Obstetrics & Gynecology:
Accuracy of Real-Time Polymerase Chain Reaction for Toxoplasma gondii in Amniotic Fluid
Wallon, Martine MD, PhD; Franck, Jacqueline MD; Thulliez, Philippe MD; Huissoud, Cyril MD, PhD; Peyron, François MD, PhD; Garcia-Meric, Patricia MD; Kieffer, François MD
From the Service de Parasitologie, Hôpital de la Croix-Rousse, Hospices Civils de Lyon, Lyon, France; the Laboratoire de Parasitologie-Mycologie, Hôpital de la Timone, Marseille, France; the Laboratoire de la Toxoplasmose, Institut de Puériculture de Paris, Paris, France; the Service de Gynécologie Obstétrique, Hôpital de la Croix Rousse, Hospices Civils de Lyon, Lyon, France; the Service de Néonatalogie, Hôpital de la Conception, Marseille, France; the Service de Néonatologie, Institut de Puériculture, Paris, France.
Corresponding author: Martine Wallon, Service de Parasitologie, Hôpital de la Croix Rousse, 103 grande rue de la Croix Rousse, F 69004 Lyon, France; e-mail: firstname.lastname@example.org.
Financial Disclosure The authors did not report any potential conflicts of interest.
OBJECTIVE: To provide clinicians with information about the accuracy of real-time polymerase chain reaction (PCR) analysis of amniotic fluid for the prenatal diagnosis of congenital Toxoplasma infection.
METHODS: This was a prospective cohort study of women with Toxoplasma infection identified by prenatal screening in three centers routinely carrying out real-time PCR for the detection of Toxoplasma gondii in amniotic fluid. The data available were gestational age at maternal infection, types and dates of maternal treatment, results of amniocentesis and neonatal work-up and definitive infectious status of the child. We estimated sensitivity, specificity and positive and negative predictive values both overall and per trimester of pregnancy at the time of maternal infection.
RESULTS: Polymerase chain reaction analysis was carried out on amniotic fluid for 261 of the 377 patients included (69%). It was accurate with the exception of four negative results in children who were infected. Overall sensitivity and negative predictive value were 92.2% (95% confidence interval [CI] 81–98%) and 98.1% (95% CI 95–99.5%), respectively. There was no significant association with the trimester of pregnancy during which maternal infection occurred. Specificity and positive predictive values of 100% were obtained for all trimesters.
CONCLUSION: Real-time PCR analysis significantly improves the detection of T. gondii on amniotic fluid. It provides an accurate tool to predict fetal infection and to decide on appropriate treatment and surveillance. However, postnatal follow-up remains necessary in the first year of life to fully exclude infection in children for whom PCR results were negative.
LEVEL OF EVIDENCE: III
Toxoplasma gondii infection is generally mild or subclinical in healthy humans. However, when acquired during pregnancy, it exposes the fetus to a risk of congenital infection. The consequences of such infection range from severe fetal lesions diagnosed in utero or at birth, to the late development of retinal diseases in otherwise healthy children or adults.1,2 The early detection and treatment of congenital Toxoplasma infection are thought to reduce the risk of severe clinical lesions,3 and neurological4 or ocular1 damage, although the precise benefits of such treatment remain unclear.5
Screening policies to prevent or diagnose congenital toxoplasmosis differ between developed countries from no prenatal screening (eg, in the United States) to monthly prenatal testing (eg, in France).6 Worldwide variations in the prevalence of past infection among pregnant women (ie, 44% in France7 compared with 15%8 in the United States) contribute to explain these differences. In several European countries screening for Toxoplasma gondii infections in pregnancy is either mandatory or recommended by the national guidelines6 but individual screening is increasingly performed elsewhere, at the request of pregnant women or on the initiative of their obstetricians. When acute infection is detected during pregnancy, the physician must simultaneously make decisions about the most appropriate treatment, organize fetal ultrasound surveillance and deal with the anxiety of the parents. Prenatal diagnosis by polymerase chain reaction (PCR)-based methods of amniotic fluid is widely used to detect fetal infection and to guide decisions concerning the need to introduce or continue treatment with pyrimethamine and sulfonamides.9 The correct interpretation of PCR results requires an understanding of the performance and limitations of the test. Most published data were obtained in conventional PCR assays, frequently based on the B1 gene,10 which have an overall sensitivity of only 65%.9,11 Over the past 10 years, new techniques have been developed to increase the performances of PCR tests, through real-time amplification methods12 and the use of more repetitive DNA targets, such as the 529 base pair (bp) fragment described by Homan et al13 and Reischl et al.14 The information available to clinicians concerning the performance of these tests with amniotic fluid for the prenatal diagnosis of congenital toxoplasmosis remains scarce, and is limited to retrospective samples.15,16 We provide data for a large cohort of pregnant women with acute Toxoplasma infection prospectively enrolled at three centers using identical protocols for real-time PCR on amniotic fluid and for the management of pregnant patients and their children.
