Impact of Genetic Variants in ABCG2, NR1I2 and UGT1A1 on the Pharmacokinetics of Dolutegravir in Children

Abstract

Background:

Dolutegravir plasma concentrations and pharmacokinetic (PK) parameters in children display considerable variability. Here, the impact of genetic variants in ABCG2 421C>A (rs2231142), NR1I2 63396 C>T (rs2472677) and UGT1A1 (rs5839491) on dolutegravir PK was examined.

Methods:

Children defined by age and administered dolutegravir formulation had AUC₂₄ at steady state, C_max and C_24h determined. Associations between genetic variants and PK parameters were assessed using the dominant inheritance model.

Results:

The 59 children studied had a median age of 4.6 years, log₁₀ plasma HIV RNA of 4.79 (copies/mm³) and CD4+ lymphocyte count 1,041 cells/mm³; 51% were female. For ABCG2, participants with ≥1 minor allele had lower adjusted mean AUC difference (hr*mg/L) controlling for weight at entry, cohort and sex, (−15.7, 95% CI: [−32.0, 0.6], p = 0.06) and log₁₀C_max adjusted mean difference (−0.15 (95% CI: [−0.25, −0.05], p = 0.003). Participants with ≥1 minor allele had higher adjusted mean AUC difference (11.9, 95% CI: [−1.1, 25.0], p = 0.07). For UGT1A1, poor metabolizers had non-significant higher concentrations (adjusted log₁₀C_max mean difference 11.8; 95% CI: [−12.3, 36.0], p = 0.34) and lower mean log₁₀ adjusted oral clearance −0.13 L/hr (95% CI: [−0.3, 0.06], p = 0.16). No association was identified between time-averaged AUC differences by genotype for adverse events, plasma HIV RNA or CD4+ cell counts.

Conclusions:

Dolutegravir AUC₂₄ for genetic variants in ABCG2, NR1l2 and UGT1A1 varied from −25% to +33%.These findings help to explain some of the variable pharmacokinetics identified with dolutegravir in children.

Dolutegravir is an integrase strand transfer inhibitor that can be taken once daily orally by persons living with human immunodeficiency virus type-1 (HIV-1) ^1–3. Dolutegravir is a particularly attractive drug for use in young children where once daily, age-appropriate well-tolerated oral formulations extending down to infancy are limited and is the World Health Organization preferred option for first and second line treatment in children ⁴. The International Maternal Pediatric Adolescent AIDS Clinical Trials (IMPAACT) P1093 protocol (P1093) was designed to evaluate the safety, tolerability, pharmacokinetics (PK) and antiviral efficacy of dolutegravir when given in combination with an optimized background regimen to children 4 weeks to 18 years of age living with HIV-1 ^5–7. In adolescents 12 to <18 years, a dose of 1 mg/kg/day was found to give a PK profile similar to that observed in adults and enabled the drug to be licensed in children 12 years and older simultaneously with adults ⁷. Additional studies as part of the P1093 study established the safety, PK and excellent virologic suppression of dolutegravir in children when combined with an optimized background of other antiretroviral drugs that led to FDA and EMA regulatory approval down to 4 weeks of age ^5–7.

PK and pharmacodynamic studies of dolutegravir have shown that it is primarily metabolized via UDP glucuronosyltransferase family 1 member A1 (UGT1A1) and cytochrome P450 3A4 (CYP3A4). It is also a substrate for ABCB1 (also called breast cancer resistance protein; BCRP) and P-glycoprotein (P-gp) ^8–11. Elliot et al. examined the pharmacogenetic influence of variants in CYP3A4, ABCG2, UGT1A1 and NR1I2 (human pregnane X receptor, PXR) in adults receiving dolutegravir ¹². They found a significant increase in dolutegravir exposure in carriers of UGT1A1*28 associated with poor dolutegravir metabolism. Additionally, ABCG2 421C>A was associated with a 28% increase in dolutegravir C_max, and NR1I2 63396 C>T was associated with a 24% increase in dolutegravir C_max and 27% increase in AUC₂₄. Pharmacogenetic data of antiretrovirals are limited in children and to our knowledge, no studies have examined the pharmacogenetics of dolutegravir in pediatric populations. Examination of the genetic influences on drug PK in children is of particular importance since the activity of UGT1A1 and efflux transporters can vary with age. Here, we investigated the influence of genetic variants in UGT1A1, ABCG2 and NR1I2 on dolutegravir PK in children who participated in the P1093 study.

