Download Cytomegalovirus (CMV)–Encoded UL144 (Truncated Tumor

MAJOR ARTICLE
Cytomegalovirus (CMV)–Encoded UL144
(Truncated Tumor Necrosis Factor Receptor)
and Outcome of Congenital CMV Infection
Ravit Arav-Boger,1,2 Catherine A. Battaglia,3 Tiziana Lazzarotto,4 Liliana Gabrielli,4 Jian C. Zong,2
Gary S. Hayward,2 Marie Diener-West,3 and Maria P. Landini4
1
Department of Pediatrics, Division of Infectious Disease, Johns Hopkins Hospital, 2Molecular Virology Laboratories,
Sidney-Kimmel Cancer Research Center, Johns Hopkins University School of Medicine, and 3Department of Biostatistics,
Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland; 4Department of Clinical and Experimental
Medicine, Clinical Unit of Microbiology, St. Orsola Malpighi General Hospital, University of Bologna, Bologna, Italy
Background. Cytomegalovirus (CMV) is the most common congenital infection in humans. The effect of viral
strains on the outcome of congenital CMV is debated. We evaluated whether UL144 polymorphisms in amniotic
fluid from CMV-infected Italian women were associated with terminations of pregnancy, subsequent disease in
their offspring, or viral load.
Methods. The study was nested within a prenatal CMV program. We sequenced the UL144 gene from 66
amniotic-fluid samples, without knowledge of pregnancy outcome. We performed data analyses on 56 samples for
which all information was available.
Results. Genotype C was associated with termination of pregnancy (P p .03 ). Genotype B was associated with
fewer terminations of pregnancy (P p .003 ). A possible association was found between genotype C and symptomatic
disease in newborns (odds ratio, 8.81 [95% confidence interval, 0.48–164.02]; P p .05 ). There was no association
between specific genotype and the viral load in amniotic fluid. Symptomatic newborns who had the most common
UL144 genotype (B) were more likely to have higher viral loads than were asymptomatic infants (P p .003).
Conclusions. UL144 polymorphisms may be associated with the outcome of congenital CMV infection. Larger
studies should be conducted to confirm this association, before genotype analysis can be used, along with other
factors, in considering terminations of pregnancy.
Cytomegalovirus (CMV) is the most common congenital infection in humans; it affects ∼1% of newborns
in the United States and 0.2%–2.2% of newborns in
other countries [1, 2]. A wide spectrum of disease severity
has been documented in congenitally infected infants;
the majority of them are asymptomatic, but ∼10% have
disseminated disease and central nervous system (CNS)
Received 28 October 2005; accepted 3 March 2006; electronically published 7
July 2006.
Potential conflicts of interest: none reported.
Financial support: United Cerebral Palsy Research and Education Foundation;
Johns Hopkins Clinician Scientist Award (to R.A.-B.); Ministry of Public Health,
Istituto Superiore di Sanità, AIDS Program; Ministry of University, Scientific and
Technological Research; St. Orsola Malpighi General Hospital; University of Bologna
(grants to T.L., L.G., and M.P.L.).
Reprints or correspondence: Dr. Ravit Arav-Boger, Dept. of Pediatrics, Div. of
Infectious Diseases, Johns Hopkins Hospital/Park 256, Baltimore, MD 21287
([email protected]).
The Journal of Infectious Diseases 2006; 194:464–73
2006 by the Infectious Diseases Society of America. All rights reserved.
0022-1899/2006/19404-0008$15.00
464 • JID 2006:194 (15 August) • Arav-Boger et al.
involvement [3–6]. Most of the symptomatic infants who
do not die have long-term CNS sequelae, such as mental
retardation, seizures, and hearing loss [6, 7]. The estimated annual societal cost of supporting children with
congenital CMV sequelae approaches $2 billion (1991
US dollars) [8]. Available evidence suggests that both
host and viral factors—including maternal immunity,
CMV DNA load in amniotic fluid, and CMV strain—
may contribute to the outcome of congenital CMV infection [1–11]. Viral loads have recently been suggested
to be good predictive and prognostic markers of disease
severity, but it remains unknown which genes influence
viral loads [12, 13].
Genetic variability in human CMV isolates has been
well established [14–16]. Whether this variability contributes to the outcome of CMV infection in general
and to congenital CMV in particular is debated [16].
In the past, we have studied several genes as potential
virulence factors, including UL144, UL146, UL147,
US28, and glycoprotein B (gB), and we found that only polymorphisms in UL144, a cytokine receptor gene, were associated
with outcome [11]. Our interest in UL144 was piqued because
of its consistent presence in all clinical isolates but its frequent
deletion, along with up to 18 other open reading frames, in
laboratory-adapted strains [17]. Moreover, UL144 is a potential
candidate for a pathogenesis marker because it is a truncated
tumor necrosis factor (TNF)–like receptor [18]. Interest in this
gene locus is reflected in recent reports from other parts of the
world, including France, Japan, and China [19–21].
We and others [11, 14, 19, 22] have identified 3 major genotypes in UL144—A, B, and C, of which B is the most common, as well as several relatively rare recombinants of A/B and
A/C. Uncommon genotypes of UL144 have been detected more
frequently in newborns with severe congenital CMV disease
than in asymptomatic newborns [11]. To extend our initial
observations to a different geographic area, we evaluated, in a
retrospective study, whether polymorphisms of UL144 in amniotic fluid from pregnant Italian women who contracted CMV
during pregnancy were associated with termination of pregnancy or with the outcome of infection in their offspring. We
also examined the possible relationship between UL144 genotypes and viral loads.
