If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Postoperative circulating tumor DNA combined with consensus molecular subtypes can better predict outcomes in stage III colon cancers: A prospective cohort study
Department of Colorectal Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaDepartment of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, China
Department of Colorectal Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaDepartment of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, China
Department of Colorectal Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaDepartment of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, ChinaCancer Institute, Fudan University Shanghai Cancer Center, Shanghai 200032, China
Department of Colorectal Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaDepartment of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, China
Department of Colorectal Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaDepartment of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, China
Department of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, ChinaDepartment of Pathology, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaInstitute of Pathology, Fudan University, Shanghai 200032, China
Department of Colorectal Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaDepartment of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, ChinaCancer Institute, Fudan University Shanghai Cancer Center, Shanghai 200032, China
Department of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, ChinaDepartment of Pathology, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaInstitute of Pathology, Fudan University, Shanghai 200032, China
Department of Colorectal Surgery, Fudan University Shanghai Cancer Center, Shanghai 200032, ChinaDepartment of Oncology, Shanghai Medical College, Fudan University, Shanghai 200032, China
ctDNA was detectable in 15.9% pre-chemo and 8.9% post-chemo sample.
•
About 54% of positive ctDNA cases recurred, while 21.3% of negative cases recurred.
•
Sixty percent of CMS4 tumors recurred, while only 21% of CMS1-3 tumors recurred.
•
Redefined risk of ctDNA, CMS and clinical risk is a better prognostic biomarker.
•
High-/mid-risk cases were 14.6 and 5.3 times more likely to recur than the low-risk.
Abstract
Background
Precise methods for risk stratification to guide adjuvant chemotherapy for stage III colon cancers are needed. Here, we combined circulating tumor DNA (ctDNA) with consensus molecular subtype (CMS) to improve risk stratification in stage III colon cancers.
Methods
We conducted a prospective, observational cohort study of 165 patients with stage III colon cancers. Somatic variants in tumor tissues and plasmas collected pre- and post-chemo were detected via a targeted sequencing panel of 197 cancer-related genes. CMSs classification was characterized using a targeted RNA sequencing panel of 788 genes.
Results
We analyzed 151 pre-chemo and 124 post-chemo plasmas, while 130 patients were CMSs classified. ctDNA was detectable in 15.9% pre-chemo and 8.9% post-chemo samples. Significantly worse recurrence-free survival (RFS) was seen if ctDNA was detectable in pre-chemo samples (hazard ratio [HR], 3.585; P < 0.001) or in post-chemo samples (HR, 3.337; P = 0.005). Pre-chemo ctDNA (HR, 5.538; P < 0.001) and post-chemo ctDNA status (HR, 3.272; P = 0.037) remained independently associated with RFS in multivariate analysis. According to the redefined recurrence risk stratification, mid-risk patients (ctDNA-negative with CMS4/T4 or N2 tumors) were 5.3 times (HR, 5.269; P = 0.025) more likely to relapse than low-risk patients (ctDNA-negative with CMS1-3/T3N1 tumors), while high-risk patients (ctDNA-positive) were 14.6 times (HR, 14.590; P < 0.001) more likely to relapse.
Conclusions
Postoperative ctDNA indicating residual disease, combined with CMSs classification and clinical risk reflecting the intrinsic characteristics of tumors, can redefine risk stratification of stage III colon cancers and better predict relapse.
Colorectal cancer is the third most commonly diagnosed cancer worldwide and the second leading cause of cancer-related deaths, with over 1.9 million newly diagnosed cases annually [
]. Despite the implementation of screening, improvement of treatment modalities and advances in drug discovery, the 5-year mortality rate for patients with colorectal cancer remains at approximately 40% [
]. Radical resection of primary tumor followed by adjuvant chemotherapy is currently the standard treatment for patients with stage III colon cancers. Adjuvant chemotherapy with 5-fluorouracil and oxaliplatin based regimen has been proved effective to improve recurrence-free survival and overall survival [
], while the duration of combined chemotherapy can be adjusted according to the clinical risk categories from 3 months for low-risk group (T1-3 and N1) to 6 months for high-risk group (T4 and/or N2) [
Microsatellite instability in patients with stage III colon cancer receiving fluoropyrimidine with or without oxaliplatin: an ACCENT pooled analysis of 12 adjuvant trials.
Multicenter international society for immunotherapy of cancer study of the consensus immunoscore for the prediction of survival and response to chemotherapy in stage III colon cancer.
Prognostic and predictive value of the Immunoscore in stage III colon cancer patients treated with oxaliplatin in the prospective IDEA France PRODIGE-GERCOR cohort study.
Ann Oncol : official journal of the European Society for Medical Oncology.2020; 31: 921-929
High-risk Stage III colon cancer patients identified by a novel five-gene mutational signature are characterized by upregulation of IL-23A and gut bacterial translocation of the tumor microenvironment.
