Patient characteristics

A total of 62 patients underwent ctDNA analysis following colorectal cancer surgery between June 2023 and June 2025. Among them, 56 patients had ctDNA evaluated within 4 to 10 weeks postoperatively, with median turnaround time from blood collection to ctDNA result was 9 (range 6–13) days. All samples were successfully analyzed, resulting in 100% quality control pass rate. Of these 56 patients, 18 were ctDNA-positive and 38 were ctDNA-negative. Median follow-up was 13 (range 1–26) months. Three patients (2 ctDNA-positive and 1 ctDNA-negative) were censored due to follow-up at external institutions (Fig. 1). When ctDNA was detected, additional imaging examinations (CT scan, liver MRI and PET if possible) were performed to detect early recurrence. Patient characteristics of the 53 patients excluding 3 patients who were lost to follow-up are detailed in Table 1. Most patients had an ECOG performance status of 0; however, approximately 60% had one or more comorbidities.

Fig. 1Table 1 Patient characteristicsDetails of ctDNA-positive cases (N = 16)

Among the 16 patients, median follow-up duration was 9.5 months (range 1–26 months). Imaging within one month of the assay result detected radiological recurrence in 9 patients. The postoperative treatment courses of the 16 patients with ctDNA positivity are summarized in Fig. 2. Among the 9 patients with early recurrence, metastatic sites included the liver (n = 4), lymph nodes (n = 4), and peritoneum (n = 1). All of them received systemic chemotherapy for metastatic disease following the diagnosis of recurrence and three successfully underwent curative metastasectomy. The remaining 6 patients received CAPOX as adjuvant chemotherapy (pStage III/IV: 3/3), with a median follow-up duration of 9.5 months (range 6–24 months). To date, 5 patients (83.3%) remain disease-free and 1 patient (16.7%) had thoracic spine metastasis 22 months after curative primary surgery (#2 in Fig. 2). In contrast, 1 patient (pStage IV) who did not receive adjuvant chemotherapy developed peritoneal recurrence at 4 months postoperatively and was transitioned to best supportive care (#1 in Fig. 2).

Fig. 2

Details of ctDNA-positive patients (n = 16). The start point for each bar in the swimmer’s plot corresponds to the date of curative-intent surgery. *A 5-fluorouracil, leucovorin, oxaliplatin, and irinotecan combined with bevacizumab (FOLFOXIRI + Bmab), B 5-fluorouracil, leucovorin, and irinotecan combined with bevacizumab (FOLFIRI + Bmab), C 5-fluorouracil, leucovorin, and irinotecan combined with ramucirumab (FOLFIRI + Rmab)

Surveillance of ctDNA-negative cases

Although ctDNA testing after colorectal cancer surgery is available through self-funding at our institution, repeat testing was performed in some cases upon patient request. Among the 16 patients with available follow-up, 3 underwent serial ctDNA assessments (#2, 3, and 7 in Fig. 2). In cases #2 and #7, imaging studies were performed immediately following the postoperative ctDNA-positive results, but no overt recurrent lesions were identified. However, due to a high suspicion of subclinical micrometastases, both patients received adjuvant CAPOX chemotherapy. Subsequent ctDNA testing demonstrated conversion to ctDNA negativity. Case #2 experienced bone metastasis at 22 months postoperatively and underwent palliative radiotherapy. In contrast, case #7 has remained recurrence-free to date. In case #3, imaging performed based on the postoperative ctDNA-positive result revealed liver metastasis, and the patient received one cycle of FOLFOXIRI plus bevacizumab, followed by surgical resection of the metastatic lesion. ctDNA remained positive one month after the hepatectomy despite the absence of radiological recurrence. Therefore, the patient underwent seven cycles of CAPOX chemotherapy. A third ctDNA test following completion of CAPOX confirmed conversion to negativity, and the patient remains disease-free to date.

Tumor fraction and metastatic sites

Figure 3 shows the association between tumor fraction values in ctDNA-positive patients after curative-intent surgery and the sites of recurrence, which were mostly liver and lymph nodes.

