Validation of a 46-gene expression signature in early-stage non-small-cell lung cancer

Edward S. Kim, MD Chair, Solid Tumor Oncology and Investigational Therapeutics Donald S. Kim Distinguished Chair for Cancer Research Levine Cancer Institute Carolinas HealthCare System

Raphael Bueno, MD Associate Chief, Thoracic Surgery Brigham and Women's Hospital Professor of Surgery Harvard Medical School Jennifer Saam, PhD Medical Science Liaison for Personalized Medicine Medical Services, Myriad Genetic Laboratories

Agenda

7:00 AM - 7:25 AM Dr. Edward Kim: Welcome; Integrated Prognosis in Early Stage, Resectable Lung Adenocarcinoma

7:25 AM - 7:50 AM Dr. Raphael Bueno: Validation of a Proliferation-based Expression Signature as Prognostic Marker in Early Stage Lung Adenocarcinoma

7:50 AM - 8:00 AM Dr. Jennifer Saam: myPlanTM Lung Cancer Overview

Integrated Prognosis in Early Stage, Resectable Lung Adenocarcinoma

Edward S. Kim, MD Chair, Solid Tumor Oncology and Investigational Therapeutics Donald S. Kim Distinguished Chair for Cancer Research Levine Cancer Institute Carolinas HealthCare System

Early-Stage Lung Cancer Patients Have Poor Outcomes • 35%-50% recurrence rates even in patients with no nodal or

other metastatic involvement at time of surgery

• No proven benefit of adjuvant chemotherapy in stage IA patients

• One controversial study suggests adjuvant chemotherapy provides benefit in Stage IB patients >4cm

• Defined need for objective information to identify early-stage patients at greater or less risk of mortality

Old Staging vs. New Staging

5th edition • T1 <3 cm • T2 >3 cm

7th edition • T1a <2 cm • T1b >2 to 3 cm • T2a >3 to 5 cm • T2b >5 to 7 cm

Surgical Resection

IA: CT Observation

IB: CT Observation or adjuvant chemotherapy1

II: Adjuvant chemotherapy

1 High-risk patients are defined by poorly differentiated tumors, vascular invasion, wedge resection, tumors > 4 cm, visceral pleural involvement

Current Treatment Guidelines for Stage I-II NSCLC

Pisters KMW, Evans WK, Azzoli CG, et al. Cancer Care Ontario and American Society of Clinical Oncology adjuvant chemotherapy and adjuvant radiation therapy for stages I-IIIA resectable non–small-cell lung cancer guideline. J Clin Oncol. 2007;25:5506-5518.

Treatment Dilemma

Is there benefit to molecular understanding of the aggressiveness of

early stage lung cancer?

Cell Cycle Progression (CCP) Score

• RNA expression profile of cell cycle genes measured by multiplex, quantitative RT-PCR

• Hypothesis based gene selection

• Validated in multiple tumor types – Prostate – Lung – Breast

Cell Cycle Progression (CCP) Score

• Why Cell Cycle Progression Genes? –Cell cycle genes are the major component of most

prognostic signatures –Measures cancer aggressiveness based on cell division

• CCP Score –RNA expression profile based on:

• 31 genes induced during cell cycle • 15 housekeeping genes for normalization

–Validated on FFPE preserved surgical specimens –For adenocarcinoma

Cell Cycle Progression Genes

• BUB1B • RRM2 • DLGAP5 • BIRC5 • KIF20A • PLK1 • TOP2A • TK1 • C18orf24 • RAD54L • CDCA3

• CDC20 • ORC6L • CDCA8 • CENPM • RAD51 • CEP55 • DTL • PRC1 • PBK • ASF1B

• FOXM1Z • CDzqKN3 • CDC2 • KIF11 • KIAA0101 • NUSAP1 • CENPF • ASPM • MCM10 • PTTG1

Verification Data: Cohorts (Stage I and II patients)

1. Publicly available microarray data – Director’s consortium (US) N=256 – GSE31210 (Japan) N=204

2. Clinical FFPE samples from surgical resection

– MD Anderson cohort N=207 – Institute of European Oncology cohort (Milan, IT) N=174

Adenocarcinoma Has a Wide CCP Score Distribution

CCP Distribution in Adenocarcinoma

CCP Score Predicts Lung Cancer Death

• CCP is significant both as a univariate predictor and in a multivariate model in all three cohorts

• CCP provides additional information to usual clinical factors

Study N Deaths

HR* 95% CI

CCP univariate p-values

CCP multivariate

p-values

Director’s Consortium 256 71 2.02

(1.29-3.17) 0.0001 0.002

GSE31210 204 25 2.16 (1.32-3.53) 0.001 0.003

MDACC-IEO 381 62 1.92 (1.18-3.10) 0.0003 0.007

HR are reported for the interquartile range Events are cancer-related death within 5 years of surgery

Multi Variate Model: MDACC and IEO Combined

DS ~ CCP + Age + Gender + Smoking Status

+ TNM Stage + Adjuvant Treatment

+ Tumor Size + Pleural Invasion +

Cohort + Stage: Treatment

Interaction terms for CCP: Stage, and CCP:cohort were tested but were not significant at the 5% level

