Stool DNA Testing For Colon Cancer
Steven Itzkowitz, MD, FACP, FACG, AGAF
Professor of Medicine
Mount Sinai School of Medicine
New York, NY
Tests that detect Adenomas and Cancer: (structural)
• Flexible sigmoidoscopy q 5 yrs
• Colonoscopy q 10 yrs
• Barium enema (air contrast) q 5 yrs
• CT Colonography q 5 yrs
CRC Screening Guidelines:
Average-Risk Adults Over Age 50
(ACS, US Multi-Society Task Force, ACR)
Tests that primarily detect Cancer: (stool-based)
• Fecal occult blood test (FOBT) q 1 yr
• Fecal immunochemical test (FIT) q 1 yr
• Stool DNA test (sDNA) interval uncertain
Levin et al. CA-Cancer J Clin 58:130, 2008
Why Stool DNA Tests?• Colonoscopy is becoming the preferred CRC
screening test.
• However, barriers to colonoscopy include:
– Organizational: access (USA, abroad); capacity
– Patient-associated: discomfort, fear, embarrassment, inconvenience (work absence; patient escort; child care)
• Therefore, non-invasive tests may greatly facilitate CRC screening efforts.
• DNA is theoretically a more specific analyte than blood for stool-based detection.
Rationale for Stool DNA Testing: Mucocellular Layer
Colon cancer Normal colon
Courtesy: David Ahlquist, MD, Mayo Clinic
CarcinomaEarly
adenoma
Intermediate
adenoma
Late
adenoma
APC (10)
Normal
mucosa
Molecular Markers of Colon Carcinogenesis
Chromosomal Instability (e.g. FAP)•Aneuploidy
•LOH
•Tumor suppressor gene mutations
Microsatellite Instability (e.g. HNPCC)•Hypermethylation/mutation of DNA MMR genes
•Target gene alterations (TGFbRII; others)
K-ras (3) DCC/18q genes P53 (8)
70-85%
15% Long-DNA (DIA)
BAT26
Version 1 Stool DNA Test:
Collection Kit (with freezer pacs)
Stool DNA Testing: Early Studies
Study Sensitivity Specificity
Ahlquist ‘00 91% (20/22) 93% (26/28)
Tagore ‘00 63% (33/52) 98.2% (111/113)
Syngal ‘02,‘03 62% (40/65) --
Brand ‘02 69% (11/16) --
Calistri ‘03 62% (33/53) 97% (37/38)
Syngal ‘06 63% (43/68) --
These studies:
• used the same multi-target DNA panel (Version 1)
• paved the way for a large average-risk pop’n screening study
sDNA is Better than FOBT in
Average-Risk Individuals2,507 asymptomatic, average-risk subjects over age 50
Fecal DNA assay compared to Hemoccult-II
PreGenPlus Assay:
• 22 Mutations - APC (10), K-ras (3), p53 (8), BAT-26
• DNA integrity assay (DIA)
sDNA Hemoccult-II
Cancer (n=31) 51.6 % 12.9%
Adenomas
• HGD (n=40) 32.5% 15.0%
• Villous (n=133) 18.0% 9.8%
• >1 cm (n=214) 10.7% 10.3%
Normal colon (n=1423) 5.6% 4.8%
Imperiale, Ransohoff, Itzkowitz, et al. NEJM 351:2704, 2004
(p=0.003)
Patient Preferences
(Based on Imperiale Study)
• Preferred strategy among 4,042 patients in the multicenter study (84% response rate)
• Stool DNA received the same or higher mean ratings than FOBT for prep- and test-related features.
• Stool DNA received higher ratings than colonoscopy for all prep- and test-related features except accuracy.
• Preferred test:
– Stool DNA: 45%
– FOBT: 32%
– Colonoscopy: 15%
– No preference: 8%
Schroy et al, Am J Prev Med 28:208, 2005
sDNA is Better than FOBT in
Average-Risk Individuals
Conclusions:
1. sDNA more sensitive than Hemoccult-II for CRC
2. sDNA similar specificity as Hemoccult-II
3. But, DIA performance lower than expected• DNA degraded in transit, despite use of freezer pacs
and overnight shipping.
