As the current regimens cannot eradicate HIV infection, persons living with HIV are doomed to assume chronic therapies with often very demanding daily schedules. In the long run, this may lead to reduced adherence to therapy, to incomplete suppression of viral replication and to the emergence of resistant viruses that can negate the patients’ efforts [1–3]. Furthermore, with continued exposure to antiretroviral drugs, patients have begun to experience new adverse effects including body fat redistribution, dislipidaemia, diabetes, insulin resistance and osteopenia [4,5], and the fear of unexpected complications such as cardiovascular diseases has been raised . A variety of strategies has emerged to try to help the long-term management of highly active antiretroviral therapy (HAART). Researchers have focused their attention on the development of simpler regimens [7–9]. An alternative approach involves various types of structured treatment interruption (STI) . Several different mechanisms have been claimed for STI. It has been postulated that STI could enhance specific immune responses and allow control of viral replication, in the absence of continuous HAART, in seroconverters . In those with advanced treatment failure, it has been suggested that STI could favour the overgrowth of wild-type virus in the peripheral circulation [12,13] and thus facilitate the ‘disappearance’ of resistance mutations [14,15]. In chronically infected patients without virologic failure, STI has been explored as a way to boost HIV-specific immune responses [16,17] and, more recently, to reduce drug exposure, promote adherence and minimize drug-related morbidity [18,19].
Various strategies have been explored, including fixed intervals for on and off periods and specific thresholds, either immunological or virological, for reinitiation of therapy.
To date, there are a number of potential risks for STI or even pulse therapy that have not been assessed: the risk of exposing patients to differential drug levels, which may enhance selection of resistant mutants, the effects of re-seeding of viral reservoirs, the real extent of the assumption that less drug is associated with less toxicity or better adherence, and the overall outcome compared with continuous HAART.
Finally the use of STI and pulse therapy may be strictly linked with questions raised by new treatment recommendations : what to do with individuals who began therapy at a CD4 cell level above the currently recommended threshold, and how to manage those patients who, under HAART, have gained a CD4 cell count far above this threshold.
The controlled, prospective and randomized trial described here involves a cohort of chronically HIV infected individuals taking effective HAART and attempts to address most of these open questions. The study used an individualized pulse therapy strategy, driven by CD4 cell count, and this was compared with conventional continuous HAART.
Study design and patients
The BASTA study is a randomized, open-label, controlled trial that compares two different treatment strategies. Results at 64 weeks of follow-up are reported.
The patients, recruited from one single centre in Northern Italy, had to be older than 18 years, with laboratory documentation of HIV-1 infection, a confirmed (at least two consecutive tests) CD4 cell count > 800 × 106 cells/l, a confirmed plasma HIV-1 RNA level < 50 copies/ml and a stable HAART (at least three drugs). Patients were excluded if they had ever received any immunomodulatory drugs and if they were pregnant or breast-feeding women. Eligible patients were randomly assigned in a 1:2 ratio to continue their ongoing treatment or to stop it. Patients randomized in the STI arm were kept off therapy until their CD4 cell count dropped ≤ 400 × 106 cells/l, irrespective of their viral load. Once their CD4 cell count declined to this limit, they resumed their previous therapy until it had increased to ≥ 800 × 106 cells/l again. Another interruption of therapy was proposed at this time-point. STI were, therefore, individualized with the aim to keep CD4 cell count > 400 × 106 cells/l with the shortest possible exposure to drugs.
The modification of HAART treatment was permitted at any time after randomization in the event of drug-associated toxicity or clinical progression of HIV disease, according to the US Centers for Disease Control and Prevention (CDC) 1993 revised classification system for HIV infection category C clinical conditions.
All patients provided their informed consent and the trial was approved by the Institutional Review Board of our hospital.
Monitoring and laboratory testing
Consecutive HIV-infected patients referring to our outpatient clinic who met the enrolment criteria entered the study. Patients were followed up every month for the first 2 months and every other month afterwards, with clinical assessment and routine laboratory monitoring. CD4 cell counts and HIV RNA plasma levels were determined at baseline, at weeks 4 and 8, and thereafter every 8 weeks. Unscheduled visits and laboratory tests were performed according to the clinical needs (i.e., evidence of adverse events or viral rebound).