PATIENTS AND METHODS
The subjects involved in the present study were among those
enrolled in a prenatal diagnosis program for CMV in Italy [10].
Samples of amniotic fluid were obtained during 1990–2003
from women who were identified as being at risk of transmitting
CMV and who agreed to participate in the program. On the
basis of published data, all pregnant women were counseled
regarding the risk of vertical transmission, possible fetal and
neonatal damage, and the availability of fetal investigations.
Sixty-six women agreed to participate in the study and underwent amniocentesis. After written informed consent was
obtained, 20 mL of amniotic fluid was collected by transabdominal amniocentesis under continuous ultrasound guidance
at 21–23 weeks of gestation [10]. On the basis of clinical and
virological results (e.g., viral loads), women decided either to
continue their pregnancies to delivery or to terminate their
pregnancies. CMV infection in newborns was determined by
the isolation of CMV from urine during the first week of life.
Infected newborns were classified as having symptomatic disease or asymptomatic infection on the basis of the presence of
physical and laboratory findings. One or more of the following
findings determined symptomatic disease: retardation of intrauterine growth; hepatosplenomegaly; petechiae/purpura; thrombocytopenia (!100,000 platelets/mm3); jaundice with direct bilirubin levels of 13 mg/dL; alanine aminotransferase levels of
180 U/L; pneumonia; neurologic involvement, including microcephaly, lethargy/hypotonia, poor sucking, seizures, and
sensorineural defects (i.e., chorioretinitis and deafness); and
abnormal neuroimaging results (periventricular hyperechogenicity, intracranial calcifications, ventriculomegaly, and hyperechogenicity of lenticolostriatal vessels). In aborted fetuses,
macroscopic and histologic tissue examination was performed
for multiple tissues.
The shell-vial procedure [23] was used to isolate CMV from
samples of amniotic fluid and urine. Inoculated cells were fixed
24–48 h after inoculation and were stained using an indirect
immunofluorescence assay with a monoclonal antibody reacting with the CMV IE1 and EA genes. The methods for DNA
extraction and quantitative polymerase chain reaction (PCR)
have been described elsewhere [10, 24]. DNA from samples of
amniotic fluid was used for the amplification of UL144. The
PCR conditions and primers have been published elsewhere
[11]. PCR products were gel purified and sequenced using the
BigDye Terminator Cycle Sequencing Kit (Perkin-Elmer Applied Biosystems). An ABI 310 automated sequencer (Applied
Biosystems) was used to analyze the sequencing products. The
sequencing alignment and phylogenetic analyses used the multiple-alignment algorithms in the Megalign program package
for proteins (Lasergene version 7; DNAStar). There were no
identifiers on any of the samples for which PCR and sequence
data were available.
To assess bivariate relationships among genotype, viral load,
and either symptomatic disease or termination of pregnancy,
analyses were performed using x2 and Fisher’s exact tests. A
test-based 95% confidence interval (CI) was constructed for
the unadjusted odds ratio (OR) when cell counts were small
in the cross-tabulation table. A correction of 0.5 was added to
each cell count when a zero cell was present. Unadjusted and
adjusted ORs for exposure and associated 95% CIs were obtained using simple and multiple logistic regressions, respectively. All exposure covariates that were significantly associated
with outcome in the bivariate analyses (P ! .05 ) were included
in the multivariate analysis. Analyses to assess differences in
the relative frequency of genotypes over time were obtained
using the x2 goodness-of-fit test. All statistical analyses were
performed using Stata software (version 8.0; StataCorp).
RESULTS
Sixty-six samples of amniotic fluid were obtained from 66
pregnant women during 1990–2003 and were available for
genotype analysis. We were able to amplify the UL144 genotype in 56 of 66 samples (GenBank accession numbers
DQ275474–DQ275528). All 10 samples from which UL144 was
not amplified were obtained from mothers of asymptomatic
newborns. Data analysis was therefore performed on the 56
samples typed. Of the enrolled women, 25 terminated their
pregnancies, 1 had an intrauterine death, and 30 had an uneventful delivery. Of the 25 women who decided to terminate
CMV-Encoded UL144 in Congenital Infection • JID 2006:194 (15 August) • 465
Table 1.
Detailed results of the study population.
