Efficacy and safety of “unboosting” atazanavir in a randomized controlled trial among HIV-infected patients receiving tenofovir DF

Abstract

Objectives: To assess safety and efficacy of a switch to unboosted atazanavir (ATV) among HIV-infected adults receiving ATV/ritonavir (r) and tenofovir disoproxil fumarate (TDF).

Methods: HIV-infected adults with viral load (VL) <40 copies/mL at screening and <150 copies/mL consistently for ≥3 months while receiving a regimen including ATV/r and TDF were randomized to continue ATV/r 300/100 mg daily (control) or change to ATV 400 mg daily (switch), while maintaining their TDF backbone. The primary outcome was proportion of subjects without treatment failure (regimen switch or VL > 200 copies/mL twice consecutively) at 48 weeks.

Results: Fifty participants (46 male, median age 47 years) were randomized, 25 to each arm. At week 48, treatment success occurred in 76% in the control arm and 92% in the switch arm (ITT, p = 0.25). ATV trough levels at week 9 were higher in controls (median 438 ng/mL) than in the switch arm (median 124 ng/mL) (p = 0.003), as was total bilirubin at week 48 (median 38 μmol/L and 28 μmol/L, respectively; p = 0.02). Estimated glomerular filtration rate (eGFR) decreased in the control arm (p = 0.007), but did not change in the switch arm. At week 48, eGFR was higher in the switch arm (median 96 mL/min) than in the control arm (median 85 mL/min) (p = 0.035), but the arms were similar with respect to fasting glucose, C-reactive protein, and lipid parameters.

Conclusions: Switching from ATV/r to unboosted ATV appears to be safe and effective in selected virologically suppressed patients receiving TDF-containing regimens, and may have favorable effects on bilirubin and renal function.

Keywords:

• Atazanavir
• Unboosting
• Tenofovir
• Ritonavir
• HIV
• Nephrotoxicity
• eGFR

Introduction

Once HIV viral load (VL) suppression has been achieved with antiretroviral regimens including atazanavir 300 mg/ritonavir 100 mg (ATV/r) daily, discontinuing ritonavir and increasing ATV to 400 mg daily (“unboosting” ATV) is an attractive treatment simplification option. In this setting, discontinuing ritonavir may be considered to manage or prevent ritonavir-related metabolic toxicity (e.g. lipid or glucose abnormalities)^{1, 2} or tolerability issues, as well as to reduce the potential for clinically significant drug-drug interactions. This strategy has been found to be safe and effective in randomized controlled trials among subjects receiving an abacavir/lamivudine backbone.^3–5 However, limited prospective data are available for ATV unboosting in regimens containing tenofovir disoproxil fumarate (TDF).^{6, 7} Because co-administration of TDF may lower ATV plasma levels,⁸ some treatment guidelines recommend against using unboosted ATV with TDF.⁹ However, retrospective data indicate that, among virologically suppressed patients, TDF-based regimens were not independently associated with treatment failure after switching to unboosted ATV.¹⁰ Prospective studies suggest that the combination of unboosted ATV and TDF is safe and effective as a switch strategy when guided by therapeutic drug monitoring⁶ or pharmacogenetic assessment⁷; however, neither therapeutic drug monitoring nor pharmacogenetic testing is widely available to clinicians in most jurisdictions. In the absence of either of these tests, the ATV unboosting strategy with TDF-containing regimens has not previously been examined in a randomized controlled trial.

We conducted a randomized controlled trial to assess the effects of a switch to unboosted ATV among HIV-infected patients virologically suppressed on a regimen including ATV/r and TDF. Since both TDF and ATV/r, but not unboosted ATV, have been associated with renal impairment in cohort studies,^{11, 12} we also examined whether unboosting ATV might improve renal safety.

