Given a long enough lifespan, perhaps the brain would degenerate to a point such that we'd all develop Alzheimer's disease if we lived long enough. There's some basis for this notion; many investigators accept it. But careful analysis shows a differential effect for neurofibrillary tangles and amyloid plaques that we think discriminates aging from AD. There is a slow accumulation of neurofibrillary tangles in vulnerable areas of the brain (entorhinal cortex, hippocampus, and other medial temporal lobe structures) beginning around age 50—if you look hard enough you can find a few tangles in almost everyone of that age or older. Even in people with absolutely normal cognitive abilities, the older they are, the more likely we are to see tangles in these vulnerable regions. It is very likely that the tangles mark neuronal degeneration in these regions.
We postulate that, in the normal human lifespan, this slow accumulation of neurofibrillary tangles does not reach a point where it would cause dementia. If we lived to 200, maybe, but at 100 years I doubt it causes dementia. Could this slow process of neuronal degeneration, marked by tangles, in vulnerable medial temporal lobe structures be responsible for some of the normal cognitive changes that healthy older adults experience, such as, "I can't remember the name of that person" or "I process information more slowly"? Perhaps so. But in and of itself, this slow process of tangle accumulation does not result in frank dementia.
We believe that there is a disease process that marks the onset of AD that occurs in addition to this very gradual, age-related neurodegeneration in restricted areas such as the entorhinal cortex and hippocampus. This is manifested by the deposition of amyloid in the cerebral cortex. Abnormalities of amyloid generation or metabolism that result in cerebral deposition mark the disease. That's what we think. We find that cognitively healthy older adults can live into the ninth, even the tenth decade of life without cerebral amyloid deposition, so that amyloid deposition is not part of truly healthy aging but instead reflects a disease process. These concepts are open for debate. Perhaps the reason Alzheimer's is so powerfully age-associated is because this age-related accumulation of neurofibrillary tangles predisposes the brain to developing this disease. Nevertheless, our conceptual basis is that age alone is insufficient—even if we all lived to be 100, we would not necessarily develop Alzheimer's. Amyloid deposition triggers the disease.
GS: This is the concept. Now the practice?
JM: Consider more than 25 years of cognitive test performance data and clinical experience. Both show that in the absence of amyloid deposition, cognitive abilities of older adults can be quite good. There's a perception that we inevitably decline in memory and thinking abilities as we get older. I grant you that there is a bit of that, but the amount of decline accepted as "normal" often is exaggerated. That's in part because in many studies comparing performance on cognitive test batteries of 20-year-olds, 40-year-olds, 60-year-olds, and 80-year-olds, the 80-year-old group includes people with unrecognized very mild AD who lower the group's performance. If we remove the prodromal AD people who contaminate that sample, then cognitive performance is surprisingly good. We see very little cognitive decline over 25 years in truly healthy individuals as we follow them.
There are limitations, of course, to our studies. It's terrific to do longitudinal studies in the same group of individuals year after year after year, and find little decline. On the other hand, we are likely examining motivated, cognitively engaged people who are not representative of the general population of older adults. However, even if our sample is selected, it does demonstrate that it clearly is possible to have stable cognitive performance even late in life.
For physicians, here's what this research observation means in practice: From a cognitive standpoint, older adults who do not develop Alzheimer's disease or another dementia can do all the things they need to do to maintain their independence. We should expect that; that should be the norm. If we are fortunate enough to escape Alzheimer's, our cognitive abilities allow us to make decisions and live independently—and that's, of course, what people want. When we mentally project what we will be like in our retirement years, we don't see ourselves as demented, dependent, and disabled. We see ourselves as mentally intact. Isn't that the image you have?
JM: And that's possible. That's our goal. In contrast, when people begin to lose their ability to maintain independence from memory and thinking problems, that point indicates early disease.
In our studies, early disease is marked by a precipitous drop in neuropsychologic test performance. It is not a gradual decline like you would think if Alzheimer's disease were accelerated aging. There's a clear inflection point. This contrasts with the gradual clinical onset of the symptomatic disease. Very gradually individuals begin to lose the ability to do the things they once did. Others gradually have to take over for them. That change, from being cognitively independent to being less able to do the particular things they once did—that's the mark of disease.
GS: It's fair to say you're famous for your skill in diagnosing…
JM: I thought it was for my looks. Gosh darn…
GS: That, too, but I wouldn't be so blunt. You are known for your skill in diagnosing Alzheimer's very early. What's your secret?
