Abstract
Information on the virologic durability of modern antiretroviral regimens is important to clinicians. We aimed to describe virologic durability of first-line integrase strand transfer inhibitor (INSTI)-, nonnucleoside reverse transcriptase inhibitor (NNRTI)-, or protease inhibitor (PI)-based antiretroviral regimens. This was a retrospective study of antiretroviral-naïve patients that initiated first-line antiretroviral regimens with two nucleoside reverse transcriptase inhibitors and an INSTI, NNRTI, or PI between January 2006 and June 2016. The outcome was time to virologic failure, which was assessed by Kaplan–Meier survival analysis and Cox regression models. There were 780 patients (median age = 37 years [interquartile range (IQR) = 30–45], 93.3% male, 56.2% Caucasian, median HIV duration = 1.8 years [IQR = 0.4–5.4], baseline log10 viral load [VL]=4.6 [IQR = 4.1–5.1], and baseline CD4+ cell count = 320 cells/µl [IQR = 217–440]). In total, 189/780 were on a third agent INSTI, 339/780 on a third agent NNRTI, and 252/780 on a third agent PI. Kaplan–Meier survival probability revealed longer time to virologic failure for INSTI, followed by NNRTI then PI (p < 0.001). Multivariable Cox regression revealed that being on an INSTI regimen (aHR = 0.27; 95%CI = 0.18–0.41) or NNRTI regimen (aHR = 0.64; 95%CI = 0.47–0.87) versus PI regimen, frequent VL testing (per year), (aHR = 0.64; 95%CI = 0.47–0.87), and duration of ART (aHR = 0.22; 95%CI = 0.17–0.30) (years) were inversely associated with time to virologic failure, and log10 of baseline VL (aHR = 1.94; 95%CI = 1.58–2.39 per log10) increased risk. Virologic failure was delayed and virologic durability prolonged for INSTI- compared to NNRTI- and PI-based regimens, supporting current antiretroviral therapy guidelines.
Keywords
Introduction
HIV continues to be an important global disease with approximately 36.7 million people living with HIV (PLHIV) by the end of 2016. 1 Additionally, at the end of 2016, there were approximately one million HIV-related deaths. 2 Fortunately, in the past two decades there have been tremendous advances in HIV treatment and AIDS-related deaths have declined. 3 Globally, AIDS-related death rates in 2014 were 28.5% fewer than those in 2009 and 41.4% fewer than those reported in 2005, respectively.1,2,4 This is primarily due to aggressive scaling up of combination antiretroviral therapy (cART) programs. Current guidelines suggest that HIV is to be treated as early as possible regardless of CD4+ cell count and baseline viral load (VL).5,6 Since cART has been shown to be highly effective, recent emphasis promotes the 90–90–90 ‘Cascade of HIV Care,’ which ultimately aims to ensure HIV-diagnosed individuals reach virologic suppression.7,8 The target set forth by the Cascade of HIV Care is that 90% of PLHIV will know their diagnosis, 90% of those diagnosed will be on cART, and 90% will have reached virologic suppression. 7
As evident from the Cascade of HIV Care, the most important step is virologic suppression. First of all, in order to achieve virologic suppression, one requires effective and potent antiretroviral drugs and one must be adherent to drug regimens and to clinic visits. 7 If virologic suppression is achieved and maintained for a sufficient amount of time, then the CD4+ cell count can continually rise improving one’s health and quality of life. 9 Therefore, viral suppression is critical for survival for PLHIV.9,10 Moreover, full virologic suppression reduces the probability of drug resistance mutations and a need for complex cART regimens, which can be burdensome. 11 Finally, many studies have shown that cART with virologic suppression has been shown to prevent HIV transmission and is the basis of treatment as prevention.12–14
The standard cART regimens recommended to treat antiretroviral-naïve HIV-positive patients consist of two nucleoside reverse transcriptase inhibitors (NRTIs) combined with a third antiretroviral agent which has typically been from the protease inhibitor (PI), nonnucleoside reverse transcriptase inhibitor (NNRTI), or integrase strand transfer inhibitor (INSTI) classes. 15 A recent update to the Guidelines for the Use of Antiretroviral Agents in Adults and Adolescents Living with HIV (2017) concluded that the use of INSTI-based regimens be the only regimen for antiretroviral-naïve patients. 16 This is due to the higher tolerability and performance seen in INSTI compared to other third agent drug classes from clinical trials.16–19
It is important for clinicians to assess and understand the virologic durability of the various modern first-line cART regimens in their day-to-day clinical practice, particularly regarding the performance of the third agent in cART regimens. In our study, we aimed to assess the virologic durability for cART regimens with the third agent class being either an INSTI, NNRTI, or PI, in HIV-positive patients who were antiretroviral naïve.
