Feasibility of Methylated CLIP4 in Stool for Early Detection of Colorectal Cancer: A Training Study in Chinese Population

Background Early detection of colorectal cancer (CRC) and precancerous lesion is vitally important for mitigating CRC morbidity and mortality. Aberrant DNA methylations in certain promoter regions have been identified to be closely associated with CRC development and progression, suggesting their potential as diagnostic biomarkers for early detection. In this study, we evaluated the performance of methylated CLIP4 in stool specimens as a potential biomarker for CRC detection. Methods A total of 321 subjects out of 365 enrolled participants were included in the final analysis, including 154 CRC patients, 23 advanced adenoma (AA) patients, 49 small polyp (SP) patients, and 95 healthy controls. CLIP4 methylation level was examined by qPCR with bisulfite converted DNA purified from approximately 5 g stool specimen. Results Methylated CLIP4 test showed high sensitivities of 78.3% (95% CI: 55.8%–91.7%) and 90.3% (95% CI: 84.2%–94.3%) for detecting AA and CRC, respectively, with a specificity of 88.4% (95% CI: 79.8%–93.8%). CLIP4 methylation level discriminated AA and CRC patients from control subjects with area under the curve values of 0.892 (95% CI: 0.795–0.988) and 0.961 (95% CI: 0.938–0.983). Further analysis indicated no significant difference in sensitivities among different ages, genders, stages, locations, sides, tumor sizes and differentiation statuses. Conclusions Methylated CLIP4 showed a strong potential as a noninvasive biomarker for early CRC detection.


BACKGROUND
Colorectal cancer (CRC) remains the third most commonly diagnosed cancer types, and the second most common cause of cancer-related deaths worldwide in 2020 (1). In China, new CRC cases and death in 2020 were 555,477 and 286,162, accounting for approximately 29% of the global disease burden. Rankings of CRC rose from the fifth most common cancer before 2015 to the second in both sexes. Incidence rate of CRC in China exhibited a substantial upward trend in the past decades, and agestandardized mortality rate also endured an upward swing (2). Meanwhile, significant and sustained declines in both incidence and mortality for adults over 50 years old have occurred in the United States, owing to increased awareness of screening, especially by colonoscopy (3,4).
Multiple CRC screening methods have been developed over the years, and each has its own advantages and disadvantages. For adults over 50 years old in the US, routine fecal occult blood test (FOBT) or fecal immunochemical test (FIT) (5), sigmoidoscopy, colonoscopy, computed tomography colonography or stool DNA (sDNA) test are recommended for CRC screening (3,4). Similar screening methods are recommended for high-risk group over 40 years old and lowto-average-risk group over 50 in China (6). As the gold standard, colonoscopy has higher sensitivity and specificity than stoolbased tests, especially for precancerous lesion and early stage CRC. However, the population coverage in China is still insufficient. The compliance rates for colonoscopy and FOBT remained at low levels of 4.01% and 11.01%, respectively, even in Shanghai, one of the most developed Chinese cities in the past decade (7)(8)(9). A large screening campaign of 182,927 participants with high-risk for CRC from 16 Chinese provinces only increased the compliance rate for colonoscopy to 14.0%. History of FOBT or colonic polyp, family history of CRC and high level of education were found to be associated with the increased participation (10).
In the meantime, existing stool-based tests providing noninvasive and high-compliance alternatives bear their own drawbacks. FOBT or FIT for fecal hemoglobin detection is affordable, but their performance is unsatisfactory due to low sensitivity in detecting advanced colorectal neoplasia. Cologuard, the first stool-based CRC screening test approved by the US Food and Drug Administration (FDA), demonstrated relatively high sensitivity and specificity (11). However, a high list price of $649 due to its complex operations associated with multiple assays per test makes it difficult to promote among uninsured and/or lowand moderate-income population. Therefore, intensive efforts have been made to develop more accurate and cost-effective screening tests.
DNA methylation is an epigenetic mechanism of gene regulation. Aberrant DNA methylation has been observed in all cancer types including CRC (12). Therefore, it has emerged as a class of important biomarkers with more diagnostic values than mutation markers for early CRC detection (13). A number of methylated genes have been proposed as CRC biomarker candidates in previous studies. Several among them have been incorporated into one-marker or multi-marker commercial tests, such as methylated SEPT9, SDC2, SFRP2, VIM, BMP3, and NDRG4. Another new candidate, CLIP4, is a member of CAP-Gly domain containing linker protein (CLIP) family involved in plus-end binding of microtubule, and has been implicated in immune response-related biological processes, cell migration and viability in certain cancer metastases (14,15). Hypermethylation of CLIP4 in plasma has been shown for cancer types such as CRC and gastric cancer. Further studies in multiplex blood-based methylation tests validated its potential as another promising biomarker for CRC (16)(17)(18)(19). However, the performance of methylated CLIP4 (mCLIP4) in stool samples for CRC detection has never been reported. The aim of this study was to evaluate the feasibility of stool mCLIP4 as a biomarker for early CRC detection.

