Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jun 17.
Published in final edited form as: J Alzheimers Dis. 2023;95(4):1609–1622. doi: 10.3233/JAD-230317

Precision Medicine for Preventing Alzheimer’s Disease: Analysis of the ADAPT Study

Sid E O’Bryant a,b,*, Fan Zhang a,b, Leigh A Johnson a,c, James Hall a,b, Melissa Petersen a,b, Esther S Oh d,e,f, Constantine G Lyketsos e,f, Robert A Rissman g,h
PMCID: PMC13270990  NIHMSID: NIHMS2082024  PMID: 37718801

Abstract

Background:

The Alzheimer’s Disease Anti-inflammatory Prevention Trial (ADAPT) was the first-ever large-scale anti-inflammatory prevention trial targeting Alzheimer’s disease.

Objective:

The overall goal of this study was to evaluate predictive blood biomarker profiles that identified individuals most likely to be responders on NSAID treatment or placebo at 12 and 24 months.

Methods:

Baseline (n = 193) and 12-month (n = 562) plasma samples were assayed. The predictive biomarker profile was generated using SVM analyses with response on treatment (yes/no) as the outcome variable.

Results:

Baseline (AUC = 0.99) and 12-month (AUC = 0.99) predictive biomarker profiles were highly accurate in predicting response on Celecoxib arm at 12 and 24 months. The baseline (AUC = 0.95) and 12-month (AUC = 0.9) predictive biomarker profile predicting response on Naproxen were also highly accurate at 12 and 24 months. The baseline (AUC = 0.93) and 12-month (AUC = 0.99) predictive biomarker profile was also highly accurate in predicting response on placebo. As with our prior work, the profiles varied by treatment arm.

Conclusions:

The current results provide additional support for a precision medicine model for treating and preventing Alzheimer’s disease.

Keywords: Alzheimer’s disease, bioinformatics, biomarkers, clinical trial, inflammation, precision medicine, prevention, proteomics

INTRODUCTION

To date, Alzheimer’s disease (AD) therapeutic trials (and approved therapies) utilize a “one-size-fits-all” approach assuming that each individual has same likelihood of response to a specific intervention strategy. Dementias have an annual healthcare cost similar to that of heart disease and more than cancer [1]; however, while death rates due to heart disease have declined, death rates due to dementia have steadily increased [2]. We previously proposed that AD is not homogeneous, but rather, a heterogeneous condition with different likelihoods of response to different treatments in different disease subgroups [3, 4]. Here we test this precision medicine approach assessing targeted nonsteroidal anti-inflammatory drug (NSAID) therapy to prevent AD by leveraging samples and data from the Alzheimer’s Disease Anti-inflammatory Prevention Trial (ADAPT) [5].

While a precision medicine approach targeting specific subpopulations of patients most likely to respond to a given therapy has been proposed for AD [3, 6, 7], few studies have explicitly tested the paradigm. Precision medicine [8] is heavily reliant on biomarker driven therapy, an approach that has led to significant advancements in cancer therapeutics [9]. In 2001, Spear and colleagues [10] estimated that efficacy rates of cancer therapeutics were about 25%. Subsequently, significant improvements have been observed through the use of biomarker driven therapy [11] following the development of trastuzumab for the treatment of specific patients with a particular biomarker positive form of breast cancer [12]. In our previous work, a proteomic-based precision medicine approach identified treatment responders with high accuracy when examining previously conducted trials in AD [3, 4] and Parkinson’s disease [13].

The FDA defines a “Predictive Biomarker” as “a biomarker used to identify individuals who are more likely than similar patients without the biomarker to experience a favorable or unfavorable effect from a specific intervention or exposure [14].” It is our view that the failure of clinical trials targeting AD is largely due to the fact that “most medical treatments are designed for the ‘average patient; as a one-size-fits-all approach” [8, 17]. This approach does not consider the substantial biological heterogeneity between patients [6, 18]. If a treatment was appropriate and effective specifically for only 10% AD patients, all previously utilized trial designs would likely fail to identify the specific group of treatment-responsive patients. A predictive biomarker has a high likelihood of identifying the specific subset of patients likely to respond to a specific therapy, thereby facilitating success by limiting enrollment to individuals most likely to benefit from an intervention. Following previously published methods [4, 14, 15], we sought to utilize baseline and 12-month stored blood samples from the ADAPT study to evaluate a predictive biomarker with the specific context of use to identify cognitively normal older adults most likely to experience protection against cognitive decline (responder) after 12 or 24 months of exposure to an NSAID therapy (prevention).

