Assessment of IL-28: rs12979860 and rs8099917 Polymorphisms in a Cohort of Cuban Chronic HCV Genotype 1b Patients

Abstract

Hepatitis C virus (HCV) is a significant global public health problem with >185 million infections worldwide. A series of genome-wide association studies (GWAS) has identified IL-28B polymorphisms as a predictor of sustained virologic response (SVR), as well as spontaneous clearance in chronic HCV genotype 1 patients. The objective of this work was to evaluate the prevalence of IL-28B rs12979860 and rs8099917 polymorphisms in Cuban chronic HCV patients. The study cohort included 73 chronic HCV patients treated with concomitant administration of CIGB-230 and nonpegylated IFN-α plus ribavirin (non-pegIFN-α/R) antiviral therapy. The genotype distribution of IL-28B rs12979860CC, -CT, and -TT was 29, 41, and 30%, respectively, and the distribution for rs8099917TT, -TG, and -GG was 63, 31, and 5%, respectively. The allele frequencies for rs12979860C and -T alleles were 51 and 49%, respectively, and for rs8099917G and -T alleles, the values were 21 and 79%, respectively. SVR rates were 55, 42, and 35% for rs12979860CC, -CT, and -TT, respectively, and 52, 30, and 25% for rs8099917TT, -GT, and -GG, respectively. The combined assessment of both single nucleotide polymorphisms (SNPs) resulted in 3 major genotypes (rs12979860CC/rs8099917TT, rs12979860CT/rs8099917TT, and rs12979860CT/rs8099917GG) with a frequency of 30.1, 21.9, and 20.5%, respectively. In patients with heterozygous variant rs12979860CT, the additional genotyping of rs8099917 contributed to increase the SVR rate. It is concluded that in Cuban HCV-infected patients, the responder homogeneous variant rs8099917TT is the most frequent genotype. The simultaneous genotyping of 2 IL-28B SNPs could improve the prediction of SVR contributing to better therapeutic decisions and treatment management.

INTRODUCTION

HCV is a significant global public health problem, and recent estimates show an increase in seroprevalence over the last 15 yr, to 2.8%, with >185 million infections.¹ HCV viruses are classified into 6 genotypes, with genotype 1 as the most widespread and common, comprising 83.4 million cases (46.2% of all HCV cases).^{1, 2}

In 2009, a series of GWAS identified polymorphisms (SNPs) in the IL-28B gene, a gene coding for IFN-λ3 as a predictor of SVR, as well as spontaneous clearance in chronic HCV genotype 1 patients treated with pegIFN-α/R.^3–6 Among all studied IL-28B SNPs, rs12979860 and rs8099917 were found to have the stronger association with positive HCV treatment outcome in chronic HCV genotype 1 patients.^{7, 8} These 2 SNPs have been reported to be in strong linkage disequilibrium (LD) with rs12979860, showing a robust population frequency, and its genotyping seems to be enough to predict SVR, particularly the rs12979860CC variant with the strongest SVR-predictive value.^{4, 9, 10} At the same, other groups reported that combined genotyping of rs12979860 and rs8099917 could significantly improve the prediction of SVR in carriers of heterozygous nonresponder variant rs12979860CT. ^{11, 12}

Additionally, other variables have been found as predictive for SVR against HCV genotype 1, e.g., pretreatment of HCV RNA level, Ishak fibrosis score, body mass index (BMI), and pretreatment ratio of serum γ-glutamyl transferase/alanine transaminase (γ-GT/ALT), among others.^{13, 14}

However, despite the estimated prevalence of HCV in Cuba (0.6–0.8%, with genotype 1b as the most common^{15, 16}), the prevalence among those patients of the mentioned SNPs remained to be characterized, as well as the other variables mentioned as predictive for SVR after therapy. In this work, we describe the genotypes and allele frequencies of IL-28B SNPs rs12979860 and rs8099917 in 73 treatment-naïve, chronic HCV genotype 1b patients enrolled in a Phase II clinical trial conducted at the National Institute of Gastroenterology (Havana, Cuba).¹⁷ In this clinical study, they were administered with a DNA-based therapeutic vaccine candidate against HCV¹⁸—CIGB-230—combined with a non-pegIFN-α/R therapy. Prognostic markers for successful outcome of therapy in chronic HCV patients were assayed by using logistic model.

