Abstract
Importance:
Observational studies have associated use of aspirin and selective cyclooxygenase inhibitors with decreased recurrence and improved survival in patients with colon cancer. While randomized trials have not shown benefit across all patients, these findings suggest that select subgroups may benefit from their use.
Objective:
To investigate the predictive value of postoperative circulating tumor DNA (ctDNA) for survival outcomes with adjuvant celecoxib alongside conventional chemotherapy.
Design, Setting, and Participants:
Patients were enrolled in Cancer and Leukemia Group B (now Alliance)/Southwest Oncology Group 80702, a phase III randomized trial conducted between 2010–2015 assessing adjuvant celecoxib vs. placebo and 3 vs. 6 months of adjuvant 5-fluorouracil, leucovorin, and oxaliplatin for stage III colon cancer. Patients consented to biospecimen collection and had ctDNA analysis performed.
Exposures:
Postoperative ctDNA positivity was determined using a clinically validated, tumor-informed 16-plex PCR-NGS assay (Signatera™, Natera, Inc.) performed between surgery and initiation of adjuvant therapy.
Main Outcomes:
Primary outcomes were disease-free survival (DFS) and overall survival (OS). Survival by ctDNA status and adjuvant celecoxib use were assessed as part of a post hoc companion study with pre-specified statistical analysis plan.
Results:
Among 940 patients (median follow-up=6.0 years, mean age=60.8 years, 54.7% male, 23.6% with prior low-dose aspirin use), 767 (81.6%) were ctDNA negative and 173 (18.4%) were ctDNA positive. ctDNA positivity was highly prognostic of worse DFS (reference: ctDNA negativity, adjusted hazard ratio [HR]=6.12, 95% confidence interval [CI] 4.66–8.03) and OS (adjusted HR=5.86, 95% CI 4.19–8.19). In patients with ctDNA positivity, celecoxib was associated with improved DFS (adjusted HR=0.61, 95% CI 0.42–0.89) and OS (adjusted HR=0.62, 95% CI 0.40–0.96) compared to placebo. Among patients with ctDNA negativity, celecoxib did not provide survival benefit, though the interaction was not significant (Pinteraction=0.41 and 0.33 for DFS and OS, respectively). When restricting the analysis to a subgroup of patients with high-quality samples, the interaction test was significant (Pinteraction=0.038). These findings persisted when stratifying patients by microsatellite instability status and PIK3CA mutational status.
Conclusion and Relevance:
These results suggest the potential for ctDNA status to inform clinical decision-making among patients who should consider adjuvant celecoxib in addition to conventional chemotherapy.
INTRODUCTION
Colorectal cancer (CRC) remains one of the most prevalent and morbid cancers in the United States.1 Ongoing advancements in diagnosis, surveillance, and treatment have improved survival.2 However, ~20–40% of patients with resected stage III colon cancer recur,3–5 highlighting the need for tailored therapeutic approaches based on risk stratification. Targeted treatments offer promise as adjuncts to conventional chemotherapy, maximizing treatment benefit while limiting toxicity.
There is considerable evidence that aspirin and nonsteroidal anti-inflammatory drugs (NSAIDs) protect against development of colorectal adenomas and subsequent progression to CRC.6,7 Proposed pathways include both cyclooxygenase (COX; prostaglandin-endoperoxide synthase, PTGS)-dependent and COX-independent mechanisms.8 Observational studies have shown that aspirin or NSAID use after CRC diagnosis is associated with lower recurrence risk.9,10
The Cancer and Leukemia Group B (CALGB; now part of the Alliance for Clinical Trials in Oncology) and Southwest Oncology Group Trial 80702 (CALGB/SWOG 80702; NCT01150045) was developed, in part, to prospectively evaluate adjuvant celecoxib.11 This randomized phase III trial, using a 2×2 factorial design, assigned patients with stage III colon cancer after curative resection to receive either three or six months of adjuvant fluorouracil, leucovorin, and oxaliplatin (FOLFOX) and either celecoxib 400mg daily or placebo for three years. These outcomes have been reported.5,11,12 Although the difference in disease-free survival (DFS) between celecoxib and placebo arms did not reach protocol-defined statistical significance, the hazard ratio (HR) of 0.89 suggested a potential benefit in a subset of patients.
Interest has grown in circulating tumor DNA (ctDNA) as a minimally-invasive biomarker in CRC.13,14 Tumor-informed approaches show promise in the postoperative setting for detecting molecular residual disease (MRD) with high sensitivity and specificity.15 However, despite ctDNA’s well-established prognostic value, its role in guiding treatment remains unclear.16 Understanding how ctDNA can be used to inform clinical decision-making remains an area of active interest.
Herein, we examine the prognostic value of ctDNA status for patients with stage III colon cancer and evaluate its predictive utility for identifying patients who may benefit from the addition of adjuvant celecoxib to FOLFOX.
METHODS
Study Design
This was a post hoc analysis with a pre-specified statistical analysis plan (SAP) of patients within CALGB/SWOG 80702 (protocol and SAP provided in Supplement 1 and 2, respectively). The initial trial design has been described.11 Briefly, patients were recruited from National Cancer Trials Network-participating centers across the United States and Canada between 2010–2015. A minimum of 5 years of outcome follow-up was defined within the protocol which was completed in August 2020. Patients were eligible if they underwent curative-intent resection with pathologically confirmed node-positive or N1c disease. Exclusions included baseline NSAID use greater than twice weekly or 325-mg aspirin use more than three times weekly. Low-dose aspirin ≤100-mg daily was permitted. Patients included in this analysis consented separately to correlative science studies involving biospecimen collection, including tumor tissue and peripheral blood. A CONSORT diagram is provided in eFigure 1. Institutional Review Board approval was obtained at all participating centers, and participants provided written consent.
ctDNA Sequencing and Analysis
In patients with adequate biospecimens, a retrospective ctDNA assessment was performed on plasma samples using a clinically validated, tumor-informed multiplex-PCR NGS assay (Signatera™, Natera, Inc.). Plasma (median 2mL, range 0.6–5.3mL) used for ctDNA testing was collected at a single timepoint after surgery and prior to initiation of adjuvant FOLFOX. Patients were excluded if the date of blood draw was unknown, prior to the date of surgery, or greater than 12 weeks after the date of surgery. Tumor tissue was obtained from resection specimens.
