Objective Limited data exist on attendance and additional yield of 2-sample faecal immunochemical testing (FIT) screening during multiple rounds. We therefore conducted a population-based colorectal cancer screening trial comparing attendance and yield of repeated 1-sample and 2-sample FIT screenings.
Design Two randomly selected groups of average-risk subjects aged 50–74 years were invited for two rounds of either 1-sample (n=5007) or 2-sample (n=3197) FIT (OC-sensor Micro) screening. The test was considered positive if at least one sample was positive (cut-off 50 ng/mL; 10 µg haemoglobin/g).
Results The cumulative attendance rate was similar for repeated 1-sample and 2-sample FIT screenings (1-sample FIT: 68.1%; 2-sample FIT: 67.1%, p=0.368). The positivity rate in the second round was lower for 1-sample FIT (6.2%, 95% CI 5.4% to 7.2%) than for 2-sample FIT (8.4%, 95% CI 7.1% to 9.8%, p=0.007), whereas the detection rate of advanced neoplasia (AN, 1-sample FIT: 1.9%, 95% CI 1.2% to 2.2%; 2-sample FIT: 1.7%, 95% CI 1.2% to 2.5%, p=0.861) and the positive predictive value (1-sample FIT: 32%, 95% CI 24% to 40%; 2-sample FIT: 21%, 95% CI 15% to 29%, p=0.075) did not differ. After two rounds of screening, the cumulative diagnostic yield of AN for 1-sample FIT was 29.3 per 1000 invitees, compared with 34.0 for 2-sample FIT (p=0.241).
Conclusions Using 2-sample FIT instead of 1-sample FIT does not result in a higher detection rate of AN in the second round of repeated FIT screening. Furthermore, both strategies lead to a similar yield of AN over two rounds. These findings imply that 1-sample FIT screening is preferred over 2-sample FIT screening.
- COLORECTAL CARCINOMA
- COLORECTAL CANCER SCREENING
Statistics from Altmetric.com
Significance of this study
What is already known on this subject?
Faecal immunochemical testing (FIT) is superior to guaiac-based faecal occult blood testing in terms of higher attendance and superior detection rate.
FIT screening requires successive screening rounds for optimal programme sensitivity.
Screening by means of two-sample FIT increases test sensitivity (ie, reduces the risk of missed advanced lesions).
One-sample and two-sample FIT screenings are associated with a similar adherence in a first screening round.
In a first screening round, two-sample FIT screening leads to a higher detection rate of advanced neoplasia (AN) than one-sample FIT screening.
What are the new findings?
Two-sample FIT screening is associated with a stable and high attendance during repeated screening rounds.
The detection rate of AN of two-sample FIT screening decreases significantly in the second round compared with the first screening round.
Both strategies lead to a similar yield of AN over two rounds.
How might it impact on clinical practice in the foreseeable future?
These findings imply that one-sample FIT screening is preferred over two-sample FIT screening.
