Li, Yan

Outpatient clinical notes are a rich source of information regarding drug safety. However, data in these notes are currently underutilized for pharmacovigilance due to methodological limitations in text mining. Large language models (LLMs) like Bidirectional Encoder Representations from Transformers (BERT) have shown progress in a range of natural language processing tasks but have not yet been evaluated on adverse event (AE) detection. We adapted a new clinical LLM, University of California – San Francisco (UCSF)‐BERT, to identify serious AEs (SAEs) occurring after treatment with a non‐steroid immunosuppressant for inflammatory bowel disease (IBD). We compared this model to other language models that have previously been applied to AE detection. We annotated 928 outpatient IBD notes corresponding to 928 individual patients with IBD for all SAE‐associated hospitalizations occurring after treatment with a non‐steroid immunosuppressant. These notes contained 703 SAEs in total, the most common of which was failure of intended efficacy. Out of eight candidate models, UCSF‐BERT achieved the highest numerical performance on identifying drug‐SAE pairs from this corpus (accuracy 88–92%, macro F1 61–68%), with 5–10% greater accuracy than previously published models. UCSF‐BERT was significantly superior at identifying hospitalization events emergent to medication use ( P < 0.01). LLMs like UCSF‐BERT achieve numerically superior accuracy on the challenging task of SAE detection from clinical notes compared with prior methods. Future work is needed to adapt this methodology to improve model performance and evaluation using multicenter data and newer architectures like Generative pre‐trained transformer (GPT). Our findings support the potential value of using large language models to enhance pharmacovigilance.

AbstractINTRODUCTIONIncreasing evidence suggests that amyloid reduction could serve as a plausible surrogate endpoint for clinical and cognitive efficacy. The double‐blind phase 3 DIAN‐TU‐001 trial tested clinical and cognitive declines with increasing doses of solanezumab or gantenerumab.METHODSWe used latent class (LC) analysis on data from the Dominantly Inherited Alzheimer Network Trials Unit 001 trial to test amyloid positron emission tomography (PET) reduction as a potential surrogate biomarker.RESULTSLC analysis categorized participants into three classes: amyloid no change, amyloid reduction, and amyloid growth, based on longitudinal amyloid Pittsburgh compound B PET standardized uptake value ratio data. The amyloid‐no‐change class was at an earlier disease stage for amyloid amounts and dementia. Despite similar baseline characteristics, the amyloid‐reduction class exhibited reductions in the annual decline rates compared to the amyloid‐growth class across multiple biomarker, clinical, and cognitive outcomes.DISCUSSIONLC analysis indicates that amyloid reduction is associated with improved clinical outcomes and supports its use as a surrogate biomarker in clinical trials.Highlights We used latent class (LC) analysis to test amyloid reduction as a surrogate biomarker. Despite similar baseline characteristics, the amyloid‐reduction class exhibited remarkably better outcomes compared to the amyloid‐growth class across multiple measures. LC analysis proves valuable in testing amyloid reduction as a surrogate biomarker in clinical trials lacking significant treatment effects.

AbstractINTRODUCTIONStudies suggest distinct differences in the development, presentation, progression, and response to treatment of Alzheimer's disease (AD) between females and males. We investigated sex differences in cognition, neuroimaging, and fluid biomarkers in dominantly inherited AD (DIAD).METHODSThree hundred twenty‐five mutation carriers (55% female) and one hundred eighty‐six non‐carriers (58% female) of the Dominantly Inherited Alzheimer Network Observational Study were analyzed. Linear mixed models and Spearman's correlation explored cross‐sectional sex differences in cognition, cerebrospinal fluid (CSF) biomarkers, Pittsburgh compound B positron emission tomography (11C‐PiB PET) and structural magnetic resonance imaging (MRI).RESULTSFemale carriers performed better than males on delayed recall and processing speed despite similar hippocampal volumes. As the disease progressed, symptomatic females revealed higher increases in MRI markers of neurodegeneration and memory impairment. PiB PET and established CSF AD markers revealed no sex differences.DISCUSSIONOur findings suggest an initial cognitive reserve in female carriers followed by a pronounced increase in neurodegeneration coupled with worse performance on delayed recall at later stages of DIAD.

