A case of multidrug resistance in the central nervous system

INTRODUCTION

The central nervous system (CNS) becomes infected in primary HIV infection, particularly targeting perivascular macrophages and microglial cells. ¹ This infection can damage neurons leading to a spectrum of neurocognitive impairment. This may result in HIV-associated dementia. There has been a continued improvement in antiretroviral therapy (ART), which has aided longevity and significantly reduced the incidence of severe HIV-related neurocognitive impairment. ² However, HIV-1 may persist in the CNS, despite adequate ART, as not all drugs penetrate the CNS equally and not all HIV-associated dementia responds to ART. ³

The importance of adherence to medications and the need for three active drugs is an important component in the efficacy of ART. ^4,5 We report this case to highlight the possibility of resistant virus developing in the CNS despite peripheral viral suppression. This case discusses severe cognitive decline in a man with HIV-1 infection on a fully active regimen for five years. This is a very unusual case and at the time of writing there were no other reports of CNS ART failure in the absence of plasma failure.

CASE REPORT

This 55-year-old man, with HIV-1 infection (subtype B), was admitted to hospital with Staphylococcus aureus meningitis in June 2004. He had previous exposure to zidovudine, didanosine, tenofovir, lamivudine, saquinavir and efavirenz. Poor adherence to medication had resulted in persistent low-level viraemia for five years. At this time his CD4 count was 150 × 10⁶/L (35%) (previous nadir CD4 count 170 × 10⁶/L [34%]) and plasma HIV RNA level was 3.26 log₁₀ RNA copies/mL (baseline at diagnosis 5.2 log₁₀ RNA copies/mL). A genotypic resistance profile amplified a number of key reverse transcriptase (RT) mutations including M41L, L210W and T215Y. The viral population was fully susceptible to all non-nucleoside RT inhibitors (NNRTI) and protease inhibitors (PIs). A new regimen of Combivir^® (zidovudine and lamivudine), ritonavir-boosted lopinavir and efavirenz was commenced. Two months later, virological suppression was achieved and maintained for the following four years. During this period, in order to reduce the potential for long-term toxicity, a drug switch was made from zidovudine to abacavir with no change in plasma virological control. It should be noted that throughout this time the patient was in full-time employment and completely independent in daily life.

In April 2008, he was admitted to hospital with a sudden marked cognitive decline. Over the course of several weeks, he had become increasingly confused and emotionally labile with significant deterioration in mobility. On examination, he was disoriented in time and place, with loss of attention, recall and ability to follow complex commands. He scored 5 out of 30 on the Folstein test. ⁶ There was no loss of peripheral power, sensation or reflex response. Clinical, biochemical or bacteriological investigations revealed no cause. Serology for syphilis, hepatitis B and C was negative. The plasma viral load was still undetectable and CD4 count was 760 × 10⁶/L (44%). Magnetic resonance imaging (MRI) of the brain showed extensive, diffuse, ill-defined areas of increased T2-weighted signal intensity in the white matter of the cerebral hemispheres. The report suggested HIV-associated encephalopathy or progressive multifocal leukoencephalopathy. Cerebrospinal fluid (CSF) analysis revealed a protein level of 0.97 g/L (normal range 0.15–0.4), glucose level in the normal range, white blood cells 20 cells/μL (mostly lymphocytes) and erythrocytes <1 cell/μL. Cryptococcal antigen, syphilis, JC virus, BK virus, picornavirus, cytomegalovirus, Epstein–Barr virus, herpes simplex virus type 1 and 2 infections were all excluded in the CSF. Direct microscopy for acid-fast bacilli was negative and the mycobacterial culture grew no organisms. Cytopathological examination did not reveal tumour cells. HIV RNA was measured in the CSF revealing 3.52 log₁₀ RNA copies/mL. Consequently, HIV encephalopathy was the working diagnosis. Zidovudine was added to the antiretroviral regimen.

