Download Evaluation of an automated immunochemical fecal occult blood test

2420
Evaluation of an Automated Immunochemical Fecal
Occult Blood Test for Colorectal Neoplasia Detection
in a Chinese Population
Wai Man Wong, M.D.1
Shiu Kum Lam, M.D.1
Kwan Lok Cheung, B.Sc1
Teresa Sze Man Tong, B.Sc1
Paul Rozen, MBBS2
Graeme P. Young, M.D.3
Kin Wah Chu4
Judy Ho4
Wai Lun Law, M.S4
Hiu Ming Tung4
Hok Kwok Choi4
Yee Man Lee4
Kam Chuen Lai1
Wayne H. C. Hu1
Chi Kuen Chan1
Man Fung Yuen, M.D.1
Benjamin Chun-Yu Wong, M.D.1
1
Department of Medicine, University of Hong
Kong, Queen Mary Hospital, Hong Kong.
2
Department of Gastroenterology, Tel Aviv Medical Center, Tel Aviv University, Tel Aviv, Israel.
3
Department of Gastroenterology and Hepatology,
Flinders Medical Centre, Flinders University of
South Australia, Adelaide, Australia.
4
Department of Surgery, University of Hong Kong,
Queen Mary Hospital, Hong Kong.
Supported by the Simon K. Y. Lee Gastroenterology
Research Fund, University of Hong Kong, Queen
Mary Hospital, Hong Kong.
The authors thank members of the World Organization for Digestive Endoscopy/World Health Organization/American Society for Gastrointestinal Endoscopy Outreach Program for the International
Promotion of Colorectal Cancer Screening for their
advice on the development of the study protocol
and its evaluation; Fujirebio, Inc. (Tokyo, Japan) for
providing the Magstream 1000 machine and test
kits; Seekers Biochemical Ltd. (Hong Kong) for
technical support and facilitation; and Ms. Fiona
M. Y. Fung, Ms. Vicky Y. K. Ho, and the endoscopy
nurses from the Departments of Medicine and
Surgery, Queen Mary Hospital, Hong Kong, for their
© 2003 American Cancer Society
BACKGROUND. Most commercial fecal occult blood tests (FOBT) used for colorectal
carcinoma screening of Western populations are guaiac-based, manually developed, subjective, and sensitive to dietary components. Preliminary studies demonstrated the unsuitability of these tests for screening a Chinese population. The
goal of the current study was to evaluate the performance characteristics of a
human hemoglobin–specific automated immunochemical FOBT, the Magstream
1000/Hem SP (Fujirebio, Inc., Tokyo, Japan), in a Chinese population referred for
colonoscopy.
METHODS. Two hundred fifty consecutive patients who were referred for colonoscopy and met the study inclusion criteria provided samples for the immunochemical FOBT (without dietary restrictions) from two successive stool specimens. Tests
were developed with an automated instrument that had an adjustable sensitivity
threshold. The sensitivity, specificity, and positive predictive value for detecting
colorectal adenomas and carcinomas were calculated according to the manufacturer’s instructions over a range of sensitivity levels.
RESULTS. At the optimal threshold level, the sensitivity, specificity, and positive
predictive value for detection of significant colorectal neoplasia (adenomas ⱖ 1.0
cm and carcinomas) were 62%, 93%, and 44%, respectively. The test was easy to
use, and results did not depend on operator experience.
CONCLUSIONS. The automated immunochemical FOBT used in the current study
was a robust, convenient, and useful tool for colorectal carcinoma screening in the
study population. Cancer 2003;97:2420 – 4. © 2003 American Cancer Society.
DOI 10.1002/cncr.11369
KEYWORDS: colon carcinoma, adenoma, colorectal neoplasia, screening.