Imperiale et al. NEJM 351:2704, 2004
Improving the Stool DNA Test:
“Version 2”
• Better DNA stabilization
– Adding EDTA-containing buffer to stool
significantly increases the recovery of DNA 1
• Improved DNA extraction method
– Gel-based extraction (instead of beads)
enhances DNA recovery 2
• New markers
– Methylation markers (eg. vimentin) 3
1 Olson et al, Diagn Mol Pathol 14:183, 20052 Whitney et al. J Mol Diagn 6:386, 20043 Chen et al. J Natl Cancer Inst 97:1124, 2005
Carcinoma
(MSS)
Early
adenoma
Intermediate
adenoma
Late
adenoma
APC
Normal
mucosa
PATHWAYS OF COLON CARCINOGENESIS
Chromosomal Instability (e.g. FAP)•Aneuploid; LOH; Tumor suppressor gene mutations
Microsatellite Instability (e.g. HNPCC)•Mutation/loss of DNA MMR genes; diploid
•Mutations of key target genes (eg, TGFbRII)
K-ras DCC/18q genes p5370-85%
15%
CpG Island Methylation; CIMP (e.g. HPS)•DNA methylation inhibits key gene expression
•BRAF oncogene mutation
Sessile Serrated Polyp
(SSP; SSA)
Carcinoma
(MSI-H)
Carcinoma
(MSI)
15%
New Stool Collection Kit
(with buffer)
Results of Version 1 Assay
(MuMu22+DIA)
Version 1(Imperiale, NEJM, „04)
Version 1.1*(with buffer and gel capture)
No. Positive
%Positive
No. Positive
% Positive
Sensitivity:
•All markers 16/31 51.6% 29/40 72.5%
•MuMu22 16/31 51.6% 17/40 42.5%
•DIA 1/31 3.2% 26/40 65.0%****(p<0.0001)
Analyzing the original Version 1 markers, the DNA
stabilizing buffer & gel capture increased sensitivity for CRC
(51.6% -> 72.5%), especially DIA (3.2% -> 65%)
* Itzkowitz et al. Clin Gastroenterol Hepatol 2007, 5:111
Version 2: Two Markers
Sensitivity (n=40) Specificity (n=122)
No.
Positive
%
(95% C.I.)
No.
Positive
%
(95% C.I.)
DY (DIA) 26 65.0 (49.5-77.9) 9 92.6 (86.6-96.1)
Vimentin 29 72.5 (57.2-83.9) 16 86.9 (79.8-91.8)
Vim + DY 35 87.5 (73.9-94.5) 22 82.0 (74.2-87.8)
Vimentin methylation + DY resulted in optimal sensitivity
(87.5%) & specificity (82.0%)
Itzkowitz et al. Clin Gastroenterol Hepatol, 2007, 5:111
Sensitivity of Version 2:
by Cancer Stage
No.
