|Home | About | Journals | Submit | Contact Us | Français|
PC and RWP compiled literature reviews, drafted and finalized the manuscript. All authors participated in conceptualizing, reviewing and editing the manuscript.
Progressive multifocal leukoencephalopathy (PML) caused by the polyomavirus, JC virus (JCV), is one of the most dreaded complications of HIV-1 infection. Unlike other opportunistic infections, PML may present while blood CD4+ T cells remain above AIDS-defining levels and while patients receive combined antiretroviral therapy (cART), either shortly after starting or, more rarely, during chronic successful treatment. PML can be suspected by typical presentation with focal neurological deficits and corresponding demyelinating lesions at magnetic resonance imaging (MRI), while definitive diagnosis requires identification of JCV in cerebrospinal fluid (CSF) or brain tissue. While there is no specific treatment, reversal of immunosuppression by cART leads to clinical and MRI stabilization in 50-60% of PML patients and JCV clearance from CSF. A proportion of cART-treated patients develop inflammatory lesions, which may either accompany a favorable outcome or associate with clinical worsening. The reasons for variability in PML natural history and treatment responses are largely undefined, and more specific and rational approaches to management are sorely needed.
Progressive multifocal leukoencephalopathy (PML) is an infection of the central nervous system (CNS) by the polyoma virus, JC virus (JCV), that destroys oligodendrocytes and their myelin processes. Outside of the context of HIV-1 infection, PML is rare and develops as a complication of conditions with compromised immunity, such as hematologic malignancies and immunesuppressive treatments [1, 2]. Interest in PML has recently increased by its presentation in patients receiving humanized antibody-based immunomodulatory treatments, such as natalizumab [3, 4], efalizumab  and, possibly, rituximab . However, it is in HIV-1 infection that PML remains most common and its biology, diagnosis and treatment issues have been most extensively studied.
In this review we summarize the epidemiology, biology and clinical presentation of HIV-1-related PML, and discuss current views on management, emphasizing areas in need of better information and understanding.
Antiretroviral treatment has reduced the incidence of PML in HIV-1 infection and its mortality. Before the advent of combined antiretroviral therapy (cART), PML developed in 3-7% of HIV-1-infected patients [7, 8] and comprised up to 18% of fatal CNS diseases . This frequency has decreased substantially in the current treatment era, though not to the same extent as other CNS opportunistic infections (CNS-OIs) [10, 11]. In the Eurosida cohort, PML incidence decreased from 0.7 per person year of follow-up in 1994 to 0.07 in 2001-2002, with the lowest annual decrease of incidence, of 31%, among all CNS-OIs . Unlike the other major CNS-OIs, PML may develop in patients with blood CD4+ T-cell counts above 200 cells/μL, in patients initiating cART, and, more rarely, even in stably treated patients with full viral suppression [12, 13].
PML survival has increased substantially over the last 10 years, from 0-30% at 1 year in the pre-cART period to 38-62% following cART introduction [12-16]. However, this increase appears to be the smallest among all AIDS-defining diseases [11, 17, 18]. In the Antiretroviral Therapy Cohort Collaboration, mortality for PML was indeed 10-fold higher compared to all AIDS-related diseases , and, in a 2005 French survey, PML accounted for 14% of all AIDS-related deaths, second only to non-Hodgkin's lymphoma .
JCV is a small, ubiquitous DNA virus belonging to the family of human polyomaviruses (Figure 1). JCV infection is common worldwide. Both inhalation and ingestion of contaminated water or food have been suggested as major modes of human transmission [20-22]. The seroprevalence of JCV among healthy persons varies among studies, likely depending on the technique employed. In a large British survey, the overall seroprevalence was 35% by hemoagglutination inhibition assay against VP1, with figures rising with age - from 11% in children below 5 years to 50% at 60-69 years . Figures were higher in other studies using the more sensitive enzyme immunoassays [24-26], with an IgG seroprevalence of 58% in a recent Swiss study on 400 healthy blood donors, also showing an increasing trend with age .
