Cidofovir added to HAART improves virological and clinical outcome in AIDS-associated progressive multifocal leukoencephalopathy

Abstract

Objectives 

To analyse the virological and clinical efficacy of cidofovir combined with highly active antiretroviral therapy (HAART) in AIDS-related progressive multifocal leukoencephalopathy (PML).

Design 

Multicentre observational study of consecutive HIV-positive patients with histologically or virologically-proven PML. Group A, 26 patients treated with HAART; group B, 14 patients treated with HAART plus cidofovir 5 mg/kg intravenously per week for the first 2 weeks and alternate weeks thereafter. JC virus DNA was quantified in cerebrospinal fluid (CSF) by PCR.

Results 

Baseline virological, immunological and clinical characteristics were homogeneous between the groups. In one case cidofovir was discontinued because of severe proteinuria. There was no significant difference in HIV RNA responses and changes in the number of CD4 cells between group A and B. After 2 months of therapy, five out of 12 (42%) patients from group A and seven out of eight (87%) from group B reached undetectable JC virus DNA in the CSF (Chi-square P = 0.04); moreover, 24% of group A and 57% of group B patients showed neurological improvement or stability (P = 0.038). One-year cumulative probability of survival was 0.67 with cidofovir and 0.31 without (log-rank test, P = 0.01). Variables independently associated with longer survival were the use of cidofovir, HAART prior to the onset of PML, a baseline JC virus DNA load in CSF < 4.7 log₁₀ copies/ml, and a baseline Karnofsky performance status ≥ 60.

Conclusions 

In AIDS-related PML, cidofovir added to HAART is associated with a more effective control of JCV replication, with improved neurological outcome and survival compared with HAART alone.

Introduction

Progressive multifocal leukoencephalopathy (PML), the demyelinating disease caused by JC virus (JCV), is an important cause of morbidity and mortality in patients with AIDS. Several reports show a reduced incidence of PML in the era of highly active antiretroviral therapy (HAART), although the reduction seems less evident than that of other opportunistic diseases ^[1,2]. Furthermore, potent antiretroviral therapies have been shown to significantly prolong survival of patients affected by AIDS-related PML, even in large cohort studies ^[3–5]. Nevertheless, not all HIV-infected patients with this neurological disorder benefit from HAART and cases of PML occurring after potent antiretroviral therapies had begun have also been reported ^[6,7]. Cidofovir is an antiviral compound which inhibits the replication of Papovaviridae in vitro^[8]. Reports of clinical benefit after treatment of PML-affected patients with this compound are anecdotal ^[9–12]. The aim of the present study was to assess whether treatment with cidofovir in combination with HAART is more effective in terms of virological and clinical response than HAART treatment alone in individuals with AIDS-related PML.

Patients and methods

Patients and clinical data collection

Patients were recruited from three large Italian tertiary centres for Infectious Diseases. HIV-infected individuals with a confirmed diagnosis of PML, who were exposed to HAART with or without cidofovir for at least 4 weeks were included in the study. Diagnosis was required to rely either on the concomitant presence of a compatible clinical and neuroradiological picture and of JC virus DNA detected in the cerebrospinal fluid (CSF) by PCR, or on the clinical and radiological picture and the presence of characteristic histopathological features in brain tissue. These included the classic triad of demyelination, oligodendrocytes with enlarged, hyperchromatic nuclei and enlarged bizarre astrocytes. Other viral infections [cytomegalovirus (CMV), herpes simplex virus, varicella zoster virus, Epstein–Barr virus] were excluded by PCR assays of the CSF. Patients were classified in two groups according to the treatment received: group A consisted of patients treated with HAART (three or more antiretroviral drugs), and patients treated with HAART and cidofovir were recorded in group B. Data were collected retrospectively from clinical records. In the three clinical centres, patients routinely receive a standardized complete neurological assessment at each hospitalization. During follow-up, neurological examination was performed mostly at monthly intervals, but in every case after 2 ± 0.5 and after 6 ± 0.5 months. Therefore, results from baseline, month 2 and month 6 neurological examinations were retrieved for analysis. Follow-up neurological examination was categorized in comparison to baseline as follows: progressed, worsening of previously detected focal signs and/or appearance of new focal signs and/or alteration of consciousness; improved, improvement or resolution of previously detected neurological signs; stable, lack of improvement or progression. Patients with improvement or stability of the neurological picture were defined further as clinical responders; those with neurological progression were defined as clinical non-responders.

Semiquantitative JC virus DNA assay in the CSF

JCV DNA was quantified in all CSF samples by a central laboratory. A semi-quantitative assay was used that was a slight modification of a previously described PCR technique with nested primers from the large T antigen region ^[13]. Briefly, an external standard of a known concentration of recombinant JCV DNA (Mad-1 strain) was serially diluted two-fold in parallel with the study sample. Samples were extracted on resin columns, (QIAamp, Qiagen, Hilden, Germany), amplified and electrophoresed on 2% agarose as described previously ^[13,14]. Gels were analysed by the Gel Doc 2000 documentation technique (Biorad, Segrate, Italy). The last positive dilution of the clinical sample was multiplied by the lowest detectable concentration of the standard and by 100 to give the final concentration per ml CSF. In preliminary experiments, 2000 copies of the standard were detected in 20 out of 20 assays, and 1000 copies were detected in eight out of 20 assays. Using the Poisson distribution, the detection limit of the assay was 1600 copies/ml. Therefore, for statistical purposes, an assay in which no JCV DNA was detected was given a value of 1599 JCV DNA copies/ml.

