PDF
Citation
Print
INTRODUCTION
Colorectal cancer (CRC) is the fourth most common malignancy and the fifth leading cause of cancer-related deaths globally among the 36 major cancers.1 The incidence of CRC is five times higher in North America, Europe, and Japan, and four times higher in South Korea and China compared to regions with the lowest incidence.1 In South Korea, CRC ranked second in incidence and third in mortality among the top ten major cancers in 2022. The US Multi-Society Task Force on CRC recommends a colonoscopy every 10 years and an annual fecal immunochemical test (FIT) as Tier 1 options for CRC screening.2 Although colonoscopy is the gold standard, it is limited by low compliance, high cost, challenges in bowel preparation, and its invasive nature with potential complications.3 FIT is used widely in organized CRC screening programs, but it also has limitations, including low participation rates, difficulties in sample collection, and a low positive predictive value (PPV) for colorectal neoplasia (CRN). Therefore, expanding CRC screening modalities to less invasive yet more effective methods is necessary.
CRC development is associated with genetic and epigenetic changes. Hypermethylation of CpG islands in gene promoters is a common epigenetic alteration in human carcinogenesis.3 DNA methylation biomarkers are stable in various sample types and offer high diagnostic sensitivity. Therefore, stool DNA methylation tests may improve CRC detection.3,4 Methylation of Syndecan-2 (meSDC2) is a frequent epigenetic alteration in CRC carcinogenesis, detectable in tissue, stool, and blood samples.5 Recently, a stool DNA-based meSDC2 test was approved for commercial use in South Korea because of its high sensitivity and specificity for CRC.3,5 A recent prospective study revealed its diagnostic performance and value in a high-risk CRC population.6 The use of this test has increased post-approval owing to its non-invasive nature and greater convenience compared to colonoscopy, and promising diagnostic performance compared to FIT.3,5,6 Nevertheless, real-world post-approval performance data are lacking.
The effectiveness of any CRC screening modality depends on the diagnostic performance and patient adherence, which can vary across populations. Currently, no data exist on adherence and diagnostic performance of stool DNA-based meSDC2 testing in average-risk individuals because previous studies primarily involved a case-control design5 or high-risk populations.6 Programmatic CRC screening― system-wide, organized screening for all members of a healthcare plan―typically targets average-risk individuals. Thus, understanding the diagnostic performance of the stool DNA-based meSDC2 test in this population is a key step toward developing a population-based CRC screening strategy using stool DNA testing.
This study evaluated the diagnostic performance of the stool DNA-based meSDC2 test for CRN in an average-risk population within a multicenter, real-world setting.
SUBJECTS AND METHODS
1. Study design
This retrospective, multicenter, consecutive cohort study included all asymptomatic, average-risk individuals who underwent a stool DNA-based meSDC2 test (EarlyTect®ColonCancer; Genomictree, Inc., Daejeon, Korea) between February 2019 and July 2023 at 18 medical institutions. The meSDC2 stool test was approved for commercial use in August 2018 in South Korea. In January 2019, online and offline education sessions were conducted across various medical institutions, covering the assay procedure, instrument usage, variable definitions, and the importance of data quality. This study was approved by the Institutional Review Board of Kyung Hee University Hospital at Gangdong, Seoul, South Korea (IRB No. KHNMC 2023-10-016), as well as by all data-providing institutions. A public Institutional Review Board approved the primary care institutions without an Institutional Review Board. Informed consent from each individual was waived because the study was retrospective in nature, and anonymous clinical data without personal identifiers were used.
2. Participants
All participants were asymptomatic, average-risk adults aged 30–80 years who underwent a meSDC2 stool test during the study period. In clinical practice, the meSDC2 stool test was recommended for individuals seeking CRC screening during routine health check-ups who wanted to avoid the inconvenience of a colonoscopy or bowel preparation. This test was also recommended for CRC screening in individuals at high risk of complications during a colonoscopy because of poor general condition, comorbid diseases, or previous failure of cecal intubation.
The age criteria in this study were consistent with the labeling from the Ministry of Food and Drug Safety of South Korea for this test. Individuals were excluded if they had any of the following high-risk factors of CRC: a previous history of CRC, inflammatory bowel disease, hereditary CRC syndromes, or signs or symptoms suggestive of CRC, such as colorectal bleeding, iron-deficiency anemia, recent changes in bowel habits, or significant weight loss. Individuals were also excluded from data analysis if they submitted inadequate stool samples (<2 g of stool, samples submitted >4 weeks after collection, or samples collected during the bowel preparation for colonoscopy) or had incomplete clinical information. An average-risk population for CRC was defined as the screening-eligible population, excluding those at higher risk based on a prior CRC diagnosis, family history of CRC, or presence of inflammatory bowel disease or Lynch syndrome.7 Therefore, all individuals included in this study were considered average-risk based on the exclusion criteria. The definition of family history has varied across studies and is often unavailable in retrospective data. Thus, individuals with a family history of CRC were not excluded.
