Human cytomegalovirus (HCMV) long-term shedding and HCMV-specific immune response in pregnant women with primary HCMV infection.

Human cytomegalovirus (HCMV) shedding has been extensively investigated in newborns and in young children, however, much less is known about it in immunocompetent adults. Shedding of HCMV was investigated in saliva, vaginal secretions and urine of pregnant women experiencing primary infection along with the development of the HCMV-specific immune response. Thirty-three pregnant women shed HCMV DNA in peripheral biological fluids at least until one year after onset of infection, while in blood HCMV DNA was cleared earlier. Significantly higher levels of viral load were found in vaginal secretions compared to saliva and urine. All subjects examined two years after the onset of infection showed a high avidity index, with IgM persisting in 36% of women. Viral load in blood was directly correlated with levels of HCMV-specific IgM and inversely correlated with levels of IgG specific for the pentameric complex gH/gL/pUL128L; in addition, viral load in blood was inversely correlated with percentage of HCMV-specific CD4+ and CD8+ expressing IL-7R (long-term memory, LTM) while viral load in biological fluids was inversely correlated with percentage of HCMV-specific CD4+ and CD8+ effector memory RA+(TEMRA). In conclusion, viral shedding during primary infection in pregnancy persists in peripheral biological fluids for at least one year and the development of both antibodies (including those directed toward the pentameric complex) and memory T cells are associated with viral clearance.

Virologic and immunologic monitoring of cytomegalovirus to guide preemptive therapy in solid-organ transplantation.

Control of human cytomegalovirus (HCMV) infection during the posttransplant period was investigated in 134 solid-organ transplant recipients by monitoring in parallel virologic and immunologic parameters for at least 1 year of follow-up. Virologic monitoring was achieved by determining HCMV DNAemia with real-time PCR, using the threshold of 300 000 DNA copies/mL blood as a cutoff for starting preemptive therapy. Immunologic monitoring included measurement of HCMV-specific CD4+ and CD8+ T cells by cytokine flow cytometry, using HCMV-infected dendritic cells as a stimulus. HCMV infection was diagnosed in 110 (82%) and required treatment in 49 (36%) patients. At 12 months after transplantation 'protective' immunity (≥0.4 CD4+ and CD8+ HCMV-specific T cells/µL blood) was achieved in 115/129 (89%) patients. During the entire study period, 122 patients reconstituting HCMV-specific CD4+ and CD8+ T-cell immunity at 60 days posttransplant onward were able to control HCMV infection, except for one patient who developed HCMV disease because of a rejection episode. Patients reconstituting HCMV-specific CD8+ only did not control HCMV infection. In conclusion, the presence of both HCMV-specific CD4+ and CD8+ T cells ≥ 0.4/µL blood appears to be protective against HCMV disease. This result does not apply to patients undergoing antirejection treatment, or reconstituting HCMV-specific CD8+ T cells only.

Prognostic markers of symptomatic congenital human cytomegalovirus infection in fetal blood.

OBJECTIVE: To identify fetal cord blood prognostic markers of symptomatic congenital human cytomegalovirus infection (HCMV). DESIGN: Retrospective observational study. SETTING: Fetal medicine unit in Milan and Medical virology unit in Pavia, Italy. POPULATION: HCMV-infected and -uninfected fetuses of mothers with primary HCMV infection during the period 1995-2009. METHODS: Overall, 94 blood samples from as many fetuses of 93 pregnant women experiencing primary HCMV infection were examined for multiple immunological, haematological and biochemical markers as well as virological markers. Congenital HCMV infection was diagnosed by detection of virus in amniotic fluid, and symptomatic/asymptomatic infections were determined by ultrasound scans, nuclear magnetic resonance imaging, histopathology or clinical examination at birth. Blood sample markers were retrospectively compared in symptomatic and asymptomatic fetuses with congenital infection. MAIN OUTCOME MEASURES: A statistical analysis was performed to determine the value of each parameter in predicting outcome. RESULTS: Univariate analysis showed that most nonviral and viral markers were significantly different in symptomatic (n = 16) compared with asymptomatic (n = 31) fetuses. Receiver operator characteristics analysis indicated that, with reference to an established cutoff for each marker, the best nonviral factors for differentiation of symptomatic from asymptomatic congenital infection were ß(2) -microglobulin and platelet count, and the best virological markers were immunoglobulin M antibody and DNAaemia. ß(2) -Microglobulin alone or the combination of these four markers reached the optimal diagnostic efficacy. CONCLUSIONS: The determination of multiple markers in fetal blood, following virus detection in amniotic fluid samples, is predictive of perinatal outcome in fetuses with HCMV infection.

