5/22/19 Colorectal Cancer An Evidence Based Approach to Colorectal - - PDF document

5/22/19 1

An Evidence Based Approach to Colorectal Cancer Screening

Jay Ryan, M.D. Associate Professor of Medicine UCSF 5/22/2019

No Disclosures

Colorectal Cancer

• Lifetime incidence of 6%
• Prevalence of 1% in asymptomatic patients aged 50-75
• Common cause of cancer death, 2nd in men, 3rd in women
• Well defined precursor lesion (adenoma) with long lag time

until the development of cancer

• Reasonable target for screening

USPSTF CRC Screening Recommendations (2016)

Tier 1 CRC Screening Tools

• Colonoscopy screening every 10 yr
• Adenomas on index colon prompt 3-5 yr surveillance colonoscopy
• Polyp revoval prevents cancer
• Reduces the lifetime incidence of CRC by 85-90%
• Annual FIT screening
• Measures human hemoglobin in stool sample
• Positive tests prompt colonoscopy
• FIT screening not indicated in those with up-to-date colonoscopy,

known prior adenomas, or family history of CRC

5/22/19 2

Colonoscopy

• Nearly 100% sensitive for the detection of cancer, 91%

for polyps

• National Colon Polyp Study and subsequent studies

have shown that colonoscopy decreases colon cancer risk from 6% to <0.5% and can prevent death from colon cancer

Colonoscopy of Asymptomatic Patients Aged 50-75

37.7% have colorectal neoplasia: 27% TA <10 mm 5% TA >10 mm 3% Villous adenoma 1.7% High grade dysplasia/CIS 1.0% Invasive cancer

Lieberman, NEJM, 2000

Performance characteristics of colonoscopy vs FIT in CRC screening?

Colonoscopy: Gold Standard for polyp and cancer detection FIT: Newer test whose performance characteristics are being defined

All screening tests are subjected to Bayesian constraints

5/22/19 3

Bayesian analysis: Theoretical test, 99.5% sensitive and specific

• 1000 pts from high risk (50%) population:

– 500 true pos, 5 false pos – Predictive value 500/505

• 1000 pts from low risk (0.5%) population:

– 5 true pos, 5 false pos – Predictive value 5/10

Bayesian analysis: Similar test, 90% sensitive and specific

• 1000 pts from high risk population:

– 500 true pos, 100 false pos – Predictive value 500/600 = 83%

• 1000 low risk patients:

– 5 true pos, 100 false pos – Predictive value 5/105 = 4%

Practical requirements of annual fecal screening tests for CRC

• Highly sensitive for CRC and maybe for large polyps
• Should reduce the need for colonoscopy
• Should be <6-8% positive in asymptomatic screening

aged individuals

Why a 6-8% fecal test positivity rate?

• Over 10 yr [1- (0.94)10] = (1 - 0.53) or 47% eventually will be

positive and undergo colonoscopy

• 2.5% of screening patients every year undergo symptom

generated colonoscopy (25% over 10 yrs)

• Total colonoscopies over 10 yr estimated at (47% + 25%) 72%

in FIT screening programs

• HEDIS benchmark for screening is 71%

5/22/19 4

Sensitivity of FIT, Lee, et al

Overall sensitivity of FIT for CRC was 77% in all included studies by Meta analysis

Lee, et al, Ann Int Med 2014

5/22/19 5

2-y follow-up (91%) vs colonoscopy (71%) sensitive

Exclude studies not examining asymptomatic, screening aged patients and those not using colonoscopic controls Exclude studies not examining asymptomatic screening aged patients and those not using colonoscopic controls Exclude studies with FIT+ rates of over 10%

Varying the FIT Cutoff Alters Cancer Specificity and the Percent Positivity

Study Levi (2007) N 1204 Park (2010) N 770 DeWijk (2012) N 1256 Cutoff % Pos Adv Ad Cancer % Pos Adv Ad Cancer % Pos Adv Ad Cancer 50 17% NR 72% 14.2% 44.1% 12/13 (92.3%) 10% 35.4% 7/8 (88%) 75 12.5% NR 67% 12.3% 37.3% 12/13 (92.3%) 6.6% 31% 6/8 (75%) 100 11.6% NR 61% 11.3% 33.9% 12/13 (92.3%) 5.6% 29.2% 6/8 (75%) 125 9.8% NR 53% 10% 28.8% 11/13 (84.6%) 150 9.4% NR 53% 7.9% 27.1% 11/13 (84.6%)

Lee, Ann Int Med 2014

5/22/19 6

Hi Quality FIT Studies: Positive <10%, screening age, colonoscopically controlled

Study N % Positive Sens AA Sens CRC Levi (2007) 1000 9.4% NR 53% Morikawa (2005) 21,805 5.6% 27.1% 65.8% Chiu (2013) 18,296 7.3% 28% 78.6% Brenner (2013) 2235 5.0% 23.4% 60.0% Brenner (2013) 2235 5.0% 20.4% 53.3% Brenner (2013) 2235 5.0% 25.7% 73.3% Lee, Ann Int Med 2014

16.2% Pos 7.0% Pos

Incidence of Colonic Neoplasia by Group

5/22/19 7

Is colonoscopic screening feasible?

