背景：

随着人口老龄化进程加速，痴呆的发病率与社会经济负担显著上升。然而，常规临床评估和结构性神经影像学在阿尔茨海默病（AD）、额颞叶痴呆（FTD）、血管性痴呆（VaD）等痴呆亚型的准确鉴别中仍存在局限，尤其对早发型、家族性或非典型病例的诊断敏感性不足。脑脊液Aβ/Tau定量、Aβ/Tau–PET显像、全外显子测序（WES）及重复扩增检测（如C9orf72、NOTCH2NLC）等技术的应用，虽已深化了对痴呆分子生物学和遗传学基础的认识，但迄今鲜有纵向研究能够系统整合家族史、认知功能及影像生物标志物和基因组数据，以量化遗传风险对认知功能衰退和脑区萎缩的影响。本研究旨在检验以家族史阳性或携带致病基因致病/可能致病（P/LP）变异为标志的遗传风险，是否加速痴呆患者的认知衰退和区域性脑萎缩，并在此基础上构建简明的临床特征预测模型，以识别高危人群并指导个性化干预。

方法：

第一部分：

本研究纳入695例临床诊断痴呆患者，均完成脑脊液Aβ/Tau定量或Aβ/Tau–PET显像、100×深度WES及C9orf72、NOTCH2NLC、HTT、FMR1重复扩增检测，并进行APOE基因分型、MMSE、MoCA、ADL、AVLT及不同认知域评估，结合65个脑区MRI体积分割。研究内容包括：（1）家族史阳性与P/LP变异携带的流行病学分布；（2）有/无家族史患者组间认知及脑体积变化差异；（3）P/LP携带者与非携带者的上述比较。采用线性混合效应模型，调整随访时间、遗传状态、家族史、性别、受教育年限及基线年龄，评估部分纵向认知与脑体积变化，并对基因变异携带者开展临床—影像—基因型关联分析。

第二部分：

本研究纳入601例家族史阳性痴呆患者，按发病时间先后分为476例建模队列和125例时间性验证队列。预测因子包括发病年龄、APOE ε4携带状态及家族史特征（受累亲属数及遗传模式）。在建模队列中构建多因素逻辑回归模型，并通过ROC曲线下面积（AUC）与Hosmer–Lemeshow拟合优度检验评估区分度与校准度，随后在验证队列中进行外部验证。

结果：

第一部分：

695例患者（中位发病年龄60岁［IQR 53–66］，46%男性）中，AD 424例、FTD 89例、朊蛋白病（Prion disease）18例、VaD 42例、其他神经退行性疾病（ONDD）31例、非AD型（未分类）痴呆91例；家族史阳性236例（34.0%），P/LP变异58例（8.4%）。FH⁺与FH⁻组在基线及纵向认知评分（MMSE、MoCA、ADL）无统计学差异（FDR > 0.05）。两组整体MRI体积无特异性差异。P/LP携带者较非携带者发病年龄显著提前（中位54.5 vs. 60.0岁，p<0.001），虽MMSE、MoCA衰退速率相当，携带者ADL衰退更快（β=+3.66 Z/年，p=3.7×10⁻⁴），MRI显示脑实质体积损失更快，包括双侧海马及左顶下小叶等表现出显著萎缩，伴随侧脑室和脑脊液体积增大。对APP、PSEN1/2及NOTCH3、COL4A2、FOXC1等相关基因的关联分析，提示携带者可合并脑淀粉样血管病、FTD样或CJD样临床表现及癫痫等非典型特征。

第二部分：

在家族史阳性患者中，建模队列（中位发病年龄66岁），10.3%的患者检出P/LP基因变异。多因素逻辑回归分析显示，发病年龄<55岁（OR=2.56，p=0.0098）、受累亲属数>2例（OR=3.32，p=0.0039）、父母疾病状态（OR=4.72，p=0.015）及家族中存在早发病例（OR=2.61，p=0.0096）均与P/LP变异携带显著正相关，而APOE ε4携带与P/LP变异携带显著负相关（OR=0.36，p=0.0041）。该模型在建模队列的ROC曲线下面积（AUC）为0.776（95% CI 0.701–0.853），在时间性验证队列（中位发病年龄58岁）中AUC为0.781（95% CI 0.647–0.914），校准良好。

结论：

家族史阳性患者整体认知及结构MRI进展与无家族史患者无差异，而P/LP携带者发病显著提前、认知功能衰退加速，并在双侧海马及左侧顶下小叶出现更明显的萎缩，伴随脑室和脑脊液体积快速扩张。我们构建了一项多因素逻辑回归模型（预测变量：发病年龄、APOE ε4 载体状态和家族史评分），用于预测家族史阳性痴呆患者是否携带致病基因致病/可能致病（P/LP）变异。该模型在预测P/LP突变携带方面表现出较高的区分能力，可为精准基因筛查和早期干预提供有力支持。

Background:

Population aging is driving a surge in dementia, yet routine clinical evaluation and structural imaging frequently misclassify its subtypes—particularly in early‐onset or familial cases. Although CSF Aβ/Tau assays, Aβ/Tau PET, whole‐exome sequencing and repeat‐expansion testing have deepened our biological and genetic understanding, few longitudinal studies have integrated pedigree, biomarker, imaging and genomic data to quantify how inherited risk shapes cognitive decline and regional brain atrophy. In this study, we examine whether inherited risk, defined by a positive family history or by pathogenic/likely pathogenic (P/LP) variants, accelerates cognitive decline and region‐specific brain atrophy, and we develop a simple, clinically based prediction model to identify high‐risk individuals and inform personalized interventions.

