Research paper 115
Optimizing sampling device for the fecal immunochemical test increases colonoscopy yields in colorectal cancer screening Yanqin Huanga, Qilong Lib, Weiting Gea, Yue Hua, Shanrong Caia, Ying Yuana, Suzhan Zhanga and Shu Zhenga The fecal immunochemical test (FIT) that quantifies hemoglobin concentration is reported to be better than qualitative FIT and the reason for its superiority has not been interpreted. To evaluate and understand the superiority of quantitative FIT, a representative randomly selected population (n = 2355) in Jiashan County, China, aged 40–74 years was invited for colorectal cancer screening in 2012. Three fecal samples were collected from each participant by one optimized and two common sampling devices, and then tested by both quantitative and qualitative FITs. Colonoscopy was provided independently to all participants. The performances of five featured screening strategies were compared. A total of 1020 participants were eligible. For screening advanced neoplasia, the positive predictive value (PPV) and the specificity of the strategy that tested one sample dissolved in an optimized device by quantitative FIT [PPV = 40.8%, 95% confidence interval (CI): 27.1–54.6; specificity = 96.8%, 95% CI: 95.7–98.0] were significantly improved over the strategy that tested one sample dissolved in the common device by qualitative FIT (PPV = 14.1%, 95% CI: 8.2–19.9; specificity = 87.9%, 95% CI: 85.8–89.9), whereas the sensitivity did not differ (39.2 and 37.3%, P = 0.89). Similar disparity in performance was observed between the strategies using qualitative FIT to test one sample dissolved
Introduction Colorectal cancer (CRC) is one of the leading causes of cancer morbidity and mortality worldwide. However, early diagnosis and early treatment of CRC by screening is one of the most successful achievements of human beings in combatting cancer. In the 1990s, three randomized controlled screening trials in the USA, UK, and Denmark independently found that population screening by the guaiac fecal occult blood test (gFOBT) reduces CRC mortality (Mandel et al., 1993; Jorgensen et al., 2002; Scholefield et al., 2002). In 2010, the Centers for Disease Control of the USA stated that the incidence of CRC across the country has decreased at an annual rate of 1% since the 1990s and CRC screening was estimated to contribute toward more than half of this reduction in incidence (Mandy et al., 2011). In the years after the randomized controlled trials, gFOBT was replaced by the fecal immunochemical test 0959-8278 Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
in optimized (PPV = 29.5%, 95% CI: 18.1–41.0; specificity = 95.3%, 95% CI: 94.0–96.7) versus common sampling devices. High sensitivity for advanced neoplasia was observed in the strategy that tested two samples by qualitative FIT (52.9%, 95% CI: 39.2–66.6). Quantitative FIT is better than qualitative FIT for screening advanced colorectal neoplasia. However, the fecal sampling device might contribute most significantly toward the superiority of quantitative FIT. European Journal of Cancer Prevention 25:115–122 Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved. European Journal of Cancer Prevention 2016, 25:115–122 Keywords: cancer screening, colorectal cancer, fecal immunochemical test, sampling bias a 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 and bJiashan Institute of Cancer Prevention and Treatment, Zhejiang Province, China
Correspondence to Suzhan Zhang, MD, Jiefang Road No. 88, Hangzhou 310009, Zhejiang Province, China Tel: + 86 571 877 84608; fax: + 86 571 872 14404; e-mail: [email protected] Received 29 July 2014 Accepted 18 February 2015
(FIT) for hemoglobin as the latter is more sensitive and easy to use. Now, FIT has been recommended in all CRC screening guidelines. Various types of FIT products have been developed and used for CRC screening. The commercially available FIT products can largely be classified into two major forms: the quantitative FIT, which quantifies hemoglobin concentration, and the qualitative FIT, which determines the existence of blood in stool samples. Qualitative FIT is cheap (0.5–13 $/test) and is used widely worldwide, especially in some developing countries. Quantitative FIT is relatively expensive (20–50 $/test) and is mainly used in developed countries such as Japan, Israel, and Europe. Several studies (Levi et al., 2006; Park et al., 2010; Rabeneck et al., 2012) have evaluated the screening performance of different FIT products, most of which have focused on comparisons of screening performance for gFOBT and FIT. Only a few studies have focused on the differences between quantitative FIT and qualitative FIT. DOI: 10.1097/CEJ.0000000000000154
Copyright r 2016 Wolters Kluwer Health, Inc. All rights reserved.
