SEX-SPECIFIC COMPOSITE SCALES FOR LONGITUDINAL STUDIES OF INCIPIENT ALZHEIMER’S DISEASE

Abstract

The impact of incipient Alzheimer's disease (AD) on cognition and function differs in men and women. This may be relevant to cohort studies and clinical trials investigating the earliest stages of disease, where sex-specific global scales may be more sensitive, and statistically powerful, measures of progression. We derived optimal composite assessment scales from component scales available in the AD Neuroimaging Initiative (ADNI) cohort database restricting to cognitively normal and MCI subjects by CSF Aβ biomarker positive for early AD. Maximally sensitive composites were constructed separately for men and women using standard formulas to determine the weighted sum of component measures that has the greatest mean to standard deviation ratio (MSDR) and therefore greatest statistical power in longitudinal studies [Ard et al., Pharmaceutical Statistics. 2016]. We considered composites constructed from all possible combinations of the Clinical Dementia Rating Scale Sum of Boxes (CDR-sob) and 8 psychometric scales available in ADNI. For reference, we compared the statistical power of the composites with the ADNI-PACC as outcome measures in a clinical trial. The ADNI-PACC is an approximation of the Prodromal Alzheimer's Cognitive Composite which sums component measures without regard to sex. Twenty four month data were available for 54 men (age 54.6 to 87.8; mean MMSE 27.3) and 84 women (age 55.2 to 85.6; mean MMSE 27.4). Improvement in MSDR was negligible after a total of 5 components was included in the composites (five component scales were both parsimonious and near maximal in terms of statistical power). The CDR-sob was the most sensitive measure in both the men and women. Beyond this, the sex-specific composites differed. The male composite included the Clock Drawing, Rey-AVLT Immediate Learning and 30 Minute Delayed Recall, and Logical Memory Delayed Recall, and the female composite included the Logical Memory Immediate Recall, Boston Naming, Animal Category Fluency, and a timed test calculated as the Trails B minus Trials A. In terms of statistical power, the sex-specific composites consistently out-performed composites constructed without regards to sex (not shown) as well as the ADNI-PACC (Figure).