以Cox模型來探討膀胱癌的預後因子

Abstract

背景：膀胱癌(bladder cancer)為泌尿道腫瘤中最常見的癌症，膀胱癌的死亡率雖不是很高，發生率卻頗高，尤其台灣西南沿海烏腳病流行地區的北門、學甲、布袋、義竹最高，發生率甚至是台灣其他地區的2-3倍。因此膀胱癌的診斷及治療已是現代醫學及公共衛生上不容小覷的問題。 雖然大部分的膀胱癌病患有不錯的預後，但某些病患確有無法解釋的高致死率與腫瘤復發及轉移。目前膀胱癌的治療最大的挑戰就是膀胱癌本身沒有一個很好的腫瘤標記來預測其存活跟復發及轉移的時間。目前有少數研究顯示，利用分子生物學與基因分析的方式，能找出預測膀胱癌病患的存活與復發的指標。但是這些分析方式僅停留於實驗室階段，且較為昂貴、耗時，不容易於臨床上應用。因此在臨床治療方面，我們急切需要較簡便且能準確的預測膀胱癌病患，經手術切除後，其存活、復發及轉移時間的指標，以期能降低這類癌症對國人的威脅。 目標：本研究希望可以透過簡單且易取得的臨床參數，找出膀胱癌病患經手術切除後，影響其無病存活與腫瘤相關存活時間的相關因子。 方法：自1998年至2005年，收集來自於奇美醫院，經診斷為膀胱尿路上皮癌，並接受根治性手術的病患，共269位。所收集的基本資料包括病人的年齡、性別、是否居住住烏腳病(BFD)的流行地區。 臨床病理參數的部分包括檢體邊緣是否有癌細胞侵犯、腫瘤是否有不正常的分化(Differentiation)、腫瘤是否有壞死(Necrosis) 、腫瘤的分期(T,N)、組織學分級(WHO)、腫瘤乳突化的程度(papillary) 、腫瘤侵犯的方式(I, T, N)、是否有淋巴或血管侵犯(LVI)、神經侵犯(NI) 及高倍數下有絲分裂的個數(MF10)。利用所收集到的基本資料及臨床參數，將病患分組，來比較各組間病患無病的存活機率及腫瘤相關的存活機率是否不同。 統計方法以Kapaln-Meier estimate估計不同組別的存活函數，再以Log-rank test或Wilcoxon test來檢定單一因子對膀胱癌患者經手術切除之預後的影響。之後再以Cox 比例風險模型來進行多變量分析，透過適當的模式選擇，找出對膀胱癌病患無病存活與腫瘤相關存活時間有影響的因子。 結果：以無病的存活時間來看，檢體邊緣有癌細胞侵犯(p<0.0001)、較高的組織學分級(p=0.0422)、非結節狀的侵犯方式(p=0.0006)及較高的高倍數下有絲分裂個數(p=0.0138)，有比較低的無病存活機率，並達到統計學上的意義。透過Cox比例風險模型，發現檢體邊緣是否有癌細胞侵犯 (p<0.0001)及組織學分級(p=0.0692)是無病存活時間的獨立預後因子，並達到顯著的意義。 以腫瘤相關的存活時間來看，較高的T分期(p<0.0001)、有淋巴結侵犯(p=0.0124)、較高的組織學分期(p=0.0036)、較低的腫瘤乳突化程度(<90%) (p<0.0001)、非結節狀的侵犯方式(p=0.0025)、有淋巴或血管的侵犯(p=0.0211)、有神經的侵犯(p=0.0166)及較高的高倍數下有絲分裂個數(≧10)(p=0.0006)，有比較低的腫瘤相關存活機率，並達到統計學上的意義。在Cox 比例風險模型，發現僅有T分期(p<0.0001)、高倍數下有絲分裂的個數(p=0.0231)及年齡(p=0.0216)是腫瘤相關存活時間的獨立預後因子，並達到顯著的意義。 結論：透過存活分析的方式，我們可以找出膀胱尿路上皮癌的病患，經手術切除 腫瘤後，影響無病存活時間及腫瘤相關存活時間的因子，以提供臨床醫師 作為治療的參考。

Background Bladder cancer (BC) is one of the most common cancer worldwide. Its incidence is even higher at so-called “Black-Foot Disease(BFD)”endemic areas in southern Taiwan, including Pei-men, Hsieh-chia, Pu-tai and Ichu that lead to a remarkable public health problem. Although the majority of bladder cancer behaves in indolent course, however, a subset of patients suffers from unexpected lethal recurrence and metastasis. Currently, the challenge in the management of BC is the lack of ideal prognostic model. Accordingly, it is highly desirable to develop a prognostic system to aid the determinate of treatment strategies for BC with especial emphasize on those at BFD endemic areas. Objectives To find out easily assessed clinical markers to predict the prognosis of BC in terms of disease-free and disease-specific survival after surgical resection. Methods We retrospectively reviewed the medical records of 269 patients with bladder cancer after surgical resection in a medical center in Southern Taiwan, and who had transurethral resection of bladder tumor (182) or radical cystectomy (87) between 1998 and 2005. Data collected including demographic features (age, sex, and whether living in the BFD endemic area) and clinical characteristics (pattern of invasion, tumor differentiation, tumor necrosis, T-stage, N-stage, WHO grade, lymphovascular invasion, perineural invasion, mitotic figures etc.) We compared the disease-free survival and cancer-specific survival of patients with the demographic and clinical parameters mentioned above Kaplan-Meier method was used to estimate the survival by different clinical features, of which the statistical significance was tested by using Log-rank test or Wilcoxon test. Multivariate Cox-regression was further applied to identify the significant covariates after adjusting by other covariates with a series of model selections. Results With respect to disease-free survival, positive surgical margin (p<0.0001), higher histological grade (p=0.0422), adverse invasion pattern (p=0.0006), as well as higher mitotic activity (p<0.0138) significantly affected disease-free survival. However, only positive margin (p<0.0001) and histological grade (p=0.0692) remained prognostically independent in the multivariate Cox regression model. Furthermore, numerous factors including increment of pT status (p<0.0001), the presence of nodal metastasis (p=0.0124), higher histological grade (p=0.0036), less papillary component (<90%)(p<0.0001), adverse invasion pattern (p=0.0025), the presence of lymphovascular invasion (p=0.0211), perineurial invasion (p=0.0166), and higher mitotic activity (p=0.0006) significantly predicted inferior cancer-specific survival In the multivariate analysis, the increment of pT status (p<0.0001), along with higher mitotic activity (p=0.0231), and old patient age (p=0.0216) significantly predicted inferior outcome. Conclusions Throughout this analysis, we have identified numerous significant prognosticators which identified patients at higher risk of disease relapse after surgical treatment. These can provide further information to adjust clinical management.