Abstract
The Face Name Associative Memory Exam (FNAME) was introduced into the NIH Toolbox as part of the ARMADA study and establishes normative data for diverse participants, ages 64 to 85+, and proposes cutoff scores between biomarker positive versus negative (+/−) groups. The FNAME was administered to 257 participants across the clinical spectrum with 122 having amyloid biomarkers. Linear regression explored the association between demographics and FNAME and between amyloid (+/−) groups. Receiver operating characteristic curves (ROC) identified performance thresholds that best discriminated between biomarker (+/−) individuals. Lower FNAME scores occurred in males, older ages, Black/African Americans, Hispanics, and biomarker‐positive participants. ROC analyses demonstrated acceptable accuracy (0.73 to 0.77) but only when combined with clinical status. The diagnostic discrimination of amyloid positivity was acceptable but not excellent, suggesting the FNAME may be a better screening indicator of clinical status rather than amyloid deposition in cognitively normal individuals. Normative data are provided.
Keywords: Alzheimer's disease, cognition, dementia, mild cognitive impairment, neuropsychology, NIH Toolbox
1. INTRODUCTION
Aging increases our vulnerability to cognitive decline and dementia. 1 Differentiating normal age‐related cognitive change from the earliest stages of Alzheimer's disease (AD) is even more relevant as treatments become available. 2 While paper and pencil‐based assessments are traditionally the gold standard, there is a growing movement toward developing digital tests that might capture subtle or more nuanced cognitive performance 3 , 4 to aid in more refined diagnostic discrimination. Many digital tools are still experimental and yet to be validated for clinical use. Some of these digital tools were developed to identify the earliest cognitive changes associated with AD biomakers. 3 , 5 , 6 Some were created to test the feasibility of remote testing 7 , 8 and some were designed to enable frequent measurement for the purpose of gaining more reliable information on intraindivual change over time. 5 , 9
The multisite study, Advancing Reliable Measurement in Alzheimer's Disease and Cognitive Aging (ARMADA), was launched to validate an iPad administration of the NIH Toolbox for Assessment of Neurological and Behavioral Function (NIHTB) in older individuals ages 64 to 85+ across diverse ethnic and racial groups. 10 The Face Name Associative Memory Exam (FNAME), an associative memory test sensitive to the earliest biomarker changes in AD, 6 was introduced into ARMADA at year 2. Studies have indicated that performance on the FNAME correlates with other measures of episodic memory, 6 , 11 , 12 has been translated into various languages, 13 , 14 , 15 , 16 and has been shown to differentiate clinically normal (CN) individuals from those with diagnostic designations of subjective cognitive decline 17 , 18 and mild cognitive impairment (MCI). 17 Furthermore, the FNAME differentiated cognitively asymptomatic CN individuals with and without biomarker evidence of AD. 6 , 11 , 12 , 13
The purpose of this study was to establish baseline performance data (ie, means and standard deviations [SDs]) on the FNAME iPad version of the NIHTB in diverse (White, Black/African American, and Hispanic) participants 64 to 85+ and to determine if biomarker status was related to FNAME performance. Normative scores stratified by age, sex, education, and race are provided for participants who were CN, for those with MCI/AD, and for those with positive/negative amyloid biomarkers. We also provide more filtered normative data stratified by age ranges for sex and race. Performance thresholds of FNAME component scores are proposed across the biomarker sample that best discriminated between positive and negative groups.
2. METHODS
2.1. Participants
ARMADA study participants were recruited from nine academic sites across the United States including Mayo Jacksonville, University of Michigan, University of Wisconsin‐Madison, University of Pittsburgh, Emory University, Massachusetts General Hospital/Harvard Medical School, Northwestern University (study lead site), University of California San Diego, and Columbia University. Across all of these sites, a total of 257 participants completed the FNAME at a single timepoint. Participants ranged in age from 64 to 94 and included people who were CN (n = 181), had MCI (n = 71), or were diagnosed with early dementia due to AD (n = 5). Cognitive status was based on a global Clinical Dementia Rating (CDR) score (CN: CDR = 0.0, MCI: CDR = 0.5, AD: CDR = 1.0 or higher) administered by trained clinicians at each ARMADA site, as part of the Alzheimer's Disease Research Centers’ Uniform Data Set or other NIH‐funded independent research protocols. We combined MCI and AD participants into one diagnostic group because clinical severity was relatively equivalent (mean CDR = 0.7, SD = 0.5). Participants who took the FNAME were divided into three age groups for normalization: 64 to 74 (n = 130), 75 to 84 (n = 85), and 85+ (n = 42). A subset (n = 122) of the 257 participants underwent positron emission tomography (PET) imaging (n = 85) or cerebrospinal fluid (CSF) collection (n = 37) for amyloid biomarkers. Biomarker participants included CN (n = 103), MCI (17) and AD (n = 2) participants. All participants took the FNAME as part of the NIHTB on an iPad, in the clinic, with a study coordinator assisting in test administration.
RESEARCH IN CONTEXT
Systematic Review: The authors reviewed the literature on PubMed for use of the FNAME in research and clinical contexts, as well as studies that provided normative data and diagnostic discrimination among diagnoses of cognitively normal, subjective cognitive decline, mild cognitive impairment, and Alzheimer's disease (AD), and among biomarkers of AD. Literature regarding the use of the FNAME adapted for different languages were reviewed. Relevant citations are provided.
Interpretation: This study provides norms for FNAME performance across three main age groups, 64 to 74, 75 to 84, and 85+, and demographic norms for sex, age, education, race, and ethnicity. Linear regressions explored the association between FNAME in biomarker and non‐biomarker groups. Receiver operating characteristic analyses demonstrated acceptable discriminatory accuracy (0.73 to 0.77) but only when combined with clinical status.
Future Directions: Future work will explore the longitudinal performance of FNAME, its ability to predict decline, and whether novel composite measures can be derived to better predict biomarker status.
2.2. Standard protocol approvals, registrations, and patient consents
The Institutional Review Boards at Mayo Jacksonville, University of Michigan, University of Wisconsin‐Madison, University of Pittsburgh, Emory University, Massachusetts General Hospital, Northwestern University, University of California San Diego, and Columbia University approved the ARMADA Study. Written informed consent was obtained from all participants prior to study procedures.