METHODS

Study Population:

The participants in this study are a subset of children participating in IMPAACT P1093 consisting of cohorts defined by age and dolutegravir formulations ⁵ (Supplemental Table S1). Each cohort was enrolled in two stages. In Stage I, participants underwent intensive PK sampling and were monitored for the safety and tolerability of dolutegravir for 4 weeks. Once a treatment dose was accepted, additional participants were enrolled (Stage II) and all were followed for evaluation of safety, tolerability and efficacy. The pharmacogenetic analysis only included participants from Stage 1 enrollment as only they had the intensive PK data necessary for the generation of steady state PK parameters. Children whose parents refused genetic testing and those for whom insufficient peripheral blood mononuclear cells were available were excluded from the study. All remaining P1093 participants with genomic data receiving dolutegravir film coated tablets, dispersible and granules are included in the analyses.

Pharmacokinetics:

Participants in this analysis had blood sampled pre-dose and at 1, 2, 3, 4, 6, 8 and 24 hours post-dosing during a visit 5–10 days after starting dolutegravir. Dolutegravir plasma concentrations were determined using liquid chromatography with tandem mass spectrometry, as previously described ¹³. Week 2 included steady-state intensive PK sampling to determine the 24- hour area-under-the-curve (AUC₂₄, hr*mg/L), maximum concentration (C_max, ng/mL) 24 hour trough concentration (C_24h, ng/mL), and oral clearance (CL/F). These dolutegravir steady-state pharmacokinetic parameters were calculated using Phoenix WinNonlin version 8·3 (Certara USA, Inc., Princeton, New Jersey, USA) and were performed in real-time at the IMPAACT Pharmacology Support Laboratory at the University of Alabama at Birmingham ⁵.

Methods for genotyping:

SNP determinations for ABCG2 421 C>A (rs2231142) and NR1I2 63396 C>T (rs2472677) were performed using the rhAmp SNP genotyping system (Integrated DNA Technologies) that uses a two enzyme system coupled with RNA-DNA primers to interrogate target SNPs. For rs2231142, allele primer 1: rhAmp-F/CTCTGACGGTGAGAGAAAACTTA- CrAGTTC/GT1; allele primer 2: rhAmp-Y/ CTCTGACGGTGAGAGAAAACTTACrAGTTC/GT1; and locus primer: GCCCTTGGAGTCTGCCACTTTATrCCAGA/GT3. For rs2472677, allele primer 1: rhAmp-F/TCAACTTTTTTGTGCCATATTTTTTCrUGATT/GT4; allele primer 2: rhAmp-Y/ TCAACTTTTTTGTGCCATATTTTTTCrUGATT/GT4; and locus primer: GCCATATTACATTCGGAAGACTTATTCTATTrCCTGT/GT3. Polymorphisms in UGT1A1 (rs5839491) were determined by real-time polymerase chain reaction as previously described ¹⁴.

Genotype Data:

Samples for all eligible participants were genotyped successfully for genetic variants ABCG2 421 C>A (rs2231142), NR1I2 63396 C>T (rs2472677), and UGT1A1 (rs5839491). Genetic variants for ABCG2 and NR1I2 are standard single nucleotide polymorphisms (SNPs) whereas UGT1A1 is coded in terms of TA repeats, with each participant having two records for number of repeats (e.g., 5/6). UGT1A1 was mapped and grouped using these levels of metabolizers: (1) Normal (extensive) metabolizers, which included *1/*1, *1/*36, and *36/*36, (2) Intermediate metabolizers, which included *1/*28, *1/*37, *36/*28, *36/*37, and *1/*6, and (3) Poor metabolizers, which included *28/*28, *28/*37, *37/*37, *6/*6. (*36 corresponds to five repeats, *1 corresponds to six, *28 corresponds to seven, and *37 corresponds to eight. Additionally, *36 leads to increased promoter activity, *1 is the dominant allele, and *28 and *37 leads to decreased promoter activity.)