Whole exome sequencing (WES) was previously performed using formalin-fixed paraffin-embedded (FFPE) tumor specimens and matched normal tissue or peripheral blood as a source of germline DNA.17 FASTQ files from the WES analysis were then analyzed to design a multiplex PCR assay targeting 16 patient-specific, somatic single-nucleotide variants.18 Samples with at least two tumor-specific variants above the ctDNA calling confidence threshold were defined as ctDNA positive. ctDNA concentration was reported in mean tumor molecules per mL (MTM/mL).15
Endpoints
The study endpoints were DFS, defined as time from randomization to disease recurrence or any-cause death, and overall survival (OS), defined as time from randomization to any-cause death. Recurrence was monitored by serial history, physical examination, and carcinoembryonic antigen levels every 3 months for 3 years following randomization then every 6 months for 6 years after randomization, or until disease recurrence, whichever came first. All patients also had surveillance imaging of the chest, abdomen, and pelvis every 6 months for 3 years, then annually for 3 years, or until disease recurrence.
Patient Factors
Baseline demographic factors, including age, sex, body mass index (BMI), race, and ethnicity, as well as clinical factors, including Eastern Cooperative Oncology Group Performance Status (ECOG PS), low-dose aspirin use, T stage, N stage, tumor location, and days from surgery to ctDNA blood draw, were obtained via a combination of patient self-report and chart review. Assigned oral agent (celecoxib vs. placebo) and duration of FOLFOX were determined at the time of randomization. Tumor genomic profiles, including KRAS, BRAF, and PIK3CA mutation status, along with microsatellite instability (MSI) status, were determined via WES. KRAS and BRAF status were assigned based on expert-curated annotation of individual variants within each gene and predicted functional effect with respect to activation of RAS/MAPK pathway signaling. Criteria for PIK3CA mutation and MSI status determination have been previously described.17 PIK3CA wildtype tumors and those harboring only variants of uncertain functional significance were classified together.
Statistical Analysis
Baseline factors were compared by ctDNA status. Kaplan-Meier estimates were used to evaluate time-to-event outcomes, and log-rank tests were used to compare survival outcomes between patients according to ctDNA status and assigned oral agent. Multivariable Cox proportional hazard models for DFS and OS were constructed to compare treatment groups among both patients who were ctDNA positive and negative, adjusting for potential confounders determined a priori based on expert knowledge. Patients with missing covariates (10/940 in the analyzed cohort) were excluded in adjusted analyses. Interaction effects between ctDNA status and assigned oral agent were determined using multiplicative models.
Pre-planned subgroup analyses by MSI status (MSI high [MSI-H] vs. non-MSI-H) and PIK3CA status (wildtype vs. mutant) were performed given known interactions with COX-inhibition.17,19 Quantitative ctDNA effects were explored among patients who were ctDNA positive.
Data collection and statistical analyses were conducted by the Alliance Statistics and Data Management Center (SAS, version 9.4; SAS Institute, Inc; R, version 4.1.2; R Foundation for Statistical Computing). Statistical significance was defined as 2-sided P<0.05, except for interaction testing, where P<0.10 was deemed statistically significant among subgroups. Hazard ratios (HRs) were reported with 95% confidence intervals (CIs).
RESULTS
Among 2526 patients enrolled in CALGB/SWOG 80702, 940 with valid ctDNA status were included in this study (eFigure 1). These patients had a median follow-up time of 6.0 years (95% CI: 6.0–6.0). Overall, 173 patients (18.4%) were ctDNA positive, consistent with prior reports.14,20 Baseline characteristics of patients with ctDNA results were similar to those of the overall trial cohort (eTable 1). These characteristics, stratified by ctDNA status, are presented in Table 1. Patients with ctDNA positivity were more likely to be male, T4 or N2 stage, possess a BRAF mutation, and be KRAS wildtype. Notably, a slightly higher proportion of patients were randomized to celecoxib among those with ctDNA positivity vs. ctDNA negativity (57.2% vs. 48.9%, respectively).
Table 1.
Clinical and Pathological Characteristics of Stage III Colon Cancer Patients, Stratified by Postoperative ctDNA Status (N=940).