] showed promising effect on prognosis prediction for patients with stage III colorectal cancers, the treatment decision-making are still mainly based on the TNM stage and clinical risk factors. In recent years, circulating tumor DNA (ctDNA) has been detected in the cell-free component of peripheral blood samples in advanced colorectal cancers and many other solid tumors [
]. Several previous studies have suggested that in patients with stage I–III colorectal cancer, postoperative ctDNA was a valuable biomarker to predict minimal residual disease (MRD) after radical resection, thus redefining patients risk outcome groups [
]. In addition, consensus molecular subtypes (CMSs) provide an integrated framework to capture the intrinsic heterogeneity of colorectal cancer at the gene expression level. The CMSs classification holds potentials in clinical application for predicting prognosis and treatment response of patients with colorectal cancer at different stages [
]. Currently, the prognostic effect of ctDNA in colorectal cancer still requires further studies and more evidence; additionally, its relationship with clinical risk factors or tumor molecular features is largely unknown.
Here, we conducted a prospective observational study in patients with stage III colon cancers to investigate the clinical validity of postoperative ctDNA collected before and after adjuvant chemotherapy for recurrence prediction. The combined analyses of ctDNA status with CMSs classification and clinical risk groups were performed to redefine recurrence risk for patients with stage III colon cancers.
2. Material and methods
2.1 Study design and participants
This prospective observational single-center study (ChiCTR1800018754) recruited consecutive patients with stage III colon cancers treated at the Fudan University Shanghai Cancer Center (FUSCC). Key eligibility criteria included patients pathologically diagnosed with stage III colon cancer located at least 12 cm from anal verge with radical resection, with no evidence of metastatic disease on computed tomography (CT) of the chest, abdomen and pelvis before surgery, received a treatment plan of standard adjuvant chemotherapy for 3–6 months without early recurrence or death, and were able to provide at least 4 mL plasma. Patients who had previously been diagnosed with another malignant neoplasm, received neo-adjuvant therapy, or were unable to tolerate adjuvant chemotherapy for 3 months were excluded from the study.
The chemotherapy regimens, including XELOX(Oxaliplatin 130 mg/m2 IV day 1, capecitabine 1000 mg/m2 twice daily PO for 14 days, repeat every 3 weeks), FOLFOX (Oxaliplatin 85 mg/m2 IV day 1, Leucovorin 400 mg/m2 IV day 1, 5-FU 400 mg/m2 IV bolus on day 1, followed by 2400 mg/m2 CIV 46h, repeat every 2 weeks) and Xeloda (1250 mg/m2 twice daily PO for 14 days, repeat every 3 weeks), were chosen by the clinicians who were blinded to the ctDNA results. We classify the clinical risk group into two categories based on standard criteria, with low-risk patients defined as those with pT3 and pN1 disease according to the pTNM staging system, and high-risk patients as those with pT4 and/or pN2 disease. All patients had surveillance CT scans 4–6 weeks after the completion of adjuvant chemotherapy. Thereafter, patients were followed up intensively according to guidelines of National Comprehensive Cancer Network. Clinicopathological data, as well as postoperative follow-up and surveillance information, were also collected (Supplementary Table S1).
The study protocol was approved by the ethical committee at FUSCC (1808189–14), and was conducted in full compliance with the principles of the Helsinki Declaration and national regulations. Informed consent form was signed by all patients before any procedure. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. All authors had access to the study data and had reviewed and approved the final manuscript.
2.2 Sample processing and data analysis
The Roche AVENIO Tumor Tissue and ctDNA Surveillance Kits (Roche) covering 197 cancer-related genes were used for analyses. CMSs classification was characterized using a targeted RNA sequencing panel of 788 genes. Detailed tumor tissue mutational analysis, ctDNA analysis, gene expression analysis by targeted RNA sequencing and CMSs classification are described in Supplementary materials and methods.
2.3 Statistical analysis
Data were analyzed from December, 2020, through May, 2021. Differences in baseline characteristics between patients with ctDNA-positive and ctDNA-negative findings were assessed using the two-sided Pearson χ2 test or Fisher exact test for categorical variables as appropriate. The primary outcome measure was recurrence-free survival (RFS), which as measured from the date of surgery to the first documented radiological recurrence or death due to colon cancer. Data were censored at the last follow-up or noncolon cancer-related death. Survival analysis was performed using the Kaplan–Meier method with the log-rank test. Cox proportional hazards regression models were used to assess the association of ctDNA and clinicopathological factors with RFS. Multivariate analysis was performed with all variables without interaction with others. To assess the performance of ctDNA and clinical risk measurements in predicting outcomes, we calculated an area under the curve (AUC) with the receiver operating characteristics (ROC) method using the R package survAUC [
]. All analyses were performed with SPSS 22.0 (SPSS, Chicago, IL) and R 3.1.2 software (Institute for Statistics and Mathematics, Vienna, Austria). Two-sided P < 0.05 was considered significant and all confidence intervals (CIs) were stated at the 95% confidence level.
