Molecular studies of cerebrospinal fluid in human immunodeficiency virus type 1 associated opportunistic central nervous system diseases an update

Journal of NeuroVirology, 8(suppl. 2): 122 128, 2002 c 2002 Taylor & Francis ISSN 1355 0284/02 $12.00+.00 DOI: 10.1080/13550280290167957 Molecular studies of cerebrospinal fluid in human immunodeficiency virus type 1 associated opportunistic central nervous system diseases an update Paola Cinque, Simona Bossolasco, Arabella Bestetti, Serena Sala, Chiara Pierotti, and Adriano Lazzarin Clinic of Infectious Diseases, San Raffaele Scientific Institute, Milano, Italy Although the incidence of opportunistic central nervous system (CNS) diseases has markedly declined in developed countries following the advent of highly active antiretroviral therapies (HAARTs), they still represent a major diagnostic and therapeutic challenge over the world. The application of nucleic acid amplification techniques to the study of cerebrospinal fluid (CSF) has contributed substantially to their diagnosis. The detection of specific microbial genomes in the CSF is now the preferred test for some CNS opportunistic diseases, such as progressive multifocal leukoencephalopathy or cytomegalovirus encephalitis. More recent developments of these techniques are the quantitative amplification techniques and postamplification studies. Quantification of nucleic acids in CSF is an important aid both at the time of diagnosis, for the interpretation of positive findings, and during patient follow-up. Postamplification analyses can provide important information with regard to clinical patient management, e.g., detection of genotypic resistance to antimicrobial drugs, and in the attempt to elucidate disease epidemiology and pathogenesis. Journal of NeuroVirology (2002) 8(suppl. 2), 122 128. Keywords: cerebrospinal fluid; HIV; nucleic acid amplification; opportunistic infections; polymerase chain reaction; progressive multifocal leukoencephalopathy CNS opportunistic diseases Opportunistic diseases of the central nervous system (CNS) are a major cause of mortality in human immunodeficiency virus (HIV)-infected persons. These are caused by viruses, including herpesviruses cytomegalovirus (CMV), herpes simplex virus type 1 or 2 (HSV-1, HSV-2), varicella-zoster virus (VZV) and the polyomavirus JC virus (JCV); bacteria and mycobacteria, such as Mycobacterium tuberculosis; and fungi or parasites, the most frequent of which are Cryptococcus neoformans and Toxoplasma gondii. Address correspondence to Paola Cinque, Clinic of Infectious Diseases, San Raffaele Hospital, Via Stamira d Ancona, 20, 20127 Milano, Italy. E-mail: paola.cinque@hsr.it Received 23 August 2002; revised 12 September 2002; accepted 20 September 2002. In addition, non-hodgkin lymphomas (NHLs) frequently localize within the brain. With some prevalence differences, depending on geographic areas, CMV encephalitis (CMV-E) and CNS toxoplasmosis have been reported to be the most frequent opportunistic CNS diseases, followed by lymphoma, cryptococcosis, and progressive multifocal leukoencephalopathy (PML). CNS diseases have significantly decreased in both incidence and prevalence in developed countries, following the advent of highly active antiretroviral therapies (HAARTs) (Sacktor et al, 2001); however, they still represent a great threat among the majority of HIV-infected patients over the world, who have no or only limited access to treatments. On the other hand, CNS opportunistic infections are still frequently observed in the western world, as first acquired immunodeficiency syndrome (AIDS) manifestation in antiretroviralnaïve patients, in patients who fail to respond or are

P Cinque et al 123 not adherent to treatment, and even in patients showing optimal immunological and virological response to therapy, as observed in some cases of PML. During the last decade, remarkable progress has been done in the fields of diagnostics and therapeutics for opportunistic CNS diseases. Already in the pre-haart era, the development and licensing of new drugs, as well as the assessment of prophylaxis and treatment strategies in clinical trials, have substantially contributed to control of these diseases. On the other hand, development and clinical validation of new diagnostic procedures, first of all those based on molecular amplification of microbial genomes in the cerebrospinal fluid (CSF), have enabled early and specific diagnosis of the majority of these infection. Nucleic acid amplification techniques in CSF and most recent developments Molecular amplification techniques are based on in vitro amplification of even small quantities of target nucleic acid molecules to considerably larger amounts (over 10 6 copies), which can be visualized by common laboratory procedures. The polymerase chain reaction (PCR) is the most popular amplification method, but a number of other techniques