Following primary infection, JCV establishes a persistent infection. There is evidence supporting both the urinary tract and the bone marrow as potential peripheral sites of this persistence. JCV is shed in urine of approximately 30% of immunologically normal individuals [26-28], and, recently, viral DNA has been demonstrated by polymerase chain reaction (PCR) in bone marrow of 13% of HIV-negative and 47% of HIV-1-positive patients without PML . The CNS has also been suggested as a possible site of JCV persistence. The virus might reach the CNS at the time of primary infection or during its peripheral persistence phase and establish latent or low level benign CNS infection. In support of this hypothesis, PCR studies have shown JCV DNA in the brain of patients without PML [30, 31], including, recently, detection of JCV DNA fragments in oligodendrocytes and astrocytes . However, efforts to identify viral proteins or DNA by in-situ techniques have generally been unsuccessful [31-33].
A proportion of patients with PML have detectable JCV DNA in plasma (see below) and the analysis of the highly variable regulatory region of JCV usually shows similar, highly-rearranged sequences in CSF and blood, but not in urine, of PML patients [34-36]. These observations support the hematogenous spread of JCV to the CNS from bone marrow or a related source at the time of PML, although, theoretically, they are also compatible with virus reactivation in the CNS and subsequent release into blood.
JCV-specific immunity is likely critical in controlling persistent JCV infection. CD4 cell responses have been detected in healthy subjects who excrete JCV in urine , and JCV-speci?c cytotoxic T lymphocytes were demonstrated in 73% of JCV-seropositive, immunocompetent subjects . Conversely, the loss of the immune control against JCV is certainly key in favoring viral reactivation, either in periphery or in the CNS, and allowing viral propagation in the CNS, lytic infection of oligodendrocytes and brain disease. The importance of host defenses is clear from both the context of PML, with immune defects, and its remission after reversion of immune defect , most notably by cART [12-16]. Indeed, PML remission after initiation of cART is often associated with restoration of JCV-specific CD4 and CD8 T-cell and B-cell responses in blood [37, 40, 41] and CSF [42, 43].
The PML brain shows multiple areas of demyelination, varying in size and stage of evolution. Initial foci of demyelination expand and may coalesce into larger areas that in advanced cases can evolve into cavitary necrosis. All CNS regions may be involved, although spinal cord lesions are rare .
Oligodendrocytes sustain productive-lytic infection accompanied by characteristic histopathology (Figure 2). Their nuclei become enlarged and densely basophilic, filled with eosinophilic inclusions, and stain immunocytochemically and by in situ hybridization for JCV gene proteins and nucleic acid. Astrocytes may also appear enlarged, sometimes with multiple or multilobate hyperchromatic nuclei, at times resembling neoplastic cells – the so-called “bizarre” astrocytes , and also harbor JCV gene products. Foamy macrophages, though not specific for PML, are commonly present and are a response to myelin breakdown. While clinically and pathologically less conspicuous, JCV has also been identified in cerebellar granule cells, either within the context of PML, or in isolation in patients presenting with ataxia and cerebellar atrophy . Thus, while usually considered as sparing neurons, JCV can also cause an infectious cerebellar degeneration in the context of immunosuppression related to its tropism for this particular class of neurons.
One of the classical features of PML is the paucity or absence inflammation . However, inflammatory forms of PML have been observed with increasing frequency following the introduction of cART. These are characterized by either diffuse or focal perivascular mononuclear infiltrates comprised mostly of CD8+ T lymphocytes and monocyte/macrophages (Figure 3); B lymphocytes, plasma cells and CD4+ T-cells are found only in small numbers [46-49]. This inflammatory PML may be the initial presenting disease phenotype or, more frequently, evolve after initiation of cART (see below).
Classically, PML presents insidiously with focal neurological deficits that vary depending on lesion localization. Thus, there may be hemisensory defects or hemiparesis (following involvement of parietal or frontal lobes), visual field loss and hemianopsia (occipital lobes or optic radiations), aphasia (language-dominant hemisphere), or ataxia and dysmetria (cerebellar hemispheres and peduncles). These often begin as ‘partial’ deficits that worsen with time (for example, arm weakness evolving to hemiparesis), reflecting the spread of individual lesions concentrically or along white matter tracts. Approximately 20 percent of patients develop seizures, in association with lesions adjacent to the cortex . While cognitive deficits are reported in up to one third of the cases, isolated dementia without motor, sensory or visual deficits is rare [51, 52]. Headache and fever are usually absent.
