, according to an investigation published Oct. 9 in . Using sex-specific cut scores to define verbal memory impairment improves diagnostic accuracy and “may result in earlier detection of memory impairment in women and avoid false diagnoses in men,” wrote , assistant project scientist in psychiatry at the University of California, San Diego, and colleagues.
A diagnosis of aMCI generally requires a verbal memory deficit. Ample research demonstrates a female advantage on verbal memory tests, but normative data for these tests usually do not adjust for sex. Dr. Sundermann and colleagues previously showed that among men and women with aMCI and similar disease burden, women perform better on tests of verbal memory. Given these results, the investigators initiated a new study to test the hypothesis that using sex-specific norms and cut scores to identify memory impairment improves the accuracy of aMCI diagnosis, compared with non–sex-specific norms and cut scores.
An examination of ADNI data
Dr. Sundermann’s group extracted cross-sectional data from the(ADNI) database in October 2016. They included participants without dementia for whom neuropsychologic and Alzheimer’s disease pathologic marker data were available at baseline. They excluded patients with a non-aMCI diagnosis based on typical and sex-specific criteria.
The researchers’ primary outcome was the Rey Auditory Verbal Learning Test (RAVLT). They also determined the presence or absence of the APOE e4 allele for each participant. Biomarker outcomes included the CSF ratio of hyperphosphorylated tau (p-tau) to beta-amyloid (A-beta), and cortical A-beta deposition.
Dr. Sundermann and colleagues applied the Jak/Bondi actuarial neuropsychologic diagnostic method to baseline data. This method relies on six neuropsychologic tests, including the RAVLT Learning and Delayed Recall. They subsequently derived two sets of normative data for the RAVLT outcomes in a normative sample of 1,620 patients enrolled in the Mayo Clinic Study of Aging (MCSA). The latter patients were considered normal controls at baseline and at least two follow-up visits at 15 months apart. One set of normative data was specific for age and education, and the other was specific for age, education, and sex. Dr. Sundermann’s group next applied the typical Jak/Bondi method and the sex-specific Jak/Bondi method to all ADNI participants’ data.
Biomarker analysis supported the hypothesis
The researchers included 985 participants (453 women) in their final sample. Approximately 94% of the population was white. Mean age was 72.9 years, and mean education duration was 16.3 years. Overall, women had a significantly lower mean age (71.9 years vs. 73.6 years), significantly fewer mean years of education (15.7 years vs. 16.7 years), and a significantly higher mean Mini-Mental State Examination score (28 vs. 28.1) compared with men. Compared with men’s scores, women’s scores on the RAVLT Learning (mean 42.3 vs. mean 35.6) and Delayed Recall (mean 6.2 vs. mean 4.5) were significantly higher.
When Dr. Sundermann and colleagues used typical cut scores, the frequency of aMCI diagnosis was significantly higher in men. Using sex-specific cut scores eliminated this sex difference, however. Among men, 184 (35%) were categorized as true positive, 293 (55%) as true negative, and 55 (10%) as false positive. No men were categorized as false negative. Among women, 120 (26%) were categorized as true positive, 288 (64%) as true negative, and 45 (10%) as false negative. No women were categorized as false positive.
The likelihood of cortical amyloid positivity in false negative women was 3.6 times greater than in true negative women but did not differ from that in true positive women. The likelihood of positivity for the CSF p-tau/A-beta ratio in false negative women was more than two times higher than in true negative women but did not differ from that in true positive women. The likelihood of having an APOE e4 allele in false negative women was almost fivefold higher than in true negative women but did not differ from that in true positive women.
The likelihood of cortical amyloid positivity in false positive men was less than that in true positive men (odds ratio [OR], 0.45) but did not differ from that in true negative men. The likelihood of positivity for CSF p-tau/A-beta ratio in false positive men was significantly less than in true positive men (OR, 0.47) but did not differ from that in true negative men. The likelihood of having the APOE e4 allele in false positive men was lower than that in true positive men (OR, 0.63) and higher than that in true negative men (OR, 1.50), but not significantly.
Results have implications for treatment
“If these results are confirmed, they have vital implications,” Dr. Sundermann said in a press release. “If women are inaccurately identified as having no problems with memory and thinking skills when they actually have MCI, then treatments are not being started, and they and their families are not planning ahead for their care or their financial or legal situations. And for men who are inaccurately diagnosed with MCI, they can be exposed to unneeded medications along with undue stress for them and their families.”
Among the limitations that the investigators acknowledged was the study’s cross-sectional, rather than longitudinal, design. In addition, the ADNI population that the researchers examined is a convenience sample of predominantly white and well-educated volunteers. The results therefore may not be generalizable to the broader U.S. population, wrote the authors.
Grants from the National Institutes of Health funded the study. Several of the investigators reported receiving honoraria from various pharmaceutical companies such as Mylan. One investigator sits on the editorial board for Neurology.