Sample Collection
The original plan was to perform stool mCLIP4 test on 400 participants at the Affiliated Hospital of Xuzhou Medical University, comprising 200 CRC patients, 100 polyp patients and 100 subjects with no evidence of disease (NED). The inclusion criteria consisted of the following: 18 years old or older, no history of CRC, no pregnant woman; and all participants must have undergone complete colonoscopies by trained physicians. Standard operation was followed for all colonoscopy examinations where endoscope reached cecum. Participants with abnormal colonoscopy results should have pathological diagnoses. Pathology analysis was first done independently by two trained pathologists. If both agreed on the same diagnosis, no further evaluation was needed. If their diagnoses did not agree, evaluation by a third pathologist was required for the final determination of diagnosis. All pathologists involved were at or above the level of associated chief pathologist. During stool sample collection, transferring urine into the collection tube was avoided, and no diarrhea sample was collected. All samples were collected before purgative bowel preparation for colonoscopy. Approximately 5 g of solid specimen was collected from whole stool and preserved in 25 mL of preservative buffer (Suzhou VersaBio Technologies Co., Ltd., Kunshan, China) in a 50 mL tube to stabilize human genomic DNA. Stool samples were stored at room temperature for at most 7 days before being transferred to −80°C for longterm preservation and storage.
Until the submission of this manuscript, 365 stool specimens were collected, among which 11 were excluded due to insufficient sample information and another three were excluded due to repeated sampling. Of the remaining 351 specimens evaluated by mCLIP4 test, 30 samples were excluded due to insufficient DNA indicated by low ACTB levels (see data analysis). As a result, the final analysis included 321 specimens collected from 154 CRC patients, 23 patients with advanced adenomas (AA, an adenoma measuring ≥ 10 mm in size, with high-grade dysplasia, or with ≥ 25% villous features), 49 with small polyps (SP, non-advanced adenoma or hyperplastic polyp) and 95 NED control subjects ( Figure 1).
Fresh-frozen CRC tissues (n=28) and paired adjacent paracancerous tissues (n=28) were collected at the time of surgery at the Affiliated Hospital of Xuzhou Medical University. The details of age and gender distribution of tissue samples were described in Supplementary Table 1. All tissue samples were stored at −80°C until use.
This study was performed according to the principles of the Helsinki Declaration and approved by the Institutional Review Board of the Affiliated Hospital of Xuzhou Medical University (Ethics Committee reference number: XYKY2020-KL156-01). All participants have acknowledged and signed the informed consent.

