Liquid biopsies using circulating tumor DNA for surveillance of gastrointestinal cancers in Hispanics: first real-world data report

Abstract

Background

Malignant tumors release circulating tumor DNA (ctDNA) into the bloodstream, providing insights into tumor-specific mutations and pathways driving cancer progression. ctDNA testing is currently approved as a type of liquid biopsy to monitor disease burden and detect minimal residual disease (MRD). This study aimed to evaluate the adoption of ctDNA testing in a community oncology practice and assess the overall diagnostic performance of ctDNA and its association with disease progression in stage IV colorectal cancer (CRC), as determined by imaging studies.

Patients and methods

This retrospective study analyzed the medical records of 88 patients with gastrointestinal cancers (80 CRC, 5 gastric, 3 esophageal) who underwent ctDNA molecular testing between January 2020 and April 2022. Electronic medical records from patients aged ≥21 years who had two or more ctDNA tests with concurrent imaging studies or a pathology-confirmed CRC, gastric cancer, or esophageal cancer diagnosis were evaluated.

Results

At baseline, 47 (53.4%) patients had negative and 41 (46.6%) had positive results. Most patients had CRC (90.1%). In stage IV CRC, ctDNA was increasing before radiologic progression in all documented cases (100%), with a median lead time of 2.5 months (range 0.5-15 months). In early-stage CRC (I-III), ctDNA preceded radiologic progression in 40% of cases, with a median lead time of 6 months (range 6-10 months).

Conclusions

Using real-world data, we report the first-time results of the ctDNA testing adoption in a community oncology setting among patients with gastrointestinal cancers, predominantly CRC. Our findings suggest that integration of ctDNA testing may support disease monitoring in routine clinical practice.

Highlights

• •This is the first real-world study of ctDNA gastrointestinal Hispanic cancer monitoring.
• •ctDNA concords with disease confirmed by imaging in stage IV CRC.
• •Routine ctDNA testing may improve disease monitoring.

Introduction

Gastrointestinal cancers are responsible for a significant disease burden among United States Hispanics, particularly Puerto Rican Hispanics (PRH).^1,2 From 2016 to 2020, the age-adjusted incidences of colorectal cancer (CRC), gastric cancer, and esophageal cancer among PRH were 35.0, 5.8, and 2.5 per 100 000 population, respectively.² Meanwhile, the age-adjusted mortality rates for these cancers were 12.5, 2.2, and 1.8 per 100 000, respectively. In comparison, the overall United States population had incidences of 38.1, 6.5, and 2.8 per 100 000, with mortality rates of 13.9, 3.5, and 2.0 per 100 000, respectively.² Despite having similar incidence and mortality rates, PRH remain underrepresented in clinical studies, including those evaluating innovative surveillance methods.3, 4, 5, 6, 7 Furthermore, the genetic admixture of PRH can affect cancer susceptibility, progression, and response to treatment, highlighting the importance of research that addresses biological mechanisms contributing to gastrointestinal cancers in this underserved population.^8,9

Traditional biopsy methods are often invasive and do not necessarily provide a comprehensive description of tumor heterogeneity and disease progression. To account for this limitation, circulating tumor-specific DNA (ctDNA), a type of liquid biopsy, has emerged as a promising alternative. When cells die, they release DNA fragments into the bloodstream, and these ctDNA fragments offer a minimally invasive means to monitor tumor biology and disease burden.^10,11 The amount of ctDNA found within the bloodstream heavily depends on the overall tumor biology and burden within the patient.¹² Measuring ctDNA allows for predicting therapy response and resistance while monitoring the real-time dynamics of cancer.¹² Furthermore, as ctDNA is also shed from multiple metastatic sites, it can provide molecular insights into tumor heterogeneity,^11,12 allowing for longitudinal monitoring of tumor burden and the detection of minimal residual disease (MRD) using a blood test.¹³

In 2019, Signatera™ was granted the Breakthrough Device Designation by the Food and Drug Administration.¹⁴ This assay is used as a post-surgical intervention tool to detect and quantify ctDNA in the blood of patients with gastrointestinal, bladder, breast, and ovarian cancer, among other types, and also as part of disease surveillance and monitoring of treatment response.¹⁵ Despite these, current CRC surveillance guidelines recommend blood carcinoembryonic antigen (CEA) testing alongside other procedures, such as endoscopy, colonoscopy, and imaging studies including computed tomography (CT).^16,17 While CEA is used as a blood biomarker for treatment response and surveillance, it has lower sensitivity when compared with newer methods, such as ctDNA, which offers greater precision and early detection capabilities.16, 17, 18, 19

Our study examined the real-world utilization of ctDNA testing (Signatera™, Austin, TX) within a gastrointestinal oncology practice in a community oncology hospital among patients with CRC, gastric cancer, or esophageal cancer. The study aimed to (i) describe the adoption of ctDNA in clinical oncology practice and (ii) assess the overall diagnostic performance of ctDNA testing and its association with disease progression in stage IV CRC patients.

