Predictors of CD4 cell recovery following initiation of antiretroviral therapy among HIV-1 positive patients with well-estimated dates of seroconversion.

OBJECTIVES: To investigate factors that predict speed of recovery and long-term CD4 cell count in HIV-1 seroconverters initiating combination antiretroviral therapy (cART), and to quantify the influence of very early treatment initiation. We make use of all pre-treatment CD4 counts, because analyses using only a single observation at initiation may be subject to biases. METHODS: We used data from the CASCADE (Concerted Action on SeroConversion to AIDS and Death in Europe) multinational cohort collaboration of HIV-1 seroconverters. We analysed pre- and post-treatment data of patients with seroconversion dates estimated January 2003-March 2014 (n = 7600 for primary analysis) using a statistical model in which the characteristics of recovery in CD4 counts are determined by multiple predictive factors. Secondary analyses were performed incorporating uncertainty in the exact timing of seroconversion to allow more precise estimation of the benefit of very early treatment initiation. RESULTS: 'True' CD4 count at cART initiation was the strongest predictor of CD4 count beyond 3 years on cART. Allowing for lack of complete certainty in the date of seroconversion, CD4 recovery was more rapid for patients in whom treatment was initiated within 4 months. For a given CD4 count, higher viral load (VL) at initiation was strongly associated with higher post-treatment CD4 recovery. For other patient and drug characteristics, associations with recovery were statistically significant but small in magnitude. CONCLUSIONS: CD4 count at cART initiation is the most important factor in predicting post-treatment recovery, but VL provides substantial additional information. If cART is initiated in the first 4 months following seroconversion, recovery of CD4 counts appears to be more rapid.

Demographic and HIV-specific characteristics of participants enrolled in the INSIGHT Strategic Timing of AntiRetroviral Treatment (START) trial.

OBJECTIVES: The risks and benefits of initiating antiretroviral treatment (ART) at high CD4 cell counts have not been reliably quantified. The Strategic Timing of AntiRetroviral Treatment (START) study is a randomized international clinical trial that compares immediate with deferred initiation of ART for HIV-positive individuals with CD4 cell counts above 500 cells/µL. We describe the demographics, HIV-specific characteristics and medical history of this cohort. METHODS: Data collected at baseline include demographics, HIV-specific laboratory values, prior medical diagnoses and concomitant medications. Baseline characteristics were compared by geographical region, gender and age. RESULTS: START enrolled 4685 HIV-positive participants from 215 sites in 35 countries. The median age is 36 years [interquartile range (IQR) 29-44 years], 27% are female, and 45% self-identify as white, 30% as black, 14% as Latino/Hispanic, 8% as Asian and 3% as other. The route of HIV acquisition is reported as men who have sex with men in 55% of participants, heterosexual sex in 38%, injecting drug use in 1% and other/unknown in 5%. Median time since HIV diagnosis is 1.0 year (IQR 0.4-3.0 years) and the median CD4 cell count and HIV RNA values at study entry are 651 cells/µL (IQR 584-765 cells/µL) and 12,754 HIV RNA copies/mL (IQR 3014-43,607 copies/mL), respectively. CONCLUSIONS: START has enrolled a diverse group of ART-naïve individuals with high CD4 cell counts who are comparable to the HIV-positive population from the regions in which they were enrolled. The information collected with this robust study design will provide a database with which to evaluate the risks and benefits of early ART use for many important outcomes.

Early antiretroviral treatment reduces risk of bacille Calmette-Guérin immune reconstitution adenitis.

