Impact of three Decades of Antiretroviral Therapy in a Longitudinal Population Cohort Study

Abstract

Background

We have used a comprehensive HIV population to characterize antiretroviral therapy (ART), drug class selection, pill burden, drug costs and health outcomes over the entire span of the HIV epidemic.

Methods

Antiretroviral (ARV) use (drugs, classes, formulations) and both the laboratory and clinical outcomes (HIV-1 RNA, CD4⁺ T-cell count and mortality) were determined for all patients in Southern Alberta, Canada, at each year-end between 1986 and 2017. Pill burden, cumulative drug exposure and costs were calculated for each year.

Results

The number of ARV-treated patients increased from 29.6% (77/260) in 1989 to 93.4% (1,814/1,943) in 2017. Regimen selection showed continuous adjustments for toxicity, resistance, pill burden and adherence. Dramatic improvements in outcomes were seen. In 1997, 22.4% of treated patients had an undetectable viral load, this has been consistently around 90% since 2010 (92.7% in 2017). While HIV-related annual mortality rate declined from 11.0% in 1994 to 0.1% in 2017, all-cause mortality remained relatively stable from 1997 onwards. ART pill burden escalated in 1997 (12.4/day), then decreased to 2.1/day in 2016. Mean ART cost increased in 1997 (CAN$905/month/regimen in 1997, $1,223 in 2016). Mean cumulative lifetime exposure to protease inhibitors is 5.98 ±4.9 and to nucleoside reverse transcriptase inhibitors 8.8 ±6.2 years.

Conclusions

Our findings demonstrate not only the immense burden that HIV has imposed on both patients and society, but also the substantial benefit of ART on patient outcomes. They show that research, patient engagement and programme support can with time minimize the harmful long-term effects of HIV-infection.

Introduction

Following the discovery in 1984 that AIDS is caused by a retrovirus (now called HIV), over 27 antiretroviral agents (ARVs) have been licensed for its treatment. The aim of antiretroviral therapy (ART) is focused on controlling HIV replication, thereby preventing loss of CD4 lymphocytes, preserving immunity, improving morbidity and decreasing mortality [1]. As new ARVs and new ARV classes became available, combination therapy developed progressing from mono to dual-therapy and then to triple therapy in 1996 within the context of highly active or combination antiretroviral therapy (HAART) [2,3]. Drug resistance from incomplete viral suppression necessitated, for a brief period, the use of multiple drugs and ARV classes (sometimes called mega-HAART or giga-HAART) [4]. Most current ART regimens combine two nucleoside/ nucleotide analogues acting as reverse transcriptase inhibitors (NRTIs) with a third drug from another class [1]. With the availability of new and more potent drugs and classes of ARVs with higher genetic barriers to resistance, simpler regimens including dual therapy are again being pursued to reduce both cost and cumulative ARV exposure [5].

Little longitudinal data covering the span of the entire HIV epidemic has been published on changing patterns of ARV use and ARV classes, pill burden, drug costs, health outcomes and the resulting burden of decades of ARV exposure for many of the patients in current care. We have used a well characterized, geographically defined entire HIV population receiving care in southern Alberta to characterize ART uptake, drug class selection and treatment outcomes between 1986 and 2017. These data show how, by continuously accessing new therapeutic options and adjusting treatment approaches over three decades, despite changes in pill burden, associated drug costs and extensive exposure to ARVs, ART has achieved and sustained dramatic improvement in HIV population health.

Methods

Since 1986, the Southern Alberta HIV clinic (SAC) has been the sole provider, under universal free health care, of provincially funded ART for all those living with HIV in southern Alberta, Canada. The nearest alternative HIV care programme is located 300 km (180 miles) away. ARVs have always been dispensed exclusively from an in-clinic pharmacy with its dispensing data being regularly reconciled with the SAC database. We used cross-sectional year-end analysis in this clinical cohort. ARV use for all patients on current or past ART between 1986 and 2017 was determined at each year-end. ARV drugs, ARV classes, formulations, pill burden, drug exposure and costs were calculated for each year. Pharmacological boosters were not counted as individual agents. As part of a government mandated programme, performance metrics such as HIV-1 RNA, CD4⁺ T-cell count, mortality and ART costs (adjusted to 2016 CDN$) were annually extracted from the SAC database. For viral load we used the last reported measurement, which was almost always within the preceding 6 months (patients who had moved, died or otherwise been lost to follow-up were not included). We define late presentation with a CD4⁺ T-cell count <350/μl and/or an AIDS diagnosis. This work was undertaken as a quality assurance project as specified by the University of Calgary Bioethics committee.

