The Synergic Effect of AT(N) Profiles and Depression on the Risk of Conversion to Dementia in Patients with Mild Cognitive Impairment

Neuropsychiatric symptoms (NPSs) encompass behavioral and psychological changes such as mood and anxiety disturbances, apathy, perceptual changes, sleep problems, agitation and aggression, and are associated with poorer cognitive and functional outcomes and faster progression to severe dementia. NPSs can manifest early in preclinical and prodromal stages of Alzheimer's disease (AD), including mild cognitive impairment (MCI), predicting higher conversion risk to dementia. Advances in in vivo AD biomarkers (CSF, plasma, imaging, genomics) allow quantification of core AD-related brain changes. The 2018 NIA-AA AT(N) research framework biologically defines AD independent of clinical syndrome, categorizing biomarkers as Aβ plaques (A), pathologic tau (T), and neurodegeneration (N). Individuals with MCI can be classified as having underlying AD pathology if A and T are abnormal. Few studies have examined associations between AD-core biomarkers and NPSs in MCI, and none have addressed interaction effects between AT(N) profiles and NPSs on risk of conversion to dementia. The study aims to evaluate the predictive value of combining CSF-based AT(N) profiles with NPSs (NPI-Q) using survival analysis to estimate conversion to dementia in MCI, and to test additive and multiplicative interactions between AT(N) profiles and NPSs on conversion risk.

Design and participants: Prospective cohort study of 500 patients diagnosed with mild cognitive impairment (MCI) evaluated at the ACE Alzheimer Center Barcelona memory clinic (single site) between 2016 and 2022. Participants underwent lumbar puncture (LP) for CSF AD-core biomarker quantification and baseline assessment of neuropsychiatric symptoms (NPSs) using the NPI-Q within 5 months. Annual follow-up visits were conducted. Consensus diagnoses were provided by a multidisciplinary team. Clinical and cognitive assessments: Collected demographics (age, sex, years of education). Cognitive evaluations included the Spanish MMSE, memory component of the 7-Minute Test, NPI-Q (baseline), Hachinski Ischemia Scale, Blessed Dementia Scale, Clinical Dementia Rating (CDR), and the ACE neuropsychological battery (NBACE). MMSE and NBACE were performed at all visits; NPI-Q only at baseline. NPS assessment: NPI-Q administered by neurologist/geriatrician using caregiver report for the last month. NPSs considered present/absent; to ensure stable estimates, only symptoms with prevalence >5% were analyzed: depression/dysphoria (50.4%), anxiety (43.6%), apathy (37.6%), irritability/lability (35.4%), and nighttime behaviors (20.6%). Low-prevalence items (agitation/aggression, delusion, hallucination, euphoria/elation, disinhibition, aberrant motor behavior, eating/appetite problems) were excluded. MCI subtyping: Classified as amnestic vs non-amnestic and possible vs probable MCI. Amnestic MCI indicated age/education-adjusted memory impairment (NBACE cut-offs); non-amnestic showed deficits in other domains. Probable MCI indicated absence of significant comorbidities; possible MCI indicated presence of comorbidities potentially contributing to deficits. Outcome—conversion to dementia: Converters were those who developed dementia during follow-up. Dementia etiologies were assigned per standard criteria: NIA-AA 2011 for AD, NINDS-AIREN for vascular dementia, Neary criteria for FTD, and McKeith criteria for LBD. Conversion definitions followed prior work. Non-converters remained MCI throughout follow-up. CSF biomarker collection and assays: LPs conducted under fasting conditions following Alzheimer's Biomarkers Standardization Initiative pre-analytical recommendations. CSF collected in polypropylene tubes, processed (centrifuged 2000×g, 10 min, 4°C within 2 h), aliquoted, stored at −80°C. Aβ1-42, total tau (T-tau), and phosphorylated tau (p181-tau) measured via ELISA (INNOTEST; n=252) or CLEIA on Lumipulse G 600 II (n=248). CSF results were not used for initial clinical diagnosis; clinicians were blinded to CSF status. AT(N) profile classification: Participants categorized into four AT(N) groups using CSF biomarkers: Normal (A−T−N−), Brain amyloidosis (A+T−N−), Prodromal AD (A+T+N−, A+T+N+, A+T−N+), and Suspected non-AD pathology (SNAP: A−T+N−, A−T−N+, A−T+N+). Assay-specific cut-offs: ELISA—Aβ1-42 < 676 pg/mL (A+), p181-tau > 58 pg/mL (T+), T-tau > 367 pg/mL (N+); CLEIA—Aβ1-42 < 796 pg/mL (A+), p181-tau > 54 pg/mL (T+), T-tau > 412 pg/mL (N+). APOE genotyping: DNA extracted from blood (Chemagic system). APOE genotypes from Axiom SP array; carriers defined by ≥1 ε4 allele. Ethics: LP consent approved by Hospital Clínic i Provincial de Barcelona ethics committee; procedures adhered to Spanish biomedical laws and Declaration of Helsinki. Informed consent obtained from all participants. Statistical analysis: Conducted in STATA 15. Group differences across AT(N) profiles assessed via ANOVA or chi-squared tests. Associations of demographics/clinical variables with NPS presence assessed by t-tests/chi-squared tests. Differences between converters vs non-converters assessed similarly. Main analyses used Cox proportional hazards models to estimate hazard ratios (HRs) for conversion, including AT(N) profiles (reference: Normal) and five NPSs, adjusted for age, sex, education, baseline MMSE, APOE ε4 status, MCI subtype (amnestic yes/no), MCI status (probable yes/no). Interaction analyses evaluated additive and multiplicative interactions between AT(N) (contrasting Normal vs Prodromal AD) and each NPS using dummy variables and methods per VanderWeele and Knol. Significant interactions led to estimation/plotting of marginal means of cumulative hazard functions for combinations of AT(N) and NPSs. Two-tailed alpha p<0.05.

