Fecal DNA testing for colorectal neoplasia in IBD: could it be as simple as a stool study?

Kisiel JB, Yab TC, Nazer Hussain FT, et al. Stool DNA testing for the detection of colorectal neoplasia in patients with inflammatory bowel disease. Aliment Pharmacol Ther 2013;37:546–554.Patients with long-standing inflammatory bowel disease (IBD) in the colon are at increased risk for dysplasia and colorectal cancer (Cancer 1992;70[suppl 5]:1313–1316; Gastroenterology 1991;100[5 Pt 1]:1241). Owing to this increased risk, IBD patients typically undergo surveillance colonoscopy to detect colorectal neoplasia (CRN): Colorectal cancer, high-grade dysplasia (HGD), or low-grade dysplasia found in random biopsies of colonic mucosa. New techniques such as chromoendoscopy with targeted biopsies have shown to be an alternative to improve the yield of surveillance colonoscopy in these patients. Despite these advances, there remain limitations to current surveillance tools, which include undersampling and unknown ideal frequency of surveillance. Stool assay of exfoliated DNA markers for CRN may represent a noninvasive adjunctive tool for the detection of CRN in IBD patients. In non-IBD patients, the feasibility of detecting tumor-specific products in the stool has been demonstrated (Gastroenterology 2005;128:192–206; Cancer 2006;106:277–283). Prior tissue-based studies have shown that there are various molecular alterations, such as acquired mutations or abnormal methylations, associated with IBD-CRN.The primary aims of this study were to assess the discriminant value of 5 mutation markers (p53, APC, BRAF, K-ras, and PIK3CA) and 4 methylation markers (VIM, BMP3, EYA4, septin 9) based on DNA extracted from tissue specimens, and to assess the feasibility of stool DNA testing for detection of premalignant and malignant IBD-CRN. A single-center archive of IBD-CRC cases and IBD controls were used to identify tissue specimens. Cases and controls were matched for age, gender, disease duration, anatomic extent, and primary sclerosing cholangitis (PSC) status. DNA was then extracted from paraffin-embedded tissues and amplified using real-time polymerase chain reaction.The 25 case and 25 control patients in the tissue study were well-matched in terms of clinical characteristics including age, gender, duration and extent of disease, and PSC cases. DNA sequencing for the case samples revealed 14 mutation markers across 3 genes (APC, K-ras, and p53; no mutations were identified on BRAF or PIK3CA) with p53 found to be the most informative marker. The aggregate sensitivity of using all mutation markers was 60% with a specificity of 100%; there were no mutations found in control tissues. Receiver operating characteristic curves for each of the methylation markers demonstrated higher areas under the curve for methylated EYA4 (mEYA4), VIM (mVIM), and BMP3 (mBMP3).Given the results of the tissue study, case and control sets were analyzed for detection of CRN by stool assay of 3 methylated genes: BMP3, Vimentin, and EYA4. The fourth methylated gene assayed was NDGR4, which was selected because of high discrimination for sporadic CRN in studies performed after the tissue study. A total of 23 cases and 220 controls were found to be eligible. Of these, 19 case patients with biopsy-proven CRN and 35 control patients without CRN submitted stool. In a comparison of the case and control sets, case patients were significantly more likely to have extensive disease involvement and longer disease duration. There was no difference in disease activity between the 2 groups.Of the 19 cases with confirmed CRN, 9 patients had cancer with a medium size of 2.3 cm, 4 had HGD, and 6 had low-grade dysplasia. All 4 methylated genes showed high discrimination for cancer. Stool assay of mBMP3 alone was 100% sensitive for CRC, 84% for all CRN, and 91% specific. The combination of mBMP3 and mNDRG4 was 100% sensitive for both CRC and HGD with 89% specificity. In comparison of patients with ulcerative colitis (UC) and Crohn’s disease, Marker levels were not different between UC and Crohn’s disease (CD) patients. In multivariate analyses including age, gender, disease duration, disease extent, and presence of PSC, methylation markers for CRN detection remained significant.CommentOver the past decade, there have been several sets of guidelines regarding colon cancer screening and surveillance in patients with UC (Gut 2002;51[suppl 5]:V10–V12; Gastroenterology 2003;124:544–560; Gastrointest Endosc 2006;63:546–557; Inflamm Bowel Dis 2005;11:314–321; Am J Gastroenterol 2010;105:501–524; Gut 2010;59:666–689). These recommendations all include the use of colonoscopy. Unfortunately, the prevalence of colonoscopy surveillance to detect CRN ranges from 25% to 64% (Gastroenterology 2010;139:1511–1518; Gastrointest Endosc 2007;65:432–439; World J Gastroenterol 2009;15:226–230). Recent