A Study Of A NGS-SNPs Assay On Individual Identification Of Tumor Tissues

Tracing to source of tumor tissues is a challenge in clinical tumor treatment at home and aboard.The accurate correspondence of patients and the resected tumor tissues is crucial to accurate diagnosis and treatment of tumor patients.Specimens mislabeling,tissues mix-up and cross-contamination happen occasionally in department of pathology because of high cancer incidence,which not only lead to medical tangle impacting normal functioning of hospital but also perplex clinical tumor treatment impacting advance of clinical oncology.With increasing morbidity and mortality of tumor,there are more and more forensic cases involving tumor tissues.Many studies reported that mutations rate of STRs widely used in forensic investigations was high in tumor tissues resulting in different genotypes in comparison with that in normal tissues,which lead to false exclusionary conclusion.And the result of STRs mutations in tumor tissues from multiple systems in our laboratory confirmed the phenomenon.Thus,a kind of biomarker with low mutations is needed to be detected to perform source identification of tumor tissues.Compared with STRs,SNPs with low mutations rate(～10^-8)and small amplicon（～100bp）is more suitable for forensic identification of tumor tissues.However,due to the bi-allelic nature,more SNPs need to be genotyped to obtain similar power of STRs to perform individual identification.The next generation sequencing（NGS）could detect hundreds of SNPs in several samples simultaneously to obtain higher forensic power,which benefit from its capacity.Therefore,the SNPs in tumor tissues were detected by NGS in this study to explore SNPs mutations in tumor tissues and provide a new method for source identification of tumor tissues.Part 1 Evaluation of Precision ID Identity Panel on Ion Torrent PGM^TM sequencing platformObjective:To perform evaluation of the NGS-SNPs assay in our experiment.Methods:Genomic DNA in normal tissues from cancer patients was extracted and quantified.Libraries were constructed using Precision ID Identity Panel including 4 major steps:amplification of target regions in genomic DNA,partial digestion of primer sequence,ligation of DNA barcode and purification,and quantification of libraries.The libraries were mixed in equivolume to perform template preparation and sequencing according to the user guide.The raw sequencing data was processed on Ion Torrent Suite Server v5.0.2.The HID＿SNP＿Genotyper v4.3.2 was used to obtain SNPs genotype.The evaluation of the assay included locus coverage balance,locus strand balance,allelic balance and background noises.Control DNA 2800M was detected 3 times and compared with results of other laboratory to verify accuracy and reproducibility of the NGS-SNPs assay.The population data of the NGS-SNPs assay in Hebei Han was investigated using genotypes from normal tissues to provide basis for application of the assay.Results:For locus coverage balance,average coverage from autosomal SNPs was almost twice that from Y-SNPs.A total of 5 autosomal SNPs and 1Y-SNP were identified as outliers according to the respective mean±2SD.For locus strand balance,the average strand balance across all loci was 0.50±0.06.The threshold 0.50±0.20 was used to screen out imbalance loci,two SNPs were the outliers.For allelic balance,the major allele reads frequencies of all homozygote were above 90%.And the major allele reads frequencies of most loci were 50%～60%.Heterozygote imbalance was seen at 5 SNPs rs7520386,rs876724,rs214955,rs917118,and rs430046 with heterozygote ratio of 0.30,0.60,0.59,0.59 and 0.65,respectively.For background noises,it was significantly higher in alternative homozygote than in reference homozygote and heterozygote.Incorrect alleles were the major component of background noises.Most of the quality check flags were from low coverage（COV,41.08%）and strand imbalance（PPC,32.99%）.The genotypes of control DNA 2800M were the same in three detection,and the same with genotypes from other laboratory.In the 90 autosomal SNPs,the minimum allele frequencies were from0.083 to 0.490,which showed low polymorphism of some loci in Hebei Han.There was no significant deviation from HWE and LD at any locus after Bonferroni corrections（P=0.05/90=0.0026 and