DOI: 10.1111/hiv.12272 HIV Medicine (2016), 17, 188 --27
© 2015 British HIV Association
ORIGINAL RESEARCH
Does gender or mode of HIV acquisition affect virological response to modern antiretroviral therapy (ART)? P Saunders,1* AL Goodman,1,2* CJ Smith,3 N Marshall,1 JL O’Connor,3 FC Lampe3 and MA Johnson1 Royal Free London NHS Foundation Trust, London, UK, 2Department of Infection and Immunity, University College London, London, UK and 3Research Department of Infection and Population Health, University College London, London, UK
1
Objectives
Previous UK studies have reported disparities in HIV treatment outcomes for women. We investigated whether these differences persist in the modern antiretroviral treatment (ART) era. Methods
A single-centre cohort analysis was carried out. We included in the study all previously ART-naïve individuals at our clinic starting triple ART from 1 January 2006 onwards with at least one follow-up viral load (VL). Time to viral suppression (VS; first viral load < 50 HIV-1 RNA copies/mL), virological failure (VF; first of two consecutive VLs > 200 copies/mL more than 6 months post-ART) and treatment modification were estimated using standard survival methods. Results
Of 1086 individuals, 563 (52%) were men whose risk for HIV acquisition was sex with other men (MSM), 207 (19%) were men whose risk for HIV acquisition was sex with women (MSW) and 316 (29%) were women. Median pre-ART CD4 count and time since HIV diagnosis in these groups were 298, 215 and 219 cells/μL, and 2.3, 0.3 and 0.3 years, respectively. Time to VS was comparable between groups, but women [adjusted hazard ratio (aHR) 2.32; 95% confidence interval (CI) 1.28–4.22] and MSW (aHR 3.28; 95% CI 1.91–5.64) were at considerably higher risk of VF than MSM. Treatment switches and complete discontinuation were also more common among MSW [aHR 1.38 (95% CI 1.04–1.81) and aHR 1.73 (95% CI 0.97–3.16), respectively] and women [aHR 1.87 (95% CI 1.43–2.46) and aHR 3.20 (95% CI 2.03–5.03), respectively] than MSM. Conclusions
Although response rates were good in all groups, poorer virological outcomes for women and MSW have persisted into the modern ART era. Factors that might influence the differences include socioeconomic status and mental health disorders. Further interventions to ensure excellent response rates in women and MSW are required. Keywords: antiretroviral therapy, gender inequality, HIV, HIV acquisition, men who have sex with men, men who have sex with women, treatment outcomes, virological control Accepted 11 March 2015
Women comprise almost a third of those living with HIV in the UK [2]. In some previous studies in Europe, women have been reported to have slower times to virological suppression, higher rates of virological failure and increased rates of discontinuation of treatment [3,4]. However, a study in Uganda found better outcomes in women [5] and a large Danish cohort study found no difference between genders in terms of response to antiretroviral therapy (ART) [6].
Introduction Despite major breakthroughs in treatment, HIV continues to cause major morbidity for both men and women [1]. Correspondence: Dr Anna Goodman, c/o Ian Charleson HIV Centre, Royal Free Hospital, Pond Street, London NW3 2QG, UK. Tel: 0207 188 1285; fax: 0207 188 3146; e-mail: [email protected] *These authors contributed equally to the paper.
18
MSM, MSW and women: differences in first-line ART outcomes
Differences in treatment outcomes between genders might be expected as a result of physical differences between genders; for example, pregnancy, breast feeding and perhaps differences in pharmokinetics as a result of differences in size, adiposity and side-effect profiles. Such physical differences may influence both choice of ART and effective dosing. Universal antenatal testing for HIV might be expected to lead to earlier diagnoses of HIV infection in women, and in resource-poor settings this is often the case. However, women and heterosexual men [men who have sex with women (MSW)] tend to be diagnosed with HIV infection later than men who have sex with men (MSM) in the UK [2]. In addition, women with HIV infection comprise a different population in the UK, with differences in ethnic origin, country of birth and socioeconomic status when compared with men with HIV infection [2]. There is also some evidence of gender differences in compliance with medications in other disease areas [7]. The previous studies of gender difference in ART outcomes were commenced some years ago now, and we considered the possibility that any gender differences may now have narrowed. This may have occurred as a result of improvements in ART leading to drugs with better side-effect profiles that are easier to take, and the increasing evidence of the safety of ART in pregnancy. In the UK, guidelines for treatment in pregnancy no longer recommend changes in treatment regimes in the majority of cases. In addition, findings from other countries may not be generalizable to a UK HIV-diagnosed population. Men with HIV infection comprise disparate population groups in the UK with differing profiles in terms of demographic factors, such as country of birth, between MSM and MSW. In the UK, MSW with HIV infection appear to be more similar to women living with HIV in their sociodemographic characteristics and ethnicity [2]. There are a smaller number of studies comparing MSM with MSW [8], and we felt a comparison of those in our care would add to the knowledge and understanding of the treatment of HIV infection in the UK. As a large singlecentre study with a diverse group of patients, the Royal Free HIV Cohort Study is well placed to analyse the impact of gender and mode of sexual transmission on response to ART.
