Introduction
The introduction of highly active anti-retroviral therapy (HAART) for the treatment of HIV-1 infection has had a dramatic effect on AIDS-related morbidity and mortality in the industrialized world [1]. However, the majority of HIV-1-infected individuals live in sub-Saharan Africa and Asia [2], where, due to financial constraints, HAART is not an option and, therefore, its efficacy remains unproven.
There is enormous geographical sequence variability in HIV-1 isolates and this diversity may impact on infectivity, clinical manifestations and, possibly, response to therapy. The majority of African subtypes (A, C and D) belong to the 'M' (major) group of HIV-1 isolates, which also includes subtype B (prevalent in Europe and North America). The M group is believed to have diverged from a point source following zoonotic introduction from a chimpanzee host [3]. DNA sequence analyses of env, gag or pol have been used to determine subtype [4].
Information on the outcome of infection by different subtypes has only recently started to emerge. For example, data from the Kyamulibwa cohort in Uganda [5], may suggest that patients infected with subtype D strains progress to death more rapidly than those with subtype A infections, although other data suggest that subtypes A, B, C or D have no bearing on disease progression [6].
Concerns exist regarding the susceptibility of African HIV-1 subtypes to anti-retroviral agents designed, tested and validated against the European and North American subtype B strain. Certain non-B subtypes appear as susceptible as B subtypes strains to HAART [7,8], although some subtypes (e.g. D [9] and F [10] may be less susceptible in vitro. There are few supporting clinical data on African patients treated with anti-retrovirals, and these are mostly from studies on mono- or dual-therapy [11-13]. For example, the HIVNET 012 [11] study from Uganda showed that the risk of HIV-1 transmission from mother to child during the first 14-16 weeks of life was lowered by both zidovudine and nevirapine monotherapy, the latter by nearly 50% in a breast-feeding population. However, the absence of regular HAART in African cohorts has limited our understanding of potential efficacy in these populations.
The development of drug-resistance is one major problem associated with not responding to, or 'failing', therapy, leading to resistance testing becoming a key component of care in many major western HIV centres [14]. The development of resistance by some African subtypes to HAART has been reported by several groups. In the HIVNET-006 study [12], 20% of women receiving nevirapine monotherapy developed the K103N mutation, associated with non-nucleoside reverse transcriptase inhibitor (NNRTI) resistance. In addition, similar patterns of resistance mutations have been found in Brazilian A and F subtypes, in comparison with B subtypes [7]. African strains of HIV-1 harbour polymorphisms at variance to the consensus B subtypes at sites within pol which may impact on the development of drug resistance [15,16]. Polymorphisms at codons not associated with resistance in an un-subtyped cohort have been associated with the development of resistance [17]. However, there are no data on the influence of baseline polymorphisms on the development of resistance and the outcome to therapy in African patients.
Nevertheless, despite the absence of data from Africa, it is possible to study the impact of HAART on emigrant African HIV-1-positive patients treated overseas. For example, a computerized database at St. Mary's Hospital details over 4000 HIV-1 positive patients studied since 1982, of which 371 patients were infected in Africa, the majority in Uganda. The objective of this study was to investigate whether baseline polymorphisms in HIV-1 reverse transcriptase (RT) and protease in these HIV-1 infected African patients impacted on their virological response to HAART, and resulted in alternative drug-resistance pathways.
Patients and methods
Patients
The HIV-1 clinical database at St. Mary's was used to identify patients infected in sub-Saharan Africa with non-B viral subtypes. On first presentation at the clinic, all patients complete a questionnaire that includes details of 'ethnic status', 'country of origin' and 'risk factor'. These questionnaires were used to identify those who were 'Black African', heterosexual and born in Africa. In order to guarantee the accuracy of the database, corresponding patient notes were also examined. From these groups only patients who were drug-naive prior to starting a HAART regime, who were maintained on the same regime for either 1 year or until failure, and who had at least one viral load measurement pre- and post-therapy were selected for the study.
