Abstract
We analyzed the reverse transcriptase genotypes of human immunodeficiency virus type 1 subtype C viruses isolated from 23 patients in Botswana treated with didanosine-based regimens. The K65R mutation was selected either alone or together with the Q151M, S68G, or F116Y substitution in viruses from seven such individuals. The results of in vitro passage experiments were consistent with an apparent increased propensity of subtype C viruses to develop the K65R substitution.
Non-subtype-B human immunodeficiency virus (HIV) is prevalent throughout sub-Saharan Africa and in eastern Asia (12) and may now represent ≈15 to 20% of all new infections in Western countries (4, 17). Although current evidence suggests that all HIV subtypes probably respond in an equivalent fashion to antiretroviral drugs (ARVs), there is increasing evidence that distinct pathways toward the development of drug resistance may emerge among viruses of different subtypes. In part, this may be based on differences in codon usage and baseline sequences that can vary extensively among viral subtypes. This is an issue that is of importance to patients but which also has public health importance, given that drug-resistant variants of HIV can be sexually transmitted (16).
Subtype C is the most prevalent non-B subtype worldwide and is the most common subtype in Botswana (12). Previous work has shown that distinct patterns of codon usage can result in the appearance of specific mutations associated with resistance to nonnucleoside reverse transcriptase (RT) inhibitors. For example, a V106M mutation appears to be more prevalent among subtype C viruses than among subtype B viruses, for which a V106A substitution is more common at this position (3). Studies on mother-to-child transmission of HIV type 1 (HIV-1) reported that viruses from a higher proportion of women infected with HIV-1 subtype C or D developed resistance to nevirapine (NVP) than was found for women infected with subtype A viruses (7). In addition, an L89M/I substitution may be preferentially associated with resistance to protease inhibitors for subtype C viruses (1, 6).
Although the K65R substitution can cause extensive cross-resistance among currently used nucleoside reverse transcriptase inhibitors (NRTIs), this mutation has been observed relatively rarely among viruses from subtype B-infected individuals who received antiviral drugs (9, 10, 15, 18). For example, K65R was observed in viruses from a relatively small proportion of individuals treated with tenofovir (TDF) together with lamivudine (3TC) and efavirenz (EFV) over long periods (8). Interestingly though, d4T was shown to select for K65R in culture with subtype C viruses (9). Recent increases in the frequency of K65R in Western countries are attributed to extensive TDF use in recent years (11, 13). We recently demonstrated that subtype C clinical isolates, unlike those of subtype B, can select for the K65R substitution within 12 weeks when exposed to increasing concentrations of TDF in tissue culture (2). These observations led us to evaluate the incidence of K65R in Botswana patients who received ARV therapy in the context of both first- and second-line regimens provided by the National Antiretroviral Treatment Program.
In January 2002, Botswana officially initiated a national antiretroviral treatment program providing highly active antiretroviral therapy (HAART) within the public health system. The government of Botswana offers free ARVs to all citizens if they have a CD4 cell count of less than 200 cells/μl or an AIDS-defining condition. HIV-infected adults who qualify for treatment begin with the three-drug regimen of zidovudine (ZDV) plus 3TC plus EFV if male or ZDV plus 3TC plus NVP if female, due to potential teratogenicity associated with EFV. In cases of therapeutic failure, second-line regimens that include various combinations of didanosine (ddI), stavudine (d4T), and lopinavir/ritonavir, which has now replaced nelfinavir (NFV), and third-line regimens containing saquinavir/ritonavir are available.
Before January 2002 and as per national practice and international guideline recommendations for developing countries, most patients initiated therapy with ddI/d4T-based regimens as the nucleoside backbone. In this study, we evaluated the incidence of K65R in viruses from 23 Botswana patients who received ddI/d4T-based therapy in the context of both first- and second-line regimens provided by the National Antiretroviral Treatment Program. All of these patients experienced treatment failure, i.e., rising viral loads and/or diminishing CD4 counts, while on combination regimens that included ddI. Ten of these individuals had initiated treatment with ddI-based regimens (9 with ddI/d4T and 1 with ddI/ZDV) together with either EFV or NVP as nonnucleoside reverse transcriptase inhibitor, while 13 had started therapy with ZDV/3TC/NVP or ZDV/3TC/EFV prior to switching to ddI- and/or d4T-containing regimens. The HAART history was available for all of these patients (Table 1).