Validation of a DNAemia cutoff for preemptive therapy of cytomegalovirus infection in adult hematopoietic stem cell transplant recipients.

A randomized trial comparing a DNAemia cutoff of 10 000 copies per ml whole blood and first pp65 antigenemia positivity for initiation of preemptive therapy of human cytomegalovirus (HCMV) infection in adult hematopoietic stem cell transplant recipients was completed. DNAemia was chosen for cutoff definition since it is more automatable and standardizable than antigenemia, and more closely reflects the actual viral replication. The primary end point of the study was to compare the number of patients treated in the two arms. A total of 83 patients (42 in the DNAemia, and 41 in the antigenemia arm) were enrolled in the study. The incidence of HCMV infection, as detected by the relevant randomization assay (76% in the DNAemia versus 85% in the antigenemia arm), was comparable in the two arms, whereas the number of patients treated was significantly lower in the DNAemia arm (63 versus 80%, P=0.02). A single patient in the DNAemia arm suffered from biopsy-proven HCMV gastric disease diagnosed in the absence of detectable virus in blood. The incidence of graft-versus-host disease, and transplantation-related mortality did not differ between the two arms. In conclusion, our study shows that the use of a cutoff significantly reduces the number of patients requiring antiviral treatment, thus sparing unnecessary drug administration.

Neutralizing and ELISA IgG antibodies to human cytomegalovirus glycoprotein complexes may help date the onset of primary infection in pregnancy.

BACKGROUND: Definition of onset for primary human cytomegalovirus (HCMV) infection during pregnancy is critical for several reasons, including diagnosis of pre-conceptional infections and definition of gestational age at the time of infection. OBJECTIVE: To determine the onset of primary HCMV infection, differential kinetics of antibodies neutralizing infection of epithelial and fibroblast cells, as well as ELISA IgG antibodies to HCMV glycoprotein complexes (gC) gH/gL/pUL128L, gH/gL/gO, and gB were exploited and compared with conventional assays. STUDY DESIGN: In a series of 40 pregnant women with primary HCMV infection and ascertained HCMV-related mild clinical symptoms, the kinetics of different types of neutralizing and ELISA IgG antibodies were investigated with the aim of establishing criteria for dating the onset of primary infection in pregnant women without clinical symptoms. RESULTS: IgG antibodies to gB and gH/gL/pUL128L, as well as antibodies neutralizing infection of epithelial cells appeared early after infection onset (within 2-3 weeks) and increased rapidly, whereas antibodies to gH/gL/gO and antibodies neutralizing infection of fibroblasts appeared later (>30 days) and increased slowly. Both the conventional diagnostic assays (IgG, and IgM antibody, and IgG avidity index) and the novel assays for determination of antibody responses directed against HCMV gC allowed the definition of an algorithm indicating the onset of primary HCMV infection in asymptomatic women within a period of 1-2 months. CONCLUSION: New neutralization and ELISA IgG assays to HCMV gC provide additional tools for dating the onset of primary infection in pregnancy.

HCV genotyping by three methods: analysis of discordant results based on sequencing.