Methods

This prospective, randomized, 48-week, open-label study enrolled HIV-1 infected adults (≥19 years of age) receiving ATV/r 300/100 mg, TDF 300 mg, and either emtricitabine 200 mg or lamivudine 300 mg, all taken once daily, for ≥3 months. Eligible participants had plasma VL < 40 copies/mL ≥ twice consecutively including the screening value, and <150 copies/mL continuously for ≥3 months prior to study enrollment. Subjects were excluded if they had evidence of resistance to nucleoside reverse transcriptase inhibitors (NRTIs) or protease inhibitors (PIs) on previous genotypic resistance tests (with the exception of isolated M184 V/I, which would not be expected to markedly impact study regimen efficacy¹³) or if they were receiving concomitant proton pump inhibitors, rifampin, St. John’s wort, or garlic supplements.

The trial was conducted at a large HIV outpatient clinic in a major urban centre in Canada (the John Ruedy Immunodeficiency Clinic at St. Paul’s Hospital, Vancouver, British Columbia). The protocol and informed consent form were approved by the Providence Health Care/ University of British Columbia Research Ethics Board. The study protocol is registered at ClinicalTrials.gov (identifier NCT01351740).

Design and intervention

Eligible participants were randomized 1:1 to continue ATV/r 300/100 mg daily (control arm) or change to ATV 400 mg daily (switch arm), while maintaining their current TDF-containing backbone.

Procedures and assessments

Participants had the following tests and assessments at study baseline and at weeks 12, 24, 36, and 48: review of clinical adverse events and concomitant medications, CD4 cell count and fraction, HIV-RNA (COBAS Ampliprep Taqman HIV-1 assay, Roche Diagnostic Systems), total bilirubin, creatinine, estimated glomerular filtration rate (eGFR, calculated using the MDRD equation¹⁴), phosphorus, random urine albumin-creatinine ratio (UACR), fasting glucose and lipids (total cholesterol [TC], high-density lipoprotein [HDL] and low-density lipoprotein [LDL] cholesterol, triglycerides [TG]), apolipoprotein B (apoB), and high-sensitivity C-reactive protein (hsCRP). Non-HDL cholesterol, TC/HDL ratio, atherogenic index of plasma (AIP, log[TG/HDL])¹⁵, and LDL/ apoB¹⁶ were calculated.

A timed plasma sample for measurement of ATV pre-dose trough was collected once at or after study week 4. ATV levels were measured using a validated LC-MS/MS method with a lower limit of quantitation of 50 ng/mL.¹⁷ Results were not available in real time, as one of the aims of the study was to assess whether unboosting ATV could safely be accomplished without therapeutic drug monitoring guidance.

Outcome measures

The primary objective of the study was to compare the two arms with respect to the proportions of subjects experiencing treatment success at 48 weeks. Treatment failure was defined as either regimen change for any reason or VL > 200 copies/mL twice consecutively >2 weeks apart.¹⁸

Key secondary objectives were to compare the two treatment arms with respect to ATV trough levels and the proportions of subjects with ATV trough levels below 150 ng/mL (the previously established therapeutic target)^{19, 20} and below 50 ng/mL (the lower limit of quantitation of the assay) at 4 to 8 weeks. Additional secondary objectives were to compare the two strategies with regard to adverse events and changes from baseline to 48 weeks in CD4 cell count (absolute and fraction), total bilirubin levels, and renal and metabolic laboratory parameters.

Statistical analyses

The sample size calculations were based on the software program PASS© (http://www.ncss.com/pass.html). Assuming 85% of individuals in the control group stay suppressed, a sample size of 50 allowed us to detect a difference between arms of 32% or more at 48 weeks using a 1-sided test with an alpha of 0.05 and 0.80 power.

Wilcoxon rank sum test was conducted to compare demographics, laboratory parameters, and ATV levels between arms. Fisher’s exact test was used to compare arms with respect to the proportions of subjects with treatment success at week 48. Wilcoxon signed rank test was conducted to assess within-arm changes from baseline to weeks 24 and 48. All analyses were performed using SAS version 9.4 (SAS, Cary, North Carolina, United States) with a level of significance set at 0.05.

Results

Baseline

Fifty subjects were enrolled between August 2011 and November 2013, of whom 30 (60%) were antiretroviral-naïve prior to starting their ATV/r- and TDF-based regimen. Twenty-five were randomized to the control arm and 25 to the switch arm. At baseline, the groups were similar with respect to demographic characteristics, time on antiretrovirals, and laboratory parameters, with the exception of TG, TC/HDL ratio, and AIP, which were higher in the group randomized to switch to unboosted ATV (Table 1).