3. Results
3.1 Clinicopathological characteristics, CMSs classification and ctDNA detection
Patient enrolment and the study design are presented in Fig. 1. 165 patients were enrolled from Aug. 2018 to Dec. 2019. Fourteen patients were excluded from further analysis due to unqualified cfDNA of pre-chemo blood samples (n = 7) or stage IV disease at diagnosis (n = 7), leaving 151 for the analysis. Another 27 patients were further excluded from post-chemo ctDNA analysis due to unwillingness (n = 13), inadequate chemotherapy (n = 12), unqualified cfDNA (n = 1), or death prior to chemo completion (n = 1). Thus, only patients receiving chemotherapy without early progression or death are evaluated for analyses of post-chemo ctDNA. CMSs classification was applied to a total of 134 patients' tissue specimens, while 4 patients’ specimen cannot be classified into any of the four CMS subtypes (Supplementary Figure S1).
Fig. 1Patient enrolment, sample collections, and evaluable population. A) Flowchart of patient enrolment. B) Sample collections and detections schema. cfDNA, cell-free DNA; FFPE,formalin-fixed, paraffin-embedded; NGS, next generation sequencing; ctDNA, circulating tumor DNA; CMS, consensus molecular subtype.
The clinicopathological characteristics of 151 eligible patients were summarized in Supplementary Table S2. The patients’ median age was 61 years (range 25–86) and 40.4% were female. Right-sided tumors were found in 28.5% of the patients (43/151). Preoperative CEA levels were elevated in 59/151 (39.1%) of patients. 51.7% (78/151) of the tumors were pathologically diagnosed as T4 and 33.8% (51/151) as N2, leading to that 66.2% tumors were classified as high-risk according to NCCN guidelines. Median number of lymph nodes dissected was 18, and 12 patients had fewer than 12 lymph nodes examined. CMSs classification was possible on 130 patients. 17 (11.3%) patients were classified as CMS1, 70 (46.4%) as CMS2, 13 (8.6%) as CMS3 and 30 (19.9%) as CMS4. 141 patients (93.4%) were treated with oxaliplatin-based therapy (XELOX or FOLFOX), 7 (4.6) received adjuvant Xeloda alone, and 3 did not receive adjuvant chemotherapy. 92.0% (139/151) of the patients completed at least 3 months of adjuvant chemotherapy and 108 (71.5%) patients completed 6 months of adjuvant chemotherapy.
In the primary tumor, somatic variants were found in 136 patients. The detailed mutational landscape is presented in Supplementary Figure S2, together with corresponding clinicopathological features. A tumor-specific mutation was detected in 24 of 151 (15.9%) pre-chemo plasma samples (ctDNA-positive) and 11 of 124 (8.9%) post-chemo plasma samples. Chemotherapy cleared ctDNA of 10 of 19 patients (52.6%) with positive findings in pre-chemo samples and ctDNA in 2 of 105 patients (1.9%) with negative findings in pre-chemo plasma turned positive in post-chemo samples. Baseline clinicopathological characteristics and their association with pre-chemo and post-chemo ctDNA status are shown in Supplementary Table S2. Patients with N2 disease (P = 0.001), poor differentiation (P = 0.015) and high-risk disease were more likely to be tested as ctDNA positive in the pre-chemo samples; while patients under 60 years old were associated with positive post-chemo ctDNA (P = 0.005).
3.2 ctDNA status and RFS
During a median follow-up of 33.5 months (95%CI 33.0–34.2), 39 patients had radiological recurrence, and 9 patients died of colon cancer. Pre-chemo ctDNA was detectable in 13/39 (33.3%) of patients with recurrence. Patients with detectable pre-chemo ctDNA had an increased risk of recurrence (HR, 3.585; 95% CI, 1.838–6.991; Log rank P < 0.001) (Fig. 2A). Kaplan–Meier estimates of 3-year RFS for patients with positive pre-chemo ctDNA findings were 45.8%, while those with ctDNA-negative findings were 77.1%. Recurrence occurred in 26 of 127 patients (20.5%) who had negative pre-chemo ctDNA findings, while 13 of 24 (54.2%) patients with positive pre-chemo ctDNA experienced relapse (Fig. 2B).
Fig. 2Recurrence-free survival (RFS) of patients according to circulating tumor DNA (ctDNA) status. Kaplan–Meier estimates of RFS (A) and comparison of recurrence rate (B) for 151 patients with stage III colon cancer stratified by pre-chemo ctDNA status. Kaplan–Meier estimates of RFS (C) and Comparison of recurrence rate (D) for 124 patients with stage III colon cancer stratified by post-chemo ctDNA status.
The ctDNA status of the post-chemo samples was also strongly associated with RFS (HR, 3.337, 95% CI, 1.379–8.077; Log rank P = 0.005) (Fig. 2C). The 3-year RFS for cases with detectable post-chemo ctDNA was 45.5%, while the 3-year RFS for those with undetectable ctDNA was 72.7%. Recurrence occurred in 28 of 111 patients (24.8%) with negative post-chemo ctDNA findings, while 6 of 11 patients (54.5%) with positive findings experienced relapse (Fig. 2D).
The RFS outcomes were compared according to the pre-chemo and post-chemo ctDNA status (Supplementary Figure S3). Post-chemo ctDNA was not associated with RFS in both subsets of patients with negative findings (Supplementary Figure S3A) and positive findings (Supplementary Figure S3B) in pre-chemo ctDNA (P > 0.05).