116 European Journal of Cancer Prevention 2016, Vol 25 No 2
Nevertheless, some scientists (Allison et al., 2014) have pointed out that quantitative FIT might be better than qualitative FIT because of the automated detection procedure, the adjustable cut-off level, and the precise quality control system of quantitative FIT. The first version of European guidelines for quality assurance in CRC screening and diagnosis (Halloran et al., 2012) also states the advantages of quantitative FIT. However, the superiority of quantitative FIT has not yet been confirmed because of the limited number of studies that have focused on comparisons of performance characteristics of quantitative and qualitative FIT.
the user manual of the manufacturer, the spiral tip of the sampling stick should be inserted into at least three different parts of the feces and dipped into the preservative buffer. The predefined minimum hemoglobin concentration for a positive result of the qualitative FIT is 400 ng of hemoglobin in the 2 μl preservative. The mass of feces tested in the sampling tube of the qualitative FIT is undetermined and these types of fecal sampling device are used widely; thus, the accessory sampling device of the qualitative FIT is a common sampling device.
We previously conducted a screening trial to compare the positive predictive values (PPVs) of quantitative and qualitative FIT for colorectal neoplasia in a rural population (n = 9000) in China. The result showed that the PPV of quantitative FIT for advanced colorectal neoplasia was significantly improved compared with that of qualitative FIT (Huang et al., 2014). However, we could not analyze differences in their sensitivities and specificities as only those with positive FIT were subjected to colonoscopy in that study. We then designed another screening trial in an attempt to validate the results of the first screening trial as well as to determine which part of FIT contributes toward its superior screening ability. In this screening trial, all participants were invited for colonoscopy.
This study started on 15th March and ended on 24th December 2012. The study population was from Datong, a medium sized (total population = 4719) village in Jiashan County. The inclusion criteria included residents aged 40–74 years in Datong. Those unsuitable for colonoscopy, including patients with severe illness (coronary heart disease and paralysis) and pregnant women, were excluded. All of the participants (n = 2611) in Datong aged 40–74 years were enrolled for the screening. We first obtained a list of names of all the residents in Datong from the local population registry. Invitation letters were sent to every family by a village officer. A work team that included a primary care physician, a nurse from the Huiming Health Service Center, and a village officer visited every family. The work team explained the study to every potential eligible participant. The written informed consent forms were signed by individuals (n = 2355) who agreed to participate in the study at the first visit. 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 guidelines in the Declaration of Helsinki (October 2008).
Methods The FIT materials
The quantitative FIT is an OC-Sensor automated fecal blood analyzer. The detecting machine with its accessory fecal sampling device was manufactured by Eiken Chemical Co., Ltd (Tokyo, Japan). The sampling device uses a stick stretching out from the inner side of the lid to sample feces. When the sampling device is sealed with the lid, the top part of the sampling stick precisely passes through a tiny hole. The tiny hole allows only 10 μg of feces attached to the groove of the spiral tip to enter the sealed chamber. The sealed chamber includes 2 μl of preservative buffer. According to the user manual of the sampling device, the sampling stick should be scratched over the entire surface of the feces and placed back in the sealed chamber. The OC-Sensor automated fecal blood analyzer has a calibration system that produces reliable test results for feces solutions, with hemoglobin concentrations ranging from 50 to 2000 ng/ml. The accessory sampling device of the quantitative FIT was designated as an optimized sampling device. The qualitative FIT is the XIAOBAOKANG immunogold labeling dipstick manufactured by W.H.P.M. Co., Ltd (Beijing, China). The feces sampling device and the preservative buffer for the qualitative FIT were provided by W.H.P.M. Co., Ltd. The sampling device is a cylindrical tube filled with 2 μl of preservative buffer and is sealed with a lid. A stick with spirals on the tip also stretches out from the inner side of the lid. According to
Population and study design
At the first visit, two fecal sampling devices (an optimized one and a common one) were distributed simultaneously to each participant (n = 2355). Instructions on the use of each sampling device were provided by the primary care physician. A questionnaire sheet documenting the medical history of the eligible participant was filled out by the work team at the first visit. All of the participants were asked to submit two fecal samples from the same excreta within 2 weeks after receiving the fecal sampling devices. No diet restriction was required. Lowtemperature storage of the fecal samples was not mandatory during the delivery of the samples as the time span from the collection of fecal samples to the FIT test was usually no more than 12 h. After the first sample collection, another common sampling device for qualitative FIT was distributed to the participants by the village officer when they came to the village officer’s home to submit the first two fecal samples. The third fecal sample was collected in the same way. The first and the second sample collections started on 23th March and ended on 29th May. We excluded individuals (n = 242) who failed
Copyright r 2016 Wolters Kluwer Health, Inc. All rights reserved.
