Assembling the evidence…
Midlife hypertension (HTN) contributes to cognitive decline in Alzheimer's disease (AD). However, the exact effect of late-life HTN on the AD pathology and on cognitive decline is still controversial. Here, we aimed to assess the impact of HTN and AD pathology in cognitively unimpaired (CU) individuals over 65 years of age on longitudinal cognitive decline. We evaluated 637 CU individuals from two independent cohorts (475 CU individuals from the ADNI cohort; and 162 CU individuals from the TRIAD cohort), with a follow-up of up to 6 years. Linear mixed-effects models showed that HTN and Aβ acted together to promote longitudinal cognitive decline, especially memory loss, in a synergistic way, with a dose-dependent association of blood pressure and Aβ pathology. Hence, HTN in late-life confers additional risk for cognitive decline, particularly for memory loss, in CU individuals at risk of developing dementia due to AD and is a potential modifiable risk factor even at older age.
HTN and Aβ acted together to promote longitudinal cognitive decline, especially memory loss, in a synergistic way, with a dose-dependent association of blood pressure and Aβ pathology.
causal and dose-response wording exceeds observational interactions with uncorrected multiplicity and differing cohort effect magnitudes
Cox-proportional hazard models supported that the Aβ+HTN+ had the highest risk of clinical progression (adjusted HR = 7.79, p < 0.001).
adjusted time-to-event analysis supports ranking, though main-text event counts and confidence intervals are absent
HTN in late-life confers additional risk for cognitive decline, particularly for memory loss, in CU individuals at risk of developing dementia due to AD and is a potential modifiable risk factor even at older age.
prevalent hypertension without onset, treatment, or control data cannot establish causal risk or modifiability
HTN in late-life had no impact in predicting the baseline levels nor the longitudinal trajectory of CSF p-tau181/Aβ1-42 ratio.
a nonsignificant model without equivalence bounds or power analysis does not establish absence of an effect
Derived from the full evaluation — not a separate score.
Strengths
The memory interaction is evaluated in ADNI and then tested in TRIAD, with Aβ defined using CSF in one cohort and PET in the other. This supplies cross-cohort evidence for the direction of the association.
↳ Methods, Biomarkers Assessment; Results, Validation of Findings; Table 2
Mixed-effects models incorporate random intercepts and slopes and examine memory, language, executive, and visuospatial trajectories rather than relying only on a global score.
↳ Methods, Cognitive Assessment and Statistical Analysis; Results, Figure 1 and Table 2
The discussion addresses harmful, protective, and U-shaped late-life blood-pressure findings and contrasts studies reporting biomarker associations with studies reporting null results.
↳ Introduction, paragraph 2; Discussion, paragraphs 2 and 5
Limitations
The abstract and conclusion move from an observational interaction to claims that hypertension promotes decline and that treatment may enhance available Alzheimer’s therapies. No intervention or treatment-response analysis supports that inference.
↳ Abstract; Discussion, conclusion
Hypertension is classified from a recorded diagnosis or antihypertensive use regardless of onset or duration. Treatment status, blood-pressure control, and likely misclassification are not evaluated in the primary exposure definition.
↳ Methods, Participants and Clinical Assessment; Discussion, limitations paragraph
Table 2 contains ancillary point estimates outside their stated confidence intervals and a malformed coefficient, while multiplicity across domains and secondary analyses is not addressed. These issues warrant correction and temper confidence in presentation quality.
↳ Table 2; Methods, Statistical Analysis
The two-cohort design, longitudinal mixed-effects modeling, and domain-specific outcomes support a genuine contribution at the lower end of the well-supported band. Methodological Rigour is bounded by the move from adjusted observational interactions to causal and treatment-oriented conclusions, compounded by a coarse prevalent-hypertension definition. The paper is structurally clear and engages contradictory literature, but several stated limitations are not carried through to the final claims. Table 2’s ancillary confidence-interval inconsistencies and the lack of multiplicity handling further temper confidence without directly overturning the reported memory interaction.
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Sound3.0
Confidence mediumThe study extends prior work through domain-specific longitudinal analyses and a second cohort using a different Aβ ascertainment method. The advance is meaningful but incremental because related vascular-risk interactions had already been reported and the cohort-specific interaction magnitudes differ substantially.
“We recently showed that multiple VRFs, including hypertension, act together with amyloid β and tau pathologies to promote cognitive decline.”
Adjusted mixed-effects models with random intercepts and slopes, attrition checks, sensitivity analyses, and independent-cohort replication support the analytic execution. Binary prevalent hypertension lacks onset, duration, treatment-control, and misclassification assessment, while the treatment-oriented causal conclusion triggers a 2.8 ceiling.
“Proper HTN treatment is a potential strategy to enhance the efficacy of currently available treatments for AD.”
The research question, modeling sequence, domain results, and validation analysis are easy to follow. The abstract and conclusion nevertheless use causal, synergistic, dose-response, and modifiability language that can lead readers beyond what the observational interaction establishes.
“HTN and Aβ acted together to promote longitudinal cognitive decline, especially memory loss, in a synergistic way.”
The discussion engages protective, harmful, U-shaped, biomarker, and treatment findings from the late-life hypertension literature. Its limitations are specific but are not consistently carried through to the general treatment and anti-Aβ combination claims.
“Both cohorts excluded individuals at high cerebrovascular risk, limiting generalizability to broader populations.”
Caveats4 of 4 checks
Several ancillary values are internally inconsistent, although the confidence intervals for the two headline memory-interaction estimates contain their reported point estimates. These issues reduce confidence in table preparation without establishing that the central result is invalid.
Ethics approval, consent, funding, data access, and extensive conflicts are declared. The separate competing-interest declaration is inconsistent with the fuller conflict-of-interest section.
Flags: 3 declared / 5 total
57 references in manuscript 57 of 57 checkable references found in an index
No retraction notice found in Retraction Watch.
Sources: Retraction Watch ✓
Where this paper’s evidence sits on the path from initial observation to real-world use.
The study supplies observational hypothesis-generating evidence rather than a tested intervention. It does not establish that treating hypertension changes memory trajectories or improves anti-Aβ treatment efficacy.
“Proper HTN treatment is a potential strategy to enhance the efficacy of currently available treatments for AD.”
AI-generated, human-governed. Something look off? Contact us to request a review.