The Influence of Baseline Alzheimer's Disease Severity on Cognitive Decline and CSF Biomarkers in the NILVAD Trial

Abstract

We examined the effects of a dihydropyridine calcium channel blocker nilvadipine with anti-inflammatory properties on cognition and cerebrospinal fluid (CSF) biomarkers by baseline Alzheimer's disease (AD) severity. Exploratory analyses were performed on the dataset (n = 497) of a phase III randomized placebo-controlled trial to examine the response to nilvadipine in AD subjects stratified by baseline AD severity into very mild (MMSE ≥ 25), mild (MMSE 20-24) and moderate AD (MMSE < 20). The outcome measures included total and subscale scores of the Alzheimer's Disease Assessment Scale Cognitive 12 (ADAS-Cog 12), the Clinical Dementia Rating Scale sum of boxes (CDR-sb) and the AD composite score (ADCOMS). Cerebrospinal fluid biomarkers Aβ38, Aβ40, Aβ42, neurofilament light chain (NFL), neurogranin, YKL-40, total tau and P181 tau (ptau) were measured in a subset of samples (n = 55). Regression analyses were adjusted for confounders to specifically examine the influence of nilvadipine and baseline AD severity on cognitive outcomes over 78-weeks. Compared to their respective placebo-controls, nilvadipine-treated, very mild AD subjects showed less decline, whereas moderate AD subjects showed a greater cognitive decline on the ADAS-Cog 12 test and the ADCOMS. A lower decline was observed after nilvadipine treatment for a composite memory trait in very mild AD subjects and a composite language trait in mild AD subjects. Cerebrospinal fluid Aβ42/Aβ40 ratios were increased in mild AD and decreased in moderate AD patients treated with nilvadipine, compared to their respective controls. Among moderate AD subjects, levels of ptau, total tau, neurogranin and YKL-40 increased in subjects treated with nilvadipine compared to placebo. These studies suggest that baseline AD severity influenced the treatment outcome in the NILVAD trial and that future clinical trials of nilvadipine should be restricted to mild and very mild AD patients. Trial Registration: NCT02017340 Registered 20 December 2013, https://ichgcp.net/clinical-trials-registry/NCT02017340 EUDRACT Reference Number 2012-002764-27 Registered 04 February 2013, https://www.clinicaltrialsregister.eu/ctr-search/search?query=2012-002764-27.

Keywords: cerebrospinal fluid Aβ42/Aβ40 ratios; cognitive decline; exploratory analysis; mild Alzheimer's disease; nilvadipine.

Copyright © 2020 Abdullah, Crawford, Tsolaki, Börjesson-Hanson, Olde Rikkert, Pasquier, Wallin, Kennelly, Ait-Ghezala, Paris, Hendrix, Blennow, Lawlor and Mullan.

Figures

Figure 1

Further stratification of the mild AD group by increasing the increment of MMSE scores by 1 starting from ≥20 up to ≥25. Mean ± SE (for MMSE ≥ 20 n = 154 for nilvadipine and n = 157 for placebo; ≥21 n = 125 for nilvadipine and 136 for placebo; ≥22 n = 103 for nilvadipine and n = 117 for placebo; ≥23 n = 70 for nilvadipine and n = 99 for placebo; ≥ 24 n = 57 for nilvadipine and n = 68 for placebo; ≥25 n = 36 for nilvadipine and n = 44 for placebo). Mean change from baseline for the total ADAS-Cog 12 scores show the least decline among nilvadipine-treated subjects compared to placebo-treated subjects with MMSE score of ≥25.

Figure 2

Grouping of ADAS-cog and CDR-sb sub-scales to examine specific cognitive traits (A) Sub-scales of the ADAS-cog 12 test were grouped based on traits for memory, language and praxis to account for the topography of tissue loss in AD depending on the stage of disease. (B) Sub-scales of the ADAS-cog 12 and the CDR-sb were analyzed by PCA, which resulted in grouping of sub-scales into four factors that explained most of the variance in the dataset. Composite variables were then generated, which included sub-scales identified in each factor by PCA, and were named factors 1 through 4. Note, factor 1 also contains the orientation sub-scale from the ADAS-Cog in addition to the ones from the CDR-sb.