Materials and methods
Study population
Patient data were collected retrospectively from those who had seen a physician at the Maple Leaf Medical Clinic (MLMC) prior to 31 December 2016. MLMC has provided HIV primary care and community-specific specialty care, including hepatitis C, to over 3900 HIV-positive and 7100 HIV-negative patients. Among the 3900 HIV-positive patients ever seen at MLMC, 2689 were actively receiving care as of 31 December 2016, defined as having at least one doctor visit in the last two years. Of the 2689 patients in care at MLMC, 2558 (95.1%) were on a cART regimen and 2267 (88.6%) had an undetectable VL (<40 copies/ml).
The inclusion criteria for this analysis were having confirmed HIV, being 16 years or older, being antiretroviral naïve and starting a first-line cART regimen between 1 January 2006 and 31 December 2016. The first-line cART regimen must have consisted of a three-drug regimen consisting of two NRTIs and a third agent from a different class. Regimens were limited to those used today in a modern cART era. The NRTI backbone could have included tenofovir disoproxil fumarate (TDF), tenofovir alafenamide, or abacavir (ABC) combined with lamivudine (3TC) or emtricitabine (FTC). Third agents were either an INSTI (raltegravir [RAL], elvitegravir/c cobicistat [EVG/c], dolutegravir [DTG]), an NNRTI (nevirapine [NVP], efavirenz [EFV], etravirine [ETV], rilpivirine [RLP]), or a PI (darunavir boosted with ritonavir [DRV/r] or darunavir boosted with cobicistat [DRV/c], atazanavir boosted with ritonavir [ATV/r] or atazanavir [ATV], lopinavir boosted with ritonavir [LPV/r]). While ETV is typically not used for treatment-naïve patients in our clinic, it is prescribed for its high tolerability and once a day use. Similarly, while unboosted ATV is not common in HIV treatment-naïve patients, it was common in our clinic when prescribed simultaneously with ABC/3TC.
NRTI backbone switches and third agent switches within their respective classes were allowed and still considered as the first-line cART regimens. Patients must have stayed on their first-line cART regimens for at least 14 days and have had at least one available VL test within six months after their first antiretroviral regimen initiation to be included in the analyses. Exclusion criteria included having a baseline VL <400 copies/ml (as some patients were transferred from other clinics), being on any study/placebo medications, discontinuing and switching the third agent class or addition of a fourth drug to the initial three-drug regimen prior to day 14 on their first-line cART regimens.
Statistical analysis
The primary outcome in our study was time to virologic failure. Viral suppression was defined as VL <40 copies/ml or VL <50 copies/ml if done before 1 January 2011. Virologic failure was defined if at least one of the following occurred: (1) failure to achieve viral suppression by six months or (2) after achieving viral suppression, having a detectable VL on two consecutive occasions at least 14 days apart or a single VL ≥200 copies/ml.
The covariate of interest was third antiretroviral agent class, INSTI, NNRTI versus PI. Other covariates included baseline age, gender, ethnicity, time since HIV diagnosis, duration of cART, baseline VL, baseline CD4+ cell count, baseline HCV status, baseline HBV status, and frequency of VL testing (number of tests per year).
Summary statistics were presented for the entire study population and for the INSTI, NNRTI, and PI-based regimen groups separately using frequencies and proportions for categorical variables and medians and interquartile ranges (IQRs) for continuous variables.
