Characteristics of heterosexually-acquired compared to homosexually-acquired HIV and implications for clinical practice: results from the Australian HIV Observational Database

Abstract

Heterosexuals living with HIV report feeling additional HIV stigma compared to homosexual men, which may affect clinical outcomes. Yet, beyond routinely collected surveillance data, little is known about the characteristics of individuals who acquire HIV heterosexually and clinical outcomes by mode of sexual acquisition have not been directly compared. Using data from the Australian HIV Observational Database, we compared clinical characteristics of those with heterosexually-acquired (Het-HIV) to homosexually-acquired HIV (Hom-HIV) to investigate any differences and their implications for clinical management. 513 Het-HIV and 1467 Hom-HIV patients were included and contributed 3,127 and 9,457 person-years of follow-up, respectively. Compared with Hom-HIV, Het-HIV were more often born outside Australia (62.5% vs 39.9%, p<0.001), less likely to have Hepatitis C (4.8% vs 7.8%, p=0.029) and had lower median CD4 counts at diagnosis (292 vs 450 cells/µL, p<0.001) and cART initiation (270 vs 340 cells/µL, p<0.001). Despite these lower CD4 counts, there were no significant differences between groups for time to the major clinical endpoints of cART initiation, viral suppression, virological failure or all-cause mortality. Het-HIV had a lower risk of loss-to-follow-up than Hom-HIV (aHR 0.78; 95% CI 0.64–0.95). Further studies examining factors associated with, and interventions to inform retention in care are required.

Introduction

The World Health Organisation (WHO) estimates that over 36.9 million people live with HIV globally, the majority of whom were infected via heterosexual sex (World Health Organisation, 2019). In Australia, as in other resource-rich countries, men who have sex with men (MSM) represent the highest risk group for HIV acquisition, accounting for almost two-thirds of new HIV diagnoses in 2018 (Kirby Institute, 2018). Heterosexuals accounted for approximately one-quarter of new Australian HIV diagnoses in the same period (Kirby Institute, 2018).

Recent Australian surveillance data indicates that the epidemiology of HIV acquisition in Australia is changing. While new HIV diagnoses in MSM are decreasing overall, modest increases in numbers of new HIV diagnoses in heterosexuals are being observed. In 2017, 238 (25%) new HIV diagnoses across Australia were heterosexually-acquired, an increase from 209 (21%) in 2016. This increase was mostly observed among Australian-born heterosexual men, in whom notifications have increased by 45% over the last decade (Kirby Institute, 2018).

There is evidence that heterosexuals living with HIV face additional challenges which may affect their clinical care. These challenges include feeling additional HIV stigma (Brener et al., 2013) and that existing HIV services are more tailored to gay men (Antoniou et al., 2012). Such challenges have been associated with poorer health outcomes (Logie & Gadalla, 2009), treatment uptake and adherence (Antoniou et al., 2012; Brener et al., 2013).

However, beyond routine surveillance data little is known about the characteristics of individuals who acquire HIV heterosexually and clinical outcomes have not been previously compared by mode of sexual acquisition of HIV. We sought to compare the clinical characteristics of those with heterosexually-acquired (Het-HIV) with homosexually-acquired HIV (Hom-HIV) and consider the implications of any differences identified on clinical management and health promotion messaging.

Methods

We included all patients enrolled in the Australian HIV Observational Database (AHOD) who were diagnosed between January 1997 and March 2018 and reported only heterosexual exposure or only homosexual exposure as their likely mode of HIV acquisition. AHOD is an observational cohort of 4270 HIV-positive individuals attending 30 sexual health clinics, high-caseload general practices and tertiary referral centres throughout Australia and New Zealand (Australasia) for HIV clinical care. Ethics approval was obtained from the University of New South Wales Australian Ethics Committee and written informed consent to upload de-identified clinical information was obtained from patients. A detailed description of AHOD has previously been published (Australian HIV Observational Database, 2002). Patients were excluded if they reported homosexual contact and injecting drug use, injecting drug use only, blood products or “other” as their likely mode of HIV acquisition. Individuals identifying as transgender or bisexual were also excluded.