METHODS

Participants

Participants were enrolled in ADAPT [5]. A full characterization of the sample has been published [5]. ADAPT was a multi-site, randomized, double-blind, parallel assignment trial conducted over 45 months. The aim of ADAPT was to examine if specific nonsteroidal anti-inflammatory medications (Naproxen and Celecoxib) could prevent the onset of AD dementia or attenuate cognitive decline. A total of 2,625 participants were recruited between March of 2001 and December of 2004 and randomized into one of the following treatment arms: Celecoxib (200 mg b.i.d per day), Naproxen sodium (220 mg b.i.d. per day), or placebo. A total of 145 completed (512 randomized) the placebo arm, 99 completed (343 randomized) the Celecoxib arm and 98 completed (337 randomized) the Naproxen arm. Of note, on December 17, 2004, the study was suspended prior to completion due to the National Cancer Institute – sponsored Adenoma Prevention with Celecoxib (APC) trial announced significantly increased cardiovascular risk with celecoxib. On March 31, 2005, the ADAPT Steering Committee made the suspension permanent.

This study protocol was reviewed and approved by the UNTHSC IRB protocols UNTHSC 2016–128 & 2020–125. Each participant (or his/her legal representative) signed written informed consent to participate in the study.

ADAPT inclusion and exclusion criteria

Inclusion criteria

Participants were aged 70 or older with a first degree relative who suffered age-related memory loss, senility, dementia, or AD.

Exclusion criteria

Personal history of cognitive impairment or dementia; hypersensitivity to aspirin, ibuprofen, celecoxib, naproxen, or other NSAIDs; use of anti-coagulant medication; current alcohol abuse or dependence; history of peptic ulcer disease with bleeding or obstruction; clinically significant liver or kidney disease.

Assays

Blood sample collection and banking in ADAPT was initiated well after recruitment began but before it was completed. As such, while some participants provided sample before randomization the majority did not. Banked ADAPT samples were mailed frozen to be assayed at the University of North Texas Health Science Center Institute for Translational Research (ITR) biomarker core.

Sample collection

Blood from each study participant was collected in EDTA coated tubes, centrifuged for 10 min at 1,200 RPM within one hour of collection and dispensed into 0.5 ml aliquots that were immediately frozen.

Sample preparation

Previously unthawed samples were prepared for proteomic assay using the Hamilton Robotic StarPlus system, which is an automated liquid handling workstation that improves the quality of assays, QA/QC monitoring, and proteomic capacity in the ITR biomarker core. Re-aliquoting, when necessary, was conducted with the Hamilton easyBlood robotic system.

Sample assay

Plasma samples were assayed on a multi-plex biomarker assay platform using electrochemiluminescence (ECL) technology, which uses labels that emit light when electrochemically stimulated thereby improving sensitivity of detection of analytes even at low concentrations. ECL assays were conducted using our previously published proteomic panel [3, 19–22] that included the following proteins: fatty acid binding protein, beta 2 microglobulin, pancreatic polypeptide, CRP, ICAM-1, thrombopoeitin, α2 macroglobulin, exotaxin 3, tumor necrosis factor α, tenascin C, interleukin (IL)-5, IL-6, IL-7, IL-10, IL-18, I-309, Factor VII, VCAM 1, TARC, and SAA. The ITR laboratory has assayed over n > 20,000 samples on these markers using this system. Inter- and intra-assay variability has been excellent [19, 20, 22]. Average CVs (>3,000 samples) for these assays are all <10% with the majority being <=5% [19, 20].

Statistical analyses

The predictive biomarker profile was generated using all proteomic data through support vector machine (SVM) analyses. SVM is based on the concept of decision planes that defines decision boundaries and is primarily a classifier method that performs classification tasks by constructing hyperplanes in a multidimensional space that separates cases of different class labels. SVM analyses have the capacity of simultaneously considering a large volume of data to generate an overall profile (e.g., over and under-expression of each values of the ECL-measured proteins) that most accurately classifies multiple outcomes rather than only binary outcomes [23]. The SVM analyses were conducted with the e1071 package (v1.6–8) in R (v3.4.2). Predictive biomarkers guide therapeutic decisions and, therefore, accuracy must be as high as possible (particularly positive predictive value). Therefore, future studies will seek to determine if a smaller, optimized, predictive biomarker profile can be generated. However, given the heterogeneity of the various biological pathways for cognitive decline, we elected to use the full targeted panel that we have previously shown to be highly accurate in detecting AD [19].