Results

The number of HIV-infected patients actively attending SAC at each year-end between 1986 and 2017 along with their ART use are shown (Figure 1). In 1989, soon after the introduction of ART monotherapy, 77 of 260 (29.6%) patients were receiving treatment, while by 2017 this had increased to 1,814 (93.4%). The changing distribution over 30 years of ARV-naive, ARV-exposed on treatment interruptions (TI) and those on ART is also shown. The use in 1992, of dual-therapy (initially restricted to clinical trials) was followed in late 1996 by increasing use of triple-therapy. Triple-therapy has dominated treatment since, with only rare mono- or dual-therapy use. The use of greater than four antiretroviral agents (the mega-HAART era) peaked in 2003 in 6.5% of treated patients (3.8% of the entire population). After the initial enthusiasm for triple therapy ART in 1997, TI became common and while TI has since declined, it has not completely disappeared despite compelling trial evidence and is most often driven by patient choice [6]. Three drug regimens have become almost universal with a recent trend to the reintroduction of dual-therapy (4.4%) and a few legacy patients (1.4%) on complex therapies due to resistance.

Figure 1

ARV exposure and use in active patients at year end

The changing preferences for specific ARV drug classes (reflecting both clinical trial and wider use after licensure) is shown (Figure 2). These changes demonstrate not only the availability and use of new classes, but also changing knowledge of class side effects, vulnerabilities and improvements within a class. The changing preferential uses of the protease inhibitors (PIs)- and non-nucleoside reverse transcriptase inhibitor (NNRTI) classes reflect these concerns as well as the recent surge in use of the integrase strand transfer inhibitor (INSTI) class [7,8]. Fusion inhibitor (FI) use effectively stopped in 2009 with the introduction INSTIs; however, rare C-C-motif-chemokine-receptor 5 antagonists (CCR5) use persists. NRTI use remains common, with 92% of current patients on at least one NRTI.

Figure 2

All ARV use within each class in active patients at year end

The clinical benefit of ART in the SAC population on both the accepted surrogate markers of HIV disease severity (HIV-1 RNA and CD4⁺ T-cell count) and on yearly all-cause mortality (subdivided into AIDS-related or non-AIDS-related deaths) is shown (Figures 3, 4 and 5). The expanded use of triple-ART in 1997 is followed by improved viral load suppression in the population, climbing CD4⁺ T-cell counts and a sustained decreasing mortality from AIDS. The HIV-related annual mortality rate in SAC patients declined from 11.0% in 1994 to 3.6% in 1997 and to 0.1% in 2017.

Figure 3

Plasma HIV-1 RNA (copies/ml) in antiretroviral-treated patients at year-end

Figure 4

CD4⁺ T-cell count (cells/μl) in all active patients at year end

Figure 5

Yearly mortality and causes of death in Southern Alberta Clinic Cohort

The changing pill burden and the average cost per regimen of the ART used to achieve these gains is shown (Figure 6). HAART initially incurred a high pill burden (particularly for some in the mega-HAART era between 2002 and 2004) but has recently decreased mostly due to newer co-formulated products. While the ‘per patient’ cost has remained stable, the programme cost has increased dramatically, similar to other jurisdictions due to increased use within the population [9].

Figure 6

Number of ARV pills and monthly mean cost of ART in 2016 CDN$ (1990–2016)

The legacy of past ARV use, in years of drug exposure to each class of ARV, in current SAC patients is displayed (Figures 7). In some, the exposure has lasted for over 25 years with daily ART, resulting in a cumulative exposure of many kg of a purified class of ARVs. The mean exposure to INSTIs is 0.21 kg (95% CI 0.18, 0.24), NNRTIs is 1.29 kg (95% CI 1.22, 1.36), PIs is 2.55 kg (95% CI 2.37, 2.72) and NRTIs is 2.49 kg (95% CI 2.39, 2.58) in patients taking each respective A RV.