Methods Study population Individuals included in this analysis are participants in the Royal Free HIV Cohort Study. This is an observational research database containing information on all HIV-
© 2015 British HIV Association
19
positive individuals who have ever attended the Ian Charleson Day Centre, an out-patients’ HIV Clinic at the Royal Free Hospital, London, UK. Information on demographics, all ART start and stop dates and reasons for stopping, in-patient hospital admissions, new AIDS-related events and serious non-AIDS-related events are extracted through a 100% notes audit by a trained researcher on an annual basis. Laboratory data, including CD4 counts and viral loads, are transferred electronically from the appropriate department. Baseline genotype to guide treatment choice was available from 2004. Data for this analysis were available until 1 March 2012.
Inclusion criteria Previously antiretroviral-naïve individuals starting combination ART for the first time were considered. Combination ART was defined as a regimen consisting of at least three antiretrovirals. As we wished to consider whether gender differences existed in the most recent years, we only considered individuals who started ART for the first time from 1 January 2006 onwards. Finally, to be included, individuals were required to have at least one follow-up viral load measured after the start of ART. As there has previously been evidence of marked differences in response to ART in different risk for acquisition groups, we divided the study population into three gender/HIV acquisition groups for analysis: men whose primary risk factor for acquiring HIV was through sex with other men (MSM), men whose primary risk factor for HIV acquisition was through sex with women(MSW) and women with heterosexually acquired HIV infection. As the number of individuals with non-sexually acquired HIV infection in this cohort is small, these individuals’ data were excluded.
Statistical analyses Individuals were followed from the time of starting ART (baseline) until the date of their last available viral load measurement, or 1 March 2012, whichever occurred first. Primary analyses considered time to virological suppression and time to virological failure. Virological suppression was defined as the first viral load after starting ART that was < 50 HIV-1 RNA copies/mL. Virological failure was defined as the first of two consecutive viral loads > 200 copies/mL after more than 6 months of ART, regardless of whether virological suppression had previously been achieved. Further, two approaches for accounting for treatment discontinuation were considered in this analysis: one where individuals were censored on the date of complete ART discontinuation (so only virological failures while
HIV Medicine (2016), 17, 18 --27
20
P Saunders et al.
receiving ART were considered) and a second where any virological failure was considered, regardless of whether the individual remained on treatment or not. Hazard ratios (HRs) for the association between gender/HIV acquisition group and the virological endpoints were calculated using Cox proportional hazards models. Multivariable models were adjusted for: age at the start of ART; time from diagnosis to the start of ART; pre-ART viral load; pre-ART CD4 count; calendar date of starting ART; nucleoside reverse transcription inhibitor (NRTI) backbone type [emtricitabine (FTC)/tenofovir (TDF) and other]; ‘third’ drug type [protease inhibitor (PI) only, nonnucleoside reverse transcription inhibitor (NNRTI) only and other]; and ethnicity (white, black and other). Next, time to any ART switch (switch or addition at least one new antiretroviral to the regimen) and time to complete ART discontinuation were considered as outcomes. Treatment switches did not include dose or formulation changes, switching from lamivudine (3TC) to emtricitabine (FTC), splitting fixed dose combination tablets into their constituent parts or vice versa. Complete antiretroviral discontinuation was defined as a complete stop of all ART for at least 2 weeks (to exclude classifying intermittent adherence as treatment discontinuation). In the UK, access to health care and all ART is free, and no stock-outs have taken place at the Royal Free Hospital. Cox proportional hazards models investigating the association between gender/HIV acquisition group and ART modification/ discontinuation were performed similarly to those for the virological endpoints. Finally, changes in CD4 cell count after 12 and 24 months were summarized according to gender/HIV acquisition group. The 12-month CD4 count was defined as the measurement taken in the window 9–15 months after starting ART that occurred closest to 12 months. The 24-month (window 18–30 months) CD4 count was calculated similarly. We conducted a number of sensitivity analyses to investigate the robustness of our results. Firstly, we altered the definition of virological failure in a number of ways; changing the viral load cut-off from 200 to both 50 and 1000 copies/mL; reducing the time after which virological failure was assessed from 6 months post-ART to 4 months post-ART and ignoring virological failures that occurred in the first year of treatment. We repeated all multivariable Cox proportional hazards analyses without adjusting for ethnicity. We stratified analyses by baseline viral load (above or below 100 000 copies/mL) and by time since HIV diagnosis (greater or less than 1 year). Finally, we repeated all analyses excluding the women who started ART while pregnant. All analyses were performed using SAS version 9.3 (SAS Institute Inc, Cary, NC).
