- Email: [email protected]
Plausibility of an extensive use of stool DNA test for screening advanced colorectal neoplasia
Journal Pre-proofs Plausibility of an extensive use of stool DNA test for screening advanced colorectal neoplasia Jiayi Mu, Yanqin Huang, Shanrong Cai, Qilong Li, Yongmao Song, Ying Yuan, Suzhan Zhang, Shu Zheng PII: DOI: Reference:
S0009-8981(19)32163-1 https://doi.org/10.1016/j.cca.2019.12.001 CCA 15951
To appear in:
Clinica Chimica Acta
Received Date: Revised Date: Accepted Date:
4 November 2019 28 November 2019 3 December 2019
Please cite this article as: J. Mu, Y. Huang, S. Cai, Q. Li, Y. Song, Y. Yuan, S. Zhang, S. Zheng, Plausibility of an extensive use of stool DNA test for screening advanced colorectal neoplasia, Clinica Chimica Acta (2019), doi: https://doi.org/10.1016/j.cca.2019.12.001
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2019 Published by Elsevier B.V.
Plausibility of an extensive use of stool DNA test for screening advanced colorectal neoplasia Jiayi Mu1, Yanqin Huang1, Shanrong Cai1, Qilong Li2, Yongmao Song1, Ying Yuan1, Suzhan Zhang1, Shu Zheng1 1.Cancer Institute (Key Laboratory of Cancer Prevention and Intervention, China National Ministry of Education; Key Laboratory of Molecular Biology in Medical Sciences, Zhejiang Province), The Second Affiliated Hospital, Zhejiang University School of Medicine. 2.Jiashan Institute of Cancer Prevention and Treatment, Jiaxing, Zhejiang, China Corresponding author: Full name: Huang Yanqin, Email: [email protected] Tel:8657187784608 Fax:8657187214404 Address: No.88 Jiefang Road, Hangzhou, Zhejiang province, China Co-corresponding author: Full name: Shu Zheng, Email: [email protected] Tel:8657187784501 Fax:8657187214404 Address: No.88 Jiefang Road, Hangzhou, Zhejiang province, China Running Title: FIT-DNA test for CRC screening in China Funding Source: The National Key R&D Program of China No. 2016YFC1302803
Declaration of interest statement All Authors declare no conflicting interest Author’s contribution: Shu Zheng, Suzhan Zhang , Yanqin Huang designed the study and finally approved the manuscript; Ying Yuan and Yongmao Song supervised data analysis of colonoscopy and pathological results, and revised the manuscript; Qilong Li checked the surveillance data; Jiayi Mu and Shanrong Cai conducted statistical analysis; Yanqin Huang, Qilong Li, Shanrong Cai arranged collection of the screening data and the surveillance data; Jiayi Mu and Yanqin Huang drafted the manuscript and assisted all other parts of the manuscript preparation.
Abstract Purpose: FIT-DNA test is supposed to be highly sensitive for advanced colorectal neoplasms and is advocated in some developed countries, but lack extensive use in developing countries. Methods: A case control study on stool DNA test for colorectal neoplasms patients was conducted from March 2016 to October 2017 in China. We recruited CRC, colorectal neoplasms and normal controls from ambulatory patients and screening attendees in communities. The stool DNA was tested by a molecular panel similar as ColoGuard in addition to fecal immunochemical test(FIT) in a blinded manner. A risk scoring system was used to determine the positiveness of tests with histological diagnosis as its reference standard. Results: Eligible subjects included 203 colorectal cancer (CRC), 49 advanced adenoma (AA), 156 non-advanced adenoma(NAA) and 431 normal controls(NC). The FIT-DNA kit detected 97.5% CRC (n=198, 95% CI=95.4-99.7) and 53.1% AA (n=26, 95% CI=39.1-67.0), with specificity of 89.1% (95%CI=86.2-92.0) in NC and 88.1% (95%CI=85.5-90.7) in non-advanced controls. The FIT embedded in the kit alone identified 94.6% (n=192, 95% CI=91.5-97.7) CRC and 36.7% AA (n=18, 95% CI=23.2-50.2). Consistency of KRAS mutation, BMP3 methylation, NDRG4 methylation in 26 paires stool DNA and CRC tumor DNA were 80.9%, 71.4% and 81.8%, respectively Conclusion: At the sacrifice of significantly decreased specificity, a FIT-DNA kit may has better sensitivity than FIT for predicting advanced colorectal adenoma, but not for predicting colorectal cancer. More evidences are needed for the extensive use of FIT-DNA testing.
Keyword: Colorectal cancer, cancer screening tests, early detection of cancer, biomarkers
Introduction Colorectal cancer is one of the leading causes of cancer morbidity and mortality both in western countries and in China[1]. Large-scale of colonoscopy and fecal occult blood screening tests had been shown to significantly reduce the mortality and incidence of colorectal cancer in large randomized population screening trials[2-5]. Fecal immunochemical test (FIT) was recommended as an alternative to chemical based fecal blood tests which use chemicals like guaiac or benzidine[6]. Automated fecal blood analyzer was developed to facilitate the detection of a large scale processing of fecal samples during mass screening[7]. However, FIT still misses 20-30% of colorectal cancer, 70-80% of advanced colorectal adenoma and misidentifies approximately 10% of normal controls in the overall screening population as patients[8]. Direct screening by colonoscopy achieves the best sensitivity and specificity but yields over 70% of the screened subjects have no significant finding. To overcome the shortcomings of FIT and colonoscopy, many new biomarkers have been developed to predict colorectal cancer but none of them can exceed the role that FIT plays in colorectal cancer screening so far[9]. One of the major advances of new screening products was a FIT-DNA test kit, ColoGuard, which combined stool DNA test and FIT. It was found that ColoGuard had significantly better sensitivity than FIT in predicting and did not sacrifice too much of specificity. Its superiority over FIT was further demonstrated in an independent study in Alaska. However, skepticism remains with respect to the rationale of their intricate risk predicting system. It was difficult to discern individual contributions from FIT and DNA markers under such a scoring system. It was also unclear that whether the superior sensitivity of such a FIT-DNA test system could be replicated independently by other products. Here we report the findings of a case control study, which was designed to evaluate the analytical and screening performances of a FIT-DNA test kit similar to ColoGuard which was developed and applied in China.
Method 1. Study design All laboratory and clinical results were blinded until the end of the study and the data were
analyzed using an independent analyzer. The analytical performances included the distribution of gene mutation and methylation in stool DNA and the concordance of gene mutation and methylation between stool and tumor tissue DNA. The screening performances included sensitivity and specificity of the test kit for predicting CRC and advanced colorectal neoplasia. The FIT-DNA test kit was manufactured by New Horizon Health Technology Corporation Limited in Hangzhou China and marketed by the name “ColoClear” in China. A case control study was designed. Two medical centers were selected for recruiting case and control subjects, one is The Second Affiliated Hospital Zhejiang University School of Medicine (SAHZU), the other one is Cancer Prevention and Treatment Institute of Jiashan county (CPTIJ). The study was reviewed and approved by the Institutional Review Board of SAHZU in May 2015 and reported following the guidelines in STARD checklist. 2. Included and excluded criteria for the study population Colorectal cancer patients were mainly recruited from the center of SAHZU, whereas colorectal polyp patients and normal controls were mostly recruited from the center of CPTIJ. Include criteria for CRC patients were from 40 to 85 years of age at the time of diagnosis, histologically diagnosed with colon or rectal adenocarcinoma. The exclusion criteria for CRC patients were: patients with family adenomatous polyposis syndrome; patients with inflammatory bowel diseases (IBD) and patients with gastrointestinal malignancies except for colorectal cancer that might cause unexpected malignant cells in gastrointestinal tract. The inclusion criteria for colorectal polyp patients were also from 40 to 85 years of age at the time of diagnosis, underwent complete colonoscopy, confirmed diagnosis of colorectal polyp by colonoscopy and histological examination within 6 months after fecal sample collection. The requirements of age and fecal sample collection for control subjects recruiting were the same as case subjects. Normal control subjects should have complete colonoscopy with cecum intubation. Minimum of 5 grams fecal sample were collected prior to the surgical removal of tumor tissue from the intestinal tract for CRC patients and prior to the bowel preparation for colonoscopy examination for colorectal polyp patients. 3. Laboratory and clinical procedures The study procedure for each enrolled subject started from fecal sample collection. Subjects from SAHZU were selected from inpatients who then permitted fecal donations. Subjects from
CPTIJ were recruited only from community who had been scheduled for colonoscopy. All subjects have signed informed consent prior to sample collection. All laboratory procedures associated with stool DNA testing and FIT were performed in the laboratory of New Horizon Health. Briefly, fecal samples preserved by desiccation were dissolved with buffer, and immediately subject to fecal immunochemical assay. Fecal samples preserved in nucleic acid stabilization buffer were first vortexed thoroughly, followed by centrifugation and supernatant collection by standard laboratory techniques. The gene mutation and methylation detection kits as well as the risk prediction algorithm were also developed by NHH. The Ct values of KRAS gene mutation, NDRG4 and BMP3 methylation, as well as results of the fecal immunochemical test were then fed into the risk prediction model, which provided a risk score as a single output. If the score was equal to or greater than 165, the test was called “positive”. If the score was less than 165, the test was called “negative”. The details about the risk scoring system are provided in supplemental files. Colonoscopy and histological diagnosis were the reference standard to determine the accuracy of the test kit for verifying screening performances. The histology of tumor was reviewed by pathologists in SAHZU. All of the pathologists followed the diagnostic criteria of the WHO 2010 classification of gastrointestinal tumor. The stages of CRC were determined by attending surgery and histological diagnosis was made according to AJCC 7th edition of CRC TNM staging system. If multiple lesions throughout colon and rectum were found, only the most advanced lesion was evaluated. Distal lesion was defined as lesion distal to spleen flexure (including) of colon. At the end of the study, paraffin embedded tumor samples were acquired for the subjects whose stool DNA came out with positive DNA mutation or methylation results. The mutation and methylation status of DNA markers in tumor DNA were then used as reference standard for evaluating the analytical performances of the stool DNA tests. 4. Statistical analysis To evaluate the analytical performances, Ct values were exported from ABI real-time fluorescent quantitative PCR (RT-PCR) system. The analyzed Ct was the arithmetic average of three RT-PCR replications. The Ct values of beta-actin, which represent the total amount of human DNA in the stool sample, were compared by subgroup of patients. The Ct values of mutations and methylation of detected genes were normalized by the Ct values of beta-actin and
B2M respectively. The differences of the Ct values in subgroups were estimated by 95% confidence interval under normal distribution. The concordance of gene mutation and methylation in stool and tumor tissue DNA was estimated by kappa value. McNemar's test was used to analyze its sensitivity and specificity for predicting colorectal cancer and advanced colorectal neoplaisa.