PATIENTS AND METHODS
This study was found to be exempt from institutional review board approval by the Comite de Protection des Personnes Sud-Est, Hotel-Dieu Hospital Lyon. Pregnant women with acute Toxoplasma infections were enrolled at three French centers (Paris, Lyon, and Marseille) that routinely perform real-time PCR analysis on amniotic fluid for the antenatal diagnosis of Toxoplasma infection. They had been identified during the monthly retesting that is mandatory for all French pregnant women identified as susceptible in the first trimester of pregnancy. The inclusion criteria were proven Toxoplasma infection acquired at an accurately known gestational age, in pregnancies ending between January 1, 2006, and December 31, 2007, and the availability of follow-up data concerning the infection status of the fetus or the child. Case management during pregnancy and delivery conformed to a protocol established by the three centers and described elsewhere.1
In each center, the prenatal diagnosis of fetal infection in pregnant women with confirmed seroconversion was performed prospectively throughout the study period, by PCR analysis of amniotic fluid samples. Ten mL of amniotic fluid were centrifuged (20 mn at 6,000 g) and DNA was extracted from 200-microliter pellets, with the QIAmp DNA mini kit (Qiagen, Courtaboeuf, France) or the High Pure PCR Template Preparation kit (Roche, Meylan, France) according to the manufacturer's instructions. The system was designed to amplify the repetitive 529 bp DNA fragment of T. gondii (Genbank NoAF 487550) through real-time PCR, with detection by fluorescence resonance energy transfer on the LightCycler instrument (Roche Molecular Biochemicals, Meylan, France). Results were classified as positive or negative.
Definitive infectious status was assessed postnatally by serologic follow-up based on immunosorbent agglutination assays for specific immunoglobulin (Ig)A and IgM and on immunofluorescent antibodies techniques or dye test for IgG. Congenital Toxoplasma infection was established by the demonstration of specific IgM, IgA, or IgG synthesis.17 Positive parasitological or histological findings on examination of fetal tissue were considered conclusive for fetal deaths. The absence of congenital toxoplasmosis was defined as negative results in tests for specific IgG antibodies in an untreated child.
The following data were available for analysis: gestational age, in weeks, at maternal infection; types and dates of antiparasitic treatment during pregnancy; results of fetal ultrasound scans; results of amniotic fluid analysis and the time between maternal infection and amniocentesis (if performed) and the definitive infection status of the children.
We estimated sensitivity, specificity, and positive and negative predictive values for PCR results, overall and for the trimester of pregnancy during which maternal seroconversion occurred (first trimester: before 15 weeks of gestation, second trimester from 15 to 26 weeks, and third trimester, after 26 weeks of gestation). We calculated confidence intervals (CIs) for sensitivity and specificity by the binomial exact method. Discrete variables are expressed as numbers of cases and percentages and were analyzed with the χ2 test. For continuous variables, results are expressed as medians with the interquartile range and were compared by Kruskal-Wallis nonparametric tests. P<.05 was considered significant. Calculations were carried out with NCSS2000 software (Statistical Solutions, Kaysville, UT).
A total of 399 mother–child pairs were enrolled over the study period. Twenty of these pairs were excluded owing to uncertainty concerning gestational age at the time of maternal infection (six patients), an absence of pathological examination (four patients: two abortions for toxoplasmosis, one for Down syndrome, and one stillbirth) or loss-to-follow up before the postnatal confirmation of congenital infection status on the basis of reference criteria (10 cases). Only the firstborn of twins was included in the analysis (two sets of twins in the sample). We analyzed a total of 377 mother-child pairs, including three fetal losses, two induced abortions for toxoplasmosis and 372 liveborn children (181 boys and 191 girls) (Fig. 1). In total, 182 patients were enrolled in Lyon, 117 in Paris, and 78 in Marseille.