Time to virologic failure was assessed using survival analysis with Kaplan–Meier curves for the three third class agents (INSTI, NNRTI versus PI) and Cox proportional hazard regression models. Four Kaplan–Meier curves were plotted assessing time to virologic failure: (1) between all third agent classes, (2) between INSTI and NNRTI, (3) between NNRTI and PI, and (4) between INSTI and PI. Patients were censored if they died, were lost to follow-up, had a third agent class switched, were no longer on a three-drug cART regimen (drugs removed from or added to the regimen), were transferred out of our clinic or after seven years of follow-up (follow-up cutoff date). In order to meet the proportional hazard assumption, a time-dependent interaction term was added, years on ART×log (time to virologic failure). 20
Results
Patient demographics
In total, there were 780 patients: 189 (24.2%) on a third agent INSTI; 339 (43.5%) on a third agent NNRTI regimen; and 252 (32.3%) on a third agent PI, respectively. The median age was 37 years (IQR = 30, 45) and 728 (93.3%) patients were male. Four hundred and thirty-eight (56.2%) patients were Caucasian, 67 (8.6%) were Black, 54 (6.9%) were Hispanic, and 53 (6.8%) were Asian. Median time between HIV diagnosis and start of cART was 1.8 years (IQR = 0.4, 5.4). The median duration of cART was 2.5 years (IQR = 1.1, 4.4) years. The baseline log10 HIV VL was 4.6 copies/ml (IQR = 4.1, 5.1) and baseline CD4+ cell count was 320 copies/µl (IQR = 217, 440). Other sociodemographic and clinical characteristics of the study population are presented in Table 1.
Sociodemographic and clinical variables of the study population.
Statistically significant results (p < 0.05) are indicated by an asterisk *.
ART: antiretroviral therapy; HBV: hepatitis B virus; HCV: hepatitis C virus; IDU, injecting drug use; INSTI: integrase strand transfer inhibitor; IQR: interquartile range; MSM: men who have sex with men; NNRTI: nonnucleoside reverse transcriptase inhibitor; PI: protease inhibitor.
The most common backbone prescribed was TDF+FTC in 590 (75.6%), followed by ABC + 3TC in 153 (19.6%), respectively. Of those on a third agent INSTI, 77 (40.7%) were on EVG/c, 66 (34.9%) were on DTG, and 46 (24.3%) were on RAL. Of those on an NNRTI third agent, 273 (80.5%) were on EFV, 21 (6.2%) were on ETV, 19 (5.6%) were on NVP, and 26 (7.7%) were on RLP. Of those on a third agent PI, 79 (31.3%) were on ATV/r or ATV, 90 (35.7%) were on DRV/r or DRV/c, and 83 (32.9%) were on a LPV/r. The antiretroviral data are presented in Table 2.
Drug regimens.
Statistically significant results (p < 0.05) are indicated by an asterisk *.
ABC: abacavir; ARVs: antiretrovirals; ATV: atazanavir; ATV/r: atazanavir boosted with ritonavir; DRV/c: darunavir boosted with cobicistat; DRV/r: darunavir boosted with ritonavir; DTG: dolutegravir; EFV: efavirenz; ETV: etravirine; EVG/c: elvitegravir boosted with cobicistat; FTC: emtricitabine; INSTI: integrase strand transfer inhibitor; LPV/r: lopinavir boosted with ritonavir (LPV/r); NNRTI: nonnucleoside reverse transcriptase inhibitor; NVP: nevirapine; PI: protease inhibitor; RAL: raltegravir; RLP: rilpivirine; TAF: tenofovir alafenamide; TDF: tenofovir; 3TC: lamivudine.
Kaplan–Meier survival analysis
Kaplan–Meier survival analysis assessed time to virologic failure among the three third agent drug classes, INSTI, NNRTI, and PI and compared them using log-rank testing (Figure 1). When assessing time to virologic failure for all three third agent classes (PI, NNRTI, and INSTI), time to virologic failure occurred quickest in those on a third agent PI, followed by NNRTI and INSTI, respectively (P < 0.001). When comparing time to virologic failure between those on a third agent NNRTI or INSTI, there was no significant difference in time to virologic failure in those on an INSTI compared to an NNRTI (p = 0.2911). When comparing time to virologic failure between those on a third agent INSTI or PI, time to virologic failure was significantly shorter in those on a third agent PI (p < 0.001). Finally, when comparing time to virologic failure between those on a third agent NNRTI or PI, time to virologic failure was significantly short in those on a PI compared to an NNRTI (p < 0.001).