Qualitative Quantitative Testing methodologies
Copyright r 2016 Wolters Kluwer Health, Inc. All rights reserved.
FIT, fecal immunochemical test; Hb, hemoglobin.
Quantitative
Automated readout ≥ 40 μg Automated readout ≥ 10 μg Positive criteria
Two qualitative FIT
Color visible on both test and control strips for either sample Qualitative
Common Predefined at 200 ng Hb/ml fecal solution Color visible on both test and control strips Qualitative Optimized Adjusted to 40 μg Hb/g feces Optimized Adjusted to 10 μg Hb/g feces
Optimized Predefined at 200 ng Hb/ml fecal solution (equal to 40 μg Hb/g feces) Color visible on both test and control strips
1
One qualitative FIT Optimized qualitative FIT
1 1
Table 1
The technicians in the Huiming Health Service Center tested the fecal samples immediately on receipt. Among the three fecal samples, the first two were delivered at the same time: one in an optimized sampling device and another in a common sampling device. The fecal samples in the optimized sampling devices were tested using an OC-Sensor occult blood analyzer according to the manufacturer’s instruction. Before it was tested by the OCSensor blood analyzer, 200 μl of the feces solution in the optimized sampling device was obtained for simultaneous testing by qualitative FIT. The fecal solution in the common sampling device as well as the third fecal sample that was submitted a couple of days later were all tested by qualitative FIT. We analyzed the influence of different preservative buffers on the test results of qualitative FIT and no effect was observed. The details of FIT testing and interference analysis have been described in our previous publication (Huang et al., 2013).
The five screening strategies evaluated
Laboratory analyses
Quantitative FIT ≥ 10 μg
Quantitative FIT ≥ 40 μg
Altogether, we collected three fecal samples from the same participant using the two types of fecal sampling device: one in an optimized sampling device and two in common sampling devices. On the basis of the differences in the sampling devices, the numbers of samples being tested, the positive criteria defined, and the testing methodologies used by FIT, a total of five different screening strategies (Table 1) were generated.
1
After the completion of fecal sample collection, colonoscopy invitations were sent to all eligible individuals (n = 2355) by the endoscopists in Jiashan Institute of Cancer Prevention and Treatment. All of the endoscopists involved in this study had performed at least 500 colonoscopies per year for more than 3 consecutive years before the study. The sample donors were informed about the FIT test results by the village officer. However, the endoscopists were completely blinded to the test results of FIT. A precolonoscopy consultation was scheduled by the endoscopists for every participant who agreed to undergo colonoscopy. The endoscopists assessed the health status of the participants in the consultation and evaluated the risks of colonoscopy for these participants. However, there was no exclusion during this process. All of the participants received a blood test for hepatitis B virus and a bag of bowel-cleaning drug (50 g MgSO4) after the consultation. The nurses in Jiashan Institute of Cancer Prevention and Treatment instructed the participants on how to prepare for colonoscopy. Colonoscopy was performed without sedation a week after the consultation. Histological examination was required if polypoid and distinguishable flat lesions were found during colonoscopy. We did not assess the adverse effects of the study on the participants.
Fecal sample numbers Sampling device Positive cut-off
to submit any of the required fecal samples during the sample collection period.
2 Common Predefined at 200 ng Hb/ml fecal solution
Optimizing FIT for CRC screening Huang et al. 117
118 European Journal of Cancer Prevention 2016, Vol 25 No 2
The reference standard
Colonoscopy findings and histologically verified lesions were the reference standards of our study. For the screening colonoscopy, the endoscopists required an average withdrawal time of 6 min and cecal intubation. The lesions were resected or at least biopsied and were sent to an independent center for clinical diagnosis. Uncertain diagnostic results were further verified by experienced pathologists from the Second Affiliated Hospital, Zhejiang University School of Medicine. Advanced adenomas were adenomas with sizes at least 1 cm in diameter, those with villous (≥25%) differentiation, or those with high-grade dysplasia. Advanced neoplasia included a combination of CRC and advanced adenoma. Screening-relevant neoplasia comprised a combination of CRC and all adenomas. Statistical analyses
The screening performances (sensitivities, specificities, predictive values, and likelihood ratios) of the six strategies were compared. The 95% confidence interval was estimated for the uncertainty of analysis. The χ2-test was used to determine the significance of differences for different strategies. All of the statistical calculations were carried out using SPSS 17.0 software (SPSS Inc., Chicago, IL, USA).