3. Data collection
A uniform case report form, including the definitions of all variables, was used for data abstraction. The following medical information was collected: institution and physician data (hospital location, physician specialty, and hospital type), patient data (age, sex, body mass index [BMI], smoking status, alcohol use, family history of CRC, and prior colonoscopy), meSDC2 test results, and colonoscopy data (size, location, number, and histology of CRNs, withdrawal time, cecal intubation rate, and bowel preparation quality).
In clinical practice, only individuals with a positive meSDC2 stool test were recommended to undergo a confirmatory colonoscopy. All individuals were included in the primary analysis regardless of the colonoscopy preparation quality. Only the subsequent adequate colonoscopy was included in the analysis if the initial diagnostic colonoscopy was inadequate.
The meSDC2 test results were independently compared with the colonoscopy and pathology findings as reference standards. Based on the complete colonoscopy and pathology results, individuals were categorized as having any CRN, advanced neoplasia (AN), or CRC. AN was defined as CRC, advanced adenoma, or both, with advanced adenomas being defined as adenomas ≥10 mm in size, with high-grade dysplasia, or with ≥25% villous components. The lesion sizes were obtained from pathology reports and endoscopic estimates. Right-sided lesions were defined as those located at or proximal to the splenic flexure. Left-sided lesions were defined as those distal to the splenic flexure. The most advanced or largest lesion was used for classification when a participant had multiple CRN types.
4. Stool DNA-based meSDC2 test
The stool collection procedure has been described in previous studies.8,9 All participants were required to provide at least 2 g of stool using a plastic collection tube provided in the kit, with no dietary or medication restrictions. The collected fecal samples were sent to a central laboratory from each hospital according to the standard operating procedure, within four weeks of collection, and at ambient temperature. The stool DNA was isolated using either the GT Nucleic Acid PREP Kit II (Genomictree, Inc.) or the Chemagic 360-D instrument with the Chemagic DNA Stool 3k Kit H24 (PerkinElmer, Waltham, MA, USA), as reported elsewhere.6,9 The DNA concentration was measured using the Qubit dsDNA BR Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA).
The methylation status of SDC2 DNA in the stool samples was determined using a real-time polymerase chain reaction.9 The meSDC2 tests were conducted with two DNA concentrations (0.5 ng and 0.1 ng) diluted in 2 μg of meSDC2-negative stool DNA.6,8,10 The reproducibility, repeatability, lot-to-lot variation, mean CT, standard deviation, and coefficient of variation (%) have been reported in previous studies.6,8,10 The test positivity was determined based on a predefined cutoff value.6,8,10
5. Statistical analysis
The primary study endpoints included the test positivity rate, adherence to diagnostic colonoscopy, and the PPV of any CRN for the meSDC2 stool test. No statistical calculation was performed for the sample size because this was not a randomized controlled trial. Based on previous studies,5,8,10 a meSDC2 test positivity rate of approximately 4% was assumed, and it was anticipated that at least 60% of test-positive participants would undergo colonoscopy. Therefore, at least 120 colonoscopy results were expected from 5,000 meSDC2 stool tests. With 5,000 tests, including 200 positive and 4,800 negative cases with a 4% positive rate, it was calculated that the power to detect a difference between the two groups at the 5% significance level can be expected to exceed 90%.
Continuous variables were presented as the mean ± standard deviation (SD) or as medians with interquartile ranges (IQRs), and compared using the Student’s t-test or Wilcoxon rank-sum test, as appropriate. The categorical variables were reported as numbers (percentages) and compared between subgroups using Fisher’s exact test or X2 tests. All statistical tests were two-sided, and a p-value <0.05 was considered significant. Statistical analyses were conducted using the Statistical Package for the Social Sciences Korea plus 28.0 network version (IBM Co., Armonk, NY, USA).
RESULTS
A total of 4,968 tests were completed during the study period; 58 participants were excluded from data analysis (Fig. 1) for the following reasons: inadequate stool sample (n=1), insufficient clinical data (n=8), age <30 years (n=4), and age >80 years (n=45).
