There is evidence that screening with the guaiac fecal

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2013;11: Association Between Early Stage Colon Neoplasms and False-negative Results From the Fecal Immunochemical Test HAN MO CHIU,*, YI CHIA LEE,*, CHIA HUNG TU,*, CHIEN CHUAN CHEN,*, PING HUEI TSENG,*, JIN TUNG LIANG, CHIA TUNG SHUN, JAW TOWN LIN,,# and MING SHIANG WU*,, ** *Department of Internal Medicine, Health Management Center, Department of Surgery, and Department of Pathology, National Taiwan University Hospital, Taipei; School of Medicine, Fu Jen Catholic University, New Taipei City; # Department of Internal Medicine, E-Da Hospital, Kaohsiung; and **Department of Primary Care, College of Medicine, National Taiwan University, Taipei, Taiwan BACKGROUND & AIMS: METHODS: RESULTS: CONCLUSIONS: The fecal immunochemical test (FIT) can identify patients with advanced colorectal neoplasms, but it also has a high rate of false-negative results. It would be helpful to characterize colorectal neoplasms that are not detected by FIT to aid in development of new tests. We characterized colorectal neoplasms from patients who had negative results from the FIT. We analyzed data from 18,296 subjects who were screened for colorectal cancer by colonoscopy and the FIT at the Health Management Center of National Taiwan University Hospital from September 2005 through September We identified 4045 subjects with colorectal neoplasms (3385 with nonadvanced adenomas, 632 with advanced adenomas, and 28 with cancer). We analyzed the sensitivity of the FIT in identifying these patients, along with information on lesion size, location, and morphology. The FIT identified patients with nonadvanced adenomas, advanced adenomas, and cancer with sensitivity values of 10.6% (95% confidence interval [CI], 10.2% 12.3%), 28.0% (95% CI, 24.6% 31.7%), and 78.6% (95% CI, 58.5% 91.0%), respectively. The FIT detected proximal advanced adenomas and nonpolypoid lesions with lower levels of sensitivity than distal advanced adenomas; it had a high false-negative rate in detection of adenomas <15 mm (adjusted odds ratio, 2.85; 95% CI, ) and nonpolypoid adenomas (adjusted odds ratio, 2.15; 95% CI, ), after adjusting for demographic characteristics, colonoscopy findings, and potential confounders. The FIT produced a higher percentage of falsenegative results in detection of carcinoma in situ and T1 cancer than in T2-T4 cancers (66.7% sensitivity vs 100%; P.049). The FIT produces a high rate of false-negative results for patients with small or nonpolypoid adenomas. Early-stage cancers are associated with a high rate of false-negative results from the FIT. Keywords: Colon Cancer; Early Detection; Database Analysis; Screening Comparison. See editorial on page 839. There is evidence that screening with the guaiac fecal occult blood test (gfobt) effectively reduces colorectal cancer (CRC) mortality. 1 This prompted the implementation of fecal-based population screening programs in countries with a modest-to-high prevalence of CRC but with resources, manpower, and endoscopy capacity constraints. 2 The fecal immunochemical test (FIT) for hemoglobin detects human hemoglobin in the feces and thus can specifically detect bleeding from the lower gastrointestinal tract. Compared with gfobt, FIT can detect earlier neoplasms, owing to its high sensitivity to advanced adenomas and early cancers, user-friendly design of sampling devices and no need of dietary restrictions with resultant greater participation, and adjustable cutoff that can be determined by individual programs according to their own available endoscopic capacities. 3,4 Accordingly, an increasing number of countries and regions are implementing FIT for CRC screening programs. 5 High programmatic efficacy depends largely on high sensitivity of the screening test and high participation/adherence rates. Because of the fact that current Abbreviations used in this paper: aor, adjusted odds ratio; CI, confidence interval; CRC, colorectal cancer; FIT, fecal immunochemical test; gfobt, guaiac fecal occult blood test by the AGA Institute /$

2 July 2013 FIT-NEGATIVE ADVANCED COLORECTAL NEOPLASMS 833 uptake of screening in populations at risk remains low in some regions, it is crucial to improve the sensitivity of the screening test. 2,6 Improved detection of cancerous lesions confined to the mucosal or submucosal layers can contribute to more favorable outcomes and improved overall CRC survival. 