Impact of variant-level batch effects on identification of genetic risk factors in large sequencing studies

Genetic studies have shifted to sequencing-based rare variants discovery after decades of success in identifying common disease variants by Genome-Wide Association Studies using Single Nucleotide Polymorphism chips. Sequencing-based studies require large sample sizes for statistical power and therefore often inadvertently introduce batch effects because samples are typically collected, processed, and sequenced at multiple centers. Conventionally, batch effects are first detected and visualized using Principal Components Analysis and then controlled by including batch covariates in the disease association models. For sequencing-based genetic studies, because all variants included in the association analyses have passed sequencing-related quality control measures, this conventional approach treats every variant as equal and ignores the substantial differences still remaining in variant qualities and characteristics such as genotype quality scores, alternative allele fractions (fraction of reads supporting alternative allele at a variant position) and sequencing depths. In the Alzheimer’s Disease Sequencing Project (ADSP) exome dataset of 9,904 cases and controls, we discovered hidden variant-level differences between sample batches of three sequencing centers and two exome capture kits. Although sequencing centers were included as a covariate in our association models, we observed differences at the variant level in genotype quality and alternative allele fraction between samples processed by different exome capture kits that significantly impacted both the confidence of variant detection and the identification of disease-associated variants. Furthermore, we found that a subset of top disease-risk variants came exclusively from samples processed by one exome capture kit that was more effective at capturing the alternative alleles compared to the other kit. Our findings highlight the importance of additional variant-level quality control for large sequencing-based genetic studies. More importantly, we demonstrate that automatically filtering out variants with batch differences may lead to false negatives if the batch discordances come largely from quality differences and if the batch-specific variants have better quality.

Impact of Variant-level Batch Effects on Identification of Genetic Risk Factors in Large Sequencing Studies

Background: Genetic studies have shifted to sequencing-based rare variants discovery after decades of success in identifying common disease variants by Genome-Wide Association Studies using Single Nucleotide Polymorphism chips. Sequencing-based studies require large sample sizes for statistical power but often inadvertently introduce batch effects because samples are typically collected, processed, and sequenced at multiple centers. Conventionally, batch effects are first detected and visualized using Principal Components Analysis and then controlled by including batch covariates in the disease association models. For sequencing-based genetic studies, because all variants included in the association analyses have passed quality control measures, this conventional approach treats every variant as equal and ignores the substantial differences still remaining in variant qualities and characteristics such as genotype quality scores, alternative allele fractions (fraction of reads supporting alternative allele at a variant position) and sequencing depths. Results: In the Alzheimer’s Disease Sequencing Project (ADSP) exome dataset of 9,904 cases and controls, we discovered hidden variant-level differences between sample batches of three sequencing centers and two exome capture kits. Although sequencing centers were included as a covariate in our association models, we observed differences at the variant level in genotype quality and alternative allele fraction between samples processed by different exome capture kits that significantly impacted both the confidence of variant detection and the identification of disease-associated variants. Furthermore, we found that the association signals of a subset of top disease risk variants came exclusively from samples processed by one exome capture kit that was more effective at capturing the alternative alleles compared to the other kit. Conclusions: Our findings highlight the importance of additional variant-level quality control for large sequencing-based genetic studies. More importantly, we demonstrate that automatically filtering out variants with batch differences may lead to false negatives if the batch discordance came largely from quality differences and if the variants from one batch had better quality scores.

