Patient-specific Genetic Factors Predict Treatment Failure in Sofosbuvir-treated Patients With Chronic Hepatitis C

Catrina M. Loucks; Jennifer J. Lin; Jessica N. Trueman; Britt I. Drögemöller; Galen E. B. Wright; Wan-Chun Chang; Kathy H. Li; Eric M. Yoshida; Jo-Ann Ford; Samuel S. Lee; Pam Crotty; Richard B. Kim; Bandar Al-Judaibi; Ute I. Schwarz; Alnoor Ramji; Jeanette F. Farivar; Edward Tam; Lori Lee Walston; Colin J. D. Ross; Bruce C. Carleton

Disclosures

Liver International. 2022;42(4):796-808.

Abstract and Introduction

Abstract

Background & Aims: According to pivotal clinical trials, cure rates for sofosbuvir-based antiviral therapy exceed 96%. Treatment failure is usually assumed to be because of virological resistance-associated substitutions or clinical risk factors, yet the role of patient-specific genetic factors has not been well explored. We determined if patient-specific genetic factors help predict patients likely to fail sofosbuvir treatment in real-world treatment situations.

Methods: We recruited sofosbuvir-treated patients with chronic hepatitis C from five Canadian treatment sites, and performed a case-control pharmacogenomics study assessing both previously published and novel genetic polymorphisms. Specifically studied were variants predicted to impair CES1-dependent production of sofosbuvir's active metabolite, interferon-λ signalling variants expected to impact a patient's immune response to the virus and an HLA variant associated with increased spontaneous and treatment-induced viral clearance.

Results: Three hundred and fifty-nine sofosbuvir-treated patients were available for analyses after exclusions, with 34 (9.5%) failing treatment. We identified CES1 variants as novel predictors for treatment failure in European patients (rs115629050 or rs4513095; odds ratio (OR): 5.43; 95% confidence interval (CI): 1.64–18.01; P = .0057), replicated associations with IFNL4 variants predicted to increase interferon-λ signalling (eg rs12979860; OR: 2.25; 95% CI: 1.25–4.06; P = .0071) and discovered a novel association with a coding variant predicted to enhance the activity of IFNL4's receptor (rs2834167 in IL10RB; OR: 1.81; 95% CI: 1.01–3.24; P = .047).

Conclusions: Ultimately, this work demonstrates that patient-specific genetic factors could be used as a tool to identify patients at higher risk of treatment failure and allow for these patients to receive effective therapy sooner.

Introduction

Lay Summary

In this study, we asked whether patient-specific genetic factors help predict patients likely to fail sofosbuvir treatment, and we found that patients with certain genetic variants in CES1, which encodes an enzyme required for sofosbuvir activation, increase the rate of sofosbuvir failure. Additionally, a genetic variant that increases interferon lambda signalling, which is an important part of the immune response to the virus, increases the chance of sofosbuvir failure. In the future, identifying patients most likely to fail treatment with a predictive genetic test would allow for pre-emptive prediction of the need for retreatment, thus reducing the burden on both the healthcare system and the patient.

Sofosbuvir directly inhibits the hepatitis C viral polymerase and leads to sustained virological response-based cure rates greater than 96% according to pivotal clinical trials assessing relatively homogenous groups of chronic hepatitis C patients.^[1] When treatment failure does occur, virological resistance to antiviral treatment via resistance-associated substitutions is often assumed to be the culprit, yet the majority of patients with known resistance-associated substitutions still achieve a sustained virological response. Clinical risk factors can also increase the difficulty of eradicating the virus with sofosbuvir-based treatment (eg advanced liver disease, prior hepatitis C treatment experience and infection with viral genotype 3);^[1] however, these factors are not sufficient to predict who will fail treatment.

Significant attention has been given to investigating the role of viral genetic variation on treatment failure, yet apart from recently identified genetic variation specific to viral genotype 3a,^[2] viral genetic variation reducing sofosbuvir effectiveness is rarely seen clinically.^[3] In contrast, the potential role of patient-specific genetic factors (ie intrinsic genetic-based patient-specific factors) on treatment failure has generally been ignored.