SETTING: Two centres in Soweto and Cape Town, South Africa. OBJECTIVE: To assess the effects of timing of initiation of antiretroviral treatment (ART) and other factors on the risk of bacille Calmette-Guérin (BCG) related regional adenitis due to immune reconstitution inflammatory syndrome (BCG-IRIS) in human immunodeficiency virus (HIV) infected infants. DESIGN: HIV-infected infants aged 6-12 weeks with CD4 count ≥25% enrolled in the Children with HIV Early Antiretroviral Therapy (CHER) Trial received early (before 12 weeks) or deferred (after immunological or clinical progression) ART; infants with CD4 count <25% all received early ART. All received BCG vaccination after birth. Reactogenicity to BCG was assessed prospectively during routine study follow-up. RESULTS: Of 369 infants, 32 (8.7%) developed BCG-IRIS within 6 months of starting ART, 28 (88%) within 2 months after ART initiation. Of the 32 cases, 30 (93.8%) had HIV-1 RNA > 750 000 copies/ml at initiation. Incidence of BCG-IRIS was 10.9 and 54.3 per 100 person-years (py) among infants with CD4 count ≥25% at enrolment receiving early (at median age 7.4 weeks) vs. deferred (23.2 weeks) ART, respectively (HR 0.24, 95%CI 0.11-0.53, P < 0.001). Infants with CD4 count <25% receiving early ART had intermediate incidence (41.7/100 py). Low CD4 counts and high HIV-1 RNA at initiation were the strongest independent risk factors for BCG-IRIS. CONCLUSIONS: Early ART initiation before immunological and/or clinical progression substantially reduces the risk of BCG-IRIS regional adenitis.

Daily co-trimoxazole prophylaxis in severely immunosuppressed HIV-infected adults in Africa started on combination antiretroviral therapy: an observational analysis of the DART cohort.

BACKGROUND: Co-trimoxazole prophylaxis can reduce mortality from untreated HIV infection in Africa; whether benefits occur alongside combination antiretroviral therapy (ART) is unclear. We estimated the effect of prophylaxis after ART initiation in adults. METHODS: Participants in our observational analysis were from the DART randomised trial of management strategies in HIV-infected, symptomatic, previously untreated African adults starting triple-drug ART with CD4 counts lower than 200 cells per muL. Co-trimoxazole prophylaxis was not routinely used or randomly allocated, but was variably prescribed by clinicians. We estimated effects on clinical outcomes, CD4 cell count, and body-mass index (BMI) using marginal structural models to adjust for time-dependent confounding by indication. DART was registered, number ISRCTN13968779. FINDINGS: 3179 participants contributed 14 214 years of follow-up (8128 [57%] person-years on co-trimoxazole). Time-dependent predictors of co-trimoxazole use were current CD4 cell count, haemoglobin concentration, BMI, and previous WHO stage 3 or 4 events on ART. Present prophylaxis significantly reduced mortality (odds ratio 0.65, 95% CI 0.50-0.85; p=0.001). Mortality risk reduction on ART was substantial to 12 weeks (0.41, 0.27-0.65), sustained from 12-72 weeks (0.56, 0.37-0.86), but not evident subsequently (0.96, 0.63-1.45; heterogeneity p=0.02). Variation in mortality reduction was not accounted for by time on co-trimoxazole or current CD4 cell count. Prophylaxis reduced frequency of malaria (0.74, 0.63-0.88; p=0.0005), an effect that was maintained with time, but we observed no effect on new WHO stage 4 events (0.86, 0.69-1.07; p=0.17), CD4 cell count (difference vs non-users, -3 cells per muL [-12 to 6]; p=0.50), or BMI (difference vs non-users, -0.04 kg/m(2) [-0.20 to 0.13); p=0.68]. INTERPRETATION: Our results reinforce WHO guidelines and provide strong motivation for provision of co-trimoxazole prophylaxis for at least 72 weeks for all adults starting combination ART in Africa. FUNDING: UK Medical Research Council, the UK Department for International Development, the Rockefeller Foundation, GlaxoSmithKline, Gilead Sciences, Boehringer-Ingelheim, and Abbott Laboratories.

Nevirapine/zidovudine/lamivudine has superior immunological and virological responses not reflected in clinical outcomes in a 48-week randomized comparison with abacavir/zidovudine/lamivudine in HIV-infected Ugandan adults with low CD4 cell counts.

