Early Alzheimer’s and cognitive decline: the root-cause workup

Start here: what this is and is not

If you or someone close to you has noticed memory slipping, has been told it is mild cognitive impairment, or has an early Alzheimer's diagnosis and has been sent home with little more than a follow-up date, this guide is for you. It is also for people with a strong family history, or a known ApoE4 result, who want to act before there are symptoms.

I am a hospital doctor, a certified functional medicine practitioner, and trained in Dr Dale Bredesen's protocol for cognitive decline. It is the centre of my practice. I am going to describe what the workup tests and why, walk through the subtypes Bredesen describes, be specific about cost and time, show what a realistic trajectory looks like, and be honest about the evidence.

One thing first, because the field attracts overstatement and this clinic has been criticised in the past for language it no longer uses. I do not promise to reverse Alzheimer's disease. The aim of the programme is to halt further progression. Some people improve, most often those who start early and young. Some hold steady. Some continue to decline. I will explain what determines which, and I will not tell you a firmer story than the evidence supports.

Why a root-cause workup at all

The standard NHS pathway for suspected dementia is set out in NICE guideline NG97: a GP assessment with basic bloods, referral to a memory clinic, cognitive testing, usually a brain scan, and a diagnosis. It is a good pathway for what it is designed to do, which is to establish whether dementia is present and of what type. It is not designed to ask why it is happening in this person, or what could be changed.

That is not a criticism of memory clinics. It reflects the fact that until recently there was very little to offer once the diagnosis was made. Two things have changed that.

The first is the accumulation of evidence that a large share of dementia risk is modifiable. The 2024 Lancet Commission on dementia identified fourteen risk factors, including hearing loss, high LDL cholesterol, depression, physical inactivity, diabetes, hypertension, obesity, smoking, excess alcohol, social isolation, air pollution, traumatic brain injury, untreated vision loss and low education, and estimated that addressing them could prevent or delay around forty-five per cent of cases. The FINGER trial in Finland showed that a two-year multidomain programme of diet, exercise, cognitive training and vascular risk management improved or maintained cognitive function compared with standard advice, in a randomised trial of over twelve hundred people.

The second is Bredesen's argument, set out in a series of papers from 2014 onwards, that Alzheimer's disease is not one process with one cause but the end result of several, which differ between people, and that identifying and addressing the specific contributors in an individual is the rational response. The evidence for that argument is discussed honestly further down. But the premise, that decline is usually multifactorial and the factors can be found, is now close to mainstream.

What the Bredesen workup actually tests

The protocol calls the initial evaluation a 'cognoscopy', by analogy with a colonoscopy: a structured look for the contributors that can be found early. In practice it is a long consultation, a wide set of investigations, and a results session in which the findings are grouped into the patterns below. This is what I test, and why.

Cognitive baseline

• A validated cognitive screen, usually the Montreal Cognitive Assessment (MoCA), repeated at intervals so that change is measured rather than guessed at. The MoCA is a screening tool with real limits, and where a fuller picture is needed I arrange formal neuropsychological testing.
• Previous memory clinic reports, imaging and correspondence, which I ask to see before we meet.

Metabolic

• Fasting glucose, fasting insulin, HbA1c and a calculated measure of insulin resistance. Insulin resistance is so closely tied to Alzheimer's that some researchers have called the disease type 3 diabetes. It is the single most common finding in the people I see, and it is treatable.
• A full lipid profile with ApoB and lipoprotein(a). High LDL in mid-life is one of the Lancet Commission's fourteen risk factors.
• Liver and kidney function, uric acid, and body composition.

Inflammatory and immune

• High-sensitivity CRP, ESR, and homocysteine. Homocysteine is worth singling out: raised levels are associated with brain atrophy, they are usually driven by B vitamin status, and the VITACOG trial showed that B vitamin treatment slowed atrophy in people with mild cognitive impairment and raised homocysteine.
• Markers of gut-derived inflammation and intestinal permeability where the history points there, and a comprehensive stool analysis where digestive symptoms are present.
• Autoimmune screening where indicated.

Nutritional

• Vitamin B12 with methylmalonic acid, folate, vitamin D, the omega-3 index, red cell magnesium, zinc and copper and their ratio, selenium, and vitamin E. Several of these have direct trial or observational evidence for cognition; all of them are cheap to correct.

