台灣卵巢癌全外顯子分析與存活率的關係

Abstract

背景與動機 卵巢癌的發生和死亡存在種族差異與亞型差異。本研究旨在探索台灣女性卵巢癌患者腫瘤的體細胞單核甘酸變異 (single nucleotide variants，SNVs)，是否可以作為患者預後的預測子，並探討其是否具有種族特異性或亞型特異性。

方法: 本研究以全外顯子定序技術分析腫瘤體細胞突變。我們以Cox風險比例模型和加速失效時間模型分析SNV和死亡風險及復發風險的關係，並透過邏輯斯迴歸模型了解SNV和患者對鉑類藥物反應之間的關係。同時計算次要等位基因，以及以卡方檢定或費雪精準檢定以確定SNV在特定亞型中的顯著性。

結果: 我們識別出17個SNV，可以作為全亞型的死亡風險預測子(如rs17120062、rs3746435和rs75323205)、透明細胞 (Clear cell carcinoma，CCC)亞型（如rs3746435、rs17120062等）、子宮內膜樣 (Endometrioid carcinoma，EC)亞型（如rs36097019、rs6788448等）和漿液 (Serous carcinoma，SC) 亞型（如rs3810481、rs2108622等）的死亡風險預測子。存活分析的log rank test皆顯示攜帶變異者和未攜帶者在存活曲線存在顯著差異。此外，我們鑑定了20個SNV，可以作為復發風險預測子 (如rs57115249、rs7047726和rs2279218等)，這些SNV在存活分析中也展現了顯著的差異。我們也識別出18個SNV，可以作為對藥物敏感的預測子(如rs529208、rs56727079和rs57115249等)。

在比較公開資料庫中的等位基因頻率時，我們發現這些SNV在東亞人群中普遍高於白人，但在本研究樣本中的頻率卻相對較低，暗示台灣卵巢癌患者中可能存在特異性的基因變異，這些變異可能是導致卵巢癌發生率和死亡率種族差異的潛在因素。本研究還確定了5個用於區分不同亞型的SNV。

結論: 本研究成功鑑定了多個與卵巢癌預後密切相關，而且可用於區分不同亞型的SNV。這些SNV的等位基因頻率揭示了台灣卵巢癌的種族特異性變異。上述這些SNV坐落的13個基因 (ALPK2、LOXL4、SLC28A1、TNFRSF10D、ATP13A4、IDO2、IGHV1-2、PLXNB2、CYP2D6、EGFLAM 、C19orf53、CYP4F2和UBAP2) 已有少數與卵巢癌相關的研究。在與預後有顯著相關的變異中，有六個變異在不同的分析模型中都有重複地出現在CCC亞型以及全亞型的組別中，分別為rs3746435、rs17120062、rs11547731、rs1065852、rs1983864和rs584855，對應的基因名稱為MYH7B、TEKT4、PLXNB2、CYP2D6、LOXL4和ANKRD60。這些基因目前有少數與卵巢癌或其他癌症相關的研究，這些基因可能可以為在台灣較為常見的CCC亞型卵巢癌患者提供重要的治療和診斷洞見。

這些發現顯示了本探索性分析的重要性。這些鑑定出的SNV為後續深入的功能性研究奠定了基礎，它們具有成為癌症診斷和治療的關鍵生物標記的潛力，並有機會直接作為癌症治療的新型藥物靶點。此外，這些SNV的發現為臨床決策提供了新的參考依據，可以幫助優化治療策略和提高預後預測的準確性。

Background and Motivation: Ovarian cancer exhibits ethnic and subtype disparities in incidence and mortality. This study aimed to explore whether somatic single nucleotide variants (SNVs) in Taiwanese female ovarian cancer patients could serve as prognostic predictors and to investigate their ethnic specificity or subtype specificity.

Methods: The study analyzed tumor somatic mutations using whole-exome sequencing. We employed Cox proportional hazards and accelerated failure time models to analyze the relationship between SNVs and risks of death and recurrence. Logistic regression was used to understand the relationship between SNVs and patient response to platinum-based drugs. Minor allele frequencies were calculated, and chi-square or Fisher''s exact tests were used to determine the significance of SNVs in specific subtypes.

Results: We identified 17 SNVs as predictors of mortality risk in all subtypes (e.g., rs17120062, rs3746435, rs75323205), in clear cell carcinoma (CCC) subtype (e.g., rs3746435, rs17120062), in endometrioid carcinoma (EC) subtype (e.g., rs36097019, rs6788448), and in serous carcinoma (SC) subtype (e.g., rs3810481, rs2108622). Log-rank tests showed significant differences in survival curves between carriers and non-carriers of these variants. Furthermore, 20 SNVs were identified as predictors of recurrence risk (e.g., rs57115249, rs7047726, rs2279218), showing significant differences in survival analysis. Additionally, 18 SNVs were identified as predictors of drug sensitivity (e.g., rs529208, rs56727079, rs57115249). When comparing allele frequencies in public databases, these SNVs were generally higher in East Asians than in Caucasians but were relatively lower in our study sample, suggesting the presence of specific genetic variations in Taiwanese ovarian cancer patients that may contribute to ethnic differences in incidence and mortality rates. The study also identified 5 SNVs for differentiating subtypes.

Conclusion: This study successfully identified multiple SNVs related to ovarian cancer prognosis and useful for differentiating subtypes. The allele frequencies of these SNVs reveal ethnic-specific variations in Taiwanese ovarian cancer. The 13 genes hosting these SNVs (ALPK2, LOXL4, SLC28A1, TNFRSF10D, ATP13A4, IDO2, IGHV1-2, PLXNB2, CYP2D6, EGFLAM, C19orf53, CYP4F2, and UBAP2) have limited studies related to ovarian cancer. Among the prognostically significant variants, six (rs3746435, rs17120062, rs11547731, rs1065852, rs1983864 and rs584855 corresponding to MYH7B, TEKT4, PLXNB2, CYP2D6、LOXL4 and ANKRD60 , respectively) were consistently observed in different analytical models in CCC and all subtype groups. These genes, linked to ovarian cancer or other cancers in a few studies, may provide critical therapeutic and diagnostic insights for CCC subtype, prevalent in Taiwan.

These findings highlight the importance of this exploratory analysis. The identified SNVs lay the groundwork for subsequent functional studies, holding potential as key biomarkers for cancer diagnosis and treatment, and as novel drug targets. Moreover, these discoveries provide new references for clinical decision-making, helping optimize treatment strategies and enhance the accuracy of prognosis prediction.