Amniotic fluid
Patient
UL144
genotype
Symptoms, autopsy findings
Late sequelae
Virus
isolation
qPCR
result,
copies/mL
Isolation of CMV
in urine during the
first week of life
11
12
A1
B
Asymptomatic
Asymptomatic
…
Bilateral hearing loss
…
Positive
!9.4 ⫻ 102
Positive
1.0 ⫻ 103
14
15
16
17
A1
A1
A1
A2
Asymptomatic
Asymptomatic
Asymptomatic
Petechiae, thrombocytopenia
…
…
…
Monolateral hearing loss
Positive
…
…
Positive
1.0 ⫻ 103
1.2 ⫻ 104
!9.4 ⫻ 102
1.9 ⫻ 105
Positive
Positive
Positive
18
19
B
C/B
Hepatitis
ToP, disseminated infection, cerebral ventriculomegaly, hyperechogenic bowel
…
…
Positive
Positive
1.8 ⫻ 106
1.4 ⫻ 106
Positive
20
B
Cerebral ventriculomegaly, diffused
microcalcified cerebral areas
Bilateral hearing loss,
psychomotor retardation
Positive
1.8 ⫻ 106
Positive
21
C
ToP, disseminated infection, hyperechogenic cerebral areas
…
Positive
6.5 ⫻ 105
NA
22
B
Diffused microcalcified cerebral areas
Positive
9.0 ⫻ 105
Positive
23
B
ToP, disseminated infection, DandyWalker syndrome
Monolateral hearing loss,
psychomotor retardation
…
Positive
5.5 ⫻ 105
NA
24
25
B
A/C
Hepatitis, high C-reactive protein
Thrombocytopenia, diffused microcalcified cerebral areas
…
Strabismus
Positive
Positive
1.8 ⫻ 106
1.7 ⫻ 106
Positive
26
A/C
ToP, disseminated infection, cerebral ventriculomegaly, hyperechogenic bowel,
pneumonitis, hepatitis
ToP, disseminated infection
…
Positive
4.0 ⫻ 104
NA
NA
NA
Positive
Positive
NA
Positive
28
C
…
Positive
3.1 ⫻ 105
29
30
C
B
ToP, disseminated infection
ToP, disseminated infection, cerebral ventriculomegaly, pneumonitis
…
…
Positive
Positive
1.7 ⫻ 105
1.0 ⫻ 106
31
B
…
Positive
3.0 ⫻ 106
NA
32
C
ToP, disseminated infection, cerebral ventriculomegaly, hepatosplenomegaly,
pneumonia
Diffused microcalcified cerebral areas,
cerebral ventriculomegaly, hepatosplenomegaly
Bilateral hearing loss,
psychomotor retardation
Positive
2.9 ⫻ 105
Positive
33
34
B
A1
Intrauterine growth restriction
ToP, disseminated infection, pneumonitis, agenesis of corpus callosum,
hepatomegaly
IQ !70
…
Positive
Positive
4.0 ⫻ 105
1.1 ⫻ 106
Positive
NA
35
36
B
A1
Hepatitis
ToP, disseminated infection, hepatosplenomegaly, pneumonia
…
…
Positive
Positive
3.7 ⫻ 105
9.1 ⫻ 105
Positive
37
38
39
40
B
C
B
A1
Hyperechogenic bowel
Intrauterine deatha
ToP, disseminated infection
ToP, disseminated infection
…
…
3.7 ⫻ 103
Positive
…
…
…
…
Positive
Positive
!9.4 ⫻ 102
NA
NA
11.0 ⫻ 106
NA
41
42
A2
A1
Diffused microcalcified cerebral areas
ToP, disseminated infection, pneumonia
…
…
Positive
Positive
1.0 ⫻ 105
1.1 ⫻ 105
Positive
43
44
A2
A1
ToP, disseminated infection, spleen, lung
ToP, disseminated infection, pneumonia,
diffused microcalcified cerebral areas,
kidney, pancreas
…
…
Positive
Positive
11.0 ⫻ 106
NA
1.7 ⫻ 105
NA
5.2 ⫻ 105
NA
NA
NA
(continued)
Table 1. (Continued.)
Amniotic fluid
Patient
UL144
genotype
Late sequelae
Virus
isolation
qPCR
result,
copies/mL
Isolation of CMV
in urine during the
first week of life
Bilateral hearing loss
Positive
11.0 ⫻ 106
Positive
Symptoms, autopsy findings
45
B
Diffused microcalcified cerebral areas,
chorioretinitis
46
B
Asymptomatic
…
…
3.7 ⫻ 103
Positive
47
48
B
B
Asymptomatic
Asymptomatic
…
…
…
…
5.20 ⫻ 102
!9.4 ⫻ 102
Positive
Positive
49
50
B
B
Asymptomatic
Asymptomatic
…
Bilateral hearing loss
Positive
Positive
1.8 ⫻ 106
4.9 ⫻ 104
Positive
51
52
A1
B
Asymptomatic
Asymptomatic
…
…
…
Positive
!9.4 ⫻ 102
Positive
1.3 ⫻ 104
53
C
ToP, disseminated infection, pneumonia,
kidney, pancreas, liver
…
Positive
3.0 ⫻ 105
Positive
NA
54
B
ToP, disseminated infection, pneumonia,
kidney
…
Positive
2.9 ⫻ 106
NA
55
A/B
…
Positive
1.2 ⫻ 106
NA
56
A1
ToP, disseminated infection, pneumonia,
kidney, liver, heart, thymus
ToP, disseminated infection, pneumonia,
kidney
…
Positive
2.7 ⫻ 105
NA
57
C
…
Positive
1.4 ⫻ 105
NA
58
A1
ToP, disseminated infection, pneumonia,
cerebral ventriculomegaly, liver, pancreas, kidney
Asymptomatic
…
…
!9.4 ⫻ 102
Positive
59
60
A1
B
ToP, disseminated infection, kidney
ToP, disseminated infection, hepatosplenomegaly, lung, pancreas
…
…
Positive
Positive
1.6 ⫻ 105
1.7 ⫻ 105
NA
61
62
B
A2
Asymptomatic
ToP, lung, liver, kidney
…
…
Positive
Positive
2.4 ⫻ 105
4.2 ⫻ 105
Positive
63
68
B
A/C
Asymptomatic
ToP, disseminated infection, lung,
pancreas, kidney
…
…
Positive
Positive
3.5 ⫻ 105
1.9 ⫻ 105
Positive
69
70
72
73
A1
A2
B
B
ToP, disseminated infection
Asymptomatic
Asymptomatic
Asymptomatic
…
…
…
…
Positive
…
Positive
Positive
1.5
6.5
2.5
5.0
⫻
⫻
⫻
⫻
NA
Positive
105
102
105
104
Positive
NA
NA
NA
Positive
Positive
NOTE. CMV, cytomegalovirus; NA, not available; qPCR, quantitative polymerase chain reaction; ToP, termination of pregnancy.
a
Results at autopsy were fetal hydrops, edema, and hepatosplenomegaly.
their pregnancies, 10 did so on the basis of both pathologic
ultrasound findings and high viral loads in amniotic fluid, 4
on the basis of both positive CMV results in fetal blood and
high viral loads in amniotic fluid, 3 on the basis of borderline
ultrasound findings and high viral loads in amniotic fluid, and
8 on the basis of high viral loads in amniotic fluid.
