DOI 10.1515/cclm-2013-0232 Clin Chem Lab Med 2013; 51(11): 2173–2180
Yanqin Huang, Weiting Ge, Viktoriya London, Qilong Li, Shanrong Cai, Suzhan Zhang and Shu Zheng*
Diagnostic inconsistency of faecal immunochemical tests for haemoglobin in population screening of colorectal cancer Abstract Background: There is currently very little data available on the consistency of quantitative and qualitative faecal immunochemical test (FIT) for colorectal cancer screening. Methods: A representative random population (n = 1889, 40–74 year olds) in Jiashan, China was invited for FIT screening in 2012. Faecal samples were collected by a single specimen collection device and simultaneously tested by a quantitative FIT (OC-SENSOR, OC) and two qualitative FITs (FIT A and FIT B with intrinsic positive haemoglobin cut-off concentrations of 20 μg Hb/g faeces and 40 μg Hb/g faeces, respectively). The observational criteria for a positive result of the qualitative FIT were set according to the density of the colour appearing in the test strip. The results produced by the quantitative and qualitative FIT for each sample were compared. κ coefficient was used to measure consistency. Results: A total of 1368 (72.4%) individuals returned faecal samples. Both FIT A and FIT B precisely identified all faecal samples with haemoglobin concentration above 100 μg Hb/g faeces, but the overall consistency was poor for OC & FIT A (κ = 0.32, 95% CI 0.20–0.44) and was moderate for OC & FIT B (κ = 0.74, 95% CI 0.64–0.85). A more favourable consistency (κ = 0.64, 95% CI 0.57–0.72) was achieved when a different positive criterion was employed for FIT A. Conclusions: The diagnostic inconsistency between quantitative and qualitative FITs mainly exists in the faecal samples with low haemoglobin concentrations. Refining the criterion for a positive result may be a feasible way to improve the accuracy of qualitative FIT. Keywords: colorectal cancer screening; faecal immunochemical test; haemoglobin; quality assurance. *Corresponding author: Shu Zheng, Jiefang Road No. 88, Hangzhou 310009, China, Phone: +86 57187784501, Fax: +86 57187214404, E-mail: [email protected]
Yanqin Huang, Weiting Ge, Shanrong Cai, Suzhan Zhang and Shu Zheng: Cancer Institute (Key Laboratory of Cancer Prevention and Intervention, China National Ministry of Education; Key Laboratory of Molecular Biology in Medical Sciences, Zhejiang Province, China), The Second Affiliated Hospital, Zhejiang University School of Medicine, P.R. China Viktoriya London: Case Western Reserve School of Medicine, USA Qilong Li: Jiashan Institute of Cancer Prevention and Treatment, Zhejiang Province, P.R. China
Introduction Faecal occult blood test (FOBT) has been widely used as a screening tool for colorectal cancer. Early products of FOBT use chemical reagents to detect peroxidase activity of haem, by which the presence of haemoglobin is determined. The guaiac-based FOBT (gFOBT) is the typical test of the early FOBT products. However, the later developed faecal immunochemical test (FIT) for haemoglobin has been shown to be better than the original gFOBT [1]. It has better sensitivity and specificity for predicting colorectal neoplasms in an asymptomatic population [2]. It does not require diet restriction and only needs one faecal sample for testing, therefore, it significantly improves the compliance rate of colorectal cancer screening [3, 4]. There are currently two major forms of commercially available FIT products [5]: the quantitative FIT and the qualitative FIT. A lot of comparative studies have evaluated the screening performances of different FOBT products in colorectal cancer screening programmes. However, most of the studies focused on analysis of one type of FIT versus the gFOBT [6–15]. Few studies analysed the differences in the diagnostic performance of different types of FIT products. In 2009, Hundt et al. [15] first compared six qualitative FITs in a colorectal screening programme and observed a large difference in the performance characteristics of the different qualitative FITs. Brenner et al. [16] measured diagnostic consistency of the same six qualitative FITs and reported that the definition of the cut-offs for qualitative
2174 Huang et al.: Diagnostic inconsistency of FIT FIT was a critical issue of the test. Recently, some other experts [17–19] also expressed concerns about the diagnostic consistency of different types of FITs. Theoretically, the variant performances of different FIT products can be caused by the characteristics of the screening populations, the specimen devices, the faecal preservatives as well as the testing methodologies. Until now, very little has been studied about the effect that the testing methodologies have on the diagnostic consistency of different types of FIT products. In attempt to evaluate the effects of the testing metho dologies on diagnostic consistency, we designed a comparative study between a quantitative FIT that is widely used in Western Europe [8], Israel [20] and Japan [21] and two qualitative FITs that are widely used in China [22, 23]. In this study, we collected 1368 faecal samples from a representative random sample of population by a single specimen collection device which enclosed with the same type of faeces preservative and simultaneously tested each sample by the three FIT products mentioned above.