1. Baseline characteristics
Table 1 lists the baseline clinical characteristics of the 4,910 participants included in this analysis. The mean age was 51 years, with 48.7% and 51.3% aged 30–49 years and 50–80 years, respectively. The cohort had a male predominance (63.1%). Among the participants, 36.2%, 42.7%, and 40.4% were obese (BMI ≥25 kg/m²), alcohol drinkers, and current or former smokers, respectively. In addition, 5.9% had a positive family history of CRC, and 34.8% had no history of colonoscopy within the past five years.
2. Characteristics according to meSDC2 stool tests
Among the 4,910 participants included in the analysis, 249 tested positive, yielding an overall test-positive rate of 5.1% (Fig. 1). Table 2 lists the clinical characteristics according to meSDC2 stool test results. Participants with a positive test were more likely to be older (p<0.001), smokers (p=0.002), and have a family history of CRC (p=0.007). In particular, 49.8% of the participants with a positive test had no history of colonoscopy within the past five years.
3. Colonoscopy data
One hundred and fifty-two participants (61.0%) underwent diagnostic colonoscopy, with a median delay of 61.5 days (IQR 28.0–355.3) after receiving a positive test result. The remaining 97 participants (39.0%) did not undergo a diagnostic colonoscopy (Table 3). Of those who did, the colonoscopy data were available for 121 participants (79.6%) because the remaining 20.4% had procedures performed at other hospitals where medical records could not be accessed (Fig. 1).
The colonoscopy quality was high, with a 100% cecal intubation rate, 97.5% adequate bowel preparation, and a mean withdrawal time of 11.2 min. Neoplastic lesions were observed in 48 participants (39.7% PPV for CRN). Of these, 15 had AN (12.4% PPV for AN), and three were diagnosed with CRC (2.5% PPV for CRC). Right-sided neoplasia was present in 25 of the 48 participants with CRN (52.1% PPV for proximal CRN).
4. Predictors of colorectal neoplasia
Table 4 lists the clinical predictors of any CRN and AN among participants with a positive meSDC2 stool test. A family history of CRC was significantly associated with any CRN (p=0.029) and AN (p=0.003). Older age was associated only with any CRN (p=0.039).
DISCUSSION
Most programmatic CRC screening is based on stool tests such as FIT, whereas colonoscopy remains the dominant screening modality for opportunistic CRC screening. Although FIT is a non-invasive and simple screening test that does not require dietary or medication restriction, it is limited by its relatively lower sensitivity for detecting CRC, false negatives, reduced sample stability under high temperatures, the inconvenience of stool sampling, and the need for annual testing. Therefore, emerging tests, such as stool DNA tests, could offer higher sensitivity for CRN and allow for less frequent testing. Compared to the multi-target stool DNA test, the meSDC2 stool test provides several notable advantages (Supplementary Table 1), suggesting its potential as a promising tool for programmatic CRC screening. The major clinical implication of this study is that the diagnostic performance of a stool DNA-based meSDC2 test was evaluated in the average-risk population for CRC, suggesting this test as a potential screening modality in programmatic CRC screening.
The stool DNA-based meSDC2 test shows promise for CRN detection because methylation occurs in precancerous lesions or early-stage CRCs.5,8,10 To the best of the authors’ knowledge, this is the second largest real-world study evaluating the clinical performance of a stool DNA-based meSDC2 test in an asymptomatic, average-risk population. Most previous studies on the meSDC2 stool test were either case-control in design5,8,10 or focused on high-risk populations.6,9 Therefore, real-world, multicenter studies such as the present study are necessary to determine its utility in average-risk, asymptomatic individuals.11 In this study, 5.1% of the participants tested positive on the meSDC2 stool test, and the PPV was 39.7% and 12.4% for any CRN and AN, respectively. Among the participants with a positive meSDC2 test, those with a family history of CRC were significantly more likely to have CRN or AN. Therefore, individuals with a family history of CRC who are reluctant to undergo a colonoscopy could be recommended for the meSDC2 stool test.
Importantly, the overall PPV of the meSDC2 test for CRC in the present cohort was 2.5%, which is 2.5 times higher than the 1% PPV reported for multi-target stool DNA testing in real-world settings.12 A next-generation multi-target stool DNA test, designed to improve the specificity and reduce false positives compared to the original test approved in 2014, reported PPVs for AN and CRC of 11.1% and 3.4%, respectively.13 Thus, the diagnostic performance of the meSDC2 stool test for CRN appears promising.
