Women and HIV: is it really a biological difference?

Polly Clayden, HIV i-Base

The April meeting of the European AIDS Treatment Group’s ECAB (European Community Advisory Board) was the first dedicated exclusively to women’s HIV treatment issues.

Women are now the fastest-growing HIV-infected group in Europe and the US. In industrially developed nations, as a rule, women with HIV fare less well than men – they are diagnosed later, more likely to be admitted to emergency, less likely to use PCP prophylaxis, less likely to use HAART and more likely to die from AIDS. In addition to this they are less likely to enrol in clinical trials and therefore we know less about their specific treatment needs.

In 1993 the FDA altered the guidelines for drug study and research to include women in phase I, and early phase II trials. Prior to this they were generally excluded from these stages due to the possible risks of studying women of reproductive potential. However it is during these phases that clinical tolerance of a drug is assessed, it’s metabolism and short-term pharmacokinetics. As a result the availability of information about drug performance in women and the possibility of biological gender differences that would influence the relationship between dose and efficacy, tolerability or adverse reactions historically has been scarce.

Even in phase III there are many potential obstacles to women participating in clinical trials:

  • Hormonal variations
  • Availability of more convenient cohort ie. male
  • Conflicting domestic pressures and caring for other infected family members
  • Clinics originally founded to care for homosexual men, so on-site expertise pertinent to women (women-friendly clinics) may be limited
  • Physicians believe that women have less interest in participating in trials

The WIHS cohort

ÔIn the US for decades, like any other disease, if men and women are both affected the studies are in men’ was one of Dr Kathryn Anastos’s comments about the situation, although she added that, to be fair, with HIV in the developed world, historically this asymmetry merely reflected the nature of the epidemic.

The WIHS (Women’s Interagency HIV Study), cohort is the largest group of HIV-infected women studied to date with consortia from Brooklyn, the Bronx, Chicago, LA, Northern California and Washington DC. Her own clinic in the Bronx NYC, serves an area with the highest case of HIV-infected women in the developed world. She emphasised that WIHS is a Ôcohort study, not a clinical trial, cohort studies are viewed as rather …lower down in the hierarchy of evidence based medicine.’

The WIHS database contains data (as of September 99) on 2068 HIV-positive and 560 HIV-negative women over a median follow-up time of 4.0 years. In addition to their regular clinic visits and Ôcontinuous outcome assessment’, women attend a twice-yearly 2-4 hour interview and examination in which they are assessed by interview questionnaire on their behaviour, medical, obstetrical/gynaecological history, quality of their health services, demographics and psychosocial issues. They have a physical and gynaecological examination, a lipodystrophy examination, and assessment of their medications, participant samples are taken: blood (virologic, immunologic etc) and other (CVL, urine and saliva), which are stored in local and national repositories. Of the HIV+ women studied to date the baseline characteristics are: median age 36 years; 56% African-American, 23% Latina; median CD4 count (cells/mm3) 330; median viral load 22,000 copies/ml (NASBA assay) [1].

In her presentation from the WIHS cohort Dr Anastos addressed the burning question, ‘Are there biological differences …at work, or is the poorer prognosis in women more likely to be a socio-economical and socio-cultural problem?’ and looked at biological factors that could possibly influence clinical performance and could be integrated into studies. Besides gender the WIHS have also performed a number of analyses to look at the possible influence of race on outcome, but she explained ‘anything that addresses possible biological determinants is a political minefield, particularly in the US’.

Association of Race and Gender with HIV-1 RNA Levels and Immunologic Progression [2]

In this retrospective analysis WIHS collaborated with MACS (Multi-Center AIDS Cohort Study [3] – one of the seminal US studies that looked at disease progression in homosexual men), using stored blood samples. Although the two cohorts were separated by over a decade, which meant WIHS samples were designed to preserve the virus and MACS samples were preserved in the normal way, study populations were matched as closely as possible:

WIHS: 1268 women; HIV-positive 1994-1995; not reporting current use of ARV vs. MACS: 1604 men; HIV+ homosexual 1983-1984

Table1: Cohort characteristics

Median age WIHS MACS
36.6 33.5
White 18% 88%
African/American 57% 4%
Latina/o 22% 6%
Other 3% 1%
% with IDU history 43% 13%
% symptomatic 30% 7%

In the WIHS cohort HIV-RNA was measured (using NASBA) on one-year reposited samples to a lower limit of detection of 4000 or 400 cps/ml and statistical methods imputated. And in the MACS cohort b-DNA was measured on 10-year reposited samples to a lower limit of detection of detection of 500 copies and converted to RT-PCR equivalents.