Seroconversion occurred during the first trimester in 157 patients (41.7%), the second trimester in 126 patients (33.4%), and the third trimester in 94 patients (24.9%). Maternal treatment is shown, by trimester of infection, in Table 1; 277 women received spiramycin only and 62 initially received spiramycin but were transferred subsequently to pyrimethamine-sulfonamide (in most cases [75.8%, 47 of 62] owing to positive fetal diagnosis). In most of the 339 women treated initially or solely with spiramycin, seroconversion occurred in the first (157 of 157) or second (125 of 126) trimester. The median time between the estimated date of infection and the initiation of spiramycin treatment was 4 weeks (2.9–6), with a significantly shorter interval for late infections (Table 1). Pyrimethamine-sulfonamide was prescribed alone in nine women in whom seroconversion occurred in the third (eight cases) or second (one case) trimester. Twenty-nine women (8%) in whom seroconversion occurred during the 8th (16 cases) or 9th month of gestation13 received no treatment.
Congenital infection was diagnosed in five fetuses and 101 liveborn children (106 of 377, 28.1%, 95% CI 4–33%) on the basis of predefined standard criteria. The rates of congenital infection were 4.5% (7 of 157, 95% CI 2–9%) in the first trimester, 31.7% (40 of 126, 95% CI 24–41%) in the second trimester, and 62.8% (59 of 94, 95% CI 52–73%) in the third trimester. The overall rates of congenital infection were 35% (41 of 117, 95% CI 27–44%) in Paris, 26.9% (49 of 182, 95% CI 21–34%) in Lyon and 20.5% (16 of 78, 95% CI 13–31%) in Marseille (P=.08). Infection was diagnosed at birth in 75.2% (76 of 101, 95% CI 60–79%) of children on the basis of specific IgM and IgA detection (48 cases), the detection of IgM only (20 cases), or the detection of IgA only (eight cases). In seven children (6.6%), infection was diagnosed between the ages of 1 and 9 months based on the detection of IgG, IgM, or IgA synthesis. In the remaining children, congenital infection was confirmed by the persistence of specific IgG at the age of 1 year. In total, 21 out of 101 infected neonates (20.8%) had one or more clinical manifestations: 10 had intracranial lesions (calcifications in nine cases and moderate ventricular dilation in one case), nine had retinochoroiditis, and two had both calcifications and retinochoroiditis. Ocular lesions were diagnosed at birth, except one that was diagnosed at 6 months of age. They involved the macula in three children.
Amniocentesis was performed in 261 (69.2%) women, including 79.4% with first-trimester infections, 85.7% of those with second-trimester infections and 27.7% of those with third-trimester infections (Fig. 2). Amniotic fluid was sampled at a median gestational age of 23 weeks (interquartile range 19.1–30.1) overall, 19.1 weeks (interquartile range 18.3–20.3) in first trimester infections, 27.4 weeks (interquartile range 21.7–31.3) in second trimester infections and 36.4 (interquartile range 34.9–38.6) in third trimester infections (Table 1). Most women (257 of 261, 98.5%) underwent amniocentesis at least 4 weeks after seroconversion and the median interval between seroconversion and amniocentesis was 8.6 weeks (interquartile range: 6.7 to 12.1 weeks). The median duration of spiramycin treatment before amniocentesis was 4.3 weeks and the duration of treatment decreased with increasing gestational age at the time of maternal infection (Table 1). Three patients received pyrimethamine-sulfonamide for 3 to 19 days before sampling.
Polymerase chain reaction results were positive for 43 of the 47 (91.5%, 95% CI 79–97%) infected liveborn children for whom amniocentesis was performed. It was also positive for four nonlive births: two in utero deaths at 18 and 19 weeks after first trimester infection with positive histological or parasitological findings on necropsy and two fetuses who underwent abortion at 22 and 28 weeks after maternal infection at respectively 7 and 16 weeks and had confirmed congenital toxoplasmosis at postmortem examination.
Table 2 provides details for the four infected liveborn children for whom PCR analysis of the amniotic fluid was negative. None of them received pyrimethamine-sulfonamide in utero. In these children, the diagnosis of congenital infection was based on the detection of specific IgM at birth, alone or with other criteria; examinations carried out at birth and at the age of 1 year were normal. Polymerase chain reaction results were negative in all 210 noninfected fetuses.
