GASTROENTEROLOGY 2008;135:380 –399
`
`REVIEWS IN BASIC AND CLINICAL
`GASTROENTEROLOGY
`
`BASIC AND CLINICAL
`GASTROENTEROLOGY
`
`REVIEWS IN
`
`Screening, Surveillance, and Primary Prevention for Colorectal Cancer: A
`Review of the Recent Literature
`
`CHARLES J. KAHI,*,‡ DOUGLAS K. REX,* and THOMAS F. IMPERIALE*,§
`
`*From the Division of Gastroenterology and Hepatology, Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana; ‡The Richard L.
`Roudebush VA Medical Center, Indianapolis, Indiana; and §Regenstrief Institute, Inc, Indianapolis, Indiana
`
`Colorectal cancer (CRC) remains a major clinical and
`
`public health challenge, with 148,810 new cases and
`49,960 deaths expected in the United States in 2008.1 The
`field of CRC research is dynamic and expanding in several
`directions, encompassing areas of clinical and outcomes
`research, epidemiology, public health, and molecular sci-
`ences. In this review, we summarize important develop-
`ments in CRC screening and surveillance over the past
`several years and discuss the present state of the art of
`this field.
`
`Risk Factors for Colorectal Neoplasia
`Metabolic Syndrome
`According to the National Cholesterol Education
`Program’s Adult Treatment panel III, metabolic syn-
`drome is the presence of 3 or more of the following
`factors: hypertension (blood pressure of 130/85 mm Hg
`or greater), central adiposity (waist circumference greater
`than 102 cm in men or greater than 88 cm in women or
`a body mass index [BMI] greater than 27 [kg/m2]), low
`high-density lipoprotein (HDL) cholesterol (HDL ⬍40
`mg/dL in men or ⬍50 mg/dL in women), hypertriglyc-
`eridemia (150 mg/dL or greater), and impaired glucose
`tolerance (fasting serum glucose of 110 mg/dL or great-
`er).2 Colorectal neoplasia has been associated with mark-
`ers of glucose and insulin control; insulin resistance,
`which is the cornerstone of the metabolic syndrome, may
`be the mechanism by which several risk factors (obesity,
`diabetes mellitus, [lack of] fitness) affect colorectal carci-
`nogenesis.3,4
`Four of the most recent studies of metabolic syndrome
`and CRC are summarized in Table 1.5– 8 These studies
`comprise different study populations and different study
`designs but use the same or comparable definitions of
`metabolic syndrome, similar methods of analysis, and
`either adenoma or cancer as outcomes. The study find-
`ings are quite consistent: either the metabolic syndrome
`or its components increase the risk for colorectal neopla-
`sia (both adenomas and cancer) by approximately 50%.
`The effect of metabolic syndrome on neoplasia risk ap-
`
`pears to be greater in men than in women. The relation-
`ship between metabolic syndrome and large bowel loca-
`tion of neoplasia reported by Chiu et al6 is interesting
`and requires validation in analyses of other populations.
`
`Cigarette Smoking
`The epidemiologic evidence that cigarette smok-
`ing increases the risk of CRC was elegantly reviewed by
`Giovannucci in 2001.9 An association between colorectal
`neoplasia and cigarette smoking is supported by several
`studies, with the association more consistently estab-
`lished for smoking and adenomas, including large ade-
`nomas, than for cancer.9 –24 Recently, the bulk of the
`evidence supports an association with CRC as well. With
`men having begun smoking several decades earlier than
`women, the temporal pattern of the studies supports an
`induction period of 3– 4 decades between exposure and
`the diagnosis of CRC.9 Despite the volume of studies,
`several questions remain unanswered: What is the rela-
`tionship between dose and duration and risk of neopla-
`sia? Which persons are most susceptible to the effects of
`cigarette smoking? Is smoking associated to specific sub-
`groups of cancer, perhaps having one or more prevalent
`mutations? By how much and how quickly does risk drop
`after quitting smoking?
`Table 2 summarizes recent selected endoscopic and
`population-based studies on smoking and risk for colo-
`rectal neoplasia.25–29 The 5 studies use different study
`designs: cohort, case-control, and cross-sectional, with
`sample sizes ranging from 1154 to 146,877 individuals.
`All 5 use multivariable analysis, which provides the inde-
`
`Abbreviations used in this paper: CAD, coronary artery disease;
`CCSA, colon cancer-specific antigen; CTC, computed tomographic
`colonography; FDR, first-degree relative; gFOBT, guaiac-based fecal
`occult blood testing; HNPCC, hereditary nonpolyposis CRC; iFOBTs,
`immunochemical fecal occult blood tests; MMR, mismatch repair;
`MSI, microsatellite instability.
