Ardaman Shergill, MD, and Aparna Raj Parikh, MD, explore ongoing clinical investigations evaluating the use of circulating tumor DNA in the treatment of early-stage colon cancer.
Colorectal cancer (CRC) represents about 7.9% of all new cancer cases and is the second leading cause of cancer death in the United States. In 2022, about 150,000 new cases will occur and the 5-year survival is projected to be 65.1%.1 Likely due to screening efforts via colonoscopy, approximately 80% of patients with CRC present with early-stage disease;1 however, about 25% to 40% of patients with early-stage disease treated with curative intent have recurrence of their cancer despite definitive treatmen.2 Hence, therapies that improve upon current treatments, as well as strategies that improve earlier detection of recurrences, have the potential to significantly influence patient outcomes. In this article we discuss the role of circulating tumor DNA (ctDNA) in management and surveillance of early-stage colon cancer.
Currently, staging of nonmetastatic colon cancer is done pathologically. Clinical or pathologic risk factors such as obstruction or perforation on presentation, presence of lymphovascular or perineural invasion, presence of T4 disease, presence of lymph node involvement, or presence of tumor deposits predict for high risk of recurrence in stage I, II, or III disease. Historically, these factors have been used to determine the need for adjuvant chemotherapy to decrease the risk of disease recurrence.3,4 However, these factors have been identified only through retrospective analysis, limiting their real prognostic impact. Carcinoembryonic antigen (CEA) is the most commonly used prognostic and predictive blood-based biomarker in the management of CRC.
Recent studies have examined the role of circulating tumor DNA (ctDNA) as a measure of minimal residual disease (MRD). MRD detection refers to the least amounts of cancer that may be present in a patient after completion of definitive therapy but are not yet detectable radiographically.5 Stratification for risk of recurrence based on the clinicopathologic criteria is imprecise, and it does not accurately identify patients with MRD, which may result in over- or undertreatment of patients.6-8 Results of recent studies have shown that ctDNA is a promising biomarker for MRD for colon cancer because in colon cancer, compared with other types of cancers, there are overall higher rates of “shedding” of tumor fragments in circulation.9-11
Primary therapy for nonmetastatic colon cancer is with resection of the primary tumor along with the removal of regional lymph nodes. Surgery alone is curative in about 50% of patients, even those with stage III disease.6,8 Based on traditional clinicopathologic criteria, however, adjuvant chemotherapy is recommended for all patients with stage III colon cancer even if they have had curative resection because these criteria do not take residual disease into consideration; this leads to possible overtreatment of some patients.
Oxaliplatin-based adjuvant chemotherapy has improved 5-year disease-free survival (DFS) to about 67% to 70% in patients with resected stage III disease. The benefit of the addition of oxaliplatin in elderly patients has been questioned.12,13 In many cancer registries, patients with stage I or II colon cancer have 5-year survival rates of greater than 90% and about 80%, respectively.13,14 But based on current clinicopathologic stratification criteria alone, the questions of who benefits most from adjuvant chemotherapy and which patients have the highest risk of disease recurrence are still unanswered. For example, no adjuvant chemotherapy is currently recommended for patients with stage I disease; however, about 5% of such patients will have recurrence, and about 10% to 20% of patients with stage II disease who do not get chemotherapy will have recurrence.15 Approximately 15% and 30% of patients with stage II and III disease, respectively, experience recurrence despite completing appropriate treatment.12 These data highlight the limitations of current standards and underscore the need for improved patient selection for adjuvant therapy and better monitoring.
ctDNA is proving to be a promising biomarker in this setting. Many studies now have reported a correlation between the presence of ctDNA in plasma after the completion of definitive treatment and cancer recurrence.11,16-23 In general, 95% to 100% of patients with persistently detectable ctDNA recur if no systemic therapy is offered.11,17,20,24,25
Cell-free DNA (cfDNA) refers to extracellular DNA fragments detectable in various body fluids, including plasma, urine, cerebrospinal fluid, and saliva. DNA is released due to necrosis, apoptosis, active secretion, and autophagy as well as other forms of cell death (such as phagocytosis). In healthy individuals, the majority of cfDNA in plasma is released from hematopoietic cells. Cancer cells also release detectable cfDNA fragments, and these fragments have the genetic and epigenetic alterations that are unique to the tumor from which they originated.