We compared differences in key clinical characteristics between groups using either the Wilcoxon rank-sum test or a two-tailed χ² test significantat the α=0.05 level. Patient characteristics evaluated are listed in Table 1.

Table 1.

Patient characteristics by mode of HIV acquisition

Variable	Homosexual exposure	Heterosexual exposure	P-value
n	1467	513
Age
at diagnosis	35.7 (29.2–43.5)	35.5 (28.9–46.2)	0.280A
at enrolment	39.3 (32.3–46.9)	40.1 (32.6–49.5)	0.080A
Time in follow-up
total	9,457	3,127
median	5.4 (2.1–8.9)	5.0 (1.9–8.1)	0.062A
Country of birth
Australia	729 (60.1%)	166 (37.5%)	<.001B,C
other	484 (39.9%)	277 (62.5%)
missing	254	70
Most recent HBV serologyD
positive	45 (3.8%)	17 (3.9%)	0.937B,C
negative	1132 (96.2%)	418 (96.1%)
missing	290	78
Most recent HCV serologyE
positive	105 (7.8%)	22 (4.8%)	0.029B,C
negative	1246 (92.2%)	439 (95.2%)
missing	116	52
CD4 count
at diagnosis	450 (290–635)	292 (122–574)	<.001A
missing	588	211
at enrolment	518 (374–700)	460 (297–640)	<.001A
missing	170	72
at cART initiation	340 (220–506)	270 (154–410)	<.001A
missing	356	105
Viral load
at diagnosis	61,075 (12,106–204,000)	63,463 (9,333–160,000)	0.276A
<400	37 (4.3%)	18 (6.0%)	0.288B,C
≥400	833 (95.7%)	283 (94.0%)
missing	597	212
at enrolment	74 (49–8,735)	50 (40–3,562)	0.076A
<400	797 (62.6%)	302 (67.7%)	0.058B,C
≥400	477 (37.2%)	144 (32.3%)
missing	193	67
at cART initiation	64,016 (14,212–179,469)	38,450 (3,306–139,552)	<.001A
<400	108 (9.9%)	63 (15.8%)	0.002B,C
≥400	984 (90.1%)	337 (84.3%)
missing	375	113
Past AIDS diagnosisF
yes	154 (10.5%)	77 (15.0%)	0.006B
no	1313 (89.5%)	436 (85.0%)
Treatment initiation
year	2009 (2003–2012)	2009 (2004–2012)	0.090A
no ART	128	30

^AWilcoxon’s rank sum test

^Bχ²-test

^Cexcluding “missing” category

^DHBV serology defined as Hepatitis B surface antigen

^EHCV serology defined as Hepatitis C antibody or Hepatitis C PCR (if previous positive Hepatitis C antibody)

^FPast AIDS diagnosis defined as having a history of AIDS at time of enrolment to AHOD

Cox proportional hazard models were used to investigate times to combination antiretroviral therapy (cART) initiation, first viral suppression, first treatment failure, first cART change, all-cause mortality and loss-to–follow-up (LTFU).To eliminate the effect of any factors other than mode of HIV acquisition on outcomes, multivariate models were adjusted a priori for the factors outlined in the footnotes of Table 2.

Table 2.

Clinical end-point univariate- and covariate-adjusted Cox proportional hazard ratios.

End-point Time to:	Hazard ratio (hetero vs homosexual exposure)	95% CI	P-value
All-cause mortality
univariate	0.79	0.43–1.45	0.447
multivariateA	0.75	0.40–1.41	0.373
Loss-to-follow-up
univariate	0.80	0.66–0.97	0.025
multivariateA	0.78	0.64–0.95	0.014
ART initiation
univariate	1.17	1.05–1.30	0.004
multivariateB	1.04	0.93–1.16	0.501
Viral suppression
univariate	0.98	0.86–1.11	0.735
multivariateC	0.95	0.83–1.08	0.445
Virological failure
Univariate	0.88	0.68–1.14	0.321
multivariateC	1.10	0.83–1.45	0.518
1st treatment change
Univariate	1.03	0.90–1.17	0.700
multivariateC	0.97	0.83–1.13	0.677

^AAdjusted for age at cohort enrolment, country of birth, HCV infection, HBV infection, CD4 cell count at cohort enrolment, HIV viral load at cohort enrolment, year of ART initiation and clinical care setting.

