Molecular Residual Disease and Recurrence in Rectal Cancer Patients Undergoing Upfront Surgery: A Prospective Cohort Study.

METHODS

Study Design and Participants
The GALAXY study is a prospective, large-scale, nationwide registry designed to monitor ctDNA status for patients with clinical stage II to IV CRC undergoing complete surgical resection. The key eligibility criteria for participation in the GALAXY study were (1) histopathologically confirmed colorectal adenocarcinoma, (2) clinical stage II to IV, (3) scheduled curative-intent resection, (4) patient age ≥20 years, and (5) Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) of 0 or 1. Patients diagnosed with other malignancies within the past 5 years were excluded. All participants provided written informed consent. Blood samples for ctDNA analysis were collected presurgery and at 4, 12, 24, 36, 48, 72, and 96 weeks after surgery until recurrence (whichever occurred first). Computed tomography imaging was performed every 6 months after surgery. The detailed study design, methods, and protocol have been previously published.16 This study was approved by the institutional review board of the National Cancer Center, Japan, and authorized by the head of each participating institution. The study is registered in the Japan Registry of Clinical Trials (UMIN000039205) and was conducted in accordance with the Declaration of Helsinki.
The inclusion criteria for patients included in the analysis reported herein (GALAXY cohort B1) included confirmed pathologic stage II and III rectal cancer, R0 resection margins, availability of clinical follow-up data, and at least 1 ctDNA time point in the MRD or surveillance window. Patients enrolled in the interventional phase III trials (ALTAIR and VEGA) associated with CIRCULATE Japan, and patients whose sample(s) failed quality control (QC) were excluded.
The GALAXY study comprises 4 distinct cohorts based on disease stage and treatment setting. Cohort A includes patients with clinical stage II to III colon or rectosigmoid cancer scheduled for R0 resection. Cohort B (focus of this analysis) includes patients with clinical stage II to III rectal cancer (excluding rectosigmoid) scheduled for R0 resection, including those with lateral lymph node metastasis. Cohort C includes patients with clinical stage IV or relapsed (M1) colorectal cancer undergoing curative-intent resection. Cohort D includes patients with pT1 colorectal cancer following local resection who are scheduled for additional resection due to noncurative pathologic features (eg, deep submucosal invasion, LVI, poor differentiation, or high-grade tumor budding).

Tumor-informed ctDNA Testing
A clinically validated, personalized, tumor-informed 16-plex polymerase chain reaction (PCR)-next-generation sequencing assay (SignateraTM, Natera Inc.) was used for the detection and quantification of ctDNA in blood samples as previously described.16 Briefly, formalin-fixed, paraffin-embedded tumor tissue from surgical resection or biopsy samples and matched normal DNA extracted from a peripheral blood sample were processed for whole-exome sequencing to target up to 16 tumor-specific somatic single-nucleotide variants (SNVs) to track in the associated patient’s plasma. Cell-free DNA was extracted from patient plasma (median 9.4 mL, range 1.7–12.8 mL) at a given time point and was used to detect ctDNA. Plasma samples with at least 2 tumor-specific variants detected above a predefined confidence threshold were defined as ctDNA-positive. ctDNA concentration was reported as mean tumor molecules per mL of plasma (MTM/mL).

Statistical Analysis
The primary endpoint was DFS, which was defined as the time between the date of landmark and the date of diagnosis with recurrence or death due to any cause or the latest radiologic assessment. Recurrence was determined based on diagnostic imaging or other diagnostic procedures if imaging was not confirmative (eg, colonoscopy to diagnose local recurrence). The χ2 test was used to compare categorical variables. Survival analyses were carried out using R software v4.4.0 using packages survival and survminer. The survival distribution was estimated using the Kaplan-Meier method. Differences between the groups were tested using the log-rank test. A multivariable Cox proportional hazards model was used to assess prognostic factors associated with DFS (coxph and cox.zph). Clinically relevant cutoffs were applied for demographic variables, wherever appropriate.
The MRD window was defined as 2 to 10 weeks postsurgery before the start of any adjuvant therapy. MRD analyses were landmarked at the date of the MRD time point to account for immortal time bias. The surveillance window was defined as the time from 4 weeks post-ACT or the end of the MRD window if the patient had no ACT until the last follow-up or recurrence. The surveillance analyses were landmarked at 10 weeks postsurgery.
The secondary endpoint was ctDNA clearance after ACT, which was defined as the time elapsed from the MRD-window result until the date of the first negative ctDNA result, irrespective of subsequent ctDNA status. The exploratory endpoint was molecular recurrence analysis. These analyses were performed using R software v4.4.0, and Cox regression was used to compare cumulative incidence function differences between the ACT and observation groups. The ctDNA clearance and molecular recurrence analyses were landmarked at 2 months postsurgery, and were adjusted for age, sex, and performance status. ctDNA clearance analyses were landmarked from the dates of blood collection at 3-month, 6-month, and MRD-window time points. P values <0.05 were considered statistically significant.

