基因標記的演變在骨髓化生不良症候群之臨床應用與預後意義

Abstract

簡介 骨髓化生不良症候群是一群相當異質性的骨髓疾病。其特色是骨髓功能異常而導致無效的造血，同時有轉變成急性血癌的風險。骨髓化生不良症候群的分類是根據1982年英美法合作小組準則與世界衛生組織於2001、2008與2016年修訂而定。目前已知包括年紀，骨髓芽細胞的多寡，血紅素，白血球，血小板數目以及染色體的變化等是重要的預後因子，因而建構成了國際預後積分系統與國際預後積分系統修正版這兩種評分系統。但是，臨床上仍有一些病人儘管具有類似的國際預後積分系統或是國際預後積分系統修正版的評分，他們的預後卻迥然不同，這是重要的臨床問題。 近年來，關於骨髓化生不良症候群的基因變異的陸續被發現，其中一些帶有臨床預後的意義。台大醫院的血癌研究團隊先前研究發現，將單套染色體變化併入國際預後積分系統修正版，可以更好的評估骨髓化生不良症候群病人的存活率。除此之外，許多在診斷時用Sanger定序方式檢測出來的基因變化已發現與臨床特徵相關。此外有些基因變異在疾病變化過程中相當穩定，有些則否。然而，使用次世代定序分析來完整的檢驗基因突變在疾病狀況改變時的角色及其相關文獻仍然相當少。 目的 這個研究的目的是用次世代基因序列分析，來研究骨髓化生不良症候群的分子基因突變標記在初診斷與疾病惡化時的演變。我們希望更進一步研究基因突變的變化型態所帶有之預後意義。 方法 我們收集了臺大醫院150例新診斷，且保留有一系列骨髓檢體的原發性骨髓化生不良症候群病患，這些病患皆保存有完整的臨床資料。其中，70位病人出現疾病的進展，有23位病人惡化為高風險的骨髓化生不良症候群的惡化，有47位病患惡化為急性白血病。我們從骨髓的單核細胞抽取脱氧核醣核酸，並以次世代序列分析54個目標基因的變異狀況。我們比較初診斷與疾病惡化時基因突變型態的不同，以及分析其與臨床表徵，治療反應，以及預後之間的關係。 結果 有疾病惡化的病患與維持穩定的病患相比，不論是在診斷時的檢驗數據或是染色體風險，都沒有明顯差異。在有疾病惡化的病患之中，初診斷時，最常見的基因突變是ASXL1 基因 (初診斷與惡化時分別為34.3% 與35.7%)、RUNX1基因 (22.9% 與27.1%)、SRSF2基因 (18.6% 與18.6%)、U2AF1基因 (18.6% 與20.0%)、STAG2基因 (15.7%與17.1%)、與DNMT3A基因(15.7% 與12.9%)突變。在疾病惡化時其基因突變的分布基本上與初診斷時類似。若分析基因標記在疾病惡化時的新增加的演變(genetic evolution，包含染色體核型變化的演變或基因突變的轉換)，最常見的是在疾病惡化時出現第八對染色體相關的核型變化 (16.1%)，接著是NRAS基因 (15.8%) 與RUNX1基因 (14.0%) 的突變。在70位有疾病變化的病患中，有48 (68.6%)位病人有基因標記的演變: 15 (21.4%) 位病人只有染色體核型變化的演變，18 (25.7%) 位病人只有基因突變的轉換，還有15 (21.4%)位病患則是同時擁有染色體的核型演變與基因突變的轉換。當考慮到基因突變的數量與疾病嚴重程度的關係，我們發現，在初診斷時其兩者關係並不顯著，但在疾病惡化時，轉變為急性骨髓性白血病的病患其擁有的基因突變數量，遠比轉變為較高風險骨髓化生不良症候群要顯著的多。另外，無論原本疾病嚴重程度為何，惡化為急性骨髓性白血病的病患，其預後最差，其次是惡化成高風險骨髓化生不良症候群的病患。另外，染色體核型變化的演變也是一個重要的預後因子。有趣的是，有cohesin突變的病患，若在疾病惡化時方有此突變，預後比初診斷時帶有此突變要好。至於血液幹細胞移植對有分子基因標記演變的病患並無法明顯改善其預後。 我們使用SciClone來分析基因突變的型態，可以根據分子標記演變的模式，將基因突變分成兩群 :第一群基因突變 [DNMT3A、NRAS、RUNX1、WT1、IDH2、splicing因子(包含有SFSR2、 SF3B1與U2AF1) 與TET2突變] 其突變頻率傾向在疾病惡化時增加，或是在病人惡化時會額外得到這群突變。而第二群基因突變 (ASXL1、SETBP1、STAG2、BCOR與 TP53 突變) 的突變頻率則是在疾病惡化時傾向維持或是減少，或是病人會在惡化時失去這些突變。我們發現，第一群基因突變在次發性急性骨髓性白血病比較常見，相反的，第二群基因突變則是在高風險的骨髓化生不良症候群中較多。在第二群基因突變中，有兩個基因(TP53與SETBP1)分別與非常不好與不好風險的染色體核型變化有相關。我們可以根據病人初診斷時這兩群基因突變的分布，將病人分為不同的三組:第一組病人有第一群+/-第二群的基因突變，第二組病人只有第二群的基因突變，第三組病人則沒有這兩群的基因突變。第一組病人有高的比例有好的風險的染色體核型變化，相反的，第二組病人有比較多的非常不好風險的染色體核型變化。在整體存活率部分，第三組病人的預後最好，接著是第一組病人，第二組病人最差。 結論 總結來說，超過一半的骨髓化生不良症候群的病患在疾病惡化時會有分子基因標記的演變，而有此演變具有獨特的臨床表徵且將對預後會有影響，並且會消弭血液幹細胞移植可能帶來的好處。發展新的治療方式，對於具有分子標記演變的病患是重要的臨床議題。