Table 1 Baseline characteristics of study participants

	Control arm (N = 25)	Switch arm (N = 25)	Total (N = 50)
Age (years)	46 (39–53)	48 (42–52)	47 (40–53)
Male, n (%)	23 (92%)	23 (92%)	46 (92%)
Ethnicity, n (%)
Caucasian	17 (68%)	18 (72%)	35 (70%)
Aboriginal	4 (16%)	2 (8%)	6 (12%)
African	1 (4%)	0 (0%)	1 (2%)
Other	3 (12%)	5 (20%)	8 (16%)
BMI, kg/m^2	25.7 (21.2–29.7)	23.3 (22.3–26.7)	23.5 (22.2–29.0)
Current regimen, n (%)
TDF/FTC	23 (92%)	18 (72%)	41 (82%)
TDF/3TC	2 (8%)	7 (28%)	9 (18%)
Time on antiretrovirals (months)	36.8	55.8	44.1
	(23.1–59.5)	(28.7–124.1)	(26.6–81.8)
Genotypic resistance, n (%)
3TC/FTC	1 (4%)	3 (12%)	4 (8%)
PI	1 (4%)	0 (0%)	1 (2%)
Concomitant medications, n (%)
Metformin	2 (8%)	0 (0%)	2 (4%)
Atorvastatin	1 (4%)	0 (0%)	1 (2%)
Rosuvastatin	2 (8%)	2 (8%)	4 (8%)
CD4 absolute (cells/μL)	575	570	570
	(440–730)	(470–660)	(460–700)
CD4 fraction	0.33 (0.25–0.38)	0.31 (0.26–0.35)	0.32 (0.25–0.37)
Total bilirubin (μmol/L)	43	48	45
	(34–59)	(37–57)	(34–58)
Creatinine (μmol/L)	79 (70–89)	75 (68–80)	76 (70–85)
eGFR (mL/min)	88 (78–105)	98 (90–105)	95 (79–105)
Phosphorus (mmol/L)	0.98	1.04	0.98
	(0.88–1.09)	(0.90–1.13)	(0.89–1.11)
UACR (mg/mmol)	1.25	0.85	1.00
	(0.50–6.90)	(0.50–1.90)	(0.50–2.95)
Fasting glucose (mmol/L)	5.30	5.20	5.30
	(4.80–6.30)	(4.70–5.80)	(4.75–5.80)
Total cholesterol (mmol/L)	4.29	4.26	4.29
	(3.81–4.69)	(3.66–4.92)	(3.75–4.87)
LDL cholesterol (mmol/L)	2.29	2.27	2.29
	(1.82–2.72)	(1.83–3.01)	(1.83–2.86)
HDL cholesterol (mmol/L)	1.28	1.23	1.25
	(1.04–1.65)	(0.98–1.47)	(1.04–1.54)
Non-HDL cholesterol (mmol/L)	2.78 (2.41–3.32)	3.05 (2.56–3.83)	2.97 (2.46–3.65)
TC/HDL*	3.10 (2.60–3.80)	3.75 (3.25–4.40)	3.50 (2.90–4.20)
TG (mmol/L)*	1.21 (1.05–1.71)	1.76 (1.28–2.47)	1.39 (1.07–2.16)
AIP*	−0.124	0.248	0.028
	(−0.190 to 0.071)	(−0.101 to 0.359)	(−0.151 to 0.321)
C-reactive protein (mg/L)	0.80 (0.30–2.80)	0.70 (0.40–1.20)	0.70 (0.30–1.40)
Apo B (g/L)	0.79 (0.68–0.97)	0.83 (0.71–1.04)	0.81 (0.69–1.00)
LDL/apoB	2.97 (2.89–3.37)	2.91 (2.41–3.17)	2.96 (2.79–3.17)

Notes: Data shown are median (quartile 1 – quartile 3) unless otherwise indicated.

