1. Microsatellite instability (MSI) in stage II and III colon cancer treated with 5FU-LV or 5FU-LV and irinotecan (PETACC 3-EORTC SAKK 60/00 trial).
S. Tejpar, F. Bosman, M. Delorenzi, R. Fiocca, P. Yan, D. Klingbiel, D. Dietrich, E. Van Cutsem, R. Labianca, A. Roth
Abst. ID: 4001

2. Disclosures I receive research funding from
I have been a paid consultant to
Pfizer
I receive research funding from
Disclosures ok, right format?

3. Primary endpoint 5-year DFS rate of stage III (1)
Trial: Randomized Phase III Comparing Biweekly Infusional Fluorouracil/Leucovorin Alone or With Irinotecan in the Adjuvant Treatment of Stage II-III Colon Cancer: PETACC 3-EORTC SAKK 60/00
Primary endpoint 5-year DFS rate of stage III (1)
56.7% with Iri/LV5FU2
54.3% with LV5FU2
log-rank p=
3278 patients randomized, stage II and III
pathological material of 1564 patients collected
Large TR program: alterations in p53, SMAD4, thymidylate synthetase, TERT, KRAS, BRAF, MSI and 18qLOH
Outcomes used for the TR study:
RFS at 3 and 5 years, OS median 68 mo follow up .
Were the stage II high risk stage II, why were they included?
(1) J Clin Oncol May 18

4. Microsatellite instability (MSI) is the mutational signature found in colorectal cancers (CRCs) that evolve as a result of inactivation of the DNA mismatch repair (MMR) system.
Prognostic:
Multiple retrospective studies and a meta-analysis(1) have shown that MSI H tumors have a more favorable stage-adjusted prognosis compared to MSS (2).
(1) combined hazard ratio for overall survival: 0.65 (95% CI, 0.59 to 0.71).
J Clin Oncol 23:609–618, 2005
Current Opinion in Oncology 2009, 21

5. Predictive Irinotecan
Predictive 5FU
+ Elsaleh H et al (1)
Benefit of 5FU for MSI-H
- Ribic CM et al (2) :
retrospective, 570 stage II + III, W/0 5FU
Benefit of 5FU only in MSS
+/- Kim GP et al (3):
retrospective, 542 stage II + III, W/O 5FU
neutral.
-Sargent DA et al (4).
retrospective 1000 stage II+III, W/O 5FU
Predictive Irinotecan
+Bertagnolli et al (5)
702 stage III patients, 5FU/IFL
Benefit of Irinotecan in MSI-H
Ribic CM et al (2) : retrospective analysis, 570 stage II and III patients, untreated and treated 5FU.
16% MSI H
All patients: 5Y DFS according to MSI: yes (p=0.04)
All patients : 5 Y OS according to MSI : no (p=0.07)
No stage effect
In multivariate models for OS adjusted for the stage and grade
All patients: 0.61 [95 percent confidence interval, 0.38 to 0.96]; P=0.03)
In treated???
In untreated: 0.32 [95 percent confidence interval, 0.14 to 0.75]; P=0.008)
5FU treatment improved OS only in MSS , controlled for stage and grade, test for interaction
Kim GP et al (3): retrospective analysis NASPB trials, 542 stage II and III patients, untreated and treated 5FU.
18% MSI H
All patients: RFS according to MSI: trend (p=0.1)
All patients : OS according to MSI : no (p=0.67)
No stage correction done, No association of MSI with stage
failed to identify an interaction between MSI status and FULV therapy (but anyhow no effect of Msi on RFS or OS, power?)
Bertagnolli et al (5)
702 stage III patients, 5FU, IFL
13% MSI H
All cases , OS by MSI: no (HR: 0.86 (95% CI, 0.57 to 1.29; log-rank P = .44)
All cases, DFS by MSI: no (HR was 0.61 (95% CI, 0.57 to 0.64; log-rank P= .15)
Per treatment arm 5YDFS:
in IFL patients: 0.57; 95% CI, 0.42 to (p=0.03)
In 5FU patients: 0.76; 95% CI, 0.64 to 0.88 (p=0.8)
hazard ratio interaction between tumor status and treatment, 0.51; likelihood ratio P = .117
So also significant 5YDFS seen in IFL patients, contrary to sargent, not in 5FU. Specific interaction with iri?
Lancet 355: , (2) N Engl J Med 349:247–257, 2003.
(3) J Clin Oncol 25:767–772, (4) abstract #4008 ASCO 2008, J Clin Oncol 2008; 26 (15).
(5) J Clin Oncol.27: , 2009 5

