Title: Recent Findings on Test Performance
1Recent Findings on Test Performance
- Brian P. Mulhall, MD MPH
- Walter Reed Army Medical Center
2Colorectal Cancer ScreeningViable Tests
- FOBT
- Rehydrated and unrehydrated
- Immunochemical FOBT
- Barium Enema
- Flexible Sigmoidoscopy
- Colonoscopy
- Virtual Colonoscopy
- Stool DNA
The Past
The Present.
The Future?
3Fecal Occult Blood Tests
4Fecal Occult Blood Tests
- FOBTs
- GUAIAC-based
- Hemoccult II
- Hemoccult SENSA
- Fecatwin S
- Coloscreen III/VPI
- Hemoccult Wipe
- Immunochemical tests
- FECA-EIA
- Monohaem
- HemeSelect
- HemoQuant
n81
Gopalswamy, N., et al. Arch Fam Med 3 1043,
1994.
5Test PerformanceFOBTs
- GUAIAC-based
- Hemoccult II
- Hemoccult SENSA
- Fecatwin S
- Coloscreen III/VPI
- Hemoccult Wipe
- Hemofec
- Hemopreuve
- Unrehydrated/Rehydrated
6Test PerformanceImmunochemical FOBTs
- IFOBTs
- FECA-EIA
- Hemolex
- Monohaem
- HemeSelect
- HemoQuant
- Hemoblot
- BM-test Colon
- Immudia-HemSp
- OC-hemodia
- Immunohemostick
- RPHA
- Fecal calprotectin
- Magstream 1000
- FlexSure OBT
7FOBT vs. IFOBT
- Large VA-based population study
- Varied risk for CRC, 98 male
- IFOBT (FlexSure OBT) vs. FOBT (HemoccultSENSA)
- 2965 tests ordered each 1410/1369 returned
- 48 return rate
- 66 positive results referred for colonoscopy
(50 completed)
- Test Performance
- PPV Adenoma PPV Adenoma 1 cm or Cancer
- IFOBT ? 58 17
- FOBT ? 59 30
Ko, C.W., et al. FOBT in a General Medicine
Clinic Comparison of GUAIAC and IFOBT. Am J Med,
115 111-114, 2003.
8Barium Enema
9Test PerformanceBarium Enema
- Infrequently used in practice
- Limited data on impact on outcomes
- Varied studies
- Symptomatic or high-risk populations (Sensitivity
? 85-90)
- Work-up bias
- Retrospective cohort studies (Sensitivity ?
70-96)
- Biased
- Surveillance populations (Sensitivity ? 30-81
NPS Sens 50 polyps 1 cm)
- Selection bias
- Tagged polyp study (specificity ? 96-98 for
polyps 7-10 mm or CA)
- Data for hyperplastic versus adenomatous polyps
not reported
10Endoscopy
11Test PerformanceFlexible Sigmoidoscopy
- An accepted standard for CRC screening
- Permits detection and removal of distal polyps
- Insertion beyond the sigmoid in
- Proximal lesions may be missed
- In average-risk patients with distal adenomas
- Half had lesions beyond the reach of the
sigmoidoscope
- Autopsy studies show 50-70 adenomas are
proximal
- Overall sensitivity ? 60-70
- Sensitivity for distal lesions ? 85-95
- Specificity ? 85-98
12Test PerformanceColonoscopy
- The Gold Standard for diagnosing polyps and
CRC
- Visualizes the entire colon
- Examination completed ? 80-98
- Allows identification, removal of culprit
lesions
- Prevents 60-80 colorectal cancers
- Sensitivity
- Cancer ? 79-100
- Polyps ? 73-100 (85 polyps 5 mm 92 7 mm)
- Miss rate for polyps 8 mm ? 0-26 Polyps mm ? 15-30
- Specificity
- Assumed to be 100
- Likely in the range of 90-95
13Stool DNA
14(No Transcript)
15Test PerformanceStool DNA
- About 50 of CRCs have K-ras mutations
- False positives in IBD and lymphocytic colitis
- New multi-target assays have been developed
- Uses K-ras, APC, p53, BAT-26, L-DNA/DIA
- Improved test performance over single mutations
- Sensitivity in high-risk groups
- Adenomas ? 47-82
- Cancers ? 64-91
- Screening populations (n2,507)
- Sensitivity for advanced adenomas ? 15
- Specificity in patients with polyps ? 95
16Test PerformanceVirtual Colonoscopy
- Evolving technology
- Wide range of results
- 2-D vs. 3-D
- Software and hardware
- Radiologist experience
- Prevalence of population
- Sensitivity for polyps
- 10 mm ? 32-96
- 6-9 mm ? 41-94
- Specificity for polyps
- 10 mm ? 74-98
- 6-9 mm ? 63-95
17Virtual Colonoscopy
18Virtual Colonoscopy StudiesAll with n 100
19V.C. Study Results Per Patient Analysis
(1) 1 reader (2) 2 readers (3) 3 readers
20The Pickhardt Study -Methods
- Multi-center Enrollment
- Adults ages 50-79
- Adults with FH CRC ages 40-79
- Preparation
- Fleet PhosphoSoda Bisacodyl
- 500 ml barium with clear diet
- Solid-stool tagging
- 120 mL of diatrozoate meglumine and diatrozoate
sodium
- Liquid opacification