| ctDNA status | ||||
|---|---|---|---|---|
| Characteristica | Total (N=940) |
Negative (N=767) |
Positive (N=173) |
P-value |
| Age, mean years (SD) | 60.9 (10.8) | 60.8 (10.9) | 61.2 (10.3) | 0.699 |
| Sex | 0.003 | |||
| Male | 514 (54.7%) | 402 (52.4%) | 112 (64.7%) | |
| Female | 426 (45.3%) | 365 (47.6%) | 61 (35.3%) | |
| BMI, kg/m2 | 0.415 | |||
| <18.5 | 19 (2.0%) | 16 (2.1%) | 3 (1.7%) | |
| 18.5 – 24.9 | 260 (27.7%) | 219 (28.6%) | 41 (23.7%) | |
| 25.0 – 29.9 | 338 (36.0%) | 267 (34.8%) | 71 (41.0%) | |
| ≥30 | 323 (34.4%) | 265 (34.6%) | 58 (33.5%) | |
| Aspirin use | 0.107 | |||
| No | 718 (76.4%) | 594 (77.4%) | 124 (71.7%) | |
| Yes | 222 (23.6%) | 173 (22.6%) | 49 (28.3%) | |
| T stage | 0.001 | |||
| T1 or T2 | 148 (15.8%) | 131 (17.1%) | 17 (9.9%) | |
| T3 | 662 (70.7%) | 543 (71.0%) | 119 (69.2%) | |
| T4 | 127 (13.6%) | 91 (11.9%) | 36 (20.9%) | |
| Missing | 3 | 2 | 1 | |
| N stage | <0.001 | |||
| N1 | 667 (71.0%) | 578 (75.4%) | 89 (51.4%) | |
| N2 | 273 (29.0%) | 189 (24.6%) | 84 (48.6%) | |
| Missing | ||||
| Race | 0.503 | |||
| White | 774 (82.3%) | 628 (81.9%) | 146 (84.4%) | |
| Black or African American | 105 (11.2%) | 85 (11.1%) | 20 (11.6%) | |
| Asian | 37 (3.9%) | 32 (4.2%) | 5 (2.9%) | |
| All others or not reported | 24 (2.6%) | 22 (2.9%) | 2 (1.2%) | |
| Ethnicity | 0.128 | |||
| Hispanic or Latino | 47 (5.0%) | 41 (5.3%) | 6 (3.5%) | |
| Not Hispanic or Latino | 880 (93.6%) | 713 (93.0%) | 167 (96.5%) | |
| Unknown/Not reported | 13 (1.4%) | 13 (1.7%) | 0 (0.0%) | |
| ECOG Performance Status | 0.103 | |||
| 0 | 662 (70.4%) | 549 (71.6%) | 113 (65.3%) | |
| 1–2 | 278 (29.6%) | 218 (28.4%) | 60 (34.7%) | |
| Tumor location | 0.990 | |||
| Left | 434 (46.5%) | 353 (46.4%) | 81 (46.8%) | |
| Right / transverse | 493 (52.8%) | 402 (52.9%) | 91 (52.6%) | |
| Multiple | 6 (0.6%) | 5 (0.7%) | 1 (0.6%) | |
| Missing | 7 | 7 | 0 | |
| KRAS status | 0.064 | |||
| Wildtype | 532 (56.6%) | 445 (58.0%) | 87 (50.3%) | |
| Mutant | 408 (43.4%) | 322 (42.0%) | 86 (49.7%) | |
| BRAF status | 0.053 | |||
| Wildtype | 810 (86.2%) | 653 (85.1%) | 157 (90.8%) | |
| Mutant | 130 (13.8%) | 114 (14.9%) | 16 (9.2%) | |
| MSI status | 0.287 | |||
| Non-MSI-High | 843 (89.7%) | 684 (89.2%) | 159 (91.9%) | |
| MSI-High | 97 (10.3%) | 83 (10.8%) | 14 (8.1%) | |
| PIK3CA status | 0.232 | |||
| Wildtype | 171 (23.2%) | 31 (18.9%) | 202 (22.4%) | |
| Mutant | 566 (76.8%) | 133 (81.1%) | 699 (77.6%) | |
| Missing | 30 | 9 | 39 | |
| Assigned FOLFOX Duration | 0.498 | |||
| 3 months | 478 (50.9%) | 386 (50.3%) | 92 (53.2%) | |
| 6 months | 462 (49.1%) | 381 (49.7%) | 81 (46.8%) | |
| Assigned Oral Agent | 0.048 | |||
| Celecoxib | 474 (50.4%) | 375 (48.9%) | 99 (57.2%) | |
| Placebo | 466 (49.6%) | 392 (51.1%) | 74 (42.8%) | |
| Days from surgery to ctDNA blood draw, mean (SD) | 44.4 (10.7) | 44.2 (10.8) | 45.3 (10.6) | 0.202 |
All data reported as no. (%), unless otherwise specified.
Abbreviations: ctDNA: circulating tumor DNA; SD: standard deviation; BMI: body mass index; ECOG: Eastern Cooperative Oncology Group; MSI: microsatellite instability; FOLFOX: folinic acid, fluorouracil, and oxaliplatin.
ctDNA Status and Celecoxib Use
Across all patients, ctDNA status was highly prognostic of survival. Compared to patients who were ctDNA negative, patients who were ctDNA positive had shorter DFS (adjusted HR=6.12, 95% CI 4.66–8.03) and OS (adjusted HR=5.86, 95% CI 4.19–8.19). The point estimates for 3-year DFS for patients with ctDNA negativity and positivity were 86.5% and 33.7%, respectively, and for 5-year OS, 91.5% and 52.6%, respectively (Figure 1). Subgroup analyses using standard clinicopathologic features, including T4 and N2 disease, did not demonstrate significant differences in the association between ctDNA status and DFS (eFigure 2).
Figure 1.
Disease-Free Survival (DFS) and Overall Survival (OS) Estimates, by Postoperative Circulating Tumor DNA (ctDNA) status.
[A] DFS
[B] OS
Disease-free survival and overall survival, stratified by postoperative ctDNA status. Data cutoff was 6 years after time of randomization.
Among patients who were ctDNA positive, celecoxib use was associated with improved DFS compared to placebo, with an estimated 3-year DFS of 41.0% vs. 22.6%, respectively (adjusted HR=0.61, 95% CI 0.42–0.89; Figure 2A, Table 2). In contrast, among patients who were ctDNA negative, there was no statistically significant difference in DFS (estimated 3-year DFS, 87.4% and 85.6%, respectively; adjusted HR=0.76, 95% CI 0.53–1.09). However, the interaction test comparing the effect of celecoxib vs. placebo by ctDNA status was not statistically significant (P=0.41).
Figure 2.