C
olorectal carcinoma (CRC) is a leading cause of cancer mortality
in the United States, Europe, and Westernized Asian cities such as
Hong Kong.1 As a result, CRC is becoming a major medical and
economic burden for China. It is believed that most cases of sporadic
colorectal carcinoma arise from preexisting adenomatous polyps. By
screening asymptomatic, average-risk individuals, curable carcinomas and removable adenomatous polyps can be detected; detection
can lead to interruption of the natural history of the disease or an
increased chance of curing it.2
Three large-scale randomized trials (1 each in the United States,
the United Kingdom, and Denmark) using a guaiac-based fecal occult
assistance and for providing care to the patients in
the study.
Hospital, Pokfulam Road, Hong Kong; Fax: (011)-8522872-5828; E-mail: [email protected]
Address for reprints: Dr. Benjamin C.-Y. Wong, Department of Medicine, University of Hong Kong, Queen Mary
Received December 10, 2002; revision received
January 8, 2003; accepted January 28, 2003.
Automated Immunochemical FOBT/Wong et al.
blood test (FOBT) reported a reduction in CRC mortality of 15–33% on an intent-to-screen basis.3–5 These
trials also demonstrated that when screening was repeated annually or biennially, an even higher level of
sensitivity was achieved (i.e., program sensitivity in
detecting colorectal neoplasia at a treatable stage improved with each screening cycle). In addition, screening can reduce the incidence of CRC by identifying
large adenomatous polyps so that they can be removed.2,6
In Hong Kong, the incidence of CRC has increased
significantly over the past decade, and CRC currently
is the second most common carcinoma (after lung
carcinoma) in Hong Kong. In 1999, the age-standardized rates of CRC per 100,000 males and females were
37.8 and 28.2, respectively.7 These rates are comparable to the age-standardized rates per 100,000 persons
for white males in the United States (43.9) and for the
populations of Japan (39.5) and Australia and New
Zealand combined (45.8).8,9 CRC is a growing medical
problem in Hong Kong and in sectors of the Chinese
population that are becoming Westernized.10
The FOBT is the simplest, cheapest noninvasive
and effective CRC screening method available. There
are 2 types of commercially available FOBTs: 1) guaiac
tests that detect the pseudoperoxidase activity of
heme; and 2) immunochemical tests that detect either
the globin protein of human hemoglobin or other
blood products. Guaiac-based FOBTs are used more
frequently and have been validated extensively for
screening in North America and most of Europe. However, the guaiac tests are not human hemoglobin–
specific and therefore require the patient to avoid red
meat, certain uncooked vegetables, vitamin C, and
nonsteroidal anti-inflammatory drugs (NSAIDs) before and during the fecal sample collection period.
These restrictions can affect patient compliance as
well as the accuracy of the tests. Extensive screening
experiences in Japan suggested that a two-day immunochemical FOBT that did not require dietary restriction was more suitable for the study populations than
was the guaiac test.11,12
Most commercial or in-house guaiac-based and
immunochemical FOBTs are interpreted by the visual
determination of color changes after manual addition
of reagent solutions to the stool samples. In research
settings, methods of test interpretation may be well
standardized; however, if an FOBT is to be widely
used, a more objective method is essential for avoiding interobserver variation.
The Magstream 1000/Hem SP automated system
(Fujirebio, Inc., Tokyo, Japan) is an immunologic
FOBT that is easy to use and well suited to centralized
management of a high test volume, because it elimi-
2421
nates human bias in the interpretation of results. The
assay evolved from the Immudia-Hem SP test (Fujirebio, Inc.). The hemagglutination technique using
chicken red cells in the Immudia-Hem SP test was
replaced by magnetically induced agglutination,
which was made possible by the attachment of rabbit
antibodies against human hemoglobin to particles of
gelatin containing ferrite and gum arabic.13 This modification led to faster agglutination and produced a
more stable agglutinate, which allowed automated result interpretation by the Magstream 1000 instrument.14 The assay is suitable for high-volume testing
and general population screening.
After our initial experience with a sensitive guaiac
FOBT, in which we found the test unsuitable for our
Chinese study population because of its low specificity
for significant neoplasia, we performed the current
study to evaluate the Magstream 1000/Hem SP system.