Molecular Analysis in Patients with Idiopathic Erythrocytosis

Background: Erythrocytosis corresponds to an increase in the red blood count, hemoglobin (Hb) concentration and hematocrit (Htc) above the reference range adjusted to age, sex and living altitude. Clinical and molecular data on patients suffering from primary erythrocytosis are sparse. Due to its rare incidence, primary erythrocytosis frequently represents a challenge for the diagnosis. The management of such patients is demanding as no clear standards of treatment have been defined. Objective: We aimed to analyze the clinical and molecular features of patients with idiopathic erythrocytosis at our clinic. Methods: We evaluate in our internal registry patients encoded as erythrocytosis (primary or secondary) which were currently in follow-up. All data was obtained from the clinical records of the patients. For patient selection, we focused on those in which a JAK2 determination was available. We identified 102 patients from them 78 patients were JAK2 positive and were excluded in this analysis. 24 patients were initially selected as possible idiopathic erythrocytosis, of which 8 patients were excluded due to possible secondary causes. Patients in whom the diagnosis of primary erythrocytosis was suspected were studied with a 13 genes NGS panel. The analyzed genes included EPOR (exons 7,8), VHL (orf including exon 1'), EGLN1, EPAS1, EPO (including several regulatory regions), JAK2 (exons 12-16), BPGM, HBB, HBA1, HBA2, HIF3A, OS9, SH2B3 (somatic). Only gene variants of the categories pathogenic, likely pathogenic and variant of unknown significance (VUS) according to ACMG guidelines 2015 are mentioned. Results: We identified 16 patients (15%) that did not fulfill the criteria for polycythemia vera nor had a secondary underlying cause of polycythemia and were diagnosed as idiopathic erythrocytosis. The median age was 32 years (r 20-76) and 81% (13/16) were males. The median Hb-value was 173 g/l (r 142-192), Htc 50% (r 43-65%). The median erythrocyte concentration was 5.87 G/l (r 5.18-6.57 G/l). No thromboembolic complications occurred in any of these patients either before or after diagnosis. Eleven patients (69%) were initially treated with venesection, which was interrupted during the follow-up except in two cases. The median follow-up in our clinic was 43 months (r 1 - 132). The median follow up since the interruption of venesection was 23 months (r 12-106). There was no evolution to leukemia, a myeloproliferative disorder or any kind of tumor. In the results of the molecular analysis we found that the EPO: c.-1306C>A, rs1617640 reference allele was never homozygous. Indeed, all 16 patients showed an alternative allele: nine patients were heterozygous and 8 homozygous. EPO c.*772G>T, rs551238 was similarly distributed: no patients homozygous for the reference allele, 8 heterozygous and 8 patients homozygous for the alternative allele. In 5 patients other gene variants were also detected: in two patients EGLN1: c.1088T>G, p.(Leu363Arg) (the nucleotide is highly conserved, VUS, likely pathogenic), the other three were VHL: c.340+648T>C (reported as quite frequent, so more likely VUS, but it is within the alternate exon 1 region described by Lenglet et al. thus further studies are recommended), EPO: c.-1206C>A (not rare, VUS), EPO: c.*656G>A (downstream variant, very rare, unknown significance) respectively. No other mutations were identified. Conclusion: In our clinic idiopathic erythrocytosis showed a population of patients predominantly constituted of young men, had an indolent evolution, neither presented with thrombotic complications before or after the diagnosis, nor with a hematological transformation. The high prevalence of EPO gene variant underlines the role of this association. The five patients with combined mutations did not show a different clinical pattern. The use of NGS to evaluate patients with unclear erythrocytosis might allow an improvement in clinical management avoiding unnecessary interventions. Disclosures Jalowiec: Novartis: Other: Travel grant; Amgen: Other: Travel grant; Pfizer: Other: Travel grant.

Polymorphism of the C-509 gene of TGFB1 and the risk of obliterative atherosclerosis of lower limb arteries

Genetic factors play an important role in the development of obliterative atherosclerosis of arteries in the lower extremities. There are numerous data in the literature on polymorphic variants of the TGFB1 gene associated with the risk of development of cardiovascular diseases, but some of them are contradictory. The present study aimed at elucidating the role ofpolymorphism of509C/T genes of TGFB1 in the development of obliterative atherosclerosis in arteries of the lower extremities. The material included 328 samples of DNA from 203 unrelated subjects of Russian ethnicity residing in Central Russia (177 men and 23 women) with obliterative atherosclerosis of arteries in the lower extremities hospitalized to the Dpt. of Vascular Surgery and 125 age-matched controls. Genotyping of polymorphism of the C-509T (rs1800469) gene of TGFB1 was performed by the PCR-PDRF method with the use of oligonucleotide primers (Sintol Co., Moscow). The patients with obliterative atherosclerosis showed an increased frequency of the occurrence of alternative allele of the C-509T gene of TGFB1 in comparison with control subjects (r=0,0001). Simultaneously, the frequency of homozygous genotypes of wild-type alleles of -509CC decreased (OR=0,58, 95%CI=0,37-0,92, P=0,02) while the frequency of homozygotes with an alternative allele -509TT increased (OR=4,93, 95%CI=5,20-11,03, P=0,00004). The difference in the frequency of heterozygotes 509CT was absent (OR=0,81, 95%CI=0,52-1,27, P=0,35). Thus, polymorphism of the C-509T gene of TGFB1 is a significant predictor ofpredisposition to the development of obliterative atherosclerosis in arteries of the lower extremities.

Stress Erythropoiesis and Genetic Regulation of Fetal Hemoglobin in Inherited Bone Marrow Failure Syndromes