HIV RNA measurements

Plasma HIV RNA was measured using a branched DNA (bDNA) assay (Quantiplex 2.0, Chiron, Emeryville, California, USA) with a detection limit of 500 copies/ml. HIV RNA levels in CSF were determined either by an ultrasensitive reverse transcription (RT)–PCR (Roche, Branchburg, New Jersey, USA), with a detection limit of 20–50 copies/ml, or by NASBA (Organon Teknika, Baxtel, The Netherlands) with a detection limit of 80 copies/ml. For statistical purposes plasma HIV RNA values below the limit of detection of the bDNA assay were computed as 499 copies/ml and CSF values below the limit of detection of either RT–PCR or NASBA were computed as 79 copies/ml.

Statistical analysis

Viral concentrations in body fluids were transformed in log units/ml before calculations. Differences between proportions were analysed by the Chi-square test, differences between continuous variables by Student's t test or the Mann–Whitney U test, as appropriate. Survival analysis was performed using Kaplan–Meier curves. Cox proportional hazards regression was used to detect the independent association of variables with survival. Two-sided P values ≤ 0.05 were considered statistically significant. All analyses were performed using the Statistica 5.0 software package (Statsoft, Padova, Italy).

Results

Baseline characteristics of patients

Of 45 patients observed from September 1996 to February 2000, 40 fulfilled the inclusion criteria; five were excluded from analysis because of lack of diagnostic criteria. Diagnosis relied on the clinical and magnetic resonance imaging picture in all cases, a detectable JCV DNA in CSF in 38 cases and was histologically confirmed in four cases (post-mortem). Twenty-six patients were given HAART (group A) and 14 were given both HAART and cidofovir (group B). Baseline characteristics are summarized in Table 1. There was a higher proportion of injecting drug users and a lower proportion of heterosexual HIV transmission in group A. The other variables were homogeneous, including factors demonstrated previously to be associated with prolonged survival in AIDS-associated PML ^[15]. To exclude that the anti-CMV and anti-herpes simplex virus activity of cidofovir accounted for a better prognosis in the cidofovir-treated group, infection by Herpesviridae in the central nervous system was excluded by PCR of the CSF and clinical end-organ disease was excluded by clinical and fundoscopic examination in all patients.

Table 1: Baseline characteristics of patients according to treatment group.

Treatments and toxicity

Group A patients were treated with two nucleoside reverse transcriptase inhibitors (NRTI) plus one protease inhibitor (PI) (23 cases) and one or two NRTI plus one non-nucleoside reverse transcriptase inhibitor (NNRTI) plus one PI (three cases). Group B individuals were treated as follows: two NRTI plus one PI (11 cases), two NRTI plus one NNRTI plus one PI (two cases) and two NRTI plus two PI (one case) and they were treated concomitantly with intravenous cidofovir (5 mg/kg per week the first 2 weeks and every other week thereafter) in association with probenecid. The median number of cidofovir cycles was eight (range, 4–30). There was a single World Health Organization grade 3 toxicity (proteinuria), that led to permanent discontinuation of cidofovir and reverted after drug interruption. In three other cases (21%) there were delays in the administration of the drug due to mild proteinuria (WHO grade 1).

Neurological response

After 2 months of therapy, in an intention to treat analysis (using a noncompleter = non-responder principle) six out of 25 (24%) group A and eight out of 14 (57%) group B patients showed clinical response (P = 0.038). At the on-treatment analysis, responders were six out of 15 (40%) in group A and eight out of 14 (57%) in group B (P = non-significant). After 6 months, by intent-to-treat analysis, responders were seven out of 24 (29%) in group A and six out of 12 (50%) in group B (P = non-significant); on-treatment analysis showed six out of nine (66%) responders in group A and seven out of eight (87%) in group B.

Concentration of JCV DNA in CSF

The median baseline concentration of JCV in CSF was similar in the two groups (see Table 1). Two months after therapy (range, 45–75 days), JCV DNA showed a mean (± SD) change of + 0.07(± 0.77)log₁₀ copies/ml in group A and −1.27(± 1.36)log₁₀ copies/ml in group B (Mann–Whitney U test;P = 0.05). By this time, in five out of 12 (42%) tested patients from group A and in seven out of eight (87%) from group B JCV DNA in the CSF had become undetectable (P = 0.04) (see Fig. 1).

Fig. 1.: Proportion of PML patients with undetectable JCV DNA levels (< 1600 copies/ml) in CSF at diagnosis and after therapy. Dashed line, group A patients (HAART only); continuous line, group B patients (cidofovir plus HAART). Month indications are approximated ± 0.5 month. Difference of proportions at 2 months, P = 0.04.

