Background: The Memory Associative Test TMA-93 examines visual relational binding, characteristically affected in early-AD stages.
Objective: We aim to validate the TMA-93 by biomarkers determination and compare its diagnostic characteristics with the Free and Cued Selective Reminding Test (FCSRT).
Methods: Retrospective analysis of a Biobank database. Patients' records initially consulted for memory complaints, scored MMSE≥22, had TMA-93 and FCSRT tested, and AD biomarker determination (Amyloid-PET or CSF), either positive or negative, were selected. As cutoffs, we considered the 10-percentile for TMA-93 (P10/TMA-93), and "total free recall" (TFR) 21/22, total recall (TR) 43/44, and Cued Index < 0.77 for FCSRT from previous Spanish validation and normative studies. Diagnostic utilities were calculated using ROC curves and compared by the DeLong method. We studied if one test improved the other test's prediction, following a forward stepwise logistic regression model.
Results: We selected 105 records: 64 "positive" and 41 "negative" biomarkers. TMA-93 total score diagnostic utility (AUC = 0.72; 95%CI:0.62-0.82) was higher than those of the FCSRT: TFR (AUC = 0.70; 95%CI: 0.60-0.80), TR (AUC = 0.63; 95%CI:0.53-0.74), and Cued Index (AUC = 0.62; 95%CI:0.52-0.73). The P10/TMA-93 cutoff showed 86%sensitivity, similar to that of the most sensitive FCSRT cutoff (TFR21/22, 89%) and 29%specificity, lower than that of the most specific FCSRT cutoff (Cued Index < 0.77, 57%). 32.8%of the positive-biomarker group scored above CI/0.77 but below p10TMA-93. The addition of TMA-93 total score to FCSRT variables improved significantly the biomarkers results' prediction.
Conclusion: TMA-93 demonstrated "reasonable" diagnostic utility, similar to FCSRT, for discriminating AD biomarker groups. TMA-93 total score improved the AD biomarker result prediction when added to FCSRT variables.
Keywords: Alzheimer’s disease; TMA-93; amyloid-PET; biomarker; cerebrospinal fluid; free and cued selective reminding test.