Disease-Free Survival (DFS) and Overall Survival (OS) Estimates, by Assigned Oral Agent (Celecoxib vs. Placebo) and Postoperative Circulating Tumor DNA (ctDNA) Status.
[A] DFS
[B] OS
Disease-free survival and overall survival, stratified by assigned oral agent (celecoxib or placebo) and postoperative ctDNA status in 2×2 fashion. Data cutoff was 6 years after time of randomization.
Table 2.
Survival Outcomes Based on Use of Celecoxib vs. Placebo and Postoperative ctDNA Status.
| Events / At Risk | Adjusted HR (95% CI)a | Adjusted P-valuea | Pinteractiona | |
|---|---|---|---|---|
| Disease-Free Survival | ||||
| ctDNA Negative | 0.413 | |||
| Placebo | 72 / 387 | Ref | ||
| Celecoxib | 58 / 371 | 0.76 (0.53, 1.09) | 0.131 | |
| ctDNA Positive | ||||
| Placebo | 57 / 74 | Ref | ||
| Celecoxib | 60 / 98 | 0.61 (0.42, 0.89) | 0.011 | |
| Overall Survival | ||||
| ctDNA Negative | 0.330 | |||
| Placebo | 40 / 387 | Ref | ||
| Celecoxib | 36 / 371 | 0.85 (0.54, 1.36) | 0.499 | |
| ctDNA Positive | ||||
| Placebo | 44 / 74 | Ref | ||
| Celecoxib | 40 / 98 | 0.62 (0.40, 0.96) | 0.034 | |
Adjusting for age, sex, aspirin usage, T stage, N stage, ECOG Performance Status, tumor location, BRAF status, KRAS status, MSI status, and days from surgery to ctDNA blood draw.
HR, 95% CI, and P-value were calculated from multivariable Cox proportional hazard models, adjusted for the covariates above. P value for interaction between ctDNA status and oral agent (celecoxib vs. placebo) was derived using the likelihood ratio test.
Abbreviations: HR, hazard ratio; CI, confidence interval; ctDNA, circulating tumor DNA.
Similarly, celecoxib use, when compared to placebo, correlated with improved OS among patients who were ctDNA positive, with an estimated 5-year OS of 61.6% and 39.9%, respectively (adjusted HR=0.62, 95% CI 0.40–0.96), but not among patients who were ctDNA negative (91.8% and 91.3%, respectively; adjusted HR=0.85, 95% CI 0.54–1.36). Again, the interaction effect was not statistically significant (P=0.33).
Subgroup Analysis
Selected subgroup analyses were performed to assess potential effect modifiers of ctDNA status and survival from celecoxib use. Stratification of patients by MSI status showed that among patients with non-MSI-H tumors, ctDNA positivity was associated with an estimated 3-year DFS of 39.7% in the celecoxib arm and 22.3% in the placebo arm (adjusted HR=0.63, 95% CI 0.42–0.93; Figure 3A,B, eTable 2). Among patients who were ctDNA negative with non-MSI-H tumors, there was no significant survival benefit with celecoxib as compared to placebo (adjusted HR=0.76, 95% CI 0.52–1.11). Similarly, in patients with MSI-H tumors, celecoxib was associated with improved DFS for those who were ctDNA positive (adjusted HR=0.05, 95% CI 0.00–0.59), but not ctDNA negative (adjusted HR=1.25, 95% CI 0.37–4.23).
Figure 3.
Disease-Free Survival (DFS) by Celecoxib vs. Placebo and Circulating Tumor DNA (ctDNA) Status Among Select Subgroups of Interest.
[A] Non-MSI-High
[B] MSI-High
[C] PIK3CA Wildtype
[D] PIK3CA Mutant
Disease free survival, stratified by assigned oral agent (celecoxib or placebo) and postoperative ctDNA status in 2×2 fashion. Shown are subgroups of patients with A: non-microsatellite instability high tumors (non-MSI-High), B: microsatellite instability high tumors (MSI-High), C: PIK3CA wildtype tumors, and D: PIK3CA mutant tumors. Data cutoff was 6 years after time of randomization.
We further examined the relationship between ctDNA and celecoxib according to primary tumor PIK3CA mutation status, as activating or gain-of-function PIK3CA mutations are known to predict survival with adjuvant celecoxib.17 These were present in 202/940 (21.5%) patients. Among those with PIK3CA wildtype tumors, celecoxib was associated with improved DFS among patients who were ctDNA positive (adjusted HR=0.64, 95% CI 0.42–0.98; Figure 3C,D, eTable 2) but not ctDNA negative (adjusted HR=0.80, 95% CI 0.55–1.18). Similarly, among patients with PIK3CA-activated tumors, celecoxib use was also associated with improved DFS compared to placebo, with an estimated 3-year DFS of 44.4% vs. 15.4%, respectively (adjusted HR=0.19, 95% CI 0.06–0.58). Conversely, patients who were ctDNA negative with PIK3CA-activated tumors did not derive a DFS benefit from celecoxib (adjusted HR=0.85, 95% CI 0.33–2.24).
Sensitivity Analysis
Given variability in tumor-derived ctDNA shedding and potential differences in archival plasma specimen quality, we conducted sensitivity analyses based on additional ctDNA metrics. We first evaluated the significance of ctDNA concentration among patients who were ctDNA positive, reasoning that lower ctDNA concentration might correlate with lower sensitivity for the detection of micrometastatic disease. We evaluated receiver operating characteristic curves to identify a ctDNA concentration cutoff of 1 MTM/mL, which optimizes discrimination for identifying patients at risk of mortality within the present cohort and is consistent with findings from Signatera™ internal testing (data not published). Among patients who were ctDNA positive, 60% had ctDNA concentration ≥1 MTM/mL. Compared to 0<MTM/mL<1, MTM/mL≥1 was associated with worsened DFS (HR=2.10, 95% CI 1.43–3.10; eFigure 3A), and OS (HR=2.11, 95% CI 1.33–3.36; Figure 3B). Furthermore, improved DFS in patients who received celecoxib vs. placebo was observed for high ctDNA concentration (MTM/mL≥1; HR=0.57, 95% CI 0.37–0.90) but not low ctDNA concentration (0<MTM/mL<1; HR=0.54, 95% CI 0.28–1.03) subgroups (eFigure 3C).
