Population-based colorectal cancer screening by fecal immunochemical testing over multiple rounds van der Vlugt, M.

Transcription

1 UvA-DARE (Digital Academic Repository) Population-based colorectal cancer screening by fecal immunochemical testing over multiple rounds van der Vlugt, M. Link to publication Citation for published version (APA): van der Vlugt, M. (2017). Population-based colorectal cancer screening by fecal immunochemical testing over multiple rounds General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons). Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: or a letter to: Library of the University of Amsterdam, Secretariat, Singel 25, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible. UvA-DARE is a service provided by the library of the University of Amsterdam ( Download date: 15 Jan 2019

2 Interval Colorectal Cancer Incidence Among Subjects Undergoing Multiple Rounds of Fecal Immunochemical Testing Manon van der Vlugt*, Esmée J. Grobbee*, Patrick M.M. Bossuyt, Amanda Bos, Evelien Bongers, Wolfert Spijker, Ernst J. Kuipers, Iris Lansdorp-Vogelaar, Manon C.W. Spaander, Evelien Dekker *shared first authorship Gastroenterology, 2017

3 CHAPTER ABSTRACT Background & Aims Among subjects screened for colorectal cancer (CRC) by the guaiac fecal occult blood test, interval cancers develop in 8% to 55% of the subjects. Data are limited on how many persons screened by fecal immunochemical tests (FIT), over multiple rounds, develop interval cancers. In the Netherlands, a pilot FIT-based biennial CRC screening program was conducted between 2006 and 201. We collected and analyzed data from the program on CRCs detected during screening (SD-CRC) and CRCs not detected within the screening program (non SD-CRC; such as FIT interval cancers, colonoscopy interval cancers and cancer in nonparticipants). Methods Screenees with a negative FIT result received a letter explaining that no blood had been detected in the stool sample and were re-invited, if eligible, for screening biennially. Screenees with a positive FIT result (hemoglobin concentration of 10 µg Hb/g feces) were invited for consultation and scheduled for colonoscopy; results were collected. After the fourth round of FIT screening, the cohort was linked to the Netherlands Cancer Registry, through March 31, 2015; participant characteristics, data on tumor stage, location (at time of resection), and survival status were collected for all identified CRC cases. A reference group comprised all persons with CRC diagnosed in the Netherlands general population during the same period, in the same age range (50 76 years), who had not been offered CRC screening. The median time between invitations (2.37 years) was used as a cutoff to categorize participants within the FIT interval cancer category. We compared participant characteristics, tumor characteristics, and mortality among subjects with SD-CRC and with non SD-CRC. Results A total of 27,30 eligible individuals were invited for FIT screening, of whom 18,716 (69%) participated at least once. Of these, 3005 (16%) had a positive result from the FIT in 1 of the screening rounds. In total, CRC was detected in 261 participants: 116 SD-CRCs and 15 non SD-CRCs (27 FIT interval cancers, 9 colonoscopy interval cancers, and 109 CRCs in nonparticipants). The FIT interval cancer proportion after 3 completed screening rounds was 23% Participants with SD-CRC had more earlystage tumors than participants with non SD-CRCs (P <.001). Of persons with SD-CRC and FIT interval cancers, significantly higher proportions survived (89% and 81% respectively), compared to persons with colonoscopy interval cancers (% survival) and non-participants with CRC (60% survival) (P <.001). Conclusions In an analysis of data from a pilot FIT-based biennial screening program, we found that among persons screened by FIT, 23% developed FIT interval cancer. FIT therefore detects CRC with 77% sensitivity. The proportion of FIT interval cancers in FIT screening appears to be lower than that with guaiac fecal occult blood testing. 6