6. Questions we addressed
Incidence of MSI-H in stage II (n= 395) and III (n= 859)
Prognostic effect of MSI-H?
All our patients treated with at least 5FU (w/o irinotecan)
Stage specific prognostic effects of MSI-H?
Is MSI-H predictive for irinotecan ? 6

7. Methods MSI
MSI was analyzed at 10 different microsatellite loci containing mono- or di-nucleotide repeated sequences. The panel consisted of the five markers from the Bethesda reference panel, with the addition of five markers which were also suggested during the International Workshop on HNPCC in (BAT-25, BAT-26, D2S123, D5S346, TGFBR2, BAT-40, D17S787, D18S69, D17S250 and D18S58)
The amplified PCR products were analyzed using the automated ABI Prism Sequencer Model 3100 Genetic Analyzer (Applied Biosystems, Foster City, USA). A locus was called unstable if unequivocal instabilities were seen in the tumour sample in comparison to the paired normal DNA of the same patient.
The analysis was successful in 1327/1405 assessable FFPE blocks
MSI was graded as high (MSI-H) when 30% or more of markers were unstable, low when 10-20% of markers were unstable and stable when all markers were negative. For the statistical analysis MSI-L and MSS populations were pooled and identified as MSS 7

8. Statistical Methods
Survival curves (and rates) were determined using the Kaplan-Meier estimator and compared using the log-rank test.
Frequencies were compared using chi-squared contingency table tests and chi-squared goodness-of-fit tests.
Hazard Ratios (with 95% Confidence Intervals plus p-values from the Wald-Test) were computed with (multivariate) proportional hazard models (Cox regression), and adjusted for stage where indicated.

9. Results: Frequency 22% (86/395) 12% (104/859) <0.0001
Frequency Analysis
Stage II
Stage III
P value
MSI-H
22%
(86/395)
12%
(104/859)
<0.0001
in accordance with recent literature:
J Clin Oncol.27: , 2009 : 13.2% in 700 stage III
#4008 ASCO 2008, J Clin Oncol 2008; 26 (15):12-19% in 1020 pooled stage II+III

10. MSI as a suppressor of lymph node and distant metastasis?
Frequency Analysis
Stage II
Stage III
Stage IV
MSI-H
22%
(86/395)
12%
(104/859)
3.5% *
*Br J Cancer (2):

11. Results: Univariate Associations- Clinical
Marker
Frequency Analysis
MSI-H
MSS
P value
Stage (II)
85/188 (45.2%)
306/1045 (29.3%)
<0.0001
N stage (N0 vs N1-2)
Higher proportion of MSI in N0 tumors
T stage (T1/2 vs T3 vs T4)
The higher the T stage the higher proportion of MSI
0.037
Grade ( 1-2)
137/188 (72.9%)
942/1045 (94%)
Site (Right vs Left)
143/188 (76%)
344/1045 (33%)
Age (≤ 60 )
125/188 (66.5%)
486/1045 (46.5%)
Sex (female)
80/188 (42.6%)
437/1045 (41.8%)
0.91
For comparison:Bertagnolli:
Relationships Between Tumor MMR Status and Clinicopathologic Factors Overall median follow-up was 6.65 years.. Patients with MMR-D/MSI-H tumors had significantly more proximal tumors ( 2 P < .0001), poorly/undifferentiated tumors ( 2 P < .0001), and tumors containing extracellular mucin (>50%; 2 P < .0001). The nodal ratio (number of positive nodes divided by number of nodes sampled) was significantly lower among patients with MMR-D/MSI-H tumors (Satterthwaite t test P = .005), and women had more MMR-D/MSI-H tumors ( 2 P = .03). There were no significant differences between MMR-D and MMR-I tumors by age, treatment, or number of positive lymph nodes
Also compare to Kim et al (Allegra) 11