Of the 371 African patients listed on the database, 79 either had undetectable viral load after 1 year of therapy or experienced virological rebound (`failed') prior to this time. These patients were divided, accordingly, into 'responders' and 'non-responders'. The cohort was divided further into those receiving NNRTI- and those receiving protease inhibitor (PI)-based HAART, to determine any difference in outcome of non-B infected patients on these therapies. Five patients had undetectable viral load on HAART after 1 year but rebounded after 15, 16, 17, 18 and 22 months. They were categorized as 'responders' for the analysis on outcome. The majority of patients were from Uganda (33), Zambia (eight) and Congo (eight). The rest were from Zimbabwe (five), Kenya (four), Rwanda (three), Ethiopia (three), Tanzania (three), Nigeria (two), South Africa (two), Malawi (two), Botswana (one), Cote d'Ivoire (one), Somalia (one), Eritrea (one), Angola (one) and Cameroon (one).
Methods
Baseline pre-therapy plasma samples and those from patients 'failing' post-therapy were recovered from storage and sequenced in pol using the ViroSeq vII. HIV Genotyping System (Applied Biosystems, Foster City, California, USA), according to the manufacturers instructions. Briefly, RNA was extracted from pelleted HIV-1 from patient plasma using isopropanol/ethanol precipitation, converted to cDNA using kit reagents and then amplified by polymerase chain reaction (PCR). The amplicon constituted approximately 1500 base pairs of pol, including all of protease and the first 320 codons of RT. The PCR product was sequenced in both the 5′ prime and 3′ prime directions using seven different primers for each sample (four upstream, three downstream) and Big Dye terminator cycle sequencing chemistry. All amplification and sequencing reactions were carried out on a 9600 thermal cycler (Perkin Elmer, Warrington, UK). After ethanol purification, the sequences were read on an ABI 310 Genetic Analyser. 'ViroSeq HIV Analysis' software (ABI) was used to compare protease and the first 320 codons of RT with the consensus sequence for subtype B HIV-1.
In order to confirm that the patients were infected with 'non-B' viral subtypes, phylogenetic analysis by means of Jukes and Cantor neighbour-joining methodology was used to compare patient pol sequences with well-characterized isolates from the Los Alamos database (http://www.hiv-web.lanl.gov./). Certain sequences did not appear to group within one subtype and were submitted to the National Centre for Biotechnology Information (NCBI) (http://www.ncbi.nlm.nih.gov/retroviruses/subtype/subtype.html) for subtyping using the 'BLAST' algorithm. Two African patients were infected with subtype B HIV-1 and, in both cases, the individuals had reported homosexuality as their risk factor. As a result, both these patients and all African male homosexuals were, retrospectively, excluded from the analysis.
Statistical analysis
Comparison of outcome within the cohort used paired and independent Student's t test for viral load and CD4 cell data. The impact of codon changes within pol was examined using the χ2 test and Fisher's exact test for discrete variables.
Results
African cohort response to HAART
Seventy-nine patients were identified as a 'non-B' group, defined as being 'Black African', born in Africa, not homosexual and having continued a HAART regime either for 1 year or until virological failure, and being initially drug-naive. Patients were grouped according to whether the triple-drug HAART regime contained either a PI or NNRTI in conjunction with a double nucleoside reverse transcriptase inhibitor (NRTI) backbone. Thirty-nine patients received PI-based and 40 NNRTI-based therapy. Of the former, eight received indinavir, seven received ritonavir, 16 received nelfinavir, six received saquinavir and two received a ritonavir/saquinavir combination as part of the HAART regime. Of the NNRTI group, 36 received nevirapine and four received efavirenz. There was no difference in the baseline viral load (P = 0.80) between patients receiving PIs and those receiving NNRTIs, although the former had a lower CD4 cell count (P = 0.01) (Table 1), which may reflect earlier prescribing trends as PIs were introduced into clinical practice before NNRTIs.
African patients who had undetectable viral load for at least 1 year on a single HAART regime were classed as 'responders'. Failing patients whose viral load rebounded towards baseline, or those whose viral load had never been undetectable within 1 year, were classed as 'non-responders'. Although a higher percentage of the NNRTI-receiving patients responded to therapy, this was not significant (82.5 versus 69.2% for PI-based HAART, P = 0.26). There was no significant difference in outcome to therapy according to country of origin (Table 1).