ctDNA refers to short tumor-derived DNA fragments detectable in plasma. Factors including tumor burden, anatomic site of tumors, shedding characteristics of the tumor, and recent trauma or surgery can all influence the detection of ctDNA.26-28 Several ctDNA assays are currently in practice and trials (Table 1). These can be divided into tumor-agnostic and tumor-informed assays. Tumor-agnostic assays (eg, Guardant REVEAL) are broad panel–based assays that look for genomic alterations and aberrant DNA methylation patterns known to occur in a given tumor. Advantages of tumor-agnostic assays can include logistical simplicity (because of occasional limited availability of tumor tissue) and faster turnaround time. Tumor-informed assays (eg, Signatera, SafeSeqS) utilize whole-exome sequencing (WES) or targeted sequencing of the tumor. The assay is designed specifically for a given patient’s tumor, and it detects genomic alterations unique to that tumor. Tumor-informed assays have high sensitivity and low rates of false positivity. However, tumor-informed assays have a longer turnaround time, which can mean results are not available in time to influence clinical decision-making if the assays were ordered post surgery.
The fraction of ctDNA in plasma varies widely in patients with cancer, from less than 0.1% to greater than 10%, depending on burden of tumor, tumor DNA shedding, and anatomic site of the tumors.28-31 Notably, recent trauma, including surgery, can cause elevated cfDNA levels for up to 4 weeks post event, and thus it can influence ctDNA results. The currently available techniques cannot distinguish trauma-induced cfDNA from ctDNA that would indicate MRD. Hence, the optimal time to check ctDNA may be at least about 4 weeks post surgery. ctDNA results checked sooner than 4 to 6 weeks post surgery may need to be repeated to ensure accuracy.32
ctDNA presence has been studied in several adjuvant studies to predict risk of recurrence as well as to guide adjuvant therapy management (Table 2). Most recently, the DYNAMIC-II study results showed that analysis of ctDNA in resected stage II colon cancer may reduce the use of adjuvant chemotherapy in low-risk patients by 50%, without affecting the risk of disease recurrence.23 Using the tumor-informed Safe-SeqS assay, patients were randomized 2:1 to have treatment decisions guided by ctDNA or standard clinicopathologic features. The 2-year recurrence-free survival (RFS) rates were statistically equivalent in the ctDNA-guided adjuvant therapy group compared with the standard management group: 93.5% vs 92.4%; absolute difference, 1.1%; 95% CI, –4.1 to 6.2; noninferiority margin, –8.5%. The 3-year RFS rate was 92.5% in ctDNA-negative patients and 86.4% in ctDNA-positive patients (HR, 1.83; 95% CI, 0.79-4.27). The risk of recurrence in stage II, low-risk disease by historic clinicopathologic features was 96.7%, indicating that chemotherapy may be avoided in these patients with low-risk disease. These study results suggest that there may be low-risk, ctDNA-negative patients with resected stage II colon cancer in whom chemotherapy may be avoided.