© 2015 British HIV Association
Results Study population There were 1190 previously ART-naïve individuals who started ART at the Royal Free Hospital from 1 January 2006 onwards, of whom 39 men (20 who acquired HIV through injecting drug use, three through receipt of blood products, two by other routes and 14 by unknown routes) and 11 women (eight who aquired HIV through injecting drug use and three with unknown routes of infection) were excluded as they had acquired HIV through nonsexual routes. Of the remaining 1140 individuals, we excluded 29 individuals as they did not start with at least three antiretroviral drugs and 25 individuals as they did not have at least one viral load measurement after starting ART. Therefore, 1086 individuals were included, of whom 563 (52%) were MSM, 207 (19%) were MSW and 316 (29%) were women. The median follow-up time was similar between groups: 3.0 [interquartile range (IQR) 1.7–4.6] years for MSM, 2.8 (IQR 1.3–4.3) years for MSW and 2.9 (IQR 1.4–4.6) years for women. The median (IQR) numbers of follow-up viral load measurements per individual were also similar at 11 (7–17), 10 (5–17) and 12 (7–17), respectively. The characteristics at the time of starting ART are given in Table 1. Of note, women and MSW were more likely to be of black African ethnicity and have a pre-existing AIDS diagnosis, a lower pre-ART CD4 count and a shorter time since HIV diagnosis. There were also differences in the types of ART regimens prescribed: MSM were more likely to receive a FTC/TDF backbone, whereas women were more likely to receive a zidovudine/3TC backbone and to be on a PI-based regimen.
Virological endpoints High virological suppression rates were seen in all groups. By 6 months post-ART, 70.1% [95% confidence interval (CI) 66.2–73.9%] of MSM, 66.1% (95% CI 59.4–72.7%) of MSW and 74.1% (95% CI 69.2–79.0%) of women had achieved this outcome. By 12 months these figures had increased to 88.8% (95% CI 86.2–91.4%), 84.1% (95% CI 78.9–89.2%) and 85.1% (95% CI 81.1–89.1%), respectively (P = 0.35; log rank test). In a Cox proportional hazards model (Table 2), after adjustment for potential confounders, there was no evidence of an association between gender and time to virological suppression (P = 0.21). Next, we considered virological failure. Here, we found that virological failure was more common among MSW and women. When censoring at the time of complete ART discontinuation (Fig. 1a), 2.6% of MSM, 6.9% of MSW and
HIV Medicine (2016), 17, 18 --27
MSM, MSW and women: differences in first-line ART outcomes
21
Table 1 Characteristics at the time of starting antiretroviral therapy (ART) MSM n Pregnant when starting ART Age at ART initiation (years) [median (IQR)] Time from diagnosis to ART initiation (years) [median (IQR)] Ethnicity White Black African Other Year started ART [median (IQR)] Nadir CD4 count (cells/μL) [median (IQR)] CD4 count at ART initiation (cells/μL) [median (IQR)] VL at ART initiation (log10 copies/mL) [median (IQR)] Previous AIDS diagnosis NRTI backbone FTC/TDF ABC/3TC ZDV/3TC TDF/3TC Other 2 NRTIs < 2 NRTIs > 2 NRTIs PI/NNRTI/II component EFV NVP Other NNRTI LPV ATV/r DRV/r SQV/r Fos-APV/r Dual PI Unboosted PI 3 NRTIs PI + NNRTI RAL
MSW
563 0 (0.0) 39 (33, 44) 2.3 (0.2, 5.1)
Women
207 0 (0.0) 41 (35, 50) 0.3 (0.1, 2.5)
459 (81.5) 7 (1.2) 97 (17.2) 10/2008 (06/2007, 04/2010) 262 (184, 348) n = 524 298 (200, 411) n = 524 4.9 (4.4, 5.3) n = 511 70 (12.4)
316 32 (10.1) 37 (32, 43) 0.3 (0.1, 3.6)