Methods:

Part I: Longitudinal Impact of Family History and Pathogenic/Likely Pathogenic Variants on Cognitive Decline and Brain Atrophy in the PUMCH Dementia Biomarker Cohort

We enrolled 695 dementia patients who completed CSF Aβ/Tau or Aβ/Tau‐PET, 100× WES with repeat‐expansion assays, APOE genotyping, standardized cognitive testing (MMSE, MoCA, ADL, AVLT, executive measures) and 65‐region MRI volumetry. We then (1) described the prevalence and distribution of positive family history and P/LP variant carriage; (2) compared cognitive trajectories and 65‐region volumetric differences between family‐history–negative (FH-) and –positive (FH+) groups; and (3) compared the same measures between P/LP carriers and non‐carriers. Longitudinal changes were modeled with linear mixed‐effects—adjusting for follow‐up time, genetic status, family history, sex, education and baseline age—and rare‐variant carriers underwent detailed clinico-imaging-genotype correlation.

Part II: Predictive Accuracy of a Clinical Model for Carriage of Pathogenic/Likely Pathogenic Variants in Patients with Dementia and a Positive Family History at PUMCH

A total of 601 Chinese patients with dementia and a family history were enrolled at Peking Union Medical College Hospital, with 476 in a retrospective derivation cohort and 125 in a temporal validation cohort. Predictive factors included age at onset, APOE ε4 status, and family history characteristics. Model performance was assessed using discrimination and calibration metrics.

Results:

Part I:

At baseline, the 695 biomarker‐confirmed dementia patients (median onset age 60 y [IQR 53–66], 46% male) included 424 AD, 89 FTD, 18 prion disease, 42 VaD, 31 ONDD and 91 non-AD cases; among them, 236 (34.0%) had a positive family history and 58 (8.4%) carried P/LP variants.

Among FH⁺ patients, there were on average 1.14 affected first-degree, 0.32 second-degree and 0.35 early-onset relatives (53% maternal inheritance). FH⁺ and FH⁻ patients showed no statistically significant differences in baseline cognitive scores or longitudinal time‐effect trajectories—either cognitively or in regional MRI atrophy patterns (all FDR > 0.05).

P/LP variants were present in 58 patients (8.4%); carriers’ median age at onset was 54.5 vs. 60.0 years in non-carriers (p < 0.001). While MMSE and MoCA declined at similar rates, carriers showed faster ADL deterioration (β = +3.66 Z/year, p = 3.7×10⁻⁴), accelerated parenchymal atrophy most pronounced in bilateral hippocampus and the left inferior parietal lobule, and more rapid lateral ventricular enlargement and CSF volume increase. Genotype–phenotype analyses of APP, PSEN1/PSEN2, and other dementia-related genes (e.g., AD7B  VCP  MAPT  NOTCH3, COL4A2, FOXC1) in the AD-biomarker cohort revealed atypical features—cerebral amyloid angiopathy, FTD-like presentations, CJD-like syndromes, and epilepsy.

Part II:

In the derivation cohort (median age at onset 66 years), 10.3% carried Pathogenic/Likely Pathogenic Variants. Among patients with dementia, those with age at onset < 55 years (OR 2.56, p = 0.0098), more than two affected relatives (OR 3.32, p = 0.0039), parental disease status (OR 4.72, p = 0.015), and early-onset cases in the family (OR 2.61, p = 0.0096) were positively associated with P/LP Variant carriage, whereas APOE ε4 carriage was inversely associated (OR 0.36, p = 0.0041). The model achieved an area under the curve of 0.776 (95% CI, 0.701–0.853) in the derivation cohort and 0.781 (95% CI, 0.647–0.914) in the validation cohort (median age at onset 58 years), with adequate calibration.

Conclusion:

While FH⁺ patients did not exhibit worse outcomes, P/LP carriers develop symptoms 5.5 years earlier than non-carriers and show faster functional decline (ADL) and pronounced parenchymal loss in the hippocampi and inferior parietal lobule alongside rapid ventricular and CSF expansion.A clinical prediction model demonstrated strong predictive performance for P/LP Variant carriage, supporting its clinical utility in guiding genetic testing. Further research is needed to refine the model.