Repetitive screening with guaiac-based faecal occult blood tests (gFOBTs) reduces colorectal cancer (CRC)-related mortality.1 This effect of gFOBT screening on CRC mortality persists for many years.2 Today, the faecal immunochemical test (FIT) is replacing the classical gFOBT as a screening method, based on acceptability and effectiveness leading to a superior detection rate (DR) of subjects with AN.3 ,4 The possibility of adjusting the positivity rate (PR) of FIT to meet available resources provides a significant additional benefit to FIT screening.5–7 Population-based FIT screening is therefore currently being implemented in several countries, including the Netherlands.4 ,8
AN can bleed intermittently and therefore may be missed with a single stool sampling. Screening by means of a two-sample FIT increases test sensitivity (ie, reduces the risk of missing advanced lesions). In the first round of this study, comparing one-sample versus two-sample FIT screening showed no differences in attendance rate yet a significantly higher DR of AN with first round two-sample FIT screening was reported.8 Two-sample FIT screening thus seems more effective than one-sample FIT screening in a first screening round. It is unknown whether this advantage persists over repeated screening rounds, a prerequisite for optimal screening by means of FIT. Until now, data on successive two-sample FIT screening are lacking. Trials pertain to a single round of FIT screening using multiple samples, and trials comparing one-sample and two-sample FIT screenings with regards to attendance and diagnostic yield are lacking.7 ,9–13 The aim of this study was to determine attendance and diagnostic yield of repeated two-sample FIT screening. Furthermore, we aimed to compare these data with repeated one-sample FIT screening.8
Details about the design of this ongoing population-based CRC screening programme have been described previously, as the first round of this trial comparing one-sample and two-sample FIT screenings has been published previously.3 ,8 ,14 In brief, demographic data of all individuals between the ages of 50–74 years in the southwest of the Netherlands were obtained from municipal population registers. Two random samples were taken from the target population by a computer-generated algorithm (Tenalea, Amsterdam, the Netherlands). Allocation to one-sample or two-sample FIT screening occurred prior to invitation. In the repeated two-sample FIT screening group, analysis of the first FIT in the second screening round allowed us to assess screening performance with two-sample FIT in the first round and one-sample FIT in the second round (scenario A). Since there was no CRC screening programme at the time of the trial in the Netherlands, the target population was screening-naïve when first approached. Individuals with a history of IBD or CRC, as well as those who had undergone a colonoscopy, sigmoidoscopy or barium contrast enema in the last 3 years and those with an estimated life expectancy of less than 5 years were excluded from the study. Recruitment took place between November 2006 and May 2011. All subjects in the two-sample FIT group were invited after a 2-year interval. The one-sample FIT group consisted of one group invited after a 2-year interval and one group invited after a 3-year interval. Since no difference was observed in DR and positive predictive value (PPV) for AN between the two groups, we did not expect that the difference in interval would lead to differences in test characteristics in the second round.14
Faecal immunochemical test (FIT)
One or two FITs (OC-sensor Micro, Eiken Chemical, Japan) were sent by mail. Subjects were referred for colonoscopy if at least one sample tested positive (haemoglobin concentration ≥50 ng/mL; 10 µg haemoglobin/g faeces). Subjects with a positive FIT were scheduled for colonoscopy within 4 weeks. In case of incomplete colonoscopy, a CT-colonoscopy was performed. All colonoscopies were performed by experienced endoscopists (>1000 colonoscopies performed). The maximum reach of the endoscope, the quality of bowel preparation, as well as the characteristics and location of any polyps were recorded. All removed polyps were evaluated by experienced GI pathologists. Patients with a positive colonoscopy entered a surveillance programme according to the guideline of the Dutch Society of Gastroenterology, while patients with a negative colonoscopy were referred back to the screening programme, but were considered not to require FIT screening for 10 years. Screen-detected cancers were defined as cancers identified at colonoscopy performed after a positive test result.
Analyses were performed for two rounds of one-sample FIT (cut-off 50 ng/mL; FIT50), two rounds of two-sample FIT (if at least one of both tests was positive at a cut-off of 50 ng/mL), and first round screening with two-sample FIT and the second round with one-sample FIT (only analysing the first performed FIT at a cut-off of 50 ng/mL; scenario A).
For each screening round, we calculated the attendance rate, the PR, the DR of CRC and advanced adenomas (together defined as AN), and the PPV for CRC and advanced adenomas. The attendance rate was calculated by dividing the number of participants by all eligible subjects (defined as all invitees minus the excluded subjects). The PR was defined as the proportion of participants having a positive test result. The DR was defined as the proportion of participants having AN. This was calculated as the number of screened individuals with AN divided by all screened individuals with an analysable FIT. When more than one lesion was present, the screenee was classified according to the most advanced lesion. The number needed to screen (NNscreen) was calculated as the number of analysable FITs needed to find one AN or CRC. The PPV describes the number of AN among screenees with a positive FIT, who underwent a colonoscopy or in case the colonoscopy was incomplete a CT-colonography. The number needed to scope (NNscope) describes the number of colonoscopies to find one screenee with an AN or CRC.