After adjusting for confounding variables and varied by CD4 category, HIV-RNA levels were 31-50% lower in the WIHS samples than in the MACS. Stronger differences occurred when adjusting for gender differences in CD4 cell counts – among HIV-negative cohorts MACS CD4 cell counts were -71 cells lower than WIHS. Adjusting for the difference in CD4 counts did not alter the results of the analysis. Of note, consistent results were found when removing pregnant women (N=17) from the analysis, or including all WIHS women in the analysis. Participants of colour had significantly lower (-41%) HIV-RNA and those with a history of IDU, and less immunosupressed individuals had significantly lower HIV-1 RNA.

Table 2: Association with race and gender with HIV-1 RNA

WIHS women relative to MACS men
  • CD4 <200; +13%
  • 200≤ CD4 <350: – 32%
  • 350≤ CD4 <500: – 50%
  • 500< CD4: -46%
Whites relative to non-white
  • +41%

The mean annual CD4 count decline among WIHS and MACS participants was faster in women and faster in whites.

Table 3: CD4 decline

Gender/cohort Ethnicity N Mean CD4 slope/SE
WIHS of colour 359 -77/15.5
white 78 -115/20.7
MACS of colour 94 -31/16.9
white 780 -70/6.2
White vs. colour -41.3/19.0

The women (N=10) who ovulated showed statistically significant differences ie viral load went down during cycle

Table 4: viral load and menstrual cycle

Phase of cycle Viral load p-value
Early follicular 3.4 log
Periovulatory 3.26 log .07
Multi-luteal 3.20 log .03

Seeing these results, it may be important for physicians to record at what stage a woman is in her menstrual cycle when taking blood samples for viral load.

Although overall these differences are quite significant it is important to remember that in the general (HIV-negative) population CD4 counts are higher in women (but similar amongst African-Americans, Latinos and whites). Among WIHS and MACS participants CD4 decline was slower in those participants of colour compared to white participants. In WIHS/MACS comparisons (seroprevalent cohorts) women’s CD4 cell counts decline faster, but it is unclear whether in seroincident cohorts CD4 cell counts decline with parity to men or faster.

Clinical questions arising from these data

The later part of Dr Anastos presentation comprised of a series of questions as to the possible clinical significance of these findings and how it should impact on future trials or analyses, because, Ôof course there’s more that we misunderstand than we understandÉ’

If gender differences in HIV-RNA are real, do we need to re-examine prior analysis of untreated natural history? Most importantly – does progression to AIDS or death in different demographic groups occur at similar time for primary infection at a similar median CD4 count and HIV-RNA? And do these differences affect the response to treatment?

When treatment is introduced in different demographic groups, is HAART more or less effective immunologically, virologically and clinically? Does, for example lower HIV-RNA in women and people of colour result in more complete virological response to HAART and better clinical outcomes?

Why have gender or ethnic differences in disease manifestations not been properly documented? Is it because there are no differences, the differences are too slight to be revealed in small studies or are the studies themselves flawed – seroincident cohorts have not had enough women and seroprevalent cohorts have looked mostly at late stage disease ie. AIDS to death? And is stratification by CD4 a valid proxy for duration of HIV infection in groups of mixed gender? What is the prognostic value of these lab markers and should we make adjustments for differences in CD4 counts in seroprevalent cohorts of women and men?

It is possible even likely that the provision and potency of HAART obliterates the clinical differences (disease progression and death) possibly present in different demographic groups as may occur with age-related difference in disease progression. But without appropriate studies it is impossible to be certain, and closer following of women needs to be undertaken in all trial phases to understand possible differences in tolerability and side effects (such as metabolic changes and nevirapine rash).

If potency proves to overcome virological gender differences, and biological factors are of little significance to prognosis in different demographic groups, we must ensure that trial design, clinical care and social/adherence support, best addresses the disparity between different groups.


Clearly gender differences require more attention within studies. Issues of differences in side-effects and toxicities may require variations in management between men and women (and in different races). However, interpatient variabilities may be similar or even greater than those seen between genders or races and the potency of antiretroviral therapy may obliterate such differences. It would seem more likely that rather than biological it is conflicting psychosocial and socio-economic factors which may be at work in producing the differences observed. Poor prognosis and even side-effect profiles are likely related to social and medical (ie. adherence) support. All HIV-infected patients deserve access to state-of-the-art treatment and diagnostics, friendly clinics and good support systems. Clinics must ensure that service provision is accessible and of an equally excellent standard for all.


  1. Barkan, et al Epidemiology 1998;9:117-125
  2. Anastos, et al J AIDS May 2000-05-25
  3. Mellors, et al Ann Int Med 1997

Links to other websites are current at date of posting but not maintained.