The sensitivity of the PCR test was 92.2% (47 of 51, 95% CI 81–98%) overall, with no significant differences as a function of the trimester during which maternal infection occurred (P=.94). The highest estimate of sensitivity was obtained for infections occurring during the second trimester (96.7%; Table 3). Overall estimates did not differ between centers (P=.64).
The overall negative predictive value was 98.1%. A slight decrease in negative predictive value from the first (99%) to the third trimester (82%) was observed, but this difference was not significant (P=.9). There were no false positive results, so specificity and positive predictive values of 100% were obtained for all trimesters.
Our study was designed to provide clinicians with data concerning the accuracy of real-time PCR analysis on amniotic fluid for the prediction of congenital toxoplasmosis. The parallel inclusion at three centers made it possible to include a large number of patients managed according to the same protocol over a relatively short period. The risk of selection bias was minimized by the prospective enrollment of consecutive patients, prospective data collection and a high rate of follow-up. This enabled us to obtain recent, valid data concerning the routine performance of a 529 bp PCR assay on fresh amniotic fluid carried out in similar conditions at all three sites.
Overall predictive values were satisfactory, as there were only four discordant results, evenly distributed between trimesters. Overall sensitivity was significantly higher than the 64% recorded in our previous study11 and previous estimates from other large multicenter postnatal follow-up studies, which range from 69 to 71%.9,10 Many factors, including the selection of primers and probes, DNA target, reaction volume, amplification and amplicon detection methods, may influence sensitivity.18 Several successive changes to the testing process in recent years may account for the better performance of the PCR technique used here. For example, in this study, the final volume of amniotic fluid used for extraction was five times greater than in our previous study.11 Further improvement may have resulted from the introduction of real-time PCR, which is more reliable and faster than conventional endpoint PCR12 and detects smaller amounts of Toxoplasma DNA, although retrospective analyses of 100 and 30 amniotic fluid samples19,20 reported no clinically significant gain in sensitivity associated with this technique. Further increases in sensitivity were also probably achieved by detecting a 529 bp DNA fragment with a much higher copy number (200–300 copies) than the formerly used B1 gene (35 copies in the parasite genome)13,14,21 and by using fluorescent probes.22 Published reports on the routine performance of 529 bp PCR assays on amniotic fluid have been restricted to comparisons with the B1 gene assay, taking no account of the clinical status of the fetus,23 or have been based on the retrospective re-analysis of samples.15,16 Our study has the advantages of being based on the analysis of a larger number of amniotic fluid samples analyzed without freezing.
All four infected children with negative PCR results for amniotic fluid remained free of symptoms at the age of 1 year, suggesting possible milder infection and lower parasitic load, which would account for the negative results obtained by PCR. The negative results were not a result of the sample being taken too soon after infection because the interval between infection and testing was at least 5 weeks.
Previous studies of conventional PCR assays have reported an effect of gestational age at the time of maternal infection on the sensitivity of the test.9,11 The absence of such an association in our study may be due to the lower rate of false negative results in our study and the relatively small number of fetuses infected through first-trimester and third-trimester infections for which amniocentesis was performed. However, the ability to identify fetuses infected as a consequence of second-trimester infection was also reported in our previous study based on conventional PCR.11 This finding is important, because these fetuses are considered to be exposed to the highest risk of severe immediate lesions if infected24 justifying additional ultrasound examinations as well as pyrimethamine-sulfonamide treatment.25
The rates of infection by trimester in our study were consistent with expected prevalence rates. Our estimates for negative and positive predictive values therefore provide valid indicators for clinical decision-making. A positive PCR result provides proof of fetal infection, making it possible to take appropriate decisions concerning fetal ultrasound surveillance and prenatal and postnatal treatment.