Figure 7

Cumulative lifetime exposure in years to each ARV class in current patients

Discussion

We present well-documented longitudinal data of ARV use in an entire HIV-infected patient population for the complete time frame of ART availability. This data describes the history of local ARV use and reflects knowledge, drug availability and our application of knowledge over 30 years in a developed country with free access to use available ARVs. The clinical benefit, as seen by both surrogate markers and reduced deaths from improved treatment options, the demands on the patients (pill burden) and ART costs are presented as the therapeutic response to an evolving epidemic unfolded. The results are not presented for comparison to others. In our cohort collaborations, our results (for example as cohort NA # 5 [10]) show similarity to other programmes, but we believe the data are unique due to the coverage of everyone in HIV care in a population and the ability to follow ARV use over 32 years [10]. We wished to document together the temporal changes in ART use and ARV choices, the factors associated with changes in ART, the impact initially of ART on clinical and subsequently on surrogate markers, along with pill burden and costs over 32 years of use. Finally, we show the burden of lifelong ARV drug exposure on many of those now still in care.

The limited potency in the pre-HAART era of the first ARVs and approaches such as monotherapy or dual therapy with early NRTI (such as zidovudine, didanosine, dideoxycytidine, stavudine, lamivudine) is seen with high proportion of patients with CD4⁺ T-cell counts <75 cells/μl and high death rates with many being on a TI due to intolerance of their ART. These figures highlight and reflect the very high morbidity and mortality of untreated HIV infection even in the developed world.

The dramatic advances achieved by the HAART regimens from late 1996 onwards, is seen in both the surrogate (CD4⁺ T-cell counts and the newly introduced metric HIV-1 RNA) and clinical markers (death). Palella et al. [11], although studying a population with more advanced HIV disease, found a comparable decline in HIV-related mortality with use of HAART. The improvement in treatment outcomes reflects the convergence of more effective and less toxic ARV regimens, more flexible treatment options, better knowledge surrounding HIV and its therapy, and enhanced use of prophylaxis and comorbidity management [12]. Early HAART was characterized by high pill burden with the initial use of PIs and two NRTIs. Regimens soon changed with the option to use first-generation NNRTIs with the NRTIs [13,14]. By the end of 1997, triple therapy was used in nearly half (49%) of all HIV-infected patients in southern Alberta (72% of treated patients).

Complete viral suppression of viral replication is essential for avoiding resistance, as was recognized in the initial guidelines for resistance testing published in 1998 [15]. Patient adherence as well as drug potency increasingly attracted clinical attention and effort in attempt to avoid developing resistance and poor outcomes [16,17]. The initial enthusiasm for HAART, however, was tempered as both toxicities and ARV failures from resistance were identified [7,8]. This resulted in more patients remaining ART-naive (21%) and an increase of the popularity of TI (25%) by 2002. ARV toxicities included lipodystrophy (most often seen with earlier PIs), mitochondrial toxicity from some NRTIs, central nervous system effects from NNRTIs, as well as cardiovascular, bone, renal and metabolic effects. Increasing reports of such toxicities likely influenced the ART choices being used [7,18–20]. From 1999 onwards, a small number (4.5%) of patients accumulating resistance, required the use of ‘mega-HAART’ with four or often more ARVs from several classes to achieve viral suppression.

During the past 30 years, the timing for starting ART has fluctuated dramatically. In the pre-HAART era concern about the initiation of ART centred on the toxicity and limited potency of available drugs. This resulted in restricted use, except for the most profoundly immune compromised in an attempt prolong life. In the very early HAART era, enthusiasm for treatment peaked with widespread use after the suggestion that a cure could be achieved within years of suppressive ART [21]. However, recognition of long surviving cellular reservoirs and increasing appreciation of ARV toxicities reduced enthusiasm to use HAART until clinically needed [7,18–20,22]. Drug interactions, difficult regimens, pill burden and subsequent decrease in quality of life also diminished enthusiasm for lifelong ART [6,13]. In 2009 observational cohort data supporting very early treatment with ART again increased its use [23]. The confirmation in a randomized controlled study of the value of early ART and its endorsement in many guidelines has also consolidated this approach [24]. Despite the compelling evidence for the early and comprehensive use of ART, it has now, for diverse reasons, plateaued at 93% of the population in care [25].