3.3 ctDNA status combined with CMSs classification/clinical risk and RFS
In terms of CMSs classification for prognostic analysis, CMS1, CMS2 and CMS3 were analyzed as one subgroup compared with CMS4. Kaplan–Meier analysis indicated that patients with CMS4 subtype of tumors had a significantly increased risk to experience occurrence (HR, 2.145; 95% CI, 1.054–4.363; Log rank P = 0.031) (Fig. 3A). In the subgroup of CMS1-3 tumors, patients with negative findings in pre-chemo ctDNA have significantly lower risk of relapse (HR, 6.249; 95% CI, 2.576–15.156; Log rank P < 0.001) (Fig. 3C), with 85.1% of patients not experiencing recurrence (Fig. 3B). In the CMS4 tumor subgroup, patients with positive pre-chemo ctDNA have a higher risk of relapse (HR, 3.894; 95% CI, 1.041–14.573; Log rank P = 0.028) (Fig. 3D), with 75.0% experiencing recurrence (Fig. 3B).
Fig. 3Recurrence-free survival(RFS) outcomes according to consensus molecular subtypes (CMSs) classification and circulating tumor DNA (ctDNA) status. A) Kaplan–Meier estimates of RFS for 130 patients with stage III colon cancer stratified by CMSs classification. B) Comparison of recurrence rate for patients stratified by pre-chemo ctDNA status in subsets of patients with CMS1-3 and CMS4 tumors. Kaplan–Meier estimates of RFS for patients stratified by pre-chemo ctDNA status in the subset of patients with CMS1-3 tumors (C) and CMS4 tumors (D).
In terms of clinical risk, Kaplan–Meier analysis showed that patients with high-risk tumors had an increased risk of recurrence compared to patients with low-risk tumors (HR, 3.251; 95% CI, 1.361–7.764; Log rank P = 0.005) (Fig. 4A). Subgroup analysis according to clinical risk was further conducted. In the low-risk subgroup, pre-chemo ctDNA status was not associated with RFS (Log rank P = 0.073) (Fig. 4C) with recurrence rate of 10.2% and 50.0% in patients with negative and positive findings in pre-chemo ctDNA, respectively (Fig. 4B) . In the high-risk subgroup, patients with positive findings in pre-chemo ctDNA had a significantly increased risk of relapse (HR, 2.672; 95% CI, 1.311–5.445; Log rank P = 0.005) (Fig. 4D), with 54.5% of patients experiencing recurrence compared to 26.9% of recurrence rate in patients with negative pre-chemo ctDNA (Fig. 4B).
Fig. 4Recurrence-free survival(RFS) outcomes according to clinical risk and circulating tumor DNA (ctDNA) status. A) Kaplan–Meier estimates of RFS for 151 patients with stage III colon cancer stratified by clinical risk. B) Comparison of recurrence rate for patients stratified by pre-chemo ctDNA status in subsets of patients with clinical low-risk and high-risk tumors. Kaplan–Meier estimates of RFS for patients stratified by pre-chemo ctDNA status in the subset of patients with low-risk tumors (C) and high-risk tumors (D).
3.4 Clinicopathological variables and multivariate analysis
Univariate and multivariate analysis were shown in Table 1. In 130 patients with pre-chemo ctDNA detected and CMSs classified, univariate analysis showed that T stage (HR, 2.956; 95% CI, 1.372–6.369; P = 0.006), CMSs classification (HR, 2.145; 95% CI, 1.054–4.363; P = 0.035), clinical risk (HR, 4.716; 95% CI, 2.561–10.992; P = 0.004) and pre-chemo ctDNA status (HR, 5.306; 95% CI, 2.561–10.992; P < 0.001) were significantly associated with RFS. Following multiple variables adjustment in the Cox proportional hazards regression model, pre-chemo ctDNA status had the strongest independent association with RFS (HR, 5.538; 95% CI, 2.487–12.331; P < 0.001), followed by clinical risk (HR, 3.290; 95% CI, 1.100–9.843; P = 0.033), CMSs classification (HR, 3.015; 95% CI, 1.306–6.962; P = 0.010) and tumor location (HR, 2.489; 95% CI, 1.006–6.158; P = 0.048).
Table 1Univariate and multivariate Cox analysis of recurrence-free survival (RFS) by clinicopathological variables and circulating tumor DNA (ctDNA) status.
P-value in bold denotes statistically significant.