`© 2008 by the AGA Institute
`0016-5085/08/$34.00
`doi:10.1053/j.gastro.2008.06.026
`
`Geneoscopy Exhibit 1050, Page 1
`
`

`

`GASTROENTEROLOGY
`BASICANDCLINICAL
`
`REVIEWSIN
`
`August 2008
`
`REVIEW OF RECENT STUDIES OF PREVENTION OF CRC 381
`
`Table 1. Summary of Selected Studies on Metabolic Syndrome and Risk of Colorectal Neoplasia
`
`First author,
`year (ref)
`
`Ahmed,
`2006 (5)
`
`Study population
`
`14,109 subjects from the
`Atherosclerosis Risk in
`Communities (ARIC)
`multicenter study)
`
`Study
`design
`
`Criteria for metabolic
`syndrome
`
`Outcomes
`
`Type of risk model
`
`Main findings
`
`Cohort
`
`ATP III
`
`Colorectal cancer
`
`Multiple logistic model,
`adjusted for age, gender,
`family history of CRC,
`physical activity, NSAID
`use, aspirin use,
`smoking, alcohol use,
`hormone replacement
`use
`
`Multiple logistic model,
`adjusted for age,
`smoking, exercise,
`alcohol use, multivitamin
`use, and consumption of
`fruits and vegetables
`
`MS associated with increased risk
`of CRC (age and gender
`adjusted RR, 1.49; 95% CI:
`1.0–2.4), which attenuated
`after multivariate adjustment
`(RR, 1.39; 95% CI: 0.9–2.2).
`Adjusted risk was increased in
`men (RR, 1.78; 95% CI: 1.02–
`3.6) but not in women (RR,
`1.16; 95% CI: 0.6–2.2)
`BMI ⬎27 kg/m2 (RR, 1.4; 95% CI:
`1.1–1.7) and diabetes (RR, 1.5;
`95% CI: 1.1–2.0) were
`associated with CRC;
`hypertension and
`hypercholesterolemia were not.
`
`Colorectal cancer
`
`Sturmer,
`2006 (8)
`
`22,071 healthy male
`physicians initially ages
`40–84 years
`
`Cohort
`
`Kim,
`2007 (7)
`
`3584 consecutive subjects
`undergoing screening
`colonoscopy
`
`Cross-
`sectional
`
`BMI of ⱖ27 kg/m2, total
`cholesterol of ⱖ240
`mg/dL or use of lipid-
`lowering drugs, blood
`pressure of ⱖ130/85
`mm Hg or use of
`antihypertensives, and
`a diagnosis of
`diabetes mellitus
`Modified ATP III criteria
`
`Colorectal adenoma
`
`Multiple logistic model,
`adjusted for age, gender,
`smoking, alcohol use
`
`Chiu,
`2007 (6)
`
`4277 consecutive ethnic
`Chinese who had
`screening or surveillance
`colonoscopy as part of a
`medical health checkup
`
`Cross-
`sectional
`
`Modified ATP III criteria,
`modified Asian criteria
`(HDL cholesterol of
`⬍40 mg/dL, waist
`circumference ⱖ90
`cm for men, ⱖ80 cm
`for women)
`
`Colorectal neoplasia,
`anatomic location
`
`Multiple logistic model,
`adjusted for age, gender,
`BMI, smoking, alcohol
`use, previous adenoma,
`family history of CRC
`
`MS, metabolic syndrome.
`
`17% of subjects with adenomas
`and 11% of those without
`adenomas had MS. MS
`associated with increased risk
`of adenoma: OR, 1.51; 95% CI:
`1.19–1.93. Waist circumference
`was an independent risk factor
`for adenoma: OR, 1.39; 95% CI:
`1.15–1.68
`MS associated with increased risk
`of any neoplasia (OR, 1.35;
`95% CI: 1.05–1.73), proximal
`neoplasia (OR, 1.62; 95% CI:
`1.14–2.30), synchronous
`lesions (OR, 2.15; 95% CI:
`1.40–3.31), and synchronous
`lesions both proximal and distal
`(OR, 2.30; 95% CI: 1.42–3.72).
`
`pendent effect of smoking after adjustment for covariates
`such as age, sex, BMI, and others.