HIV-1 RNA and CD4+ cell counts:

Time-averaged AUCs for HIV-1 RNA and CD4+ cell counts were calculated using the trapezoidal rule for each participant through study week 24 for log₁₀ HIV RNA-load (collected at weeks 0, 2, 4, 12, 24) and CD4+ cell counts (collected at weeks 0, 12 and 24). Differences in time averaged AUC for HIV-1 RNA and CD4+ by genotype levels were tested using a 2-sample t-test with unequal variances. Sustained viral suppression was defined as a viral load <400 copies/mL at week 2 and with all subsequent viral load measurements at <400 copies/mL through week 24, were summarized with counts and percentages, and tested using an unadjusted Poisson regression with a robust variance estimate.

Adverse events:

All new adverse events of grade 3 or higher up to 24 weeks were used as the outcome measure. Adjusted and unadjusted incidence rate ratios were used to compare genotype levels and estimated using a Poisson regression with a dispersion parameter and time offset. Adjustment variables included cohort, weight, and sex.

Statistical Methods:

Baseline refers to measurements collected before or on the date of study entry, except for CD4+ cell counts and CDC staging, which included measurements up to two days after study entry. PK parameters were estimated from intensive PK samples collected at 5–10 days after study drug initiation. Unadjusted and adjusted mean differences for AUC₂₄ (μg*hr/mL), log₁₀ C_max (μg/mL), oral clearance at steady-state (L/hr) and log₁₀ C_min (μg/mL) by the genetic groups were estimated using linear regression. Adjustment variables included cohort, weight, and sex. ABCG2 421 C>A (rs2231142) and NR1I2 63396 C>T (rs2472677) genetic predictors were modeled with the dominant inheritance model, with the major allele as the reference group. The dominant model was selected to replicate the methods of the study by Elliot et al ⁹. In that paper different inheritance models were considered with the dominant model presented. We only considered the dominant model since our sample size is small making model selection difficult.

Analyses were based on participants with available data (complete case analysis) which assumes missing data are missing completely at random. No adjustments were made for multiple testing. Analyses were conducted with SAS 9.4.

RESULTS

Baseline Characteristics:

The baseline characteristics of the 59 children whose families consented to genetic analysis and for whom PBMC were available for pharmacogenetic analysis and the 68 children who were in P1093 but not part of this analysis are shown in Supplemental Table S2. Of the 59 children in the pharmacogenetic study, 30 (51%) were females and had a median (Q1, Q3) age of 4.6 (1.6, 10.0) years. Children from the United States comprised 21 (36%) of participants with the remainder of the children being from countries in Africa, Thailand and Brazil. The median (Q1, Q3) log₁₀ plasma HIV RNA was 4.79 (4.15, 5.31) and the median (Q1, Q3) CD4+ lymphocyte percentage was 23 (18, 31). Children had been on another antiretroviral regimen a median (Q1, Q3) of 2.8 (1.2, 6.4) years prior to study entry.

Allelic frequencies:

For ABCG2 the major allele C and minor allele A frequencies were 88.1% and 11.9%, respectively, while for NR1I2 the major allele frequency C was 60.2%, and the minor allele frequency T was 39.8% (Table 1A). Of the 59 participants genotyped for UGT1A1, 38 (64.4%) were normal (extensive) metabolizers, 16 (27.1%) intermediate metabolizers and 5 (8.5%) poor metabolizers (Table 1B).

Table 1.

Major and minor allelic frequencies for ABCG2 and NR1I2, and distribution UGT1A1 variants by metabolism levels.

A. Major and minor allelic frequencies for ABCG2 and NR1I2
Gene	Total Number of Alleles	Major Allele	Number of Major Alleles	Major Allele Frequency	Minor Allele	Number of Minor Alleles	Minor Allele Frequency
ABCG2	118	C	104	88.1%	A	14	11.9%
NR1I2	118	C	71	60.2%	T	47	39.8%

B: Frequencies of UGT1A1 variants by metabolism levels.
Metabolism Level	Type of Metabolizer	n	Percent
Normal	*1/*1	25	42.4%
	*1/*36	4	6.8%
	*36/*36	9	15.3%
Intermediate	*1/*28	16	27.1%
Poor	*28/*28	5	8.5%

Association of pharmacokinetics with ABCG2, NR1I2 and UGT1A1 genotypes:

ABCG2 421 C>A (rs2231142): Applying the predetermined dominant model, for ABCG2 the unadjusted mean AUC₂₄ difference was −12.80 (95% Confidence Interval (CI): [−28.29, 2.69], p-value = 0.11); controlling for weight at entry, cohort and sex, the mean adjusted difference was −15.69 (95% CI: [−31.97, 0.58], p-value = 0.06) (Table 2). The log₁₀C_max showed the greatest difference for the ABCG2 genotypes with the unadjusted mean C_max difference being −0.13 (95% CI: [−0.22, −0.03], p = 0.010); and the adjusted mean difference was −0.15 (95% CI: [−0.25, −0.05], p = 0.003). Consistent with the finding for log₁₀C_max for ABCG2, the unadjusted mean log₁₀C_min difference was −0.14 (95% CI: [−0.34, 0.06], p = 0.17); and the adjusted mean difference was −0.17 (95% CI: [−0.36, 0.02], p = 0.09). Additionally, the mean difference in log₁₀ adjusted oral clearance was 0.10 L/hr (adjusted 95% CI: [−0.03, .23]; p = 0.13) greater for those with the A allele. Findings when using a Co-Dominant Model are shown in Supplemental Table S3.

Table 2:

Week 2 AUC₂₄, log₁₀C_max, log₁₀C_min and log₁₀ Clearance. Mean differences by ABCG2 and NR1I2 genotype (Dominant Model). Adjustment variables included drug formulation cohort, weight, and sex.

							Regression Model
				Unadjusted			Adjusted
Genotype	PK Parameter	Gene Levels	n	Mean Difference	CI	P-value	Mean Difference	CI	P-value
ABCG2	AUC24(μg*h/mL)	C/C	46
		A/A or C/A	12	−12.80	(−28.29, 2.69)	0.11	−15.69	(−31.97, 0.58)	0.06
	Cmax(μg/mL)	C/C	46
		A/A or C/A	12	−0.13	(−0.22, −0.03)	0.010	−0.15	(−0.25, −0.05)	0.003
	Cmin(μg/mL)	C/C	46
		A/A or C/A	12	−0.14	(−0.34, 0.06)	0.17	−0.17	(−0.36, 0.02)	0.09
	log Clearance (L/hr)	C/C	47
		A/A or C/A	12	0.06	(−0.14, 0.26)	0.54	0.10	(−0.03, 0.23)	0.13
NR1I2	AUC24(μg*h/mL)	C/C	21
		T/T or C/T	37	11.63	(−1.38, 24.64)	0.08	11.92	(−1.14, 24.99)	0.07
	Cmax(μg/mL)	C/C	21
		T/T or C/T	37	0.05	(−0.04, 0.13)	0.27	0.05	(−0.03, 0.14)	0.22
	Cmin(μg/mL)	C/C	21
		T/T or C/T	37	0.11	(−0.06, 0.28)	0.20	0.11	(−0.04, 0.27)	0.16
	log Clearance (L/hr)	C/C	22
		T/T or C/T	37	−0.03	(−0.19, 0.14)	0.77	−0.08	(−0.18, 0.02)	0.14

NR1I2 63396 C>T (rs2472677):

For NR1I2 applying the dominant model, the unadjusted mean AUC₂₄ difference was 11.63 (95% CI: [−1.38, 24.64], p-value = 0.08); and the mean adjusted difference was 11.92 (95% CI: [−1.14, 24.99], p = 0.07) (Table 2). Although the log₁₀C_max and log₁₀C_min were in the same direction there was no significant difference. The unadjusted mean log₁₀C_max difference was 0.05 (95% CI: [−0.04, 0.13], p = 0.27); and the adjusted mean difference was 0.05 (95% CI: [−0.03, 0.14], p = 0.22). Similarly for log₁₀C_min the unadjusted mean difference was 0.11 (95% CI: [−0.06, 0.28], p-value = 0.20); and the adjusted mean difference was 0.11 (95% CI: [−0.04, 0.27], p-value = 0.16). Additionally, the mean difference in log₁₀ adjusted oral clearance was −0.08 L/hr (adjusted CI: −0.18, 0.02; p = 0.14) lower for those with the T allele (Table 2). Of note, for SNPs in ABCG2 and NR1I2, the direction and magnitude of the differences were similar to results that averaged the PK outcomes across all weeks (data not shown).

