======================================================================== HSEM 3010 Spring 2007 Clinical Trials Assignment 4 Due: February 26, 2007 Answer the following for the paper of the SMART Study Group: CD4+ Count-Guided Interrpution of Antiretroviral Treatment (N. Engl J. Med (2006) v. 355, pp 2283-2296). See also the Editorial, NEJM (2006) v. 355, pp 2359-2361. 1. What was the motivation for doing this clinical trial? Toxicity, side effects, and cost of drugs that decrease the chances of progression of HIV/AIDS. If patients could make more sparing use of these drugs without increasing their chances of opportunistic infections, death, and other signs of HIV progression they would have fewer side effects and costs of treatment would be lower. 2. What was the design? This was basically a two-group randomized multicenter clinical trial. Participants were assigned to either (1) drug conservation (DC) or (2) viral suppression (VS). In the DC group, the therapy was determined by counts of CD4+ cells in the blood (higher CD4+ implied decreases in antiretroviral therapy). In the VS group, drugs were prescribed based on 2003 guidelines by an NIH-convened panel. This was not a double blind design - neither patients nor their doctors were masked to treatment assignment. 3. What patients were eligible for this trial? HIV-infected patients with a CD4+ count greater than 350 per mm^3, age > 13, not pregnant or breast-feeding; willing to comply with the assigned study regimen; willing and able to give written informed consent. 4. Why is the CD4+ cell count important? CD4+ cells are an indicator of immune function. A CD4+ cell count below 250 mm^3 means the person is at high risk of an opportunistic infection or other complications. 5. What was the primary endpoint ? Time to first occurrence of opportunistic disease or death. 6. The sample size target for the trial was 6000 patients. What was the basis for that? They estimated that they would need 6000 participants in order to have 80% power of detecting a 17% reduction in the rate of the combined endpoint of opportunistic disease or death (OD + D), (two-sided alpha = 0.05). The main reason that such large numbers were needed was that the event rates were expected to be relatively low. They expected 910 events total after patients had been in the study 6 years. The 17% reduction was expected to be in the drug-conservation (DC) group, as compared to the VS group. 7. In the Statistics section of the Methods, it says: "Randomization was stratified according to clinical site with the use of permuted blocks of random sizes." What does this mean, and why did they do it? How many clinical sites were there? Where were they? In 'permuted block' designs, the randomization schedule is divided into relatively short segments (lengths randomly chosen) of sizes, for example, 2, 4, and 6. The numbers of assignments to each group within each block are equal. This ensures that at any point in the schedule, there will be approximate balance between the treatment groups. There were 318 sites in 33 countries on all continents except Antarctica. 8. The Data and Safety Monitoring Board terminated this trial early. What factors went into their decision? What is your understanding of the "O'Brien-Fleming boundary and the Lan- DeMets spending function" ? The trial stopped enrollment at its 6th meeting (Jan 10, 2006) "because of a safety risk in the DC group and because it appeared very unlikely that superiority of the DC groups would be demonstrated when all the data were in." The Lan-DeMets spending function is a means of specifying the monitoring boundaries for the study in such a way that (1) it is unlikely that the study will be stopped very early, and (2) the probability that at least one of th boundaries is reached at some point during the study, if the null hypothesis is true, is still close to 5%. The basic problem here is, if you perform statistical tests a number of times, the chances that you will come up with a 'significant' answer is greater than if you just perform one test at the end. The 'Lan-DeMets boundaries' are specified so that the chance of crossing the boundary at least once is approximately equal to (but a little bit less than) the prespecified alpha. 9. Do you agree with the DSMB's decision to stop early ? Why or why not? What are some reasons that might be considered to continue the study longer? The data look quite convincing, especially Figure 2A. I think they were right to stop. Of course you can always learn more by continuing a study for a longer time. Especially you may be able to learn more about which patients are more likely to benefit from the treatment being tested. However, it is not ethical to continue to give people the inferior treatment if it is very unlikely that the differences between the groups are just due to chance. 10. What does Figure 2 tell you? What does 'censored observation' mean? Figure 2A is strongly indicative of a beneficial effect for the VS group - after 44 months, the cumulative incidence in the drug- conservation group was more twice as high as that in the VS group, and the difference between the two cumulative incidence curves was getting wider. The other panels in figure 2 provide weaker evidence, but in all cases, the event rates were lower in the VS group. A 'censored observation' is the time of followup in the study for someone who did not have the event of interest. The idea is, if the followup were extended indefinitely, that person would eventually have the event; since the person was not observed for a long enough time to have an event, we say that his/her event time was 'censored'. 