Prognosis of HIV-positive patients treated with antiretroviral therapy: comparative analyses and treatment strategies

Highly effective treatments consisting of combinations of different drugs are now available for HIV-positive people and have dramatically reduced the risk of acquired immune deficiency syndromes (AIDS) and death since their introduction in 1996. Increasing numbers of patients have been treated for many years, but outcomes in these people remain unclear, patterns of cause of death have changed, and new predictors of outcomes are emerging. Randomised clinical trials are the best way to examine treatment effects, but most trials in the field rely on outcomes that are measurements from blood tests (rather than AIDS and death, which are of more interest to patients and doctors) and recruit small numbers of patients followed for a short length of time. Trials often exclude sicker patients, and cannot estimate the effectiveness of drugs in important subgroups such as women, those aged >50 years, or ethnic minorities. Many questions relating to treatment strategies cannot be, or have not been, answered in trials. Therefore information about long term outcomes and effects of treatment in patient subgroups must largely rely on large collaborations of HIV cohorts - studies in which groups of HIV positive people are followed over time.

This application seeks funding for the ART Cohort Collaboration (ART-CC), an international collaboration that combines data from 19 HIV cohort studies from Europe and North America. It is coordinated by a small team at Bristol University, and administered by a steering committee of representatives from the contributing studies with expertise in HIV medicine, clinical epidemiology and medical statistics, and patient representatives. We will work within European and North American regional collaborations, which will provide considerable value for money because we can take advantage of their data management systems, and also ensure that we have a distinct and unique research agenda. We expect to combine data on at least 100,000 patients. We will work with collaborations in Southern and West Africa in order to compare outcomes among patients treated in these settings, in which resources for treatment and health care are limited, with outcomes among patients treated in Europe and North America. We will update our datasets annually during the period covered by the application.

Our objectives are: (A) Examine the prognosis of patients on treatment for AIDS, deaths from all causes and deaths from specific causes, focusing on prognosis over many years, the role of new factors that predict AIDS and death, and prognosis from some years after starting treatment; (B) Examine differences in prognosis between regions and settings including the roles of gender and ethnic group; (C) Estimate life expectancy, defined as the average number of additional years a person will live, from the time of starting treatment and separately for groups of patients with different characteristics; (D) Conduct collaborative work on prognosis in low income settings, such as sub-Saharan Africa, where the majority of HIV-positive people live. We will use statistical methods that mimic randomised trials using cohort data to: (E) Investigate strategies for regimen modification (switching) in patients in whose blood HIV-1 virus can be detected, and examine risk factors for and prognosis according to the type of regimen change; (F) Estimate the adverse effects of missed doses or interruptions to treatment on clinical outcomes; (G) Examine the best time to start treatment for HIV in patients who also require treatment for other specific infections that occur when the immune system has been damaged and that affect the central nervous system.

Based on our past experience, we expect the results of this research to be of direct relevance to the care of HIV-positive people and to be incorporated into treatment guidelines. We will provide online risk and life expectancy calculators, as well as summaries for patients, on the study web site.

Antiretroviral therapy (ART) dramatically reduces rates of AIDS and death in HIV-positive people in Europe, North America and low-income settings, particularly those who start early and attain viral suppression. However mortality rates still exceed those in the general population, and there are increasing numbers of older patients and those treated for many years. Predictors of prognosis are changing to include factors beyond conventional markers of HIV disease progression: long term outcomes remain unclear. This application seeks funding for a collaboration of 19 European and North American HIV cohort studies. Working within regional collaborations, we plan to combine data on >100000 treated patients and compare outcomes with those in Southern and West Africa. Objectives are: A) Examine prognosis for AIDS and deaths from all and specific causes among patients on ART, focusing on non-HIV biomarkers, additional CD4 measures, co-infections, new measures of and interactions with HIV virus burden, and HIV-1 subtype; B) Investigate heterogeneity in outcomes between regions and settings; C) Estimate life expectancy overall and in subgroups; D) Conduct collaborative work on prognosis in low-income settings (comparison of mortality in S. Africa, Europe and N. America allowing for incomplete death ascertainment, and prognosis for HIV-2 in W. Africa). We will use causal inference methods to examine optimal treatment strategies including E) strategies for regimen modification (switching) in patients with virological failure of first ART regimen; F) effect of non-adherence to ART on treatment failure and progression to AIDS and death; G) optimal timing of ART in patients presenting with specific CNS AIDS-defining opportunistic infections such as cryptococcal meningitis and toxoplasmic encephalitis. We expect our research results to be of direct relevance to the care of HIV-positive people and to be published in high impact journals and incorporated in treatment guidelines.

Our research programme aims to provide information that will be directly useful to patients and health professionals making decisions about antiretroviral therapy (ART). An improved understanding of factors predicting prognosis among treated patients will help doctors and other health professionals improve and optimise care for high-risk patients and guide patients on ways to improve their prognosis. Our previous work has directly influenced, and been highly cited in, national and international treatment guidelines on treatment of HIV-positive people, and the objectives of our current application address questions of direct relevance to formulation of future guidelines. We will respond to requests for additional analyses from those writing guidelines (for example, the World Health Organisation (WHO)).

Prognostic models are important in enabling policy-makers and those planning health services to predict levels of care that will be needed by an increasing and aging population of treated HIV-positive people. Defining the burden of morbidity and mortality associated with HIV in the era of combination ART will help future planning of both health service provision and social welfare. ART is an expensive treatment that has to be taken for life, but costs may reduce as some drugs come off-patent in the coming years. Our work will assist in studies of the optimum choice of initial ART regimen, balancing effects on rates of AIDS and death, regimen durability, and costs.

Our work on heterogeneity will provide insights into the influence of health systems on outcomes, and an improved understanding of differences between patient groups (for example, according to sex or ethnicity) will enable targeting of groups in which outcomes are sub-optimal, for example because of imperfect adherence, presence of modifiable risk factors or a need for improved treatment of comorbidities.

Estimates of life-expectancy will enable health professionals to answer one of the most common questions asked by patients considering starting ART, or changing or interrupting their regimen - "How long have I got to live?". We will work with the insurance industry to promote life insurance for treated HIV-positive people, by providing estimates in the form required for actuarial calculations.

Comparisons between outcomes of ART in high and low income settings are key to identifying problems of delivering ART in low income settings (for example, very high early mortality), as well as demonstrating that excellent outcomes can be achieved with much more limited resources than are available in Europe or North America.

Randomised clinical trials (RCTs) underpin decisions about optimal treatments and treatment strategies, but many questions about ART cannot be, or have not yet been, addressed in RCTs. The best available alternative is to conduct analyses of cohort studies using methods that mimic RCTs and can hence provide causal inferences, based on strong but clearly defined assumptions. We will use these methods to answer questions about effects of ART regimen modification (switching), non-adherence to ART, and the optimal timing of ART. Results of this work will again be directly useful to health professionals and patients making treatment decisions.