JM: This is a result of the methods developed by the Alzheimer's Disease Research Program at Washington University, which was initiated by Leonard Berg and colleagues. The key has been to keep the clinical distinctions—diagnosis, classification, and staging of severity—separate from the neuropsychologic batteries. We have developed many important findings from neuropsychologic tests, but test performance does not enter into our diagnosis. Keeping clinical and neuropsych assessments separate has allowed us to compare how our participants do on these independent measures as they go forward in time. One doesn't confound the other.
The clinical diagnosis rests on clinically meaningful change for that individual patient; that is, it is face-valid. The main point is a change from what that particular person once was able to do from a cognitive standpoint, the premise being, again, that healthy aging is relatively stable. If there's a decline in that person's performance of his or her accustomed activities, then that triggers the suspicion of dementia. We spend a great deal of time trying to find out what the person once did, what were his or her activities. Everyone's background is different. You have to reference the cognitive change in terms relevant for that person, because we all have different strengths and skills, and hence, different levels of usual cognitive abilities.
I remember my first case as I was coming on board here in 1984. One of our healthy participants came for his fourth annual assessment; he had been absolutely cognitively normal at his first three. He was an 84-year-old, retired, very highly accomplished investment banker who still maintained his household finances and investment portfolio and was active in investment clubs. His wife said at this fourth time of assessment that he had experienced "a little memory change" in the past year. I was struck that her example of the change was that, for the first time, her husband did not prepare their income tax return; he just put the material in a folder and gave it to their accountant. I've never done my income taxes. For me, that would not be a change, but, of course, for him it was. He died four months after that assessment from pneumonia; his brain was full of Alzheimer's disease pathology.
GS: So this was a case of incipient Alzheimer's.
JM: Yes. His wife noticed a change in an area in which he was accomplished. You asked why we are famous, and it's because of our emphasis on the individual's performance now in comparison with what he or she once did in areas they were accustomed to doing.
To know what they once did, we need someone who knows the patient well. No test is going to tell us that. We can find out how they do on the Mini-Mental State Exam (MMSE), but that doesn't say what they once were doing in their life. Of course, we see the subject, but we spend as much or more time with the informant. We ask: Is he driving, does he shop, does he cook, does he launder, what are his hobbies?
Anther face-valid assessment we are using (on top of the standard neuropsychologic measures) is to establish with the informant an account of something the subject has done recently. Say last week they went to the rush-hour showing of a current movie, finished it at 7:30, and went out for dinner. We'll ask the informant about that, and then we'll ask the person and note how much the subject remembers of this episode.
By face-valid I mean relevant to the patient. We base the diagnosis on this clinical information; a person's test performance matters less. Most investigators rely on tests to sort, categorize, and stage people. However, someone can score 23 on the Mini-Mental State, but perhaps that's "normal" for them (no change) because they're bad test-takers, or poorly educated. Granted, most people with a Mini-Mental State score of 23 are demented; that's why it works. But not everyone.
GS: You once said that in your experience of comparing results of neuropsych tests and clinical interviews, the skilled clinician always picked up the earliest changes before any psychometric changes showed up.
JM: That's right. When I talk about this precipitous drop in the cognitive test performance pattern—remember, the neuropsychologic performance does not enter into our diagnosis—it actually is occurring probably a year or so after our clinical diagnosis. The investment banker I mentioned performed normally on his cognitive test battery at the time we made the diagnosis based on his wife saying he had these vague changes. Now, we're not always right; the diagnosis is not always easy or correct! In fact, I remember back in 1987, I was wrong once.
GS: Your humility is appreciated!
JM: It takes some experience to sort out what the informant is saying and to link it to what the subject is doing. Sometimes the informant is in denial or is unobservant, even though the subject may be clearly demented.
GS: Can bickering with the spouse confound this assessment?