Results A total of 2008 (85.3%) individuals were studied; each of them submitted at least one optimized sampling device with feces, whereas 2115 (89.8%) individuals returned at least one common sampling device with feces. The positive rate of the strategy that tested two fecal samples dissolved in the common sampling device by qualitative FIT (12.6%) was significantly (P < 0.001) higher than that of any of the other strategy groups. The positive rate of the strategy that tested one fecal sample dissolved in the optimized sampling device by quantitative FIT with a positive cut-off of 10 μg Hb/g feces (7.8%) was significantly higher than that of the strategy using onesample quantitative FIT with a cut-off of 40 μg (3.0%, P < 0.001) and the strategy that tested one fecal sample dissolved in the optimized sampling device by qualitative FIT (3.7%, P < 0.001), but it was similar to that of the strategy that tested one fecal sample dissolved in the common sampling device by qualitative FIT (7.7%, P = 0.855). We excluded participants (n = 1272) who refused colonoscopy, participants (n = 8) with incomplete colonoscopy (without cecal intubation), and participants (n = 5) in whom colorectal polyps were found, but who refused histological examination. We also excluded participants (n = 38) who underwent thorough examinations of their colon and rectum, but did not submit any fecal samples. Figure 1 shows the flow diagram of the study. After exclusion, a total of 1020 participants were included in the analysis. Each of these participants submitted at
least one fecal sample and had explicit diagnosis of their colon and rectum. The total numbers of patients detected with CRC, advanced adenoma, and nonadvanced adenoma were 10 (0.98%), 42 (4.1%), and 107 (10.5%), respectively. There were 455 men (median age: 55 years) and 565 women (median age: 53 years). The characteristics of the population are shown in Table 2. Among the participants who were included in the analysis, 49 did not submit any optimized sampling device with feces. Twelve participants did not submit any common sampling device with feces. Two participants did not submit any fecal sample at the first sample collection, but submitted a sample at the second sample collection. Accordingly, 971 participants were included in the five screening strategies for the strategy that tested one fecal sample dissolved in the optimized sampling device by quantitative FIT with a positive cut-off of 40 μg Hb/g feces, 971 for the same strategy with a cut-off of 10 μg Hb/g feces, 971 for the strategy that tested one fecal sample dissolved in the optimized sampling device by qualitative FIT, 1008 for the strategy that tested one fecal sample dissolved in the common sampling device by qualitative FIT, and 1010 for the strategy that tested two fecal samples dissolved in common sampling devices by qualitative FIT, respectively. Tables 3 and 4 show the true-positive, false-positive, false-negative, and true-negative rates for predicting advanced neoplasia and screening-relevant neoplasia of the five screen strategies. Although there were overlaps between the 95% confidence intervals of some strategies, the PPVs and specificities of the screening strategies that tested fecal samples dissolved in optimized sampling devices were apparently higher than those of the screening strategies that tested fecal samples dissolved in common sampling devices, irrespective of whether the testing methodology was quantitative or qualitative. The strategy that tested one fecal sample dissolved in the optimized sampling device by quantitative FIT with a positive cut-off of 40 μg and the strategy that tested one fecal sample dissolved in the common sampling device by qualitative FIT each tested one fecal sample, both with an adjusted cut-off or a predefined cut-off of 40 μg/g feces. The differences were within those for fecal sampling devices and testing methodologies. The PPV and the specificity for detecting both advanced neoplasia and screening-relevant neoplasia of the former strategy were significantly increased over the latter strategy, but the sensitivity was the same. For the strategy that tested one fecal sample dissolved in the common sampling device by qualitative FIT and the strategy that tested one fecal sample dissolved in the optimized sampling device by quantitative FIT with a positive cut-off of 40 μg, both strategies tested one fecal sample, both with the same testing methodology and the same cut-offs; a difference only existed in the sampling device. The specificity for detecting advanced neoplasia
Copyright r 2016 Wolters Kluwer Health, Inc. All rights reserved.
