Background

Randomized clinical trials support the switch both to the fixed dose combination (FDC) of rilpivirine/tenofovir disoproxil fumarate/emtricitabine from a ritonavir-boosted protease inhibitor (PI/r) and to regimens based on integrase strand transfer inhibitors (InSTI) from any kind of antiretroviral therapy. In the SPIRIT Study switching from a regimen based on a PI/r to the FDC of rilpivirine/tenofovir/emtricitabine was not inferior to continuing the PI/r, with virological success in 93,7% and 89,9% of patients at 48 weeks [1]. Non-inferiority was shown also for switching from the FDC of efavirenz/tenofovir disoproxil fumarate/emtricitabine to rilpivirine/tenofovir alafenamide/emtricitabine [2], as well as for switching from a PI/r to raltegravir [3], from a regimen based on a non-nucleoside reverse transcriptase inhibitor (NNRTI) to the FDC of elvitegravir/cobicistat/emtricitabine/tenofovir disoproxil fumarate [4] or from any regimen to the FDC of dolutegravir/abacavir/lamivudine [5]. By contrast, in the STRATEGY-PI Study, switching from a PI/r to the FDC of elvitegravir/cobicistat/emtricitabine/tenofovir disoproxil fumarate was superior in terms of virological success (93,8% vs. 87,1%) and improved patients related outcomes [6], while switching from lopinavir/ritonavir to raltegravir (without changing the nucleoside backbone) was associated with a higher incidence of virological failure [7].

However, these switch strategies have never been compared in randomized clinical trials and it is very unlikely that such a trial will be performed.

The aim of this study was two-fold: first, we aimed at identifying factors associated with the choice of a FDC of rilpivirine/tenofovir/emtricitabine vs. an InSTI-based switch regimen; secondly, we aimed at investigating the durability of rilpivirine/tenofovir/emtricitabine and of InSTI-based switch regimens (when associated to tenofovir disoproxil fumarate/emtricitabine, TDF/FTC) in virologically suppressed HIV-infected patients in clinical practice.

Methods

Retrospective cohort study on patients followed at the Infectious Diseases Department of the San Raffaele Scientific Hospital in Milan (Italy). Data recorded in the database of the Infectious Diseases Department of the San Raffaele Hospital (IDD-HSR) were used for the analyses. At their first visit in our clinic, subjects provide written informed consent to include their clinical and laboratory data in the IDD-HSR for scientific purposes. The study was approved by the ethics committee of the San Raffaele Scientific Institute.

Eligible patients were those who started rilpivirine or an InSTI (both along with TDF and FTC) with <50 HIV-RNA copies/mL and had at least one HIV-RNA assessed while receiving the study regimen. Patients who switched to one of the study regimens since 2008 (date of availability of raltegravir in Italy) were included in the analyses. Patients were followed up to virological failure (VF) or discontinuation of any drug or data freezing (8th September 2015), whichever occurred first.

Estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration Equation (CKD-EPI) [8] and the liver fibrosis FIB-4 index was calculated as described [9].

HIV-RNA was quantified by using the kinetic PCR molecular system (kPCR, Versant HIV-1 RNA kPCR 1.0; Siemens HealthCare Diagnostics, Tarrytown, NY, USA) up to March 2014 and by using the Abbot Real-Time PCR (Abbott Molecular, Des Plaines, IL, USA) thereafter. The kPCR assay gives three possible outputs: (i) a quantitative result for HIV-RNA values of ≥37 copies/mL; (ii) a semiquantitative result (detectable below 37 copies/mL) when HIV RNA is detectable but not precisely quantifiable; (iii) a qualitative result (‘undetectable’) when no signal can be detected. The Abbot Real-Time PCR assay gives also three possible outputs: (i) a quantitative result for HIV RNA values of ≥40 copies/mL; (ii) a semiquantitative result (detectable below 40 copies/mL) when HIV RNA is detectable but not precisely quantifiable; (iii) a qualitative result (‘undetectable’) when no signal can be detected.

