1. Molecular mechanisms of resistance to anti EGFR based therapies in colorectal cancer
Alberto Bardelli
Institute for Cancer Research and Treatment
University of Torino - Medical School

2. DISCLOSURES Founder: Horizon Discovery (Cambridge, UK)
Consultant: Merck-Serono, Amgen

3. Mutations and the cancer genome
Mutations and resistance to therapies in CRCs
Parallel clinical trials in cells, mice and patients

4. Bert Vogelstein (1988) NEJM 1988; 319:525-532.
“Cancer is, in essence, a genetic disease. Although cancer is complex, and environmental and other nongenetic factors clearly play a role in many stages of the neoplastic process, the tremendous progress made in understanding tumorigenesis in large part is owing to the discovery of the genes, that when mutated, lead to cancer.”
Bert Vogelstein (1988)
NEJM 1988; 319:

5. Cancer: a genetic disease

6. DNA IS DIGITAL
Tumour
Normal
Mutation

7. Tyrosine kinome mutations
Residue is evolutionarily conserved
Mutations of equivalent residues in other kinases are pathogenic
Bardelli et. Al., Science: 300;949 (2003)

8. Mutational lansdscapes of cancer genomes
PIK3CA
TP53
PIK3CA
TP53
APC
KRAS
Wood et al., Science : 318 (2007)

9. The genetic bases of response and resistance to EGFR therapies
BRAF
PIK3CA

10. Parallel clinical trials in cells, mice and patients
Drug Y
Mutation X

11. EGFR-targeted therapies in CRCsTK
Inhibitors 3
Anti-ligand-
blocking
Antibodies 2
Ligand–
toxin
Conjugates 4
Antibody–
toxin
Conjugates 5
Anti-HER1/EGFR- blocking antibodies 1
Noonberg SB, Benz CC. Drugs 2000;59:753–67

12. Who will benefit from treatment with antibodies targeting EGFR in mCRCs ?
Responders (15-20%)
Non-Responders
Bardelli and Siena, J Clin Oncol 2010

13. Cetuximab Panitumumab
EGFR Mutations
EGFR Gene Copy Number
EGFR Protein expression (IHC)
EGFR
GSK
S6K
AKT
PDK
p85
PI3K
MAPK
MEK
Ras
Raf
Ras
Ras
Ras
Ras
Ras
Ras
Ras
Ras
Ras
Ras
Ras
SOS
Raf
Raf
Raf
Raf
Raf
Raf
Raf
Raf
Raf
Raf
Raf
Grb2
PI3K
PI3K
PI3K
PI3K
PI3K
PI3K
PI3K
PI3K
Shc
MEK
MEK
MEK
MEK
MEK
MEK
MEK
MEK
MEK
MEK
p85
p85
p85
p85
p85
p85
p85
PTEN
MAPK
MAPK
MAPK
MAPK
MAPK
MAPK
MAPK
MAPK
MAPK
DUSPs
PDK
PDK
PDK
PDK
PDK
GSK
GSK
AKT
AKT
AKT
AKT
S6K
S6K
S6K
Moroni et al Lancet Oncology 2005

14. mCRC patients treated with panitumumab or cetuximab, N=114
*P<0.05 (P=.011)
Mutated KRAS
34/113 (30%)
Wild-Type KRAS
79/113 (70%)
Responders
2/34 (6%)*
22/79 (28%)*
Non Responders
32/34 (94%)*
57/79 (72%)*
BRAF mutational status on
Wild-Type KRAS tumors (N=79)
**P<0.05 (P=.029)
Mutated BRAF
11/79 (14%)
Wild-Type BRAF
68/79 (86%)
Responders
0/11 (0%)**
22/68 (32%)**
Non Responders
11/11 (100%)**
46/68 (68%)**
Benvenuti et al., Cancer Research. 2007
Di Nicolantonio et al., J Clin Oncol. 2008 14

15. KRAS-NRAS mutated (35-45%) BRAF/PIK3CA mutated
Responder (15%)
KRAS/PIK3CA mutated
KRAS-NRAS mutated (35-45%)
BRAF/PIK3CA mutated
BRAF mutated (8%)
20-25% ???
PIK3CA mutated and/or PTEN loss (15-20%)
Bardelli and Siena, J Clin Oncol 2010

