University of California, San Francisco Logo

University of California, San Francisco | About UCSF | Search UCSF | UCSF Medical Center

Home > Global Health Literature Digest > Lack of Effectiveness
Lack of effectiveness of cellulose sulfate gel for the prevention of vaginal HIV transmission
Global Health Sciences Literature Digest
Published December 16, 2008
Journal Article

Van Damme L, Govinden R, Mirembe, et al. Lack of effectiveness of cellulose sulfate gel for the prevention of vaginal HIV transmission. N Engl J Med 2008;359:463-72.

In Context

Women account for over 60% of the 22 million people living with HIV in sub-Saharan Africa.(1) Although male condoms and circumcision are effective methods to prevent HIV transmission, these methods remain under the control of men. Cellulose sulfate (CS), a vaginal microbicide, has demonstrated in vitro inhibitory action against HIV and prevents simian HIV infection in macaques.(2) Clinical studies among women found CS to be safe and well tolerated.(3)


To measure the efficacy of CS in reducing HIV transmission


Clinical settings in Cotonou, Benin; Chennai, India; Durban, South Africa; Kampala, Uganda; and Mudhol and Jhamkandi, Karnataka, India

Study Design

Randomized, double-blind, placebo-controlled trial


A total of 1,428 HIV-negative women


New infections of HIV, gonococcal or Chlamydia infections


Participants were randomly assigned to receive 6% CS gel or placebo gel in a 1:1 ratio. Women were asked to insert the gel 1 hour prior to intercourse using a 3.5-ml single-use applicator for every act of intercourse and to do this over a 12-month period. Women who became pregnant during the study remained as participants but discontinued the use of the product until the end of the pregnancy.

Follow-up occurred at 1, 3, 6, 9, and 12 months. HIV antibody testing was performed at each follow-up visit or at any visit where the product was discontinued. Polymerase chain reaction (PCR) tests from baseline were used to test blood from any woman who tested HIV-positive at 3 months. Cervical specimens were collected quarterly and tested for gonorrhea and Chlamydia infections using strand displacement amplification (SDA). Laboratory staff was trained at the central laboratory (Institute of Tropical Medicine, Antwerp, Belgium). All positive HIV and SDA tests and a 10% sample of SDA-negative tests were verified at the central laboratory.

A sample size of 2,574 participants would provide 80% power to detect the effectiveness of CS using a one-sided test and a significance level of 0.025. A total of 66 infections were anticipated. These calculations were based on a 12-month cumulative incidence of 4% in the placebo group and a 50% reduction in infection in the CS group. Complete follow-up was estimated at 80%.

Interim analysis to evaluated effectiveness and harm was undertaken once 50% of the estimated number of HIV infections occurred. The Lan-DeMets spending function with O'Brien-Fleming boundaries was to be used to preserve the type I error of concluding effectiveness with the use of a one-sided alpha level of 0.025. A one-sided P value of 0.10 in the direction of harm was set as the point for the data monitoring board to provide guidance regarding the study's continuation.

Placebo and CS gels were divided into three groups of differing colors to ensure blinding. Randomization was done using a validated SAS program in equal-sized groups and conducted by a person who was not involved with the study.

An intention-to-treat analysis was used. The effectiveness was measured using the log-rank test, stratified by study site. The proportional hazards model was used to calculate the hazards ratios (HRs) and 95% confidence limits. Cumulative probabilities of infection were done using the Kaplan-Meier method. Incidence rates were calculated as the number of events divided by person-years of follow-up. The date of HIV infection was estimated as the mid-point between the first positive and most recent negative test. Women without infection were censored at 380 days; for women not completing the study, the censor date was the date of their last test.


Enrollment began on July 30, 2005 and ended on January 25, 2007. Of the 2,985 women screened, 1,428 enrolled and were randomly assigned to the study group. Three women who had positive HIV tests at 3 months and positive baseline PCR tests were excluded from analysis, and another 27 women did not have any HIV test results. Of the remaining 1,398 women, 706 received CS and 692 received placebo. There were 141 women who were lost to follow-up and 24 who discontinued before the study was stopped.

Baseline characteristics and reported rates of condom and gel use did not differ between the two groups. Reported adherence to gel was 87.4% in the CS group and 86.7% in the placebo gel group. Use of gel when condoms were not used was lower (45.9% in the CS group and 45.7% in the placebo group). Interim analysis was conducted on data collected through December 18, 2006. At that time, there were 35 infections, 24 of which were in the CS group (HR 2.23 P=0.02). Based on these results, the study was halted. There were an additional six HIV infections that were not included in the interim analysis. With these additional cases, the incidence rate in the CS group was 5.29 per 100 person-years and 3.33 per 100 person-years in the placebo group. The relative hazard was 1.61 (95% CL: 0.86-3.01, P=0.13).

Rates of gonorrhea were similar in the two groups but rates of Chlamydia infection were lower in the CS group. Pregnancy rates also did not differ (incidence of 21.8 in the CS group and 23.1 in the placebo group, P=0.74).


There is no evidence of the effectiveness of CS gel in the prevention of male-to-female HIV transmission. CS may increase the risk of HIV acquisition.

Quality Rating

All of the criteria for a well-designed and conducted trial were met.

Programmatic Implications

Although this study did not demonstrate the effectiveness of CS in prevention of HIV, its role in increasing the risk of transmission is less clear. At the time of the interim analysis, the risk of infection in the CS group was statistically significantly higher than that in the placebo group, but, with the addition of a few previously unreported infections, the difference was no longer significant. Given the low rate of gel use in the absence of condom use, it is difficult to determine the independent effect of gel. Poor adherence in other trials also may have prevented detecting a difference between treatment arms.(4,5) Although an effective transmission prevention tool that is under the control of women remains elusive, continued efforts to develop such a method should be made.


  1. 2007 AIDS Epidemic Update. Geneva: Joint United Nations Programme on HIV/AIDS.
  2. Saifuddin M, Doncel GF, Tsai L, Gettie A, Bhom R, Cheng-Mayer C. Intravaginal administration of 6% cellulose sulfate gel prevented systemic infection in rhesus macaques in a multiple dose R5/X4 SHIV vaginal challenge model. Presented at the Late Breaker Session, Microbicides 2008 Conference, Delhi, India, February 24-27, 2008.
  3. Mauck C, Weiner DH, Ballagh S, et al. Single and multiple exposure tolerance study of cellulose sulfate gel: a Phase I safety and colposcopy study. Contraception 2001;64:383-91.
  4. Johansson E. Results of phase III Carraguard trial. Presented at the Microbicides 2008 Conference, Delhi, India, February 24-27, 2008.
  5. Watson-Jones D, Weiss HA, Rusizoka M, et al. Effect of herpes simplex suppression on incidence of HIV among women in Tanzania. N Engl J Med 2008;358:1560-71.