The number of CD4 lymphocytes was measured by flow cytometry. Plasma HIV RNA was measured through a quantitative essay (HIV RNA 3.0 bDNA, Chiron, Emeryville, California, USA) with a lower limit of quantification of 50 copies/ml. All other variables were measured by means of standard laboratory tests.
During the follow-up, HIV-related diseases and adverse events were looked for systematically at all scheduled and unscheduled visits. AIDS was diagnosed according to category C of the CDC classification system for HIV infection.
Adverse events were graded in severity from 1 to 4 according to the World Health Organization (WHO) grading system. Plasma lipid profile was evaluated according to the American Heart Association (AHA) guidelines. All blood tests were performed in fasting state. A grade 4 (WHO) or 3 (AHA) toxicity were reasons for treatment discontinuation.
The primary outcome measure was the proportion of subjects maintaining at each time-point a CD4 cell count > 400 × 106 cells/l.
Secondary end-points of the study were to verify the possibility of steadily discontinuing antiretroviral therapy in patients that started it with baseline immunological values higher than those currently recommended by international guidelines for HIV treatment ; to identify predictive variables of the possibility of safely discontinuing antiretroviral therapy; to verify the dynamic of CD4 cell loss and HIV replication after treatment interruption; and to study the occurrence of adverse events
Primary analysis was performed on an intention-to-treat basis that included all available data, irrespective of discontinuation of the study treatment or withdrawal from the trial. In addition, a secondary per-protocol analysis was performed.
The primary outcome measure (the proportion of patients with CD4 cell counts > 400 × 106 cells/l) was analysed by means of chi-square or Fisher's exact test, which were also used to analyse all other categorical variables. Time to treatment failure was estimated using Kaplan–Meier product-limit estimates (presented graphically). The log-rank test was used to assess the difference between the survival curves for each treatment, while ANOVA test and Student's t-test were used for continuous variables unless the variables were not normally distributed, in which case the Kruskal–Wallis and Mann–Whitney U tests were used. Logistic regression analyses (forward stepwise model) were used to evaluate the relationship between variables and outcome. All analyses were performed taking into account multiplicity effects. Comparisons between pairs of groups were performed only in the case of a multiple test yielding a statistically significant result. All test were two-sided and a P value < 0.05 was regarded as significant.
The trial was powered to establish equivalence between the two arms. Equivalence was defined on the basis of the 95% confidence interval. To assess equivalence the 95% confidence interval for the primary variable should lie between ±15% at each time point (4, 8, 16, 24, 32, 40, 48, 56 and 64 weeks). Taking into account these limits and with the assumption of a failure rate < 5% in the control arm, approximately 70 patients were required.
All analyses were performed with the SPSS statistical software package for Windows, version 10.0 (SPSS, Chicago, Illinois, USA).
Sixty-nine patients were randomized, 46 in the STI arm and 23 in the control group. All of them were included in the intention-to-treat and safety analyses. The baseline characteristics of the study patients were well balanced among the two treatment groups (Table 1). In the STI group, all patients but one received a triple drug therapy, the remaining patient was on a four drug regimen (Table 2). None of the patients in the control group showed a decrement of CD4 cell counts below the limit of 400 × 106 cells/l over the 64-week follow-up. Conversely, in the STI group, the number of patients reaching this end-point was one after 4 weeks, one at 8 weeks, two at 16 weeks, one at 24 weeks, three at 32 weeks, three at 40 weeks and one at 48 weeks. All patients regained a CD4 cell count > 400 × 106 cells/l in the following 8 weeks. Consequently, at each time point, the maximum proportion of patients with the primary endpoint of the study ranged from 0% (at 56 and 64 weeks) to 6.6%. The largest 95% confidence interval for the difference between groups was, therefore, observed after 32 and 40 weeks (−0.5 to 9.3). The two treatment strategies, therefore, had an equivalent result. To obtain this result, therapy was restarted in 13 (28%) patients in the STI group. All but one patient (who started therapy twice) resumed it just once. A prompt immune recovery was observed in all cases, with a mean CD4 cell count of 686 × 106 cells/l (SD, 142) after 4 weeks and of 720 × 106 cells/l (SD 85; lowest measure 601 × 106 cells/l) after 8 weeks of restart. In 72% of these patients, HIV RNA returned below the detection limit (50 copies/ml) within 16 weeks and in the remaining patients within 24 weeks.