Materials and methods The quantitative FIT The quantitative FIT (OC-SENSOR, OC for short) was manufactured by Eiken Chemical Co., Ltd (Tokyo, Japan). All of the reagents used by OC were provided by the Nine-Strong Bioengineering Co., Ltd (Beijing, China). The OC is a fully automated instrument for quantitative measurement of faecal haemoglobin concentrations, requiring only a manual insertion of 10 sample tubes into the tube bed, one sample at a time. A paper print-out of the sample reading results is automatically generated in seven minutes after the initiation of reading. The mass of the tested faeces in each Eiken sampling device is 10 mg and the volume of the preservative buffer in Eiken sampling device is 2 mL. According to the manufacturer’s user manual, the OC produces reliable results in the range of 10–400 μg Hb/g faeces with an accuracy of 0.2 μg Hb/g faeces. The quality control process of the quantitative FIT involves regenerating the calibration curve for each batch of reagents and testing standards to make sure that the test results are within the permissible scope.
the manufacturers, the minimum haemoglobin concentrations for positive results are 100 ng haemoglobin per 1 mL faecal sample solution (ng/mL) for FIT A and 200 ng/mL for FIT B. The mass of faeces dissolved in the perservative of the specimen device which was used in our study was 10 mg and the volume of the preservative in a single specimen device was 2 mL, hence the minimum positive haemoglobin concentrations of FIT A and FIT B can be expressed as 20 μg Hb/g faeces and 40 μg Hb/g faeces, respectively. All the products used were within their quality guarantee periods. The quality control process of the qualitative FIT is to make sure that the control strip always display a dark red colour in every test, otherwise the test is regarded as null.
Population and faeces sampling We invited 1889 adults, aged 40–74 years, who reside in the Datong village in Jiashan County, Zhejiang Province, China to participate in a FIT-based colorectal cancer screening programme between the dates of March 26th to April 20th, 2012. Specimen collecting devices were distributed to participants and the participants were taught to sample and preserve their faecal samples by the nurses from a local community health service centre. Every participant was asked to submit one faecal sample from a single bowel movement. No diet restriction was required. Faeces were sampled with a serrated probe attached to the lid of the specimen device and only the faeces clinging to the grooves of the probe were brought into the chamber and dissolved in preservative buffer. Each specimen device was labelled with an order number and the name of the sample donor. A trained villager collected faecal samples from participants and sent them to the local community health service centre at 10:00 and 16:00 on a daily basis. A professional inspector in the local community health service centre tested each faecal sample by OC, FIT A and FIT B at the same time and on the same day the faecal sample was received. Personal and medical information of the sample donors was blind to the inspector. This study was approved by the Institutional Review Board on Medical Research, the Second Affiliated Hospital, Zhejiang University School of Medicine in accordance with the ethical principles in declaration of Helsinki (Oct 2008). A written informed consent was obtained from each subject before the faecal sampling tube was distributed.