61 (29.5) 102 (49.3) 44 (21.3) 07/2008 (05/2007, 01/2010) 180 (50, 277) n = 185 215 (56, 321) n = 185 4.9 (4.2, 5.5) n = 207 67 (32.4)
49 (15.5) 195 (61.7) 72 (22.8) 06/2008 (01/2007, 10/2009) 202 (90, 275) n = 278 219 (98, 304) n = 278 4.6 (3.8, 5.2) n = 278 69 (21.8)
471 (83.7) 52 (9.2) 9 (1.6) 7 (1.2) 4 (0.8) 18 (3.2) 2 (0.4)
147 (71.0) 33 (15.9) 10 (4.8) 5 (2.4) 6 (2.9) 6 (2.9) 0 (0.0)
203 (64.2) 47 (14.9) 49 (15.5) 1 (0.3) 7 (2.2) 8 (2.5) 1 (0.3)
229 (40.7) 19 (3.4) 6 (1.1) 109 (19.4) 61 (10.8) 71 (12.6) 23 (4.1) 16 (2.8) 3 (0.5) 2 (0.4) 1 (0.2) 13 (2.3) 10 (1.8)
95 (45.9) 10 (4.8) 1 (0.5) 48 (23.2) 18 (8.7) 17 (8.2) 5 (2.4) 7 (3.4) 2 (1.0) 1 (0.5) 0 (0.0) 2 (1.0) 1 (0.5)
119 (37.7) 14 (4.4) 1 (0.3) 107 (33.9) 27 (8.5) 20 (6.3) 12 (3.8) 4 (1.3) 3 (1.0) 1 (0.3) 0 (0.0) 3 (1.0) 5 (1.6)
All entries are n (%) unless otherwise stated. MSM, men who have sex with men; MSW, men who have sex with women; IQR, interquartile range; VL, viral load; NRTI, nucleoside reverse transcriptase inhibitor; FTC, emtricitabine; TDF, tenofovir; ABC, abacavir; 3TC, lamivudine; ZDV, zidovudine; PI, protease inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor; SQR, saquinavir; II, integrase inhibitor; EFV, efavirenz; NVP, nevirapine; /r, ritonavir-boosted; LPV, lopinavir; ATV, atazanavir; DRV, darunavir; Fos-APV, fos-amprenavir; RAL, raltegravir.
10.0% of women had experienced virological failure by 18 months (P < 0.0001; log rank test). When ignoring treatment discontinuation (Fig. 1b), these figures were 5.2% 9.4% and 19.0%, respectively (P < 0.0001). In a Cox proportional hazards model (Table 2), we found that, after adjustment for potential confounders, MSW had more than three times the rate of virological failure [adjusted hazard ratio (aHR) 3.64; 95% CI 1.69–7.83] and women more than four times the rate (aHR 4.74; 95% CI 2.29–9.84; P = 0.0001) compared with MSM. When ignoring treatment discontinuations, although attenuated, the strong associations remained, with MSW having two and a half times (aHR 2.32; 95% CI 1.28–4.22) and women having more than three times (aHR 3.28; 95% CI 1.91–5.64) the rate of virological failure compared with MSM (P < 0.0001). When comparing women with MSW, women
© 2015 British HIV Association
had a 30% higher rate of virological failure than MSW (aHR 1.3; 95% CI 0.7–2.44) and this effect was increased to 41% (aHR 1.41; 95% CI 0.86–2,33) when treatment discontinuations were ignored.
Treatment modifications/discontinuations After 12 months of ART, 26.9% of MSM, 33.2% of MSW and 41.5% of women had made at least one change to their antiretroviral regimen (P < 0.0001; log rank test; Fig. 2a). MSM were most likely to make a treatment switch for toxicity reasons or patient choice, non-MSM men were most likely to switch for toxicity, patient choice or treatment failure and women were most likely to stop for toxicity or patient choice, or because they were only receiving ART to prevent mother-to-child transmission
HIV Medicine (2016), 17 18 --27
© 2015 British HIV Association
Results are from a Cox proportional hazards regression model. Multivariable analyses adjusted for: age at start of ART, time from diagnosis to start of ART, pre-ART viral load, pre-ART CD4 count, calendar date of starting ART, nucleoside reverse transcriptase inhibitor backbone type (emtricitabine/tenofovir or other), third drug type (protease inhibitor only, nonnucleoside reverse transcriptase inhibitor only or other) and ethnicity (white, black African or other). ART, antiretroviral therapy; MSM, men who have sex with men; MSW, men who have sex with women.