RESULTS

Patient Characteristics
A total of 1017 patients with rectal cancer were enrolled in the GALAXY study between May 2020 and June 2024. For the current analysis, 767 patients were excluded due to enrollment into other cohorts (A, C, and D) or interventional phase III trials, receipt of neoadjuvant therapy (cohort B2), confirmed pathologic stage 0/I or IV disease, incomplete resection, or indeterminate residual tumor status (RX) (Fig. 1). The final analysis included 250 patients with stage II to III who underwent upfront surgery were included in this analysis, with a total of 2254 plasma time points (Fig. 1). Based on the availability of plasma samples, 246 patients had available plasma samples within the MRD window, 223 were evaluable at the 3-month postsurgery time point, and 182 were evaluable at 6-month postsurgery (Fig. 1).
At the baseline (before any treatment, including surgery), 97.14% (238/245) of patients were ctDNA positive. In the MRD window, 14.2% (35/246) were ctDNA positive (MRD positive). Supplementary Table 1, Supplemental Digital Content 1, http://links.lww.com/SLA/F640 summarizes patient characteristics overall as well as relative to ctDNA status in the MRD window. Patients with higher pathologic stage (III vs II) and pathologic N-stage (N1-N2 vs N0) had significantly higher rates of ctDNA positivity in the MRD window (Supplementary Table 1, Supplemental Digital Content 1, http://links.lww.com/SLA/F640). In addition, 127 patients (50.8%) had tumors located in the upper rectum (Ra), and 123 (49.2%) had tumors located in the lower rectum (Rb), and ctDNA-positivity rates were similar between locations. Patients with RAS-mutated tumors tended to have a higher ctDNA-positivity rate in the MRD window than those with wild-type RAS, although the differences did not reach statistical significance (Supplementary Table 1, Supplemental Digital Content 1, http://links.lww.com/SLA/F640). Of note, cohort B1 included only 2 patients with BRAF-mutated tumors and 3 patients with MSI-H tumors, precluding any analysis of these subsets.

