Comparative Risk of Alzheimer Disease and Related Dementia Among Medicare Beneficiaries With Rheumatoid Arthritis Treated With Targeted Disease-Modifying Antirheumatic Agents | Dementia and Cognitive Impairment | JAMA Network Open | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  Risk of Alzheimer Disease and Related Dementia in Patients Treated With Tofacitinib, Tocilizumab, or Tumor Necrosis Factor Inhibitors (TNFi) vs Abatacept After 1:1 Propensity Score Matching, Medicare Data 2007-2017
Risk of Alzheimer Disease and Related Dementia in Patients Treated With Tofacitinib, Tocilizumab, or Tumor Necrosis Factor Inhibitors (TNFi) vs Abatacept After 1:1 Propensity Score Matching, Medicare Data 2007-2017

Analysis 1 indicates an as-treated follow-up approach; analysis 2, an as-started follow-up approach incorporating a 6-month induction period; analysis 3, incorporating a 6-month symptom to diagnosis period; analysis 4, alternate outcome definition; and HR, hazard ratio.

Figure 2.  Cumulative Incidence of Alzheimer Disease and Related Dementia in Patients Treated With Tofacitinib, Tocilizumab, or Tumor Necrosis Factor Inhibitors (TNFi) vs Abatacept After 1:1 Propensity Score Matching, Medicare Data 2007-2017
Cumulative Incidence of Alzheimer Disease and Related Dementia in Patients Treated With Tofacitinib, Tocilizumab, or Tumor Necrosis Factor Inhibitors (TNFi) vs Abatacept After 1:1 Propensity Score Matching, Medicare Data 2007-2017

Analysis 1 indicates an as-treated follow-up approach; analysis 2, an as-started follow-up approach incorporating a 6-month induction period; analysis 3, incorporating a 6-month symptom to diagnosis period; analysis 4, alternate outcome definition.