Treatment response definition

The predictive algorithm was built on a dichotomous outcome of improved/stable versus declined defined by change in Mini-Mental Status Examination (MMSE) scores from baseline to 12 months and or to 24 months, which was based on our previously published methods [4, 14]. Incident dementia (yes/no), based on consensus panel diagnosis at 24 months after baseline was used as a dichotomous outcome variable in models. Analyses were run by treatment arm to determine drug-specific profiles by examining variable importance plots which was based on our previously published methods [4, 14]. Diagnostic accuracy was examined using ROC analyses with area under the curve (AUC), sensitivity (SN), and specificity (SP) values calculated.

RESULTS

For this study, plasma assays were derived from n = 193 baseline pre-randomization samples (Celecoxib n = 60, Naproxen n = 51, Placebo n = 82) and n = 562 12-month (Celecoxib n = 157, Naproxen n = 159, Placebo n = 246) samples across treatment arms. Table 1 shows descriptive characteristics for ADAPT participants by treatment arm.

Table 1.

Demographic characteristics of the ADAPT trial participants split by study arm at baseline and 12 months

Total Celecoxib Naproxen Placebo
0 months
 No. randomized 193 60 51 82
Age percentiles
 50 73.6 73.2 73 74.1
 25, 75 71.7, 77.5 71.3, 77.1 71.6, 77.0 72.4, 78.8
 0, 100 70.0, 86.4 70.1, 85.4 70.2, 83.9 70.0, 86.4
Sex, %
 Female 40.4 40 37.3 42.7
 Male 59.6 60 62.7 57.3
Education, %
 Less than high school 0 0 0 0
 High school degree 23.8 16.7 21.6 30.5
 College, no degree 48.7 55 49 43.9
 College degree 27.5 28.3 29.4 25.6
Incdem, %
 0 97.9 98.3 98 97.6
 1 2.1 1.7 2 2.4
12 months
 No. randomized 562 157 159 246
Age percentiles
 50 74.1 74.2 74.1 74.1
 25, 75 71.8, 77.5 71.8, 77.2 71.8, 78.0 71.8, 77.7
 0, 100 70.0, 88.8 70.0, 87.1 70.0, 87.4 70.0, 88.8
Sex
 Female 45.7 45.9 50.9 42.3
 Male 54.3 54.1 49.1 57.7
Education, %
 Less than high school 0 0 0 0
 High school degree 24.6 21.7 23.3 27.2
 College, no degree 47.7 52.2 46.5 45.5
 College degree 27.8 26.1 30.2 27.2
Incdem, %
 0 97.3 95.5 97.5 98.4
 1 2.7 4.5 2.5 1.6
Open in a new tab

Celecoxib arm

Baseline proteomics

On Celecoxib, when measuring change in MMSE scores from baseline to 12 months, 40 participants were in the stable category with 15 decliners. When measuring MMSE change from baseline to 24 months, 31 were in the stable category with 19 decliners. The predictive biomarker algorithm (with an optimal SVM-based cut-off score of 0.886) reached an area under the curve (AUC) of 99.5% for detecting 12-month outcomes with a sensitivity (SN) of 1.00 and specificity (SP) of 0.93 (Fig. 1). The same algorithm (with an optimal SVM-based cut-off score of −0.23) reached an AUC of 95% for detecting 24-month outcomes with baseline proteomics with an SN of 1.00 and SP of 0.47 (Fig. 2).

Fig. 1.

Fig. 1.

Open in a new tab

Baseline predictive biomarker for Celecoxib arm predicting 12-month outcomes.

Fig. 2.

Fig. 2.

Open in a new tab

Baseline predictive biomarker for Celecoxib arm predicting 24-month outcomes.

