AIDS dementia complex (ADC) occurs in approximately 20% of patients with advanced HIV-1 infection . Response to antiretroviral therapy is variable  and one explanation for this problem may be the presence of resistance to antiretroviral therapy in the blood and central nervous system (CNS). Because ADC response to antiretroviral therapy may take several months, knowledge of potential drug resistance prior to commencement of therapy may be advantageous. However, there is little detailed knowledge of resistance patterns in blood and how this relates to the CNS.
The CNS and blood compartments may behave differently virologically. HIV-1 isolated from post-mortem brains is virologically distinct from that in blood, spleen and other organs [3–5]. Similarly, HIV-1 isolates from CSF can be genotypically and phenotypically distinct from paired blood specimens [6–10]. Furthermore, CSF viral load often differs from that of the blood [6,11–13]. There is also evidence that the severity of symptomatic HIV neurological disease may progress despite combination drug regimens being virologically efficacious systemically .
We therefore considered that antiretroviral resistance patterns would differ between the CSF and blood compartments, particularly in those patients with increased rates of viral replication within the CNS. We analysed the frequency and positions of mutations conferring resistance to four reverse transcriptase (RT) inhibitor drugs [zidovudine (ZDV), didanosine (ddI), zalcitabine (ddC) and lamivudine (3TC)] in paired CSF and blood from 49 patients with HIV-1 infection.
Specimens and study population
Fifty-three paired (same day) CSF and plasma specimens were collected from 49 HIV-1 infected patients undergoing investigations for a variety of CNS disorders. Samples were collected between 1994 and 1997 from patients who were well characterized clinically and part of a large CSF library. Inclusion in this study was based on the availability of sufficient, suitably stored CSF and blood. Four patients had repeat samples and were considered independent observations as the specimens were collected at least 1 month apart for different clinical reasons. All specimens were stored at −70°C within 6 h of collection. To determine blood brain barrier (BBB) impairment, the CSF to serum albumin ratio multiplied by 1 × 103 was calculated; a value > 6.5 was considered abnormal . HIV-1 plasma and CSF RNA concentrations were measured with the commercially available reverse transcription–PCR assay Amplicor HIV-1 Monitor (Roche Diagnostics, Branchburg, NJ, USA).
Spearman's correlation test was used to evaluate the relationship between plasma and CSF HIV-1 RNA levels in 39 of the 53 patients. Where results were at the limits of detection of the assay, 400 copies/ml was substituted for the lower limit, 750 000 copies/ml was used for results of greater than 750 000 copies/ml for the purpose of analysis. All results were then logarithmically transformed.
The following information was recorded: age, date of HIV-1 diagnosis, diagnosis of CNS disorders, presence or history of an AIDS defining illness, antiretroviral treatment history and CD4 lymphocyte counts.
Resistance genotypes were detected simultaneously using the reverse hybridization principle according to the manufacturer's instructions . The line probe assay (LiPA HIV-1 RT, Murex Biotechnology, Kent, UK) incorporates nested PCR using 50 μl CSF, plasma or serum. The primer sequences were: RT-9, 5′- GTACAGTATTAGTAGGACCTACACCTGTC-3′; RT-1, 5′-CCAAAAGTTAAACAATGGCCATTGA CAGA-3′; RT-4, 5′-AGTTCATAACCCATCCA AA-3′; RT-12, 5′-ATCAGGATGGAGTTCATAA CCCATCCA-3′. The resultant biotinylated DNA fragments of approximately 640 base pairs are hybridized to oligonucleotide probes immobilized as parallel lines in membrane strips. The 18 bands contain 33 probes detecting well characterized wild-type, mutation and several third letter base polymorphisms at or near codons 41, 69, 70, 74, 184, 214 and 215. Although the assay has greater sensitivity than the DNA sequencing method particularly in the detection of mixtures – detecting 2–8% of the total genome compared with 10–25% with a limited number of probes – novel mutations occurring in the RT region will not be detected resulting in null signals, i.e. no band forms for that codon. The limit of detection is defined at 1000 HIV RNA copies/ml.