Optimizing FIT for CRC screening Huang et al. 119
Fig. 1
Total population in the village n = 4719 Accepted n = 1071 Complete n = 1020 IC n = 8
Eligible participants (age 40−74) n = 2611
NB n = 5 No feces sample submitted n = 38
Refused screening n = 256 Accepted screening n = 2355
Colonoscopy
Refused n = 1284
FIT screening
The second sample collection
The first sample collection
The common sampling device The common sampling device
The optimized sampling device
Completed n = 2008
Quantitative FIT
Qualitative FIT
Qualitative FIT
≥10 μg Hb/g n = 157 Completed colonoscopy Yes n = 118 IC n = 1 No n = 38 NB n = 0
Optimizing sampling device for the fecal immunochemical test increases colonoscopy yields in colorectal cancer screening Yanqin Huanga, Qilong Lib, Weiting Gea, Yue Hua, Shanrong Caia, Ying Yuana, Suzhan Zhanga and Shu Zhenga The fecal immunochemical test (FIT) that quantifies hemoglobin concentration is reported to be better than qualitative FIT and the reason for its superiority has not been interpreted. To evaluate and understand the superiority of quantitative FIT, a representative randomly selected population (n = 2355) in Jiashan County, China, aged 40–74 years was invited for colorectal cancer screening in 2012. Three fecal samples were collected from each participant by one optimized and two common sampling devices, and then tested by both quantitative and qualitative FITs. Colonoscopy was provided independently to all participants. The performances of five featured screening strategies were compared. A total of 1020 participants were eligible. For screening advanced neoplasia, the positive predictive value (PPV) and the specificity of the strategy that tested one sample dissolved in an optimized device by quantitative FIT [PPV = 40.8%, 95% confidence interval (CI): 27.1–54.6; specificity = 96.8%, 95% CI: 95.7–98.0] were significantly improved over the strategy that tested one sample dissolved in the common device by qualitative FIT (PPV = 14.1%, 95% CI: 8.2–19.9; specificity = 87.9%, 95% CI: 85.8–89.9), whereas the sensitivity did not differ (39.2 and 37.3%, P = 0.89). Similar disparity in performance was observed between the strategies using qualitative FIT to test one sample dissolved
Introduction Colorectal cancer (CRC) is one of the leading causes of cancer morbidity and mortality worldwide. However, early diagnosis and early treatment of CRC by screening is one of the most successful achievements of human beings in combatting cancer. In the 1990s, three randomized controlled screening trials in the USA, UK, and Denmark independently found that population screening by the guaiac fecal occult blood test (gFOBT) reduces CRC mortality (Mandel et al., 1993; Jorgensen et al., 2002; Scholefield et al., 2002). In 2010, the Centers for Disease Control of the USA stated that the incidence of CRC across the country has decreased at an annual rate of 1% since the 1990s and CRC screening was estimated to contribute toward more than half of this reduction in incidence (Mandy et al., 2011). In the years after the randomized controlled trials, gFOBT was replaced by the fecal immunochemical test 0959-8278 Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
in optimized (PPV = 29.5%, 95% CI: 18.1–41.0; specificity = 95.3%, 95% CI: 94.0–96.7) versus common sampling devices. High sensitivity for advanced neoplasia was observed in the strategy that tested two samples by qualitative FIT (52.9%, 95% CI: 39.2–66.6). Quantitative FIT is better than qualitative FIT for screening advanced colorectal neoplasia. However, the fecal sampling device might contribute most significantly toward the superiority of quantitative FIT. European Journal of Cancer Prevention 25:115–122 Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved. European Journal of Cancer Prevention 2016, 25:115–122 Keywords: cancer screening, colorectal cancer, fecal immunochemical test, sampling bias a 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 and bJiashan Institute of Cancer Prevention and Treatment, Zhejiang Province, China
Correspondence to Suzhan Zhang, MD, Jiefang Road No. 88, Hangzhou 310009, Zhejiang Province, China Tel: + 86 571 877 84608; fax: + 86 571 872 14404; e-mail: [email protected] Received 29 July 2014 Accepted 18 February 2015
(FIT) for hemoglobin as the latter is more sensitive and easy to use. Now, FIT has been recommended in all CRC screening guidelines. Various types of FIT products have been developed and used for CRC screening. The commercially available FIT products can largely be classified into two major forms: the quantitative FIT, which quantifies hemoglobin concentration, and the qualitative FIT, which determines the existence of blood in stool samples. Qualitative FIT is cheap (0.5–13 $/test) and is used widely worldwide, especially in some developing countries. Quantitative FIT is relatively expensive (20–50 $/test) and is mainly used in developed countries such as Japan, Israel, and Europe. Several studies (Levi et al., 2006; Park et al., 2010; Rabeneck et al., 2012) have evaluated the screening performance of different FIT products, most of which have focused on comparisons of screening performance for gFOBT and FIT. Only a few studies have focused on the differences between quantitative FIT and qualitative FIT. DOI: 10.1097/CEJ.0000000000000154
Copyright r 2016 Wolters Kluwer Health, Inc. All rights reserved.