MATERIALS AND METHODS

Study Population

The study cohort included 73 treatment-naïve patients, positive to plasma HCV RNA, genotype 1b, with diagnosed chronic hepatitis by liver biopsy and no other documented cause of liver disease, who were included in the Phase II clinical trial (Protocol Code: IG/VHI/HC/0701; Public Register Code: RPCEC00000074) conducted at the National Institute of Gastroenterology. Detailed information about the clinical study design can be found in Amador et al.¹⁷ In brief, all patients received conventional IFN-α-2b (3 × 10⁶ units s.c., 3×/wk) and ribavirin for 48 wk. Patients were randomized in 5 groups, according to CIGB-230 treatment. The control group received 12 vaccine placebo inoculations. Two groups received 6 inoculations of CIGB-230, 1 starting simultaneously with the antiviral treatment as early add-on treatment (E6) and the other starting on wk 12 of therapy as late add-on treatment (L6). The remaining 2 groups were inoculated 9 times with CIGB-230, 1 as early add-on treatment (E9) and the other as late add-on treatment (L9). In our study, only patients that could be classified as SVR, determined as undetectable HCV RNA 6 mo after completion of therapy, or as non-SVRs were included in the analysis. The final number of patients in each group was as follows: control group (n = 28), E6 (n = 9), L6 (n = 12), E9 (n = 16), and L9 (n = 8). The demographic characteristics of patients are shown in Table 1. The study was approved by the Ethics Committee of the National Institute of Gastroenterology and the National Regulatory Authority [Center for State Control of Drugs, Equipment and Medical Devices (CECMED), Havana, Cuba]. A written, informed consent for genetic testing was obtained from all participants.

TABLE 1.

Demographic Characteristics of Healthy Donors and Chronic HCV Cohorts

Variable	Chronic HCV cohort	Healthy donors
Age	46 (26–59)	36 (18–63)
Sex, male (%)	32 (44)	363 (90)
Skin color
White	45	187
Black	13	75
Mestizo	15	138

Genomic DNA Extraction and SNPs

Blood samples were collected in EDTA anticoagulation Vacutainer tubes. Genomic DNA was purified from EDTA anti-coagulated peripheral blood using Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA).

Alleles of rs12979860 were assessed by PCR sequencing. First, an amplification PCR step was performed on an iCycler Thermal Cycler (Bio-Rad, Hercules, CA, USA) using primers rs12979860 forward, TCCGCTAAGCAGGACAGATTG, and rs12979860 reverse, CCACCGGGTTCCCATACAC. In brief, 50–100 ng genomic DNA was amplified with 5 pmol forward and reverse primers in a total reaction volume of 25 µl. The PCR temperature profile comprised 94°C for 5 min; 30 cycles of 94°C for 15 s, 57.5°C for 25 s, and 72°C for 60 s; and a final extension at 72°C for 5 min. The resulting PCR products were bidirectionally sequenced in a 3730xl DNA Analyzer (Thermo Fisher Scientific, Waltham, MA, USA).

The genotyping of the rs8099917 site was tested using TaqMan SNP Genotyping Assays (Cat. #4351379; Thermo Fisher Scientific) on a 7900HT instrument (Thermo Fisher Scientific). In brief, the thermal cycling conditions were held at 95°C for 10 min, denatured at 92°C for 15 s, and annealed/extended at 60°C for 1 min. Automated allele calling was performed using SDS software (Thermo Fisher Scientific).

Statistical Analysis

Statistical analysis was performed with R 2.11.0 (The R Project for Statistics Computing; https://www.r-project.org/).¹⁹ Comparison of categorical variables was assessed in contingency tables by Pearson's χ² test and Fisher's exact test. P < 0.05 was considered statistically significant. Continuous variables were represented as mean and sd and compared using the Mann-Whitney test or Student’s t test. Univariate and multivariate logistic regression models were used to assess the relationship among the SVR variable and demographic, clinical, and IL-28B SNPs variables. The odds ratio (OR) and the 95% confidence interval (CI) were calculated. Multivariate logistic regression analysis was used to identify the association of SVR with covariates rs12979860, rs8099917, age, sex, fibrosis stage, skin color,²⁰ BMI, and pretreatment ratio of serum γ-GT/ALT.²¹ Structure of LD was analyzed with Haploview 4.2 (Broad Institute, Cambridge, MA, USA).

