肺癌長年位居台灣癌症死因之首。即使隨著低劑量電腦斷層（LDCT）篩檢的推廣，愈來愈多患者得以在早期診斷並接受手術切除，臨床上仍觀察到一個令人困惑的現象：部分看似已成功治療的早期肺癌患者，仍會在數年內復發甚至轉移。

目前臨床主要依賴病理分期、影像學特徵與少數分子標誌來評估預後風險，但這些工具對於腫瘤生物學異質性的掌握仍相當有限。高風險患者未能及早接受加強治療與密集追蹤，而低風險患者卻可能承受不必要的輔助治療與心理壓力。如何在疾病早期即精準辨識真正具有高復發潛能的族群，成為精準醫療長期未解、卻極為關鍵的臨床問題。

二、台灣癌症登月計畫：建置國人癌症多體學藍圖

為回應此一挑戰，台灣自 2015 年起即積極投入癌症多體學研究。我們率先與美國國家癌症研究院（NCI）展開癌症組織蛋白體合作，並於 2016 年受邀加入時任美國副總統 Joe Biden 推動的「癌症登月計畫（Cancer Moonshot）」，成為「國際癌症蛋白基因體學聯盟（ICPC）」的成員。

在中央研究院「政策額度計畫」的支持下，「台灣癌症登月計畫」正式啟動，聚焦台灣常見且具高度臨床需求的癌症類型，整合中研院、台灣大學的技術團隊和多家醫學中心的醫學專家，於國內首度系統性導入蛋白基因體學（proteogenomics）大數據策略。透過整合基因體、轉錄體、蛋白體、磷酸化蛋白體與完整臨床追蹤資料，我們希望能全面解碼國人癌症的成因、異質性與演化軌跡，為精準醫療奠定科學基礎。

台灣登月計畫第一期的早期肺癌研究成果，於 2020 年發表於 Cell 並登上封面，成為全球首度針對東亞族群解析非吸菸肺癌分子機制的深度多體學研究。隨後於 2025 年，透過台美跨國合作，完成涵蓋歐美與亞洲多族群的大規模肺腺癌蛋白基因體研究，成果刊登於 Cancer Cell，建立了全球首個跨族群肺癌分子圖譜藍圖。

三、發現「類晚期」新亞型：改寫早期肺癌的認知

透過病人腫瘤組織蛋白質體分析，我們發現即使同屬病理分期上的早期肺腺癌，腫瘤在蛋白體層次仍可進一步分群。其中一群患者呈現出一種前所未見的分子特徵：其蛋白質體表現高度類似病人腫瘤組織，卻發生在早期的第一期。我們將此新族群命名為「類晚期（late-like）」亞型。這類腫瘤在訊息傳遞路徑、侵襲與轉移相關蛋白的活化程度上，與真正的晚期腫瘤極為相近。後續臨床追蹤結果顯示，這些患者的復發風險明顯高於其他早期患者，復發比例甚至接近晚期肺癌。

這項發現不僅揭示蛋白質體分類捕捉了傳統病理分期無法分辨的關鍵生物學差異，也顯示在 EGFR 突變為主的早期肺癌族群中，腫瘤早期即存在高度異質性，為「預測早期肺癌預後」開啟全新的可能性。

四、從組織到血液：非侵入式預測的新工具

儘管以腫瘤組織為基礎的多體學分析可提供深度的生物學洞見，但其在實務上尚未成熟轉化為臨床常規檢測。為了提升技術的可應用性與產業轉譯潛力，我們進一步將研究策略由腫瘤組織延伸至血液分析，並開發出一套可用於預測肺癌復發風險的非侵入式檢測方法。

透過系統性分析，我們篩選出四個可於血液中穩定偵測的蛋白質生物標誌，並建立風險計算模型。患者僅需抽取極少量血液，即可計算個人化的復發風險分數，用以評估術後數年內復發的可能性。此一工具可協助臨床醫師在手術後即進行風險分層：對高風險患者及早規劃輔助治療與密集追蹤，對低風險患者則避免不必要的過度治療，真正落實精準醫療的精神。

五、邁向產業化：專利布局與檢測套組開發

在國科會「健康大數據永續平台計畫」的支持下，我們以「早期肺癌高復發風險預測」為目標，系統性推動研究成果的產業轉譯。全球專利調研顯示，肺癌檢測市場需求快速成長，但相關專利布局仍相對分散，具備高度創新與切入空間。