Fig. 3

Tumor Fraction and metastatic sites

Details of ctDNA-negative cases (N = 37)

Among the 38 patients with postoperative ctDNA negativity, 37 were available for follow-up. The distribution of pathological stage was I/II/III/IV: 3 (8.1%)/8 (21.6%)/19 (51.4%)/7 (18.9%), with median follow-up of 13 months (range 1–24 months). All 3 pStage I patients did not receive adjuvant chemotherapy and have remained disease-free to date. Among the 8 pStage II patients, none received adjuvant chemotherapy. One patient (#3;12.5%) developed pulmonary metastasis and was treated with systemic chemotherapy, while the other 7 (#4, 9, 13, 16, 19, 22, 27; 87.5%) have not had recurrence. In the 19 pStage III patients, 9 (#5, 7, 18, 20, 28, 31, 32, 34, 36; 47.4%) underwent observation only, 9 (#12, 14, 15, 17, 21, 25, 33, 35, 37; 47.4%) received capecitabine monotherapy, and 1 (#26; 5.3%) received CAPOX. Of those under observation, 1 (#5; 11.1%) died of a non-cancer cause, and the other 8 (88.9%) have remained disease-free. Among the patients treated with capecitabine alone, 1 (#23; 11.1%) developed peritoneal dissemination; the rest (88.9%) have had no recurrence. The patient who received CAPOX also remains recurrence-free. Of the 7 pStage IV patients, 6 (#6, 8, 11, 24, 29, 30; 85.7%) were followed without adjuvant therapy and 1 (#23; 14.3%) received capecitabine monotherapy. 2 patients (#6, 8; 33.3%) developed recurrence (lymph node and lung metastasis, respectively) at 9 months after curative-intent surgery and are currently receiving systemic chemotherapy. The other 4 (66.7%) have not relapsed. The patient treated with capecitabine (#23) developed pulmonary recurrence and is receiving systemic therapy. For patients with negative postoperative ctDNA results, de-escalation to capecitabine monotherapy or observation alone was administered, particularly in elderly patients. (Fig. 4).

Fig. 4

Details of ctDNA-negative patients (n = 37)

Surveillance of ctDNA-negative cases

Among the 37 patients with ctDNA negativity, 6 underwent serial ctDNA assessments (#5, 10, 15, 18, 19, 23, 26 in Fig. 4). Case #5 underwent ctDNA testing twice: the first at 4 weeks and the second at 4 months after curative-intent surgery. The second test was positive; however, imaging studies performed in response to the positive ctDNA result did not reveal recurrence, and the patient was managed with continued surveillance. The patient died of a non-cancer-related cause 8 months after surgery. Case #23 received adjuvant capecitabine monotherapy. Postoperative ctDNA testing was performed at 4 weeks and again at 4 months following the initial curative-intent resection, with both results returning negative. However, at 6 months postoperatively, surveillance CT identified pulmonary metastasis. The patient subsequently underwent surgical resection of the metastatic lesion. Adjuvant chemotherapy was resumed thereafter, and the patient remains disease-free. Cases #10, #15, #18, #19, and #26 underwent ctDNA testing twice, and both results were negative. These patients have had no recurrence after the primary surgery.

Recurrence sites—ctDNA-positive vs. -negative

Among the 16 patients with postoperative ctDNA positivity, recurrence was diagnosed based on ctDNA findings in 11 patients (positive predictive value 68.8%) (Fig. 5). Among these, 4 (36.4%) had liver metastases, 4 (36.4%) had lymph node metastases, 2 (18.2%) had peritoneal dissemination, and 1 (9.1%) had thoracic spine metastasis. No pulmonary metastasis was observed in this group. The median interval from curative-intent surgery to detection of liver metastasis was 2 months (range 0–3 months).

Fig. 5

Recurrence sites based on ctDNA positivity: a ctDNA-positive cases (N = 11); b ctDNA-negative cases (N = 5)

In contrast, among the 37 ctDNA-negative patients with follow-up, 5 (13.2%) were diagnosed with recurrence (negative predictive value 86.5%). Of these, 3 (60.0%) had pulmonary metastases, 1 (20.0%) had lymph node metastasis, and 1 (20.0%) had peritoneal dissemination. No liver metastases were detected in this group (Fig. 5). The median interval between curative-intent surgery and detection of lung metastasis was 6 months (range 3–10 months).