Furthermore, in this study, those patients who had ctDNA-positive disease and then received chemotherapy seemed to derive a benefit from treatment based on the low rate of disease recurrence that they experienced. It should be noted that among all patients who received adjuvant chemotherapy, a higher percentage of patients in the ctDNA-guided group received an oxaliplatin-based doublet compared with those in the standard-management group (62% vs 10%) per treating physician’s choice. This may reflect a bias based on known prior knowledge of prognostic significance of ctDNA positivity, and historical data suggesting benefit of adding oxaliplatin to adjuvant fluorouracil (5-FU) alone for patients with high-risk disease. Among patients with ctDNA positivity, the 3-year RFS rate was 92.6% among those who received an oxaliplatin-based regimen and 76% among those who received 5-FU alone. Overall, this study shows that in stage II colon cancer that has been resected, ctDNA may be used to choose to de-escalate or even avoid therapy in lower-risk patients with postsurgery ctDNA-negative disease, but we have not yet answered all questions regarding de-escalation of adjuvant therapy in patients with high-risk stage II colon cancer. Furthermore, while numerically better RFS was seen in ctDNA-positive patients who were treated with an oxaliplatin-based regimen compared with 5-FU alone, large studies are needed to define the relative effect of adding oxaliplatin to adjuvant treatment in this setting.23
In prior stage II and III observational studies and pooled analyses, inferior 5-year RFS (38.6% vs 85.5%; P <.001) and overall survival (64.6% vs 89.4%;
P < .001) rates in patients with detectable ctDNA after the completion of definitive therapy have been seen, which hints at the prognostic impact of ctDNA in the management of early-stage colon cancer.19,20,24 In a study of patients with stage II colon cancer who were treated with chemotherapy and were ctDNA positive immediately after completing adjuvant chemotherapy, inferior RFS was noted compared with those having ctDNA negativity following adjuvant chemotherapy (HR, 11; 95% CI, 1.8-68, P = .001). Additionally, ctDNA also outperformed CEA in predicting radiographic recurrence; 85% of patients with positive ctDNA had radiographic recurrence vs only 41% of the patients with CEA elevation (P = .0003).19,20,24
Similar studies evaluating the role of ctDNA in postoperative settings showed that patients with ctDNA-positive disease had a higher risk of recurrence compared with patients with ctDNA-negative disease (HR, 7.2; P <.001). Similarly, positive ctDNA immediately after adjuvant chemotherapy and during surveillance was associated with 17 times (HR, 17.5; 95% CI, 5.4-56.5; P <.001) and 40 times (HR, 43.5; 95% CI, 9.8-193.5; P <.001) increased risk of cancer relapse, respectively, compared with negative ctDNA. A single test obtained postoperatively had sensitivity of about 41%; serial testing improved sensitivity to 88%. Specificity was 98% serially. In multivariable analyses, ctDNA was independently associated with the risk of cancer recurrence after adjusting for known clinicopathologic risk factors.11,28
In addition, tumor-agnostic assays, such as Guardant Health’s REVEAL assay, have emerged as important prognostic tools for the management of early-stage colon cancer. The REVEAL assay detects genomic changes and epigenomic signatures related to aberrant DNA methylation; this helps in detecting ctDNA but does not require any sequencing of tumor tissue, and so may be a valuable option especially in cases where tumor cellularity may be low. Aparna Raj Parikh, MD, and colleagues, tested feasibility of using a tumor-uninformed, plasma-only MRD ctDNA assay in 103 patients with stage I to IV colon cancer undergoing curative-intent surgery. Samples were collected 4 weeks after surgery and 4 weeks after completion of adjuvant therapy. The “landmark” time point was defined as when samples were collected approximately 1 month after completion of definitive therapy (surgery alone or surgery and chemotherapy in those who needed adjuvant therapy). Overall, 84 patients were evaluable, of whom 70 were evaluable for “landmark” analysis. Seventeen out of 70 patients (24%) had detectable ctDNA after definitive therapy. Two of the 17 patients had less than
