1. Humblet ASCO 2007 – Draft CONFIDENTIAL Association of skin toxicity (ST) severity with clinical outcomes and health-related quality of life (HRQoL) with panitumumab (Pmab) Yves Humblet, 1 Marc Peeters, 2 Salvatore Artale, 3 Alain Hendlisz, 4 Bart Neyns, 5 A. Sobrero, 6 Michael Wolf, 7 Michael Woolley, 7 Rafael Amado, 7 Eric Van Cutsem 8 Abstract #4038 1 St-Luc University Hospital, Brussels, Belgium; 2 Ghent University Hospital, Ghent, Belgium; 3 Ospedale Niguarda Ca’ Granda, Milan, Italy; 4 Jules Bordet Institute, Brussels, Belgium; 5 AZ Vrije Universiteit Brussel, Brussels, Belgium; 6 Ospedale San Martino, Genoa, Italy; 7 Amgen Inc., Thousand Oaks, CA; 8 University Hospital Gasthuisberg, Leuven, Belgium

2. Humblet ASCO 2007 – Draft CONFIDENTIAL Introduction Skin toxicity severity has been shown to be associated with efficacy in patients with metastatic colorectal cancer (mCRC) receiving anti- epidermal growth factor receptor (EGFr) therapy Skin toxicity severity has been shown to be associated with efficacy in patients with metastatic colorectal cancer (mCRC) receiving anti- epidermal growth factor receptor (EGFr) therapy Panitumumab is a fully human monoclonal antibody directed against the EGFr and is indicated for use in patients with mCRC who progressed on or following standard chemotherapy 1,2 Panitumumab is a fully human monoclonal antibody directed against the EGFr and is indicated for use in patients with mCRC who progressed on or following standard chemotherapy 1,2 This phase 3, randomized controlled trial demonstrated that panitumumab plus best supportive care (BSC) improved progression-free survival (PFS) compared with BSC alone in chemorefractory mCRC patients 3 This phase 3, randomized controlled trial demonstrated that panitumumab plus best supportive care (BSC) improved progression-free survival (PFS) compared with BSC alone in chemorefractory mCRC patients 3 We conducted an exploratory analysis to evaluate the association of skin toxicity severity with PFS, overall survival (OS), disease-related symptoms and HRQOL with panitumumab treatment We conducted an exploratory analysis to evaluate the association of skin toxicity severity with PFS, overall survival (OS), disease-related symptoms and HRQOL with panitumumab treatment

3. Humblet ASCO 2007 – Draft CONFIDENTIAL RANDOMIZE Study Design 1:1 PanitumumabPD* Follow-up 6.0 mg/kg Q2W + BSC BSCPD* Follow-up Optional Panitumumab Crossover Study Stratification: ECOG score: 0-1 vs. 2 ECOG score: 0-1 vs. 2 Geographic region (Western Europe vs. Central and Eastern Europe vs. the rest of world) Geographic region (Western Europe vs. Central and Eastern Europe vs. the rest of world) Primary Endpoint Progression-free survival (per blinded central radiology assessment of modified RECIST criteria) Progression-free survival (per blinded central radiology assessment of modified RECIST criteria) Secondary Endpoints Overall survival time and best overall objective response (central radiology; testing prespecified) Overall survival time and best overall objective response (central radiology; testing prespecified) Safety, disease-related symptoms, and HRQoL Safety, disease-related symptoms, and HRQoL *PD= progressive disease

4. Humblet ASCO 2007 – Draft CONFIDENTIAL Metastatic colorectal adenocarcinoma (mCRC) Metastatic colorectal adenocarcinoma (mCRC) ECOG score 0 to 2 ECOG score 0 to 2 Radiologic documentation of disease progression after fluoropyrimidine, irinotecan, and oxaliplatin Radiologic documentation of disease progression after fluoropyrimidine, irinotecan, and oxaliplatin – During or within 6 months following most recent chemotherapy regimen – 2 or 3 prior regimens EGFr membrane staining on ≥ 1% tumor cells (IHC, central laboratory) EGFr membrane staining on ≥ 1% tumor cells (IHC, central laboratory) Adequate hematologic, renal, and hepatic function Adequate hematologic, renal, and hepatic function No symptomatic brain metastases No symptomatic brain metastases No chemotherapy within 30 days before randomization No chemotherapy within 30 days before randomization No prior anti-EGFR agent therapy No prior anti-EGFR agent therapy Key Eligibility Criteria