Study population characteristics
The diagnoses of patients were: various stages of ADC (n = 15), cryptococcal meningitis (n = 11); cytomega lovirus CNS neurological disease (n = 5); progressive multifocal leukoencephalopathy (n = 4); and HIV-1 related neuropathies (n = 6). In the remainder of patients (n = 12) the lumbar puncture was performed for a variety of reasons: headache, seizures or back pain but the lumbar puncture associated investigations and subsequent clinical assessment did not reveal any significant findings. The male to female ratio was 24 : 1. The mean age was 38 years (range, 23–52 years). The median CD4 lymphocyte count was 30 × 106/l (range, 0–778 × 106/l). The mean time from HIV-1 diagnosis was 9.5 years (range, 3–14 years). Mean plasma and CSF HIV-1 RNA levels on 25 of the 31 patients were 5.35 log10 (median 4.95 log10; range, 2.6–5.9 log10) and 4.92 log10 (median, 3.67 log10; range, 2.6–5.9 log10) respectively. No significant correlation was found between plasma and CSF HIV-1 RNA levels (P = 0.84;r = 0.04).
HIV-1 RT sequences of both CSF and plasma specimens could be determined in 31 of 53 (58%) patients. Twenty-one of 53 (40%) CSF specimens failed to amplify compared with five serum samples (9%); these were classified as null signals and the patients’ paired samples were not used in further analysis. Thirteen of the 21 CSF samples had viral load data. Of the 13, 11 had viral loads < 400 copies/ml; three of the five serum samples also had < 400 copies/ml virus. Twenty-one of the remaining 31 (68%) specimens had identical genotypic profiles in the CSF and blood compartments; the other 10 (32%) paired specimens demonstrated obvious RT sequence differences between the CSF and blood compartments. There was no correlation between the presence of different or identical resistance profiles across compartments and evidence of blood–brain barrier impairment indicated by the CSF : serum albumin ratios (P = 0.917). Of the 21 patients who had identical resistance patterns in both compartments, 12 were at the same stage of HIV infection as the 10 discordant profile patients based on CD4 cell counts.
No CSF specimen tested contained amino acid changes at position 74 (Leu74Val) conferring cross-resistance to ddI and ddC. This mutation was only present as part of a mixed population in the blood of one patient. Four (12.9%) patients demonstrated genotypic resistance to 3TC characterized by a mutation at position 184 (Met184Val). Three of the four patients showed identical 184 mutation in both compartments and one patient showed resistance in blood but not in the CSF.
Results for the 10 patients with clear differences identified between both compartments are shown in Table 1. In this group there is a higher prevalence of mutations conferring resistance in the blood compartment than in the CSF. This group of patients also has a higher incidence of ADC diagnoses (seven out of 10) compared with the concordant group (three out of 21).
Patients 1, 2 and 8 showed high level ZDV resistance in their blood compartment, with fewer mutations in their CSF. However their time on ZDV varied considerably: with patient 2 had a 60 month treatment history and patient 8 had been taking ZDV for 1 month. Patient 8 also had a mixed population at codon 74 without having been exposed to ddI or ddC. Patients 3 and 7 demonstrated higher level resistance in their CSF sample than their blood but patient 7 had null signals in both blood and CSF. Patients 4 and 5 were sequential samples from the same patient taken 9 months apart. After 9 months, of the three ZDV resistance codons showing mixed populations, two showed wild-type populations where the third changed to mutant without change in antiretroviral drug resistance. A mutant population developed at codon 69 and a null signal was now detectable as an identifiable resistance mutation. Patient 6 demonstrated a mixed population of wild-type and resistant virus at position 70 (Lys70Arg) conferring resistance to ZDV in CSF but remained wild-type in the blood after 14.5 months of ZDV/ddI combination therapy. Patients 9 and 10 showed high level ZDV resistance in blood but remained sensitive to ZDV therapy in the CSF. Both patients 9 and 10 had null signals in their CSF.