have been applied to CSF, including the nucleic acid sequence based amplification (NASBA), the branched DNA, and the hybrid capture assay. The great advantages of these methods have been their extraordinary sensitivity and their rapidity. These features, upon application to CSF, have been crucial for a rapid and etiological diagnosis of HIV-related opportunistic CNS diseases. In contrast, conventional diagnostic methods, e.g., microscopy and culture of CSF, or antigen detection, are less sensitive and diagnostically useful only in some diseases, such as cryptococcal meningitis. The demonstration of a specific intrathecal antibody synthesis is also of little diagnostic help in AIDS patients, whereas brain biopsy, because of its invasiveness, is usually not well accepted by both patients and their physicians. Since the first descriptions at the beginning of the 1990s, there has been important technical improvement of nucleic acid amplification methods in CSF analysis. Efforts have been spent to improve sensitivity and rapidity. Techniques other than PCR, such as nucleic acid sequence based amplification (NASBA) or branched DNA, have been applied to CSF, and costand time-saving strategies have often be adopted, e.g., multiplex PCR assays, which enable simultaneous assessment of different genomes. In addition, quantitative amplification methods have been developed for measuring nucleic acids amounts in CSF, and molecular techniques for accurate characterization of microbial genomes following their amplification in CSF, have become largely accessible. Quantitative methods are a precious support at the time of diagnosis, and also useful for subsequent patient follow-up. A variety of semiquantitative and quantitative nucleic acid amplification methods are in use (Preiser et al, 2000). The former include those based on limiting dilution of samples before amplification, or comparison of the extent of amplification between samples and external standards at known nucleic acid concentration. The main disadvantage of these procedures is that they do not take into account the possible differences in amplification efficiency between samples and/or standards. Quantitative techniques allow a more accurate estimate of nucleic acids levels, through coamplification in the same tube of the target and an internal standard at known concentration, which enables control of amplification efficiency. New automated real-time PCR protocols, such as those based on the TaqMan and the LightCycler technology, are further improving both accuracy and dynamic range of quantification. Compared to classical methods, where the amounts of DNA are measured at the end of amplification, when amplification efficiency is reduced, in real-time PCR, the products of amplification are recorded during its exponential phase, where the amount of PCR product correlates to the initial quantity of template. This is accomplished by the use of probes that generate a fluorescent signal at each cycle of amplification, upon hybridization with the target molecules (Bustin, 2000). Postamplification analyses for genomic characterization take advantage of the large amounts of DNA that can be produced from CSF. In HIV-infected patients with CNS diseases, these studies can provide different kinds of information, for instance for epidemiological purposes, to investigate mechanisms of disease, or to genotypically detect resistance to antimicrobial drugs, as in the case of CMV or M. tuberculosis infections. Nucleotide sequencing is the most accurate method to collect information on genome composition. Automated procedures have been developed during recent years, thus making sequencing relatively easy to carry out. Another technique that has been used for the study of CSF is the restriction fragment length polymorphism (RFLP), in which restriction enzymes are used to digest nucleic acids into fragments of different size, which can then be visualized by gel electrophoresis and a number of hybridization-based techniques. Clinical applications Nucleic acid amplification based assays are nowadays routinely performed in most diagnostic laboratories and have become the test of choice for some HIV-related CNS infections, such as CMV encephalitis or PML. Opportunistic viral CNS diseases, as well as Epstein-Barr virus (EBV)-associated CNS lymphomas, have particularly taken advantage from the application of nucleic acid amplification to the CSF. However, these techniques have also been shown to

