Secondary Logo

Journal Logo

Clinical Science

Brief Report: Sex Differences in Outcomes for Individuals Presenting for Third-Line Antiretroviral Therapy

Godfrey, Catherine MD, FRACPa; Hughes, Michael D. PhDb; Ritz, Justin MSb; Coelho, Lara MDc; Gross, Robert MD, MSCEd; Salata, Robert MDe; Mngqibisa, Rosie MB ChBf; Wallis, Carole L. PhD, MScg; Mumbi, Makanga. E. MDh; Matoga, Mitch MDi; Poongulali, Selvamuthu MBChBj; Van Schalkwyk, Marije MDk; Hogg, Evelyn BAl; Fletcher, Courtney V. PharmDm; Grinsztejn, Beatriz MD, PhDc; Collier, Ann C. MDn, on behalf of the A5288 team

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: June 1, 2020 - Volume 84 - Issue 2 - p 203-207
doi: 10.1097/QAI.0000000000002324
  • Free



Antiretroviral treatment (ART) options for people living with HIV (PLH) are constrained in low- and middle-income countries (LMIC), and fewer drugs are available to individuals experiencing treatment failure. Typically, there is one recommended first-line regimen, either a non-NRTI (NNRTI)-based regimen, or more recently a dolutegravir (DTG)-based regimen, with 2 NRTIs. Second-line regimens have used a boosted protease inhibitor (PI), either lopinavir/ritonavir or atazanavir/ritonavir, with modifications of the other components of the regimen. Modifications can be made but are dependent on availability of the desired drugs or drug combinations. Third-line regimens may be unavailable or difficult to access, sometimes requiring application to central authorities such as ministries of health. Given these circumstances, identifying and addressing barriers to adherence allows for targeted interventions that may enhance treatment success.

A5288 was an open-label phase IV, prospective interventional strategy study at 19 urban sites in 10 LMIC in Africa (Kenya, Malawi, South Africa, Uganda, and Zimbabwe), Latin America (Brazil, Haiti, and Peru), and Asia (India and Thailand) evaluating treatment options for individuals experiencing confirmed virological failure (VF) on their PI-based second-line regimen.1We report here results from a secondary analysis of differences in outcomes by sex in this diverse study population.


Real-time HIV drug resistance results, treatment history, and historical drug resistance (when available) results were used to assign participants to one of 4 treatment cohorts. Participants with no LPV/r resistance and who were susceptible to at least one NRTI were assigned to cohort A. They continued their PI backbone, but NRTIs could be modified. Participants with LPV/r resistance and/or resistance to NRTIs were assigned to cohorts B or C, which prescribed regimens including darunavir/ritonavir and raltegravir, with either etravirine or optimized NRTIs. In cohort D, representing individuals with the most complex resistance profile, the best regimen was constructed using any study-provided or locally available agents. Ritonavir boosting was included in every cohort. This analysis combines cohorts B, C, and D (BCD) as all 3 involved newer ART regimens and numbers were too small in individual cohorts for meaningful analysis. Virological suppression was defined as HIV-1 RNA ≤200 c/mL at week 48, and VF was defined as confirmed HIV-1 RNA≥1000 c/mL at/after week 24. Toxicity and adherence interventions were managed as per the site's standard of care. Sex differences by cohort group (A or BCD) were evaluated using χ2 tests, in participant characteristics using the Wilcoxon rank-sum test stratified by cohort group, and in outcomes using Cochran–Mantel–Haenszel Tests, logistic regression or proportional hazards models adjusted for cohort group. Differences were further evaluated in models adjusting for cohort group, country, baseline age, weight, and HIV-1 RNA and CD4 count, characteristics considered to have high potential to impact outcomes.


The demographics of the A5288 participants are presented elsewhere.1 Women represented 47% of the total study population (258/545). They were more likely than men to have received nevirapine(NVP) (70% vs. 60%, P = 0.003) largely driven by the fact that more women than men had previous NVP use in cohort A (69% vs. 47%). Women were less likely to have received efavirenz (49% vs. 64%, P < 0.001) before study entry. More women were assigned to cohort A (62% vs. 44%) and fewer to cohorts BCD (38% vs. 56%) (P < 0.001). Women were slightly younger than men (median 39 vs. 43 years, P < 0.001) with slightly lower weight (median 56.3 vs. 59.0 kg, P = 0.020), but had similar CD4 counts (median 172 vs. 179 cells/mm3, P = 0.84) and HIV-1 RNA levels (median 4.3 vs. 4.5 log10 copies/mL, P = 0.38).