Viral and host characteristics of pregnant women who agreed
to participate in the study are detailed below; expanded characteristics of both pregnant women and their offspring are
described in table 1. All women were white, and the age at the
time of amniocentesis was recorded for 30 women with term
deliveries and ranged from 20 to 41 years. Of the 56 women
who underwent amniocentesis and for whom we had both viral
load and genotype data, CMV was present in 45 samples of
amniotic fluid (80.4%). We used the viral-load threshold cutoff
in amniotic fluid of 1 ⫻ 10 5 copies/mL determined by Lazzarotto et al. [10, 24], and we found that a viral load in amniotic
fluid of ⭓1 ⫻ 10 5 copies/mL was present in 39 (69.6%) of the
samples.
Of the samples of amniotic fluid evaluated, the following distribution of UL144 genotypes was determined: B, 25 (44.6%);
A1, 14 (25.0%); C, 7 (12.5%); A2, 5 (8.9%); and A/C recombinant, 3 (5.4%), with 1 (1.8%) each of recombinants A/B and
C/B. Each genotype was analyzed in comparison with all other
CMV-Encoded UL144 in Congenital Infection • JID 2006:194 (15 August) • 467
Table 2.
Association of UL144 genotype and viral load in amniotic fluid with outcome of cytomegalovirus (CMV) infection.
Isolation of CMV
Amniotic fluid qPCR
in amniotic fluid
result, copies/mL
UL144 recombinant
UL144 genotype B
UL144 genotype C
UL144 genotype A1
genotypes
UL144 genotype A2
Association
result
Non–A/B,
!1 ⫻ 105
⭓1 ⫻ 105
B
Non-B
C
Non-C
A1
Non-A1
A2
Non-A2
A/B, A/C, B/C
A/C, B/C
2 (94.7)
3 (7.9)
35 (92.1)
14 (36.8)
24 (63.2)
7 (18.4)
31 (81.6)
8 (21.1)
30 (79.9)
4 (10.5)
34 (89.5)
5 (13.2)
33 (86.8)
9 (50.0)
14 (77.8)
4 (22.2)
11 (61.1)
7 (38.9)
0 (0)
18 (100)
6 (33.3)
12 (66.7)
1 (5.6)
17 (94.4)
0 (0)
18 (100)
Positive
Negative
36 (5.3)
Infection, no. (%)
Symptomatic
Asymptomatic
Pearson’s x2
P
OR (95% CI)
9 (50.0)
15.49
28.2
2.91
3.79
0.98
0.37
0.11
!.001
!.001
.09
.05
.32
.54
.11
18.00 (3.30–98.27)
40.83 (8.08–206.40)
0.371 (0.12–1.18)
8.81 (0.48–164.02)
0.533 (0.15–1.87)
2.00 (0.21–19.31)
6.07 (0.32–115.58)
a
a
NOTE. CI, confidence interval; OR, odds ratio; qPCR, quantitative polymerase chain reaction.
a
Before the calculation of OR and CI, 0.5 was added to each cell count. The 2-sided Fisher’s exact test results for genotype C and for recombinant genotypes were 0.08 and 0.13, respectively.
Table 3. Association of UL144 genotypes with isolation of cytomegalovirus (CMV) and quantitative polymerase chain reaction
(qPCR).
Result of amniotic
fluid testing
Isolation of CMV
Positive
Negative
Pearson’s x2
P
qPCR
!1 ⫻ 105 copies/mL
⭓1 ⫻ 105 copies/mL
Pearson’s x2
P
C
Non-C
B
Non-B
6 (85.7)
1 (14.3)
39 (79.6)
10 (20.4)
21 (84.0)
4 (16.0)
24 (77.4)
7 (22.6)
0.15
.70
0.15
.70
0.38
.54
0.38
.54
1 (14.3)
16 (32.6)
8 (32.0)
9 (29.0)
6 (85.7)
33 (67.4)
17 (68.0)
22 (71.0)
0.98
.32
0.98
.32
0.06
.81
0.06
.81
NOTE. Data are no. (%), unless otherwise indicated.
genotypes. We considered genotype recombinants A/B, A/C,
and C/B to be 1 group in comparison with the separate genotypes A1, A2, B, and C.
All 30 live newborns had a positive urine culture; 12 (40%)
were symptomatic, and 18 (60%) were asymptomatic. All terminations of pregnancy and the single intrauterine death were
defined as symptomatic on the basis of viral loads, ultrasound
findings, and macroscopic and histological evidence of multiple organ involvement. We also performed subset analysis to
compare symptomatic versus asymptomatic newborns and terminations of pregnancy versus asymptomatic newborns. CMVassociated late sequelae were observed in 9 (30%) of the newborns and included monolateral and bilateral hearing loss,
psychomotor retardation, strabismus, and an IQ of !70.
We investigated the associations among UL144 genotypes,
viral loads, and symptomatic disease (table 2). Culture positivity
of and high viral load (⭓1 ⫻ 10 5 copies/mL of virus) in amniotic fluid were significantly associated with symptomatic disease (OR, 18.00 [95% CI, 3.30–98.27]; P ! .001; and OR, 40.83
[95% CI, 8.08–206.40]; P ! .001, respectively).