TABLE 1.
Treatment history of 23 Botswana patients treated with ddI/d4T-based regimens
| Patient no. | CD4 countb | VLb | Initial regimen (durationd) | Second and/or third regimen, where applicable (mo) | Last regimen | Duration of treatment with ddI/d4T (mo) | Presence of K65R at genotyping |
|---|---|---|---|---|---|---|---|
| 1 | 140 | 5.21 | ZDV/3TC/NVP (10) | ddI/d4T/NFV (15) | ZDV/3TC/NFV | 15 | − |
| 2 | 158 | 4.39 | ZDV/3TC/NVP (14) | ddI/d4T/NFV (9) | ddI/d4T/NFV | 9 | − |
| 3 | 78 | 4.49 | ddI/d4T/EFV (19) | ZDV/3TC/NFV (6) | ABC/3TC/NFV | 19 | + |
| 4 | 62 | 5.87 | ZDV/3TC/NVP (36) | ddI/d4T/NFV (5) | SQV/RTV/ABC | 5 | − |
| 5 | 214 | 5.87 | ZDV/3TC/NVP (ND) | ZDV/3TC/RTV/SAQ (9) | ddI/d4T/NFV | 15 | − |
| 6 | 1 | 5.87 | ZDV/3TC/EFV (2) | ddI/d4T/EFV (4) | ddI/d4T/EFV | 4 | + |
| 7a | 7 | 5.87 | ddI/d4T/NVP | ddI/d4T/NVP | ddI/d4T/NVP | 4 | + |
| 8 | NAc | NA | ddI/d4T/NVP (7) | ZDV/3TC/NFV (10) | ZDV/3TC/EFV | 7 | − |
| 9 | 199 | 5.70 | ZDV/3TC/EFV (12) | ddI/d4T/EFV (9) | ddI/d4T/NFV | 13 | − |
| ZDV/3TC/EFV (6) | |||||||
| 10a | 0 | 5.87 | ddI/d4T/NVP | ddI/d4T/NVP | ddI/d4T/NVP | 4 | + |
| 11 | 45 | 5.87 | ZDV/3TC/NVP (14) | ddI/d4T/NFV (23) | ddI/d4T/NFV | 23 | − |
| 12 | 344 | 5.87 | ddI/d4T/EFV (3) | ZDV/3TC/NFV | ZDV/3TC/NFV | 3 | − |
| 13 | 2 | 5.39 | ddI/d4T/EFV (15) | ZDV/3TC/NFV(6) | ZDV/3TC/NFV | 15 | + |
| 14 | 4 | 5.46 | ZDV/3TC/NVP (16) | ddI/d4T/NFV(6) | ddI/d4T/NFV | 6 | − |
| 15a | 4 | 5.47 | ddI/d4T/NVP | ddI/d4T/NVP | ddI/d4T/NVP | 15 | + |
| 16 | 20 | 5.79 | ddI/ZDV/NVP (11) | ZDV/3TC/NVP (6) | ZDV/3TC/NVP | 11 | + |
| 17 | 9 | 5.87 | ZDV/3TC/EFV (16) | ddI/d4T/NFV (9) | ddI/d4T/NFV | 9 | − |
| 18 | 0 | 5.87 | ddI/d4T/EFV (4) | ZDV/3TC/NFV (10) | ZDV/3TC/NFV | 4 | − |
| 19 | NA | 5.76 | ddI/d4T/NVP | ddI/d4T/NVP (6) | ZDV/3TC/NVP | 6 | − |
| 20 | 8 | 5.79 | ZDV/3TC/NVP (18 days) | ddI/d4T/NVP (6), ddI/3TC/NFV (9) | ABC/ddI/LPVr | 19 | − |
| 21 | 126 | 5.82 | ZDV/3TC/NVP (22) | ddI/d4T/NFV (8) | ddI/d4T/NFV | 8 | − |
| 22 | 2 | 4.66 | ZDV/3TC/NVP (15) | ddI/d4T/NFV (16) | ddI/d4T/NFV | 16 | − |
| 23 | 68 | 5.87 | ZDV/3TC/NVP (10) | d4T/3TC/NVP (2) | ddI/ZDV/NFV | 14 | − |
This patient remained on ddI/d4T/NVP throughout the follow-up for 4 months (patients 7 and 10) or 15 months (patient 15).