5. Humblet ASCO 2007 – Draft CONFIDENTIAL Assessments Skin toxicity was measured by:Skin toxicity was measured by: –CTCAE grading criteria version 3.0 (modified for dermatology toxicities) and –Modified Dermatology Life Quality Index (mDLQI) (See Table 1) CRC symptoms were measured by the NCCN/FACT CRC symptom index (FCSI) (See Table 1)CRC symptoms were measured by the NCCN/FACT CRC symptom index (FCSI) (See Table 1) HRQoL was measured by the EQ-5D and EORTC QLQ-C30 Global QoL subscale (See Table 1)HRQoL was measured by the EQ-5D and EORTC QLQ-C30 Global QoL subscale (See Table 1)

6. Humblet ASCO 2007 – Draft CONFIDENTIAL Table 1. Patient Reported Outcomes PRO Instrument MeasurementDescription Assessment Schedule Modified Dermatology Life Quality Index (mDLQI) Problems due to skin condition 3 items measuring problems and bother with skin condition. Scores range from 0 to 100 with higher scores indicating less problems/bother Baseline, Q2W during treatment, and at the 30-day safety follow-up visit after the last protocol directed therapy NCCN/FACT Colorectal Symptom Index (FCSI) CRC Symptoms A brief symptom index comprising of the most clinically relevant (clinician rated) symptoms for assessing symptomatic response to treatment for CRC. Scores range from 0 to 100 with higher scores indicating less symtomatology EUROQOL (EQ5D) Health Index HRQOL Provides a preference-weighted assessment of overall QOL across five dimensions: mobility, self– care, usual activity, pain or discomfort, and anxiety or depression. Scores range from 0 to 1 (0 = death, 1 = perfect health) Baseline, monthly during treatment, and at the 30-day safety follow-up visit after the last protocol directed therapy EUROQOL (EQ5D) Visual Analog Scale (VAS) HRQOL Measures overall health status. Scores range from 0 to 100 with 0 representing worst imaginable health state and 100 representing best imaginable health state EORTC QLQ-C30 Global Health Status/QOL Scale HRQOL 2 items of the EORTC QLQ-C30 measuring overall health status and overall QOL. Scores range from 0 to 100 with higher scores indicating better HRQOL

7. Humblet ASCO 2007 – Draft CONFIDENTIAL Overall Treatment Effect PRO Analysis set included all patients who had at least 1 post-baseline PRO assessment PRO Analysis set included all patients who had at least 1 post-baseline PRO assessment Association of skin toxicity with efficacy and PROs Analyses of skin toxicity by CTCAE grading criteria and mDLQI were restricted to patients in the panitumumab arm who were progression-free for at least 28 days to allow for onset of skin toxicity and Analyses of skin toxicity by CTCAE grading criteria and mDLQI were restricted to patients in the panitumumab arm who were progression-free for at least 28 days to allow for onset of skin toxicity and – had  grade 1 skin toxicity (for CTCAE) or – had at least 1 post-baseline PRO assessment (mDLQI) Dichotomization of PFS for degree of being bothered was explored using all possible cut points of mDLQI (based on Cox model) Dichotomization of PFS for degree of being bothered was explored using all possible cut points of mDLQI (based on Cox model) Hazard ratios (HR) adjusted for randomization factors and correlation analyses were used to determine relationships between PFS and OS to skin toxicity, and skin toxicity to colorectal cancer symptoms and overall HRQoL Hazard ratios (HR) adjusted for randomization factors and correlation analyses were used to determine relationships between PFS and OS to skin toxicity, and skin toxicity to colorectal cancer symptoms and overall HRQoL Statistical Analyses

8. Humblet ASCO 2007 – Draft CONFIDENTIAL Association of skin toxicity with efficacy and PROs (cont.) Missing PRO data were imputed using 2 imputation methods:Missing PRO data were imputed using 2 imputation methods: –The last value carried forward (LVCF) method, in which missing observations were replaced with the LVCF, or zero value carried forward at death; and –The slope method, in which a forward linear interpolation of observed data was utilized to impute missing data Statistical Analyses (cont.)