BMI, body mass index; TDF, tenofovir disoproxil fumarate; FTC, emtricitabine; 3TC, lamivudine; PI, protease inhibitor; eGFR, estimated glomerular filtration rate by MDRD equation; UACR, urine albumin to creatinine ratio; LDL, low-density lipoprotein; HDL, high-density lipoprotein; TC, total cholesterol; TG, triglycerides; AIP, atherogenic index of plasma (log[TG/HDL]); apoB, apolipoprotein B.

* p < 0.05 for comparison between arms.

Efficacy results

In the control arm, six subjects discontinued study treatment early: two withdrew for reasons unrelated to study medications, one was lost to follow-up, one discontinued ATV/r due to diarrhea, and two discontinued TDF due to suspected renal toxicity (one with increasing creatinine, decreasing eGFR, and hypophosphatemia; and one due to granular casts in urine and increasing UACR). In the switch arm, two subjects discontinued early: one withdrew and one experienced virologic failure (VL >200 copies/mL at week 16; no genotypic resistance; VL resuppressed to <40 copies/mL after ATV/r was restarted). At week 48, the proportions of subjects without treatment failure were 76% (19/25) (95% confidence interval [CI] 59–93%) in the control arm and 92% (23/25) (95% CI 81–100%) in the switch arm (intent to treat, missing = failure; p = 0.25). By on-treatment analysis, the proportions with treatment success were 86% (19/22) (95% CI 72–100%) in the control arm and 96% (23/24) (95% CI 88–100%) in the switch arm (p = 0.34).

Pharmacokinetic results

ATV levels were obtained a median of 64 days (quartile 1 – quartile 3 [IQR] 60–75) after baseline in 12 control and 24 switch participants. The remaining subjects did not consent to drug level testing, mainly for reasons of inconvenience (e.g. work schedule or distance from study clinic). Samples were drawn a median of 24 h (range 16.4–28 h) after the previous ATV dose. ATV trough levels were higher in the control arm (median 438 ng/mL [IQR 311–732]) than in the switch arm (median 124 ng/mL [IQR 71–259]) (p = 0.003). ATV levels were <50 ng/mL in 1/12 (8%) subjects in the control arm and 5/24 (21%) in the switch arm, and <150 ng/mL in 3/12 (25%) subjects in the control arm and 14/24 (58%) in the switch arm. The subject in the switch arm who experienced virologic failure at week 16 had an ATV trough level of 131 ng/mL at week 8.

Safety results

Clinical adverse events

The only drug-related clinical adverse event that was at least moderate in severity was diarrhea that led to discontinuation in one patient in the control arm. No significant drug-related clinical adverse events were observed in the switch arm.

Bilirubin

A significant decrease in total bilirubin was observed between baseline and week 48 in the switch arm but not in the control arm (Table 2). Therefore, at the end of the study, total bilirubin was higher in the control arm (median 38 μmol/L [IQR 29–49]) than in the switch arm (median 28 μmol/L, [IQR 21–36]) (p = 0.02) (Table 3). The adult reference interval for total bilirubin is <20 μmol/L.