124 P Cinque et al be useful for diagnosis of CNS tuberculosis, in which enhance sensitivity of detection and remarkably reduce the times for diagnosis, compared to conventinal culture. Sensitive assays have also been developed for the detection of C. neoformans DNA, although this diagnosis is still commonly achieved by traditional methods. Despite some encouraging reports, molecular methods for detection of T. gondii DNA have not yet reached, in most people hands, satisfactory levels of sensitivity for diagnosis of toxoplasmosis (Bastien, 2002). In addition to providing a means for diagnosis, nucleic acid amplification techniques have also been useful to recognize and define clinically unusual or not yet well-characterized CNS diseases, such as those caused by HSV-1, HSV-2, or VZV, or CMV ventriculoencephalitis itself (Cinque et al, 1997). Examples of these techniques in CSF studies will be discussed for CMV-E, PML, and CNS lymphoma, with emphasis on the most recently acquired information on nucleic acid quantification in CSF and postamplification analyses. CMV encephalitis In the pre-haart era, CMV-E was reported in as many as one third of HIV-infected patients at postmortem examination. Following HAART, this disease has become quite rare in parallel with the dramatic reduction of extracerebral CMV infections. However, CMV-E is still observed in severely immunocompromised HIV-infected patients and it remains as a paradigm of successful application of molecular techniques in the CSF. Although a diffuse encephalopathy is the clinical correlate of CMV-E, its clinical presentation may not be specific. Among traditional virological techniques, CSF virus isolation of CMV and CMV antigen detection lacked sensitivity. On the other hand, brain biopsy was generally not useful because of the absence of radiologically detectable focal lesions. In contrast, CSF PCR for the detection of CMV DNA was shown to be a highly reliable method of diagnosing CMV-E. A number of studies comparing CSF PCR with histopathological, clinical, or virological findings found high sensitivity and specificity values (Table 1). Because of its high diagnostic reliability and speed, nucleic acid amplification based assays are nowadays of widespread use for diagnosing CMV-related CNS lesions in HIV-infected patients (Cinque et al, 1998). A limitation of qualitative PCR methods, however, was the observation that CMV DNA could be detected in both extensive and mild CNS infections. Clinical presentation was not always informative, because of the frequent confounding presence of concomitant CNS pathogens. Quantitative analysis of CMV DNA in CSF has actually been shown to be useful in order to distinguish the severity of the brain lesions. Earlier studies performed by semiquantitative CSF PCR showed that patients with CMV ventriculoencephalitis at postmortem examination had higher levels of CMV DNA as compared to patients with discrete intracerebral foci of parenchymal necrosis (Arribas et al, 1995). More recently, we have confirmed the association between ventriculoencephalitis and high CSF CMV DNA levels by both a quantitative colorimetric PCR assay (Bestetti et al, 2001) and real-time PCR (Bossolasco et al, 2002b). Furthermore, extensive ventriculoencephalitis could be distingueshed by focal periventricular lesions, the latter being associated with lower CMV DNA levels in CSF. Another important use of CMV DNA level measurement in CSF is in the evaluation of virological response to anti-cmv drugs. Patients with CMV- E treated with ganciclovir, foscarnet, or a combination of these two drugs may show decreased CSF CMV DNA levels in response to treatment. However, by using a real-time PCR assay, we have recently observed that, following standard 3-week treatment regimens, CSF CMV DNA levels remain detectable in a remarkable proportion of patients with CMV-E. CSF viral load decrease is usually less marked than in plasma and, in parallel, the clinical response to therapy is also generally poor (Bossolasco et al, 2002b). The study of CMV DNA genes coding for target proteins for antiviral agents (e.g., UL55/polymerase) or necessary for antiviral drug phosphorylation (e.g., UL97/phosphotranspherase) is useful in order to identify virus mutations associated with resistance to antiviral drugs. CMV-E often occurs in patients with a long history of CMV infection and anti-cmv treatments, and both clinical and virological responses of CMV-E to antiviral agents are unsatisfactory. It has therefore been hypothesized that CNS infections may be caused by resistant CMV strains. UL97 mutations conferring resistance to ganciclovir have actually been identified in the CSF of CMV-E patients treated for several months, suggesting that resistant strains might have emerged outside the CNS and then invaded this organ upon intensive replication (Wolf et al, 1995). Progressive multifocal leukoencephalopathy PML is a progressive demyelinative disease induced by oligodendrocyte infection with the polyoma virus JCV. Although clinical and radiological presentation can be supportive, a diagnosis of etiology can only be achieved by virus demonstration in brain tissue or in CSF. Because there is no effective therapy for PML, patients and clinicians are often reluctant to perform brain biopsy, and this procedure has largely been replaced in the clinical practice by CSF PCR for JCV as standard diagnostic method. A number of studies were performed to assess the diagnostic reliability of CSF PCR in HIVassociated PML. Despite a virtually 100% specificity, JCV sequences could be demonstrated by PCR in only two thirds of the patients (Table 1). Because higher rates of detection were found in the more advanced stages of disease, it is now common