BACKGROUND: Correct genotyping of hepatitis C virus (HCV) RNA-positive serum samples may have important clinical and therapeutic implications. OBJECTIVES: Three methods were compared to improve accuracy of HCV genotyping. STUDY DESIGN: A panel of 144 HCV RNA-positive sera prospectively tested by a modified Okamoto's type-specific reverse transcription-nested polymerase chain reaction (RT-nPCR) (Okamoto H, Tokita H, Sakamoto M, Kojima M, Iizuka H, Mishiro S. J Gen Virol 1993; 74: 2385-2390) was retrospectively analyzed by two recently described methods which were reported to identify all HCV types and the majority of HCV subtypes: (i) a restriction fragment length polymorphism (RFLP) analysis of PCR products amplified from the 5' untranslated region (5'UTR) of the viral genome (Pohjanpelto P, Lappalainen M, Widell A, Asikainen K, Paunio M. Clin Diagn Virol 1996; 7: 7-16); and (ii) a type-specific RT-nPCR relevant to the core region (Ohno T, Mizokami M, Wu R, Saleh M, Ohba K, Orito E, Mukaide M, Williams R, Lau J. J Clin Microbiol 1997; 35: 201-207). The panel (according to results given by the modified Okamoto's method) consisted of: (i) 105 sera belonging to five different HCV subtypes; (ii) 20 specimens containing a mixture of > or = 2 genotypes; and (iii) 19 untypeable clinical samples. RESULTS: There was agreement of the three methods for 78/144 (54.2%) blood samples, whereas discordant results were obtained for the remaining 66 samples, 56 of which could be typed by sequencing. Of these, 51 (91.7%) were correctly typed by RFLP, 37 (66.0%) by Ohno's and 27 (48.2%) by the modified Okamoto's procedure. The overall genotyping sensitivity of each method over the total number of 134 samples whose genotype was ascertained, was 96.2% for RFLP, 85.8% for Ohno's and 78.3% for the modified Okamoto's procedure. CONCLUSIONS: RFLP analysis, notwithstanding some limitations in subtyping efficiency of genotype 1 samples, appears superior to the two RT-nPCR methods because: (i) it is able to type a larger number of samples; (ii) it is more efficient in identifying genotypes 2a/c, which are widespread in Italy; (iii) it is highly sensitive (together with Ohno's method) in recognizing genotypes 3 and 4.

Quantification of human cytomegalovirus DNA in peripheral blood polymorphonuclear leukocytes of immunocompromised patients by the polymerase chain reaction.

Human cytomegalovirus (HCMV) DNA amplification by the polymerase chain reaction (PCR) was utilized previously for successful monitoring of HCMV infections in immunocompromised patients. However, analysis of an extended series of clinical samples revealed the relatively frequent presence of PCR inhibitors. Hence, the need for availability of an internal control of the reaction allowing identification of false negative results. Similarly, an internal standard appeared necessary for quantification of viral DNA in clinical samples. For this purpose, we constructed a recombinant DNA molecule which could be amplified by the same set of primers used for HCMV DNA amplification. Coamplification of the recombinant DNA molecule and clinical samples proved to be a simple and reliable method for verifying sample competence for amplification. In addition, coamplification of serial known amounts of the same molecule, used as internal standard, and test sample, allowed quantification of viral DNA in polymorphonuclear leukocyte samples. Quantitative monitoring of HCMV infection and antiviral treatment may provide critical indications as to whether and when to initiate or discontinue antiviral treatment in immunocompromised patients with systemic HCMV infections.

Rapid detection of enteroviral RNA in cerebrospinal fluid (CSF) from patients with aseptic meningitis by reverse transcription-nested polymerase chain reaction.

The aim of this study was to compare conventional enterovirus isolation with rapid detection of enteroviral RNA by a reverse transcription-nested polymerase chain reaction (RT-nPCR) method amplifying the 5' nontranslated region of the enteroviral genome in specimens from patients with aseptic meningitis. Reference enterovirus strains and clinical enterovirus isolates were analyzed to evaluate assay sensitivity and specificity. All known enteroviral serotypes tested, but one (echovirus type 22), were detected by RT-nPCR. A series of unrelated viral isolates as well as CSF samples from patients with meningitis/encephalitis or neurological syndromes unrelated to enterovirus infection were included as controls. A total of 47 specimens (31 CSF, 12 rectal swabs, 4 throat swabs) from 30 patients with aseptic meningitis were available for the study. Of the 31 CSF samples tested from 30 patients, 17 from 17 patients (54.8%) were positive by RT-nPCR, while only 10 from 10 patients (32.2%) were positive by culture. Thus, RT-nPCR allowed diagnosis of enterovirus meningitis in 7 additional patients compared to cell culture. The cytopathic effect was observed 5-15 days after inoculation of CSF specimens onto cell cultures, while direct detection of viral RNA in CSF samples by RT-nPCR permitted diagnosis of enteroviral meningitis within 1-2 days. On the whole, viral isolation was positive in 12/47 (25.5%) specimens, whereas viral RNA was detected by RT-nPCR in 11 additional samples (23/47, 48.9%). Specimens of the control group were consistently negative by both viral isolation and RT-nPCR. Restriction endonuclease analysis of PCR products (RFLP) was applied to differentiate poliovirus (PV) from non-polio enteroviruses (NPEV). All enterovirus strains detected in clinical samples (n = 23) were identified as NPEV by RFLP. Clinical isolates were typed by neutralization as echovirus type 30 (n = 6), while 6 were not typed. In conclusion, detection of enteroviral RNA in CSF by RT-nPCR allows: i) rapid diagnosis of enteroviral meningitis; ii) increased sensitivity with respect to virus isolation; iii) differentiation between PV and NPEV infections of the central nervous system.