To account for variable quality in plasma samples, we restricted analyses to a subgroup of patients that had high-quality samples available during ctDNA testing, with optimal plasma volume and cell-free DNA extraction (defined as ≥11.25ng and 5–15ng/mL per sample; thresholds based on internal Natera testing standards, data not published). This “optimal” quality control group consisted of 65.7% (618/940) of the total cohort, of whom 20.0% were ctDNA positive. Analysis of this subgroup showed a stronger DFS benefit for adjuvant celecoxib vs. placebo in patients who were ctDNA positive (adjusted HR=0.49, 95% CI 0.31–0.78; eTable 3, eFigure 4) as compared to the overall cohort, but not among patients who were ctDNA negative (adjusted HR=0.97, 95% CI 0.62–1.50). Notably, in this high-quality subgroup, the interaction test comparing the effect of celecoxib vs. placebo between patients who were ctDNA positive and negative was significant (P=0.038).
DISCUSSION
In this post hoc analysis of a randomized study of stage III colon cancer, we found that patients with postoperative ctDNA positivity experienced improved DFS and OS with adjuvant celecoxib alongside conventional chemotherapy, whereas those who were ctDNA negative did not benefit. These findings persisted independent of baseline demographic, pathologic, and molecular factors, including MSI and PIK3CA mutation status. While the interaction between ctDNA status and treatment group did not reach statistical significance in the full cohort, it did in a subanalysis restricted to high-quality ctDNA results. Furthermore, celecoxib use yielded an 18.4% improvement in estimated 3-year DFS in patients who were ctDNA positive, compared to 1.8% in those who were ctDNA negative, suggesting a potentially clinically meaningful effect. These findings are the first to implicate MRD, as assessed by ctDNA, as a potential predictor of therapeutic response to adjuvant COX-inhibition in CRC.
Previous studies have reported the negative prognostic role of MRD in locally advanced CRC.14,21,22 Our findings are aligned, demonstrating a ~6-fold higher risk of recurrence and death among patients who were ctDNA positive vs. negative. This underscores ctDNA’s value as a prognostic marker for disease recurrence and survival and highlights its ability to identify high-risk patients in the adjuvant setting.
Given strong correlative findings, efforts to leverage ctDNA to guide clinical decision-making are ongoing,13,14,20,23 with most studies focusing on intensification or de-escalation of adjuvant chemotherapy. Analyses examining ctDNA status and CALGB/SWOG 80702’s second randomization factor of 3 vs. 6 months of adjuvant FOLFOX are underway and will be reported separately. To our knowledge, this study represents the first evidence supporting MRD as a determinant for adjunctive strategies, specifically COX-inhibition, in the adjuvant management of CRC, thereby expanding its utility beyond guidance of conventional chemotherapies.
Recent randomized studies of adjuvant COX-inhibition among all-comers with CRC, including CALGB/SWOG 80702, failed to achieve statistical significance. However, point estimates suggest that a subset of patients may benefit.11,24,25 Gain-of-function mutations in PIK3CA occur in 15–20% of CRC and promote tumor growth via COX-dependent PI3 kinase/AKT pathways.26,27 Previous work from this cohort showed that patients with PIK3CA-mutated tumors may derive benefit from adjuvant celecoxib.17 The recently reported ALASCCA trial also demonstrated benefit from adjuvant aspirin in this group, leading to a change in the standard of care for this subset of patients.16,28 We performed a subgroup analysis of patients with PIK3CA wildtype tumors and found that the predictive value of ctDNA remained and was nearly undiminished in effect size. This suggests that PIK3CA status and ctDNA are independent predictors of celecoxib response, and that MRD may signal alternative oncogenic pathways susceptible to therapeutic COX-inhibition.
These findings, when interpreted alongside ALASCCA, pose important questions regarding the optimal integration of COX-inhibition in adjuvant therapy for locally advanced colon cancer. The choice of aspirin versus selective COX-2 inhibitor in this setting remains uncertain. Though COX-2-dependent antineoplastic pathways have previously been implicated (hence the use of celecoxib in CALGB/SWOG 80702),7,10,11 ALASCCA and other studies have demonstrated meaningful benefit using aspirin in patients with tumors harboring PIK3CA activating mutations. However, whether the benefit between celecoxib and aspirin are similar in this population is unknown. Further, this study is the first to evaluate the impact of adjuvant COX-inhibition based on ctDNA status. Whether the observed benefit in patients who are ctDNA positive would be seen with aspirin is unknown and would require additional studies, either a prospective clinical trial or substudy using currently ongoing adjuvant aspirin therapy trials with blood collections.
There are several mechanisms that may explain our findings. Tumorigenesis is strongly shaped by the tumor microenvironment (TME), which contains immune cells that can both facilitate and suppress tumor growth.29–32 MRD may reflect the aggregate TME that serves as a key target of celecoxib’s anti-inflammatory effects. This is corroborated by our findings of a quantitative relationship between ctDNA concentration and survival, as well as previous works linking systemic inflammation after curative resection with poor prognosis in CRC.33,34 COX-inhibition also suppresses angiogenesis and proliferation,8,35 processes that are crucial for the outgrowth of residual micrometastatic disease reflected by ctDNA. Finally, PTGS2 is known to be expressed in tumor cells, and regular aspirin use has been shown to reduce the risk of developing CRCs that overexpress PTGS2.7 It is therefore possible that ctDNA status may indicate tumors directly responsive to COX-inhibition. Further research is needed to elucidate these mechanistic relationships.