本技術已於 2025 年 8 月通過美國臨時專利申請，並同步推進台灣與國際正式專利布局。在產業端，我們串接台灣本土抗體原料廠與 GMP 製造體系，完成「早期肺癌高復發風險預測 ELISA 檢測套組」的小量試產。初步驗證結果顯示，模型所預測的高風險族群確實具有顯著較高的復發機率，而低風險族群則展現良好預後，顯示其臨床應用潛力。

六、台灣原創的精準醫療里程碑

「早期肺癌高復發風險預測套組」的關鍵價值，在於成功將高維度的多體學大數據，轉化為低侵入、高靈敏度、可於臨床落地的血液檢測工具。對醫師而言，這是一項即時且客觀的輔助決策指標；對患者而言，僅需一次抽血，便能在手術當下掌握未來數年的復發風險，爭取更早的預防與治療時機。

此成果不僅展現台灣在精準醫療領域的自主研發能量，也是一項具備明確技轉與產業化路徑的創新典範。此外，我們亦於 2022 年與國家高速網路與計算中心生醫團隊共同建置「台灣癌症登月多體學智識庫」，以肺癌為起點，透過動態視覺化平台呈現，未來將擴展至乳癌等癌別，並以「資料不攜出」與「非專屬技轉」模式開放，為產官學界提供高品質、多層次的研究資源，協助台灣在癌症精準檢測與新藥開發上建立關鍵利基。

（115年度TBF學術講座、中研院化學所　陳玉如特聘研究員）

---

Blood-based Prediction of Lung Cancer Recurrence Risk: A Precision Medicine Milestone from the Taiwan Cancer Moonshot Program

1. An Unmet Clinical Need in Early-Stage Lung Cancer: Who Will Relapse?

Lung cancer has long been the leading cause of cancer-related mortality in Taiwan. With the implementation of low-dose computed tomography (LDCT) screening, an increasing number of patients are now diagnosed at an early stage and receive curative surgical resection. Nevertheless, a perplexing clinical reality remains: a substantial proportion of patients with early-stage lung cancer still experience disease recurrence or metastasis within a few years after surgery.

Current clinical risk assessment relies primarily on pathological staging, imaging features, and a limited set of molecular markers. However, these tools provide only a coarse representation of tumor biology and often fail to capture the underlying heterogeneity that drives aggressive behavior. As a result, some high-risk patients miss the opportunity for early intervention and intensified surveillance, while low-risk patients may be exposed to unnecessary adjuvant therapy and psychological burden. Identifying patients with a genuine risk of recurrence at an early stage remains one of the most critical unmet needs in precision oncology.

2. The Taiwan Cancer Moonshot Program: Creating a Multi-Omics Ecosystem for Precision Medicine

To address this challenge, Taiwan has made sustained investments in large-scale cancer multi-omics research. Since 2015, we have collaborated with the U.S. National Cancer Institute (NCI) on cancer proteomics initiatives, and in 2016 were invited to participate in the 「Cancer Moonshot」 program launched by then–Vice President Joe Biden. Taiwan subsequently became an active member of the 「International Cancer Proteogenome Consortium (ICPC)」.

With support from the Academia Sinica-initiated National Policy Program, the Taiwan Cancer Moonshot Program was formally launched to focus on major cancers prevalent in the Taiwanese population. By integrating the technology-centric teams at Academia Sinica, National Taiwan University and the clinical experts from multiple medical centers, the program pioneered the adoption of proteogenomics-based big-data strategies in Taiwan, combining genomics, transcriptomics, proteomics, phosphoproteomics, and comprehensive clinical annotation. The overarching goal is to decode the molecular etiology and patient heterogeneity of cancer in Taiwanese patients and to translate these insights into actionable precision-medicine strategies.

The first phase of the program culminated in a landmark study on early-stage lung cancer, published in Cell in 2020 and featured on the journal cover. This work represented the world’s first deep multi-omics analysis of non-smoking lung cancer in East Asian populations, uncovering potential disease mechanisms unique to this group. In 2025, through close Taiwan–U.S. collaboration, the program further completed the first large-scale proteogenomic study of lung adenocarcinoma spanning both Asian and Western populations, published in Cancer Cell. Together, these studies established the first cross-ethnic molecular blueprint of lung cancer.