BACKGROUND: Triple nucleoside reverse transcriptase inhibitor regimens have advantages as first-line antiretroviral therapy (ART), avoiding hepatotoxicity and interactions with anti-tuberculosis therapy, and sparing two drug classes for second-line ART. Concerns exist about virological potency; efficacy has not been assessed in Africa. METHODS: A safety trial comparing nevirapine with abacavir was conducted in two Ugandan Development of Antiretroviral Therapy in Africa (DART) centres: 600 symptomatic antiretroviral-naïve HIV-infected adults with CD4 counts <200 cells/microL were randomized to zidovudine/lamivudine plus abacavir or nevirapine (placebo-controlled to 24-week primary toxicity endpoint, and then open-label). Documented World Health Organization (WHO) stage 4 events were independently reviewed and plasma HIV-1 RNA assayed retrospectively. Exploratory efficacy analyses are intention-to-treat. RESULTS: The median pre-ART CD4 count was 99 cells/microL, and the median pre-ART viral load was 284 600 HIV-1 RNA copies/mL. A total of 563 participants (94%) completed 48 weeks of follow-up, 25 (4%) died and 12 (2%) were lost to follow-up. The randomized drug was substituted in 21 participants (7%) receiving abacavir vs. 34 (11%) receiving nevirapine (P=0.09). At 48 weeks, 62% of participants receiving abacavir vs. 77% of those receiving nevirapine had viral loads <50 copies/mL (P<0.001), and mean CD4 count increases from baseline were +147 vs. +173 cells/microL, respectively (P=0.006). Nine participants (3%) receiving abacavir vs. 16 (5%) receiving nevirapine died [hazard ratio (HR) 0.55; 95% confidence interval (CI) 0.24-1.25; P=0.15]; 20 receiving abacavir vs. 32 receiving nevirapine developed new or recurrent WHO 4 events or died (HR=0.60; 95% CI 0.34-1.05; P=0.07) and 48 receiving abacavir vs. 68 receiving nevirapine developed new or recurrent WHO 3 or 4 events or died (HR=0.67; 95% CI 0.46-0.96; P=0.03). Seventy-one participants (24%) receiving abacavir experienced 91 grade 4 adverse events compared with 130 events in 109 participants (36%) on nevirapine (P<0.001). CONCLUSIONS: The clear virological/immunological superiority of nevirapine over abacavir was not reflected in clinical outcomes over 48 weeks. The inability of CD4 cell count/viral load to predict initial clinical treatment efficacy is unexplained and requires further evaluation.

Routine versus clinically driven laboratory monitoring of HIV antiretroviral therapy in Africa (DART): a randomised non-inferiority trial.