Although genotypes A1, A2, and B occurred in both symptomatic and asymptomatic newborns, genotype C and all recombinants were found exclusively in symptomatic newborns
and terminations of pregnancy. A possible association was
noted for genotype C and symptomatic disease (OR, 8.81 [95%
CI, 0.48–164.02]; P p .05, Pearson’s x2; P p .08, 2-sided
Fisher’s exact test). When we grouped recombinant genotypes
containing C with genotype C, we found a statistically significant association between these genotypes and outcome (OR,
15.47 [95% CI, 0.86–279.01]; P p .01, 2-sided Fisher’s exact
test). The probability of having genotype B was 61.1% in
asymptomatic newborns, compared with 36.8% in symptomatic newborns (P p .09). A statistically significant association
was determined between the presence of genotype C CMV and
termination of pregnancy (P p .03 ). There was increased odds
of terminating pregnancy when genotype C CMV was detected
in amniotic fluid (OR, 8.7 [95% CI, 1.3–59.4]) (table 3). The
OR for genotype C, after adjustment for culture positivity
(P p .70) and for viral loads of ⭓1 ⫻ 10 5 copies/mL in amniotic fluid (P p .32), was 12.87 (95% CI, 0.92–178.40) and
8.95 (95% CI, 0.76–105.20), respectively. We also found that
the percentage of women terminating their pregnancies in the
presence of non-B UL144 genotypes was 64.5%, compared with
24.0% for genotype B (P p .003) (table 4). The odds of a
woman terminating her pregnancy were reduced by 83% (OR,
0.17 [95% CI, 0.05–0.56]) when genotype B CMV was detected
in amniotic fluid. After adjustment for both culture positivity
and ⭓1 ⫻ 10 5 copies/mL of DNA amniotic fluid, the adjusted odds that a woman infected with genotype B CMV would
terminate her pregnancy, compared with women infected with
non–genotype B CMV, were reduced by 90% (OR, 0.10 [95%
CI, 0.02–0.42]). In subset analysis of symptomatic versus
asymptomatic newborns and terminations of pregnancy versus
asymptomatic newborns (table 5), findings were consistent with
those of the original analysis, in which symptomatic newborns
and terminations of pregnancy were grouped together under
the category of “symptomatic” (table 2). Viral isolation and
loads in amniotic fluid were significantly associated with outcome. There were higher odds of infection with C genotype or
C and recombinant genotypes containing C in symptomatic
than in asymptomatic newborns (OR, 4.83 and 8.81, respectively) and in terminations of pregnancy than in asymptomatic
Table 4. Unadjusted and adjusted odds ratios (ORs) and 95% confidence intervals
(CIs) for termination of pregnancy, by genotype.
Termination, no. (%)
OR (95% CI)
Yes
No
P
Unadjusted
Adjusteda
B
Non-B
C
6 (24)
20 (64.5)
6 (85)
19 (76)
1 (35.5)
1 (14.3)
.003
0.17 (0.05–0.56)
0.10 (0.02–0.42)
.03
8.70 (0.97–77.20)
10.9 (0.74–160.40)
Non-C
20 (40.8)
29 (59.2)
Genotype
a
Adjusted for isolation of cytomegalovirus and quantitative polymerase chain reaction.
CMV-Encoded UL144 in Congenital Infection • JID 2006:194 (15 August) • 469
Table 5. Subset analysis of symptomatic vs. asymptomatic liveborn infants and terminations of pregnancy vs. asymptomatic liveborn infants.
Terminations vs.
asymptomatic
Symptomatic vs.
asymptomatic
Characteristic
OR (95% CI)
P
OR (95% CI)
P
Viral isolation
qPCR
11.00 (1.16–103.94)
38.50 (3.75–395.40)
.036
.002
25.00 (2.76–226.03)
42.00 (6.80–259.44)
.004
.000
Genotype B
Genotype C
1.27 (0.28–5.87)
4.83 (0.18–128.80)
.757
.213
0.19 (0.05–0.71)
11.73 (0.06–222.88)
.014
.103
Genotype C and recombinant
C genotypes
8.81 (0.38–201.40)
.15
20.08 (1.08–371.67)
.005
Genotype A1
0.080 (0.004–1.520)
.06
0.89 (0.25–3.22)
.858
Genotype A2
Recombinant genotypes
3.40 (0.27–42.44)
4.83 (0.18–128.80)
.342
.40
1.42 (0.12–16.91)
7.40 (0.37–146.54)
.783
.133
NOTE. For the calculation of odds ratios (ORs) and 95% confidence intervals (CIs), 0.5 was added to
each cell of the 2 ⫻ 2 table. qPCR, quantitative polymerase chain reaction.
newborns (OR, 11.73 and 20.08, respectively). Genotype B was
associated with fewer terminations of pregnancy.
A graphic display of the association between UL144 genotypes, given symptomatic and asymptomatic disease versus log
viral load in amniotic fluid, is shown in figure 1. For all UL144
genotypes, symptomatic disease was associated with higher viral
loads in amniotic fluid. We performed a subgroup analysis of
the 25 mothers enrolled who had UL144 genotype B, to determine the relationship between symptomatic disease in newborns and the viral load in amniotic fluid. We determined that
the percentage of symptomatic newborns with viral loads in
amniotic fluid of ⭓1 ⫻ 10 5 copies/mL was 92.9%, compared
with 36.4% of those with asymptomatic infection (P p .003).