7 Improved detection of noncancerous advanced adenomas can further prevent the occurrence of CRC and thus reduce its incidence. Despite the launch of various screening programs, interval cancers and missed proximal colon cancers are well-recognized problems, irrespective of fecal-based or colonoscopy-based screening programs. Previous studies have shown that interval cancers in gfobt-based screening are more likely to arise in the proximal colon, and recent studies of colonoscopy-based screening programs also showed a lack of protection against proximal colon cancer Whether this is caused by substandard detection of proximal neoplasms by colonoscopy or fecal tests or by biologically different neoplasms that may escape from screening test and program remains unclear but is worth investigating. Only limited data are available about the diagnostic performance of the FIT with regard to different stages, anatomic sites, and endoscopic morphology of colorectal neoplasms By using a prospective cohort of 18,296 Chinese subjects who underwent screening colonoscopy and FIT during a 5-year period, we attempted to study the characteristics of false-negative FIT colorectal neoplasias and to determine whether the test s performance can vary according to different stage, locations, and morphology of neoplasms. Methods Study Subjects We prospectively enrolled consecutive asymptomatic adult individuals who underwent screening colonoscopy as part of thorough health check-ups at the Health Management Center of National Taiwan University Hospital between September 2005 and September The details of this health check-up program have been reported previously. 16 Only subjects older than 50 years were included in the study cohort. Subjects who had CRC and received colectomy and incomplete colonoscopy and did not submit fecal samples were excluded from the analysis. The study design is shown in Figure 1. The institutional review board and ethical committee of our hospital approved this study. Reporting of the study results followed the STARD guidelines. 17 Figure 1. Study flow diagram (Standards for Reporting of Diagnostic Accuracy flow diagram). Fecal Immunochemical Test None of the subjects were advised for dietary or medication restriction before testing. A one-step commercial FIT kit with a serrated tipped sampling probe as a component of the sample collection device cap (OC-LIGHT V-PC50 and V-PH80; Eiken Chemical Co, Ltd, Tokyo, Japan) was given to the participants, and a single-spot sample was used for testing. All participants were asked to collect their fecal samples at home by using the sampler in at least 6 different areas of the feces, put into the collection tubes, store them in the refrigerator 1 day before colonoscopy, and then to submit them on the day of colonoscopy. After collection of the fecal samples, they were sent to the laboratory, and all the tests were completed within the day of colonoscopy. The manufacturer s quoted hemoglobin cutoff was 50 ng hemoglobin/ml buffer, which was equivalent to 10 g hemoglobin/g feces. Three certified laboratory staff members in Department of Laboratory Medicine (laboratory is certified by ISO 15189) of our institute managed the collected fecal samples and performed all the tests. The laboratory staff members were blinded to the colonoscopic findings. Colonoscopy The procedures for performing colonoscopy and for recording the results and the histologic evaluation of biopsy specimens have been described previously. 18 Seven colonoscopists who had prior experience with at least 6000 colonoscopies performed colonoscopic examinations with a standard colonoscope (CF-260AI or CF-H260AZI; Olympus Medical Systems Corp, Ltd, Tokyo, Japan) and were blinded to the FIT results. All colonoscopists were asked to spend at least 6 minutes during colonoscopy withdrawal and inspection. Noninvasive neoplasms were resected endoscopically, and participants with invasive lesions or advanced cancers were referred for surgery. Definitions Colonic neoplasms were classified according to World Health Organization criteria. Advanced colorectal neoplasia was defined as lesions with one of the following characteristics: 10 mm in diameter, having a villous component, or high-grade dysplastic lesions or carcinoma in situ, and invasive cancers. The invasive depth of the primary tumor was staged according to the American Joint Committee on Cancer 7th edition as follows: Tis, carcinoma in situ, intraepithelial or invasion of lamina propria; T1, invades submucosa; T2, invades muscularis

3 834 CHIU ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 7 propria; T3, invades through the muscularis propria into pericolorectal tissues; T4, penetrates to the surface of the visceral peritoneum or directly invades or is adherent to other organs or structures. The location of the colorectal neoplasia was defined according to its anatomic distribution. The colon above the level of the splenic flexure (including splenic flexure) was defined as the proximal colon. Synchronous lesions denoted concurrent proximal and distal lesions of certain categories of neoplasm. When multiple lesions were present, subjects were categorized according to the highest severity of lesion in the proximal and distal colon. The Paris endoscopic classification was applied to define the morphology of advanced colorectal neoplasms. Lesions with 0-Is and 0-Ip morphologies were classified as polypoid lesions, whereas those with flat (0-IIa) and depressed (0-IIc, 0-IIa IIc, 0-IIc IIa, and 0-Is IIc) morphologies were classified as nonpolypoid. 