Table 1.  Select Baseline Characteristics of Patients Included in the Study Cohort After 1:1 Propensity Score Matching, Medicare Data 2007-2017
Select Baseline Characteristics of Patients Included in the Study Cohort After 1:1 Propensity Score Matching, Medicare Data 2007-2017
Table 2.  Incidence Rates for Alzheimer Disease and Related Dementia Across 4 Analysis Schemes
Incidence Rates for Alzheimer Disease and Related Dementia Across 4 Analysis Schemes
1.
Cummings  JL, Morstorf  T, Zhong  K.  Alzheimer’s disease drug-development pipeline: few candidates, frequent failures.   Alzheimers Res Ther. 2014;6(4):37. doi:10.1186/alzrt269 PubMedGoogle ScholarCrossref
2.
Yiannopoulou  KG, Anastasiou  AI, Zachariou  V, Pelidou  SH.  Reasons for failed trials of disease-modifying treatments for Alzheimer disease and their contribution in recent research.   Biomedicines. 2019;7(4):E97. doi:10.3390/biomedicines7040097 PubMedGoogle Scholar
3.
Hebert  LE, Weuve  J, Scherr  PA, Evans  DA.  Alzheimer disease in the United States (2010-2050) estimated using the 2010 census.   Neurology. 2013;80(19):1778-1783. doi:10.1212/WNL.0b013e31828726f5 PubMedGoogle ScholarCrossref
4.
Godyń  J, Jończyk  J, Panek  D, Malawska  B.  Therapeutic strategies for Alzheimer’s disease in clinical trials.   Pharmacol Rep. 2016;68(1):127-138. doi:10.1016/j.pharep.2015.07.006 PubMedGoogle ScholarCrossref
5.
Iqbal  K, Liu  F, Gong  CX.  Alzheimer disease therapeutics: focus on the disease and not just plaques and tangles.   Biochem Pharmacol. 2014;88(4):631-639. doi:10.1016/j.bcp.2014.01.002 PubMedGoogle ScholarCrossref
6.
Panza  F, Logroscino  G, Imbimbo  BP, Solfrizzi  V.  Is there still any hope for amyloid-based immunotherapy for Alzheimer’s disease?   Curr Opin Psychiatry. 2014;27(2):128-137. doi:10.1097/YCO.0000000000000041 PubMedGoogle ScholarCrossref
7.
Bennett  DA, Schneider  JA, Arvanitakis  Z,  et al.  Neuropathology of older persons without cognitive impairment from two community-based studies.   Neurology. 2006;66(12):1837-1844. doi:10.1212/01.wnl.0000219668.47116.e6 PubMedGoogle ScholarCrossref
8.
Castellani  RJ, Perry  G.  The complexities of the pathology-pathogenesis relationship in Alzheimer disease.   Biochem Pharmacol. 2014;88(4):671-676. doi:10.1016/j.bcp.2014.01.009 PubMedGoogle ScholarCrossref
9.
Desai  RJ, Varma  VR, Gerhard  T,  et al.  Targeting abnormal metabolism in Alzheimer’s disease: the Drug Repurposing for Effective Alzheimer’s Medicines (DREAM) study.   Alzheimers Dement (N Y). 2020;6(1):e12095. doi:10.1002/trc2.12095 PubMedGoogle Scholar
10.
Pushpakom  S, Iorio  F, Eyers  PA,  et al.  Drug repurposing: progress, challenges and recommendations.   Nat Rev Drug Discov. 2019;18(1):41-58. doi:10.1038/nrd.2018.168 PubMedGoogle ScholarCrossref
11.
Thambisetty  M.  Understanding mechanisms and seeking cures for Alzheimer’s disease: why we must be “extraordinarily diverse.”   Am J Physiol Cell Physiol. 2017;313(4):C353-C361. doi:10.1152/ajpcell.00111.2017 PubMedGoogle ScholarCrossref
12.
Berger  ML, Sox  H, Willke  RJ,  et al.  Good practices for real-world data studies of treatment and/or comparative effectiveness: recommendations from the joint ISPOR-ISPE Special Task Force on real-world evidence in health care decision making.   Value Health. 2017;20(8):1003-1008. doi:10.1016/j.jval.2017.08.3019 PubMedGoogle ScholarCrossref
13.
Smith  GD, Ebrahim  S.  Data dredging, bias, or confounding.   BMJ. 2002;325(7378):1437-1438. doi:10.1136/bmj.325.7378.1437PubMedGoogle ScholarCrossref
14.
Taylor  DH  Jr, Østbye  T, Langa  KM, Weir  D, Plassman  BL.  The accuracy of Medicare claims as an epidemiological tool: the case of dementia revisited.   J Alzheimers Dis. 2009;17(4):807-815. doi:10.3233/JAD-2009-1099 PubMedGoogle ScholarCrossref
15.
Solomon  A, Ngandu  T, Soininen  H, Hallikainen  MM, Kivipelto  M, Laatikainen  T.  Validity of dementia and Alzheimer’s disease diagnoses in Finnish national registers.   Alzheimers Dement. 2014;10(3):303-309. doi:10.1016/j.jalz.2013.03.004 PubMedGoogle ScholarCrossref
16.
Kivipelto  M, Ngandu  T, Laatikainen  T, Winblad  B, Soininen  H, Tuomilehto  J.  Risk score for the prediction of dementia risk in 20 years among middle aged people: a longitudinal, population-based study.   Lancet Neurol. 2006;5(9):735-741. doi:10.1016/S1474-4422(06)70537-3 PubMedGoogle ScholarCrossref
17.
Barnes  DE, Beiser  AS, Lee  A,  et al.  Development and validation of a brief dementia screening indicator for primary care.   Alzheimers Dement. 2014;10(6):656-665.e1. doi:10.1016/j.jalz.2013.11.006PubMedGoogle ScholarCrossref
18.
Albrecht  JS, Hanna  M, Kim  D, Perfetto  EM.  Predicting diagnosis of Alzheimer’s disease and related dementias using administrative claims.   J Manag Care Spec Pharm. 2018;24(11):1138-1145. doi:10.18553/jmcp.2018.24.11.1138 PubMedGoogle Scholar
19.
Brookhart  MA, Patrick  AR, Dormuth  C,  et al.  Adherence to lipid-lowering therapy and the use of preventive health services: an investigation of the healthy user effect.   Am J Epidemiol. 2007;166(3):348-354. doi:10.1093/aje/kwm070 PubMedGoogle ScholarCrossref
20.
Schneeweiss  S, Seeger  JD, Maclure  M, Wang  PS, Avorn  J, Glynn  RJ.  Performance of comorbidity scores to control for confounding in epidemiologic studies using claims data.   Am J Epidemiol. 2001;154(9):854-864. doi:10.1093/aje/154.9.854 PubMedGoogle ScholarCrossref
21.
Kim  DH, Schneeweiss  S, Glynn  RJ, Lipsitz  LA, Rockwood  K, Avorn  J.  Measuring frailty in Medicare data: development and validation of a claims-based frailty index.   J Gerontol A Biol Sci Med Sci. 2018;73(7):980-987. doi:10.1093/gerona/glx229PubMedGoogle ScholarCrossref
22.
Rosenbaum  PR, Rubin  DB.  The central role of the propensity score in observational studies for causal effects.   Biometrika. 1983;70(1):41-55. doi:10.1093/biomet/70.1.41 Google ScholarCrossref
23.
Rassen  JA, Shelat  AA, Myers  J, Glynn  RJ, Rothman  KJ, Schneeweiss  S.  One-to-many propensity score matching in cohort studies.   Pharmacoepidemiol Drug Saf. 2012;21(suppl 2):69-80. doi:10.1002/pds.3263 PubMedGoogle ScholarCrossref
24.
Austin  PC.  Some methods of propensity-score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations.   Biom J. 2009;51(1):171-184. doi:10.1002/bimj.200810488 PubMedGoogle ScholarCrossref
25.