Introduction Myelodysplastic syndrome (MDS) is a heterogeneous group of clonal bone marrow (BM) disorders. It is characterized by ineffective hematopoiesis leading to BM failure and a high risk of progression to acute leukemia. The classification of MDS is according to either the French-American-British Cooperative Group Criteria developed in 1982 or the World Health Organization criteria in 2001, 2008, and 2016. Several prognostic markers, including the age, BM blast percentage, hemoglobin level, white blood cell and platelet counts, and cytogenetics, have been proposed. The International prognostic scoring (IPSS) and revised international prognostic scoring system (IPSS-R) both have great predictive values for probability of leukemia transformation and overall survival. However, there are still some patients with similar IPSS or IPSS-R scores but with discrete prognosis. Recently, many gene mutations have been detected in MDS patients and some of them have prognostic implications. The myeloid research team at the National Taiwan University Hospital (NTUH) has found that integration of monosomal karyotype to IPSS-R can better risk-stratify MDS patients into different groups. Besides, several gene mutations detected at diagnosis by Sanger sequencing have been found to correlate well with clinical and biological features and be either stable or unstable during clinical evolution. However comprehensive mutation analysis of a large-cohort of MDS patients by next generation sequencing in serial studies are still lacking. Specific aims The aims of this study were to investigate the dynamic clonal architecture by mutation analyses of paired samples obtained at diagnosis and subsequent disease progression during follow-ups using targeted next generation sequencing. Furthermore, we speculated if the evolution profile has any prognostic impact. Methods Totally, 150 de novo MDS patients diagnosed according to the 2016 WHO classification at the NTUH were recruited in this study. Among them, 70 patients had disease progression: 23 patients progressed to high risk MDS and 47 progressed to AML. The DNA extracted from BM mononuclear cells was amplified by polymerase chain reaction and then analyzed by targeted next generation sequencing for gene mutations. The mutation patterns obtained at diagnosis and disease progression from the same patient were compared. The findings were correlated with clinical features, treatment response, and clinical outcome. Result For clinical characteristics, such as laboratory data at diagnosis or cytogenetic risk stratification base on IPSS, patients with or without disease progression were comparable. The most prevalent gene mutations at both diagnosis and progression were similar, including ASXL1 (34.3% and 35.7%, at diagnosis and at progression, respectively), RUNX1 (22.9% and 27.1%), SRSF2 (18.6% and 18.6%), U2AF1 (18.6% and 20.0%), STAG2 (15.7% and 17.1%), and DNMT3A mutations (15.7% and 12.9%). The most common acquisition of genetic evolution was the one involving chromosome 8 (16.1%), followed by NRAS (15.8%), and RUNX1 mutations (14.0%). Among the 70 patients with disease progression, 48 (68.6%) patients had genetic evolution: 15 (21.4%) patients had only cytogenetic evolution, 18 (25.7%) had only mutation shift, and 15 (21.4%) had both cytogenetic evolution and mutation shift. The patients with disease progression had a higher frequency of mutation shift (either acquisition or loss) than those with stable disease (47.1% vs. 30.0%, P=0.043). The gene mutation numbers had no association with the disease status at diagnosis, but at progression, patients with acute myeloid leukemia (AML) had more gene mutation numbers than those with high risk MDS. Progression to AML rather than to high risk MDS, and presence of cytogenetic evolution or genetic evolution predicted poorer overall survival (OS) after progression, irrespective of the original disease status. Intriguingly, patients with cohesin mutations acquired at progression had a significant better OS than those with the mutations at diagnosis. The patients with genetic evolution did not benefit from hematopoietic stem cell transplantation. We used SciClone to analyze the clonal architecture of gene mutations at diagnosis and at disease progression, and identified two clusters of gene mutations. Cluster 1 gene mutations [DNMT3A, NRAS, RUNX1, WT1, IDH2, splicing factors (including SRSF2, SF3B1, and U2AF1), and TET2 mutations] tended to have increasing VAF or be acquired at progression. On the other hand, cluster 2 gene mutations (ASXL1, SETBP1, STAG2, BCOR, and TP53 mutations) tended to be stationary, decreasing, or lost at disease progression. At diagnosis, the cluster 1 gene mutations were significantly more prevalent in secondary AML but the cluster 2 gene mutations were more likely to present in high risk MDS status. Two cluster 2 gene mutations, TP53 and SETBP1, were associated with very poor and poor-risk cytogenetics. We risk-stratified patients into three groups at diagnosis: the group 1 patients with cluster 1 ± cluster 2 gene mutations, the group 2 patients with only cluster 2 gene mutations, and the group 3 patients with neither cluster 1 nor cluster 2 gene mutations. The group 1 patients had significantly more good-risk cytogenetics; on the contrary, the group 2 patients harbored significantly more very poor-risk cytogenetics. For OS, the group 3 patients had the longest OS, followed by the group 1 and group 2 patients. Conclusion In conclusion, more than half of MDS patients experienced genetic evolution during disease progression and the presence of the genetic evolution had poor prognostic impact, not remediated by HSCT. It is necessary to develop new treatment strategy, such as target or immunotherapy, for patients with genetic evolution.