MATERIALS AND METHODS
Patient Population
From May 2001 to June 2002, 250 consecutive patients
at Queen Mary Hospital (Hong Kong) who required
colonoscopy for the investigation of gastrointestinal
symptoms or colonic polyp surveillance were recruited for study. Patients with a previously positive
FOBT, history of overt gastrointestinal bleeding, active
rectal bleeding, menstruation, hematuria, and known
ulcerative colitis were excluded. Patients were asked to
begin fecal sampling for the FOBT at least five days
before hospital admission to ensure that two samples
were collected before bowel preparation commenced.
No dietary restriction was required. Medications such
as aspirin and NSAIDs were withdrawn one week before preparation of the stool tests. The study was approved by a local ethics committee, and all examinees
provided written informed consent.
Patients were asked to prepare fecal samples from
two consecutive stool specimens using the collection
kit provided by the manufacturer (Fujirebio, Inc.).
Each collection kit contained two collection tubes, two
sample collection probes, identification labels, a patient instruction sheet, and stool collection papers in a
resealable plastic bag. The collection tube consisted of
a translucent tube and a push-fit lid, which was held
in place by a small adhesive label. The outer edge of
the lid was hollow, and the internal diameter of the lid
decreased in steps until it reached a soft septum seal.
Stool specimens were collected with a sample probe
by repeated scraping until the specimen collection
line was stained. The probe then was pushed through
the septum seal, and the tip of the probe was immersed in buffered saline solution, which contained
2422
CANCER May 15, 2003 / Volume 97 / Number 10
rabbit serum and sodium azide (a preservative). Collection tubes were returned to the hospital on the day
of colonoscopy. Samples were collected with dropper
tips that were provided by the manufacturer. Each tip
had a shielded point to pierce through the bottom of
the collection tube. Sample aliquots (25 ␮L) were
transferred to microtiter plates, which then were
placed in a rack on the side of the Magstream 1000 and
automatically carried into the machine.
The Magstream 1000/Hem SP immunoassay system is a microprocessor-controlled analytic system for
automatic processing of the Hem SP magnetic particle
immunoassay. The system used an immunologic indirect agglutination method to detect fecal occult
blood with rabbit anti– human hemoglobin, which
was attached to magnetic gelatin particles. Agglutination of the magnetic gelatin particles was accelerated
by the use of magnetic force to draw particles together
at the bottom of a V-shaped well. The plate then was
tilted to 60° from horizontal to allow free magnetic
particles to run down the slope of the well and form a
measurable line. In the presence of human hemoglobin, particles remained in the shape of a tight button
or a short line during tilting. The length of the line was
determined optically using reflected 690-nm light and
was measured in standard units (SU) (1 SU ⫽ 0.03
mm). The cutoff recommended by the manufacturer
was 100 SU, but the cutoff level was adjustable. Samples with values ⬎ 100 SU were deemed negative,
whereas samples with values ⱕ 100 SU were deemed
positive. The system had a sensitivity of 20 mg hemoglobin per liter or 0.1– 0.2 mg hemoglobin per gram of
feces. The positive and negative controls (25 ␮L each)
were purified human hemoglobin and buffer diluent
containing azide, respectively.
Correlation with Clinical Findings
Colonoscopy was performed without knowledge of
FOBT results. All participating endoscopists had at
least six years of experience in colonoscopy. If colorectal polyps were detected, the polyp site was recorded and polypectomy performed. Polyps were examined histologically, and the size and histologic type
of each polyp were recorded. The locations and histologies of carcinomas also were recorded. Colonoscopy was incomplete in 2 patients (1%) due to obstructing tumors of the sigmoid colon.
Statistical Analysis
The presence of an adenomatous polyp of any size or
a carcinoma was considered a positive finding on
colonoscopy. Patients were classified by largest detected adenoma or by the presence of carcinoma.