VF was defined as two consecutive measurements of HIV-RNA >50 copies/mL or a single HIV-RNA >50 copies/mL followed by ART modification; an unconfirmed HIV-RNA >50 copies/mL (i.e. one measurement >50 copies/mL preceded and followed by a measurement <50 copies/mL), not followed by ART modification, was defined as viral blip. Residual viremia was defined as any detectable HIV-RNA below 50 copies/mL, as assed by Siemens kPCR or by Abbot Real-Time PCR.

Time spent with residual viremia was calculated as a proportion of time with residual viremia on observed follow-up. If between 2 observation the viremia changed from undetectable to residual or vice-versa, the time spent considered in this interval was the half. The mathematical formula was:

$$ {T}_{\%}=\frac{\sum_{i=1}^{i=j}\left(\frac{t_i-{t}_{i-1}}{a}\right)}{t_{tot}}\cdot 100 $$

where t is the time length of interval i, j is the last observation and t tot is the patient’s cumulative follow-up. If, during the i th interval, viremia changed from undetectable to residual or vice-versa, then a = 2, else a = 1.

Treatment failure (TF) was define as either VF or discontinuation (for any length) of any drug of the regimen, for any reason. Causes of change in the regimen were reviewed independently by two clinicians; discordances were discussed and reconciled.

Descriptive data are expressed as median (interquatile range) of frequency (%), as appropriate.

Chi-square and Mann-Whitney tests were used to evaluate differences between the two groups for categorical and continuous variables respectively. Durability was assessed by the Kaplan-Meier curve and compared by Log-rank test. Multivariable logistic regression was used to identify predictors of opting for an InSTI- rather than a rilpivirine-based regimen and a multivariable Cox regression model was used to identify factors independently associated with TF.

All of the statistical tests were two-sided at 5% level, and were performed using SAS Software (release 9.2; SAS Institute).

Results

Six hundred seventy-five patients (466 switched to a rilpivirine-, 209 switched to an InSTI-based regimen [18% dolutegravir, 39% raltegravir, 43% elvitegravir/cobicistat]), on antiretroviral therapy since 6.6 [3.3–14.1] years and with HIV-RNA <50 copies/mL since 3.1 [1.1–5.6] years, were included in the analysis; their baseline characteristics are illustrated in Table 1 and in the Additional file 1: Table S1. Dolutegravir was used at a daily dose 50 mg in all cases, raltegravir at the standard dose of 400 mg twice daily and elvitegravir/cobicistat was always co-formulated with TDF/FTC).

Table 1 Baseline characteristics
Full size table

After adjusting for age, gender, HIV risk factor, higher HIV-RNA value before starting ART, nadir and baseline CD4+ count, baseline triglycerides and cholesterol, history of failure to nucleoside reverse transcriptase inhibitors (NRTIs) or to non-NRTIs (NNRTIs), opting for an InSTI- rather than a rilpivirine-based regimen was more likely in subjects co-infected with HCV (OR = 2.16; 95%CI = 1.26–3.71; p = 0.015), with longer exposure to antiretroviral therapy (OR = 1.05 per year longer; 95%CI = 1.01–1.09; p = 0.042) and shorter time with undetectable viremia (OR = 0.93 per year longer; 95%CI = 0.86–0.99; p = 0.041), treated with protease inhibitors (PIs) vs NNRTIs (OR = 5.14; 95%CI = 3.09–8.95;p = 0.003), and treated with regimens not based on a PI/r and not based on a NNRTI vs those treated with NNRTIs (OR = 5.67; 95%CI = 2.22–14.38;p = 0.032).

