.6 log₁₀) genome copies/ml, and the 95% sensitivity was 4,000 (3.6 log₁₀) genome copies/ml. With extraction protocol two, CMV DNA-specific PCR was positive for 100% (15/15) of the spots loaded with the three highest concentrations, positive for 90% (9/10), 70% (7/10), and 50% (5/10) of the spots loaded with, respectively, 10³, 5 × 10², and 10² genome copies/ml, and positive for 0% (0/10) of the spots loaded with 10 genome copies/ml. The 45% sensitivity of the assay was 189 (2.28 log₁₀) genome copies/ml, and the 95% sensitivity was 2,000 (3.3 log₁₀) genome copies/ml.

Guthrie cards of 76 children were analyzed. Fourteen were from neonates who had CMV congenital infection (proven by PCR or found to be culture positive in a urine sample obtained at birth), of which 8 were asymptomatic at birth and 6 were symptomatic; 20 were from noninfected neonates (proven by PCR or who were culture negative according to a urine sample obtained at birth); and 42 were from children (age 1 to 72 months) diagnosed with hearing loss (n = 35) or with other symptoms compatible with CMV congenital infection (n = 7). All Guthrie cards collected on the third day of life were retrieved from the Regional Screening Laboratories after obtaining parents' consent. Similar results were obtained with both protocols (Table 1): albumin gene amplification was positive for all extracts, and CMV DNA was detected in 100% (14/14) of the DBS from CMV-infected neonates (patients A1 to A14), in 0% (0/20) of the DBS from noninfected neonates, in 34% (12/35) of the DBS from children with hearing loss (patients B1 to B12), and in 28% (2/7) of the DBS from children with other symptoms (patients C1 and C2). With protocol one, the PCR was positive for only one duplicate for two DBS: in case A7, DNA extraction efficiency was poor (low albumin load), and in case A14, the CMV viral load was very low (16 [1.20 log₁₀] genome copies/10⁵ cells).

TABLE 1.

Clinical data and CMV DNA-specific quantitative PCR results in 28 children with CMV congenital infection

Patient	Clinical symptom(s) related to CMV infection	CMV PCR result (protocol 1)	CMV PCR result (protocol 2)	CMV DNA copies (log10)/105 cells (protocol 1)
A1	None	+/+	+	460 (2.66)
A2	None	+/+	+	120 (2.08)
A3	None	+/+	+	140 (2.15)
A4	None	+/+	+	360 (2.55)
A5	None	+/+	+	NDc
A6	None	+/+	+	70 (1.85)
A7	None	+/−	+	450 (2.65)
A8	None	+/+	+	57 (1.76)
A9	IUGR,b ventriculomegaly	+/+	+	185,000 (5.27)
A10	Intracerebral calcificationsa	+/+	+	110,000 (5.04)
A11	Ventriculomegalya	+/+	+	4,000 (3.60)
A12	Hepatomegaly, petechiae	+/+	+	6,520 (3.81)
A13	Deafness	+/+	+	720,000 (5.86)
A14	Hepatomegaly, thrombocytopenia	+/−	+	16 (1.20)
B1	Deafness	+/+	+	600 (2.78)
B2	Deafness	+/+	+	230 (2.36)
B3	Deafness	+/+	+	1,200 (3.08)
B4	Deafness	+/+	+	980 (2.99)
B5	Deafness	+/+	+	120,000 (5.08)
B6	Deafness	+/+	+	21,000 (4.32)
B7	Deafness	+/+	+	7,050 (3.85)
B8	Deafness	+/+	+	1,450 (3.16)
B9	Deafness	+/+	+	2,750 (3.44)
B10	Deafness	+/+	+	7,050 (3.85)
B11	Deafness	+/+	ND	7,500 (3.88)
B12	Deafness	+/+	+	165 (2.22)
C1	Mental retardation	+/+	+	ND
C2	Hepatomegaly, thrombocytopenia	+/+	+	1,350 (3.13)

^aIn these two cases Guthrie cards were analyzed with fetal blood collected at termination of pregnancy.

^bIUGR, in utero growth retardation.

^cND, not done.

