Background
Since 2002, single nucleotide polymorphisms (SNPs) located in the interferon-λ genetic region (
IFNL) on chromosome 19q13 are associated with antiviral protection [
1,
2]. In humans, polymorphisms of the interferon-λ3 gene
(IFNL3) or located near
IFNL3 (mainly rs12979860, rs4803217, rs8099917, rs12980275) became established genetic markers of spontaneous and PEGylated interferon-based antiviral drug-induced hepatitis C virus (HCV) clearance [
3‐
8].
In 2013, Prokunina-Olsson et al. [
9] discovered the interferon-λ4 gene (
IFNL4) located on chromosome 19q13.2 upstream of
IFNL3 that harbours the dinucleotide frame-shift variant rs368234815 (TT/ΔG), named initially ss469415590. The polymorphism rs368234815, located within the first
IFNL4 exon, is classified as a deletion/insertion genetic variation (TT/ΔG). The
IFNL4 rs368234815 polymorphism controls a generation of the functional protein – interferon-λ4 (IFN-λ4) [
9]. The ancestral
IFNL4 ΔG allele creates an open reading frame for IFN-λ4, whereas the alternative
IFNL4 TT variant eliminates IFN-λ4 [
9].
IFNL4 rs368234815 influences the signalling pathway between interferon-λ3 receptor 1 (IFN-λR1) and interferon-stimulated response element (ISRE). Overexpression of
IFNL4 suppresses
IFNL3 induction and promoter activation [
10]. Transient
IFNL4 overexpression in a hepatoma cell line induced signal transducer and activator of transcription (STAT)1/STAT2 phosphorylation and expression of interferon-stimulated genes (ISG) [
9]. Low intracellular expression of IFN-λ4 induces IFN-λ expression, leading to the Janus activated kinase (JAK)-STAT signalling and expression of ISG [
11]. The favourable outcome of HCV infection correlates with the inability to encode IFN-λ4 [
9,
10].
There is a high linkage disequilibrium (LD) between rs368234815 and rs12979860, previously designated
IFNL3 rs12979860 [
12]. At present, we know that rs12979860 lies in the intron 1 of
IFNL4 and has to be related to
IFNL4. Since 2013, we know that rs12979860 is associated with HCV spontaneous or therapeutic eradication due to its high LD with
IFNL4 rs368234815 (TT/ΔG) [
9]. In Ferraris et al. [
11],
IFNL4 ∆G showed a 100% correlation with the TT genotype of rs12979860. Compared to rs12979860, rs368234815 appeared similarly [
9,
13] or more strongly [
14‐
16] associated with HCV clearance. Unfavourable (variant) alleles of both rs12979860 and rs368234815 were also associated with reduced clearance of RNA viruses from the respiratory tract [
17].
IFNL3 rs4803217, altering
IFNL3 mRNA stability [
3], is also in a substantial LD with
IFNL4 rs368234815. However, rs4803217 seems to be less strongly associated with HCV clearance
than IFNL4 rs368234815 [
18].
An association between
IFNL polymorphisms and HCV clearance was studied mainly in non-uremic subjects [
4‐
10,
15,
16,
18,
19]. The meta-analysis by Xie et al. [
20] revealed that the rs368234815 TT/TT genotype correlated with the sustained virologic response (SVR) to treatment with PEGylated interferon plus ribavirin in HCV-1/4-infected Caucasian patients but not in HCV-2/3-infected Caucasian patients. Conversely, the rs368234815 ΔG/ΔG genotype was linked to treatment failure in Caucasian patients, regardless of the HCV genotype. The rs368234815 ΔG allele was also a negative predictor of efficacy of direct-acting antivirals applied in HCV patients [
21‐
23].
Reports concerning mentioned above associations in patients suffering from haemodialysis (HD)-dependent end-stage renal disease (ESRD) are scarce [
24‐
27], and, to our knowledge, were not conducted for
IFNL4 rs368234815, currently designated as the most promising predictor of HCV resolution [
18]. Uremic milieu changes the expression of genes [
28]. Therefore, it is worthy of investigating whether the impact of
IFNL polymorphic variants on HCV clearance is comparable in HD patients to that observed in non-uremic populations.