116 European Journal of Cancer Prevention 2016, Vol 25 No 2
Nevertheless, some scientists (Allison et al., 2014) have pointed out that quantitative FIT might be better than qualitative FIT because of the automated detection procedure, the adjustable cut-off level, and the precise quality control system of quantitative FIT. The first version of European guidelines for quality assurance in CRC screening and diagnosis (Halloran et al., 2012) also states the advantages of quantitative FIT. However, the superiority of quantitative FIT has not yet been confirmed because of the limited number of studies that have focused on comparisons of performance characteristics of quantitative and qualitative FIT.
the user manual of the manufacturer, the spiral tip of the sampling stick should be inserted into at least three different parts of the feces and dipped into the preservative buffer. The predefined minimum hemoglobin concentration for a positive result of the qualitative FIT is 400 ng of hemoglobin in the 2 μl preservative. The mass of feces tested in the sampling tube of the qualitative FIT is undetermined and these types of fecal sampling device are used widely; thus, the accessory sampling device of the qualitative FIT is a common sampling device.
We previously conducted a screening trial to compare the positive predictive values (PPVs) of quantitative and qualitative FIT for colorectal neoplasia in a rural population (n = 9000) in China. The result showed that the PPV of quantitative FIT for advanced colorectal neoplasia was significantly improved compared with that of qualitative FIT (Huang et al., 2014). However, we could not analyze differences in their sensitivities and specificities as only those with positive FIT were subjected to colonoscopy in that study. We then designed another screening trial in an attempt to validate the results of the first screening trial as well as to determine which part of FIT contributes toward its superior screening ability. In this screening trial, all participants were invited for colonoscopy.
This study started on 15th March and ended on 24th December 2012. The study population was from Datong, a medium sized (total population = 4719) village in Jiashan County. The inclusion criteria included residents aged 40–74 years in Datong. Those unsuitable for colonoscopy, including patients with severe illness (coronary heart disease and paralysis) and pregnant women, were excluded. All of the participants (n = 2611) in Datong aged 40–74 years were enrolled for the screening. We first obtained a list of names of all the residents in Datong from the local population registry. Invitation letters were sent to every family by a village officer. A work team that included a primary care physician, a nurse from the Huiming Health Service Center, and a village officer visited every family. The work team explained the study to every potential eligible participant. The written informed consent forms were signed by individuals (n = 2355) who agreed to participate in the study at the first visit. 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 guidelines in the Declaration of Helsinki (October 2008).
Methods The FIT materials
The quantitative FIT is an OC-Sensor automated fecal blood analyzer. The detecting machine with its accessory fecal sampling device was manufactured by Eiken Chemical Co., Ltd (Tokyo, Japan). The sampling device uses a stick stretching out from the inner side of the lid to sample feces. When the sampling device is sealed with the lid, the top part of the sampling stick precisely passes through a tiny hole. The tiny hole allows only 10 μg of feces attached to the groove of the spiral tip to enter the sealed chamber. The sealed chamber includes 2 μl of preservative buffer. According to the user manual of the sampling device, the sampling stick should be scratched over the entire surface of the feces and placed back in the sealed chamber. The OC-Sensor automated fecal blood analyzer has a calibration system that produces reliable test results for feces solutions, with hemoglobin concentrations ranging from 50 to 2000 ng/ml. The accessory sampling device of the quantitative FIT was designated as an optimized sampling device. The qualitative FIT is the XIAOBAOKANG immunogold labeling dipstick manufactured by W.H.P.M. Co., Ltd (Beijing, China). The feces sampling device and the preservative buffer for the qualitative FIT were provided by W.H.P.M. Co., Ltd. The sampling device is a cylindrical tube filled with 2 μl of preservative buffer and is sealed with a lid. A stick with spirals on the tip also stretches out from the inner side of the lid. According to
Population and study design
At the first visit, two fecal sampling devices (an optimized one and a common one) were distributed simultaneously to each participant (n = 2355). Instructions on the use of each sampling device were provided by the primary care physician. A questionnaire sheet documenting the medical history of the eligible participant was filled out by the work team at the first visit. All of the participants were asked to submit two fecal samples from the same excreta within 2 weeks after receiving the fecal sampling devices. No diet restriction was required. Lowtemperature storage of the fecal samples was not mandatory during the delivery of the samples as the time span from the collection of fecal samples to the FIT test was usually no more than 12 h. After the first sample collection, another common sampling device for qualitative FIT was distributed to the participants by the village officer when they came to the village officer’s home to submit the first two fecal samples. The third fecal sample was collected in the same way. The first and the second sample collections started on 23th March and ended on 29th May. We excluded individuals (n = 242) who failed
Copyright r 2016 Wolters Kluwer Health, Inc. All rights reserved.
