急性骨髓性白血病中不同年齡層基因變異之比較性分析

Abstract

論述重點 本研究旨在使用次世代基因定序技術（NGS），比較分析台灣兒童與成人急性骨髓性白血病（AML）患者的基因變異分佈情形，藉由鑑別不同年齡層在基因層面的異同，探討疾病潛在的致病機制。儘管急性骨髓性白血病在臨床上被視為同一類型的疾病，然而越來越多研究顯示，其分子致病機轉可能隨年齡而有所差異。兒童與成人 AML 在基因變異頻譜上呈現明顯不同，反映出兩者可能受到不同的生物致病機制所驅動。AML 為一種在臨床表現、治療反應與分子特徵上均具高度異質性的血液惡性腫瘤。隨著基因體技術的進展，成人 AML 的風險分層與治療策略已獲得明顯進展，包括特定基因變異的鑑定與標靶治療的應用。然而相較之下，兒童 AML 的研究發展相對不足。目前國際診斷與風險分層指引大多根據成人族群研 究資料建構，對兒童族群的適用性可能有限。因此，發展針對兒童 AML 的基因變頻譜分析，對於未來提升診斷準確性有助益。本研究透過比較兒童與成人 AML 患者的基因變異頻譜，期望鑑別 AML 相關之年齡特異性基因變異頻率 。藉由分析年齡相關與族群特異性變異分布，本研究有助於促進更精準的風險分層、優化治療策略，並支持發展針對兒童 AML 族群的精準醫療模式，進而提升整體治療成果與臨床應用價值。 方法 本研究納入 103 位診斷為急性骨髓性白血病（AML）之兒童患者，收案期間橫跨 12 年。AML 的診斷經由骨髓抹片細胞形態學檢查及流式細胞儀免疫表現型分析確認 。自診斷時取得之骨髓檢體中萃取基因體 DNA，並進行 52 個基因的目標定序分析，使用 Ovation Target Enrichment 系統（NuGEN）捕捉外顯子區域。這些基因選自先前有關兒童 AML 的研究，涵蓋主要致病路徑，包括訊息傳導、轉錄因子、表觀遺傳調控與細胞週期等。由於訊息傳導相關突變在兒童 AML 中較為常見，而表觀遺傳相關突變則多見於成人，故本研究依據此差異作為基因選擇的依據，以探索與年齡相關的基因變異特徵。原始定序資料經 FastQC 與Trimmomatic 進行品質控制 ， 並以 BWA 對齊至人類參考基因組 hg19， 接續使用 GATK 進行重校正與重複讀段標記。變異辨識以 VarScan 執行，並依品質與族群頻率進行過濾。所獲得之兒童 AML 基因變異資料與台大成人 AML 患者群進行比較， 並進一步以 TCGA 成人 AML公開資料集進行驗證。此外，亦依年齡進行分層分析，以觀察不同年齡層的突變模式變化。本研究兒童 AML 資料亦與公開的 TARGET 兒童 AML 資料集進行比較，以評估族群間基因變異頻率之差異。 結果與討論 本研究顯示平均每位患者可偵測到 4 個與疾病相關的基因變異。最常見的突變為 ZBTB7A（58%），其餘依序為 KMT2C（53.4%）、PML（44.3%）與 PTPN14（36.4%），突顯部分基因在本族群中的高頻突變現象。進一步比較兒童與成人 AML 患者的基因變異分佈，可見 兩群體在突變頻率上具有明顯差異。兒童族群中 WT1 突變比例顯著較高（16.5% vs. 9.5%，FDR q = 0.023） ， 而成人則以 DNMT3A （19.7% ， q = 0.001）、 NPM1 （23.4% ， q = 2.4×10⁻⁶）、IDH1/2 及 TET2 等突變為主，反映不同年齡層 AML 在致病機轉上可能存在差異。透過年齡分層分析，進一步觀察到基因突變分佈隨年齡變化的趨勢：兒童 AML 患者以訊息傳導路徑相關突變為主，而成人則以表觀遺傳與轉錄調控失衡為主，顯示 AML 在不同年齡階段可能由不同的生物機制所驅動。此外，將本研究結果與國際急性骨髓性白血病資料庫進行比較， 發現台灣兒童 AML 患者展現出獨特的基因變異特徵。ZBTB7A 突變在本研究兒童患者中高達49.5%，遠高於 TARGET 資料中的 0.7% （FDR q = 5.32×10⁻²²），WT1 亦顯著偏高 （14.6% vs. 3.3%，q = 0.014） ，呈現明顯的族群差異。整體而言，本研究結果不僅揭示 AML 於不同年齡 層的突變異質性，也強調考量年齡與族群背景，採取個別化診斷與風險分類策略的臨床重要性。 結論 台灣兒童急性骨髓性白血病的基因頻譜有其獨特性及多樣性，其基因變異頻率與成人患者及國外資料庫呈現明顯差異。本研究結果支持急性骨髓性白血病具有年齡與族群相關的變異特徵 ， 強調未來臨床診斷與治療策略應考量年齡與族群差異， 並朝向個別化醫療方向發展。