There were, however, marked differences in the way patients in the STI group responded to treatment interruption. The predictive value of several variables on the subsequent possibility to stay off therapy was analysed. Considered variables included sex, risk factor for HIV infection, CDC classification stage, age at randomization, number of previous treatment regimens, type of drugs (class) included in last HAART, time on antiretroviral treatment, nadir CD4 cell count value, CD4 cell count at the start of first antiretroviral therapy, CD4 cell count at randomization (start of STI), HIV RNA level before first HAART, and time with HIV RNA below the detection limit (50 copies/ml). When entered in a logistic regression analysis, the only parameter significantly associated with the possibility to stay off therapy and, therefore, to prolong STI was the CD4 cell count nadir (B − 1.749; P = 0.001).
A post-hoc analysis was, therefore, performed after grouping patients in the STI group according to their CD4 cell count nadir. Four categories of patients were selected: group A, patients with a nadir CD4 cell count < 200 × 106 cells/l, considered as highly compromised patients (eight patients); group B, patients with a nadir CD4 cell count 200–350 × 106 cells/l, considered to be a proposed (but not yet completely accepted) immunological status indicating the need of antiretroviral therapy (seven patients); group C, patients with a nadir CD4 cell count 350–500 × 106 cells/l, considered as suitable for a more conservative approach to antiviral therapy (18 patients); and group D, patients with a nadir CD4 cell count > 500 × 106 cells/l, considered as patients in whom antiretroviral therapy was formerly, but no longer, indicated (13 patients).
The follow-up for this analysis was limited to 12 months as only the first STI period was considered and because all patients in one group interrupted the off-therapy period after 10 months. The duration of first STI was significantly different (P < 0.0001, log-rank test) among these groups. Differences are depicted in Fig. 1. The mean time off-therapy (time to reach the immunological end-point) for patients in group A was 5.38 months and for no patient was it longer than 40 weeks (P = 0.0027 versus group B and P < 0.001 versus group C and D). In group B, the mean time off-therapy was 11.7 months with 57% of patients reaching 16 months (P = 0.02 versus group C and D). In groups C and D, the mean time off-therapy was of 15.1 and 16 months, respectively, and 88.9% and 100% of patients in either group stayed off therapy for the whole follow-up period (P not significant). As in these two groups (C and D) the endpoint of the analysis (≤ 400 × 106 cells/l) was overlapping or below their nadir CD4 cell count, it was also evaluated if, at the end of the first 12 months period off therapy, patients’ CD4 cell count was superior to the count at the moment they started their first antiviral therapy. In group C the mean 12-month CD4 cell count was 727 × 106 cells/l compared with 465 × 106 cells/l at the beginning of therapy; in group D the same figures were 831 and 663 × 106 cells/l, respectively. In both cases, the difference was statistically significant (P < 0.002, paired t-test).
The dynamic of CD4 cell loss was different among groups (P < 0.01, Anova) too. In group A, there was a rapid decline of CD4 cell counts within the first 4 weeks, with a mean CD4 cell loss of 366 × 106 cells/l. In contrast, the mean CD4 cell loss in the same period of time was 144, 179 and 137 × 106 cells/l in groups B, C and D, respectively (Fig. 2). Consequently, mean CD4 cell counts were significantly different in group A when compared with those in group D (P < 0.01 from week 4 on) or in group C (P < 0.01 from week 8 on). Marginal differences were also observed between group A and B (P < 0.05 from week 24 on). The mean CD4 cell counts did not significantly differ at any time between groups B and C, while group D outperformed these two last groups (P < 0.01 from week 8 on). The mean CD4 cell loss at week 48 off-therapy was 470 × 106 cells/l (9.8 × 106 cells/l per week) in group B, 395 × 106 cells/l (8.2 × 106 cells/l per week) in group C and 386 × 106 cells/l (8.0 × 106 cells/l per week) in group D (Fig. 2).
The immunological results did not parallel the virological findings. HIV RNA rebound was, indeed, more pronounced in patients with lower CD4 nadir, but differences were merely statistically significant (P < 0.05) between group A and D from the 24th week on (Fig. 3).
According to the intention-to-treat analysis, no differences were observed when the variation of metabolic parameters, namely fasting total cholesterol and triglycerides, were compared at different time points between the STI and the control group (independent sample t-test). However, when variations were compared within a single group, triglycerides levels after 16 weeks in the STI group were significantly lower (P < 0.001, paired t-test) than baseline levels. Similarly cholesterol levels after 16, 32 and 48 weeks were significantly lower (P < 0.006, paired t-test) (Fig. 4). In the control group, on the contrary, no differences reached statistical significance.