A
Test strip FIT A Test positive (Grade 3) FIT B Test negative
The qualitative FITs The two qualitative FITs were purchased from two large manufacturers in China and were labelled as FIT A (Acon Co., Ltd, Hangzhou, China) and FIT B (W.H.P.M Co., Ltd, Beijing, China). Both products were certificated by State Food and Drug Administration of China. Both of the qualitative FITs are approximately 10 cm long, 3 mm wide and 1.5 mm thick. They are designed with a sample suction part at one end, a testing strip in the middle, with a holder located on the opposite end (Figure 1A). According to the user manual provided by
Control strip
Holding patch
Test patch
Sample sucking patch
B G0 G1 G2 G3 G4 G5 G6 G7 G8 G9 G10
Colour grades for FIT A Colour grades for FIT B
Figure 1 The qualitative FIT and the colour grade ladders. (A) The surface view of the qualitative FIT with their functional domain illustrated. (B) The colour grades ladders for FIT A and FIT B.
Huang et al.: Diagnostic inconsistency of FIT 2175
Faecal immunochemical test for haemoglobin Before inserting a specimen device with faeces into the OC, we pipetted out 500 μL of faecal solution from the specimen device and evenly distributed the faecal solution into two wells in a clean 96-well plate. Water suction ends of FIT A and FIT B test strips were inserted in the wells right after pipetting the sample solutions. The inspector read the result in 5–10 min for each qualitative FIT and recorded the result by taking a picture of the qualitative FIT and the specimen device. The test results of OC were printed out on a paper sheet, then were entered into a computer Excel spread sheet, in accordance to the order numbers and the sample donors. The inspector strictly abided by the quality control stipulations recommended by the manufacturers. All the tests of the standards were within the permissible scopes of the quality requirements for OC.
Measurement of the influence of Eiken preservative buffer to the qualitative FIT The preservative buffer in the specimen device was from Eiken Chemical Co., Ltd. The influence of Eiken preservative buffer on the test results of the qualitative FIT was not previously known. We acquired 20 fresh peripheral blood samples from the clinical laboratory of the Second Affiliated Hospital, Zhejiang University School of Medicine. First, we repeatedly diluted each blood sample by double distilled H2O until the haemoglobin concentration of the diluted sample was around 10 μg Hb/mL (10 ng Hb/μL). Next, a total of 800 μL Eiken preservative buffer and 800 μL preservative buffer provided by the manufacturers of FIT A were evenly pipetted into eight wells of a clean 96-well plate, 200 μL for each well. Aliquots of 1 μL, 2 μL, 4 μL and 8 μL of the diluted blood sample were then pipetted into four wells which were filled with Eiken preservative buffer and into another four wells which were filled with preservative buffer for FIT A. After gently mixing the liquid for a few seconds, the sample suction end of FIT A was insert into each well and the test result was measured within 5–10 min after the insertion. FIT B was also tested by the same method. Finally, the test results of the qualitative FIT with Eiken preservative buffer and that testing with the preservative buffer provided by the manufacturers were compared.
Positive criteria for the qualitative FIT In order to evaluate the relationship between the colour appearing on the test strips of the qualitative FIT and the haemoglobin concentrations of the faecal sample, we tentatively built up 10 “grades” of positive criteria for the qualitative FIT by measuring the density of the colour in the test strips of the products. Photographs of the tested qualitative FIT were loaded on a computer by Firework CS5 (Adobe Systems Incorporated, USA) software. The red part of the control strip was cropped out from the picture and the transparency index of the control strip picture was adjusted from 10% to 100%, at a 10% intervals, thus producing a colour ladder from faint pink to dark red with 10 different grades (Figure 1B). The colour bands on the control strips of FIT A and FIT B were uniformly dark red. Therefore, the colour ladders of FIT A and FIT B were quite similar. The 10 different colour
grades on the ladder were then numbered from G1 to G10, with G1 indicating the faintest pink colour and G10 corresponding to the darkest colour. G0 indicated no colour change in the test strip. The test result of each sample was recorded as G0 to G10 by comparing the test strip colour to the developed colour ladders. If a colour grade (from G1 to G10) was designated as the observational positive cut-off grade of a qualitative FIT, the faecal sample with a test strip colour darker than the designated colour grade was regarded as positive case. We selected 50 test strip photographs of the qualitative FITs which displayed different colour grades (27 G0, 8 G1, 6 G2, 5 G3 and 4 G4) to test the reproducibility of the positivity observations. We invited five students to grade the diagnosis of the 50 pictures by referencing the colour ladders on a computer monitor. The colour grades assigned by the students for each sample photograph were then compared to the grades assigned to the same photograph by the inspector from the local community health service centre. κ coefficients were calculated to measure their diagnostic consistency.
