ctDNA dynamics and prognostic value during first-line treatment of metastatic colorectal cancer: Analysis from the phase 3 PARADIGM study.

3573 Background: The phase 3 PARADIGM trial showed overall survival (OS) and tumor response advantages of anti-EGFR (panitumumab, PAN) over anti-VEGF (bevacizumab, BEV) therapy with FOLFOX6 in first-line metastatic colorectal cancer (mCRC) despite similar progression-free survival (PFS), highlighting limitations of imaging-based PFS as a surrogate for clinical benefit. We investigated whether ctDNA dynamics provide more refined prognostic information complementary to baseline (BL) ctDNA. Methods: Plasma samples collected from 556 patients (pts) with mCRC paired at pre-treatment (BL) and post-treatment (after treatment discontinuation) were profiled for ctDNA. Distribution of the highest variant allele fraction was modeled using a Gaussian mixture model. At pre- and post-treatment, pts were classified into 3 groups—High, Mid, and Low—according to ctDNA levels. OS, post-treatment survival (PTS), and PFS were compared across 9 ctDNA dynamic categories (eg, High→Low, High→Mid, High→High) based on changes from pre- to post-treatment. Results: Lower BL ctDNA was significantly associated with longer OS (median OS [95% CI]: Low, n = 97, 50.7 mo [44.0–65.2]; Mid, n = 159, 34.1 mo [26.9–37.1]; High, n = 300, 27.8 mo [26.2–31.3]). BL ctDNA level also correlated with tumor volume and mutation burden. Pts with lower ctDNA at treatment discontinuation had longer PTS. Incorporating ctDNA dynamics provided additional prognostic value. Regardless of BL ctDNA level, pts whose post-treatment ctDNA decreased to Low had the best survival (eg, High→Low: 50.9 mo [35.8–61.0]), while pts with persistently high ctDNA showed the poorest survival (High→High: 23.3 mo [18.2–27.8]). The prognostic value of ctDNA dynamics was consistent across treatment arms. Among pts with High BL ctDNA (PAN, n = 152; BEV, n = 148), a greater proportion achieved Low ctDNA at post-treatment with PAN vs BEV (29 [19.1%] vs 16 [10.8%], p = 0.053), whereas the proportions achieving Mid ctDNA were similar (52 [34.2%] vs 58 [39.2%], p = 0.40). In the High→Low/Mid ctDNA subsets, treatment discontinuation following curative resection was more frequent with PAN vs BEV (High→Low: 18/29 [62.1%] vs 7/16 [43.8%]; High→Mid: 11/52 [21.2%] vs 5/58 [8.6%]). OS and PTS trends were favorable with PAN vs BEV (OS; High→Low: 61.0 vs 43.1 mo, HR 0.49, 95% CI 0.23–1.04; High→Mid: 33.6 vs 26.8 mo, HR 0.74, 95% CI 0.50–1.11, PTS; High→Low: 55.1 vs 39.0 mo, HR 0.48, 95% CI 0.22–1.03; High→Mid: 20.6 vs 19.4 mo, HR 0.72, 95% CI 0.48–1.07). Conclusions: ctDNA dynamics add prognostic value beyond BL ctDNA level. In pts with high BL ctDNA, PAN yielded more conversions into low ctDNA levels and discontinuations following curative resection, with OS/PTS trends favoring PAN. These findings suggest that ctDNA dynamics may better reflect long-term clinical benefit than PFS, although standardized thresholds and prospective validation are needed. Clinical trial information: NCT02394834 ; NCT02394795 .

Authors

Institutions

Publication Details

Journal
Journal of Clinical Oncology
Published
2026-05-27
DOI
https://doi.org/10.1200/jco.2026.44.16_suppl.3573
Primary Topic
Cancer Genomics and Diagnostics
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

ctDNA dynamics and prognostic value during first-line treatment of metastatic colorectal cancer: Analysis from the phase 3 PARADIGM study.

Junpei Soeda, Ikuo Mori, Katsuya Tsuchihara, Hiroyuki Uetake et al.
Journal of Clinical Oncology
Cancer Genomics and Diagnostics
article

ctDNA dynamics and prognostic value during first-line treatment of metastatic colorectal cancer: Analysis from the phase 3 PARADIGM study.