Importance The Clinical Dementia Rating (CDR) is a well-validated instrument widely used to detect and stage dementia due to Alzheimer disease. The digital Electronic Clinical Dementia Rating (eCDR) can be remotely self-administered and automatically scored, with potential to facilitate efficient dementia screening and staging. Objective To evaluate the association of the eCDR with the CDR and other in-clinic assessments for screening older adults for cognitive impairment. Design, Setting, and Participants This multisite, cross-sectional study used baseline data from a longitudinal, observational study from 2020 to 2023, including up to 3 years of follow-up. Participants were enrolled from 3 Alzheimer Disease Research Centers and the Brain Health Registry. Participants (aged ≥55 years, with a study partner, and no acute or unstable major medical conditions) were recruited during in-clinic visits or by automated emails. Exposures Participants completed the Uniform Data Set Version 3 (UDS; including the CDR) in supervised clinical research settings, and then completed the eCDR remotely, online and unsupervised, using their own device. Main Outcomes and Measures The primary outcomes were eCDR scores (item; categorical box and global; continuous box and global), CDR scores (item; categorical box and global), and UDS assessment scores. Associations were evaluated using linear and logistic regressions. Results A total of 3565 participants were contacted, and 288 were enrolled. Among 173 participants with item-level data (mean [SD] age, 70.84 [7.65] years; 76 women [43.9%]), eCDR to CDR concordance was 90% or higher for 33 items (63%) and 70% to 89% for 13 items (25%). Box (domain) level concordance ranged from 80% (memory) to 99% (personal care). The global score concordance rate was 81%. κ statistics were fair to moderate. Among 206 participants with box and global scores (mean [SD] age, 71.34 [7.68] years; 95 women [46.1%]), eCDR continuous global score was associated with CDR global (categorical) score with an area under the receiver operating characteristic curve of 0.79 (95% CI, 0.70-0.87). Correlations between eCDR and in-clinic UDS assessments were similar to those between CDR sum of box scores and the same in-clinic assessments. Conclusions and Relevance These findings suggest that the eCDR is valid and has potential use for screening and assessment of older adults for cognitive and functional decline related to Alzheimer disease. Instrument optimization and validation in diverse cohorts in remote settings are crucial for evaluating scalability and eCDR utility in clinical research, trials, and health care settings.

AbstractBackgroundDyadic (self and study partner) subjective reports of an individual’s recent cognitive and functional decline are an efficient approach to facilitate identification of older adults with, and at risk for, Mild Cognitive Impairment (MCI). Such methods can be easily adapted for remote administration, but few studies have investigated the ability of remotely‐collected, unsupervised dyadic subjective reports to accurately detect MCI and cognitive decline.MethodsIn 289 older adults enrolled in the Brain health Registry (BHR), with clinically‐confirmed diagnoses of Cognitively‐unimpaired (CU) or MCI, we identified the optimal combination of remote measures to accurately distinguish CU from MCI. We then externally validated our predictive model in 217 participants enrolled in a separate cohort, Validation of Online Methods to Predict Cognitive Decline (eVAL). Predictors included Cogstate Brief Battery One Card Learning (OCL) accuracy; self‐ and study partner‐reported memory concern (SMCs); performance on online adaptations of Everyday Cognition Scale (ECog, self‐ and study partner‐report versions), Geriatric Depression Scale Short Form (GDS); family history of AD, age, gender, and education. We calculated 10‐fold cross validated area under the ROC curves (cvAUC), 95% confidence intervals (CI), and the percentage of participants correctly classified (accuracy) for all models. Likelihood ratio (LR) tests were used to compare model fits.ResultsIn the BHR cohort, models including dyad‐report ECog, family history of AD, GDS, and Cogstate OCL scores distinguished CU from MCI with the highest accuracy (cvAUC = 0.90; 95% CI = 0.83, 0.92; accuracy = 85.5%). Models excluding study partner‐report ECog distinguished diagnostic groups with moderate accuracy (cvAUC = 0.83; 95% CI = 0.78, 0.85; accuracy = 75.9%) When applied to the validation cohort, the optimal training model including dyadic report performed with moderate accuracy ( Table 2 ). In the validation cohort, inclusion of study partner (but not self) report SMC, in addition to dyadic‐report ECog, improved model fit (LR test p<0.01).ConclusionsRemotely‐collected dyadic subjective measures, along with other remote measures, accurately distinguish CU from MCI. Subjective cognitive and functional decline and concerns about memory problems independently contribute to diagnostic discrimination. Results support future utilization of remote dyadic subjective measures to efficiently identify older adults with MCI for prodromal AD clinical trials, and in clinical care settings.

