Clinically meaningful breakthroughs awaited

Three simultaneously published reviews of Alzheimer’s disease (AD) in Annals of Internal Medicine outlined progress in research-level diagnostic procedures and conceptual understanding but showed little meaningful change in routine clinical diagnosis or treatment over the last decade.

Separate literature reviews on brief cognitive tests, biomarkers, and treatment options were conducted by Howard Fink, MD, MPH, of the Minneapolis VA Health Care System, and coauthors, with literature searches through Nov. 2019, including studies that exhibited low to medium risk of bias.

In an accompanying editorial, Raj Shah, MD, and David Bennett, MD, both of Rush University in Chicago, noted that “overall, better instruments to detect MCI and mild dementia are needed in the primary care toolkit, biomarkers need to be clinically actionable with clear patient benefits before advocating widespread clinical use, and better therapeutics that target the full range of clinical manifestations of Alzheimer dementia are urgently needed.”

Brief Cognitive Tests

Fink and colleagues focused their review of cognitive testing on the ability to distinguish clinical Alzheimer’s-type dementia from mild cognitive impairment (MCI) or normal cognition and included 57 studies. They found single cognitive tests were sensitive and specific for distinguishing clinical Alzheimer’s-type dementia from normal cognition. These included:

  • Clock-drawing (median sensitivity 0.79 and specificity 0.88).
  • Mini-Mental State Examination (median sensitivity and specificity 0.88 and 0.94).
  • Montreal Cognitive Assessment (median sensitivity and specificity 0.94 and 0.94).
  • Brief Alzheimer Screen (median sensitivity and specificity 0.92 and 0.97).
  • List delayed recall (median sensitivity and specificity 0.89 and 0.94).
  • Category fluency (median sensitivity and specificity 0.92 and 0.89).

Many brief, single cognitive tests accurately distinguish clinical Alzheimer’s-type dementia from normal cognition but are less accurate distinguishing mild impairment from normal cognition or Alzheimer’s from MCI, the researchers said. The editorialists echoed this point: “Brief cognitive tests distinguish frank dementia from normal cognition but provide limited discrimination in the space among normal cognition, MCI, and mild dementia — the space where simple tools to aid the clinical history are most needed,” Shah and Bennett wrote.


Review of biomarkers included brain imaging and cerebrospinal fluid (CSF) studies.

The imaging review of 15 studies included:

  • Amyloid positron emission tomography (PET): median sensitivity and specificity 0.91 and 0.92.
  • Fluorodeoxyglucose (FDG)–PET: median sensitivity and specificity 0.89 and 0.74.
  • Magnetic resonance imaging (MRI): median sensitivity and specificity (medial temporal lobe atrophy) 0.91 and 0.89.
  • Single-photon emission computed tomography: median sensitivity and specificity 0.64 and 0.83.

A review of 9 studies of CSF markers showed that individual CSF biomarkers and ratios had moderate sensitivity (range 0.62 to 0.83) and specificity (range 0.53 to 0.69). In direct comparisons, beta-amyloid 42/phosphorylated tau (p-tau) ratio, total tau (t-tau)/beta-amyloid 42 ratio, and p-tau seemed more accurate than beta-amyloid 42 or t-tau alone.

A biological definition of Alzheimer’s has been proposed that includes pathology and biomarkers, the editorialists noted. “Biomarkers in PET and CSF seem up to the task,” they wrote. “However, these biomarkers are not actionable therapeutically, despite evidence suggesting that they lead to changes in clinical practice,” since there are no approved medications that depend on biomarker results, they added. “Although these biomarkers are highly useful in clinical research and may be useful in limited clinical settings, they have not been shown to have a major effect on our patients’ health outcomes.”


The authors considered treatment of both cognitive and functional issues (55 studies) and psychiatric symptoms (12 studies) in their review. “Evidence was mostly insufficient on drug treatment of behavioral and psychological symptoms of dementia and on supplements for all outcomes,” they noted.

Across Alzheimer’s severity, mostly low-strength evidence suggested that, compared with placebo, cholinesterase inhibitors produced:

  • Small average improvements in cognition (median standardized mean difference, 0.30).
  • No difference to small improvement in function.
  • No difference in the likelihood of moderate or better improvement in global clinical impression.

Add-on memantine in moderate-to-severe AD showed low to insufficient strength evidence that it inconsistently improved cognition and global clinical impression, but not function. “Similar to results of a 2008 systematic review, cholinesterase inhibitors and memantine showed slightly reduced short term worsening of cognition and function, of uncertain clinical significance,” Shah and Bennett observed.

A limitation of the reviews was the relatively small numbers of included studies in all three domains, with many excluded due to the requirement for low- to medium-risk of bias. “It is concerning how many of the peer reviewed articles published since 2008 could not be included in the systematic reviews due to high risk of bias,” Shah and Bennett noted.

Other limitations included study heterogeneity and unclear applicability to primary care settings, the editorialists added. “From a human-centered design perspective, a key issue at this time is how to improve the physician–patient relationship for persons living with Alzheimer dementia without the overemphasis on diagnostic tools and biomarkers, while recognizing the limitations of current treatment options,” they wrote.

  1. Brief cognitive tests distinguish frank dementia from normal cognition but provide limited discrimination in the space involving normal cognition, mild cognitive impairment, and mild dementia, where simple tools are most needed. Biomarkers can biologically define Alzheimer’s disease, but their clinical utility may be limited.

  2. Similar to results of a 2008 systematic review, cholinesterase inhibitors, and memantine showed slightly reduced short-term worsening of cognition, of uncertain clinical significance, while data on psychiatric treatments and supplements were mostly insufficient.

Paul Smyth, MD, Contributing Writer, BreakingMED™

Reviews were based on research conducted by the Minnesota Evidence-based Practice Center under contract to the AHRQ.

Fink had no disclosures.

The editorial was supported by the National Institutes of Health, the Health Resources and Services Administration, and the Illinois Department of Public Health.

Bennett reported grants from Biogen, grants from Neurovision, personal fees from AbbVie, personal fees from Takeda, and personal fees from Origent Data Sciences.

Cat ID: 33

Topic ID: 82,33,282,485,730,33,192,255,362,925