19 Statistical Analysis To determine the performance of FIT, we used test results and colonoscopic findings to calculate the sensitivity, specificity, accuracy, positive predictive value, negative predictive value, and the corresponding 95% confidence intervals (CIs). Numeric variables were compared with an independent sample t test or the Mann Whitney U test. The Pearson 2 test or Fisher exact test was used for categorical variables. The Cochrane Armitage trend test was applied for comparing the sensitivity of advanced adenomas, Tis plus T1 cancers, and T2 to T4 cancers. When comparing the sensitivities of the FIT between polypoid and nonpolypoid advanced neoplasms, subjects with concurrent polypoid and nonpolypoid lesions were excluded from the analysis. We constructed 2 multivariable logistic regression models for all neoplasms and for nonadvanced and advanced adenomas individually to evaluate the association between false-negative FIT and colonoscopic findings. Model 1 contained colonoscopic findings, including lesion location, size, and morphology (for advanced adenomas only), whereas model 2 (the full model) contained not only the colonoscopic findings in model 1 but also the major confounders, including antiplatelet or anticoagulant agent use, platelet counts, and hemoglobin concentration, as we and others have investigated in previous studies. 20,21 Gender was also included in this model, because previous studies have shown a gender difference in FIT results. 22 Other factors that might have influenced colonic transit time, including age, gender, and smoking and drinking habits, were also included. The backward elimination method was used for these regression models. The results were expressed as adjusted odds ratios (aors) and 95% CIs, with ORs greater than 1 representing an increased risk of false-negative FIT results. All analyses were performed by using the SPSS statistical program, version 15.0 (SPSS, Inc, Chicago, IL). The reported P values were for 2-sided statistical tests, and P value.05 was considered statistically significant. Results Demographics of the Study Population A total of 18,296 subjects with complete FIT and colonoscopy data were enrolled in the study period for analysis (Figure 1). No subjects had perforation caused by colonoscopy. Of these subjects, 10,828 (59.2%) were male; 10,741 (58.7%) were years old, 5416 (29.6%) ranged from years of age, and 2139 (11.7%) were older than 70 years. The mean age was 59.8 years (standard deviation, 7.6 years). The FIT was positive in 1330 subjects (7.3%), and colonoscopic findings revealed nonadvanced adenomas in 3385 (18.5%), advanced adenomas in 632 (3.5%), cancer in 28 (0.15%), and invasive cancer in 23 (0.13%) of the subjects. Test Performance of the Fecal Immunochemical Test The test performance of the FIT in relation to the characteristics of the lesions is shown in Table 1. The sensitivity of FIT for nonadvanced adenomas, advanced adenomas, and cancer was 10.6%, 28.0%, and 78.6%, respectively. The sensitivity of FIT for advanced polypoid neoplasm (31.1%) was significantly higher than that of nonpolypoid ones (21.1%) (P.001). Fecal Immunochemical Test Sensitivity in Relation to Neoplasm Stage FIT sensitivity showed stage-dependence and was 28.0% for advanced adenomas, 66.7% for Tis plus T1 cancers, and 100% for T2 to T4 cancers (P for trend.001). The difference Table 1. Diagnostic Performance of FIT in Association With Neoplasm Characteristics and Stage Colonoscopic findings Positive test, N (%) Negative test, N (%) Performance measurements, % (95% CI) Sensitivity Specificity PPV NPV No. needed to scope, N (95% CI) All neoplasms 578 (14.3) 3467 (85.7) 14.3 ( ) 94.7 ( ) 43.5 ( ) 79.6 ( ) 2.29 ( ) Nonadvanced neoplasm 379 (11.2) 3006 (88.8) 10.6 ( ) 93.6 ( ) 28.5 ( ) 82.3 ( ) 3.51 ( ) Advanced neoplasm 199 (30.2) 461 (69.8) 30.2 ( ) 93.6 ( ) 15.0 ( ) 97.3 ( ) 6.67 ( ) Polypoid only 149 (31.1) 328 (69.9) 31.1 ( ) 93.4 ( ) 11.2 ( ) 98.1 ( ) 8.92 ( ) Nonpolypoid only 30 (21.1) 112 (78.9) 21.1 ( ) 92.9 ( ) 2.25 ( ) 99.3 ( ) 44.4 ( ) Advanced adenoma 177 (28.0) 455 (72.0) 28.0 ( ) 93.5 ( ) 13.3 ( ) 97.3 ( ) 7.52 ( ) Polypoid only 141 (30.1) 328 (69.9) 30.1 ( ) 93.3 ( ) 10.6 ( ) 98.1 ( ) 9.43 ( ) Nonpolypoid only 24 (18.1) 109 (81.9) 18.5 ( ) 92.8 ( ) 1.80 ( ) 99.6 ( ) 55.6 ( ) Cancer 22 (78.6) 6 (21.4) 78.6 ( ) 92.8 ( ) 1.65 ( ) 99.9 ( ) 60.6 ( ) Invasive cancer 17 (73.9) 6 (26.1) 73.9 ( ) 92.8 ( ) 1.28 ( ) 99.9 ( ) 78.1 ( ) Tis T1 cancer 12 (66.7) 6 (33.3) 66.7 ( ) 92.8 ( ) 0.90 ( ) 99.9 ( ) 111 ( ) T2-T4 cancer 5 (100.0) 0 (0.0) 100 ( ) 92.8 ( ) 0.38 ( ) ( ) 133 ( ) NPV, negative predictive value; PPV, positive predictive value.

