C.P. Belani 1 , T. Brodowicz 2 , P. Peterson 3 , W. John 3 , G. Scagliotti 4

C.P. Belani1, T. Brodowicz2, P. Peterson3, W. John3, G. Scagliotti 4

1Penn State Cancer Institute, Hershey, PA USA;

2Medical University, Vienna, Austria; 3Eli Lilly and Co., IN, USA; 4University of Torino, Orbassano, Italy

Nonsquamous Histology: A Predictor of Efficacy for Pemetrexed Treatment. An Analysis of Phase III Trials

Using Treatment-by-Histology Interaction (THI) in Advanced NSCLC

The identification of predictive factors may allow a The identification of predictive factors may allow a tailored- approach to the treatment of patientstailored- approach to the treatment of patients

Treatment-by-histology interaction (THI) analyses Treatment-by-histology interaction (THI) analyses predicts the relative efficacy of a specific treatment predicts the relative efficacy of a specific treatment versus a comparatorversus a comparator

Randomized phase III study of pemetrexed vs. docetaxel Randomized phase III study of pemetrexed vs. docetaxel indicated a predictive role for histology in NSCLC, as indicated a predictive role for histology in NSCLC, as well as a differential treatment effect by histology for well as a differential treatment effect by histology for pemetrexedpemetrexed11

We report the results of THI analyses performed on two We report the results of THI analyses performed on two large randomized trials evaluating the effects of large randomized trials evaluating the effects of pemetrexed.pemetrexed.

11 Peterson, et al. WCLC. 2007; P2-328, Seoul, Korea.Peterson, et al. WCLC. 2007; P2-328, Seoul, Korea.

Role of Histology as a Predictive Factor

Study Design – Pemetrexed/Cisplatin vs. Gemcitabine/CisplatinStudy Design – Pemetrexed/Cisplatin vs. Gemcitabine/Cisplatin

Randomization factors: • Stage • PS • Gender • Histo vs. cyto dx • Brain mets. hx

Cisplatin 75 mg/m2 d1 +Pemetrexed 500 mg/m2 d1(N=862)

Median OS Cis Pemetrexed 10.3 mos Cis Gemcitabine 10.3 mos

HR=0.94(95% CI: 0.84–0.1.05 )Cis/Pem noninferior vs Cis/Gem

Cisplatin 75 mg/m2 d 1 +Gem 1.25 mg/m2 d 1 & 8(N=863)

B12, folate, and dexamethasone given in both arms

1:1 Randomization

Double-blind, Placebo-controlled, Multicenter, Phase III Trial

# Nonsquamous pts : CisPem 71.7%, CisGem 72.4%

Each cycle repeated q3weeks up to 6 cycles

Study Design – Maintenance Pemetrexed vs. PlaceboStudy Design – Maintenance Pemetrexed vs. Placebo

Stage IIIB/IV NSCLC PS 0-1 4 prior cycles of gem,

doc, or tax + cis or carb, with CR, PR, or SD

Randomization factors: gender PS stage best tumor response to

induction non-platinum induction

drug brain mets

Pemetrexed 500 mg/m2 (q21d) + BSC (N=441)

Median PFS Pemetrexed 4.04 mos Placebo 1.97 mos

HR=0.599 (95% CI: 0.49–0.73)p <0.00001

Placebo (d1, q21d) + BSC (N=222)

B12, folate, and dexamethasone given in both arms

2:1 Randomization

Double-blind, Placebo-controlled, Multicenter, Phase III Trial

# Nonsquamous pts : Pemetrexed 74%, Placebo 70.3%

Treatment-by-Histology AnalysisTreatment-by-Histology Analysis

Presence of THI implies that histologic- subtype will be predictive of the treatment effect

Analysis of the 2 studies performed to determine the treatment effect for pemetrexed relative to the control arm (non-squamous histology vs. those with squamous histology)

THI analyses, prospectively planned for both the studies utilized adjusted Cox proportional hazard models of PFS & OS

The interaction parameter Exp (B3) is a ratio of two hazard ratios:

HR for non-squamous over squamous (for Pemetrexed patients) ---------------------------------------------------------------------------------------------------------- HR for non-squamous over squamous (for Comparator arm patients)

Exp (B3) can also be re-written as the following ratio:

HR for Pemetrexed over the Comparator (for Non-squamous patients) -------------------------------------------------------------------------------------------------------- HR for Pemetrexed over the Comparator (for Squamous patients)

So the interaction HR Exp (B3) is the treatment effect for the non-squamous subgroup divided by the treatment effect for the squamous subgroup.

The test for interaction has null hypothesis “H0: Exp (B3) = 1.00”.

When we say that there is a "treatment by histology interaction", we are saying that the treatment effect varies significantly between the subgroups, i.e. that Exp (B3) differs significantly from 1.00.

The Interaction Test

Treatment-by-Histology InteractionsTreatment-by-Histology Interactions

Pemetrexed/Cisplatin vs Gemcitabine/Cisplatin

Pemetrexed versus Placebo*

Efficacy Parameter

Nonsquamous

n=1252

Squamous

n=473

Nonsquamous

n=482

Squamous

n=181

PFS adjusted HR (95% CI)

0.95

(0.84, 1.06)

1.36

(1.12, 1.65)

0.47

(0.37, 0.60)

1.03

(0.71, 1.49)THI HR

(95% CI)

p-value

0.70

(0.56, 0.88)

0.002

0.46

(0.30, 0.72)

0.001OS adjusted HR (95% CI)

0.84

(0.74, 0.96)

1.23

(1.00, 1.51)

0.66

(0.49, 0.88)

1.28

( 0.85, 1.93) THI HR

(95% CI)

p-value

0.69

(0.54, 0.87)

0.002

0.52

(0.31, 0.86)

0.011

* PFS data independently reviewed; OS data is preliminary

Median PFS, mos Median OS, mos

Pemetrexed plus Cisplatin versus Gemcitabine plus Cisplatin

Pem/Cis Gem/Cis p-value Pem/Cis Gem/Cis p-valueNonsquamous

N=12525.3 5.0 0.349 11.0 10.1 0.011

Squamous N=473

4.4 5.5 0.002 9.4 10.8 0.050

Pemetrexed vs Placebo*

Pem Placebo p-value Pem Placebo p-valueNonsquamous

N=4824.4 1.8 <0.00001 14.4 9.4 0.005

Squamous

N=1812.4 2.5 0.896 9.6 11.9 0.231

Efficacy by HistologyEfficacy by Histology

* PFS data independently reviewed (N=430 nonsquamous; N=151 squamous); OS data is preliminary

ConclusionsConclusions

These two large, randomized, phase III studies, provide evidence of significant interaction between NSCLC histology and a pemetrexed treatment effect, regardless of the control arm treatment

The consistency of these results across studies confirms that the treatment advantage for pemetrexed in patients with non-squamous histology is reproducible

Non-squamous histology is predictive of the improved efficacy of pemetrexed in patients with NSCLC