Our study’s primary purpose was to recognize the
IFNL3/IFNL4 impact on spontaneous HCV elimination in HD patients with particular attention paid to
IFNL4 rs368234815 polymorphism functionally associated with the outcome of HCV infection [
9]. Firstly, we aimed to investigate whether
IFNL4 rs368234815 is associated with spontaneous HCV clearance in ESRD patients on regular HD treatment. If so, whether genotyping of rs368234815 in HD subjects can be more useful as a predictor of spontaneous HCV resolution than other tested
IFNL3/IFNL4 polymorphisms (near
IFNL3 rs12980275,
IFNL3 rs4803217,
IFNL4 rs12979860, near
IFNL4 rs8099917). We also planned to check whether an analysis of haplotypes of tested
IFNL3/IFNL4 SNPs could be useful in predicting HCV infection outcome in HD patients. Additionally, a retrospective survival analysis was performed concerning
IFNL4 rs368234815 polymorphism. For a better understanding of genetic mechanisms underlying differences in analysed phenotypes possible attributed to
IFNL (spontaneous HCV clearance, survival), we used in silico methods for prediction of alterations in transcription factor (TF) binding sites (TFBS) caused by the tested polymorphisms.
Patients and methods
Patients
The study included 161 HD subjects with persistently positive anti-HCV antibodies. They were enrolled between January 2009 and November 2018 in the Wielkopolska region of Poland. The same patients were also tested in this study for near
IFNL3 rs12980275,
IFNL3 rs4803217,
IFNL4 rs12979860, and near
IFNL4 rs8099917. Some of them were included in our previous studies indicating associations of these SNPs with spontaneous HCV clearance in HD subjects [
24,
26]. We used the current data of rs12980275, rs4803217, rs12979860, and rs8099917 to compare the predictability of spontaneous HCV clearance using the newly tested rs368234815 and previously established genetic markers of HCV resolution.
Enrolled patients were never treated with anti-HCV medications. Among them, 68 (42.2%) spontaneously resolved HCV infection, whereas 93 (57.8%) were persistently HCV RNA positive.
Eighty (49.7%) HD subjects showed positive antibodies against core antigen (anti-HBc) of hepatitis B virus (HBV); 10 (6.2%) HD individuals presented surface HBV antigen (HBsAg) positivity; 7 (4.3%) patients were persistently HBV DNA positive. All HD patients tested negative for antibodies to the human immunodeficiency virus (anti-HIV-1/HIV-2).
We obtained demographic, clinical, and baseline laboratory data of HD patients from physicians of HD facilities.
Survival studies
All anti-HCV positive patients genotyped for IFNL4 rs368234815 were included in the retrospective longitudinal survival study. In this group, we analysed survival probability from the onset of renal replacement therapy (RRT) to the last data collection (1st - 10th November 2020) concerning IFNL4 rs368234815 genotypes.
Genotyping
Chromosomal localization of tested
IFNL3 and
IFNL4 polymorphisms is shown in Supplementary Fig. 1A. of the Additional file
1.
IFNL4 rs368234815 polymorphism (TT/TT, ΔG/TT, ΔG/ΔG) was genotyped by a polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) method, as recently described by Pouryasin et al. [
29]. Genotyping of rs4803217, rs12980275, rs8099917, and rs12979860 polymorphisms was carried out by a high-resolution melting curve analysis (HRM) as described in our previous studies [
24,
26]. Supplementary Table 1 in the Additional file
1 shows conditions for identifying polymorphisms genotyped by PCR-RFLP or HRM.
For quality control, approximately 10% of the randomly chosen samples were re-genotyped using the same genotyping method; the concordance rate was 100%. Among 161 tested samples, two (1.2%) failed the genotyping of rs12980275 and rs4803217 each, four (2.5%) – rs12979860, and one (0.6%) - rs8099917. We excluded samples that failed the genotyping from statistical analyses.
Prediction of transcription factor binding sites
Potential regulatory impacts of the rs4803217, rs12980275, rs8099917, rs12979860, and rs368234815 through modifications of the TFBS motifs were assessed with the experimental ENCODE TFBS ChIP-seq data [
30] and in silico prediction of DNA-binding sites collected in HOCOMOCO version 9 [
31], JASPAR CORE version 5.0 ALPHA 2016 [
32], and CIS-BP version 1.02 [
33] databases with FIMO software version 4.11.1 [
34].
