探討惡性軟組織肉瘤的新穎免疫分類法

Abstract

軟組織惡性肉瘤與其他癌症最大不同之處在於惡性肉瘤可以源發於任何人體的器官與部位而且有超過一百種以上的病理組織型態。這樣的異質性讓惡性肉瘤的分類、預後的判斷以及治療的選擇面臨諸多挑戰。 目前惡性肉瘤較常分類的方式仍是利用組織型態來做區別。不過在我們與亞洲其他國家合作收集了276位晚期血管惡性肉瘤的研究中發現，不同的原發部位對於預後有顯著的影響 (表淺性的血管惡性肉瘤相對於內臟型的血管惡性肉瘤其對於存活的危險比(hazard ratio) 為0.63, p = 0.004)。此結果反映出的是即使為同一種組織型態中，不同的原發部位仍然會影響臨床的預後。不過，即使我們把惡性肉瘤的研究侷限在一個組織器官內，仍然可能會有不同的預後。在我們與三大洲國際上共五個醫學中心的合作計畫收納了61位源發於心臟的惡性肉瘤研究中，我們發現血管惡性肉瘤有傾向有較差的預後(危險比 1.84, p = 0.07)。以上的研究顯示，我們可能難以單純就組織型態或是原發部位的分類就能掌握惡性肉瘤病患的預後，我們必須從更接近腫瘤生物學相關的角度來切入了解。 癌症免疫治療的成功讓腫瘤免疫微環境 (tumor immune microenvironment)的研究再度獲得了重視。利用轉譯體為基礎的演算法目前已經可以直接將綜合各種細胞種類的RNA (bulk RNA-Seq) 拆解以及量化成各種不同的免疫細胞的比例。因為腫瘤免疫微環境是宿主與腫瘤交會反應的地方，因此我們假設可以利用腫瘤免疫微環境內的免疫細胞分布特性讓惡性肉瘤有更好的分類結果。我們與法國的研究者合作，包含了最開始的試驗設計以及利用該團隊所設計的運算法來分析惡性肉瘤的腫瘤免疫微環境。法國團隊分析結果顯示在綜合了各種組織型態的惡性肉瘤群體中，可以根據免疫細胞種類的差異，將所有的病患區分為五個免疫相關分類，分別為A至E五個組別。其中E組為免疫細胞浸潤最多的而且也發現B細胞不論是在哪種組織型態或是原發部位都對於預後都有顯著的正面影響 (p < 0.001)。而我們臺大醫院所提供的惡性肉瘤檢體分析也證實了以上的分析結果。另外也發現免疫組別 E 對於目前的免疫前哨抑制劑pembrolizumab有治療效果預測的能力 (p = 0.026)。 我們也希望知道腫瘤免疫微環境之組成是否可以幫助我們預測晚期惡性肉瘤病患接受其他非免疫藥物治療的效果預測性以及預後。我們收集總共34 個在我們所發起的”第 Ib/II 期單組檢測樂衛瑪 (lenvatinib) 合併賀樂維 (eribulin) 用於晚期惡性脂肪肉瘤以及惡性平滑肌肉瘤的臨床試驗”的檢體先用Nanostring nCounter 來做mRNA 的量化，而後再用之前提及的轉譯體演算法分析腫瘤免疫微環境中的免疫細胞比例。我們發現腫瘤免疫微環境在接受這些藥物治療後，樹突細胞會有顯著增加 (p = 0.037)。而且某些特定的免疫細胞變化也與腫瘤對於治療的控制的不惡化存活期有顯著的關係。這些結果顯示腫瘤免疫微環境中的免疫細胞是與惡性肉瘤與治療的預後有相關性的。 我們的研究顯示了腫瘤免疫微環境內的免疫細胞分類或許可以對現行已有的惡性肉瘤分類模式提供額外的訊息。我們會持續進行藉由免疫的特色來分類惡性肉瘤的研究也希望對於這個困難治療的癌症提供更精準的治療選擇。

Soft tissue sarcoma (STS), unlike many other cancer types, could occur in any anatomical location and are associated with more than 100 different types of histology. This plethora of variations challenges a proper classification that could dictate both prognosis and treatment. Histopathology is the most common method proposed to classify STS. However, in our study in collaboration with seven Asian countries that collected 276 advanced angiosarcoma patients, primary site was the most important prognostic factor for survival (hazard ratio (HR) of cutaneous vs visceral angiosarcoma 0.63, p = 0.004), suggesting that within one histology type different primary anatomical locations have differential clinical outcomes. On the other hand, not all STS in the same anatomical location have similar outcomes. In our study that collaborated with 5 medical centers across 3 continents, among the 61 patients with primary cardiac sarcoma, angiosarcoma tended to have worse survival as compared to other histologies (HR 1.84, p = 0.07). The sophisticated complexity between histology and anatomical location in STS provides reasons to pursue biological-based factors other than histology and anatomy to assist in the classification of STS patients. The success of immunotherapy in oncology has brought forth attention to the tumor immune microenvironment (TIME). The development of transcriptomic-based algorithms that could decipher bulk RNA sequencing data into different immune cell types also improves our ability to understand TIME. Because the TIME represents the interaction between host and tumor, we hypothesized that features in the TIME may be able to classify STS irrespective of histology or anatomical location. We collaborated with French researchers and together, we conceptualized the idea to use cell type-based transcriptomic approach to decipher the TIME in multiple STS histology subtypes. Based on immune cell type distribution, STS patients were classified into five (A to E) different sarcoma immune classes (SIC) with different prognosis. The E cluster is the most inflamed. Further, B cell signature was prognostic for survival across different histologies and anatomical locations (p < 0.001). These findings were confirmed by NTUH cohorts. We also demonstrated the predictive value of the E cluster patients who were more likely to benefit from the treatment of immune checkpoint inhibitor (p = 0.026). To further explore whether TIME may also be prognostic or predictive for advanced STS patients receiving non-immunotherapeutic agents, a total of 32 clinical specimens from our clinical trial “A single-arm phase Ib/II study of the combination of lenvatinib and eribulin in advanced adipocytic sarcoma and leiomyosarcoma (LEADER)” were selected for mRNA quantification by Nanostring nCounter platform. We found that the treatment of lenvatinib and eribulin could increase the quantity of dendritic cells in the TIME (p = 0.037) and changes in different immune cell types were associated with progression-free survival duration. These results suggested that the immune cell components of TIME were associated with prognosis in advanced STS. Our studies have shown that TIME may form a backbone for a novel classification which could bolster current classifications that are based on histology and anatomy. Our future works will refine the model of immunologic classification for STS, and thereby to help develop precision medicine of this difficult tumor.