Time to virologic failure between the three third agent classes (INSTI, NNRTI versus PI) in treatment-naïve patients initiating their first-line cART regimens between 1 January 2006 and 31 December 2016. (a) INSTI-based regimens showed a significantly greater time to viremia versus an NNRTI- and PI-based regimen (p<0.0001); (b) INSTI-based regimens showed a nonsignificantly greater time to viremia versus an NNRTI-based regimen (p = 0.29); (c) INSTI-based regimens showed a significantly greater time to viremia versus a PI-based regimen (p = 0.0003); (d) NNRTI-based regimens showed a significantly greater time to viremia versus a PI-based regimen (p<0.0001).
Univariate and multivariable Cox regression analyses
Univariate Cox regression analyses results are presented in Table 3 and revealed that being on a third agent INSTI (aHR = 0.46; 95%CI = 0.30–0.68) or NNRTI (aHR = 0.56; 95%CI = 0.41–0.75) reduced the risk of virologic failure compared to being on a PI. In terms of other covariates, baseline VL (aHR = 1.88; 95%CI = 1.54–2.31 per log10) was associated with an increased risk of virologic failure, whereas frequent VL testing was associated with a decreased risk (aHR = 0.67; 95%CI = 0.60–0.75), while age (aHR = 0.98; 95%CI = 0.97–0.99) and duration of cART (aHR = 0.75; 95%CI = 0.68–0.82) were also associated with a decreased risk. Other covariates were not significantly associated with virologic failure.
Multivariate analysis.
*P < 0.05 is considered statistically significant.
ART: antiretroviral therapy; cART: combination antiretroviral therapy; CI: confidence interval; HBV: hepatitis B virus; HCV: hepatitis C virus; IDU: injecting drug use; INSTI: integrase strand transfer inhibitor; MSM: men who have sex with men; NNRTI: nonnucleoside reverse transcriptase inhibitor; PI: protease inhibitor; VL: viral load.
Multivariable Cox regression (Table 3) also revealed that being on an INSTI (aHR = 0.27; 95%CI = 0.18–0.41) and NNRTI (aHR = 0.64; 95%CI = 0.47–0.87) decreased risk of virologic failure compared to being on a third agent PI. Similarly, frequent VL testing (aHR = 0.64; 95%CI = 0.47–0.87) and duration of cART (aHR = 0.22; 95%CI = 0.17–0.30) decreased risk of virologic failure, whereas a greater log10 of baseline VL (aHR = 1.94; 95%CI = 1.58–2.39 per log10) increased risk. The time-dependent interaction term, years on ART×log (time to virologic failure) was also significant (aHR = 0.20; 95%CI = 0.13–0.31).
Discussion
In our study, we found that time to virologic failure differed for antiretroviral-naïve patients starting their first cART regimen among the three third agent classes (INSTI, NNRTI, and PI), with INSTI-based regimens having the most prolonged virologic durability. In the Kaplan–Meier survival analysis, INSTI-based regimens were not significantly more durable than NNRTI-based regimens, while in the multivariable model, they were significantly more durable. Both INSTI- and NNRTI-based regimens outperformed PI-based regimens virologically.
Our study results are not surprising, as INSTIs have been described as a superior third agent in cART; many guidelines are leaning towards them being used as the first-line cART regimens to achieve the end goal of viral suppression.7,15,21 Raltegravir was the first Federal Drug Administration-approved INSTI in 2007 and since then three more have been approved: DTG, EVG, and bictegravir.21,22 INSTIs have been shown to contribute to faster viral suppression than other forms of antiretrovirals, and this has been shown as early as two weeks for patients, both antiretroviral-naïve and antiretroviral-experienced, initiating an INSTI-based regimen.23–26 Another study showed INSTIs were effective at suppressing the HIV-RNA within the first four weeks of treatment in both antiretroviral-naïve patients or switch patients. 27 Additionally, INSTIs have been shown to be more beneficial in achieving viral suppression in patients who have already experienced a prior virologic failure and are therefore the preferred regimen in patients switching drug class from an NNRTI- or PI-based regimen.21,28,29
Like many other studies, we found that baseline HIV VL significantly increased risk of virologic failure.30–33 Adjusting for baseline VL may be the factor that influenced the significance of virologic failure when comparing INSTI- to NNRTI-based regimens. Like a couple of other studies, more frequent VL testing had a decreased risk of virologic failure.7,34 This could be for a couple of reasons. People who have more frequent VLs may have better engagement in care leading to better virologic responses or if the VL is not suppressing, the clinicians can act quickly and change regimens.