`The study by Lieberman et al, a colonoscopy-based study,
`examined the effect of several candidate risk factors on the
`risk of advanced neoplasia in a cohort of 3121 asymptom-
`atic patients aged 50–75 years from 13 Veterans Affairs
`medical centers.27 Using a multivariate model that included
`family history of CRC, BMI, physical activity, smoking,
`alcohol use, and several dietary components, the investiga-
`tors found that the effect of smoking on advanced neopla-
`sia (odds ration [OR], 1.85; 95% confidence interval [CI]:
`1.33–2.58) was comparable with having a first-degree rela-
`tive (FDR) with CRC (OR, 1.66; 95% CI: 1.16–2.35). In a
`retrospective, cross-sectional analysis of 1988 persons un-
`dergoing screening colonoscopy, Anderson et al found that
`cigarette smoking increased the risk for any colorectal neo-
`plasia (OR, 1.89; 95% CI: 1.42–2.51) and for significant
`neoplasia (OR, 2.26; 95% CI: 1.56–3.27) that was predomi-
`nantly left-sided.26
`Results of recent population-based studies have shown
`somewhat inconsistent results. The findings of the case-
`control study by Verla-Tebit et al are consistent with
`those of several earlier epidemiologic studies that sup-
`port the 30 –35 year induction period between exposure
`to cigarette smoking and CRC.29 The results also suggest
`
`that risk reduction requires at least 20 years and increases
`with increasing duration of smoking cessation. In addi-
`tion to the findings described in Table 2, the study by
`Akhter et al, which studied only men, found that longer
`smoking duration, age of 18 or younger at onset of
`smoking, and consumption of 20 or more cigarettes per
`day significantly increased the risk of CRC, with risk
`ratios ranging from 1.46 to 1.86.25
`In the study from the Women’s Health Initiative,28
`which is a pooled analysis of participants in the observa-
`tional study and 3 clinical trials, the risk of rectal cancer
`was increased with longer smoking duration and its con-
`founder, older age at smoking cessation; however, the
`risk of colon cancer was not increased. This study has
`limitations,
`including self-reported smoking exposure
`that did not allow for changes in smoking behavior after
`initial reporting and a rate of cigarette smoking that was
`lower than US women of similar ages.
`In summary, the majority of evidence indicates that
`CRC is a tobacco-associated malignancy. In the United
`States, it has been estimated that as many as 1 in 5
`CRCs is attributable to cigarette smoking.11,13,14,20 The
`magnitude of the increase in risk for CRC and large
`adenoma appears to be the same as having an FDR
`with CRC. It would be useful to have a way to estimate
`
`Geneoscopy Exhibit 1050, Page 2
`
`

`

`382 KAHI, REX, AND IMPERIALE
`
`GASTROENTEROLOGY Vol. 135, No. 2
`
`Table 2. Summary of Selected Studies on Cigarette Smoking and Risk of Colorectal Neoplasia
`
`Study population
`
`Study design
`
`Outcomes
`
`Type of risk model
`
`Main findings
`
`First author,
`year (ref)
`
`Lieberman,
`2003 (27)
`
`BASIC AND CLINICAL
`GASTROENTEROLOGY
`
`REVIEWS IN
`
`3121 asymptomatic patients
`aged 50-75 years from 13
`Veterans Affairs medical
`centers
`
`Cross-sectional
`
`Advanced neoplasia
`(CRC or advanced
`polyps)
`
`Multiple logistic regression that adjusted
`for age, family history, BMI, physical
`activity, alcohol use, NSAID use,
`certain dietary features
`
`Anderson,
`2003 (26)
`
`1988 persons aged 40 and
`older undergoing screening
`colonoscopy
`
`Cross-sectional
`
`Significant neoplasia
`(CRC, advanced
`polyps, or ⬎2
`adenomas of any
`size)
`
`Multiple logistic regression that adjusted
`for age, alcohol consumption,
`exercise, BMI, ethnicity, education,
`and consumption of fruits and
`vegetables
`
`Verla-Tebit,
`2006 (29)
`
`540 patients with incident
`CRC and 614 population-
`based, matched to cases
`by 5-year age group, sex,
`county of residence
`
`Case-control
`
`CRC
`
`Akhter,
`2007 (25)
`
`25,279 Japanese men
`recruited when aged 40-64
`years
`
`Cohort (7 years
`of follow-up)
`
`CRC
`
`Paskett,
`2007 (28)
`
`146,877 women’s Health
`Initiative participants
`
`Cohort (mean
`of 7.8 years
`of follow-up)
`
`CRC
`
`Multiple logistic model that adjusted for
`age, sex, history of CRC in first-degree
`relatives, BMI, alcohol use, physical
`activity, fruit and vegetable intake, red
`meat consumption, NSAID use,
`previous endoscopy of the large bowel,
`education level, and use of hormone
`replacement therapy
`
`Proportional hazards model that adjusted
`for age, family history of CRC,
`education, BMI, alcohol use, time
`spent walking per day, and
`consumption frequency of fruits, green-
`yellow vegetables, and red meat
`
`Proportional hazards model that adjusted
`for age, ethnicity, study type
`(observational or clinical trial) study
`arm, family history of CRC, total
`physical activity metabolic equivalents,
`alcohol use, NSAID use, hormone
`therapy use, colonoscopy, diabetes,
`waist circumference, certain dietary
`features
`
`Current smoking was a risk factor for
`advanced neoplasia (OR, 1.85; 95% CI:
`1.33-2.85) and was of comparable
`magnitude to having an FDR with CRC
`(OR, 1.66; 95% CI: 1.16–2.35).