Walker  AM, Patrick  AR, Lauer  MS,  et al.  Tool for assessing the feasibility of comparative effectiveness research.   Comp Effect Res. 2013;3:11-20. doi:10.2147/CER.S40357Google Scholar
26.
Franklin  JM, Rassen  JA, Ackermann  D, Bartels  DB, Schneeweiss  S.  Metrics for covariate balance in cohort studies of causal effects.   Stat Med. 2014;33(10):1685-1699. doi:10.1002/sim.6058 PubMedGoogle ScholarCrossref
27.
Austin  PC, Lee  DS, Fine  JP.  Introduction to the analysis of survival data in the presence of competing risks.   Circulation. 2016;133(6):601-609. doi:10.1161/CIRCULATIONAHA.115.017719 PubMedGoogle ScholarCrossref
28.
Andrew  MK, Tierney  MC.  The puzzle of sex, gender and Alzheimer’s disease: why are women more often affected than men?   Womens Health (Lond). 2018;14:1745506518817995. doi:10.1177/1745506518817995 Google Scholar
29.
Tublin  JM, Adelstein  JM, Del Monte  F, Combs  CK, Wold  LE.  Getting to the heart of Alzheimer disease.   Circ Res. 2019;124(1):142-149. doi:10.1161/CIRCRESAHA.118.313563 PubMedGoogle ScholarCrossref
30.
Fayosse  A, Nguyen  D-P, Dugravot  A,  et al.  Risk prediction models for dementia: role of age and cardiometabolic risk factors.   BMC Med. 2020;18(1):107. doi:10.1186/s12916-020-01578-x PubMedGoogle ScholarCrossref
31.
Wang  SV, Verpillat  P, Rassen  JA, Patrick  A, Garry  EM, Bartels  DB.  Transparency and reproducibility of observational cohort studies using large healthcare databases.   Clin Pharmacol Ther. 2016;99(3):325-332. doi:10.1002/cpt.329 PubMedGoogle ScholarCrossref
32.
Fralick  M, Kesselheim  AS, Avorn  J, Schneeweiss  S.  Use of health care databases to support supplemental indications of approved medications.   JAMA Intern Med. 2018;178(1):55-63. doi:10.1001/jamainternmed.2017.3919 PubMedGoogle ScholarCrossref
33.
Patorno  E, Schneeweiss  S, Gopalakrishnan  C, Martin  D, Franklin  JM.  Using Real-world data to predict findings of an ongoing phase IV cardiovascular outcome trial: cardiovascular safety of linagliptin versus glimepiride.   Diabetes Care. 2019;42(12):2204-2210. doi:10.2337/dc19-0069 PubMedGoogle ScholarCrossref
34.
Chou  RC, Kane  M, Ghimire  S, Gautam  S, Gui  J.  Treatment for rheumatoid arthritis and risk of Alzheimer’s disease: a nested case-control analysis.   CNS Drugs. 2016;30(11):1111-1120. doi:10.1007/s40263-016-0374-z PubMedGoogle ScholarCrossref
35.
Zhou  M, Xu  R, Kaelber  DC, Gurney  ME.  Tumor necrosis factor (TNF) blocking agents are associated with lower risk for Alzheimer’s disease in patients with rheumatoid arthritis and psoriasis.   PLoS One. 2020;15(3):e0229819. doi:10.1371/journal.pone.0229819 PubMedGoogle Scholar
36.
Ray  WA.  Evaluating medication effects outside of clinical trials: new-user designs.   Am J Epidemiol. 2003;158(9):915-920. doi:10.1093/aje/kwg231 PubMedGoogle ScholarCrossref
37.
Schneeweiss  S.  A basic study design for expedited safety signal evaluation based on electronic healthcare data.   Pharmacoepidemiol Drug Saf. 2010;19(8):858-868. doi:10.1002/pds.1926 PubMedGoogle ScholarCrossref
38.
Wallach  JD, Sullivan  PG, Trepanowski  JF, Steyerberg  EW, Ioannidis  JP.  Sex based subgroup differences in randomized controlled trials: empirical evidence from Cochrane meta-analyses.   BMJ. 2016;355:i5826. doi:10.1136/bmj.i5826PubMedGoogle Scholar
39.
Fukuyama  T, Tschernig  T, Qi  Y, Volmer  DA, Bäumer  W.  Aggression behaviour induced by oral administration of the Janus-kinase inhibitor tofacitinib, but not oclacitinib, under stressful conditions.   Eur J Pharmacol. 2015;764:278-282. doi:10.1016/j.ejphar.2015.06.060 PubMedGoogle ScholarCrossref
40.
Dale  RC.  Interleukin-6 blockade as rescue therapy in autoimmune encephalitis.   Neurotherapeutics. 2016;13(4):821-823. doi:10.1007/s13311-016-0471-1 PubMedGoogle ScholarCrossref
41.
Perry  VH, Holmes  C.  Microglial priming in neurodegenerative disease.   Nat Rev Neurol. 2014;10(4):217-224. doi:10.1038/nrneurol.2014.38 PubMedGoogle ScholarCrossref
42.
Hannestad  J, Gallezot  JD, Schafbauer  T,  et al.  Endotoxin-induced systemic inflammation activates microglia: [11C]PBR28 positron emission tomography in nonhuman primates.   Neuroimage. 2012;63(1):232-239. doi:10.1016/j.neuroimage.2012.06.055 PubMedGoogle ScholarCrossref
43.
Butchart  J, Brook  L, Hopkins  V,  et al.  Etanercept in Alzheimer disease: a randomized, placebo-controlled, double-blind, phase 2 trial.   Neurology. 2015;84(21):2161-2168. doi:10.1212/WNL.0000000000001617 PubMedGoogle ScholarCrossref
44.
Yoshida  K, Solomon  DH, Kim  SC.  Active-comparator design and new-user design in observational studies.   Nat Rev Rheumatol. 2015;11(7):437-441. doi:10.1038/nrrheum.2015.30 PubMedGoogle ScholarCrossref
45.
Kern  DM, Lovestone  S, Cepeda  MS.  Treatment with TNF-α inhibitors versus methotrexate and the association with dementia and Alzheimer’s disease.   Alzheimers Dement (N Y). 2021;7(1):e12163. doi:10.1002/trc2.12163 PubMedGoogle Scholar
46.
Tweedie  D, Sambamurti  K, Greig  NH.  TNF-alpha inhibition as a treatment strategy for neurodegenerative disorders: new drug candidates and targets.   Curr Alzheimer Res. 2007;4(4):378-385. doi:10.2174/156720507781788873 PubMedGoogle ScholarCrossref
47.
Torres-Acosta  N, O’Keefe  JH, O’Keefe  EL, Isaacson  R, Small  G.  Therapeutic potential of TNF-α inhibition for Alzheimer’s disease prevention.   J Alzheimers Dis. 2020;78(2):619-626. doi:10.3233/JAD-200711 PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    1 Comment for this article
    The last steps of a very long way
    Fatih Tufan, Assoc. Prof. | Istanbul Aydin University, Florya Medical Park Hospital, Department of Geriatrics
    I congratulate the authors for this relevant and important paper. Recent studies indicate that the onset of clinical symptoms and signs of Alzheimer disease occur decades after the onset of histopathological damage. Various potentially beneficial interventions to date have not led to any benefit, likely because they were administered to relatively older people at risk of Alzheimer disease or those with mild cognitive impairment. Notably, these interventions mostly target inflammation and not  damage. Therefore, most of the damage might have occurred years before the study in this relatively older patient population of the present study.
    CONFLICT OF INTEREST: None Reported
    Views 4,771
    Citations 0
    Original Investigation
    Neurology
    April 8, 2022