studies investigating the low rate of surveillance have found that factors associated with higher compliance include the absence of significant comorbidity and more than 3 UC-related outpatient visits (Gastroenterology 2010;139:1511–1518). Other factors affecting low surveillance rate may include patient concerns regarding risks of colonoscopy and the inconvenience of bowel prep or fasting before the procedure.The availability of a noninvasive test may serve to increase compliance with colorectal cancer screening and surveillance in IBD patients. The most common noninvasive approach to screening in the general population has been the detection of occult blood, which lacks sensitivity and specificity, and is unreliable in IBD patients given the inflammatory nature of the disease. Many cancers or premalignant adenomas do not bleed and can be missed (JAMA 1990;263:76–78; JAMA 1993;269:1262–1267). The genetics of CRC provide the basis for analysis of fecal DNA (Science 1992;256:102–105; Clin Chim Acta 2002;315:157–168). Altered DNA from a CRN arises from the neoplasm itself as opposed to circulation and is released via exfoliation. Neoplasm-specific DNA alterations have been well characterized (Eur J Gastroenterol Hepatol 1992;5:205–213; Eur J Gastroenterol Hepatol 1993;5:219–225). From this, a stool assay system was developed to identify a spectrum of DNA alterations that are associated with CRN (Gastroenterology 2000;119:1219–1227).Several studies have demonstrated the feasibility of detecting tumor-specific products in the stool (Gastroenterology 2005;128:192–206; Cancer 2006;106:277–283). Stool analyzed in blinded fashion from patients with CRC, adenomas, or normal colon revealed even an early generation fecal DNA assay was 91% sensitive for cancer and 82% for adenomas >1 cm with a specificity of 91% (Gastroenterology 2000;119:1219–1227). A large, prospective, multicenter study in >4000 average-risk asymptomatic patients over the age of 50 showed that fecal DNA test had a significantly higher sensitivity compared with stool Hemoccult test (N Engl J Med 2004;351:2704–2714).The authors of this study have addressed an interesting and important question. Patients with UC are at increased risk for developing CRC; however, the rate of surveillance colonoscopies within this group is highly variable. Even with current surveillance techniques, colonoscopy is an imperfect practice. The authors found 2 of the 9 CRC cases with positive stool results were missed on colonoscopy and diagnosed only after colectomy. The use of fecal DNA testing could be integral to the care of IBD patients. The design of this study was a major strength. In the tissue study, cases and controls were well-matched and the authors were able to determine which methylation markers were highly discriminant for IBD-CRN as opposed to relying on and extrapolating from prior studies performed with sporadic, non-IBD cases. They were also able to confirm prior observations in IBD cases with regard to mBMP3 and mEYA4. Based on the results of the tissue study, the authors demonstrated that fecal DNA testing was capable of achieving high detection rates for both CRC and dysplasia.The authors were able to achieve the target sample size to address the primary endpoints. The power to analyze subgroups, however, was limited, and the authors acknowledge this. A particular area of interest would have been to determine any differences in stool DNA markers between flat, dysplastic lesions from other dysplastic lesions. The finding of flat HGD or carcinoma on endoscopic biopsies is an indication for colectomy. Although controversial, there is some evidence to suggest that flat, low-grade dysplasia is also an indication for colectomy because of the high rate of progression to HGD or cancer (Am J Gastroenterol 2002;97:922–927; Gastroenterology 2003;125:1311–1319; Gut 2003;52:1127–1132; Dis Colon Rectum 2002;45:615–620; Gastroenterology 2004; 127:950–956). The sensitivity for low-grade dysplasia in the fecal DNA assay used in this study was at best 67%. Furthermore, owing to limited power in the subgroup analyses, the lack of an effect upon marker levels by disease duration, severity and extent of disease, and presence of concomitant PSC could be falsely negative.Overall, fecal DNA testing seems to be a promising alternative to colonoscopy for screening and surveillance purposes. Stool assay may be useful for early detection of IBD-CRN, and could also help to guide the frequency of colonoscopy surveillance to tailor individual risk. Larger studies are needed, however, to confirm the results demonstrated in this study, as well as to provide the power for greater sensitivity in subgroup analyses. Kisiel JB, Yab TC, Nazer Hussain FT, et al. Stool DNA testing for the detection of colorectal neoplasia in patients with inflammatory bowel disease. Aliment Pharmacol Ther 