Table 2 Change in laboratory parameters from baseline to week 48

	Control arm (ATV/r) (N = 19)		Switch arm (unboosted ATV) (N = 23)
	Change	P value	Change	P value
CD4 absolute (cells/mm^3)	90	0.06	40	0.68
	(−30 to 200)		(−90 to 100)
CD4 fraction	0.11	0.013	0.12	<0.001
	(−0.02 to 0.26)		(0.06 to 0.19)
Total bilirubin (μmol/L)	−5	0.73	−18.5	0.002
	(−19 to 20)		(−33.0 to 4.0)
Creatinine (μmol/L)	7	0.004	0	0.19
	(3 to 10)		(−2 to 7)
eGFR (mL/min)	−6	0.007	0	0.36
	(−13 to −3)		(−7.5 to 3)
Phosphorus (mmol/L)	−0.04	0.96	0.00	0.99
	(−0.16 to 0.17)		(−0.09 to 0.08)
UACR (mg/mmol)	0.00	0.95	0.00	0.57
	(−0.70 to 0.40)		(−0.80 to 0.40)
Fasting glucose (mmol/L)	−0.10	0.18	0.20	0.30
	(−1.00 to 0.40)		(−0.50 to 1.10)
Total cholesterol (mmol/L)	0.01	0.42	−0.16	0.27
	(−0.12 to 0.48)		(−0.67 to 0.30)
LDL cholesterol (mmol/L)	0.13	0.59	−0.11	0.90
	(−0.14 to 0.29)		(−0.56 to 0.60)
HDL cholesterol (mmol/L)	0.06	0.30	0.07	0.55
	(−0.04 to 0.26)		(−0.21 to 0.24)
Non-HDL cholesterol (mmol/L)	0.00	0.41	−0.35	0.07
	(−0.15 to 0.40)		(−0.77 to 0.20)
TC/HDL	0.00	0.97	−0.50	0.08
	(−0.20 to 0.20)		(−0.90 to 0.08)
TG (mmol/L)	0.01	0.94	−0.90	<0.001
	(−0.34 to 0.37)		(−1.42 to −0.49)
AIP	0.02	0.82	−0.30	<0.001
	(−0.19 to 0.13)		(−0.42 to −0.12)
C-reactive protein (mg/L)	0.05	0.999	0.10	0.27
	(0.00 to 0.70)		(−0.30 to −0.70)
ApoB (g/L)	0.07	0.07	−0.04	0.41
	(−0.06 to 0.14)		(−0.20 to 0.12)
LDL/apoB	−0.17	0.21	0.22	0.08
	(−0.39 to 0.09)		(−0.11 to 0.55)

Notes: Data shown are median (quartile 1 – quartile 3) changes. P values are shown for within-arm changes between baseline and week 48. Bold font indicates significant p values (<0.05).

ATV, atazanavir; ATV/r, atazanavir/ritonavir; eGFR, estimated glomerular filtration rate; UACR, urine albumin to creatinine ratio; LDL, low-density lipoprotein; HDL, high-density lipoprotein; TC, total cholesterol; TG, triglycerides; AIP, atherogenic index of plasma (log[TG/HDL]); apoB, apolipoprotein B.

Table 3 Laboratory parameters at week 48

	Control arm (N = 19)	Switch arm (N = 23)	P value for comparison between arms
CD4 absolute (cells/μL)	650	515	0.19
	(530–750)	(460–710)
CD4 fraction	0.33 (0.28–0.41)	0.33 (0.26–0.37)	0.77
Total bilirubin (μmol/L)	38	28	0.02
	(29–49)	(21–36)
Creatinine (μmol/L)	81 (72–91)	76 (68–84)	0.125
eGFR (mL/min)	85 (73–90)	96 (83–106)	0.035
Phosphorus (mmol/L)	0.88	1.02	0.29
	(0.77–1.15)	(0.91–1.10)
UACR (mg/mmol)	0.90	1.00	0.82
	(0.50–2.40)	(0.50–1.90)
Fasting glucose (mmol/L)	5.00	5.30	0.12
	(4.60–5.30)	(4.80–5.80)
Total cholesterol (mmol/L)	4.17	3.99	0.33
	(3.79–5.18)	(3.57–4.75)
LDL cholesterol (mmol/L)	2.42	2.30	0.925
	(1.67–2.87)	(1.78–2.95)
HDL cholesterol (mmol/L)	1.33	1.19	0.15
	(1.13–1.84)	(0.93–1.46)
Non-HDL cholesterol (mmol/L)	2.89	2.84	0.735
	(2.37–3.65)	(2.14–3.38)
TC/HDL	3.40 (2.70–3.80)	3.40 (2.60–4.30)	0.57
TG (mmol/L)	1.49 (0.88–1.70)	0.88 (0.73–1.13)	0.17
AIP	0.003	−0.105	0.62
	(−0.399 to 0.125)	(−0.237 to 0.027)
C-reactive protein (mg/L)	1.00	0.90	0.43
	(0.70–1.80)	(0.50–1.40)
Apo B (g/L)	0.79 (0.63–1.01)	0.81 (0.59–0.93)	0.83
LDL/apoB	2.84 (2.68–3.12)	3.05 (2.97–3.27)	0.16

Notes: Data shown are median (quartile 1 – quartile 3). Bold font indicates significant p values (<0.05).

eGFR, estimated glomerular filtration rate by MDRD equation; UACR, urine albumin to creatinine ratio; LDL, low-density lipoprotein; HDL, high-density lipoprotein; TC, total cholesterol; TG, triglycerides; AIP, atherogenic index of plasma (log[TG/HDL]); apoB, apolipoprotein B.