BACKGROUND: HIV antiretroviral therapy (ART) is often managed without routine laboratory monitoring in Africa; however, the effect of this approach is unknown. This trial investigated whether routine toxicity and efficacy monitoring of HIV-infected patients receiving ART had an important long-term effect on clinical outcomes in Africa. METHODS: In this open, non-inferiority trial in three centres in Uganda and one in Zimbabwe, 3321 symptomatic, ART-naive, HIV-infected adults with CD4 counts less than 200 cells per microL starting ART were randomly assigned to laboratory and clinical monitoring (LCM; n=1659) or clinically driven monitoring (CDM; n=1662) by a computer-generated list. Haematology, biochemistry, and CD4-cell counts were done every 12 weeks. In the LCM group, results were available to clinicians; in the CDM group, results (apart from CD4-cell count) could be requested if clinically indicated and grade 4 toxicities were available. Participants switched to second-line ART after new or recurrent WHO stage 4 events in both groups, or CD4 count less than 100 cells per microL (LCM only). Co-primary endpoints were new WHO stage 4 HIV events or death, and serious adverse events. Non-inferiority was defined as the upper 95% confidence limit for the hazard ratio (HR) for new WHO stage 4 events or death being no greater than 1.18. Analyses were by intention to treat. This study is registered, number ISRCTN13968779. FINDINGS: Two participants assigned to CDM and three to LCM were excluded from analyses. 5-year survival was 87% (95% CI 85-88) in the CDM group and 90% (88-91) in the LCM group, and 122 (7%) and 112 (7%) participants, respectively, were lost to follow-up over median 4.9 years' follow-up. 459 (28%) participants receiving CDM versus 356 (21%) LCM had a new WHO stage 4 event or died (6.94 [95% CI 6.33-7.60] vs 5.24 [4.72-5.81] per 100 person-years; absolute difference 1.70 per 100 person-years [0.87-2.54]; HR 1.31 [1.14-1.51]; p=0.0001). Differences in disease progression occurred from the third year on ART, whereas higher rates of switch to second-line treatment occurred in LCM from the second year. 283 (17%) participants receiving CDM versus 260 (16%) LCM had a new serious adverse event (HR 1.12 [0.94-1.32]; p=0.19), with anaemia the most common (76 vs 61 cases). INTERPRETATION: ART can be delivered safely without routine laboratory monitoring for toxic effects, but differences in disease progression suggest a role for monitoring of CD4-cell count from the second year of ART to guide the switch to second-line treatment. FUNDING: UK Medical Research Council, the UK Department for International Development, the Rockefeller Foundation, GlaxoSmithKline, Gilead Sciences, Boehringer-Ingelheim, and Abbott Laboratories.

Virological and immunological outcomes at 3 years after starting antiretroviral therapy with regimens containing non-nucleoside reverse transcriptase inhibitor, protease inhibitor, or both in INITIO: open-label randomised trial.

BACKGROUND: Antiretroviral therapy has greatly reduced HIV mortality and morbidity. However, the best sequence of regimens and implications of initial regimen for long-term therapeutic success are not well defined. METHODS: In INITIO, a large international randomised trial, we compared antiretroviral therapy with two nucleoside analogue reverse transcriptase inhibitors (didanosine+stavudine) plus either a non-nucleoside reverse transcriptase inhibitor (efavirenz, EFV) or a protease inhibitor (nelfinavir, NFV), or both (EFV/NFV), in patients with HIV-1 infection who had not previously received antiretroviral drugs. Primary outcomes were proportion with undetectable HIV RNA in plasma, and change in CD4 count from baseline at 3 years. Analyses were by intention-to-treat. This study is registered as an International Standard Randomised Controlled Trial, number ISRCTN44582462. FINDINGS: We followed up 911 participants (297 EFV, 311 NFV, 303 EFV/NFV). At 3 years, the proportion with HIV RNA less than 50 copies per mL was highest in the EFV group (188 [74%] EFV, 162 [62%] NFV, 155 [62%] EFV/NFV; p=0.004). Mean (95% CI) increases in CD4 count were 316x10(6) cells per L (288-343) for EFV, 289x10(6) cells per L (262-316) for NFV, and 274x10(6) cells per L (231-291) for EFV/NFV (p=0.1). Fewer participants in the EFV group than in the other groups stopped adequate antiretroviral therapy for more than 30 days (p=0.005). Participants in the EFV/NFV group had shorter time to stopping the initial regimen (p<0.0001) and to a treatment modifying adverse event (p=0.04) than those in the other groups. INTERPRETATION: Starting antiretroviral therapy with a three-drug/two-class regimen including efavirenz was better than starting with regimens including nelfinavir or efavirenz plus nelfinavir in terms of virological suppression and durability of the initial regimen. The shorter time on adequate antiretroviral therapy or to a treatment-modifying adverse event might explain the absence of additional benefit for the four-drug regimen.

A randomized controlled trial of the value of phenotypic testing in addition to genotypic testing for HIV drug resistance: evaluation of resistance assays (ERA) trial investigators.