SF-VAMC GI Unit 44,000 screening age patients

• 1994: Commitment to colonoscopic screening strategy
• 1996: Only 57 screening colonoscopies
• 1998: Direct access scheduling by GI nurses
• 1999: Telephone scheduling by GI nurses
• 1999: Elimination of routine clinic visits for path FU
• 1997-2003: Marked increase in exams for even minimal chronic symptoms

(de facto screening)

• 2002-2005: Steady state reached at 76-79% with CRC screening from

reminder data

CRC Screening at the SFVAMC 1994-Present

Cancers 1998-2011

25 Cancers in those with prior colonoscopy: 12 diagnosed outside surveillance interval 13 cancers Dx within surveillance interval:

• 4 recurrence at large polyp site
• 3 following normal colonoscopy at OSH
• 6 likely missed lesions at SFVA

5/22/19 8

SFVAMC CRC Screening Results

• Fecal testing has never been programmatically utilized at the SFVAMC
• Efforts spent chasing down fecal tests have rather been spent

improving GI unit efficiencies

• Colonoscopic screening is currently in steady state at 83% among

screening aged SFVAMC patients and is nearly completely protective from the development of CRC

• Colon cancer at the SFVAMC is now vanishingly rare and the paucity
• f new diagnoses has resulted in multiple investigatory site visits by

the US Inspector General

USPSTF CRC Screening Recommendations (2016)

Tier 2 CRC screening options

• Flexible sigmoidoscopy every 5 yr:
• 70.3% sensitive for cancer, 70% sensitive for advanced adenomas
• 70% sensitive for small adenomas
• Prompt colonoscopy in 35% of cases in a 10 yr span
• Inexpensive test can be done by general practitioners
• CT Colography: 3D reconstruction of CT images every 5 yr
• 84-91% sensitive for cancer, 70% sensitive for advanced adenomas
• Neither sensitive nor specific for small lesions <8 mm (considered negative!)
• Prompt colonoscopy in 11-22% of cases in a 10 yr span
• Expensive test with some radiation exposure
• Cologuard: Fecal DNA plus FIT every 3 yr:
• 92% sensitive for cancer, 42% for advanced adenomas
• No utility in the detection of small adenomas
• False positives prompt colonoscopy in 45% of cases in a 10 yr span

Relative Costs:

80% colon acceptance 25% symptom generated colon/10 yr Cost of colonoscopy is denoted by X

• Colonoscopy every 10 yr: 0.8 colon (0.8X)

0.8X

• FIT every yr:

0.7 colon (0.7X) + FIT admin 0.8X + ca cost

• Cologuard every 3 yr:

0.7 colon (0.7X) + CG admin 0.8X + ca cost

• Flex Sig every 5 yr:

0.35 colon (0.35X) + 2 Flex (0.5X) 0.85X + ca cost

• CT Colon every 5yr:

0.45 colon (0.45X) + 2 CT (1X) 1.45X + ca cost

5/22/19 9

Conclusions

• Among CRC screening strategies, direct colonoscopy is a practical

screening method exhibits superior sensitivity, cancer prevention, and resource utilization compared to other strategies

• Fecal testing does not appreciably reduce colonoscopy utilization but

delays its implementation, yet misses 30% of cancers. The vast majority of positive fecal tests are false positive.

• CT colography is an expensive Tier 2 screening tool whose

performance in real world studies has not been fully evaluated.

• Flex Sig is falling out of favor but is likely better than Fecal Testing

Special Consult Considerations

• Request for colonoscopy in patient with FIT+ despite

negative screening colon 2 yr ago. No anemia or symptoms.

• Request for colonoscopy in patient with FIT+ despite

negative screening colon 2 yr ago. No anemia or symptoms.

• If the majority of positive FIT+ are false positive, nearly

all positive FIT in those with up-to-date colonoscopy are false positive.

• Recommendation: Please discontinue Fecal testing

Special Consult Considerations

• Request for colonoscopy in patient with negative

screening colon 2 yr ago because his spouse was dxed with CRC and he is worried about cancer. No anemia or symptoms.

Special Consult Considerations

5/22/19 10

• Request for colonoscopy in patient with negative

screening colon 2 yr ago because his spouse was dxed with CRC and he is worried about cancer. No anemia or symptoms.

• Recommendation: Please tell this patient not to worry
• anymore. A complication from an unindicated

colonoscopy is very difficult to defend!

Special Consult Considerations

• Request for colonoscopy due to new onset constipation or

a solitary episode of hematochezia. Patient with screening colon 2 yr ago showing no neoplasia. No anemia or other symptoms.

Special Consult Considerations

• Request for colonoscopy due to new onset constipation or

a solitary episode of hematochezia. Patient with screening colon 2 yr ago showing no neoplasia. No anemia or other symptoms.

• Most CRC sx manifest in the distal colon. Recommend

examine distal colon with Flex Sig

Special Consult Considerations