RESULTS

The genotype distribution and allele frequencies of IL-28B rs12979860 and rs8099917 are shown in Table 2. The overall genotype distributions of IL-28B rs12979860CC, -CT, and -TT were 29, 41, and 30%, respectively, and the distributions for rs8099917TT, -TG, and -GG were 63, 32, and 5%, respectively. The SNP allele frequencies were calculated from genotype data. For rs12979860C and -T alleles, the frequencies were 51 and 49%, respectively, and for rs8099917G and -T alleles, the values were 21 and 79%, respectively. The allele frequencies were analyzed according to skin color stratification, although statistically significant differences were not found.

TABLE 2.

Genotype Distribution and Allele Frequencies (%) of IL-28B rs12979860 and rs8099917 in Chronic HCV Cohort

IL-28B SNP	Skin color (n)	Genotype				Allele frequency, %	OR (CI 95%)	P
rs12979860		CC	CT	TT	C	T
	White (45)	15	21	9	57	43
	Black (13)	1	6	6	30	70	2.8 (0.6–14.6)	0.12
	Mestizo (15)	6	4	5	53	47	1.1 (0.2–4.2)	1
	All (73)	22	31	20	51	49
rs8099917		TT	GT	GG	G	T
	White (45)	26	15	4	25	75
	Black (13)	8	5	0	19	81	0.5 (0.05–3.2)	0.7
	Mestizo (15)	12	3	0	0.1	0.9	0.4 (0.03–2.2)	0.3
	All (73)	46	23	4	21	79

The combined assessment of both SNPs resulted in 3 major genotypes (rs12979860CC/rs8099917TT, rs12979860CT/rs8099917TT, and rs12979860CT/rs8099917GG) with frequencies of 30, 22, and 21%, respectively. The remaining genotypes for combined loci were less common, and variants rs12979860CC/rs8099917GT, rs12979860CC/rs8099917GG, and rs12979860CT/rs8099917GG were not observed. In Fig. 1, the relation of SVR rate with genotype frequencies is shown. Among single genotypes, the homozygous rs12979860CC and rs8099917TT were the major responder variants, followed by rs12979860CT and rs12979860TT. Combined assessment revealed that in chronic HCV patients carrying the homozygous rs12979860CC genotype, the additional genotyping of rs8099917 did not have a significant effect on the prediction of SVR rate; in fact, all homozygous rs12979860CC was also genotyped as homozygous rs8099917TT. Nevertheless, in patients with heterozygous variant rs12979860CT, additional genotyping of rs8099917 contributed to increase the SVR rate, mainly when combined with the rs8099917TT variant. The single rs12979860CT achieved 41.9% of SVR, and when combined with rs8099917TT, the percent was increased to 50%, although the difference was not statistically significant. Furthermore, when homozygous variant rs12979860TT was combined with rs8099917TT, an increase in the percent of SVR form 35 to 50% was observed. On the contrary, when rs12979860CT was combined with rs8099917TG, a slight decrease in SVR rate was observed, from 45% for rs12979860CT to 33% for rs12979860CT/rs8099917TG.

Figure 1.

Contribution of single rs12979860 and rs8099917 genotyping and of combined genotypes on SVR rate in a genotype 1 chronic HCV cohort.

Characteristics of chronic HCV genotype 1b patients enrolled in the clinical study are shown in Table 3. Among all of the 73 patients, 32 (43.8%) achieved SVR. From univariate statistical analysis, the factors favoring SVR were in a less fibrosis stage (F0–F2) and complete early virological response (cEVR). cEVR is another term, used to guide HCV treatment, and indicates HCV RNA undetectable after 12 wk of treatment. The SVR rates associated with IL-28B genotypes were 55, 42, and 35% for rs12979860CC, -CT, and -TT, respectively, and 52, 30, and 25% for rs8099917TT, -GT, and -GG, respectively. Our analysis showed an almost statistically significant relation between rs8099917TT and SVR (P = 0.06) and a less significant relation in the case of the rs12979860CC variant (P = 0.2).

