Objectives: To establish the identification method of tumor tissue origin based on commonly used STR typing kits.
Methods: ForenSeqTM DNA Signature Prep kit was used to detect the typing of 27 autosomal STR loci in 55 paired tumor tissue samples (tumor tissue paired with normal tissue of the same individual) and 75 unrelated individual whole blood samples. The genotyping data of full sibling pairs and parent-child pairs of 55 tumor tissues were simulated. The number of total identical alleles (An) and identity by state (IBS) scores were calculated within the paired carcinoma (PC), the tumor-unrelated individual (UI), the tumor-simulated full sibling (FS) and the tumor-simulated parent-offspring (PO) groups. The tumor tissue origin identification and prediction models of 8 commonly used STR typing kits were established based on the statistical results obtained above, and an attempt was made to establish a specific model for tumor tissue origin identification. The accuracy, sensitivity and specificity were verified and evaluated using the additional 23 paired tumor tissue samples.
Results: (1) In any kit, there was no statistically significant difference in the number of loci shared with 0 identical allele (A0) between the PC and PO groups. The number of loci shared with 1 identical allele (A1), 2 identical alleles (A2), and IBS scores were statistically significant different between the PC group and the UI, FS and PO groups. (2) The An and IBS scores of different STRs varied in different groups. The A2 levels of 13 STRs (CSF1PO, D12S391, D19S433, D20S482, D2S1338, D3S1358, D4S2408, D7S820, D8S1179, FGA, TH01, TPOX, vWA) in PC group were higher than those of other STR loci. The A2 levels of two STRs (D6S1043, Penta E) in UI group were significantly lower than those of other STR loci. (3) The tumor tissue origin identification and prediction models of 8 commonly used STR typing kits and the identification model of tumor tissue origin with 15 STR loci (15-STRs) were successfully established, with sensitivity of 100%, specificity of 97.56%-99.88%, and accuracy of 97.59%-99.89%. Among them, the 15-STRs model had 100% sensitivity, 99.88% specificity, and 99.89% accuracy, which were higher than those of commonly used commercial kits.
Conclusions: This study successfully establishes the tumor tissue origin identification methods with 8 commonly used STR typing kits, which expands the application of tumor tissue origin identification. In addition, the differences of different loci in the identification of tumor tissue origin were compared, and 15 STR loci which were particularly suitable for the identification of tumor tissue origin were selected, providing the data basis for the establishment of tumor origin tracing kits in future.
目的: 建立基于常用STR分型试剂盒的肿瘤组织身源鉴定方法。方法: 采用ForenSeqTM DNA Signature Prep试剂盒检测55例配对肿瘤组织样本（肿瘤组织和同一个体正常组织成对）以及75例无关个体全血样本27个常染色体STR基因座的分型情况，并模拟55例肿瘤组织的全同胞、亲子对分型数据，统计成对肿瘤（paired carcinoma，PC）、肿瘤-无关个体（tumor-unrelated individual，UI）、肿瘤-全同胞（tumor-simulated full sibling，FS）与肿瘤-亲子（tumor-simulated parent-offspring，PO）的共有等位基因个数（number of total identical alleles，An）及状态一致性（identity by state，IBS）评分。以上述统计结果作为参照，建立8个常用STR分型试剂盒的肿瘤组织身源鉴定预测模型，并尝试构建一个专用于肿瘤组织身源鉴定的模型。使用另外23例配对肿瘤组织样本的检测结果对鉴定模型的准确性、灵敏度及特异度进行验证与评估。结果: （1）在任一试剂盒中，全不同基因座数量（A0）在PC组与PO组之间差异无统计学意义。1个相同基因座数量（A1）、2个相同基因座数量（A2）和IBS评分在PC组与UI、FS、PO组之间差异均有统计学意义。（2）不同STR基因座的An与IBS评分在不同组别存在差异，其中，13个STR基因座（CSF1PO、D12S391、D19S433、D20S482、D2S1338、D3S1358、D4S2408、D7S820、D8S1179、FGA、TH01、TPOX、vWA）的A2在PC组均高于其他STR基因座；2个STR基因座（D6S1043、Penta E）的A2在UI组低于其他STR基因座。（3）成功构建了8个常用STR分型试剂盒的肿瘤组织身源鉴定预测模型以及15个STR基因座的肿瘤组织身源鉴定模型（15-STRs），灵敏度均达100%，特异度为97.56%~99.88%，准确度为97.59%~99.89%。其中，15-STRs模型的灵敏度为100%，特异度为99.88%，准确率为99.89%，高于常用商业化试剂盒。结论: 本研究成功建立了8个常用STR分型试剂盒的肿瘤组织身源鉴定方法，拓展了肿瘤组织身源鉴定的应用范围。通过比较不同基因座在肿瘤组织身源鉴定中的差异，筛选出了15个特别适用于肿瘤组织身源鉴定的STR基因座，为未来肿瘤组织溯源的试剂盒构建提供了数据基础。.
Keywords: forensic genetics; identity by state (IBS); individual identification; next-generation sequencing; number of total identical alleles; prediction model; short tandem repeat (STR); tumor tissues.