    Comparative Risk of Alzheimer Disease and Related Dementia Among Medicare Beneficiaries With Rheumatoid Arthritis Treated With Targeted Disease-Modifying Antirheumatic Agents

    Author Affiliations
    • 1Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts
    • 2Clinical and Translational Neuroscience Section, Laboratory of Behavioral Neuroscience, National Institute on Aging, Baltimore, Maryland
    • 3Center for Pharmacoepidemiology and Treatment Science, Ernest Mario School of Pharmacy, Rutgers University, New Brunswick, New Jersey
    • 4Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland
    • 5Division of Rheumatology, Inflammation, and Immunity, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts
    JAMA Netw Open. 2022;5(4):e226567. doi:10.1001/jamanetworkopen.2022.6567
    Key Points

    Question  Is use of targeted disease-modifying antirheumatic drugs associated with risk of Alzheimer disease and related dementia (ADRD)?

    Findings  In this cohort study including 22 569 propensity score–matched patient pairs, initiation of inhibitors of Janus-kinase, interleukin-6, or tumor necrosis factor was not associated with reduced risk of ADRD compared with initiation of abatacept, a T-cell activation inhibitor.

    Meaning  These results do not support advancing targeted disease-modifying antirheumatic drugs as disease modifying candidates for ADRD.

    Abstract

    Importance  Cytokine signaling, including tumor necrosis factor (TNF) and interleukin (IL)-6, through the Janus-kinase (JAK)–signal transducer and activator of transcription pathway, was hypothesized to attenuate the risk of Alzheimer disease and related dementia (ADRD) in the Drug Repurposing for Effective Alzheimer Medicines (DREAM) initiative based on multiomics phenotyping.

    Objective  To evaluate the association between treatment with tofacitinib, tocilizumab, or TNF inhibitors compared with abatacept and risk of incident ADRD.

    Design, Setting, and Participants  This cohort study was conducted among US Medicare fee-for-service patients with rheumatoid arthritis aged 65 years and older from 2007 to 2017. Patients were categorized into 3 cohorts based on initiation of tofacitinib (a JAK inhibitor), tocilizumab (an IL-6 inhibitor), or TNF inhibitors compared with a common comparator abatacept (a T-cell activation inhibitor). Analyses were conducted from August 2020 to August 2021.

    Main Outcomes and Measures  The main outcome was onset of ADRD based on diagnosis codes evaluated in 4 alternative analysis schemes: (1) an as-treated follow-up approach, (2) an as-started follow-up approach incorporating a 6-month induction period, (3) incorporating a 6-month symptom to diagnosis period to account for misclassification of ADRD onset, and (4) identifying ADRD through symptomatic prescriptions and diagnosis codes. Hazard ratios (HRs) with 95% CIs were calculated from Cox proportional hazard regression after adjustment for 79 preexposure characteristics through propensity score matching.

    Results  After 1:1 propensity score matching to patients using abatacept, a total of 22 569 propensity score–matched patient pairs, including 4224 tofacitinib pairs (mean [SD] age 72.19 [5.65] years; 6945 [82.2%] women), 6369 tocilizumab pairs (mean [SD] age 72.01 [5.46] years; 10 105 [79.4%] women), and 11 976 TNF inhibitor pairs (mean [SD] age 72.67 [5.91] years; 19 710 [82.3%] women), were assessed. Incidence rates of ADRD varied from 2 to 18 per 1000 person-years across analyses schemes. There were no statistically significant associations of ADRD with tofacitinib (analysis 1: HR, 0.90 [95% CI, 0.55-1.51]; analysis 2: HR, 0.78 [95% CI, 0.53-1.13]; analysis 3: HR, 1.29 [95% CI, 0.72-2.33]; analysis 4: HR, 0.50 [95% CI, 0.21-1.20]), tocilizumab (analysis 1: HR, 0.82 [95% CI, 0.55-1.21]; analysis 2: HR, 1.05 [95% CI, 0.81-1.35]; analysis 3: HR, 1.21 [95% CI, 0.75-1.96]; analysis 4: HR, 0.78 [95% CI, 0.44-1.39]), or TNF inhibitors (analysis 1: HR, 0.93 [95% CI, 0.72-1.20]; analysis 2: HR, 1.02 [95% CI, 0.86-1.20]; analysis 3: HR, 1.13 [95% CI, 0.86-1.48]; analysis 4: 0.90 [95% CI, 0.60-1.37]) compared with abatacept. Results from prespecified subgroup analysis by age, sex, and baseline cardiovascular disease were consistent except in patients with cardiovascular disease, for whom there was a potentially lower risk of ADRD with TNF inhibitors vs abatacept, but only in analyses 2 and 4 (analysis 1: HR, 0.76 [95% CI, 0.50-1.16]; analysis 2: HR, 0.74 [95% CI, 0.56-0.99]; analysis 3: HR, 1.03 [95% CI, 0.65-1.61]; analysis 4: HR, 0.45 [95% CI, 0.21-0.98]).