Association of Postoperative ctDNA Status With DFS
The median follow-up period was 22.0 months (range: 1.2–36 mo). Among 35 patients who were ctDNA positive in the MRD window, 71.4% (25/35) experienced a DFS event (recurrence or death), compared with only 13.3% (28/211) of patients who were ctDNA negative in the MRD window, [hazard ratio (HR): 9.96, 95% CI: 5.76–17.2, P<0.0001], demonstrating a 12-month DFS of 35.0% (95% CI: 19.48–50.9%) versus 89.5% (95% CI: 84.3%–93%), respectively (Fig. 2A). This trend was observed for both pathologic stages II and III (stage II: HR: 27.14, 95% CI: 6.4–115.04, P<0.0001, Supplementary Fig. 1A, Supplemental Digital Content 1, http://links.lww.com/SLA/F640; stage III: HR: 9.44, 95% CI: 7.79–11.42, P<0.0001, Supplementary Fig. 1B, Supplemental Digital Content 1, http://links.lww.com/SLA/F640). Of the 28 ctDNA-negative patients during the MRD window who experienced a DFS event, 5 died due to other causes, 15 experienced recurrence in the lung, 2 in the peritoneum, 2 in the liver, and 1 in the brain, and 3 had LN/local recurrence. Notably, of the 23 ctDNA-negative patients with radiologic recurrence, 14/20 with available surveillance time points turned ctDNA positive. In the multivariate analysis for DFS in patients with pathologic stage II to III disease, ctDNA positivity in the MRD window was the most significant prognostic factor associated with increased risk for recurrence (HR: 7.82, 95% CI: 4.17–14.6, P<0.001, Fig. 2B). The only other clinicopathological risk factor traditionally used for staging and prognostication that was significantly associated with DFS was pathologic N stage (HR: 2.92, 95% CI: 1.33–6.4, P=0.007, Fig. 2B).
At the 3-month (70–112 d) postsurgery time point, 19 patients were ctDNA positive. Of those, 73.7% (14/19) experienced recurrence, whereas 15.2% (31/204) of patients who were ctDNA negative at this time point recurred (HR: 7.98, 95% CI: 4.22–15.08, P<0.0001, Fig. 2C), demonstrating a 12-month DFS of 23.7% (95% CI: 7.58%–44.7%) versus 86.3% (95% CI: 80.5%–90.5%), respectively (Fig. 2C). Of the 31 patients who were ctDNA negative but experienced an event, 4 patients died from other causes, 15 had recurrence in lung, 4 in liver, 3 in peritoneum, 5 patients had LN/Local recurrence. Similarly, ctDNA positivity at the 6-month postsurgery (160–200 d) time point was predictive of inferior DFS (HR: 15.16, 95% CI: 6.18–37.18, P<0.0001, Fig. 2D).
In the surveillance window, 17.2% (40/233) of patients were ctDNA positive at least at one time point, and the remaining 82.7% (193/233) were ctDNA-negative at all time points tested. Compared with patients who were serially ctDNA-negative, patients with ctDNA positivity at any time point during the surveillance window were nearly 25-fold more likely to recur (HR: 24.95, 95% CI: 12.02–51.80, P<0.0001, Fig. 2E), with a 24-month DFS of 15.40% (95% CI: 4.61%–23.10%) versus 94.60% (95% CI: 89.60%–97.20%; P<0.0001). Furthermore, in the multivariate analysis, ctDNA positivity during the surveillance window was the strongest prognostic factor associated with poor DFS (HR: 31.71, 95% CI: 12.88–78.10, P<0.001, Fig. 2F). The only other clinicopathological factor that was prognostic of DFS was pathologic N stage (HR: 3.46, 95% CI: 1.36–8.80, P=0.009, Fig. 2F).

Association of Postoperative ctDNA Status and Levels With Recurrence Site
Among patients who recurred during follow-up, 17 had recurrence in the liver, 18 in the lung, 5 in the lymph node, 4 had local recurrence, 3 in the peritoneum, and 1 in the brain. In the MRD window, ctDNA positivity was observed in 88.2% (15/17) of patients with liver recurrence, 80% (4/5) with lymph node recurrence and 50% (2/4) with local recurrence, whereas ctDNA positivity in the MRD window was observed in 33.3% (1/3) of patients with peritoneal recurrence, 16.7% (3/18) with lung metastases and 0% (0/1) with brain recurrence. Liver metastases were most common in ctDNA-positive patients, while lung metastases were most common in ctDNA-negative patients. In the longitudinal setting, ctDNA positivity was observed in all (100%, 29/29) patients with liver, local, lymph node, and peritoneal dissemination recurrences at any point postsurgery. On the other hand, ctDNA positivity in the longitudinal setting was observed in 55.6% (10/18) lung metastases (Supplementary Table 2, Supplemental Digital Content 1, http://links.lww.com/SLA/F640).