Hormonal

• A full thyroid panel: TSH, free T4, free T3 and antibodies. Subclinical hypothyroidism is a recognised, reversible contributor to cognitive symptoms.
• Oestradiol, progesterone and testosterone, interpreted in context. In women the timing of menopause and any hormone therapy is directly relevant, which I have written about in the perimenopause guide.
• Morning cortisol, DHEA-S and pregnenolone.

Toxic

• Blood heavy metals (lead, mercury, arsenic, cadmium) where there is any exposure history.
• Urinary mycotoxins only where the history includes water-damaged housing or workplaces and the pattern fits. This is one of the more contested parts of the protocol, and I treat the result as a prompt to investigate the environment rather than as a diagnosis.
• A careful history of alcohol, anaesthetics, medication with anticholinergic burden, and air quality.

Vascular

• Blood pressure, including home readings; ApoB and lipoprotein(a) as above; and review of any MRI for small vessel disease. Where cardiovascular risk is high I involve the GP or a cardiologist.

Infectious

• Herpes simplex and other herpes virus serology; tick-borne serology where there is exposure; and a dental history, because periodontal pathogens have a plausible and increasingly studied link to Alzheimer's pathology. I do not run speculative infection panels.

Genetic

• ApoE genotype. Roughly one in four people carry one copy of the ApoE4 variant, which raises lifetime risk; two copies is rarer and raises it substantially. Knowing your status is genuinely useful for planning, and I discuss the implications before testing rather than afterwards.
• Where relevant, variants affecting B vitamin metabolism and detoxification, which inform treatment but do not by themselves cause anything.

Sleep and the senses

• Screening for obstructive sleep apnoea, which is common, damaging to the brain, and treatable. A sleep study where the screen is positive.
• Hearing and vision. Both are on the Lancet Commission's list, both are cheap to address, and both are routinely overlooked.

Imaging and newer blood biomarkers

• An MRI with volumetric analysis of the hippocampus where one has not been done, usually via the memory clinic.
• Plasma phosphorylated tau (p-tau217) and related blood tests are now available privately in the UK and are changing how early Alzheimer's pathology can be detected. They are useful, they are not yet a substitute for a specialist diagnosis, and I will explain what a result would and would not tell you before ordering one.

The six subtypes

Bredesen groups the findings into subtypes, each describing a dominant driver. In practice most people show more than one, and the subtypes are a way of organising treatment rather than fixed categories. These are his terms, with what each means clinically.

The reason the subtypes matter is that the treatment for one can be irrelevant, or wrong, for another. Aggressive carbohydrate restriction helps a glycotoxic pattern and can worsen an atrophic one in a thin, under-nourished older person. That is the practical argument for testing before treating, and it is why I resist giving generic advice before the results are in.

What the workup costs and how long it takes

I would rather you had the numbers now than discovered them later.

Those fees include review of your records before we meet, a written summary and plan afterwards, and email support for a week after the initial consultation. Full details are on the fees page.

Testing is charged separately and is the largest additional cost. The range is wide because it depends on what the history points to: a person with a clear metabolic picture and recent NHS bloods needs far less than someone with a possible toxic exposure. I give a written estimate before anything is ordered, sequence the most informative tests first, and use NHS results wherever they exist rather than repeating them.

Time. The first cycle, from initial consultation to a settled programme, usually takes two to three months. The programme itself is then reviewed at intervals over six to twelve months, and the first honest reading of whether it is working comes at around six months, when cognitive testing is repeated. It is a long-term way of living rather than a course of treatment.

Commitment. This is a demanding programme: diet, exercise, sleep, cognitive activity, supplementation and follow-up have to hold together over months, usually with a partner or family member involved. Adherence is one of the largest determinants of outcome, and it is the part I cannot do for you.

What the evidence shows, and what it does not

Being straight about this is the most useful thing I can do on this page.

Bredesen's first paper, in 2014, described ten patients, nine of whom improved on a personalised programme. A 2016 paper described a further ten, with documented improvement in cognitive testing and in some cases in hippocampal volume. A 2018 paper described a hundred patients treated by several clinicians, with improvement reported across the group. In 2022, Toups and colleagues published a prospective trial of twenty-five people with mild cognitive impairment or early dementia treated for nine months with a precision-medicine protocol; eighty-four per cent improved on the MoCA, with improvements in secondary cognitive measures as well.