    Conclusions and Relevance  This cohort study did not find any association of risk of ADRD in patients treated with tofacitinib, tocilizumab, or TNF inhibitors compared with abatacept.

    Introduction

    Alzheimer disease and related dementias (ADRD) remain a looming public health challenge, with current prevalence of more than 5 million in the US and with extremely limited disease-modifying treatment options.1-3 The traditional drug discovery approach of identifying targets based on experimental animal models that recapitulate the pathological features of ADRD, including amyloid plaque and neurofibrillary tangle pathologies, has had limited success.4-6 As a result, researchers increasingly recognize a need to identify early molecular triggers preceding accumulation of pathogenesis and clinical symptoms to aid in drug discovery for ADRD.7,8

    We proposed an alternative approach to drug discovery in ADRD in the Drug Repurposing for Effective Alzheimer Medicines (DREAM) initiative,9 which is an ongoing multidisciplinary collaborative study aimed at identifying drug repurposing candidates for ADRD. Briefly, we generate testable hypotheses based on multiomics phenotyping of AD to identify genetic regulators of abnormal metabolic pathways associated with ADRD neuropathogenesis.10,11 Next, we identify existing medications with US Food and Drug Administration approval that act on the identified targets and pursue them as repurposing candidates. Finally, in a hypothesis refinement step, we conduct rigorous patient-level pharmacoepidemiologic analyses using routinely collected health care data to evaluate the association between exposure to the repurposing candidates and incident ADRD, avoiding common pitfalls of noninterventional studies, including immortal time bias and confounding by indication. Our results may lead to novel hypotheses about biological pathways and drug targets associated with ADRD that merit testing in relevant experimental models.

    We recently defined a hypothetical network of interacting and intersecting metabolic pathways in AD, linked to dysregulation in brain glycolysis, the Alzheimer Disease Aberrant Metabolism (ADAM) network.9 We nominated genetic regulators of metabolic and signaling reactions in the ADAM network as plausible AD drug targets. In this network, cytokine signaling, including tumor necrosis factor (TNF) and interleukin (IL)-6, through the Janus-kinase–signal transducer and activator of transcription (JAK/STAT) pathway was nominated as 1 such drug target for pharmacoepidemiologic analyses in the DREAM study.

    In this study, we describe results comparing the risk of ADRD in Medicare beneficiaries with rheumatoid arthritis (RA) who were treated with any of 3 targeted synthetic or biologic disease-modifying antirheumatic drugs (TDMARDs) identified in the hypothesis generation step as repurposing candidates: tofacitinib (a JAK inhibitor), tocilizumab (an IL-6 inhibitor), and TNF inhibitors, compared with an active comparator abatacept (a T-cell activation inhibitor) that is used for similar indication but was not found to be associated with changes in the JAK/STAT pathway in our omics studies. We prespecified all our hypotheses and design choices in our prior publication9 to safeguard against publication bias and data dredging.12,13

    Methods

    This cohort study was approved by the Brigham and Women’s Hospital institutional review board. All the analyses were conducted using anonymized patient data, therefore the institutional review board waived the requirement for informed consent. The full study protocols for patient-level analyses in Medicare claims were preregistered on ClinicalTrials.gov prior to data analysis (ClinicalTrials.gov Identifiers: NCT04529902, NCT04529876, and NCT04529863) and contain detailed information on implementation, including all codes that were used to identify study variables to allow for independent replication and validation. This study is reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

    Data Source

    We used Medicare fee-for-service claims data from 2007 through 2017. Medicare Part A (hospitalizations), B (medical services), and D (prescription medications) claims are available for research purposes through the Centers for Medicare & Medicaid Services (CMS).

    Study Cohort

    A new user, active comparator, observational cohort study design was used in which patients were required to have 365 days of continuous enrollment in Medicare parts A, B, and D before initiation date of study medications of interest. Date of initiation of the TDMARD was defined as the cohort entry date, and the 365-day period prior to cohort entry was defined as the baseline period. Based on the TDMARD initiated, patients were identified in 3 cohorts: tofacitinib vs abatacept, tocilizumab vs abatacept, and TNF inhibitors vs abatacept. We restricted entry to patients with at least 1 diagnosis code of RA during the baseline period and required no prior use of the study TDMARDs in each comparison. To focus on incident events, we excluded patients with existing diagnoses of ADRD any time prior to and including cohort entry date. Patients with a nursing home admission 365 days prior to and including the cohort entry date were excluded, as medication records for short nursing home stays are unavailable in Medicare claims. eFigure 1 in the Supplement shows the study design.

    Outcome Measurement

    The outcome of interest was time to incident ADRD and was operationalized based on diagnosis codes recorded on inpatient or outpatient claims of AD, vascular dementia, senile, presenile, or unspecified dementia, or dementia in other diseases classified elsewhere (eTable 1 in the Supplement). Medicare claims–based dementia identification is reported to have a positive predictive value in the range of 65% to 78% when validated against a structured in-home dementia assessment.14 We evaluated time to incident AD as a secondary outcome of interest.

    Alternative Analytic Approaches

    Pharmacoepidemiologic investigations focused on ADRD risk face numerous uncertainties, including informative censoring, reverse causality bias, and outcome misclassification. To address these concerns, we used the following 4 alternative analyses with equal priority (eFigure 2 in the Supplement).

    Analysis 1: As-Treated Follow-up Approach

    In analysis 1, the follow-up started on the day following the cohort entry date and continued until first of the following events: outcome, treatment discontinuation or switch to the comparator treatment, insurance disenrollment, death, or administrative end point (December 2017). A 90-day grace period after the end of the expected days’ supply of the most recently filled prescription was used to define the treatment discontinuation date to accommodate suboptimal adherence during treatment periods.

    Analysis 2: As-Started Follow-up Approach Incorporating a 6-Month Induction Period

    In analysis 2, we incorporated a 6-month induction period after the cohort entry date before beginning the follow-up for ADRD and followed patients for a maximum of 3 years regardless of subsequent treatment changes or discontinuation, similar to an intent-to-treat approach in randomized clinical trials. This follow-up approach addresses concerns related to informative censoring, which occurs if patients discontinue or if physicians deprescribe the treatments under consideration because of memory problems associated with ADRD but there is no claim for this diagnosis until after the treatment is discontinued. By disregarding early events, the 6-month induction period addresses reverse causation bias concerns if treatment decisions were driven by early disease symptoms, but disease diagnosis is only recorded shortly after treatment initiation.

    Analysis 3: Incorporating a 6-Month Symptoms to Diagnosis Period

    In analysis 3, we assigned an outcome date that was 6 months before the first recorded ADRD date and excluded last 6 months of follow-up for those who were censored without an event to account for the possibility that ADRD symptoms appeared some time before a diagnosis was indicated in insurance records, which could lead to misclassification of ADRD onset, and followed patients in an as-treated follow-up scheme.