2.3. AD biomarker collection
Several ARMADA sites collected AD biomarkers via PET or lumbar puncture using previously described methods. 10 Each site was asked to designate cut‐points for both PET and CSF data as dichotomous values (amyloid negative or positive) based on site‐specific criteria. This method allowed for the harmonization of various collection procedures, tracers, and biomarker type for statistical analyses. CSF assays used Luminex or Elisa methods and cutoffs were determined as an amyloid‐beta 42/tau ratio. The PET imaging ligands were either Pittsburgh Compound B or florbetapir and cutoffs were derived from either a distribution volume ratio or a standardized uptake value ratio in aggregate cortical regions for frontal association, precuneus, posterior cingulate, and lateral parietal areas using cerebellar cortex as the reference tissue. Since it was not possible to control for time lapse between site biomarker collection and ARMADA participation, a time limit was set at 24 months for negative results and no time limit for positive results. 10
2.4. FNAME (Face Name Associative Memory Exam)
The FNAME test was initially developed by Rentz et al. 6 and consists of one encoding phase and three memory phases (Face Recognition, Name Recall, and Name Recognition) see Figure 1.
FIGURE 1.
Examples of each phase of the FNAME as seen by the participants on the iPad.
During the encoding phase, twelve face‐name pairs were presented serially. To ensure that the participants were paying attention to learning the face‐name pairs, the participants chose a button on the screen stating that the name either “fits” or “doesn't fit” the face. This initial encoding phase is not scored. Following a 25 to 30‐min delay during which other NIHTB tests were completed, the memory phase was tested. Participants were asked to choose which face they learned previously among two distractors of matching age, sex, and race (Face Recognition). Then, the previously learned face was presented with a keyboard on the screen and the participant was asked to select the first letter of the name previously paired with the face (Name Recall). Finally, the target face was presented with three names underneath (target name, a same‐sex name with a different face previously presented, and an age‐ and sex‐matched foil name). The participant was asked to select the correct name (Name Recognition). Participants were scored on accuracy (correct Face Recognition, Name Recall, and Name Recognition) with a maximum score of 12 points for each component. A higher score indicates higher accuracy. A total Summary Score was calculated by adding together each component score for a total of 36 points.
Implementing the FNAME into the NIHTB for the ARMADA study required optimizing the image resolution and ensuring proper sequencing within the battery so as to not overlap with any other delayed memory tests.
2.5. Statistical analyses
The demographic descriptives included the whole cohort (n = 257) and amyloid biomarker group (n = 122) and CN‐only participants within each group. For the continuous variables of age, education, and FNAME component scores, we calculated means and standard deviations. For categorical variables of sex, race, ethnicity, and amyloid status, we report frequency and percent. To determine whether there were demographic differences between those who underwent biomarker testing (n = 122) from those who did not (n = 135), a Wilcoxon rank‐sum test and Pearson's chi‐squared test was used. Post hoc tests were performed using row‐wise Fisher's exact test and were adjusted for multiple comparisons with the false discovery rate method in order to determine whether those with biomarker information differed by race or sex.
Linear regression was used to determine the association between demographics and performance on the FNAME including the Summary Score, Face Recognition, Name Recall, and Name Recognition with sex, age, education, race, ethnicity, and diagnostic group. Analyses were performed across four groups: the whole sample of 257 participants; a subset of 181 CN‐only individuals; those with amyloid biomarkers (n = 122); and CN‐only participants with amyloid biomarkers (n = 103). For normative purposes, means and standard deviations were calculated for each FNAME component score across the four groups.
The next objective was to compare FNAME performance between amyloid‐positive versus ‐negative groups. Linear regressions, adjusting for sex, age, education, race, and diagnosis determined the association between amyloid status and FNAME scores. Ethnicity was excluded as a covariate because all participants with available amyloid data were non‐Hispanic. Independent t‐tests were used to determine the mean score differences on the FNAME between amyloid‐positive and ‐negative groups. Pearson correlations were used to determine the strength of the relationship between FNAME and amyloid status. Given that this was a multisite study, we reran a covariate sensitivity analysis controlling for site.
Receiver operating characteristic (ROC) curve analysis was used to identify thresholds at which the FNAME best discriminated between biomarker‐positive versus biomarker‐negative participants. The ROC approach was conducted by first fitting a logistic model regressing the indicator for biomarker positivity versus negativity on each FNAME score to estimate the area under the curve (AUC) and predicted probability of being biomarker positive. We ran separate models for each FNAME component score. We then used the predicted probabilities, number of true and false positives and negatives, sensitivity, and specificity output from logistic models run as input to the %rocplot macro in SAS statistical software to generate ROC plots. 19 The discriminatory ability of the AUC value was evaluated based on the following standard criteria: ≥ 0.70 = acceptable; ≥ 0.80 = excellent; ≥ 0.90 = outstanding. 20 The threshold for discriminating biomarker positivity from this analysis were defined as the FNAME scores suggested by Youden's index, or the sum of sensitivity and specificity − 1. 21 We adjusted the threshold suggested by Youden's index for unequal proportions of biomarker‐positive and ‐negative patients. 22
For all analyses, two‐tailed P < 0.05 was considered statistically significant. All data were analyzed using R except for the ROC analyses, which used SAS.
3. RESULTS
3.1. Demographics
Table 1 provides a summary of the demographics by age ranges for the whole sample (n = 257), CN participants in the whole sample (n = 181), and those participants with amyloid biomarkers across all diagnostic groups (n = 122) and the CN group only (n = 103).
TABLE 1.
Demographics by age group.