UGT1A1:

Comparing intermediate metabolizers to normal the unadjusted AUC₂₄ (μg*h/mL) was 15.83 less for the intermediate metabolizers (unadjusted CI: −30.01, −1.64; p = 0.020 using the normal (extensive) metabolizers as the reference (Model 1); following adjustment for drug formulation cohort, weight and sex the AUC₂₄ (μg*h/mL) was −18.52 for the intermediate metabolizers (adjusted CI: −34.23, −2.81; p = 0.021). Analyses of poor metabolizers was limited because only 5 study participants were identified in this category. Unexpectedly, the intermediate metabolizers had the lowest AUC₂₄, C_max and C_min (Table 3). However, when the intermediate and normal (extensive) metabolizers are combined (Model 2), as expected the poor metabolizers have the highest drug concentrations with an unadjusted mean difference in AUC₂₄ of 11.97 (95% CI: [−10.68, 34.62], p = 0.30); the adjusted AUC₂₄ mean difference was 11.84 (95% CI: [−12.27, 35.95], p = 0.34). Similarly for log₁₀C_max comparison, the unadjusted mean difference was 0.07 (95% CI: [−0.07, 0.22], p = 0.32); and the adjusted mean difference was 0.07 (95% CI: [−0.08, 0.22], p =0.37), while the unadjusted mean log₁₀C_min difference was 0.23 (95% CI: [−0.06, 0.52], p = 0.11); and the adjusted mean difference was 0.17 (95% CI: [−0.12, 0.45], p = 0.25). Additionally, the mean difference in log₁₀ adjusted oral clearance was −0.13 L/hr (adjusted CI: −-0.32, 0.06; p = 0.16) lower for poor metabolizers (Table 3).

Table 3:

Week 2 AUC₂₄, log₁₀C_max, log₁₀C_min and log₁₀ Clearance. Mean differences by UGT1A1 variants Model 1 using the normal metabolism as the reference. Model 2 uses normal and intermediate as reference. Adjustment variables included drug formulation cohort, weight, and sex.

							Regression Model
				Unadjusted			Adjusted
Genotype	PK Parameter	Gene Levels	n	Mean Difference	CI	P-value	Mean Difference	CI	P-value
UGT1A1	AUC24(μg*h/mL)	Normal	38
Model 1		Intermediate	15	−15.83	(−30.01, −1.64)	0.029	−18.52	(−34.23, −2.81)	0.021
		Poor	5	7.49	(−14.65, 29.63)	0.51	9.48	(−13.68, 32.64)	0.42
	Cmax(μg/mL)	Normal	38
		Intermediate	15	−0.13	(−0.22, −0.04)	0.004	−0.14	(−0.24, −0.04)	0.005
		Poor	5	0.04	(−0.10, 0.17)	0.60	0.05	(−0.09, 0.20)	0.48
	Cmin(μg/mL)	Normal	38
		Intermediate	15	−0.18	(−0.36, −0.00)	0.047	−0.19	(−0.38, −0.00)	0.045
		Poor	5	0.18	(−0.10, 0.46)	0.21	0.14	(−0.13, 0.42)	0.31
	log Clearance (L/hr)	Normal	38
		Intermediate	16	0.06	(−0.12, 0.25)	0.52	0.08	(−0.05, 0.20)	0.23
		Poor	5	0.03	(−0.27, 0.32)	0.85	−0.13	(−0.31, 0.06)	0.19
Model 2	AUC24(μg*h/mL)	Normal/Intermediate	53
		Poor	5	11.97	(−10.68, 34.62)	0.30	11.84	(−12.27, 35.95)	0.34
	Cmax(μg/mL)	Normal/Intermediate	53
		Poor	5	0.07	(−0.07, 0.22)	0.32	0.07	(−0.08, 0.22)	0.37
	Cmin(μg/mL)	Normal/Intermediate	53
		Poor	5	0.23	(−0.06, 0.52)	0.11	0.17	(−0.12, 0.45)	0.25
	log Clearance (L/hr)	Normal/Intermediate	54
		Poor	5	0.01	(−0.28, 0.30)	0.95	−0.13	(−0.32, 0.06)	0.16

Association of genetic variants in ABCG2, NR1I2 and UGT1A1 with virologic suppression and CD4+ cell count:

For the 53 study participants, there was no association between time-averaged AUC differences by genotype for plasma HIV RNA levels or CD4+ cell counts (Table 4 and Supplemental Table S4). Scatter plots of week 2 CD4+ count time-averaged AUC₂₄ differences by genotype are presented in the Supplemental Figure. Eighteen (34%) of the 53 children achieved sustained virologic suppression during the course of the study. Although not statistically significant, those children with genetic variants in ABCG2 and UGT1A1 associated with the highest dolutegravir concentrations were most likely to achieve sustained viral suppression (Table 5 and Supplemental Figure).