Viral load (VL) (log) and CD4 counts (no. of cells/μl) evaluated at baseline (before HAART experience). A value of >750,000 HIV-1 RNA copies/ml was considered to be 5.87 logs.
NA, information not available.
Duration of regimen is given in months except where otherwise indicated.
Sequencing of extracted DNA, performed by Bayer TruGene or by Associated Biomedical Systems technology to determine genotypic changes in RT associated with drug resistance, phylogenetic analysis, and confirmation of subtype C infection were performed as previously described (5, 6). Selection of resistance to ddI and d4T (each) using subtype C as well as subtype B clinical isolates was studied with culture, beginning with concentrations of ddI and d4T of 5 and 0.1 μM, respectively, and in combination with 1 μM ddI-0.1 μM d4T. RT assays were performed weekly to assess viral replication. Cells and virus supernatants were harvested for passage, and drug concentrations were increased at each step. Genotyping was performed at select passages to evaluate the time to development of drug resistance.
Frequency of K65R among viruses from ddI-treated patients.
The results in Table 2 show that viruses from 7 of the 23 patients who failed ddI-based regimens possessed the K65R substitution after a median exposure to combination ddI/d4T therapy of only 8 months (range, 4 to 18 months). For four of these patients, viruses developed K65R while the patient was still on ddI/d4T at the time of genotyping. For three of the seven patients, viruses developed only K65R, while for four others, viruses also developed Q151M, F116Y, and S68G. The association of K65R and M184V was seen for two patients who had experienced both ddI/d4T- and ZDV/3TC-based regimens. In contrast, for 9 of 13 patients who received 3TC/ZDV as initial therapy, viruses mostly developed thymidine-associated mutations, e.g., M41L, D67N, K70R, T215Y/F, and K219E/Q, while for 3 of 13, viruses had no NRTI resistance (not shown). The presence of thymidine-associated mutations in viruses from patients who first experienced ZDV/3TC may explain the nonemergence of K65R while the patient was subsequently receiving ddI/d4T, due to presumed antagonism among these mutations (13).
TABLE 2.
Association of K65R with other RT mutations in viruses from patients infected with HIV-1 subtype C and treated with ddI/d4T
| Patient no. | NRTI mutation(s) | Time of first appearance of K65R (after commencement of ddI/d4T-based therapy) (mo) | Polymorphisms |
|---|---|---|---|
| 3 | K65R, S68G, F116Y, Q151M, M184V | 18 | V35T, E36A, T39E, S48T, T69I, D123S, S163C, K166Q, K173T, Q174R, T200A, I202V, E203K, Q207A, H208Y, R211E, L214F, V245M, D250E, R277K, E291D, V292I, I293V, D324E |
| 6 | K65R, S68G, K70R, V75I, F116Y, Q151M | 4 | V35Q, T39D, S48T, V60I, T69I, I135T, E138A, I142T, Q151L, K173A, Q174K, D177E, I178V/L, G196E, T200A, Q207E, R211K, V245K, D250E, R277K, E291D, V292I, I293V, V317A, Q334N |
| 7 | K65R | 4 | V35T, T39K, S48E, T69N, E122K, D123E, I135M, K173T/I, T200A, Q207E, R211K, L214F, V245Q, P272A, R277K, K281R, T286A, E291D, V292I, I293V, E312F, V314P |
| 10 | K65R | 4 | T39E, E40K, D67G, T69N, E122K, K173A, G177G/R, T200A, Q207E, R211K, L214F, H221H, L228L/R, V245E, D250E, E291D, V292I, I293V, I326V, Q334H |
| 13 | K65R, S68G, K70R, F77L, Y115F, F116Y, Q151M, K219E | 6 | V35T, E36A, T39E, T58N, V75A, D123S, R172K, K173T, D177E, I178M, T200A, Q207K, R211K, L214F, P225H, V245Q, E248D, E291D, V292I, I326V, Q334D |
| 15 | K65R | 12 | V35T, T39E, S48T, D123G, K173E, D177G, T200A, Q207K, R211K, L214F, V245Q, E248D, P272S, T286A, L289I, E291D, V292I, Q334N |
| 16 | K65R, M184V | 11 | T27S, V35T, T39E, S48T, D123G, T139K, S162C, K173A, Q174K, D177E, I178L, T200A, Q207D, R211K, L214F, L228R, V245Q, P272S, E291D, V292I, I293V, Q334N |
In vitro preferential selection of K65R in HIV-1 subtype C viruses.