The OR of a symptomatic newborn having ⭓1 ⫻ 10 5 copies/
mL of virus in amniotic fluid was 22.8 (95% CI, 2.1–244.9)
times higher than that of an asymptomatic newborn. A similar
subgroup analysis could not be performed for genotype C,
because there were no asymptomatic newborns with this genotype. No statistically significant associations were found between UL144 genotypes A1, A2, or recombinants A/B, A/C, and
C/B and outcomes of pregnancy or virologic parameters (data
not shown).
To determine whether the relative frequency of genotypes
changed over time, the date of amniocentesis was determined.
Test dates preceding January 1998 were excluded from this
analysis (n p 4). The remaining 52 test dates were then divided
into 6-month intervals, beginning with 1 January 1998. Using
a x2 goodness-of-fit test, we determined that the distribution
of genotypes did not significantly vary over time, thereby providing no evidence that the appearance of uncommon genotypes are responsible for outbreaks of severe disease (P p
.759; data not shown).
470 • JID 2006:194 (15 August) • Arav-Boger et al.
DISCUSSION
Our genotype analysis of UL144 in amniotic fluid from Italian
women who contracted CMV during pregnancy suggests that
polymorphisms, specifically genotypes B and C, may be associated with the outcome of congenital CMV disease. For the
common genotype B, viral load predicted outcome, but infection with one of the uncommon UL144 genotypes was associated with severe disease, irrespective of viral load.
In a recent US cohort of newborns with congenital CMV
infection, we found and reported that UL144 polymorphism
was associated with outcome. Among 23 newborns, 13 of whom
were symptomatic and 10 of whom were asymptomatic, genotype B was found in both groups, whereas genotypes C, A1,
and A/C were found exclusively in symptomatic newborns [11].
Of the 5 genes that were sequenced (UL144, UL146, UL147,
US28, and gB), only polymorphisms in the UL144 gene appeared to be associated with the outcome of congenital CMV
infection during the immediate neonatal period. We hypothesized that human CMV genotypes that have not recently coevolved with the host population are less well tolerated and are
therefore associated with a worse outcome. We were, therefore,
interested in studying UL144 polymorphisms during pregnancy
in a different geographic region, to determine whether they
could predict the outcome of congenital CMV infection in
newborns.
The distribution of genotypes in the Italian population that
we report shows, again, that genotype B is the most common
genotype in the population (73.9% in the US cohort and 44.6%
in the Italian cohort). The distribution of the uncommon genotypes shows some differences (table 6). Among 56 samples
in the Italian cohort, 38 were from symptomatic newborns (25
terminations, 1 intrauterine death, and 12 newborns), and 18
Figure 1. Box plots comparing log10 cytomegalovirus load in amniotic
fluid (in copies per milliliter) for symptomatic and asymptomatic newborns,
by UL144 genotype. For each genotype, symptomatic newborns are shown
as black bars on the left, and asymptomatic newborns (when present)
are shown as bars shaded on the right. Genotypes for symptomatic
newborns were genotype B (n p 14 ), genotype C (n p 7), genotype A1
(n p 8), genotype A2 (n p 4), and recombinant genotypes (n p 5 ). Genotypes for asymptomatic newborns were genotype B (n p 11), genotype
C (n p 0), genotype A1 (n p 6), genotype A2 (n p 1 ), and recombinant
genotypes (n p 0).
were from asymptomatic newborns. Genotypes B, A1, and A2
were found in both symptomatic and asymptomatic newborns,
whereas genotypes C and the recombinants A/B, A/C, and C/
B were found exclusively in symptomatic newborns. We note
that the newborns in our US study cohort were African American; therefore, the distribution of polymorphisms may have
been related to genetic susceptibility between the 2 different
populations.
Our results show that the odds of having genotype B CMV
in the amniotic fluid of asymptomatic newborns was ∼20%
higher than that of symptomatic newborns (P p .09 ). For genotype B CMV, a severe outcome was associated with higher
viral loads in amniotic fluid, specifically those ⭓1 ⫻ 10 5 copies/
mL (P p .003). We note that there was no association between
genotype B and symptomatic disease; therefore, the high viral
load could be secondary to the presence of symptomatic disease
and may not be related to UL144 genotype. For genotypes other
than B, viral loads in amniotic fluid were not predictive of
outcome, although our analyses were limited by the small sample sizes of these groups
Similar to genotype B, genotypes A1 and A2 appeared in
both symptomatic and asymptomatic newborns, but no significant associations were found between outcome and virologic
parameters. Genotypes C and all recombinants (A/B, A/C, and
C/B) were exclusively found to be associated with severe disease,
in either live newborns or terminations of pregnancy. Although
the recombinant genotypes were not statistically associated with
symptomatic severe disease, a possible association was found
for genotype C and symptomatic CMV disease (OR, 8.81 [95%
CI, 0.48–164.02]; P p .05, Pearson’s x2; P p .08, 2-sided Fisher’s exact test).
We found that genotypes B and C had opposite effects on
termination of pregnancy, both of which were statistically significant. Although genotype C was significantly associated with
termination of pregnancy (P p .03), fewer terminations were
noted for genotype B (P p .003). The odds of termination of
pregnancy increased in the presence of non-B and non-C genotypes (OR, 5.76 [95% CI, 1.78–18.67] and OR, 8.7 [95% CI,
0.97–77.92], respectively). When we analyzed the role of possible confounding due to virologic parameters, these associations remained statistically significant. After adjustments for
viral load, both ORs moved farther away from the null value,
which suggests that the relationship between genotype and viral
load may be underestimated without adjustment.