Renal parameters

Serum creatinine increased and eGFR decreased significantly from baseline to week 48 in the control arm, while neither changed significantly in the switch arm (Table 2). No significant changes in serum phosphorus or UACR were observed in either arm. At week 48, eGFR was lower in the control arm (median 85 mL/min [IQR 73–90]) than in the switch arm (median 96 mL/min [IQR 83–106]) (p = 0.035), but the arms did not differ with respect to creatinine, phosphorus, or UACR (Table 3).

Metabolic parameters

No significant changes were observed in either arm with respect to fasting glucose, LDL cholesterol, HDL cholesterol, apoB, LDL/apoB, or hsCRP (Table 2). TC, non-HDL cholesterol, TC/HDL, TG and AIP decreased significantly in the switch but not in the control arm at week 24 (data not shown); however, the observed decreases in the switch arm did not maintain statistical significance at week 48 with the exception of TG and AIP. At week 48, the two arms were similar with respect to fasting glucose, hsCRP, and all lipid parameters (Table 3). During the study, no subjects changed or added any hypoglycemic or lipid-lowering medications.

Discussion

In this 48-week randomized controlled trial of virologically suppressed subjects receiving ATV/r with TDF and either lamivudine or emtricitabine, virologic suppression was maintained in 76% who continued ATV/r and 92% who switched to unboosted ATV (p = 0.25). As expected, both trough ATV levels (at a median of 9 weeks) and bilirubin levels (at week 48) were lower among subjects randomized to unboosted ATV than among those who continued ATV/r. Serum creatinine increased and eGFR decreased in the control arm, while remaining stable in the switch arm. Significant changes were not seen in either arm with respect to fasting glucose, hsCRP, or lipid parameters, with the exception of favorable changes in TG and AIP in the switch arm only.

Previously, a switch to unboosted ATV was shown to be virologically non-inferior to remaining on ATV/r in studies using non-TDF backbones including abacavir/lamivudine.^3–5 In our study using a TDF backbone, one virologic failure occurred in the ATV unboosting arm, while three subjects who continued ATV/r subsequently changed therapy because of intolerance or toxicity.

In this study as in earlier studies,^3–5 unboosting ATV was associated with less hyperbilirubinemia than remaining on ATV/r. This is not surprising, as hyperbilirubinemia is associated with higher plasma ATV levels,^{6, 21–23} and these would be expected to be lower in the absence of ritonavir, as was the case in our study. The magnitude of bilirubin change observed in the switch arm would be considered clinically significant: median bilirubin levels decreased from moderately elevated (Grade 2) to mildly elevated (Grade 1) in the switch arm, while remaining moderately elevated (Grade 2) in the control arm.²⁴

Despite the lower ATV trough levels observed in the switch arm of our study, only one subject in that arm experienced virologic failure, without emergence of ATV resistance. Data from previous clinical studies indicate that TDF has a negligible effect on ATV trough levels even in the absence of ritonavir boosting.^{25, 26} In addition, the generally accepted target ATV trough level of >150 ng/mL was not established among patients like those in our study, who are already virologically suppressed and have no or minimal background NRTI or PI resistance.^{19, 20, 25} In patients without PI resistance, like those in our study, previous investigators have shown a lack of association of ATV levels with virologic efficacy.²³ Our observation of good virologic efficacy despite ATV trough levels below 150 ng/mL has also been described in other studies,^{22, 27} suggesting this target trough level may not be applicable to all patient populations.

