Gender, Ethnicity, and Clinical Trials
Since the first reported AIDS case in 1981, hundreds of thousands of people have participated in research related to this epidemic. From trials of new drugs to surveys and questionnaires, the thirst for knowledge has driven a massive effort to recruit people with HIV into such studies. During the early years, attention focused on the community that appeared hardest hit -- predominately white gay men. In retrospect, it is clear that gay men of color, particularly African-Americans, were also heavily impacted, and many remember that the disease was affecting injection drug users and their partners as early as the late seventies, (and of course, women were also there from day one). The fact that HIV traverses demographic boundaries has now become clear, but clinical trial participation is still skewed so that white gay men are over-represented compared to other affected populations. That raises important questions about how complete our knowledge of HIV really is, questions to which recent -- and not so recent -- studies have provided some disturbing answers. Clearly, there are issues of fairness and potential racial, sexual and socioeconomic discrimination when it comes to representation in clinical trials. There are also potentially important differences in genetics and physiology (how our bodies work) that may have profound implications for the health of people living with HIV.
Viral Load, T-cells, and Disease Progression
Perhaps the most glaring example of a study conducted solely in one population (and then applied to everyone else) is John Mellor's study linking viral load levels and T-cell counts to the rate of disease progression. The current US HIV treatment guidelines' recommendations on starting drug therapy are based heavily on this study, which was conducted almost exclusively in white gay men. In fact, the ethnic breakdown of study participants was not even provided in the widely cited publication by Mellors in the Annals of Internal Medicine in 1997. The study population was simply identified as "1,604 men infected with HIV." The reasons for this bias may not have been intentionally racist. The study was conducted in what is called the MACS (Multi-AIDS Center Study) cohort, which began in 1984 when the full demographic spread of HIV infection was not appreciated. The MACS' focus on gay men in major cities was understandable and continues to provide an invaluable source of regularly donated blood samples for testing and analysis. However, similar studies focusing on women (the Women's Interagency HIV Study, or WIHS) did not start enrolling participants until 1994, and only after considerable community advocacy efforts.
Illustrating that this is more than just an academic problem, a recent analysis by Kathryn Anastos, MD directly compared viral load and T-cell count data from MACS and WIHS. Ananstos found that viral load levels in women averaged 32% to 50% lower than those in men. Other differences uncovered by Anastos were that white participants averaged viral loads that were 42% higher than those of their counterparts of color. Study participants with a history of injection drug use had average viral loads 23% lower than those with other risk factors for HIV transmission.
When it came to T-cell counts, MACS and WIHS comparisons turned up even more disturbing surprises. Overall, women experienced more significant declines in T-cell counts each year than men, losing an average of 45 more T-cells from the blood annually. Among women, however, the T-cell decline was more rapid in white women than in women of color, averaging 115 cells per year versus 77. The difference was 70 cells per year in white men versus a 31 cell per year decrease in men of color. The same trend held true for participants of color overall, who on an average lost 41 fewer T-cells a year than their white counterparts.
Although Anastos tried to eliminate any factors that might bias the results, it is hard to draw firm conclusions from these preliminary data when it comes to the speed of disease progression among differing groups. It is noteworthy, however, that Anastos' data support a military study from 1994 that noted potentially slower disease progression among African-Americans. A 1992 Johns Hopkins University study that focused solely on gay men also uncovered slower disease progression rates among participants of color. One finding that is slightly askew from these results was reported in a 1997 study published in the Journal of Infectious Diseases. Among 545 men in the military, researchers uncovered no significant differences in viral load levels between African-Americans, Hispanics, and Caucasions.
It is reasonable to wonder if lower viral load in women translates into altered disease progression. Unfortunately, a widely publicized study by Johns Hopkins investigators reported that progression appeared to occur at the same rate in women despite lower viral load levels. This once again raises the question of whether basing treatment guidelines on the MACS study may have been premature. Even more worrying, a 1994 study of male and female intravenous drug users with HIV concluded: "These data suggest a greater risk for death for women compared to men and IDUs, despite lesser degrees of immunosuppression." One firm conclusion can be drawn from these data: we need to know more.
Drug Metabolism and Side Effects
Several studies have shown that there can be gender and ethnicity-related differences in the way drugs are processed, or metabolized, in the body. Yet this remains an understudied area, with extraordinarily little data available on blood levels of HIV drugs in women compared to men or among various ethnic groups. The same problem applies to variations in the risk of drug side effects.
In 1995, researchers from Holland reported that women metabolized AZT (Retrovir) 42% slower than men. In a large US trial that included ddI (Videx), women were more likely to require dose reductions of the drug but still seemed to get the same health benefits. This suggest slower metabolism of ddI. More dramatically, a study of the NNRTI delavirdine (Rescriptor) found that drug blood levels were 1.8 times higher in women than in men. Another recent report noted that women are eight times more likely than men to develop a severe rash from the NNRTI nevirapine (Viramune). This suggests that there may be gender differences in the metabolism of this NNRTI as well, although the labeling claims otherwise.
In terms of side effects, a study of the protease inhibitor ritonavir (Norvir) reported more nausea, vomiting, depression and fatigue among women than men. And recent studies of lipodystrophy have found that women are more likely to have central fat accumulation (initially dubbed "protease belly"), whereas men seem more likely to have higher levels of fat and lower levels of HDL, a blood fat that can lead to cardiovascular disease.
Although the underlying reasons are almost certainly different, there may be similar issues related to ethnicity. One example is a recent finding that people of Asian decent appear to be at greater risk for the side effect of rash from the NNRTI's (Viramune, Rescriptor, and Sustiva). Hispanics have also been reported to have a slightly elevated risk of NNRTI related rash compared to African-Americans and Caucasions.