Table 1 Examples of clinical application of nucleic acid amplification techniques to the study of cerebrospinal fluid in HIV-1-related opportunistic central nervous system diseases Qualitative PCR Quantitative PCR Genome sequencing Diagnostic Diagnostic Diagnostic Clinical Disease sensitivity specificity significance Main findings significance Main findings Significance Monitoring anti-cmv therapy UL97: identification of mutations associated with ganciclovir resistance Interpretation of positive findings; monitoring antiviral therapy 62 100% 89 100% Method of choice CMV-DNA levels: higher in extensive than in localized lesions; decrease following anti-cmv therapy Cytomegalovirus encephalitis VP-1: mainly epidemiological RR: insights into mechanisms of disease VP-1: distinction of 7 main genotypes RR: identification of rearranged sequences Monitoring response to HAART JCV DNA levels: remain stable or increase with disease progression; decrease following HAART 72 100% 92 100% Method of choice for noninvasive diagnosis Progressive multifocal leukoencephalopathy Intepretation of positive EBV DNA findings CNS lymphoma 83 100% 88 100% Highly supportive EBV DNA levels: higher in patients with systemic NHL > with PCNSL > without brain lymphoma; lower in patients receiving antiherpes drugs Note. HAART, highly active antiretroviral therapy; RR, noncoding regulatory region. 125

126 P Cinque et al practice that a second test is performed if the first is negative (Cinque et al, 1997). The use of HAART was associated with increased survival of PML patients, due to stabilization of their disease. Interestingly, clearance of JCV DNA from CSF was frequently observed in patients receiving HAART, in association with clinical and radiological stabilization (Miralles et al, 1998; Giudici et al, 2000). JCV DNA has been quantified in the CSF by a number studies using both semiquantitative and quantitative methods. In general, a large variation of JCV DNA levels has been observed in CSF, from nondetectable to over 8 log 10 genome equivalents/ml (Taoufik et al, 1998). Within individual patients, the JCV DNA load may either increase or remain stable over the duration of the disease, although fluctuations of the virus burden observed upon repeated sampling suggest a variable viral shedding during intermediate disease stage (Taoufik et al, 1998; Eggers et al, 1999). No significant correlation has been demonstrated between CSF JCV DNA load and CD4+ cell counts or plasma HIV-1 RNA load, although PML patients with >200 CD4+ cells/µl at the time of diagnosis tend to have low JCV DNA CSF levels (Taoufik et al, 1998). Neither has any association been demonstrated with the topographic location or the total volume of the brain lesions (Taoufik et al, 1998; Garcia De Viedma et al, 2002). A number of reports have described an association between high JCV DNA load in CSF and short survival, with values above 4 to 5 log 10 genome equivalents/ml being predictive of poor prognosis (Taoufik et al, 1998; Yiannoutsos et al, 1999). In contrast, we have observed no association between survival and JCV DNA level in CSF samples drawn at the time of PML diagnosis, but only in samples taken during follow-up (Bossolasco et al, 2002a). Because CSF DNA levels may decrease following HAART, it is likely that these apparently discrepant results are due to different timing of CSF sampling. Postamplification analyses of JCV sequences obtained from CSF have mainly studied two virus regions: the major viral capsid protein (VP-1) gene and the noncoding regulatory region. According to the VP-1 region sequences, at least seven major JCV genotypes can be distinguished. PML was reported as more frequent in patients with genotype 2, but it is actually recognized in association with all genotypes, with differences more likely related to geographical distribution of JCV genotypes (Agostini et al, 1997). Sequencing of this region is actually mainly used for epidemiological purposes, although the antigenic function of VP-1 would suggest that polymorphisms in this region may be important in PML pathogenesis. The noncoding regulatory region consists of a hypervariable sequence and, because of its direct role on virus transcription and replication, its molecular structure might be related to various aspects of viral pathogenesis (Jensen and Major, 2001). Both VP-1 and regulatory region sequences have been recently studied in CSF of PML patients treated with HAART, in order to identify