`Current smokers were more likely to have
`any neoplasia (OR, 1.89; 95% CI,
`1.42–2.51) and significant neoplasia
`(OR, 2.26; 95% CI: 1.56–3.27). Risk of
`significant neoplasia was greater for
`smokers than for those with a family
`history of CRC.
`Compared with nonsmokers, smokers for
`ⱖ40 years had increased risk (OR,
`1.92; 95% CI: 1.13–3.28). Among
`smokers ⱖ30 years, risk was greater
`among women (OR, 3.5; 95% CI: 1.29–
`9.52) than men (OR, 1.15; 95% CI:
`0.69–1.91). Risk reduction observed
`after ⱖ20 years of quitting smoking
`and was significant for ⱖ40 years (OR,
`0.46; 95% CI: 0.21–0.98).
`Compared with never smokers, the risk of
`CRC was increased for past smokers
`(RR, 1.73; 95% CI: 1.04–2.87) and
`current smokers (RR, 1.47; 95% CI:
`0.93–2.34). Among current smokers, a
`greater number of cigarettes smoked
`per day and an earlier age of smoking
`onset were associated with a
`significant linear increase in CRC risk.
`Current smokers had increased risk for
`rectal cancer (HR, 1.95; 95% CI: 1.10–
`3.47) but not colon cancer (HR, 1.03;
`95% CI: 0.77–1.38).
`
`the incremental effect of smoking on risk of advanced
`neoplasia that considers sex; age of smoking onset;
`degree and duration of cigarette consumption; and, for
`former smokers, time since smoking cessation. Greater
`consideration should be given to cigarette smoking
`when considering whether, when, and how best to
`screen patients.
`
`Coronary Artery Disease
`In a study from Hong Kong, Chan et al com-
`pared the prevalence of colorectal neoplasia in 206
`subjects with angiographically proven coronary artery
`disease (CAD), 208 subjects whose angiogram did not
`show CAD, and an age- and sex-matched control group
`of 207 subjects who were asymptomatic (other than
`having functional dyspepsia with a normal upper en-
`doscopy) but who did not have angiography.30
`Colonoscopy was scheduled within 8 weeks after eligi-
`bility was determined or after revascularization. En-
`doscopists were blinded to CAD status.
`The prevalence of advanced neoplasia in the CAD-
`positive, CAD-negative, and control groups was 18.4%,
`8.7%, and 5.8%, respectively (P ⬍ .001), whereas the prev-
`alence of cancer was 4.4%, 0.5%, and 1.4%, respectively (P
`
`⫽ .02). After adjustment for age and sex, CAD remained
`associated with advanced neoplasia (OR, 2.51; 95% CI:
`1.43– 4.35). Of interest, both metabolic syndrome and
`cigarette smoking were strong independent predictive
`factors for the positive association between CAD and
`advanced neoplasia, meaning that persons who were
`smokers and/or had the metabolic syndrome were much
`more likely to develop both conditions.
`Although it is not clear that the CAD-positive group
`was free of symptoms of signs of early CRC, this study
`links CAD with advanced neoplasia and is consistent
`with previously published studies.31,32 It is unclear
`whether and to what extent the association would
`remain after further adjustment for other confounding
`factors. Nevertheless, most likely because of a common
`set of risk factors that includes cigarette smoking,
`waist circumference, diabetes, and others, CAD ap-
`pears to be a marker for colorectal neoplasia. Although
`the prevalence of advanced neoplasia in persons with
`CAD suggests the need for earlier or more aggressive
`CRC screening, the extent to which CAD as a comorbid
`condition may reduce the benefits of screening re-
`quires careful consideration.
`
`Geneoscopy Exhibit 1050, Page 3
`
`

`

`GASTROENTEROLOGY
`BASICANDCLINICAL
`
`REVIEWSIN
`
`August 2008
`
`REVIEW OF RECENT STUDIES OF PREVENTION OF CRC 383
`
`Diabetes Mellitus
`Previous studies have shown that the risk of CRC
`is higher among persons with diabetes, although this
`finding is not consistent among studies nor is the con-
`tribution of confounding factors to the increased risk
`well established. In a population-based cohort study,
`Limburg et al identified incident cases of CRC among
`1975 type 2 diabetic individuals and compared them with
`what was expected from the general population.33 Overall
`risk of CRC was increased among diabetic individuals
`(standardized incidence ratio⫽ 1.39; 95% CI: 1.03–1.82).