Results 1. Study population The study population was recruited from 1st May 2015 to 30th October 2017. During the time of open enrollment, a total of 858 subjects were enrolled as participants and donated fecal samples. All of them were followed up until final diagnosis was made. Among cancer cases, only three subjects were excluded(One diagnosed as gastrointestinal stromal tumor, one diagnosed with neuroendocrine tumor and one belongs to familial adenomatous polyposis syndrome). Nine subjects from community were found colorectal polyps under colonoscopy but no histological result was available. One subject was diagnosed with ulcerative colitis. Six subjects were not cecum intubated. Thus, 19 subjects were excluded. A total of 839 subjects were finally included for analysis. Among them, 628 (74.9%) subjects were recruited from CPTIJ and 211 (25.1%) subjects from SAHZU. There were 203 (24.2%) colorectal adenocarcinoma patients, 49 (5.8%) advanced adenoma patients, 129 (15.4%) non-advanced adenoma patients, 27 (3.2%) non-adenomatous polyp patients and 431 (51.4%) subjects with normal colon and rectum. The average age and standard deviation for subgroups of colorectal carcinoma, advanced adenoma, non-advanced adenoma, non-adenomatous polyp and normal control were 62.0±11.7, 60.4±9.6, 59.8±8.2, 56.9±9.8 and 57.3±8.6, respectively. Taking advanced neoplasia as the cases and others as the controls, the average age of the case group was 4 years older (61.7±11.3 vs 57.8±8.6, P<0.01) than that of the control group. Male subjects accounted for 63.2% of the case group but only 48.3% of the control subjects. Among the 203 cancer cases, 29 rectum cancer cases were unable to be evaluated by TNM staging system due to preoperative neoadjuvant chemotherapy (n=27) and rectum cancer relapse(n=2). Distal lesion presented in 70.4%(143/203), 83.7%(36/49), 60.5%(78/129) and 59.3%(16/27) of the colorectal carcinomas, advanced adenomas, non-advanced adenomas and non-adenomatous polyps, respectively. Table 1 shows the
characteristics of the study population. 2. Analytical performances of the FIT-DNA test kit The internal control of KRAS mutation quantitation, β-actin, was sufficiently detected by RT-PCR in all fecal samples. The automatically generated Ct values of β-actin in stool DNAs were between 8.22 to 24.98 with a mean value of 19.2 and standard deviation of 3.1. The mean Ct value (17.2, 95%CI=16.7-17.7) of β-actin in subgroup of carcinoma patients was significantly lower than that of advanced adenoma patients (19.9, 95%CI=19.2-20.6), none advanced adenoma patients (20.3, 95%CI=19.9-20.8), none adenomatous polyp patients (20.5, 95%CI=19.5-21.5) and normal subjects (19.7, 95%CI=19.5-20.0). Compared with β-actin, the Ct values of B2M methylation which was the internal control of NDRG4 and BMP3 markers, were much lower (mean Ct 33.7, standard deviation 3.4). However, judged by the results of QPCR, the quantity of B2M methylated DNA was well correlated with the quantity of β-actin in the same fecal DNA. The correlation coefficient (r) of the Ct values of β-actin and B2M was 0.914 (P<0.001). Figure 1 shows the expression level of β-actin (blue dot) in coordination with the expression level of methylated B2M (red dot) for each stool DNA sample. Figure 2 shows a comparison of the Ct values of beta-actin in qPCR for the stool DNA samples from different subgroups. The average Ct values of beta-actin in the stool DNA samples from carcinoma patients is significantly lower than that of the rest subgroups. According to the detected CT values of KRAS, NDRG4 and BMP3, we selected 25 fecal samples for which the corresponding colorectal cancer tissues of the sample donors were acquired. Tumor DNA were extracted and tested for KRAS mutation, BMP3 and NDRG4 methylation, as well as β-actin and B2M quantitation, by the same type of test kit. A positive result of mutation and methylation was denoted by the difference of Ct values between KRAS mutation and β-actin quantitation, BMP3/NDRG4 methylation and methylated B2M quantitation respectively. A maximum CT difference of 5 was regarded as a positive result. By taking the test results of tumor tissue DNA as reference standard, we presented the true positives, true negatives, false positives and false negatives of stool DNA tests for predicting tumor tissue DNA abnormalities in table 2. NDRG4 methylation had either the best true positive (81.8%) or the worst false positive (57.1%). The kappa values of KRAS mutation, BMP3 and NDRG4 methylation between tissue DNA and stool DNA were 0.429 (P=0.022), 0.386 (P=0.045) and 0.233 (P=0.189),
respectively. 3. Performances of the FIT-DNA test kit compared with FIT By the predefined cutoff risk score of 165, a total of 294 subjects were test positive in the FIT-DNA test. The multi-target stool DNA test successfully detected 198 out of the 203 colorectal carcinoma patients and 26 out of the 49 advanced adenoma patients. The sensitivity of the test for colorectal cancer and advanced adenoma were 97.5% (198/203, 95% CI=95.4-99.7) and 53.1% (26/49, 95% CI= 39.1-67.0), respectively. A combined sensitivity for predicting advanced colorectal neoplasia was 88.9% (224/252, 95% CI=85.0-92.8). The false negative occurred in five colorectal carcinoma patients. Two of them were at TNM stage II, another two were at stage IV and one was a rectal cancer patient right after neoadjuvant therapy. Among the 49 advanced adenoma patients, 10 patients had adenomas with diameter larger than 2 cm. The multi-target stool DNA test detected 90% of those large adenomas. However, it missed 22 advanced adenomas that were below 2 cm in diameters. As for non-advanced adenomas, it only detected 18 out of the total 129 patients (14%, 95 CI=8.0-19.9). There were 40 colorectal carcinomas that located proximal to spleen flexture in the study. The FIT-DNA test kit detected all 40 proximal carcinomas. All of the missed carcinomas (n=5) located at the distal part. Specificity of the FIT-DNA test varied over the definition of control subjects. If the subjects without any neoplastic lesion were taken as the controls, 384 out of the total 431 subjects were identified as normal. The specificity was 89.1% (95% CI=86.2-92.0). If only the subjects without advanced neoplastic lesion were taken as the controls, the multi-target stool DNA test identified 517 out of the total 587 control subjects as normal. The specificity was 88.1% (95% CI=85.5-90.7). The FIT in the FIT-DNA test kit was taken out for performance analysis. FIT alone detected 94.6% (192/203, 95% CI=91.5-97.7) of all colorectal cancer cases and 36.7% (18/49, 95% CI=23.2-50.2) of all advanced adenomas. The specificity of FIT were 97.9% (422/431, 95% CI=96.6-99.3) for the control subjects without any neoplasitic lesion and 96.8% (568/587, 95% CI=95.3-98.2) for the control subjects without any advanced lesion. The sensitivity margin of the FIT-DNA test over FIT were 2.9% for predicting colorectal cancer and 16.4% for predicting advanced adenoma (Figure 3). Table 3 shows the sensitivities and specificities of the FIT-DNA test for predicting different type of colorectal neoplasia.