JM: Another anecdote. One of our participants is a monozygotic twin, and his twin brother a few years ago developed Alzheimer's disease, so we're waiting for the shoe to drop for this fellow. Recently, his wife said: "I'm beginning to see changes; he is showing poor judgment, making bad investments, and I really think he's impaired, and I need to step in." Turns out they were having marital discord. She wanted more control over the finances; he was fine. However, secondary gain or ulterior motives are rare in our informants. More often, it may be difficult for them to report the changes, especially if the informant is a child; it's hard for them to say: "Dad's slipping." They want to think: "He's retired now, maybe that's why." Sometimes it goes the other way, especially if they come from a family where there's Alzheimer's disease. Informants then can be oversensitive to even normal changes, for example: "She left the stove burner on" when that happened only once months ago. To be clinically relevant, it must be a consistent change. We all can experience occasional memory lapses. So it does take skill to tease out the relevant information. I admit, even though we think we're good at recognizing the disease very early, there are situations where we remain uncertain.
GS: In those very early stages, how do you distinguish Alzheimer's from all those related, and unrelated, forms of dementias?
JM: The changes at the very earliest stages of Alzheimer's disease are the same changes you see in people with more advanced Alzheimer's disease, just milder. There are characteristic memory problems in everyday life of Alzheimer's patients. They repeat the same question over and over: "What time do we have to go there?" Well, we have to be at the dinner party at six. A few minutes later: "What time do we have to go there?" Over and over. Subjects misplace items in funny places or they have difficulty with cognitive processing.
Here is an example: An older driver begins to have problems in making left-hand turns. They're at the stop sign or traffic light, and they've got to process the flow of traffic in two different directions simultaneously. They focus on only one direction, pull out, and get hit. (see also Duchek et al., 2003.)
There are characteristic features we hear in mild Alzheimer's disease that are also apparent in very mild Alzheimer's disease, just less pronounced. Moreover, related forms of dementia have distinguishing features. For example, if a person has vivid visual hallucinations at this very mild stage, we consider dementia with Lewy bodies.
GS: In politics, at least, intelligence information is frequently wrong and hard to verify. Have you studied whether your informants are accurate?
JM: Yes, see Cacchione et al., 2003 and Carr et al., 2000. We've found that the self-reported memory complaints of our participants are much less predictive of whether they go on to become demented than when the informant says the participant has a memory problem.
The patient error cuts both ways. Sometimes, the most high-functioning, highly educated people complain bitterly, e.g.: "I forgot the Latin name for this particular flower." They're convinced they're losing it, but are still functioning fine. Think back to the independence: they're still doing the things they want to do. Maybe they have lost some memory, but it doesn't interfere and no one else can detect it. Those kinds of complaints don't predict future dementia.
On the other hand, if the collateral says they're losing it, that's important. Generally, the patients who have early dementia lack insight. They don't complain. "Do you have a memory problem?" "No." That sums up one study we did: The informant prediction of a memory problem is much more accurate than the self-report or the lack of self-report of the subject.
In the Cacchione study, we asked the informant practical questions: "If you gave your wife/father a short shopping list of five items—don't write it down, just say it—are they likely to remember?" We don't actually go shopping with the subject, but we ask them a five-item memory test. We compare the subjects' performance on the five-item recall test with the informant's prediction, and the fit is very high.
Again, I admit it takes some experience to elicit the information, because the collaterals may offer other explanations for the changes. They may say, "I think he's more forgetful, but that's because he's worried about our son who lost his job," or something like that. The collateral may not ascribe the changes to dementia, but what they predict about the subject performance is accurate.
GS: The clinical dementia rating (CDR) was invented in this center. What is it and how widely is it being used?
JM: The CDR is a dementia staging instrument to judge dementia severity (Morris et al.,1991). It's interesting, Gabrielle, that the CDR often is considered to be a diagnostic test. If you are rated as CDR 0.0, you have no dementia; if you're rated as 0.5 or beyond, you have dementia. In that way it's diagnostic, but the main purpose is to stage dementia severity.
It has been widely adopted in many dementia research programs and clinical trials, even internationally. Many people know it and use it because it's face-valid. It gives all of us an idea of where people are in their dementia course, and we can relate to it. We developed a protocol that we use for the informant and subject interviews. We've truncated it to make it more adaptable for other centers, but because it is structured, it's possible to train other people to use it in a standardized fashion. So the designation of CDR 1 in San Diego is the equivalent to a CDR 1 in Chicago, and thus we can compare demented individuals across sites. As a clinical instrument, physicians and non-physicians can use it. Nurses actually are more accurate, because physicians tend to think they know better than the semi-structured interview, but nurses adhere to the way it's presented. So it's something that we can all train on and compare.
GS: Other respected clinicians have said they can't pin down that CDR 0.5 stage as well as your group. Is this a matter of experience, training, viewpoint? What's the problem?