`However, the increased risk was present among men only,
`both overall (SIR, 1.67; 95% CI: 1.16 –2.33) and proxi-
`mally (SIR, 1.96; 95% CI: 1.16 –3.10). Furthermore, cur-
`rent and former cigarette smokers were at higher risk for
`CRC than diabetic individuals who never smoked.
`In addition to increasing baseline risk for colorectal
`neoplasia, insulin resistance also increases the risk for
`recurrent neoplasia. In an analysis from the Polyp Pre-
`vention Trial, Flood et al compared fasting insulin and
`glucose levels in 375 subjects with and 375 subjects
`without recurrent adenoma.34 After adjustment for age,
`sex, BMI, and intervention group, risk for recurrent ad-
`enoma was higher for subjects in the highest quartile
`compared with the lowest quartile: OR, 1.56; 95% CI:
`1.00 –2.43 for insulin; OR, 1.49; 95% CI: 0.95–2.31 for
`glucose. The highest quartile of glucose was associated
`with advanced adenoma as well: OR, 2.43; 95% CI: 1.23–
`4.79. The strength of the associations between high fast-
`ing glucose and risk of recurrent adenoma increased
`when the analysis was restricted to subjects with no
`family history of CRC.
`These studies support other research in which diabetes
`has been associated with increased risk for CRC and are
`consistent with a larger body of literature that links
`insulin resistance, metabolic syndrome, and coronary ar-
`tery disease with colorectal neoplasia. Understanding
`both the mechanisms leading to neoplasia and the inde-
`pendent contribution of each of these factors to ad-
`vanced adenoma and CRC risks requires further study.
`Although the literature is replete with data on risk
`factors for CRC and adenoma, most established risk
`factors are not incorporated into current screening guide-
`lines. Current guidelines stratify risk with age and family
`history alone. Age is used only as a threshold factor,
`although CRC incidence increases with age in an approx-
`imately linear fashion. The risk of CRC in average-risk
`persons doubles by 10 years—approximately the same
`increase in risk as having an FDR with CRC.35 We need a
`way to integrate all risk factors (age, sex, family history,
`cigarette smoking, metabolic syndrome, and others)
`quantitatively to estimate absolute risk for CRC and
`advanced adenoma. One study has integrated age, sex,
`and BMI to estimate the risk of advanced neoplasia
`anywhere in the large intestine.36 Another study used age,
`
`sex, and most advanced distal finding to estimate the risk
`of advanced proximal neoplasia.37 Both systems require
`validation and further development before they can be
`applied to clinical practice. Furthermore, the effect of
`more extensive risk stratification on screening remains to
`be determined. On the one hand, providing risk-specific
`information to patients and providers has the potential
`to improve screening rates and screening efficiency. On
`the other hand, if risk stratification is perceived as mak-
`ing CRC screening too complicated, there is the potential
`to adversely affect further uptake of screening. Whether
`incorporating several factors with modest relative risks
`would add significantly to using age, sex, and family
`history alone is also important to determine.
`
`Screening Colonoscopy
`Several recent studies have described the findings
`of screening colonoscopy in an asymptomatic average-
`risk population.38 – 43 Table 3 summarizes the study char-
`acteristics of this body of literature. Although the study
`objectives, settings, and designs vary, the variation does
`not necessarily preclude comparing the findings.
`Descriptively, the studies are from Japan, Poland, Is-
`rael, Korea, and the United States. The number of per-
`sons analyzed varies from 994 to 43,042. The mean age
`ranges from 48.2 years in a study in which 57% of sub-
`jects were younger than 50 years old, to 62.2 years. The
`proportion of men ranges from 0% in a screening study
`of military women to 72% in a Japanese study of asymp-
`tomatic adults who participated in a comprehensive
`health examination.
`The endoscopic findings, expressed as the proportion
`of persons according to the most advanced histology, are
`shown in Table 4. Despite differences in the study pop-
`ulations, the fraction of persons with no colorectal neo-
`plasia is consistent, ranging from 75% to 83%. Ranges of
`persons with nonadvanced adenoma, advanced adenoma,
`and cancer are 8.9%–16.5%, 3%– 6.3%, and 0%–1.3%, re-
`spectively, with the variation largely because of age and
`sex.