Data analysis The positive rates for faecal haemoglobin based on OC measurements were calculated at different haemoglobin cut-off concentrations. The faecal samples with haemoglobin concentrations above or below a certain set concentration measured by OC testing were further analysed by comparing them to the test results of FIT A and FIT B. κ coefficient was used to measure diagnostic consistency of the quantitative and qualitative FITs. The 95% confidence interval (95% CI) of κ was calculated. Statistical analysis was done by SPSS 17.0 software (IBM, USA).
Results We compared the test results of FIT A and FIT B for different concentrations of blood samples which were dissolved in Eiken preservative buffer and the preservative buffer provided by the manufacturers of the qualitative FIT. For the diluted blood samples with the same haemoglobin concentration, no difference was found in the test strips of the qualitative FITs done with Eiken faecal buffer versus the FIT done with the preservative buffer provided by the manufacturers of the qualitative FIT. We measured the inter-consistency of the positivity observation for 50 photographs of 50 faecal samples. The κ coefficients of the observed colour grades between the inspector from the local community health service centre and the five students were 0.91 (95% CI 0.81–1.00), 0.85 (95% CI 0.73– 0.97), 0.91 (95% CI 0.81–1.00), 0.79 (95% CI 0.65–0.92) and 0.91 (95% CI 0.81–1.00), respectively. A total of 1368 (72.4%) subjects returned 1368 faecal samples, one sample per subject, that were collected mostly during the second and third day after receiving the specimen collecting device. There were 687 male and 681 female
a μg Hb/g means haemoglobin concentrations of the faeces sample detected by OC; bOC test samples (%) means the number of samples with haemoglobin concentrations above or below certain levels which were defined by OC testing and its percentage in the number of total samples. cFIT A test samples (%) means that the samples with certain haemoglobin concentrations defined by OC were tested by FIT A and the test results of FIT A were subclassified into 11 groups according to darkness of the colour appearing in the test trips of FIT A.
5 (26.3) 5 (14.7) 5 (11.1) 5 (6.6) 0 (0) 0 (0) 0 (0) 5 (0.4) 13 (68.4) 15 (44.1) 15 (33.3) 15 (19.7) 0 (0) 0 (0) 0 (0) 15 (1.1) 17 (89.5) 22 (64.7) 22 (48.9) 22 (29.0) 0 (0) 0 (0) 0 (0) 22 (1.6) 18 (94.7) 24 (70.6) 24 (53.3) 25 (32.9) 3 (0.2) 3 (0.2) 3 (0.3) 28 (2.1) 19 (100.0) 28 (82.4) 29 (64.4) 33 (43.4) 11 (0.9) 9 (0.7) 9 (0.7) 44 (3.2) 19 (100.0) 31 (91.2) 33 (73.3) 43 (56.6) 18 (1.4) 13 (1.0) 13 (1.1) 61 (4.5) 19 (100.0) 34 (100.0) 39 (86.7) 58 (76.3) 34 (2.6) 22 (1.8) 20 (1.6) 92 (6.7) 19 (100.0) 34 (100.0) 45 (100.0) 73 (96.1) 69 (5.3) 52 (4.1) 45 (3.7) 142 (10.4) 19 (100.0) 34 (100.0) 45 (100.0) 76 (100.0) 250 (19.4) 220 (17.5) 193 (15.8) 326 (23.8) ≥ 100 ≥ 60 ≥ 40 ≥ 20
Yanqin Huang, Weiting Ge, Viktoriya London, Qilong Li, Shanrong Cai, Suzhan Zhang and Shu Zheng*
Diagnostic inconsistency of faecal immunochemical tests for haemoglobin in population screening of colorectal cancer Abstract Background: There is currently very little data available on the consistency of quantitative and qualitative faecal immunochemical test (FIT) for colorectal cancer screening. Methods: A representative random population (n = 1889, 40–74 year olds) in Jiashan, China was invited for FIT screening in 2012. Faecal samples were collected by a single specimen collection device and simultaneously tested by a quantitative FIT (OC-SENSOR, OC) and two qualitative FITs (FIT A and FIT B with intrinsic positive haemoglobin cut-off concentrations of 20 μg Hb/g faeces and 40 μg Hb/g faeces, respectively). The observational criteria for a positive result of the qualitative FIT were set according to the density of the colour appearing in the test strip. The results produced by the quantitative and qualitative FIT for each sample were compared. κ coefficient was used to measure consistency. Results: A total of 1368 (72.4%) individuals returned faecal samples. Both FIT A and FIT