Table 4:

Genotype differences for time-averaged AUC₂₄ log₁₀ plasma HIV RNA and CD4+ cell count.

					Time-Average AUC
		Log10 HIV RNA-load				CD4 Count
Genotype	Gene Level	n	Estimate	95% CI	P-value	n	Estimate	95% CI	P-value
ABCG2	C/C	43				46
	A/A or C/A	10	0.00	(−0.46, 0.47)	0.99	9	−86.90	(−787.24, 613.44)	0.79
NR1I2	C/C	20				21
	T/T or C/T	33	0.17	(−0.36, 0.70)	0.51	34	132.59	(−325.13, 590.31)	0.56
UGT1A1 (Model 1)	Intermediate	13				13
	Extensive	36	−0.44	(−0.82, −0.06)	0.024	37	95.19	(−423.18, 613.56)	0.71
	Poor	4	−0.09	(−0.43, 0.25)	0.57	5	124.39	(−653.00, 901.79)	0.73
UGT1A1 (Model 2)	Extensive/Intermediate	49				50
	Poor	4	0.24	(−0.13, 0.60)	0.19	5	53.00	(−685.92, 793.82)	0.86

P-values and 95% Confidence intervals were computed using a two-sample t-test for the differences in mean

Table 5:

Percent of participants with sustained plasma HIV RNA suppression by genotype.

Gene	Genetic variant	% suppressed
ABCG2	C/C	16/43 (37.21%)
	A/A or C/A	2/10 (20.00%)
NR1I2	C/C	7/20 (35.00%)
	T/T or C/T	11/33 (33.33%)
UGT1A1 (Model 1)	Normal	11/36 (30.56%)
	Intermediate	5/13 (38.46%)
	Poor	2/4 (50.00%)
UGT1A1 (Model 2)	Normal/Intermediate	16/49 (32.65%)
	Poor	2/4 (50.00%)

Adverse events:

There were 37 grade 3 adverse events reported among study participants within the 24 weeks of follow-up. There was no statistical difference in the risk of having an adverse event among the different genotypes evaluated (data not shown).

DISCUSSION

Dolutegravir, due to its efficacy, low rate of adverse effects and limited drug-drug interactions is currently recommended for use as first- and second-line treatment for children, adults and pregnant women. Notably, in both the ODYSSEY study and IMPAACT P1093 no child experienced a serious adverse event attributed to dolutegravir. However, despite the efficacy and low rate of serious adverse events, a notable portion of children experienced failure over time, raising concern for a possible role for sub-therapeutic drug concentrations. In the “ODYSSEY” trial conducted predominantly in Africa in children and adolescents dolutegravir-based ART was associated with superior virologic outcomes compared to standard of care (that included, most commonly, efavirenz- or a boosted protease inhibitor-based regimens) ¹⁵. Of note in this study, in the dolutegravir group 13% of participants experienced virologic failure by 96 weeks. The findings of the IMPAACT P1093 PK study in children 4 weeks to <6 years are similar to the ODYSSEY trial. In the IMPAACT study at 48 weeks, overall 85% of participants achieved an HIV-1 viral load <400 copies per mL and 63% achieved a viral load <50 copies per mL. Of note, many children achieved virologic suppression after difficulty with adherence was addressed. Thus, there continues to be suboptimal viral suppression in some children where adherence counseling combined with the use of host genetics might be useful to guide treatment in order to achieve the highest efficacy with the least toxicity.

We identified that the allelic variant NR1I2 63396 C>T (rs2472677) is associated with increased dolutegravir concentrations in children of this cohort. NR1I2 is a nuclear receptor transcription factor that encodes a protein involved with the regulation of the cytochrome P450 gene CYP3A4, binding to the response element of the CYP3A4 promoter ¹⁶. Associated with the induction of CYP3A4 and CYP2B6 promoter activity, NR1I2 variants also have the potential to complicate antiretroviral drug interactions ¹⁶. Additionally, variants in NR1I2 have been associated with the more rapid progression to AIDS in a cohort of adults living with HIV-1 in Brazil ¹⁷.