possible genomic markers associated with survival. No polymorphisms clearly associated with neurovirulence have thus far been found, although an association has been reported between the presence of CSF regulatory region archetype sequences and long survival (Sala et al, 2001; Pfister et al, 2001). CNS lymphoma Another virus-related CNS disease in HIV-infected patients is primary CNS lymphoma (PCNSL), which is almost always associated with the presence of EBV in tumor cells (MacMahon et al, 1991). Furthermore, secondary CNS localizations of systemic NHL (CNS- NHL) are most frequently associated with EBV. Studies in patients with histologically proven CNS lymphoma reported a striking association between the presence of this complication and EBV DNA detection in CSF. EBV DNA detection in CSF is now regarded as a valuable tool for noninvasive diagnosis of primary CNS lymphoma. This test is often used in association with brain thallium-201 single positron emission computed tomography (SPECT), which has been shown to be of value to distinguish lymphoproliferative lesions, characterized by high tallium uptake, from other focal brain lesions. Despite a high diagnostic specificity of this test was shown by studies comparing CSF with postmortem findings, it was not infrequent in clinical practice to find EBV DNA in CSF of patients without clinical evidence of lymphoma (Table 1) (Cinque et al, 1997). In some of these cases, EBV DNA was actually detected days or months before the lymphoma manifested itself clinically. In other cases, where no evidence of lymphoma could be demonstrated, false-positive EBV findings might have theoretically resulted from either sliding of virus or latently infected lymphocytes through an impaired blood-csf barrier or by EBV reactivation infection in the CNS. We have recently shown that real-time quantitative PCR can be useful to discriminate a clinically significant positive EBV finding, characterized by high viral loads, from incidental findings. By the use of this technique, patients with PCNSL or CNS-NHL had significantly higher CSF EBV DNA levels than patients with systemic but no CNS lymphoma or with other CNS diseases (Bossolasco et al, 2002c). Interestingly, we also observed that patients with CNS lymphoma receiving ganciclovir for a concomitant CMV infection had significantly lower CSF EBV DNA levels than untreated lymphoma patients. Because, theoretically, only active, but not latent, EBV infection is inhibited by this drug, these observations suggest that active EBV replication can take place in the brain of patients with lymphoma. From a clinical point of view, these findings also indicate that low or negative EBV DNA values should be interpreted cautiously in patients with suspected lymphoma, because falsenegative results might actually be due to the effect of antiherpes drugs.

P Cinque et al 127 Final remarks The applications of molecular methods to the study of HIV-associated CNS infections are in continuous evolution. An array of nucleic acid amplification techniques is nowadays applicable to the CSF in order to establish an etiological diagnosis of most of these complications. Quantitative methods have provided a valuable additional tool for their clinical management, whereas postamplification techniques can enable precise genome characterization following their recovery in the CSF. On the other hand, further tech- nical developments are still needed to increase the diagnostic sensitivity of molecular techniques in certain CNS infections, e.g., toxoplasmosis, and, more in general, to further reduce costs and times required for examination. In addition, because most diagnostic laboratories use in-house-developed assays, the diagnostic reliability of molecular methods would enormously benefit from assay standardization. In this regard, quality control programmes involving distribution of molecular standards have become a priority, to provide a means towards optimal assay efficiency and consistency of results among laboratories. References Agostini HT, Ryschkewitsch CF, Mory R, Singer EJ, Stoner GL (1997). JC virus (JCV) genotypes in brain tissue from patients with progressive multifocal leukoencephalopathy (PML) and in urine from controls without PML: increased frequency of JCV type 2 in PML. J Infect Dis 176: 1 8. Arribas JR, Clifford DB, Fichtenbaum CJ, Commins DL, Powderly WG, Storch GA (1995). Level of cytomegalovirus (CMV) DNA in cerebrospinal fluid of subjects with AIDS and CMV infection of the central nervous system. J Infect Dis 172: 527 531. Bastien P (2002). Molecular diagnosis of toxoplasmosis. Trans R Soc Trop Med Hyg 96(suppl 1): S205 S215. Bestetti A, Pierotti C, Terreni M, Zappa A, Vago L, Lazzarin A, Cinque P (2001). Comparison of three nucleic acid amplification assays of cerebrospinal fluid for diagnosis of cytomegalovirus encephalitis. J Clin Microbiol 39: 1148 1151. Bossolasco