2013;37:546–554. Patients with long-standing inflammatory bowel disease (IBD) in the colon are at increased risk for dysplasia and colorectal cancer (Cancer 1992;70[suppl 5]:1313–1316; Gastroenterology 1991;100[5 Pt 1]:1241). Owing to this increased risk, IBD patients typically undergo surveillance colonoscopy to detect colorectal neoplasia (CRN): Colorectal cancer, high-grade dysplasia (HGD), or low-grade dysplasia found in random biopsies of colonic mucosa. New techniques such as chromoendoscopy with targeted biopsies have shown to be an alternative to improve the yield of surveillance colonoscopy in these patients. Despite these advances, there remain limitations to current surveillance tools, which include undersampling and unknown ideal frequency of surveillance. Stool assay of exfoliated DNA markers for CRN may represent a noninvasive adjunctive tool for the detection of CRN in IBD patients. In non-IBD patients, the feasibility of detecting tumor-specific products in the stool has been demonstrated (Gastroenterology 2005;128:192–206; Cancer 2006;106:277–283). Prior tissue-based studies have shown that there are various molecular alterations, such as acquired mutations or abnormal methylations, associated with IBD-CRN. The primary aims of this study were to assess the discriminant value of 5 mutation markers (p53, APC, BRAF, K-ras, and PIK3CA) and 4 methylation markers (VIM, BMP3, EYA4, septin 9) based on DNA extracted from tissue specimens, and to assess the feasibility of stool DNA testing for detection of premalignant and malignant IBD-CRN. A single-center archive of IBD-CRC cases and IBD controls were used to identify tissue specimens. Cases and controls were matched for age, gender, disease duration, anatomic extent, and primary sclerosing cholangitis (PSC) status. DNA was then extracted from paraffin-embedded tissues and amplified using real-time polymerase chain reaction. The 25 case and 25 control patients in the tissue study were well-matched in terms of clinical characteristics including age, gender, duration and extent of disease, and PSC cases. DNA sequencing for the case samples revealed 14 mutation markers across 3 genes (APC, K-ras, and p53; no mutations were identified on BRAF or PIK3CA) with p53 found to be the most informative marker. The aggregate sensitivity of using all mutation markers was 60% with a specificity of 100%; there were no mutations found in control tissues. Receiver operating characteristic curves for each of the methylation markers demonstrated higher areas under the curve for methylated EYA4 (mEYA4), VIM (mVIM), and BMP3 (mBMP3). Given the results of the tissue study, case and control sets were analyzed for detection of CRN by stool assay of 3 methylated genes: BMP3, Vimentin, and EYA4. The fourth methylated gene assayed was NDGR4, which was selected because of high discrimination for sporadic CRN in studies performed after the tissue study. A total of 23 cases and 220 controls were found to be eligible. Of these, 19 case patients with biopsy-proven CRN and 35 control patients without CRN submitted stool. In a comparison of the case and control sets, case patients were significantly more likely to have extensive disease involvement and longer disease duration. There was no difference in disease activity between the 2 groups. Of the 19 cases with confirmed CRN, 9 patients had cancer with a medium size of 2.3 cm, 4 had HGD, and 6 had low-grade dysplasia. All 4 methylated genes showed high discrimination for cancer. Stool assay of mBMP3 alone was 100% sensitive for CRC, 84% for all CRN, and 91% specific. The combination of mBMP3 and mNDRG4 was 100% sensitive for both CRC and HGD with 89% specificity. In comparison of patients with ulcerative colitis (UC) and Crohn’s disease, Marker levels were not different between UC and Crohn’s disease (CD) patients. In multivariate analyses including age, gender, disease duration, disease extent, and presence of PSC, methylation markers for CRN detection remained significant. CommentOver the past decade, there have been several sets of guidelines regarding colon cancer screening and surveillance in patients with UC (Gut 2002;51[suppl 5]:V10–V12; Gastroenterology 2003;124:544–560; Gastrointest Endosc 2006;63:546–557; Inflamm Bowel Dis 2005;11:314–321; Am J Gastroenterol 2010;105:501–524; Gut 2010;59:666–689). These recommendations all include the use of colonoscopy. Unfortunately, the prevalence of colonoscopy surveillance to detect CRN ranges from 25% to 64% (Gastroenterology 2010;139:1511–1518; Gastrointest Endosc 2007;65:432–439; World J Gastroenterol 2009;15:226–230). Recent studies investigating the low rate of surveillance have found that factors associated with higher compliance include the absence of significant comorbidity and more than 3 UC-related outpatient visits (Gastroenterology 2010;139:1511–1518). Other factors affecting