User login
A new European consensus statement offers expert guidance on which biomarkers to use for patients presenting with cognitive complaints.
Led by Giovanni B. Frisoni, MD, laboratory of neuroimaging of aging, University of Geneva, and director of the memory clinic at Geneva University Hospital, the multidisciplinary task force set out to define a patient-centered diagnostic workflow for the rational and cost-effective use of biomarkers in memory clinics.
The new algorithm is part of a consensus statement presented at the Congress of the European Academy of Neurology 2023. An interim update was published in June in Alzheimer’s and Dementia.
Which biomarker?
Many biomarkers can aid diagnosis, said Dr. Frisoni; the challenge is choosing which biomarker to use for an individual patient.
A literature-based search, he said, yields a number of recommendations, but the vast majority of these are either disease based or biomarker based. The task force notes that “in vivo biomarkers enable early etiological diagnosis of neurocognitive disorders. While they have good analytical validity, their clinical validity and utility are uncertain.”
“When you have a patient in front of you, you don’ t know whether they have Alzheimer’s disease,” Dr. Frisoni said.
“You have a differential diagnosis to make, and you have a number of biomarkers – a number of weapons in your armamentarium – you have to choose. You can’t use all of them – we would like to, but we cannot.”
He added that trying to determine from the literature which biomarker is most appropriate given individual clinical conditions and all of the potential combinations is impossible.
“You will not find evidence of the comparative diagnostic value and the added diagnostic value” of one test vs, another, he noted.
“Is CSF [cerebrospinal fluid] better than amyloid PET in a particular clinical situation? What do I gain in terms of positive and negative predictive value in all the possible clinical conditions that I encounter in my clinical practice?”
Dr. Frisoni said the reality is that clinicians in memory clinics end up using biomarkers that are “based on clinical opportunities.”
For instance, “if you have a proficient nuclear medic, you use PET a lot.” In contrast, “if you have a proficient laboratory medic,” CSF markers will be favored – a situation that he said is “not ideal” and has resulted in large discrepancies in diagnostic approaches across Europe.
Harmonizing clinical practice
In a bid to harmonize clinical practice, 22 European experts from 11 European scientific societies and the executive director of Alzheimer Europe set out to develop a multidisciplinary consensus algorithm for the biomarker-based diagnosis of neurocognitive disorders in general, rather than specific neurocognitive disorders.
They used the Delphi method, in which a systematic literature review of the literature was followed by the drafting of a series of clinical statements by an executive board. These were then presented to the expert panel. If a majority consensus was reached on a given statement, it was considered closed. Questions for which there was no consensus were revised and presented to the panel again. The process was repeated until a consensus was reached.
A total of 56 statements underwent six rounds of discussion. A final online meeting led to the development of a diagnostic algorithm for patients who attend memory clinics for cognitive complaints.
The algorithm features three potential assessment waves. Wave 1 defines 11 clinical profiles that are based on the results of clinical and neuropsychological assessments, blood exams, brain imaging, and, in specific cases, electroencephalography. Wave 2 defines first-line biomarkers based on Wave 1 clinical profiles, and Wave 3 defines the second-line biomarker based on Wave 2 biomarker results.
When a patient’s clinical profile suggests Alzheimer’s disease and, in undefined cases, cerebrospinal fluid biomarkers are used first line. When CSF is inconclusive, 18-fluorodeoxyglucose positron emission tomography (FDG-PET) is used second line.
When the clinical profile suggests frontotemporal lobar degeneration or motor tauopathies, FDG-PET is first line and CSF biomarkers second line in atypical metabolic patter cases. When the clinical profile suggests Lewy body disease, dopamine transporter SPECT is first line and cardia I23I-metaiodobenzylguanidine scintigraphy is second line.
Dr. Frisoni noted that the panel strongly recommends performing biomarker tests for patients younger than 70. For those aged 70-85 years, biomarker testing is only recommended for patients with specific clinical features. For patients older than 85, biomarker testing is recommended only in “exceptional circumstances.”
Dr. Frisoni noted that the consensus document has a number of limitations.
“First of all, we could not capture all the theoretical possible combinations” of potential diagnosis and relevant biomarker tests. “There are so many that it’s virtually impossible.”
He also noted that the agreement among the panel for the use of some markers was “relatively low” at “barely 50%,” while for others, the agreement was approximately 70%.
The consensus document also does not explicitly address patients with “mixed pathologies,” which are common. In addition, it does not include emerging biomarkers, such as neurofilament light polypeptide levels, an indicator of axonal compromise.
“Last, but not least,” Dr. Frisoni said, the consensus document requires validation.
“This is a paper and pencil exercise. We, as self-appointed experts, can recommend ... whatever we want, but we must check whether what we write is applicable, feasible.”
In other words, it must be determined whether the “real patient journey” fits with the “ideal patient journey” set out in the consensus document.
This kind of validation, Dr. Frisoni said, is “usually not done for this type of exercise,” but “we want to do it in this case.”