B precisely identified all faecal samples with haemoglobin concentration above 100 μg Hb/g faeces, but the overall consistency was poor for OC & FIT A (κ = 0.32, 95% CI 0.20–0.44) and was moderate for OC & FIT B (κ = 0.74, 95% CI 0.64–0.85). A more favourable consistency (κ = 0.64, 95% CI 0.57–0.72) was achieved when a different positive criterion was employed for FIT A. Conclusions: The diagnostic inconsistency between quantitative and qualitative FITs mainly exists in the faecal samples with low haemoglobin concentrations. Refining the criterion for a positive result may be a feasible way to improve the accuracy of qualitative FIT. Keywords: colorectal cancer screening; faecal immunochemical test; haemoglobin; quality assurance. *Corresponding author: Shu Zheng, Jiefang Road No. 88, Hangzhou 310009, China, Phone: +86 57187784501, Fax: +86 57187214404, E-mail: [email protected]
Yanqin Huang, Weiting Ge, Shanrong Cai, Suzhan Zhang and Shu Zheng: Cancer Institute (Key Laboratory of Cancer Prevention and Intervention, China National Ministry of Education; Key Laboratory of Molecular Biology in Medical Sciences, Zhejiang Province, China), The Second Affiliated Hospital, Zhejiang University School of Medicine, P.R. China Viktoriya London: Case Western Reserve School of Medicine, USA Qilong Li: Jiashan Institute of Cancer Prevention and Treatment, Zhejiang Province, P.R. China
Introduction Faecal occult blood test (FOBT) has been widely used as a screening tool for colorectal cancer. Early products of FOBT use chemical reagents to detect peroxidase activity of haem, by which the presence of haemoglobin is determined. The guaiac-based FOBT (gFOBT) is the typical test of the early FOBT products. However, the later developed faecal immunochemical test (FIT) for haemoglobin has been shown to be better than the original gFOBT [1]. It has better sensitivity and specificity for predicting colorectal neoplasms in an asymptomatic population [2]. It does not require diet restriction and only needs one faecal sample for testing, therefore, it significantly improves the compliance rate of colorectal cancer screening [3, 4]. There are currently two major forms of commercially available FIT products [5]: the quantitative FIT and the qualitative FIT. A lot of comparative studies have evaluated the screening performances of different FOBT products in colorectal cancer screening programmes. However, most of the studies focused on analysis of one type of FIT versus the gFOBT [6–15]. Few studies analysed the differences in the diagnostic performance of different types of FIT products. In 2009, Hundt et al. [15] first compared six qualitative FITs in a colorectal screening programme and observed a large difference in the performance characteristics of the different qualitative FITs. Brenner et al. [16] measured diagnostic consistency of the same six qualitative FITs and reported that the definition of the cut-offs for qualitative
2174 Huang et al.: Diagnostic inconsistency of FIT FIT was a critical issue of the test. Recently, some other experts [17–19] also expressed concerns about the diagnostic consistency of different types of FITs. Theoretically, the variant performances of different FIT products can be caused by the characteristics of the screening populations, the specimen devices, the faecal preservatives as well as the testing methodologies. Until now, very little has been studied about the effect that the testing methodologies have on the diagnostic consistency of different types of FIT products. In attempt to evaluate the effects of the testing metho dologies on diagnostic consistency, we designed a comparative study between a quantitative FIT that is widely used in Western Europe [8], Israel [20] and Japan [21] and two qualitative FITs that are widely used in China [22, 23]. In this study, we collected 1368 faecal samples from a representative random sample of population by a single specimen collection device which enclosed with the same type of faeces preservative and simultaneously tested each sample by the three FIT products mentioned above.