© 2015 British HIV Association
ORIGINAL RESEARCH
Does gender or mode of HIV acquisition affect virological response to modern antiretroviral therapy (ART)? P Saunders,1* AL Goodman,1,2* CJ Smith,3 N Marshall,1 JL O’Connor,3 FC Lampe3 and MA Johnson1 Royal Free London NHS Foundation Trust, London, UK, 2Department of Infection and Immunity, University College London, London, UK and 3Research Department of Infection and Population Health, University College London, London, UK
1
Objectives
Previous UK studies have reported disparities in HIV treatment outcomes for women. We investigated whether these differences persist in the modern antiretroviral treatment (ART) era. Methods
A single-centre cohort analysis was carried out. We included in the study all previously ART-naïve individuals at our clinic starting triple ART from 1 January 2006 onwards with at least one follow-up viral load (VL). Time to viral suppression (VS; first viral load < 50 HIV-1 RNA copies/mL), virological failure (VF; first of two consecutive VLs > 200 copies/mL more than 6 months post-ART) and treatment modification were estimated using standard survival methods. Results
Of 1086 individuals, 563 (52%) were men whose risk for HIV acquisition was sex with other men (MSM), 207 (19%) were men whose risk for HIV acquisition was sex with women (MSW) and 316 (29%) were women. Median pre-ART CD4 count and time since HIV diagnosis in these groups were 298, 215 and 219 cells/μL, and 2.3, 0.3 and 0.3 years, respectively. Time to VS was comparable between groups, but women [adjusted hazard ratio (aHR) 2.32; 95% confidence interval (CI) 1.28–4.22] and MSW (aHR 3.28; 95% CI 1.91–5.64) were at considerably higher risk of VF than MSM. Treatment switches and complete discontinuation were also more common among MSW [aHR 1.38 (95% CI 1.04–1.81) and aHR 1.73 (95% CI 0.97–3.16), respectively] and women [aHR 1.87 (95% CI 1.43–2.46) and aHR 3.20 (95% CI 2.03–5.03), respectively] than MSM. Conclusions
Although response rates were good in all groups, poorer virological outcomes for women and MSW have persisted into the modern ART era. Factors that might influence the differences include socioeconomic status and mental health disorders. Further interventions to ensure excellent response rates in women and MSW are required. Keywords: antiretroviral therapy, gender inequality, HIV, HIV acquisition, men who have sex with men, men who have sex with women, treatment outcomes, virological control Accepted 11 March 2015
Women comprise almost a third of those living with HIV in the UK [2]. In some previous studies in Europe, women have been reported to have slower times to virological suppression, higher rates of virological failure and increased rates of discontinuation of treatment [3,4]. However, a study in Uganda found better outcomes in women [5] and a large Danish cohort study found no difference between genders in terms of response to antiretroviral therapy (ART) [6].
Introduction Despite major breakthroughs in treatment, HIV continues to cause major morbidity for both men and women [1]. Correspondence: Dr Anna Goodman, c/o Ian Charleson HIV Centre, Royal Free Hospital, Pond Street, London NW3 2QG, UK. Tel: 0207 188 1285; fax: 0207 188 3146; e-mail: [email protected] *These authors contributed equally to the paper.
18
MSM, MSW and women: differences in first-line ART outcomes
Differences in treatment outcomes between genders might be expected as a result of physical differences between genders; for example, pregnancy, breast feeding and perhaps differences in pharmokinetics as a result of differences in size, adiposity and side-effect profiles. Such physical differences may influence both choice of ART and effective dosing. Universal antenatal testing for HIV might be expected to lead to earlier diagnoses of HIV infection in women, and in resource-poor settings this is often the case. However, women and heterosexual men [men who have sex with women (MSW)] tend to be diagnosed with HIV infection later than men who have sex with men (MSM) in the UK [2]. In addition, women with HIV infection comprise a different population in the UK, with differences in ethnic origin, country of birth and socioeconomic status when compared with men with HIV infection [2]. There is also some evidence of gender differences in compliance with medications in other disease areas [7]. The previous studies of gender difference in ART outcomes were commenced some years ago now, and we considered the possibility that any gender differences may now have narrowed. This may have occurred as a result of improvements in ART leading to drugs with better side-effect profiles that are easier to take, and the increasing evidence of the safety of ART in pregnancy. In the UK, guidelines for treatment in pregnancy no longer recommend changes in treatment regimes in the majority of cases. In addition, findings from other countries may not be generalizable to a UK HIV-diagnosed population. Men with HIV infection comprise disparate population groups in the UK with differing profiles in terms of demographic factors, such as country of birth, between MSM and MSW. In the UK, MSW with HIV infection appear to be more similar to women living with HIV in their sociodemographic characteristics and ethnicity [2]. There are a smaller number of studies comparing MSM with MSW [8], and we felt a comparison of those in our care would add to the knowledge and understanding of the treatment of HIV infection in the UK. As a large singlecentre study with a diverse group of patients, the Royal Free HIV Cohort Study is well placed to analyse the impact of gender and mode of sexual transmission on response to ART.