16. Sartore-Bianchi A et al., PLOS One 21010

17. Siena; Di Nicolantonio and Bardelli JNCI 2009

18. KRAS, NRAS, or BRAF mutations are non overlapping, while PIK3CA mutations may occur concomitantly with any of the above
Janakiraman M et al., Cancer Res; 70(14) July 15, 2010

19. From gene targeted therapies to mutant targeted therapies
Example 1: PIK3CA mutations
Example 2: KRAS mutations

20. PIK3CA mutations and resistance to anti EGFR MoAbs ?
Sartore-Bianchi A et al., Cancer Res YES
Prenen et al., Clin Cancer Res NO

21. Different role for individual PIK3CA mutations on the response to EGFR MoAbs in mCRCs
Zhao and Vogt PNAS 2008

22. Sample characteristics
Effects of KRAS, BRAF, NRAS and PIK3CA mutations on the efficacy of cetuximab plus chemotherapy in chemotherapy-refractory metastatic colorectal
Sample characteristics
Total number of samples successfully assessed
969/1000 (97%)
Type of tissue sample
Primary tumor
790/969 (81.5%)
Metastasis
118/969 (12.2%)
Missing
61/969 (6.3%)
Total number of chemotherapy-refractory tumors
717/969 (74%)
Treatment type in chemotherapy-refractory tumors
Panitumumab monotherapy
16/717 (2.2%)
Cetuximab monotherapy
43/717 (6%)
Cetuximab + chemotherapy
658/717 (91.8%)
De Roock et al., EU Consortium Lancet Oncology, 2010

23. Multivariate Cox regression analysis of overall survival in the unselected and KRAS wild-type population
Unselected population
KRAS wild-type population
Genotype
Adjusted hazard ratio OS
(95% CI)
LRT
p-value
Adjusted hazard ratio OS
(95% CI)
LRT
p-value
KRAS
(mutant vs. wild-type)
1.87
( )
<0.0001 NC NC
PIK3CA exon 9
(mutant vs. wild-type)
1.08
( )
0.67
1.27
( )
0.39
PIK3CA exon 20
(mutant vs. wild-type)
1.57
( )
0.14
3.69
( )
0.0055
BRAF
(mutant vs. wild-type)
2.68
( )
2.97
( )
<0.0001
NRAS
(mutant vs. wild-type)
1.81
( )
0.069
1.96
( )
0.042
De Roock et al., EU Consortium
Lancet Oncology, 2010

24. From gene targeted therapies to mutant targeted therapies
Example 1: PIK3CA mutations
Example 2: KRAS mutations

25. mCRC patients N=114 *P<0.05 (P=.011) Mutated KRAS 34/113 (30%)
Wild-Type KRAS
79/113 (70%)
Responders
2/34 (6%)*
22/79 (28%)*
Non Responders
32/34 (94%)*
57/79 (72%)*
Cancer Res 2007;67(6):2643–8 & J Clin Oncol. 2008; 26: 25

26. KRAS mutations: clinical results from cetuximab treated mCRC
Moroni Lancet Oncol 2005 n=31
Lièvre Clin Cancer Res 2006 n=30
Di Fiore Br J Cancer 2007 n=59
Frattini Br J Cancer 2007 n=27
Benvenuti Cancer Res 2007 n=48
Khambata-Ford J Clin Oncol 2007 n=80
De Roock ASCO Proc 2007 n=37
Finocchiaro ASCO Proc 2007 n=81
Response rate:
analysis of 8 studies available in PubMed or from ASCO
RAS mutated (7.0%)
RAS mutated (43.9%)
wt (93.0%)
wt (56.1%)
Responders (n=82)
Non-Responders (n=312)

27. KRAS mutations
Smith G, et al., British Journal of Cancer (2010), 1 –11
GEP
GDI
GDP
GTP
GTP
GDP
RAS
(inactive)
RAS
(active)
Effectors:
RAF/MAPK/ERK
PI3K/AKT
GAP Pi
Farnesyl Geranylgeranyl

