Semantic Fluency Predicts Survival in Memory Clinic Patients

Key Takeaways

• Among patients referred to a tertiary memory clinic for dementia workup, poorer baseline semantic fluency was associated with shorter survival, and 48% died during follow-up.
• Semantic fluency carried the strongest mortality signal among CERAD-NAB measures and remained the only significant cognitive test in multivariable modeling.
• The association between semantic fluency and survival persisted after separate adjustment for MMSE and amyloid PET status.
• In the repeat-testing subgroup (163/583 patients, 28%), greater decline on semantic fluency was associated with shorter survival (HR 0.66, 95% CI 0.53-0.83).
• In the 218-patient imaging subgroup, worse semantic fluency tracked with lower left-sided FDG-PET uptake, while global amyloid load was not associated with semantic fluency or survival.

Prognosis in patients undergoing memory clinic evaluation often has to be discussed before a diagnostic label or biomarker profile is fully settled. Semantic fluency offers a practical entry point for that conversation because the one-minute task is already embedded in routine cognitive testing for suspected neurodegenerative impairment. The open question is whether that familiar bedside measure carries prognostic information beyond broader cognition and imaging biomarkers.

At a tertiary referral memory clinic in Freiburg, Germany, investigators retrospectively identified 583 patients referred for dementia diagnostic workup who completed the Consortium to Establish a Registry for Alzheimer’s Disease neuropsychological assessment battery (CERAD-NAB) and had retrievable vital status in 07/2024 in the Alzheimer’s Research & Therapy study of semantic fluency and survival. Semantic fluency was defined as the German CERAD-NAB animals task with a 1-minute time limit and was compared with 10 other CERAD-NAB measures. Investigators used age-adjusted Cox models, a confirmatory multivariable model, separate models adding the Mini-Mental Status Examination (MMSE) or amyloid positron emission tomography (amyloid PET) status, and subgroup analyses for repeat testing and combined amyloid PET and fluorodeoxyglucose positron emission tomography (FDG-PET).

Among the 583 patients, 280 had died by a median follow-up of 8.0 years. Better baseline semantic fluency was associated with lower mortality, HR 0.71 (95% CI 0.63-0.80), Bonferroni-corrected p < 0.001, and Harrell’s c reached 0.67 when age and semantic fluency were combined. Semantic fluency showed the strongest prognostic value among the CERAD-NAB measures tested.

Semantic fluency remained the only significant CERAD-NAB cognitive predictor in the multivariable Cox model and remained associated with mortality after separate adjustment for MMSE and amyloid status. Greater decline on repeat testing was also linked to shorter survival in the repeat-testing subgroup, HR 0.66 (95% CI 0.53-0.83), p < 0.001, in the analysis of longitudinal semantic fluency and PET biomarkers. In the PET subgroup, worse semantic fluency correlated with lower left temporal, dorsolateral frontal, and posterior parietal FDG uptake but not with amyloid load, and left inferior frontal gyrus pars opercularis uptake was also associated with survival, HR 0.68 (95% CI 0.55-0.84).

These findings describe prognostic associations from a retrospective single-center registry rather than causal effects. Cases with missing data were excluded, generalizability beyond this tertiary referral memory clinic is uncertain, semantic fluency was the only executive measure specifically tested, and the FDG incremental analyses were exploratory, with the strongest added-value signal not remaining Bonferroni-significant. For North American clinicians, the practical relevance is the use of a familiar CERAD task, although the effect estimates came from a German tertiary referral population.

The authors concluded that semantic fluency emerged as a survival marker in memory clinic patients; in the full text, they report that it remained associated with survival in separate age-adjusted models controlling for MMSE or for amyloid status, while anterior cortical glucose metabolism carried a weaker complementary signal and, in the imaging subgroup of 218 patients, global amyloid load was not associated with semantic fluency or survival.

Clinician Questions

Which memory clinic patients were represented in the semantic fluency survival analysis?

The cohort came from a tertiary referral memory clinic in Freiburg, Germany, and included patients referred for dementia diagnostic workup who completed CERAD-NAB testing and had retrievable vital status. Filed diagnoses were available in 75% of patients and included mild cognitive impairment, Alzheimer disease, frontotemporal dementia, mixed dementia, dementia with Lewy bodies, depression, and vascular dementia, indicating that the analysis was not limited to a single diagnosis.

How was semantic fluency measured in this memory clinic cohort?

Semantic fluency was measured with the German CERAD-NAB animals task using a 1-minute time limit, and performance was analyzed as an age-, sex-, and education-adjusted z score. Investigators compared that measure with 10 other CERAD-NAB tests rather than treating semantic fluency in isolation.

What brain regions tracked with poorer semantic fluency on FDG-PET?

Poorer semantic fluency was linked to lower FDG uptake in left inferior and middle temporal cortex, dorsolateral frontal cortex, and posterior parietal regions. The same analysis did not show a correlation with global amyloid load, which helps distinguish the regional FDG signal from the amyloid findings.

How wide was the survival separation across semantic-fluency performance levels?

Median survival ranged from 6.39 years in the worst-performing semantic-fluency quartile to 10.40 years in the best-performing quartile. Those quartiles were used for display, while the main analysis treated semantic fluency as a continuous predictor.