Limitations
This study has several limitations. First, it is a post hoc analysis of existing clinical trial data and may be confounded by lack of randomization within ctDNA subgroups. However, we accounted for many demographic, clinical, and molecular factors that were considered confounders,11 and demonstrated that these were relatively balanced between patients who were ctDNA positive and negative. Second, ctDNA was only measured at one timepoint, prior to initiation of therapy, preventing assessment of ctDNA clearance with adjuvant therapy, as has been previously studied.14,20 However, we did utilize a tumor-informed ctDNA assay, which demonstrates higher analytical sensitivity and specificity than tumor-agnostic approaches.36 Additionally, the majority (65.7%) of the cohort had optimal quality samples. Lastly, clinical trial enrollees may not represent the broader disease population, which could limit the generalizability of our results. Although CALGB/SWOG 80702 was a multicenter study that recruited from both community and academic centers across North America, baseline demographics from the overall cohort and ctDNA subset evaluated in this study demonstrated an overrepresentation of patients who were White, non-Hispanic, and used low-dose aspirin.
CONCLUSIONS
Postoperative ctDNA positivity is strongly associated with decreased survival in stage III colon cancer, and adjuvant celecoxib conveys a survival benefit among those who were ctDNA positive. ctDNA may help inform decision making by identifying a subset of patients that benefits most from adjuvant COX-inhibition alongside conventional chemotherapy and offer opportunities for personalized approaches in the treatment of CRC.
Supplementary Material
KEY POINTS.
Question:
Does postoperative circulating tumor DNA (ctDNA) status predict survival among patients with stage III colon cancer treated with adjuvant celecoxib?
Findings:
In this prospective cohort analysis, a clinically validated, tumor-informed assay was used to determine ctDNA status in 940 patients with stage III colon cancer. ctDNA negativity was highly associated with improved survival. Patients with ctDNA positivity demonstrated a survival benefit with adjuvant celecoxib, whereas patients with ctDNA negativity did not.
Meaning:
These findings underscore the prognostic value of ctDNA and suggest it may help identify a subset of patients that benefit from adjuvant celecoxib alongside conventional chemotherapy.
ACKNOLWEDGEMENTS
Funding/Support:
This manuscript is the result of funding in whole or in part by the National Cancer Institute of the National Institutes of Health (NIH) under Award Numbers U10 CA180821, U10 CA180882, U24 CA196171 (to the Alliance for Clinical Trials in Oncology); UG1CA189830, UG1CA233180, UG1CA233339, UG1CA233290, U10CA180863 and CCS 707213 (CCTG); U10CA180794, U10CA180820 (ECOG-ACRIN); U10CA180868 (NRG Oncology); U10CA180888 and UG1CA233163 (SWOG). Additional support provided to the Alliance for Clinical Trials in Oncology Foundation can be found on their website: https://pubs.alliancefound.org/acknowledgments. It is subject to the NIH Public Access Policy.
Through acceptance of this federal funding, NIH has been given a right to make the Author Accepted Manuscript publicly available in PubMed Central upon the Official Date of Publication, as defined by NIH.
In addition, the work was supported by Karen Guo Colon Cancer Fund, Project P, Douglas Gray Woodruff Chair in Colorectal Research Fund, Stone Research Fund, NIH T32CA009001 (Zhang), and in part by the American Cancer Society Clinical Research Professor Award CRP-24-1185864-01-PROF (Ogino). Also supported in part by Pfizer and Natera through the Alliance for Clinical Trials in Oncology Foundation, as part of a correlative science proposal (CSC0218) approved by the National Clinical Trials Network Core Correlative Sciences Committee. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Role of the Funder/Sponsor:
Nonfederal sponsors did not contribute to design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication. The National Cancer Institute provided grant and administrative support to the trial, was involved in the design of the study and in the review and final approval of the manuscript.
Conflict of Interest Disclosures:
GQZ, JAM, TT, LP, CM, JPV, MZ, YT, PaK, FC, PhK, SK, SO, and AFS report no relevant conflicts of interest.
QS reports consulting/advisory role at Yiviva Inc, Regeneron Pharmaceuticals, Inc., Hoosier Cancer Research Network, Kronos Bio, Mirati Therapeutics Inc. Genmab US, Inc., HopeAI, Inc. BMS, AbbVie, BeOne Medicines USA, Inc and Exelixis, and institutional research support from Regeneron, BMS, Roche/Genentech, Janssen, Novartis, and MPAACT.
AS reports consulting/advisory role at Natera, Merus, Takeda, Ipson, Guardant, Pfizer, Regeneron/Sanofi, Catalyst Pharma, and KLJ associates, institutional research support from Hutchison MediPharma/Takeda, Merck, Verastem Oncology, Turning Point Therapeutics, Gritstone, Bolt Therapeutics, BMS, Pfizer, Astellas, Oncologie, Macogenics, Seattle Genetics, Amgen, Daiichi, Lilly, Jacobio, AstraZeneca, Genentech, Jazz Pharma, and Agenus, and travel/speaking fees from Cure CRC Summit, Great Debates, OncLive, OSCO, ASCO, ACPMP, Cholangiocarcinoma Summit, Cholangiocarcinoma Foundation, ASCO Advantage, and Takeda.
BT reports research support from Abbvie, Adaptimmune, AstraZeneca, Bristol Myers Squibb, Exelixis, Genentech, and Tvardi.