12-month proteomics

The predictive biomarker algorithm using 12-month proteomics reached an AUC of 99.5% for detecting 12-month outcomes with a SN of 1.00 and SP of 0.87 (with an optimal cut-off score of 0.817) (Fig. 3). When this same algorithm was applied to detect 24-month outcomes, it reached an AUC of 99% with an SN of 1.00 and SP of 0.90 (with an optimal cut-off score of 0.62) (Fig. 4). When baseline proteomics were applied to predict incident dementia at 24 months (n = 59 without dementia; n = 1 incident dementia), the predictive biomarker reached an AUC of 96% with a SN of 0.85 and SP of 1.00 (with an optimal cut-off score of −0.999) (Fig. 5).

Fig. 3.

Fig. 3.

Open in a new tab

12-month predictive biomarker for Celecoxib arm predicting 12-month outcomes.

Fig. 4.

Fig. 4.

Open in a new tab

12-month predictive biomarker for Celecoxib arm predicting 24-month outcomes.

Fig. 5.

Fig. 5.

Open in a new tab

Baseline predictive biomarker for Celecoxib arm predicting incident dementia at 24 months.

Naproxen arm

Baseline proteomics

On Naproxen, after 12 months 30 participants remained stable with 18 decliners based on MMSE change from baseline. After 24 months, 25 remained stable with 11 decliners. The predictive biomarker yielded an AUC of 95% with a SN of 1.00 and SP of 0.78 (with an optimal cutoff-score of 0.69) (Fig. 6). The same algorithm (with an optimal cut-off score of 0.753) reached an AUC of 99% for detecting 24-month outcomes with an SN of 1.00 and SP of 0.91 (Fig. 7).

Fig. 6.

Fig. 6.

Open in a new tab

Baseline predictive biomarker for Naproxen arm predicting 12-month outcomes.

Fig. 7.

Fig. 7.

Open in a new tab

Baseline predictive biomarker for Naproxen arm predicting 24-month outcomes.

12-month proteomics

The predictive biomarker algorithm reached an AUC of 99% with a SN of 1.00 and SP of 0.94 (with an optimal cut-off score of 0.254) for detecting 12-month outcomes (Fig. 8). When the predictive biomarker algorithm using 12-month proteomics was applied to detect outcomes at 24 months the AUC reached 1.00 with a SN of 1.00 and SP of 1.00 (optimal cut-off score of 0.628) (Fig. 9). When baseline proteomics were applied to predict incident dementia at 24 months (n = 50 without dementia; n = 1 incident dementia) in this treatment arm, the predictive biomarker reached an AUC of 100% with a SN of 1.00 and SP of 1.00 (with an optimal cut-off score of −0.983) (Fig. 10).

Fig. 8.

Fig. 8.

Open in a new tab

12-month predictive biomarker for Naproxen arm predicting 12-month outcomes.

Fig. 9.

Fig. 9.

Open in a new tab

12-month predictive biomarker for Naproxen arm predicting 24-month outcomes.

Fig. 10.

Fig. 10.

Open in a new tab

Baseline predictive biomarker for Naproxen arm predicting incident dementia at 24 months.

Placebo arm

Baseline proteomics

On Placebo, after 12 months, 47 remained stable with 27 decliners, while after 24 months 37 remained stable with 25 decliners. The predictive biomarker yielded an AUC of 93% with a SN of 1.00 and SP of 0.70 (with an optimal cutoff-score of 0.653) (Fig. 11). The same algorithm (with an optimal cut-off score of 0.165) reached an AUC of 93% for detecting 24-month outcomes with an SN of 1.00 and SP of 0.52 (Fig. 12).

Fig. 11.

Fig. 11.

Open in a new tab

Baseline predictive biomarker for Placebo arm predicting 12-month outcomes.

Fig. 12.

Fig. 12.

Open in a new tab

Baseline predictive biomarker for Placebo arm predicting 24-month outcomes.

12-month proteomics

The predictive biomarker algorithm reached an AUC of 99% with a SN of 1.00 and SP of 0.82 (with an optimal cut-off score of 0.426) for detecting 12-month outcomes (Fig. 13). When the predictive biomarker algorithm using 12-month proteomics was also applied to detect outcomes at 24 months, the AUC reached 98% with a SN of 1.00 and SP of 0.42 (optimal cut-off score of −0.097) (Fig. 14). When baseline proteomics were applied to this trial arm predict incident dementia at 24 months (n = 80 without dementia; n = 2 incident dementia), the predictive biomarker reached an AUC of 100% with a SN of 1.00 and SP of 1.00 (with an optimal cut-off score of −0.997) (Fig. 15).

