Changes in Cerebral Function Parameters in HIV Type 1-Infected Subjects Switching to Darunavir/Ritonavir Either as Monotherapy or with Nucleoside Analogues

E ditor: In recent years nucleoside-sparing treatment simplification in HIV-1-infected subjects established on combination antiretroviral therapy (cART) has been investigated in an attempt to reduce therapy-related toxicities. ^{1 –3} However, concern exists that utilization of such strategies may result in cognitive sequelae, owing to reduced central nervous system (CNS) antiviral activity. ^4,5 The aim of this 48-week study, performed within the MONET clinical trial, ⁶ was to assess prospective changes in cerebral function parameters in HIV-1-infected subjects switching antiretroviral therapy to either darunavir/ritonavir alone (DRVmono) or with nucleoside analogues (DRVnrti).

The study was conducted at two London hospitals between 2007 and 2009. Ethical approval was obtained via the National Research Ethics Service. HIV-1-infected adults, receiving stable cART [including two nucleoside analogues + a boosted protease inhibitor (PI)] with a plasma HIV RNA level <50 copies/ml for at least 3 months prior to entry were enrolled. Exclusion criteria included active neurological disease, current major depression or psychosis, recent head injury, current use of recreational drugs, or alcohol abuse. Eligible subjects were randomized one to one, in an open-label fashion, to receive darunavir 800 mg and ritonavir 100 mg both once daily (arm 1: DRVmono) or darunavir 800 mg and ritonavir 100 mg once daily plus any two nucleoside analogues (arm 2: DRVnrti). Patients were then followed prospectively with a medical review at 3-monthly intervals.

Cerebral function was assessed at baseline and week 48 via a validated, computerized neurocognitive assessment (CogState) ⁷ and cerebral proton magnetic resonance spectroscopy (MRS). Neurocognitive data were analyzed according to recommendations of the manufacturer. MRS data were acquired by single voxel examination in the frontal white matter, frontal gray matter, and right basal ganglia, using a double spin echo point resolved spectroscopy (PRESS) sequence with the following settings: echo time (TE) 36 ms, repetition time (TR) 3000 ms, 2048 data points, spectral width of 2500 Hz, and 128 data acquisitions. Cerebral metabolites [including N-acetylaspartate (NAA), myoinositol (mI), and choline (Cho)] were measured and expressed as ratios to cerebral creatine (Cr). Associations between parameters and study treatment arms were evaluated using linear regression.

Overall, six subjects were enrolled (three assigned to each treatment arm) and five completed all study procedures. The mean age was 44 years (SD 5) and 80% were male. All patients were receiving boosted PI-based therapy (see Table 1) but were naive to darunavir at study entry. Mean baseline plasma CD4⁺ cell count was 535 cells/μl (SD 172). Over 48 weeks, mean score improvements were observed in six of the eight neurocognitive tasks assessed. These included identification speed (0.7 log₁₀ ms faster), visual learning (accuracy increased by 0.23 arc. proportion correct), and executive function (error rate reduced from 45 to 34, see Table 1). Analysis of MRS findings revealed reductions in cerebral metabolite markers of inflammation (Cho/Cr and mI/Cr ratios) in all cerebral locations assessed (maximum reduction of mI/Cr ratio in frontal gray matter). No consistent changes in NAA/Cr ratios were observed. No association between study treatment arm and cerebral function parameter improvements (p > 0.06 all values) or between the observed changes in Cho/Cr or mI/Cr ratios and neurocognitive test scores were observed.

In this small, prospective study, we have observed improvements in cerebral function parameters over 48 weeks, in five subjects switching therapy to either DRVmono or DRVnrti. Features of the neurocognitive improvement included speed tasks, learning, accuracy, and executive functioning. Additionally, we have observed reductions in markers of cerebral inflammation in three cerebral locations between baseline and week 48. The clinical meaning of these changes requires further elucidation. They may represent the early CNS manifestations of switching to an antiretroviral regimen with improved CNS antiviral activity, or may be related to cessation of drugs associated with CNS mitochondrial toxicities, such as didanosine and abacavir. ⁸ Darunavir, despite being approximately 95% protein-bound, appears to cross the blood–brain barrier to concentrations exceeding the median inhibitory concentration required for HIV-1 antiviral activity in cerebrospinal fluid in the majority of subjects. ^9,10 This is in contrast to the cerebrospinal fluid concentrations observed with alternative PIs including atazanavir and saquinavir, ^11,12 which the majority of subjects were receiving at baseline.

The cerebral metabolites Cho and mI are present in glial cells and become elevated upon cell membrane injury or with glial activation and increased concentrations of these compounds correlate with advanced HIV-1 disease and dementia. ¹³ It is therefore reassuring that following enrollment to this study, favorable Cho and mI metabolite shifts were observed, representing reductions in cerebral inflammation. In this study, we did not observe any meaningful increase in NAA, a cerebral metabolite representing neuronal integrity. ¹⁴ Reduced NAA/Cr ratios are observed in advanced disease stages, including AIDS dementia complex and severe neurocognitive impairment ^15,16 and increases in NAA/Cr ratio following initiation of cART have previously been described in antiretroviral-naive individuals. ¹⁷ It is likely, however, that the absence of any change to this ratio in our study was due to a high mean plasma CD4⁺ cell count (535 cells/μl) and lengthy duration of virological suppression prior to study entry.

Improvements in neurocognitive test scores may be attributed to a learning effect, whereby performance improves upon retesting. To minimize this effect, however, subjects underwent a full practice test prior to study entry. Furthermore, as simultaneous improvements in metabolite markers of cerebral inflammation were also observed, it is likely the changes represent real cerebral function improvements, rather than a learning effect. We did not observe any relationship between changes in neurocognitive test scores and cerebral metabolite ratios, which may reflect the small numbers in this study.

Despite 256 subjects being enrolled within the parent MONET study, owing to competitive recruitment around Europe and the time taken to gain regulatory approval for this substudy in the United Kingdom, unfortunately very few subjects participated. However, as the majority of UK MONET subjects were enrolled in this substudy, recruitment bias was minimal. Due to the study size, it was not possible to perform multiple adjustments for potential clinical parameters and education level, which may have an influence upon neuropsychological performance. In summary, this work demonstrates that objective, noninvasive tools can practically be administered and prospectively assess cerebral function in HIV-1-infected individuals and can be utilized in future HIV-1 treatment trials.