| Whole group (N = 257) | CN group (N = 181) | Amyloid group (N = 122) | Amyloid CN group (N = 103) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Component | 64 to 74 N = 130 a | 75 to 84 N = 85 a | 85+ N = 42 a | 64 to 74 N = 99 a | 75 to 84 N = 55 a | 85+ N = 27 a | 64 to 74 N = 77 a | 75 to 84 N = 34 a | 85+ N = 11 a | 64 to 74 N = 67 a | 75 to 84 N = 28 a | 85+ N = 8 a |
| Education | 15.74 (3.73) | 14.22 (5.21) | 16.10 (3.82) | 16.26 (3.08) | 16.00 (3.23) | 17.00 (2.15) | 17.14 (2.19) | 17.24 (2.47) | 17.91 (1.70) | 17.10 (2.15) | 17.39 (2.39) | 18.00 (1.51) |
| Sex | ||||||||||||
| Female | 82 (63%) | 59 (69%) | 20 (48%) | 67 (68%) | 43 (78%) | 13 (48%) | 46 (60%) | 27 (79%) | 3 (27%) | 42 (63%) | 23 (82%) | 2 (25%) |
| Male | 48 (37%) | 26 (31%) | 22 (52%) | 32 (32%) | 12 (22%) | 14 (52%) | 31 (40%) | 7 (21%) | 8 (73%) | 25 (37%) | 5 (18%) | 6 (75%) |
| Race | ||||||||||||
| White | 83 (64%) | 45 (53%) | 33 (79%) | 63 (64%) | 32 (58%) | 23 (85%) | 62 (81%) | 24 (71%) | 11 (100%) | 53 (79%) | 20 (71%) | 8 (100%) |
| Black or African/American | 33 (25%) | 21 (25%) | 5 (12%) | 27 (27%) | 16 (29%) | 4 (15%) | 15 (19%) | 10 (29%) | 0 (0%) | 14 (21%) | 8 (29%) | 0 (0%) |
| Other | 14 (11%) | 19 (22%) | 4 (9.5%) | 9 (9.1%) | 7 (13%) | 0 (0%) | ||||||
| Ethnicity | ||||||||||||
| Not Hispanic | 113 (87%) | 61 (72%) | 36 (86%) | 89 (90%) | 43 (78%) | 26 (96%) | 77 (100%) | 34 (100%) | 11 (100%) | 67 (100%) | 28 (100%) | 8 (100%) |
| Hispanic | 17 (13%) | 24 (28%) | 6 (14%) | 10 (10%) | 12 (22%) | 1 (3.7%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) |
| Amyloid Status | ||||||||||||
| Negative | 67 (87%) | 27 (79%) | 7 (64%) | 63 (94%) | 25 (89%) | 6 (75%) | 67 (87%) | 27 (79%) | 7 (64%) | 63 (94%) | 25 (89%) | 6 (75%) |
| Positive | 10 (13%) | 7 (21%) | 4 (36%) | 4 (6.0%) | 3 (11%) | 2 (25%) | 10 (13%) | 7 (21%) | 4 (36%) | 4 (6.0%) | 3 (11%) | 2 (25%) |
| Diagnosis | ||||||||||||
| CN | 99 (76%) | 55 (65%) | 27 (64%) | 99 (100%) | 55 (100%) | 27 (100%) | 67 (87%) | 28 (82%) | 8 (73%) | 67 (100%) | 28 (100%) | 8 (100%) |
| MCI/AD | 31 (24%) | 30 (35%) | 15 (36%) | 10 (13%) | 6 (18%) | 3 (27%) | ||||||
Abbreviations: AD, Alzheimer's disease; CN, cognitively normal; MCI, mild cognitive impairment.
Mean (SD); n (%).
Table 2 provides the means and standard deviations of the FNAME component scores stratified by age ranges, sex, race, ethnicity, and amyloid biomarker status for the above‐named four groups. To aid in clinical interpretation, Tables S1‐4 provide more refined normative data for each FNAME component score further stratified by age range for race and sex across all four groups.
TABLE 2.
FNAME scores by demographics
| Whole Group (N = 257) | CN Group (N = 181) | |||||||
|---|---|---|---|---|---|---|---|---|
| Demographic | Summary Score a | Face Recognition a | Name Recall a | Name Recognition a | Summary Score a | Face Recognition a | Name Recall a | Name Recognition a |
| Sex | ||||||||
| Female | 19.7 (7.02) | 9.52 (2.54) | 2.98 (2.7) | 7.2 (2.84) | 21.86 (6.12) | 10.21 (1.9) | 3.59 (2.76) | 8.06 (2.48) |
| Male | 17.41 (6.46) | 8.59 (2.91) | 2.19 (2.08) | 6.62 (2.69) | 20.05 (5.67) | 9.59 (2.07) | 2.81 (2.25) | 7.66 (2.5) |
| Age group | ||||||||
| 64 to 74 | 20.83 (6.92) | 9.74 (2.51) | 3.36 (2.68) | 7.73 (2.79) | 22.99 (5.83) | 10.47 (1.73) | 3.96 (2.74) | 8.56 (2.45) |
| 75 to 84 | 17.42 (6.74) | 8.99 (2.86) | 2.25 (2.41) | 6.19 (2.71) | 19.95 (6.2) | 9.89 (2.07) | 2.96 (2.58) | 7.09 (2.5) |
| 85+ | 15.55 (4.99) | 7.81 (2.55) | 1.45 (1.25) | 6.29 (2.37) | 17.74 (3.88) | 8.56 (1.95) | 1.85 (1.26) | 7.33 (1.98) |
| Race | ||||||||
| White | 20.07 (6.82) | 9.45 (2.57) | 3.02 (2.66) | 7.59 (2.72) | 22.36 (5.77) | 10.19 (1.81) | 3.67 (2.76) | 8.49 (2.26) |
| Black/African American | 19.24 (6.1) | 9.61 (2.36) | 2.85 (2.19) | 6.78 (2.61) | 20.85 (5.23) | 10.13 (1.97) | 3.26 (2.19) | 7.47 (2.23) |
| Other | 12.86 (5.23) | 7.27 (3.11) | 0.92 (1.38) | 4.68 (2.12) | 14.62 (5.91) | 8.31 (2.44) | 1.19 (1.72) | 5.12 (2.75) |
| Ethnicity | ||||||||
| Not Hispanic | 20.11 (6.68) | 9.56 (2.52) | 3.09 (2.56) | 7.46 (2.74) | 22.3 (5.55) | 10.28 (1.8) | 3.7 (2.6) | 8.32 (2.28) |
| Hispanic | 13.17 (4.62) | 7.45 (2.91) | 0.85 (1.1) | 4.87 (1.92) | 14.3 (4.32) | 8.17 (2.19) | 0.91 (1.12) | 5.22 (2.13) |
| Amyloid Group (N = 122) | CN Amyloid Group (N = 103) | |||||||
|---|---|---|---|---|---|---|---|---|
| Demographic | Summary score a | Face recognition a | Name recall a | Name recognition a | Summary score a | Face recognition a | Name recall a | Name recognition a |
| Sex | ||||||||
| Female | 23.04 (6.18) | 10.46 (1.84) | 4.25 (2.74) | 8.33 (2.79) | 24.33 (5.19) | 10.75 (1.51) | 4.63 (2.68) | 8.96 (2.22) |
| Male | 19.48 (6.97) | 9.33 (2.96) | 2.85 (2.37) | 7.3 (2.8) | 21.92 (5.05) | 10.42 (1.65) | 3.36 (2.36) | 8.14 (2.18) |
| Age group | ||||||||
| 64 to 74 | 22.64 (6.48) | 10.14 (2.29) | 4.03 (2.71) | 8.47 (2.66) | 24.15 (5.12) | 10.67 (1.52) | 4.43 (2.64) | 9.04 (2.16) |
| 75 to 84 | 20.65 (7.29) | 9.97 (2.7) | 3.65 (2.67) | 7.03 (3.1) | 22.79 (5.7) | 10.68 (1.68) | 4.11 (2.7) | 8 (2.34) |
| 85+ | 18.36 (4.92) | 9.45 (2.02) | 1.82 (1.72) | 7.09 (2.34) | 20.38 (3.42) | 10.12 (1.55) | 2.38 (1.69) | 7.88 (1.81) |
| Race | ||||||||
| White | 21.86 (6.92) | 10.02 (2.48) | 3.77 (2.81) | 8.06 (2.84) | 23.81 (5.35) | 10.72 (1.44) | 4.28 (2.77) | 8.81 (2.26) |
| Black/African American | 21.08 (5.79) | 10.08 (1.98) | 3.52 (2.16) | 7.48 (2.79) | 22.27 (4.76) | 10.32 (1.94) | 3.82 (2.08) | 8.14 (2.08) |
| Amyloid status | ||||||||
| Negative | 22.99 (5.45) | 10.53 (1.63) | 4.03 (2.59) | 8.43 (2.41) | 23.61 (5.04) | 10.72 (1.45) | 4.21 (2.59) | 8.67 (2.24) |
| Positive | 15.48 (8.57) | 7.62 (3.67) | 2.24 (2.66) | 5.62 (3.54) | 22.22 (7.31) | 9.67 (2.35) | 3.89 (3.22) | 8.67 (2.24) |
Abbreviation: CN, cognitively normal.