Treatment interruption was well tolerated. Two patients presented adverse events during treatment interruption. A 49-year-old man had a symptomatic piastrinopenia (platelets 1 × 109 cells/l) after 4 months of STI that resolved only 4 months after the treatment resumption. This patient had also had piastrinopenia before starting his first antiviral therapy more than 5 years previously. The second patient was a young woman who had been taking steady antiretroviral therapy for longer than 4 years and had had a viral load below the detection limit for almost 2 years; she developed a generalized lymphoadenopathy with intermittent fever. Symptoms appeared after a few weeks from treatment interruption and therapy had to be restarted after 16 weeks because the patient became pregnant. After delivery, treatment was interrupted again without any inconvenient.
In the control group only one patient had to stop her therapy because of the persistence of dizziness caused by the assumption of efavirenz. Another five patients asked to stop treatment. One patient was lost to follow-up.
Current antiretroviral drugs cannot eradicate HIV infection and persons living with HIV are often faced with very demanding, lifelong therapies. Frequent dosing, high pill number, food restrictions, risk of medication unmasking HIV status through an alteration of body shape, and long-term toxicities are common drawbacks of most antiretroviral regimens. These factors have been widely recognized as causes of poor adherence [21,22], which may result in the emergence of drug-resistant viruses, a frequent pathway for treatment failure. Reduced exposure to antiretroviral drugs could limit these pitfalls. We explored this possibility on a cohort of individuals with substantial immune recovery following HAART regimen. Our approach was very conservative in order to limit the risk of disease progression in our patients. Only patients with a very marked immune response to HAART (twice the lower normal limit for CD4 cell counts) were enrolled and the immunological threshold to resume therapy was set to the lower normal limit of CD4 cell counts for HIV-uninfected adults. Under these assumptions, patients enrolled in the STI group performed as well as similar patients on continuous HAART, who formed the control group. None of the patients, neither in the STI group nor in the control one, showed a disease progression or any AIDS-defining event. At each time point, the proportion of subjects in the STI group that had a CD4 cell count decline below the defined threshold for resuming HAART was modest and not statistically different from the control group. In all cases, the 95% confidence interval for this difference was smaller than ±10%. Furthermore, in all patients who resumed therapy, there was a marked ability to re-expand CD4 T cells in a short period of time to levels comparable to those obtained before STI. Overall, pulse therapy seems an alternative strategic option for a wide variety of chronically infected individuals responding to HAART. However, according to our results, it seems also possible to predict which patients would benefit most from such a strategy. In other studies on uncontrolled cohorts [23,24], the only predictive factor for CD4 cell depletion after treatment interruption was the rate of CD4 cell increment under HAART. On the contrary, we observed a high predictive power of the nadir CD4 cell count. A possible explanation for this difference could be related to the selection criteria, such as the inclusion, in our study, of patients with lower nadir counts, or of patients with higher levels of immune reconstitution while on HAART. While differences from those of other studies can be explained, it is unclear why patients with lower nadir CD4 cell counts had to resume therapy in a significantly higher proportion and within a significantly shorter period of time than patients with higher CD4 nadir, despite the fact that they had similar pre-STI CD4 cell levels. CD4 cell depletion was not significantly related to the magnitude of viral load rebound, nor to the pre-STI CD4 cell count or to the time on an effective HAART. Therefore, this finding suggests that, independently from the extent of immune reconstitution under HAART, once a low CD4 cell level is reached, the immune system response to HIV replication is impaired. The immunological determinants of this situation are beyond the scope of the present study; however, some clinical implications deserve, in our opinion, further discussion. Currently implemented guidelines for antiretroviral therapy, in the attempt to balance the risk of toxicity against the benefits of preserving or improving immune function, recommend a more conservative approach and consider starting HAART when the CD4 cell count is 200–350 × 106 cells/l . Studies considering hard end-points, such as disease progression to AIDS, support this approach . Our data confirm the high risk of rapid deterioration of the immunological status once therapy is withheld in patients with nadir CD4 cell counts < 200 × 106 cells/l, but also identify a ‘gray zone’ containing those patients whose nadir CD4 cell count was 200–350 × 106 cells/l. These patients, after treatment interruption, maintained an optimal immunological status for a significantly longer period than those whose nadir CD4 cell count was < 200 × 106 cells/l, yet 57% showed a CD4 cell count fall to < 400 × 106 cells/l over a 64-week period. This proportion was statistically different (P = 0.02) from the nearly 90% of patients with a nadir CD4 cell count 350–500 × 106 cells/l that maintained, over the same period, a CD4 cell count > 400 × 106 cells/l. With this respect, starting HAART when CD4 cell count falls to just above 350 × 106 cells/l seems to confer some immunological advantages and, consequently, such a consideration could be a part of the decision-making process for starting antiretroviral therapy in therapy-naive patients.