HIV RNA and CD4 responses

The baseline plasma HIV RNA concentration was comparable between groups (see Table 1). The proportion of patients ever reaching < 500 HIV RNA copies/ml of plasma during follow-up was 71% in group A and 75% in group B (P = non-significant). Median baseline CSF HIV RNA levels were similar in the two groups (see Table 1). The mean (± SD) changes of HIV RNA in CSF after 2 ± 0.5 months of therapy were −0.45 ± 1.36 log₁₀ copies/ml in group A and −0.50 ± 1.41 log₁₀ copies/ml in group B (P = non-significant).

Baseline CD4 cell counts were homogeneous between the groups (see Table 1). Three months after therapy for PML, there was a mean (± SD) increase of 26 ± 88 × 10⁶ CD4 cells/l in group A and 58 ± 85 × 10⁶ cells/l in group B (P = non-significant), with 50% and 60% showing > 100 × 10⁶ CD4 cells/l at this time point, respectively (P = non-significant).

Survival analysis

The follow-up was censored at 31 March 2000. Twenty-one patients had died until this date (17 in group A and four in group B). All deaths were related to PML. Kaplan–Meier analysis showed that the group of patients treated with cidofovir in addition to HAART had a longer survival time than the group treated with HAART alone (see Fig. 2). The cumulative proportion of patients surviving after 1 year was 0.31 in the HAART group and 0.67 in the HAART plus cidofovir group (log-rank, P = 0.01). Baseline variables significantly associated with longer survival in univariate analysis were a CSF JCV load < 50 000 (4.7 log₁₀) DNA copies/ml (P = 0.017), a Karnofsky performance status ≥ 60 (P = 0.029), beginning HAART before the onset of PML (P = 0.046) and baseline CD4 cell counts > 100 × 10⁶/l (P = 0.05). HIV RNA levels in plasma and CSF, sex, age, HIV transmission categories and a history of previous AIDS-defining events did not influence survival. Follow-up variables significantly associated with a longer survival were a clinical response at 2 months (P = 0.007) and reaching undetectable levels of JCV DNA in the CSF (P = 0.01). The treatment group and all baseline variables significantly associated with survival were included in a multivariate model. Cox regression analysis showed that the use of cidofovir (P = 0.005), a baseline JCV DNA < 4.7 log₁₀ copies/ml (P = 0.011), HAART before the onset of PML (P = 0.027) and a baseline Karnofsky ≥ 60 (P = 0.041) were independently associated with longer survival.

Fig. 2.: Kaplan–Meier curves for patient survival according to treatment group. Continuous line, HAART plus cidofovir; dashed line, HAART only (log rank, P = 0.01).

Discussion

Our findings demonstrate that in HIV-seropositive patients with PML there is an overall virological and clinical advantage of treatment with cidofovir in addition to HAART over treatment with HAART alone. Major cidofovir-related toxicity was an uncommon event. With cidofovir there was a more rapid decline of JCV DNA in the CSF, an important prognostic marker of PML ^[14–17]. Moreover, a significantly higher percentage of patients on cidofovir showed clinical response, defined as a lack of neurological progression, after 2 months of therapy. There was a similar trend at 6 months that was not statistically significant probably because the number of patients was smaller. Finally, patients in the cidofovir plus HAART treatment group survived significantly longer than those in the HAART only group.

Cidofovir has an established activity and clinical efficacy against CMV and other herpesviruses. It is also the antiviral compound that produces the most significant inhibition of Papovaviridae in vitro^[8]. Several case reports have suggested that the intravenous administration of this agent may be clinically useful in AIDS-associated PML ^[9–12]. In agreement with our findings, a study on 10 patients showed that those on cidofovir and HAART cleared JC virus more rapidly from the CSF than those treated with HAART alone ^[18].

An obvious limitation of our study is its observational nature, which raises the question of possible bias in favour of one treatment group. Nevertheless, baseline characteristics, in particular variables proved previously to be relevant in terms of disease prognosis, were similar in both treatment groups. Also, HIV RNA and CD4 cell responses to treatment did not differ between groups, which demonstrates that effectiveness of HAART was not a bias in favour of one group. Finally, no patient from either group had evidence of CMV infection or end organ disease at baseline and all deaths were PML-related, so that the established anti-CMV activity of cidofovir should not have had a major influence on patients’ survival.

Despite the demonstrated benefit, cidofovir with HAART was not always effective. Two other baseline prognostic factors, a high JCV load in the CSF and a low Karnofsky performance status, were independently predictive of a shorter survival. This indicates that prompt diagnosis and early treatment of the disease may be important factors for obtaining the maximal benefit from this therapy.

To establish definitively safety and efficacy of adding cidofovir to HAART, a formal prospective, randomized, possibly blinded clinical trial is now needed. Such a study will require a multi-centre and probably multi-national effort to reach sufficient statistical power and it is unlikely to be completed in the near future. Meanwhile, we believe that the results with regard to safety and efficacy from the present study are convincing enough to recommend the use of HAART plus cidofovir for the treatment of AIDS-associated PML.