Mean (SD).
Since biomarker information was only available for a subset of our participants, we tested whether there were demographic differences between those with (n = 122) and without (n = 135) biomarkers. We found that participants with biomarkers were younger and more educated and the frequency across race, ethnicity, and diagnosis differed between the groups. The biomarker group had a greater number of White participants and a smaller number of participants of other races. However, the frequency of Black/African American participants who underwent biomarker testing (20%) was not statistically different from those without biomarkers (25%). Those with biomarkers had a greater number of non‐Hispanic (100%) and CN participants (84%) compared to those without biomarkers (non‐Hispanic 65%; CN 58%). There were no sex differences between the groups (see Table S5).
3.2. Demographic differences in FNAME performance in the whole group
Regression analyses (see Table 3) revealed that Summary Score performances were statistically lower in males versus females (estimate = −0.11, P = 0.036), older age versus younger age (estimate = −0.24, P < 0.001), Black/African American versus White participants (estimate = −0.13, P = 0.011), and Hispanic versus non‐Hispanic participants (estimate = −0.23, P = 0.009). Those with MCI/AD also performed worse than those who were CN (estimate = −0.46, P = < 0.001). There was no association with education across any groups.
TABLE 3.
Standardized betas and 95% confidence intervals of linear regression models
| Whole group (N = 257) | CN Group (N = 181) | |||||||
|---|---|---|---|---|---|---|---|---|
| Model 1: Summary Score | Model 2: Face Recognition | Model 3: Name Recall | Model 4: Name Recognition | Model 5: Summary Score | Model 6: Face Recognition | Model 7: Name Recall | Model 8: Name Recognition | |
| Sex (Male) | −0.11 * (−0.21 to −0.01) | −0.10 (−0.21 to 0.01) | −0.11 (−0.22 to 0.00) | −0.07 (−0.17 to 0.04) | −0.16 * (−0.29 to −0.03) | −0.13 (−0.27 to 0.01) | −0.15 * (−0.29 to −0.01) | −0.12 (−0.25 to 0.02) |
| Age | −0.24 *** (−0.33 to −0.14) | −0.18 ** (−0.28 to −0.07) | −0.25 *** (−0.35 to −0.14) | −0.19 *** (−0.28 to −0.09) | −0.34 *** (−0.46 to −0.21) | −0.30 *** (−0.43 to −0.17) | −0.30 *** (−0.43 to −0.17) | −0.25 *** (−0.38 to −0.12) |
| Education | −0.05 (−0.19 to 0.10) | −0.02 (−0.19 to 0.14) | −0.08 (−0.24 to 0.09) | −0.02 (−0.18 to 0.13) | 0.03 (−0.12 to 0.19) | −0.02 (−0.19 to 0.15) | 0.05 (−0.12 to 0.22) | 0.05 (−0.11 to 0.21) |
| Race (Black/African American) | −0.13 * (−0.23 to −0.03) | −0.05 (−0.16 to 0.07) | −0.11 (−0.22 to 0.00) | −0.18 *** (−0.29 to −0.08) | −0.15 * (−0.29 to −0.02) | −0.07 (−0.21 to 0.07) | −0.11 (−0.25 to 0.03) | −0.21 ** (−0.34 to −0.07) |
| Race (Other) | −0.10 (−0.27 to 0.06) | −0.09 (−0.27 to 0.09) | −0.04 (−0.22 to 0.15) | −0.13 (−0.31 to 0.04) | −0.16 (−0.34 to 0.01) | −0.10 (−0.29 to 0.08) | −0.09 (−0.28 to 0.09) | −0.21 * (−0.40 to −0.03) |
| Ethnicity (Hispanic) | −0.23 ** (−0.41 to −0.06) | −0.14 (−0.34 to 0.05) | −0.29 ** (−0.49 to −0.09) | −0.18 (−0.36 to 0.01) | −0.31 ** (−0.50 to −0.11) | −0.29 ** (−0.50 to −0.08) | −0.25 * (−0.46 to −0.05) | −0.24 * (−0.44 to −0.04) |
| Diagnosis (MCI/AD) | −0.46 *** (−0.57 to −0.36) | −0.40 *** (−0.52 to −0.29) | −0.33 *** (−0.45 to −0.22) | −0.45 *** (−0.56 to −0.34) | ||||
| Amyloid Group (N = 122) | CN amyloid group (N = 103) | |||||||
|---|---|---|---|---|---|---|---|---|
| Model 9: Summary Score | Model 10: Face Recognition | Model 11: Name Recall | Model 12: Name Recognition | Model 13: Summary Score | Model 14: Face Recognition | Model 15: Name Recall | Model 16: Name Recognition | |
| Amyloid status (Positive) | −0.07 (−0.23 to 0.09) | −0.17 * (−0.34 to −0.01) | −0.01 (−0.20 to 0.19) | −0.01 (−0.18 to 0.16) | −0.01 (−0.20 to 0.19) | −0.16 (−0.36 to 0.05) | 0.00 (−0.20 to 0.21) | 0.09 (−0.11 to 0.29) |
| Sex (Male) | −0.23 ** (−0.38 to −0.09) | −0.22 ** (−0.37 to −0.07) | −0.22 * (−0.39 to −0.04) | −0.16 * (−0.31 to −0.01) | −0.26 * (−0.46 to −0.06) | −0.15 (−0.36 to 0.06) | −0.24 * (−0.44 to −0.03) | −0.23 * (−0.43 to −0.03) |
| Age | −0.14 * (−0.28 to 0.00) | 0.01 (−0.13 to 0.15) | −0.18 * (−0.35 to −0.02) | −0.17 * (−0.31 to −0.02) | −0.23 * (−0.42 to −0.03) | −0.07 (−0.28 to 0.13) | −0.20 * (−0.40 to −0.01) | −0.25 * (−0.44 to −0.05) |
| Education | 0.14 (0.00 to 0.28) | 0.20 ** (0.05 to 0.35) | 0.00 (−0.17 to 0.17) | 0.16 * (0.01 to 0.31) | 0.09 (−0.11 to 0.30) | 0.07 (−0.14 to 0.29) | 0.00 (−0.21 to 0.20) | 0.17 (−0.04 to 0.37) |
| Race (Black/African American) | −0.11 (−0.24 to 0.03) | −0.02 (−0.16 to 0.13) | −0.12 (−0.28 to 0.05) | −0.13 (−0.27 to 0.02) | −0.17 (−0.37 to 0.02) | −0.10 (−0.30 to 0.10) | −0.14 (−0.34 to 0.06) | −0.17 (−0.37 to 0.03) |
| Diagnosis (MCI/AD) | −0.55 *** (−0.70 to −0.39) | −0.47 *** (−0.63 to −0.30) | −0.36 *** (−0.55 to −0.17) | −0.56 *** (−0.72 to −0.40) | ||||
Abbreviations: AD, Alzheimer's disease; CN, cognitively normal; MCI, mild cognitive impairment.