There was no association between a specific genotype and
the ability to culture CMV from either amniotic fluid or urine,
which suggests that they all were able to grow well in cell
culture. Therefore, we agree with other researchers who reported that all genotypes can be transmitted from mothers to
newborns [19].
The results of our study differ, with regard to UL144 genotypes and their association with symptomatic disease, from
those of a report by Picone et al. [19], in which a distribution
of UL144 genotypes in a cohort of CMV-infected newborns
Table 6. Distribution of UL144 genotypes in the
US and Italian cohorts.
Cohort, UL144
genotype (%)
Symptomatic
newborns
Asymptomatic
newborns
US (n p 23)
B (73.9)
A1 (4.3)
7
1
10
0
A2 (0.0)
C (17.4)
0
4
0
0
A/C (4.3)
1
0
A/B (0.0)
B/C (0.0)
Total (%)
0
0
13 (56.5)
0
0
10 (43.5)
B (44.6)
14
11
A1 (25.0)
A2 (8.9)
C (12.5)
8
4
7
6
1
0
A/C (5.4)
A/B (1.8)
3
1
0
0
Italian (n p 56)
B/C (1.8)
Total (%)
1
38 (67.9)
0
18 (32.1)
CMV-Encoded UL144 in Congenital Infection • JID 2006:194 (15 August) • 471
and immunocompromised adults (organ or bone-marrow transplant recipients) is described. Although we agree that all UL144
genotypes are transmitted to the fetus, Picone et al. defined 2
groups in their study—severely and nonseverely symptomatic fetuses (defined by the presence of, at most, only 1 extracerebral ultrasound feature)—on the basis of ultrasound findings determined during second- and third-trimester evaluations. There were 23 terminations of pregnancy in that study,
all of which were based on abnormal ultrasound results, but
histopathologic data were not presented, and viral loads were
not determined. Of 11 newborns delivered at term, 9 were
asymptomatic and 2 were symptomatic, yet all were grouped
as severely symptomatic.
Picone et al. [19] concluded that all UL144 genotypes are
found in both symptomatic and asymptomatic infections.
However, a better classification of outcome would help in the
analysis of the results of that study. In the present article, we
report a possible association between genotype C and symptomatic disease in newborns (P p .05 , Pearson’s x2) and a significant association between genotype C and termination of
pregnancy (P p .03). Additionally, genotype B was associated
with fewer terminations of pregnancy (P p .003 ). The limitations of the present study are the small sample size of the
nested cohort, which led to wide CIs (i.e., imprecise estimates
of ORs). In addition, we acknowledge that, because of the
uncommon presence of genotype C and recombinants in the
amniotic fluid, we were limited in our interpretations of the
association between these genotypes and symptomatic CMV
disease.
We also recognize potential enrollment bias. Of the pregnant
women approached for enrollment into the study, ∼50% refused amniocentesis. It is unclear whether these refusals were
based on severity of disease, socioeconomic status, and/or religious considerations. Given the documented risks to the fetus
of amniocentesis, is it possible that evidence of more severe
disease leads to a pregnant woman’s decision to consent to
amniocentesis. The numbers of enrolled women in our Italian
cohort support this theory: 38 (67.9%) of the neonates were
classified as being symptomatic, compared with only 18 (32.1%)
who were classified as being asymptomatic. In addition, we
were unable to amplify the specific UL144 genotype in 10 samples of amniotic fluid. The outcome of these 10 pregnancies
was asymptomatic infection. It is likely that we were unable to
amplify UL144 from these samples because of the low number
of DNA copies detected and the use of a less-sensitive PCR
sequencing technique.
In conclusion, the present results indicate that UL144 genotypes C and B seem to have opposite effects on the outcome
of congenital CMV disease. We believe that these findings need
to be replicated and confirmed before they can begin to affect
clinical practice. Although there may be difficulties in obtaining
472 • JID 2006:194 (15 August) • Arav-Boger et al.
a larger cohort of samples, future studies should be conducted
to investigate further this controversial relationship between
UL144 genotypes and congenital, symptomatic CMV disease.
A recent elegant study showed that UL144 binds to the B and
T lymphocyte attenuator (BTLA) and may inhibit T cell proliferation [25]. Although all genotypes seem to bind the BTLA,
the effect that this has on T cell responses in congenital CMV
is still unclear. UL144 shows sequence similarity to the TNFa receptor superfamily, with 37% amino acid sequence similarity to herpesvirus entry mediator (HVEM). The viral ligand
for HVEM is herpesvirus gD, and its cellular counterparts are
LIGHT and lymphotoxin-a [26, 27]. Like UL144, HVEM interacts with BTLA via cysteine-rich domain 1, but whether it
may also affect TNF signaling pathways by another mechanism
is unknown.
To our knowledge, our study is the first to examine a possible
association between CMV genotypes other than gB and viral
load. Future studies will be important in identifying the genetic
elements that affect viral loads.
References
1. Stagno S. Cytomegalovirus. In: Remington JS, Klein JO, eds. Infectious
diseases of the fetus and newborn infant. 4th ed. Philadelphia: WB
Saunders, 1995:312–53.
2. Barbi M, Binda S, Primache V, Clerici D. Congenital cytomegalovirus
infection in a northern Italian region. NEOCMV Group. Eur J Epidemiol 1998; 14:791–6.
3. Arav-Boger R, Pass RF. Diagnosis and management of cytomegalovirus
infection in the newborn. Pediatr Ann 2002; 31:719–25.