However, the active triphosphorylated metabolite of sofosbuvir is formed in a five-step process involving five enzymes (carboxylesterase 1 [CES1], cathespin A [CTSA], histidine triad nucleotide binding protein 1 [HINT1], cytidine/uridine monophosphate kinase 1 [CMPK1] and NME1-NME2 readthrough [NME1-NME2]), where genetic variation that alters the enzymatic activity of any of these would change the rate of formation of the active metabolite and therefore affect the antiviral effectiveness of the drug. Notably, the first step of sofosbuvir (the prodrug) activation is mediated by CES1^[4] that has known patient-specific genetic variation affecting drug outcome. In particular, CES1 variation predicts toxicity in patients taking capecitabine, drug plasma levels of dabigatran and increased antiplatelet activity of clopidogrel: all of which are metabolized by CES1-dependent metabolism similar to sofosbuvir.^[5–10] Furthermore, patient-specific genetic factors that contribute to variable immune responses to viral infection (eg through activation of interferon lambda 4 [IFNL4] signalling) have been linked to treatment failure across various treatment regimens.^[11–14]

In this work, we focused on the role of patient-specific genetic factors predicted to decrease CES1-based metabolism (and consequently predicted to reduce the production of sofosbuvir's active metabolite), increase interferon-λ signalling (through IFNL4 activation or upregulation of its receptor component interleukin 10 receptor subunit beta [IL10RB]), or defining a classical human leucocyte antigen (HLA ) allele associated with increased spontaneous^[14–16] and treatment-induced^[13] viral clearance (HLA-DQB1*03:01).

Retreatment of patients with prior sofosbuvir-based treatment failure is largely successful when treatment regimens are modified (eg extended treatment duration to increase overall drug exposure, the addition of ribavirin to boost antiviral action, and/or the use of complementary direct-acting antivirals to reduce the impact of viral escape mutations). Therefore, further optimization of sofosbuvir-based treatment regimens using patient-specific genetic factors would ensure that patients are placed on effective therapy sooner, and before failure occurs, thus reducing the need for retreatment and leading to cost savings for the healthcare system. This is increasingly important given that outcomes are primarily predicted based on highly selected patients included in pivotal clinical trials, and real-world treatment situations will continue to encompass more heterogeneous patients, treated with a variety of regimens. Thus, identifying additional predictive factors for treatment failure will better allow for patient-centred decisions to increase the likelihood of treatment success for individual, and unique, patients.

1 2 3 4

Next Section

Abstract and Introduction
Methods
Results
Discussion
References
Sidebar

Table 1. Clinical and demographical characteristics of sofosbuvir-treated patients before treatment initiation
Baseline characteristic	Treatment failure (n = 34)	Treatment success (n = 325)	P value*
Age, median (IQR)^a	60 (56–64)	60 (54–64)	.68
Sex, no. female (% female)	8 (23.5)	116 (35.7)	.19
Body mass index (BMI), median (IQR)^b	27 (24–33)	27 (24–30)	.41
Significant fibrosis, no. yes (% yes)^c	30 (96.8)	279 (91.2)	.49
Viral RNA level (log₁₀IU/mL), median (IQR)^d	6.3 (5.7–6.5)	6.0 (5.4–6.5)	.13
Treatment naïve, no. naïve (% yes)	19 (55.9)	202 (62.2)	.47
Genetic ancestral contributions, median (IQR)
PC 1	−0.006 (−0.007 to -0.006)	−0.007 (−0.007 to -0.006)	.72
PC 2	0.019 (0.019–0.020)	0.019 (0.018–0.020)	.28
PC 3	0.007 (0.006–0.008)	0.007 (0.006–0.008)	.95
PC 4	−0.005 (−0.007 to −0.003)	−0.005 (−0.006 to −0.003)	.16
Viral genotype, no. (%)
Genotype 1	19 (55.9)	248 (76.3)	.013
Genotype 2	4 (11.8)	23 (7.1)	.30
Genotype 3	9 (26.5)	47 (14.5)	.081
Genotype 4	0 (0.0)	4 (1.2)	1.00
Genotype 5	1 (2.9)	0 (0.0)	.095
Genotype 6	1 (2.9)	1 (0.3)	.18
Mixed	0 (0.0)	2 (0.6)	1.00
Treatment regimen, no. (%)
Sofosbuvir/ledipasvir	10 (29.4)	222 (68.3)	1.5 × 10 ⁻⁵
Sofosbuvir/ledipasvir + RBV	0 (0)	9 (2.8)	1.00
Sofosbuvir/velpatasvir	3 (8.8)	21 (6.5)	1.00
Sofosbuvir/velpatasvir + RBV	0 (0)	3 (0.9)	1.00
Sofosbuvir/daclatasvir + RBV	0 (0)	2 (0.6)	1.00
Sofosbuvir/daclatasvir + PR^e	0 (0)	1 (0.3)	1.00
Sofosbuvir/elbasvir/grazoprevir	0 (0)	3 (0.9)	1.00
Sofosbuvir/velpatasvir/voxilaprevir	0 (0)	2 (0.6)	1.00
Sofosbuvir/velpatasvir/voxilaprevir + RBV	0 (0)	1 (0.3)	1.00
Sofosbuvir + RBV^e	12 (35.3)	40 (12.3)	1.1 × 10⁻³
Sofosbuvir + PR^e	5 (14.7)	13 (4.0)	0.011
Sofosbuvir/simeprevir^e	4 (11.8)	8 (2.5)	0.019
Treatment regimen considered standard of care, no. (%)	13 (38.2)	264 (81.2)	2.9 × 10⁻⁷
Treatment weeks, no. (%)^f
8	3 (9.1)	68 (20.9)	.11
12	19 (57.6)	185 (56.9)	1.00
24	9 (27.3)	71 (21.8)	.51
Other (4, 16 or 20 weeks)	2 (6.1)	1 (0.3)	.02