Materials and methods The quantitative FIT The quantitative FIT (OC-SENSOR, OC for short) was manufactured by Eiken Chemical Co., Ltd (Tokyo, Japan). All of the reagents used by OC were provided by the Nine-Strong Bioengineering Co., Ltd (Beijing, China). The OC is a fully automated instrument for quantitative measurement of faecal haemoglobin concentrations, requiring only a manual insertion of 10 sample tubes into the tube bed, one sample at a time. A paper print-out of the sample reading results is automatically generated in seven minutes after the initiation of reading. The mass of the tested faeces in each Eiken sampling device is 10 mg and the volume of the preservative buffer in Eiken sampling device is 2 mL. According to the manufacturer’s user manual, the OC produces reliable results in the range of 10–400 μg Hb/g faeces with an accuracy of 0.2 μg Hb/g faeces. The quality control process of the quantitative FIT involves regenerating the calibration curve for each batch of reagents and testing standards to make sure that the test results are within the permissible scope.
the manufacturers, the minimum haemoglobin concentrations for positive results are 100 ng haemoglobin per 1 mL faecal sample solution (ng/mL) for FIT A and 200 ng/mL for FIT B. The mass of faeces dissolved in the perservative of the specimen device which was used in our study was 10 mg and the volume of the preservative in a single specimen device was 2 mL, hence the minimum positive haemoglobin concentrations of FIT A and FIT B can be expressed as 20 μg Hb/g faeces and 40 μg Hb/g faeces, respectively. All the products used were within their quality guarantee periods. The quality control process of the qualitative FIT is to make sure that the control strip always display a dark red colour in every test, otherwise the test is regarded as null.
Population and faeces sampling We invited 1889 adults, aged 40–74 years, who reside in the Datong village in Jiashan County, Zhejiang Province, China to participate in a FIT-based colorectal cancer screening programme between the dates of March 26th to April 20th, 2012. Specimen collecting devices were distributed to participants and the participants were taught to sample and preserve their faecal samples by the nurses from a local community health service centre. Every participant was asked to submit one faecal sample from a single bowel movement. No diet restriction was required. Faeces were sampled with a serrated probe attached to the lid of the specimen device and only the faeces clinging to the grooves of the probe were brought into the chamber and dissolved in preservative buffer. Each specimen device was labelled with an order number and the name of the sample donor. A trained villager collected faecal samples from participants and sent them to the local community health service centre at 10:00 and 16:00 on a daily basis. A professional inspector in the local community health service centre tested each faecal sample by OC, FIT A and FIT B at the same time and on the same day the faecal sample was received. Personal and medical information of the sample donors was blind to the inspector. This study was approved by the Institutional Review Board on Medical Research, the Second Affiliated Hospital, Zhejiang University School of Medicine in accordance with the ethical principles in declaration of Helsinki (Oct 2008). A written informed consent was obtained from each subject before the faecal sampling tube was distributed.