Qualitative Quantitative Testing methodologies
Copyright r 2016 Wolters Kluwer Health, Inc. All rights reserved.
FIT, fecal immunochemical test; Hb, hemoglobin.
Quantitative
Automated readout ≥ 40 μg Automated readout ≥ 10 μg Positive criteria
Two qualitative FIT
Color visible on both test and control strips for either sample Qualitative
Common Predefined at 200 ng Hb/ml fecal solution Color visible on both test and control strips Qualitative Optimized Adjusted to 40 μg Hb/g feces Optimized Adjusted to 10 μg Hb/g feces
Optimized Predefined at 200 ng Hb/ml fecal solution (equal to 40 μg Hb/g feces) Color visible on both test and control strips
1
One qualitative FIT Optimized qualitative FIT
1 1
Table 1
The technicians in the Huiming Health Service Center tested the fecal samples immediately on receipt. Among the three fecal samples, the first two were delivered at the same time: one in an optimized sampling device and another in a common sampling device. The fecal samples in the optimized sampling devices were tested using an OC-Sensor occult blood analyzer according to the manufacturer’s instruction. Before it was tested by the OCSensor blood analyzer, 200 μl of the feces solution in the optimized sampling device was obtained for simultaneous testing by qualitative FIT. The fecal solution in the common sampling device as well as the third fecal sample that was submitted a couple of days later were all tested by qualitative FIT. We analyzed the influence of different preservative buffers on the test results of qualitative FIT and no effect was observed. The details of FIT testing and interference analysis have been described in our previous publication (Huang et al., 2013).
The five screening strategies evaluated
Laboratory analyses
Quantitative FIT ≥ 10 μg
Quantitative FIT ≥ 40 μg
Altogether, we collected three fecal samples from the same participant using the two types of fecal sampling device: one in an optimized sampling device and two in common sampling devices. On the basis of the differences in the sampling devices, the numbers of samples being tested, the positive criteria defined, and the testing methodologies used by FIT, a total of five different screening strategies (Table 1) were generated.
1
After the completion of fecal sample collection, colonoscopy invitations were sent to all eligible individuals (n = 2355) by the endoscopists in Jiashan Institute of Cancer Prevention and Treatment. All of the endoscopists involved in this study had performed at least 500 colonoscopies per year for more than 3 consecutive years before the study. The sample donors were informed about the FIT test results by the village officer. However, the endoscopists were completely blinded to the test results of FIT. A precolonoscopy consultation was scheduled by the endoscopists for every participant who agreed to undergo colonoscopy. The endoscopists assessed the health status of the participants in the consultation and evaluated the risks of colonoscopy for these participants. However, there was no exclusion during this process. All of the participants received a blood test for hepatitis B virus and a bag of bowel-cleaning drug (50 g MgSO4) after the consultation. The nurses in Jiashan Institute of Cancer Prevention and Treatment instructed the participants on how to prepare for colonoscopy. Colonoscopy was performed without sedation a week after the consultation. Histological examination was required if polypoid and distinguishable flat lesions were found during colonoscopy. We did not assess the adverse effects of the study on the participants.
Fecal sample numbers Sampling device Positive cut-off
to submit any of the required fecal samples during the sample collection period.
2 Common Predefined at 200 ng Hb/ml fecal solution
Optimizing FIT for CRC screening Huang et al. 117
118 European Journal of Cancer Prevention 2016, Vol 25 No 2
The reference standard
Colonoscopy findings and histologically verified lesions were the reference standards of our study. For the screening colonoscopy, the endoscopists required an average withdrawal time of 6 min and cecal intubation. The lesions were resected or at least biopsied and were sent to an independent center for clinical diagnosis. Uncertain diagnostic results were further verified by experienced pathologists from the Second Affiliated Hospital, Zhejiang University School of Medicine. Advanced adenomas were adenomas with sizes at least 1 cm in diameter, those with villous (≥25%) differentiation, or those with high-grade dysplasia. Advanced neoplasia included a combination of CRC and advanced adenoma. Screening-relevant neoplasia comprised a combination of CRC and all adenomas. Statistical analyses
The screening performances (sensitivities, specificities, predictive values, and likelihood ratios) of the six strategies were compared. The 95% confidence interval was estimated for the uncertainty of analysis. The χ2-test was used to determine the significance of differences for different strategies. All of the statistical calculations were carried out using SPSS 17.0 software (SPSS Inc., Chicago, IL, USA).