Before the introduction of cART, PML was almost invariably fatal. Median survival from the diagnosis was only of a few months, although continuous progression over one year or more was occasionally observed among patients with relatively high CD4 T-cell counts . In cART-treated patients, disease stabilization is observed in approximately half of the cases (see below).
Establishing a definitive diagnosis of PML is not only useful for clinical studies, but important for individual patient management. This is necessary in atypical cases, and, in the more typical cases, it prevents the need to revisit diagnosis in the face of disease progression, helps physicians to proceed rapidly and with certainty in therapy, and guides prognosis for patient and family decisions. One can consider three ‘stages’ to diagnosis of PML: clinical suspicion, radiological identification and etiological confirmation by CSF or tissue analysis (Table 1)[54-56]. The first of these relies on the character and temporal evolution of focal neurological symptoms and signs as outlined above, along with the setting of disease susceptibility.
The second stage in diagnosis entails detection and characterization of brain lesions by neuroimaging, preferably magnetic resonance imaging (MRI). MRI shows characteristic white matter lesions in brain areas corresponding to the clinical deficits. Because lesions involve demyelination, they are usually hyperintense on T2-weighted and FLAIR MRI sequences, but also hypointense on T1-weighted sequences, indicating white matter destruction. The latter helps distinguish PML from other pathologies, primarily HIV-1 encephalopathy, with more diffuse central white matter changes that are not detected on T1 sequences. While lesions can develop in any part of the brain, including the deep grey matter, they are most common in the subcortical white matter, the white matter of the cerebellar peduncles or hemispheres and in the brain stem (Figure 4). In the classical, non-inflammatory form, there is either no or only minimal contrast-enhancement, and no mass effect, unlike cerebral toxoplasmosis and primary CNS lymphoma [57, 58]. Inflammatory forms may be defined radiologically by the presence of contrast enhancement, edema, mass effect or displacement of normal structures, at times making distinction from the other opportunistic infections more difficult (Figure 3)[46, 59, 60].
Newer imaging techniques may provide additional diagnostic information. By diffusion-weighted imaging (DWI) sequences, which read the degree of diffusion of water molecules in a lesion area, more recent lesions and the advancing edge of established lesions show high intensity signal, reflecting active infection and cell swelling (low water diffusion). In contrast, older lesions and the center of large lesions exhibit low intensity signal, marking areas of necrosis and reparative gliosis (high water diffusion)(Figure 5)[61-63]. The degree of diffusion can also be expressed numerically using the Apparent Diffusion Coefficient (ADC) maps (Figure 5).
With proton MR spectroscopy (MRS), PML lesions usually show elevated choline (Cho) or Cho/creatine ratio, related to demyelination, and decreased N-acetyl-aspartate (NAA), or NAA/creatine ratio, reflecting neuronal and axonal damage. The NAA/creatine ratio is usually lower in PML than in other white matter lesions, and it may help differentiate PML from HIV-1 encephalopathy. Lactate and lipid signals, likely reflecting necrosis, are also frequently elevated (Figure 5)[64, 65].
Etiological diagnosis of PML usually relies first on detection of JCV DNA in CSF by PCR (Table 1). Among HIV-1-infected, cART untreated patients with neurological diseases, the diagnostic sensitivity of this technique was of 72–92% and specificity of 92–100% . Thus, a positive result is regarded as diagnostic in the appropriate clinical context. Because the rate of JCV DNA detection increases with progression of PML, lumbar puncture is usually repeated if initial PCR analysis is negative but suspicion remains high. However, the likelihood of detecting JCV in CSF may be reduced - down to 58% - in patients presenting with PML in the setting of cART . The level of JCV DNA in CSF varies largely among PML patients, between the lower limit of detection - usually around 102 copies/mL - to over 107 copies/mL . In untreated patients, the magnitude of JCV DNA load in CSF added additional prognostic information, with higher copy numbers predicting shorter survival [68, 69]. However, the correlation with survival is less clear for those who develop PML while on cART in whom high copy number is not always unfavorable [69, 70]. In addition, patients with lower CD4 T-cell counts generally exhibit higher CSF JCV DNA levels .
When efforts to identify JCV DNA in the CSF fail, brain biopsy is required to achieve an etiological diagnosis of PML. PML is identified by the characteristic tissue histopathology described above, and virological confirmation is obtained by identification of JCV proteins by immunohistochemistry, JCV DNA by in situ nucleic acid hybridization or JC virions by electron microscopy [1, 33, 71](Figure 2, Table 1).