A
Test strip FIT A Test positive (Grade 3) FIT B Test negative
The qualitative FITs The two qualitative FITs were purchased from two large manufacturers in China and were labelled as FIT A (Acon Co., Ltd, Hangzhou, China) and FIT B (W.H.P.M Co., Ltd, Beijing, China). Both products were certificated by State Food and Drug Administration of China. Both of the qualitative FITs are approximately 10 cm long, 3 mm wide and 1.5 mm thick. They are designed with a sample suction part at one end, a testing strip in the middle, with a holder located on the opposite end (Figure 1A). According to the user manual provided by
Control strip
Holding patch
Test patch
Sample sucking patch
B G0 G1 G2 G3 G4 G5 G6 G7 G8 G9 G10
Colour grades for FIT A Colour grades for FIT B
Figure 1 The qualitative FIT and the colour grade ladders. (A) The surface view of the qualitative FIT with their functional domain illustrated. (B) The colour grades ladders for FIT A and FIT B.
Huang et al.: Diagnostic inconsistency of FIT 2175
Faecal immunochemical test for haemoglobin Before inserting a specimen device with faeces into the OC, we pipetted out 500 μL of faecal solution from the specimen device and evenly distributed the faecal solution into two wells in a clean 96-well plate. Water suction ends of FIT A and FIT B test strips were inserted in the wells right after pipetting the sample solutions. The inspector read the result in 5–10 min for each qualitative FIT and recorded the result by taking a picture of the qualitative FIT and the specimen device. The test results of OC were printed out on a paper sheet, then were entered into a computer Excel spread sheet, in accordance to the order numbers and the sample donors. The inspector strictly abided by the quality control stipulations recommended by the manufacturers. All the tests of the standards were within the permissible scopes of the quality requirements for OC.
Measurement of the influence of Eiken preservative buffer to the qualitative FIT The preservative buffer in the specimen device was from Eiken Chemical Co., Ltd. The influence of Eiken preservative buffer on the test results of the qualitative FIT was not previously known. We acquired 20 fresh peripheral blood samples from the clinical laboratory of the Second Affiliated Hospital, Zhejiang University School of Medicine. First, we repeatedly diluted each blood sample by double distilled H2O until the haemoglobin concentration of the diluted sample was around 10 μg Hb/mL (10 ng Hb/μL). Next, a total of 800 μL Eiken preservative buffer and 800 μL preservative buffer provided by the manufacturers of FIT A were evenly pipetted into eight wells of a clean 96-well plate, 200 μL for each well. Aliquots of 1 μL, 2 μL, 4 μL and 8 μL of the diluted blood sample were then pipetted into four wells which were filled with Eiken preservative buffer and into another four wells which were filled with preservative buffer for FIT A. After gently mixing the liquid for a few seconds, the sample suction end of FIT A was insert into each well and the test result was measured within 5–10 min after the insertion. FIT B was also tested by the same method. Finally, the test results of the qualitative FIT with Eiken preservative buffer and that testing with the preservative buffer provided by the manufacturers were compared.
Positive criteria for the qualitative FIT In order to evaluate the relationship between the colour appearing on the test strips of the qualitative FIT and the haemoglobin concentrations of the faecal sample, we tentatively built up 10 “grades” of positive criteria for the qualitative FIT by measuring the density of the colour in the test strips of the products. Photographs of the tested qualitative FIT were loaded on a computer by Firework CS5 (Adobe Systems Incorporated, USA) software. The red part of the control strip was cropped out from the picture and the transparency index of the control strip picture was adjusted from 10% to 100%, at a 10% intervals, thus producing a colour ladder from faint pink to dark red with 10 different grades (Figure 1B). The colour bands on the control strips of FIT A and FIT B were uniformly dark red. Therefore, the colour ladders of FIT A and FIT B were quite similar. The 10 different colour
grades on the ladder were then numbered from G1 to G10, with G1 indicating the faintest pink colour and G10 corresponding to the darkest colour. G0 indicated no colour change in the test strip. The test result of each sample was recorded as G0 to G10 by comparing the test strip colour to the developed colour ladders. If a colour grade (from G1 to G10) was designated as the observational positive cut-off grade of a qualitative FIT, the faecal sample with a test strip colour darker than the designated colour grade was regarded as positive case. We selected 50 test strip photographs of the qualitative FITs which displayed different colour grades (27 G0, 8 G1, 6 G2, 5 G3 and 4 G4) to test the reproducibility of the positivity observations. We invited five students to grade the diagnosis of the 50 pictures by referencing the colour ladders on a computer monitor. The colour grades assigned by the students for each sample photograph were then compared to the grades assigned to the same photograph by the inspector from the local community health service centre. κ coefficients were calculated to measure their diagnostic consistency.