Results A total of 2008 (85.3%) individuals were studied; each of them submitted at least one optimized sampling device with feces, whereas 2115 (89.8%) individuals returned at least one common sampling device with feces. The positive rate of the strategy that tested two fecal samples dissolved in the common sampling device by qualitative FIT (12.6%) was significantly (P < 0.001) higher than that of any of the other strategy groups. The positive rate of the strategy that tested one fecal sample dissolved in the optimized sampling device by quantitative FIT with a positive cut-off of 10 μg Hb/g feces (7.8%) was significantly higher than that of the strategy using onesample quantitative FIT with a cut-off of 40 μg (3.0%, P < 0.001) and the strategy that tested one fecal sample dissolved in the optimized sampling device by qualitative FIT (3.7%, P < 0.001), but it was similar to that of the strategy that tested one fecal sample dissolved in the common sampling device by qualitative FIT (7.7%, P = 0.855). We excluded participants (n = 1272) who refused colonoscopy, participants (n = 8) with incomplete colonoscopy (without cecal intubation), and participants (n = 5) in whom colorectal polyps were found, but who refused histological examination. We also excluded participants (n = 38) who underwent thorough examinations of their colon and rectum, but did not submit any fecal samples. Figure 1 shows the flow diagram of the study. After exclusion, a total of 1020 participants were included in the analysis. Each of these participants submitted at
least one fecal sample and had explicit diagnosis of their colon and rectum. The total numbers of patients detected with CRC, advanced adenoma, and nonadvanced adenoma were 10 (0.98%), 42 (4.1%), and 107 (10.5%), respectively. There were 455 men (median age: 55 years) and 565 women (median age: 53 years). The characteristics of the population are shown in Table 2. Among the participants who were included in the analysis, 49 did not submit any optimized sampling device with feces. Twelve participants did not submit any common sampling device with feces. Two participants did not submit any fecal sample at the first sample collection, but submitted a sample at the second sample collection. Accordingly, 971 participants were included in the five screening strategies for the strategy that tested one fecal sample dissolved in the optimized sampling device by quantitative FIT with a positive cut-off of 40 μg Hb/g feces, 971 for the same strategy with a cut-off of 10 μg Hb/g feces, 971 for the strategy that tested one fecal sample dissolved in the optimized sampling device by qualitative FIT, 1008 for the strategy that tested one fecal sample dissolved in the common sampling device by qualitative FIT, and 1010 for the strategy that tested two fecal samples dissolved in common sampling devices by qualitative FIT, respectively. Tables 3 and 4 show the true-positive, false-positive, false-negative, and true-negative rates for predicting advanced neoplasia and screening-relevant neoplasia of the five screen strategies. Although there were overlaps between the 95% confidence intervals of some strategies, the PPVs and specificities of the screening strategies that tested fecal samples dissolved in optimized sampling devices were apparently higher than those of the screening strategies that tested fecal samples dissolved in common sampling devices, irrespective of whether the testing methodology was quantitative or qualitative. The strategy that tested one fecal sample dissolved in the optimized sampling device by quantitative FIT with a positive cut-off of 40 μg and the strategy that tested one fecal sample dissolved in the common sampling device by qualitative FIT each tested one fecal sample, both with an adjusted cut-off or a predefined cut-off of 40 μg/g feces. The differences were within those for fecal sampling devices and testing methodologies. The PPV and the specificity for detecting both advanced neoplasia and screening-relevant neoplasia of the former strategy were significantly increased over the latter strategy, but the sensitivity was the same. For the strategy that tested one fecal sample dissolved in the common sampling device by qualitative FIT and the strategy that tested one fecal sample dissolved in the optimized sampling device by quantitative FIT with a positive cut-off of 40 μg, both strategies tested one fecal sample, both with the same testing methodology and the same cut-offs; a difference only existed in the sampling device. The specificity for detecting advanced neoplasia
Copyright r 2016 Wolters Kluwer Health, Inc. All rights reserved.
Optimizing FIT for CRC screening Huang et al. 119
Fig. 1
Total population in the village n = 4719 Accepted n = 1071 Complete n = 1020 IC n = 8
Eligible participants (age 40−74) n = 2611
NB n = 5 No feces sample submitted n = 38
Refused screening n = 256 Accepted screening n = 2355
Colonoscopy
Refused n = 1284
FIT screening
The second sample collection
The first sample collection
The common sampling device The common sampling device
The optimized sampling device
Completed n = 2008
Quantitative FIT
Qualitative FIT
Qualitative FIT
≥10 μg Hb/g n = 157 Completed colonoscopy Yes n = 118 IC n = 1 No n = 38 NB n = 0