    Analysis 4: Alternate Outcome Definition

    In analysis 4, the outcome was defined using a combination of diagnosis code and at least 1 prescription claim for a symptomatic treatment (ie, donepezil, galantamine, rivastigmine, and memantine) occurring within 6 months of each other, with the outcome date assigned to the second event in the sequence, and patients were observed in an as-treated follow-up scheme. Use of medication records to identify dementia had a positive predictive value exceeding 95% in a previous validation study.15

    Covariates

    All covariates were measured in the 365-day baseline period preceding cohort entry. They included: sociodemographic factors, including age, sex, race, and receipt of low-income subsidy; risk factors for ADRD, including diabetes, stroke, and depression16-18; lifestyle factors, such as smoking, as well as use of preventive services, like screening mammography and vaccinations, to account for healthy-user confounding19; measures of health care services use before cohort entry, including number of prescriptions filled, number of emergency department visits, hospitalizations, and physician office visits, to minimize the potential of differential surveillance bias,20 and a frailty indicator21; RA-related treatments, including number of nonbiologic DMARDs, number of TDMARDs, and steroid and opioid use to account for differences in RA activity at baseline; and other comorbid conditions and comedications (eTable 2 and eTable 3 in the Supplement). Race data from Medicare claims are based on voluntary report from enrollees to the Social Security Administration. These data are collected by CMS for administrative reasons and released for research purposes.

    Statistical Analysis

    We used propensity score (PS)22 matching to account for measured confounding in this study. The PSs were calculated as the projected probability of initiating the exposure of interest vs the reference drug conditional on baseline covariates using multivariable logistic regression separately for each comparison. For all of our analyses, initiators of each exposure of interest were matched with initiators of abatacept based on their PS using a nearest-neighbor algorithm within a caliper of 0.025 on the natural scale of the PS.23,24 Multiple diagnostics for PS analysis were evaluated including PS distributional overlap before and after matching to ensure comparability of these groups25 and balance in each individual covariate between 2 treatment groups using standardized differences.26

    In the PS-matched sample, incidence rates, along with 95% CIs, for the outcome were estimated for the treatment and reference groups. We calculated cumulative incidence using cumulative incidence functions that account for competing risk by death and provided cause-specific hazard ratios (HRs) from Cox proportional hazards regression models.27 Prespecified subgroup analyses were conducted based on age, sex, and baseline cardiovascular disease, as there is evidence of potentially heterogenous pathogenesis of ADRD based on these factors.28-30 Statistical significance was determined as 95% CIs that did not cross 1. Statistical analyses were performed in the Aetion Evidence Platform version 4.30 (Aetion), including R version 3.4.2 (R Project for Statistical Computing), which has been scientifically validated by accurately repeating a range of previously published studies31 and by replicating32 or projecting clinical trial findings.33

    Results
    Cohort Characteristics

    After 1:1 PS matching to patients using abatacept, a total of 22 569 PS-matched patient pairs, including 4224 tofacitinib pairs (mean [SD] age 72.19 [5.65] years; 6945 [82.2%] women), 6369 tocilizumab pairs (mean [SD] age 72.01 [5.46] years; 10 105 [79.4%] women), and 11 976 TNF inhibitor pairs (mean [SD] age 72.67 [5.91] years; 19 710 [82.3%] women), were assessed (Table 1; eFigure 3, eTable 2 and eTable 3 in the Supplement). Diabetes and hypertension were commonly observed in all 3 study cohorts. Importantly, nearly three-fourths of patients had no use of any previous TDMARDs (ie, TDMARD naive) in the TNF inhibitor cohort; while only 34% of patients in the tofacitinib cohort and 27% of patients in the tocilizumab cohort were TDMARD naive, indicating that these 2 drugs are frequently used as second-line medications. All characteristics were well-balanced after PS matching, with standardized differences <0.1 (Table 1).

    Incidence Rates of ADRD

    Rates of incident ADRD ranged from a high of 14 to 18 per 1000 person-years in analysis 2 (as-started analysis) to a low of 2 to 4 per 1000 person-years in analysis 4, which used a highly specific but less sensitive outcome definition requiring ADRD treatment in addition to diagnosis codes (Table 2). Rates were generally similar within analysis schemes across the 3 treatment cohorts.

    Comparative Risk of ADRD

    We found no evidence of differences in the risk of ADRD associated with tofacitinib (analysis 1: HR, 0.90 [95% CI, 0.55-1.51]; analysis 2: HR, 0.78 [95% CI, 0.53-1.13]; analysis 3: HR, 1.29 [95% CI, 0.72-2.33]; analysis 4: HR, 0.50 [95% CI, 0.21-1.20]), tocilizumab (analysis 1: HR, 0.82 [95% CI, 0.55-1.21]; analysis 2: HR, 1.05 [95% CI, 0.81-1.35]; analysis 3: HR, 1.21 [95% CI, 0.75-1.96]; analysis 4: HR, 0.78 [95% CI, 0.44-1.39]), or TNF inhibitors (analysis 1: HR, 0.93 [95% CI, 0.72-1.20]; analysis 2: HR, 1.02 [95% CI, 0.86-1.20]; analysis 3: HR, 1.13 [95% CI, 0.86-1.48]; analysis 4: 0.90 [95% CI, 0.60-1.37]) compared with abatacept (Figure 1; eFigure 4 in the Supplement). Estimates across all 4 analysis schemes generally were in agreement, indicating limited association of various assumptions with study results, although 95% CIs were wide for analysis 4 owing to fewer events. There were no differences in the cumulative incidence of ADRD in any of the 3 exposure groups compared with the abatacept exposed group (Figure 2). Results for the secondary outcome of AD were also consistent (eTable 4 in the Supplement).

    Subgroup Analyses

    Overall, while precision was limited owing to small event counts for tofacitinib and tocilizumab cohorts, no differences were observed consistently across the 4 analysis schemes for any subgroups for these 2 comparisons (eFigure 5 and eFigure 6 in the Supplement). For the TNF inhibitor vs abatacept comparison, results from all subgroups were consistent with the primary analysis. However, for the subgroup of patients with baseline cardiovascular disease, point estimates indicated a lower incidence of ADRD with TNF inhibitors in 2 of 4 analyses (analysis 1: HR, 0.76 [95% CI, 0.50-1.16]; analysis 2: HR, 0.74 [95% CI, 0.56-0.99]; analysis 3: HR. 1.03 [95% CI, 0.65-1.61]; analysis 4: HR, 0.45 [95% CI, 0.21-0.98]) (eFigure 7 in the Supplement).

    Discussion

    In this population-based cohort study designed to evaluate a prespecified hypothesis generated based on a multiomics approach, we found no reduction in the risk of ADRD for users of tofacitinib, tocilizumab, or TNF inhibitors compared with an active comparator, abatacept. These results were consistent across a range of analytic approaches, providing clear evidence that various design decisions had limited impact on the overall conclusion.