9. Humblet ASCO 2007 – Draft CONFIDENTIAL Results Disposition and Patient Analysis Sets Panitumumab Plus BSC BSC Alone Total Randomized, n 231232 PRO All Enrolled Analysis Set, n 207184 Patients Completing EQ-5D, n Week 4 Week 4189129 Week 8 Week 811147 Week 12 Week 129114 Week 16 Week 16627 Patients Completing FACT CRC Subscale, n Week 4 Week 4190130 Week 8 Week 811248 Week 12 Week 129014 Week 16 Week 16627 Patients Completing mDLQI Subscale, n Week 4 Week 4189128 Week 8 Week 811148 Week 12 Week 129013 Week 16 Week 16626

10. Humblet ASCO 2007 – Draft CONFIDENTIAL Demographics and Disease Characteristics PRO All Enrolled Analysis Set Panitumumab Plus BSC (N=207) BSC Alone (N=184) Sex – n (%) Men 137 (66) 119 (65) Median (min, max) age – years 61 (27, 79) 63 (32, 83) Age Categories – years < 65 ≥ 65 124 (60) 83 (40) 109 (59) 75 (41) ECOG status – n (%) 0-1 ≥ 2 a 185 (89) 22 (11) 160 (87) 24 (13) Number of metastatic sites – n (%) 1-2 3-5 147 (71) 60 (29) 126 (69) 56 (31) Prior adjuvant chemotherapy – n (%) 75 (36) 69 (38) Prior chemotherapy – n (%) At least 2 lines At least 3 lines 206 (100) 77 (37) 184 (100) 66 (36) a One patient had ECOG score of 3

11. Humblet ASCO 2007 – Draft CONFIDENTIAL The mDLQI-Bother Score Indicated That Panitumumab Patients Were More Bothered By Their Skin Condition Mean (+/- SE) Overall Treatment Effect

12. Humblet ASCO 2007 – Draft CONFIDENTIAL Difference (95% CI) in HRQOL Scores (EQ-5D) at Prespecified Timepoints for Panitumumab minus BSC Patients (Primary PRO Analysis) LVCF imputation method used for missing PRO data The yellow line at -0.08 represents a minimum clinical important difference (MCID) anchored to a ½ point change in ECOG score Although panitumumab patients were more bothered by their skin conditions, HRQOL scores were numerically in favor of panitumumab Although panitumumab patients were more bothered by their skin conditions, HRQOL scores were numerically in favor of panitumumab Difference in EQ-5D Scores (panitumumab- BSC) Panitumumab N=207 BSC N=184

13. Humblet ASCO 2007 – Draft CONFIDENTIAL Difference (95% CI) in CRC Symptom Scores (FCSI) at Prespecified Timepoints for Panitumumab minus BSC Patients (Primary PRO Analysis) LVCF imputation method used for missing PRO data The yellow line at -3.94 represents a minimum clinical important difference (MCID) anchored to a ½ point change in ECOG score Although panitumumab patients were more bothered by their skin condition, disease-related symptom scores were numerically in favor of panitumumab Although panitumumab patients were more bothered by their skin condition, disease-related symptom scores were numerically in favor of panitumumab Difference in FCSI scores (panitumumab- BSC) Panitumumab N=207 BSC N=184 MCID

14. Humblet ASCO 2007 – Draft CONFIDENTIAL Time to Onset of Worst Severity Skin Toxicity Among all patients:Among all patients: –The median time (95% CI) to most severe skin toxicity was 15 (14,16) days (Kaplan-Meier method) The worst severity of skin toxicity may occur beyond 28 days, thus lead-time bias cannot be completely eliminatedThe worst severity of skin toxicity may occur beyond 28 days, thus lead-time bias cannot be completely eliminated

15. Humblet ASCO 2007 – Draft CONFIDENTIAL Progression-Free Survival by Worst Severity of Skin Toxicity in Panitumumab Patients (Landmark Analysis) Panitumumab patients with worst grade of 2-4 had better progression free survival vs those with a worst grade of 1Panitumumab patients with worst grade of 2-4 had better progression free survival vs those with a worst grade of 1 hazard ratio (grade 2-4:1)=0.66 p=0.01 Landmark Analysis set included patients who were progression free for ≥ 28 days Association of Skin Toxicity with Outcomes

16. Humblet ASCO 2007 – Draft CONFIDENTIAL Overall Survival by Worst Severity of Skin Toxicity in the Panitumumab Patients (Landmark Analysis) hazard ratio (grade 2-4:1)=0.68 p=0.03 Panitumumab patients with worst grade of 2-4 had better overall survival vs those with a worst grade of 1Panitumumab patients with worst grade of 2-4 had better overall survival vs those with a worst grade of 1 Landmark Analysis set included patients who were progression free for ≥ 28 days