Deaths: 62/381 Univariate Multivariate

Age 0.03 0.12

Gender 0.002 0.01

Smoking 0.32 0.99

Stage 0.004 0.15

Treatment 0.52 0.13

Tumor Size 0.007 0.39

Pleural Inv. 0.01 0.009

Cohort 0.43 0.61

Stage:Treatment NA 0.09

CCP 0.0003 0.007

CCP HR* (95% CI) 2.10 (1.39-3.17) 1.92 (1.18-3.10)

* Standardized Hazard Ratio

MDACC and IEO Combined

• In this cohort, gender, pleural invasion and CCP are significant in both the univariate and multivariate models

CCP Score by Stage

•There is a wide CCP score distribution across stages •CCP score adds significant information to stage

stage IA, n=184 stage IB, n=153 stage II, n=44

Derivation of a Combined Score (CCP+Stage) • Analysis of untreated patients from Director’s Consortium

cohort (DCC) and the clinical data set (MDACC/IEO). • The two variables are weighted by the hazard ratios from a

Cox PH analysis with 5-year disease-specific survival.

CONFIDENTIAL

Prognostic score = 0.33 * CCP score + 0.52 * stage

Stage was used as numerical variable (1=IA, 2=1B, 3=IIA, 4=2B)

5-year Risk of Lung Cancer Specific Mortality Based on Prognostic Score

Prognostic Score

Data on file.

5-year Lung Cancer Mortality Risk MDACC-IEO NSCLC Cohort

Summary • CCP expression score is prognostic for lung cancer survival in

3 lung adenocarcinoma data sets.

• A combination of pathological stage and the CCP expression score, which produces a prognostic score (PS) that is a more effective predictor of post-surgical risk of cancer specific death than pathological stage alone.

• A more precise risk assessment can give better guidance in balancing treatment related risks with disease-specific survival.

Validation of a Proliferation-based Expression Signature as Prognostic Marker in Early Stage Lung Adenocarcinoma

Raphael Bueno, MD Associate Chief, Thoracic Surgery Brigham and Women's Hospital Professor of Surgery Harvard Medical School

Methods Design: • Patient clinical and outcome data were blinded. • Pre-determined Statistical Analysis Plan (SAP) archived

with files of finalized clinical data and CCP scores prior to unblinding.

Patient Population: • Stage I-II NSCLC ADC by 7th edition IASLC staging • Complete surgical resection • No neo-adjuvant treatment • No adjuvant chemotherapy or radiation within 12 weeks

of surgery • Brigham and Women’s and Royal Infirmary at Edinburgh (RIE) Samples: • FFPE tumor specimens • Processed in the CLIA-certified laboratory • 650 patients • 152 events

Demographic Information

Table 1. Demographic and clinical data for the two patient cohorts in the validation study * Pleural invasion data were not available for 17 patients.

CCP and Stage Are Significant Predictors of 5-year Lung Cancer Mortality

Table 2. The CCP score is a significant prognostic marker after adjustment for clinical variables. Results from univariate and multivariate Cox proportional hazards analysis. * Hazard ratio is reported per interquartile range of the CCP score. # Hazard ratio is reported per cm, rounded to the nearest mm. Multivariate analysis, and univariate analysis of pleural invasion, included 633 patients with 147 events. All other univariate analyses included 650 patients with 152 events.

Significance of Prognostic Score

• Prognostic Score, Combining Stage and CCP Captures Significant Information that is not Captured by Stage Alone

• Prognostic score is predictive of 5-year mortality

Table 3. The Prognostic Score captures significant prognostic information that is not provided by stage alone. Significance of the Prognostic Score and stage in univariate and bivariate models. Analyses included 650 patients with 152 events. * Hazard ratio is reported per interquartile range of the PS score.

5-year Lung Cancer Mortality as Predicted by the Prognostic Score

Low versus High Prognostic Score

• Patients in the Low PS group had significantly more favorable 5-year survival than patients in the High PS group

Low PS (n=328) High PS (n=322) P-Value = 3.8 x 10-6

5-year Lung Cancer Mortality Risk BWH-RIE NSCLC ADC

Prognostic Score Risk Stage Min Mean Max

IA 11% 18% 34% IB 17% 28% 43% IIA 27% 42% 62% IIB 38% 60% 68%

Summary

• Prognostic Score (PS) was significantly predictive of 5-year lung cancer mortality, and was significantly more predictive than stage alone (P-Value 2.8×10-11)

• Patients in the low PS group had significantly more favorable 5-year survival than patients in the high PS group (P-Value = 3.8 x 10-6)

• PS improves risk stratification compared to pathological stage alone.

PS Group Survival Mortality Low Risk 82% 18% High Risk 65% 35%

Jennifer Saam, PhD Medical Science Liaison for Personalized Medicine Medical Services, Myriad Genetic Laboratories

myPlanTM Lung Cancer Overview

• myPlanTM Lung Cancer is a RNA expression profile of cell cycle genes measured by multiplex, quantitative RT-PCR, validated for stage I and II non-small cell lung adenocarcinoma without pre-operative radiation therapy and /or chemotherapy.

• FFPE blocks or slides are sent to Myriad where the analysis is

performed in a CLIA-certified laboratory.

• Patient results are sent as either a High or Low myPlanTM Lung Cancer Prognostic Score, which predicts the risk of 5-year lung cancer specific mortality.

• myPlanTM Lung Cancer will be available in the United States via an

early access Clinical Experience Program mid-November 2013.

Questions?

Result example