* p < 0.05; ** p < 0.01; *** p < 0.001.
Among the CN‐only group, regression analyses revealed that Summary Score performances were similar, with worse performances among males versus females (estimate = −0.16, P = 0.017), older age versus younger age (estimate = −0.34, P < 0.001), Black/African American versus White participants (estimate = −0.15, P = 0.021), and Hispanic versus non‐Hispanic participants (estimate = −0.31, P = 0.002). No significant difference was found for years of education.
For each FNAME component test, a lower performance on Facial Recognition was associated with older versus younger age (estimate = −0.18, P = 0.001) and diagnosis of MCI/AD versus CN (estimate = −0.40, P = < 0.001). Lower performances on Name Recall were associated with age (estimate = −0.25, P = < 0.001), ethnicity (estimate = −0.29, P = 0.004), and diagnosis (estimate = −0.33, P = < 0.001). Finally, performance on Name Recognition was associated with age (estimate = −0.19, P = < 0.001), race (Black/African American participants; estimate = −0.18, P = 0.001), and diagnosis (estimate = −0.45, P = < 0.001). Performance was not found to significantly differ by sex or education on any component test.
3.3. Amyloid status and FNAME performance
Regression analyses in those with amyloid biomarkers (see Table 3) revealed that the Summary Score, when adjusted for sex, age, race, education, and diagnosis was not associated with amyloid status (amyloid‐positive estimate = −0.07, P = 0.408). However, in the amyloid biomarker group, lower Summary Scores were noted demographically in males versus females (estimate = −0.23, P = 0.002), older versus younger age (estimate = −0.14, P = 0.047), and MCI/AD versus CN (estimate = −0.55, P < 0.001). Among the CN‐only group, there was no association between the Summary Score and amyloid status but the same pattern of demographic differences was found.
On individual FNAME component tests (see Table 3), linear regression models adjusting for sex, age, race, education, and diagnosis revealed that amyloid status was associated with Face Recognition (amyloid‐positive estimate = −0.17, P = 0.042) but not with Name Recall (amyloid‐positive estimate = −0.01, P = 0.951) or Name Recognition (amyloid‐positive estimate = −0.01, P = 0.921). Lower performances across both component tests were noted in males versus females and those diagnosed with MCI/AD versus CN. Lower performance was associated with lower years of education on Face Recognition and Name Recognition, and with older age on Name Recall and Name Recognition. Linear regression models for the CN‐only group adjusted for sex, age, education, and race revealed no association with amyloid status on any FNAME component test.
Independent t‐tests comparing FNAME performance with amyloid‐positive and ‐negative cutoffs revealed that the Summary Score was lower in amyloid‐positive individuals compared to those who were amyloid negative (t = 3.86, P < 0.001), as was Face Recognition (t = 3.57, P = 0.001), Name Recall (t = 2.82, P = 0.008), and Name Recognition (t = 3.47, P = 0.002). Pearson correlations exploring the strength of the relationships between FNAME performance and amyloid status were moderate with amyloid‐positive individuals having a lower Summary Score (r = −0.43, P = < 0.001), and lower scores for Face Recognition (r = −0.46, P = < 0.001) and Name Recognition (r = −0.38, P = < 0.001). Name Recall was also statistically significant (r = −0.25, P = 0.004), but the relationship with amyloid status was not as strong.
The covariate sensitivity analysis controlling for site revealed that ethnicity was no longer significant on FNAME component scores in models 1 and 3 (see Table 3) but remained significant in models (5 to 8) in the CN group. On further exploration, we discovered that a majority of Hispanic participants came from only one site. Therefore, the relationship between site and FNAME and between ethnicity and FNAME was multi‐collinear, thus suppressing effects regarding ethnicity.
Independent t‐tests exploring FNAME scores within the CN‐only amyloid group revealed no differences between amyloid‐positive and ‐negative individuals on the Summary Score (t = 0.56, P = 0.592), Face Recognition (t = 1.33, P = 0.218), Name Recall (t = 0.29, P = 0.776) and Name Recognition (t = 0.00, P = 0.996). Nor were there any significant relationships on Pearson correlation coefficients between amyloid status and Summary Score (r = −0.07, P = 0.452), Face Recognition (r = −0.19, P = 0.052), Name Recall (r = −0.03, P = 0.726), and Name Recognition (r = −0.00, P = 0.996).
3.4. ROC analyses exploring FNAME score discrimination thresholds between amyloid‐positive and ‐negative individuals
Receiver operating characteristic (ROC) curves were used to identify thresholds at which FNAME component tests (Summary Score, Face Recognition, Name Recall, and Name Recognition) best discriminated between amyloid‐positive and ‐negative individuals (see Table 4).
TABLE 4.
FNAME component summary statistics and receiver operating characteristics curve adjusted thresholds in the amyloid group (N = 122).
| FNAME component | Possible score range | Mean (SD) [min, max] | ROC adjusted threshold a |
|---|---|---|---|
| Summary score | 0 to 36 | 21.70 (6.69) [4, 34] | 17 |
| Face recognition | 0 to 12 | 10.03 (2.38) [1, 12] | 10 |
| Name recall | 0 to 12 | 3.72 (2.68) [0, 11] | 2 |
| Face name recognition | 0 to 12 | 7.94 (2.83) [1, 12] | 5 |
Adjusted for unequal sample sizes between biomarker positive and negative patients.