17. Humblet ASCO 2007 – Draft CONFIDENTIAL Progression-Free Survival by Minimum Post-Baseline mDLQI Score in the Panitumumab Patients (Landmark Analysis) Panitumumab patients with a minimum post-baseline mDLQI score ≤ 66.667(more bothered by their skin toxicity)had longer progression free survivalPanitumumab patients with a minimum post-baseline mDLQI score ≤ 66.667 (more bothered by their skin toxicity) had longer progression free survival hazard ratio (≤66.667:>66.667)=0.41 p<0.0001 Landmark Analysis set included patients who were progression free for ≥ 28 days; at least 1 post-baseline PRO assessment was also required. mDLQI nadir timing: mean – 30 days, median – 14 days, 75th Percentile – 28 days

18. Humblet ASCO 2007 – Draft CONFIDENTIAL Overall Survival by Minimum Post-Baseline mDLQI Score in the Panitumumab Patients (Landmark Analysis) hazard ratio (≤66.667:>66.667)=0.35 p<0.0001 Panitumumab patients with a minimum post-baseline mDLQI score ≤ 66.667(more bothered by their skin toxicity)had longer overall survivalPanitumumab patients with a minimum post-baseline mDLQI score ≤ 66.667 (more bothered by their skin toxicity) had longer overall survival Landmark Analysis set included patients who were progression free for ≥ 28 days; at least 1 post-baseline PRO assessment was also required. mDLQI nadir timing: mean – 30 days, median – 14 days, 75th Percentile – 28 days

19. Humblet ASCO 2007 – Draft CONFIDENTIAL Minimum Post-Baseline mDLQI Bother Score Association with Median Post-Baseline Patient-Reported Outcomes EQ-5D index EQ-5D VAS EORTC Global FCSI mDLQIBother-0.24(p=0.0006)-0.17(p=0.01)-0.08(p=0.25)-0.13(p=0.06) EQ-5D index 0.61(p<.0001)0.60(p<.0001)0.71(p<.0001) EQ-5D VAS 0.79(p<.0001)0.70(p<.0001) EORTC Global 0.74(p<.0001) PRO All Enrolled Analysis set, panitumumab patients only. Pearson correlation coefficients are shown with p-values in parentheses. NOTES: Bother = the mDLQI “bother” item; EQ-5D Index = EQ-5D Index; EQ-5D VAS = EQ-5D Visual Analog Store; EORTC Global = EORTC QLQ-C30 Global Quality of Life Scale; FCSI = FACT/NCCN Colorectal Cancer Symptom Index In the panitumumab patients, lower mDLQI scores (more skin bother) were associated with higher HRQOL scores (improved HRQoL)In the panitumumab patients, lower mDLQI scores (more skin bother) were associated with higher HRQOL scores (improved HRQoL)

20. Humblet ASCO 2007 – Draft CONFIDENTIAL Conclusions As expected, panitumumab patients reported being more bothered by their skin conditions than BSC patients; however, CRC symptoms and HRQOL scores trended in favor of panitumumab vs BSC As expected, panitumumab patients reported being more bothered by their skin conditions than BSC patients; however, CRC symptoms and HRQOL scores trended in favor of panitumumab vs BSC – Based on the minimal clinical important difference, a negative effect with panitumumab treatment could be excluded From this exploratory analysis of data from a phase 3 trial, improved PFS, overall survival, CRC symptomalogy, and HRQOL were associated with worse skin toxicity as measured by the CTCAE grading scale and mDLQI. From this exploratory analysis of data from a phase 3 trial, improved PFS, overall survival, CRC symptomalogy, and HRQOL were associated with worse skin toxicity as measured by the CTCAE grading scale and mDLQI. These findings support the role of skin toxicity severity as a pharmacodynamic marker of on-target activity that appears to be associated with clinical benefit. Further analyses are being conducted to explore the predictive value of early onset of skin toxicity severity (see abstract #4134, poster #P5 by Berlin et al.) These findings support the role of skin toxicity severity as a pharmacodynamic marker of on-target activity that appears to be associated with clinical benefit. Further analyses are being conducted to explore the predictive value of early onset of skin toxicity severity (see abstract #4134, poster #P5 by Berlin et al.)

21. Humblet ASCO 2007 – Draft CONFIDENTIAL References 1.Foon KA, Yang X-D, Weiner LM, et al. Preclinical and clinical evaluations of ABX-EGF, a fully human anti-epidermal growth factor receptor antibody. Int J Radiat Oncol Biol Phys 2004:58:984-990. 2.Vectibix TM Prescribing Information, Amgen Inc. Thousand Oaks CA; 2006. 3.Van Cutsem E, Peeters M, Siena S, et al. An open-label, randomized, phase 3 clinical trial of panitumumab plus best supportive care versus best supportive care in patients with chemotherapy- refractory metastatic colorectal cancer. J Clin Oncol 2007;25:1658- 1664.