Over the course of several weeks his clinical condition improved dramatically, enabling him to return to full independence. His plasma viral load remained undetectable and the CD4 count was 550 × 10⁶/L (44%). Unfortunately, eight months later, in January 2009, he had a similar deterioration in his cognitive state. He was re-admitted to hospital confused, incontinent of urine and faeces, and completely dependent for all activities of daily living. He had travelled to Egypt in the preceding months. He was febrile and, although not able to comply fully with examination, was noted to have dysarthric speech, lip-smacking and dyskinetic movements of his upper and lower limbs. The Folstein score was again low, with three out of 30 achieved. Initially, empirical intravenous antibiotics (ceftriaxone 2 g twice daily and benzylpenicillin 2.4 g 4-hourly) and acyclovir (10 mg/kg, 8-hourly) were commenced to treat bacterial meningitis and herpes encephalitis. Computed tomography of the brain showed areas of low attenuation in the white matter of both frontal lobes and no intracranial mass was evident. Further evaluation with MRI showed similar but more extensive changes when compared with the previous scan in September 2008. This was again in keeping with HIV encephalopathy. This was further supported by an electroencephalogram showing background slow wave activity across both sides of the brain. CSF analysis revealed elevated protein (1.04 g/L), normal glucose, five white bloods cells/μL and <1 erythrocyte/μL. Infective agents, as in the previous episode, were again excluded from CSF and serum, including Brucella, Borrelia and arboviruses. Despite viral suppression in blood, the CSF HIV viral load was 3.72 log₁₀ RNA copies/mL. Genotyping of this sample showed an extensive pattern of resistance with multiple RT mutations, including M41L, D67N, M184I, L210W and T215Y. Key NNRTI mutations were also identified: K101Q and Y188L. Several PI mutations were amplified, including I84V and L90M. A new regimen was commenced containing maraviroc, etravirine, raltegravir, nitonavir–boosted darunavir and enfuvirtide. Within two weeks he began to make a clinical recovery with improvement in orientation and communication. The patient is now alert and oriented with a Folstein score of 30 out of 30. He is fully independent in activities of daily living and considering a return to work. His plasma viral load remains undetectable and the CSF viral load in August 2009 was also undetectable.

DISCUSSION

The most notable aspect of this case was finding detectable virus in the CSF despite an undetectable plasma viral load. There was a divergent resistant population developing in the CNS coinciding with the plasma viral suppression, as evidenced by the very different resistance profiles. This patient had a history of adherence problems and small populations of resistant virus may have been present, which migrated to CNS during the early staphylococcal meningitis illness. A key feature of HIV populations in the CSF is that initially they may be identical to that in the plasma, but as the infection progresses the viral populations diverge, with the greatest divergence being observed in patients with HIV-associated neurocognitive impairment. ^7,8

Problems with adherence are associated with subtherapeutic drug concentrations. ^4,5 However, the drug concentrations may not be the only factor as the regimen may have had inadequate CNS penetration. Concerns have been raised in the literature about the potential for limited penetration of some antiretroviral drugs through the blood–brain barrier and that this could lead to incomplete suppression of viral replication in the CNS. ⁹ Such occurrence would have important implications here, where elevated viral replication levels can be associated with HIV-related tissue damage and neurological disease. Letendre et al. ^10,11 devised and recently updated the CNS penetration-effectiveness (CPE) score, which grades each drug from 1 (poor) to 4 (good) on the basis of its ability to both work in and penetrate the CNS. ^10,11 Findings from the CHARTER study show that the CPE score correlates highly with the likelihood of detectable HIV RNA in CNS. In those receiving an ART regimen with CPE score greater or equal to 9, 9% had detectable viral load in CSF compared with 46% in those on a regimen with a CPE score of 3 or below. ¹¹ Although there was no difference in neurocognitive function, this may be due to virus levels in CSF being low or perhaps a subtle deterioration may become more clinically apparent with time on a regimen with reduced CPE score. In this case, the regimen at the time of the first episode of neurocognitive impairment had a CPE score of 11. This suggests that although this regimen was able to penetrate the CNS, selective resistance may have resulted in poor virological control within this compartment. The addition of zidovudine after this first episode of neurocognitive decline improved the CPE score to 15. Following the second episode, a new regimen was constructed containing agents from new drug classes to which he had no previous exposure. This regimen containing several fully active drugs with good CNS penetration (CPE score 12) has resulted in full neurocognitive recovery, which hopefully will be sustained.

Resistant virus may develop as a result of subtherapeutic drug levels. This is due to patient factors, such as non-adherence, or pharmacological factors, such as inadequate drug penetration. The CSF appears to be a distinct compartment for virological infection with HIV-1 and thus may develop a separate viral population. ¹² The trafficking of immune cells between CSF and plasma may have facilitated resistant virus entry into his CNS compartment. ¹ There are also some observational data supporting the hypothesis that resistant strains can develop selectively in the CNS. ¹² Several studies comparing CSF and plasma sequences have shown different resistant mutations in the two compartments, with mutations found in CSF but not in plasma. ^12–14 However, Bestetti et al. ⁹ found that failure in CSF only occurred along with failure in plasma. The present case study reports plasma virological suppression with detectable resistant virus in another site in the CSF.

CONCLUSION

This case highlights the ability for drug-resistant mutations to develop in the CNS compartment while plasma virus remains suppressed. It also highlights the need for more routine use of CNS viral load measurement and possible use of resistance profiling in this compartment.