28. Meta-analysis of 3 Chemotherapy Refractory Datasets
NCIC CTG dataset
from CO.17 trial
Leuven dataset
from clinical trials: EVEREST, BOND, SALVAGE, BABEL
Italian dataset:
from clinical trials mentioned above
from non-trial patients with advanced, irinotecan-refractory CRC considered suitable to receive an EGFR MAb

29. KRAS Mutation Status and Therapy by Dataset
Number of patients (%)
Dataset
NCIC CTG
Leuven
Italian
Kras results and treatment information available
394
282
125
Kras mutation status
G13D
20 (5)
20 (7)
8 (6)
Other mutation
144 (37)
102 (36)
24 (19)
Wild-type
230 (58)
160 (57)
93 (74)
Treatment
Cetuximab monotherapy
199 (50.5%)
33 (11.7%)
15 (12%)
Panitumumab monotherapy
0 (0%)
0 (%)
23 (18.4%)
Cetuximab + chemotherapy
249 (88.3%)
87 (63.6%)
No cetuximab or panitumumab
195 (49.5%)

30. Baseline Patient Characteristics by Tumour KRAS status
G13D Mutation
(N = 48)
Other mutations
(N = 270)
Wild type KRAS
(N = 483)
p-value*
Age – median (range) in year
65.5 ( )
62.0 ( )
62.0 ( )
.79
<65
23 ( 47.9)
157 ( 58.1)
287 ( 59.4)
≥65
25 ( 52.1)
113 ( 41.9)
192 ( 39.8)
Missing
0 ( 0.0)
0 ( 0.0)
4 ( 0.8)
Gender Female
22 ( 45.8)
109 ( 40.4)
161 ( 33.3)
.06
Male
26 ( 54.2)
161 ( 59.6)
322 ( 66.7)
ECOG performance status 0
12 ( 25.0)
54 ( 20.0)
118 ( 24.4)
.45 1
166 ( 61.5)
264 ( 54.7) 2
7 ( 14.6)
30 ( 11.1)
48 ( 9.9)
3 ( 6.3)
20 ( 7.4)
53 ( 11.0)
Site of primary Rectum only
10 ( 20.8)
57 ( 21.1)
116 ( 24.0)
.61
Colon
38 ( 79.2)
213 ( 78.9)
366 ( 75.8)
1 ( 0.2)
Number of prior chemotherapy regimens 0
3 ( 1.1)
8 ( 1.7)
.80
5 ( 10.4)
17 ( 6.3)
25 ( 5.2)
13 ( 27.1)
74 ( 27.4)
156 ( 32.3) 3
16 ( 33.3)
93 ( 34.4)
151 ( 31.3) 4
56 ( 20.7)
87 ( 18.0) ≥5
4 ( 8.3)
25 ( 9.3)
47 ( 9.7)
2 ( 0.7)
9 ( 1.9)
Treatment Mono Cetuximab
91 ( 33.7)
146 ( 30.2)
.34
Mono panitumumab
5 ( 1.9)
15 ( 3.1)
Cetuximab + chemotherapy
105 ( 38.9)
209 ( 43.3)
No cetuximab or panitumumab
69 ( 25.6)
113 ( 23.4)
* between biomarker positive and negative groups from chi-square test for categorical variables and t-test for continuous variables.

31. KRAS G13D Mutation status as a prognostic factor for OS in patients not treated with Cetuximab or Panitumumab?
Proportion alive 20 40 60 80
100
0.0
5.0
10.0
15.0
Time from randomization (months)
KRAS subset
Median OS (months)
Wild-type
4.5
G13D mutation
3.6
Other Mutation
4.7
De Roock et al JAMA 2010