4 July 2013 FIT-NEGATIVE ADVANCED COLORECTAL NEOPLASMS 835 Table 2. Sensitivity of FIT for Advanced Neoplasm in Association With Anatomic Location and Lesion Morphology Proximal/distal (N) Proximal colon (95% CI) Distal colon (95% CI) P value Advanced neoplasm 72/ ( ) 34.3 ( ).013 Polypoid 51/ ( ) 31.6 ( ).001 Nonpolypoid 16/ ( ) 24.3 ( ).001 Advanced adenoma 65/ ( ) 31.6 ( ).001 Polypoid 47/ ( ) 31.2 ( ).001 Nonpolypoid 15/ ( ) 20.6 ( ).001 Cancerous neoplasm 8/ ( ) 82.3 ( ).441 Invasive cancer 5/ ( ) 82.4 ( ).172 Tis T1 cancer 5/ ( ) 70 ( ).846 T2-T4 cancer 3/7 100 ( ) 100 ( ).206 NOTE. Subjects who had concurrent proximal and distal or polypoid and nonpolypoid advanced neoplasms were excluded from this analysis. was also significant when comparing advanced adenomas with Tis plus T1 (P.001) and Tis plus T1 with T2 to T4 (P.049). Fecal Immunochemical Test Sensitivity for Advanced Neoplasms in Relation to Location and Morphology Subjects with concurrent proximal and distal or concurrent polypoid and nonpolypoid advanced neoplasms were excluded from this analysis. As shown in Table 2, the sensitivity of FIT was significantly lower for proximal lesions than for overall advanced neoplasms (proximal vs distal, 24.1% vs 34.3%; P.013). After stratification with morphology, the sensitivity was significantly lower for proximal polypoid advanced neoplasms compared with distal ones (27.7% vs 31.6%; P.001) and proximal nonpolypoid advanced neoplasms compared with distal ones (16.2% vs 24.3%; P.001). The sensitivity was also significantly lower for proximal or distal nonpolypoid advanced neoplasms compared with polypoid ones at correspondent location (P.001 for both). The trend was similar when proximal and distal, polypoid and nonpolypoid advanced adenomas were compared. For invasive cancers, although there was a trend toward lower sensitivity for proximal lesions, the result was not statistically significant (62.4% vs 82.4%; P.17). Multivariable Analysis of Factors Associated With a False-negative Test Result All neoplasms. Proximal location was positively associated with false-negative results in model 1 (aor 1.27; 95% CI, ) and model 2 (aor 1.27; 95% CI, ). Synchronous lesions were inversely associated with falsenegative results in model 1 (aor 0.71; 95% CI, ) but became insignificant in model 2 (aor 0.78; 95% CI, ). The lesion size was inversely associated with false-negative results, and aor was 0.90 (95% CI, ) in both model 1 and model 2 per 1-mm increments in size (Figure 2 and Supplementary Table 1). Nonadvanced adenoma. For nonadvanced adenomas, synchronous locations were inversely associated with falsenegative results on the FIT in model 1 (aor 0.72; 95% CI, ) and model 2 (aor 0.70; 95% CI, ), and a smaller size ( 5 mm) was associated with false-negative results in both model 1 (aor 1.65; 95% CI, ) and model 2 (aor 1.57; 95% CI, ) (Supplementary Table 2). Advanced adenoma. As shown in Figure 3, no significant differences existed regarding lesion location. Small lesion size ( 15 mm) was associated with false-negative results in models 1 (aor 2.72; 95% CI, ) and 2 (aor 2.85; 95% CI, ). These trends were consistent even when we used the 20-mm or 10-mm cutoff for the advanced adenoma lesion size (not shown). Lesions with nonpolypoid morphology were more likely to have false-negative results in models 1 (aor 2.13; 95% CI, ) and 2 (aor 2.15; 95% CI, ) (Supplementary Table 3). Discussion Improving the sensitivity of the fecal tests, especially to early advanced neoplasms, could increase the effectiveness of CRC screening programs. In our large prospective study, we found lower sensitivity of FIT for advanced neoplasms of early stage, small size, nonpolypoid morphology, and proximal location. Our findings suggest that the benefits of a single FIT screening may differ on the basis of tumor characteristics and highlighted the existence of lower FIT performance in subgroups. These findings also highlight the importance of future studies exploring whether this limitation really leads to less protection against proximal colon cancer in FIT-based screening programs. Such studies might also help clarify whether FIT screening programs could be improved by being modified (by increasing the number of samples or shortening screening intervals, for example) and could lead to development of new screening tests with better ability to detect such neoplasms. Most FIT-based screening programs currently recommend annual or biennial screening with 1 or 2 fecal samples under a predetermined cutoff. 