4. Boppana SB, Fowler KB, Britt WJ, Stagno S, Pass RF. Symptomatic
congenital cytomegalovirus infection in infants born to mothers with
preexisting immunity to cytomegalovirus. Pediatrics 1999; 104:55–60.
5. Stagno S, Pass RF, Cloud G, et al. Primary cytomegalovirus infection
in pregnancy: incidence, transmission to fetus, and clinical outcome.
JAMA 1986; 256:1904–8.
6. Fowler KB, Stagno S, Pass RF. Maternal age and congenital cytomegalovirus infection: screening of two diverse newborn populations,
1980–1990. J Infect Dis 1993; 168:552–6.
7. Guerra B, Lazzarotto T, Quarta S, et al. Prenatal diagnosis of symptomatic congenital cytomegalovirus infection. Am J Obstet Gynecol
2000; 183:476–82.
8. Demmler GJ, Infectious Diseases Society of America, Centers for Disease Control. Summary of a workshop on surveillance for congenital
cytomegalovirus disease. Rev Infect Dis 1991; 13:315–29.
9. Fowler KB, Stagno S, Pass RF. Maternal immunity and prevention of
congenital cytomegalovirus infection. JAMA 2003; 289:1008–11.
10. Lazzarotto T, Varani S, Guerra B, Nicolosi A, Lanari M, Landini MP.
Prenatal indicators of congenital cytomegalovirus infection. J Pediatr
2000; 137:90–5.
11. Arav-Boger R, Willoughby RE, Pass RF, et al. Polymorphisms of the
cytomegalovirus (CMV)–encoded tumor necrosis factor–a and b-chemokine receptors in congenital CMV disease. J Infect Dis 2002; 186:
1057–64.
12. Boppana SB, Fowler KB, Pass RF, et al. Congenital cytomegalovirus
infection: association between virus burden in infancy and hearing
loss. J Pediatr 2005; 146:817–23.
13. Lanari M, Lazzarotto T, Venturi V, et al. Neonatal cytomegalovirus
blood load and risk of sequelae in symptomatic and asymptomatic
congenitally infected newborns. Pediatrics 2006; 117:e76–83.
14. Bale JF Jr, Petheram SJ, Robertson M, Murph JR, Demmler G. Human
cytomegalovirus a sequence and UL144 variability in strains from infected children. J Med Virol 2001; 65:90–6.
15. Meyer-Konig U, Vogelberg C, Bongarts A, et al. Glycoprotein B genotype correlates with cell tropism in vivo of human cytomegalovirus
infection. J Med Virol 1998; 55:75–81.
16. Pignatelli S, Dal Monte P, Rossini G, Landini MP. Genetic polymorphisms among human cytomegalovirus (HCMV) wild-type strains. Rev
Med Virol 2004; 14:383–410.
17. Cha TA, Tom E, Kemble GW, Duke GM, Mocarski ES, Spaete RR.
Human cytomegalovirus clinical isolates carry at least 19 genes not
found in laboratory strains. J Virol 1996; 70:78–83.
18. Benedict CA, Butrovich KD, Lurain NS, et al. Cutting edge: a novel
viral TNF receptor superfamily member in virulent strains of human
cytomegalovirus. J Immunol 1999; 162:6967–70.
19. Picone O, Costa JM, Chaix ML, Ville Y, Rouzioux C, Leruez-Ville M.
Human cytomegalovirus UL144 gene polymorphisms in congenital
infections. J Clin Microbiol 2005; 43:25–9.
20. He R, Ruan Q, Xia C, et al. Sequence variability of human cytomegalovirus UL144 open reading frame in low-passage clinical isolates.
Chin Med Sci J 2004; 19:293–7.
21. Murayama T, Takegoshi M, Tanuma J, Eizuru Y. Analysis of human
22.
23.
24.
25.
26.
27.
cytomegalovirus UL144 variability in low-passage clinical isolates in
Japan. Intervirology 2005; 48:201–6.
Lurain NS, Kapell KS, Huang DD, et al. Human cytomegalovirus
UL144 open reading frame: sequence hypervariability in low-passage
clinical isolates. J Virol 1999; 73:10040–50.
Gleaves CA, Smith TF, Shuster EA, Pearson GR. Rapid detection of
cytomegalovirus in MRC-5 cells inoculated with urine specimens by
using low-speed centrifugation and monoclonal antibody to an early
antigen. J Clin Microbiol 1984; 19:917–9.
Lazzarotto T, Gabrielli L, Foschini MP, et al. Congenital cytomegalovirus infection in twin pregnancies: viral load in the amniotic fluid
and pregnancy outcome. Pediatrics 2003; 112:e153–7.
Cheung TC, Humphreys IR, Potter KG, et al. Evolutionarily divergent
herpesviruses modulate T cell activation by targeting the herpesvirus
entry mediator cosignaling pathway. Proc Natl Acad Sci USA 2005;
102:13218–23.
Marsters SA, Ayres TM, Skubatch M, Gray CL, Rothe M, Ashkenazi
A. Herpesvirus entry mediator, a member of the tumor necrosis factor
receptor (TNFR) family, interacts with members of the TNFR-associated factor family and activates the transcription factors NF-kappaB
and AP-1. J Biol Chem 1997; 272:14029–32.
Granger SW, Rickert S. LIGHT-HVEM signaling and the regulation of
T cell-mediated immunity. Cytokine Growth Factor Rev 2003; 14:289–96.
CMV-Encoded UL144 in Congenital Infection • JID 2006:194 (15 August) • 473