4 FIT INTERVAL CANCERS IN COLORECTAL CANCER SCREENING INTRODUCTION Colorectal cancer (CRC) is one of the leading causes of cancer-related deaths in the Western World [1,2]. Approximately 50% of patients die from the disease [3]. Survival is strongly related to tumor stage at time of diagnosis, with a 5-year survival of 9% for stage I CRC to 8% for stage IV CRC [3]. Population-based CRC screening programs enable detection of CRC at an earlier stage (screendetected CRC;SD-CRC). Screening with fecal occult blood tests (FOBT) has been shown to reduce CRC-related mortality [-6]. Although FOBT screening is effective, not all CRCs will be detected within a screening program. A number of screening invitees will develop CRC in time although their cancer was not detected by FOBT: these are non-screen-detected CRCs (non-sd-crc) [7]. These cancer cases partially consist of interval cancers: cancer cases detected after a negative screening examination and before the date of the next recommended screening [7]. Such cancers can be missed by the first-line screening test or during colonoscopy following a positive FOBT [7]. Other cancer cases will be detected in those who did not participate in screening. Monitoring the incidence of interval cancers is regarded as an indicator of program sensitivity and FIT sensitivity. So far, most studies have reported interval cancers of programs using guaiacbased FOBT, showing high proportions of interval cancers. In such programs, the proportion of interval cancer cases ranged from 8% to 55% [5,6,8,9]. These studies have shown better survival rates in screen-detected cancers than in interval cancers [10]. As fecal immunochemical tests have a better diagnostic accuracy than guaiac-based FOBTs (gfobts), it is conceivable that interval cancer proportions are lower than those reported in gfobt-based programs [11]. In this study, we report FIT interval cancers over multiple rounds of FIT screening. METHODS/MATERIALS Population and design Since 2006, 2 pilot programs of biennial FIT-based CRC screening have been conducted in the southwest and northwest regions of the Netherlands. These 2 cohorts were combined for a fourth pilot round of screening in 201. Details about the design of these CRC screening programs have been reported previously [12,13]. In 2006, we selected the initial 2 cohorts based on postal code areas within our regions, which were both rural and urban areas with a known average uptake in breast cancer screening. Since 2006, we used the same inclusion and exclusion criteria for both cohorts. In short, demographic data of all invitees between 50 and 7 years living in the target areas were obtained from municipal population registers. See Supplementary Table 1 for baseline invitee characteristics per region. No national screening program had been implemented at the start of this pilot program; the target population was screening-naive when first contacted. In the Netherlands, a national FIT-based CRC screening program has been gradually initiated from January 201 onward. Invitees for the study cohort were not invited for the national CRC screening program. Selected persons were invited for each consecutive round, except for those who had moved out of the area, those that had passed the upper age limit, institutionalized people, those with an 65

5 CHAPTER estimated life expectancy of less than 5 years, those unable to give informed consent, and those who had tested positive in a previous screening round and had undergone a colonoscopy. In our information leaflet, persons with a history of inflammatory bowel disease or CRC were advised not to participate CRC screening and to report this to the screening organization. Characteristics Date of birth, sex, and postal codes of all invitees were collected from the municipal population register. Socioeconomic status was based on the social status scores that are available through the Netherlands Institute of Social Research ( The postal code of the invitee was used to evaluate social status. The social status score of a postal code area was based on the unemployment rate, educational level, average income and position on the labor market. The average score is 0 and the 2006 standard deviation in the Netherlands was used to assign invitees to 1 of 3 categories: high (status scores >0.96), average (status scores between and 0.96) and low SES (status scores <-0.96). The first available postal code of the invitee was used to categorize invitees. Invitations and stool tests Because of organizational and logistical issues, the time interval between rounds differed (median time between invitations was 2.37 years; interquartile range (IQR) 2.01 to 2.76). Invitations were sent between June 2006 and December 201. Two different brands of stool tests were used in our cohort. In the first, second, and third round all invitees received an OC-Sensor (Eiken Chemical Co, Tokyo, Japan). In the fourth round (executed between March 201 and February 2015), all invitees were randomly allocated to receive either an OC-Sensor or a FOB-Gold (Sentinel Diagnostics SpA, Milan, Italy) [1]. A hemoglobin (Hb) value of 10 µg Hb/g feces was used as the positivity threshold; this corresponds to 50 ng/ml buffer for OC-Sensor and 58 ng/ml buffer for FOB-Gold. Test results Screenees with a negative FIT result received a letter explaining that no blood had been detected in the stool sample and no follow-up was needed at that time. It was emphasized that the FIT is not 100% sensitive and that vigilance for symptoms of CRC remained important. Participants were instructed to contact their general practitioner in case of symptoms, despite the negative test result. Screenees with a positive FIT result were invited for a consultation at the outpatient clinic to discuss the test result. During the intake at the outpatient clinic, for all persons who were FIT-positive, a colonoscopy was scheduled within weeks. In some cases, patient-related circumstances lead to postponing the colonoscopy. All endoscopies were performed within 6 months and therefore, a timeframe of 6 months after a positive FIT was used for the definition of screen-detected cancers. Follow-up colonoscopy and colorectal lesions The colonoscopy was performed per international quality standards; quality parameters were collected in a database [15]. To describe the quality of the bowel preparation, a -point scale was 66