Discussion Using stool DNA to predict colorectal neoplasia is a challenge in terms of its accuracy and reliability[10-12]. Here we evaluated performances of a new FIT-DNA test kit, for predicting advanced colorectal neoplasia in Chinese patients and normal controls. We observed high sensitivities of the test kit for predicting colorectal cancer (97.5%, 95%CI=95.4-99.7) and advanced adenoma (53.1%, 95%CI=39.1-67.0) in this case control study. The specificity of the FIT-DNA test kit was 89.1% for the normal controls and was 88.4% for the none-advanced controls. However, the FIT in the FIT-DNA test kit alone detected 94.6% of CRC and 36.7% of advanced adenoma. The specificity of FIT in our study were high at 97.9% for the normal controls and 97.1% for the none-advanced controls. The beneficial margin of the FIT-DNA test kit over FIT reflected mainly in detecting advanced adenoma. Further analysis of the concordance between the molecular test results of tumor DNA and that of stool DNA revealed good consistency of positive for KRAS mutation (80.9%), BMP3 methylation (71.4%) and NDRG4 (81.8%). Our study suggested the positive role of stool DNA test in improving predictive sensitivity of FIT for predicting advanced adenoma, but may not for predicting colorectal cancer. The significantly decreased specificity of FIT-DNA test compared with FIT also makes the plausibility of an extensive use of stool DNA test indefensible. The FIT-DNA test, ColoGuard, was undoubtedly a milestone of stool based molecular testing in colorectal cancer screening. The developers of ColoGuard publicized the results of a large population screening trial in 2014[13]. The sensitivity of ColoGuard in that trial was 92.3% for predicting colorectal cancer and 42.4% for predicting advanced precancerous neoplasia. However, the performances of FIT that embedded in the test kit of ColoGuard was not mentioned. They presented the screening performances of an independent FIT for comparison. The sensitivity of the independent FIT were significantly lower than the FIT-DNA test for both colorectal cancer (92.3% Vs 73.8%, P<0.001) and advanced neoplasia (42.4% Vs 23.8%, P=0.004). Unlike the ColoGuard study, our study compared the screening performances of a FIT-DNA test kit with the FIT that embedded in the FIT-DNA test kit intending to reveal the contributions of stool DNA testing that beyond FIT. The results shown that the stool DNA testing in this FIT-DNA test kit
contributed approximately one third of the sensitivity for predicting advanced adenoma whereas the contribution of the stool DNA testing over FIT for predicting CRC was not as significant as that for predicting advanced adenoma. The performances of ColoGuard were later tested in another study that was carried out in Alaska[14]. The Alaska study included 661 subjects with 10 cases of colorectal cancer and a total of 92 cases of screening relevant neoplasia (SRN), which was more close to the design of our study. When compared with the Alaska study, the kit in our study detected 97.5% of CRC cases while ColoGuard detected 100% of CRC cases. ColoGuard detected 42.7% while the FIT-DNA kit detected 53.1% of advanced adenomas. For the none-advanced controls, the specificity of ColoGuard in Alaska study was 91% (95% CI, 88%-93%) while the specificity of the kit in our study was 88.1% (95% CI=85.5-90.7). Generally, the screening performances of the FIT-DNA kit were quite similar with that of ColoGuard. One of the notable aspects of the FIT-DNA kit was that it used immunogold FIT dipsticks to replace ELISA tests for hemoglobin in ColoGuard. FIT dipsticks alone in our study detected 94.6% of all CRC cases. The high sensitivity of FIT dipsticks was a little unexpected but it was still reasonable. Because the screening performs of FIT were highly fluctuating under different screening scenarios[8]. ColoGuard is a commercially marketing product with technical secret in it. Therefore, few information with respect to the analytical performances of ColoGuard other than that written in the user manual is publicly available. Here we analyzed the analytical performances of the stool DNA testing part of the FIT-DNA test kit. Our major concern was the consistency of detected mutation and methylation in stool sample DNA and the corresponding tumor tissue DNA of the sample donor. A comparison of gene abnormalities between stool DNA and tumor DNA may not be a perfect method for verifying the reliability of a stool DNA testing. It is because that there was possible interference DNA from other cells that may occur in stool samples. Yet it might be the best way. By comparing the internally controlled Ct values for 25 pairs of stool and tissue samples and taking the detected results of tumor tissue DNA as reference standard, we observed acceptable sensitivity for KRAS mutation(17/21, 80.9%), BMP3 methylation (5/7, 71.4%) and NDRG4 methylation (9/11, 81%) detection for the kit in stool DNA testing. However, the specificity of these tests was not favorable. False positives were presented in nearly 1/4 of KRAS mutation and BMP3 methylation tests. The false positive for NDRG4 methylation was even worse that nearly 60% of subjects with negative tests in tumor DNA were identified with positive tests in
stool DNA. A Korea study[15] also found good sensitivity for BMP3 and NDRG4 methylation but unfavorable specificity. The causes of false positive results are unclear. The interference of other human cells or cell debris in stool sample may cause a false positive result. Technical problems like specificity of primers and probes may cause false positive results, too. Cotter et al.[16] studied the subjects that tested positive by ColoGuard but had no significant finding under colonoscopy. They followed up 1050 subjects for a medium of 4 years and found the risks of those subjects were not increased. It supported our finding from another angle that stool DNA testing is prone to false positive and should be further improved. There were some shortcomings in our study. Firstly, the evidence level of a case control study is apparently inferior to a population based screening trial. Consequently, we cannot guarantee that the kit could perform as good in population screening as that in our study. Secondly, a selection bias may exist in recruiting advanced adenoma patients from SAHZU. The sizes of the advanced adenoma recruited from SAHZU were generally over 2 cm in diameter, which were much larger than that from screening attendees in community. It may explain that the observed sensitivity of the kit for predicting advanced adenoma in our study was higher (53.1%) than that of ColoGuard in both the screening trial (42.4%) reported in 2014 and the Alaska study (42.7%). Although our study did not support an extensive use of FIT-DNA test for colorectal cancer screening. A FIT-DNA test kit is promising in increasing detective sensitivity of colorectal precancerous lesions and promoting public awareness of colorectal cancer screening. A comparison of the detective mutation and methylation in tumor DNA and stool DNA for the same CRC patient revealed a notable consistency in mutation and methylation positive patients. It demonstrated that the high sensitivity of the kit for screening advanced adenoma was largely resulted from the identification of molecular markers in stool DNA. Although there were contradictive conclusions[17, 18] about the cost effectiveness of FIT-DNA test for CRC screening. Since there are always people that favor a more sensitive and convenient screening test, a FIT-DNA test kit might be further developed.
References
1.
Chen W, Zheng R, Baade PD, Zhang S, Zeng H, Bray F et al. Cancer statistics in China, 2015. CA Cancer J Clin 2016; 66:115-132.
2.
Mandel JS, Church TR, Ederer F, Bond JH. Colorectal cancer mortality: effectiveness of biennial screening for fecal occult blood. J Natl Cancer Inst 1999; 91:434-437.
3.
Jorgensen OD, Kronborg O, Fenger C. A randomised study of screening for colorectal cancer using faecal occult blood testing: results after 13 years and seven biennial screening rounds. Gut 2002; 50:29-32.
4.
Scholefield JH, Moss S, Sufi F, Mangham CM, Hardcastle JD. Effect of faecal occult blood screening on mortality from colorectal cancer: results from a randomised controlled trial. Gut 2002; 50:840-844.
5.
Holme O, Schoen RE, Senore C, Segnan N, Hoff G, Loberg M et al. Effectiveness of flexible sigmoidoscopy screening in men and women and different age groups: pooled analysis of randomised trials. BMJ 2017; 356:i6673.
6.
Halloran SP, Launoy G, Zappa M, International Agency for Research on C. European guidelines for quality assurance in colorectal cancer screening and diagnosis. First Edition--Faecal occult blood testing. Endoscopy 2012; 44 Suppl 3:SE65-87.
7.
Gies A, Cuk K, Schrotz-King P, Brenner H. Direct Comparison of Diagnostic Performance of 9 Quantitative Fecal Immunochemical Tests for Colorectal Cancer Screening. Gastroenterology 2018; 154:93-104.
8.
Lee JK, Liles EG, Bent S, Levin TR, Corley DA. Accuracy of fecal immunochemical tests for colorectal cancer: systematic review and meta-analysis. Ann Intern Med 2014; 160:171.