We performed the computational analysis on the GenBank DNA sequences (contig NT_011109.17) [
35] adjacent to SNP positions. FASTA sequences, one per each SNP allele, were used as an input for the FIMO. To minimize false rates, we calculated the background file directly from input sequences with MEME-suit fasta-get-Markov script.
We analysed results via a self-developed Python script to find locations of ENCODE ChIP-seq validated TFBS and FIMO predicted motif BS overlapping SNP positions. A
p-value < 0.0005 and q-value < 0.05 were selected as the cut-off values for reliable predictions. Motifs matched in both orientations were also analysed concerning perfect reverse complement forming, thus increasing the true positive match probability. We identified TF annotation information from the SwissProt database [
36] and filtered statistically significant differentially bound hits of
Homo sapiens motifs.
Statistical methods
The results are presented as percentages for categorical variables or medians (range) for continuous variables because they were usually non-normally distributed as determined by the Shapiro–Wilk test.
We compared demographic, clinical, and laboratory data in anti-HCV positive HD subjects stratified by HCV RNA testing results. Continuous variables were compared using the Mann-Whitney test. The Chi-squared test or exact Fisher’s test was used for comparison of categorical variables, as appropriate.
We analysed the Hardy–Weinberg equilibrium (HWE) to compare the observed genotype frequencies to the expected ones using the Chi-square test (P > 0.05 with df = 1 for balance).
We tested polymorphisms for trends in association using Fisher’s test if two groups were compared or the Cochran-Armitage test if three groups were compared (Ptrend). Genotype (Pgenotype) and allelic (Pallelic) distributions were compared between the tested groups using Pearson’s Chi-squared test or Fisher’s exact test, as appropriate.
Odds ratios (OR) and 95% confidence intervals (CIs) for OR were calculated to quantify how strongly the presence or absence of the tested allele or genotype is associated with the presence or absence of selected phenotypes of HD patients. Pearson’s Chi-squared test was used for statistical evaluation of OR. All probabilities were two-tailed. We applied logistic regression to assess the significance of the IFNL4 genotype, among other possible determinants of HCV clearance. A ROC curve was plotted to show the AUC as a measure of the accuracy of the model.
This study’s statistical power was estimated by Quanto 1.2.4, under the “unmatched case-control” study design, with the present sample sizes and genetic effects that were observed in the study. All analyses were performed with a type I error rate of 0.05.
The survival studies were retrospectively performed from each patient’s individual RRT onset (from 15 June 1983 to 27 July 2018) to each patient’s outcome data (1st - 10th November 2020). Patients who underwent renal transplantation were analysed if they returned to haemodialysis treatment. All-cause, cardiovascular, infection-related, and neoplasm-related reasons for death were analysed. These analyses were also performed using dominant, recessive, and additive models of inheritance. Due to a relatively small number of HCV exposed patients, this analysis was planned as a preliminary one. The Kaplan-Meier method with the log-rank test was solely used to estimate differences in the cumulative proportion surviving, characterizing the genotype groups in each inheritance model.
Statistical analyses of the data mentioned above were performed using Statistica version 12 (Stat Soft, Inc., Tulsa, OK, USA). P-values less than 0.05 were considered significant. However, a Bonferroni correction was used for significance in detailed association analyses involving rs368234815 because such associations were not previously evaluated in HD patients.
Pair-wise LD between
IFNL3 and
IFNL4 polymorphisms was computed as both D′ and r
2 using the genotype data from the tested sample and the Haploview 4.2 software [
37] (Supplementary Fig. 1B in the Additional file
1). Haplotypes were estimated using the mentioned Haploview 4.2 software (sliding windows method) and statistically analysed if their incidence in the examined group was at least 1%. Statistical significance was assessed using the 1000-fold permutation test. Only
P-values corrected by the test rules were considered significant.
Discussion
IFNL4 rs368234815 polymorphism is recognized as the most potent predictor of HCV infection outcome in the general population [
18]. To our knowledge, this paper, for the first time, describes associations between
IFNL4 rs368234815 polymorphism and spontaneous HCV clearance in HD patients who substantially differ from the general population due to persistent uremic status and altered immune competence [
39]. In this study, we applied the PCR_RFLP method for genotyping of rs368234815 polymorphism, which was recently described by Pouryasin et al. [
29]. Among Iranian non-uremic patients with chronic HCV infection tested with this method, the rs368234815 ΔG/ΔG frequency was 17.3% [
29]. Caucasian HD patients of Polish origin with persistently positive testing for HCV RNA revealed a similar frequency of the ΔG/ΔG genotype (16.1%).