Other factors may exert a greater impact on ATV plasma levels than the drug interaction with TDF. Clearly adherence is a critical consideration. While adherence was not formally assessed, it was expected that these preselected, virologically suppressed patients would continue to be highly adherent during the study. Adherence would not be expected to differ between arms, since both pill burden and dosing frequency remained the same (i.e. both ATV 300 mg/ritonavir 100 mg and unboosted ATV 400 mg are given as two pills once daily). This assumption was confirmed to be correct in a post hoc analysis of prescription refill records (using methodology as previously described²⁸) among subjects who were on study medications for at least 24 weeks: 92% of subjects in the switch arm (22/24) and 95% in the control arm (20/21) were 90% adherent or better (data not shown). Thus, adherence is not likely to explain the observed differences in ATV plasma levels between the two arms.

Genetic polymorphisms have been shown to significantly influence ATV pharmacokinetics, their diverse distribution no doubt contributing to the wide interpatient variability observed in ATV plasma levels and the incidence of hyperbilirubinemia.^{21, 27, 29} In a pilot randomized controlled study of patients switching from ATV/r to unboosted ATV with TDF/emtricitabine, adjusting the ATV dosing regimen according to the patient’s genetic profile resulted in higher ATV concentrations, but did not affect safety or virologic outcomes at week 48.⁷ Further research needs to be done in the area of pharmacogenetic-based ATV dosing, but at the present time, human genetic testing (aside from HLA-B*5701 allele testing) is not widely available to guide clinical management.

A potential benefit of the lower ATV plasma levels when ritonavir is discontinued might be an amelioration renal toxicity, which may be related to ATV crystallization in the kidneys.^{11, 12, 30, 31} However, the association between ATV plasma levels and renal toxicity is less clear than for hyperbilirubinemia.³² Unfortunately, plasma ATV levels were not available for either of the two subjects in the control arm of our study who discontinued study treatment early due to suspected nephrotoxicity, so we are unable to comment on the association between ATV levels and renal toxicity based on our study results.

On the other hand, ATV levels may not be the main contributing factor to renal toxicity among patients also taking TDF. In the ARIES study using an abacavir backbone, eGFR did not change over 108 weeks in either the ATV/r or the unboosted ATV arm.⁴ This suggests that the favorable eGFR changes seen in the switch arm of our study (i.e. eGFR stabilization as compared to worsening in the ATV/r arm) may have been related to an interaction between ATV/r and TDF, rather than to unboosting ATV alone. In ASSURE, as in our study, eGFR decreased (although not significantly in ASSURE) in subjects who remained on TDF/emtricitabine /ATV/r,⁵ again suggesting a negative renal impact of continued exposure to the combination of TDF and ATV/r. This is consistent with our finding of nephrotoxicity in two subjects randomized to continue ATV/r, but none of those who switched to unboosted ATV. While renal toxicity is not clearly associated with ATV plasma levels, it has been clearly associated with higher tenofovir plasma levels.^33–35 Patients receiving TDF with ritonavir-boosted PIs have higher tenofovir exposure (as compared to TDF with nonnucleoside reverse transcriptase inhibitors or integrase inhibitors),³⁶ and have been shown, in large cohort studies, to be at the greatest risk of renal dysfunction.^{11, 12} This provides a potential explanation for our finding of improvements in renal function markers after switching from ritonavir-boosted ATV to unboosted ATV; however, in a cross-sectional pharmacokinetic study of HIV-infected patients receiving TDF, unboosted ATV was associated with higher tenofovir concentrations than boosted ATV, although the mechanism of this effect is unclear.³⁶ Overall, while we observed an improvement in renal function among our study subjects who switched to unboosted ATV as compared to those who remained on ATV/r, it remains uncertain whether this is related to lower levels of ATV, lower levels of tenofovir, or some other etiology.

In the ARIES and Induma studies with non-TDF backbones, unboosting ATV was associated with significant improvements in fasting TC and TG,^{3, 4} TC/HDL,⁴ and non-HDL cholesterol.³ In our study, decreases in non-HDL cholesterol and TC/HDL (p < 0.1 for both) and significant decreases TG and AIP were seen in the switch arm, but not in the control arm. AIP is a useful biomarker of cardiovascular risk because it correlates closely with small, atherogenic LDL and HDL particles.¹⁵ However, the clinical significance of the observed decreases in TC/HDL, TG, and AIP is unclear, given that those parameters were higher in the switch arm at baseline. As in previous studies examining the ATV unboosting strategy, we did not observe a beneficial effect on hsCRP.^{5, 37}

