I wonder whether the segregation power of this assay would be further strengthened if a marker of gliosis (GFAP, Iba1) or neuronal cell lysis (neurofilament) were added to the set.

View all comments by Florian Muller

Very interesting. I suggest PET versus SPET studies.

View all comments by alessio dalla libera

I am wildly enthusiastic about the ADNI results to date and am confident that much more will emerge that can be tested in subsequent studies.

From my perspective, the one item of greatest relevance to industry that needs clarification is whether ADNI 1 data is sufficient to specify, at least provisionally, standardized approaches to MRI data. I emphasize this since we heard at the ADNI Data Presentations meeting that structural MRI could provide a biomarker of 25 percent drug effect on disease progression in a one-year trial using fewer than 100 subjects per arm.

Anyone wanting to implement this for internal decision-making would need to specify the exact measure and analytic plan. So, at some point it would be nice to say that ADNI 1 has provided us with a standard approach to set our primary measure. Others could be secondary. One could then include in ADNI 2 a test of whether the standardized approach we took holds up as the most robust and well-behaved measure in a subsequent study.

Obviously, if we all come to believe that the FDG-PET data is clear enough to...  Read more

I am wildly enthusiastic about the ADNI results to date and am confident that much more will emerge that can be tested in subsequent studies.

From my perspective, the one item of greatest relevance to industry that needs clarification is whether ADNI 1 data is sufficient to specify, at least provisionally, standardized approaches to MRI data. I emphasize this since we heard at the ADNI Data Presentations meeting that structural MRI could provide a biomarker of 25 percent drug effect on disease progression in a one-year trial using fewer than 100 subjects per arm.

Anyone wanting to implement this for internal decision-making would need to specify the exact measure and analytic plan. So, at some point it would be nice to say that ADNI 1 has provided us with a standard approach to set our primary measure. Others could be secondary. One could then include in ADNI 2 a test of whether the standardized approach we took holds up as the most robust and well-behaved measure in a subsequent study.

Obviously, if we all come to believe that the FDG-PET data is clear enough to support standardization around a single common approach, all the better. My sense was that more work remained to be done to have a reasonable expectation of a clear consensus in terms of setting a primary approach to generating and analyzing the data.

View all comments by William Potter

ADNI is a scientific project with scientific goals, and one of the goals is to find the best methods to use for clinical trials. We are accomplishing this. The ADNI data for structural MRI now show very clearly that the greatest rate of change is in the hippocampal area and that measurements of tissue in this area have the highest statistical power to detect change. This does not mean that this region is the best or only region that will be affected by a treatment. Hence, other measurements of the brain, including whole brain volume, should be included and considered.

Having consensus conferences to help standardize analytical methods would be a good idea, and I'd be happy to participate. But it has never been ADNI's role to define a standard or suggest a specific method. Here are some of the problems:

1. The field continues to emerge.

2. We only have one-year data so far. With two-year data, the results could be different.

3. Scientists are refining their methods as we speak.

4. There are all kinds of commercial and intellectual property issues. ADNI can be aware...  Read more

ADNI is a scientific project with scientific goals, and one of the goals is to find the best methods to use for clinical trials. We are accomplishing this. The ADNI data for structural MRI now show very clearly that the greatest rate of change is in the hippocampal area and that measurements of tissue in this area have the highest statistical power to detect change. This does not mean that this region is the best or only region that will be affected by a treatment. Hence, other measurements of the brain, including whole brain volume, should be included and considered.

Having consensus conferences to help standardize analytical methods would be a good idea, and I'd be happy to participate. But it has never been ADNI's role to define a standard or suggest a specific method. Here are some of the problems:

1. The field continues to emerge.

2. We only have one-year data so far. With two-year data, the results could be different.

3. Scientists are refining their methods as we speak.

4. There are all kinds of commercial and intellectual property issues. ADNI can be aware of these issues, but we cannot solve them.

5. It would be difficult to identify a method for brain structural analysis to recommend as the "standard" FDA method. There are so many competing methods. It is unclear at this point which would be the best method for clinical trials with patient populations different from those in ADNI, and each treatment may have its own different effects on the brain.

That said, I have suggested that the PET and MRI groups consider setting up consensus conferences to discuss how to standardize acquisitions, processing, and analysis for PET and MRI data for clinical trials. It will take time to achieve consensus. ADNI can lead continued discussion on standardization of such methods, but this is an expansion of the original goals of ADNI.

We are happy to expand, but with limited or even diminishing financial resources, we have to focus on our major goals:

1. Developing new scientific information.

2. Comparing different methods in different ways.

3. Developing clinical scenarios that help inform clinical trials.

4. Developing methods for early detection of AD.

Finally, I don't think we need a "standard method" right now. It would be nice, but we don't need it. Each trial that goes to the FDA specifies its methodology, and the FDA is very flexible about this. The ADNI data set provides a range of reasonable choices.

View all comments by Michael Weiner

It seems clear that modern scanning methods have essentially replaced the tissue diagnostic criteria of the autopsy in defining the anatomical changes that accompany clinical Alzheimer's dementia. It also appears, at least to an outsider, that what are believed to be early precursors to frank clinical disease might also be identified. If these results are confirmed and extended, they might be effective ways to monitor responses to therapy, an inability that now greatly hampers drug development.

But here are my concerns. Are these imaging methods capable of detecting the earliest pathologic changes that lead to clinical dementia? This is a hard question to address, since we don’t know what these earliest changes are, or when and where in the brain they occur. It seems that what is now needed are reliable and patho-physiologically relevant biomarkers. Measuring levels of Aβ and tau in the CSF has been correlated with existing clinical dementia, but why do we believe that they are reliable reporters of the earliest stages of the disease?

I am also surprised by the...  Read more

It seems clear that modern scanning methods have essentially replaced the tissue diagnostic criteria of the autopsy in defining the anatomical changes that accompany clinical Alzheimer's dementia. It also appears, at least to an outsider, that what are believed to be early precursors to frank clinical disease might also be identified. If these results are confirmed and extended, they might be effective ways to monitor responses to therapy, an inability that now greatly hampers drug development.

But here are my concerns. Are these imaging methods capable of detecting the earliest pathologic changes that lead to clinical dementia? This is a hard question to address, since we don’t know what these earliest changes are, or when and where in the brain they occur. It seems that what is now needed are reliable and patho-physiologically relevant biomarkers. Measuring levels of Aβ and tau in the CSF has been correlated with existing clinical dementia, but why do we believe that they are reliable reporters of the earliest stages of the disease?

I am also surprised by the absence of any discussion of blood-based biomarkers in the ADNI program. Are serum samples being collected and systematically stored for future studies? Surely any widespread clinical testing program will have to be based on more accessible clinical material than cerebrospinal fluid.

View all comments by Vincent Marchesi

Reply to Michael Weiner
From a scientific viewpoint, I agree with Mike that we should remain open as to the best structural MRI method. But from an industry viewpoint, if the field can agree on a standard measure, it would make planning and powering registration studies much simpler. Gaining widespread industry appreciation of the value of ADNI 2 will be facilitated if we can agree that a "good enough" measure has been identified which we will see replicated and validated in this subsequent study.

View all comments by William Potter

Reply to William Potter
All of us in ADNI recognize the importance of developing standard methods for performing clinical AD trials. We've made considerable headway with 1) standardized MRI acquisition, 2) standardized FDG-PET acquisition, and 3) use of the Luminex platform for blood/CSF biomarkers. Furthermore, following the meeting in Seattle there has been considerable discussion among ADNI Core leaders and others to set up some working groups that would address the standardization issues. We'll get there.

View all comments by Michael Weiner