6 FIT INTERVAL CANCERS IN COLORECTAL CANCER SCREENING used: excellent, good, fair, poor. Excellent and good were considered as sufficient and fair and poor were considered as insufficient bowel preparation. Since 2013, the Boston Bowel preparation scale (BBPS) was used with BBPS 6 considered as sufficient bowel preparation. Cecal intubation was confirmed by photo documentation of cecal landmarks. All endoscopists were experienced, and had performed at least 1,000 colonoscopies. Advice regarding surveillance colonoscopy after removal of adenomatous polyps, large ( 10 mm) serrated lesions or cancer was given to the participant according to the Dutch Guideline Colonoscopy Surveillance [16]. Data on the location, size, macroscopic aspects, and morphology, as well as details on the technique for polypectomy and endoscopic assessment of radicality were recorded for all colorectal lesions detected during colonoscopy. Tumor location was categorized as either proximal or distal to the splenic flexure. Collected lesions were evaluated by an experienced gastrointestinal pathologist, using the Vienna criteria [17]. Cancers were staged per the seventh edition American Joint Committee on Cancer classification [18]. Screen-detected cancers, non-screen detected cancers and the general population After finishing the fourth screening round, the cohort was linked to the Netherlands Cancer Registry, managed by the Netherlands Comprehensive Cancer Organisation (up to date until March 31, 2015). Since 1989, the Netherlands Cancer Registry registers all participants diagnosed with cancer in the Netherlands and provides a unique and fully covered database. Data on tumor stage, location (at time of resection) and survival status were collected for all identified CRC cases. Definition of SD-CRC and non-sd-crc are depicted in Table 1. A colonoscopy interval cancer within a FIT screening program was defined as a cancer diagnosed after negative colonoscopy after a positive FIT within the surveillance interval. We defined a reference group consisting of all persons with CRC diagnosed in the Netherlands general population during the same period and in the same age range (50-76 years) who had not been offered CRC screening. The upper range of 76 years was chosen based on the median time between invitations and the upper age limit for eligibility ( years). Participants with CRC detected at a scheduled surveillance colonoscopy as well as CRC in screening participants with a positive FIT who declined to undergo a subsequent colonoscopy are reported separately. The median time between invitations (2.37 years; IQR ) was used as a cutoff to categorize participants within the FIT interval cancer category. If a participant had a negative FIT and developed a CRC but had not yet been invited for the consecutive round this was categorized as a FIT interval cancer. If a participant had a negative FIT and was not invited for a consecutive round (due to having passed the upper age limit, or having moved out of the area) but developed a CRC within the 2.37 years interval (the median time between invitations), this was categorized as a FIT interval cancer. Those who were not compliant with the FIT screening program were not defined as an interval CRC, because there cannot be an interval CRC if the individual did not undergo initial testing [7] and was not included in the analysis. Data on the general population and on nonparticipants (because there was no informed consent) were anonymously analyzed and subsequently delivered to us by the Netherlands Comprehensive 67

7 CHAPTER Cancer Organisation. Data on follow-up time, in days after diagnosis, and SES are not available for these groups. Data-analysis The proportion of FIT interval cancers was calculated by dividing the number of FIT interval cancers by the sum of SD-CRC and FIT interval cancers. The proportion was calculated over 3 completed screening rounds so SD-CRC cases detected within the first 3 screening rounds were selected and the total number of FIT interval cancers occurring after these rounds. FIT CRC sensitivity was calculated as 1-FIT interval cancer proportion. Median follow-up for participants with SD-CRC, FIT interval cancers, and colonoscopy interval cancers was calculated from date of diagnosis to death or considered censored at the end of follow-up (March 31, 2015). Median follow-up could not be calculated for the general population and for nonparticipants because no data on follow-up in days after diagnosis was available nor the exact date of diagnosis; only the year of diagnosis of these cases was provided. Survival after diagnosis was only estimated for participants with SD-CRC, FIT interval cancers, and colonoscopy interval cancers. Differences in proportions between groups were evaluated for statistical significance using the χ2-test statistic. Kaplan-Meier was used to estimate for survival with accompanying confidence intervals. P-values <.05 were considered to indicate statistically significant differences. Data analysis was performed using SPSS 23 for Windows (Chicago, Ill). Ethics approval Approval for the study was provided by the Dutch National Health Council (WBO 26267, , and , The Hague, The Netherlands). All authors had access to the study data and reviewed and approved the final manuscript. Table 1. Definitions of cancers Terminology SD-CRC non-sd-crc Definition Screen-detected colorectal cancer defined as CRCs detected at colonoscopy following a positive FIT result Non-screen-detected colorectal cancer defined as all CRCs that were not diagnosed at a colonoscopy following a positive FIT result and are divided into three groups: FIT interval cancer defined as cancers diagnosed between screening rounds after negative FIT before the next FIT was due Colonoscopy interval cancer within a FIT screening program defined as cancers diagnosed after negative colonoscopy after a positive FIT within the surveillance interval Colorectal cancer in nonparticipants defined as cancers diagnosed in those who did not respond to their FIT invitation. General population The reference group was defined as all persons diagnosed with CRC in the Netherlands population during the same period and in the same age range (50-76 years), whom were not offered CRC screening. 68