JM: Two issues here. One is that the CDR 0.5 group encompasses a fairly heterogeneous group of subjects. Not everyone in the mild cognitive impairment (MCI) stage is going to have Alzheimer's disease, but we can identify those who will pretty well. In fact, we just label these individuals as very mild Alzheimer's. There are also some people who are CDR 0.5 who we don't think are going to have Alzheimer's disease, or whom we're uncertain about. As I told you, some cases are really difficult, so we do not label them as AD. Many people do not understand that we don't call every 0.5 person demented.
The other issue is: why do we call it CDR 0.5 and other people call it MCI? We haven't done this systematically, but I think two reasons apply. First, unlike most dementia centers, we rely simply on the clinical information, the informant-based interview. Most other centers will do a consensus conference with a neuropsychologist, neurologist, psychiatrist, geriatrician. We don't care how they did on the neuropsychological battery; we don't encompass that in our diagnosis. Our call comes from the change we talked about, and we're very comfortable because our data shows that's very predictive. I don't think other people take as much time as we do probing an informant, and instead they'll rely on the test performance. It's a different emphasis in how we get the relevant information, and it's coming back to face-valid, patient-oriented rather than test performance.
The second consideration is a threshold phenomenon. Different physicians have different levels of comfort in calling people demented. This is not an inconsequential issue. In California, individuals diagnosed with very mild dementia are reported to the motor vehicles department. There are consequences with long-term health care insurance. Many physicians don't want to deal with these issues until necessary, so they are concerned about labeling.
We've just started a study measuring the impact on the patient and on the family of giving the diagnosis of AD. We look at practical issues—driving, insurance, employability for those who are employed, genetic testing—as well as the emotional impact. Our hypothesis is that the same issues apply regardless of whether the individual is CDR 0.5 versus CDR 1 or 2. We don't think that CDR 0.5 people are more likely to suffer from being told, but no one has looked at what is the impact of saying, "You have dementia" at a very early stage.
At the same time, there are advantages to diagnosing early. That's a time when the subject can still make decisions. "I want so-and-so to be my durable power-of-attorney. Here are my advance directives. If I become incapacitated, I want to go to live with my daughter, or I want to stay in assisted living with my friend." They can be involved in their planning. If they need to revise their will, this is the time. Most importantly, we think that's where treatment ought to be introduced. So are there labeling concerns? Yes. Are there advantages to diagnosing early? Yes.
GS: Your threshold of clinical detection now is CDR 0.5, but before that there's a preclinical phase during which pathology builds up silently. You believe the brain already sustains significant damage in that time. What will it take to push the diagnosis into this phase?
JM: Preclinical detection is where we're devoting a great deal of our resources. There is lots of evidence to suggest that AD pathology begins years or decades before any clinical symptoms appear. A good model of that is Down's syndrome, where plaques can develop in the second decade of life, but dementia doesn't appear until the forties or later. There now are promising drug discovery efforts, most based on the amyloid hypothesis of Alzheimer's disease, for mechanism-based therapeutic interventions. If one or more of these strategies prove effective and safe, then it may be possible to arrest the pathologic cascade incited by amyloid deposition.
But by the time dementia is expressed, even at the CDR 0.5 stage, already there has been substantial neuronal damage, synaptic loss, nerve cell loss. Let's assume we have an effective disease-modifying drug: Currently, we can't introduce the drug until the person is diagnosed, when the brain has substantial damage. Even if the treatment arrests the disease, the person's cognitive abilities are unlikely to be restored. So it makes sense to try to detect the disease in the presymptomatic stage. Once we have disease-modifying therapies, ideally we would intervene in the presymptomatic stage, before substantial brain damage has occurred and before dementia is expressed. Now we're not talking about treatment; we're talking about prevention.
I think the dementia phase is the end-stage of the disease process. We need to look at the antecedent factors that cause the disease, the initial brain changes, and identify people developing these changes.
These dual purposes—disease-modifying therapy and preclinical detection—put into our reach the chance to actually prevent Alzheimer's disease. I think that's the real excitement, the real promise in the field.
With that long preamble, what are we doing? We have a number of lines of inquiry—cognitive changes, biochemical changes in blood plasma and in cerebral spinal fluid, and imaging correlates—to look at what indicates the very beginning of the brain disease. This January, we started a major study of adult children of people who had Alzheimer's disease. In subjects age 45 and older, we're monitoring their clinical, cognitive, behavioral, blood, spinal fluid, and imaging features. We plan to follow them and see what indicates change in the brain. Hopefully in the next decade, we'll have these disease-modifying drugs to treat those people and see if we can't prevent the expression of Alzheimer's disease.