Negative results are also clinically useful, regardless of the trimester during which the mother becomes infected. In first- and second trimester infections they allow to reassure the parents, as they are associated with a 99% probability that the child will not be infected. It may even be possible to stop maternal treatment and decrease fetal ultrasound follow-up. In third-trimester infections they are associated with an 82% probability of the child not having the disease. Pyrimethamine-sulfonamide treatment in such circumstances would unnecessarily expose five fetuses in every six identified to drugs that are known to have side effects. However a negative PCR result does not mean that there is no need to check for the absence of biological and clinical signs at birth and to carry out regular tests to monitor the disappearance of IgG during the first year of life. Such tests are particularly important in cases of late maternal infections owing to the high residual risk of infection, which would be less likely to result in severe immediate lesions compared with congenital infections resulting from seroconversions earlier in pregnancy but as likely to lead to long-term complications.1
Another interesting finding of our study was that 20% and 16% of patients acquiring Toxoplasma infection in the first and second trimesters, respectively, and up to 72% of those infected after 24 weeks, did not undergo amniocentesis, despite the recommendations repeatedly issued in our protocol. We were unable to determine the proportion of such cases in which sampling was refused by the patient and the proportion in which it was not offered by the physician. There may be obstetric reasons for not carrying out amniocentesis: to avoid premature labor26 or cord trauma, or the late recognition of maternal infection (at delivery). The risk of congenital infection is often thought to be higher for infections occurring during the third trimester (this risk was 63% in our study). This might also account for the decision not to risk sampling, but instead to initiate presumptive pyrimethamine- sulfonamide treatment, with the associated risks of treating children that are not actually infected and potentially erasing biological markers of infection at birth in infected neonates. The decision not to carry out late amniocentesis may also have been influenced by reports of an estimated sensitivity significantly lower than that reported here.10,11
Clinicians should be made aware of the excellent performance of current PCR techniques for the analysis of amniotic fluid. These tests provide important information during all three trimesters and negative results can reassure parents, particularly early in gestation, when the fears of severe fetal lesions are highest. Positive results can also be used to guide decisions relating to pyrimethamine-sulfonamide treatment. Polymerase chain reaction testing should therefore be offered whenever obstetrically possible.
Real-time PCR analysis significantly improves the detection of T. gondii on amniotic fluid. It provides an accurate tool to predict fetal infection and decide on appropriate treatment and surveillance. Amniotic fluid sampling should be recommended whenever obstetrically possible. Postnatal follow-up remains however necessary in the first year of life to fully exclude infection in children for whom PCR results were negative.
1. Kieffer F, Wallon M, Garcia P, Thulliez P, Peyron F, Franck J. Risk factors for retinochoroiditis during the first 2 years of life in infants with treated congenital toxoplasmosis. Pediatr Infect Dis J 2008;27:27–32.
2. Wallon M, Kodjikian L, Binquet C, Garweg J, Fleury J, Quantin C, et al. Long-term ocular prognosis in 327 children with congenital toxoplasmosis Pediatrics. 2004;113:1567–72.
3. Foulon W, Villena I, Stray-Pedersen B, Decoster A, Lappalainen M, Pinon JM, et al. Treatment of toxoplasmosis during pregnancy: a multicenter study of impact on fetal transmission and children's sequelae at age 1 year. Obstet Gynecol 1999;180:410–5.
4. Gras L, Wallon M, Pollak A, Cortina-Borja M, Evengard B, Hayde M, et al. Association between prenatal treatment and clinical manifestations of congenital toxoplasmosis in infancy: a cohort study in 13 European centres. Acta Paediatr 2005;94:1721–31.
5. SYROCOT (Systematic Review on Congenital Toxoplasmosis) study group, Thiébaut R, Leproust S, Chêne G, Gilbert R. Effectiveness of prenatal treatment for congenital toxoplasmosis: a meta-analysis of individual patients' data. Lancet 2007;369:115–22.
7. Berger F, Goulet V, Le Strat Y, Desenclos JC. Toxoplasmosis in pregnant women in France: trends in seroprevalence and incidence, and associated factors, 1995–2003 [in French]. Bull Epidemiol Hebd 2008;14–5:117–21.
8. Jones JL, Kruszon-Moran D, Wilson M. Toxoplasma gondii infection in the United States, 1999–2000. Emerg Infect Dis 2003;9:1371–4.
9. Thalib L, Gras L, Romand S, Prusa A, Bessieres MH, Petersen E, et al. Prediction of congenital toxoplasmosis by polymerase chain reaction analysis of amniotic fluid. BJOG 2005;112:567–74.
11. Romand S, Wallon M, Franck J, Thulliez P, Peyron F, Dumon H. Prenatal diagnosis using polymerase chain reaction on amniotic fluid for congenital toxoplasmosis. Obstet Gynecol 2001;97:296–300.