2 The findings from our study have several implications for balancing FIT sensitivity and specificity and current screening algorithms. First, missed advanced neoplasms, especially the cancerous ones, may lead to the occurrence of interval cancers. Interval cancer is an important performance indicator reflecting the program sensitivity, and such cancers exist in both endoscopy-based and fecal-based screening programs, as reported in previous studies. 8,12,23 In our current study, we revealed that proximally located advanced adenomas, or tumors with nonpolypoid morphology, and early-stage cancers were less susceptible to detection by FIT. Regarding the natural course of colorectal neoplasms and the perfect FIT sensitivity to detect T2 cancers and its 60% sensitivity to detect

5 836 CHIU ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 7 Figure 2. Characteristics of all neoplasms in association with false-negative FIT results. ratios in model 1 and model 2 are aors. Model 1 was adjusted for endoscopic findings, including lesion location and size. Model 2 was adjusted for endoscopic findings, age, sex, body mass index, drinking or smoking habits, use of antiplatelet or anticoagulant agents, low platelet counts, low hemoglobin concentration, and family history of CRC. Tis and T1 cancers, as demonstrated in this study, these testnegative advanced adenomatous or cancerous lesions theoretically could be detected at an earlier stage at the upcoming screening rounds with a short screening interval and good compliance. Governments and screening organizers need to devote more resources to motivate the public to regularly participate in screening tests. Second, screening organizers may apply different strategies such as lowering cutoff, increasing numbers of fecal samples, and shortening of screening intervals to improve programmatic sensitivity. Although previous studies have shown that these approaches may improve single FIT sensitivity, their effectiveness on programmatic sensitivity or long-term outcome needs further investigation. 24,25 Moreover, cost-effectiveness is another important concern and should be carefully evaluated before implementing to the population program. 26 Finally, whether development of improved and affordable fecal DNA tests in combination with FIT will reduce the current false-negative rates in FIT-resistant screening subgroups represents a critical issue deserving further attention. Another intriguing finding in our study was the lower sensitivity of FIT in detecting nonpolypoid and proximal advanced adenomas (Table 2, Supplementary Table 4). Although the significance of proximal location decreased after adjustment for size and morphology, it is caused by proximal-prone location of the nonpolypoid neoplasms. Whether nonpolypoid colorectal neoplasms carry a higher risk of malignant potential has been debated for 2 decades since its initial report by Japanese researchers. 27,28 Today, such neoplasms have been detected and reported worldwide and are known to be more frequently found in the proximal colon; it is also known that they carry a higher risk for malignant transformation and invasiveness at a relatively smaller size when compared with their polypoid counterparts. 18,29 Detecting such neoplasms by screening may help prevent interval cancer and improve the effectiveness of screening programs. It is still not well understood why these lesions are less sensitive to detection by the FIT. We speculated that their smaller surface areas in contact with feces, the sparser vasculature in the mucosa, and the more advanced degree of hemoglobin degradation during bowel passage for proximally located lesions provide partial explanations, and this issue is worthwhile of further investigation. Our study has several strengths. First, our analyses were adjusted for other factors that previously reported to be associated with false-negative or false-positive results but failed to conduct multivariate analysis owing to the limited number of cases of advanced neoplasms. 