Fig. 13.

Fig. 13.

Open in a new tab

12-month predictive biomarker for Placebo arm predicting 12-month outcomes.

Fig. 14.

Fig. 14.

Open in a new tab

12-month predictive biomarker for Placebo arm predicting 24-month outcomes.

Fig. 15.

Fig. 15.

Open in a new tab

Baseline predictive biomarker for Placebo arm predicting incident dementia at 24 months.

DISCUSSION

Here we provide proof-of-concept findings in support of a precision medicine approach to preventing cognitive decline by leveraging the ADAPT study data. The proteomic profile based predictive blood biomarker was highly accurate in predicting response across both treatment arms and on placebo. In recent work [3], we found that treatment-specific blood-based biomarkers could predict treatment response in the previously conducted ADCS NSAID trial for individuals diagnosed with AD. Here we build on that approach for use in prevention studies.

Profiling biological pathways associated with neurodegenerative disease has been posited to highlight novel pathways for therapeutics [24], with inflammation being a major implicated pathway [3, 25, 26]. In animal models, inflammation has been linked to AD-like pathology [27, 28] and anti-inflammatory compounds have been shown to reduce pathology and improve cognition [29, 30]. Multiple cohort studies support a link between inflammation and AD [31, 32] with a meta-analysis of 175 published studies (pooled sample size >13,000) demonstrate alterations in multiple inflammatory markers (including IL6, CRP, and TNF) among AD patients [33]. Additionally, a meta-analysis of nine published longitudinal studies (pooled sample size = 14,654) found a protective effect of NSAID use in terms of AD development with a relative risk of 0.27 (95% CI = 0.13–0.58) associated with long-term use [34]. Despite the strong underlying rationale, randomized clinical trials of NSAIDs to treat or prevent AD have not met predefined trial outcomes [5, 35].

We propose that these trials were in fact successful for specific subgroups of patients. Further, given the pleiotropic effects of most drugs, it is possible that the identification of the most effective predictive biomarkers will consider multiple pathways, some non-inflammatory. In fact, we recently generated a predictive biomarker to identify treatment response on rosiglitazone for AD in the REFLECT trials [4] as well as treatment response to α-tocopherol or deprenyl for Parkinson’s disease in the DATATOP trial [15]. In each of these trials, a predictive biomarker that considered multiple biological pathways was highly accurate in predicting outcomes on specific treatments, as well as in placebo arms, as was seen here.

It is noteworthy that the most predictive profiles varied by timeframe of biomarker collection (i.e., baseline versus 12 months) and treatment arm. The top 5 markers in the variable importance plots for the predictive biomarker on Celecoxib were: baseline – IL10, Eotaxin3, A2M, SAA, and TNC; 12-months – FABP3, TNC, B2M, TPO, and IL5. Therefore, only a single marker overlapped in the top 5. On Naproxen the top 5 markers were as follows: Baseline – TPO, IL5, IL7, I309, and IL6; 12-months – TPO, IL6, CRP, FABP3, and FVII. Therefore, only 2 of the top 5 overlapped in this therapeutic. On Placebo the top 5 markers were as follows: Baseline – SAA, CRP, sICAM, FABP3, and A2M; 12-months – SAA, FABP3, Eotaxin3, TNC, and A2M. Therefore, it is possible that a 3-protein algorithm of SAA, FABP3, and A2M may hold utility for a predictive algorithm for placebo (i.e., general response biomarker), which will be examined in the future.

In our recent examination of proteomic profiles of neurodegeneration in the HABS-HD study, profiles spanned markers of metabolic dysfunction, cardiovascular disease, inflammation as well as AD pathology (amyloid) and neurodegeneration (NfL). These markers varied by disease severity (i.e., control versus MCI versus dementia) as well as ethnicity [36]. Therefore, advancement of a precision medicine approach needs to take into account individual risk factors, comorbidities as well as race/ethnicity. Additional predictive biomarkers are being investigated, but these expanded blood markers (e.g., glucose, HbA1c, GLP-1, insulin, Aβ40, Aβ42, total tau, and NfL) were not available in this dataset. These biological subgroups are currently being examined epidemiologically against the full AT(N) framework biomarkers in the Health & Aging Brain – Health Disparities (HABS-HD) study.