The rate of increase of CD4 cell count on therapy was greater for patients on PI- than NNRTI-based HAART (mean increase of 16.2 versus 11.3 × 106 cells/l per month, P = 0.07), which may be a reflection of the lower baseline CD4 cell values in the former. Responders in both NNRTI- and PI-HAART groups achieved greater rates of increase in CD4 cell count (11.7 and 17.4 × 106 cells/l per month, respectively), than the non-responders (6.8 and 10.7 × 106 cells/l per month, respectively) (Table 1).
Association between subtype and response to therapy
The most prevalent subtypes in the cohort were A (25 of 68), C (23 of 68) and D (13 of 68). Other subtypes found included AC (one), AD (one), G/AG (three), H (one). One sample was untypeable. Of the 25 patients infected with subtype A virus, seven failed (28.0%), three of 23 (13.0%) patients with subtype C failed and three of 13 (23.1%) patients with subtype D virus failed (Table 1). There were no significant differences in outcome to PI or NNRTI regimes within or between subtypes (P > 0.05), as determined by Fisher's exact test.
Association between subtypes A, C and D and individual polymorphisms
Baseline pre-therapy samples were sequenced to produce an analysis of pol comprising all 99 codons of protease and codons 1 to 320 of RT. A full sequence was obtained in 68 of 69 African baseline samples, with one sequence producing a result in protease only. Sequencing revealed 133 different polymorphisms within the patient cohort in pol (96 in RT and 37 in protease), with a mean per patient of 9.0 in protease and 22.3 in RT. Certain polymorphisms predominated within specific subtypes. In RT (Fig. 1a), at least 80% of subtype A isolates contained polymorphisms at codons V35, Q102, K122, D123, C162, K173, D177, V179, Q207, R211, V245, R277 and T286. The majority of subtype C isolates were polymorphic at the same codons (except R277), and also at E36, T39, S48, T200, A272, E291 and V292. More than 80% of subtype D isolates were polymorphic at codons V35, Q102, K122, C162, D177, Q207, R211, V245, D250 and A272.
In protease (Fig. 1b), the majority of subtype A isolates were polymorphic at codons I13, E35, M36, R41, G57, H69 and L89. More than 80% of isolates were polymorphic at M36, R41, G57, H69 and L89 for subtype C isolates, and also at codons I15, L19 and I93. Subtype D viruses were 100% polymorphic at codons R41 and G57, and at least 50% of isolates at I13, M36, L63 and I64.
The impact of baseline polymorphisms on outcome to therapy
Patients were divided according to whether they had responded to NNRTI-based HAART (NNRTI/R), responded to PI-based HAART (PI/R), failed or not responded to NNRTI-based HAART (NNRTI/NR) or failed PI-based HAART (PI/NR). There were 29, 21, eight and 11 patients in these four groups, respectively. The distributions of each identified polymorphism within the four treatment-outcome groups were compared using Fisher's exact test, to determine whether any individual polymorphisms impacted on therapeutic response (Tables 2 and Table 3). The NNRTI responders and non-responders had a mean of 31.5 and 32.3 polymorphisms in pol, respectively. The PI responders and non-responders had a mean of 30.4 and 31.1 polymorphisms, respectively. Fourteen polymorphisms occurred at codons associated with resistance in subtype B strains (L10, K20, M36, L63, A71, V77, V82 and L90 in protease, and K70, A98, V106, M184, T215 and K219 in RT). No individual polymorphism was significantly (P > 0.05) associated with the 'non-responder' treatment group for any HAART regime. In particular, pre-therapy polymorphisms at codons associated with drug-resistance did not associate with the 'non-responder' patients.
The development of resistance mutations in African patients not responding to therapy
In order to investigate HIV-1 variants escaping therapeutic control, pol sequences at the time of failure were studied. Genotypic analyses were carried out at failure on 23 sequences from 18 individuals (eight NNRTI non-responders and 10 PI non-responders) who experienced viral rebound on HAART.