The 5.1% positivity rate of the meSDC2 test was 2.8 times lower than the 14.1% reported in multi-target stool DNA testing,12 suggesting that the meSDC2 test may be a more efficient method for CRC screening with fewer false positives. In the Korean national CRC screening program, the pooled PPV of FIT for CRC over 14 years was only 2.1%,14 which is much lower than the PPV observed for the meSDC2 test in the present study. In a recent Chinese study, the meSDC2 test had a higher area under the curve (AUC) value for CRC than FIT, but the difference was not statistically significant.11 Hence, head-to-head comparisons between the meSDC2 test, FIT, and multi-target stool DNA tests in the same population are needed. In this study, 49.8% of the participants testing positive for meSDC2 had not undergone a colonoscopy within the past five years. This figure is similar to previous reports using multi-target stool DNA tests.12 Thus, stool DNA tests such as meSDC2 or multi-target DNA tests may provide a valuable gateway to colonoscopy engagement for individuals reluctant to undergo a colonoscopy. Moreover, colonoscopy following a positive meSDC2 test may result in high-quality procedures. The overall PPV for any CRN in the present cohort was 39.7%, which is approximately 15% higher than the benchmark adenoma detection rate (ADR) for average-risk screening colonoscopies.15 A meta-analysis of real-world observational studies, involving 31 studies and 3,644,561 participants, reported ADRs of only 26.5% overall, 34.4% after fecal occult blood testing, and 26.6% for primary colonoscopy screening.16 These findings reflect the suboptimal quality of a colonoscopy in average-risk populations in real-world settings.
The high PPV in the present study is likely due to CRN enrichment in the population tested (i.e., those with positive meSDC2 results) and the high quality of the colonoscopy exams. The quality indicators in this study, such as a 100% cecal intubation rate, 97.5% adequate bowel preparation, and a mean withdrawal time of 11.2 min, may reflect more careful examinations performed by endoscopists aware of the positive meSDC2 test results. Unfortunately, 39% of patients with a positive meSDC2 test result did not undergo a diagnostic colonoscopy, despite the potential risk of CRC. This may be due to patient refusal of an invasive procedure or a lack of understanding about the implications of a positive test result. Such selection bias could have resulted in an overestimation of the PPV because 39% did not undergo a colonoscopy, and 20.4% were excluded because the medical records from outside hospitals were unavailable. In a recent, prospective, multicenter, community-based study, similarly, the colonoscopy completion rate was only 32.6% after a positive meSDC2 test.11 The low colonoscopy uptake after a positive meSDC2 test is similar to the low acceptance rates of colonoscopy observed following positive FIT results. These findings highlight the need for widespread education for physicians and patients to promote optimal use of the meSDC2 stool test in CRC screening. Several evidence-based interventions could be implemented to enhance the colonoscopy completion rates following a positive meSDC2 stool test, including patient navigation programs, mitigation of the structural barriers to colonoscopy, such as cost and scheduling difficulties, and the strategic use of incentives and motivation.17,18
As a potential biomarker for CRN, the meSDC2 serum test showed 87.0% sensitivity and 95.2% specificity for CRC detection.5 Nevertheless, stool samples may be more appropriate for CRC screening because neoplastic cells exfoliate into the colon lumen before vascular invasion occurs during carcinogenesis.3 Recently, the meSDC2 stool test showed excellent performance for CRC detection, with a sensitivity, specificity, and AUC of 90.2%, 90.2%, and 0.90, respectively (95% confidence interval [CI] 0.88–0.93).8 A multicenter, prospective study in Korea focused on a high-risk CRC population.6,9 The meSDC2 stool test detected 20 CRCs out of 1,124 participants, with a sensitivity, specificity, and AUC of 95%, 81.5%, 8.6%, respectively, and a negative predictive value (NPV) of 99.9%. A study in Taiwan that included 138 participants (62 CRC patients and 76 healthy participants) reported sensitivity, specificity, PPV, NPV, and AUC values for the meSDC2 stool test of 77.4%, 88.2%, 84.2%, 82.7%, and 0.87 (95% CI 0.76–0.89), respectively.19 In a Chinese study of 497 participants (196 CRC patients, 122 with adenomas, and 179 normal participants), the AUC of the meSDC2 stool test for CRC was 0.92 (95% CI 0.89–0.95), and 81.1% of CRC cases were detected at a specificity of 93.3%.20 Nevertheless, both studies were limited by their case-control design and small sample sizes.19,20
Recently, the meSDC2 stool test was also introduced commercially in China as “Colosafe®,” with its performance evaluated in three Chinese hospitals.21 Among the 1,110 participants who underwent testing and reference colonoscopy, the sensitivity and specificity for CRC (n=359) were 83.8% and 98.0%, respectively. The test also showed 42.1% sensitivity for advanced adenomas (AA, n=38) and 87.0% sensitivity for early-stage CRC.21 Nevertheless, the results are limited because the study population was a highly selected Chinese cohort, with a high CRC detection rate (32.3%). In another study of 1,035 high-risk Chinese individuals, the sensitivity of the meSDC2 test for CRC and AA was 87.5% and 40.0%, respectively. The PPV for CRC, AA, and AN was 16.1%, 29.9%, and 46.0%, respectively.22 This study was also limited by the use of an arbitrary definition of a high-risk group, which included factors such as a family history of CRC in first-degree relatives, history of polyps, or a history of two or more of the following: chronic diarrhea, chronic constipation, mucoid bloody feces, chronic appendicitis or appendectomy, chronic cholecystitis or cholecystectomy, and psychiatric trauma.21 Hence, the performance of the meSDC2 stool test depends on the study population. Accordingly, further studies will be needed to determine the optimal target population for this promising application.