OBJECTIVES: To assess the clinical utility of phenotypic resistance testing in addition to genotypic resistance testing among HIV-1-infected patients experiencing virologic failure and with limited therapeutic options. DESIGN: Multicenter randomized trial. METHODS: Patients were eligible if a decision had been made to switch antiretroviral therapy, the most recent HIV-1 RNA plasma viral load (VL) exceeded 2000 copies/mL, and the clinician was unable to select a potent regimen of 3 or more drugs without access to a resistance test. Subjects were randomized to genotypic resistance testing alone (G arm) or to genotypic plus phenotypic testing (G + P arm). Patients had access to resistance testing at any time during follow-up (minimum of 1 year) according to the original allocation. The primary end point was change in plasma VL from baseline at 12 months. RESULTS: Three hundred eleven patients were recruited between February 2000 and July 2001. At baseline, mean VL and CD4 count were 4.23 log10 copies/mL and 275 cells/mm, respectively, and subjects had previous exposure to a mean of 7.7 antiretroviral drugs. There was no appreciable difference between the study arms in the drug regimens prescribed after randomization. Mean reduction in VL load at 12 months was similar in the 2 arms (G: 1.37 log10 reduction, G + P: 1.28 log10 reduction; P = 0.77), as was the proportion of subjects with VL <50 copies/mL (G: 35%, G + P: 27%). CONCLUSION: The study did not demonstrate added value of phenotypic resistance testing in conjunction with genotypic resistance testing in patients with limited therapeutic options.

A comparison of two methods for the estimation of precision with incomplete longitudinal data, jointly modelled with a time-to-event outcome.

Several methods for the estimation and comparison of rates of change in longitudinal studies with staggered entry and informative drop-outs have been recently proposed. For multivariate normal linear models, REML estimation is used. There are various approaches to maximizing the corresponding log-likelihood; in this paper we use a restricted iterative generalized least squares method (RIGLS) combined with a nested EM algorithm. An important statistical problem in such approaches is the estimation of the standard errors adjusted for the missing data (observed data information matrix). Louis has provided a general technique for computing the observed data information in terms of completed data quantities within the EM framework. The multiple imputation (MI) method for obtaining variances can be regarded as an alternative to this. The aim of this paper is to develop, apply and compare the Louis and a modified MI method in the setting of longitudinal studies where the source of missing data is either death or disease progression (informative) or end of the study (assumed non-informative). Longitudinal data are simultaneously modelled with the missingness process. The methods are illustrated by modelling CD4 count data from an HIV-1 clinical trial and evaluated through simulation studies. Both methods, Louis and MI, are used with Monte Carlo simulations of the missing data using the appropriate conditional distributions, the former with 100 simulations, the latter with 5 and 10. It is seen that naive SEs based on the completed data likelihood can be seriously biased. This bias was largely corrected by Louis and modified MI methods, which gave broadly similar estimates. Given the relative simplicity of the modified MI method, it may be preferable.

Age as a determinant of survival in HIV infection.

Age is a major determinant of mortality for many diseases including HIV infection, yet the effect of age is rarely studied directly. In this article, we review what is known about the effect of age at seroconversion on HIV disease progression and survival prior to the widespread use of HAART before describing appropriate methods for adjusting for background mortality in more detail. We then investigate the impact of HAART on the effect of age at seroconversion on mortality and consider the estimation of the age effect in seroprevalent cohorts with regard to lack of knowledge of the true age at infection. Finally, we discuss mechanisms by which age at seroconversion might impact on disease progression and death. Throughout, we use published results by the Collaborative Group on AIDS Incubation and HIV Survival (CGAIHS), and published results and data from the Concerted Action on SeroConversion to AIDS and Death in Europe (CASCADE) for illustration.

Analysis of failure time hierarchical data in the presence of competing risks with application to oral contraceptive pill use in Egypt.