P=0.05/4005=0.00001248,respectively）.The combined discrimination power（CDP）was 1-2.568×10^-34,and the combined power of exclusion（CPE）was 0.99999993 for trio paternity testing and 0.99990027 for duo paternity testing in Hebei Han.A total of 8haplotypes were observed from 73 male samples,which were assigned as 8haplogroups including O3,N,O2,C,Q,O,D,E according to the number of observation.Part 2 SNPs in tumor tissues genotyped by the NGS-SNPs assayObjective:To find law of SNPs mutations in tumor tissues and investigate application of the NGS-SNPs assay to source identification of tumor tissues by detecting SNPs in tumor tissues using the NGS-SNPs assay.Methods:Genomic DNA in tumor tissues was extracted and quantified to perform libraries construction,template preparation and sequencing according to the manufacturer’s instructions.The raw data was processed in the same way with part 1.The differences of locus coverage balance,locus strand balance,allelic balance and background noises in normal tissues and tumor tissues were analyzed.SNPs mutations were detected by comparing the genotypes in tumor tissues and normal tissues.The unrelated individuals pairs（UIP）,parent-offspring pairs（POP）and full sibling pairs（FSP）were imitated based on allele frequencies according to binomial distribution.The IBS score and number of loci with 2 alleles（A2）,1 allele（A1）and 0 allele（A0）sharing within tumor-normal pair（TNP）,UIP,POP and FSP were calculated,respectively.Results:1. Comparison of sequencing data from tumor tissues with normal tissuesThe tendency of locus coverage balance in tumor tissues was consistent with that in normal tissues.The average locus coverage in tumor tissues was slightly higher than that in normal tissues.A total of 3 autosomal SNPs and 1Y-SNPs were identified as outliers according to respective mean±2SD.There was no significant difference of locus strand balance in tumor tissues and normal tissues.The locus strand balance was seen in the same two loci in tumor tissues and normal tissues.The major allele reads frequencies in tumor tissues were significantly different from that in normal tissues,especially the allelic imbalance in heterozygote.The major allele reads frequencies for homozygote in tumor tissues were above 90%.The major allele reads frequencies for heterozygote in tumor tissues were 50%～90%.There were genotypes with heterozygote imbalance at all SNPs.The quality check flags were major allele frequency（MAF,58.51%）,COV（11.63%）and PPC（16.96%）.There were significant difference in background noises only at several loci in tumor tissues and normal tissues,such as rs1355336,rs938283,rs733164.2.Establishment of a strategy to identify source of tumor tissues by the NGS-SNPs assayThere was only one mutations type at autosomal SNPs in tumor tissues in comparison with normal tissues--loss of heterozygosity（LOH）.There was no mutation at Y-SNPs.According to the distribution of A₀,A₁,A₂ and IBS score in TNP,UIP,POP and FSP,the threshold of 67 in A₂ was regarded to exclude tumors and unrelated individuals and 68 was regarded to confirm tumors and source individuals.The threshold of 157 in IBS score was regarded to exclude the relationship of tumors and unrelated individuals and 178 was regarded to confirm the relationship of tumors and source individuals.The frequencies of major allele reads of mutations in tumor tissues were 90%-95%by further analysis of these genotypes.The results of Sanger sequencing showed that the genotypes were heterozygote in these SNPs,and the peak height of the two alleles were significantly different.Thus,the genotype accuracy would be improved if the threshold of heterozygote and homozygote in tumor tissues was setted to 95%.Sanger sequencing is needed.The NGS-SNPs assay could generate a loyal representation of DNA samples.Based on the law of ratio of tumor cells and stromal cells and major allele frequencies in sequencing result,we established a new method to perform identification of tumor tissues using the NGS-SNPs assay.First,the ratio of tumor cells and stromal cells in a tumor tissue was evaluated.Second,Genomic DNA in the tumor tissue and a reference sample was exacted and quantified to perform NGS sequencing and obtain SNPs genotypes.The DNA of the tumor tissue was sequenced two times.Third,the genotypes of the reference sample and