Clinical Manifestations of Discordance between Cerebrospinal Fluid and Plasma HIV‐1 Loads

Sir—The demonstration of discordance between HIV-1 evolution in CSF and plasma (with greater drug resistance occurring in the CSF, as described by Tashima et al. [1]) underscores the importance of recognizing the CNS as a distinct compartment. This is particularly important with respect to obscure white matter encephalopathy, as is illustrated by the following description of a woman who developed progressive encephalopathy that was characterized by a high CSF HIV-1 load, despite the presence of reasonably well-controlled plasma viremia, and who experienced remission of encephalopathy coincident with a reduction in the CSF HIV-1 load. In September 2000, a 34-year-old woman was seen at Jacobi Medical Center (Bronx, NY); she had a CD4 cell count of 2 cells/mm and an HIV-1 load of 58,194 copies/mL (as determined by RT-PCR; Roche Amplicor), and she had previously developed cryptococcal meningitis in 1994. She manifested tremulousness and marked cognitive impairment characterized by difficulty understanding instructions and poor memory. When she began receiving antiretroviral treatment (ART), her cognitive function improved. In December 2001, despite having an increased CD4 cell count of 192 cells/mm and a reduced HIV-1 load of 2199 copies/mL, she experienced neurological deterioration that recapitulated her original presentation. Brain MRI showed diffuse abnormalities in the subcortical and deep white matter. Examination of the CSF revealed a WBC count of 12 cells/mm (98% lymphocytes), a protein level of 123 mg/dL, and a glucose level of 58 mg/dL. The results of cultures of CSF samples for fungi, mycobacteria, and bacteria were negative, as were the results of serologic tests for cryptococcal antigen and syphilis. Further neurological deterioration ensued. In March 2002, examination of the CSF revealed a WBC count of 46 cells/mm (98% lymphocytes), a protein level of 148 mg/dL, and a glucose level of 50 mg/dL. Results of CSF PCR for the detection of JC virus and cytomegalovirus were negative, as were the results of repeated cultures and serologic tests. Because of its ability to penetrate the CNS, efavirenz was added to the patient’s regimen of lopinavir/ritonavir, lamivudine, and abacavir, despite the persistence of a K103N mutation that resulted from prior receipt of efaviranz. The patient experienced transient improvement in her condition, but, by June 2002, she experienced further deterioration despite maintaining a CD4 cell count of 212 cells/mm. In addition to cognitive dysfunction, she manifested hyperreflexia with sustained clonus and nearly constant fine myoclonus, exacerbated by action, that involved her extremities and trunk and prevented ambulation. During a 2week period, her condition worsened. The patient could no longer ambulate. She was disoriented and confused, as was evidenced by her speech, which was characterized by nearly incomprehensible muttering. Examination of the CSF revealed the following findings: a WBC count of 156 cells/mm (80% lymphocytes), a glucose level of 37 mg/dL, and a protein level of 200 mg/dL; results of repeated CSF cultures, serologic tests, and PCR (for the detection of JC virus, herpes simplex virus, and varicella-zoster virus) were negative. The plasma HIV-1 load was 7392 copies/mL. By contrast, her CSF HIV-1 load was 266,667 copies/mL. Biopsy of nondominant frontal white matter obtained from the patient’s brain showed HIV encephalitis without viral inclusions, necrosis, demyelination, or the presence of microorganisms. The ART that the patient was receiving was intensified, according to her plasma HIV-1 genotype. Zidovudine and tenofovir were added to therapy, and administration of efavirenz was stopped. The patient’s condition improved remarkably. In July 2002, she was alert, oriented, and ambulatory, and she had fluent speech; she had no abnormal movements or sustained clonus. Examination of the CSF revealed a WBC count of 20 cells/mm (100% lymphocytes), a protein level of 112 mg/dL, and a glucose level of 48 mg/dL. The CSF HIV1 load decreased to 248 cells/mm, and the plasma HIV-1 load was !50 copies/mL. Levels of HIV-1 in plasma typically are substantially higher than are those in CSF, as demonstrated by 5 cases reported by Tashima et al. [1] and by additional cases reported elsewhere [2, 3]. Among patients with dementia, CSF HIV-1 levels are elevated, compared with those of patients without dementia [3, 4]; CSF HIV-1 levels may also exceed plasma levels in up to 80% of patients with dementia [5]. Logically, the occurrence of discordance between CSF and plasma virus loads and genotypes would presage the occurrence of discordance between the virologic responses of CSF and plasma to ART. The effectiveness of ART, with respect to HIV encephalopathy, may not be adequately characterized by the plasma virus load. Further studies will determine the role of measuring the CSF HIV-1 load and determining the HIV-1 genotype for patients with HIV encephalopathy. It is fortunate that this patient had a favorable response without the benefit of a CSF HIV-1 genotype.