Despite being a randomized controlled trial, the conclusions of this study are limited by its small sample size and open-label design. The interpretation of the ATV plasma level results is hampered by missing data for almost one-third of participants. In addition, UACR is not a good marker of tubular toxicity, so the etiology of the reduction in renal function observed in the control arm is unclear. Furthermore, since all subjects in the study received TDF, the results cannot be extrapolated to regimens including the newer formulation, tenofovir alafenamide (TAF). Limited data are available regarding potential drug interactions between TAF and ATV.³⁸

In summary, unboosting ATV in TDF-containing regimens appears to be safe and effective for at least one year in selected virologically suppressed patients, and may have favorable effects on bilirubin, some lipid parameters, and renal function.

Disclosure statement

Marianne Harris has received grants or research support from AbbVie, Bristol-Myers Squibb Canada, and Gilead Sciences Canada Inc.; and has received honoraria for advisory boards, similar committees, and speaking engagements from Gilead Sciences Canada Inc., Merck Canada Inc., and ViiV Healthcare.

Mark Hull has received grants or research support from AbbVie, Bristol-Myers Squibb Canada, and Gilead Sciences Canada Inc.; and has received honoraria for speaking engagements and/or consultancy meetings from Bristol-Myers Squibb Canada, Gilead Sciences Canada Inc., Merck Canada Inc., Ortho-Janssen, Pfizer, and ViiV Healthcare.

Silvia A. Guillemi has received honoraria for advisory boards, similar committees, and speaking engagements from ViiV Healthcare; Bristol-Myers Squibb (Canada); Abbvie; Gilead Sciences Canada Inc.; Janssen; and Merck Canada Inc.

P. Richard Harrigan has received grant support and/or served as a consultant for Merck Canada Inc., Gilead Sciences Canada Inc., and ViiV Healthcare.

The other authors have no potential conflicts of interest to disclose.

Notes on contributors

Marianne Harris is the clinical research advisor in the British Columbia Centre for Excellence in HIV/AIDS; clinical assistant professor in the Faculty of Medicine, Department of Family Practice, University of British Columbia; and associate member of the Division of AIDS, Department of Medicine, University of British Columbia. Her primary research interests are HIV clinical trials and complications of antiretroviral therapy.

Bruce Ganase is a clinical trials coordinator in the AIDS Research Program, a collaboration between St. Paul's Hospital and the University of British Columbia.

Birgit Watson is a research assistant in the Laboratory Program at the British Columbia Centre for Excellence in HIV/AIDS. Her research interests include the pharmacokinetics of antiretroviral drugs.

Mark W Hull is a research scientist at the British Columbia Centre for Excellence in HIV/AIDS and clinical associate professor in the Division of AIDS, Department of Medicine, University of British Columbia. His research interests include HIV/hepatitis C co-infection.

Silvia A Guillemi is a director of Clinical Education at the British Columbia Centre for Excellence in HIV/AIDS; clinical assistant professor in the Faculty of Medicine, Department of Family Practice, University of British Columbia; and associate member of the Division of AIDS, Department of Medicine, University of British Columbia. Her research interests include co-morbidities in HIV (osteoporosis, metabolic disorders, neurocognitive impairment, and others) and antiretroviral management.

Wendy Zhang is a biostatistician in the British Columbia Centre for Excellence in HIV/AIDS, providing statistical analysis support for the Drug Treatment Program.

Ramesh Saeedi is a medical biochemist and clinical assistant professor in the Department of Pathology & Laboratory Medicine, University of British Columbia. Her research interests are lipid disorders and their relation to cardiovascular disease.

P Richard Harrigan is adirector of the Laboratory Program at the British Columbia Centre for Excellence in HIV/AIDS; associate professor in the Department of Medicine at the University of British Columbia; and member of the Division of AIDS, Department of Medicine, University of British Columbia. His research focuses primarily on HIV drug efficacy, drug resistance, human and viral parameters that influence HIV disease progression.

Acknowledgements

The authors would like to thank Sean Ling and Benita Yip for assistance in the preparation of this manuscript.