Findings were analyzed based on the number of de-
TABLE 1
Patient Demographic Data, Indications for Endoscopy, and
Endoscopy Findings (Classified by Worst Neoplasia)
Demographic data
M/F (%)
Mean age ⫾ SD (range) (years)
Indication for colonoscopy (%)
History of colonic polypsa
History of cured colorectal carcinomaa
Family history of colorectal neoplasiaa
Abdominal pain
Anemia
Alteration of bowel habits
Other
Findings at colonoscopy (%)
Normal
Inflammatory polyps
Hyperplastic polyps
Adenoma
Cecum and ascending colon
Transverse colon
Descending and sigmoid colon
Rectum
Carcinoma
Cecum and ascending colon
Transverse colon
Descending and sigmoid colon
Rectum
134/116 (54/46)
59.5 ⫾ 9.5 (40–81)
68 (27)
38 (15)
28 (11)
16 (6)
28 (11)
51 (20)
21 (8)
165 (66)
2 (1)
12 (5)
14 (6)
12 (5)
29 (12)
7 (3)
4 (2)
0 (0)
2 (1)
1 (1)
SD: standard deviation.
a
Patients were asymptomatic.
tected neoplasms (adenomas or carcinomas), adenomas only, and clinically significant neoplasms (adenomas ⱖ 1.0 cm in diameter or carcinomas). The
sensitivity, specificity, and positive predictive values of
the Magstream 1000/Hem SP system were calculated
using the cutoff provided by the manufacturer as well
as other cutoff values. To evaluate the effect of time to
development on sensitivity, we arbitrarily divided the
study population into two halves according to time to
development. The statistical methods used included
the Student t test, chi-square test, and Fisher exact
test; differences were estimated by comparing confidence intervals from paired tests. P ⬍ 0.05 was considered statistically significant. All P values were twosided.
RESULTS
Demographic and clinical data for the 250 patients
were tabulated (Table 1). The mean patient age was
59.5 years. Fifty-four percent of patients were male,
and all were ethnically Chinese. Seventy-three (54%)
patients were asymptomatic (personal history of colonic polyps, personal history of cured colorectal carcinoma, and family history of colorectal neoplasia).
Sixty-three patients were found to have adenomatous
Automated Immunochemical FOBT/Wong et al.
2423
TABLE 2
Positivity Rate (at Various Cutoff Values) of Magstream 1000/Hem SP for Detection of Colorectal Carcinoma, Adenoma, and Significant
Colorectal Neoplasiaa in 250 Chinese Patients Who Underwent Colonoscopy
Cutoff (SUb)
Endpoint (%)
50
60
70
80
90
100
Overall positivity (n ⫽ 250)
Positivity in normal colon (n ⫽ 165)
Carcinomas (n ⫽ 7) detected
Large adenomas (n ⫽ 19) detected
Significant neoplasms (n ⫽ 26) detectedc
All adenomas (n ⫽ 63) detected
Specificity of significant neoplasia (n ⫽ 165)
PPV for carcinoma
PPV for significant neoplasia
15 (6)
2 (1)
4 (57)
4 (21)
8 (31)
8 (13)
163 (99)
4/15 (27)
8/15 (53)
29 (12)
8 (5)
7 (100)
6 (32)
13 (50)
13 (21)
157 (95)
7/29 (24)
13/29 (45)
36 (14)
12 (7)
7 (100)
9 (47)
16 (62)
16 (25)
153 (93)
7/36 (19)
16/36 (44)
43 (17)
17 (10)
7 (100)
9 (53)
16 (65)
16 (29)
148 (90)
7/43 (16)
16/43 (37)
44 (18)
18 (11)
7 (100)
9 (53)
16 (65)
16 (29)
147 (89)
7/44 (16)
16/44 (36)
51 (20)
22 (13)
7 (100)
10 (53)
17 (65)
18 (29)
143 (87)
7/51 (14)
17/51 (33)
SU: standard units; PPV: positive predictive value.
a
Includes adenomas ⱖ 10 mm and carcinomas.
b
1 SU ⫽ 0.03 mm.
c
Equivalent to overall detection sensitivity.
polyps (19 had adenomas ⱖ 1.0 cm), and 7 had CRC.