Data analysis The positive rates for faecal haemoglobin based on OC measurements were calculated at different haemoglobin cut-off concentrations. The faecal samples with haemoglobin concentrations above or below a certain set concentration measured by OC testing were further analysed by comparing them to the test results of FIT A and FIT B. κ coefficient was used to measure diagnostic consistency of the quantitative and qualitative FITs. The 95% confidence interval (95% CI) of κ was calculated. Statistical analysis was done by SPSS 17.0 software (IBM, USA).
Results We compared the test results of FIT A and FIT B for different concentrations of blood samples which were dissolved in Eiken preservative buffer and the preservative buffer provided by the manufacturers of the qualitative FIT. For the diluted blood samples with the same haemoglobin concentration, no difference was found in the test strips of the qualitative FITs done with Eiken faecal buffer versus the FIT done with the preservative buffer provided by the manufacturers of the qualitative FIT. We measured the inter-consistency of the positivity observation for 50 photographs of 50 faecal samples. The κ coefficients of the observed colour grades between the inspector from the local community health service centre and the five students were 0.91 (95% CI 0.81–1.00), 0.85 (95% CI 0.73– 0.97), 0.91 (95% CI 0.81–1.00), 0.79 (95% CI 0.65–0.92) and 0.91 (95% CI 0.81–1.00), respectively. A total of 1368 (72.4%) subjects returned 1368 faecal samples, one sample per subject, that were collected mostly during the second and third day after receiving the specimen collecting device. There were 687 male and 681 female
a μg Hb/g means haemoglobin concentrations of the faeces sample detected by OC; bOC test samples (%) means the number of samples with haemoglobin concentrations above or below certain levels which were defined by OC testing and its percentage in the number of total samples. cFIT A test samples (%) means that the samples with certain haemoglobin concentrations defined by OC were tested by FIT A and the test results of FIT A were subclassified into 11 groups according to darkness of the colour appearing in the test trips of FIT A.
5 (26.3) 5 (14.7) 5 (11.1) 5 (6.6) 0 (0) 0 (0) 0 (0) 5 (0.4) 13 (68.4) 15 (44.1) 15 (33.3) 15 (19.7) 0 (0) 0 (0) 0 (0) 15 (1.1) 17 (89.5) 22 (64.7) 22 (48.9) 22 (29.0) 0 (0) 0 (0) 0 (0) 22 (1.6) 18 (94.7) 24 (70.6) 24 (53.3) 25 (32.9) 3 (0.2) 3 (0.2) 3 (0.3) 28 (2.1) 19 (100.0) 28 (82.4) 29 (64.4) 33 (43.4) 11 (0.9) 9 (0.7) 9 (0.7) 44 (3.2) 19 (100.0) 31 (91.2) 33 (73.3) 43 (56.6) 18 (1.4) 13 (1.0) 13 (1.1) 61 (4.5) 19 (100.0) 34 (100.0) 39 (86.7) 58 (76.3) 34 (2.6) 22 (1.8) 20 (1.6) 92 (6.7) 19 (100.0) 34 (100.0) 45 (100.0) 73 (96.1) 69 (5.3) 52 (4.1) 45 (3.7) 142 (10.4) 19 (100.0) 34 (100.0) 45 (100.0) 76 (100.0) 250 (19.4) 220 (17.5) 193 (15.8) 326 (23.8) ≥ 100 ≥ 60 ≥ 40 ≥ 20