Two meta-analyses have reported the pooled diagnostic performance of meSDC2 for CRC detection.23,24 In a meta-analysis of 11 published articles, the pooled diagnostic odds ratio and AUC for meSDC2 were 52.46 (95% CI 30.43–90.45) and 0.94 (95% CI 0.92–0.96), respectively.23 This analysis, however, included meSDC2 results from stool, plasma, serum, and bowel lavage fluid. Another meta-analysis of 12 studies reported a pooled sensitivity and specificity of 0.81 (95% CI 0.74–0.86) and 0.95 (95% CI 0.93–0.96), respectively.24 This study also included stool and blood-based meSDC2 tests, which limits its applicability to stool testing alone. Therefore, new meta-analyses focusing exclusively on meSDC2 stool tests are warranted. Novel stool DNA biomarker panels that combine methylated SDC2 with other markers, such as SFRP2, KRAS mutations, and hemoglobin,25 or methylated SDC2 and SEPT9,26 have shown excellent diagnostic performance for CRC detection. These findings suggest that combining meSDC2 with other biomarkers may enhance the accuracy of CRC screening. The global incidence of early-onset CRC has been increasing in recent years, and its prognosis is generally less favorable than that of late-onset CRC.27,28 Therefore, there is a growing need for tailored screening strategies targeting the younger population. The role of non-invasive meSDC2 stool test may be gaining attention, particularly in younger populations who are less inclined to undergo invasive colonoscopy.
This study had some limitations. First, its retrospective design is a significant limitation because the test-negative participants did not undergo a colonoscopy in clinical practice. Nevertheless, this design allowed an examination of the largest number of cases in a real-world setting. Second, despite the inclusion of a multicenter, hospital-based cohort ranging from primary to tertiary care, it remains uncertain whether the data can represent the general population eligible for mass CRC screening from a different population. Furthermore, the study population was composed almost entirely of Korean individuals. Therefore, validation in other racial and ethnic groups from large-scale validation studies will be necessary to evaluate its effectiveness in that context. Third, the performance of the meSDC2 test was not compared with FIT, a widely used tool for mass CRC screening. Currently, there is only one direct comparison between the meSDC2 stool test and FIT.29 In that Chinese study,29 among 111 CRC cases, the sensitivities of the meSDC2 stool test and FIT were 79.3% (95% CI 70.5–86.4%) and 93.7% (95% CI 87.4–97.4%), respectively. Given that the previous meta-analysis has reported a pooled sensitivity of 78% for FIT in detecting CRC,30 the markedly higher sensitivity observed in the previous Chinese study indicates the need for further validation. Considering the low positivity rate (only 1.1–2.0%) of quantitative FIT in Korea’s National Screening Program31,32 and the pooled PPV of FIT for CRC,18 the meSDC2 stool test may offer advantages over FIT. Nevertheless, comparative studies between the two are needed. Fourth, the relatively small number of CRC and AN cases in the present study limits its statistical power for assessing the test performance, particularly in estimating the PPV of AN according to size, histologic features, and location. This limitation may be explained by the fact that this study focused on an average-risk, asymptomatic population, which naturally has a low prevalence of CRC and AN compared to high-risk groups. Fifth, although this study targeted an average-risk population, individuals with a family history of CRC were not excluded. A family history of CRC is typically defined as having ≥1 first-degree relative diagnosed before 60 years of age or ≥2 first-degree relatives diagnosed at any age. Individuals with a family history of CRC were not excluded from this study because such detailed information was not available in this retrospective dataset. Finally, the sensitivity and specificity of the meSDC2 test could not be calculated because the test-negative participants did not undergo colonoscopy, and these performance metrics have been reported in previous case-control studies.8
In conclusion, this large real-world study of the meSDC2 stool test in asymptomatic, average-risk individuals for CRC showed a high PPV for CRN, which may serve as a complementary modality in programmatic CRC screening. Nevertheless, these findings need to be confirmed in other racial and ethnic populations.
XML Download