By PCR, JCV DNA can also be detected in blood, both in mononuclear cells (PBMC) and plasma. JCV DNA has been found in PBMC of 10-60% of PML patients but also in up to 20% of immunocompromised controls [72-74]; copy numbers were generally below 103 per 106 cells but did not differ between patients and controls . In plasma, JCV is detected in 16-40% of patients with PML, but only occasionally in controls without PML [3, 73, 75]; the viral load seems to vary widely in PML patients (from 102 to 106 copies/mL), although is most often low, i.e., below 103 copies/mL in controls (and personal observation). These figures indicate that JCV DNA detection in blood is unlikely to provide a sensitive and specific diagnostic tool for PML, though more systematic study is required in light of the wide range of published results. In urine, JCV DNA is detected in approximately one-third of both healthy and immunocompromised individuals without PML [26-28, 73, 76, 77], at a wide range of copy number, and thus is not useful in PML diagnosis.
Because of the high seroprevalence of JCV, the presence of JCV-specific serum IgG is unlikely to be useful in PML diagnosis. However, demonstration of JCV-specific intrathecal antibody synthesis might provide additional diagnostic support. Intrathecal production of antibodies to JCV-VP1 was observed in 76% of PML cases but in only 11% of HIV-1-infected controls without PML, with a significantly higher antibody specificity index in the former .
JCV-specific T-cell immunity has been assessed in PML by different technical approaches, including T-cell lymphoproliferation (studying the CD4 cell responses), cytotoxic T cell (CTL) chromium release and CD8+ tetramer identification [37, 40, 79], as well as by functional cytokine secretion and Elispot assays, measuring both CD4 and CD8 responses [41, 80]. At the time of diagnosis, most of the patients with HIV-1-related PML show low, most often undetectable, responses [37, 40, 41]. Although the diagnostic sensitivity and specificity of these assays has not been formally assessed, the presence of responses in some PML patients and, conversely, lack of responses in patients without the disease, suggest that these assessments are unlikely to become diagnostically useful.
To provide a subject designation for planning and evaluating treatment interventions, we have developed a classification system of HIV-1-associated PML, which uses two principal components: cART context and immuno-virological response, and inflammatory features (Table 2). Inflammation characterizes the cART-associated Immune Reconstitution Inflammatory Syndrome (IRIS), and the presence of inflammation in PML justifies the combined term PML-IRIS . By definition, the general term, IRIS, comprises three elements: 1. immune-reconstitution, meaning a decrease of plasma HIV-1 RNA with or without an increase of CD4 T cells temporarily associated with the start of cART; 2. tissue inflammation; and, 3. clinical disease or worsening that would not be expected from the natural course of the disease [82, 83]. In PML, this occurs in two settings. The first is when symptomatic PML is treated with cART and inflammation develops in relation to the existing PML lesions (paradoxical IRIS). In this setting, it is important to distinguish a favorable immune reconstitution, associated with clinical stabilization, from harmful IRIS associated with clinical worsening . The second setting is when patients manifest PML after initiating cART and an inflammatory picture is found on MRI (unmasking IRIS).
Complicating this picture, asymptomatic patients initiating cART may develop PML after a few weeks of therapy, but without showing the characteristic inflammatory picture on MRI [12, 81]. The mechanism underlying PML development in these cases is unknown, but a ‘provocative effect’ of cART on JCV replication is invoked. These forms might be termed immune reconstitution (IR)-PML, but without the immunopathological connotation of the inflammatory syndrome terminology.
Thus, there is a spectrum of PML with respect to setting of incidence that includes both unmasking PML-IRIS and IR-PML in asymptomatic patients following initiation of cART; and both paradoxical PML-IRIS and PML with therapeutically useful and clinically benign inflammation when PML is treated with cART. It will be useful to further characterize these states with respect to JCV replication and immune responses, imaging characteristics and other biomarkers to more clearly guide clinicians in treatment approaches.
The current main approach to treatment of HIV-1-related PML involves cART with the objective of reversing the immunological defect that interferes with the normal host response to JCV. This is thus an ‘indirect’ approach to PML treatment, but it is the only one that has proven effective. While a number of antiviral and/or immunomodulant agents have been proposed or used as more specific PML treatments, none has proven effective after greater scrutiny (Table 3).