Positive
%
(95% CI)
Total: 35/40 87.5 (73.9-94.5)
• Stage I 6/8 75.0 (40.9-92.8)
• Stage II 9/10 90.0 (59.6-98.2)
• Stage III 16/17 94.1 (73.0-99.0)
• Stage IV 4/5 80.0 (37.6-96.4)
• DY+Vim detected the vast majority of CRC regardless of
tumor stage
Itzkowitz et al. Clin Gastroenterol Hepatol, 2007, 5:111
Version 2 Detects CRC
Regardless of Location
PV1 DY Vim DY + Vim
Right (n=11) 54.5% 36.4% 72.7% 90.9%
Left (n=29) 79.3% 75.9% 72.4% 86.2%
P value NS 0.03 NS NS
• DY preferentially detected distal CRC
• Vim detected CRC regardless of location
• Therefore, DY+Vim detected the majority of CRC’s
regardless of location
Itzkowitz et al. Clin Gastroenterol Hepatol 5:111,2007
Version 2:
Patient Satisfaction Survey
Percent
Male 41%
Age >60 yrs 40%
Perform the test; easy/very easy 97%
Open the preservative bottle; easy/very easy 96%
Add the preservative to specimen; easy/very easy 100%
Very comfortable performing the test 93%
Would repeat test if doctor recommended it 84%
Itzkowitz et al. Clin Gastroenterol Hepatol 5:111, 2007
Stool DNA Test - Version 2
CRC NL Sensitivity Specificity
Phase 1a 40 122 88% (74-95) 82% (74-88)
Phase 1b 42 241 86% (72-93) 73% (67-78)
TOTAL 82 363 83% (73-90) 82% (77-85)
Note: 6/7 (86%) adenomas with HGD/CIS were also positive
Itzkowitz et al, Am J Gastroenterol 103:2862, 2008
2nd MultiCenter sDNA Study
sDNA
Positive
(%)
Hemoccult-II
Positive (%)
HOSensa
Positive (%)
P value
SDT-1 20 (14-26) 11 (6-16) 21 (15-27) NS
SDT-2 46 (38-55) 16 (10-22) 24 (17-31) <0.001
Cancer 58 (36-80) 47 (25-70) 63 (41-85) NS
Adenoma >1 cm 46 (35-54) 10 (4-15) 17 (9-24) <0.001
Normal 16 (8-24) 4 (1-11) 5 (1-13) 0.03
• 3,764 asymptomatic, average-risk subjects over age 50; 22 centers
• Stool DNA assay compared to Hemoccult-II & HemoccultSensa
Stool DNA test:
• SDT-1: MuMu22+DIA
• SDT-2: K-ras, APC scan, methyl-vimentin (better adenoma markers)
Ahlquist et al. Ann Intern Med 149:441, 2008
Stool DNA Test Sensitivity for
Screen-Relevant Neoplasia (n=142)
0
10
20
30
40
50
60
70
1 2 3 1 2 3 1Stool n
Hemoccult HemoccultSensa Stool DNA
* P<0.0001 vs Hemoccult or HemoccultSensa *
%
Ahlquist et al, Ann Int Med 149:441, 2008
New Stool DNA Methylation Markers
Marker Sensitivity Specificity
Cancer Adenoma
SFRP2 63-94% 12-62% 77-100%
SFRP1 84% 100% 86%
NDRG4 53-61% -- 93-100%
TFPI2 76% 21% 79-93%
New Stool DNA Assay:
Digital Melt Curve Assay
DMC Exact V. 1.1 Hemoccult-II Hemoccult-Sensa
Sensitivity (AAP)* 59% 26% 7% 15%
Specificity 92% 100% 92% 92%
• Analyzed 27 advanced adenomas with k-ras mutation
Zou et al. Gastroenterology 136:459, 2009
• Adenomas >2 cm: 8/10 (80%)
• Adenomas with HGD: 5/5 (100%)
Stool DNA: Cost Effectiveness
With Perfect Adherence Reduction in
CRC Incidence
Reduction in
CRC Mortality
No screening -- --
FOBT 49% 66%
sDNA test (V 2.0) q 3 yrs 43% 63%
FIT 66% 78%
Colonoscopy 73% 80%
• FOBT ($15), FIT ($22), Stool DNA ($300), C’scopy ($920)
• FIT dominated other stool tests.
• sDNA V2 (with 100% adherence) more effective when per-
cycle FIT adherence fell below 50%
Parekh et al. Aliment Pharmacol Ther 27:697, 2008
Conclusions
• Newer stool DNA tests:
• Are much less complex
• Are less expensive
• Can theoretically be run by local laboratories
• Are showing promise for detecting important
adenomas
• The future:
• newer assays/markers under development
• reducing cost