1 year of follow up. Fifteen out of 15 patients with positive ctDNA and at least 1 year of clinical follow up had cancer recurrence (positive predictive value, 100%; HR, 11.28; P <.0001). The single time point sensitivity of this assay was 55.6% and single time point specificity was 100.0%. Longitudinal specimens improved sensitivity from 55.6% to 69.0% (n = 20/29 patients with positive ctDNA; HR, 12.26, P <.0001). In patients who had positive surveillance ctDNA results in a test obtained within 4 months of clinical recurrence, the observed sensitivity improved to 90.9% (n = 20/22). Notably, in this study, serum CEA levels did not predict recurrence well (positive predictive value, 53.9%; HR, 1.84; P = .18). Hence, plasma-based tumor-agnostic assays may have very promising clinical utility in MRD detection.17
The prospective CIRCULATE-Japan trial (jRCT1031200006) is an adaptive platform study that has 3 complementary clinical trials encompassing both de-escalation and escalation treatment strategies for patients with CRC, with selection (or enrollment) into each trial based on their ctDNA results. More than 3000 patients with CRC with stage I to IV disease will enroll. The tumor-
informed Signatera assay is used for MRD detection.33 The 3 studies are:
Preliminary results from the GALAXY observational cohort study showed that ctDNA positivity after surgery was predictive of benefit from adjuvant chemotherapy and that changes in the ctDNA in response to adjuvant chemotherapy predicted improved outcomes. Further, those patients who were ctDNA negative did not seem to benefit from adjuvant chemotherapy. The risk of disease relapse in patients with resected stage II and III colon cancer correlated well with ctDNA positivity at 1 month after surgery (HR, 13.3; P <.001). ctDNA clearance was observed in 68% of the patients, and in those patients the survival outcomes were similar to the outcomes of those who were ctDNA negative post surgery (HR, 0.8; P = .60). On the other hand, patients who were ctDNA positive despite adjuvant chemotherapy had 15.8 times increased risk of cancer recurrence than those who were ctDNA negative (P ≤.001). Patients with negative ctDNA 4 weeks after surgery had good survival outcomes regardless of adjuvant chemotherapy administration, with a DFS of approximately 95% at 12 months. The results were reported after a median follow-up of only 11.4 months. These data indicate the potential of ctDNA technologies in guiding patient selection for adjuvant therapy; however, longer follow-up, especially at completion of study accrual, is needed.34
Similarly, in a study of Danish and Spanish patients who had stage III CRC, ctDNA detection postoperatively and after adjuvant chemotherapy was strongly predictive of recurrence (HR, 7.0; 95% CI, 3.7-13.5; P <.001; and HR, 50.76; 95% CI, 15.4-167; P <.001, respectively). Recurrence rate was 80% in those who had positive ctDNA postoperatively and after adjuvant chemotherapy. Only those patients who cleared their ctDNA permanently did not relapse. This study also evaluated the correlation between the rate of rise of ctDNA and survival prognosis (HR, 2.7; 95% CI, 1.1-6.7; P = .039). There were 2 growth patterns: slow (25% mean ctDNA increase per month) and fast (143% mean ctDNA increase per month). The median lead time of detection of recurrence using every-3-month ctDNA detection in addition to standard imaging techniques was 9.8 months. However, importantly, standard imaging recommendations outside of the United States incorporate less-frequent images than those commonly used in this country, which may have influenced the lead time reported in this study.35
The goal of monitoring patients who have achieved what is, clinically, “no evidence of disease,” is to detect early recurrence and enable intervention early enough to improve their outcomes. Currently, the follow-up for patients who have completed definitive therapy for colon cancer incorporates clinical exams, serial serum CEA levels, periodic CT scans, and periodic colonoscopy evaluations.4 As noted above, studies using serial ctDNA monitoring seem to outperform those involving traditional CEA or radiographic surveillance. In one such study, serial ctDNA testing predicted cancer recurrence up to 16.5 months before radiologic imaging (mean, 8.7 months earlier; range, 0.8-16.5).10 Several studies have compared the sensitivity and specificity of ctDNA vs CEA in predicting cancer relapse, and the outcomes indicated that ctDNA significantly outperformed CEA and imaging.11,20,35 However, whether ctDNA-based early detection of colon cancer recurrence improved overall survival in patients is unclear and is a subject of future studies.
