Factors associated with 10 years of continuous viral load suppression on HAART

The primary goal of highly active antiretroviral therapy (HAART) in HIV-infected individuals is sustained viral load (VL) suppression which, in turn, leads to several benefits. Immune recovery is evidenced by CD4 cell gains, a reduction in activated CD8 cells, decreased immune activation, and enhanced responsiveness to vaccines [1‐3]. Effective HAART can also mitigate opportunistic infections and the development of acquired immunodeficiency syndrome (AIDS) events [4]. Successful HIV treatment is also associated with a reduction in HIV-associated non-AIDS diseases, including cancer and cardiovascular, renal, and liver disease [5, 6]. Recent studies have demonstrated life expectancy approaching that of HIV-uninfected persons for those with low rates of virologic failure and relatively preserved CD4 counts [7]. In addition, sustained VL suppression enhances the durability of the treatment regimen and prevents the development of antiretroviral resistance [8].

Unfortunately, not all patients are able to maintain VL suppression on HAART. In the US, approximately 40 % of HIV-infected persons have a suppressed VL [9]. Recent cohort studies report VL suppression rates ranging between 65 and 80 % on a variety of HAART regimens, whereas newer HAART regimens have suppression rates approaching 90 % in clinical trials [10‐13]. There are many patient-related and treatment-related factors associated with virologic failure early in the course of HAART. Unlike many other chronic conditions, adherence is of major importance as the probability of VL suppression is diminished in patients with <95 % adherence [14]. In addition, suboptimal adherence can lead to increased drug resistance and reduced quality of life and survival [15, 16].

Sustained VL suppression is essential to achieve the best possible treatment outcomes. Because a significant proportion of patients do not achieve this goal, we examined factors associated with ≥10 years of VL suppression in the US Military HIV Natural History Study (NHS). We hypothesize that both patient and treatment-related factors will be associated with long-term VL suppression, and analyses included medication adherence and characteristics of the early HAART era such as prior therapy with single or dual agents. Longitudinal treatment outcomes spanning the entire HAART era were also analyzed to evaluate CD4 trajectories and development of new AIDS events during long-term HAART.

A total of 276 participants met the inclusion criteria (Table 1). Participants were predominately male with similar proportions of Caucasians and African Americans in each group. The median age at HIV diagnosis was 32 and 30 years in the CS and NCS groups, respectively (p = 0.065). A greater proportion of NCS participants (90 %) were treated in the early HAART era (1996–1999) compared with the CS group (66 %; p < 0.001). The median CD4 count at first HAART was higher in the CS compared with the NCS group (375 vs 261 cells/uL; p < 0.001) and the median VL at first HAART was 4.4 log₁₀ copies/mL and 4.5 log₁₀ copies/mL, respectively (p = 0.048). First HAART regimens differed among the groups (p < 0.001), with NCS having a higher proportion of unboosted PI use compared with CS participants (76 % vs 51 %). AIDS outcomes prior to first HAART were similar for CS (8 %) and NCS participants (12 %; p = 0.399). During longitudinal follow-up, groups had a median of 1 treatment interruption [0,1], with a higher proportion of any treatment interruption observed in NCS (57 %) compared to CS (17 %) participants (p < 0.001). NCS participants also had a greater median number of HAART regimens (7 [4, 9] vs 3 [2, 5]; p < .001) and a higher proportion of new AIDS events by 10 years of HAART (13 vs 5 %; p = 0.032) compared with CS participants. In a subgroup analysis of participants with VL data using assays with a limit of detection <50 copies/mL during HAART, the proportion with a suppressed VL <50 copies/mL was higher in the CS group (2399 of 2550 VL values, 94 %) compared to the NCS group (1545 of 2221, 70 %; p < 0.001).

Over the past 20 years, the availability of HAART has effectively transformed HIV from a uniformly fatal condition to that of a chronic disease. Contemporary treatment is associated with improved life expectancy, with one study finding no elevated mortality risk in patients with CD4 counts above 500 cells/uL who maintained VL suppression [17]. Since long-term VL suppression is required for optimal HIV disease outcomes, we aimed to determine the factors associated with sustained virologic control on HAART in a large military cohort.

The inability of many patients to maintain VL suppression can be explained in part by the era in which participants began HAART. A previous study in our cohort reported VL suppression during the early HAART era (1996–2000) compared to the late HAART era (2000–2007), with 58 and 85 % VL suppression at 5 years for those treated in the early versus late HAART era, respectively [10]. Similarly, the Antiretroviral Therapy (ART) Cohort Collaboration analyzed treatment naïve participants who began HAART from 1995 to 2003 and observed an increase in VL suppression in later years, with 58 and 83 % for those initiating HAART between 1995–1996 and 2002–2003, respectively [18]. Our current study with a longer duration of follow-up after HAART initiation (≥10 years) affirms that treatment during the early HAART era is a risk for non-continuous VL suppression.