9.
Schreuders EH, Grobbee EJ, Spaander MC, Kuipers EJ. Advances in Fecal Tests for Colorectal Cancer Screening. Curr Treat Options Gastroenterol 2016; 14:152-162.
10.
Force USPST, Bibbins-Domingo K, Grossman DC, Curry SJ, Davidson KW, Epling JW, Jr. et al. Screening for Colorectal Cancer: US Preventive Services Task Force Recommendation Statement. JAMA 2016; 315:2564-2575.
11.
Knudsen AB, Zauber AG, Rutter CM, Naber SK, Doria-Rose VP, Pabiniak C et al. Estimation of Benefits, Burden, and Harms of Colorectal Cancer Screening Strategies: Modeling Study for the US Preventive Services Task Force. JAMA 2016; 315:2595-2609.
12.
Lin JS, Piper MA, Perdue LA, Rutter CM, Webber EM, O'Connor E et al. Screening for Colorectal Cancer: Updated Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA 2016; 315:2576-2594.
13.
Imperiale TF, Ransohoff DF, Itzkowitz SH, Levin TR, Lavin P, Lidgard GP et al. Multitarget stool DNA testing for colorectal-cancer screening. N Engl J Med 2014; 370:1287-1297.
14.
Redwood DG, Asay ED, Blake ID, Sacco PE, Christensen CM, Sacco FD et al. Stool DNA Testing for Screening Detection of Colorectal Neoplasia in Alaska Native People. Mayo Clin Proc 2016; 91:61-70.
15.
Park SK, Baek HL, Yu J, Kim JY, Yang HJ, Jung YS et al. Is methylation analysis of SFRP2, TFPI2, NDRG4, and BMP3 promoters suitable for colorectal cancer screening in the Korean population? Intest Res 2017; 15:495-501.
16.
Cotter TG, Burger KN, Devens ME, Simonson JA, Lowrie KL, Heigh RI et al. Long-term Follow-up of Patients Having False-Positive Multitarget Stool DNA Tests after Negative Screening Colonoscopy: The LONG-HAUL Cohort Study. Cancer Epidemiol Biomarkers Prev 2017; 26:614-621.
17.
Barzi A, Lenz HJ, Quinn DI, Sadeghi S. Comparative effectiveness of screening strategies for colorectal cancer. Cancer 2017; 123:1516-1527.
18.
Berger BM, Levin B, Hilsden RJ. Multitarget stool DNA for colorectal cancer screening: A review and commentary on the United States Preventive Services Draft Guidelines. World J Gastrointest Oncol 2016; 8:450-458.
Tables and Figures
Table 1 Characteristics of the cases and the controls Case group
Control group
Colorectal
Advanced
Non-advanced
Normal
carcinoma
colorectal neoplasia
neoplasia *
subjects
203
49
156
431
199
39
2
0
4
10
154
431
male
124
36
95
190
female
79
13
61
241
40-49
28
6
24
90
50-59
50
19
52
161
60-69
71
16
59
134
>70
54
8
21
46
proximal
40
13
60
-
distal
163
36
96
-
I
23
-
-
-
II
57
-
-
-
III
61
-
-
-
IV
34
-
-
-
28
-
-
-
Center SAHZU CPTIJ Gender
Age
location
Stage
Not available
Non-advanced Neoplasia *: 129 non-advanced neoplasia, 27 benign polyp
Table 2 Accuracy of using stool DNA to predict tumor tissue DNA markers Stool DNA
Tissue DNA (-)
Stool DNA
Tissue DNA (+) TP
FN
TN
17 KRAS mutation
21
3 4
4
(80.9) 7
13 2
18
(71.4) 11
6 2
(81.8)
5 (72.2)
9 NDRG4 methylation
1 (75.0)
5 BMP3 methylation
FP
14
8 (42.9)
* TP=true positive; FN=false negative; TN=true negative; FP=false positive;
Table 3 Comparison of sensitivity and specificity for advanced colorectal neoplasia of the multi-target stool DNA test kit and FIT kit FIT-DNA
FIT-DNA
FIT
FIT
True positive
Sensitivity (95% CI)
True positive
Sensitivity (95% CI)
203
198
97.5(95.4-99.7)
192
94.6(91.5-97.7)
Stage I
23
23
100
23
100
Stage II
57
55
96.5(91.7-100)
54
94.7(88.9-100)
Stage III
61
61
100
60
98.4(95.2-100)
Stage IV
34
32
94.1(86.2-100)
31
91.2(81.6-100)
Neoadjuvant
26
25
96.2(88.8-100)
22
84.6(70.7-98.5)
Proximal
39
38
97.4(92.5-100)
37
94.9(87.9-100)
Distal
164
160
97.6(95.2-99.9)
155
94.5(91.0-98.0)
Advanced adenoma
49
26
53.1(39.1-67.0)
18
36.7(23.2-50.2)
Large adenoma (≥2cm)
10
9
90.0(71.4-100)
5
50.0(19.0-81.0)
High dysplasia
18
12
66.7(44.9-88.4)
11
61.1(38.6-83.6)
Villous differentiation
19
10
52.6(30.2-75.1)
6
31.6(10.7-52.5)
Proximal
13
4
30.8(5.7-55.9)
2
15.4(0-35.0)
Distal
36
22
61.1(45.2-77.0)
16
44.4(28.2-60.7)
Multiple polyps
13
8
61.5(34.1-88.9)
5
38.5(11.1-65.94)
Sample size
Positive
Specificity (95% CI)
Positive
Specificity (95% CI)
Normal controls
431
47
89.1(86.9-92.5)
9
97.9(96.6-99.3)
None advanced controls
587
70
88.1(85.5-90.7)
19
96.8(95.3-98.2)
None advanced neoplasia
129
18
86.1(85.4-86.7)
8
93.8(93.3-94.3)
Benign polyp
27
5
81.5(66.5-96.4)
2
92.6(82.5-100)
Sample size Colorectal cancer
Advanced neoplasia
Figure 1 Figure legend The blue line is piled up by 839 blue dots that represent the detected quantity of beta-actin in stool DNA samples of the 839 study subjects. The X axle of a blue dot links to the index number of a stool sample. The Y axle of a blue dot shows the detected quantity of beta-actin sequences. The chart shows the maximal quantity of beta-actin sequences detected in stool DNA samples is 10 times larger than the minimum quantity of beta-actin detected. The red dot represents the detected quantity of methylated B2M sequences in stool DNA samples. The 839 blue dots in the chart vertically linked to a total of 839 red dots. The vertically linked pair of red and blue dots represents the detected quantity of beta-actin sequences and methylated B2M sequences for the same stool DNA sample, respectively. The chart shows a general consistency between the detected quantity of beta-actin sequences and methylated B2M sequences. The correlation coefficient is 0.914 (P<0.001).
Figure 2 Figure legend A comparison of the Ct values of beta-actin in qPCR for the stool DNA samples from subgroups of normal subjects, none adenomatous patients, none advanced adenoma patients, advanced adenoma patients and carcinoma patients. The vertical black bar consists of black dots. Each black dot represent the Ct value of beta-actin for the stool DNA sample of a subject. There are three horizontal lines in the middle area of every vertical black bar. The longer line represents the average Ct value of a patient subgroup. The above and below short lines represent the 95%confidence interval of the average Ct value. It shows that the average Ct values of beta-actin in the stool DNA samples from carcinoma patients is significantly (P<0.001) lower than that of the rest subgroups which means there were more human cell debris in feces of colorectal cancer patients. No significant difference was observed in any other comparative groups.
Figure 3 Figure Legend This chart shows the beneficial margin of adding stool DNA test to FIT was larger for predicting advanced adenoma (AA) than that for predicting normal control (NC), non-advanced adenoma (NA) and colorectal cancer (CRC). Statistical significance (P<0.001) was observed in the comparison of the beneficial margins for predicting AA and CRC, but failed to be observed in other comparisons.
Highlights: l l l l
Adding a panel of stool DNA molecular tests to fecal immunochemical test (FIT) significantly decreases the specificity of colorectal cancer screening. When compared with FIT, FIT-DNA test may increases sensitivity for predicting advanced adenoma but not for predicting colorectal cancer. The consistency of mutation and methylation between stool DNA and tumor DNA of the stool sample donor is good. Stool DNA testing is still a promising technology for colorectal cancer screening.
Author’s contribution: Shu Zheng, Suzhan Zhang , Yanqin Huang designed the study and finally approved the manuscript; Ying Yuan and Yongmao Song supervised data analysis of colonoscopy and pathological results, and revised the manuscript; Qilong Li checked the surveillance data; Jiayi Mu and Shanrong Cai conducted statistical analysis; Yanqin Huang, Qilong Li, Shanrong Cai arranged collection of the screening data and the surveillance data; Jiayi Mu and Yanqin Huang drafted the manuscript and assisted all other parts of the manuscript preparation.