Within the NNRTI group, five of eight failing patients developed new resistance-associated mutations in RT (D67N, K70R, L74V, A98G, Y181C, M184V, Y188F, G190A, L210W, T215Y and K219Q), and five of eight developed changes in codons not known to be associated with resistance (K43N, K49R, I94L, T139K, P157S, A158T, I202V, E204K, D218E and V293I). Four out of eight NNRTI-treated patients lost polymorphisms on therapy, all at codons not associated with resistance (E40D, I195T, K223R, K281R, E291D, V317G). Two out of eight patients not receiving PIs gained (G16A, D60E) and four of eight lost (Q7H, I13V, I15V, V82I) polymorphisms in protease, two of which (D60E and V82I) have been associated with resistance.
Four out of ten patients who failed PI-based HAART developed resistance mutations in protease (L10I, K20R, L33I, M46L, G48V, I54V, L63P, V82A and N88S) and two of 10 developed changes not associated with resistance (I15V, A22V, I62V, I64V and N83D). Seven of ten PI-treated patients showed changes at 27 codons in RT, predominantly at codons not associated with resistance, but also at K70R, V75I, Q151M, M184V and K219Q.
Discussion
The data presented are encouraging for those arguing for the implementation of HAART in Africa. We demonstrate that patients infected with non-B strains of HIV-1 respond well to HAART (whether PI- or NNRTI-based), and that there is no virological reason for therapy not to succeed in these individuals.
Sixty (75.6%) individuals remained undetectable after 1 year of HAART. There were no differences in outcome associated with the country of origin, the subtype or the therapy group, although it was noted that the PI-receiving cohort had a greater rate of CD4 cell increase on therapy (P = 0.07). However, the average baseline CD4 count in these patients was lower than in the NNRTI-group (P = 0.01), and this may correlate with the response while on therapy.
One hundred and thirty-three polymorphisms were identified in pol (37 in protease and 96 in RT), with a mean of 9.0 in protease and 22.3 in RT per patient. Many of these codon changes were associated with specific sub-types. Analysis of sequences from clades A, C and D shows that many polymorphisms in both RT (Fig. 1a) and protease (Fig. 1b) were found in nearly all isolates of a specific sub-type. The functional significance of these polymorphic sites on both viral replicative fitness and drug resistance is currently unknown. However, the correlation of baseline polymorphisms with clinical outcome presented in this paper reveals no apparent selective advantage.
Pre-treatment HIV-1 pol sequences and viral subtypes were analysed with respect to clinical outcome on different regimes. Neither individual polymorphisms in RT or protease, nor subtypes (a marker of more general polymorphic variability) were associated with failure on any therapy. Surprisingly, this remained true when polymorphisms were present at baseline at codons associated with resistance in subtype B viruses, particularly in protease. These data agree with those previously reported in an un-subtyped cohort [18]. Despite the large number of polymorphisms identified at baseline in both RT and protease, there was no link between the number of codon changes and response to therapy, a factor that has previously been reported as being predictive of resistance [19,20]. Although no individual mutation was associated with therapeutic value, further analyses of these data to identify the impact of specific combinations of polymorphisms is currently being undertaken.
If a patient is poorly adherent to a HAART regime this would result in the application of weak drug pressure. Any rebound in viral load is, therefore, likely to contain a smaller proportion of drug-resistance-associated mutations relative to a patient who was fully adherent. Only nine of 18 patients in the 'non-B' cohort developed mutations at codons known to be associated with genotypic resistance at the time of therapy failure, possibly suggesting that a weak selective pressure was being applied in the other half of this cohort. These results may be interpreted as reflecting a problem with adherence or, alternatively, there may be a pharmacological explanation. Further studies to measure drug levels and assess adherence are necessary to explain fully these findings. However, if patient compliance accounts for these differences, clinics will have to reassess their methods of patient education and support. This has implications for those advocating therapeutic protocols in sub-Saharan countries, as well as on the management of HIV-positive African patients in the industrialized world. Virologically, there are excellent reasons why African patients should benefit from the significant advantages conferred by HAART.
Acknowledgements
We acknowledge the support of the Jefferiss Research Trust.
References
1. Palella FJJ, Delaney KM, Moorman AC. et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med 1998, 338: 853-860.
2. Pisani E, Schwartlander B, Cherney S. et al. Report on the Global HIV/AIDS Epidemic. Geneva: UNAIDS; 2000.
3. Gao F, Bailes E, Robertson DL. et al. Origin of HIV-1 in the chimpanzee Pan troglodytes troglodytes. Nature 1999, 397: 436-441.