GS: A next-generation natural history study. What are you doing on imaging?
JM: We're doing structural and functional MRI. The big, new excitement is amyloid imaging (see ARF related news story), and we are pursuing a couple of protocols with that. We've been collaborating with William Klunk and his group at the University of Pittsburgh. By the way, Bill was an M.D./Ph.D. student in the same Washington University laboratory where I did a neuropharmacology fellowship, so I take full credit for his work.
GS: Not only the most beautiful, but also the brightest!
JM: All right, Bill deserves some credit, also! He did some of it. Seriously, he's made a major contribution to the field and has been a terrific collaborator.
GS: How are you using the PIB marker?
JM: We first have to replicate the Pittsburgh study to demonstrate at another site that PIB discriminates Alzheimer's cases from nondemented controls. Then we want to use it in the nondemented controls to determine if we can identify those people who are beginning to develop amyloid plaques and what happens to them clinically.
GS: That would be a way of finally testing the amyloid hypothesis definitively.
JM: No question. From a therapeutic standpoint, it's important because now we don't have to rely on the treatment group in clinical trials declining less on cognitive tests over a year or more. Instead, serial brain images may indicate whether the drug affects the amyloid burden. I'm being very simplistic, but I do think PIB is going to spark-plug therapeutic monitoring.
GS: If PIB proves its worth and can be tied to a more convenient, long-lived radioligand like technetium or something like that, would that in the long run replace a lot of the neuropsychology testing that is now being done?
JM: It could.
GS: It's still provocative to many people that you say "silent" neuropathology means a person has Alzheimer's disease, i.e.,
people who died of other causes with plaques and tangles but no symptoms whatsoever.
How does tau fit into your picture? You've told us how tau pathology relates to aging versus Alzheimer's. Still, given that tau clearly marks neurodegeneration, I'm surprised there's so much less going on around tau in terms of its diagnostic or therapeutic value. Do you think phospho-tau will be useful as an antecedent biomarker? In what way can this important protein serve as a handle?
JM: It's a complicated question, but fortunately, my level of ignorance will lead me to simplistic responses. Without question, tau abnormalities alone can result in neurodegeneration. However, hyperphosphorylated tau and tangles are a feature of many neurodegenerative disorders. They are not pathognomonic for Alzheimer's disease, but instead represent neuronal and synaptic degeneration from multiple causes. Dementia caused by primary tauopathies does occur, as shown by Trojanowski and Lee with FTDP 17, which Hutton and Hardy first described, but that's a small number of people. More often, tangles represent the end result of neuronal degeneration incited by other causes. In the amyloid cascade hypothesis, amyloid deposition sets off the neurotoxic cascade, which causes neurodegeneration as marked by neurofibrillary tangles. Now, if we had therapeutic strategies targeted toward tau that somehow prevented the neurons from degenerating even in the presence of the amyloid, that would be very promising. But…
GS: …we don't?
JM: There are tau-related drug discovery efforts, but more is going on with amyloid immunotherapy, secretase inhibitors, and antiaggregation agents.
GS: About the heterogeneity of Alzheimer's. It's quite a varied disease and many researchers call it a syndrome. Also, people with AD often have other conditions to boot, so it's complex. Do these differences among patients really matter that much when it comes down to the essentials, that is, progression of the dementia, placement in a nursing home, and death?
JM: In the large majority of cases, I think not. Conceptually, it's very interesting: Is it a single disease or multiple diseases? As we begin to diversify our research populations, we are finding that causes or factors that we think are important for developing Alzheimer's disease in one ethnic group or one cultural group may be very different in another. So there is variability, no question. But in practice, once the disease is expressed—and again I go back to dementia being the end-phase of Alzheimer's disease—then the course really is predictable. In terms of overall outcomes, it doesn't matter much whether they have coexistent depression, coexistent stroke, coexistent Parkinson's disease. Now, are individuals all the same? No. Some become agitated, some develop hallucinations, so yes, there's variability, but it doesn't matter with the major endpoints. (See also Villareal et al., 2003.)
GS: Too bad our time is up. I have 55 more questions! We must continue this conversation another time.