TABLE 3.

Baseline Characteristics of the Chronic HCV Cohort According to Treatment Outcome

Variable	Total (n = 73)	SVR (n = 32)	N-SVR (n = 41)	P
Sex, n (%)				0.16
Male	32 (43.8)	11 (34.3)	21 (51.2)
Female	41 (56.2)	21 (65.7)	20 (48.8)
Fibrosis stage, n (%)				0.008
F0–2	52 (71.2)	28 (87.5)	24 (58.5)
F3–4	21 (28.8)	4 (12.5)	17 (41.5)
Age, yr (mean ± sd)	46.4 (7.4)	46.2 (7.4)	46.5 (7.5)	0.83
BMI, kg/m2 (mean ± sd)	26.4 (3.9)	26.4 (4.0)	26.4 (4.0)	0.94
Skin color				0.14
White	45 (61.6)	23 (71.8)	22 (53.6)
Black/mestizo	28 (38.4)	9 (28.2)	19 (46.4)
EVR, n (%)	21 (28.7)	19 (59.3)	2 (4.8)	7.71 × 105

N-SVR, Non-SVR; P, significance value.

The associations between SVR and IL-28B genotypes in the chronic HCV cohort stratified by clinical groups are shown in Table 4. Among all groups, L6 resulted with higher SVR, with 58%, followed by groups E9, control, E6, and L9 with 50, 46, 33, and 12.5%, respectively. Coincidently, the poorly responder group L9 showed the less frequency (25%) for rs8099917TT, with significant differences (Fisher’s exact test P < 0.05) when compared with the other groups.

TABLE 4.

Association between SVR and IL-28B Genotypes in the Chronic HCV Cohort Stratified by Groups Included in the Clinical Study

IL-28B SNP	Genotype	SVR % (n)
		Control group	E6 group	E9 group	L6 group	L9 group
rs12979860
	CC	67 (9)	25 (4)	50 (4)	100 (3)	0 (2)
	CT	42 (12)	67 (3)	43 (7)	40 (5)	25 (4)
	TT	29 (7)	0 (2)	60 (5)	50 (4)	0 (2)
rs8099917
	TT	53 (19)	33 (6)	60 (10)	67 (9)	0 (2)
	TG	33 (9)	50 (2)	50 (4)	0 (2)	17 (6)
	GG	–	0 (1)	0 (2)	100 (1)	–
	All	46 (28)	33 (9)	50 (16)	58 (12)	12.5 (8)

According to different studies, the prediction of clinical outcome using molecular biomarkers will drive advances in risk, diagnosis, and prognosis, which will be an integral part of physician–patient shared decisionmaking, improving clinical care and patient outcomes.²² We explored the possibility to improve outcome prediction of HCV therapy by using multivariate analysis, which includes different clinical and molecular biomarkers as independent variables. For this purpose, a logistic regression developed by statistician Davis Cox²³ in 1958 was modeled. This binary logistic model is used to estimate the probability of a binary response based on 1 or more predictor or independent variables. The logistic model was adjusted for covariates age, fibrosis, γ-GT/ALT, skin color, sex, and IL-28B SNPs. Among all studied independent variables, only fibrosis, γ-GT/ALT, and sex were included in the stepwise logistic model for SVR prediction. The final model was described by the following equation:

$$\log \left(odds\right)=-0.73+1.57\ast fibrosis+1.03\ast sex-1.2\ast \gamma \text{-}GT/ALT.$$

Reference categories for covariate fibrosis were histologic stage F3–4 and for male individuals. Based on these coefficients, the ORs for fibrosis, sex, and γ-GT/ALT were 4.8, 2.8, and 0.3, respectively. The model was able to predict correctly 81.7% out of a total of 73 samples, with 87.1% sensitivity and 77.5% specificity.