For the Summary Score, Youden's index that discriminated amyloid‐positive from ‐negative individuals suggested a threshold of 16/36 points (AUC = 0.77; sensitivity = 0.62; specificity 0.91), and adjusting for unequal sample sizes increased it to 17/36 points (see Figure 2A). For the Face Recognition test, Youden's index suggested a threshold of 9/12 points to discriminate between amyloid‐positive versus ‐negative individuals (AUC = 0.75; sensitivity = 0.67; specificity 0.80), though after adjusting for unequal sample sizes of biomarker‐positive and ‐negative groups, this threshold was increased to 10/12 points (see Figure 2B). For Name Recall, Youden's index suggested a threshold of 1/12 points (AUC = 0.73; sensitivity = 0.52; specificity 0.84), and adjusting for unequal sample sizes increased it to 2/12 points (see Figure 2C). For Name Recognition, Youden's index suggested a threshold of 4/12 points (AUC = 0.73; sensitivity = 0.48; specificity 0.93), and adjusting for unequal sample sizes increased it to 5/12 points (see Figure 2D).
FIGURE 2.
Receiver operatic characteristic curves for FNAME tests discriminating between biomarker‐positive versus biomarker‐negative patients. (A) Summary Score, (B) Face Recognition, (C) Name Recall, (D) Name Recognition.
The FNAME Summary Score and each of the individual FNAME component tests had an AUC range between 0.73 and 0.77, demonstrating acceptable predictive diagnostic accuracy. The Youden's threshold provides the optimal cut‐point with higher scores more likely to predict those who were amyloid negative while scores below the cut‐point may indicate a greater probability of amyloid positivity (see Table 4).
4. DISCUSSION
The FNAME is a cross‐modal associative memory test designed to be sensitive to memory changes associated with the earliest stages of Alzheimer's disease. 6 , 11 , 12 The FNAME has been shown to differentiate CN individuals from those with amnestic MCI (aMCI), 17 and also cognitively asymptomatic CN individuals with and without biomarker evidence of AD. 6 , 11 , 12 , 13 The FNAME was added to the NIHTB in the ARMADA research protocol at year two. Normative scores stratified by age, sex, education and race were provided for CN; those with MCI/AD; and those with positive/negative amyloid biomarkers. Filtered normative data stratified by age ranges for sex and race were also provided. Performance thresholds of FNAME component scores were proposed across the biomarker sample that best discriminated between positive and negative groups.
The FNAME is an episodic memory test that taps into a frequent complaint among older individuals, namely, an inability to remember names associated with faces. As expected, we found that participants diagnosed with MCI/AD consistently performed worse than participants diagnosed as CN, 12 , 17 , 23 suggesting that the FNAME might be useful in differentiating between age‐related cognitive decline and those in early stages of a neurodegenerative disease. 6 , 11 , 12 , 13 , 17 Comparable to other memory studies, males performed worse than females and those of an older age performed worse than younger individuals. 24 , 25 , 26 Despite educational equivalence in the ARMADA cohort, racial differences were found on the FNAME, with Black/African American and Hispanic individuals 16 , 23 , 27 having lower scores than their White and Non‐Hispanic counterparts despite concerted efforts to make the FNAME stimuli racially diverse. Other research investigators who have adopted the FNAME found that the face‐name stimuli needed to be modified to their country of origin. 15 , 16 , 18 , 23 This raises the question as to whether FNAME performance among the ARMADA Black/African American and Hispanic individuals may be an artifact of the test stimuli rather than indicative of lower cognitive performance. Furthermore, despite providing the FNAME in Spanish, our Hispanic individuals chose to take the test in English. This may have driven some of the lower performances in this cohort, since English was not their primary language. Finally, recruitment of diverse individuals who were administered the FNAME was hampered by the COVID pandemic, thus reducing sample size. Nevertheless, this study provides important normative data in Black/African American and Hispanic adults on the NIHTB FNAME, which can be useful both clinically and in research.
The AUC for the FNAME Summary Score's ability to discriminate between amyloid‐positive versus ‐negative groups was acceptable (AUC ≥ 0.70) but not excellent (AUC ≥ 0.80). Given the particularly high specificity scores across all the FNAME component tests, the FNAME will likely excel at predicting amyloid‐negative rather than amyloid‐positive cases. Knowledge regarding potential amyloid status, whether negative or positive, 28 was found to be important information to patients and families and directly impacted clinical care. 29 Unlike expensive PET scans or cerebral spinal fluid studies, the FNAME is a simple, non‐invasive, cost‐effective test that can be potentially useful in the primary care arena for guiding diagnostic decisions about triaging to specialty clinics and ultimately, future clinical care. However, given that the discriminatory ability was not excellent, we would caution against using the FNAME Summary Score as a diagnostic test at this time.
When examining amyloid status alone without consideration of other demographic variables, we found a significant relationship between FNAME performance and amyloid positivity across all FNAME component scores. In contrast to previous findings, 6 , 11 , 12 the FNAME in ARMADA was not able to differentiate between positive and negative amyloid buildup among CN‐only adults. A closer examination of the ARMADA biomarker cohort revealed that 63% of the 19 individuals with MCI/AD had positive amyloid biomarkers, while only 9% of the 103 CN individuals were amyloid positive. The small number of amyloid‐positive CN participants in this cohort likely influenced the amyloid/FNAME relationship. A larger study of FNAME performance in CN individuals with amyloid biomarkers (ie, the Anti‐Amyloid in Asymptomatic Alzheimer's Disease trial), also found the FNAME unable to discriminate between amyloid‐positive and ‐negative individuals at baseline. However, when combined with other visual tests in the Computerized Cognitive Composite (C3), the baseline Summary C3 Composite Score significantly discriminated biomarker groups comparable to the Preclinical Alzheimer Cognitive Composite. 30 This suggests that a summary or composite score, rather than any individual test, may be more useful for detecting early cognitive changes. 31 , 32 , 33 Another recent longitudinal study of FNAME in CN individuals with amyloid biomarkers found equivalent performance across amyloid groups at baseline but those who were amyloid positive declined over time and were not able to benefit from repeated practice exposure as their non‐amyloid CN counterparts. 34 As more longitudinal data emerges in the ARMADA cohort, we will be able to examine amyloid status along with other neuropsychological nuances of FNAME performance that could guide clinical and research use of the test.
In addition to the Summary Score that captures associative memory across all three phases of memory performance (Face Recognition, Name Recall, and Name Recognition), we explored whether any individual FNAME component scores in the ARMADA cohort were associated with amyloid deposition across the clinical spectrum. Linear regression models adjusted for demographics suggested amyloid‐positive participants in ARMADA performed worse on Face Recognition than participants who were amyloid negative. It is well known that recognition memory is usually preserved in aging and often differentiates AD‐like memory changes from other non‐AD etiologies. 35 Recognition discriminability, or the capacity to distinguish targets from distractors, is significantly poorer in aMCI with biomarker evidence of AD than other forms of recognition indices. 36 Visual recognition also predicted progression from aMCI to AD dementia with a sensitivity of 80% and specificity of 90%. 37 The Behavioral Pattern Separation‐Object test, designed to discriminate between novel and familiar visual object patterns, also differentiated amyloid‐positive from amyloid‐negative groups. 30 All of these studies suggest that visual discrimination, or the failure to accurately recognize objects and faces, may be uniquely capturing specific aspects of memory vulnerable to early biomarker changes. Thus, declines in visual recognition on the FNAME or other visual memory tests may be a useful indicator for identifying individuals at risk for future cognitive decline.