Colonoscopies were normal for 165 patients. Two of
the seven patients with CRC had obstructing lesions
that prevented a complete colonoscopic examination.
The test had a mean time to development of 5.9 days
(median, 3 days; range, 0 –33 days) from the first day of
sample collection. Mean time to development was
similar for FOBTs with positive results and FOBTs with
negative results. All patients returned their samples;
none defaulted. The collection kit was easy to use, and
there was no leakage during delivery.
Test Sensitivity in Detection of Colorectal Carcinoma and
Significant Colorectal Neoplasia
All cases of CRC were detected by the Magstream
1000. At the standard cutoff (100 SU), the sensitivity,
specificity, and positive predictive value of the Magstream 1000 for detection of CRC were 100%, 87%,
and 14%, respectively. For significant colorectal
neoplasms (adenomas ⱖ 1.0 cm in diameter or carcinomas), the detection rate was 65% at the standard cutoff. There was no difference in the sensitivity of significant neoplasia detection between the
early and late stool specimen development groups
(P ⫽ not significant) (Table 2).
Test Sensitivity in Detection of Adenomas
The mean adenoma size for all patients was 0.7 cm
(range, 0.2–3.0 cm). Mean adenoma size was significantly larger for patients with positive Magstream
1000 results compared with patients with negative
results (1.2 cm vs. 0.58 cm, P ⬍ 0.001). The sensitivity
of the Magstream 1000 for detection of all adenomas
was 29% using the standard cutoff of 100 SU (Table 2).
Effect of Cutoff Value
Sensitivities, specificities, and positive predictive values were calculated for different cutoffs. The detection
rate of CRC remained at 100% even when the cutoff
was decreased to 60 SU. At a cutoff of 70 SU, the
sensitivity, specificity, and positive predictive value for
detection of significant colorectal neoplasia were 62%,
93%, and 44%, respectively; however, the detection
rate for large adenomas (size, ⱖ 1 cm) decreased from
53% to 47% (Table 2).
DISCUSSION
In the current study, a one-time FOBT had a positive
predictive value, sensitivity, and specificity of 33– 44%,
62– 65%, and 87–93%, respectively, for the detection of
significant colorectal neoplasia in a mixed group of
symptomatic and asymptomatic individuals; results
depended on the development threshold chosen.
These results are suitable for screening and are consistent with results from other studies that used immunochemical tests to screen larger populations.15–18
The test used in the current study is easy to perform,
and results are independent of operator experience.
Although the number of patients was small and all
patients were recruited after the decision to perform
colonoscopy had been made, we were able to obtain
useful information regarding the accuracy of the Magstream 1000/Hem SP FOBT in a group of Chinese
patients.
One of the main advantages of using an immuno-
2424
CANCER May 15, 2003 / Volume 97 / Number 10
chemical FOBT was that the test lacked a dietary restriction requirement, which would not have been
widely acceptable to the local population. In general,
the Chinese eat a variety of food items and consume a
large amount of fruits and vegetables every day; such
a diet produced a high false positive rate for the guaiac
test, despite the adoption of a three-day delayed development period (Wong et al. 2003, unpublished
data). In addition, extensive screening experience in
Japan suggested that a two-day immunochemical
FOBT was more suitable for that population than was
the guaiac test.11,12
At the standard cutoff, the positivity rate for samples from patients with “normal” colonoscopy findings was 13%; this rate is quite high and would lead to
an excessive colonoscopy rate. However, the population that we studied was not a true screening population, and 46% of examinees had abdominal symptoms. Lowering the cutoff value to 70 SU decreased
the false positive rate to 7%, but this improvement was
offset slightly by decreased sensitivity in the detection
of large adenomas (down from 53% to 47%). We believe that these undetected lesions eventually would
be identified with annual retesting. A large pilot study
of truly asymptomatic patients could help to determine the optimal cutoff value in the study population.