21The Pickhardt Study Diagnostic Methods
- Optical Colonoscopy
- Standard videoendoscope
- 17 experienced endoscopists
- 14 GI, 3 GS
- Polyps measured by calibrated linear probe
- Segmental unblinding
- Polyps resected
- Sent for pathology
- Procedural times recorded
- Questionnaires provided
- Virtual Colonoscopy
- Rectal catheter
- Patient-controlled insufflation
- Breath-holding
- Supine and prone
- GE Lightspeed or Ultra CT
- Reconstruction interval 1 mm
- Viatronix 3D Colon 1.2
- Virtual fly-through
- Correlation with 2-D images
- Polyps measured by electronic calipers
- Extra-colonic findings recorded
- Analyzed by experienced radiologists
22The Pickhardt Study -Statistical Analysis
- OC Reference Standard
- Primary outcome
- Polyps 6 mm
- Advanced neoplasia
- Polyps 10 mm
- HGD, TV, or cancer
- Polyp Matching VC/OC
- True positive
- Same/adjacent segment
- Diameters equal /- 50
- True positive by size
- Polyp of size on VC OC
- Analysis using
- McNemars test
- Fishers exact test
- Chi-square tests
- Paired t-tests
- Interobserver reliability
- 100 random VCs
- 2nd read by blinded radiologist
- Kappa defined
23Pickhardt Results
- 1233/1253 patients completed VC/OC
- 1201 average risk
- Adenomas
- 10 mm? 3.9
- 8 mm ? 6.7
- 6 mm ? 13.6
- Malignant ? 0.4
- Advanced polyps
24Pickhardt Results Test Performance
- Polyps on VC not OC
- 55 polyps
- 21 TAs 6 mm
- Advanced Neoplasms
- Sensitivity
- VC ? 91.5
- OC ? 88.1
- Colon Cancer
- Sensitivity
- VC ? 100 (2/2)
- OC ? 50 (1/2)
25Pickhardt Results
- Extra-colonic findings (rates are half of
previous reports)
- Potentially high clinical importance ? 4.5
- 7/56 had extra-colonic cancer or AAA (12.5)
- Potentially moderate clinical importance ? 13.5
- Nephrolithiasis (7.9) or cholelithiasis (5.6)
- Procedural time
- VC ? 14.1 minutes (in CT suite)
- OC ? 31.5 minutes (64.4 additional min.s in
recovery)
- Satisfaction (81.5 questionnaires returned)
- Greater discomfort VC 54.3 vs. OC 38.1
- More acceptable(convenience) VC 68.3 vs. OC
24.1
26Pickhardt ResultsSummary Statistics
- VC for polyps 8 mm
- Sensitivity Specificity 90
- 1 in 6 patients will get OC
- Kappa 0.75
- VC OC for CT/radiologist time
- Quicker for patients
- VC less comfortable than OC
- But more acceptable
- Bottom-line
- VC better than OC for detection of adenomas
27Pickhardt Study Questions/Issues
- What is the gold standard in this study?
- OC was the reference standard
- But sensitivity defined for VC and OC. Perhaps
this is reasonable, but
- Proceduralists ? Staff radiologists vs.
Staff/Fellows GI/Surgeons
- What about data for hyperplastic polyps?
- Specificity of VC for any polyps
- Polyps 10 mm ? 97.4
- Polyps 8 mm ? 95.0
- Polyps 6 mm ? 84.5
- What about small advanced polyps?
- Only 0.1 polyps
- But what about with 6 or 7 mm polyps?
- Test positive rate at 8 mm is 13.5 (1 in 6 get
colonoscopy)
- At 6 mm, 1 in every 3 patients get colonoscopy
(where spec 80)
VC HPs 5-8 mm would get VC q 2-3 years
All OC HPs would get next OC at 10 years
28Test PerformanceOther Studies
- FDG-PET
- Dark lumen MR colography
- Chromoendoscopy
29Test PerformanceThe Bottom-Line
30Test PerformanceThe Bottom-Line
40 100
50 95
80 100
31Summary Comments
- No test has been established as the Gold
Standard
- Tests
- FOBTs and IFOBTs
- Wide range of sensitivity and specificity
- Lower sensitivity, but has demonstrated benefit
- Barium Enema
- Has acceptable sensitivity and specificity
- Suffers from limited usage, inadequate study,
inconvenience
- Sigmoidoscopy
- Has acceptable sensitivity and specificity
- Limited exam of colon, usage/training suffers
colonoscopys role
32Summary Comments
- Tests
- Colonoscopy
- Has high sensitivity and specificity (?)
- Miss rate, incomplete examinations and
complications
- Virtual Colonoscopy
- Has a range of sensitivity and specificity (high
specificity)
- Hard-/Software, inter-observer reliability,
inability for intervention
- Stool DNA
- Has a range of sensitivity, specificity
- Limited number of studies, assay dependent, lab
dependent
33BIBLIOGRAPHY
- See References in Notes View