^BAdjusted for age at diagnosis, country of birth, HCV infection, HBV infection, CD4 cell count at diagnosis, HIV viral load at diagnosis, year of diagnosis and clinical care setting.

^CAdjusted for age at ART initiation, country of birth, HCV infection, HBV infection, CD4 cell count at ART initiation, HIV viral load at ART initiation, year of ART initiation and clinical care setting.

Time windows for included CD4 and HIV viral load measurements spanned from 90 days prior to 30 days after HIV diagnosis or cohort enrolment (as appropriate). For cART initiation, the included time window was from 180 days prior to 14 days after. Measurements outside these windows were excluded.

Results

A total of 1980 patients were included in the analysis: 513 Het-HIV, 1467 Hom-HIV who contributed 3127 and 9457 person-years of follow-up (PYFU), respectively. There was no significant difference in age between the groups either at diagnosis or enrolment to AHOD (Table 1).

As shown in Table 1, compared with Hom-HIV, Het-HIV were significantly more often born outside Australia and significantly less likely to have current Hepatitis C, but not Hepatitis B infection. Het-HIV had significantly lower CD4 counts at both diagnosis and cART initiation compared to Hom-HIV and were significantly more likely to have had a previous AIDS diagnosis at time of enrolment. There was no significant difference in viral load between groups at diagnosis although Het-HIV had significantly lower viral loads at cART initiation compared to Hom-HIV.

Het-HIV commenced cART at significantly lower CD4 counts than Hom-HIV and after 5 years of therapy, despite CD4 reconstitution, Het-HIV median CD4 counts remained significantly lower than those of Hom-HIV (550 v 618 cells/µL, p=0.012) (Figure 1A). To determine whether this could be due to a cohort effect within the data set, we conducted a sub-analysis, including only patients who started cART in 2007 or later. This revealed a similar pattern and again, significantly lower CD4 counts in the Het-HIV compared to the Hom-HIV group after 5 years of treatment (490 v 661 cells/µL, p=0.001) (Figure 1B).

Figure 1.

Median CD4 response after 5 years of treatment among the entire cohort.; Figure 1B. Median CD4 response after 5 years of treatment for those commencing cART in 2007 or later.

At 5 years after treatment initiation there was no significant difference in rates of undetectable viral load in Het-HIV compared to Hom-HIV (92.4% v 88.7%, p=0.219) (Figure 2A. A similar sub-analysis of those commencing cART from 2007 onwards also found no significant difference in rates of viral suppression between groups (96.8% v 97.9%, p=0.63) (Figure 2A).

Figure 2.

Viral load response to treatment, % undetectable (<400 copies/ml).; Figure 2B. Viral load response to treatment, limited to patients starting cART in 2007 or later.

There was no significant difference between Hom-HIV and Het-HIV for time to cART initiation, time to viral suppression, time to virological failure, time to first treatment change or time to all-cause mortality between groups (Table 2). However, Het-HIV were significantly less likely to be LTFU than Hom-HIV (Table 2).

Discussion

This study is the first to compare and report the clinical characteristics and outcomes of HIV-infected adults in Australasia according to mode of sexual transmission. In addition to limited data available from routine surveillance, our findings suggest that Het-HIV were less likely to have current Hepatitis C infection, but not Hepatitis B infection. Het-HIV were also less likely to be lost to follow up than Hom-HIV. Aside from this finding, there were no differences between groups for times to the major clinical endpoints of treatment initiation, viral suppression, treatment failure, first treatment change or all-cause mortality.

The other results of our study are largely consistent with known surveillance data on heterosexually-acquired HIV in Australia (Kirby Institute, 2018), including that Het-HIV were more likely to be born outside Australia and diagnosed later, with lower CD4 counts than Hom-HIV.

