This study combined data from two Egyptian cohorts. The first included 3994 individuals from a large epidemiological study in a village of the Nile delta, the details of which have been described elsewhere 
. Briefly, in this cohort 491 individuals (12.3%) were seropositive for HCV and 305 of them had positive PCR test. None of the seropositive individuals have been treated at the time of the study. Ninety-six RNA samples had previously been sequenced in this cohort, and 95 (99%) of these samples were of HCV genotype 4 
. From the overall cohort, we selected 188 independent unrelated HCV-infected subjects defined on the basis of two positive serological tests 
. Individuals with positive PCR test for viral RNA were further classified as “chronically infected”, whereas those with negative results were considered to have cleared the virus spontaneously. Subjects of this cohort were not tested for HBV infection, but it could be noted that several studies in Nile Delta villages, including this one, reported a rather low prevalence on HBV infection (2%–8%) 
. The second cohort included 73 patients with symptomatic acute HCV infection recruited from Cairo hospitals in the context of a study that has been described elsewhere 
. In these subjects, spontaneous clearance was defined as the loss of HCV RNA from the serum in the absence of treatment, based on two consecutive negative viral PCR results, together with subsequent negative tests, including at least one more than one year after the acute episode. Chronic infection was defined as the presence of HCV RNA in the serum in all PCR tests over a period of at least one year after the acute episode. Twenty-seven RNA samples from this cohort had previously been sequenced, and 24 (89%) were of HCV genotype 4 
. Subjects of this cohort were tested for HBV infection, as previously described 
, and out of the 73 present patients, only one was found to be HBV infected. Finally, the overall sample of 261 subjects consisted of 130 subjects with spontaneous clearance and 131 with chronic infection. All these subjects were not tested for HIV infection, but HIV seroprevalence was found to be quite low (<0.1%) in the general Egyptian population aged of 15 to 49 years (http://www.unaids.org/en/regionscountries/countries/egypt
) as well as among injection drug users (~0.6%) 
. The study was approved by the Institutional Review Board of the Egyptian Ministry of Population and Health. Signed informed-consent documents were obtained from all study participants before they entered the study.
We genotyped a total of nine SNPs within the IL28B
genomic region (). We first included the three SNPs (rs12980275, rs12979860, rs8099917) most frequently reported to be associated with HCV clearance in previous studies. We then explored the long-range linkage disequilibrium (LD) (500 kilobases on either side of IL28B
) of these three SNPs, using the European CEU population available data from HapMap (http://www.hapmap.org
) and 1000 Genomes project 
databases. We selected the two most distant SNPs (rs958039, rs576832) in LD, with an r2
>0.3, with any of the three initial SNPs. Within the chromosome 19 region defined by these two SNPs, we included three additional regularly spaced SNPs (rs4803217, rs4803222, rs10424607) in LD, with an r2
>0.3, with any of the three initial SNPs. Finally, we also selected the non synonymous SNP rs8103142 not present in the public databases but reported to be in strong LD with rs12979860 in European populations 
. Illumina GoldenGate genotyping was carried out with VeraCode technology (Illumina). For genetic markers, we applied the following quality control criteria: call rate <90%, Hardy-Weinberg P
-value<0.005 or low-quality genotype clustering. No individuals were excluded on the basis of genotyping call rate.
SNPs genotyped in the IL28B genomic region and their association with spontaneous clearance.
Univariate tests of association between the genotyped SNPs and the clearance/persistence status of HCV infection were carried out using PLINK software 
. Multivariate analysis was performed with the LOGISTIC procedure implemented in SAS software v.9.2 (SAS Institute, Cary, North Carolina, USA). Pairwise LD between SNPs and haplotype frequencies were estimated with PLINK 
and Haploview 4.1 (http://www.broad.mit.edu/mpg/haploview