点击显示 收起
Population Council, New York, New York
Population Council, Johannesburg, Infectious Disease Epidemiology Unit, School of Public Health and Family Medicine, University of Cape Town, Cape Town
Department of Medical Microbiology, Medical University of Southern Africa, South Africa
The article by Fleming and Richardson, entitled "Some Design Issues in Trials of Microbicides for the Prevention of HIV Infection" [1], highlights some of the challenges in conducting microbicide clinical trials. Fleming comments on these challenges from the perspective of one who is (1) lead statistician for the HIV Prevention Trials Network, which shortly thereafter initiated a Phase 2b study of 2 potential microbicides, and (2) a consultant to the US Food and Drug Administration. We focus specifically on Fleming and Richardson's position with regard to including a third arm, an unblinded-control group, within a trial.
First, Fleming and Richardson assert that one of the major merits of the use of an unblinded-control group is that it will provide an estimate of "condom migration" (a term used in the microbicide field to describe a change in condom usespecifically, a decrease that is due to the availability of other HIV-prevention options, a decrease that may, in turn, affect the "real world" effectiveness of a given microbicide). However, true microbicide effectiveness and condom migration will not be addressed by the "condom-only" arm of a clinical trial, because (1) the study population will be very highly selected and, therefore, totally unrepresentative of all at-risk women in the study communities and (2) there is little volition involved in women's decisions to use or not use the microbicide in the context of a trial, whereas study participants are instructed to use the product and are thoroughly counseled on safe sex. Furthermore, concern about condom migration seems inappropriate in sub-Saharan Africa, where condom use remains low despite high rates of disease [2] and educational programs about prevention [3]. Mathematical modeling has shown that condom migration would have an insignificant impact on the spread of HIV if the preexisting condom use were low [4]. Because a clinical trial is a controlled experiment, we should not expect that the outcome is directly representative of what will happen in the real world, should a microbicide become approved [5].
Second, adding an unblinded-control group increases a trial's cost by more than half, because of the loss of power for multiple-comparison penalties [6]. Trials using unblinded controls are usually conducted when no reasonable placebo is available. Although a true placebo may be impossible to obtain, reasonable placebos for microbicide trials do exist. Furthermore, the possible effect of the placebos is likely to be very minor, in comparison with the effect that we would expect an efficacious microbicide to have.
Third, the use of an unblinded-control group threatens the study because all groups are not treated exactly the same [5], and this disparity may introduce condom use that is different from that in the gel-using group(s), the very phenomenon that the use of an unblinded-control group is designed to quantify. This disparity is problematic specifically because the lack of surrogate end points for HIV means that neither the use of condoms nor the use of microbicides can be reliably assessed. There exists the risk that an unblinded-control group will perceive themselves as benefiting less from being included within a clinical trialbecause they receive no gel whereas the other 2 groups do receive geland this perception may both introduce gel sharing and increase loss to follow up in the unblinded-control, nongel-using group, which, in turn, has the potential to introduce bias into the evaluation of the microbicide's impact on the risk for HIV.
Finally, the efficacy of a product has a major impact on the motivation to use it. Although the urgent need for an available, effective microbicide is greatest where the products are currently being tested, the goal is to make them available globally. The use of a microbicide that has been proven to be efficacious will vary by location, because of differences in attitudes and behavior. However, the use of a microbicide that has unknown efficacy (e.g., when it is being tested in a clinical trial) will likely be different than the use of a microbicide whose efficacy is known (e.g., once efficacy has been quantified and the product is on the market). This difference will cause a difference in effectiveness. Real-world effectiveness, therefore, cannot be reliably assessed before efficacy is known but, rather, should be researched by using other trial designs in post-marketing studies.
In summary, many of Fleming and Richardson's arguments for including an unblinded-control arm within a trial overlook obvious concerns about how the data generated by such an arm should be interpreted. As a result, the most compelling reasons for including an unblinded-control arm are reduced to academic interests regarding individual behaviors in the context of clinical trials. Given the urgent need for the identification and mass promotion of an effective microbicide, there may be better uses of time and resources than including an unblinded-control arm within a trial. Current microbicide trials should address efficacy, in an effort to pave the way for future trials focusing on effectiveness. We thank Fleming and Richardson for raising some interesting points about this important topic.
References
1. Fleming TR, Richardson BA. Some design issues in trials of microbicides for the prevention of HIV infection. J Infect Dis 2004; 190:66674. First citation in article
2. Eaton L, Fisher AJ, Aaro LE. Unsafe sexual behavior in South African youth. Soc Sci Med 2003; 56:14965. First citation in article
3. Hartung TK, Nash J, Ngubane N, Fredlund VG. AIDS awareness and sexual behaviour in a high HIV prevalence area in rural northern Kwazulu-Natal, South Africa. Int J STD AIDS 2002; 13:82932. First citation in article
4. Foss AM, Vickerman PT, Heise L, Watts CH. Shifts in condom use following microbicide introduction: should we be concerned AIDS 2003; 17:122737. First citation in article
5. Stein ZA, Myer L, Susser M. The design of prophylactic trials for HIV: the case for microbicides. Epidemiology 2003; 14:803. First citation in article
6. Coplan PM, Mitchnick M, Rosenberg ZF. Regulatory challenges in microbicide development. Science 2004; 304:19112. First citation in article