Diagnostic significance and clinical impact of quantitative assays for diagnosis of human cytomegalovirus infection/disease in immunocompromised patients.

In recent years several assays have been developed for quantitation of human cytomegalovirus (HCMV) in blood of immunocompromised (transplanted and AIDS) patients. It is currently agreed that the only reliable indication of the degree of dissemination of HCMV infection/disease is the measurement of HCMV in blood. Diagnosis of HCMV end-organ disease (organ localizations) often does not benefit from quantitation of virus in blood, but requires detection and quantification of virus in samples taken locally. The most important and clinically useful diagnostic assays for HCMV quantitation in blood are: i) viremia, quantifying infectious HCMV carried by peripheral blood leukocytes (PBL); ii) pp65-antigenemia, quantifying the number of PBL positive for HCMV pp65 in the nucleus; iii) circulating cytomegalic endothelial cell (CEC) viremia (CEC-viremia) measuring the number of circulating CEC carrying infectious HCMV (during the antigenemia assay); iv) leuko- and plasma-DNAemia, quantifying the number of HCMV genome equivalents present in PBL or plasma, respectively, by quantitative polymerase chain reaction (Q-PCR). Other less widely used assays are: i) determination of immediate early and late gene transcripts (mRNA) to detect active viral infection; ii) in situ hybridization to detect viral nucleic acid (DNA or RNA) in tissue sections or cell smears; iii) in situ PCR to detect a low DNA copy number in single cells. Monitoring of HCMV infection/disease in transplant recipients and AIDS patients has established threshold values for different assays above which HCMV-related clinical symptoms are likely to appear. These values are approximately 10 for viremia, 100 for antigenemia and 1,000 GE for leukoDNAemia, and are valid for both solid organ and bone marrow transplant recipients as well as AIDS patients, whereas presence of even a single circulating CEC is sufficient to suggest the presence of a disseminated HCMV infection with potential organ involvement. Monitoring of antiviral treatment of HCMV infection/disease with either ganciclovir or foscarnet has aimed at keeping virologic parameters below the threshold values reported above. On the other hand, rising levels of the same virologic parameters during antiviral treatment have mostly revealed emergence of resistant HCMV strains to either ganciclovir (mutations in the UL97 or DNA polymerase gene) or foscarnet (mutations in the UL54 gene) or both drugs (double resistance with both types of mutations). Rapid assays for chemosensitivity testing of virus directly in clinical specimens have been developed to allow timely (4-6 days) detection of resistance to a drug and provide clinicians with the rationale for shifting to an alternative treatment.

Preemptive therapy of EBV-related lymphoproliferative disease after pediatric haploidentical stem cell transplantation.

The treatment of Epstein-Barr virus (EBV)-related post-transplant lymphoproliferative disease (PTLD) after hematopoietic stem cell transplantation (HSCT) is still unsatisfactory. We conducted a prospective trial to evaluate the impact of routine EBV surveillance and preemptive treatment with the anti-CD20 monoclonal antibody rituximab on the development of PTLD in pediatric recipients of extensively T-cell depleted HSCT from an HLA-haploidentical relative. Twenty-seven patients were included in the surveillance program, 12 developed EBV DNA positivity, with 8 of 12 presenting with sustained viral DNA levels requiring treatment with rituximab. Treatment was well tolerated, and induced clearance of EBV DNA in all patients. However, 4/8 patients showed a new increase in EBV load, coincident with the emergence of CD20(-)/CD19(+) B cells in peripheral blood, accompanied by overt PTLD in 3 patients. The latter cleared PTLD after receiving donor EBV-specific cytotoxic T-lymphocytes (CTLs), and persist in remission at a median 30-month follow-up. EBV-specific T-cell frequency, undetectable at time of EBV DNA positivity, was restored by T-cell therapy to levels comparable with controls. We conclude that preemptive therapy with rituximab is safe, but only partly effective in haplo-HSCT recipients. Patients who progress to PTLD under rituximab treatment can be rescued permanently by infusion of EBV-specific CTLs.