Similarly, as in non-uremic subjects [
14,
15], rs368234815 was associated with spontaneous HCV clearance also in HD patients. On the same group of anti-HCV positive HD patients, we were able to show that major homozygosity in rs368234815 (TT/TT) similarly predicts spontaneous HCV clearance in the dominant model of inheritance as other tested
IFNL3/
IFNL4 polymorphisms already associated with HCV clearance in HD subjects. All tested
IFNL polymorphisms (including rs368234815) indicated a 2.4–2.8-fold higher frequency of spontaneous HCV clearance in HD patients characterized by the major homozygosity in analysed polymorphisms.
The usefulness of the
IFNL4 rs368234815 TT/TT genotype in the predictability of spontaneous HCV clearance was superior to other tested
IFNL polymorphisms only in the additive inheritance model. In non-uremic subjects, the rs368234815 TT/TT genotype indicated 3.6-fold (Swiss Caucasians [
15]) – 12-fold (American white women [
14]) higher frequency of spontaneous HCV clearance compared with the ∆G/∆G genotype. In HD patients, the TT/TT genotype predicted a 6.4-fold higher probability of spontaneous HCV clearance. Therefore, uremic conditions do not seem to influence substantially
IFNL4 rs368234815 expression concerning spontaneous HCV resolution. Additionally, an identification of the homozygosity in the variant ∆G allele of rs368234815 in HD subjects means a more potent prediction of persistent HCV infection that it is observed in the case of the variant homozygosity of other tested
IFNL3/
IFNL4 polymorphisms.
A question was raised whether the haplotype of unfavourable rs368234815 ∆G allele and favourable major allele of any other
IFNL3/
IFNL4 SNPs could be associated with more frequent spontaneous HCV clearance. Data of O’Brien et al. [
18] excluded such a possibility for the haplotype of a favourable major allele of
IFNL3 rs4803217 [
40,
41] and the unfavourable
IFNL4 rs368234815 ∆G allele concerning a day 28 anti-HCV response in African American individuals. A recent study by Vergara et al. [
42] demonstrated in the European ancestry group that subjects with the haplotype rs368234815∆G_rs4803221C were 1.7 times more likely to clear HCV than individuals showing rs368234815∆G_rs4803221G haplotype. Also, the haplotype rs368234815∆G_rs8099917T was associated with a 1.6 times higher frequency of HCV resolution compared to rs368234815∆G_rs8099917G haplotype. There were no differences in association analysis of these two haplotypes concerning spontaneous HCV clearance in African ancestry [
42]. In the examined European HD patients, only haplotypes containing exclusively major alleles of tested
IFNL3/
IFNL4 SNPs were associated with spontaneous HCV resolution. In HD patients, similarly to in African Americans [
42], there were no significant differences in spontaneous HCV clearance frequency if haplotype was composed of rs368234815∆G and major versus variant allele of tested
IFNL3/
IFNL4 SNPs (Supplementary Table 4 in the Additional file
1). However, a relatively small group of the examined patients could participate in a lack of significance. It has to be noted that novel SNPs were recently identified in the
IFNL region of which rs4803221 (and already known rs8099917) appear contributors in a primary signal of association represented by
IFNL4 rs368234815 SNP [
42].
Spontaneous HCV elimination is associated with the induction of virus-specific CD4
+ and CD8
+ T cell responses and, to a lesser extent, with the generation of neutralising antibodies targeting the HVR1 region of HCV envelope glycoprotein 2 [
43]. The rs368234815 impact on HCV infection outcome is still investigated. The
IFNL4 rs368234815 TT allele, which promotes spontaneous HCV clearance, is associated with the expression of the transcript JN806227. The latter generates a prematurely terminated IFN-λ4 protein with a lack of activity [
9,
10]. The
IFNL4 ∆G allele transcript produces full-length IFN-λ4, of which IFN-λ4 -70Pro is more active, while IFN-λ4-70Ser shows diminished activity. This difference depends on nonsynonymous variant
IFNL4 rs117648444 A/G (Pro70Ser) in exon 2.