EMO reports research support (institution) from Genentech/Roche, BioNTech, AstraZeneca, Arcus, Elicio, Parker Institute, NIH/NCI, Digestive Care, Break Through Cancer, Agenus, Amgen, and Revolution Medicines, consulting/DSMB fees (uncompensated) from Arcus, Amgen, AstraZeneca, Ability Pharma, Alligator BioSciences, Pfizer, Agenus, BioNTech, Ipsen, Ikena, Merck, Moma Therapeutics, Novartis, Astellas, BMS, Revolution Medicines, Regeneron, and Tango, consulting/DSMB fees (compensated) from Leap Therapeutics, travel fees from BioNTech and Arcus, and support from Abbvie (spouse).
MG reports research support from Janssen and Sunbird Bio, and consulting fees from Nerviano Medical Sciences.
AJ, SS, and AA are full-time employees of Natera with stock or options to own stock.
JAN reports research support from Natera, consulting fees from Leica Biosystems, and speaking fees from Bristol Myers Squibb.
Footnotes
Trial Registration: ClinicalTrials.gov Identifier: NCT01150045
REFERENCES
- 1.Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA A Cancer J Clinicians. 2024;74(1):12–49. doi: 10.3322/caac.21820 [DOI] [Google Scholar]
- 2.Brouwer NPM, Bos ACRK, Lemmens VEPP, et al. An overview of 25 years of incidence, treatment and outcome of colorectal cancer patients. Intl Journal of Cancer. 2018;143(11):2758–2766. doi: 10.1002/ijc.31785 [DOI] [Google Scholar]
- 3.Tsikitis VL, Larson DW, Huebner M, Lohse CM, Thompson PA. Predictors of recurrence free survival for patients with stage II and III colon cancer. BMC Cancer. 2014;14(1):336. doi: 10.1186/1471-2407-14-336 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.André T, Boni C, Navarro M, et al. Improved Overall Survival With Oxaliplatin, Fluorouracil, and Leucovorin As Adjuvant Treatment in Stage II or III Colon Cancer in the MOSAIC Trial. JCO. 2009;27(19):3109–3116. doi: 10.1200/JCO.2008.20.6771 [DOI] [Google Scholar]
- 5.André T, Meyerhardt J, Iveson T, et al. Effect of duration of adjuvant chemotherapy for patients with stage III colon cancer (IDEA collaboration): final results from a prospective, pooled analysis of six randomised, phase 3 trials. The Lancet Oncology. 2020;21(12):1620–1629. doi: 10.1016/S1470-2045(20)30527-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Chan AT, Arber N, Burn J, et al. Aspirin in the Chemoprevention of Colorectal Neoplasia: An Overview. Cancer Prevention Research. 2012;5(2):164–178. doi: 10.1158/1940-6207.CAPR-11-0391 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Chan AT, Ogino S, Fuchs CS. Aspirin and the Risk of Colorectal Cancer in Relation to the Expression of COX-2. N Engl J Med. 2007;356(21):2131–2142. doi: 10.1056/NEJMoa067208 [DOI] [PubMed] [Google Scholar]
- 8.Jänne PA, Mayer RJ. Chemoprevention of Colorectal Cancer. N Engl J Med. 2000;342(26):1960–1968. doi: 10.1056/NEJM200006293422606 [DOI] [PubMed] [Google Scholar]
- 9.Ng K, Meyerhardt JA, Chan AT, et al. Aspirin and COX-2 Inhibitor Use in Patients With Stage III Colon Cancer. JNCI: Journal of the National Cancer Institute. 2015;107(1). doi: 10.1093/jnci/dju345 [DOI] [Google Scholar]
- 10.Chan AT, Ogino S, Fuchs CS. Aspirin Use and Survival After Diagnosis of Colorectal Cancer. JAMA. 2009;302(6):649–658. doi: 10.1001/jama.2009.1112 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Meyerhardt JA, Shi Q, Fuchs CS, et al. Effect of Celecoxib vs Placebo Added to Standard Adjuvant Therapy on Disease-Free Survival Among Patients With Stage III Colon Cancer: The CALGB/SWOG 80702 (Alliance) Randomized Clinical Trial. JAMA. 2021;325(13):1277. doi: 10.1001/jama.2021.2454 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Grothey A, Sobrero AF, Shields AF, et al. Duration of Adjuvant Chemotherapy for Stage III Colon Cancer. N Engl J Med. 2018;378(13):1177–1188. doi: 10.1056/NEJMoa1713709 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Tie J, Cohen JD, Lahouel K, et al. Circulating Tumor DNA Analysis Guiding Adjuvant Therapy in Stage II Colon Cancer. N Engl J Med. 2022;386(24):2261–2272. doi: 10.1056/NEJMoa2200075 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Nakamura Y, Watanabe J, Akazawa N, et al. ctDNA-based molecular residual disease and survival in resectable colorectal cancer. Nat Med. 2024;30(11):3272–3283. doi: 10.1038/s41591-024-03254-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kalashnikova E, Aushev VN, Malashevich AK, et al. Correlation between variant allele frequency and mean tumor molecules with tumor burden in patients with solid tumors. Molecular Oncology. 2024;18(11):2649–2657. doi: 10.1002/1878-0261.13557 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.National Comprehensive Cancer Network. NCCN Clincal Practice Guidelines in Oncology. Colon Cancer. https://www.nccn.org/professionals/physician_gls/pdf/colon.pdf [Google Scholar]
- 17.Nowak JA, Twombly T, Ma C, et al. Improved Survival With Adjuvant Cyclooxygenase 2 Inhibition in PIK3CA -Activated Stage III Colon Cancer: CALGB/SWOG 80702 (Alliance). JCO. 2024;42(24):2853–2859. doi: 10.1200/JCO.23.01680 [DOI] [Google Scholar]