Postinfectious inflammatory disorders: subgroups based on prospective follow-up.

BACKGROUND: Acute disseminated encephalomyelitis (ADEM) refers to a monophasic acute multifocal inflammatory CNS disease. However, both relapsing and site-restricted variants, possibly associated with peripheral nervous system (PNS) involvement, are also observed, and a systematic classification is lacking. OBJECTIVE: To describe a cohort of postinfectious ADEM patients, to propose a classification based on clinical and instrumental features, and to identify subgroups of patients with different prognostic factors. METHODS: Inpatients of a Neurologic and Infectious Disease Clinic affected by postinfectious CNS syndrome consecutively admitted over 5 years were studied. RESULTS: Of 75 patients enrolled, 60 fulfilled criteria for ADEM after follow-up lasting from 24 months to 7 years. Based on lesion distribution, patients were classified as encephalitis (20%), myelitis (23.3%), encephalomyelitis (13.3%), encephalomyeloradiculoneuritis (26.7%), and myeloradiculoneuritis (16.7%). Thirty patients (50%) had a favorable outcome. Fifteen patients (25%) showed a relapsing course. Poor outcome was related with older age at onset, female gender, elevated CSF proteins, and spinal cord and PNS involvement. All but two patients received high-dose steroids as first-line treatment, with a positive response in 39 (67%). Ten of 19 nonresponders (53%) benefited from high-dose IV immunoglobulin; 9 of 10 had PNS involvement. The data were not controlled. CONCLUSIONS: A high prevalence of "atypical variants" was found in this series, with site-restricted damage or additional peripheral nervous system (PNS) involvement. Prognosis and response to steroids were generally good, except for some patient subgroups. In patients with PNS involvement and steroid failure, a favorable effect of IV immunoglobulin was observed.

Comparative quantitation of human cytomegalovirus DNA in blood leukocytes and plasma of transplant and AIDS patients.

A new method for the quantitation of human cytomegalovirus (HCMV) DNA was used to determine the levels of viral DNA in parallel in 120 blood leukocyte (leukoDNAemia) and plasma (plasmaDNAemia) samples from 8 heart or heart-lung transplant patients and 17 AIDS patients with disseminated HCMV infection. PlasmaDNAemia was consistently associated with leukoDNAemia in both groups of patients. However, at least in the transplant patients, plasmaDNAemia was not necessarily associated with clinical symptoms, appearing later and disappearing earlier than leukoDNAemia during the course of infection. Quantitative mean levels of leukoDNAemia were mostly higher than those of plasmaDNAemia in both transplant and AIDS patients. However, in the absence of antiviral treatment, plasmaDNAemia levels were significantly higher in AIDS patients than in transplant recipients, whereas leukoDNAemia levels were not significantly different between the two groups of patients. A significant correlation was found between leukoDNAemia and plasmaDNAemia in AIDS patients, as well as in transplant recipients, although to a lesser degree. However, from a diagnostic standpoint, quantitative determination of plasmaDNAemia appears to represent a much less sensitive parameter than that of leukoDNAemia (or antigenemia) for monitoring HCMV infections and antiviral treatment.

Detection of human cytomegalovirus DNA: how, when and where?

Although several methods have been utilized for the detection and quantification of human cytomegalovirus (HCMV) DNA, all of them can be divided into three groups: (i) detection of HCMV DNA directly in tissues by in situ hybridization or in situ polymerase chain reaction (PCR); (ii) detection of HCMV DNA in cell or tissue lysates by hybridization with DNA or RNA probes differently labelled-labels were progressively modified in order to provide an increasing sensitivity (hybridization products were revealed by radioactive, colorimetric or chemiluminescent procedures); (iii) detection of HCMV DNA in cell or tissue lysates by qualitative (single-step and nested) and quantitative (semiquantitative, competitive or noncompetitive) PCR. The selection of the methods to be employed depends primarily on the clinical situation which must be evaluated. Clinical samples for HCMV genome detection must vary accordingly.

Congenital rubella infection following rubella outbreak in northern Italy, 2002: need for an effective vaccination programme.