Methods Study population Individuals included in this analysis are participants in the Royal Free HIV Cohort Study. This is an observational research database containing information on all HIV-
© 2015 British HIV Association
19
positive individuals who have ever attended the Ian Charleson Day Centre, an out-patients’ HIV Clinic at the Royal Free Hospital, London, UK. Information on demographics, all ART start and stop dates and reasons for stopping, in-patient hospital admissions, new AIDS-related events and serious non-AIDS-related events are extracted through a 100% notes audit by a trained researcher on an annual basis. Laboratory data, including CD4 counts and viral loads, are transferred electronically from the appropriate department. Baseline genotype to guide treatment choice was available from 2004. Data for this analysis were available until 1 March 2012.
Inclusion criteria Previously antiretroviral-naïve individuals starting combination ART for the first time were considered. Combination ART was defined as a regimen consisting of at least three antiretrovirals. As we wished to consider whether gender differences existed in the most recent years, we only considered individuals who started ART for the first time from 1 January 2006 onwards. Finally, to be included, individuals were required to have at least one follow-up viral load measured after the start of ART. As there has previously been evidence of marked differences in response to ART in different risk for acquisition groups, we divided the study population into three gender/HIV acquisition groups for analysis: men whose primary risk factor for acquiring HIV was through sex with other men (MSM), men whose primary risk factor for HIV acquisition was through sex with women(MSW) and women with heterosexually acquired HIV infection. As the number of individuals with non-sexually acquired HIV infection in this cohort is small, these individuals’ data were excluded.
Statistical analyses Individuals were followed from the time of starting ART (baseline) until the date of their last available viral load measurement, or 1 March 2012, whichever occurred first. Primary analyses considered time to virological suppression and time to virological failure. Virological suppression was defined as the first viral load after starting ART that was < 50 HIV-1 RNA copies/mL. Virological failure was defined as the first of two consecutive viral loads > 200 copies/mL after more than 6 months of ART, regardless of whether virological suppression had previously been achieved. Further, two approaches for accounting for treatment discontinuation were considered in this analysis: one where individuals were censored on the date of complete ART discontinuation (so only virological failures while
HIV Medicine (2016), 17, 18 --27
20
P Saunders et al.
receiving ART were considered) and a second where any virological failure was considered, regardless of whether the individual remained on treatment or not. Hazard ratios (HRs) for the association between gender/HIV acquisition group and the virological endpoints were calculated using Cox proportional hazards models. Multivariable models were adjusted for: age at the start of ART; time from diagnosis to the start of ART; pre-ART viral load; pre-ART CD4 count; calendar date of starting ART; nucleoside reverse transcription inhibitor (NRTI) backbone type [emtricitabine (FTC)/tenofovir (TDF) and other]; ‘third’ drug type [protease inhibitor (PI) only, nonnucleoside reverse transcription inhibitor (NNRTI) only and other]; and ethnicity (white, black and other). Next, time to any ART switch (switch or addition at least one new antiretroviral to the regimen) and time to complete ART discontinuation were considered as outcomes. Treatment switches did not include dose or formulation changes, switching from lamivudine (3TC) to emtricitabine (FTC), splitting fixed dose combination tablets into their constituent parts or vice versa. Complete antiretroviral discontinuation was defined as a complete stop of all ART for at least 2 weeks (to exclude classifying intermittent adherence as treatment discontinuation). In the UK, access to health care and all ART is free, and no stock-outs have taken place at the Royal Free Hospital. Cox proportional hazards models investigating the association between gender/HIV acquisition group and ART modification/ discontinuation were performed similarly to those for the virological endpoints. Finally, changes in CD4 cell count after 12 and 24 months were summarized according to gender/HIV acquisition group. The 12-month CD4 count was defined as the measurement taken in the window 9–15 months after starting ART that occurred closest to 12 months. The 24-month (window 18–30 months) CD4 count was calculated similarly. We conducted a number of sensitivity analyses to investigate the robustness of our results. Firstly, we altered the definition of virological failure in a number of ways; changing the viral load cut-off from 200 to both 50 and 1000 copies/mL; reducing the time after which virological failure was assessed from 6 months post-ART to 4 months post-ART and ignoring virological failures that occurred in the first year of treatment. We repeated all multivariable Cox proportional hazards analyses without adjusting for ethnicity. We stratified analyses by baseline viral load (above or below 100 000 copies/mL) and by time since HIV diagnosis (greater or less than 1 year). Finally, we repeated all analyses excluding the women who started ART while pregnant. All analyses were performed using SAS version 9.3 (SAS Institute Inc, Cary, NC).