the tumor tissue were checked and determined respectively.It need to be noted that in the tumors with ratio of tumor cells and stromal cells less than 20%,the loci with F_MARvalue of homozygote in 90%-95%in two sequencing genotyped heterozygote and validated by Sanger sequencing.Fourth,compare the genotypes of the tumor tissue and the reference tissue to draw a conclusion by the rule:if they are consistent,it is supported that the two samples are from the same person;if they are inconsistent,it is supported that the two samples are not from the same person.For the tumors with percentage of stromal cells less than 10%,it is likely to detetct mutations showing F_MAR more than 95%,the A₂ and IBS score could be used to determine the source of tumors.Part 3 Validation of applied research on detection of specimens from cancer patients using the NGS-SNPs assayObjective:To explore applicability of the NGS-SNPs assay to specimens from cancer patients and investigate the population data of the NGS-SNPs assay in Hebei Han to provide basis for its application.Methods:Sensitivity studies were performed by detecting control DNA2800M at different input amounts by the NGS-SNPs assay to determine the least amount of input DNA and provide reference for detection of trace materials.Genomic DNA of the peripheral blood from cancer patients was detected by the NGS-SNPs assay to investigate the impact of cell-free DNA on genotypes,which provides instructions for detection of peripheral blood from cancer patients in cases.DNA of formalin-fixed tissues was extracted by Gene Read DNA FFPE Kit to investigate the impact of formalin fixation on genotypes,which provides instructions on methods of DNA extraction of formalin-fixed paraffin embedded tissues.DNA in different degraded degrees was detected by the NGS-SNPs assay to investigate the detection ability of the assay for degraded samples.A case of source identification of a tumor tissue was detected to validate the applicability of the NGS-SNPs assay to identification of tumor tissues.Results:1.The performance of locus coverage balance,locus strand balance and heterozygote balance were poor,even happened locus drop-out,allele drop-in and allele drop-out when input DNA<50 pg.It is suggested that input DNA was at least more than 100 pg,because whole and correct genotypes were detected when input DNA was 100 pg.However,considering low input DNA had potential effects on genotyping,the optimal input DNA would be 200pg～1 ng.2. The genotypes of peripheral blood from cancer patients were consistent with genotypes of normal tissues indicating that peripheral blood from cancer patients could be used as normal samples in cases.3. It did not detect the impact of artifect on SNPs genotypes by formalin fixation from DNA of formalin-fixed tissues exacted by Gene Read DNA FFPE Kit.It is suggested that DNA extraction from paraffin blocks of tumor tissues uses the Gene Read DNA FFPE Kit.4. The detection rate of the NGS-SNPs assay for the mild and moderate degraded samples with degradation index 1 to 10 was 100%,which was significantly superior to CE-STR.Detection rate for moderate degraded samples could be improved by increasing input DNA.The detection ability for severe degraded samples was limited.5. The result of a case involving source identification of a tumor tissue detected by the NGS-SNPs assay showed that the SNPs genotypes of the tumor tissue and peripheral blood from the appraised individual were consistent,which supported the tumor tissue was from the appraised individual.Conclusions:The Precision ID Identity Panel on Ion PGM sequencing platform is a well-performed and reliable assay to get accurate SNPs genotypes.However,it can benefit from some improvements.There is only one mutation type in tumor tissues SNPs--loss of heterozygosity（LOH）.There are significant differences of background noises and major allele reads frequencies in tumor tissues and normal tissues because of SNPs mutation.Based on the law of ratio of tumor cells and stromal cells and major allele reads frequencies in sequencing results,a new stragety to identify the source of tumor tissues by the NGS-SNPs assay is established.The validation of the stragety indicated that it provides a new idea and basis for source identification of tumor tissues in forensic investigation.