On the opposite extreme, we confirm previously reported results [18,24]. Treatment interruptions in patients who started therapy with very high CD4 cell counts (i.e. > 500 × 106 cells/l) are a safe practice that allows a dramatic reduction of drug intake without exposing patients to the risk of clinically significant immune deterioration. On the contrary, over the 64-week follow-up, we observed a maintenance of CD4 cells at a higher level than these patients had at the time of their first HIV therapy. A pulse therapy with long ‘off’ periods followed by short ‘on’ periods is, therefore, foreseeable for them.
Tolerability drawbacks of pulse therapy seem to be limited. A re-seeding of viral reservoirs is probably the rule during STI, and clinical consequences such as generalized lymphoadenopathy may occur but are generally limited in time and seriousness. More important seems to be the recurrence of adverse events related to the active replication of HIV. We observed a case of severe piastrinopenia in a patient that already had this manifestation before initiating HAART. Such patients may cautiously be excluded from therapeutic programmes with STI. However, patients willingly accepted drug suspension and several individuals in the control group of our study, who were aware of this possibility, withdrew their consent to continue therapy. This might constitute a problem in designing controlled studies for pulse therapy .
In summary, this study suggests that prolonged STI in patients whose virus is fully suppressed with antiretroviral therapy and in whom marked immune reconstitution had been obtained is generally safe if the patients are properly monitored. This strategy could be a valid alternative for the long-term management of HIV infection. The main predictor of CD4 cell decline is the nadir CD4 cell count, which allows the identification of patients at greater risk of rapid immune depletion. These findings contribute to the open debate on rethinking treatment strategy in patients who started their first HAART with marginal indications (i.e., CD4 cell count > 500 × 106 cells/l), but also outline possible risks of a late start of therapy (i.e., CD4 cell count < 350 × 106 cells/l). Pulse therapy warrants further careful prospective evaluation, especially to investigate virological and clinical outcomes in the very long term.
We gratefully acknowledge the collaboration and technical assistance of C. Pisoni, D. Nozza, A. Castelli, S. Cavagna, D. Colombo, J. Cortinovis, M. L. Innocenti, M. Mussetti, M. M. Pini.
1. Paterson DL, Swindells S, Mohr J, Brester M, Vergis EN, Squier C, et al. Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med
2. Lerner BH, Gulick RM, Dubler NN. Rethinking non-adherence: historical perspectives on triple-drug therapy for HIV disease. Ann Intern Med
3. Stone VE. Strategies for optimizing adherence to highly active antiretroviral therapy: lessons from research and clinical practice.
CID 2001, 33
4. Carr A, Cooper DA. Adverse effects of antiretroviral therapy. Lancet
5. Moyle GJ, Gazzard BG. A risk-benefit assessment of HIV protease inhibitors. Drug Safety
6. Pellicelli AM, Palmieri F, Boumis E, Soinazzola F, Riamada A, De Martino G, D'Amato C. Patterns of heart disease in HIV infected patients with and without opportunistic infections: a color Doppler echocardiographic study. XII International Conference on AIDS.
Geneva, June 1998 [abstract 60995].