We have shown that 10 of 31 patients with a variety of neurological disorders demonstrated clear differences in resistance patterns between their CSF and blood compartments. In clinical terms, a higher occurrence of ADC diagnoses in the discordant resistance profile group (seven out the 10) compared with that of the patients with identical resistance profiles is noted (three out of 21). This observation cannot be attributed to advanced stage HIV disease or to impairment of the blood–brain barrier in the discordant group. Other studies have shown that increased or continuous viral replication in the presence of subtherapeutic concentrations of antiretroviral drugs may promote the selection of drug resistant mutants . This may be particularly important for patients with ADC receiving antiretroviral drugs in doses which yield low drug concentrations in the CNS. This may be true for some of the patients (patients 3,6 and 7) but it cannot be the whole explanation. CSF concentrations of antiretroviral drugs are known to vary among individuals; perhaps in some individuals the concentrations are adequate to prevent viral escape.
These results may also make it possible to give an approximation of how often resistance mutations occur in the CSF when they are present in the blood. While the lumbar punctures were performed for clinical reasons and not to assess resistance patterns, it is instructive that three of the 16 patients with resistance in the blood (19%) did not have resistance in the CSF. In the 10 ADC patients, eight had resistance mutations in the blood and two of these eight did not have resistance in the CSF. Conversely, of the 15 patients who did not have resistance mutations in the blood, two demonstrated resistance in the CSF and both of these patients had ADC. These data further suggest the importance of resistance testing in both compartments, especially in ADC patients.
There are however, a number of limitations to this study. First, the assay system itself. HIV-1 RNA from CSF could not be amplified from 40% of patients. Viral load data shows low levels of RNA in the CSF samples as the probable reason for amplification failure. There was also a problem with null signals occurring, perhaps due to non-consensus mutations at or around these codons, a problem reported previously . Second, we have no data on phenotypic antiretroviral drug susceptibility of the isolates. This is also problematic because the selected isolate identified may not be the dominant quasispecies conferring resistance in vivo. However, other studies have shown that these assays do correlate with phenotypic resistance . Third, the analysis of CSF resistance profiles may not reflect what is occurring in the brain; however, a strong correlation between CSF viral load and the severity of ADC has been identified [12,20,21].
In summary, these data have clinical implications: it may be useful to determine the resistance profiles of both CSF and blood before selecting or changing an antiretroviral therapeutic regimen, this is especially true for patients with ADC.
1. McArthur JC, Hoover DR, Bacellar H. et al. Dementia in AIDS patients: incidence and risk factors.
Neurology 1993, 43: 2245 –2252.
2. Siditis J, Garsonis C, Price RW. et al. Zidovudine for the treatment of AIDS dementia complex: results of a placebo-controlled trial.
Ann Neurol 1993, 33: 344 –349.
3. Sheehy N, Desselberger U, Whitwell H, Ball JK. Concurrent evolution of regions of the envelope and polymerase genes of human immunodeficiency virus type 1 observed during zidovudine (AZT) therapy.
J Gen Virol 1996, 77: 1071 –1081.
4. Wong JK, Ignacio CC, Torriani F. et al. In vivo compartmentalization of human immunodeficiency virus; evidence from the examination of pol sequences from autopsy tissue.
J Virol 1997, 71: 2059 –2071.
5. McGavin CH, Land S, Sebire KL, Hooker DJ, Gurusinghe AD, Birch CJ. Syncytium-inducing phenotype and zidovudine susceptibility of HIV-1 isolated from post-mortem tissue.
AIDS 1996, 10: 47 –53.