32. OS Predictive Analysis by KRAS status: EGFR Mab Monotherapy vs no EGFR Mab
KRAS G13D Mutation
Other KRAS Mutation
KRAS Wild-type 20 40 60 80
100
0.0
5.0
10.0
15.0 20 40 60 80
100
0.0
5.0
10.0
15.0
20.0 20 40 60 80
100
0.0
5.0
10.0
15.0
20.0
HR 0.98 (0.70 to 1.38)
p=0.91
HR 0.56 (0.42 to 0.73)
p<0.0001
Proportion alive
Proportion alive
Proportion alive
HR 0.23 (0.09 to 0.61)
p=0.002
Time from randomization (months)
Time from randomization (months)
Time from randomization (months)
Monotherapy with cetuximab or panitumumab
No Treatment with cetuximab or panitumumab
De Roock et al JAMA 2010

33. Molecular bases of G12V versus G13D mediated resistance to cetuximab in cellular and animal models

34. Parallel clinical trials in cells, mice and patients
Drug Y
Mutation X

35. Isogenic models of tumour progression
Knock-out of cancer genes
Knock-in of oncogenic mutations wt
p53 -/-
Ras / Raf
PI3K
EGFR
Homologous recombination A B
The technology has already been validated into a murine mouse model. We have succeeded in fact in obtaining MLP 29 mutated in the codon…. A
Isogenic cells carrying
cancer mutations B

36. Mutation-specific pharmacogenomic profiles+
Parental cell line
Knock-in cell line
Drug screening
Incubate cells with drugs
Mutated genotype selective drug
Drug with no selectivity
Wild genotype selective drug
Di Nicolantonio; Arena et al., PNAS 2008
Di Nicolantonio et al., J Clin Invest, 2010

37. Experimental design Cellular model Measure drug response
Gene targeting (Knock-in approach)
KRAS: G12D, G12V, G12C, G12A, G12S, G12R, G13D
BRAF: V600E, PIK3CA: E545K (exon 9), H1047R (exon 20)
Biochemical validation (pathway activation)
Measure drug response

38. ITR P AAV-KRas-12V A ITR P Neo ITR AAV-KRas-13D Knock-in G12V
NotI
ITR
Neo P
LoxP
AAV-KRas-12V
G12V (G35>G/T) A
NotI
G13D (G38>G/A)
NotI
ITR
LoxP P
Neo
LoxP
LoxP
ITR
AAV-KRas-13D
Knock-in G12V
(or G12D / G12C) B
Homologous recombination
KRAS WT CRC cells
Supplementary Figure 1. Targeted knock-in (KI) of KRAS cancer mutations in SW48 colorectal cancer cells.
(A) Structure of AAV targeting constructs. AAV vectors carrying oncogenic alleles either in the 5’ (KRAS G12V) or the 3’ arm (KRAS G13D) were used to introduce the relevant genetic alterations in human somatic cells by homologous recombination. P, SV40 promoter; Neo, geneticin-resistance gene; ITR, inverted terminal repeat; triangles, loxP sites. The nucleotide and aminoacid changes corresponding to the oncogenic alleles are indicated. (B) The G12V and G13D alleles were introduced in the genome of SW48 colorectal cancer cells by AAV-mediated targeted homologous recombination. (C) The expression of the introduced genetic alterations in the targeted cells was determined by RT-PCR and sequencing of the KRAS transcript. C
SW48 KRAS WT
Knock-in G13D
SW48 KRAS G12V
SW48 KRAS G13D

39. chemotherapy in cellular models
KRAS G12V or G13D and
chemotherapy in cellular models
De Roock et al JAMA 2010

40. ceruximab in cellular models
KRAS G12V and G13D and
ceruximab in cellular models
De Roock et al JAMA 2010

41. Cetuximab delays growth of SW48 tumor xenografts
De Roock et al JAMA 2010

42. Cetuximab does not affect growth of G12V tumors, but inhibits the growth of G13D tumor xenografts
SW48 KRAS G12V
SW48 KRAS G13D
Start of treatment
Start of treatment
De Roock et al JAMA 2010