S0009-8981(19)32163-1 https://doi.org/10.1016/j.cca.2019.12.001 CCA 15951
To appear in:
Clinica Chimica Acta
Received Date: Revised Date: Accepted Date:
4 November 2019 28 November 2019 3 December 2019
Please cite this article as: J. Mu, Y. Huang, S. Cai, Q. Li, Y. Song, Y. Yuan, S. Zhang, S. Zheng, Plausibility of an extensive use of stool DNA test for screening advanced colorectal neoplasia, Clinica Chimica Acta (2019), doi: https://doi.org/10.1016/j.cca.2019.12.001
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2019 Published by Elsevier B.V.
Plausibility of an extensive use of stool DNA test for screening advanced colorectal neoplasia Jiayi Mu1, Yanqin Huang1, Shanrong Cai1, Qilong Li2, Yongmao Song1, Ying Yuan1, Suzhan Zhang1, Shu Zheng1 1.Cancer Institute (Key Laboratory of Cancer Prevention and Intervention, China National Ministry of Education; Key Laboratory of Molecular Biology in Medical Sciences, Zhejiang Province), The Second Affiliated Hospital, Zhejiang University School of Medicine. 2.Jiashan Institute of Cancer Prevention and Treatment, Jiaxing, Zhejiang, China Corresponding author: Full name: Huang Yanqin, Email: [email protected] Tel:8657187784608 Fax:8657187214404 Address: No.88 Jiefang Road, Hangzhou, Zhejiang province, China Co-corresponding author: Full name: Shu Zheng, Email: [email protected] Tel:8657187784501 Fax:8657187214404 Address: No.88 Jiefang Road, Hangzhou, Zhejiang province, China Running Title: FIT-DNA test for CRC screening in China Funding Source: The National Key R&D Program of China No. 2016YFC1302803
Declaration of interest statement All Authors declare no conflicting interest Author’s contribution: Shu Zheng, Suzhan Zhang , Yanqin Huang designed the study and finally approved the manuscript; Ying Yuan and Yongmao Song supervised data analysis of colonoscopy and pathological results, and revised the manuscript; Qilong Li checked the surveillance data; Jiayi Mu and Shanrong Cai conducted statistical analysis; Yanqin Huang, Qilong Li, Shanrong Cai arranged collection of the screening data and the surveillance data; Jiayi Mu and Yanqin Huang drafted the manuscript and assisted all other parts of the manuscript preparation.
Abstract Purpose: FIT-DNA test is supposed to be highly sensitive for advanced colorectal neoplasms and is advocated in some developed countries, but lack extensive use in developing countries. Methods: A case control study on stool DNA test for colorectal neoplasms patients was conducted from March 2016 to October 2017 in China. We recruited CRC, colorectal neoplasms and normal controls from ambulatory patients and screening attendees in communities. The stool DNA was tested by a molecular panel similar as ColoGuard in addition to fecal immunochemical test(FIT) in a blinded manner. A risk scoring system was used to determine the positiveness of tests with histological diagnosis as its reference standard. Results: Eligible subjects included 203 colorectal cancer (CRC), 49 advanced adenoma (AA), 156 non-advanced adenoma(NAA) and 431 normal controls(NC). The FIT-DNA kit detected 97.5% CRC (n=198, 95% CI=95.4-99.7) and 53.1% AA (n=26, 95% CI=39.1-67.0), with specificity of 89.1% (95%CI=86.2-92.0) in NC and 88.1% (95%CI=85.5-90.7) in non-advanced controls. The FIT embedded in the kit alone identified 94.6% (n=192, 95% CI=91.5-97.7) CRC and 36.7% AA (n=18, 95% CI=23.2-50.2). Consistency of KRAS mutation, BMP3 methylation, NDRG4 methylation in 26 paires stool DNA and CRC tumor DNA were 80.9%, 71.4% and 81.8%, respectively Conclusion: At the sacrifice of significantly decreased specificity, a FIT-DNA kit may has better sensitivity than FIT for predicting advanced colorectal adenoma, but not for predicting colorectal cancer. More evidences are needed for the extensive use of FIT-DNA testing.
Keyword: Colorectal cancer, cancer screening tests, early detection of cancer, biomarkers
Introduction Colorectal cancer is one of the leading causes of cancer morbidity and mortality both in western countries and in China[1]. Large-scale of colonoscopy and fecal occult blood screening tests had been shown to significantly reduce the mortality and incidence of colorectal cancer in large randomized population screening trials[2-5]. Fecal immunochemical test (FIT) was recommended as an alternative to chemical based fecal blood tests which use chemicals like guaiac or benzidine[6]. Automated fecal blood analyzer was developed to facilitate the detection of a large scale processing of fecal samples during mass screening[7]. However, FIT still misses 20-30% of colorectal cancer, 70-80% of advanced colorectal adenoma and misidentifies approximately 10% of normal controls in the overall screening population as patients[8]. Direct screening by colonoscopy achieves the best sensitivity and specificity but yields over 70% of the screened subjects have no significant finding. To overcome the shortcomings of FIT and colonoscopy, many new biomarkers have been developed to predict colorectal cancer but none of them can exceed the role that FIT plays in colorectal cancer screening so far[9]. One of the major advances of new screening products was a FIT-DNA test kit, ColoGuard, which combined stool DNA test and FIT. It was found that ColoGuard had significantly better sensitivity than FIT in predicting and did not sacrifice too much of specificity. Its superiority over FIT was further demonstrated in an independent study in Alaska. However, skepticism remains with respect to the rationale of their intricate risk predicting system. It was difficult to discern individual contributions from FIT and DNA markers under such a scoring system. It was also unclear that whether the superior sensitivity of such a FIT-DNA test system could be replicated independently by other products. Here we report the findings of a case control study, which was designed to evaluate the analytical and screening performances of a FIT-DNA test kit similar to ColoGuard which was developed and applied in China.
Method 1. Study design All laboratory and clinical results were blinded until the end of the study and the data were
analyzed using an independent analyzer. The analytical performances included the distribution of gene mutation and methylation in stool DNA and the concordance of gene mutation and methylation between stool and tumor tissue DNA. The screening performances included sensitivity and specificity of the test kit for predicting CRC and advanced colorectal neoplasia. The FIT-DNA test kit was manufactured by New Horizon Health Technology Corporation Limited in Hangzhou China and marketed by the name “ColoClear” in China. A case control study was designed. Two medical centers were selected for recruiting case and control subjects, one is The Second Affiliated Hospital Zhejiang University School of Medicine (SAHZU), the other one is Cancer Prevention and Treatment Institute of Jiashan county (CPTIJ). The study was reviewed and approved by the Institutional Review Board of SAHZU in May 2015 and reported following the guidelines in STARD checklist. 2. Included and excluded criteria for the study population Colorectal cancer patients were mainly recruited from the center of SAHZU, whereas colorectal polyp patients and normal controls were mostly recruited from the center of CPTIJ. Include criteria for CRC patients were from 40 to 85 years of age at the time of diagnosis, histologically diagnosed with colon or rectal adenocarcinoma. The exclusion criteria for CRC patients were: patients with family adenomatous polyposis syndrome; patients with inflammatory bowel diseases (IBD) and patients with gastrointestinal malignancies except for colorectal cancer that might cause unexpected malignant cells in gastrointestinal tract. The inclusion criteria for colorectal polyp patients were also from 40 to 85 years of age at the time of diagnosis, underwent complete colonoscopy, confirmed diagnosis of colorectal polyp by colonoscopy and histological examination within 6 months after fecal sample collection. The requirements of age and fecal sample collection for control subjects recruiting were the same as case subjects. Normal control subjects should have complete colonoscopy with cecum intubation. Minimum of 5 grams fecal sample were collected prior to the surgical removal of tumor tissue from the intestinal tract for CRC patients and prior to the bowel preparation for colonoscopy examination for colorectal polyp patients. 3. Laboratory and clinical procedures The study procedure for each enrolled subject started from fecal sample collection. Subjects from SAHZU were selected from inpatients who then permitted fecal donations. Subjects from
CPTIJ were recruited only from community who had been scheduled for colonoscopy. All subjects have signed informed consent prior to sample collection. All laboratory procedures associated with stool DNA testing and FIT were performed in the laboratory of New Horizon Health. Briefly, fecal samples preserved by desiccation were dissolved with buffer, and immediately subject to fecal immunochemical assay. Fecal samples preserved in nucleic acid stabilization buffer were first vortexed thoroughly, followed by centrifugation and supernatant collection by standard laboratory techniques. The gene mutation and methylation detection kits as well as the risk prediction algorithm were also developed by NHH. The Ct values of KRAS gene mutation, NDRG4 and BMP3 methylation, as well as results of the fecal immunochemical test were then fed into the risk prediction model, which provided a risk score as a single output. If the score was equal to or greater than 165, the test was called “positive”. If the score was less than 165, the test was called “negative”. The details about the risk scoring system are provided in supplemental files. Colonoscopy and histological diagnosis were the reference standard to determine the accuracy of the test kit for verifying screening performances. The histology of tumor was reviewed by pathologists in SAHZU. All of the pathologists followed the diagnostic criteria of the WHO 2010 classification of gastrointestinal tumor. The stages of CRC were determined by attending surgery and histological diagnosis was made according to AJCC 7th edition of CRC TNM staging system. If multiple lesions throughout colon and rectum were found, only the most advanced lesion was evaluated. Distal lesion was defined as lesion distal to spleen flexure (including) of colon. At the end of the study, paraffin embedded tumor samples were acquired for the subjects whose stool DNA came out with positive DNA mutation or methylation results. The mutation and methylation status of DNA markers in tumor DNA were then used as reference standard for evaluating the analytical performances of the stool DNA tests. 4. Statistical analysis To evaluate the analytical performances, Ct values were exported from ABI real-time fluorescent quantitative PCR (RT-PCR) system. The analyzed Ct was the arithmetic average of three RT-PCR replications. The Ct values of beta-actin, which represent the total amount of human DNA in the stool sample, were compared by subgroup of patients. The Ct values of mutations and methylation of detected genes were normalized by the Ct values of beta-actin and
B2M respectively. The differences of the Ct values in subgroups were estimated by 95% confidence interval under normal distribution. The concordance of gene mutation and methylation in stool and tumor tissue DNA was estimated by kappa value. McNemar's test was used to analyze its sensitivity and specificity for predicting colorectal cancer and advanced colorectal neoplaisa.