Problems of practical interest in the analysis of data on contraceptive use, from Demographic and Health Surveys (DHS), include the estimation of the cause-specific probability of discontinuation by time t (the cumulative incidence function), in the presence of other competing causes and the evaluation of the effect of covariates on the cause-specific hazards of discontinuation. Methods of analysis of failure time data with competing risks are by now fairly well developed in the case of a simple random sample. However, the data from the DHS are clustered by geographical areas and include multiple episodes per woman. For a marginal (population average) approach, we propose using methods developed for simple random samples with standard errors calculated using a double bootstrap to take account of the clustered hierarchical nature of the data. In the conditional approach, the cause-specific hazards are modelled as log-linear functions of the covariates conditional on random effects of clusters and women, using a three-level multinomial discrete-time logit model. The methods are applied to data from Egypt 1992 DHS on the oral contraceptive pill use.

Impact of missing data due to drop-outs on estimators for rates of change in longitudinal studies: a simulation study.

Many cohort studies and clinical trials are designed to compare rates of change over time in one or more disease markers in several groups. One major problem in such longitudinal studies is missing data due to patient drop-out. The bias and efficiency of six different methods to estimate rates of changes in longitudinal studies with incomplete observations were compared: generalized estimating equation estimates (GEE) proposed by Liang and Zeger (1986); unweighted average of ordinary least squares (OLSE) of individual rates of change (UWLS); weighted average of OLSE (WLS); conditional linear model estimates (CLE), a covariate type estimates proposed by Wu and Bailey (1989); random effect (RE), and joint multivariate RE (JMRE) estimates. The latter method combines a linear RE model for the underlying pattern of the marker with a log-normal survival model for informative drop-out process. The performance of these methods in the presence of missing data completely at random (MCAR), at random (MAR) and non-ignorable (NIM) were compared in simulation studies. Data for the disease marker were generated under the linear random effects model with parameter values derived from realistic examples in HIV infection. Rates of drop-out, assumed to increase over time, were allowed to be independent of marker values or to depend either only on previous marker values or on both previous and current marker values. Under MACR all six methods yielded unbiased estimates of both group mean rates and between-group difference. However, the cross-sectional view of the data in the GEE method resulted in seriously biased estimates under MAR and NIM drop-out process. The bias in the estimates ranged from 30 per cent to 50 per cent. The degree of bias in the GEE estimates increases with the severity of non-randomness and with the proportion of MAR data. Under MCAR and MAR all the other five methods performed relatively well. RE and JMRE estimates were more efficient(that is, had smaller variance) than UWLS, WLS and CL estimates. Under NIM, WLS and particularly RE estimates tended to underestimate the average rate of marker change (bias approximately 10 per cent). Under NIM, UWLS, CL and JMRE performed better in terms of bias (3-5 per cent) with the JMRE giving the most efficient estimates. Given that markers are key variables related to disease progression, missing marker data are likely to be at least MAR. Thus, the GEE method may not be appropriate for analysing such longitudinal marker data. The potential biases due to incomplete data require greater recognition in reports of longitudinal studies. Sensitivity analyses to assess the effect of drop-outs on inferences about the target parameters are important.

Comparison of a point mutation assay with a line probe assay for the detection of the major mutations in the HIV-1 reverse transcriptase gene associated with reduced susceptibility to nucleoside analogues.