12. Costa JM, Pautas C, Ernault P, Foulet F, Cordonnier C, Bretagne S. Real-time PCR for diagnosis and follow-up of Toxoplasma reactivation after allogeneic stem cell transplantation using fluorescence resonance energy transfer hybridization probes. J Clin Microbiol 2000;38:2929–32.
13. Homan WL, Vercammen M, De Braekeleer J, Verschueren H. Identification of a 200- to 300-fold repetitive 529 bp DNA fragment in Toxoplasma gondii
, and its use for diagnostic and quantitative PCR. Int J Parasitol 2000;30:69–75.
14. Reischl U, Bretagne S, Krüger D, Ernault P, Costa JM. Comparison of two DNA targets for the diagnosis of Toxoplasmosis
by real-time PCR using fluorescence resonance energy transfer hybridization probes. BMC Infect Dis 2003;3:7.
15. Filisetti D, Gorcii M, Pernot-Marino E, Villard O, Candolfi E. Diagnosis of congenital toxoplasmosis: comparison of targets for detection of Toxoplasma gondii
by PCR. J Clin Microbiol 2003;41:4826–8.
16. Kasper DC, Sadeghi K, Prusa AR, Reischer GH, Kratochwill K, Förster-Waldl E, et al. Quantitative real-time polymerase chain reaction for the accurate detection of Toxoplasma gondii
in amniotic fluid. Diagn Microbiol Infect Dis 2009;63:10–5.
17. Lebech M, Joynson DH, Seitz HM, Thulliez P, Gilbert RE, Dutton GN, et al. Classification system and case definitions of Toxoplasma gondii infection in immunocompetent pregnant women and their congenitally infected offspring. European Research Network on Congenital Toxoplasmosis. Eur J Clin Microbiol Infect Dis 1996;15:799–805.
18. Bastien P, Procop GW, Reischl U. Quantitative real-time PCR is not more sensitive than “conventional PCR.” J Clin Microbiol 2008;46:1897–900.
19. Costa JM, Ernault P, Gautier E, Bretagne S. Prenatal diagnosis of congenital toxoplasmosis by duplex real-time PCR using fluorescence resonance energy transfer hybridization probes. Prenat Diagn 2001;21:85–8.
20. Ordinaire I, Simon A, Fréalle E, Soula F, Valat AS, Rouland V, et al. Real-time quantitative PCR for toxoplasmosis diagnosis [in French]. Ann Biol Clin (Paris) 2005;63:67–73 French.
21. Edvinsson B, Lappalainen M, Evengård B, ESCMID Study Group for Toxoplasmosis. Real-time PCR targeting a 529-bp repeat element for diagnosis of toxoplasmosis. Clin Microbiol Infect 2006;12:131–6.
22. Nagy B, Bán Z, Beke A, Nagy GR, Lázár L, Papp C, et al. Detection of Toxoplasma gondii
from amniotic fluid, a comparison of four different molecular biological methods. Clin Chim Acta 2006;368:131–7.
23. Cassaing S, Bessières MH, Berry A, Berrebi A, Fabre R, Magnaval JF. Comparison between two amplification sets for molecular diagnosis of toxoplasmosis by real-time PCR [published erratum appears in J Clin Microbiol 2006;44:4295]. J Clin Microbiol 2006;44:720–4.
24. Dunn D, Wallon M, Peyron F, Petersen E, Peckham C, Gilbert R. Mother-to-child transmission of toxoplasmosis: risk estimates for clinical counselling. Lancet 1999;353:1829–33.
25. Romand S, Chosson M, Franck J, Wallon M, Kieffer F, Kaiser K, et al. Usefulness of quantitative polymerase chain reaction in amniotic fluid as early prognostic marker of fetal infection with Toxoplasma gondii.
Am J Obstet Gynecol 2004;190:797–802.
26. Zalud I, Janas S. Risks of third-trimester amniocentesis. J Reprod Med 2008;53:45–8.
This article has been cited 1 time(s).
Clinical Infectious DiseasesCongenital Toxoplasma Infection: Monthly Prenatal Screening Decreases Transmission Rate and Improves Clinical Outcome at Age 3 YearsClinical Infectious Diseases
© 2010 The American College of Obstetricians and Gynecologists