In the analyses adjusted for cohort group, fewer women achieved virological suppression at week 48 {odds ratio [OR] [95% confidence interval (CI)] 0.64 [0.43 to 0.96], P = 0.030}, and more women than men experienced confirmed VF during the median follow-up of 72 weeks (hazard ratio [HR] (95% CI) 1.48 (1.08 to 2.03), P = 0.014) (Table 1). In the multivariable analyses, these associations were attenuated {[HIV-1 RNA ≤200 copies/mL at week 48: aOR (95% CI) 0.87 (0.54 to 1.39), P = 0.56]; [VF: aHR (95% CI) 1.37 (0.97 to 1.93), P = 0.075]}.

Results From Models Assessing Each of the Outcomes and Groups Specified

Table 1 also displays differences in clinical outcomes. Women were more likely than men to have grade ≥3 signs and symptoms (20% vs. 9%), an effect that was seen in both cohort A and cohorts BCD (HR (95% CI) 1.87 (1.17 to 2.99), P = 0.009). This effect persisted in the multivariable analysis [aHR (95% CI) 1.67 (1.01 to 2.74), P = 0.044]. Sex differences in grade 3+ s/s in cohort A were seen among participants on LPV (26% for women vs. 16% for men) and ATV (19% for women vs. 12% for men). There were no significant differences by sex in multivariable analyses of serious adverse events, grade ≥3 diagnoses, or grade ≥3 laboratory abnormalities.

Participants experiencing a grade ≥3 sign/symptom during the first 48 weeks of follow-up had a lower probability of achieving virological suppression at week 48 regardless of cohort group. Specifically, among the 51 participants who experienced a grade ≥3 sign/symptom during the first 48 weeks of follow-up, only 15 (29%) achieved HIV-1 RNA ≤200 c/mL at week 48, whereas among the 494 participants who did not have a grade ≥3 sign/symptom during this time, 334 (68%) achieved suppression ≤200 c/mL at week 48 (P < 0.001). This association was seen in both women (15% vs. 44%) and men (25% vs. 51%) in cohort A and in both women (50% vs. 86%) and men (80% vs. 90%) in cohorts BCD. Experiencing a grade ≥3 sign/symptom, however, only partially explained the sex difference in virological suppression: The OR for sex moved toward 1.00 changing from 0.64 to 0.69 with adjustment for grade ≥3 sign/symptom (and the adjusted OR shown in Table 1 moved from 0.87 to 0.93).

Table 2 displays a descriptive summary of resistance outcomes. The proportion of men and women who experienced VF with mutations was similar in both cohort groups: 18% of women experienced VF in cohort A with a new resistance associated mutation (vs. 16% of men), whereas 2% of women in cohorts BCD had VF with a new resistance mutation (vs. 4% of the men). The difference between sexes in the overall VF rate therefore primarily reflected the fact that women were more likely than men to have VF without a new resistance mutation (38% of women vs. 25% of men in cohort A; 6% of women vs. 3% of men in cohorts BCD).

Summary of Confirmed Virologic Failures and Changes in Mutation Patterns, by Cohort Group and Sex


We noted important differences between men and women with respect to signs and symptoms in the 2 cohort groups. There was no difference between the sexes when we looked at more easily quantifiable metrics such as abnormal laboratory values, serious adverse events, and newly established diagnoses. More women than men had a suboptimal virological response and experienced VF, primarily reflecting more women than men experiencing VF without new drug resistance mutations. Of note, all participants in this study experienced VF twice before study entry. Data from treatment programs in LMIC suggest that women are more likely to access therapy and, in contrast to the findings of our study, achieve virological suppression on their first regimens.2 Importantly, in our study, more women than men had an NVP-based first-line regimen, a regimen that is more likely to lead to VF than an efavirenz-based regimen.3 Drivers of regimen failure are complex and include both biological and personal factors. For example, historical use of single-dose NVP for the prevention of vertical transmission has led to significant NNRTI resistance in women; in 2019, the WHO reported levels of pretreatment drug resistance nearly twice as high among women as among men.4

We do not have good data on access to PI-based therapy in LMIC. The excess in pretreatment resistance, documented above, may mean that more women than men require second-line regimens. In addition, individuals experiencing VF of a second regimen may represent a specific subset of people that have problems with adherence and may be different from those initiating ART. Importantly, we do not have data on stigma or mental health issues that may affect women's adherence differently than men's.