Abstract 2401 Patients with inherited bone marrow failure syndromes (IBMFS) frequently have manifestations of what has been called “stress erythropoiesis”. This includes macrocytosis (increased mean cell volume, MCV), increased fetal hemoglobin (Hb F) and erythropoietin (Epo) levels higher than predicted by the degree of anemia (red blood cell count, RBC). In patients with hemoglobinopathies (sickle cell disease, thalassemia, and hereditary persistence of fetal hemoglobin), Hb F levels are regulated by 3 quantitative trait loci (QTL), located at HBSIL-MYB on chromosome 6q, BCLIIA on chromosome 2p and XMN1-Gg representing HBB cluster on chromosome 11p. The role of these QTLs in the elevated Hb F levels in patients with an IBMFS has not been previously reported. Percent Hb F was measured by HPLC in blood from 97 untransplanted individuals with an IBMFS. Absolute Hb F (g/dL) was calculated by multiplication of Hb F% times total Hb in order to include data from transfused patients, and log-transformed to approximate a normal distribution. Epo levels were also log-transformed due to the wide range (8 to 1800 mU/mL). DNA was extracted from leukocytes, and candidate regions were amplified and genotyped by TaqMan. The candidate QTLs were evaluated by genotyping of tagging single nucleotide polymorphisms (SNPs): five for HBSIL-MYB, two for BCLIIA, and one for XMN1-Gg. Data were modeled using a generalized linear model (GLM), appropriate for data with a constant coefficient of variation. There were 31 patients with Diamond-Blackfan anemia (DBA), 35 with dyskeratosis congenita (DC), 25 with Fanconi anemia (FA), and 6 with Shwachman-Diamond syndrome (SDS). Hb F was elevated in 70% of the total group of patients: 48% of DBA, 83% of DC, 76% of FA, and 83% of SDS. In the pooled group of 97 IBMFS patients, 68 (70%) had Hb F >1 % (upper limit of normal), 59 (61%) were macrocytic, 55 (57%) were anemic for age, and 70 (77%) had elevated Epo. The frequencies of heterozygosity or homozygosity for the alternative alleles for the QTLs were 50% for HBSIL-MYB, >90% for BCLIIA, and 52% for XMN1-Gg. The multivariate model for Hb F in the total goup of IBMFS included the alternative allele for the XMN1-Gg SNP (p = 0.04), younger age (p<0.001), male sex (p=0.04), and increased Epo (p<0.001). In this model, the alternative allele for the XMN1-Gg QTL was associated with a 32% increase in the level of Hb F. Subset analyses indicated that the strongest association of the XMN1-Gg QTL was in FA and DC (increased Hb F by 68% and 48% respectively, p-values 0.02 and 0.09) and had no effect in DBA (decreased Hb F by 18%, p = 0.6). Data including the other QTLs were not significant. These results suggest that the alternative allele at XMN1-Gg is associated with the increased level of Hb F in FA and DC, but not DBA, after adjustment for age, sex, and Epo level. A low level of Hb F should not exclude the diagnosis of an IBMFS in a patient who has other signs of stress erythropoiesis (anemia with increased MCV and Epo), since that patient may not have the variant allele associated with increased Hb F. The degree of elevation of Hb F in FA and DC depends on the alleles at the XMN1-Gg QTL. A strength of this study is the sample size of almost 100 patients with an IBMFS who are well-characterized. A limitation is that the number within each syndrome is still small; the role of the other QTLs may be identified in future larger studies. Of major interest is that this is the first study to show regulation of Hb F by the same QTL in FA and DC as the common hemoglobinopathies, thus linking Hb F regulation across disparate hematologic disorders. Disclosures: No relevant conflicts of interest to declare.

CD38 Gene Polymorphisms Contribute to Genetic Predisposition to B-Cell Chronic Lymphocytic Leukemia.

Genetic background and abnormal immune response are supposed to play a role in the pathogenesis of B-cell chronic lymphocytic leukaemia (B-CLL). CD38 is a leukocyte activation antigen and ectoenzyme involved in numerous immune functions as well as important prognostic factor in B-CLL. The aim of this study was to investigate potential association of two single nucleotide polymorphism (SNPs) located in CD38 gene at position 182 (C/G) in intron 1 and at position 418 (C/T) in exon 3 with genetic susceptibility to B-CLL. Furthermore, we wanted to verify whether these SNPs influence CD38 expression or clinical outcome of B-CLL patients. Genotyping of 365 Caucasian individuals (163 B-CLL patients and 202 healthy controls) was performed by restriction fragment length polymorphism (RFLP) and allele-specific polymerase chain reaction (ASPCR) assays. CD38 gene expression was quantified by real-time RT-PCR and CD38 protein expression was measured by flow cytometry in circulating B-CLL cells from 50 patients and normal B and T lymphocytes from 30 healthy controls. We found that allele frequency and distribution for both CD38 SNPs differed significantly in patients and controls with increased prevalence of alternative alleles detected in B-CLL (allele 182G found in 34.7% B-CLL patients vs. 13.4% controls, p<0.0001, and allele 418T found in 4.6% B-CLL patients vs. 0.5% controls, p=0.0002). Logistic regression analysis regarding SNP at position 128 showed that the 182CG heterozygotic genotype conferred risk of B-CLL (odds ratio, OR=3.8, 95% confidence interval, CI=2.4–6.1) that was further increased in alternative allele homozygotes (OR=7.3, 95%CI=2.8–19.1), p<0.0001. Similarly, compared to the individuals with wild-type 418CC genotype, carriers of heterozygotic 418CT genotype and alternative allele 418TT homozygotes were at higher risk of B-CLL (OR=7.3 and OR=240, respectively, p=0.032). Interestingly, we did not observe any differences in CD38 gene or protein expression in B-CLL or normal lymphocytes related to genotypes at position 128 or 418, however, this could be due to low numbers of rare genotypes found among patients included to the expression tests. Finally, the survival analysis showed that SNP at position 128 of CD38 gene influenced overall survival (OS) of B-CLL patients, with the best prognosis observed in carriers of 128CG genotype (p=0.05). In conclusion, our data indicate that inherited SNPs located in the CD38 gene are associated with genetic susceptibility to B-CLL in Caucasian patients, and may have some impact on prognosis of this malignancy.