Several clinical trials are underway to further investigate the clinical utility of ctDNA (Table 3). These include testing risk of recurrence in various cohorts of patients with CRC, adjuvant therapy modification based on ctDNA levels, and overall impact of these ctDNA-based modifications on survival of patients. Some of these trials are PEGASUS (NCT04259944), IMPROVE-IT (NCT03748680), CIRCULATE-US (NRG-GI008; NCT05174169), VEGA (jRCT1031200006), COBRA (NCT04068103), DYNAMIC-III (ACTRN-12615000381583), TRACC (NCT04050345), MEDOCC-CrEATE (NL6281/NTR6455), and CIRCULATE-PRODIGE 70 (NCT04120701) to name several.
CIRCULATE-US is a large cooperative-group clinical trial, which will use the Signatera assay for MRD detection in patients with stage II and III colon cancer who have undergone resection. Since adjuvant chemotherapy does not cure all patients with colon cancer, this is an effort to improve outcomes by modifying therapy based on ctDNA detection postoperatively. Patients with ctDNA-negative disease will be randomized to standard-of-care (SOC) adjuvant chemotherapy or observation with standard close serial evaluations including clinical examinations, imaging, and CEA as well as ctDNA monitoring. Those with ctDNA-positive disease will be randomized to SOC chemotherapy or intensified chemotherapy with leucovorin, fluorouracil, irinotecan, and oxaliplatin (ie, FOLFIRINOX).
The BESPOKE (NCT04264702), ACT3 (NCT03803553), and COBRA studies were among the first ctDNA/colon cancer trials in the United States to open at a national level. BESPOKE is a prospective observational study in patients with stage I to IV CRC who undergo curative surgery and are followed for up to 2 years to examine the impact of tumor-informed ctDNA testing on adjuvant treatment decisions. COBRA is a phase 2/3 NRG Cooperative Group trial evaluating a plasma-only ctDNA detection approach to select high-risk patients with stage IIA colon cancer to receive chemotherapy following resection. ACT3 is evaluating the escalation of a different treatment—either chemotherapy with leucovorin, fluorouracil, and irinotecan (ie, FOLFIRI) or molecularly driven therapy—in patients who are ctDNA positive after chemotherapy for stage III disease. ACT3 is also evaluating how to modify therapy for those who continue to be ctDNA positive despite completing standard adjuvant therapies. However, additional questions remain to be answered in ongoing and future studies, including determining the best agents for treatment if ctDNA positivity persists and the impact of treating molecularly detected disease vs reinitiating therapy only when there is radiographic or other evidence of recurrent disease.
It is important to note that our outcomes are only as good as our tests, all of which must be used in the appropriate disease settings. We must be cautious about the possibility of false-negative ctDNA results, because if testing is done only immediately after surgery, or prior to 4 weeks post surgery, it may be impossible to say whether or not there is MRD: A patient could have truly negative MRD status or, alternatively, ctDNA titers might not high enough for detection or they could have low-shedding disease. If ctDNA is used for clinical decision-making or prediction, we recommend baseline testing prior to surgery in early-stage disease. In addition, consider repeating tests that indicate undetectable ctDNA post-op, 2 to 4 weeks later, to avoid false-negative results. When a decision is being made to escalate or de-escalate therapy, we want to be confident with the specificity of any given test.
As discussed here, a significant body of literature suggests that ctDNA-guided risk stratification in patients with resected early-stage colon cancer outperforms the current clinicopathologic criteria–based risk stratification. Hence, ctDNA monitoring in early-stage colon cancer is a novel emerging strategy that is likely to change current clinical practice with regard to monitoring, early detection of recurrence, and, potentially, adjuvant therapy selection.
Disclosure: The authors have no significant financial interest in or other relationship with the manufacturer of any product or provider of any service mentioned in this article.
Ardaman Shergill, MD1; and Aparna Raj Parikh, MD2,3
1The University of Chicago Medicine, Chicago, IL
2Massachusetts General Hospital, Boston, MA
3Dana-Farber Cancer Center, Boston, MA
Ardaman Shergill, MD
The University of Chicago Medicine
Section of Hematology/Oncology
5841 S Maryland Ave
Chicago, IL 60637