For the cumulative test characteristics, we combined counts of the two screening rounds to acquire new numerators and denominators. For the calculation of the diagnostic yield over two rounds we considered the total number of advanced lesions and all individuals who had been eligible at least once over two screening rounds (eg, invitees of 75 years and older during the second round were considered eligible invitees). This strategy most closely mimics population-based screening programmes. An ex post calculation of the power based on differences in diagnostic yield with one-sample and two-sample FIT screenings was performed by comparing β and the SE. Differences in proportions between groups were analysed by χ2 testing. Differences in means between groups were tested using Student's t test. Proportions were displayed with 95% CI. The cumulative attendance was defined as the number of eligible invitees attending at least once. To assess differences in attendance rate between the two rounds of two-sample FIT screening, a generalised estimating equation was used to account for clustering at the level of the invitee. All p values were two-sided and considered statistically significant if <0.05. All tests were conducted using SPSS V.20.0.
For one-sample FIT, in total 4151 individuals were invited for a second round of screening. In total 13.8% of invitees were not invited again for the second screening round due to a positive test during the first screening (n=241), becoming 75 years of age or older (n=299), death (n=77) or moving out of the region (n=76). Another 1.8% of invitees were excluded (64 individuals met one of the exclusion criteria, seven had moved away and three had died). Of the 4077 eligible invitees in total 63.2% attended a second round of screening. The test was analysable in 99.7%. The majority of participants of the first screening round also attended the second round (91.1%; 95% CI 89.9% to 92.2%). The cumulative attendance rate over two rounds was 68.1%.
For two-sample FIT, a total of 13.2% of invitees of the first round were not invited again for the second screening round (239 subjects had tested positive during the first screening, 115 individuals had become 75 years of age or older, 43 had died, and the remaining 24 subjects had moved out of the region). In total 2636 average-risk subjects were invited for the second screening round of which 2.2% were excluded (51 individuals met one of the exclusion criteria, 5 had moved away and 1 had died). Out of 2579 eligible invitees, 61.3% (95% CI 59.4% to 63.2%) responded to the second round two-sample FIT invitation. The attendance rate of two-sample FIT screening was stable over the two screening rounds (61.3% in both rounds, p=0.992). The two FIT samples were analysable in 99.9% (figure 1). The cumulative attendance rate over two rounds was comparable to one-sample FIT screening (67.1%, p=0.368).
Test characteristics of two rounds of one-sample and two-sample FIT screenings
The test characteristics of two rounds of one-sample FIT screening at a cut-off of 50 ng/mL (FIT50) are displayed in table 1. In the second round, the PR of two-sample FIT was higher than that of one-sample FIT (p<0.001). The DR of AN and therefore the NNscreen of one-sample versus two-sample FIT were similar in the second round (p=0.861). The PPV and NNscope for AN of one-sample and two-sample FIT were also similar in the second round (p=0.075).
The test characteristics of two rounds of two-sample FIT50 screening are also displayed in table 1. The PR of two-sample FIT screening was significantly higher in the first round compared with the second round (p<0.001). Of the participants with at least one positive test in the second round, 97.0% underwent a colonoscopy. Four subjects with a positive FIT refused colonoscopy (figure 1). In two subjects the colonoscopy was incomplete and therefore an additional CT-colonography was performed. These did not reveal abnormalities. The DR of AN of two-sample FIT screening was significantly higher in the first round compared with the second round (p<0.001). Therefore, the NNscreen was lower in the first round compared with the second round (p<0.001). Moreover, the PPV was significantly higher and the NNscope was significantly lower in the first round (p=0.011).
Test characteristics after two screening rounds
The cumulative test characteristics after two screening rounds of one-sample and two-sample FIT50 are shown in table 2.