low surveillance rate may include patient concerns regarding risks of colonoscopy and the inconvenience of bowel prep or fasting before the procedure.The availability of a noninvasive test may serve to increase compliance with colorectal cancer screening and surveillance in IBD patients. The most common noninvasive approach to screening in the general population has been the detection of occult blood, which lacks sensitivity and specificity, and is unreliable in IBD patients given the inflammatory nature of the disease. Many cancers or premalignant adenomas do not bleed and can be missed (JAMA 1990;263:76–78; JAMA 1993;269:1262–1267). The genetics of CRC provide the basis for analysis of fecal DNA (Science 1992;256:102–105; Clin Chim Acta 2002;315:157–168). Altered DNA from a CRN arises from the neoplasm itself as opposed to circulation and is released via exfoliation. Neoplasm-specific DNA alterations have been well characterized (Eur J Gastroenterol Hepatol 1992;5:205–213; Eur J Gastroenterol Hepatol 1993;5:219–225). From this, a stool assay system was developed to identify a spectrum of DNA alterations that are associated with CRN (Gastroenterology 2000;119:1219–1227).Several studies have demonstrated the feasibility of detecting tumor-specific products in the stool (Gastroenterology 2005;128:192–206; Cancer 2006;106:277–283). Stool analyzed in blinded fashion from patients with CRC, adenomas, or normal colon revealed even an early generation fecal DNA assay was 91% sensitive for cancer and 82% for adenomas >1 cm with a specificity of 91% (Gastroenterology 2000;119:1219–1227). A large, prospective, multicenter study in >4000 average-risk asymptomatic patients over the age of 50 showed that fecal DNA test had a significantly higher sensitivity compared with stool Hemoccult test (N Engl J Med 2004;351:2704–2714).The authors of this study have addressed an interesting and important question. Patients with UC are at increased risk for developing CRC; however, the rate of surveillance colonoscopies within this group is highly variable. Even with current surveillance techniques, colonoscopy is an imperfect practice. The authors found 2 of the 9 CRC cases with positive stool results were missed on colonoscopy and diagnosed only after colectomy. The use of fecal DNA testing could be integral to the care of IBD patients. The design of this study was a major strength. In the tissue study, cases and controls were well-matched and the authors were able to determine which methylation markers were highly discriminant for IBD-CRN as opposed to relying on and extrapolating from prior studies performed with sporadic, non-IBD cases. They were also able to confirm prior observations in IBD cases with regard to mBMP3 and mEYA4. Based on the results of the tissue study, the authors demonstrated that fecal DNA testing was capable of achieving high detection rates for both CRC and dysplasia.The authors were able to achieve the target sample size to address the primary endpoints. The power to analyze subgroups, however, was limited, and the authors acknowledge this. A particular area of interest would have been to determine any differences in stool DNA markers between flat, dysplastic lesions from other dysplastic lesions. The finding of flat HGD or carcinoma on endoscopic biopsies is an indication for colectomy. Although controversial, there is some evidence to suggest that flat, low-grade dysplasia is also an indication for colectomy because of the high rate of progression to HGD or cancer (Am J Gastroenterol 2002;97:922–927; Gastroenterology 2003;125:1311–1319; Gut 2003;52:1127–1132; Dis Colon Rectum 2002;45:615–620; Gastroenterology 2004; 127:950–956). The sensitivity for low-grade dysplasia in the fecal DNA assay used in this study was at best 67%. Furthermore, owing to limited power in the subgroup analyses, the lack of an effect upon marker levels by disease duration, severity and extent of disease, and presence of concomitant PSC could be falsely negative.Overall, fecal DNA testing seems to be a promising alternative to colonoscopy for screening and surveillance purposes. Stool assay may be useful for early detection of IBD-CRN, and could also help to guide the frequency of colonoscopy surveillance to tailor individual risk. Larger studies are needed, however, to confirm the results demonstrated in this study, as well as to provide the power for greater sensitivity in subgroup analyses. Over the past decade, there have been several sets of guidelines regarding colon cancer screening and surveillance in patients with UC (Gut 2002;51[suppl 5]:V10–V12; Gastroenterology 2003;124:544–560; Gastrointest Endosc 2006;63:546–557; Inflamm Bowel Dis 2005;11:314–321; Am J Gastroenterol 2010;105:501–524; Gut 2010;59:666–689). These recommendations all include the use of colonoscopy. Unfortunately, the prevalence of colonoscopy