African-Americans have long experienced a relatively unique toxicity of AZT -- discoloration of the fingernails. Often dismissed by doctors as a "cosmetic" problem, there is as yet no convincing explanation of the ethnic specificity of this side effect. The problem reportedly occurs in around 67-81% of African-Americans compared to 20-30% of whites or Hispanics. Whatever the cause, it is clear that the unequal incidence of side effects among ethnic groups is not just hypothetical. Conversely, African Americans appear to have a slightly lower risk of allergy to sulfa drugs such as Bactrim for pneumocystis carinii pneumonia (PCP). Since sulfa allergy has been linked to an eightfold increased risk of NNRTI rash, African Americans may also be at risk for this latter problem, although this has not yet been reported. This may be due to too few African-American participants in NNRTI trials. This brings us to the next issue -- how do you actually spot examples of gender and ethnicity-specific outcomes?
It's often said that there are lies, damn lies, and statistics. Grappling with the statistical shenanigans employed in clinical trials often reinforces this impression. However, there are statistical issues that bear direct relation to our level of knowledge about gender and ethnic differences in HIV. The most important is what statisticians call "power."
If a study of an HIV treatment only includes a small number of women or is not ethnically diverse, then it may be impossible to work out whether different outcomes among groups are meaningful or just due to chance alone. The more people from each group that are included, the more confident we can be that divergent results actually mean something.
Researchers represent the degree of certainty in a result as what's called a "p value." A p value of less than 0.05 indicates that the results of the study are statistically accurate and did not occur by chance. Achieving a significant p value is heavily dependent on the number of people in the groups that are being compared. If a hypothetical NNRTI study includes 100 Caucasian participants and twenty get a rash, while only ten African-Americans participate and only one gets a rash, it might look like African-Americans are less at risk for this side effect (and it might even be true). However, the small number of African American participants would make the p value greater than 0.05, indicating that the result could have been due to chance alone. When we read research studies that say the results were "significant," that almost always means that the p value was less than 0.05 and the researchers were confident that the outcome was not due to chance. Although this all sounds terribly nerdy, it's one of the fundamental reasons why diverse representation in clinical trials is important.
Obstacles to Trial Participation
While it's clear that broad representation in clinical trials is essential, there are many reasons why this has not occurred. For example, the infamous Tuskegee experiment exposed a medical establishment steeped in racism. Bill Clinton's apology not withstanding, deep-rooted mistrust of the medical and research system persists particularly in communities of color.
Other factors also come into play. Socioeconomic disparities may make travel to a clinical trial site harder for some than others. Many academic research centers are located far from African American and Hispanic neighborhoods, emphasizing the need for community-based research sites.
The ability of women to participate in research can be affected by many factors. Women often bear the burden of maintaining a household and have time constraints that make regular visits to a trial site difficult. For women with children, the only option for participation may be going to one of the few trial sites that offer childcare. Each clinical trial has specific criteria (called inclusion and exclusion criteria) for participation, and burdensome requirements for multiple forms of birth control are commonly required of all female participants.
Inclusion and exclusion criteria can also discriminate against other groups. For instance, most studies exclude persons with ALT (a liver enzyme) levels greater than four or five times the upper limit of normal. Since many current and former injection drug users have chronic hepatitis C which can cause elevated ALT levels, they are effectively excluded from these trials. Although safety concerns are certainly one reason for this exclusion, it is disturbing because we know that protease inhibitors in particular are metabolized through what's called the CYP450 pathway in the liver. Clinical trials also sometimes use current drug and alcohol users as an exclusion criteria based on preconceived notions that substance users are unreliable in showing up for scheduled visits. The result is little safety data on the use of anti-HIV drugs in a large portion of the HIV positive population.
Without community input, trial designs can also be off-putting to potential participants even if they meet the inclusion criteria. Ensuring that no one in the trial is receiving substandard care is crucial. Continued access to drugs after a trial is completed can also be important. Lack of health insurance is widespread, and African Americans, women, and Hispanics are over-represented among those lacking coverage. In this regard, there has recently been a disturbing trend in some pharmaceutical-company sponsored studies to charge people (or their insurance carrier, if they are lucky enough to have one) for all or some of the drugs they are taking. In a recent study of the experimental protease inhibitor ABT-378, participants were required to pay for the other study drugs (such as d4T and 3TC). It is hard to believe that the manufacturer of ABT-378, Abbott Pharmaceuticals, could not afford to provide these treatments for the participants.
Language barriers to trial participation can also be a problem. Joining a study requires signing an informed consent form, acknowledging the potential risks and benefits of participation. It's the right of potential trial enrollees to have the consent form provided in their first language, but in reality this is not always guaranteed.
While this represents just some of the issues to consider when it comes to gender, ethnicity and clinical trials, it illustrates that more than just inclusiveness and fair play is at stake. There are scientific issues of genetics and physiology that need to be kept in mind, and broad representation in well-designed trials is key to obtaining information that can be used by everyone with HIV as they make critical decisions about their healthcare and survival.
The Power of Community
The communities ability to make observations and raise red flags should not be underestimated. Soon after the first protease inhibitors were approved in late 1995 and early 1996, women at community forums were complaining of disturbances in their menstrual cycle that occurred after starting the drugs. Not until 1999 was a letter finally published in a medical journal describing just this problem in women taking the protease inhibitor Norvir (ritonavir). Likewise, many community-based organizations learned of the problems of fat redistribution and breast enlargement long before any reports appeared in the medical literature. If protease inhibitor studies had been better designed, and more attentive to the risk of such gender-specific toxicities, an information gap of several years might have been bridged.
Back to the Women Alive Summer 2000 Contents Page.
This article was provided by Women Alive. It is a part of the publication Women Alive Newsletter.