腫瘤免疫微環境在食道鱗狀細胞癌預後的意義

Abstract

食道鱗狀細胞癌是亞洲的一個主要癌症，佔比全球食道癌超過85%。大部分的病人在診斷時已是局部晚期，而前輔助同步化放療加上隨後的根除性食道切除術是目前局部晚期食道鱗狀細胞癌的標準治療。然而，仍有大約70%的病人會遭受癌症復發。因此，我們需要一個有效的預後生物標記，才能將病患依復發風險加以分層、甚至施以不同的治療，希望能進一步改善這群病人的預後。 人體對抗腫瘤的免疫力可以降低癌細胞的侵略性與生長，這也是包括化學治療與放射治療等許多抗癌治療策略，能達到治療腫瘤療效的重要關鍵之一。因此，免疫相關的生物標記是許多癌症重要的研究主題；針對食道鱗狀細胞癌，過去也有許多相關的研究。PD-L1是PD-1 (為一個免疫檢查點)的配體，可以造成T淋巴細胞的衰竭進而抑制免疫。PD-L1在腫瘤中的表現不但代表對腫瘤免疫力的拮抗、也代表著免疫力的存在。過去關於PD-L1的表現在食道鱗狀細胞癌預後意義雖有許多研究，其結果並不十分一致。另外，第三級淋巴結構是在慢性發炎的組織(包括腫瘤，也可視為一種慢性發炎組織)發生、由免疫細胞組成的一種淋巴組織。一般認為，這種淋巴組織在腫瘤中具有對抗腫瘤的免疫力。目前關於第三級淋巴結構在食道鱗狀細胞癌預後的研究仍相當有限。本論文研究針對接受前輔助化放療的局部晚期食道鱗狀細胞癌病患，聚焦其治療前的食道癌腫瘤免疫微環境，首先探討腫瘤細胞PD-L1、或免疫細胞PD-L1的表現，接著探討第三級淋巴結構的形成與成熟，對局部晚期食道鱗狀細胞癌病患的預後意義。 本論文研究的第一個部分探討PD-L1 的預後意義。這一部分研究收納了100位接受以每週太平洋紫杉醇和鉑金化療藥物為處方前輔助同步化放療的局部晚期食道鱗狀細胞癌病患。以治療前上消化道內視鏡取得的食道鱗狀細胞癌組織，進行PD-L1 的免疫組織化學染色，以半定量的方式分析PD-L1表現量為0、1+、2+或3+：在免疫細胞上的表現量0或1+定義為低表現、表現量2+或3+則為高表現；在腫瘤細胞上的表現量0定義為無表現、表現量1+、2+或3+則為有表現。我們發現免疫細胞PD-L1表現的高低與病患的無惡化存活以及整體存活都有強烈的正相關，而腫瘤細胞PD-L1表現的有無則與無惡化存活以及整體存活有顯著的負相關。多變項分析也確認了上述的發現。 本論文研究的第二部分，根據第一部分的研究結果提出了一個假說：腫瘤細胞表現的PD-L1與對抗癌藥物的抗藥性有相關性。我們測試五株食道鱗狀細胞癌細胞株對鉑金化療藥物的藥物敏感性與這些細胞株上PD-L1表現量的相關性。我們發現KYSE150是一株帶有PD-L1基因重複（重複數：7）的細胞株，其PD-L1表現量是五株細胞株最高的；KYSE150也五株細胞株中是對鉑金化療藥物抗藥性最高的細胞株。KYSE510是PD-L1表現量最低的，同時也是對鉑金化療藥物最敏感的細胞株。這五株細胞株PD-L1的表現量與其在一系列白金濃度下存活的比例，具有中等度的線性相關性。這個結果支持了PD-L1的表現與化療抗藥性相關的假說。然而，詳細的機轉還需要更進一步的研究。 本論文研究的第三部分探索局部晚期食道鱗狀細胞癌黏膜層的第三級淋巴結構的預後意義。這一部分研究收納了一組137位接受以每週太平洋紫杉醇和鉑金化療藥物為處方前輔助同步化放療的局部晚期食道鱗狀細胞癌病患。以治療前上消化道內視鏡取得的食道鱗狀細胞癌組織，進行CD20以及CD23的免疫組織化學染色，分析第三級淋巴結構的成熟狀態。我們發現成熟的第三級淋巴結構和對前輔助化放療的完全病理反應成顯著的負相關（p=0.031）。雖然在單變項分析中，成熟的第三級淋巴結構和不良的整體存活間只存在不顯著的相關趨勢，但在多變項分析中，成熟的第三級淋巴結構成為了獨立而顯著的不良預後因子（HR: 2.91, p<0.001）。這個結果與我們進行本研究之前的假說並不吻合。 本論文研究的第四個部分，我們嘗試驗證以內視鏡切片檢體判讀第三級淋巴結構狀態的可信度。我們利用nanoString平台的Human Pan-Cancer Immune Panel來分析44位局部晚期食道鱗狀細胞癌病患的腫瘤檢體，探討第三級淋巴結構中，無、不成熟與成熟三種狀態下，腫瘤免疫微環境的差異。我們發現第三級淋巴結構的形成與成熟相關於較活躍的免疫環境，包括較少的衰竭CD8 T細胞和較高的白血球、B細胞、T細胞、NK細胞、細胞激素、化學激素、腫瘤壞死因子、toll-like受器以及抗原呈現等免疫功能。我們接著利用對比成熟第三級淋巴結構與無第三級淋巴結構的腫瘤組織中免疫基因表現的差異，發展出第三級淋巴結構的基因標記（gene signature）。我們首先針對TCGA資料庫中食道鱗狀細胞癌的族群（82人），證實這個基因標記可以有效地分別出腫瘤組織中淋巴濾泡組織的有無。我們也針對一個接受免疫檢查站抑制劑治療的轉移復發食道鱗狀細胞癌族群（35人），證實這個基因標記可以有效地預測免疫檢查站抑制劑治療的療效。這些發現都支持我們使用內視鏡切片的腫瘤檢體所進行的分析方式，可以有效地區分食道癌腫瘤粘膜第三級淋巴結構的成熟狀態。 總結來說，PD-L1在免疫細胞或者腫瘤細胞的表現、粘膜中成熟的第三級淋巴結構是接受前輔助同步化放療局部晚期的食道鱗狀細胞癌獨立的預後生物標記。PD-L1在腫瘤細胞的表現和化療的抗藥性有正相關，這可能是PD-L1在腫瘤細胞的表現導致不良預後的機轉之一。然而，同步放化療是否對第三級淋巴結構造成傷害而導致不良的預後，則需要進一步的研究。

Esophageal squamous cell carcinoma (ESCC) is a major cancer in Asia, and accounts for more than 85% of global burden of patients diagnosed with esophageal cancer. Most ESCC patients had locally advanced cancer at diagnosis. Neoadjuvant chemoradiotherapy (CRT) followed by radical esophagectomy is a current standard treatment for locally advanced ESCC. However, about 70% of patients suffered from recurrence. Efficient prognostic biomarkers are thus important for helping stratify patients with different risks of recurrence and for developing novel treatment strategies to improve their