`The prevalence of findings in these recent studies is
`comparable with previously published screening stud-
`ies,26,37,44 – 47 with the possible exception of VA Coopera-
`tive Study No. 380, in which rates of neoplasia were
`numerically greater, reflecting the high-risk features of
`the study population, particularly the high predomi-
`nance of men.46
`These studies are a reminder that the majority of
`screening colonoscopies will show no adenomas. They
`highlight the need to identify a way to estimate absolute
`risk for individual persons so that screening colonoscopy
`may be more efficiently targeted to those with advanced
`neoplasia. Considering these more recent studies in the
`aggregate, the number of persons required to undergo
`screening colonoscopy on average is approximately 9 to
`detect 1 person with 1 or more nonadvanced adenoma,
`
`Geneoscopy Exhibit 1050, Page 4
`
`

`

`384 KAHI, REX, AND IMPERIALE
`
`GASTROENTEROLOGY Vol. 135, No. 2
`
`Table 3. Description of Screening Colonoscopy Studies
`
`First author
`(ref)
`
`Schoenfeld
`(40)
`
`Year
`
`2005
`
`BASIC AND CLINICAL
`GASTROENTEROLOGY
`
`REVIEWS IN
`
`Study objective
`
`Study population
`
`Study design
`
`Study setting
`
`Recruitment period
`
`Study findings
`
`To determine prevalence
`and location of advanced
`neoplasia
`
`Consecutive, average risk,
`asymptomatic women
`referred for screeninga
`
`Prospective,
`cross-
`sectional
`
`4 Military medical centers
`
`7/1999–12/2002
`
`Advanced neoplasia was
`distal in 1.7% (n ⫽ 25)
`and proximal in 3.2% (n
`⫽ 47). Sigmoidoscopy
`would have detected
`only 35.2% of advanced
`neoplasia.
`Sensitivity and specificity
`of 1-time iFOBT were
`65.8% and 94.6%,
`respectively, for cancer
`and 27.1% and 95.1%,
`respectively, for
`advanced neoplasia.
`Despite higher prevalence
`of neoplasia in elderly
`patients, mean
`extension in life
`expectancy was much
`lower in persons aged
`80 years or older than
`in the 50–54-year-old
`groups (0.13 vs 0.85
`years, respectively).
`Advanced neoplasia and
`to quantify was more
`prevalent in men in all
`age groups, with lower
`numbers needed to
`screen in men (range,
`10–23) than in women
`(range, 18–36).
`Prevalence of neoplasia
`increased with older
`age. Among persons
`with neoplasia, 21%–
`43% had isolated
`proximal neoplasia
`(beyond the
`sigmoidoscope).
`Adenomatous polyps were
`present in 17.9%,
`advanced adenomas in
`3.4%. Adenomas were
`more prevalent in men
`(23.6%) than in women
`(11.5%) and increased
`with age in both groups.
`
`Morikawa
`(43)
`
`2005
`
`To determine the test
`characteristics of an
`immunochemical FOBT
`
`Asymptomatic adults who
`participated in a
`comprehensive health
`examination
`
`Retrospective,
`cross-
`sectional
`
`A general hospital and its
`affiliated clinic
`
`4/1983–3/2002
`
`Lin (38)
`
`2006
`
`To compare estimated life-
`years saved with
`screening colonoscopy in
`very elderly vs younger
`persons
`
`Consecutive asymptomatic
`adults undergoing
`screening colonoscopy in
`age categories 50–54
`years (n ⫽ 1034), 75–
`79 years (n ⫽ 147), and
`⬎80 years (n ⫽ 63)
`
`Regula
`(39)
`
`2006
`
`Strul (41)
`
`2006
`
`To derive and validate a
`model for detection of
`advanced neoplasia. To
`quantify the number of
`persons needed to
`screen to detect 1
`advanced neoplasm
`
`To evaluate the prevalence
`and anatomic location of
`adenoma and carcinoma
`
`Kim (42)
`
`2007
`
`To evaluate the usefulness
`of colonoscopy to detect
`polyps
`
`Consecutive, asymptomatic
`adults age 50-66 years
`in good general health;
`and those age 40-49
`years with a family
`history of cancer of any
`type
`
`Consecutive average risk
`adults who were
`asymptomatic regarding
`cancer-related symptoms
`or alarm signs
`
`Consecutive adults who
`voluntarily underwent
`colonoscopy as part of a
`health examination
`program
`
`aIncludes only persons aged 50 years and older.
