Assessing Statistical Methods for Estimating the Cerebrospinal Fluid Early Fungicidal Activity as an Endpoint for

Monica Fuszard1, David R Boulware2, Biyue Dai1

  • 1Division of Biostatistics and Health Data Science, School of Public Health, University of Minnesota, Minneapolis, Minnesota, USA.

Abstract

Insights

The 2-step approach is preferred for estimating early fungicidal activity (EFA) in cryptococcal meningitis trials, especially when early culture sterility is common. Statistical models significantly impact EFA estimates, necessitating careful consideration for cross-study comparisons.

Area of Science:

  • Infectious Diseases
  • Clinical Trials
  • Statistical Modeling

Background:

  • Cryptococcal meningitis is a critical central nervous system infection.
  • Early fungicidal activity (EFA) is a key endpoint in phase 2 antifungal trials for cryptococcal meningitis.
  • EFA measures the rate of *Cryptococcus* yeast clearance from cerebrospinal fluid (CSF).

Purpose of the Study:

  • To systematically compare the 2-step linear regression and linear mixed model approaches for estimating EFA in cryptococcal meningitis.
  • To evaluate the performance and operating characteristics of these statistical methods.
  • To provide guidance on appropriate statistical methods for analyzing EFA data.

Main Methods:

  • Conducted a systematic literature review of EFA estimation methods in cryptococcal meningitis trials.
  • Performed simulation studies to assess method performance under various scenarios.
  • Applied both the 2-step approach and linear mixed models to phase 2 trial data.

Main Results:

  • The 2-step and mixed model approaches yielded discrepant EFA estimates.
  • The 2-step approach showed steeper estimates and larger treatment effects.
  • Linear mixed models produced smaller standard errors but were biased toward the null with early culture sterility.

Conclusions:

  • The 2-step approach is recommended for EFA estimation in cryptococcal meningitis studies, particularly when early culture sterility is observed.
  • EFA estimates and comparisons across studies must account for the statistical models and data preprocessing used.
  • Clarification of statistical methods is crucial for accurate interpretation of EFA in clinical trials.