Association of ctDNA Status in the MRD Window and ctDNA Clearance With DFS Benefit of ACT
Of the 211 patients who were ctDNA negative in the MRD window, 45.5% (96/211) received ACT versus observation (Fig. 3A). In the patients with ctDNA negativity in the MRD window, no statistically significant benefit of ACT was observed (HR: 0.59, 95% CI: 0.26–1.35, P=0.211) (Fig. 3B). On the other hand, patients who were ctDNA-positive in the MRD window derived significant benefit from ACT (HR: 0.28, 95% CI: 0.09–0.89, P=0.031) (Fig. 3C), demonstrating a median DFS of 9.33 months in the ACT group versus 5.62 months in the observation group.
Among MRD-positive patients who were treated with ACT (n=24, Fig. 3D), DFS was lower in those who did not achieve ctDNA clearance at 3 months compared with those who did, although this difference was not statistically significant (HR: 2.39, 95% CI: 0.86–6.65, P=0.096, Fig. 3E). The trend was stronger when ctDNA clearance was analyzed from MRD-window to 6-month time point (n=15), wherein patients with no ctDNA clearance had significantly inferior DFS compared with those who achieved clearance (HR: 20.44, 95% CI: 1.81–230.77, P = 0.015, Fig. 3F).

ctDNA Dynamics Patterns and Molecular Recurrence
Of the 250 patients in this cohort, the ctDNA status, both in the MRD window and at the 6-month postsurgery time point, was available for 180 patients. We implemented a landmark analysis to account for the immortal time bias and included only the patients who lived at least 6 months postsurgery in this analysis. Of the 180 patients, 2.8% (5/180) stayed ctDNA positive from MRD to 6-month time point, 87.2% (157/180) stayed ctDNA negative at both time points, 2.2% (4/180) converted from ctDNA negative in the MRD window to ctDNA positive at 6-month time point, and 7.8% (14/180) converted from ctDNA positive to ctDNA negative (Fig. 4). Compared with patients who were persistently negative, a significantly higher risk of recurrence was observed for patients who converted from ctDNA negative to ctDNA positive (HR: 8.22, 95% CI: 1.86–36.32, P=0.0055), and those who remained persistently positive (HR: 45.48, 95% CI: 14.31–144.57, P<0.0001) (Fig. 4).
We then evaluated the timeline of molecular recurrence among patients who were MRD negative but later turned ctDNA-positive before radiologic recurrence. Of the 212 MRD-negative patients, 12.3% (26/212) had molecular recurrence before radiologic recurrence. Of these, 80.8% (22/26) and 100% (26/26) turned ctDNA positive by 12- and 18-month postsurgery (Fig. 5A). Compared with the patients who remained serially negative, patients who later turned positive (molecular recurrence) had significantly inferior DFS (HR: 13.58, 95% CI: 6.01–30.70, P<0.0001, Fig. 5B), comparable to MRD-positive patients (P=0.062; Fig. 5B).

DISCUSSION
Our study provides compelling evidence that postoperative ctDNA status strongly predicts DFS in patients with stage II to III rectal cancer managed with upfront surgery. Our findings align with previous CRC studies where postsurgical ctDNA is indicative of MRD and correlates with an increased likelihood of recurrence.17 The high sensitivity of ctDNA analysis allows for the detection of residual disease that may not be evident through conventional imaging or pathologic assessment, thereby providing critical prognostic information that can influence postoperative management strategies.
Notably, our data suggest that the majority of ctDNA-negative patients at the MRD window who experienced an event had noncancer-related death or a recurrence in sites known for low ctDNA shedding. However, the majority of these patients converted positive during longitudinal ctDNA assessment. These findings are in agreement with previously reported data from the Galaxy cohort, which specifically analyzed patients with stage II to III colon cancer, wherein patients who converted positive or remained positive experienced inferior long-term outcomes.18 Our current study focuses on rectal cancer patients and highlights the importance of serial ctDNA testing post-MRD for early detection of molecular recurrence, which may prompt a timely change in treatment management.
Neoadjuvant chemoradiotherapy and total neoadjuvant therapy (TNT) have increasingly replaced upfront surgery in patients with stage II to III rectal cancer. Thus, most studies in early-stage rectal cancer have evaluated the utility of ctDNA testing for monitoring response to neoadjuvant therapy and for postsurgery prognostication.19–26 Patients receiving chemotherapy in the neoadjuvant setting tend to have poor adherence to ACT; thus, our cohort exclusively analyzed patients receiving upfront surgery, allowing true evaluation of ACT benefit. The prospective ENSEMBLE-127 and ENSEMBLE-228 trial within the CIRCULATE-Japan Study will report on the ctDNA’s value for monitoring TNT efficacy, local regrowth, and predicting prognosis. Current practice guidelines by the National Comprehensive Cancer Network (NCCN) and European Society for Medical Oncology (ESMO) recommend 4 months of ACT for patients with locally advanced rectal cancer (LARC).2,10 However, this recommendation is largely based on early studies in unselected patient populations.29 Our data suggest that ctDNA status has the potential to effectively guide ACT decisions: MRD-positive patients showed a substantial benefit from ACT, whereas MRD-negative patients did not. Identifying high-risk (MRD-positive) patients allows clinicians to make informed decisions regarding the necessity and intensity of adjuvant therapy. For MRD-negative patients, unnecessary ACT can be avoided, thereby improving patients’ quality of life and reducing health care costs. The benefit of ACT in MRD positive but not in MRD-negative patients has also been demonstrated in patients with colon cancer in the updated analysis of GALAXY and BESPOKE CRC studies, as well as the DYNAMIC trial.17,18,30,31 Upon analyzing ctDNA dynamics from the MRD window to 6 months postsurgery, patients who remained or turned ctDNA positive at the 6-month time point were observed to have significantly shorter DFS compared with those who were serially ctDNA negative. These patients could benefit from ctDNA-directed treatment changes, such as longer duration of ACT or treatment escalation. ctDNA-guided clinical trials are expected to find new treatment strategies for improving the outcomes of these patients, who do not derive long-term benefit from the standard-of-care treatment. The ongoing ALTAIR trial, a phase 3 interventional trial in the CIRCULATE-Japan platform, is evaluating the benefit of treatment escalation (trifluridine/tipiracil over placebo) on DFS in patients with CRC who have a molecular relapse after standard-of-care treatment but no clinical evidence of recurrence.32