UGT1A1 is responsible for the glucuronidation of unconjugated bilirubin to conjugated bilirubin and is involved with numerous drug-drug interactions and the metabolism of several integrase strand transfer inhibitors ¹⁸. Although the current study is limited to a small sample, children with UGT1A1 genotypes associated with poor metabolizer status had the highest concentrations of dolutegravir. These findings are consistent with what has been observed in adults ^8,9. Of note, no differences in adverse events were identified between the different genotypes.

We unexpectedly found that lower dolutegravir AUC₂₄, log₁₀C_max and log₁₀C_min was associated with the presence of at least one minor ABCG2 allele. This is the reverse of what was observed in adults. Whereas for adults ABCG2 421 C>A was associated with a 28% increase in dolutegravir C_max in the homozygous variant, we have observed that the presence of the A variant is associated with a 29% mean adjusted decrease. The reason for the disparity between what was observed in adults and what we have found in children is unclear. However, although not statistically significant, the findings are supported by a lower HIV-1 viral load and increased CD4+ lymphocyte count in children being C/C homozygous. The C>A variant in rs2231142 reportedly results in the loss of function of the protein encoded by the ABCG2 gene; the polymorphism results in glutamine replacing a lysine at position 141 which is thought to result in a structurally and functionally defective protein ^19,20. Protein expression of ABCG2 in human liver is lower in children when compared to adults and lowest in the elderly ²¹. Thus, contrary to our findings, the C>A variation would be expected to lead to an increase in dolutegravir concentrations. It is possible that the different drug formulations used in this study were differentially impacted by variants in the ABCG2. However, we found no differences in dolutegravir PK parameters when controlling for cohort which included drug formulation. Additionally, a recent study in adults from South Africa found that the rs2231137 SNP in ABCG2 had no effect on dolutegravir concentrations ²². Thus, even in adults, the impact of rs2231137 on dolutegravir PK is unclear.

This study has a number of limitations that may have impacted our findings including the small sample size, the wide age range in age of the children enrolled and the different dolutegravir drug formulations evaluated as part of this initial PK and safety study. However, adjusting for age, sex and drug formulation did not result in a large change from the original findings suggesting that neither of these variables have an important impact on the effects of genetic variants in ABCG2, NR1I2 and UGT1A1 on dolutegravir PK.

The published IMPAACT P1093 findings on dolutegravir PK in children found that when the data were analyzed either by pre-specified age cohorts or by WHO weight bands using dispersible dolutegravir tablets, the drug concentrations achieved approximated those found to be efficacious in adults and resulted in a similar rate of virologic suppression. When the data generated from the IMPAACT 1093 study are combined with that of the ODYSSEY study, a dosing schema for dolutegravir in children has been established. Dosing bands are available for young children. The dispersible tablet dose of six 5 mg tablets is recommended for children once they reach 20 kg ²³. Additionally, based on the ODYSSEY study, children weighing 20 kg if able to swallow tablets can transition directly to the 50 mg film coated tablet which is 60–80% less bioavailable than the dispersible tablets. The recommendations are based on the excellent safety profile of dolutegravir combined with the high level of virologic suppression observed in both studies. This also allows for considerable variability in dolutegravir concentrations above the therapeutic threshold. For the ABCG2 A/A or C/A and UGT1A1 poor metabolizers, we observed a mean increase in AUC₂₄ of 22% for both, while the NR1I2 T/T or C/T variant resulted in a 25% decrease in AUC₂₄. Given the acceptable variability in AUC₂₄ ranging from 49 to 70 mgxh/L noted in the IMPAACT 1093 study, the extent of variability identified with the genetic variants are similar to those observed for the whole cohort.

In summary, we investigated associations of genetic variants ABCG2 421 C>A, NR1I2 63396 C>T and UGT1A1 with dolutegravir exposure in children and adolescents. The PK parameters of time averaged AUC₂₄, C_max and C_min were derived from intensive PK assessment at week 2 following initiation of treatment. The AUC₂₄ for dolutegravir for SNPs in ABCG2, NR1I2 and UGT1A1 varied from −25% to +33%, but had no significant effects on viral load, CD4+ cells counts and adverse events. These findings help to explain inter-subject variability associated with dolutegravir concentrations in children.