© 2015 British HIV Association
Results Study population There were 1190 previously ART-naïve individuals who started ART at the Royal Free Hospital from 1 January 2006 onwards, of whom 39 men (20 who acquired HIV through injecting drug use, three through receipt of blood products, two by other routes and 14 by unknown routes) and 11 women (eight who aquired HIV through injecting drug use and three with unknown routes of infection) were excluded as they had acquired HIV through nonsexual routes. Of the remaining 1140 individuals, we excluded 29 individuals as they did not start with at least three antiretroviral drugs and 25 individuals as they did not have at least one viral load measurement after starting ART. Therefore, 1086 individuals were included, of whom 563 (52%) were MSM, 207 (19%) were MSW and 316 (29%) were women. The median follow-up time was similar between groups: 3.0 [interquartile range (IQR) 1.7–4.6] years for MSM, 2.8 (IQR 1.3–4.3) years for MSW and 2.9 (IQR 1.4–4.6) years for women. The median (IQR) numbers of follow-up viral load measurements per individual were also similar at 11 (7–17), 10 (5–17) and 12 (7–17), respectively. The characteristics at the time of starting ART are given in Table 1. Of note, women and MSW were more likely to be of black African ethnicity and have a pre-existing AIDS diagnosis, a lower pre-ART CD4 count and a shorter time since HIV diagnosis. There were also differences in the types of ART regimens prescribed: MSM were more likely to receive a FTC/TDF backbone, whereas women were more likely to receive a zidovudine/3TC backbone and to be on a PI-based regimen.
Virological endpoints High virological suppression rates were seen in all groups. By 6 months post-ART, 70.1% [95% confidence interval (CI) 66.2–73.9%] of MSM, 66.1% (95% CI 59.4–72.7%) of MSW and 74.1% (95% CI 69.2–79.0%) of women had achieved this outcome. By 12 months these figures had increased to 88.8% (95% CI 86.2–91.4%), 84.1% (95% CI 78.9–89.2%) and 85.1% (95% CI 81.1–89.1%), respectively (P = 0.35; log rank test). In a Cox proportional hazards model (Table 2), after adjustment for potential confounders, there was no evidence of an association between gender and time to virological suppression (P = 0.21). Next, we considered virological failure. Here, we found that virological failure was more common among MSW and women. When censoring at the time of complete ART discontinuation (Fig. 1a), 2.6% of MSM, 6.9% of MSW and
HIV Medicine (2016), 17, 18 --27
MSM, MSW and women: differences in first-line ART outcomes
21
Table 1 Characteristics at the time of starting antiretroviral therapy (ART) MSM n Pregnant when starting ART Age at ART initiation (years) [median (IQR)] Time from diagnosis to ART initiation (years) [median (IQR)] Ethnicity White Black African Other Year started ART [median (IQR)] Nadir CD4 count (cells/μL) [median (IQR)] CD4 count at ART initiation (cells/μL) [median (IQR)] VL at ART initiation (log10 copies/mL) [median (IQR)] Previous AIDS diagnosis NRTI backbone FTC/TDF ABC/3TC ZDV/3TC TDF/3TC Other 2 NRTIs < 2 NRTIs > 2 NRTIs PI/NNRTI/II component EFV NVP Other NNRTI LPV ATV/r DRV/r SQV/r Fos-APV/r Dual PI Unboosted PI 3 NRTIs PI + NNRTI RAL
MSW
563 0 (0.0) 39 (33, 44) 2.3 (0.2, 5.1)
Women
207 0 (0.0) 41 (35, 50) 0.3 (0.1, 2.5)
459 (81.5) 7 (1.2) 97 (17.2) 10/2008 (06/2007, 04/2010) 262 (184, 348) n = 524 298 (200, 411) n = 524 4.9 (4.4, 5.3) n = 511 70 (12.4)
316 32 (10.1) 37 (32, 43) 0.3 (0.1, 3.6)