Background The objective of this study was to investigate the differences in gene mutation profiles between pediatric and adult acute myeloid leukemia (AML) patients in Taiwan using next-generation sequencing (NGS). Although AML is classified as a single disease entity, increasing evidence indicates that its underlying biology varies substantially with age. Pediatric AML has been shown to harbor distinct genetic alterations compared to adult AML, suggesting that the leukemogenic mechanisms differ between these two groups. AML is a clinically and genetically heterogeneous hematologic malignancy. While significant progress has been made in adult AML through the application of genomic technologies to guide risk stratification and treatment, similar advances in pediatric AML have lagged behind. This is partly due to its lower incidence and the limited number of large-scale pediatric-specific studies. Consequently, most diagnostic guidelines and risk classification systems—such as those from the European LeukemiaNet—are derived primarily from adult data and may not be directly applicable to pediatric patients. This study aimed to compare the mutational landscapes of Taiwanese pediatric and adult AML patients. Through identifying age-specific mutation patterns relevant to pediatric disease biology, this study hope to contribute to improved risk stratification, inform treatment strategies, and support the development of precision medicine tailored specifically to the pediatric AML population. Methods This study enrolled 103 pediatric patients diagnosed with AML at National Taiwan University Hospital over a 12-year period. Diagnosis was confirmed by bone marrow morphology and flow cytometry immunophenotyping. Genomic DNA was extracted from diagnostic bone marrow samples and subjected to targeted sequencing of 52 genes using the Ovation Target Enrichment system (NuGEN). These genes were selected based on previous pediatric AML studies and are involved in key pathways such as signaling pathyway mutations, transcription factors, epigenetics and cell cycle control. Signaling pathway mutations are more frequently observed in pediatric AML, whereas epigenetics–related mutations are more common in adults, forming the rationale for gene selection to explore age-related mutational differences. Raw sequencing data were processed using FastQC and Trimmomatic, aligned to the hg19 reference genome with BWA, and refined with GATK. Variants were called using VarScan and filtered for quality and population rarity. Mutational profiles of pediatric patients were compared with adult AML patients from the NTUH cohort and validated using the publicly available TCGA adult AML dataset. Additional age-stratified analysis was performed to investigate how mutational patterns shift across age groups. The NTUH pediatric data were also compared with the publicly available TARGET pediatric AML dataset to assess ethnic differences. Results and Discussion The mutation analysis of our pediatric AML cohort revealed significant molecular heterogeneity, with an average of four mutations being detected per patient. The most frequently mutated genes were ZBTB7A (58%), KMT2C (53.4%), PML (44.3%), and PTPN14 (36.4%). Comparative analysis with NTUH adult AML patients revealed significant age-related differences. WT1 mutations were significantly more common in pediatric patients (16.5%) than in adults (9.5%, FDR q = 0.023), whereas DNMT3A (19.7%, FDR q = 0.001), NPM1 (23.4%, FDR q = 2.4×10⁻⁶), IDH1/2, and TET2 mutations were more frequently observed in adults. These findings suggest age-dependent differences in leukemogenic mechanisms. Age-stratified analysis further demonstrated a gradual shift in the mutational landscape across age groups—from signaling pathway-related mutations in pediatric AML to epigenetic and transcriptional deregulation in adult AML—highlighting the possibility of distinct biological processes driving AML at different life stages. Comparison with the TARGET pediatric AML dataset showed that ZBTB7A mutations were detected in 49.5% of cases, significantly higher than the 0.7% reported in TARGET (FDR q = 5.32×10⁻²²), and WT1 mutations were also more frequent (14.6% vs. 3.3%, FDR q = 0.014). These results underscore potential ethnic or regional differences in AML genetics and emphasize the importance of considering age and population background when developing risk classification and treatment strategies. Conclusion The mutational landscape of Taiwanese pediatric AML is distinct and diverse, with significant differences from adult AML and international pediatric datasets. These findings support the importance of considering age and ethnicity in AML diagnosis and treatment planning, and support the development of age-specific therapeutic approaches.