早期肺癌手術選擇之最佳化：基於腫瘤特徵的肺葉切除術、肺節切除術與楔狀切除術比較分析

Abstract

背景： 早期非小細胞肺癌（NSCLC），特別是肺腺癌的手術治療方式正在演進。傳統的肺葉切除術雖為標準，但亞肺葉切除術（SLR，包含肺節切除術與楔狀切除術）因早期篩檢的普及與保留肺功能的需求而日益受到重視。然而，最佳的手術範圍，特別是在不同亞肺葉切除技術的選擇以及處理具有臟層肋膜侵犯（VPI）等不良特徵的腫瘤方面，仍存在爭議。本論文旨在基於腫瘤特徵比較肺葉切除術、肺節切除術與楔狀切除術，以優化早期肺腺癌的手術選擇。 方法： 本研究包含四個系列研究，運用國立臺灣大學醫學院附設醫院之機構資料庫以及臺灣癌症登記中心之全國性數據（2011-2018年）。比較分析針對臨床第一期或病理T2a期（腫瘤≤3公分合併VPI）N0M0之非小細胞肺癌患者。研究中一致採用傾向評分匹配（PSM）以平衡觀察性研究中各手術組（SLR vs. 肺葉切除；肺節 vs. 楔狀切除）的基線共變量，比較其總體存活期（OS）、無病存活期（DFS）及肺癌特異性存活期（LCSS）。分析根據關鍵腫瘤特徵（如大小、實質比例（C/T ratio）、VPI狀態）進行分層。使用Cox比例風險模型確定獨立預後因子。 結果： 經PSM校正後，對於cT1N0肺腺癌，SLR的OS與DFS與肺葉切除術相當。比較兩種SLR術式於IA期肺腺癌，全國性數據分析顯示肺節切除術的OS優於楔狀切除術，此差異主要由>2公分的腫瘤驅動；對於≤2公分的腫瘤，兩者預後相當。單機構數據分析進一步指出，僅在腫瘤>2公分且C/T ratio >50%的亞組中，肺節切除術的DFS顯著優於楔狀切除術。針對pT2a(≤3cm+VPI)N0M0的特定亞組，SLR的LCSS與肺葉切除術在嚴謹匹配校正後相當。腫瘤大小、實質成分/C/T比、淋巴血管侵犯（LVI）及VPI程度（PL2 vs PL1）是重要的獨立預後因子，而手術範圍本身在校正這些因子後的影響相對較小。 結論： 早期肺腺癌手術的最佳化需要基於腫瘤特徵的個人化策略。SLR是經選擇的cT1N0病例的可行選項。肺節切除術主要適用於>2公分或具較高侵襲性影像特徵的腫瘤；楔狀切除術則適用於≤2公分的低風險腫瘤。即使存在VPI（於≤3公分N0腫瘤），SLR亦可能提供與肺葉切除術相當的癌症特異性存活。本研究結果支持根據腫瘤大小、影像學特徵及VPI、LVI等病理風險因子綜合評估，以制定最適切的手術決策。

Background: The surgical management of early-stage non-small cell lung cancer (NSCLC), particularly adenocarcinoma, is evolving. While lobectomy remains a standard, sublobar resections (SLR), including segmentectomy and wedge resection, are increasingly utilized, driven by enhanced early detection and the desire for parenchyma-sparing surgery. However, the optimal surgical extent, especially concerning the choice between different SLR techniques and the management of tumors with adverse features like visceral pleural invasion (VPI), remains debated. This thesis aimed to optimize surgical selection by comparing lobectomy, segmentectomy, and wedge resection based on tumor characteristics in early-stage lung adenocarcinoma. Methods: This research comprised a series of four studies utilizing institutional data from National Taiwan University Hospital and nationwide population-based data from the Taiwan Cancer Registry Database (2011-2018). Comparative analyses focused on patients with clinical stage I or pathological T2a (≤3cm with VPI) N0M0 lung adenocarcinoma. Propensity score matching (PSM) was consistently employed to balance baseline covariates in observational comparisons of overall survival (OS), disease-free survival (DFS), and lung cancer-specific survival (LCSS) between surgical groups (SLR vs. Lobectomy; Segmentectomy vs. Wedge Resection). Analyses were stratified by key tumor characteristics, including size, consolidation-to-tumor (C/T) ratio, and VPI status. Cox proportional hazards models identified independent prognostic factors. Results: After PSM, SLR demonstrated comparable OS and DFS to lobectomy for cT1N0 lung adenocarcinoma. Comparing SLR types for stage IA adenocarcinoma, segmentectomy showed superior OS compared to wedge resection in the population-based analysis, primarily driven by tumors >2cm; outcomes were comparable for tumors ≤2cm. Institutional data analysis further refined this, showing segmentectomy yielded better DFS only when both tumor size >2cm and C/T ratio >50% were present. For the specific subgroup of pT2a(≤3cm+VPI)N0M0 NSCLC, SLR provided comparable LCSS to lobectomy after rigorous PSM adjustment. Across analyses, tumor size, solid component/C/T ratio, LVI, and degree of VPI (PL2 vs PL1) emerged as significant prognostic factors, while surgical extent often was not an independent predictor after accounting for these characteristics in matched N0 cohorts. Conclusion: Optimizing surgical selection for early-stage lung adenocarcinoma requires a personalized approach integrating tumor characteristics. SLR is oncologically comparable to lobectomy for selected cT1N0 cases. Segmentectomy appears superior to wedge resection primarily for tumors >2cm or those exhibiting higher radiological invasiveness. Even in the presence of VPI (in ≤3cm N0 tumors), SLR may offer comparable cancer-specific survival to lobectomy. These findings support tailoring surgical extent based on a comprehensive assessment of tumor size, radiological features, and pathological risk factors like VPI and LVI.