`
`Prospective,
`cross-
`sectional
`
`Tertiary referral single
`medical center
`
`1/2002–1/2005
`
`Retrospective,
`cross-
`sectional
`
`Database from a national
`colonoscopy-based
`screening program
`
`10/2000–2004
`
`Retrospective,
`cross-
`sectional
`
`Databases of procedures
`from 1 of 6 outpatient
`gastroenterology clinics
`of a health maintenance
`organization in Tel-Aviv,
`Israel
`
`1/1996–2/2001
`
`Retrospective,
`cross-
`sectional
`
`Database of a company-
`based screening
`colonoscopy program
`
`1/2003–9/2005
`
`23 to detect an advanced adenoma, 20 for advanced
`neoplasia, and 143 for cancer. One goal of outcomes
`research in this area should be to identify a cluster of
`factors that define a subgroup at such low risk for ad-
`vanced neoplasia that screening may be either deferred or
`performed confidently with noninvasive testing. Another
`goal is to identify a high-risk subgroup among “average-
`risk” persons for which colonoscopy is preferred over
`other screening tests.
`
`Emerging Screening Modalities
`Fecal DNA
`The rationale for detecting mutated genes in
`feces of patients with CRC arose from studies pub-
`lished during the 1990s that established the following:
`(1) alterations in DNA were fairly neoplasm-specific,
`(2) colorectal neoplasms shed cells and released DNA
`
`continuously, and (3) polymerase chain reaction tech-
`nology could identify altered DNA in feces. Between
`2000 and 2004, several teams of investigators exam-
`ined a variety of fecal-based genetic markers for colo-
`rectal neoplasia.48 –52 Most of these studies were case-
`control studies that involved an advanced spectrum of
`CRC. A subgroup of these studies used a 21-compo-
`nent DNA panel where sensitivity for cancer ranged
`from 62% to 91% and from 27% to 82% for adenomas
`with a specificity ranging from 93% to 96%.48,53–55
`These studies begged the question of how this panel
`would perform in the screening setting.
`A multicenter study published in 2004 compared the
`21-component DNA panel with Hemoccult II among
`4404 asymptomatic average-risk subjects.44 A subgroup
`of 2507 subjects was analyzed, including all those with
`CRC and advanced adenomatous polyps plus a random
`
`Geneoscopy Exhibit 1050, Page 5
`
`

`

`GASTROENTEROLOGY
`BASICANDCLINICAL
`
`REVIEWSIN
`
`August 2008
`
`REVIEW OF RECENT STUDIES OF PREVENTION OF CRC 385
`
`Table 4. Endoscopic Findings of Recent Screening Studies
`
`First author (ref)
`
`Year
`
`Study No.
`
`Mean age, yr
`
`Gender
`(% men)
`
`No neoplasia,
`n (%)
`
`Nonadvanced
`adenoma, n (%)
`
`Advanced
`adenoma, n (%)
`
`Advanced
`neoplasia, n (%)
`
`Adenocarcinoma,
`n (%)
`
`Schoenfeld (40)
`Morikawa (43)
`Lin (38)
`Regula (39)
`Strul (41)
`Kim (42)
`
`2005
`2005
`2006
`2006
`2006
`2007
`
`1463
`21,805
`1244
`43,042a
`994a
`4491
`
`59
`48.2
`56.2
`Not available
`62.2
`48.4
`
`0
`72
`47
`35.7
`47
`53
`
`1164 (79.6)
`17,480 (80)
`1038 (83)
`32,389 (75)
`766 (77)
`3534 (79)
`
`227 (15.5)
`3544 (16)
`151 (12.1)
`3843 (8.9)
`156 (15.6)
`804 (17.9)
`
`71 (4.9)
`648 (3)
`52 (4.2)
`2168 (5.0)
`59 (5.9)
`153 (3.4)
`
`72 (4.9)
`727 (3.3)
`55 (4.4)
`2553 (5.9)
`72 (7.2)
`153 (3.4)
`
`1 (0.1)
`79 (0.4)
`3 (0.2)
`385 (0.9)
`13 (1.3)
`0 (0)
`
`aIncludes only persons aged 50 years and older.
`
`sample of subjects with no polyps or with small tubular
`adenomas. The fecal DNA panel detected 16 of 31 can-
`cers as compared with 4 of 31 for Hemoccult II (51.6% vs
`12.9%, respectively, P ⫽ .003). Among 418 subjects with
`advanced neoplasia, the panel was positive in 76 (18.2%)
`vs 45 (10.8%) for Hemoccult II (P ⫽ .001). Among sub-
`jects with no polyps, specificity was 94% for the fecal
`DNA panel and 95% for Hemoccult II.
