Evaluating ELISPOT summary measures with criteria for obtaining reliable estimates

Area of Science:

• Immunology
• Statistical Modeling
• Biostatistics

Background:

• The ELISPOT assay quantifies T lymphocyte cytokine secretion after antigen stimulation.
• Assay endpoints, spot-forming cells (SFC), are typically estimated using arithmetic means or linear regression.
• Accurate statistical methods are crucial for reliable ELISPOT assay interpretation.

Purpose of the Study:

• To compare statistical modeling approaches for summarizing ELISPOT assay results.
• To evaluate the validity of different methods under various experimental conditions.
• To identify optimal statistical strategies for analyzing T cell responses.

Main Methods:

• Comparative analysis of statistical models including simple mean, linear regression, random effects models, and nonlinear models.
• Utilized data from the Pediatric AIDS Clinical Trial Group (PACTG) study 299.
• Conducted a simulation study to assess method performance under controlled conditions.

Main Results:

• Simple arithmetic mean is appropriate for ELISPOT assays conducted at a single effector cell concentration.
• Normalizing simple means across different concentrations is not statistically valid.
• Random effects models outperform simple means when within-subject variance is high.
• Nonlinear models provide more accurate estimation than linear regression when linearity assumptions are violated.

Conclusions:

• The choice of statistical model for ELISPOT assays depends on experimental design and data characteristics.
• Nonlinear models are recommended when data is collected over a range of effector cell concentrations.
• Collecting data across multiple concentrations allows for reassessment of optimal cell concentration post-hoc.