This study compares the performance of a line probe assay (LiPA) for the detection of the major mutations associated with reduced sensitivity to nucleoside analogues with a well characterised point mutation assay (PMA). Plasma samples obtained from patients in a trial of four reverse transcriptase inhibitors (MRC Quattro Trial) were tested by both LiPA and PMA at baseline, 32nd and 64th weeks for the presence of drug resistance associated mutations in the reverse transcriptase (RT) gene. HIV-1 RNA was extracted from plasma by the Boom method and amplified by RT-PCR prior to being tested by LiPA or PMA. Assay discrepancies were further investigated by sequencing of the RT gene. Of 275 samples available from 98 trial subjects, 246 samples were successfully amplified by PCR and analysed by LiPA and PMA for six mutations. Of the 1476 individual codons analysed, LiPA successfully assayed 1444 (97.8%) and PMA gave a result with 1418 (96.1%). LiPA failed to give a result for 32 codons from 22 samples and PMA failed with 58 codons from 38 samples. Gross differences between the two assays, in which one scored a codon as wild-type only and the other as mutant only or vice versa, occurred at 28 codons analysed (1.9%) representing 26 samples from 20 subjects. Sequencing of 22 of the 26 samples confirmed the LiPA result in nine cases, the PMA result in 11 and detected a novel variant at codon 215 in four cases. The PMA and LiPA approach to the detection of the major mutations that are genotypically associated with reduced sensitivity to nucleoside analogues can correctly detect mutations in 97% of the cases.

Analysis of multivariate failure-time data from HIV clinical trials.

We illustrate the use of marginal methods for the analysis of multivariate failure-time data using a large trial in HIV infection in which the composite endpoint of AIDS or death incorporates more than 20 events with varying severity. Multivariate failure-time methods are required to investigate whether treatment delays development of new AIDS events. AIDS events can be grouped and treatment effects estimated using only the first event to occur in each group for each individual. Alternatively, all events can be included by fitting a separate baseline hazard for development of each event, and restricting treatment effects to be common within groups of events. In either case, model-based or minimum-variance estimates of the overall effect of treatment can be constructed. The covariance matrix for the treatment-effect estimates can be used in multiple testing procedures. Results from the Delta trial suggest that combination antiretroviral therapy with AZT plus either ddI or ddC may delay progression to more severe AIDS events compared to AZT monotherapy. These late events are generally untreatable and prophylaxis is not available. Trials are not generally powered to detect treatment effects on individual events making up a composite endpoint, and therefore all analyses are exploratory rather than providing definitive evidence. However, marginal multivariate models provide an easily available approach for modeling the effect of covariates on multiple disease processes, and allow the likely effects of treatment to be presented in a manner which is easily understood. They can be used in a variety of ways to explore different patterns of treatment effects and are also useful for testing multiple hypotheses regarding treatment effects on several different composite endpoints.

Randomization-based methods for correcting for treatment changes: examples from the Concorde trial.

We develop analysis methods for clinical trials with time-to-event outcomes which correct for treatment changes during follow-up, yet are based on comparisons of randomized groups and not of selected groups. A causal model relating observed event times to event times that would have been observed under other treatment scenarios is fitted using the semi-parametric approach of Robins and Tsiatis (avoiding assumptions about the relationship between treatment changes and prognosis). The methods are applied to the Concorde trial of immediate versus deferred zidovudine, to investigate how the results would have differed if no participant randomized to deferred zidovudine had started treatment before reaching ARC or AIDS. We consider issues relating to model choice, non-constant treatment effects and censoring.

The rabbit study: ritonavir and saquinavir in combination in saquinavir-experienced and previously untreated patients.

Thirteen protease inhibitor-naive patients with HIV-1 infection, and 12 patients with a median of 58 months prior treatment with saquinavir (SQV) monotherapy, were treated with SQV (400 mg twice daily) and ritonavir (RIT, 500 mg twice daily) in a study designed to assess the effect of prior treatment with SQV monotherapy on the antiretroviral activity of RIT-SQV combination therapy. Median baseline viral load and CD4+ cell counts were 155,000 and 262,000 copies/ml and 333 and 225 cells/mm3 in the naive and experienced groups, respectively. Mean viral load changes at 24 weeks were -1.63 and -0.27 log copies/ml in the naive and SQV-experienced groups, respectively (intent-to-treat analysis). Baseline genotype by point mutation assay and sequencing in the SQV-experienced group was highly predictive of virological response. Eight of 11 SQV-experienced patients had evidence of phenotypic resistance to RIT at baseline, despite previous treatment with SQV only. There was strong correlation between phenotypic resistance to RIT and the presence of the L90M mutation. We conclude that prolonged prior treatment with saquinavir monotherapy may produce cross-resistance to ritonavir and reduce the subsequent response to ritonavir-saquinavir in combination. In this study, both phenotypic resistance to ritonavir and presence of the L90M mutation predicted the viral load response to ritonavir-saquinavir.