3. Discovery of a “Late-Like” Subtype: Redefining Early-Stage Lung Cancer

Despite curative-intent surgery and standard adjuvant treatments, a subset of patients with early-stage lung cancer experience rapid recurrence. Identifying this high-risk population remains a central clinical challenge.

Through comprehensive proteomic profiling of tumor tissues, we found that early-stage lung adenocarcinomas can be further stratified at the proteome level. Notably, we identified a previously unrecognized subgroup characterized by protein expression patterns that closely resemble those of advanced-stage tumors. We termed this subgroup the “late-like” subtype. Although these tumors are classified as early stage by conventional pathology, they exhibit highly activated signaling pathways associated with invasion, metastasis, and aggressive tumor behavior. Long-term clinical follow-up revealed that patients in this subgroup have a markedly elevated risk of recurrence, with recurrence rates approaching those observed in advanced disease.

This discovery reveals that proteomics subtyping captures critical biological differences beyond traditional clinical staging systems and highlights the profound heterogeneity that exists even among early-stage, EGFR-mutant lung cancers. Importantly, it opens a new avenue for predicting prognosis at an early stage.

4. From Tumor Tissue to Blood: A Non-Invasive Tool for Risk Prediction

While tissue-based multi-omics analysis provides deep biological insights, it is not readily applicable as a routine clinical test. To translate our findings into practical use, we extended our approach from tumor tissue to blood-based analysis and developed a non-invasive method for predicting recurrence risk.

By systematically analyzing patient biospecimens, we identified four protein biomarkers that can be robustly detected in blood. These markers were integrated into a risk-scoring model that estimates the likelihood of disease recurrence within several years after surgery. Using only a minimal volume of blood, clinicians can obtain a personalized recurrence-risk score.

This tool enables risk-adapted clinical decision-making: high-risk patients may benefit from earlier adjuvant therapy and closer surveillance, while low-risk patients can be spared unnecessary treatment. In this way, the approach embodies the core principles of precision medicine.

5. Toward Industrial Translation: Patent and Diagnostic Kit Development

With support from the National Science and Technology Council’s Sustainable Health Big Data Platform Program, we have actively advanced the industrial translation of this technology, targeting the development of an early-stage lung cancer recurrence-risk prediction kit. Global patent landscape analysis indicates rapidly growing demand for lung cancer diagnostics, while existing intellectual-property coverage remains fragmented, leaving substantial room for innovation.

In August 2025, this technology was granted a U.S. provisional patent, with formal patent applications currently underway in Taiwan and other regions. On the industrial side, we have partnered with domestic antibody manufacturers and GMP-certified facilities to complete pilot production of an ELISA-based recurrence-risk prediction kit for early-stage lung cancer. Preliminary validation shows that patients classified as high risk by the model indeed exhibit significantly higher recurrence rates, whereas those in the low-risk group demonstrate favorable outcomes, underscoring the test’s translational potential.

6. A Milestone in Taiwan’s Precision Medicine Innovation

The core value of this technology lies in its successful transformation of high-dimensional multi-omics big data into a low-invasive, highly sensitive, and clinically deployable blood test. For physicians, it provides an objective and timely decision-support tool; for patients, a single blood draw at the time of surgery can offer critical insight into recurrence risk over the coming years, enabling earlier preventive or therapeutic interventions.

Beyond its clinical utility, this achievement highlights Taiwan’s capacity for original innovation in precision medicine and exemplifies a clear pathway from academic discovery to industrial translation. In parallel, we collaborated with the National Center for High-Performance Computing in 2022 to establish the Taiwan Cancer Moonshot Multi-Omics Knowledgebase. Using lung cancer as a foundational model, this platform presents large-scale cancer data through dynamic visualization interfaces and will be expanded to include other cancer types, such as breast cancer. Operated under a “data-stay-in-place” and non-exclusive technology-transfer framework, the resource is designed to support academia, industry, and government alike, helping to create strategic advantages for precision diagnostics and anticancer drug development in Taiwan.

(2026 TBF Chair in Biotechnology, Professor Yu-Ju Chen)