Of note is the accuracy of the profile approach to predicting response in the placebo group. This finding suggests that we may have identified some underlying general response biomarker, which would also be of tremendous benefit to the field. Additional work is underway to investigate this possibility examining data across multiple clinical trials as well as epidemiological cohorts. Given that the response profiles varied across all arms, as is consistent with our prior work, it is likely that there are drug-specific and general response profiles. Additional work is ongoing to seek to determine other potential reasons for the differences for the profiles at different timepoints. It is possible that these biomarkers vary as a function of the state of cognitive decline, which is being investigated. A limitation to the study is the sample size, particularly for predicting incident dementia. Future work will be undertaken to combine data from across the range of clinical trials we have examined to parse these questions out with finer detail. However, the ultimate test of this approach will be the prospective application in intervention trials.

The longitudinal findings here point to the possibility of using this approach for the generation of biomarker profiles for the context of use of surrogate outcomes. A surrogate outcome that predicts treatment response at 12 or 24 months could have substantial impact on trial design by drastically reducing trial duration. In fact, a recently FDA approved putative disease modifying drug (aducanumab) was approved because of effects on a surrogate outcome, not clinical efficacy. Surrogate outcomes are used in clinical trials across diseases (e.g., oncology, endocrinology); however, this approach has not been fully developed or implemented within clinical trials of for neurodegenerative disease. Overall, the current findings provide proof-of-concept for using NSAIDs to prevent cognitive loss among a specific subset of cognitively normal older adults. While promising, these findings must be externally validated as well as directly applied to a novel clinical trial prior to clinical application being considered.

ACKNOWLEDGMENTS

The research team thanks the local Fort Worth community and the participants of the HABS-HD study. HABS-HD Study Team: MPIs: Sid E. O’Bryant, Kristine Yaffe, Arthur Toga, Robert Rissman, and Leigh Johnson, and the HABS-HD Investigators: Meredith Braskie, Kevin King, Matthew Borzage, James R. Hall, Melissa Petersen, Raymond Palmer, Robert Barber, Yonggang Shi, Fan Zhang, Rajesh Nandy, Roderick McColl, David Mason, Bradley Christian, Nicole Philips, and Stephanie Large.

FUNDING

This work was supported by the National Institute on Aging of the National Institutes of Health under Award Numbers R01AG054073 and R01AG058533. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

CONFLICT OF INTEREST

S.E.O. has multiple patents on precision medicine for neurodegenerative diseases and is the founding scientist of Cx Precision Medicine. S.E.O. is an Editorial Board Member of this journal but was not involved in the peer-review process nor had access to any information regarding its peer-review.

The other authors declare no conflict of interest.

DATA AVAILABILITY

The data is available to the scientific community through the UNTHSC Institute for Translational Research (ITR) website.