DNA Extraction, Bisulfite Treatment and Quantitative Real-Time PCR
Tissue genomic DNA was extracted with a DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol, and purified DNA was eluted into 200 ml Buffer AE. DNA concentration was quantified with an Invitrogen NanoDrop One Microvolume UV-Vis Spectrophotometer (Thermo Fisher Scientific, Waltham, Massachusetts).
Stool specimens were first thawed for approximately 30 min at 15°C to 30°C and homogenized for 1 min on a shaking device. After centrifugation at 10,000 g for 20 min, human genomic DNA was isolated with a stool DNA extraction kit (Suzhou VersaBio Technologies Co., Ltd.) from 150 ml supernatant. Bisulfite conversion of the extracted DNA and purification of the converted DNA were performed with a bisulfite conversion kit (Suzhou VersaBio Technologies Co., Ltd.). Both kits were used according to previously published protocols (20).
Converted and purified DNA was then tested by a duplex qPCR assay. Tissue genomic DNA was tested in a single PCR reaction, and stool DNA was tested in three PCR replicates for mCLIP4 and an internal control (ACTB). The primers and probes used for mCLIP4 qPCR test were showed in Supplementary Table 2. The total reaction volume was 30 ml including 15 ml DNA. qPCR was performed on an ABI 7500 instrument (Applied Biosystems, Foster City, CA, USA) under the following conditions: initial activation at 95°C for 20 min, followed by 50 cycles at 95°C for 10 sec, 60°C for 30 sec and 72°C for 15 sec, and a final cooling to 40°C for 30 sec.
The target sequences of HCT116 and Jurkat genomic DNA were determined by bisulfite Sanger sequencing. Bisulfite-treated DNA was amplified with primers flanking the target region (Supplementary Table 2), and the expected size of the PCR product was 263 bp. PCR cycling conditions were as follows: initial activation at 95°C for 15 min, followed by 50 cycles at 95°C for 1 min, 58°C for 45 sec and 72°C for 30 sec, and a final cooling to 4°C for 1 min. PCR products were excised from agarose gels and purified with AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Hangzhou, China). Purified PCR products were ligated into pUCM-T vector (Sangon Biotech) at 16°C overnight and then transformed into competent E. coli cells (Tiangen Biotech, Beijing, China) according to the manufacturers' instructions. Randomly selected transformants were subsequently sent for Sanger sequencing with the canonical MF-13 primer by Genewiz, Inc. (Suzhou, China). The agarose gel

Limit of Detection of mCLIP4 Assay
Limit of detection (19) of mCLIP4 test was evaluated with a series of mixtures between fully methylated HCT116 genomic DNA and unmethylated Jurkat genomic DNA at different ratios (0, 6.75, 12.5, 25, 50, 60, and 70 pg fully methylated DNA out of 70 pg total DNA per qPCR reaction). The test was performed in 24 replicates for each mixture. Genomic DNA concentration was measured by an Invitrogen Qubit fluorometer and a Qubit d s D N A H S A s s a y K i t ( T h e r m o F i s h e r S c i e n t ifi c , Waltham, Massachusetts).

Data Analysis
Ct values of ACTB and mCLIP4 were obtained to validate sample processing and to determine whether mCLIP4 was detected, respectively. The results for stool specimens were considered 'valid' if all three replicate reactions for ACTB produced amplification signals and the mean Ct value was less than 40.0. To be scored positive by 3/3 algorithm, all three replicate mCLIP4 PCR reactions of a stool sample must have valid amplification curves and mean Ct value must be less than 39.0. Sensitivity was defined as the positive detection rate of CRC or AA and specificity was defined as 100% minus the positive detection rate of NED.
All statistical analyses were performed with IBM SPSS for Windows Version 22.0. Pearson chi-square test for sensitivity comparisons among groups was performance at a significant level of p < 0.05. And the differences in methylation levels were analyzed using the Mann-Whitney U test. DCt was used to determine the methylation levels of CLIP4 in tissue samples. It was defined as the difference between the Ct values of the target (mCLIP4) and the internal control gene (ACTB) to normalize for DNA amounts of tissue samples. Mean Ct values from individuals in CRC, AA and control groups were used to plot the receiver operating characteristic (21) curves and to calculate the area under the curve (22) values. Ct values of reactions returning no amplification signals were set to 50 (the maximal number of PCR cycles) for the analysis (23). Mean Ct values were also used to represent the methylation level of each plasma sample.  Table 4). Fifty-one point six percent of NED subjects and 60.4% of CRC patients were males. Across different groups, there was no significant difference for gender distribution, whereas age distribution showed significant difference ( Table 1).