- 18.Reinert T, Henriksen TV, Christensen E, et al. Analysis of Plasma Cell-Free DNA by Ultradeep Sequencing in Patients With Stages I to III Colorectal Cancer. JAMA Oncol. 2019;5(8):1124. doi: 10.1001/jamaoncol.2019.0528 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Karnes WE, Shattuck-Brandt R, Burgart LJ, et al. Reduced COX-2 protein in colorectal cancer with defective mismatch repair. Cancer Res. 1998;58(23):5473–5477. [PubMed] [Google Scholar]
- 20.Kasi PM, Aushev VN, Ensor J, et al. Circulating tumor DNA (ctDNA) for informing adjuvant chemotherapy (ACT) in stage II/III colorectal cancer (CRC): Interim analysis of BESPOKE CRC study. JCO. 2024;42(3_suppl):9–9. doi: 10.1200/JCO.2024.42.3_suppl.9 [DOI] [Google Scholar]
- 21.Tie J, Cohen JD, Wang Y, et al. Circulating Tumor DNA Analyses as Markers of Recurrence Risk and Benefit of Adjuvant Therapy for Stage III Colon Cancer. JAMA Oncol. 2019;5(12):1710. doi: 10.1001/jamaoncol.2019.3616 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Henriksen TVV, Tarazona N, Frydendahl A, et al. Serial circulating tumor DNA analysis to assess recurrence risk, benefit of adjuvant therapy, growth rate and early relapse detection in stage III colorectal cancer patients. JCO. 2021;39(15_suppl):3540–3540. doi: 10.1200/JCO.2021.39.15_suppl.3540 [DOI] [Google Scholar]
- 23.Sahin I, Lin Y, Yothers G, et al. Minimal Residual Disease–Directed Adjuvant Therapy for Patients With Early-Stage Colon Cancer: CIRCULATE-US. ONCOLOGY. 2022;(3610):604–608. doi: 10.46883/2022.25920976 [DOI] [PubMed] [Google Scholar]
- 24.Midgley RS, McConkey CC, Johnstone EC, et al. Phase III Randomized Trial Assessing Rofecoxib in the Adjuvant Setting of Colorectal Cancer: Final Results of the VICTOR Trial. JCO. 2010;28(30):4575–4580. doi: 10.1200/JCO.2010.29.6244 [DOI] [Google Scholar]
- 25.Chia JWK, Segelov E, Deng Y, et al. Aspirin after completion of standard adjuvant therapy for colorectal cancer (ASCOLT): an international, multicentre, phase 3, randomised, double-blind, placebo-controlled trial. Lancet Gastroenterol Hepatol. 2025;10(3):198–209. doi: 10.1016/S2468-1253(24)00387-X [DOI] [PubMed] [Google Scholar]
- 26.Ogino S, Nosho K, Kirkner GJ, et al. PIK3CA Mutation Is Associated With Poor Prognosis Among Patients With Curatively Resected Colon Cancer. JCO. 2009;27(9):1477–1484. doi: 10.1200/JCO.2008.18.6544 [DOI] [Google Scholar]
- 27.Manning BD, Cantley LC. AKT/PKB Signaling: Navigating Downstream. Cell. 2007;129(7):1261–1274. doi: 10.1016/j.cell.2007.06.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Martling A, Lindberg J, Hed Myrberg I, et al. Low-dose aspirin to reduce recurrence rate in colorectal cancer patients with PI3K pathway alterations: 3-year results from a randomized placebo-controlled trial. JCO. 2025;43(4_suppl). doi: 10.1200/JCO.2025.43.4_suppl.LBA125 [DOI] [Google Scholar]
- 29.Briukhovetska D, Dörr J, Endres S, Libby P, Dinarello CA, Kobold S. Interleukins in cancer: from biology to therapy. Nature Reviews Cancer. 2021;21(8):481–499. doi: 10.1038/s41568-021-00363-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Haruki K, Kosumi K, Li P, et al. An integrated analysis of lymphocytic reaction, tumour molecular characteristics and patient survival in colorectal cancer. Br J Cancer. 2020;122(9):1367–1377. doi: 10.1038/s41416-020-0780-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Väyrynen JP, Haruki K, Väyrynen SA, et al. Prognostic significance of myeloid immune cells and their spatial distribution in the colorectal cancer microenvironment. J Immunother Cancer. 2021;9(4):e002297. doi: 10.1136/jitc-2020-002297 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Väyrynen JP, Haruki K, Lau MC, et al. The Prognostic Role of Macrophage Polarization in the Colorectal Cancer Microenvironment. Cancer Immunol Res. 2021;9(1):8–19. doi: 10.1158/2326-6066.CIR-20-0527 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Cheng E, Shi Q, Shields AF, et al. Association of Inflammatory Biomarkers With Survival Among Patients With Stage III Colon Cancer. JAMA Oncol. 2023;9(3):404. doi: 10.1001/jamaoncol.2022.6911 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Chen ZY, Raghav K, Lieu CH, et al. Cytokine profile and prognostic significance of high neutrophil-lymphocyte ratio in colorectal cancer. Br J Cancer. 2015;112(6):1088–1097. doi: 10.1038/bjc.2015.61 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Sobolewski C, Cerella C, Dicato M, Ghibelli L, Diederich M. The Role of Cyclooxygenase-2 in Cell Proliferation and Cell Death in Human Malignancies. International Journal of Cell Biology. 2010;2010:1–21. doi: 10.1155/2010/215158 [DOI] [Google Scholar]
- 36.Chan HT, Nagayama S, Otaki M, et al. Tumor-informed or tumor-agnostic circulating tumor DNA as a biomarker for risk of recurrence in resected colorectal cancer patients. Front Oncol. 2023;12:1055968. doi: 10.3389/fonc.2022.1055968 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.