The development of K65R resistance to ddI and/or d4T was confirmed by tissue culture selection using HIV-1 subtype C clinical isolates (Table 3). The K65R and L74V mutations each arose within 15 to 28 weeks in two of five subtype C selections under ddI pressure alone and in three of five selections conducted using combinations of ddI and d4T. In some cases, ddI and/or d4T also selected for the appearance of the D67N and/or V75I substitution in RT. Interestingly, and as previously shown (2), none of the subtype B variants developed the K65R mutation during serial passage (week 28).
TABLE 3.
Selection of RT mutations after sequential passage with ddI and/or d4T
| Clinical isolate (subtype)a | Expt no. | Wk of passage | Highest drug concn (μM) achieved | Mutation(s) selected |
|---|---|---|---|---|
| 4742 (C) | 1 | 28 | 40, ddI | K65R |
| 2 | 24 | 5, d4T | None | |
| 3 | 17 | 10, ddI; 0.5, d4T | None | |
| 4761 (C) | 1 | 24 | 10, ddI | L74V |
| 2 | 22 | 2.5, d4T | D67N | |
| 3 | 15 | 1, ddI; 0.1, d4T | K65R | |
| BG-05 (C) | 1 | 28 | 30, ddI | K65R, D67N |
| 2 | 24 | 2.5, d4T | None | |
| 3 | 28 | 10, ddI; 0.5, d4T | K65R | |
| BG-15 (C) | 1 | 17 | 10, ddI | None |
| 2 | 17 | 0.5, d4T | None | |
| 3 | 28 | 10, ddI; 0.5, d4T | K65R, V75I | |
| Mole 18 (C) | 1 | 28 | 40, ddI | L74V |
| 2 | 22 | 0.5, d4T | None | |
| 3 | 22 | 2.5, ddI; 0.1, d4T | None | |
| 5323 (B) | 1 | 28 | 20, ddI | None |
| 2 | 28 | 2.5, ddI; 0.1, d4T | None | |
| 5512 (B) | 1 | 28 | 20, ddI | None |
| 2 | 28 | 5, ddI; 0.5, d4T | None |
The propensity of subtype C viruses to rapidly select K65R may be attributable, in part at least, to differences among coding sequences between subtype B and subtype C viruses. Nucleoside sequences at codons 64 to 66 in the RT of subtype B viruses are commonly AAG (64), AAA (65), and AAA (66), while in subtype C viruses the sequences are AAA (64), AAG (65), and AAG (66). Although single point mutations are likely always involved in selection of K65R in both subtypes B and C, it is possible that polymorphisms that underlie differences between these subtypes may create an environment in which the selection of K65R occurs more easily in viruses of subtype C.
In summary, the K65R substitution may emerge at a higher frequency for individuals infected with subtype C viruses who experienced treatment with ddI/d4T. In view of widespread ARV access in sub-Saharan countries, these findings establish a degree of concern in regard to the possibility that certain mutations, such as K65R, may emerge more rapidly in viruses of subtype C.
To be sure, these results do not diminish the importance of access to drugs, such as TDF, that are well tolerated and effective in management of HIV disease. Furthermore, most of the patients followed in this study never fully suppressed viral load due to poor adherence to the regimen, a situation that has now changed due to improved patient management. Conceivably, those patients who initiated therapy with ddI/d4T and whose viruses did not develop K65R may have been more adherent than those whose viruses did develop this substitution, but the reasons for these differences are unclear. ddI/d4T combinations have been shown to be inferior to other regimens in controlled clinical studies (14). Although use of ddI/d4T may be cost-effective in the short term, its continuing use may lead to inferior clinical outcomes and an increased overall cost of patient care.