Abbreviations: IQR, interquartile range; PC, principal component; PR, pegylated interferon and ribavirin; RBV, ribavirin.
^aAge at treatment initiation was unknown for two patients because of missing treatment start dates.
^bBMI values were missing in 14 patients.
^cDefined as a liver biopsy with a Metavir score ≥ F2, an APRI value ≥ 0.7, a FIB-4 score ≥ 1.45, or a FibroScan® result ≥ 7 kPa collected prior to treatment. Presence/absence of significant fibrosis could not be determined for 22 patients because of missing test measurements before treatment initiation or tests indicating the absence of significant fibrosis > 6 months before treatment initiation.
^dViral RNA levels prior to treatment initiation were unknown in 25 patients.
^eTreatments no longer considered standard of care in Canada.
^fTreatment duration was unknown for one patient because of a missing treatment end date.
*P values in bold indicate values under .05.

Table 2. Associations of genetic variation impacting *CES1*, *IFNL4*, *IL10RB* and *HLA-DQB1* with sofosbuvir-based treatment failure
Gene	Identifier	Functional annotation^a	Effect allele^b	Patients of all ancestries					Patients of European ancestry
Gene	Identifier	Functional annotation^a	Effect allele^b	N	EAF failures	EAF successes	OR (95% CI)	P value	N	EAF failures	EAF successes	OR (95% CI)	P value
CES1	rs115629050	Missense	A	356	0.045	0.023	2.46 (0.58–10.48)	.22	265	0.056	0.017	4.09 (0.85–19.73)	.080
	rs4513095	Splice variant	A	359	0.044	0.017	2.91 (0.65–12.99)	.16	267	0.054	0.017	5.54 (1.13–27.14)	.035
	CES1 impairment^c	—	—	356	0.091	0.040	2.86 (0.96–8.46)	.058	265	0.110	0.034	5.43 (1.64–18.01)	.0057
IFNL4	rs12979860	Intronic	T	359	0.485	0.365	2.25 (1.25–4.06)	.0071	267	0.482	0.379	2.00 (1.05–3.82)	.035
	rs74597329^d	IFNL4-ΔG/TT	G	345	0.500	0.358	2.10 (1.15–3.84)	.016	256	0.500	0.376	1.80 (0.93–3.50)	.081
	rs8099917	Upstream	G	359	0.368	0.229	2.47 (1.28–4.77)	.0069	267	0.339	0.238	2.02 (0.97–4.24)	.062
IL10RB	rs2834167	Missense	G	359	0.353	0.274	1.81 (1.01–3.24)	.047	267	0.321	0.251	1.57 (0.83–2.99)	.17
HLA-DQB1	*03:01	—	P^e	—	—	—	—	—	267	0.036	0.115	0.33 (0.08–1.42)	.14

Note: Results were obtained through logistic regression analyses assuming an additive genetic model and adjusting for treatment with regimens considered standard of care (yes/no) and the first four principal components for genetic ancestry. P values <.05 are indicated in bold.
Abbreviation: EAF, effect allele frequency.
^aBased on ensembl canonical transcripts ENST00000290200.2 (IL10RB) and ENST00000360526.3 (CES1).
^bReported on the forward strand according to genome build GRCh37 (hg19).
^c CES1 impairment represents at least one of either CES1 rs115629050 or rs4513095 that are predicted to negatively impact CES1 function.
^dProxy variant for rs368234815.
^eP: presence of HLA-DQB1*03:01 (compared to A for absence of HLA-DQB1*03:01).

Authors and Disclosures

Catrina M. Loucks^1–3, Jennifer J. Lin^1,2,4, Jessica N. Trueman^1,2, Britt I. Drögemöller⁵, Galen E. B. Wright⁶, Wan-Chun Chang^1,2, Kathy H. Li^1,2, Eric M. Yoshida⁷, Jo-Ann Ford⁷, Samuel S. Lee⁸, Pam Crotty⁸, Richard B. Kim⁹, Bandar Al-Judaibi^10,11, Ute I. Schwarz⁹, Alnoor Ramji⁷, Jeanette F. Farivar¹², Edward Tam¹³, Lori Lee Walston¹³, Colin J. D. Ross^1,4,14 and Bruce C. Carleton^1,2,4,15