11. What does Figure 3 tell you? Figure 3 says that the hazard ratios [similar to a relative risk] for the DC group versus the VS group were all bigger than 1.0 for many different subgroups of patients, and that many of the confidence intervals for the hazard ratios within subgroups were bigger than 1.0. This indicates that the treatment effect was similar regardless of age, gender, race, baseline CD4+ counts, duration of disease, etc. The one subroup where the treatment effect was closest to 1.0 was 'viral load' < 400 copies/ml, i.e., if the number of virus particles in the blood was low, the VS treatment was not very different from the DC treatment. 12. Do you find the results of this trial convincing? Explain. Yes - see answers to Q.9 and Q.11. 13. Do you think it is likely that the results of this trial will change clinical practice? Yes. This was an expedited publication in the leading medical journal, with a supportive editorial, and the results appear to be strong and definitive. A doctor who does not follow the VS regimen in patients similar to those who were entered into this study is putting himself and his patients at risk for bad outcomes. 14. Some people were disappointed with the results of this study. Who were those people, and why were they disappointed? Probably HIV/AIDS patients, who do not like the side-effects and costs of the more aggressive therapy. Also their MDs. Also third-party payers HMOs, health insurance groups. However, both patients and doctors may have been happy with the outcome of this study because it gives definitive guidelines on how patients at this stage of HIV/AIDS should be treated. 15. Who sponsored this study? Why? NIAID, the National Institute of Allergy and Infectious Disease, one of the National Institutes of Health. This was an unusually large study and various drugs were employed, which means it is not likely that the study would be undertaken by a drug company. 16. Has there been much improvement in the treatment of HIV/AIDS since the disease emerged in the early 1980s? There has been really remarkable improvement. When the disease first emerged into public consciousness in the early 1980s, there was no treatment at all. Effective antiretroviral drugs were developed in the 1980s, 1990s, and 2000s. In the early 1980s, the median survival time after HIV infection was 7+ years; it is now much longer, to the point that some patients are essentially living with a treatable chronic disease rather than an inexorably fatal disease. However more progress is needed. 17. Does the 'profit motive' play a role in medical progress? Explain. Very definitely. The new drugs for HIV/AIDS have been mostly developed by large drug companies. They do expect to make money from effective drugs because the patient population is large and continues to grow. 18. Why is HIV/AIDS difficult to prevent and treat? Why isn't there a vaccine? HIV/AIDS attacks the immune system itself. Most viral diseases infect other cells in the body, but HIV attacks CD4+ cells and T cells and disables the body's defenses against opportunistic infections and opportunistic cancers (e.g. Kaposi's sarcoma). Some vaccine trials have been carried out but the results have not been promising. 19, What are the major points made in the Editorial ? 1) Some previous small studies suggested that it might be safe to follow DC-like regimens. The DC versus VS question was important to both patients and their doctors. 2) This study was designed to settle the question of DC versus VS. 3) This study was stopped after an average of 16 months of followup, though it was projected to go to 72 months. The results were definitive. 4) "The results of the SMART study remind us of the benefit of using adequately powered randomized trials to evaluate new treatment strategies in clinical medicine." 5) It was surprising that the negative effects of DC treatment were not just in opportunistic diseases. 6) There is room for additional research in certain patient subgroups, e.g., pregnant women with HIV/AIDS. 20. HIV/AIDS is largely preventable by behavior modification. Some people argue that individuals should be responsible for the consequences of their own behavior. Others argue that the government should shift its resources more toward prevention rather than treatment. Others argue that studies like this should be done by drug companies, not by the government. What are your views? These kinds of studies are mostly not going to be done by private industry. Humans have certain human vulnerabilities, and some health conditions are beyond the control of the victim. In my view we should recognize this and not approach disease and treatment from a moralistic point of view. The fact is, a lot of people now have HIV/AIDS and as a society we should do what we can to help them, including studies like the SMART study. That said, a basic premise of public health is that prevention is better than cure - better in the sense of saving more lives and reducing pain and costs of illness and treatment. More research in prevention (behavioral and vaccine) is definitely a good idea. Public health IMPLEMENTATION of preventive approaches, especially behavioral education and modification, is maybe not so much a matter of science and research as of policy and setting of priorities. In societies less affluent than the U.S., it seems imperative to put more stress on prevention. =================================================================================