The median (interquartile range, IQR) follow-up in the rilpivirine and in the InSTI group was 16.7 (8.8–22.2) and 10.4 (5.4–19.6) months. Four (0.9%) and three (1.4%) patients showed VF at 1 year in the rilpivirine and in the InSTI group; cumulatively, 12 patients (8 [1.7%] in the rilpivirine and 4 [1.9%] in the InSTI group) showed VF during follow-up: their characteristics at VF are detailed in Table 2. TFs at 1 year were 38 (8.2%) and 14 (6.7%) in the rilpivirine and in the InSTI group. The 1-year cumulative probabilities (95%CI) of VF and TF were 0.97% (0.36%–2.62%) and 9.73% (7.21%–13.06%) in the rilpivirine group and 1.83% (0.57%–5.77%) and 8.75% (5.25%–14.4%) in the InSTI group, with no difference between groups (p = 0.328 and 0.209 for VF and TF; Fig. 1, panel A and B).

Table 2 Patients’ characteristics at virological failure
Full size table
Fig. 1
figure 1

Cumulative probabilities of virological failure (Panel a) and of treatment failure (Panel b) after switch to a rilpivirine- and or to an integrase inhibitor (InSTI)-based regimen. RPV: rilpivirine; InSTI: integrase strand transfer inhibitor; FTC: emtricitabine; TDF: ftenofovir disoproxil fumarate

Full size image

The incidence rate (IR) (95%CI) of viral blips was 4.46 (3.07–6.27) and 4.48 (2.51–7.40) per 1000 person months of follow-up in the rilpivirine and in the InSTI group (p = 0.988).

The proportion of time spent with residual viremia was comparable in the two groups (9% [IQR 0.5%–49%] and 17% [IQR 0.5%–50%] in the rilpivirine and in the InSTI group, p = 0.087).

Overall, 62/466 (13.3%) and 41/209 (19.6%) patients in the rilpivirine and in the InSTI group discontinued at least one drug of the regimen for any reason. Discontinuations were due to toxicity in 34 (7%) and in 17 (8%) patients in the rilpivirine and in the InSTI group. Of the 51 discontinuations occurred because of toxicity, 28 (55%) were deemed tenofovir toxicity (15 [3.2%] and 13 [6.2%] in the rilpivirine and in the InSTI group); other leading causes of discontinuation were non-tenofovir related toxicity or untoward drug interactions (19 [4.1%] and 4 [1.9%] in the rilpivirine and in the InSTI group) and treatment simplification (none and 12 [5.7%] in the rilpivirine and in the InSTI group).

Non-tenofovir related toxicities leading to discontinuation were: liver toxicity in 9 (2%), gastrointestinal toxicity in 5 (1%), central nervous system toxicity in 4 (1%) and undefined toxicity in one further patient in the rilpivirine group, liver toxicity in 1 (<1%) and central nervous system toxicity in 1 (<1%) in the InSTI group. Three (1.5%) patients in the InSTI group discontinued for untoward drug interactions.

Of note, discontinuations occurred in 37/82 (45%) patients who started raltegravir; 27 of these 37 patients (73%) discontinued only raltegravir; the main reason for raltegravir discontinuation was treatment simplification (12/27 [44%]).

By the multivariable Cox regression model, TF was independently associated with being on therapy with a PI vs. a NNRTI at switch (AHR = 0.52; 95%CI = 0.31–0.90; p = 0.018), baseline total/HDL-cholesterol ratio (AHR = 1.19 per 0.5-units increments; 95%CI = 1.06–1.34; p = 0.004), baseline eGFR (AHR = 0.78 per 10-units increments; 95%CI = 0.67–0.90; p = 0.001) and baseline hemoglobin (AHR = 0.78 per 1-unit increments; 95%CI = 0.64–0.94; p = 0.009), whereas treatment group (rilpivirine vs. InSTI), HCV-coinfection, nadir and baseline CD4+ cell count, time with HIV-RNA <50 copies/mL, time spent with residual viremia, years of antiretroviral therapy, failure to NRTIs, failure to NNRTIs, failure to PIs, triglycerides and FIB-4 were not. No violation of the proportional hazard assumption was detected using graphical representation (Log-log plot).