Twenty
To evaluate the analytical performance of mCLIP4 test, a series of genomic DNA solutions of different methylation levels were tested in 24 replicates. As shown in Table 2, mCLIP4 test was able to detect as low as 6.75 pg fully methylated genomic DNA (~2 copies of human genome) per PCR reaction. Defined as the concentration at which more than 95% of the replicates generated amplification signals (24), LoD of mCLIP4 test was approximately 60 pg (~18 copies of human genome) per PCR reaction.
The results of tissue samples indicated that mCLIP4 levels in all cancer tissues were higher than those in their paired adjacent paracancerous tissues (p < 0.0001, Figure 2). Out of 321 subjects diagnosed by colonoscopy and further confirmed for CRC patients by pathological analysis of surgically resected specimens, mCLIP4 was detected in 11.6% of NED ( Table 4). The overall sensitivities for detecting AA and CRC by mCLIP4 test were 78.3% (95% CI: 55.8%-91.7%) and 90.3% (95% CI: 84.2%-94.3%), respectively, with a specificity of 88.4% (95% CI: 79.8%-93.8%) ( Table 3). As shown in Figure 3, mean Ct value of each group represented the average methylation level, and a lower Ct value indicated a higher methylation level. Stool mCLIP4 levels of NED were significantly lower than those of patients with intestinal lesions, including SPs, AAs, and CRCs (p < 0.0001). There were no significant differences in mCLIP4 levels between samples of SP and stage 0 CRC patients (p > 0.05), whereas Stage I-IV CRC patient samples showed significantly higher mCLIP4 levels than those of SP patients. Differences in stool mCLIP4 levels between   (Figure 4). Further analysis showed no significant sensitivity difference among different age groups, genders, stages, locations, sides, tumor sizes and differentiation statuses (p > 0.05, Table 4).