Despite these limitations, we believe that there is an important biological basis for treatment failure in women. Specifically, the contrast in patterns for women and men between “signs and symptoms” (subjective measure of tolerability) and “diagnoses” and “laboratory abnormal values” (objective measures of tolerability) is important because it may represent sex differences in the experience of treatment. It is possible that clinicians are less likely to make changes in the face of subjective tolerability especially when regimen choices are limited.

Our study raises the possibility that tolerability issues may be driving incomplete adherence to ART in women referred for third-line therapy, with important clinical consequences. In our study, women who continued their PI-based regimens experienced more grade ≥3 signs and symptoms than men, but this was not the case for grade ≥3 diagnoses or laboratory values. Moreover, women experienced VF more frequently than men, and more women had a suboptimal virological response at week 48 than men. More women than men experienced VF without new resistance mutations which strongly suggests nonadherence to PI-based regimens in this setting.5

Differences in tolerability between the sexes may also be driving the findings in cohort A. More women than men entered cohort A (which had the least resistance at screening) after having experienced treatment failure with both an NNRTI- and a PI-based regimen. Consistent with current standard of care in LMIC, where treatment options are limited, our study required the continuation of the PI-based regimen with an emphasis on adherence counseling and support. The individuals in cohort A remained on their second-line ART regimen with only minor modifications. Cohort A was likely enriched for individuals who had past incomplete adherence, resistance, or tolerability issues. We are unable to determine which of these factors or combination of factors was responsible for the increased number of women in cohort A, a limitation of our study.

Tolerability issues that might lead to treatment discontinuation have been reported in a variety of settings. In a cross-sectional survey of self-reported nonadherence, men were more likely to report forgetting to take their medications, while women were more likely to report adverse events.6 Important sex differences in ART tolerability have been reported, and a higher rate of adverse events that lead to treatment discontinuation in women has been documented with nearly all ART classes. NVP has a well-described hepatic toxicity in women,7 leading to recommendations that its use be limited in women with CD4 counts >250 cells/mm3. Women are more likely than men to develop a rash when exposed to efavirenz.8 Nausea and vomiting have featured prominently in several reports relating to ritonavir use.9,10 The women in our study commonly experienced gastrointestinal complaints, consistent with known toxicities of ritonavir.

Sex differences in drug metabolism have been described which may lead to differences in tolerability. Ritonavir clearance is noted to be slower in women, leading to concentrations approximately a third higher than in men after adjustment for body weight.11 A large prospective clinical trial identified higher concentrations and a slower clearance of atazanavir (given with ritonavir) in women than in men.12 The authors suggested that this might have been responsible for greater rates of VF in women than men; ritonavir concentrations in that study were not measured.12 Sex differences have been reported in the metabolism of other drugs as well: Intracellular concentrations of the active triphosphate moiety of both zidovudine and lamivudine may be higher in women than men, and saquinavir (when given with ritonavir) concentrations were shown to be higher in women.13,14 It has been postulated that differences in metabolism are related to drug exposure with plasma drug concentrations in women being higher because of women's smaller size. In our study, adjustment for weight did not affect the findings, but it may be that body composition played an important role. Additionally or alternatively, the basis for higher ritonavir concentrations in women may be metabolic and/or genetic. Ritonavir is a substrate and inhibitor of cytochrome P-450 3A and P-glycoprotein, as are atazanavir and saquinavir. The oral clearance of verapamil, also a mixed CYP3A and P-glycoprotein substrate, was shown to be slower in women than in men.15 Sex-specific differences in expression/function of CYP3A and P-glycoprotein, therefore, could provide a pharmacokinetic basis for higher ritonavir concentrations in women and thereby, a pharmacodynamic basis for those higher concentrations to result in poorer tolerance in women.

Women comprise more than half of all those with HIV infection, and most women with HIV live in LMIC. Therapy for HIV is lifelong, and adherence to life-long therapy is a challenge for all PLH. Interventions designed to improve adherence may need to be tailored to specific populations and stages during the life span. Even with the planned widespread use of dolutegravir, it is likely that boosted PIs may be required at some stages of life.