Age
0.524
0.295
0.598
0.602
> 60 versus ≤ 60
0.800 (0.403–1.588)
0.682 (0.333–1.396)
0.832 (0.419–1.650)
0.823 (0.396–1.712)
Gender
0.462
0.385
0.528
0.189
Female versus male
0.762 (0.369–1.572)
0.719 (0.343–1.511)
0.792 (0.384–1.635)
0.579 (0.256–1.309)
Tumor location
0.306
0.048
0.186
0.042
Right-sided versus left-sided
1.461 (0.707–3.017)
2.489 (1.006–6.158)
1.632 (0.790–3.371)
2.454 (1.034–5.821)
Histological type
0.908
0.322
0.914
0.582
Mucinous tumors versus adenocarcinoma
1.073 (0.327–3.515)
0.500 (0.127–1.970)
1.067 (0.326–3.499)
0.687 (0.181–2.610)
T stage
0.006
0.004
T4 versus T1-3
2.956 (1.372–6.369)
NI
3.114 (1.444–6.712)
NI
N stage
0.432
0.463
N2 versus N1
1.329 (0.654–2.702)
NI
1.305 (0.642–2.652)
NI
Histological grade
0.127
0.370
0.131
0.394
Poor versus moderate/well
1.712 (0.858–3.416)
1.410 (0.665–2.990)
1.703 (0.854–3.398)
1.380 (0.658–2.892)
Lymphovascular invasion
0.383
0.479
0.407
0.907
Yes versus No
1.392 (0.662–2.927)
0.746 (0.331–1.680)
1.369 (0.651–2.880)
0.953 (0.424–2.144)
Nerve invasion
0.167
0.472
0.296
0.734
Yes versus No
1.622 (0.817–3.220)
1.349 (0.597–3.046)
1.441 (0.726–2.860)
1.146 (0.523–2.513)
CMS subgroup
0.035
0.010
0.069
0.023
CMS4 versus CMS1-3
2.145 (1.054–4.363)
3.015 (1.306–6.962)
1.931 (0.949–3.930)
2.657 (1.146–6.160)
Clinical risk
0.004
0.033
0.004
0.012
High-risk versus low-risk
4.716 (1.656–13.430)
3.290 (1.100–9.843)
4.756 (1.669–13.548)
3.957 (1.356–11.543)
ctDNA status
< 0.001
< 0.001
0.038
0.037
Positive versus negative
5.306 (2.561–10.992)
5.538 (2.487–12.331)
2.750 (1.060–7.134)
3.272 (1.071–9.992)
NI: Not included due to interaction with clinical risk. Because clinical risk is a combination of T and N stage, the latter 2 variables were not included in the multiple variable analysis. An alternative multivariable model including T and N stage but not clinical risk produces similar results (Supplementary Table S3).
a P-value in bold denotes statistically significant.
While in 120 patients with post-chemo ctDNA tested and CMSs classified, univariate analysis showed that T stage (HR, 3.114; 95% CI, 1.444–6.712; P = 0.004), clinical risk (HR, 4.765; 95% CI, 1.669–13.548; P = 0.004) and post-chemo ctDNA status (HR, 2.750; 95% CI, 1.060–7.134; P = 0.038) were significantly associated with RFS. In multivariate analysis, post-chemo ctDNA (HR, 3.272; 95% CI, 1.071–9.992; P = 0.037), together with clinical risk (HR, 3.957; 95% CI, 1.356–11.543; P = 0.012), CMSs subtype (HR, 2.657; 95% CI, 1.146–6.160; P = 0.023) and tumor location (HR, 2.454; 95% CI, 1.034–5.821; P = 0.042) still remained as independent factor for predicting RFS. Pre- and post-chemo ctDNA status, CMSs classification and clinical risk are all strong dependent prognostic factors for RFS.
3.5 Redefined risk stratification and RFS
In the subsequent analyses, postoperative ctDNA was considered positive if either pre-chemo ctDNA or post-chemo ctDNA was detectable. Combinations of postoperative ctDNA, CMSs classification and clinical risk were then analyzed to investigate how to better stratify recurrence risk for stage III colon cancer. Kaplan–Meier analysis indicated that patients with negative postoperative ctDNA, CMS1-3 and T3N1 tumors had the lowest risk for recurrence, with a 3-year RFS of 93.5%, while patients in subgroups with positive postoperative ctDNA had a poor 3-year RFS of less than 50.0%, regardless of status of CMSs classification and clinical risk (P < 0.001, Supplementary Figure S4 and Table 2).
Table 21-Year and 3-year Recurrence-free survival (RFS) for patients in subgroups stratified by circulating tumor DNA status (ctDNA), consensus molecular subtypes (CMSs) classification and clinical risk.