REFERENCES

  • [1].Hurd MD, Martorell P, Delavande A, Mullen KJ, Langa KM (2013) Monetary costs of dementia in the United States. N Engl J Med 368, 1326–1334. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [2].(2020) 2020 Alzheimer’s disease facts and figures. Alzheimers Dement 16, 391–460. [DOI] [PubMed] [Google Scholar]
  • [3].O’Bryant SE, Zhang F, Johnson LA, Hall J, Edwards M, Grammas P, Oh E, Lyketsos CG, Rissman RA (2018) A precision medicine model for targeted NSAID therapy in Alzheimer’s disease. J Alzheimers Dis 66, 97–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [4].O’Bryant SE, Zhang F, Petersen M, Johnson L, Hall J, Rissman RA (2021) A precision medicine approach to treating Alzheimer’s disease using rosiglitazone therapy: A biomarker analysis of the REFLECT Trials. J Alzheimers Dis 81, 557–568. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [5].Lyketsos CG, Breitner JCS, Green RC, Martin BK, Meinert C, Piantadosi S, Sabbagh M (2007) Naproxen and celecoxib do not prevent AD in early results from a randomized controlled trial. Neurology 68, 1800–1808. [DOI] [PubMed] [Google Scholar]
  • [6].Hampel H, O’Bryant SE, Castrillo JI, Ritchie C, Rojkova K, Broich K, Benda N, Nisticò R, Frank RA, Dubois B, Escott-Price V, Lista S (2016) PRECISION MEDICINE - The Golden Gate for detection, treatment and prevention of Alzheimer’s disease. J Prev Alzheimers Dis 3, 243–259. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Hampel H, O’Bryant SE, Durrleman S, Younesi E, Rojkova K, Escott-Price V, Corvol JC, Broich K, Dubois B, Lista S (2017) A Precision Medicine Initiative for Alzheimer’s disease: The road ahead to biomarker-guided integrative disease modeling. Climacteric 20, 107–118. [DOI] [PubMed] [Google Scholar]
  • [8].Weil AR (2018) Precision medicine. Health Aff 37, 687–687. [DOI] [PubMed] [Google Scholar]
  • [9].Jørgensen JT (2015) Clinical application of companion diagnostics. Trends Mol Med 21, 405–407. [DOI] [PubMed] [Google Scholar]
  • [10].Spear BB, Heath-Chiozzi M, Huff J (2001) Clinical application of pharmacogenetics. Trends Mol Med 7, 201–204. [DOI] [PubMed] [Google Scholar]
  • [11].Jørgensen JT (2015) Companion diagnostics: The key to personalized medicine. Foreword. Expert Rev Mol Diagn 15, 153–156. [DOI] [PubMed] [Google Scholar]
  • [12].Slamon DJ, Leyland-Jones B, Shak S, Fuchs H, Paton V, Bajamonde A, Fleming T, Eiermann W, Wolter J, Pegram M, Baselga J, Norton L (2001) Use of chemotherapy plus a monoclonal antibody against HER2 for metastatic breast cancer that overexpresses HER2. N Engl J Med 344, 783–792. [DOI] [PubMed] [Google Scholar]
  • [13].O’Bryant SE, Petersen M, Zhang F, Johnson L, Mason D, Hall J (2022) Analysis of a precision medicine approach to treating Parkinson’s disease: Analysis of the DATATOP study. Parkinsonism Relat Disord 94, 15–21. [DOI] [PubMed] [Google Scholar]
  • [14].FDA-NIH Biomarker Working Group (2016) BEST (Biomarkers, EndpointS, and other Tools) Resource Glossary. Food and Drug Administration, Silver Springs, MD. [PubMed] [Google Scholar]
  • [15].O’Bryant SE, Edwards M, Johnson L, Hall J, Villarreal AE, Britton GB, Quiceno M, Cullum CM, Graff-Radford NR (2016) A blood screening test for Alzheimer’s disease. Alzheimers Dement 3, 83–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [16].O’Bryant SE, Edwards M, Zhang F, Johnson LA, Hall J, Kuras Y, Scherzer CR (2019) Potential two-step proteomic signature for Parkinson’s disease: Pilot analysis in the Harvard Biomarkers Study. Alzheimers Dement 11, 374–382. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [17].O’Bryant SE, Xiao G, Zhang F, Edwards M, German DC, Yin X, Como T, Reisch J, Huebinger RM, Graff-Radford N, Dickson D, Barber R, Hall J, O’Suilleabhain P, Grammas P (2014) Validation of a serum screen for Alzheimer’s disease across assay platforms, species, and tissues. J Alzheimers Dis 42, 1325–1335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [18].O’Bryant SE, Ferman TJ, Zhang F, Hall J, Pedraza O, Wszolek ZK, Como T, Julovich D, Mattevada S, Johnson LA, Edwards M, Hall J, Graff-Radford NR (2019). A proteomic signature for dementia with Lewy bodies. Alzheimers Dement 11, 270–276. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [19].Cortes C, Vapnik V (1995) Support-vector networks. Mach Learn 20, 273–297. [Google Scholar]