8 FIT INTERVAL CANCERS IN COLORECTAL CANCER SCREENING RESULTS A total of 27,30 members of the target population were eligible for FIT screening, of whom 18,716 (69%) participated at least once (Figure 1). Of these, 3,005 (16%) had a positive FIT in 1 of the screening rounds and 2,762 subsequently underwent a colonoscopy (92% adherence). In the total cohort of 27,30, CRC was detected in 269 invitees: in 160 participants (60%) and in 109 who never participated (0%). The cancers detected in the 160 participants were 116 SD-CRCs (72%), 27 FIT interval cancers (17%), 9 colonoscopy interval cancers (6%), 5 CRCs in FIT-positive participants not adhering to colonoscopy (3%) and 3 CRCs detected at surveillance (2%). These last 2 groups with a total of 8 CRCs - were excluded from further analyses. FIT interval cancer proportion and FIT CRC sensitivity The FIT interval cancer proportion was 23% (27 FIT interval cancers versus 89 SD-CRC after 3 completed screening rounds), reflecting a FIT sensitivity for detecting CRC of 77%. FIT levels of participants with SD-CRCs and FIT interval cancers are shown in Supplementary Table 2; 31% of those with a SD-CRC had a positive FIT between the range of 10 to 6 µg Hb/g feces. Twelve participants (%) with a FIT interval cancer had undetectable hemoglobin concentrations at the time of the last screening. For FIT interval cancers, Supplementary Table 3 shows the quantitative FIT level per case, the total number of rounds they participated and after which round the FIT interval cancer was diagnosed. Participant demographics Table 2 summarizes the characteristics of SD-CRC and non-sd-crc separately. No significant differences were found for age at diagnosis between participants with SD-CRC and individuals with non-sd-crc (P =.83) (Table 2). A colonoscopy interval cancer was detected more often in women Figure 1. Flowchart of the rounds of FIT-based CRC screening including screen-detected and nonscreen-detected colorectal cancers among participants (participating at least once) and non-participants (never participated). 69

9 CHAPTER than in men (56% versus %), but this difference was not statistically significant. There were also no significant differences in SES among the groups (P =.76). The age and sex distribution was comparable with that of the general population (not tested for statistical significance). Among the 9 participants with a colonoscopy interval cancer, the colonoscopies were reported as complete with cecal intubation. In 6 cases sufficient bowel preparation was reported, in 3 cases this data were missing. Three participants received a colonoscopy surveillance advice, but developed a CRC within this interval. One person received an advice of a 3-year interval and 1 person a 6-year interval, and both recommendations were in agreement with the Dutch Surveillance Guideline of One person received an adjusted advice of a 1-year surveillance interval because multiple adenomas were removed at index colonoscopy. Six participants were advised a 10-year screening interval as per protocol (See Supplementary Table ). Tumor location and stage distribution Tumor location and CRC stage distribution are described in Table 3 and Figure 2. Screen-detected CRC, FIT interval cancers, and CRC in nonparticipants were mostly located in the distal colon (71%, 63%, 62%, respectively), whereas colonoscopy interval cancers were more often located in the proximal colon (78%; P =.063). The stage distribution differed significantly between the groups, with more favorable stages in participants with SD-CRC (P <.001). Stage distribution was similar for participants with FIT interval CRC and for nonparticipants with CRC (P =.39). Participants with FIT interval cancers had more advanced stages than SD-CRC (P=.019). Tumor location and stage distribution in nonparticipants with CRC were comparable to those not offered screening (see Table 3). Table 2. Characteristics of participants Non-screen-detected CRC Screendetected CRC SD-CRC FIT interval cancer Colonoscopy interval cancer CRC in nonparticipant P value CRC in general population Total CRC * Age at diagnosis >70 28 (2%) 50 (3%) 38 (33%) 6 (22%) 11 (1%) 10 (37%) 1 (11%) 6 (67%) 2 (22%) 21 (19%) 9 (5%) 39 (36%) 1651 (20%) (%) (36%) Sex.709 Male Female 73 (63%) 3 (37%) 16 (59%) 11 (1%) (%) 5 (56%) 69 (63%) 0 (37%) 286 (59%) (2%) SES score.763 Low Average High 13 (11%) 82 (71%) 21 (18%) 2 (7%) 21 (78%) (15%) 2 (22%) 6 (67%) 1 (11%) unknown unknown unknown unknown unknown unknown * cases with 2 incidents, age at diagnosis first incident selected 70