Those are encouraging results. They are also, all of them, case series or small open-label studies without a control group. Participants knew what they were receiving, the assessors knew, and people who enrol in a demanding programme are not a random sample. That design cannot rule out placebo effect, regression to the mean, or selection. A randomised controlled trial is the standard needed to settle the question, and none of sufficient size has yet reported.

Set against that, the components of the programme rest on stronger ground individually. Multidomain lifestyle intervention has randomised trial support from FINGER. B vitamins for raised homocysteine have randomised trial support from VITACOG. Treating sleep apnoea, hypertension, hearing loss, diabetes and high cholesterol are all established. So the honest position is that the programme combines interventions of varying evidence quality into a package whose overall effect has been reported in uncontrolled studies to be positive, and has not yet been tested in a controlled trial. I think that is enough to act on, particularly early, when there is the most to preserve and the least to lose. You should decide that for yourself, knowing what the evidence is.

For completeness: the anti-amyloid antibody drugs lecanemab and donanemab are licensed in the UK but have not been recommended by NICE for NHS use, on the grounds that the modest slowing of decline they produce does not justify the cost and the monitoring burden. They are not an alternative to the approach above; where someone is eligible and wants to pursue them privately, the two can run alongside each other.

A realistic patient trajectory

What follows is an illustrative composite, built from the pattern I see rather than from any single patient. I have made it unremarkable on purpose. The dramatic cases exist, but they are not what you should plan around.

1. Month 0A man in his early sixties, a retired engineer, whose wife has noticed him repeating questions and losing the thread of conversations over about eighteen months. Memory clinic diagnosis of amnestic mild cognitive impairment; MRI shows mild hippocampal volume loss; MoCA 23 out of 30. Told to come back in a year. His mother had Alzheimer’s. A free discovery call establishes he is at the stage where this approach has most to offer.
2. Month 1Three-hour initial assessment. Full history: he is a good sleeper by his own account but snores heavily; gained weight since retiring; eats well but late; walks the dog and little else; hearing has deteriorated and he has not addressed it. Testing ordered with a written estimate. Immediate changes to sleep timing, evening eating and daily walking start now.
3. Month 2Results. Fasting insulin markedly raised with HbA1c at the top of the normal range: a glycotoxic pattern. Homocysteine raised with low-normal B12. Vitamin D low. Omega-3 index low. hs-CRP mildly raised. ApoE3/4. Sleep apnoea screen positive. Thyroid, metals, and infection screening unremarkable. A two-and-a-half-hour session to go through it, and a written programme: dietary pattern for insulin sensitivity, B vitamins at treatment doses, vitamin D and omega-3 repletion, structured exercise including resistance work, cognitive activity, and referral for a sleep study and hearing assessment.
4. Months 3 to 6Sleep apnoea confirmed and CPAP started. Hearing aids fitted. Weight down, fasting insulin down by nearly half, homocysteine normalised. His wife reports he is sharper in the mornings and the repeated questions have become rare. Repeat MoCA at six months: 25.
5. Months 6 to 12Programme maintained with sixty-minute reviews. One setback after a chest infection and a fortnight of poor sleep, which resolved. MoCA at twelve months: 26. Memory clinic review records no progression.
6. BeyondHe continues the programme as a way of living, with reviews every few months. Whether the improvement holds over five or ten years is not something I can promise. What I can say is that in a condition expected to progress, not progressing is a result, and a modest measured improvement is a better one.

The cognitive test scores in that composite moved by three points over a year. That is within the range reported in the published series, and it is the scale of change to expect when things go well. It is not a cure. It is also not nothing.

Who this is for, and who it is not

The approach is best suited to people with subjective cognitive decline or mild cognitive impairment, to those with early Alzheimer's disease, and to people with a family history or known ApoE4 status who want to act before symptoms. The earlier the stage, the more there is to work with.

In moderate or advanced dementia the realistic goals are different, and I will say so on the discovery call rather than take on work that is unlikely to help. Where the assessment suggests something other than Alzheimer's, such as a vascular, Lewy body or frontotemporal process, or a reversible cause that needs urgent NHS investigation, I will make the referral. I am a hospital doctor, and recognising when something belongs with a neurologist is part of what you are paying for.

I see patients at The Maris Practice in Twickenham, at a second consulting location in Richmond, and online across the UK. A discovery call costs nothing and takes about fifteen minutes.