There is a paucity of published literature that has directly compared clinical outcomes in HIV-infected individuals by mode of HIV acquisition. However, reports from resource-rich settings generally support our findings that heterosexuals are diagnosed later (Brown et al., 2017; Mocroft et al., 2013). Interestingly, although Het-HIV commenced cART at lower CD4 counts, they did not have higher risk of all-cause mortality, even on unadjusted analysis, in contrast to findings of other large international cohort studies (Kitahata et al., 2009; Opravil et al., 2002; Sterne et al., 2009).

The lower rates of LTFU seen in Het-HIV in our study could be one explanation for this disparity. This finding has not been documented or examined in any depth previously and is perhaps surprising given Het-HIV are more likely to be overseas-born (Kirby Institute, 2018), potentially being ineligible for Australia’s universal healthcare system “Medicare” and facing language and cultural barriers to accessing care. However, there are organisations designed to engage heterosexuals living with HIV and help retain them in care, for example, the New South Wales (NSW) state-based Pozhet (Pozhet, 2019) and the NSW Multicultural HIV and Hepatitis Service for those born overseas (Multicultural HIV and Hepatitis Service, 2018). Australian publicly-funded sexual health clinics have been shown to per-form well in retaining HIV-positive individuals in care (Healey & O’Connor, 2018). Given that heterosexuals are more likely to be overseas born, they may be more likely to attend these public clinics which do not require Medicare, are free and confidential. Such clinics can also source free HIV medication for Medicare-ineligible patients through cART access programs provided by several pharmaceutical companies. The lower CD4 counts observed in Het-HIV may suggest this group could be more unwell, or symptomatic, and thus have higher rates of retention in care. Another hypothesis for our finding is that Het-HIV may be more likely to remain longer in the same geographic location due to family commitments, and Hom-HIV may be more likely to move locations and access care at clinics not participating in AHOD thus appearing to be LTFU in our analysis. Nonetheless, this finding warrants further study into predictors of engagement in care as effective retention in care is important for individual and public health outcomes (Horstmann et al., 2010).

Studies on the impact of stigma and discrimination for people living with HIV (PLHIV) have found multiple negative effects for their overall health and wellbeing (Logie & Gadalla, 2009; Tzemis et al., 2013). Despite evidence that heterosexuals experience additional HIV stigma compared to homosexuals living with HIV we found no differences between groups for times to major clinical endpoints, including all-cause mortality, in our study. A previous Australian study similarly found no differences on health and wellbeing measures in heterosexual compared with homosexual PLHIV, although these were self-reported (Brener et al., 2013). These findings suggest that there is no simple causal relation between stigma experiences and health outcomes and any association may be complicated by other issues such as minority stress or other health disadvantage (Brener et al., 2013).

Indications for treatment change, pregnancy status and adherence were not routinely collected in our study. Other limitations include missing data the self-reported likely mode of acquisition of “Heterosexual sex” which may, have included some non-gay identifying MSM.

The perception of being low-risk is a key factor accounting for lower HIV testing rates and late diagnoses among heterosexuals, compared to MSM (and other high-risk groups) in Australia (Kirby Institute, 2018). The low CD4 counts seen in heterosexuals in our study suggest that they are diagnosed late which has been reported previously (Mocroft et al., 2013; Shivaji et al., 2014). Our findings support the use of health promotion messaging to raise risk awareness in heterosexuals and other groups not historically considered to be at risk of HIV.

This study is unique in the published literature in examining differences in several important clinical outcomes between those with homosexually and heterosexually-acquired HIV. Its strengths include over 12,500 PYFU combined across groups with two decades of patients being followed up at diverse clinical settings across Australasia including Sexual Health Clinics, high-caseload General Practices and Tertiary Referral Centres reflecting “real-life” clinical care of patients living with HIV in our region. It is the first Australasian study to examine a broad range of major clinical endpoints by mode of HIV acquisition and our findings are reassuring in that most major clinical endpoints were similar between Het-HIV and Hom-HIV. Further studies are required to better characterise, and develop interventions focusing on loss-to-follow-up among individuals attending clinical services in our region.