¹BC Children's Hospital Research Institute, Vancouver, Canada
²Division of Translational Therapeutics, Department of Pediatrics, Faculty of Medicine, University of British Columbia, Vancouver, Canada
³Department of Anesthesiology, Pharmacology & Therapeutics, Faculty of Medicine, University of British Columbia, Vancouver, Canada
⁴Department of Medical Genetics, Faculty of Medicine, University of British Columbia, Vancouver, Canada
⁵Department of Biochemistry and Medical Genetics, University of Manitoba, Winnipeg, Manitoba, Canada
⁶Department of Pharmacy and Therapeutics, University of Manitoba, Winnipeg, Manitoba, Canada
⁷Division of Gastroenterology, Department of Medicine, University of British Columbia, Vancouver, Canada
⁸Liver Unit, University of Calgary Cumming School of Medicine, Calgary, Canada
⁹Division of Clinical Pharmacology, Department of Medicine, Western University, London, Canada
¹⁰Division of Transplantation, University of Rochester, Rochester, New York, USA
¹¹Department of Liver Transplantation and Hepatobiliary Surgery, King Faisal Specialist Hospital and Research Center, Riyadh, Saudi Arabia
¹²Pacific Gastroenterology Associates, Vancouver, Canada
¹³LAIR Centre, Vancouver, Canada
¹⁴Faculty of Pharmaceutical Sciences, The University of British Columbia, Vancouver, Canada
¹⁵Pharmaceutical Outcomes Program (POPi), British Columbia Children's Hospital, Vancouver, Canada

Correspondence
Bruce C. Carleton, 950 West 28th Avenue, Vancouver, BC V5Z 4H4, Canada. Email: bcarleton@popi.ubc.ca

Conflicts of Interest
Dr. Eric Yoshida has been an investigator of clinical trials sponsored by Gilead Sciences, AbbVie, Merck, Pfizer, Intercept, Madrigal, Genfit and Celgene. He has received honoraria for CME/Ad Board lectures from Gilead Canada, Merck Canada, AbbVie Canada and Intercept Canada. Jo-Ann Ford has received honoraria for lectures and advisory boards from AbbVie Canada, Gilead Canada, Merck Canada and Pfizer Canada. Dr. Samuel Lee has received research funding from, and performed consulting for, AbbVie, Gilead, Intercept, London Drugs, Merck, Oncoustics and Pendopharm. He has also received speaker fees from AbbVie, Gilead and Merck. Dr. Alnoor Ramji has received research funding from AbbVie, Allergen, Assembly Biosciences, Gilead Sciences, Janssen (J. & J.), Intercept, Galmed, Novartis and Merck; performed consulting for AbbVie, Gilead Sciences, Intercept and Merck; and received speaker fees from AbbVie, Celgene, Gilead Sciences, Janssen (J. & J.), Intercept, Novartis and Merck. Jeanette Farivar has disclosures from Gilead, Merck and AbbVie for consulting/advisory boards. Dr. Edward Tam has been an investigator for clinical studies sponsored by AbbVie, BMS, Genfit, Gilead, Intercept, Madrigal, Merck, Novartis, Pfizer and Springbank; received speaker honoraria from AbbVie, BMS, Gilead, Intercept, Merck, Novartis and PendoPharm and received honoraria for advisory board roles for AbbVie, BMS, Gilead, Intercept, Merck and Pfizer. Lori Lee Walston has received honoraria for advisory boards for AbbVie, Gilead and Merck. All other authors disclose no conflicts.

Author Contributions
Catrina M. Loucks, Jennifer J. Lin, Jessica N. Trueman, Eric M. Yoshida, Jo-Ann Ford, Samuel S. Lee, Pam Crotty, Richard B. Kim, Bandar Al-Judaibi, Ute I. Schwarz, Alnoor Ramji, Jeanette F. Farivar, Edward Tam and Lori Lee Walston carried out acquisition and preparation of data and samples. Catrina M. Loucks, Jennifer J. Lin, Britt I. Drögemöller, Galen E.B. Wright, Wan-Chun Chang, Kathy Li, Eric M. Yoshida, Samuel S. Lee, Richard B. Kim, Bandar Al-Judaibi, Alnoor Ramji, Edward Tam, Colin J. Ross and Bruce C. Carleton carried out analysis and interpretation of data. Catrina M. Loucks, Jennifer J. Lin, Jessica N. Trueman, Britt I. Drögemöller, Galen E.B. Wright, Eric M. Yoshida, Samuel S. Lee, Richard B. Kim, Bandar Al-Judaibi, Alnoor Ramji, Edward Tam, Colin J. Ross and Bruce C. Carleton were involved in study concept and design. Catrina M. Loucks and Jennifer J. Lin carried out writing of the manuscript. All authors contributed to the critical revision of the manuscript for important intellectual content. Additionally, all authors confirm that they had full access to all the data in the study and accept responsibility to submit for publication.