Our study has some clinical utility. Specifically, our findings demonstrate the better performance of INSTIs in achieving viral suppression in a real-world setting, which may further guide decision making when prescribing drugs to patients initiating a cART regimen. In addition, performance of NNRTI regimens was greater than those of a PI, as shown in the survival analysis. Therefore, should one not be able to initiate an INSTI-based regimen, NNRTIs appear to be the next preferable choice. Moreover, INSTI-based regimens may be particularly useful in patients with high baseline VLs. However, we also noted that the median duration of cART was lower for those on an INSTI-based regimen, and this may have failed to capture later failure. However, we believe that since INSTI drugs are newer than NNRTIs or PIs, and that our results are consistent with those in the literature, that our results would be the same if patients on INSTI-based regimens were followed for a longer duration of time. Also, more frequent VL testing was protective against virologic failure in our patients and incorporating an aggressive VL monitoring program should be considered to prevent risk of rebound and to enforce greater adherence to prescribed drug regimens.
Although not shown in our data, side effects of antiretrovirals may impact regimen adherence and subsequently virologic failure due to failure to take their medication. INSTIs are known to have fewer side effects than NNRTIs and PIs, respectively. 35 In 2012, a review published by Lee and Carr 35 had shown that the INSTI had the highest tolerability in patients compared to all other classes of antiretrovirals. The superior tolerability of INSTIs has been shown for short- and medium-term durations; however, data are limited on long-term efficacy of INSTIs and warrant further research. 35 In another study, analyzing the frequency and severity of drug–drug interactions (DDIs) in individuals with HIV revealed that the risk of DDIs was increased for those on boosted PIs and NNRTIs. 36 These DDIs may increase risk of side effects, which could cause a reduced adherence to their drug regimen. The only adherence surrogate measured in our study was the frequency of VL testing. Adherence does not only encompass compliance to drugs, but active participation in one’s own care via follow-up appointments and VL testing. Adherence via compliance to drugs can be shown through pill refill, questionnaire, or frequency of VL testing. Adherence has been shown to reflect and predict the success of viral suppression, where greater adherence has been associated with a greater probability of viral supression. 37 Interestingly, viral suppression can occur whether there is adherence or nonadherence to the regimen. 38 This meaning, in some cases, viral suppression may be achieved even when a patient does not follow their treatment plan. However, nonadherence can increase risk of residual virologic failure and future rebound. 39 In a recent study, it has been shown that there was greater chance of drug adherence and reduced risk of residual HIV if the third agent drug was an INSTI, then NNRTI, and then a PI, respectively. 39
Our study does not come without limitations. First, there is always the risk of bias due to unmeasured confounding, missing data, and channeling bias when using retrospective clinic data. In addition, we have tried our best to select all covariates, which may best control for our results. However, some factors which may have influenced time to virologic failure and were not considered include substance abuse and other measures of adherence to drug regimen.40,41 Also, channeling bias may be an issue since patients with presumed poorer adherence and higher VL are placed on a PI-based regimen. 42 Lastly, our study did not differentiate between low and residual level virologic failure, which is an important area to study.
Overall, clinicians are faced with the challenge of choosing and prescribing drug regimens with various characteristics and must rely on research findings to inform their decision. A drug regimen that leads to the highest probability of virologic suppression remains likely the most important factor in this decision. In our large study of patients from a large urban clinic, we found that being on an INSTI-based regimen decreased the chance of virologic failure and outperformed NNRTI- and PI-based regimens in the real world. These findings are consistent with multiple INSTI-based clinical trials and support the current antiretroviral therapy guidelines, which recommend INSTI-based regimens as the first-line cART regimens to be used in antiretroviral-naïve patients.43–47
Footnotes
Acknowledgments
We would like to thank the patients and research staff at the MLMC for making this work possible.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.