10 FIT INTERVAL CANCERS IN COLORECTAL CANCER SCREENING Table 3. Tumor location and stage distribution of CRCs Screendetected CRC SD-CRC FIT interval cancer Non-screen-detected CRC Colonoscopy CRC in nonparticipant interval cancer P value CRC in general population Total CRC Tumor location*.063 Proximal Distal Unknown Stage I II III IV Unknown 3 (29%) 82 (71%) 0 (0%) 60 (52%) 16 (1%) 37 (32%) 3 (3%) 0 (0%) 10 (37%) 17 (63%) 0 (0%) 8 (30%) 6 (22%) 9 (33%) (15%) 0 (0%) 7 (78%) 2 (22%) 0 (0%) 2 (22%) 0 (0%) 1 (11%) 6 (67%) 0 (0%) 38 (35%) 67 (62%) (%) 17 (16%) 31 (29%) 36 (33%) 23 (21%) 2 (2%) < (33%) 7290 (65%) 136 (2%) 1002 (19%) 1838 (25%) (30%) (23%) 198 (2%) * proximal (=splenic flexure and proximal) Figure 2. Stage distribution for the different types of colorectal cancer as defined by Sanduleanu [7]. For definitions see Table 1. Survival Median follow up after diagnosis for SD-CRC, FIT-interval cancers, and for colonoscopy interval cancers was 6.1 months (IQR 18.1 to 72.1 months). There was a significant difference in survival between SD-CRC, FIT interval cancers, and colonoscopy interval cancers (P <.001; Figure 3). Those with colonoscopy interval cancers had the lowest survival: survival at 2 months was 67% (95% 71

11 CHAPTER CI 35-88%) compared to 85% (95% CI 68-9%) for FIT interval cancers and 93% (95% CI 87-96%) for SD-CRCs. Survival at 36 months was 56% (95% CI 27-81%), 85% (95% CI 68-9) and 91% (95% CI 85-95%) respectively. Table list the number of deaths in the respective groups. The number of deaths among persons with FIT interval cancers was lower than among non-participants with CRC and those not offered CRC screening (19% vs 0% and 35% respectively). Figure 3. Survival curves 72

12 FIT INTERVAL CANCERS IN COLORECTAL CANCER SCREENING Table. Mortality Screendetected CRC SD-CRC FIT interval cancer Non-screen-detected CRC Colonoscopy interval cancer CRC in nonparticipant P value CRC in general population Total CRC Number of deaths 13 (11%) 5 (19%) 5 (56%) (0%) < (35%) (all-cause mortality*) Follow up months after diagnosis CRC (median(iqr)) 50 (25 to 76) 0 (15 to 63) 19 (9 to 30) - - *At end of cohort DISCUSSION This study confirms that the proportion of interval cancers in FIT-based CRC screening seems to be lower than in guaiac-based fecal occult blood testing. Over rounds of screening, fewer than 1 in cancer cases comprised of FIT interval cancers. We observed a more favorable survival for participants with FIT interval cancers than for the population that was not offered screening (ie general population). Participants with a colonoscopy interval cancer had the worst outcome, which amplifies the importance of colonoscopy quality control. We are the first to describe interval cancers in a FIT-based screening program with long-term follow-up. Our cohort is sampled from an average-risk population, comprising all age ranges commonly invited for CRC screening programs worldwide, which makes our results applicable for many countries considering CRC screening. This population was screen-naive when first approached, without the presence of any other CRC screening initiatives in the population. All data were prospectively collected and all invitees were linked to the Netherlands Cancer Registry by the Netherlands Comprehensive Cancer Organization to identify all non-sd-crcs. The success of a FIT-based CRC screening program not only depends on FIT test characteristics (sensitivity and specificity) but also on adherence to screening with FIT and to follow-up colonoscopy in persons who are FIT-positive. Within our pilot study, high participation rates and high colonoscopy adherence rates were achieved, resulting in a substantial number of persons diagnosed with SD-CRC. To fully appreciate the reported data of this research study, these adherence rates should be confirmed in the setting of an organized population-based screening program. In 201, in the Netherlands a national FIT-based CRC screening program was implemented, resulting in even higher participations rates: 73% in Excluding those for whom colonoscopy was not recommended (n = 2637) during intake, uptake of colonoscopy was 83.1% [19 ]. This confirms that the high performance indicators achieved within the pilot study could possibly also be achieved in the setting of the national organized FIT-based screening program. To fully appreciate our findings, some limitations also need to be addressed. Little information was available about nonparticipants dying of CRC in our cohort, as these data were delivered anonymously for privacy reasons. Consequently, their SES could not be assessed. Previous studies 73