- Sample characteristics: Mean age 73 years; 55% female; mean education 8 years. AT(N) distribution: Prodromal AD 42%, Normal 26%, SNAP 18%, Brain amyloidosis 14%. APOE ε4 carriers 36%. NPS prevalence: depression 50%, anxiety 44%, apathy 38%, irritability 35%, nighttime behaviors 21%. - Conversions: 224/500 (44.8%) MCI participants converted to dementia over ~2-year follow-up. Converters were older, more often amnestic and probable MCI, had lower MMSE, and higher APOE ε4 frequency. Apathy was more frequent in converters (p=0.029). Approximately 60% of converters were in the prodromal AD AT(N) group vs 26% of non-converters (p<0.001). - AT(N) as predictors (Cox model, adjusted): Compared to Normal, increased risk of conversion for SNAP (HR 1.79; 95% CI 1.04–3.06; p=0.035), Brain amyloidosis (HR 2.92; 1.72–4.97; p<0.001), and Prodromal AD (HR 4.34; 2.72–6.91; p<0.001). - NPS as predictors (adjusted): Depression (HR 1.47; 95% CI 1.06–2.20; p=0.019) and Apathy (HR 1.40; 1.05–1.86; p=0.020) predicted higher conversion risk. Anxiety, irritability/lability, and nighttime behaviors were not significant. - Covariates: Age (HR per year 1.02; p=0.033), amnestic MCI (HR 3.63; p<0.001), probable MCI (HR 1.51; p=0.006), and lower MMSE (HR 0.87 per point; p<0.001) were significant. APOE ε4 carrier status was not significant in the adjusted model (HR 0.98; p=0.904). - Interactions: Significant additive interaction between AT(N) and Depression (coefficient 3.23; 95% CI 0.20–6.26; p=0.037), indicating a synergistic increase in conversion risk among MCI with prodromal AD profile and depression beyond the sum of individual effects. This remained significant after adjusting for comorbidity with other NPSs (p=0.044). A multiplicative interaction between AT(N) and Nighttime behaviors was observed initially (coef 0.36; 95% CI 0.13–0.94; p=0.034), suggesting higher risk with nighttime behaviors in the Normal AT(N) group versus Prodromal AD; however, this lost significance when adjusting for other NPSs. - Additional observations: Nighttime behaviors differed across AT(N) groups (lowest prevalence in prodromal AD). Depression and anxiety were more common in females; irritability more common in males. No NPS was associated with APOE ε4 status. In the Normal AT(N) group, conversion rate was ~12% (noted in discussion).

The study demonstrates that CSF-derived pathological AT(N) profiles and specific neuropsychiatric symptoms—particularly depression and apathy—are independent predictors of conversion from MCI to dementia over a relatively short follow-up. The prodromal AD AT(N) group confers the highest conversion risk, followed by brain amyloidosis and SNAP, consistent with a biological gradient aligned with AD pathology burden. Crucially, the analysis reveals a synergistic additive interaction between prodromal AD AT(N) status and depressive symptoms: the combined presence leads to a higher-than-additive risk of conversion. This suggests that depressive symptoms may potentiate the clinical expression or progression of underlying AD pathology in MCI. In contrast, an initially observed multiplicative interaction with nighttime behaviors appears context-dependent and is attenuated when accounting for comorbid NPSs, indicating limited robustness. The findings refine understanding of how behavioral symptoms integrate with biological markers to influence prognosis in MCI. They align with literature suggesting that depression and apathy may relate to cerebrovascular pathways and could accelerate clinical decline, and with the concept of mixed pathologies contributing to dementia risk. The lack of association between APOE ε4 status and conversion in models including AT(N) profiles suggests that APOE’s prognostic effect may be mediated through AD biomarker abnormalities. Anxiety and irritability did not predict conversion here, reflecting heterogeneity in prior results and potential differences by setting, measurement, and follow-up duration. Overall, considering interactions between biological and behavioral factors yields more nuanced risk stratification than evaluating each in isolation, and highlights depression as a clinically accessible marker that modifies biomarker-based risk.

This study identifies, for the first time, a synergistic additive effect between CSF-based prodromal AD AT(N) profiles and depressive symptoms on the risk of conversion to dementia among MCI patients. Pathologic AT(N) profiles (especially prodromal AD) and NPSs (depression, apathy) independently increase conversion risk, with depression further exacerbating risk in those with prodromal AD biology. These insights support combined biochemical and behavioral risk assessment to identify MCI patients at highest risk and to prioritize early intervention. Future research should employ longer follow-up, repeated assessments of NPSs and biomarkers, and advanced longitudinal modeling to clarify temporal dynamics, causal pathways, and to evaluate whether targeted treatment of depression can modify progression risk in biomarker-defined MCI.

- Short follow-up period (~2 years), which may limit detection of longer-term conversion dynamics. - Restricted NPS spectrum analyzed; low-prevalence symptoms were excluded, potentially omitting relevant behavioral predictors. - Baseline-only assessment of NPSs and AT(N) status; both may change over time, and single time-point analyses may not capture dynamic interactions. - Clinical sample from a single memory clinic, which may affect generalizability. - Potential residual confounding and comorbidity among NPSs, despite adjustments.