12. Results: Univariate Associations- Molecular
Marker
Frequency Analysis
MSI H
MSS
P value
Braf mutation
45/186 (24%)
53/1037 (5%)
<0.0001
P53 intense (>45% of cells)
15/188 (.8%)
409/1045 (39%)
18q LOH
20/110 (18%)
515/675 (76%)
TS intense (75% of cells staining intensely)
121/181 (67%)
253/958 (26%)
Smad4 complete loss
11/188 (0.6%)
249/1045 (24%)
kras
55/186 (30%)
389/1028 (38%)
0.038
Telomerase >15%
50/105 (48%)
270/603 (45%)
0.664
refer to:
-MSI-high tumours were more likely to originate in the proximal colon and possess BRAF mutations and CIMP-high
status, and less likely to present with stage III or IV disease. (Ogino S et al:Gut 2009;58:90–96.)
Kim et an and Westra et al for P53 12

13. Results: Prognostic impact MSI - Univariate
HR MSI-H (95% CI)
5FU
(n= 625)
5FU/iri
(n= 608)
Both arms
Stage II
(n= 391)
RFS
( )
P= 0.043
0.296 ( )
P=0.044
0.265( ) P= OS
0.18 ( )
P= 0.095
0.143 ( )
P=0.057
0.159( )
P=0.011
Stage III
(n= 842)
0.596 ( )
P=0.064
0.815 ( )
P=0.45
0.693 ( )
P= 0.06
0.515 ( )
P=0.055
0.939 ( )
P=0.84
0.699 ( )
P= 0.12
Both stages
* P are stage corrected
0.501 ( )
P=0.0083
0.642 ( )
P=0.076
( )
P=0.0018
0.437( )
P= 0.012
0.676 ( )
P= 0.18
( )
P=0.006
RFS | SII | SIII
MSI-H | 5(85) | 29(103)
MSS | 61(306) | 285(739)
OS | SII | SIII
MSI-H | 2(85) | 21(103)
MSS | 42(306) | 211(739)

14. Results: Prognostic impact MSI - Univariate
HR MSI H (95% CI)
5FU
(n= 625)
5FU/iri
(n= 608)
Both arms
Stage II
(n= 391)
RFS
( )
P= 0.043
0.296 ( )
P=0.044
0.265( ) P= OS
0.18 ( )
P= 0.095
0.143 ( )
P=0.057
0.159( )
P=0.011
Stage III
(n= 842)
0.596 ( )
P=0.064
0.815 ( )
P=0.45
0.693 ( )
P= 0.06
0.515 ( )
P=0.055
0.939 ( )
P=0.84
0.699 ( )
P= 0.12
Both stages
* P are stage corrected
0.501 ( )
P=0.0083
0.642 ( )
P=0.076
( )
P=0.0018
0.437( )
P= 0.012
0.676 ( )
P= 0.18
( )
P=0.006
Conclusion: Effect lost in Stage III, only in stage II.
Number of evenst stage II versus III
In stage III the effect is lost , maybe due to both bad outcome of the MSIH irinotecan treated arm (1.15 HR)and improvement of the MSS (O.85 HR) .
In multivariate no effect at all in stage III for RFS, some for OS
Note also RFS and OS, both stages for arm B(5FU), quite the contrary of D Sargent 14

15. Take home message prognostic effects
When considering stage II and III combined there is a strong prognostic effect of MSI-H both on RFS and OS.
When analyzed per stage
the effect is very strong in stage II
RFS (P= ) OS (P=0.011)
but there is only a trend in stage III
RFS (P= 0.06) OS (P= 0.12)
A regression containing Stage, MSI H and their interaction shows a bordeline significant interaction both for RFS (p=0.058) and OS (p=0.051)

16. Questions we addressed
Incidence of MSI in stage II (n= 395) and III (n= 859)
Prognostic effect of MSI-H?
All our patients treated with at least 5FU (w/o irinotecan)
Stage specific prognostic effects of MSI-H?
Is MSI-H predictive for irinotecan ? 16