The changing use of the classes of ARV has varied over time. Within each class of drug however, there may be generations. The early protease inhibitors (saquinavir and indinavir) required frequent dosing and had a high pill burden with associated toxicities. The decline in PI use after the advent of NNRTIs was somewhat reversed when more potent PIs such as atazanavir and darunavir became available, which had less toxicity and with boosting could be given once a day [26]. Fusion inhibitors were only used between 2002 and 2009 in a handful of patients, while the CCR5 antagonists introduced in 2005 (licensed in 2007) are still being used in a dozen or so patients [27,28]. Their use preceded and seems to have been limited by the arrival of the INSTIs in 2006 (licensed in 2007) [13,29].

The enhanced potency and tolerability of new agents has facilitated novel approaches for individual patients whose options are not constrained by resistance, toxicity or drug-to-drug interactions. The early signals for such use can be seen in the 2017 data with return to dual therapy. Concern regarding cumulative exposure has led to NRTI-sparing (and reducing) regimens such as the use of rilpivirine and dolutegravir as dual therapy, as recently reviewed by Orkin et al. [30] This regimen was recently approved as the first two-drug regimen, single pill, once daily for maintenance treatment of virologically suppressed HIV-1 infection [31]. Dolutegravir–lamivudine dual-therapy is under investigation and used in a few patients as is ritonavir-boosted darunavir [32]. As the HIV-infected population ages, comorbidities and polypharmacy will increase the complexity of disease with therapy management, drug interactions and toxicity all playing a growing role. In the context of health economics, choices will also have to include the consideration of cost within the realm of optimized treatment.

The changing utility of markers to show disease severity and impact can also be seen. Initially mortality was the only marker in use. However, after great discussion a ‘surrogate’ marker of HIV disease, the CD4⁺ T-cell count became accepted and was the cornerstone of ARV registration trials and patient care until the HIV-1 RNA was introduced in 1997 [33]. However, the close relationship of these three markers over the epidemic is clear and validates the difficult decision in the pre-HAART era to use surrogate markers rather than clinical end points. Suppressed viral load is the surrogate marker used in the WHO 90:90:90 target.

Over the last decade, as simpler co-formulated products with higher genetic barriers and reduced toxicities became available, regimens with a lower pill burden started to be used. The first single-tablet combination dual formulation containing zidovudine and lamivudine was marketed in 1997 and many of the current preferred regimens are single tablet triple formulations [1,13]. These new potent agents, co-formulated products and use of boosting agents for the PIs have led to the observed reduction in ART pill burden (Figure 6). While these advances have achieved major goals in preserving health and survival in HIV patients, for many of those in current care it has come, in the absence of curative therapy, at the cost of years and even decades of daily ARV exposure. The result of this clinical requirement has led in many patients to lifetime ingestion of many kg of purified chemicals aimed at inhibiting viral targets. The long-term effects are still being evaluated

There are some limitations to our study. These data reflect use of ART in one programme and it may not be identical to other populations and programmes with different patient demographics, prescribing practices and ARV access. However, in comparative cohort collaborations (ART-CC and NA Accord), SAC data is highly comparable to other developed world cohorts. We use year-end figures to avoid biases from uneven annual churn in our population and this may miss some important details and significant changes of the HIV population over time [34]. We have not dissected out individual ARV use patterns or tried to distinguish precise difference between different generations of an ARV class. The increasing availability of generic ARVs and potential use of dual therapy means that future cost projections from reported figures need to be done cautiously.

Our results document the immense burden that HIV has imposed on both patients and society in our region, but we also show how application of gradual and incremental scientific advances has very significantly benefited patient care and outcomes, and can be reached within an achievable budget. The data show clearly that research, patient engagement and programme support can, with time, minimize the impact of even the most serious life-threatening epidemic.