Our analysis identified a potentially modifiable biological determinant of incomplete adherence in women. If tolerability is indeed a more important driver of incomplete adherence for women, interventions designed to improve tolerability may improve outcomes and reduce treatment failure. In our study, it is possible that lower grade signs and symptoms may have been responsible for differences in adherence and differences in outcomes between men and women. We could not fully assess the effect of intolerance in PLH assigned to cohort A because we tracked more severe adverse events and not lower grade adverse events. Future studies should proactively monitor low-level adverse events that could be addressed by more aggressive management, including treatment switches, to prevent failure due to incomplete adherence from poor tolerability. Personalized or targeted drug selections and dose modifications may also be required. Certainly, simpler and better tolerated regimens such as those that include dolutegravir could be helpful.

One of the key strengths of our study was the large proportion of women across several countries which allowed us to look critically at this information. This was a planned secondary analysis, but the study was not specifically designed to evaluate differences between men and women. In addition, the cohorts were not randomized, a potential limitation; however, the finding of “subjective intolerance” was noted across all cohorts. As such, we may have identified a potentially modifiable determinant of nonadherence in women. Understanding the specific reasons for incomplete adherence will allow targeted interventions aiming to improve adherence and HIV-related outcomes. Attention to the sex differences in the experience with ART will allow for prompt identification of emerging toxicities and mitigation of those toxicities.


1. Grinsztejn B, Hughes MD, Ritz J, et al. Third-line antiretroviral therapy in low-income and middle-income countries (ACTG A5288): a prospective strategy study. Lancet HIV. 2019;6:e588–e600.
2. PHIA. Secondary. 2019. Available at: Accessed June 27, 2019.
3. Pillay P, Ford N, Shubber Z, et al. Outcomes for efavirenz versus nevirapine-containing regimens for treatment of HIV-1 infection: a systematic review and meta-analysis. PLoS One. 2013;8:e68995–e95.
4. WHO. HIV Drug Resistance Report 2019. Geneva, Switzerland; 2019.
5. Levison JH, Orrell C, Gallien S, et al. Virologic failure of protease inhibitor-based second-line antiretroviral therapy without resistance in a large HIV treatment program in South Africa. PLoS One. 2012;7:e32144–e44.
6. Thunander Sundbom L, Bingefors K. Women and men report different behaviours in, and reasons for medication non-adherence: a nationwide Swedish survey. Pharm Pract. 2012;10:207–221.
7. Sundaram M, Srinivas CN, Solomon S, et al. Does gender and nevirapine (NVP) influence abnormal liver functions in HIV disease? J Infect. 2009;58:255–257.
8. Mazhude C, Jones S, Murad S, et al. Female sex but not ethnicity is a strong predictor of non-nucleoside reverse transcriptase inhibitor-induced rash. AIDS. 2002;16:1566–1568.
9. Squires KE, Johnson M, Yang R, et al. Comparative gender analysis of the efficacy and safety of atazanavir/ritonavir and lopinavir/ritonavir at 96 weeks in the CASTLE study. J Antimicrob Chemother. 2011;66:363–370.
10. Currier J, Bridge DA, Hagins D, et al. Sex-based outcomes of darunavir–ritonavir therapy: the GRACE (gender, race, and clinical experience) study. Ann Intern Med. 2010;153:349–357.
11. Umeh OC, Currier JS, Park JG, et al. Sex differences in lopinavir and ritonavir pharmacokinetics among HIV-infected women and men. J Clin Pharmacol. 2011;51:1665–1673.
12. Smith KY, Tierney C, Mollan K, et al. Outcomes by sex following treatment initiation with atazanavir plus ritonavir or efavirenz with abacavir/lamivudine or tenofovir/emtricitabine. Clin Infect Dis. 2014;58:555–563.
13. Ofotokun I, Chuck SK, Hitti JE. Antiretroviral pharmacokinetic profile: a review of sex differences. Gend Med. 2007;4:106–119.
14. Fletcher CV, Jiang H, Brundage RC, et al. Sex-based differences in saquinavir pharmacology and virologic response in AIDS Clinical Trials Group Study 359. J Infect Dis. 2004;189:1176–1184.
15. Krecic-Shepard ME, Barnas CR, Slimko J, et al. Gender-specific effects on verapamil pharmacokinetics and pharmacodynamics in humans. J Clin Pharmacol. 2000;40:219–230.

women; LMIC; sex differences; third-line ART; HIV; women with HIV

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.