Patients and methods

We conducted a single-center, observational, retrospective study and reviewed the medical records of 88 patients who had undergone molecular testing with ctDNA (Signatera™) during a pre-defined calendar observation window (January 2020 through April 2022) at the Dr. Isaac Gonzalez Martinez Oncology Hospital. This study included patients aged ≥21 years who had undergone two or more ctDNA tests and had a pathology-confirmed CRC, gastric cancer, or esophageal cancer diagnosis. Molecular testing was carried out as part of the patient’s clinical evaluation using the custom-built ctDNA test from Signatera™.¹⁴ A baseline blood test was followed by serial testing within the 2 years of the study window. Baseline or the index date was defined as the first ctDNA available for each patient during the observation period. This study did not include any intervention in patient management. All ctDNA tests and diagnostic imaging were ordered by treating physicians as standard of care and abstracted from electronic medical records. Due to the study’s reliance on real-time data, not all imaging studies were available for all the patients, reflecting community clinical practices, which incorporate physician’s clinical evaluation, National Comprehensive Cancer Network (NCCN) guidelines, and health insurance coverage of imaging studies. This study was approved by the Institutional Review Board of Dr. Isaac Gonzalez Martinez Oncology Hospital (date of approval: 19 September 2023). Given its retrospective design involving the review of existing medical records, the requirement for informed consent was waived.

The workflow for the ctDNA test has been established and is provided by the manufacturer.¹⁹ Assay results were reported as integer values, with a ctDNA of 0 yielding a negative result and values >0 resulting in a positive test result. For early-stage disease (stages I-III), ctDNA status was treated as a categorical variable [detectable (>0) versus undetectable (0)]. In contrast, for metastatic disease (stage IV), ctDNA was analyzed as a continuous variable, and changes were interpreted relative to each patient’s baseline value. Patients were evaluated during the observation window (January 2020 through April 2022), during which time at least two blood ctDNA samples were obtained. Clinical assessment of disease (including recurrence) was determined by all available imaging studies [CT scans and magnetic resonance imaging (MRI)] and all medical notes within 1 year of the index ctDNA result.

Statistical analysis

Descriptive statistics, including frequencies, percentages, mean, minimum, and maximum (range), were calculated to summarize the clinicopathological characteristics of the study patients. The relationships between categorical variables, such as sex, cancer type, and cancer stage, were assessed using Pearson’s chi-square test. Fisher’s exact test was used for instances where the expected cell count was less than five. A P value <0.05 was considered statistically significant. All statistical analyses were carried out using STATA 17.²⁰ A swimmer’s plot was generated in Excel to visualize changes in ctDNA status among stage IV CRC patients over time.

Results

The clinicopathological characteristics of all study patients are presented in Tables 1 and 2. A total of 88 patients had at least two ctDNA samples collected during the study period. Most patients had CRC (80 patients; 45% males), while 5 patients (40% males) had gastric cancer, and 3 patients (100% males) had esophageal cancer. Of the 80 patients with CRC, 8 patients were classified as stage I, 22 as stage II, 28 as stage III, and 22 as stage IV disease. Of the five patients with gastric cancer, two patients had stage II, and three patients had stage III disease. Lastly, there were three patients with esophageal cancer, of which one patient had stage I, one patient had stage II, and one patient had stage IV disease. There were no statistically significant associations between clinicopathological characteristics and disease stage or status at baseline. Most patients had endocrine and metabolic comorbidities (23.2%), followed by cardiovascular comorbidities (21%) (Supplementary Tables S1 and S2, available at https://doi.org/10.1016/j.esmorw.2025.100652). The most frequent somatic alteration in the sample was KRAS, present in 41% of cases (Supplementary Table S3, available at https://doi.org/10.1016/j.esmorw.2025.100652). Additional characteristics of the sample are included in Supplementary Tables S4-S11, available at https://doi.org/10.1016/j.esmorw.2025.100652.

Table 1.