  • [20].Hu WT, Chen-Plotkin A, Arnold SE, Grossman M, Clark CM, Shaw LM, McCluskey L, Elman L, Karlawish J, Hurtig HI, Siderowf A, Lee VM, Soares H, Trojanowski JQ (2010) Biomarker discovery for Alzheimer’s disease, frontotemporal lobar degeneration, and Parkinson’s disease. Acta Neuropathol 120, 385–399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].Durrenberger PF, Fernando FS, Kashefi SN, Bonnert TP, Seilhean D, Nait-Oumesmar B, Schmitt A, Gebicke-Haerter PJ, Falkai P, Grünblatt E, Palkovits M, Arzberger T, Kretzschmar H, Dexter DT, Reynolds R (2015) Common mechanisms in neurodegeneration and neuroinflammation: A BrainNet Europe gene expression microarray study. J Neural Transm 122, 1055–1068. [DOI] [PubMed] [Google Scholar]
  • [22].Heneka MT, Kummer MP, Latz E (2014) Innate immune activation in neurodegenerative disease. Nat Rev Immunol 14, 463–477. [DOI] [PubMed] [Google Scholar]
  • [23].Bemiller SM, McCray TJ, Allan K, Formica SV, Xu G, Wilson G, Kokiko-Cochran ON, Crish SD, Lasagna-Reeves CA, Ransohoff RM, Landreth GE, Lamb BT (2017) TREM2 deficiency exacerbates tau pathology through dysregulated kinase signaling in a mouse model of tauopathy. Mol Neurodegener 12, 74. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [24].Metaxas A, Vaitheeswaran R, Jensen KT, Thygesen C, Ilkjaer L, Darvesh S, Finsen B (2018) Reduced serotonin transporter levels and inflammation in the midbrain raphe of 12 month old APPswe/PSEN1dE9 mice. Curr Alzheimer Res 15, 420–428. [DOI] [PubMed] [Google Scholar]
  • [25].Lee DY, Hwang CJ, Choi JY, Park MH, Song MJ, Oh KW, Han SB, Park WK, Cho HY, Cho SY, Park HB, Song MJ, Hong JT (2017) KRICT-9 inhibits neuroinflammation, amyloidogenesis and memory loss in Alzheimer’s disease models. Oncotarget 8, 68654–68667. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [26].Sundaram JR, Poore CP, Sulaimee NHB, Pareek T, Cheong WF, Wenk MR, Pant HC, Frautschy SA, Low CM, Kesavapany S (2017) Curcumin ameliorates neuroinflammation, neurodegeneration, and memory deficits in p25 transgenic mouse model that bears hallmarks of Alzheimer’s disease. J Alzheimers Dis 60, 1429–1442. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [27].Schmidt R, Schmidt H, Curb JD, Masaki K, White LR, Launer LJ (2002) Early inflammation and dementia: A 25-year follow-up of the Honolulu-Asia Aging Study. Ann Neurol 52, 168–174. [DOI] [PubMed] [Google Scholar]
  • [28].Anthony JC, Breitner JC, Zandi PP, Meyer MR, Jurasova I, Norton MC, Stone SV (2000) Reduced prevalence of AD in users of NSAIDs and H2 receptor antagonists: The Cache County study. Neurology 54, 2066–2071. [DOI] [PubMed] [Google Scholar]
  • [29].Lai KSP, Liu CS, Rau A, Lanctôt KL, Köhler CA, Pakosh M, Carvalho AF, Herrmann N (2017) Peripheral inflammatory markers in Alzheimer’s disease: A systematic review and meta-analysis of 175 studies. J Neurol Neurosurg Psychiatry 88, 876–882. [DOI] [PubMed] [Google Scholar]
  • [30].Etminan M, Gill S, Samii A (2003) Effect of non-steroidal anti-inflammatory drugs on risk of Alzheimer’s disease: Systematic review and meta-analysis of observational studies. BMJ 327, 128. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Aisen PS, Schafer KA, Grundman M, Pfeiffer E, Sano M, Davis KL, Farlow MR, Jin S, Thomas RG, Thal LJ; Alzheimer’s Disease Cooperative Study (2003) Effects of rofecoxib or naproxen vs placebo on Alzheimer disease progression: A randomized controlled trial. JAMA 289, 2819–2826. [DOI] [PubMed] [Google Scholar]
  • [32].O’Bryant SE, Zhang F, Petersen M, Hall JR, Johnson LA, Yaffe K, Mason D, Braskie M, Barber RA, Rissman RA, Mapstone M, Mielke MM, Toga AW; HABLE Study Team (2022) A blood screening tool for detecting mild cognitive impairment and Alzheimer’s disease among community-dwelling Mexican Americans and non-Hispanic Whites: A method for increasing representation of diverse populations in clinical research. Alzheimers Dement 18, 77–87. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data is available to the scientific community through the UNTHSC Institute for Translational Research (ITR) website.

RESOURCES