17. abstract #4008 ASCO 2008, JCO 2008; 26 (Suppl 15) Pooled Data stage II + III patients : 5y DFS By MSI Status
Treated 5FU (N=512)
Untreated (N=515)
Sargent Asco 2008: We believe that the strength of the available evidence is sufficient at the present time to recommend that MSI-H stage II colon cancer patients not receive adjuvant chemotherapy given that such tumors have a favorable prognosis and exhibit resistance to 5-FU
Msi status validated as a prognostic marker in untreated patients, No suggestion of benefit from 5-FU based treatment in dMMR patients
MSI-H
MSS
HR (95% CI)
P value
5Y DFS rate
70%
67%
0.79
( )
p=0.30
MSI-H
MSS
HR (95% CI)
P value
5Y DFS rate
80%
56%
0.51
( )
p=0.009 17

18. compare ASCO 2008 with our data
Data stage II + III patients : 5y DFS By MSI Status
Treated 5FU (N=512)
Sargent
Treated 5FU (N=625)
Petacc3
Sargent Asco 2008: We believe that the strength of the available evidence is sufficient at the present time to recommend that MSI-H stage II colon cancer patients not receive adjuvant chemotherapy given that such tumors have a favorable prognosis and exhibit resistance to 5-FU
Msi status validated as a prognostic marker in untreated patients
No suggestion of benefit from 5-FU based treatment in dMMR patients
MSI-H
MSS
HR (95% CI)
P value
5Y DFS rate
70%
67%
0.79
( )
p=0.30
MSI-H
MSS
HR (95% CI)
P value
5Y RFS rate
83%
66%
0.50 ( )
p=0.0077 18

19. Predictive effects of MSI-H
All patients treated with 5FU (w/o irinotecan)
Study predictive effects for irinotecan
Data shown limited to stage III for comparison to (1), not different for stage II
(1) J Clin Oncol.27: , 2009 19

20. Results: Predictive effects of MSI
Population Stage III
rate of 5-Year RFS
95% CI N
All patients analyzed
62%
( )
N=842
Prognostic analysis
All MSI-H tumors
72%
( )
N= 103
 All MSS tumors
59%
( )
N= 739
Predictive analysis within treatment arm
FU/LV /Irinotecan
All MSI-H tumors
69%
( )
N= 49
All MSS tumors
64%
( )
N=372
Predictive analysis within tumor subtype
FU/LV
N= 54
Dirk, add p values for blue lines?. Were not in attach 20

21. Purple and red (1) J Clin Oncol.27:1814-21, 2009
Population Stage III
Rate of 5-Year RFS
Rate of 5 -Year DFS (1)
95% CI N
All patients analyzed
62%
61%
( )
( )
N=842
N= 702
Prognostic analysis
All patients with MSI-H tumors
72%
67%
( )
( )
N= 103
N= 96
 All patients with MSS tumors
59%
60%
( )
( )
N= 739
N= 606
Predictive analysis within treatment arm
FU/LV /Irinotecan
69%
76%
( )
( )
N= 49
N= 50
All patients with MSS tumors
64%
( )
( )
N=372
N= 304
Predictive analysis within tumor subtype
FU/LV
57%
(0.42_0.71)
N= 54
N= 46
P=0.03
P=0.07
Purple and red (1) J Clin Oncol.27: , 2009 21

22. Take home message treatment effects
In contrast to previous reports
(a) in stage II+III and in stage II alone the prognostic effect of MSI on RFS and OS remained significant even in pts treated with 5FU
(b) There is no evidence for an effect of the addition of irinotecan.
Read monica, did she find an effect in stage III?

23. The prognostic effect is weaker in stage III
Conclusions
Microsatellite instability is a strong prognostic factor for RFS and OS in stage II
The prognostic effect is weaker in stage III
conclusions are limited by sample size, the unbalance of MSI in the stages and multiple testing.
Taken together this may suggest stage specific biological effects of MSI.

24. retained for RFS and OS despite treatment with 5FU
Conclusions(2)
Prognostic effect
retained for RFS and OS despite treatment with 5FU
Predictive:
no evidence for an effect of the addition of irinotecan in MSI-H
Discordances with previous studies
Biomarker studies may be confounded small size studies, retrospective studies, pooled analysis
Biology: Confounded by molecular heterogeneity 24