Clinicopathological characteristics by cancer type (n = 88)

Characteristics	Colorectal cancer n (%)	Gastric cancer n (%)	Esophageal cancer n (%)	P value
Sex
Male	36 (45)	2 (40)	3 (100)	0.195
Female	44 (55)	3 (60)	0 (0)
Median age (min-max)	54 (27-80)	58 (46-73)	72 (60-77)	0.149
Cancer stagea
Stage I	8 (10)	0 (0)	0 (0)	0.453
Stage II	22 (27.5)	2 (40)	3 (60)
Stage III	28 (35)	3 (60)	1 (20)
Stage IV	22 (27.5)	0 (0)	1 (20)
Treatment status at first ctDNA test
Chemotherapy/immunotherapy	21 (26.3)	0 (0)	2 (40)	0.268
Post-surgery	20 (25)	1 (33.3)	1 (20)
Pre-surgery	11 (13.8)	2 (66.7)	1 (20)
Surveillance	28 (35)	0 (0)	1 (20)

^aStaging at baseline based on the American Joint Committee on Cancer (AJCC) TNM (tumor–node–metastasis) staging system.

Table 2.

Clinicopathological characteristics by ctDNA result (n = 88)

Characteristics	Negative (n = 47) n (%)	Positive (n = 41) n (%)	P value
Sex
Male	21 (44.7)	20 (48.8)	0.701
Female	26 (55.3)	21 (51.2)
Median age (min-max)	59 (27-80)	55 (35-78)	0.558
Cancer type
Colorectal cancer	45 (95.7)	35 (85.4)	0.224
Gastric cancer	1 (2.1)	4 (9.8)
Esophageal cancer	1 (2.1)	2 (4.9)
Cancer stagea
Stage I	5 (10.6)	3 (7.3)	0.487
Stage II	11 (23.4)	14 (34.2)
Stage III	20 (42.6)	12 (29.3)
Stage IV	11 (23.4)	12 (29.3)
Treatment status at first ctDNA test
Chemotherapy/immunotherapy	3 (6.4)	20 (48.8)	<0.001
Post-surgery	14 (29.8)	8 (19.5)
Pre-surgery	4 (8.5)	10 (24.4)
Surveillance	26 (55.3)	3 (7.3)

^aStaging at baseline based on the American Joint Committee on Cancer (AJCC) TNM (tumor–node–metastasis) staging system.

Detection of ctDNA

Overall, 47 (53.4%) patients had a negative ctDNA result at baseline, while the remaining 41 (46.6%) had a positive ctDNA result (Table 2). Statistically significant differences were found between the ctDNA result and treatment status at baseline (P < 0.001). No other statistically significant associations were found between clinicopathological characteristics and ctDNA status. Additional data not shown revealed that the mean follow-up time during the study was 7.5 ± 4.9 months, ranging from 1 to 17.1 months. While all patients had at least two measurable ctDNA reports, 26.1% (n = 23) had three ctDNA tests, 13.6% (n = 12) had four ctDNA tests, and 18.2% (n = 16) had five or more ctDNA tests during the study period. The overall diagnostic accuracy of ctDNA in predicting active disease was 97.7% [95% confidence interval (CI) 94.5% to 100.0%]. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were 95.3% (95% CI 84.2% to 99.4%), 100.0% (95% CI 92.1% to 100.0%), 100.0% (95% CI 91.4% to 100.0%), and 95.7% (95% CI 85.5% to 99.5%), respectively. Among the 47 patients with negative ctDNA results, only 2 were found to have active disease confirmed by diagnostic imaging; both had CRC. Of the 41 patients with positive ctDNA results, all were found to have active disease confirmed by diagnostic imaging (100% PPV).

Colorectal cancer

Among early-stage CRC (stages I-III) patients with available imaging data (n = 36), 10 (28%) had relapsed disease, and ctDNA was positive in only 4 patients (40%), with a median lead time of 6 months (range 6-10 months). The remaining six patients (60%) with radiologic progression had undetectable ctDNA at their last test before imaging detection. Among patients with a negative ctDNA in early stages (n = 26), late relapse occurred in 5 patients (19.2%) at a median follow-up of 8 months (range 2-18 months). Among the 22 patients with CRC, 15 (68.2%) had documented radiologic progression. In all cases (n = 15, 100%), ctDNA levels increased before radiologic evidence of progression, with a median lead time of 2.5 months (range 0.5-15.0 months) (Supplementary Table S12, available at https://doi.org/10.1016/j.esmorw.2025.100652).

Figure 1 illustrates the changes in ctDNA status among the 22 patients with stage IV CRC. Each patient is represented by a line on the plot, with active disease indicated by a filled triangle. Positive ctDNA results are shown as filled circles, while negative ctDNA results are depicted as open circles. A change in ctDNA status was defined as a transition from a positive ctDNA value (>0) to a negative value (0), or vice versa, at any time point compared with the baseline. Among these patients, three experienced a change in their ctDNA status during the follow-up period. The mean time since baseline for ctDNA tests among the 22 patients with stage IV CRC is ∼5.2 ± 2.6 months.

Figure 1.