    Multiple previous investigations using routine health care data have attempted to quantify the associations between TDMARDs, specifically TNF inhibitors, and reduction in ADRD risk, and have reported large effect sizes, ranging from 30% to 70% reduction compared with nonuse.34,35 These implausibly large effect sizes are likely attributable to combinations of various sources of bias identified in these studies, including immortal time bias, reverse causation bias, and severe confounding by indication.9 In this study, we attempted to address these biases using appropriate study design principles, including an active comparator and a new user design.36,37 We further explicitly recognized that evaluating ADRD incidence is challenging using health care claims, and there may not be a single ideal approach to address all issues. To accommodate various challenges, including potential for reverse causation bias caused by treatment selection after symptom onset but before diagnosis is made, misclassification of ADRD onset due to a time lag between symptoms and diagnosis, and limited specificity of diagnosis codes, we prespecified a series of analyses in which all these assumptions were varied to evaluate their impact on the study results. The careful attention to various sources of biases represents a key strength of our study compared with previous investigations and likely explains divergent results from previous studies. We observed results indicating potentially lower risk of ADRD with TNF inhibitors in patients with a history of cardiovascular disease in a subgroup analysis. While these results must be interpreted with caution owing to limitations associated with subgroup analyses, including possibility of type I error due to multiple hypothesis testing,38 further studies in diverse cohorts may clarify whether the interaction of cardiovascular risk factors and TNF signaling is associated with differentially modulating risk of ADRD.

    The biological rationale underlying the potential role of TDMARDs as disease-modifying ADRD treatments is based on their ability to lower inflammation in the brain or through attenuation of systemic inflammation by specific pathways. Tofacitinib, the JAK inhibitor tested in our study, is central nervous system (CNS)–penetrant and may therefore be expected to exert direct effects on the JAK/STAT signaling pathway in the brain.39 On the other hand, while some of the drugs we tested, such as tocilizumab and etanercept, have poor CNS penetration, their ability to lower systemic inflammation has been proposed as a plausible mechanism to attenuate brain microglial activation, which is implicated in the pathogenesis of ADRD.40,41 Lending support to this hypothesis are prior positron emission tomography studies in humans and nonhuman primates with systemic inflammation that have shown evidence of increased microglial activation.40,42 The potential effect of etanercept on attenuating microglial activation through lowering of systemic inflammation was also the basis for a previous phase 2 trial in AD demonstrating its safety and tolerability.43

    The interpretation of our results requires a nuanced discussion, as null findings observed in our investigation may be driven by several factors operating simultaneously. First, it is possible that TDMARDs targeting JAK, IL-6, or TNF, may truly have no causal impact on the risk or trajectory of ADRD. Second, it must be noted that we compared the risk of ADRD associated with initiation of inhibitors of these specific enzymes or cytokines with a common active comparator, abatacept, a T-cell costimulation blocker. Since abatacept has potent impact on lowering inflammation in RA, similar to the other study drugs, an alternate interpretation is that all these agents may lower ADRD risk to a similar extent. While an active comparator design complicates interpretation, we believe that it is impossible to conduct an unbiased investigation for this research question without using a truly equivalent comparator drug that is used for a similar indication and at a similar stage of the underlying illness (RA). Comparing TDMARDs with nonuse or even with nonbiologic DMARDs is subject to severe bias due to confounding, as the decision to initiate treatment with TDMARDs in old age is likely influenced by RA activity and frailty, which cannot be fully measured with claims data.44 Some active comparators, for instance methotrexate, which was used in another recent study as a comparator to TNF inhibitors,45 are likely more appropriate than nonuser comparisons owing to improved confounding adjustment. Indeed, the study comparing TNF inhibitors to methotrexate found no differences in the risk of dementia, unlike previous studies comparing TNF inhibitors with nonuse.34,35

    Limitations

    There are certain limitations inherent to the design of this study, which could also influence interpretation. First, despite accumulating the largest cohorts to date studying this question, the number of outcomes were small for tofacitinib and tocilizumab, partly owing to short mean follow-up duration, which could mean our study was underpowered to detect smaller magnitude differences or delayed treatment outcomes. It is important to note that pathogenesis of ADRD may begin many years before a clinical diagnosis. Given the pathophysiological characteristics and prolonged preclinical phase of ADRD, longer periods of treatment and/or observation may be needed to draw firmer conclusions about the null findings. Second, although we tried addressing limitations related to identifying ADRD in health care claims through careful design, there remains a possibility of bias owing to outcome misclassification. Third, while active comparator designs using an equivalent reference exposure are less prone to confounding, these designs are not free from confounding by indication. Despite all these limitations, our results likely indicate that the large signals observed in previous studies34,35 may represent design artifacts rather than true associations. The current evidence base, which is substantially enhanced by our study, does not support testing TDMARDs in prospective trials despite promising biological data.46,47

    Conclusions

    This cohort study found no differences in the risk of ADRD in patients treated with tofacitinib, tocilizumab, or TNF inhibitors compared with those treated with abatacept. Careful design choices and nuanced interpretation of results are important for generating valid, actionable evidence on drug repurposing questions explored using routine health care data.

    Back to top
    Article Information

    Accepted for Publication: February 21, 2022.

    Published: April 8, 2022. doi:10.1001/jamanetworkopen.2022.6567

    Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Desai RJ et al. JAMA Network Open.

    Corresponding Author: Rishi J. Desai, MS, PhD, Division of Pharmacoepidemiology and Pharmacoeconomics, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, 1620 Tremont St, Ste 3030-R, Boston, MA 02120 (rdesai@bwh.harvard.edu).

    Author Contributions: Dr Desai had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Desai, Varma, Gerhard, Segal, Mahesri, Schneeweiss, Thambisetty.

    Acquisition, analysis, or interpretation of data: Desai, Varma, Gerhard, Mahesri, Chin, Horton, Kim, Schneeweiss.

    Drafting of the manuscript: Desai, Mahesri, Chin, Kim, Thambisetty.

    Critical revision of the manuscript for important intellectual content: Desai, Varma, Gerhard, Segal, Mahesri, Horton, Schneeweiss.

    Statistical analysis: Desai, Mahesri, Chin, Kim.

    Obtained funding: Desai, Schneeweiss, Thambisetty.

    Administrative, technical, or material support: Varma, Mahesri, Thambisetty.

    Supervision: Schneeweiss.

    Conflict of Interest Disclosures: Dr Desai reported receiving grants from Bayer, Novartis, and Vertex outside the submitted work. Dr Gerhard reported receiving grants from Bristol-Myers Squibb and personal fees from Bristol-Myers Squibb, Merck, Pfizer, Lilly, IntraCellular Therapies, Eisai, and Alkermes outside the submitted work. Dr Segal reported receiving grants from the National Institutes of Health, Patient-Centered Outcomes Research Institute, Altarum Institute, Arnold Foundation, US Food and Drug Administration, and Bristol-Myers Squibb and personal fees from American College of Physicians and Sanofi outside the submitted work. Dr Horton reported receiving salary support from the Childhood Arthritis and Rheumatology Research Alliance, National Institute of Arthritis and Musculoskeletal and Skin Diseases, and National Center for Advancing Translational Sciences and personal fees from the American College of Rheumatology outside the submitted work. Dr Kim reported receiving grants from Pfizer, AbbVie, Roche, and Bristol-Myers Squibb outside the submitted work. Dr Schneeweiss reported receiving grants from Boehringer Ingelheim and receiving personal fees from and owning stock in Aetion and outside the submitted work. No other disclosures were reported.

    Funding/Support: The Drug Repurposing for Effective Alzheimer Medicines (DREAM) study is funded by the intramural program of the National Institute on Aging (NIA).

    Role of the Funder/Sponsor: This study is a collaborative effort between academic partners and NIA intramural program. Authors from the NIA had substantial roles in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, and approval of the manuscript; and decision to submit the manuscript for publication.