outcome. Anti-tumor immunity hinders the aggressiveness and growth of cancer and may contribute to the efficacy of various anti-cancer therapy, including chemotherapy and radiotherapy. Immune-related biomarkers have been investigated for their prognostic significance in patients with locally advanced ESCC before. PD-L1, a ligand of an immune checkpoint, programmed cell death protein 1, suppresses immunity by causing exhaustion of T cells. The expression of PD-L1 in tumor, an antagonist of anti-tumor immunity, may represent the existence of anti-tumor immunity. Multiple previous studies focusing on the expression of PD-L1 in tumors yielded conflicting results in patients with ESCC. Tertiary lymphoid structure (TLS)—aggregates of immune cells forming lymphoid structures in or around tumor tissues—is considered as an origin of antitumor immunity. The prognostic significance of TLS has been only reported in a limited number of studies involving patients of locally advanced ESCC. The current thesis, focusing on tumor immune microenvironment (TME) of ESCC, has explored the prognostic significance of PD-L1 expression on immune cells (ICs) or tumor cells (TCs) and the prognostic significance of mucosal TLS in locally advanced ESCC patients treated with neoadjuvant CRT with weekly paclitaxel/platinum chemotherapy. For the first part of the thesis, the prognostic significance of PD-L1 expression was investigated. A total of 100 locally advanced ESCC patients treated with neoadjuvant CRT with weekly paclitaxel/cisplatin chemotherapy were enrolled. Their pre-treatment ESCC tumor tissues, obtained by endoscopic biopsy, were analyzed for the expression of PD-L1 by immunohistochemistry (IHC) and scored semiquantitatively. PD-L1 expression on ICs (PD-L1 IC) was defined as low (0 or 1+) vs high (2+ or 3+), and PD-L1 expression on TCs (PC-L1 TC) was defined as negative (0) vs positive (1+~ 3+). We found that high PD-L1 IC expression was strongly associated with better progression free survival (PFS) (HR: 0.44, p=0.0025) and overall survival (OS) (HR: 0.44, p=0.0024), and positive PD-L1 TC was significantly associated with worse PFS (HR: 1.7, p=0.029) and OS (HR: 1.63, p=0.035). Multivariate analysis demonstrated both PD-L1 TC and PD-L1 IC as independent prognostic factors. For the second part of the thesis, we hypothesized that PD-L1 expression on TCs may contribute to chemoresistance, thus underlying the finding of positive PD-L1 TC as a poor prognostic factor in patients with locally advanced ESCC treated with neoadjuvant CRT. We tested the chemosensitivity to cisplatin and the expression level of PD-L1 in 5 human ESCC cells lines. We found that KYSE150, which has PD-L1 amplification (copy number=7), had higher expression of PD-L1 compared to other cell lines and was more resistant to cisplatin than other cells. KYSE510 had the lowest baseline expression among the 5 cell lines and