IFNL4 haplotype rs368234815∆G_rs117648444G produces a more active IFN-λ4 -70Pro and is captured by rs8099917 variant (G) allele, haplotype rs368234815TT_rs117648444G corresponds to lack of active IFN-λ4, and rs368234815∆G_rs117648444A generates a less active IFN-λ4-70Ser. Both latter haplotypes are tagged by the rs8099917 T allele [
9,
44]. It is suggested that one of the possible mechanisms, by which primarily intracellular IFN-λ4 deteriorates HCV elimination, is up-regulation of IFN-λR1 what impedes receptor binding of the other members of the IFN-λ family [
45]. Recent analyses [
46] have shown that
IFNL4 rs12979860, being a marker for rs368234815, is strongly associated with the HCV proteome’s amino acid variants. Like that concerning leucine at position 2224 of NS5A in HCV genotype 1b in patients with the rs12979860 CC genotype, some of these associations correspond to the pre-treatment viral load. The rs12979860 variant (T) allele correlated with lower pre-treatment viral load for all HCV genotypes, while the rs12979860 CC genotype was associated with higher mean viral load. Therefore, the difference in viral load could be partially related to changes in viral amino acids associating with the presence or absence of IFN-λ4 [
46]. It has to be noted that the rs12979860 CC genotype was simultaneously associated with spontaneous HCV clearance and higher viral load. The latter was inversely [
47] or positively [
48] correlated with spontaneous HCV elimination in clinical studies. The coincidence of rs12979860, viral load, and HCV eradication needs exploration in further investigations.
We identified that minor alleles of the rs368234815, rs12979860, and rs4803217 could cause statistically significant changes in TFBS. Therefore, the individual or combined effects of those changes may contribute to the differential
IFNL3 and
IFNL4 expression. As
IFNL3/
IFNL4 polymorphisms are well-established predictors of HCV clearance, removal or addition of the host TFBS in the presence of the minor alleles of tested polymorphisms might be at least a partial explanation of HCV persistence in the
IFNL3/
IFNL4 minor allele bearers. Our computational analysis provides a list of candidate differentially bound TFBS in tested variants. The experimental ChIP-seq study could be applied to verify their impact exerted by the minor alleles. The following example shows how critical is the need for such verification. In silico analysis revealed that the host TFBS for DNA DNMT1 is removed in the
IFNL3 rs4803217 minor allele bearers who are predisposed to persistent HCV infection ([
18,
40,
41], this study). However, CpG islands for DNMT1-induced DNA methylation were not shown in the interferon-λ genetic region [
49], so removing TFBS for DNMT1 may have no clinical relevance concerning HCV clearance. Nonetheless, the proposed putative mechanistic hypothesis describing the effects of the studied SNPs on TFBS serves as an exploratory analysis based on computational studies. Moreover, the public reference dataset is an approximation of the cellular and tissue states in studied samples.
In our study, the
IFNL4 rs368234815 ∆G/∆G genotype appeared to be associated with cancer mortality in HCV-exposed HD patients. We consider this finding as a preliminary one due to a small number of HD patients who died from cancers in this group. Additionally, such a survival analysis in the examined not HCV-exposed subjects did not reveal any relationship between neoplasm-related mortality and
IFNL4 rs368234815 SNP. On the other hand, our data are following the evidence demonstrating reduced survival in rs368234815 ∆G/∆G genotype or rs12980275 GG genotype (corresponds to rs368234815 ∆G/∆G genotype) subjects with cancers [
50,
51].
IFN-λ has a dual role in cancer. Direct antitumor effects of IFN-λ include inhibition of cell proliferation, promotion of cell apoptosis, and cell cycle arrest. Indirect antitumor effects of IFN-λ include immune cell activation and angiogenesis inhibition. However, new evidence indicates that IFN-λ can also promote oncogenesis [
52].
IFNL polymorphisms may differentiate their role in cancer antitumor effects or oncogenesis. Our in silico analyses demonstrated that neoplasm-related mortality might be due to the removal of TFBS for PLAGL1 zinc finger 1 protein, which is associated with anti-proliferative activities and tumour suppression. PLAGL1 gene is often deleted or methylated and silenced in cancer cells [
53]. This exciting aspect needs further elaboration.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.