    References
    1.
    Cummings  JL, Morstorf  T, Zhong  K.  Alzheimer’s disease drug-development pipeline: few candidates, frequent failures.   Alzheimers Res Ther. 2014;6(4):37. doi:10.1186/alzrt269 PubMedGoogle ScholarCrossref
    2.
    Yiannopoulou  KG, Anastasiou  AI, Zachariou  V, Pelidou  SH.  Reasons for failed trials of disease-modifying treatments for Alzheimer disease and their contribution in recent research.   Biomedicines. 2019;7(4):E97. doi:10.3390/biomedicines7040097 PubMedGoogle Scholar
    3.
    Hebert  LE, Weuve  J, Scherr  PA, Evans  DA.  Alzheimer disease in the United States (2010-2050) estimated using the 2010 census.   Neurology. 2013;80(19):1778-1783. doi:10.1212/WNL.0b013e31828726f5 PubMedGoogle ScholarCrossref
    4.
    Godyń  J, Jończyk  J, Panek  D, Malawska  B.  Therapeutic strategies for Alzheimer’s disease in clinical trials.   Pharmacol Rep. 2016;68(1):127-138. doi:10.1016/j.pharep.2015.07.006 PubMedGoogle ScholarCrossref
    5.
    Iqbal  K, Liu  F, Gong  CX.  Alzheimer disease therapeutics: focus on the disease and not just plaques and tangles.   Biochem Pharmacol. 2014;88(4):631-639. doi:10.1016/j.bcp.2014.01.002 PubMedGoogle ScholarCrossref
    6.
    Panza  F, Logroscino  G, Imbimbo  BP, Solfrizzi  V.  Is there still any hope for amyloid-based immunotherapy for Alzheimer’s disease?   Curr Opin Psychiatry. 2014;27(2):128-137. doi:10.1097/YCO.0000000000000041 PubMedGoogle ScholarCrossref
    7.
    Bennett  DA, Schneider  JA, Arvanitakis  Z,  et al.  Neuropathology of older persons without cognitive impairment from two community-based studies.   Neurology. 2006;66(12):1837-1844. doi:10.1212/01.wnl.0000219668.47116.e6 PubMedGoogle ScholarCrossref
    8.
    Castellani  RJ, Perry  G.  The complexities of the pathology-pathogenesis relationship in Alzheimer disease.   Biochem Pharmacol. 2014;88(4):671-676. doi:10.1016/j.bcp.2014.01.009 PubMedGoogle ScholarCrossref
    9.
    Desai  RJ, Varma  VR, Gerhard  T,  et al.  Targeting abnormal metabolism in Alzheimer’s disease: the Drug Repurposing for Effective Alzheimer’s Medicines (DREAM) study.   Alzheimers Dement (N Y). 2020;6(1):e12095. doi:10.1002/trc2.12095 PubMedGoogle Scholar
    10.
    Pushpakom  S, Iorio  F, Eyers  PA,  et al.  Drug repurposing: progress, challenges and recommendations.   Nat Rev Drug Discov. 2019;18(1):41-58. doi:10.1038/nrd.2018.168 PubMedGoogle ScholarCrossref
    11.
    Thambisetty  M.  Understanding mechanisms and seeking cures for Alzheimer’s disease: why we must be “extraordinarily diverse.”   Am J Physiol Cell Physiol. 2017;313(4):C353-C361. doi:10.1152/ajpcell.00111.2017 PubMedGoogle ScholarCrossref
    12.
    Berger  ML, Sox  H, Willke  RJ,  et al.  Good practices for real-world data studies of treatment and/or comparative effectiveness: recommendations from the joint ISPOR-ISPE Special Task Force on real-world evidence in health care decision making.   Value Health. 2017;20(8):1003-1008. doi:10.1016/j.jval.2017.08.3019 PubMedGoogle ScholarCrossref
    13.
    Smith  GD, Ebrahim  S.  Data dredging, bias, or confounding.   BMJ. 2002;325(7378):1437-1438. doi:10.1136/bmj.325.7378.1437PubMedGoogle ScholarCrossref
    14.
    Taylor  DH  Jr, Østbye  T, Langa  KM, Weir  D, Plassman  BL.  The accuracy of Medicare claims as an epidemiological tool: the case of dementia revisited.   J Alzheimers Dis. 2009;17(4):807-815. doi:10.3233/JAD-2009-1099 PubMedGoogle ScholarCrossref
    15.
    Solomon  A, Ngandu  T, Soininen  H, Hallikainen  MM, Kivipelto  M, Laatikainen  T.  Validity of dementia and Alzheimer’s disease diagnoses in Finnish national registers.   Alzheimers Dement. 2014;10(3):303-309. doi:10.1016/j.jalz.2013.03.004 PubMedGoogle ScholarCrossref
    16.
    Kivipelto  M, Ngandu  T, Laatikainen  T, Winblad  B, Soininen  H, Tuomilehto  J.  Risk score for the prediction of dementia risk in 20 years among middle aged people: a longitudinal, population-based study.   Lancet Neurol. 2006;5(9):735-741. doi:10.1016/S1474-4422(06)70537-3 PubMedGoogle ScholarCrossref
    17.
    Barnes  DE, Beiser  AS, Lee  A,  et al.  Development and validation of a brief dementia screening indicator for primary care.   Alzheimers Dement. 2014;10(6):656-665.e1. doi:10.1016/j.jalz.2013.11.006PubMedGoogle ScholarCrossref
    18.
    Albrecht  JS, Hanna  M, Kim  D, Perfetto  EM.  Predicting diagnosis of Alzheimer’s disease and related dementias using administrative claims.   J Manag Care Spec Pharm. 2018;24(11):1138-1145. doi:10.18553/jmcp.2018.24.11.1138 PubMedGoogle Scholar
    19.
    Brookhart  MA, Patrick  AR, Dormuth  C,  et al.  Adherence to lipid-lowering therapy and the use of preventive health services: an investigation of the healthy user effect.   Am J Epidemiol. 2007;166(3):348-354. doi:10.1093/aje/kwm070 PubMedGoogle ScholarCrossref
    20.
    Schneeweiss  S, Seeger  JD, Maclure  M, Wang  PS, Avorn  J, Glynn  RJ.  Performance of comorbidity scores to control for confounding in epidemiologic studies using claims data.   Am J Epidemiol. 2001;154(9):854-864. doi:10.1093/aje/154.9.854 PubMedGoogle ScholarCrossref
    21.
    Kim  DH, Schneeweiss  S, Glynn  RJ, Lipsitz  LA, Rockwood  K, Avorn  J.  Measuring frailty in Medicare data: development and validation of a claims-based frailty index.   J Gerontol A Biol Sci Med Sci. 2018;73(7):980-987. doi:10.1093/gerona/glx229PubMedGoogle ScholarCrossref
    22.
    Rosenbaum  PR, Rubin  DB.  The central role of the propensity score in observational studies for causal effects.   Biometrika. 1983;70(1):41-55. doi:10.1093/biomet/70.1.41 Google ScholarCrossref
    23.
    Rassen  JA, Shelat  AA, Myers  J, Glynn  RJ, Rothman  KJ, Schneeweiss  S.  One-to-many propensity score matching in cohort studies.   Pharmacoepidemiol Drug Saf. 2012;21(suppl 2):69-80. doi:10.1002/pds.3263 PubMedGoogle ScholarCrossref
    24.
    Austin  PC.  Some methods of propensity-score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations.   Biom J. 2009;51(1):171-184. doi:10.1002/bimj.200810488 PubMedGoogle ScholarCrossref
    25.
    Walker  AM, Patrick  AR, Lauer  MS,  et al.  Tool for assessing the feasibility of comparative effectiveness research.   Comp Effect Res. 2013;3:11-20. doi:10.2147/CER.S40357Google Scholar
    26.
    Franklin  JM, Rassen  JA, Ackermann  D, Bartels  DB, Schneeweiss  S.  Metrics for covariate balance in cohort studies of causal effects.   Stat Med. 2014;33(10):1685-1699. doi:10.1002/sim.6058 PubMedGoogle ScholarCrossref
    27.
    Austin  PC, Lee  DS, Fine  JP.  Introduction to the analysis of survival data in the presence of competing risks.   Circulation. 2016;133(6):601-609. doi:10.1161/CIRCULATIONAHA.115.017719 PubMedGoogle ScholarCrossref
    28.
    Andrew  MK, Tierney  MC.  The puzzle of sex, gender and Alzheimer’s disease: why are women more often affected than men?   Womens Health (Lond). 2018;14:1745506518817995. doi:10.1177/1745506518817995 Google Scholar
    29.
    Tublin  JM, Adelstein  JM, Del Monte  F, Combs  CK, Wold  LE.  Getting to the heart of Alzheimer disease.   Circ Res. 2019;124(1):142-149. doi:10.1161/CIRCRESAHA.118.313563 PubMedGoogle ScholarCrossref
    30.
    Fayosse  A, Nguyen  D-P, Dugravot  A,  et al.  Risk prediction models for dementia: role of age and cardiometabolic risk factors.   BMC Med. 2020;18(1):107. doi:10.1186/s12916-020-01578-x PubMedGoogle ScholarCrossref
    31.
    Wang  SV, Verpillat  P, Rassen  JA, Patrick  A, Garry  EM, Bartels  DB.  Transparency and reproducibility of observational cohort studies using large healthcare databases.   Clin Pharmacol Ther. 2016;99(3):325-332. doi:10.1002/cpt.329 PubMedGoogle ScholarCrossref
    32.
    Fralick  M, Kesselheim  AS, Avorn  J, Schneeweiss  S.  Use of health care databases to support supplemental indications of approved medications.   JAMA Intern Med. 2018;178(1):55-63. doi:10.1001/jamainternmed.2017.3919 PubMedGoogle ScholarCrossref
    33.
    Patorno  E, Schneeweiss  S, Gopalakrishnan  C, Martin  D, Franklin  JM.  Using Real-world data to predict findings of an ongoing phase IV cardiovascular outcome trial: cardiovascular safety of linagliptin versus glimepiride.   Diabetes Care. 2019;42(12):2204-2210. doi:10.2337/dc19-0069 PubMedGoogle ScholarCrossref
    34.
    Chou  RC, Kane  M, Ghimire  S, Gautam  S, Gui  J.  Treatment for rheumatoid arthritis and risk of Alzheimer’s disease: a nested case-control analysis.   CNS Drugs. 2016;30(11):1111-1120. doi:10.1007/s40263-016-0374-z PubMedGoogle ScholarCrossref
    35.
    Zhou  M, Xu  R, Kaelber  DC, Gurney  ME.  Tumor necrosis factor (TNF) blocking agents are associated with lower risk for Alzheimer’s disease in patients with rheumatoid arthritis and psoriasis.   PLoS One. 2020;15(3):e0229819. doi:10.1371/journal.pone.0229819 PubMedGoogle Scholar
    36.
    Ray  WA.  Evaluating medication effects outside of clinical trials: new-user designs.   Am J Epidemiol. 2003;158(9):915-920. doi:10.1093/aje/kwg231 PubMedGoogle ScholarCrossref
    37.
    Schneeweiss  S.  A basic study design for expedited safety signal evaluation based on electronic healthcare data.   Pharmacoepidemiol Drug Saf. 2010;19(8):858-868. doi:10.1002/pds.1926 PubMedGoogle ScholarCrossref
    38.
    Wallach  JD, Sullivan  PG, Trepanowski  JF, Steyerberg  EW, Ioannidis  JP.  Sex based subgroup differences in randomized controlled trials: empirical evidence from Cochrane meta-analyses.   BMJ. 2016;355:i5826. doi:10.1136/bmj.i5826PubMedGoogle Scholar
    39.
    Fukuyama  T, Tschernig  T, Qi  Y, Volmer  DA, Bäumer  W.  Aggression behaviour induced by oral administration of the Janus-kinase inhibitor tofacitinib, but not oclacitinib, under stressful conditions.   Eur J Pharmacol. 2015;764:278-282. doi:10.1016/j.ejphar.2015.06.060 PubMedGoogle ScholarCrossref
    40.
    Dale  RC.  Interleukin-6 blockade as rescue therapy in autoimmune encephalitis.   Neurotherapeutics. 2016;13(4):821-823. doi:10.1007/s13311-016-0471-1 PubMedGoogle ScholarCrossref
    41.
    Perry  VH, Holmes  C.  Microglial priming in neurodegenerative disease.   Nat Rev Neurol. 2014;10(4):217-224. doi:10.1038/nrneurol.2014.38 PubMedGoogle ScholarCrossref
    42.
    Hannestad  J, Gallezot  JD, Schafbauer  T,  et al.  Endotoxin-induced systemic inflammation activates microglia: [11C]PBR28 positron emission tomography in nonhuman primates.   Neuroimage. 2012;63(1):232-239. doi:10.1016/j.neuroimage.2012.06.055 PubMedGoogle ScholarCrossref
    43.
    Butchart  J, Brook  L, Hopkins  V,  et al.  Etanercept in Alzheimer disease: a randomized, placebo-controlled, double-blind, phase 2 trial.   Neurology. 2015;84(21):2161-2168. doi:10.1212/WNL.0000000000001617 PubMedGoogle ScholarCrossref
    44.
    Yoshida  K, Solomon  DH, Kim  SC.  Active-comparator design and new-user design in observational studies.   Nat Rev Rheumatol. 2015;11(7):437-441. doi:10.1038/nrrheum.2015.30 PubMedGoogle ScholarCrossref
    45.
    Kern  DM, Lovestone  S, Cepeda  MS.  Treatment with TNF-α inhibitors versus methotrexate and the association with dementia and Alzheimer’s disease.   Alzheimers Dement (N Y). 2021;7(1):e12163. doi:10.1002/trc2.12163 PubMedGoogle Scholar
    46.
    Tweedie  D, Sambamurti  K, Greig  NH.  TNF-alpha inhibition as a treatment strategy for neurodegenerative disorders: new drug candidates and targets.   Curr Alzheimer Res. 2007;4(4):378-385. doi:10.2174/156720507781788873 PubMedGoogle ScholarCrossref
    47.
    Torres-Acosta  N, O’Keefe  JH, O’Keefe  EL, Isaacson  R, Small  G.  Therapeutic potential of TNF-α inhibition for Alzheimer’s disease prevention.   J Alzheimers Dis. 2020;78(2):619-626. doi:10.3233/JAD-200711 PubMedGoogle ScholarCrossref
    ×