After two rounds two-sample FIT showed a higher cumulative PR than one-sample FIT (p<0.001). Two-sample FIT furthermore showed a lower cumulative PPV than one-sample FIT (p=0.024). No difference was seen in cumulative DR between both strategies (p=0.199). We were also able to derive from two-sample FIT the cumulative test characteristics of FIT screening with two-sample FIT in the first round and one-sample FIT in the second round. This scenario is referred to as scenario A. The cumulative test characteristics of scenario A are described in table 2 and showed a higher cumulative PR than one-sample FIT (p<0.001). No differences were seen in cumulative DR and PPV between scenario A and two-sample FIT (DR: p=0.471; PPV: p=0.095).
Table 3 displays the diagnostic yields of one-sample FIT, two-sample FIT and scenario A. The diagnostic yield after one round was 19.2 (95% CI 15.7 to 23.5) and 25.2 AN per 1000 invitees (95% CI 20.2 to 31.4), respectively. After two rounds, the diagnostic yields of one-sample FIT, two-sample FIT and scenario A increased to 29.3 (95% CI 24.9 to 34.5), 34.0 (95% CI 28.1 to 41.1) and 31.7 AN per 1000 invitees (95% CI 26.1 to 38.6), respectively. The diagnostic yields of one-sample FIT, two-sample FIT and scenario A after one and after two screening rounds was similar (table 2). With two-sample FIT 7 (95% CI 3 to 15) additional advanced lesions were found compared with scenario A, for which 43 additional colonoscopies had to be performed (NNscope=6.1). The NNscope to detect one advanced lesion was 4.3 for the second round one-sample FIT.
After one screening round a significantly higher diagnostic yield was seen with two-sample FIT than with one-sample FIT at cut-off levels of ≥75 ng/mL, ≥100 ng/mL and ≥125 ng/mL. After two rounds a higher diagnostic yield was seen with two-sample FIT screening at a cut-off level of ≥100 ng/mL and ≥125 ng/mL (table 3).
To our knowledge, this is the first population-based study to evaluate the participation and diagnostic yield of repeated two-sample FIT-based CRC screening. Two-sample FIT screening is associated with a stable and high attendance similar to screening with one-sample FIT. Our study demonstrated that second round two-sample FIT screening yields fewer AN compared with the first screening round. The cumulative diagnostic yield of two-sample FIT screening at a cut-off of 50 ng/mL or first round two-sample and a second round one-sample FIT screening (scenario A) was similar to two rounds of one-sample FIT screening. Our results support a preference for one-sample FIT screening. Given the lack of information on this matter, these data are of utmost importance for countries considering or planning the implementation of population-based FIT screening.
One of the most important factors to be taken into account when choosing a screening strategy is the acceptability of that strategy. The stable and high attendance over two rounds similar to one-sample FIT screening underlines the acceptability of a two-sample regimen. This underlines that the burden of FIT sampling of consecutive bowel movements does not impair participation to screening. Furthermore, the number of AN detected determines the effectiveness of CRC screening. The optimal number of FITs is, in this respect, very relevant. In this study we were able to compare one-sample and two-sample FIT screenings in the second round: two-sample FIT resulted in a higher PR, but a similar DR and PPV compared with one-sample FIT. In addition, the NNscope (6.1) to detect one additional advanced lesion when switching from the one-sample to the two-sample strategy in the second round was higher than the NNScope for one-sample FIT screening (4.3). Therefore, the additional value of a second test in the second round seems to be limited, since more colonoscopies are required to detect an additional advanced lesion.
We found a higher PR, but lower PPV with two-sample FIT versus one-sample FIT. The higher PPV together with a lower PR after two screening rounds would make one-sample FIT more favourable over screening with two-sample FIT, since the PPV is a measure for efficient use of colonoscopy resources. As screening colonoscopies are performed on healthy individuals, the number of unnecessary colonoscopies must be brought to a minimum, in particular since all colonoscopies carry a small risk of serious complications, such as bleedings and perforations.15 ,16 The diagnostic yield over two rounds (ie, the cumulative sensitivity of several screening rounds) did not significantly differ between one-sample and two-sample FIT50 screenings. A higher diagnostic yield was seen at higher cut-off levels with two-sample FIT screening after one and after two screening rounds. This again demonstrates that FIT screening using a low cut-off value of 50 ng/mL is highly effective. Adding a second test should only be considered at higher cut-off levels. This finding is supported by a cost-effectiveness analysis reporting that intensifying screening with one-sample FIT50 over multiple screening rounds was more cost-effective than providing two-sample FIT within one screening round.17 It was therefore recommended to increase the number of screening rounds with one-sample FIT50, before considering to increase the number of FIT samples provided per screening round.