Results 1. Study population The study population was recruited from 1st May 2015 to 30th October 2017. During the time of open enrollment, a total of 858 subjects were enrolled as participants and donated fecal samples. All of them were followed up until final diagnosis was made. Among cancer cases, only three subjects were excluded(One diagnosed as gastrointestinal stromal tumor, one diagnosed with neuroendocrine tumor and one belongs to familial adenomatous polyposis syndrome). Nine subjects from community were found colorectal polyps under colonoscopy but no histological result was available. One subject was diagnosed with ulcerative colitis. Six subjects were not cecum intubated. Thus, 19 subjects were excluded. A total of 839 subjects were finally included for analysis. Among them, 628 (74.9%) subjects were recruited from CPTIJ and 211 (25.1%) subjects from SAHZU. There were 203 (24.2%) colorectal adenocarcinoma patients, 49 (5.8%) advanced adenoma patients, 129 (15.4%) non-advanced adenoma patients, 27 (3.2%) non-adenomatous polyp patients and 431 (51.4%) subjects with normal colon and rectum. The average age and standard deviation for subgroups of colorectal carcinoma, advanced adenoma, non-advanced adenoma, non-adenomatous polyp and normal control were 62.0±11.7, 60.4±9.6, 59.8±8.2, 56.9±9.8 and 57.3±8.6, respectively. Taking advanced neoplasia as the cases and others as the controls, the average age of the case group was 4 years older (61.7±11.3 vs 57.8±8.6, P<0.01) than that of the control group. Male subjects accounted for 63.2% of the case group but only 48.3% of the control subjects. Among the 203 cancer cases, 29 rectum cancer cases were unable to be evaluated by TNM staging system due to preoperative neoadjuvant chemotherapy (n=27) and rectum cancer relapse(n=2). Distal lesion presented in 70.4%(143/203), 83.7%(36/49), 60.5%(78/129) and 59.3%(16/27) of the colorectal carcinomas, advanced adenomas, non-advanced adenomas and non-adenomatous polyps, respectively. Table 1 shows the
characteristics of the study population. 2. Analytical performances of the FIT-DNA test kit The internal control of KRAS mutation quantitation, β-actin, was sufficiently detected by RT-PCR in all fecal samples. The automatically generated Ct values of β-actin in stool DNAs were between 8.22 to 24.98 with a mean value of 19.2 and standard deviation of 3.1. The mean Ct value (17.2, 95%CI=16.7-17.7) of β-actin in subgroup of carcinoma patients was significantly lower than that of advanced adenoma patients (19.9, 95%CI=19.2-20.6), none advanced adenoma patients (20.3, 95%CI=19.9-20.8), none adenomatous polyp patients (20.5, 95%CI=19.5-21.5) and normal subjects (19.7, 95%CI=19.5-20.0). Compared with β-actin, the Ct values of B2M methylation which was the internal control of NDRG4 and BMP3 markers, were much lower (mean Ct 33.7, standard deviation 3.4). However, judged by the results of QPCR, the quantity of B2M methylated DNA was well correlated with the quantity of β-actin in the same fecal DNA. The correlation coefficient (r) of the Ct values of β-actin and B2M was 0.914 (P<0.001). Figure 1 shows the expression level of β-actin (blue dot) in coordination with the expression level of methylated B2M (red dot) for each stool DNA sample. Figure 2 shows a comparison of the Ct values of beta-actin in qPCR for the stool DNA samples from different subgroups. The average Ct values of beta-actin in the stool DNA samples from carcinoma patients is significantly lower than that of the rest subgroups. According to the detected CT values of KRAS, NDRG4 and BMP3, we selected 25 fecal samples for which the corresponding colorectal cancer tissues of the sample donors were acquired. Tumor DNA were extracted and tested for KRAS mutation, BMP3 and NDRG4 methylation, as well as β-actin and B2M quantitation, by the same type of test kit. A positive result of mutation and methylation was denoted by the difference of Ct values between KRAS mutation and β-actin quantitation, BMP3/NDRG4 methylation and methylated B2M quantitation respectively. A maximum CT difference of 5 was regarded as a positive result. By taking the test results of tumor tissue DNA as reference standard, we presented the true positives, true negatives, false positives and false negatives of stool DNA tests for predicting tumor tissue DNA abnormalities in table 2. NDRG4 methylation had either the best true positive (81.8%) or the worst false positive (57.1%). The kappa values of KRAS mutation, BMP3 and NDRG4 methylation between tissue DNA and stool DNA were 0.429 (P=0.022), 0.386 (P=0.045) and 0.233 (P=0.189),
respectively. 3. Performances of the FIT-DNA test kit compared with FIT By the predefined cutoff risk score of 165, a total of 294 subjects were test positive in the FIT-DNA test. The multi-target stool DNA test successfully detected 198 out of the 203 colorectal carcinoma patients and 26 out of the 49 advanced adenoma patients. The sensitivity of the test for colorectal cancer and advanced adenoma were 97.5% (198/203, 95% CI=95.4-99.7) and 53.1% (26/49, 95% CI= 39.1-67.0), respectively. A combined sensitivity for predicting advanced colorectal neoplasia was 88.9% (224/252, 95% CI=85.0-92.8). The false negative occurred in five colorectal carcinoma patients. Two of them were at TNM stage II, another two were at stage IV and one was a rectal cancer patient right after neoadjuvant therapy. Among the 49 advanced adenoma patients, 10 patients had adenomas with diameter larger than 2 cm. The multi-target stool DNA test detected 90% of those large adenomas. However, it missed 22 advanced adenomas that were below 2 cm in diameters. As for non-advanced adenomas, it only detected 18 out of the total 129 patients (14%, 95 CI=8.0-19.9). There were 40 colorectal carcinomas that located proximal to spleen flexture in the study. The FIT-DNA test kit detected all 40 proximal carcinomas. All of the missed carcinomas (n=5) located at the distal part. Specificity of the FIT-DNA test varied over the definition of control subjects. If the subjects without any neoplastic lesion were taken as the controls, 384 out of the total 431 subjects were identified as normal. The specificity was 89.1% (95% CI=86.2-92.0). If only the subjects without advanced neoplastic lesion were taken as the controls, the multi-target stool DNA test identified 517 out of the total 587 control subjects as normal. The specificity was 88.1% (95% CI=85.5-90.7). The FIT in the FIT-DNA test kit was taken out for performance analysis. FIT alone detected 94.6% (192/203, 95% CI=91.5-97.7) of all colorectal cancer cases and 36.7% (18/49, 95% CI=23.2-50.2) of all advanced adenomas. The specificity of FIT were 97.9% (422/431, 95% CI=96.6-99.3) for the control subjects without any neoplasitic lesion and 96.8% (568/587, 95% CI=95.3-98.2) for the control subjects without any advanced lesion. The sensitivity margin of the FIT-DNA test over FIT were 2.9% for predicting colorectal cancer and 16.4% for predicting advanced adenoma (Figure 3). Table 3 shows the sensitivities and specificities of the FIT-DNA test for predicting different type of colorectal neoplasia.