No.(%)
1-Year RFS (%)
3-Year RFS (%)
Subgroup analysis according to ctDNA, CMS classification, and clinical risk
ctDNAneg-CMS1-3-low-risk
31 (25.8)
96.8
93.5
ctDNAneg-CMS1-3-high-risk
45 (37.5)
97.8
69.4
ctDNAneg-CMS4-low-risk
8 (6.7)
100.0
87.5
ctDNAneg-CMS4-high-risk
17 (14.2)
82.4
44.1
ctDNApos-CMS1-3-low-risk
1 (0.8)
100.0
100.0
ctDNApos-CMS1-3-high-risk
13 (10.8)
69.2
38.5
ctDNApos-CMS4-low-risk
1 (0.8)
100.0
0.0
ctDNApos-CMS4-high-risk
4 (3.3)
50.0
50.0
Subgroup analysis according to recurrence risk stratified by ctDNA, CMS classification, and clinical risk
Subsequently, recurrence risk was redefined: patients with negative postoperative ctDNA, CMS1-3 and T3N1 tumors were defined as low-risk, patients with negative postoperative ctDNA and CMS4, T4 or N2 tumors were defined as mid-risk, and patients with positive postoperative ctDNA were classified as high-risk. The Kaplan–Meier analysis demonstrated that the recurrence risk of patients with stage III colon cancer in our cohort could be effectively stratified by newly defined risk stratification criteria (P < 0.001, Fig. 5A), with recurrence rate of 6.5%, 28.6% and 42.1%, respectively, for low-risk, mid-risk and high-risk patients (P < 0.001, Fig. 5B). According to redefined risk stratification criteria, mid-risk patients were 5.3 times (HR, 5.269; 95% CI, 1.230–22.575; Log rank P = 0.025) more likely to relapse than low-risk patients, while high-risk patients were 14.6 times (HR, 14.590; 95% CI, 3.227–65.956; Log rank P < 0.001) more likely to relapse.
Fig. 5Recurrence-free survival (RFS) of patients according redefined risk stratification. A) Kaplan–Meier estimates of RFS for 120 patients with stage III colon cancer stratified by redefined risk stratification. B) Comparison of recurrence rate for 120 patients with stage III colon cancer stratified by redefined risk stratification.
Furthermore, time-dependent ROC curves were used to compare the sensitivity and specificity of RFS prediction of postoperative ctDNA, CMSs classification, clinical risk and redefined risk stratification based on combination of three factors. In the ROC analysis for 1-year RFS prediction, redefined risk stratification had an area under the ROC curve (AUROC) of 0.735 (95%CI 0.580–0.890), which had a trend to better predict recurrence compared to postoperative ctDNA, CMSs classification or clinical risk alone (Supplementary Figure S5A). For 3-year RFS prediction, redefined risk stratification had a significantly higher predictive significance with AUROC of 0.656, than postoperative ctDNA alone with AUROC of 0.585 (Supplementary Figure S5B).
4. Discussion
Novel biomarkers are urgently needed to better define personalized risk of recurrence and improve clinical decision-making for patients with stage III colon cancers. ctDNA has been validated as a robust biomarker to detect MRD and CMSs classification is considered as a comprehensive biomarker to reflect intrinsic biological characteristics of tumor itself. Hereby, we reported the prognostic effect of ctDNA in patients with stage III colon cancers and demonstrated that patients with positive findings in postoperative ctDNA had significantly poor outcomes than those with negative findings. Moreover, we further explored the relationship between ctDNA and CMSs classification. To the best of our knowledge, this prospective observational study is the first to combine ctDNA with CMSs classification and clinical risk to investigate the prognostic value in order to guide clinical treatment and surveillance strategy in stage III colon cancers.
] firstly explored the prognostic value of ctDNA in stage II colon cancers and they found that positive postoperative ctDNA defines a population at very high risk of recurrence (HR, 18; P < 0.001) and initially highlighted the potential utility of ctDNA for tracking adjuvant chemotherapy effectiveness. The same research team further investigated the ability of ctDNA to improve risk stratification in patients with locally advanced rectal cancer, and demonstrated that significantly worse RFS was seen if ctDNA remained positive after chemoradiotherapy (HR, 6.6; P < 0.001), indicating ctDNA analysis could potentially be used to guide patients selection for adjuvant therapy after surgery [
]. Then they tried to determine whether serial postoperative and post-chemo ctDNA analysis could provide a real-time indication of adjuvant therapy efficacy in stage III colon cancer [
]. They found that post-chemo ctDNA analysis may define a subset of patients that remains at high risk of recurrence despite completing standard adjuvant treatment and this high-risk population may need additional therapeutic approaches. The following research on stage III colorectal cancer verified their finding [
Circulating tumor DNA in stage III colorectal cancer, beyond minimal residual disease detection, toward assessment of adjuvant therapy efficacy and clinical behavior of recurrences.
]. At the same time, studies on prognostic and surveillance value of serial detection of ctDNA, including time-points of pre- and post-surgery, and every third month afterwards during follow-up for 2–3 years, were also widely conducted in stage I-III colorectal cancers [
]. During surveillance, serial ctDNA-positive patients were more than 30 times more likely to experience recurrence than ctDNA-negative patients, and serial ctDNA analyses revealed disease recurrence up to 5–16.5 months ahead of radiological imaging.
Similar to previous study on ctDNA value in stage III colon cancers by Tie et al. [
Circulating tumor DNA in stage III colorectal cancer, beyond minimal residual disease detection, toward assessment of adjuvant therapy efficacy and clinical behavior of recurrences.