20,21,30 In this study, we reported that 816 participants with advanced neoplasms in the screening cohort concurrently received colonoscopies and FIT. Meanwhile, we also collected a drug history, especially a history of use of aspirin or nonsteroidal anti-inflammatory agents, by using a structured questionnaire, which enabled us to conduct a comprehensive multivariate analysis. Second, we used World Health Organization diagnostic criteria for colorectal neoplasms and had a dedicated staff who were unaware of the FIT results and verified all the pathology reports for the detected neoplasms. Our study did have some limitations. First, the lower sensitivity of FIT for early cancerous lesions does not represent a lower programmatic sensitivity and overall efficacy, because of good public adherence to repetitive screening and taking into account the relatively slow natural history of advanced adenomas. Previous study, however, indicated the potential impact of a so-called de novo pathway by which depressed nonpolypoid lesions transit to cancer on the reduced efficacy of CRC screening. 31 Additional study is mandatory to elucidate this. Second, we used a qualitative FIT kit, which makes it impossible to correlate test sensitivity in relation to tumor characteristics

6 July 2013 FIT-NEGATIVE ADVANCED COLORECTAL NEOPLASMS 837 Figure 3. Characteristics of advanced adenomas in association with false-negative FIT results. ratios in model 1 and model 2 are aors. Model 1 was adjusted for endoscopic findings including lesion location, size, and morphology. Model 2 was adjusted for endoscopic findings, age, sex, body mass index, drinking or smoking habits, use of antiplatelet or anticoagulant agents, low platelet counts, low hemoglobin concentration, and family history of CRC. with fecal hemoglobin concentration. Additional studies are warranted to examine whether our findings still exist in a quantitative FIT at a 100-ng/mL or higher cutoff, a commonly used threshold in many screening programs, or at a cutoff less than 50 ng/ml. Third, the timing of fecal sampling, a factor reported to influence FIT performance, was not recorded in our study. However, this is less likely to be a problem because we asked the attendees to collect fecal samples 1 day before scheduled colonoscopies and store them in the refrigerator, and all samples were analyzed on the day of colonoscopy. Finally, the study population of this study is subjects who participated in self-paid health check-up; thus, self-selection bias might exist. Conclusions Our data showed a lower sensitivity of FIT for early, proximal, and nonpolypoid advanced colorectal neoplasms. Screening organizers should further evaluate and take into account the optimal cutoff, sample numbers, screening intervals, and cost-effectiveness for future improvement of their screening programs. Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at and at doi.org/ /j.cgh References 1. Hewitson P, Glasziou P, Watson E, et al. Cochrane systematic review of colorectal cancer screening using the fecal occult blood test (hemoccult): an update. Am J Gastroenterol 2008;103: Benson VS, Patnick J, Davies AK, et al. Colorectal cancer screening: a comparison of 35 initiatives in 17 countries. Int J Cancer 2008;122: van Rossum LG, van Rijn AF, Laheij RJ, et al. Random comparison of guaiac and immunochemical fecal occult blood tests for colorectal cancer in a screening population. Gastroenterology 2008; 135: Hol L, van Leerdam ME, van Ballegooijen M, et al. Screening for colorectal cancer: randomised trial comparing guaiac-based and immunochemical faecal occult blood testing and flexible sigmoidoscopy. Gut 2010;59: Benson VS, Atkin WS, Green J, et al. Toward standardizing and reporting colorectal cancer screening indicators on an international level: the International Colorectal Cancer Screening Network. Int J Cancer 2012;130: Inventory of colorectal cancer screening activities in ICSN countries. International Cancer Screening Network Available at: html. Accessed April 17, O Connell JB, Maggard MA, Ko CY. Colon cancer survival rates with the new American Joint Committee on Cancer sixth edition staging. J Natl Cancer Inst 2004;96: Steele RJ, McClements P, Watling C, et al. Interval cancers in a FOBT-based colorectal cancer population screening programme:

7 838 CHIU ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 7 implications for stage, gender and tumour site. Gut 2012; 61: Brenner H, Hoffmeister M, Arndt V, et al. Protection from rightand left-sided colorectal neoplasms after colonoscopy: population-based study. J Natl Cancer Inst 2010;102: Bressler B, Paszat LF, Chen Z, et al. Rates of new or missed colorectal cancers after colonoscopy and their risk factors: a population-based analysis. Gastroenterology 2007;132: Singh H, Turner D, Xue L, et al. Risk of developing colorectal cancer following a negative colonoscopy examination: evidence for a 10-year interval between colonoscopies. JAMA 2006;295: Singh H, Nugent Z, Demers AA, et al. The reduction in colorectal cancer mortality after colonoscopy varies by site of the cancer. Gastroenterology 2010;139: Haug U, Kuntz KM, Knudsen AB, et al. Sensitivity of immunochemical faecal occult blood testing for detecting left- vs rightsided colorectal neoplasia. Br J Cancer 2011;104: Morikawa T, Kato J, Yamaji Y, et al. Sensitivity of immunochemical fecal occult blood test to small colorectal adenomas. Am J Gastroenterol 2007;102: Ciatto S, Martinelli F, Castiglione G, et al. Association of FOBTassessed faecal Hb content with colonic lesions detected in the Florence screening programme. Br J Cancer 2007;96: Chiu HM, Wang HP, Lee YC, et al. A prospective study of the frequency and the topographical distribution of colon neoplasia in asymptomatic average-risk Chinese adults as determined by colonoscopic screening. Gastrointest Endosc 2005;61: Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative Standards for Reporting of Diagnostic Accuracy. Clin Chem 2003;49: Chiu HM, Lin JT, Chen CC, et al. Prevalence and characteristics of nonpolypoid colorectal neoplasm in an asymptomatic and average-risk Chinese population. Clin Gastroenterol Hepatol 2009;7: Lambert R. Update on the Paris classification of superficial neoplastic lesions in the digestive tract. Endoscopy 2005;37: Levi Z, Rozen P, Hazazi R, et al. Sensitivity, but not specificity, of a quantitative immunochemical fecal occult blood test for neoplasia is slightly increased by the use of low-dose aspirin, NSAIDs, and anticoagulants. Am J Gastroenterol 2009;104: Chiang TH, Lee YC, Tu CH, et al. Performance of the immunochemical fecal occult blood test in predicting lesions in the lower gastrointestinal tract. CMAJ 2011;183: McDonald PJ, Strachan JA, Digby J, et al. Faecal haemoglobin concentrations by gender and age: implications for populationbased screening for colorectal cancer. Clin Chem Lab Med 2012; 50: Lakoff J, Paszat LF, Saskin R, et al. Risk of developing proximal versus distal colorectal cancer after a negative colonoscopy: a population-based study. Clin Gastroenterol Hepatol 2008;6: Guittet L, Bouvier V, Mariotte N, et al. Performance of immunochemical faecal occult blood test in colorectal cancer screening in average-risk population according to positivity threshold and number of samples. Int J Cancer 2009;125: van Roon AH, Wilschut JA, Hol L, et al. Diagnostic yield improves with collection of 2 samples in fecal immunochemical test screening without affecting attendance. Clin Gastroenterol Hepatol 2011;9: Chen LS, Liao CS, Chang SH, et al. Cost-effectiveness analysis for determining optimal cut-off of immunochemical faecal occult blood test for population-based colorectal cancer screening (KCIS 16). J Med Screen 2007;14: Rembacken BJ, Fujii T, Cairns A, et al. Flat and depressed colonic neoplasms: a prospective study of 1000 colonoscopies in the UK. Lancet 2000;355: Tsuda S, Veress B, Tóth E, et al. Flat and depressed colorectal tumours in a southern Swedish population: a prospective chromoendoscopic and histopathological study. Gut 2002;51: Soetikno RM, Kaltenbach T, Rouse RV, et al. Prevalence of nonpolypoid (flat and depressed) colorectal neoplasms in asymptomatic and symptomatic adults. JAMA 2008;299: Brenner H, Haug U, Hundt S. Sex differences in performance of fecal occult blood testing. Am J Gastroenterol 2010;105: Chen CD, Yen MF, Wang WM, et al. A case-cohort study for the disease natural history of adenoma-carcinoma and de novo carcinoma and surveillance of colon and rectum after polypectomy: implication for efficacy of colonoscopy. Br J Cancer 2003;88: Reprint requests Address requests for reprints to: Ming-Shiang Wu, MD, PhD, Department of Internal Medicine, National Taiwan University Hospital, 7, Chung- Shan South Road, Taipei, Taiwan. mingshiang@ntu.edu.tw; fax: Acknowledgments This study was presented during the Clinical Practice Distinguished Abstract Plenary in Digestive Disease Week 2012 (May 21, 2012, San Diego, CA; Abstract # ). Conflicts of interest The authors disclose no conflicts. Funding Partially supported by a research grant from the Department of Health of Taiwan (Center of Excellence for Cancer Research, grant DOH101-TD-C ).