Some limitations must be acknowledged. First, small differences in performance of one-sample versus two-sample FIT screening may become significant when applied to very large populations. An ex post calculation of the power showed that the observed differences in diagnostic yield between one-sample and two-sample FIT screening could have become significant at a cut-off of 50 ng/mL if the study population had been at least 2.8 times as large. From a public health perspective, this could imply a difference in diagnostic yield between one-sample and two-sample FIT screenings at a cut-off of 50 ng/mL. This should be a topic for further large population studies. Second, more awareness about CRC and advanced adenomas may have occurred over time, in particular since we started at baseline with recruitment of subjects in a screening-naïve population. No differences were however seen in attendance between the two screening rounds.
In conclusion, two-sample FIT screening is associated with a stable and high attendance. The DR of AN considerably decreases in the second round of two-sample FIT screening. Using two-sample instead of one-sample FIT during the second screening round does not result in a higher DR in a second screening round. The diagnostic yield of two rounds of two-sample FIT is similar to two rounds of one-sample FIT screening (FIT50). This implies that at low cut-off levels one-sample FIT screening should be preferred over two-sample FIT screening.
The authors thank H ‘t Mannetje for retrieval of demographic data of all potential participants in the target population, E van der Donk (Tenalea) for the random selection of invitees, all general practitioners in the region, gastroenterologists and surgeons of the Erasmus University Medical Centre, IJsselland Hospital, St Franciscus Gasthuis Hospital, Vlietland Hospital, Haven Hospital, Ikazia Hospital, Medical Centre Rijnmond-South and Albert Schweitzer Hospital, residents, secretaries, nurses and all participants of the trial.
Contributors EJK, MvB and MEvL: study concept; EJK, MvB and MEvL: study design; EJK: supervised the execution of the study; LH, AHCvR and AK: performed the retrieval of the population sample and the randomisation in collaboration with the Regional Organisation for Population Screening South-West Netherlands, Rotterdam and Tenalea, Amsterdam; J.C.I.Y. Reijerink and H. ‘t Mannetje: retrieval of the target population from the municipal registries and all mailings; AJvV, J. Francke, M. Ouwendijk, A. Heijens and N. Nagtzaam: analysis of all FIT samples; MEvL, LH and AHCvR: the database design; AHCvR and AK: data entry; A.C.M. van der Togt, and J.C.I.Y. Reijerink: coordinated the daily process; AK: drafted the report; LH, AHCvR, MvB, IL-V, WS, KI, MEvL and EJK: critical revision of the manuscript for important intellectual content; AK and C.W.N. Looman: performed the statistical analyses. All co-authors read and approved the final version of the paper.
Funding This trial was funded by the Dutch Cancer Society (EMCR 2006-3673), the Dutch Ministry of Health, Health Care Prevention Program—Implementation (ZonMw 63300022 and ZonMw 120720011), Olympus Medical Systems Europe GmbH, Hamburg, Germany, the Jacoba Foundation, and Eiken Chemical Co, Tokyo, Japan. None of them was involved in the study design, collection, analysis, and interpretation of the data and in the writing of the report.
Competing interests None declared.
Patient consent Obtained
Ethics approval The study was approved by the Dutch Ministry of Health (PG/ZP 2.727.071 and PG/ZP 2.823.158). The study letters and information brochures were approved by the Institutional Review Board at the Erasmus University Medical Centre (MEC-2005-264 and MEC-2008-029).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.