], our study confirmed the prognostic value of postoperative ctDNA. We have demonstrated that patients with positive ctDNA findings after surgery have poor outcomes with an estimated 3-year RFS of 45.8% and 45.5% for pre-chemo and post-chemo settings respectively, compared with 77.1% and 72.7% in those with a negative ctDNA finding (HR, 3.585 and 3.337, respectively; P < 0.001). However, patients with ctDNA clearance or turning positive fails to manifest significant prognostic difference in our study. The insignificant result may due to the following reasons. First, the sample size was not large enough, while only 19 patients with positive pre-chemo ctDNA were eligible for further analysis. Second, the panel and the ctDNA detection method we used is based on a tumor-agnostic approach of a relative low depth of sequencing, which has lower sensitivity compared with tumor-informed approach, especially in the setting of where mutation allele frequency (AF) is low in post-chemo ctDNA. Third, the time point of post-chemo ctDNA detection in our study is closer to chemotherapy completion (2–4 weeks in our study vs 4–6 weeks in published studies), which may affect the result. The residues of chemo drugs and inflammatory factors may interfere the ctDNA detection. Fourth, our cohort is composed of a higher proportion of high-risk patients (66.2%. 100/151). It may be difficult with current technical conditions for ctDNA to stratify such a subgroup of high possibility to relapse into a more individualized subset of worse outcome.
Studies based on different stages of CRC have all showed that positive post-surgical ctDNA is associated a significantly increased risk of relapse. Thus, the detection of MRD by post-surgical ctDNA may help guide the personalized administration of adjuvant chemotherapy. However, the sample size of previous studies are relative small and more evidence are still needed to confirm the findings. Several prospective randomized clinical trials (RCTs) of large cohorts, such as DYNAMIC (ACTRN12615000381583)/DYNAMIC-III (ACTRN12617001566325)/DYNAMIC-Rectal (ACTRN12617001560381), CIRCULATE (NCT04089631), and COBRA (NCT04068103) trials, are currently being conducted to compare ctDNA-guided duration and intensity of chemotherapy with standard of care. However, not until the survival benefit of this strategy over the standard of care is validated by RCTs, it is too early to apply ctDNA-guided chemotherapy into the clinical practice.
Although substantial studies showed that positive findings in postoperative ctDNA indicates significant poor outcome, the positive rate of ctDNA after resection is around 20%, which is lower than the recurrence rate (approximately 30%) of stage III colon cancer. Other biomarkers are needed to better predict recurrence in stage III colon cancer. CMSs classification integrates six independent transcriptomic-based subtyping systems [
], and represents the present best description of colorectal cancer heterogeneity based on tumors’ intrinsic biological characteristics. Independent of the classifier used, including original gene-expression profiling [
Immunohistochemistry-based consensus molecular subtypes as a prognostic and predictive biomarker for adjuvant chemotherapy in patients with stage II colorectal cancer.
]. CMS4 tumors has been validated associated with worse overall survival in localized colorectal cancers, representing an aggressive subtype with mesenchymal phenotype and drug-resistant feature [
]. Consistently, our study demonstrated that patients with CMS4 tumors have higher risk of relapse than patients with CMS1-3 tumors (HR, 2.145; P = 0.031). In patients with negative findings in pre-chemo ctDNA, CMS4 classification still can screen out a subset of tumors with worse RFS (HR, 2.671; P = 0.019), which may need more intensive treatment and follow-up. The tumors of this subset of patients have aggressive behavior contributing to its worse RFS, though without residual disease after resection.
Taking into account recognized prognostic factors, such as CMS classification [
Three versus 6 Months of oxaliplatin-based adjuvant chemotherapy for patients with stage III colon cancer: disease-free survival results from a randomized, open-label, international duration evaluation of adjuvant (IDEA) France, phase III trial.
Circulating tumor DNA in stage III colorectal cancer, beyond minimal residual disease detection, toward assessment of adjuvant therapy efficacy and clinical behavior of recurrences.
], and in order to make our redefined risk stratification system simplified and accurate, our study is the first to explore the prognostic value of combination of ctDNA status, CMSs classification and clinical risk. We found that the combination of three factors is a much better predictive factor than each alone. Positive findings in postoperative ctDNA is the strongest factor, while CMSs classification and clinical risk are also helpful in discriminating patients with various outcomes and guiding treatment and surveillance strategy. Specifically, newly defined low-risk patients (ctDNA-negative with T3N1 and CMS1-3 tumors), representing more than one quarter of stage III colon cancers with benign biological behavior and no residual disease, have 3-year RFS of 93.5%, which is comparable to that of stage I colon cancer. For this low-risk group of patients, it may be considered to reduce the intensity of adjuvant chemotherapy and follow-up. More prospective, multi-center data are needed to validate our findings before it can be used routinely in clinical practice. Otherwise, according to the published data, patients with negative findings in series ctDNA testing during follow-up have 2-year RFS of 87.7% in stage II–III colorectal cancers and 3-year RFS of 96.7% in stage I–III colorectal cancers. In our study, by combining with CMSs classification and clinical risk, only two time points of ctDNA testing (pre- and post-chemo) yielded the similar prognostic value with serial ctDNA testing. Our findings suggested that although the prognostic value of ctDNA is significant, the role of clinical and molecular features of tumors cannot be ignored in individualized treatment and surveillance.