8 838.e1 CHIU ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 11, No. 7 Supplementary Table 1. Results of Multivariate Logistic Regression for Subjects Older Than 50 Years for All Neoplasms Univariable Multivariable (model 1) Multivariable (model 2) Variable Age 0.83 ( ) ( ).01 Male sex (vs female) 0.83 ( ) ( ).45 Body mass index (per 1-kg/m 2 increase) 0.95 ( ) ( ).01 Smoking (vs nonsmoking) 0.68 ( ) ( ).01 Alcohol use (vs nonuse) 0.96 ( ) ( ).78 Antiplatelet drug use 0.95 ( ) ( ).55 Low (vs normal) hemoglobin 0.68 ( ) ( ).1 Low (vs normal) platelet count 0.87 ( ) ( ).39 CRC history in first-degree relative 1.04 ( ) ( ).79 Location Proximal vs distal 1.26 ( ) ( ) ( ).03 Synchronous vs distal 0.6 ( ) ( ) ( ).07 Size (mm) 0.9 ( ) ( ) ( ).01 Supplementary Table 2. Results of Multivariate Logistic Regression for Subjects Older Than 50 Years for Nonadvanced Adenomas Univariable Multivariable (model 1) Multivariable (model 2) Variable Age 0.8 ( ) ( ).01 Male sex (vs female) 0.87 ( ) ( ).96 Body mass index (per 1-kg/m 2 increase) 0.96 ( ) ( ).06 Smoking (vs nonsmoking) 0.72 ( ) ( ).02 Alcohol use (vs nonuse) 1.09 ( ) ( ).5 Antiplatelet drug use 0.93 ( ) ( ).49 Low (vs normal) hemoglobin 0.92 ( ) ( ).9 Low (vs normal) platelet count 0.75 ( ) ( ).38 CRC history in first-degree relative 1.12 ( ) ( ).46 Location Proximal vs distal 1.1 ( ) ( ) ( ).78 Synchronous vs distal 0.65 ( ) ( ) ( ).05 Size ( 5 mmvs 5 mm) 1.7 ( ) ( ) ( ).01

9 July 2013 FIT-NEGATIVE ADVANCED COLORECTAL NEOPLASMS 838.e2 Supplementary Table 3. Results of Multivariate Logistic Regression for Subjects Older Than 50 Years for Advanced Adenomas Univariable Multivariable (model 1) Multivariable (model 2) Variable Age 1.03 ( ).84 1 ( ).99 Male sex (vs female) 0.67 ( ) ( ).06 Body mass index (per 1-kg/m 2 increase) 0.95 ( ) ( ).04 Smoking (vs nonsmoking) 0.55 ( ) ( ).07 Alcohol use (vs nonuse) 0.68 ( ) ( ).98 Antiplatelet drug use 0.89 ( ) ( ).8 Low (vs normal) hemoglobin 0.58 ( ) ( ).16 Low (vs normal) platelet count 0.99 ( ) ( ).82 CRC history in first-degree relative 0.7 ( ) ( ).11 Location Proximal vs distal 1.59 ( ) ( ) ( ).14 Synchronous vs distal 0.65 ( ) ( ) ( ).37 Size ( 15 mm vs 15 mm) 2.42 ( ) ( ) ( ).01 Morphology (nonpolypoid vs polypoid) 2.03 ( ) ( ) ( ).01 Supplementary Table 4. Description of Tis and T1 Cancers Case Sex Age (y) T stage FIT Size (mm) Morphology Location 1 M 64 Tis Positive 35 0-Is Proximal T 2 F 50 Tis Positive 40 0-Is Proximal T 3 M 70 Tis Positive 20 0-Is Distal S 4 F 62 Tis Positive 50 0-Is Proximal T 5 M 55 Tis Positive 15 0-IIa IIc Distal R 6 F 67 T1 Positive 10 0-Is Distal R 7 M 56 T1 Positive 20 0-Is Distal S 8 M 61 T1 Negative 10 0-IIa IIc Proximal R 9 M 60 T1 Positive 10 0-Is IIc Distal R 10 M 58 T1 Positive 30 0-Is Proximal A 11 M 58 T1 Negative 10 0-IIa IIc Distal R 12 M 59 T1 Negative 30 0-IIa (LST-NG) Proximal T 13 M 58 T1 Positive 10 0-Is IIc Proximal A 14 F 80 T1 Negative 12 0-Is IIc Proximal A 15 F 62 T1 Positive 15 0-Is Distal S 16 M 61 T1 Negative 10 0-Is Distal S 17 M 75 T1 Positive 10 0-IIa IIc Distal S 18 F 79 T1 Negative 40 0-IIa IIc Distal S A, ascending colon; R, rectum; S, sigmoid colon; T, transverse colon.