399 - The evolution of Alzheimer's disease and dementia care | Gayatri Devi, M.D.

KEYPOINTS

• 01

  Dementia is a spectrum, and most Alzheimer's brains have mixed pathology.

  Devi, a memory-disorders neurologist practicing on Manhattan's Upper East Side for 26 years, says autopsy studies find 98–99% of Alzheimer's patients have some other concomitant brain pathology. About 40% of Alzheimer's patients eventually show Lewy body pathology and 30–40% of Lewy body patients have Alzheimer's pathology. Vascular disease is so intertwined that a 2002 Circulation paper asked, 'Is Alzheimer's stroke?'—every Alzheimer's risk factor is also a stroke risk factor.

• 02

  Prevention may need to start in the 30s and 40s.

  Because the cascade begins decades before symptoms, she argues for building anti-inflammatory habits early to keep it from being triggered. Neuroinflammation and viral infections are emerging factors; she sees stronger data for shingles vaccination, relevant since virtually everyone over 50 has had varicella.

• 03

  Standard screening tests miss high-functioning patients.

  Her patients often score 28–30 on the MMSE or MoCA 10–15 years into their condition because they compensate so well. She uses detailed neuropsychological testing: someone at the 99th percentile overall but the 10th in language or visuospatial skills has likely declined substantially.

• 04

  Forgetting names is usually normal; word-finding trouble is more concerning.

  Peter describes taking a minute to recall the name of someone he's met 30 times. Devi notes we now meet far more people than 100 years ago—some speculate we're built to remember about 150 names. Difficulty finding ordinary words is a more meaningful signal.

• 05

  Women's higher risk is tied partly to longevity and menopause.

  Life expectancy has risen from about 50 to the late 70s or early 80s; 150 years ago most women died before menopause, when estrogen production stops. Perimenopausal cognitive changes can start in the 40s unnoticed. She calls women denied HRT after the 2002 WHI a 'lost generation' now in their 70s. She prefers transdermal estrogen (gel or patch) for lower stroke and clot risk.

• 06

  Amyloid alone doesn't mean Alzheimer's; biomarkers need context.

  About 25% of community-dwelling people in their 70s, 30%+ in their 80s and about 44% at 90 have amyloid without symptoms. Blood tests are mostly standardized against amyloid scans, and she thinks they should be standardized against both amyloid and tau. About 30% of participants in Alzheimer's drug trials didn't actually have Alzheimer's pathologically.

• 07

  APOE4 raises risk but isn't deterministic.

  Devi estimates lifetime risk around 6% for a man and 10–12% for a woman with 3/3; one E4 raises it about 60% versus wild type, and 4/4 much more. PSEN1, PSEN2 and APP mutations are close to deterministic, with PSEN1 essentially 100% associated with early-onset disease.

• 08

  Slow titration of anti-amyloid antibodies may cut brain swelling and bleeding.

  Aducanumab had a greater than 40% rate of ARIA, so Devi developed a very slow titration protocol. In published data on 4/4 patients she reports about 4% ARIA. The FDA has since approved a slower titration for donanemab (350, 700, 1,050, 1,400 mg), letting her hold patients at 350 mg for four or five infusions; lecanemab starts at 10 mg/kg and insurers won't reimburse slower starts, though some 4/4 patients have had severe ARIA on 3 mg/kg.

• 09

  ARIA in E4 carriers reflects blood-vessel amyloid.

  E4 homozygotes have more cerebral amyloid angiopathy; removing amyloid from vessel walls creates 'little tears in the fabric of the blood vessels'. She monitors rigorously with MRI and treats refractory ARIA-E with steroids, as for inflammatory amyloid angiopathy.

• 10

  She favors multimodal treatment.

  Off-label MRI-guided TMS since 2008 to maintain network connectivity, management of vascular risk, shunts for normal-pressure hydrocephalus, and for patients on blood thinners for atrial fibrillation, a Watchman device so they can become eligible for antibodies.

• 11

  The benefit of anti-amyloid drugs is modest on average, so patient selection matters.

  Devi acknowledges the change on scales like the CDR sum of boxes is small, and early donanemab data showed little difference in some measures. Her approach targets high-risk people in their 50s or 60s—perhaps 5–10 years from symptoms—treated 'very low, very slow, very safe', asking whether the disease course bends. She'd still choose aducanumab despite its discontinuation and reports two exceptional responders, one stable for 17 years.