DISCUSSION

The prevalence of IL-28B genotypes rs12979860 and rs8099917 was assessed in a sample of HCV chronic-infected patients by simultaneous detection of these 2 IL-28B loci. To our knowledge, this is the first report on the distribution of IL-28B SNPs rs12979860 and rs8099917 in the Cuban HCV-infected population.

The rs12979860C allele frequency has been studied in worldwide populations. Thomas et al.³ genotyped 2371 individuals from 51 populations, and among them, there were 5 South American populations (Ticuna, Karitiana, Surui, Guihiba speakers, and Quechua) where the frequencies of rs12979860 C ranged from 20% for the Ticuna population to 82% for the Karitina population. In our sample, we found a frequency of 51% for the rs12979860C allele, which was in the range above the South American populations. The global rs12979860C allele frequency showed a striking pattern, varying markedly across ethnic groups. The highest frequencies have been found in the East Asian group (>90%), followed by Europe-American and Hispanic (60–70%) groups and finally, as the minor allele in the African-American group (40%).⁴ According to skin color stratification, we found that the rs12979860C allele was more frequent in white, followed by mestizo, and finally, in black, with 57, 53, and 30%, respectively. In our study, the allele rs8099917T was the most frequent, with a prevalence of 79% (see Table 2).

The distribution of IL-28B SNPs rs12979860 and rs8099917 obtained in Cuban HCV-infected population was similar to the distribution obtained in a study conducted by our group with 400 healthy donors from different Cuban provinces (unpublished results). It is interesting that for healthy donors, the overall genotype distribution of IL-28B rs12979860CC, -CT, and -TT was 30, 49, and 21%, respectively, and the distribution for rs8099917TT, -TG, and -GG was 65, 32, and 3%, respectively. The allele frequencies for rs12979860C and -T alleles were 55 and 45, respectively, and for rs8099917G and -T alleles were 20 and 80%, respectively. This distribution is very similar to the one obtained in the cohort of 73 chronic HCV patients (P < 0.05).

According to the findings of previous studies,^3–6 the rs12979860CC genotype predicts response, and the -CT and -TT genotypes predict nonresponse. Meanwhile, the rs8099917TT genotype predicts response, and the -GT and -GG genotypes predict nonresponse. To evaluate the use of the IL-28B SNPs to predict SVR in our cohort of genotype 1b chronic HCV patients, we compared the sensitivity and positive predictive value (PPV) for SVR. The genotype rs8099917TT had a higher sensitivity (75% CI: 56–88%) and PPV (52% CI: 34–67%) than rs12979860CC, showing a 33% sensitivity (17–52% CI), and a 45% PPV (24–67% CI). The ability to predict treatment failure was also studied; PPV for treatment failure was higher in the case of minor nonresponder homozygous rs8099917GG (75.0% CI: 19–99%) than in nonresponder rs12979860TT (65.0% CI: 40–84%), although the CIs reveal that at the population level, they could be equivalent. According to our results, the rs8099917 showed a higher PPV for treatment success and failure, which is partially in agreement with results obtained by the International Hepatitis C Genetics Consortium.²⁴ It examined a cohort of 941 previously untreated patients of European descent with genotype 1 chronic HCV and found that rs12979860CC had a higher PPV for treatment success (64% vs. 55% for rs8099917), but in contrast, for treatment failure, rs8099917GG had a higher PPV (73% vs. 69% for rs12979860). In addition, the highest percent of SVR was obtained for rs12979860CC (55%), followed by -CT and -TT variants, and in the case of rs8099917, the highest percent of SVR was observed in -TT variant (52%), followed by -TG and -GG. Similar results were obtained by others authors.^{3–6, 9, 25}