S, Bestetti A, Sala S, Brambilla AM, Viganò MG, Giudici B, Cinque P (2002a). Diagnosis and monitoring of HIV-related PML by quantitative real-time PCR for JCV DNA. Presented at the Conjoint 4th International Symposium on NeuroVirology and 10th Conference on Neuroscience of HIV infection. June 19 22, 2002, Dusseldorf, Germany. Bossolasco S, Bestetti A, Sala S, Lazzarin A, Linde A, Cinque P (2002b). Real-time polymerase chain reaction for diagnosis and treatment follow-up of herpesvirus central nervous system infections in HIVinfected patients. Presented at the Conjoint 4th International Symposium on NeuroVirology and 10th Conference on Neuroscience of HIV infection. June 19 22, 2002, Düsseldorf, Germany. Bossolasco S, Cinque P, Ponzoni M, Viganò MG, Lazzarin A, Linde A, Falk K (2002c). Epstein-Barr virus DNA load in cerebrospinal fluid and plasma of patients with AIDSrelated lymphoma. J NeuroVirol 5: 432 438. Bustin SA (2000). Absolute quantification of mrna using real-time reverse transcription polymerase chain reaction assays. J Mol Endocrinol 25: 169 193. Cinque P, Cleator GM, Weber T, Monteyne P, Sindic C, Gerna G, van Loon AM, Klapper PE (1998). Diagnosis and clinical management of neurological disorders caused by cytomegalovirus in AIDS patients. European Union Concerted Action on Virus Meningitis and Encephalitis. J NeuroVirol 4: 120 132. Cinque P, Scarpellini P, Vago L, Linde A, Lazzarin A (1997). Diagnosis of central nervous system complications in HIV-infected patients: cerebrospinal fluid analysis by the polymerase chain reaction. AIDS 11: 1 17. Eggers C, Stellbrink HJ, Buhk T, Dorries K (1999). Quantification of JC virus DNA in the cerebrospinal fluid of patients with human immunodeficiency virus-associated progressive multifocal leukoencephalopathy a longitudinal study. J Infect Dis 180: 1690 1694. Garcia De Viedma D, Diaz Infantes M, Miralles P, Berenguer J, Marin M, Munoz L, Bouza E (2002). JC virus load in multifocal progressive leukoencephalopathy: analysis of the correlation between the viral burden in cerebrospinal fluid, patient survival and the volume of neurological lesions. Clin Infect Dis 34: 1568 1575. Giudici B, Vaz B, Bossolasco S, Casari S, Brambilla AM, Luke W, Lazzarin A, Weber T, Cinque P (2000). Highly active antiretroviral therapy and progressive multifocal leukoencephalopathy: effects on cerebrospinal fluid markers of JC virus replication and immune response. Clin Infect Dis 30: 95 99. Jensen PN, Major EO (2001). A classification scheme for human polyomavirus JCV variants based on the nucleotide sequence of the noncoding regulatory region J Neuro- Virol 74: 280 287. MacMahon EM, Glass JD, Hayward SD, Mann RB, Becker PS, Charache P, McArthur JC, Ambinder RF (1991). Epstein-Barr virus in AIDS-related primary central nervous system lymphoma. Lancet 338: 969 973. Miralles P, Berenguer J, Garcia de Viedma D, Padilla B, Cosin J, Lopez-Bernaldo de Quiros JC, Munoz L, Moreno S, Bouza E (1998). Treatment of AIDSassociated progressive multifocal leukoencephalopathy with highly active antiretroviral therapy. AIDS 12: 2467 2472. Pfister L-A, Letvin N, Koralnik I (2001). JC virus regulatory region tandem repeats in plasma and central nervous system correlate with poor clinical outcome in patients with progressive multifocal leukoencephalopathy. J Virol 75: 5672 5676. Preiser W, Elzinger B, Brink NS (2000). Quantitative molecular virology in patient management. J Clin Pathol 53: 76 83. Sacktor N, Lyles RH, Skolasky R, Kleeberger C, Selnes OA, Miller EN, Becker JT, Cohen B, McArthur JC, and the Multicenter AIDS Cohort Study (2001). HIV-associated neurologic disease incidence changes: Multicenter

128 P Cinque et al AIDS Cohort Study, 1990 1998. Neurology 56: 257 260. Sala M, Vartanian J-P, Kousignian P, Delfraissy J-F, Taoufik Y, Wain-Hobson S, Gasnault J (2001). Progressive multifocal leukoencephalopathy in human immunodeficiency virus type 1-infected patients: absence of correlation between JC virus neurovirulence and polymorphisms in the transcriptional control region and the major capsid protein loci. J Gen Virol 82: 899 907. Taoufik Y, Gasnault J, Karaterki A, Pierre Ferey M, Marchadier E, Goujard C, Lannuzel A, Delfraissy JF, Dussaix E (1998). Prognostic value of JC virus load in cerebrospinal fluid of patients with progressive multifocal leukoencephalopathy. J Infect Dis 178: 1816 1820. Wolf DG, Lee DJ, Spector SA (1995). Detection of human cytomegalovirus mutations associated with ganciclovir resistance in cerebrospinal fluid of AIDS patients with central nervous system disease. Antimicrob Agents Chemother 39: 2552 2554. Yiannoutsos CT, Major EO, Curfman B, Jensen PN, Gravell M, Hou J, Clifford DB, Hall CD (1999). Relation of JC virus DNA in the cerebrospinal fluid to survival in acquired immunodeficiency syndrome patients with biopsy-proven progressive multifocal leukoencephalopathy. Ann Neurol 45: 816 821.