Estimation and comparison of rates of change in longitudinal studies with informative drop-outs.

Many cohort studies and clinical trials have designs which involve repeated measurements of disease markers. One problem in such longitudinal studies, when the primary interest is to estimate and to compare the evolution of a disease marker, is that planned data are not collected because of missing data due to missing visits and/or withdrawal or attrition (for example, death). Several methods to analyse such data are available, provided that the data are missing at random. However, serious biases can occur when missingness is informative. In such cases, one needs to apply methods that simultaneously model the observed data and the missingness process. In this paper we consider the problem of estimation of the rate of change of a disease marker in longitudinal studies, in which some subjects drop out prematurely (informatively) due to attrition, while others experience a non-informative drop-out process (end of study, withdrawal). We propose a method which combines a linear random effects model for the underlying pattern of the marker with a log-normal survival model for the informative drop-out process. Joint estimates are obtained through the restricted iterative generalized least squares method which are equivalent to restricted maximum likelihood estimates. A nested EM algorithm is applied to deal with censored survival data. The advantages of this method are: it provides a unified approach to estimate all the model parameters; it can effectively deal with irregular data (that is, measured at irregular time points), a complicated covariance structure and a complex underlying profile of the response variable; it does not entail such complex computation as would be required to maximize the joint likelihood. The method is illustrated by modelling CD4 count data in a clinical trial in patients with advanced HIV infection while its performance is tested by simulation studies.

Are HIV-infected patients with rapid CD4 cell decline a subgroup who benefit from early antiretroviral therapy?

We have developed a model to determine whether asymptomatic HIV-infected individuals who have a rapid CD4 cell decline are a subgroup who might benefit from early antiretroviral therapy. Data were obtained from a subgroup of participants in the Concorde and EACG020 trials, two randomized, double-blind, comparative trials of immediate (IMM) versus deferred (DEF) zidovudine therapy in asymptomatic HIV-infected individuals. The subgroup comprised 297 patients (IMM = 154, DEF = 143) who had at least one CD4 cell count before and after randomization. The median CD4 cell count at randomization was 491 x 10(6)/L, and the median follow-up was 61 months. The rate of CD4 decline before and after randomization was estimated using multi-level linear regression analysis, and patients were stratified into quartiles according to the rate of CD4 cell decline before randomization. Outcome measures were the development of AIDS, a 50% drop in CD4 count from the baseline, and death. A Cox proportional hazards model was used to examine whether the effect of zidovudine on disease progression varied according to the previous rate of CD4 decline. We found that a more rapid rate of CD4 decline before randomization was associated with a greater reduction in the rate of CD4 decline following IMM antiretroviral therapy (r = -0.5, P = 0.03). The greatest risk reduction in disease progression with IMM antiretroviral therapy was seen in the quartile of patients with the highest rate of CD4 decline (> or = 26 x 10(6) cells/L per 6 months) (hazards ratio (HR) = 0.61, 95% CI = 0.35-1.05). However, this effect was statistically significant in only the Concorde trial (HR = 0.48, 95% CI = 0.29-0.89). In contrast, we found no evidence in the EACG020 trial of any trend towards greater benefit in those with the most rapid CD4 cell decline. These findings suggest that asymptomatic patients with rapid CD4 cell decline are a subgroup likely to benefit from early antiretroviral therapy. This analytic approach should now be replicated in trials of combination therapy, and these should include viral load data.