4. Robertson DL, Anderson JP, Bradac JA. et al. HIV-1 nomenclature proposal. Science 2000, 288: 55-56.
5. Kaleebu P, Ross A, Morgan D. et al. Relationship between HIV-1 Env subtypes A and D and disease progression in a rural Ugandan cohort. AIDS 2001, 15: 293-299.
6. Alaeus A, Lidman K, Bjorkman A, Giesecke J, Albert J. Similar rate of disease progression among individuals infected with HIV-1 genetic subtypes A-D. AIDS 1999, 13: 901-907.
7. Caride E, Brindeiro R, Hertogs K. et al. Drug-resistant reverse transcriptase genotyping and phenotyping of B and non-B subtypes (F and A) of human immunodeficiency virus type I found in brazilian patients failing HAART. Virology 2000, 275: 107-115.
8. Shafer RW, Eisen JA, Merigan TC, Katzenstein DA. Sequence and drug susceptibility of subtype C reverse transcriptase from human immunodeficiency virus type 1 seroconverters in Zimbabwe. J Virol 1997, 71 (7): 5441-5448.
9. Palmer S, Alaeus A, Albert J, Cox S. Drug susceptibility of subtypes A,B,C,D, and E human immunodeficiency virus type 1 primary isolates. AIDS Res Hum Retroviruses 1998, 14: 157-162.
10. Apetrei C, Descamps D, Collin G. et al. Human immunodeficiency virus type 1 subtype F reverse transcriptase sequence and drug susceptibility. J Virol 1998, 72: 3534-3538.
11. Guay LA, Musoke P, Fleming T. et al. Intrapartum and neonatal single-dose nevirapine compared with zidovudine for prevention of mother-to-child transmission of HIV-1 in Kampala, Uganda: HIVNET 012 randomised trial. Lancet 1999, 354: 795-802.
12. Jackson JB, Becker-Pergola G, Guay LA. et al. Identification of the K103N resistance mutation in Ugandan women receiving nevirapine to prevent HIV-1 vertical transmission. AIDS 2000, 14: F111-F115.
13. Kroon ED, Ungsedhapand C, Ruxrungtham K. et al. A randomized, double-blind trial of half versus standard dose of zidovudine plus zalcitabine in Thai HIV-1-infected patients (study HIV-NAT 001). HIV Netherlands Australia Thailand Research Collaboration. AIDS 2000, 14: 1349-1356.
14. Hirsch MS, Brun-Vezinet F, D'Aquila RT. et al. Antiretroviral drug resistance testing in adult HIV-1 infection: recommendations of an International AIDS Society-USA Panel. JAMA 2000, 283: 2417-2426.
15. Cornelissen M, van den Burg R, Zorgdrager F, Lukashov V, Goudsmit J. pol gene diversity of five human immunodeficiency virus type 1 subtypes: evidence for naturally occurring mutations that contribute to drug resistance, limited recombination patterns, and common ancestry for subtypes B and D. J Virol 1997, 71: 6348-6358.
16. Holguin A, Rodes B, Soriano V. Protease gene analysis of HIV type 1 non-B subtypes in Spain. AIDS Res Hum Retroviruses 2000, 16: 1395-1403.
17. Lloyd RM, Lacroix J-M, Hough LM, et al. Polymorphic changes in HIV-1 pol as predictive molecular markers for resistance associated mutations. 38th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Diego, September 1998 [abstract].
18. Bossi P, Mouroux M, Yvon A. et al. Polymorphism of the human immunodeficiency virus type 1 (HIV-1) protease gene and response of HIV-1-infected patients to a protease inhibitor. J Clin Microbiol 1999, 37: 2910-2912.
19. Molla A, Korneyeva M, Gao Q. et al. Ordered accumulation of mutations in HIV protease confers resistance to ritonavir. Nat Med 1996, 2: 760-766.
20. Schmidt B, Walter H, Moschik B. et al. Simple algorithm derived from a geno-/phenotypic database to predict HIV-1 protease inhibitor resistance. AIDS 2000, 14: 1731-1738.
© 2001 Lippincott Williams & Wilkins, Inc.