There are potential limitations to our study. First, a relative modest sample size, and low event rate may limit the power of any subgroup analysis. Second, only 33.3% (13/39) of recurrences were detected by pre-chemo ctDNA analysis. The limited assay sensitivity may due to the relative low depth of sequencing of the tumor-agnostic ctDNA method we used, especially in the setting where AF is low. Third, the predictive value of ctDNA combined with CMSs classification and clinical risk was only conducted in one cohort from a single medical center, lacking internal and external validation, which limited the extrapolation of the results. Fourth, our study is a non-interventive clinical trial, and prospective, randomized clinical trials are needed to further investigate whether adjuvant chemotherapy and surveillance could be de-escalated in the newly defined low-risk group, and escalated in the high-risk group.
5. Conclusion
In conclusion, our study highlights that postoperative ctDNA status combined with CMSs classification and clinical risk can redefine risk stratification of stage III colon cancer and better predict relapse. In the era of possibly overemphasizing the prognostic value of ctDNA, the role of clinical and molecular features reflecting the intrinsic characteristics of tumors cannot be ignored. This hypothesis needs to be further explored in series of randomized studies on chemotherapy strategies guided by both ctDNA and CMSs classification.
Author contributions section
The individual contributions for this work are: Yaqi Li: Formal analysis; Investigation; Data Curation; Funding Acquisition; Writing-Original Draft; Writing-Review & Editing, Shaobo Mo: Formal analysis; Investigation; Writing-Original Draft, Long Zhang: Visualization; Validation, Xiaoji Ma: Investigation; Resources, Xiang Hu: Investigation; Resources, Dan Huang: Methodology, Binbin Lu: Data Curation; Methodology, Chonglin Luo: Data Curation; Methodology; Writing-Review & Editing, Haixiang Peng: Data Curation; Methodology, Weiqi Sheng: Conceptualization; Supervision; Funding Acquisition; Writing-Review & Editing, Sanjun Cai: Conceptualization; Supervision; Resources, Funding Acquisition; Writing-Review & Editing, Junjie Peng: Conceptualization; Supervision; Resources, Investigation; Funding Acquisition; Writing-Review & Editing.
Funding
The study was supported by grants from the National Natural Science Foundation of China (U1932145 to JP, 81972249 to WS, 82002946 to YL), Science and Technology Commission of Shanghai Municipality (18401933402 to JP), Fudan University Shanghai Cancer Center Basic and Clinical Translational Research Seed Foundation (YJZZ201802 to SC), Shanghai Clinical Science and Technology Innovation Project of Municipal Hospital (SHDC12020102 to WS), Artificial Intelligence Medical Hospital Cooperation Project of Xuhui District (2021-017 to WS), Shanghai Sailing Program (19YF1409500 to YL) and “Chenguang Program” supported by Shanghai Education Development Foundation and Shanghai Municipal Education Commission (20CG08 to YL).
Conflict of interest statement
The reagents of ctDNA detection and financial funding for this study were supported by Roche Diagnostics company. BL, CL and HP are employees of Shanghai Dunwill Medical Technology Co., Ltd, where CMS classification and ctDNA detection were conducted. The remaining authors have nothing to declare.
Acknowledgements
We would like to thank the patients and family members who gave their consent on presenting the data in this study, as well as the investigators and research staff involved in this study. We are greatly thankful to Kai Huang (Shanghai Dunwill Medical Technology Co., Ltd.) for performing the experiments for CMSs classification.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Microsatellite instability in patients with stage III colon cancer receiving fluoropyrimidine with or without oxaliplatin: an ACCENT pooled analysis of 12 adjuvant trials.
Multicenter international society for immunotherapy of cancer study of the consensus immunoscore for the prediction of survival and response to chemotherapy in stage III colon cancer.
Prognostic and predictive value of the Immunoscore in stage III colon cancer patients treated with oxaliplatin in the prospective IDEA France PRODIGE-GERCOR cohort study.
Ann Oncol : official journal of the European Society for Medical Oncology.2020; 31: 921-929
High-risk Stage III colon cancer patients identified by a novel five-gene mutational signature are characterized by upregulation of IL-23A and gut bacterial translocation of the tumor microenvironment.
Circulating tumor DNA in stage III colorectal cancer, beyond minimal residual disease detection, toward assessment of adjuvant therapy efficacy and clinical behavior of recurrences.
Immunohistochemistry-based consensus molecular subtypes as a prognostic and predictive biomarker for adjuvant chemotherapy in patients with stage II colorectal cancer.
Three versus 6 Months of oxaliplatin-based adjuvant chemotherapy for patients with stage III colon cancer: disease-free survival results from a randomized, open-label, international duration evaluation of adjuvant (IDEA) France, phase III trial.
00211-8&id=gr1.jpg){kind=link}
00211-8&id=gr2.jpg){kind=link}
00211-8&id=gr3.jpg){kind=link}
00211-8&id=gr4.jpg){kind=link}
00211-8&id=gr5.jpg){kind=link}
00211-8&id=figs1.jpg){kind=link}
00211-8&id=figs2.jpg){kind=link}
00211-8&id=figs3.jpg){kind=link}
00211-8&id=figs4.jpg){kind=link}
00211-8&id=figs5.jpg){kind=link}