Swimmer’s plot for stage IV colorectal cancer patients. Each row represents a patient with colorectal cancer and the time from the first evaluation (baseline) to the last time point with data. For each blue filled triangle, there is a positive image modality [either magnetic resonance imaging (MRI)/computed tomography (CT) or positron emission tomography (PET)–CT] denoting disease presence. Blue open triangles represent a negative image modality that shows the absence of disease. The black filled circles represent a positive ctDNA value, and the red open circles represent a negative ctDNA value. The x-axis represents the months of surveillance, with month 1 being the baseline observation (after January 2020). Each number in the y-axis represents a patient.

Discussion

This study presents the first real-world data on using ctDNA among Hispanic patients with gastrointestinal cancer, predominantly CRC. The main objective of this study was to describe the adoption of ctDNA testing in clinical oncology practice and its association with disease progression in stage IV CRC patients. At baseline, our results show that ctDNA positivity was significantly associated with clinical status: patients under surveillance or post-surgery were more frequently ctDNA negative, whereas those undergoing active therapy or pre-surgery were more likely to be ctDNA positive. This pattern is consistent with post-operative MRD data showing that ctDNA positivity after curative-intent resection confers higher recurrence risk and detects relapse earlier than imaging.^21,22

Across the entire sample, ctDNA demonstrated a high diagnostic accuracy with a sensitivity of 95.3% and a specificity of 100%. All patients with positive ctDNA results had active disease confirmed by imaging, resulting in a PPV of 100%. These findings align with previous studies in different populations that reported high PPV for ctDNA.¹¹ However, the two patients with negative ctDNA results with active disease on imaging had aggressive molecular profiles with mutations in genes such as KRAS and BRAF. This may suggest that certain genetic mutations may affect ctDNA detectability. Moreover, in the subset of patients with stage IV CRC, ctDNA positivity demonstrated concordance with disease progression.

Previous studies have shown that ctDNA can detect tumor recurrence and progression earlier than conventional imaging methods.^14,23,24 In our cohort, stratifying CRC cases by stage revealed that the predictive value of ctDNA differed between early and advanced disease. Although imaging data were not uniformly available in early stages, nearly one in five patients with persistently negative ctDNA (19.2%) experienced late relapse during a median follow-up of 8 months. This suggests that while ctDNA negativity is reassuring in the short term, longer surveillance is required to establish its long-term predictive value. These findings are consistent with the GALAXY arm of CIRCULATE-Japan, where post-surgical ctDNA positivity emerged as the most significant prognostic factor for recurrence in stage II-III CRC.²⁵ In contrast, among late-stage metastatic CRC patients, ctDNA positivity consistently preceded radiologic progression with a median lead time of 2.5 months. Similar lead-time advantages have been reported in a nationwide Danish cohort, where ctDNA positivity preceded radiologic recurrence by a median of 3 months.²⁶ Compared with traditional biomarkers such as CEA, with sensitivity and specificity ranging from 70% to 80%, ctDNA offers superior performance for clinical decision making and patient management.^16,27, 28, 29

This study demonstrates the value and feasibility of ctDNA testing in a community oncology setting in San Juan, Puerto Rico. However, we recognize that there are limitations to our study. Firstly, its retrospective design and reliance on medical record documentation restricted the completeness of available data. Secondly, incomplete and non-standardized imaging limited the uniform validation of progression events, precluding causal inferences between ctDNA and imaging outcomes. Thirdly, ctDNA and imaging intervals were not standardized but ordered at the discretion of treating physicians, introducing heterogeneity. Although Natera recommends repeating the Signatera™ test every 3 months, we relied on real-time settings for this study. While this could be interpreted as a limitation, it mirrors actual clinical practice. Fourthly, the median follow-up was short given the constraints of the study period, limiting conclusions about recurrence metrics such as PPV and NPV. Fifthly, the predominance of CRC cases and very small numbers of gastric and esophageal cancers limit generalizability across gastrointestinal malignancies. Sixthly, among stage IV CRC cases, documentation was insufficient to distinguish oligometastatic resected patients from those with active metastatic disease, which limits the interpretation of ctDNA dynamics as MRD or treatment response.

Despite the study limitations, our data provide a real-world look at the integration of ctDNA testing in a community oncology setting among Hispanic patients with gastrointestinal cancers. These findings should be interpreted as exploratory and hypothesis-generating, underscoring the need for prospective studies with standardized protocols. Although ctDNA testing is not yet universally available in routine oncology practice, our study demonstrates its adoption in community clinics outside of academic centers. Documenting this real-world use is important, as it highlights both the opportunities and challenges of ctDNA adoption. Future work will be important to include multi-institutional, prospective studies to better evaluate the integration of ctDNA testing, particularly across diverse populations and health care settings.