6. Brew BJ, Evans L, Byrne C, Pemberton L, Hurren L. The relationship between AIDS dementia complex and the presence of macrophage tropic and non-syncytium inducing isolates of human immunodeficiency virus virus type 1 in the cerebrospinal fluid.
J Neurovirol 1996, 2: 152 –157.
7. Wildemann B, Haas J, Ehrhart K. et al. In-vivo comparison of zidovudine resitance mutations in blood and CSF of HIV-1 infected patients.
Neurology 1993, 43: 2659 –2663.
8. Di Stefano M, Norkrans G, Chiodi F, Hagberg L, Nielson C, Svennerholm B. Zidovudine-resistant variants of HIV-1 in brain.
Lancet 1993, 342: 865. 865.
9. Di Stefano M, Monno L, Fiore JR. et al. Neurological disorders during HIV-1 infection correlate with viral load in cerebrospinal fluid but not with virus phenotype.
AIDS 1998, 12: 737 –743.
10. Chien JW, Valdez H, McComsey G, McClernon D, St. Clair M, Lederman MM. Presence of mutation conferring resistance to lamivudine in plasma and cerebrospinal fluid of HIV-1-infected patients.
J Acquir Immune Def Syndr 1999, 21: 277 –280.
11. Di Stefano M, Monno L, Fiore JR, Angarano G. In-vivo evidence of HIV-1 productive infection in cerebrospinal fluid of patients with HIV-1 encephalitis.
AIDS 1997, 11: 133 –135.
12. Brew BJ, Pemberton L, Cunningham P, Law MG. Levels of human immunodeficiency virus type 1 RNA in cerebrospinal fluid correlate with AIDS dementia stage.
J Infect Dis 1997, 175: 963 –966.
13. Gisslen M, Fuchs D, Svennerholm B, Hagberg L. Cerebrospinal fluid viral load, intrathecal immunoactivation, and cerebrospinal fluid monocytic cell count in HIV-1 infection
. J Acquir Immune Def Syndr 1999, 21: 271 –276.
14. Pialoux G, Fournier S, Moulignier A, Povedda J-D, Clavel F, Dupont B. Central nervous system as a sanctuary for HIV-1 infection despite treatment with zidovudine, lamivudine and indinavir.
AIDS 1997, 11: 1302 –1303.
15. Eeg-Olofsson O, Link H, Wigertz A. Concentrations of CSF proteins as a measure of blood brain barrier function and IgG synthesis within the CNS in 'normal’ subjects from the age of 6 months to 30 years.
Acta Paediatr Scand 1981, 70: 167 –170.
16. Stuyver L, Wyseur A, Rombout A. et al. Line probe assay for rapid detection of drug-selected mutations in the human immunodeficiency virus type 1 reverse transcriptase gene.
Antimicrob Agents Chemother 1997, 41: 284 –291.
17. Condra JH, Emini EA. Preventing HIV-1 drug resistance.
Science Med 1997, 4: 1 –11.
18. Koch N, Yahi N, Colson P, Fantini J, Tamalet C. Genetic polymorphism near HIV-1 reverse transcriptase resistance-associated codons is a major obstacle for the line probe assay as an alternative method to sequence analysis.
J Virol Methods 1999, 80: 25 –31.
19. Larder BA, Kellam P, Kemp SD. Zidovudine resistance predicted by direct detection of mutations in DNA from HIV-infected lymphocytes.
AIDS 1991, 5: 137 –144.
20. Ellis RJ, Hsia K, Spector SA. et al. Cerebrospinal fluid human immunodeficiency virus type 1 RNA levels are elevated in neurocognitively impaired individuals with acquired immunodeficiency syndrome.
Ann Neurol 1997, 42: 679 –688.
21. McArthur JC, McClernon DR, Cronin MF. et al. Relationship between human immunodeficiency virus-associated dementia and viral load in cerebrospinal fluid and brain.
Ann Neurol 1997, 42: 689 –698.