Discussion

In this non-randomized study, the efficacy of switching to a FDC of rilpivirine/tenofovir disoproxil fumarate/emtricitabine was similar to that of switching to an InSTI plus tenofovir disoproxil fumarate/emtricitabine-based regimen: both the virological outcomes and treatment discontinuations were similar. As regards to virological outcomes, not only the cumulative risk of virological failure was very low and similar for both regimens, but also the incidence of viral blips and the exposure to residual viremia during follow-up were not statistically different. This suggest that both of these types of switch regimen can be safely used in clinical practice.

However, among patients switched to InSTI, most discontinuation occurred in those switched to raltegravir: this suggests clinician should preferentially switch patients virologically suppressed to a once-daily regimen.

A lower risk of failure was observed in patients switched from PIs: our hypothesis is that patients switched from a PI to a PI-sparing regimen have a greater improvement in symptoms than those switched from NNRTI to another NNRTI or to a InSTI-based regimen. Indeed, the only switch study in which the superiority of the switch strategy was demonstrated was the STRATEGY-PI [6]: in this randomized clinical trial, patients who underwent treatment switch to a FDC of elvitegravir/cobicistat/tenofovir disoproxil fumarate/emtricitabine had a significant improvement in several PI-related symptoms (mainly gastrointestinal). On the contrary, in studies with comparable design, switching from a NNRTI to a FDC of elvitegravir/cobicistat/tenofovir disoproxil fumarate/emtricitabine was not superior to continuing the NNRTI [4], as well as switching from a FDC of efavirenz/tenofovir disoproxil fumarate/emtricitabine to rilpivirine/tenofovir alafenamide/emtricitabine did not resulted superior to continuing efavirenz/tenofovir disoproxil fumarate/emtricitabine [2].

Although nonspecific, hemoglobin is a marker of the general health status; it has been also associated with mortality in HIV-infected patients [10]: thus, it not surprising that in our analysis it was independently associated with TF, as it is conceivable that patients with a worse general health status are more prone to interrupt or change drugs for toxicity issues. Patients with a higher baseline total/HDL-cholesterol ratio are those with greater metabolic problems: it is likely that, in these cases, treatment was then further modified, in the attempt of normalizing the dyslipidemia. The greater risk of TF in patients with lower eGFR value is consistent with the observation that a major cause of TF in our study was toxicity due to TDF.

Overall, the results of our study confirm, in a large, unselected population, the safety and the efficacy of switch regimens based on either rilpivirine or InSTI demonstrated in randomized clinical trials [1,2,3,4,5,6], thus providing useful data for clinical decision making.

With this study, we also aimed at identifying clinical reasons for opting for an InSTI- rather than a rilpivirine-based switch strategy in everyday clinical practice; this because there are no data from randomized clinical trials guiding clinicians in this decision and the results from switch trials do not clear indicate which is the best switching strategy. The results of our analysis showed that we preferentially opted for InSTI in patients with HCV co-infection, a longer ART duration, a shorter time of HIV suppression, an ongoing treatment with PIs. Possible drivers of these preferences are likely related to the common belief that NNRTIs entail greater liver toxicity (at least among persons co-infected with HCV) [11,12,13], to the results of the STRATEGY-PI study (which showed the superiority of this strategy compared to continuing on a PI) [6] and to the belief that InSTI are more potent than NNRTIs, thus favoring this strategy in patients a shorter time of viral suppression when a decision on whether or not switching has to be taken.

Major limitations of this study are the lack of randomization and the fact that the patients switched to rilpivirine/tenofovir disoproxil fumarate/emtricitabine were different from those switched to an InSTI in many baseline clinical features; however, after adjustment for these differences, the multivariable analysis confirmed that switching to a regimen rather than the other was not independently associated to TF.

Conclusions

In our clinical practice, the durability of switch regimens based on rilpivirine and on InSTI (along with TDF and FTC) was comparable and both showed a very low probability of VF.