43. Secondary resistance to targeted therapies
Responders (15-20%)
Non-Responders
2007

44. Secondary resistance to targeted therapies
Responders (15-20%)
Non-Responders
2010

45. Parallel clinical trials in cells, mice and patients
Drug Y
Mutation X

46. Marker A Drug X Marker B Drug Y Expansion
DNA, RNA and protein extraction,
FFPE blocks stored by the pathologist
DNA, RNA and protein extraction,
FFPE blocks stored by the pathologist
Liver Met implanted s.c. in NOD SCID mice
Marker A
Drug X
Patient undergoing liver metastasectomy of CRC
Using this approach 112 samples were
succesfully engrafted since Oct 2008
A. Bertotti & L. Trusolino, Molecular Oncology, IRCC
Marker B
Drug Y
Expansion

47. Xenopatients 148 >90% 44 NUMBER OF SAMPLES p0 p1 p2 SURGERY
from the pathologist
NUMBER OF SAMPLES
FFPE blocks
148
SURGERY
DMSO
RNA later
>90% p0
engraftment
(2 mice)
DMSO
RNA later
Archive
RNA extraction
Genomic DNA extraction p1
expansion
(6 mice)
DMSO
RNA later
Snap Frozen
DMSO
RNA later
Snap Frozen
FFPE blocks 44 p2
treatment
(24 mice)
A. Bertotti & L. Trusolino, Molecular Oncology, IRCC

48. In vivo – M016

49. Understanding secondary resistance to cetuximab
Time 0: Molecular analysis using
multiple omics’ technologies (WP3)
Time x: Molecular analysis using
multiple omics’ technologies (WP3)
control
cetuximab
secondary resistance
Chronic treatment with cetuximab (0.5 mg/injection/2x/week)

50. Xenopatient M026: development of resistance

51. RESISTANCE- RESISTANCE RESISTANCE- RESISTANCE RESISTANCE- RESISTANCE RESISTANCE- RESISTANCE

52. COLTHERES “Modelling and predicting resistance to molecular
therapies in colorectal cancers”
Executive Summary:
COLTHERES is a consortium of EU-clinical centres and translational researchers who have received 6M Euros of core funding from the EU Framework-7 program to define and perform biomarker driven clinical trials to improve cancer therapy outcomes. This is a 4-year programme that will use comprehensively molecularly-annotated colon cancers as a ‘test-bed’ to define specific biomarkers of response or resistance to signalling pathway agents. This consortium is open to any Institution who wishes to determine which patients are most likely to respond to novel CRC therapies and perform rapid proof-of-concept clinical trials.

53. Consortium Members
Alberto Bardelli (University of Torino-IRCC): Cancer mutations and targeted therapies. Drug resistance mechanisms
Sabine Tejpar (University Hospital Leuven): Clinical trials with molecularly targeted therapies
Josep Tabernero (Hospital Vall d’Hebron): Clinical trials with molecularly targeted therapies
Salvatore Siena (Ospedale Niguardia): Targeted clinical trials and patient drug resistance mechanisms
Horizon Discovery: (Cambridge UK) Novel gene-targeting platform to create genetically-defined human cancer models + drug screening
Agendia : (Amsterdam) Microarrays on clinical samples and diagnosis based on molecular profiles
Rene Bernards: (NKI Amsterdam) Functional genomics, screens for drug-response modifying genes
Manel Esteller: (Barcelona) Epigenomic profiling of clinical samples
Michael Clague: (University of Liverpool) Global proteomic profiling in cancer models
Mauro Delorenzi (Swiss Institute of Bioinformatics) Bioinformatics, statistical analysis
Paul Crompton (ARTTIC Brussels) Administration and management

55. Royal Mail Stamp Issue 25 February 2003
The doctor’s perspective
Royal Mail Stamp Issue 25 February 2003

56. The patient’s perspective
EVOLUTION FROM and Anatomic and DISEASE-BASED TO A MORE PERSONALIZED-BASED MEDICINE

57. Molecular Genetics Lab:
Federica Di Nicolantonio
Sabrina Arena
Miriam Martini
Emily Crowley
Elisa Scala
Carlotta Cancelliere
Sebastijan Hobor
Davide Zecchin
Simona Lamba
Michela Buscarino
Milo Frattini
Salvatore Siena
Andrea Sartore Bianchi
Marcello Gambacorta
Livio Trusolino
Andrea Bertotti
Josep Tabernero 57