Very recently, in vitro screening of a drug library of over thousand molecules, either already approved or under clinical trial, led to the identification of several drugs active against JCV . One of these, the anti-malarial drug, mefloquine, which has favorable pharmacokinetic properties and brain penetration, has recently been brought into a phase I/II clinical trial evaluating its efficacy in decreasing CSF JCV DNA level. The trial is currently under way . It is to be hoped that this is not only effective, but also leads the way to a greater effort in drug discovery and therapeutic trial for PML.
Approximately half of HIV-1-infected PML patients receiving cART experience an arrest of disease progression [12-16, 70]. Unfortunately, neurological deficits frequently persist because of irreparable loss of brain tissue, and only a minority of patients experience functional improvement. The outcome of PML is not readily predicted at the onset or in the early stages of the disease. In some case series, poorer prognosis was associated with lower CD4+ T cell counts, higher plasma HIV-1 RNA levels, higher CSF JCV DNA level at the time of PML diagnosis, or lesion localization to the brainstem [13-16, 70, 104]. None of these variables, however, was definitely shown to strongly predict outcome.
In clinical practice, treatment strategies depend on the patient's cART status and response (Table 2). Untreated patients should start cART immediately. Similarly, treatment should be changed to an effective regimen in treated patients with continued HIV-1 viremia . More problematic is treating patients who present with PML while taking cART and with suppressed HIV-1 viremia. A recent prospective study showed survival benefit in PML patients treated with four classes of cART, including enfuvirtide , suggesting that cART intensification might be considered in patients with undetectable plasma HIV-1. Longer PML survival has also been associated with higher CNS penetration-effectiveness score of antiretroviral regimens [106, 107]. It can be speculated that fast and effective control of HIV-1 replication in the CNS might affect JCV activity by inhibiting HIV-1 transactivation of JCV early genes [108, 109] or through unknown mechanisms associated with a reduction of intrathecal immuneactivation . These issues related to antiretroviral intensification and choice of individual anti-HIV drugs clearly need further study.
As discussed above, cART-treated patients may show an exuberant inflammatory response related to the PML lesions, i.e., IRIS-PML. In HIV-1-associated IRIS outside the PML setting, corticosteroids are used to ameliorate local inflammatory reactions [82, 83]. In PML patients who respond favorably to cART, contrast enhancement or edema may not require steroid treatment if the patient is clinically stable or improving. However, in those with progressing clinical deficits and signs of inflammatory disease, corticosteroid treatment appears justified. There is little published information to guide their initiation, dosage and duration. In a recent review of 12 cases receiving steroids, a trend was observed between good PML outcome and earlier initiation or longer duration of treatment . High doses of intravenous methyl-prednisolone (500-1000 mg per day for 3-5 days), oral prednisone (1-2 mg/Kg per day) or equivalent doses of oral or intravenous dexamethasone, have improved neurological function and reduced MRI signs of inflammation in individual cases of inflammatory PML [47, 49, 81](and personal observation, Figure 4).
Both for clinical trials and individual patient management, methods and strategies that enable real-time monitoring of the effects of treatment and distinguish active/progressing from inactive/stabilized disease or disease remission are essential. Currently, this is achieved by the combination of neurological assessment, conventional MRI, and JCV DNA measurement in CSF. This approach is useful to show disease stabilization or progression over a period of weeks to months, but measures that could define the direction of disease evolution and treatment effects within a shorter time frame still need to be developed and clinically validated. Because of the frequently rapid evolution of PML, clinical assessment every 2-4 weeks and MRI and CSF examination every 4-8 weeks appear reasonable options with today's methods. Additional MRI evaluation can be recommended in case of rapid clinical deterioration, particularly when inflammatory changes associated with immune reconstitution are suspected. Clinical measures of functional change, such as the expanded disability status scale (EDSS) can help to objectively identify clinical stabilization or improvement. The MRI picture is considered inactive when there are no new lesions, and already established lesions show no increase in volume, FLAIR/T2 hyperintensity and T1 hypointensity. Preliminary observations have shown that DWI disappearance of the high-signal intensity rim at the lesion periphery, accompanying disease stabilization, may precede changes revealed by T2/FLAIR images [61, 63]. Disease stabilization is also associated with decreasing Cho/Cr at MRS, increasing NAA/Cr and myoinositol (mI)/Cr ratio, the latter signaling glial activation, and resolution of the lipid and lactate signals [64, 111].