Figure 3

Data on ADCOMS and ADAS-Cog 12 test. Nilvadipine-treated very mild AD subjects show less cognitive decline compared to controls on the ADCOMS and the ADAS-Cog 12 tests. Mean ± SE [n = 82 for moderate AD (MMSE ≤ 19) on nilvadipine, n = 94 for moderate AD on placebo, n = 118 for mild AD (MMSE 20-24) on nilvadipine, n = 113 for mild AD on placebo, n = 36 for very mild AD (MMSE ≥ 35) on nilvadipine and n = 44] for very mild AD on placebo for the change in ADAS-Cog 12 scores. There was a significant effect for the interaction between treatment, time and baseline AD severity as assessed by MMSE scores after correcting for the confounding effects of APOE, gender and education, p < 0.05. (A) Stratifications show that very mild AD subjects treated with nilvadipine have lower scores on the ADCOMS and the ADAS-Cog 12 compared to placebo after 78 weeks. post-hoc analysis stratified by time show a significant treatment effect at 78 weeks for the ADCOMS. (B) Mild AD subjects treated with nilvadipine scored similarly to their placebo controls on both the ADCOMS and the ADAS-Cog 12 (C). However, moderate AD subjects treated with nilvadipine scored higher on both the ADCOMS and the ADAS-Cog 12 at 78 weeks compared to those on placebo. *p < 0.05.

Figure 4

Nilvadipine treatment effects on cognitive traits. Very mild AD subjects show less decline on the memory trait, whereas mild AD subjects show less decline on the language trait, after nilvadipine treatment compared to placebo. Mean ± SE (n = 82 for moderate AD on nilvadipine, n = 94 for moderate AD on placebo, n = 118 for mild AD on nilvadipine, n = 113 for mild AD on placebo, n = 36 for very mild AD on nilvadipine, n = 44 for very mild AD on placebo) for the change in memory, language and praxis traits of grouped ADAS-cog 12 sub-scales. (A) There was a significant effect for the interaction between treatment, time and baseline AD severity on the memory trait. post-hoc stratifications by time show that very mild AD subjects treated with nilvadipine had significantly less decline on the memory trait compared to their controls. (B) There was also a significant interaction between treatment, time and baseline AD severity for the language trait. post-hoc stratifications by time show that mild AD subjects treated with nilvadipine had less decline on the language trait compared to the placebo-treated mild AD subjects. (C) There was no effect seen for the praxis trait. *p < 0.05.

Figure 5

Cerebrospinal fluid Aβ biomarkers. Ratios of CSF Aβ42/Aβ40 increase in nilvadipine-treated mild AD but decrease in moderate AD patients compared to their respective placebo groups. (Mean ± SE n = 9 for moderate AD on nilvadipine, n = 12 for moderate AD on placebo, n = 14 for mild AD on nilvadipine, n = 20 for mild AD on placebo). (A) Ratios of Aβ42/Aβ40 were higher in mild AD treated with nilvadipine compared to those treated with placebo (p = 0.067). There was a significant decrease in Aβ42/Aβ40 in moderate AD treated with nilvadipine compared to placebo. Also in moderate AD subjects, levels of (B) Aβ38 and (C) Aβ40 were elevated and (D) Aβ42 levels were unchanged in moderate AD subjects treated with nilvadipine. Levels of Aβ42 were non-significantly higher in mild AD treated with nilvadipine. *p < 0.05.

Figure 6

Cerebrospinal fluid tau biomarkers. Total tau and P181Tau levels were increased in nilvadipine-treated moderate AD patients compared to their respective placebo groups. (Mean ± SE n = 9 for moderate AD on nilvadipine, n = 12 for moderate AD on placebo, n = 14 for mild AD on nilvadipine, n = 20 for mild AD on placebo). Levels of (A) total tau and (B) P181 tau were increased in moderate AD subjects treated with nilvadipine. In mild AD subjects, total tau or P181 tau did not differ between nilvadipine-treated and placebo control groups.*p < 0.05.