The use of ARVs prior to HAART, typically mono- or dual-therapy, is an important characteristic of the early HAART era. We observed that ARV use prior to HAART negatively impacted both the ability to achieve long-term VL suppression and the magnitude of CD4 cell gains during HAART. Although non-HAART ARV use was likely a major reason for suboptimal treatment outcomes, other features of the early HAART era other factors likely contributed to the lower rate of treatment success during that period. For example, the early HAART era is also characterized by less potent, unboosted PI regimens, which were more commonly used for first HAART in the NCS group, and other HAART regimens associated with greater adverse effects and reduced tolerability compared to more contemporary regimens [19, 20]. In contrast to the beginnings of HAART, continuous VL suppression is more attainable with newer HAART regimens. For example, recent studies of integrase inhibitor-based regimens have demonstrated VL suppression in approximately 87 % of treatment naïve patients [21, 22]. However, longer follow-up time is necessary to determine the long-term durability of newer regimens.

Advanced HIV disease, as evidenced by low CD4 cell count and AIDS events, is associated with lower rates of VL suppression on HAART [19, 23, 24]. In our current study, we observed that new AIDS events occurring during HAART were more common in those with non-continuous VL suppression despite long-term treatment. We previously found that high cumulative VL on HAART was associated with a greater than 2-fold increase in AIDS events and impacted overall CD4 gains [11]. In our current study, the trajectory of CD4 cells gains over a 10-year period was much higher in those with continuous VL suppression. This is consistent with other studies showing maximal CD4 reconstitution occurs with sustained VL suppression [25, 26]. Thus, the long-term maintenance of VL suppression is of key importance for achieving optimal immune recovery and for the reduction of AIDS events.

Suboptimal treatment outcomes involve both clinical and non-clinical factors. For example, access and engagement in care have been identified as key areas along the HIV care continuum [9]. Some of these barriers are minimized or removed in the NHS cohort as military members and beneficiaries have free access to care and prescription medications. These benefits were demonstrated in a previous NHS study demonstrating a high proportion of VL suppression (81 %) at 1 year [10]. Recent studies in other healthcare settings report slightly lower VL suppression ranging from 70 to 78 %, which may be contributed by access to care or patient characteristics such as injection drug use or socioeconomic factors not typically present in the military population [27‐29].

Discordant HIV treatment outcomes have been observed in different racial/ethnic groups. A study of non-white participants found a VL suppression rate of 31 % compared to 54 % in white participants at 7–14 months in an urban setting [23]. Non-white ethnicity was also linked to higher number of missed appointments, injection drug use, and lower CD4 count at HAART initiation [23]. A previous study in our cohort showed that African Americans had lower odds of VL suppression at 6 and 12 months post-compared to participants of European ancestry, although time to virologic failure was no different once suppression was achieved [30]. Similarly, a negative trend for ≥10 years of VL suppression was also observed in African Americans in our current study, however only a small subset of the NHS cohort was included and additional studies are needed to evaluate potential differences according to race/ethnicity.

Adherence is a cornerstone of successful HIV treatment. We found no difference in those reporting 100 % SRA in CS and NCS participants and the overall adherence was 98 % or greater in both groups. This high degree of SRA may be somewhat explained by free access to care and medications, or possibly by other characteristics of our population such as level of education, socioeconomic status, or very low rate of injection drug use [31]. Other explanations may include potential disadvantages of SRA compared with other methods to assess adherence. For example, SRA is subject to recall and social desirability bias and patients tend to report a 10–20 % over estimation on adherence when directly compared to data obtained from electronic drug monitoring [32]. Methods such as medication event monitoring system (MEMS) track caps can be more accurate as they report exact date and times of all bottles opening. One study found that high MEMS adherence rates, but not high SRA, was associated with VL suppression [33]. However, MEMS has potential drawbacks of cost and tracking does not always directly correlate with pill ingestion [34].

The strengths of this study include long-term follow-up spanning the entire HAART era in a diverse cohort of patients with free access to care and medications. Limitations included the inability to assess adherence in the early HAART era since SRA data was not systematically captured until 2006. The high level of SRA in our study likely limited the ability to detect an association between adherence and VL suppression, since >95 % adherence is associated with optimal and sustained VL suppression [14]. However, equally high SRA in both groups allowed for the evaluation of other factors associated with ≥10 years of VL suppression in our cohort. Other potential study limitations include the retrospective study design and non-randomized treatment decisions. Differing limits of detection of VL assays used during the 18-year study period, particularly in the early HAART era, may have limited the interpretation of factors associated with continuous viral load suppression. Our results may not be generalizable to women, given the high proportion of males in our cohort, or to populations with other characteristics such as injection drug use.

Sustained VL suppression is essential to reduce long-term morbidity in HIV-infected patients, as demonstrated by the lower proportion of AIDS events and improved CD4 cell reconstitution after ≥10 years of continuous VL suppression. We identified factors negatively associated with ≥10 years of VL suppression, including high VL at first HAART, prior ARV use, HIV diagnosis in the pre-HAART era, and number of HAART regimens used. It is likely that treatment in the current era, characterized by regimens that are more potent, more convenient, and less toxic, may promote higher success rates and more durable VL suppression in the future.

AIDS, acquired immune deficiency syndrome; ARV, antiretroviral therapy; CS, continuous viral load suppression; HAART, highly active antiretroviral therapy; HIV, human immunodeficiency virus; INSTI, integrase strand transfer inhibitor; LOWESS, locally-weighted scatter plot smoothing; NCS, non-continuous viral load suppression; NHS, natural history study; NNRTI, non-nucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitors; PI, protease inhibitor; SRA, self-reported adherence; VL, viral load