Presented here are the details of a rubella outbreak that occurred in 2002 in the Lombardy region of northern Italy followed by a discussion of rubella vaccination policy in this country. From 13 maternal cases of rubella infection, congenital rubella infection was diagnosed in three fetuses and three newborns. Of the three infected fetuses, one was aborted and two died in utero, while of the three infected newborns, two were born with severe disease and one was subclinically infected. Follow-up revealed that one of the two symptomatic newborns had died at 4 months of age with disseminated rubella infection, while the other suffered from bilateral blindness and deafness and was severely retarded at 15 months of age. The remaining infant remained asymptomatic at 14 months. Congenital rubella remains a serious health problem in Italy and a successful vaccination strategy is required.

Diagnosis of human cytomegalovirus infection of the nervous system by pp65 detection in polymorphonuclear leukocytes of cerebrospinal fluid from AIDS patients.

Detection of human cytomegalovirus (HCMV) pp65 in cerebrospinal fluid (CSF) polymorphonuclear leukocytes (PMNL) was investigated in 32 CSF samples from 25 AIDS patients with neurologic symptoms. HCMV pp65-positive PMNL were detected in 9 CSF samples from 3 patients with polyradiculoneuropathy (frequency, 1.2%-20.8%) and 1 with encephalitis (frequency, 56%). pp65-positive PMNL appeared hypersegmented compared with peripheral blood PMNL. HCMV was isolated from 3 CSF samples from 2 patients with polyradiculoneuropathy. Retrospective quantitation of viral DNA by polymerase chain reaction (PCR) was in agreement with pp65 detection in 27 of 28 samples tested (96.4% concordance). Specificity and sensitivity of pp65 assay versus PCR were 100% (17/17 negative samples) and 90.9% (10/11 positive samples), respectively. Examination of CSF for presence of pp65-positive PMNL appears to be a sensitive and specific test for diagnosis of HCMV infections of the nervous system.

Polioviruses with natural recombinant genomes isolated from vaccine-associated paralytic poliomyelitis.

To determine how oral poliovirus vaccine (Sabin) strains evolve during replication in humans and to confirm the etiology of vaccine-associated paralytic poliomyelitis (VAPP), we examined 70 vaccine-derived strains isolated from VAPP cases. Two distant sequences of the poliovirus genome were targeted for a double restriction fragment length polymorphism assay (RFLP) of reverse-transcribed genomic segments amplified by PCR, an extension of the method that we described previously (Balanant et al., 1991). One (RFLP-1) was a 480-long nucleotide sequence coding for the N-terminal part of the VP1 capsid polypeptide, situated in the 5' third of the viral genome (nucleotides 2401-2880). The other (RFLP-3D1) was a 291-long nucleotide sequence coding for a part of the viral polymerase, situated near the 3' end of the genome (nucleotides 6086-6376). Strain-specific restriction profiles could be generated for different field isolates by using three restriction enzymes in each case: HaeIII, DdeI, and HpaII for RFLP-1 and HaeIII, DdeI and RsaI for RFLP-3D1. With few exceptions, the vaccine-specific RFLP profiles were found to be conserved in both regions during replication of these viruses in humans. Thus, RFLP could be used as a marker so as to identify the origin of viral isolates at both ends of their genome. Whether viral isolates were vaccine-derived was determined by using strain-specific monoclonal antibodies and RFLP-1. Among the 70 isolates, 21 of the 43 type 2 strains and 15 of the 22 type 3 strains had a recombinant genome. None of the 5 type 1 Sabin-derived isolates was found to be recombinant. Both intertypic vaccine/vaccine and vaccine/non-vaccine recombinants were detected. Partial nucleotide sequencing confirmed the RFLP results in all cases that were investigated. In one case, it was possible to predict the recombination junction site from the restriction profiles. This site was more precisely localized by sequencing. The C6203 > U nucleotide substitution, which is suspected to contribute to the reversion toward neurovirulence of the attenuated Sabin 1 strain, was detected in almost all the recombinant genomes containing Sabin 1-specific sequences at the 3' extremity. This mutation was detected by identification of the modified RsaI profile in the RFLP-3D1. The results presented in this paper suggest that recombination, alone or together with mutation, might be one of the mechanisms of the reversion toward neurovirulence of attenuated vaccine strains and of the natural evolution of poliovirus.