Despite the overall prognostic utility of ctDNA, we noted that patients with lung recurrences were less likely to have ctDNA positivity. This observation is consistent with several recent studies in CRC.33–36 Prior reports have indicated lung metastases to shed less ctDNA into the bloodstream compared with other metastatic sites, likely due to indolent tumor biology or sequestration at the site of metastasis.9 Nevertheless, for patients with CRC, lung metastases are reported to have a better prognosis compared with other sites.37,38

Although we present data from a relatively large prospective cohort of patients with stage II-III rectal cancer, our study may continue to benefit with accrual of longer follow-up to evaluate the long-term impact of ctDNA status and dynamics on overall survival (OS). In addition, this study design was observational in nature and exclusively focused on patients who underwent upfront surgery without neoadjuvant therapy, a treatment strategy commonly practiced in Japan and parts of Asia.39 Indeed, according to the Japanese Society for Cancer of the Colon and Rectum Guidelines,40 TNT is not recommended as a standard approach, with the guidelines stating that “TNT is weakly recommended against” for patients with resectable rectal cancer. As such, the majority of patients in Japan undergo surgery first, followed by adjuvant chemotherapy based on pathologic staging. This approach differs from ESMO/NCCN guidelines and from treatment paradigms in Western countries, where neoadjuvant chemoradiotherapy is standard for many patients, particularly those with locally advanced or low rectal tumors.10 Therefore, our cohort represents a typical and nationally relevant treatment pattern in Japan, and does not reflect a subset of patients who were selectively excluded from TNT. While this may limit generalizability to Western populations, it provides important real-world evidence on the utility of postoperative ctDNA in a non-TNT context, which remains highly relevant for many parts of the world where preoperative therapy is not routinely adopted, and addresses an important clinical need in Asian and Southeast Asian populations. Nonetheless, our results provide important insights into the prognostic role of postoperative ctDNA in a homogeneous cohort treated with upfront surgery, and future interventional studies are warranted to evaluate ctDNA dynamics in patients receiving neoadjuvant therapy.
In conclusion, our study demonstrates that postoperative ctDNA is a robust prognostic marker for DFS in rectal cancer patients undergoing upfront surgery. ctDNA positivity during the MRD window indicates a high risk of recurrence and a significant benefit from adjuvant chemotherapy, while ctDNA negativity suggests no benefit from additional treatment. Further prospective, interventional studies are warranted for personalized, ctDNA-guided therapeutic strategies in rectal cancer management.

Supplementary Material