Discussion Using stool DNA to predict colorectal neoplasia is a challenge in terms of its accuracy and reliability[10-12]. Here we evaluated performances of a new FIT-DNA test kit, for predicting advanced colorectal neoplasia in Chinese patients and normal controls. We observed high sensitivities of the test kit for predicting colorectal cancer (97.5%, 95%CI=95.4-99.7) and advanced adenoma (53.1%, 95%CI=39.1-67.0) in this case control study. The specificity of the FIT-DNA test kit was 89.1% for the normal controls and was 88.4% for the none-advanced controls. However, the FIT in the FIT-DNA test kit alone detected 94.6% of CRC and 36.7% of advanced adenoma. The specificity of FIT in our study were high at 97.9% for the normal controls and 97.1% for the none-advanced controls. The beneficial margin of the FIT-DNA test kit over FIT reflected mainly in detecting advanced adenoma. Further analysis of the concordance between the molecular test results of tumor DNA and that of stool DNA revealed good consistency of positive for KRAS mutation (80.9%), BMP3 methylation (71.4%) and NDRG4 (81.8%). Our study suggested the positive role of stool DNA test in improving predictive sensitivity of FIT for predicting advanced adenoma, but may not for predicting colorectal cancer. The significantly decreased specificity of FIT-DNA test compared with FIT also makes the plausibility of an extensive use of stool DNA test indefensible. The FIT-DNA test, ColoGuard, was undoubtedly a milestone of stool based molecular testing in colorectal cancer screening. The developers of ColoGuard publicized the results of a large population screening trial in 2014[13]. The sensitivity of ColoGuard in that trial was 92.3% for predicting colorectal cancer and 42.4% for predicting advanced precancerous neoplasia. However, the performances of FIT that embedded in the test kit of ColoGuard was not mentioned. They presented the screening performances of an independent FIT for comparison. The sensitivity of the independent FIT were significantly lower than the FIT-DNA test for both colorectal cancer (92.3% Vs 73.8%, P<0.001) and advanced neoplasia (42.4% Vs 23.8%, P=0.004). Unlike the ColoGuard study, our study compared the screening performances of a FIT-DNA test kit with the FIT that embedded in the FIT-DNA test kit intending to reveal the contributions of stool DNA testing that beyond FIT. The results shown that the stool DNA testing in this FIT-DNA test kit
contributed approximately one third of the sensitivity for predicting advanced adenoma whereas the contribution of the stool DNA testing over FIT for predicting CRC was not as significant as that for predicting advanced adenoma. The performances of ColoGuard were later tested in another study that was carried out in Alaska[14]. The Alaska study included 661 subjects with 10 cases of colorectal cancer and a total of 92 cases of screening relevant neoplasia (SRN), which was more close to the design of our study. When compared with the Alaska study, the kit in our study detected 97.5% of CRC cases while ColoGuard detected 100% of CRC cases. ColoGuard detected 42.7% while the FIT-DNA kit detected 53.1% of advanced adenomas. For the none-advanced controls, the specificity of ColoGuard in Alaska study was 91% (95% CI, 88%-93%) while the specificity of the kit in our study was 88.1% (95% CI=85.5-90.7). Generally, the screening performances of the FIT-DNA kit were quite similar with that of ColoGuard. One of the notable aspects of the FIT-DNA kit was that it used immunogold FIT dipsticks to replace ELISA tests for hemoglobin in ColoGuard. FIT dipsticks alone in our study detected 94.6% of all CRC cases. The high sensitivity of FIT dipsticks was a little unexpected but it was still reasonable. Because the screening performs of FIT were highly fluctuating under different screening scenarios[8]. ColoGuard is a commercially marketing product with technical secret in it. Therefore, few information with respect to the analytical performances of ColoGuard other than that written in the user manual is publicly available. Here we analyzed the analytical performances of the stool DNA testing part of the FIT-DNA test kit. Our major concern was the consistency of detected mutation and methylation in stool sample DNA and the corresponding tumor tissue DNA of the sample donor. A comparison of gene abnormalities between stool DNA and tumor DNA may not be a perfect method for verifying the reliability of a stool DNA testing. It is because that there was possible interference DNA from other cells that may occur in stool samples. Yet it might be the best way. By comparing the internally controlled Ct values for 25 pairs of stool and tissue samples and taking the detected results of tumor tissue DNA as reference standard, we observed acceptable sensitivity for KRAS mutation(17/21, 80.9%), BMP3 methylation (5/7, 71.4%) and NDRG4 methylation (9/11, 81%) detection for the kit in stool DNA testing. However, the specificity of these tests was not favorable. False positives were presented in nearly 1/4 of KRAS mutation and BMP3 methylation tests. The false positive for NDRG4 methylation was even worse that nearly 60% of subjects with negative tests in tumor DNA were identified with positive tests in
stool DNA. A Korea study[15] also found good sensitivity for BMP3 and NDRG4 methylation but unfavorable specificity. The causes of false positive results are unclear. The interference of other human cells or cell debris in stool sample may cause a false positive result. Technical problems like specificity of primers and probes may cause false positive results, too. Cotter et al.[16] studied the subjects that tested positive by ColoGuard but had no significant finding under colonoscopy. They followed up 1050 subjects for a medium of 4 years and found the risks of those subjects were not increased. It supported our finding from another angle that stool DNA testing is prone to false positive and should be further improved. There were some shortcomings in our study. Firstly, the evidence level of a case control study is apparently inferior to a population based screening trial. Consequently, we cannot guarantee that the kit could perform as good in population screening as that in our study. Secondly, a selection bias may exist in recruiting advanced adenoma patients from SAHZU. The sizes of the advanced adenoma recruited from SAHZU were generally over 2 cm in diameter, which were much larger than that from screening attendees in community. It may explain that the observed sensitivity of the kit for predicting advanced adenoma in our study was higher (53.1%) than that of ColoGuard in both the screening trial (42.4%) reported in 2014 and the Alaska study (42.7%). Although our study did not support an extensive use of FIT-DNA test for colorectal cancer screening. A FIT-DNA test kit is promising in increasing detective sensitivity of colorectal precancerous lesions and promoting public awareness of colorectal cancer screening. A comparison of the detective mutation and methylation in tumor DNA and stool DNA for the same CRC patient revealed a notable consistency in mutation and methylation positive patients. It demonstrated that the high sensitivity of the kit for screening advanced adenoma was largely resulted from the identification of molecular markers in stool DNA. Although there were contradictive conclusions[17, 18] about the cost effectiveness of FIT-DNA test for CRC screening. Since there are always people that favor a more sensitive and convenient screening test, a FIT-DNA test kit might be further developed.
References
1.
Chen W, Zheng R, Baade PD, Zhang S, Zeng H, Bray F et al. Cancer statistics in China, 2015. CA Cancer J Clin 2016; 66:115-132.
2.
Mandel JS, Church TR, Ederer F, Bond JH. Colorectal cancer mortality: effectiveness of biennial screening for fecal occult blood. J Natl Cancer Inst 1999; 91:434-437.
3.
Jorgensen OD, Kronborg O, Fenger C. A randomised study of screening for colorectal cancer using faecal occult blood testing: results after 13 years and seven biennial screening rounds. Gut 2002; 50:29-32.
4.
Scholefield JH, Moss S, Sufi F, Mangham CM, Hardcastle JD. Effect of faecal occult blood screening on mortality from colorectal cancer: results from a randomised controlled trial. Gut 2002; 50:840-844.
5.
Holme O, Schoen RE, Senore C, Segnan N, Hoff G, Loberg M et al. Effectiveness of flexible sigmoidoscopy screening in men and women and different age groups: pooled analysis of randomised trials. BMJ 2017; 356:i6673.
6.
Halloran SP, Launoy G, Zappa M, International Agency for Research on C. European guidelines for quality assurance in colorectal cancer screening and diagnosis. First Edition--Faecal occult blood testing. Endoscopy 2012; 44 Suppl 3:SE65-87.
7.
Gies A, Cuk K, Schrotz-King P, Brenner H. Direct Comparison of Diagnostic Performance of 9 Quantitative Fecal Immunochemical Tests for Colorectal Cancer Screening. Gastroenterology 2018; 154:93-104.
8.
Lee JK, Liles EG, Bent S, Levin TR, Corley DA. Accuracy of fecal immunochemical tests for colorectal cancer: systematic review and meta-analysis. Ann Intern Med 2014; 160:171.
9.
Schreuders EH, Grobbee EJ, Spaander MC, Kuipers EJ. Advances in Fecal Tests for Colorectal Cancer Screening. Curr Treat Options Gastroenterol 2016; 14:152-162.