Junpei Soeda, Ikuo Mori, Katsuya Tsuchihara, Hiroyuki Uetake, Kiwamu Akagi, Kazunori Yamanaka, Eiji Oki, Kentaro Yamazaki, Takeshi Kato, Atsushi Ochiai, Riu Yamashita, Takayuki Yoshino, Takeshi Naitoh, Kei Muro, Kouji Yamamoto, Jun Watanabe, Takeo Sato, Yoshito Komatsu, Kohei Shitara
article en

Abstract

3573 Background: The phase 3 PARADIGM trial showed overall survival (OS) and tumor response advantages of anti-EGFR (panitumumab, PAN) over anti-VEGF (bevacizumab, BEV) therapy with FOLFOX6 in first-line metastatic colorectal cancer (mCRC) despite similar progression-free survival (PFS), highlighting limitations of imaging-based PFS as a surrogate for clinical benefit. We investigated whether ctDNA dynamics provide more refined prognostic information complementary to baseline (BL) ctDNA. Methods: Plasma samples collected from 556 patients (pts) with mCRC paired at pre-treatment (BL) and post-treatment (after treatment discontinuation) were profiled for ctDNA. Distribution of the highest variant allele fraction was modeled using a Gaussian mixture model. At pre- and post-treatment, pts were classified into 3 groups—High, Mid, and Low—according to ctDNA levels. OS, post-treatment survival (PTS), and PFS were compared across 9 ctDNA dynamic categories (eg, High→Low, High→Mid, High→High) based on changes from pre- to post-treatment. Results: Lower BL ctDNA was significantly associated with longer OS (median OS [95% CI]: Low, n = 97, 50.7 mo [44.0–65.2]; Mid, n = 159, 34.1 mo [26.9–37.1]; High, n = 300, 27.8 mo [26.2–31.3]). BL ctDNA level also correlated with tumor volume and mutation burden. Pts with lower ctDNA at treatment discontinuation had longer PTS. Incorporating ctDNA dynamics provided additional prognostic value. Regardless of BL ctDNA level, pts whose post-treatment ctDNA decreased to Low had the best survival (eg, High→Low: 50.9 mo [35.8–61.0]), while pts with persistently high ctDNA showed the poorest survival (High→High: 23.3 mo [18.2–27.8]). The prognostic value of ctDNA dynamics was consistent across treatment arms. Among pts with High BL ctDNA (PAN, n = 152; BEV, n = 148), a greater proportion achieved Low ctDNA at post-treatment with PAN vs BEV (29 [19.1%] vs 16 [10.8%], p = 0.053), whereas the proportions achieving Mid ctDNA were similar (52 [34.2%] vs 58 [39.2%], p = 0.40). In the High→Low/Mid ctDNA subsets, treatment discontinuation following curative resection was more frequent with PAN vs BEV (High→Low: 18/29 [62.1%] vs 7/16 [43.8%]; High→Mid: 11/52 [21.2%] vs 5/58 [8.6%]). OS and PTS trends were favorable with PAN vs BEV (OS; High→Low: 61.0 vs 43.1 mo, HR 0.49, 95% CI 0.23–1.04; High→Mid: 33.6 vs 26.8 mo, HR 0.74, 95% CI 0.50–1.11, PTS; High→Low: 55.1 vs 39.0 mo, HR 0.48, 95% CI 0.22–1.03; High→Mid: 20.6 vs 19.4 mo, HR 0.72, 95% CI 0.48–1.07). Conclusions: ctDNA dynamics add prognostic value beyond BL ctDNA level. In pts with high BL ctDNA, PAN yielded more conversions into low ctDNA levels and discontinuations following curative resection, with OS/PTS trends favoring PAN. These findings suggest that ctDNA dynamics may better reflect long-term clinical benefit than PFS, although standardized thresholds and prospective validation are needed. Clinical trial information: NCT02394834 ; NCT02394795 .

Journal of Clinical OncologyVol. 44(16_suppl)
Kansai Medical University (JP), Shizuoka Cancer Center (JP), Osaka National Hospital (JP), Kyushu University Hospital (JP), Chiba Cancer Center (JP), Japan Research Institute (JP), Aichi Cancer Center (JP), National Hospital Organization (JP), Hokkaido University Hospital (JP), National Cancer Center Hospital East (JP), National Disaster Medical Center (JP), Takeda (Japan) (JP), Kitasato University (JP), International University of Health and Welfare (JP), Yokohama City University (JP)
No poverty
Openalex Percentile: Top 11%
Cancer Genomics and Diagnostics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.