The first-line regimen currently recommended by the Botswana guidelines on antiretroviral therapy is ZDV/3TC/EFV or NVP, which may explain why the incidence of the K65R mutation is not elevated in viruses from currently treated individuals. We publish these findings to highlight the importance of monitoring drug resistance on as widespread a basis as possible in developing countries so as to better understand the dynamics of selection of specific mutations by viruses of different subtypes and in the interests of public health.
Acknowledgments
We thank the Botswana Ministry of Health for encouragement (study approved by the National Health Research Development Committee of the Botswana Ministry of Health, no. HRU-13/18/1, vol. VI [37]), Mpho Zwinila for database management, and the Service of Infectious Diseases at the IDCC, Princess Marina Hospital, Gaborone, Botswana, for their contribution to this work.
Footnotes
Published ahead of print on 2 October 2006.
REFERENCES
- 1.Abecasis, A. B., K. Deforche, J. Snoeck, L. T. Bacheler, P. McKenna, A. P. Carvalho, P. Gomes, R. J. Camacho, and A. M. Vandamme. 2005. Protease mutation M89I/V is linked to therapy failure in patients infected with the HIV-1 non-B subtypes C, F or G. AIDS 19:1799-1806. [DOI] [PubMed] [Google Scholar]
- 2.Brenner, B., M. Oliveira, F. Doualla-Bell, D. Moisi, M. Ntemgwa, F. Frankel, M. Essex, and M. A. Wainberg. 2006. HIV-1 subtype C viruses rapidly develop K65R resistance to tenofovir in cell culture. AIDS 20:F9-F13. [DOI] [PubMed] [Google Scholar]
- 3.Brenner, B., D. Turner, M. Oliveira, D. Moisi, M. Detorio, M. Carobene, R. G. Marlink, J. Schapiro, M. Roger, and M. A. Wainberg. 2003. A V106M mutation in HIV-1 clade C viruses exposed to efavirenz confers cross-resistance to non-nucleoside reverse transcriptase inhibitors. AIDS 17:F1-F5. [DOI] [PubMed] [Google Scholar]
- 4.Chaix, M. L., D. Descamps, M. Harzic, V. Schneider, C. Deveau, C. Tamalet, I. Pellegrin, J. Izopet, A. Ruffault, B. Masquelier, L. Meyer, C. Rouzioux, F. Brun-Vezinet, and D. Costagliola. 2003. Stable prevalence of genotypic drug resistance mutations but increase in non-B virus among patients with primary HIV-1 infection in France. AIDS 17:2635-2643. [DOI] [PubMed] [Google Scholar]
- 5.Doualla-Bell, F., S. Gaseitsiwe, T. Ndungu, M. Modukanele, T. Peter, V. A. Novitsky, N. Ndwapi, G. Tendani, A. Avalos, W. Wester, H. Bussmann, P. Cardiello, R. Marlink, H. Moffat, I. Thior, M. A. Wainberg, and M. Essex. 2004. Mutations and polymorphisms associated with antiretroviral drugs in HIV-1C-infected African patients. Antivir. Chem. Chemother. 15:189-200. [DOI] [PubMed] [Google Scholar]
- 6.Doualla-Bell, F., A. Avalos, T. Gaolathe, M. Mine, S. Gaseitsiwe, N. Ndwapi, V. A. Novitsky, B. Brenner, M. Oliveira, D. Moisi, H. Moffat, I. Thior, M. Essex, and M. A. Wainberg. 2006. Impact of human immunodeficiency virus type 1 subtype C on drug resistance mutations in patients from Botswana failing a nelfinavir-containing regimen. Antimicrob. Agents Chemother. 50:2210-2213. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Eshleman, S. H., G. Becker-Pergola, M. Deseyve, L. A. Guay, M. Mracna, T. Fleming, S. Cunningham, P. Musoke, F. Mmiro, and J. B. Jackson. 2001. Impact of human immunodeficiency virus type 1 (HIV-1) subtype on women receiving single-dose nevirapine prophylaxis to prevent HIV-1 vertical transmission (HIV Network for Prevention Trials 012 study). J. Infect. Dis. 184:914-917. [DOI] [PubMed] [Google Scholar]