13 CHAPTER have shown that especially socioeconomically deprived persons are less likely to take part in CRC screening, even though they presumably are at a higher risk of developing CRC due to poor general health [20,21]. Last, their date of diagnosis could not be related to the date of invitation. It is possible that a person had already been diagnosed with CRC before being invited for screening. In such case, refraining from participation would have been a justified decision. Because FIT is more sensitive in detecting advanced neoplasia than gfobt, one would expect lower interval rates for FIT. A previous Scottish FIT-based study, however, reported a FIT interval cancer proportion of 51% using a cutoff of 80 µg Hb/g feces [22]. Italian researchers reported a FIT interval cancer proportion of 31% using a cutoff of 20 µg Hb/g feces after two rounds of FIT screening [23]. Our lower proportion of FIT interval cancers could probably be explained by the substantially lower cutoff (10 µg Hb/g feces), thereby detecting CRCs already at low Hb concentration. In our study, 31% of all SD-CRCs had low Hb concentration of < 7 µg Hb/g feces. The Scottish study investigated the effect of different cutoff levels of fecal-hb on interval cancer proportions and colonoscopy demand. It reported an interval cancer proportion of 51% at a cut-off of 80 µg Hb/g feces vs a proportion of 38% using a cutoff of 10 µg Hb/g feces with a significant increase in number of required colonoscopies for the latter cutoff [22]. Hence, choosing a very low positivity cutoff leads to a higher number of SD-CRCs and lower proportion of FIT interval cancers, at the cost of increased colonoscopy demand and increased numbers false positives [22]. Our results showed no evidence for gender differences in interval cancer rates, although previous studies have shown higher rates of interval cancers among women [8,23-25]. As in other studies, CRC was detected less frequently in women. Notably, more women had a colonoscopy interval cancer than men (56% versus %), yet this difference did not reach statistical significance. These results are in line with a Spanish study, reporting SD-CRC and interval cancers during rounds of FOBT screening mainly using gfobt and FIT in a small proportion of screenees [9]. The Scottish researchers also reported higher rates of FIT interval cancers in women, without reaching statistical significance [22,2]. Previous gfobt-based studies have shown that most SD-CRC cases were at an early stage, resulting in a better survival than observed in non-screen-detected cancer cases [,5,10,26]. In our study most screen-detected cases were stage I cancers which confirms the findings of previous studies [9,10,2,27]. In our FIT-based program, a higher percentage of stage I CRC were found than in previously reported gfobt screening programs. This could possibly be explained by the relatively low FIT cutoff and the higher sensitivity of FIT. Comparable early-stage distributions in FIT screening have been reported [22,26]. Most detected tumors were located distally, except for colonoscopy interval cancers, which were mainly located in the proximal colon. These findings are also in line with previous studies [9,28]. In contrast to previous literature reporting that prognosis of nonparticipants with CRC is poorer than that of participants with symptomatically diagnosed CRC, we did not observe this from our data [5,2]. The number of deaths was significantly lower in participants with FIT interval cancers than in nonparticipants with CRC. This is in contrast with findings of an English study, which reported similar outcomes for gfobt interval cancers and a control group [8]. Persons diagnosed with FIT 7

14 FIT INTERVAL CANCERS IN COLORECTAL CANCER SCREENING interval cancers showed better survival as compared to nonparticipants and the general population. This implies that the theoretical risk of persons seeking delayed medical consultation in case of the development of abdominal symptoms due to false reassurance caused by a recent negative FIT result seems, fortunately, to have no immediate negative effect on all-cause mortality within our cohort. These findings should be interpreted with caution because, in theory, FIT participants could have been in better health than the general population, resulting in a selection bias. In our cohort, 9 colonoscopy interval cancers were detected between 2006 and 2015, resulting in a colonoscopy interval cancer rate of 7.2%. Although the surprisingly high occurrence of colonoscopy interval cancers needs further attention, the rate is comparable to that in previously published screening studies, including another FIT-based program, reporting rates between 2% to 8 % [27,29-32]. As most colonoscopy interval cancers were detected in the proximal colon, we should consider procedural factors, especially missed lesions (for instance due to inadequate bowel preparation). In 30% of all persons diagnosed with colonoscopy interval cancers in our study, the bowel preparation score was not mentioned in the report. Because both FIT interval cancers (reflecting FIT CRC sensitivity) and colonoscopy interval cancers (partially reflecting quality of the colonoscopy) are important for the quality and success of a screening program, both should be carefully monitored. Five participants who refused colonoscopy after a positive FIT were later diagnosed with CRC. Unfortunately, no information on the reason for refusal was available. To facilitate early detection of these CRCs, additional strategies to inform participants about the need for follow-up after a positive FIT should be developed. Modifiable determinants for nonadherence to endoscopy, such as embarrassment, lack of knowledge about CRC, fear of the procedure, or inconvenience should be discussed. Other strategies could be built on sending all nonrespondents after a positive FIT result an additional information leaflet, or to offer them a consultation by phone. Overall, our results support the effectiveness of FIT screening programs with improved survival of screen-detected cancers and lower proportions of interval cancers than in guaiac-based fecal occult blood testing. Participants with colonoscopy interval cancer had the worst outcome, which amplifies the importance of colonoscopy quality control within screening programs. ACKNOWLEDGEMENTS The authors thank the registration teams of the Netherlands Comprehensive Cancer Organisation for the collection of data for the Netherlands Cancer Registry and the scientific staff of the Netherlands Comprehensive Cancer Organisation. We thank the Netherlands Organization for Health Research and Development of the Dutch Ministry of Health (ZonMW) for funding (project numbers , ). The authors thank all involved co-workers of the Foundation of Population Screening Mid-West and South-West (Bevolkingsonderzoek Midden-West, Bevolkingsonderzoek Zuid-West) for their important contributions to the study. We thank the gastroenterologists who performed all colonoscopies. 75