A particular advantage of the FNAME is its reliance on associative memory, which is particularly vulnerable to memory decline in AD. 38 From previous functional magnetic resonance imaging work in amyloid‐positive CN individuals, the failure to create an associative link between two related stimuli, such as faces and names, has been associated with temporolimbic dysfunction. This work suggests that the successful memorization and retrieval of face/name pairs requires a coordinated activity within the default mode network, which involves brain regions particularly vulnerable to amyloid deposition. 39 , 40 , 41 While amyloid is an important early biomarker along the AD cascade, in the future, determining FNAME performance in relationship to both amyloid and tau deposition may result in greater sensitivity in detecting those individuals along the AD trajectory.
Finally, the ARMADA NIHTB contains a suite of cognitive tests that were not explored in this paper. Future work will determine if there is a combination of both fluid/executive tests, such as the List Sort Working Memory Test, Flanker, Dimensional Card Sort, and FNAME that might render better sensitivity/specificity and diagnostic discriminability for early detection and diagnosis.
4.1. Limitations and future directions
While the results of this study are promising, there were obvious limitations. Our sample population only included 12 MCI/AD and 9 CN participants who were amyloid positive in a cohort of 122 with amyloid biomarkers. This low sample size, along with lower than excellent biomarker discriminability does temper enthusiasm for utilizing the FNAME as a diagnostic marker. However the FNAME's acceptable rating of discriminability and high specificity suggests it may be useful for identifying clinical status rather than amyloid burden. As blood‐based biomarkers become available, we hope to increase the number of data points in a larger community‐based sample that will allow for more robust statistical analyses to determine whether lower FNAME scores among CN individuals is predictive of developing MCI/AD in the future. We note that age was associated with the FNAME Summary Score, suggesting that age‐specific score cut‐points could improve discrimination. Our study design did not allow for estimation of age‐specific cut‐points, but this will be explored in future research. Finally, future work will also explore the longitudinal performance of FNAME, its ability to predict decline, and whether novel composite measures can be derived to better predict biomarker status.
5. SUMMARY AND CONCLUSIONS
In conclusion, the purpose of this study was to establish baseline performance for the FNAME, a cross‐modal associative memory test known to be sensitive to early memory changes related to biomarker status in AD. FNAME normative data were presented for three main age groups, 64 to 74, 75 to 84, and 85+, as well as across demographics including sex, education, race, and ethnicity, and thresholds were proposed for FNAME component scores that best discriminated between biomarker‐positive versus biomarker‐negative participants. Norms were provided across a diverse sample of White, Black/African American, and Hispanic individuals and a small sample of participants diagnosed as MCI/AD.
CONFLICT OF INTEREST STATEMENT
AS., C.F., O.R.S., H.M.K., and S.A. report no disclosures relevant to this manuscript. D.R. received salary support as a co‐PI on the ARMADA grant No. 1U2CAG057441. D.P. has been a paid consultant to FACIT, Debiopharm Beta6. D.R. has also served as a paid consultant for Biogen Idec, Digital Cognition Technologies, and Neurotrack. Author disclosures are available in the supporting information.
Supporting information
Supporting Information
Supporting Information
ACKNOWLEDGMENTS
The authors thank the study participants along with the research staff across all sites who contributed to the data used in this study. This study was funded in whole or in part with Federal funds from the National Institute on Aging, National Institutes of Health, under grant No. 1U2CAG057441 (Gershon, Weintraub) and 2P01AG036694‐11 (Sperling and Johnson).
Rentz DM, Klinger HM, Samaroo A, et al. Face Name Associative Memory Exam and biomarker status in the ARMADA study: Advancing reliable measurement in Alzheimer's disease and cognitive aging. Alzheimer's Dement. 2023;15:e12473. 10.1002/dad2.12473
REFERENCES
- 1. Gonzales MM, Garbarino VR, Pollet E, et al. Biological aging processes underlying cognitive decline and neurodegenerative disease. J Clin Invest. 2022;132(10). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Assuncao SS, Sperling RA, Ritchie C, et al. Meaningful benefits: a framework to assess disease‐modifying therapies in preclinical and early Alzheimer's disease. Alzheimers Res Ther. 2022;14(1):54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Rentz DM, Papp KV, Mayblyum DV, et al. Association of digital clock drawing with PET amyloid and tau pathology in normal older adults. Neurology. 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Rentz DM, Parra Rodriguez MA, Amariglio R, Stern Y, Sperling R, Ferris S. Promising developments in neuropsychological approaches for the detection of preclinical Alzheimer's disease: a selective review. Alzheimers Res Ther. 2013;5(6):58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Ohman F, Hassenstab J, Berron D, Scholl M, Papp KV. Current advances in digital cognitive assessment for preclinical Alzheimer's disease. Alzheimers Dement (Amst). 2021;13(1):e12217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Rentz DM, Amariglio RE, Becker JA, et al. Face‐name associative memory performance is related to amyloid burden in normal elderly. Neuropsychologia. 2011;49(9):2776‐2783. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Papp KV, Samaroo A, Hsiang‐Chin C, et al. Unsupervised mobile cognitive testing for use in preclinical Alzheimer's disease. Alzheimer Dement: Diagn, Assess Dis Monit. 2021;13:1‐10. e12243. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Thompson LI, Harrington KD, Roque N, et al. A highly feasible, reliable, and fully remote protocol for mobile app‐based cognitive assessment in cognitively healthy older adults. Alzheimers Dement (Amst). 2022;14(1):e12283. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Stawski RS, MacDonald SWS, Brewster PWH, Munoz E, Cerino ES, Halliday DWR. A comprehensive comparison of quantifications of intraindividual variability in response times: a measurement burst approach. J Gerontol B Psychol Sci Soc Sci. 2019;74(3):397‐408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Weintraub S, Karpouzian‐Rogers T, Peipert JD, et al. ARMADA: assessing reliable measurement in Alzheimer's disease and cognitive aging project methods. Alzheimers Dement. 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Amariglio RE, Frishe K, Olson LE, et al. Validation of the Face Name Associative Memory Exam in cognitively normal older individuals. J Clin Exp Neuropsychol. 2012;34(6):580‐587. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Papp KV, Amariglio RE, Dekhtyar M, et al. Development of a psychometrically equivalent short form of the Face‐Name Associative Memory Exam for use along the early Alzheimer's disease trajectory. Clin Neuropsychol. 2014;28(5):771‐785. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Alegret M, Munoz N, Roberto N, et al. A computerized version of the Short Form of the Face‐Name Associative Memory Exam (FACEmemory(R)) for the early detection of Alzheimer's disease. Alzheimers Res Ther. 2020;12(1):25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Alviarez‐Schulze V, Cattaneo G, Pachon‐Garcia C, et al. Validation and normative data of the spanish version of the face name associative memory exam (S‐FNAME). J Int Neuropsychol Soc. 2021:1‐11. [DOI] [PubMed] [Google Scholar]
- 15. Kormas C, Megalokonomou A, Zalonis I, Evdokimidis I, Kapaki E, Potagas C. Development of the Greek version of the Face Name Associative Memory Exam (GR‐FNAME12) in cognitively normal elderly individuals. Clin Neuropsychol. 2018;32(sup1):152‐163. [DOI] [PubMed] [Google Scholar]
- 16. Vila‐Castelar C, Munoz N, Papp KV, et al. The Latin American Spanish version of the Face‐Name Associative Memory Exam is sensitive to cognitive and pathological changes in preclinical autosomal dominant Alzheimer's disease. Alzheimers Res Ther. 2020;12(1):104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Kormas C, Zalonis I, Evdokimidis I, Kapaki E, Potagas C. Face‐name associative memory performance among cognitively healthy individuals, individuals with subjective memory complaints, and patients with a diagnosis of aMCI. Front Psychol. 2020;11:2173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Sanabria A, Alegret M, Rodriguez‐Gomez O, et al. The Spanish version of Face‐Name Associative Memory Exam (S‐FNAME) performance is related to amyloid burden in Subjective Cognitive Decline. Sci Rep. 2018;8(1):3828. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. SAS Institute I. Plot ROC curve with cutpoint labeling and optimal cutpoint analysis. 2021.