- 8.Gallant, J. E., S. Staszewski, A. L. Pozniak, E. DeJesus, J. M. Suleiman, M. D. Miller, D. F. Coakley, B. Lu, J. J. Toole, and A. K. Cheng. 2004. Efficacy and safety of tenofovir DF vs stavudine in combination therapy in antiretroviral-naive patients: a 3-year randomized trial. JAMA 292:191-201. [DOI] [PubMed] [Google Scholar]
- 9.Garcia-Lerma, J. G., H. MacInnes, D. Bennett, P. Reid, S. Nidtha, H. Weinstock, J. E. Kaplan, and W. Heneine. 2003. A novel genetic pathway of human immunodeficiency virus type 1 resistance to stavudine mediated by the K65R mutation. J. Virol. 77:5685-5693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Gu, Z., H. Salomon, J. M. Cherrington, A. S. Mulato, M. S. Chen, R. Yarchoan, A. Foli, K. M. Sogocio, and M. A. Wainberg. 1995. K65R mutation of human immunodeficiency virus type 1 reverse transcriptase encodes cross-resistance to 9-(2-phosphonylmethoxyethyl)adenine. Antimicrob. Agents Chemother. 39:1888-1891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Jemsek, J., P. Hutcherson, and E. Harper. 2004. Poor virologic responses and early emergence of resistance in treatment naive, HIV-infected patients receiving a once daily triple nucleoside regimen of didanosine, lamivudine, and tenofovir DF. 11th Conference on Retroviruses and Opportunistic Infections, San Francisco, Calif.
- 12.Osmanov, S., C. Pattou, N. Walker, B. Schwardlander, and J. Esparza. 2002. Estimated global distribution and regional spread of HIV-1 genetic subtypes in the year 2000. J. Acquir. Immune Defic. Syndr. 29:184-190. [DOI] [PubMed] [Google Scholar]
- 13.Parikh, U., D. Koontz, and N. Sluis-Cremer. 2004. K65R: a multinucleoside resistance mutation of increasing prevalence exhibits bi-directional phenotypic antagonism with TAM. 11th Conference on Retroviruses and Opportunistic Infections, San Francisco, Calif.
- 14.Robbins, G. K., V. De Gruttola, R. W. Shafer, L. M. Smeaton, S. W. Snyder, C. Pettinelli, M. P. Dube, M. A. Fischl, R. B. Pollard, R. Delapenha, L. Gedeon, C. van der Horst, R. L. Murphy, M. I. Becker, R. T. D'Aquila, S. Vella, T. C. Merigan, M. S. Hirsch, and the AIDS Clinical Trials Group 384 Team. 2003. Comparison of sequential three-drug regimens as initial therapy for HIV-1 infection. N. Engl. J. Med. 349:2293-2303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Roge, B. T., T. L. Katzenstein, N. Obel, H. Nielsen, O. Kirk, C. Pedersen, L. Mathiesen, J. Lundgren, and J. Gerstoft. 2003. K65R with and without S68: a new resistance profile in vivo detected in most patients failing abacavir, didanosine and stavudine. Antivir. Ther. 8:173-182. [PubMed] [Google Scholar]
- 16.Salomon, H., M. A. Wainberg, B. Brenner, Y. Quan, D. Rouleau, P. Cote, R. LeBlanc, E. Lefebvre, B. Spira, C. Tsoukas, R. P. Sekaly, B. Conway, D. Mayers, J. P. Routy, et al. 2000. Prevalence of HIV-1 resistant to antiretroviral drugs in 81 individuals newly infected by sexual contact or injecting drug use. AIDS 14:F17-F23. [DOI] [PubMed] [Google Scholar]
- 17.Snoeck, J., K. Van Laethem, P. Hermans, E. Van Wijngaerden, I. Derdelinckx, Y. Schrooten, D. A. van de Vijver, S. De Wit, N. Clumeck, and A. M. Vandamme. 2004. Rising prevalence of HIV-1 non-B subtypes in Belgium: 1983-2001. J. Acquir. Immune Defic. Syndr. 35:279-285. [DOI] [PubMed] [Google Scholar]
- 18.Stone, C., M. Ait-Khaled, C. Craig, P. Griffin, and M. Tisdale. 2004. Human immunodeficiency virus type 1 reverse transcriptase mutation selection during in vitro exposure to tenofovir alone or combined with abacavir or lamivudine. Antimicrob. Agents Chemother. 48:1413-1415. [DOI] [PMC free article] [PubMed] [Google Scholar]