10.
Force USPST, Bibbins-Domingo K, Grossman DC, Curry SJ, Davidson KW, Epling JW, Jr. et al. Screening for Colorectal Cancer: US Preventive Services Task Force Recommendation Statement. JAMA 2016; 315:2564-2575.
11.
Knudsen AB, Zauber AG, Rutter CM, Naber SK, Doria-Rose VP, Pabiniak C et al. Estimation of Benefits, Burden, and Harms of Colorectal Cancer Screening Strategies: Modeling Study for the US Preventive Services Task Force. JAMA 2016; 315:2595-2609.
12.
Lin JS, Piper MA, Perdue LA, Rutter CM, Webber EM, O'Connor E et al. Screening for Colorectal Cancer: Updated Evidence Report and Systematic Review for the US Preventive Services Task Force. JAMA 2016; 315:2576-2594.
13.
Imperiale TF, Ransohoff DF, Itzkowitz SH, Levin TR, Lavin P, Lidgard GP et al. Multitarget stool DNA testing for colorectal-cancer screening. N Engl J Med 2014; 370:1287-1297.
14.
Redwood DG, Asay ED, Blake ID, Sacco PE, Christensen CM, Sacco FD et al. Stool DNA Testing for Screening Detection of Colorectal Neoplasia in Alaska Native People. Mayo Clin Proc 2016; 91:61-70.
15.
Park SK, Baek HL, Yu J, Kim JY, Yang HJ, Jung YS et al. Is methylation analysis of SFRP2, TFPI2, NDRG4, and BMP3 promoters suitable for colorectal cancer screening in the Korean population? Intest Res 2017; 15:495-501.
16.
Cotter TG, Burger KN, Devens ME, Simonson JA, Lowrie KL, Heigh RI et al. Long-term Follow-up of Patients Having False-Positive Multitarget Stool DNA Tests after Negative Screening Colonoscopy: The LONG-HAUL Cohort Study. Cancer Epidemiol Biomarkers Prev 2017; 26:614-621.
17.
Barzi A, Lenz HJ, Quinn DI, Sadeghi S. Comparative effectiveness of screening strategies for colorectal cancer. Cancer 2017; 123:1516-1527.
18.
Berger BM, Levin B, Hilsden RJ. Multitarget stool DNA for colorectal cancer screening: A review and commentary on the United States Preventive Services Draft Guidelines. World J Gastrointest Oncol 2016; 8:450-458.
Tables and Figures
Table 1 Characteristics of the cases and the controls Case group
Control group
Colorectal
Advanced
Non-advanced
Normal
carcinoma
colorectal neoplasia
neoplasia *
subjects
203
49
156
431
199
39
2
0
4
10
154
431
male
124
36
95
190
female
79
13
61
241
40-49
28
6
24
90
50-59
50
19
52
161
60-69
71
16
59
134
>70
54
8
21
46
proximal
40
13
60
-
distal
163
36
96
-
I
23
-
-
-
II
57
-
-
-
III
61
-
-
-
IV
34
-
-
-
28
-
-
-
Center SAHZU CPTIJ Gender
Age
location
Stage
Not available
Non-advanced Neoplasia *: 129 non-advanced neoplasia, 27 benign polyp
Table 2 Accuracy of using stool DNA to predict tumor tissue DNA markers Stool DNA
Tissue DNA (-)
Stool DNA
Tissue DNA (+) TP
FN
TN
17 KRAS mutation
21
3 4
4
(80.9) 7
13 2
18
(71.4) 11
6 2
(81.8)
5 (72.2)
9 NDRG4 methylation
1 (75.0)
5 BMP3 methylation
FP
14
8 (42.9)
* TP=true positive; FN=false negative; TN=true negative; FP=false positive;
Table 3 Comparison of sensitivity and specificity for advanced colorectal neoplasia of the multi-target stool DNA test kit and FIT kit FIT-DNA
FIT-DNA
FIT
FIT
True positive
Sensitivity (95% CI)
True positive
Sensitivity (95% CI)
203
198
97.5(95.4-99.7)
192
94.6(91.5-97.7)
Stage I
23
23
100
23
100
Stage II
57
55
96.5(91.7-100)
54
94.7(88.9-100)
Stage III
61
61
100
60
98.4(95.2-100)
Stage IV
34
32
94.1(86.2-100)
31
91.2(81.6-100)
Neoadjuvant
26
25
96.2(88.8-100)
22
84.6(70.7-98.5)
Proximal
39
38
97.4(92.5-100)
37
94.9(87.9-100)
Distal
164
160
97.6(95.2-99.9)
155
94.5(91.0-98.0)
Advanced adenoma
49
26
53.1(39.1-67.0)
18
36.7(23.2-50.2)
Large adenoma (≥2cm)
10
9
90.0(71.4-100)
5
50.0(19.0-81.0)
High dysplasia
18
12
66.7(44.9-88.4)
11
61.1(38.6-83.6)
Villous differentiation
19
10
52.6(30.2-75.1)
6
31.6(10.7-52.5)
Proximal
13
4
30.8(5.7-55.9)
2
15.4(0-35.0)
Distal
36
22
61.1(45.2-77.0)
16
44.4(28.2-60.7)
Multiple polyps
13
8
61.5(34.1-88.9)
5
38.5(11.1-65.94)
Sample size
Positive
Specificity (95% CI)
Positive
Specificity (95% CI)
Normal controls
431
47
89.1(86.9-92.5)
9
97.9(96.6-99.3)
None advanced controls
587
70
88.1(85.5-90.7)
19
96.8(95.3-98.2)
None advanced neoplasia
129
18
86.1(85.4-86.7)
8
93.8(93.3-94.3)
Benign polyp
27
5
81.5(66.5-96.4)
2
92.6(82.5-100)
Sample size Colorectal cancer
Advanced neoplasia
Figure 1 Figure legend The blue line is piled up by 839 blue dots that represent the detected quantity of beta-actin in stool DNA samples of the 839 study subjects. The X axle of a blue dot links to the index number of a stool sample. The Y axle of a blue dot shows the detected quantity of beta-actin sequences. The chart shows the maximal quantity of beta-actin sequences detected in stool DNA samples is 10 times larger than the minimum quantity of beta-actin detected. The red dot represents the detected quantity of methylated B2M sequences in stool DNA samples. The 839 blue dots in the chart vertically linked to a total of 839 red dots. The vertically linked pair of red and blue dots represents the detected quantity of beta-actin sequences and methylated B2M sequences for the same stool DNA sample, respectively. The chart shows a general consistency between the detected quantity of beta-actin sequences and methylated B2M sequences. The correlation coefficient is 0.914 (P<0.001).
Figure 2 Figure legend A comparison of the Ct values of beta-actin in qPCR for the stool DNA samples from subgroups of normal subjects, none adenomatous patients, none advanced adenoma patients, advanced adenoma patients and carcinoma patients. The vertical black bar consists of black dots. Each black dot represent the Ct value of beta-actin for the stool DNA sample of a subject. There are three horizontal lines in the middle area of every vertical black bar. The longer line represents the average Ct value of a patient subgroup. The above and below short lines represent the 95%confidence interval of the average Ct value. It shows that the average Ct values of beta-actin in the stool DNA samples from carcinoma patients is significantly (P<0.001) lower than that of the rest subgroups which means there were more human cell debris in feces of colorectal cancer patients. No significant difference was observed in any other comparative groups.
Figure 3 Figure Legend This chart shows the beneficial margin of adding stool DNA test to FIT was larger for predicting advanced adenoma (AA) than that for predicting normal control (NC), non-advanced adenoma (NA) and colorectal cancer (CRC). Statistical significance (P<0.001) was observed in the comparison of the beneficial margins for predicting AA and CRC, but failed to be observed in other comparisons.
Highlights: l l l l
Adding a panel of stool DNA molecular tests to fecal immunochemical test (FIT) significantly decreases the specificity of colorectal cancer screening. When compared with FIT, FIT-DNA test may increases sensitivity for predicting advanced adenoma but not for predicting colorectal cancer. The consistency of mutation and methylation between stool DNA and tumor DNA of the stool sample donor is good. Stool DNA testing is still a promising technology for colorectal cancer screening.
Author’s contribution: Shu Zheng, Suzhan Zhang , Yanqin Huang designed the study and finally approved the manuscript; Ying Yuan and Yongmao Song supervised data analysis of colonoscopy and pathological results, and revised the manuscript; Qilong Li checked the surveillance data; Jiayi Mu and Shanrong Cai conducted statistical analysis; Yanqin Huang, Qilong Li, Shanrong Cai arranged collection of the screening data and the surveillance data; Jiayi Mu and Yanqin Huang drafted the manuscript and assisted all other parts of the manuscript preparation.