In our analysis, both SNPs complied with the Hardy-Weinberg equilibrium (rs12979860 P = 0.26; rs8099917 P = 0.8). Both SNPs also were in LD, with D′ = 0.88 and r² = 0.22. The value of r² was relatively low, as a result of the differences in the alleles’ frequencies at 2 loci, and for that reason, both SNPs could not be interchangeable for the SVR prediction. In fact, when an individual carried the rs12979860CC variant, an additional genotyping of rs8099917 did not increase the chance for SVR; nevertheless, if the individual carried rs12979860CT or -TT variants, then complementary genotyping of rs8099917 could improve the prediction of SVR (Fig. 1). In our HCV cohort, carriers of rs12979860CT represented 42% of all patients, and among them, 52% displayed the homozygous responder rs8099917TT variant. Conversely, carriers of rs12979860TT represented 27% of all patients, and among them, 40% displayed the homozygous responder rs8099917TT variant. Because of the highest PPV for SVR associated with rs8099917TT and its high prevalence found in our HCV cohort (63%), the combined determination of rs8099917 and rs12979860 is more informative for predicting SVR. In summary, we observed that combined genotyping of rs12979860 and rs8099917 SNPs increased the probability for predicting SVR in genotype 1b HCV patients treated with triple therapy IFN-α/R plus CIGB230 in the Cuban population. Our results are in agreement with previous reports^{11, 12} and contradict those by Halfon et al.,²⁶ as rs12979860CC was insufficient to predict the response to IFN.

With the analysis of SVR rate in all clinical study groups, it was found that SVR was more probable in the late add-on group L6 (58%) than in L9 (12.5%; see Table 4). We hypothesized that the low SVR rate, observed in L9, could have been caused by the nonhomogeneous distribution of responder variant rs8099917TT among all groups. These results are in agreement with claims of the critical role of population characteristics in determining treatment outcome, as several host factors have been found to be associated with response to therapy.^27–29 For this reason, we suggest a proportional distribution of patients, according to the prevalence of major responder homozygous variants rs12979860CC and rs8099917TT, for further clinical studies.

To improve outcome prediction of HCV therapy, different clinical and genomic data have been combined in logistic models for better predicting response to combination therapy with pegIFN-α/R in genotype 1 chronic hepatitis C.^{14, 30} In this sense and considering the antecedents, we modeled a logistic binary regression using age, fibrosis, γ-GT/ALT, skin color, and sex as independent variables in combination with IL-28B SNPs. The more precise and parsimonious model that better predicted SVR in our chronic HCV cohort includes only fibrosis, γ-GT/ALT, and sex. It is noteworthy that the sex was not a significant variable in univariate analysis; nevertheless, when this variable is introduced in the logistic model, the sensitivity is increased from 80 to 87.1%, with an overall prediction rate of 81.7%, probably as a result of its partial correlation with the rest of variables. According to the coefficients of the model, an individual that has moderate fibrosis (F0–2) has almost 5 times more of a chance of archiving SVR than patients with severe stage (F3–4). The other clinical variable was γ-GT/ALT, which has been studied previously as a predictor parameter. Mihm et al.^{13, 21} demonstrated a preferential virological response to IFN-α depending on HCV genotype and the ratio of pretreatment γ-GT/ALT, suggesting that when patients are infected with genotype 1b, and the pretreatment γ-GT/ALT ratio is below 0.4, they should have a good prognosis for an initial virological response to IFN-α. In our model, we obtained OR of 0.3 for pretreatment γ-GT/ALT, which means that for every unit increase in the pretreatment γ-GT/ALT ratio, the likelihood for archiving SVR decreases almost 3 times. With the use of receiver operating characteristic analysis (data not shown), we did not find a cut-off value with high sensitivity for SVR and specificity for non-SVR; in fact, below 0.4, only 19% of all SVR patients was classified as positive. Nevertheless, in our logistic model, the γ-GT/ALT ratio achieved the highest significance as a predictor parameter with P = 0.007. We suggest validation above the exploratory model in future clinical studies.

In conclusion, we found that the frequencies of rs12979860C and -T alleles were 51 and 49%, respectively. For rs8099917 alleles, the frequency difference was more striking: 79 and 21% for T and G, respectively. In this studied Cuban chronic HCV population, the responder homogeneous variant rs8099917TT is the most frequent genotype, with a higher sensitivity and PPV for SVR. The combined genotyping of 2 IL-28B SNPs could improve the prediction of SVR in HCV genotype 1b patients with allele variants rs12979860CT or -TT, which could be useful to guide therapeutic decisions and improve treatment management.