was the most chemosensitive cell line. Linear relationships can be identified between PD-L1 expression level and survival percentage of the 5 cell lines under serial concentrations of cisplatin. These findings support our hypothesis about the association of PD-L1 expression on tumor cells with resistance to chemotherapy. However, the detailed mechanisms warrant additional studies. For the third part of the thesis, to investigate the prognostic significance of mucosal TLS in locally advanced ESCC, we recruited a cohort of locally advanced ESCC patients who received neoadjuvant CRT with weekly paclitaxel/platinum chemotherapy (n= 137). Pretreatment endoscope-biopsied primary esophageal tumor tissues were analyzed for the maturation status of TLS according to the IHC-defined expression of CD20 and CD23. The status of mature TLS had significant negative association with pathological complete response to neoadjuvant CRT (p=0.031). Although mature TLS only had a trend of association with worse OS (HR: 1.45, p=0.15) in univariate analysis, mature TLS turned out to be a strong independent unfavorable prognostic factor in the multivariate cox regression analysis (HR: 2.91, p<0.001). The result rejected our initial hypothesis that TLS is a favorable prognostic factor. For the fourth part of the thesis, in order to validate the reliability of TLS status identified by endoscope-biopsied tissues, we explored the TME in tumors with no, immature, or mature TLS status of patients with locally advanced ESCC. We investigated the expression profiles of immune-related genes of ESCC tumor tissues by Pan-Cancer Immune Panel of nanoString® (n=44). We found that the maturation of TLS was associated with more active TME, including less exhausted CD8 T cells and more immune functions of leukocytes, B cells, T cells, NK cells, cytokines, chemokines, tumor necrosis factors, toll-like receptors, and antigen presentation. Based on genes with highly differentiated expression between ESCC tumors with mature TLS vs no TLS, we constructed a TLS signature and demonstrated that this TLS signature can significantly differentiate tumor tissues with from without lymphoid follicles in the ESCC cohort of TCGA database (n=82); the TLS signature can also significantly differentiate ESCC patients benefited from anti-PD-1 based immunotherapy in our cohort of metastatic or recurrent ESCC patients (n=35). These findings support that the TLS status identified by IHC on pre-treatment endoscope-biopsied tumor tissues is accurate. Overall, we found that PD-L1 IC, PD-L1 TC, and mucosal mature TLS were independent prognostic biomarkers for locally advanced ESCC patients who received neoadjuvant CRT. Positive PD-L1 TC is associated with chemoresistance, which could lead to unfavorable prognosis. It needs further investigations whether the damage from CRT changes the TME of TLS and leads to unfavorable prognosis.