March 2007 -

A Microbicide Conundrum

Minority AIDS Initiative Crisis

Prevention Takes to the Red Carpet at CROI

A Microbicide Conundrum

By David Gilden 
 

The January 31 announcement came like a bolt out of the blue: Researchers had prematurely terminated the two large efficacy trials for the microbicide cellulose sulfate. The trials were expected to show that vaginal applications of the microbicide, commercially known as Ushercell, would reduce HIV transmission during sex. But in one trial, women using cellulose sulfate paradoxically had higher rates of HIV acquisition than those using an inactive placebo gel. The Gates Foundation and USAID had poured $24 million into the trial program. Jeff Spieler, Chief of USAID’s Research, Technology and Utilization Division, Office of Population and Reproductive Health, reacted to the closure by saying, “I am surprised and disappointed by these findings given the pre-clinical effectiveness and safety profile of CS [cellulose sulfate] and its safety profile demonstrated in Phase I trials.”  
 
 

Cellulose Sulfate Building Block

Cellulose sulfate, a derivative of cotton, is a member of the group of promising microbicides technically known as “anionic polymers.” These compounds attach themselves to viral surfaces and disrupt virus-cell binding. They are active against HIV and other STDs, including such bacterial infections as gonorrhea and Chlamydia as well as herpes simplex virus. Some of the candidate microbicides, including cellulose sulfate, have contraceptive potential. They disrupt sperm membrane function, causing reduced mobility and egg penetration. Two other anionic polymers, Carraguard and PRO 2000, are also the subjects of advanced microbicide trials. 
 

The closed trials tested cellulose sulfate under particularly rigorous conditions. One trial, sponsored by CONRAD (associated with the Eastern Virginia Medical School) was recruiting 2,574 HIV-negative women in India, Benin, Uganda and South Africa. The women had to have had at least three male sex partners in the three months before enrollment and to have averaged at least three sex acts a week. Most, but not all, were commercial sex workers. Half of the trial participants applied a cellulose sulfate gel prior to having sex. The remaining participants used a placebo gel without the cellulose sulfate. Neither group knew which gel they had been given. All participants also received regular safe-sex counseling and supplies of condoms. The other trial was sponsored by Family Health International (FHI) and took place in Nigeria. This 2,160-person trial followed a similar design except that enrollees needed to have had two or more male sex partners in the previous three months. 
 
 
 
 

The CONRAD trial was the one that observed a higher rate of new HIV infections among the cellulose sulfate recipients. It is important to note that CONRAD recorded only 35 HIV transmissions in all. Both trials together were only half-enrolled when they closed. Data collection is not complete and CONRAD foresees no further release of information on its trial until next fall at the earliest. The two trials suffered from serious statistical weaknesses: HIV rates during the trials were lower than anticipated and pregnancy rates higher. The low rates of HIV transmission made it difficult to detect any effect of the microbicide unless the trials enrolled more women or followed them longer. Since pregnancy led to temporary exclusion from the trials, the resulting reductions in gel use further undercut the trials’ statistical power. . 
 

In light of the small number of HIV cases to work with, Luc Van Damme, the lead researcher for the CONRAD trial commented, “The recommendation to close was a big shock. We chose to err on the side of women. When we have the final data, it may show that cellulose sulfate was not unsafe. But even it kept going, the trial may not have been large enough to show efficacy.” 
 

HIV, Pregnancy and Trial Design

The cellulose sulfate trials were designed with the expectation that HIV transmission rates among placebo recipients would be at least four percent per year of observation. The rate instead was around 2%. At that low incidence, the trials would have had to expand enrollment several fold to demonstrate that cellulose sulfate was effective. Already, FHI had stopped enrollment at its Nigerian sites and was preparing new sites in South Africa. CONRAD had contingency plans to shift enrollment to southern Africa and increase total trial size. 
 

It has proven very difficult to anticipate the background HIV incidence rate in the populations enrolling in prevention trials. Ten years ago, large trials for nonoxynol-9 vaginal microbicides were conducted in similar populations as the present cellulose sulfate trials. The trial designs were also comparable: participants received similar safe-sex counseling, condom provision and STD treatment. Yet, the observed HIV transmission rates in the nonoxynol-9 trials were much higher: 7% to 10% – and up to 14% among one trial’s nonoxynol-9 recipients. (Nonoxynol-9 was the original big microbicide failure – see below.) 
 

High community HIV levels do not mean that there are high current rates of transmission. An HIV epidemic reaches a mature stage when most people with behavioral and biologic risks for HIV are already infected. 
 

It has proved very challenging to identify the best populations for testing microbicides. In the words of Lori Heise, director of the Global Campaign for Microbicides, “High HIV incidence occurs in newly exposed populations, where the epidemic is on the cusp. Vaccines and PrEP [pre-exposure prophylaxis] trials can recruit across settings. They can recruit men, including those who have sex with other men, and IV drug users. The current microbicide trials recruit only among women who are at risk for HIV through vaginal sex. They need women who are at high risk but consistently unable to use condoms. This is a very narrow group – even sex workers can use condoms well.”  
 

Pregnancy has proved another stumbling block. The FHI trial noted a pregnancy rate of 30% per year of observation. It also found that the women in its trial said they used their gel, whether placebo or active, for 80% of all sex acts. Reported condom use increased from 60% to 90% during the trial. A previous trial with monogamous couples in California found that cellulose sulfate compared favorably as a contraceptive with nonoxynol-9, the only vaginal agent marketed in the US. With very frequent sex, though, high pregnancy rates over fairly short periods could occur even with an average 80% usage, especially if there was a tendency to dispense with both condoms and gel on the same occasion. 
 

Pregnant women were excluded from receiving their assigned trial product because reproductive toxicity studies have yet to be conducted on cellulose sulfate. This lapse meant that women did not have the microbicide when they appear to be particularly vulnerable to HIV. Despite their suspension, the trial counted them as if they were still part of their assigned trial cohort because of the researchers’ planned “intent-to-treat” trial evaluation. “Intent-to-treat” is usually a valuable way to analyze Medical products. It takes into account the fact that, in the real world, many people stop using products due to side effects or inconvenience. This trial, though, involved involuntary discontinuations that would not have occurred in the real world. Including the pregnant women in the trial results further weakened the trials’ ability to show a benefit from cellulose sulfate. 
 
 
 

Microbicides’ Different Points of Attack  
 
 
 
 

Source: Shattock RJ and Moore JP. Inhibiting sexual transmission of HIV-1 infection. Nature Reviews Microbiology. October 2003. 
 

The Necessary Preliminaries to Advanced Human Trials

The large nonoxynol-9 trials suffered from neither of these issues. There, the problem was that heavy use of nonoxynol-9 increased the risk of acquiring HIV because it irritated the vaginal lining when applied. That increased risk was only apparent in trial participants who on average used nonoxynol-9 more than 3.5 times per day. 
 

The nonoxynol-9 results led to a shake-up in the way microbicides are developed. Laboratory reports indicating nonoxynol-9’s irritating effects became available only after the human trials had commenced. Since then, candidate microbicides have been carefully tested in the lab and in small human studies for signs that they disrupt vaginal surfaces. 
 

Nonoxynol-9 is a surfactant that dissolves viral and cell membranes. Although cellulose sulfate has no such effect, it is a potent inhibitor of blood clotting. Previous cellulose sulfate trials did not notice any signs of vaginal irritation or non-menstrual bleeding. All but two of these trials were small and short, lasting at most four weeks. The major exception is the six-month, 200-person contraception trial conducted in monogamous California couples.