`The results of this multicenter study were disappoint-
`ing because the sensitivity of the panel was lower than
`anticipated based on previous studies. The reason for the
`lower sensitivity was the nearly complete nonfunction of
`one of the most important components of the panel, the
`DNA integrity assay. The poor performance of the DNA
`integrity assay was due to DNA degradation of the long
`apoptotic DNA shed from neoplastic cells by fecal bac-
`terial endonucleases during overnight delivery of the
`specimens to the laboratory.
`In a subsequent study supported by the same manu-
`facturer, the prototype assay, along with markers of
`methylation, were tested in 40 subjects with CRC and 122
`subjects with a normal colonoscopy.56 Two improve-
`ments were made to the prototype assay: (1) use of a
`gel-based DNA capture approach, rather than the initial
`bead-based technology, which enhanced the extraction of
`DNA from feces; and (2) addition of a DNA-stabilizing
`buffer to the defecated specimen, which prevented DNA
`degradation. In addition, a new marker, methylation of
`the vimentin gene, was tested. Together, these improve-
`ments constituted the version 2 assay.
`The original panel of markers (version 1) had a sensi-
`tivity of 72.5% and specificity of 86.9%, with the DNA
`integrity assay alone having a sensitivity of 65% and
`specificity of 93%. The combination of the DNA integrity
`assay and vimentin gene methylation was considered to
`be the optimal combination, with a sensitivity of 87.5%
`and specificity of 82%. This combination of markers
`maintained its sensitivity across the disease spectrum:
`stage I (n ⫽ 8), 75%; stage II (n ⫽ 10), 90%; stage III (n ⫽
`17), 94%; and stage IV (n ⫽ 5), 80%. Importantly, older
`age was associated with a higher false-positive rate of
`vimentin methylation.
`The improved sensitivity and lower cost of the assay
`have improved the overall value of fecal DNA to the point
`that it may be considered a screening option for persons
`
`who are not high-risk and who would otherwise remain
`unscreened. However, despite the cost reduction of fecal
`DNA, it remains much more expensive than immuno-
`chemical fecal occult blood tests (iFOBTs). In addition,
`uncertainties regarding fecal DNA include determining
`an optimal screening interval, estimating its program-
`matic test characteristics, and determining sensitivity for
`advanced adenomas. In particular, the programmatic per-
`formance of iFOBT may be equal to or superior to fecal
`DNA with much lower costs.
`
`iFOBT
`Guaiac-based fecal occult blood testing (gFOBT)
`has been an evidence-based mainstay of CRC screening
`for well over a decade.57–59 However, it has several limi-
`tations including low sensitivity for cancer (and even
`lower for advanced polyps) and need for periodic testing;
`gFOBT reacts with nonhuman heme in food as well as
`blood from the upper gastrointestinal tract. iFOBTs were
`developed to improve specificity and eliminate the need
`for dietary restriction because they use one or more
`monoclonal or polyclonal antibodies to detect human
`hemoglobin.60 Over the last few years, several studies
`indicate that iFOBT is more sensitive than gFOBT for
`advanced colorectal neoplasia with no difference in spec-
`ificity.
`Morikawa et al published a retrospective analysis of
`data collected between 1983 and 2002 involving 21,805
`asymptomatic adults (mean age, 48 years; 72% men) who
`underwent one-time iFOBT with the Magstream 1000/
`Hem SP iFOBT system (Fujirebio, Tokyo, Japan) within 2
`days prior to colonoscopy.43 iFOBT detected 65.8% of all
`cancers and 27% of all advanced neoplasia, with respec-
`tive specificities of 94.6% and 95.1%. Sensitivity by Dukes’
`stage was 50% for Dukes’ stage A, 70% for Dukes’ stage B,
`and 78.3% for Dukes’ stage C or D. Of interest, the
`sensitivity of iFOBT was greater for advanced distal neo-
`plasia than for advanced proximal neoplasia (30.7% vs
`16.3%, respectively, P ⬍ .001).
`This study is one of the few to have compared a FOBT
`with the reference standard of colonoscopy.43 Although
`iFOBT missed one third of the cancers, only one speci-
`men was collected at a single point in time. Both collec-
`tion of more than one specimen and programmatic (ie,
`sequential) testing would very likely improve sensitivity
`
`Geneoscopy Exhibit 1050, Page 6
`
`

`

`386 KAHI, REX, AND IMPERIALE
`
`GASTROENTEROLOGY Vol. 135, No. 2
`
`with a small, if any, decline in specificity. The overall test
`characteristics for iFOBT are better than those for
`gFOBT as determined from other studies. However, this
`study did not directly compare gFOBT and iFOBT, so the
`true difference in test characteristics cannot be properly
`determined.
`Subsequent work by Guittet et al compared the per-
`formance of gFOBT and iFOBT among 10,673 average-
`risk persons aged 50 –74 year