AIによる大腸T1癌リンパ節転移予測

Prediction of Lymph Node Metastasis in T1 Colorectal Cancer with Artificial Intelligence

要旨●大腸T1癌は約10%にリンパ節転移を認め，内視鏡治療後の追加腸切除の要否判断が重要である．ガイドラインは低リスク群の抽出に大きく貢献する一方，高リスク群では予測精度に課題が残り，多くの患者で結果的にover-surgeryが生じている．近年，人工知能（AI）の導入によりガイドラインを上回る予測精度が示され，性別・腫瘍局在といった臨床因子の有用性も再認識された．また，病理診断の再現性の低さは根本的課題であり，AI支援デジタルパソロジーはこれを補完しうる．さらに，リンパ節転移リスクに加え，年齢・手術リスク・価値観を統合する意思決定支援AIは，T1癌治療の個別化を加速する可能性をもつ．

　T1 colorectal cancer carries an 〜10% risk of lymph node metastasis, making the crucial to decide between follow-up monitoring and additional surgical resection after endoscopic treatment. Contemporary clinical guidelines have remarkably contributed to identifying a low-risk group to be monitored ; however, their predictive accuracy for the high-risk group remains limited, resulting in overtreatment to this group. Recent advances in artificial intelligence（AI）have demonstrated a superior predictive performance compared with the guideline-based criteria and have highlighted the importance of clinical factors such as sex and tumor location. Moreover, limited reproducibility in pathological assessment represents a fundamental hurdle, for which the pathology-based AI may provide a complementary support. Decision-support AI, which integrates the lymph node metastasis risk with patient age, surgical risk, comorbidities, and individual preferences, can further personalize treatment strategies. Altogether, these developments suggest that AI-driven approaches may play an increasingly important role in optimizing decision-making for patients with T1 colorectal cancer.