Prognostic value of presurgical circulating tumor DNA (ctDNA) levels and other clinical factors in colon cancer.

N Naoya Akazawa (Department of Gastroenterological Surgery, Sendai City Medical Center Sendai Open Hospital, Sendai, Japan) S Saori Mishima K Koji Ando (Department of Surgery and Science, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan) D Daisuke Kotani Y Yoshiaki Nakamura H Hideaki Bando H Hiroya Taniguchi J Jun Watanabe T Takeshi Kato Y Yusuke Suwa (Department of Surgery, Gastroenterogical Centre, Yokohama City University Medical Center, Yokohama, Japan) K Keiji Hirata (Department of Surgery, School of Medicine, University of Occupational and Environmental Health, Kitakyushu, Japan) K Kozo Kataoka (Division of Lower GI, Department of Gastroenterological Surgery, Hyogo Medical University, Nishinomiya-Shi, Japan) A Antony Tin Y Yuefan Huang (3Natera, Inc., Austin, United States) S Susan Rojahn (Natera, Inc., Austin, TX) R Robert William Lentz (Natera, Inc., Austin, TX) A Adham A Jurdi (Natera, Inc., Austin, TX) M Minetta C. Liu T Takayuki Yoshino (National Cancer Center Hospital East, Kashiwa, Japan) E Eiji Oki

Abstract

221 Background: While post-surgical ctDNA is a known biomarker for risk of recurrence in colon cancer patients (pts), the prognostic value of presurgical ctDNA is less understood. Here, we explore whether presurgical ctDNA levels can predict outcomes and identify colon cancer pts who may benefit from earlier intervention. Methods: This retrospective study analyzed presurgical ctDNA (Signatera) data from 2,517 resectable stage I-IV colon cancer pts who enrolled in GALAXY (UMIN000039205) and did not receive neoadjuvant chemotherapy. The goal was to predict disease-free survival (DFS) by training a random forest classifier with 15 presurgical clinicopathological variables. The cohort was randomly partitioned into a training set (80%, n = 2018) and a test set (20%, n = 499) using stratified sampling to preserve the joint distribution of clinical stage and sex. The training set was used for model fitting and hyperparameter optimization via 5-fold cross-validation. The test set was used to estimate predictive performance. Variable importance was evaluated using permutation-based methods. To stabilize importance estimates, the final model was refitted on the full cohort with tuned hyperparameters. Multivariate ROC models informed the selection of thresholds for grouping patients by mean tumor molecules (MTM)/mL levels (undetectable, 0–1, >1–5, >5–55, and >55 MTM/mL). Finally, the association between ctDNA levels and early relapse (<6 months from surgery) was assessed among patients who received (n=980) or did not receive (n=1538) adjuvant chemotherapy (ACT). Results: Among the 15 presurgical clinicopathological factors assessed, ctDNA MTM/mL levels were the most important feature (100 on the relative scale) for predicting DFS, followed by TMB (58.3) and MSI status (53.6). Compared to pts with undetectable ctDNA, patients with >5–55 MTM/mL (HR, 1.70; 95% CI, 1.06–2.7, p=0.028) and >55 MTM/mL (HR, 2.36; 95% CI, 1.38–4.0, p=0.002) had a significantly inferior DFS (36-month DFS: 74.0% and 59.4%, respectively). Among all pts who relapsed (n=482), 37.1% (179/482) relapsed early. Presurgical ctDNA levels across all thresholds were significantly associated with the frequency of pts who experienced early vs. later relapse (p=1e-04), with ctDNA >5 MTM/mL associated with 6x higher odds of early relapse compared to 0 MTM/mL. ACT nearly halved the odds of early relapse (Mantel-Haenszel OR, 0.53; 95% CI, 0.37–0.76, p=0.0005), with 4.8% (47/980) of pts who received ACT relapsing early vs 7.5% (115/1538) of those who did not. Conclusions: Presurgical ctDNA levels in colon cancer pts undergoing upfront surgery may serve as a biomarker for predicting early relapse and poor DFS. Adjuvant chemotherapy significantly reduces the risk of early relapse, supporting further research into the use of longitudinal pre- and post-surgical ctDNA to guide optimal neoadjuvant and adjuvant systemic therapy. Clinical trial information: UMIN000039205 .

Article Details

Volume / Issue Vol. 44, Issue 2_suppl
Published January 10, 2026
Pages 221-221
ISSN 0732-183X
Publisher Lippincott Williams & Wilkins

Journal Info

Journal of Clinical Oncology

Lippincott Williams & Wilkins

ISSN: 0732-183X Health Sciences

Authors (20)

N

Naoya Akazawa

Department of Gastroenterological Surgery, Sendai City Medical Center Sendai Open Hospital, Sendai, Japan

S

Saori Mishima

K

Koji Ando

Department of Surgery and Science, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan

D

Daisuke Kotani

Y

Yoshiaki Nakamura

H

Hideaki Bando

H

Hiroya Taniguchi

J

Jun Watanabe

T

Takeshi Kato

Y

Yusuke Suwa

Department of Surgery, Gastroenterogical Centre, Yokohama City University Medical Center, Yokohama, Japan

K

Keiji Hirata

Department of Surgery, School of Medicine, University of Occupational and Environmental Health, Kitakyushu, Japan

K

Kozo Kataoka

Division of Lower GI, Department of Gastroenterological Surgery, Hyogo Medical University, Nishinomiya-Shi, Japan

A

Antony Tin

Y

Yuefan Huang

3Natera, Inc., Austin, United States

S

Susan Rojahn

Natera, Inc., Austin, TX

R

Robert William Lentz

Natera, Inc., Austin, TX

A

Adham A Jurdi

Natera, Inc., Austin, TX

M

Minetta C. Liu

T

Takayuki Yoshino

National Cancer Center Hospital East, Kashiwa, Japan

E

Eiji Oki