61 (29.5) 102 (49.3) 44 (21.3) 07/2008 (05/2007, 01/2010) 180 (50, 277) n = 185 215 (56, 321) n = 185 4.9 (4.2, 5.5) n = 207 67 (32.4)
49 (15.5) 195 (61.7) 72 (22.8) 06/2008 (01/2007, 10/2009) 202 (90, 275) n = 278 219 (98, 304) n = 278 4.6 (3.8, 5.2) n = 278 69 (21.8)
471 (83.7) 52 (9.2) 9 (1.6) 7 (1.2) 4 (0.8) 18 (3.2) 2 (0.4)
147 (71.0) 33 (15.9) 10 (4.8) 5 (2.4) 6 (2.9) 6 (2.9) 0 (0.0)
203 (64.2) 47 (14.9) 49 (15.5) 1 (0.3) 7 (2.2) 8 (2.5) 1 (0.3)
229 (40.7) 19 (3.4) 6 (1.1) 109 (19.4) 61 (10.8) 71 (12.6) 23 (4.1) 16 (2.8) 3 (0.5) 2 (0.4) 1 (0.2) 13 (2.3) 10 (1.8)
95 (45.9) 10 (4.8) 1 (0.5) 48 (23.2) 18 (8.7) 17 (8.2) 5 (2.4) 7 (3.4) 2 (1.0) 1 (0.5) 0 (0.0) 2 (1.0) 1 (0.5)
119 (37.7) 14 (4.4) 1 (0.3) 107 (33.9) 27 (8.5) 20 (6.3) 12 (3.8) 4 (1.3) 3 (1.0) 1 (0.3) 0 (0.0) 3 (1.0) 5 (1.6)
All entries are n (%) unless otherwise stated. MSM, men who have sex with men; MSW, men who have sex with women; IQR, interquartile range; VL, viral load; NRTI, nucleoside reverse transcriptase inhibitor; FTC, emtricitabine; TDF, tenofovir; ABC, abacavir; 3TC, lamivudine; ZDV, zidovudine; PI, protease inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor; SQR, saquinavir; II, integrase inhibitor; EFV, efavirenz; NVP, nevirapine; /r, ritonavir-boosted; LPV, lopinavir; ATV, atazanavir; DRV, darunavir; Fos-APV, fos-amprenavir; RAL, raltegravir.
10.0% of women had experienced virological failure by 18 months (P < 0.0001; log rank test). When ignoring treatment discontinuation (Fig. 1b), these figures were 5.2% 9.4% and 19.0%, respectively (P < 0.0001). In a Cox proportional hazards model (Table 2), we found that, after adjustment for potential confounders, MSW had more than three times the rate of virological failure [adjusted hazard ratio (aHR) 3.64; 95% CI 1.69–7.83] and women more than four times the rate (aHR 4.74; 95% CI 2.29–9.84; P = 0.0001) compared with MSM. When ignoring treatment discontinuations, although attenuated, the strong associations remained, with MSW having two and a half times (aHR 2.32; 95% CI 1.28–4.22) and women having more than three times (aHR 3.28; 95% CI 1.91–5.64) the rate of virological failure compared with MSM (P < 0.0001). When comparing women with MSW, women
© 2015 British HIV Association
had a 30% higher rate of virological failure than MSW (aHR 1.3; 95% CI 0.7–2.44) and this effect was increased to 41% (aHR 1.41; 95% CI 0.86–2,33) when treatment discontinuations were ignored.
Treatment modifications/discontinuations After 12 months of ART, 26.9% of MSM, 33.2% of MSW and 41.5% of women had made at least one change to their antiretroviral regimen (P < 0.0001; log rank test; Fig. 2a). MSM were most likely to make a treatment switch for toxicity reasons or patient choice, non-MSM men were most likely to switch for toxicity, patient choice or treatment failure and women were most likely to stop for toxicity or patient choice, or because they were only receiving ART to prevent mother-to-child transmission
HIV Medicine (2016), 17 18 --27
© 2015 British HIV Association
Results are from a Cox proportional hazards regression model. Multivariable analyses adjusted for: age at start of ART, time from diagnosis to start of ART, pre-ART viral load, pre-ART CD4 count, calendar date of starting ART, nucleoside reverse transcriptase inhibitor backbone type (emtricitabine/tenofovir or other), third drug type (protease inhibitor only, nonnucleoside reverse transcriptase inhibitor only or other) and ethnicity (white, black African or other). ART, antiretroviral therapy; MSM, men who have sex with men; MSW, men who have sex with women.
