R.Gomeni, PharmacoMetrica, La Fouillade, France

Date: Wednesday, October 22, 2014

Location: Mez, 5410 Page Road, Durham, NC 27003

- Reception/Networking Session 5:30-6:30 pm
- Scientific Discussion: 6:30 - 7:30 pm
- Dinner 7:30-8:45 pm
- Closing Remarks 8:45 – 9:00 pm

Discussion and Dinner

ISoP Member               $45

Student Member          $35      

Non-ISoP Member         $55

Discussion Only
ISoP Member               $25

Student Member          $15      

Non-ISoP Member      $30

 Not yet a member?  Visit and join ISoP before you register for this event.

Click here to register

BACKGROUND:  Enrichment strategies have been recommended as an effective methodology for improving the efficiency of drug development [1]. Three types of enrichment are considered:

  1. Strategies to decrease heterogeneity or “noise reduction”: Excluding patients whose disease or symptoms improve spontaneously or whose measurements are highly variable, For example, placebo response has been recognized as a major confounding factor (i.e. noise) for detecting drug effect for disease where there is a large placebo response. As a consequence, strategies for detecting and controlling the expected level of placebo response could substantially improve the signal detection ability of a clinical trial.  
  2. Prognostic enrichment strategies: Choosing patients with a greater likelihood of having a disease-related endpoint events or a substantial worsening in condition. To demonstrate the efficacy of a drug, the study population should have a reasonable number of these events or a predefined disease severity level otherwise it would not be possible to show drug efficacy as there is nothing to improve. 
  3. Predictive enrichment strategies: Choosing patients more likely to respond to the drug due to their physiology or disease characteristic.  Such selection can lead to a larger effect size and permit use of a smaller study population.

OBJECTIVES: Describes enrichment strategies for a) decreasing heterogeneity in a patient population and b) for implementing prognostic enrichment in the context of randomized controlled trials. Presents study design options for the different strategies, including advantages and disadvantages of the various designs.

METHODS: Use placebo response models to build prior knowledge on the expected time-course of placebo response and disease progression models to predict the individual disease trajectory deterioration shape. Develop probabilistic models for classifying a subject as placebo responder or slow/fast disease progressor. For each subject enrolled in a new trial, apply a Bayesian framework for combining early observations on disease with prior knowledge on disease deterioration in combination with probabilistic models to predict the individual disease status at study-end. Use this information to implement a study lead-in design where information on placebo response or predicted disease severity collected in the lead-in phase is used to randomize patients for the active treatment trial.

RESULTS: Two case-study are presented to evaluate how short-term placebo response can predict the long-term observations in Respiratory [2] and in CNS [3] therapeutic areas. The results indicated that the early (week 2) FEV1 placebo response was significantly predictive of the placebo response at week 12 and the HAMD placebo response at week 2 was significantly predictive of the HAMD score at week 8  with a satisfactory predictability defined by the area under the receiver operating characteristic curve (ROC) >75%.  Two case-study are presented for disease progression in Alzheimer’s [4] and ALS [5] diseases. The model was able to anticipate the disease severity in Alzheimer’s disease at 6 months and in the Amyotrophic Lateral Sclerosis disease at 6 and 12 months using data collected in the first months of treatment with a good predictive performances (ROC>75%).

These data were used to define a study design based on a lead-in period during which the early-collected information can be used to:  a) select a population with controlled level of placebo response and therefore more capable of responding to the treatment and b) decrease the heterogeneity in the study population, i. e. selecting patients who have a homogenous level of disease.

CONCLUSION: The results of the analysis demonstrated that the model-based approach together with placebo response and disease progression models can be successfully used to control the level of heterogeneity in a patient population enrolled in a clinical trial and to implement prognostic enrichment study design strategies. Finally, the predicted benefits associated with the selection of the fast disease progressor patients in the treatment phase indicated significant increase in the power:  fewer patients required with an increase probability of detecting a treatment effect.                   


[1]   Guidance for Industry Enrichment Strategies for Clinical Trials to Support Approval of Human Drugs and Biological Products. Draft Guidance. December 2012.

[2]   S. Yang, R. Gomeni, M. Beerahee. Does Short-Term Placebo Response Predict the Long-Term Observation? Meta-Analysis on Forced Expiratory Volume in 1 Second From Asthma Trials. The Journal of Clinical Pharmacology, 2014, In press. DOI: 10.1002/jcph.329.

[3]   R. Gomeni and E. Merlo-Pich. Bayesian modeling and ROC analysis to predict placebo responders using clinical score measured in the initial weeks of treatment in depression trials. Br J Clin Pharmacol. May;63(5):595-613, 2007.

[4]   R. Gomeni and M. Simeoni. Predicting long term response using short term data in Alzheimer’s Disease ASCPT 2010 Annual Meeting, March 17-20, 2010. Atlanta, GA, USA.

[5]   R. Gomeni R, M. Fava. The Pooled Resource Open-Access ALS Clinical Trials Consortium. Amyotrophic lateral sclerosis disease progression model. Amyotroph Lateral Scler. Frontotemporal Degener. 2014 Mar;15(1-2):119-29.