Virological remission is defined by clearance of JCV DNA in CSF [47, 69, 112], likely reflecting suppression of viral replication in brain lesions. The dynamics of decay varies, within a time frame of weeks to months, depending on initial level and, likely, the degree and dynamics of immunological restoration in individual cases . Clearance of JCV DNA can be associated with clinical stabilization but transitory extension of the signal abnormalities at MRI (Figure 6)[42, 60], which might reflect the host inflammatory response (edema that is hyperintense in T2 and FLAIR) rather than actual extension of the front of myelin loss.
Assessment of JCV-specific immunity is also likely to provide a useful tool for real-time monitoring the effects of treatment. By both CD4 T-cell lymphoproliferation and studies of CD8 T-cell activity against the JCV VP-1 protein or peptides, PML patients who survived PML showed significant anti-JCV responses both at diagnosis and during follow-up, as opposed to patients with poor prognosis [37, 40, 41]. Also systemic humoral immune responses, as assessed by ELISA at time of diagnosis, were significantly greater in survivors compared to non-survivors  and an increased JCV-specific intrathecal antibody index was shown in a small group of cART-treated patients undergoing PML remission, in parallel with JCV DNA decay in CSF .
PML remains an important complication of HIV-1 infection. It should be promptly considered in any HIV-1-infected patients presenting with insidious onset of focal neurological symptoms, both cART untreated or treated, and especially in those recently starting cART. If the diagnosis of PML is achieved, effective cART needs be immediately instituted or continued. More generally, there is a need for clearer understanding of its pathogenesis and for development of more effective therapies, in order to diminish the impact of this disease. We need to better understand the factors involved in JCV infection and host responses that convert a persistent, but seemingly benign, infection to progressive focal encephalitis, and to exploit this understanding to predict susceptibility and develop early diagnostic and therapeutic interventions. We need to better understand of host immune responses to JCV and to define differences between helpful and immunopathological reactions underlying the opposing outcomes of remission and PML-IRIS. Finally, we need to develop specific therapies for JCV and PML to supplement the effects of cART and better methods of monitoring therapeutic responses in a shorter time frame.
We would like to thank Dr. Manuela Nebuloni, Department of Pathology, Luigi Sacco Hospital, University of Milan, Italy, and Dr. Pilar Miralles, Hospital Universitario Gregorio Marañón, Madrid, Spain, for help with iconography.
IJK is supported in part by NIH grant R01 NS041198 and 047029, and K24 NS 060950
JMM is supported in part by the ‘Ministerio de Sanidad y Consumo, Instituto de Salud Carlos III, Madrid (Spain)’, Spanish Network for AIDS Research (RIS; ISCIII-RETIC RD06/006) and the Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Barcelona (Spain).
Search strategy and selection criteria
Data for this review were identified by searches of PubMed, and references from relevant articles. No date restrictions were set in the search. We also reviewed abstracts from major meetings on HIV-1 infection and neurology during the past three years. Search terms were ‘progressive multifocal leukoencephalopathy’ and ‘JC virus’. English language papers only were reviewed.
Conflict of Interest
PC is a member of the steering committee and site investigator of the Biogen Idec-sponsored trial of mefloquine for PML, has received research support from Biogen Idec and Biogen Dompè, and honoraria from Abbott, Biogen Idec, Biogen Dompè, Glaxo Smith Kline and Pfizer.
IJK is a site investigator in the Biogen Idec-sponsored trial of mefloquine for PML and has received research support from Biogen Idec and honoraria from Merck-Serono, Bristol-Myers-Squibb, Ono pharmaceuticals, Antisense and Alnylam.
SG has received honoraria from Biogen Dompè.
JMM has received honoraria and/or research grants from Abbott, Boehringer-Ingelheim, Bristol-Myers Squibb, Cubist, Gilead Sciences, Glaxo Smith Kline, Novartis, Pfizer, Roche and Theravance.
RWP is a site investigator in the Biogen Idec-sponsored trial of mefloquine for PML and has received research support from Biogen Idec and Merck and Co.