15 CHAPTER REFERENCES 1. Ferlay J, Parkin DM, Steliarova-Foucher E. Estimates of cancer incidence and mortality in Europe in Eur. J. Cancer 2010;6: Siegel R, Desantis C, Jemal A. Colorectal Cancer Statistics 201CA Cancer J Clin. 201;6: Dutch Cancer Registry. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. Minnesota Colon Cancer Control Study. N Engl J Med 1993;328: Hardcastle JD, Chamverlain JO, Robinson MH, et al. Randomised controlled trial of faecaloccult-blood screening for colorectal cancer. Lancet 1996;38: Kronborg O, Fenger C, Olsen J, et al. Randomised study of screening for colorectal cancer with faecal-occult-blood test. Lancet 1996;38: Sanduleanu S, le Clercq CM, Dekker E, et al. Definition and taxonomy of interval colorectal cancers: a proposal for standardising nomenclature. Gut 2015;6: Gill MD, Bramble MG, Rees CJ, et al. Comparison of screen-detected and interval colorectal cancers in the Bowel Cancer Screening Programme. Br J Cancer 2012;107: Garcia M, Domènech X, Vidal C, et al. Interval Cancers in a Population-Based Screening Program for Colorectal Cancer in Catalonia, Spain. Gastroenterol Res Pract 2015: Gill MD, Bramble MG, Hull MA, et al. Screendetected colorectal cancers are associated with an improved outcome compared with stage-matched interval cancers. Br J Cancer. 201;111: Tinmouth J, Lansdorp-Vogelaar I, Allison JE. Faecal immunochemical tests versus guaiac faecal occult blood tests: what clinicians and colorectal cancer screening programme organisers need to know. Gut 2015;6: Stegeman I, van Doorn SC, Mundt MW, et al. Participation, yield, and interval carcinomas in three rounds of biennial FIT-based colorectal cancer screening. Cancer Epidemiol. 2015;39: Kapidzic A, Grobbee EJ, Hol L, et al. Attendance and yield over three rounds of populationbased fecal immunochemical test screening. Am J Gastroenterol 201;109: Grobbee EJ, van der Vlugt M, van Vuuren AJ, et al. A randomised comparison of two faecal immunochemical tests in population-based colorectal cancer screening. Gut Aug 9. pii: gutjnl Rembacken B., Hassan C., Riemann J.F. et al. Quality in screening colonoscopy: position statement of the European Society of Gastrointestinal Endoscopy (ESGE). Endoscopy 2012;33: Nederlandse richtlijn coloscopie surveillance NVMDL nov sites/ Coloscopie_Surveillance_definitief_2013.pdf 17. Schlemper RJ, Riddell RH, Kato Y, et al. The Vienna classification of gastrointestinal epithelial neoplasia. Gut 2000;7: Edge SB, Compton CC. The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM. Ann Surg Oncol. 2010;17: Toes-Zoutendijk E, van Leerdam ME, Dekker E, et al. Real-Time monitoring of results during first year of Dutch colorectal cancer screening program and optimization by altering fecal immunochemical test cut-off levels. Gastroenterology;2016 Nov 2 doi: /j. gastro [Epub ahead of print] 20. Pornet C, Dejardin O, Morlais F, et al. Socioeconomic determinants for compliance to colorectal cancer screening. A multilevel analysis. J Epidemiol Community Health 2010;6: Moss SM, Campbell C, Melia J, et al. Performance measures in three rounds of the English bowel cancer screening pilot. Gut 2012;61: Digby J, Fraser C, Carey F, et al. Interval cancers using a quantitative faecal immunochemical test (FIT) for haemoglobin when colonoscopy capacity is limited. J Med Screen 2015;0:

16 FIT INTERVAL CANCERS IN COLORECTAL CANCER SCREENING 23. Zorzi M, Fedato C, Grazzini G, et al. High sensitivity of five colorectal screening programmes with faecal immunochemical test in the Veneto Region, Italy. Gut. 2011;60: Steele RJC, McClements P, Watling C, et al. Interval cancers in a FOBT-based colorectal cancer population screening programme: implications for stage, gender and tumour site. Gut 2012;61: Brenner H, Chang-Claude J, Seiler CM, et al. Interval cancers after negative colonoscopy: population-based case-control study. Gut 2012;61: Pande R, Froggatt P, Baragwanath. Survival outcome of patients with screening versus symptomatically detected colorectal cancers. Colorect Dis 2012;15: Mansouri D, McMillan DC, McIlveen E, et al. A comparison of tumour and host prognostic factors in screen-detected versus non screen-detected colorectal cancer: a contemporaneous study. Colorectal Dis. 2016;18: le Clercq CM, Sanduleanu S. Interval colorectal cancers: what and why. Curr Gastroenterol Rep. 201;16: Singh H, Nugent Z, Demers A, et al. Rate and predictors of early/missed colorectal cancers after colonoscopy in Manitoba: a population-based study. Am J Gastroenterol 2010;105: Bressler B, Paszat L F, Chen Z, et al. Rates of new or missed colorectal cancers after colonoscopy and their risk factors: a population-based analysis. Gastroenterology 2007;132: le Clercq CM, Bouwens MW, Rondagh EJ, et al. Postcolonoscopy colorectal cancers are preventable: a population-based study. Gut 201;63: Stoffel E., Erichsen R., Froslev T. et al. Clinical and molecular characteristics of postcolonoscopy colorectal cancer: a populationbased study. 77

17 CHAPTER SUPPLEMENTARY DOCUMENTS Supplementary Table 1. Basic characteristics of invitees per round per region. Round 1 Round 2 Round 3 Round MW SW MW SW MW SW MW SW Invitees 5028* Age (median, IQR) 59 (5-65) 60 (55-66) 59 (5-65) 61 (56-66) 60 (55-66) 60 (5-65) 61 (57-67) 61 (56-67) Sex (male) 285 (9) 779 (50) 981 (9) 3962 (8) 901 (9) 68 (9) 618 (9) 672 (8) *gfobt invitees excluded round 1. MW= Midwest region, SW=Southwest region Supplementary Table 2. Last measurement of Hb concentration level prior to CRC diagnosis. FIT Hb concentration (µg Hb/g feces) Screen-detected CRC; n FIT interval cancer; n 0-12 (%) (26%) (19%) (17%) (1%) (69%) - missing - 3 (11%) The 3 missing cases were FIT negative (Hb concentration level <10 µg Hb/g feces) but exact level was not documented. 78

18 FIT INTERVAL CANCERS IN COLORECTAL CANCER SCREENING Supplementary Table 3. FIT interval cancers Case Region Round incident Total rounds eligible Total rounds participated Interval diagnosis*(yr) FIT Hb concentration (µg Hb/g feces) 1 MW ,9 1,8 2 MW ,3-3 SW ,5 0, MW ,9 0 5 MW ,53-6 SW ,67,8 7 MW ,8-8 SW ,92 9,6 9 MW , MW , SW , MW , MW ,93 3,8 1 MW , MW , SW , MW , MW , MW , MW ,61 7,6 21 SW , MW , SW , SW , MW ,25 5, 26 MW ,29 7,6 27 MW ,3 0 *interval diagnosis= time between date of last negative FIT result and diagnosis of CRC.**missing FIT Hb concentration= The 3 missing cases were FIT negative (Hb concentration level <10 µg Hb/g feces) but exact level was not documented. MW= Midwest region, SW=Southwest region 79

19 CHAPTER Supplementary Table. Colonoscopy interval cancers Case Region Date FIT test 1 SW Nov SW Febr MW Jan 2007 SW Nov SW April SW May SW Nov MW Febr SW June 2008 Date Endoscopy Date diagnosis CRC Time Interval (yr) Endoscopic findings Bowel Prep score Cecal Intubation Surveillance advice Jan 2009 June ,58 8 adenoma with 1 HGD sigmoid Excellent yes 3 yr March 2010 Febr ,92 HP - yes No FU March 2007 July ,51 No findings Good yes No FU Nov 2009 Nov ,00 One TA<10 mm - yes 6 yr June 2009 June 2013,08 HP Good yes No FU July 2009 Sept ,33 >10 adenomas, 1 TVA Good yes 1 yr Dec 2008 Nov ,00 HP - yes No FU April 2012 Nov ,77 No findings Excellent yes No FU July 2008 Jan ,50 No findings Good yes No FU MW= Midwest region, SW=Southwest region, TVA= tubulovillous adenoma with low grade dysplasia, TA= tubular adenoma HP=hyperplastic polyp, HGD= high grade dysplasia. FU=follow up. 80