7. Schooley R, Ruane P, Myers R, Beall G, Lampiris H, Berger D, et al. Tenofovir disoproxil fumarate (TDF) for the treatment of antiretroviral experienced patients: a 48 week analysis of a randomized, double blind, placebo controlled study. AIDS
8. Maggiolo F, Migliorino M, Maserati R, Pan A, Rizzi M, Provettoni G, et al. Virological and immunological responses to a once-a-day antiretroviral regimen with didanosine, lamivudine and efavirenz. Antivir Ther
9. Molina JM, Ferchal F, Rancinan C, Raffi F, Rozenbaum W, Sereni D, et al. Once-daily combination therapy with emtricitabine, didanosine and efavirenz in human immunodeficiency virus-infected patients. J Infect Dis
10. Miller V. Structured treatment interruption in antiretroviral management for HIV-1. Curr Opin Infect Dis
11. Rosenberg ES, Altfeld M, Poon SF, Phillips MN, Wilkes BM, Eldridge RL, et al. Immune control of HIV-1 after early treatment of acute infection.
Nature 2000, 407
12. MillerV, Sabin C, Hertogs K, Bloor S, Martinez-Picado J, D'Aquila R, et al. Virological and immunological effects of treatment interruptions in HIV-infected patients with treatment failure. AIDS
13. Deeks SG, Wrin T, Liegler T, Hoh R, Hayden M, Barbour JD, et al. Virologic and immunologic consequences of discontinuing combination antiretroviral drug therapy in HIV-infected patients with detectable viremia. N Engl J Med
14. Devereeux HL, Youle M, Johnson MA, Loveday C. Rapid decline in detectability of HIV-1 drug resistance mutations after stopping therapy. AIDS
15. Izopet J, Massip P, Souyris C, Sandres K, Puissant B, Obadia M, et al. Shift in HIV resistance genotype after treatment interruption and short-term antiviral effect following a new salvage regimen. AIDS
16. Ruiz L, Martinez-Picado J, Romeu J, Paredes R, Zayat MK, Marfil S, et al. Structured treatment interruption in chronically HIV-1 infected patients after long-term viral suppression. AIDS
17. Hirschel B, Fagard C, Oxenius A, Gunthard HF, Garcia F. SSITT: a prospective trial of treatment interruptions in HIV infection. Ninth Conference on Retroviruses and Opportunistic Infections.
Seattle, February 2002 [abstract 528-M].
18. Krolewiecki AJ, Zala C, Gun A, Ochoa C, Perez H, Bouzas MB, et al. Treatment discontinuation in patients who started antiretroviral therapy following the 1996 IAS-USA recommendations.
A prospective randomized trial. XIV International Conference on AIDS
. Barcelona, July 2002 [abstract ThOrB1440].
19. Molina M, Gil P, Granizo JJ, Gorgolas M, Fernandez-Guerrero ML. Discontinuation of highly active antiretroviral therapy (HAART) in asymptomatic HIV-infected patients with CD4+ counts greater than 350 and viral load lower than 70 000 copies. 42nd Interscience Conference on Antimicrobial Agents and Chemotherapy.
San Diego, September 2002 [abstract H-1082].
20. Fauci AS, Bartlett JG, Goosby EP, Kates J, and the Panel on Clinical Practices for treatment of HIV Infection convened by the Department of Health and Human Services (DHHS) and the Henry J. Kaiser Family Foundation. US DHHS Guidelines for the Use of Antiretroviral Agents in HIV-infected Adults and Adolescents.
February 4, 2002. Available at http://www.hivatis.org.
21. Moyle GJ, Gazzard BG. A risk-benefit assessment of HIV protease inhibitors. Drug Safety
22. Lerner BH, Gulick RM, Dubler NN. Rethinking non-adherence: historical perspectives on triple-drug therapy for HIV disease. Ann Intern Med
23. Tebas P, Henry K, Mondy K, Deeks S, Valdez H, Cohen C, et al. Effect of prolonged discontinuation of successful antiretroviral therapy on CD4+ T cell decline in human immunodeficiency virus-infected patients: implications for intermittent therapeutic strategies. J Infect Dis
24. Parish MA, Tarwater P, Lu M, Gallant JE. Prolonged treatment interruption after immunologic response HAART. XIV International Conference on AIDS
. Barcelona, July 2002 [abstract ThOrB1439].
25. McCarthy M. New US HIV treatment guidelines urge a more conservative approach. Lancet
26. Anastos K, Barron Y, Miotti P, Weiser B, Young M, Hessol N, et al. Risk of progression to AIDS and death in women infected with HIV-1 initiating highly active antiretroviral treatment at different stages of disease. Arch Intern Med
27. Silverman WA, Altman DG. Patients’ preferences and randomised trials. Lancet
© 2004 Lippincott Williams & Wilkins, Inc.