DISCUSSION
Appropriate screening and surveillance for precancerous lesion and early stage CRC can significantly mitigate CRC mortality, and AA is the preferred target stage. Coverage of guidelinerecommended screening in the US has increased to 67%: approximately 61% and 11% of US adults over 50 underwent a colonoscopy or a stool test, respectively, contributing to substantial reduction in morbidity and mortality (4). However, the invasiveness of colonoscopy and limited medical resource per capita have resulted in low compliance rate especially in averagerisk and young adults in most countries. In this study, we provided a convenient stool DNA (mCLIP4) test as an alternative screening and potential surveillance method.
The cost-effectiveness of FIT and gFOBT, two guidelinecompliant screening methods, has been intensively investigated. Previous studies showed the sensitivities of selected commercial FIT and high-sensitivity gFOBT kits for detecting advanced colorectal neoplasia and CRC varied from 7.4% to 57.1%, with relatively high specificities between 96.8% and 98.6% (25,26). Studies suggested that FIT in consecutive 3 years could play a role in significant cost savings by replacing colonoscopy, with a risk of missing 40% to 70% AAs and 30% to 40% CRCs (27). In comparison, our case-control study showed a much higher sensitivity of 78.3% and 90.3%, respectively, for detecting AA and CRC. As a result, the risk of missed diagnosis would dramatically decrease to approximately 21.7%, 4.7% or 1.0% for AA, and 9.7%, 0.9% or 0.1% for CRC, respectively, if mCLIP4 tests were performed 1, 2 or 3 times in consecutive years, indicating its potential for early CRC screening.
Numerous studies indicated better performance of DNA methylation markers in stool than those in plasma due to a limited amount of circulating tumor DNA (ctDNA) in plasma and a substantial background of circulating free DNA (cfDNA) from other sources. This was particularly true for the ability to detect precancerous lesions and CRC at early stage. In addition to better performance, stool DNA test offered a feasible solution of at-home cancer screening in populous countries with limited medical resources per capita (28). Furthermore, unlike stool DNA test, blood-based DNA test has not been included in the guideline for routine CRC screening, since its effectiveness has yet to be demonstrated in asymptomatic screening population (29). Comparisons of the performance of the same methylated DNA markers in stool to those in plasma were conducted in several studies. Epi proColon 2.0 assay, the first blood-based mSEPT9 assay approved by FDA, showed a limited sensitivity of     (24,(32)(33)(34). In comparison, mCLIP4 assay in this study showed better performance for detecting AAs by an increase of 12% to 36% in sensitivity with similar specificity for CRC detection. Furthermore, Wang et al. showed significantly lower sensitivity of 75.6% for detecting stage IV CRC with stool mSDC2 test, implying the possible preference of mSDC2 by stages. For mCLIP4 test, no such preference was observed for stage IV CRC in our limited study. In general, multi-target methylation or methylation-mutation assays were considered capable of reducing false negative rate and improving sensitivity. Cologuard, another FDA-approved molecular diagnostic test for early CRC screening, included assays for 7 K-RAS point mutations, aberrant NDRG4 and BMP3 methylation with b-actin as a reference gene and a hemoglobin immunoassay. It demonstrated sensitivities of 42.4% for AA and 92.3% for CRC with 86.6% specificity in an extensive study (11). Our previous study also evaluated a combined assay of mSEPT9 and mSDC2, ColoDefence test, resulting in sensitivities of 66.7% for AA and 92.3% for CRC with 93.2% specificity (20). Compared to these two multiplex tests, mCLIP4 test achieved an even higher sensitivity for AA by an increase of 12% to 36% with a similarly high specificity. In addition, similar to ColoDefence test, only 5 g of stool sample was required for mCLIP4 test, a single-tube multiplex qPCR assay, leading to reduction of cost and complexity of the procedure.  Stool mCLIP4 test demonstrated the feasibility for CRC detection, especially in detecting precancerous lesions and early stage CRC in our study. However, this case-control study had several limitations. First, the main purpose of this study was to evaluate the feasibility of stool mCLIP4 test for CRC detection in a training cohort of a limited number of participants. Further validation and comparison with other existing molecular diagnostic tests in future studies could provide additional support to its potential for CRC screening and prevention. Second, due to limited enrollment, characteristics of the subjects, such as age distribution in different groups, did not reflect the true distribution in a larger population. Although mCLIP4 level did not correlate with patient age in our study, future studies with larger cohorts may better define this relationship. Third, hypermethylation of CLIP4 in plasma and/ or tissues was also found in other gastrointestinal (GI) cancers, possibly leading to false positives in CRC detection in the presence of other GI cancers. Although degradation of DNA from upper gastrointestinal tract via intestine would be expected to significantly reduce the false positive results due to other GI cancers, including stool samples from such patients as control in future studies would help address this concern. Nonetheless, our findings demonstrated that mCLIP4 stool test may be a promising tool for early CRC detection.

CONCLUSIONS
Stool methylated CLIP4 test demonstrated high sensitivities in detecting SP, AA and CRC with a high specificity. Its performance on precancerous lesions and early stage CRCs made it a promising biomarker for the early detection of colorectal neoplasms. Small amount of sample needed and single-biomarker assay may also reduce screening cost. Therefore, stool mCLIP4 test has the potential to become a convenient alternative method for early CRC screening.

DATA AVAILABILITY STATEMENT
The data sets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

ETHICS STATEMENT
This study was performed according to the principles of the Helsinki Declaration and approved by the Institutional Review Board of the Affiliated Hospital of Xuzhou Medical University (Ethics Committee reference number: XYKY2020-KL156-01). All participants have acknowledged and signed the informed consent. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
YC, GZ, YPC, SX, and YM performed the statistical analyses and drafted the manuscript. YC, YPC, ZC, XL, MY, JY, XW, ZL, MZ, and SF participated in sample collection and data analysis. GZ, SX, MZ, and SF conceived of the study and participated in the design and coordination of the study. All authors contributed to the article and approved the submitted version.