surveillance to detect CRN ranges from 25% to 64% (Gastroenterology 2010;139:1511–1518; Gastrointest Endosc 2007;65:432–439; World J Gastroenterol 2009;15:226–230). Recent studies investigating the low rate of surveillance have found that factors associated with higher compliance include the absence of significant comorbidity and more than 3 UC-related outpatient visits (Gastroenterology 2010;139:1511–1518). Other factors affecting low surveillance rate may include patient concerns regarding risks of colonoscopy and the inconvenience of bowel prep or fasting before the procedure. The availability of a noninvasive test may serve to increase compliance with colorectal cancer screening and surveillance in IBD patients. The most common noninvasive approach to screening in the general population has been the detection of occult blood, which lacks sensitivity and specificity, and is unreliable in IBD patients given the inflammatory nature of the disease. Many cancers or premalignant adenomas do not bleed and can be missed (JAMA 1990;263:76–78; JAMA 1993;269:1262–1267). The genetics of CRC provide the basis for analysis of fecal DNA (Science 1992;256:102–105; Clin Chim Acta 2002;315:157–168). Altered DNA from a CRN arises from the neoplasm itself as opposed to circulation and is released via exfoliation. Neoplasm-specific DNA alterations have been well characterized (Eur J Gastroenterol Hepatol 1992;5:205–213; Eur J Gastroenterol Hepatol 1993;5:219–225). From this, a stool assay system was developed to identify a spectrum of DNA alterations that are associated with CRN (Gastroenterology 2000;119:1219–1227). Several studies have demonstrated the feasibility of detecting tumor-specific products in the stool (Gastroenterology 2005;128:192–206; Cancer 2006;106:277–283). Stool analyzed in blinded fashion from patients with CRC, adenomas, or normal colon revealed even an early generation fecal DNA assay was 91% sensitive for cancer and 82% for adenomas >1 cm with a specificity of 91% (Gastroenterology 2000;119:1219–1227). A large, prospective, multicenter study in >4000 average-risk asymptomatic patients over the age of 50 showed that fecal DNA test had a significantly higher sensitivity compared with stool Hemoccult test (N Engl J Med 2004;351:2704–2714). The authors of this study have addressed an interesting and important question. Patients with UC are at increased risk for developing CRC; however, the rate of surveillance colonoscopies within this group is highly variable. Even with current surveillance techniques, colonoscopy is an imperfect practice. The authors found 2 of the 9 CRC cases with positive stool results were missed on colonoscopy and diagnosed only after colectomy. The use of fecal DNA testing could be integral to the care of IBD patients. The design of this study was a major strength. In the tissue study, cases and controls were well-matched and the authors were able to determine which methylation markers were highly discriminant for IBD-CRN as opposed to relying on and extrapolating from prior studies performed with sporadic, non-IBD cases. They were also able to confirm prior observations in IBD cases with regard to mBMP3 and mEYA4. Based on the results of the tissue study, the authors demonstrated that fecal DNA testing was capable of achieving high detection rates for both CRC and dysplasia. The authors were able to achieve the target sample size to address the primary endpoints. The power to analyze subgroups, however, was limited, and the authors acknowledge this. A particular area of interest would have been to determine any differences in stool DNA markers between flat, dysplastic lesions from other dysplastic lesions. The finding of flat HGD or carcinoma on endoscopic biopsies is an indication for colectomy. Although controversial, there is some evidence to suggest that flat, low-grade dysplasia is also an indication for colectomy because of the high rate of progression to HGD or cancer (Am J Gastroenterol 2002;97:922–927; Gastroenterology 2003;125:1311–1319; Gut 2003;52:1127–1132; Dis Colon Rectum 2002;45:615–620; Gastroenterology 2004; 127:950–956). The sensitivity for low-grade dysplasia in the fecal DNA assay used in this study was at best 67%. Furthermore, owing to limited power in the subgroup analyses, the lack of an effect upon marker levels by disease duration, severity and extent of disease, and presence of concomitant PSC could be falsely negative. Overall, fecal DNA testing seems to be a promising alternative to colonoscopy for screening and surveillance purposes. Stool assay may be useful for early detection of IBD-CRN, and could also help to guide the frequency of colonoscopy surveillance to tailor individual risk. Larger studies are needed, however, to confirm the results demonstrated in this study, as well as to provide the power for greater sensitivity in subgroup analyses.