Our results show that the detection threshold was slightly lower with the phenol-chloroform extraction than with the silica-based extraction (protocol one). However, clinical sensitivity values of the two extraction protocols were similar, and protocol one appeared to be more convenient for routine testing (less time-consuming and no exposure risk of phenol toxicity). CMV DNA quantification performances were therefore analyzed following the extraction method with protocol one, as follows: 50 μl of 14 prequantified CMV DNA-specific PCR positive whole-blood samples (ranging from 1,000 to 4,000,000 [3.0 to 6.6 log₁₀] genome copies/ml) from 14 infected patients who had undergone transplantation were submitted for direct extraction with a QIAamp DNA blood mini-kit and were also spotted in triplicate on Guthrie cards. Viral loads obtained from DBS were underestimated compared to those obtained from fresh whole-blood samples, with a median difference of 0.73 log₁₀ genome copies/ml (range, 0.19 to 1.18 log₁₀). Following normalization of viral loads with albumin PCR results, the median difference dropped to 0.27 log₁₀ genome copies/ml (range, 0.03 to 0.7 log₁₀), with high correlation (r = 0.94; P < 0.006, Spearman correlation).

The median normalized CMV DNA viral load was significantly higher for the six symptomatic neonates (patients A9 to A14; 58,260 [4.76 log₁₀] genome copies/10⁵ cells; range, 16 to 720,000) than for the seven asymptomatic ones (patients A1 to A8; 140 [2.14 log₁₀] genome copies/10⁵ cells; range, 57 to 460; Mann-Whitney U test, P = 0.045). The median normalized CMV DNA viral load of the 12 children with hearing loss (patients B1 to B12) was significantly higher (2,100 [3.32 log₁₀] genome copies/10⁵ cells; range, 165 to 120,000) than the median value for the 7 asymptomatic neonates (Mann-Whitney U test, P = 0.002) (Fig. 1).

FIG. 1.

Comparison of CMV DNA loads in symptomatic neonates, asymptomatic neonates, and children with CMV-related deafness. The boxes representing the CMV DNA loads detected from DBS from each group of patients extend from the 25th percentile to the 75th percentile, with a line at the median (50th percentile). The dots at each extremity represent the lowest and highest values.

The threshold values of the CMV DNA-specific PCR in DBS reported here, following any of our two DNA extraction protocols, was in the same range as those (e.g., 2,000 or 4,000 genome copies/ml) reported by two other groups (3, 14). These threshold values are much higher than those obtained using fresh whole-blood samples; indeed, the latter values usually range from 50 to 500 genome copies/ml. This difference can be explained as follows: (i) only 50 μl of blood is extracted from a whole DBS compared to the 200 to 500 μl extracted from fresh blood, and (ii) whatever the DNA extraction protocol used, it is less efficient when performed with DBS than with fresh blood. However, even if the in vitro sensitivity of CMV PCR in DBS is relatively low, its clinical sensitivity was 100% in our study and ranges from 80 to 100% in the literature (1, 2, 9, 14). CMV DNA quantification with DBS has not been reported so far. We show here that CMV DNA quantification with DBS is valuable after normalization of the viral load with the albumin gene quantification results in order to compensate for the lower efficiency of the DBS extraction. Moreover, our data indicate that normalized viral loads were significantly lower in asymptomatic neonates than in symptomatic ones (P = 0.045) and lower than in children with hearing loss (P = 0.02). These results are consistent with those published previously which indicated that 90% of neonates who subsequently developed hearing loss had a CMV viral load of over 1,000 copies/10⁵ cells (4, 10), indicating that the viral burden in neonatal blood might be a predictive marker for hearing loss development. However, according to our results, some children may still develop deafness, although their viral load is as low as those of asymptomatic children (Fig. 1).

In conclusion, we believe that quantitative real-time CMV DNA-specific PCR with DBS, following either of our two extraction protocols and after normalization, is a reliable tool for the retrospective diagnosis of CMV congenital infection. Although these results should be confirmed in a larger study, CMV viral load from DBS could help identify children who are at risk of adverse outcomes, especially CMV-induced hearing loss.