Cytel

Cytel Inc.
Clinical research services and trial technology

Type	Privately held
Industry	Clinical trial services and statistical software
Key people	Cyrus Mehta, Ph.D., Co-founder and President, Nitin Patel, Ph.D. Co-founder and CTO.
Products	East, LogXact, StatXact, SiZ, Compass
Employees	170 worldwide
Website	www.cytel.com

Headquartered in Cambridge, MA, USA, Cytel Inc. provides clinical trial design services and specialized statistical applications primarily for the biotech and pharmaceutical research markets.

Cytel specializes in adaptive trials – a rapidly emerging type of randomized clinical trial that allows modifications of ongoing trials while aiming to preserve the statistical validity and integrity of the study. Based on either frequentist or Bayesian statistics, adaptive trial designs are being increasingly accepted by government regulatory agencies including the United States Food and Drug Administration, EMEA, and MHRA in both early and later stage clinical studies.

As of January 2010, the company claims 47 of the top 50 biopharmaceutical companies as users of its software to design, simulate and analyze trials.

Background

Company founders Cyrus Mehta, Ph.D. and Nitin Patel, Ph.D. are amongst the pioneering statisticians credited for developing the underlying statistical methods behind so-called “flexible” designs: group sequential and adaptive trials.

Cytel statisticians have collectively published over 120 papers in peer-reviewed statistical and medical journals, including The Journal of the American Statistical Association, The Journal of Biopharmaceutical Statistics, Biometrics, Biometrika, Circulation and Statistics in Medicine.

Clinical Research Services

• Clinical development program planning
• Early and late stage trial design
• Drug / clinical supply forecasting
• Randomization and IVRS schemes
• Statistical Analysis Plan (SAP) development
• Statistical Programming for Tables, Listings, Graphs
• Interim and final data analysis
• DMC / DSMB organization, independent statistician representation
• Medical writing
• Migration of clinical data to CDISC standards
• Trial safety and efficacy data Integration
• Common Technical Document (CTD) preparation
• Integrated data analysis

Products

East

Clinical trial statistical software for the design, simulation and monitoring of adaptive, group sequential and fixed sample size trials. As of 2009, East 5 is in use at over 140 pharmaceutical and biotechnology companies, research centers and regulatory agencies including the FDA’s CDER, CBER and CDRH divisions.

First introduced by the Cytel Software Corporation in 1995 as “East DOS”, the name is derived from the benefit of “Early stopping” a trial due to futility: a failure of the tested treatment to demonstrate significant improvement over an existing treatment and/or placebo.

Compass

Largely the result of a co-development project with a top 3 pharmaceutical company, Compass is used by biostatisticians and clinicians to plan and design earlier stage adaptive clinical trials (traditionally known as phase 1 human tolerance and phase 2 dose-selection studies).

Compass is the first commercially-offered adaptive trial composition software with both frequentist and Bayesian methods. Other key capabilities include R code integration, trial simulation compute engines, plus various tables, charts and graphs to visualize and communicate trial design attributes.

SiZ

SiZ provides a more modern clinical trial planning approach by combining design, simulation and comparison of trial options and study data analysis within a single interface. SiZ is primarily meant for fixed sample size designs including non-inferiority and equivalence designs.

StatXact

Statistical software based on the exact branch of statistics used for small-sample categorical and nonparametric data problem solving. Utilized by statisticians and researchers in all fields of study, the StatXact package now offers 141 different statistical tests and procedures.

Initially offered in 1989 as StatXact DOS, StatXact 8 was released in 2007. The StatXact PROCs variant integrates with the popular SAS statistical software.

LogXact

A logistic regression predictive modeling software package suited particularly to cases involving small samples and / or missing data. Logistic regression is used extensively in the medical and social sciences as well as marketing applications to predict subject behavior.

First made available in 1996 under the name LogXact Turbo, LogXact 8 was introduced in 2007. The LogXact PROCs variant integrates with the popular SAS statistical software.

Locations

• Corporate headquarters, Cambridge, MA, USA
• Development and programming, Philadelphia, PA, USA
• Cytel India operations, Pune, INDIA
• Offices in Paris, FRANCE and San Jose, CA USA

References

Trial Design Methodology Development
1. Sample Size Re-estimation for Adaptive Sequential Design in Clinical Trials Ping Gao, Cyrus R. Mehta and James H. Ware
Journal of Biopharmaceutical Statistics, 2008

2. Exact Confidence Bounds Following Adaptive Group Sequential Tests
Werner Brannath, Cyrus R. Mehta and Martin Posch
Biometrics Journal, 2008 (a published reference of East by leading Pfizer statisticians)

3. Planning for an Adaptive Design: a Case Study in COPD (Chronic Obstructive Pulmonary Disease)
http://www.ncbi.nlm.nih.gov/pubmed/17080770

Statistical Methodology Development
1. Confidence Interval of the Difference of Two Independent Binomial Proportions Using Weighted Profile Likelihood
http://www.informaworld.com/smpp/content~content=a791517658

Vivek Pradhan (section a) and Tathagata Banerjee (sections b,c) Section a - Cytel Inc, Cambridge, Massachusetts, USA. Section b - Indian Institute of Management, Ahmadabad, India. Section c - Department of Statistics, University of Calcutta, Kolkata, India.
Communications in Statistics - Simulation and Computation, 2008

2. Small-sample Comparisons of Confidence Intervals for the Difference of Two Independent Binomial Proportions
Thomas J. Santner, Vivek Pradhan, Pralay Senchaudhuri, Cyrus R. Mehta, and Ajit Tamhane
Computational Statistics & Data Analysis 51 (2007) 5791 – 5799, August, 2007

3.Statistical Techniques for Comparing Measurers and Methods of Measurement: A Critical Review
John Ludbrook, Univ. of Melbourne, Parkville, Victoria, Australia.
Clinical and Experimental Pharmacology and Physiology (2002) 29, pp 527-536 + addendum

Logistic Methodology Development
1. Bias Reduction and a Solution for Separation of Logistic Regression with Missing Covariates Tapabrata Maiti a, Vivek Pradhan b a Department of Statistics and Probability, Michigan State University, East Lansing, MI 48824, USA b Cytel Inc, Cambridge, MA 02139, USA Biometrics Journal, 2008

2. Efficient Monte Carlo Methods for Conditional Logistic Regression
Cyrus R. Mehta, Nitin R. Patel and Pralay Senchaudhuri
Cytel Software Corporation Journal of the American Statistical Association, March 2000, Vol. 95, No. 449, Theory and Methods, pp 99-108