- 20. Hosmer DWLS, Sturdivant RX. Applied Logistic Regression. 3rd ed. John Wiley and Sons; 2013. [Google Scholar]
- 21. Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3(1):32‐35. [DOI] [PubMed] [Google Scholar]
- 22. Terluin B, Eekhout I, Terwee CB. The anchor‐based minimal important change, based on receiver operating characteristic analysis or predictive modeling, may need to be adjusted for the proportion of improved patients. J Clin Epidemiol. 2017;83:90‐100. [DOI] [PubMed] [Google Scholar]
- 23. Vila‐Castelar C, Papp KV, Amariglio RE, et al. Validation of the Latin American Spanish version of the face‐name associative memory exam in a Colombian Sample. Clin Neuropsychol. 2020;34(sup1):1‐12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Bleecker ML, Bolla‐Wilson K, Agnew J, Meyers DA. Age‐related sex differences in verbal memory. J Clin Psychol. 1988;44(3):403‐411. [DOI] [PubMed] [Google Scholar]
- 25. Rentz DM, Weiss BK, Jacobs EG, et al. Sex differences in episodic memory in early midlife: impact of reproductive aging. Menopause. 2017;24(4):400‐408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. van Hooren SA, Valentijn AM, Bosma H, Ponds RW, van Boxtel MP, Jolles J. Cognitive functioning in healthy older adults aged 64‐81: a cohort study into the effects of age, sex, and education. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn. 2007;14(1):40‐54. [DOI] [PubMed] [Google Scholar]
- 27. Weuve J, Barnes LL, Mendes de Leon CF, et al. Cognitive aging in black and white americans: cognition, cognitive decline, and incidence of Alzheimer disease dementia. Epidemiology. 2018;29(1):151‐159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Rabinovici GD, Gatsonis C, Apgar C, et al. Association of amyloid positron emission tomography with subsequent change in clinical management among medicare beneficiaries with mild cognitive impairment or dementia. JAMA. 2019;321(13):1286‐1294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Rentz DM, Wessels AM, Annapragada AV, et al. Building clinically relevant outcomes across the Alzheimer's disease spectrum. Alzheimers Dement (N Y). 2021;7(1):e12181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Papp KV, Rentz DM, Maruff P, et al. The computerized cognitive composite (C3) in an Alzheimer's disease secondary prevention trial. J Prev Alzheimers Dis. 2021;8(1):59‐67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Langbaum JB, Ellison NN, Caputo A, et al. The Alzheimer's Prevention Initiative Composite Cognitive Test: a practical measure for tracking cognitive decline in preclinical Alzheimer's disease. Alzheimers Res Ther. 2020;12(1):66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Lim YY, Snyder PJ, Pietrzak RH, et al. Sensitivity of composite scores to amyloid burden in preclinical Alzheimer's disease: introducing the Z‐scores of Attention, Verbal fluency, and Episodic memory for Nondemented older adults composite score. Alzheimers Dement (Amst). 2016;2:19‐26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Rentz DM, Papp KV. Commentary on Composite cognitive and functional measures for early stage Alzheimer's disease trials. Alzheimers Dement (Amst). 2020;12(1):e12012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Samaroo A, Amariglio RE, Burnham S, et al. Diminished Learning Over Repeated Exposures (LORE) in preclinical Alzheimer's disease. Alzheimers Dement (Amst). 2020;12(1):e12132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Dickerson BC, Eichenbaum H. The episodic memory system: neurocircuitry and disorders. Neuropsychopharmacology. 2010;35(1):86‐104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Goldstein FC, Loring DW, Thomas T, Saleh S, Hajjar I. Recognition memory performance as a cognitive marker of prodromal Alzheimer's disease. J Alzheimers Dis. 2019;72(2):507‐514. [DOI] [PubMed] [Google Scholar]
- 37. Didic M, Felician O, Barbeau EJ, et al. Impaired visual recognition memory predicts Alzheimer's disease in amnestic mild cognitive impairment. Dement Geriatr Cogn Disord. 2013;35(5‐6):291‐299. [DOI] [PubMed] [Google Scholar]
- 38. Werheid K, Clare L. Are faces special in Alzheimer's disease? Cognitive conceptualisation, neural correlates, and diagnostic relevance of impaired memory for faces and names. Cortex. 2007;43(7):898‐906. [DOI] [PubMed] [Google Scholar]
- 39. Buckner RL, Snyder AZ, Shannon BJ, et al. Molecular, structural, and functional characterization of Alzheimer's disease: evidence for a relationship between default activity, amyloid, and memory. J Neurosci. 2005;25(34):7709‐7717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Sperling RA, Laviolette PS, O'Keefe K, et al. Amyloid deposition is associated with impaired default network function in older persons without dementia. Neuron. 2009;63(2):178‐188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Vannini P, Hedden T, Becker JA, et al. Age and amyloid‐related alterations in default network habituation to stimulus repetition. Neurobiol Aging. 2012;33(7):1237‐1252. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information
Supporting Information