The current study provides useful data regarding
the potential application of an automated immunochemical FOBT in screening the Chinese population
for CRC. We have demonstrated that the Magstream
1000/Hem SP assay is a robust, specific, and accurate
tool for the detection of significant colorectal neoplasia and provides the basis for a large-scale screening
program in the Chinese population.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
REFERENCES
1.
2.
3.
Rozen P. Colorectal cancer: does early detection matter?
Postgrad Med J. 2001;77:289 –291.
Winawer SJ, Zauber AG, Ho MN, et al. Prevention of colorectal cancer by colonoscopic polypectomy. N Engl J Med.
1993;329:1977–1981.
Mandel JS, Bond JH, Church TR, et al. Reducing mortality
18.
from colorectal cancer by screening for fecal occult blood.
N Engl J Med. 1993;328:1365–1371.
Hardcastle JD, Chamberlain JO, Robinson MH, et al. Randomised controlled trial of faecal-occult-blood screening for
colorectal cancer. Lancet. 1996;348:1472–1477.
Kronborg O, Fenger C, Olsen J, Jorgensen OD, Sondergaard
O. Randomised study of screening for colorectal cancer with
faecal-occult-blood test. Lancet. 1996;348:1467–1471.
Mandel JS, Church TR, Bond JH, et al. The effect of fecal
occult-blood screening on the incidence of colorectal cancer. N Engl J Med. 2000;343:1603–1607.
Hong Kong Cancer Registry. Cancer incidence and mortality
in Hong Kong, 1998 –1999. Hong Kong: Hospital Authority,
2000.
Greenlee RT, Murray T, Bolden S, Wingo PA. Cancer statistics, 2000. CA Cancer J Clin. 2000;50:7–33.
Parkin DM, Pisani P, Ferlay J. Estimates of the worldwide
incidence of 25 major cancers in 1990. Int J Cancer. 1999;
80:827– 841.
Ji BT, Devesa SS, Chow WH, Jin F, Gao YT. Colorectal cancer
incidence trends by subsite in urban Shanghai, 1972–1994.
Cancer Epidemiol Biomarkers Prev. 1998;7:661– 666.
Saito H, Yoshida Y. Mass screening: Japanese perspective.
In: Young GP, Rozen P, Levin B, editors. Prevention and
early detection of colorectal cancer. London: Saunders,
1996:301–312.
Saito H. Screening for colorectal cancer: current status in
Japan. Dis Colon Rectum. 2000;43:S78 – 84.
Sohma K, Kinoshita Y, Morishita S, Fujino R, Saito T. A novel
magnetic particle agglutination in microtiter plates for rapid
detection of human T-lymphotropic virus type I antibody.
J Clin Lab Anal. 1995;9:59 – 62.
Sohma K, Konshita Y, Morishita S, Saito T. Automatic assay
of fecal occult blood test by Magstream system. J Jpn Soc
Clin Assay. 1996;21:197–201.
Allison JE, Tekawa IS, Ransom LJ, Adrain AL. A comparison
of fecal occult-blood tests for colorectal-cancer screening.
N Engl J Med. 1996;334:155–159.
St John DJ, Young GP, Alexeyeff MA, et al. Evaluation of new
occult blood tests for detection of colorectal neoplasia. Gastroenterology. 1993;104:1661–1668.
Robinson MH, Marks CG, Farrands PA, Thomas WM, Hardcastle JD. Population screening for colorectal cancer: comparison between guaiac and immunological faecal occult
blood tests. Br J Surg. 1994;81:448 – 451.
Petrelli N, Michalek AM, Freedman A, Baroni M, Mink I,
Rodriguez-Bigas M. Immunochemical versus guaiac occult
blood stool tests: results of a community-based screening
program. Surg Oncol. 1994;3:27–36.