AbstractThe temporal evolutions and relative orderings of Alzheimer disease biomarkers, including CSF amyloid-β42 (Aβ42), Aβ40, total tau (Tau) and phosphorylated tau181 (pTau181), standardized uptake value ratio (SUVR) from the molecular imaging of cerebral fibrillar amyloid-β with PET using the 11C-Pittsburgh Compound-B (PiB), MRI-based hippocampal volume and cortical thickness and cognition have been hypothesized but not yet fully tested with longitudinal data for all major biomarker modalities among cognitively normal individuals across the adult lifespan starting from 18 years. By leveraging a large harmonized database from 8 biomarker studies with longitudinal data from 2609 participants in cognition, 873 in MRI biomarkers, 519 in PET PiB imaging and 475 in CSF biomarkers for a median follow-up of 5–6 years, we estimated the longitudinal trajectories of all major Alzheimer disease biomarkers as functions of baseline age that spanned from 18 to 103 years, located the baseline age window at which the longitudinal rates of change accelerated and further examined possible modifying effects of apolipoprotein E (APOE) genotype. We observed that participants 18–45 years at baseline exhibited learning effects on cognition and unexpected directions of change on CSF and PiB biomarkers. The earliest acceleration of longitudinal change occurred for CSF Aβ42 and Aβ42/Aβ40 ratio (with an increase) and for Tau, and pTau181 (with a decrease) at the next baseline age interval of 45–50 years, followed by an accelerated increase for PiB SUVR at the baseline age of 50–55 years and an accelerated decrease for hippocampal volume at the baseline age of 55–60 years and finally by an accelerated decline for cortical thickness and cognition at the baseline age of 65–70 years. Another acceleration in the rate of change occurred at the baseline age of 65–70 years for Aβ42/Aβ40 ratio, Tau, pTau181, PiB SUVR and hippocampal volume. Accelerated declines in hippocampal volume and cognition continued after 70 years. For participants 18–45 years at baseline, significant increases in Aβ42 and Aβ42/Aβ40 ratio and decreases in PiB SUVR occurred in APOE ɛ4 non-carriers but not carriers. After age 45 years, APOE ɛ4 carriers had greater magnitudes than non-carriers in the rates of change for all CSF biomarkers, PiB SUVR and cognition. Our results characterize the temporal evolutions and relative orderings of Alzheimer disease biomarkers across the adult lifespan and the modification effect of APOE ɛ4. These findings may better inform the design of prevention trials on Alzheimer disease.

AbstractIntroductionTo address the need for remote assessments of cognitive decline and dementia, we developed and administered electronic versions of the Clinical Dementia Rating (CDR®) and the Financial Capacity Instrument‐Short Form (FCI‐SF) (F‐CAP®), called the eCDR and eFCI, respectively.MethodsThe CDR and FCI‐SF were adapted for remote, unsupervised, online use based on item response analysis of the standard instruments. Participants completed the eCDR and eFCI first in clinic, and then at home within 2 weeks.ResultsOf the 243 enrolled participants, 179 (73%) cognitively unimpaired (CU), 50 (21%) with mild cognitive impairment (MCI) or dementia, and 14 (6%) with an unknown diagnosis, 84% and 85% of them successfully completed the eCDR and eFCI, respectively, at home.DiscussionThese results show initial feasibility in developing and administering online instruments to remotely assess and monitor cognitive decline along the CU to MCI/very mild dementia continuum. Validation is an important next step.