Pros and cons
Bogdan Draganski, MD, consultant in neurology at the department of clinical neurosciences and director of the neuroimaging research laboratory, University Hospital of Lausanne (Switzerland), who cochaired the session, told this news organization that he was “swaying between two extremes” when considering the usefulness of the consensus document.
On one hand, the “reductionist approach” of breaking down a “complex issue into an algorithm” via the Delphi method risks introducing subjective bias.
He said machine learning and artificial intelligence could answer some of the questions posed by clinicians and, by extension, the statements included in the Delphi process by assessing the available data in a more objective manner.
On the other hand, Dr. Draganski said that reducing the options available to clinicians when making a differential diagnosis into the current algorithm is, pragmatically speaking, a “good approach.”
From this standpoint, the danger of using machine learning to answer clinical questions is that it “doesn’t take the responsibility” for the final decision, which means “we’re closing the loop of subjective decision-making for an individual doctor.”
He also applauded the idea of trying to provide more uniform patient assessment across Europe, although he believes “we have a long way to go” before it can deliver on the promise of personalized medicine.
Like Dr. Frisoni, Dr. Draganski noted the fact that patients with potential neurocognitive disorders often have multiple pathologies, which can include cardiovascular problems, depression, and cancer and that that could affect the choice of diagnostic biomarkers.
The second issue, he said, concerns implementation of the consensus document, which is a political decision that centers around “how politicians will define ‘uniformity’ and equal access to technological or nontechnological platforms.”
Achieving uniformity will require a pan-regional collaboration, he noted.
The task force was supported by unrestricted grants from F. Hoffmann-La Roche, Biogen International GmbH, Eisai Europe Limited, Life Molecular Imaging GmbH, and OM Pharma Suisse SA. The authors have disclosed no relevant financial relationships.
A version of this article first appeared on Medscape.com.
A new European consensus statement offers expert guidance on which biomarkers to use for patients presenting with cognitive complaints.
Led by Giovanni B. Frisoni, MD, laboratory of neuroimaging of aging, University of Geneva, and director of the memory clinic at Geneva University Hospital, the multidisciplinary task force set out to define a patient-centered diagnostic workflow for the rational and cost-effective use of biomarkers in memory clinics.
The new algorithm is part of a consensus statement presented at the Congress of the European Academy of Neurology 2023. An interim update was published in June in Alzheimer’s and Dementia.
Which biomarker?
Many biomarkers can aid diagnosis, said Dr. Frisoni; the challenge is choosing which biomarker to use for an individual patient.
A literature-based search, he said, yields a number of recommendations, but the vast majority of these are either disease based or biomarker based. The task force notes that “in vivo biomarkers enable early etiological diagnosis of neurocognitive disorders. While they have good analytical validity, their clinical validity and utility are uncertain.”
“When you have a patient in front of you, you don’ t know whether they have Alzheimer’s disease,” Dr. Frisoni said.
“You have a differential diagnosis to make, and you have a number of biomarkers – a number of weapons in your armamentarium – you have to choose. You can’t use all of them – we would like to, but we cannot.”
He added that trying to determine from the literature which biomarker is most appropriate given individual clinical conditions and all of the potential combinations is impossible.
“You will not find evidence of the comparative diagnostic value and the added diagnostic value” of one test vs, another, he noted.
“Is CSF [cerebrospinal fluid] better than amyloid PET in a particular clinical situation? What do I gain in terms of positive and negative predictive value in all the possible clinical conditions that I encounter in my clinical practice?”
Dr. Frisoni said the reality is that clinicians in memory clinics end up using biomarkers that are “based on clinical opportunities.”
For instance, “if you have a proficient nuclear medic, you use PET a lot.” In contrast, “if you have a proficient laboratory medic,” CSF markers will be favored – a situation that he said is “not ideal” and has resulted in large discrepancies in diagnostic approaches across Europe.
Harmonizing clinical practice
In a bid to harmonize clinical practice, 22 European experts from 11 European scientific societies and the executive director of Alzheimer Europe set out to develop a multidisciplinary consensus algorithm for the biomarker-based diagnosis of neurocognitive disorders in general, rather than specific neurocognitive disorders.
They used the Delphi method, in which a systematic literature review of the literature was followed by the drafting of a series of clinical statements by an executive board. These were then presented to the expert panel. If a majority consensus was reached on a given statement, it was considered closed. Questions for which there was no consensus were revised and presented to the panel again. The process was repeated until a consensus was reached.
A total of 56 statements underwent six rounds of discussion. A final online meeting led to the development of a diagnostic algorithm for patients who attend memory clinics for cognitive complaints.
The algorithm features three potential assessment waves. Wave 1 defines 11 clinical profiles that are based on the results of clinical and neuropsychological assessments, blood exams, brain imaging, and, in specific cases, electroencephalography. Wave 2 defines first-line biomarkers based on Wave 1 clinical profiles, and Wave 3 defines the second-line biomarker based on Wave 2 biomarker results.
When a patient’s clinical profile suggests Alzheimer’s disease and, in undefined cases, cerebrospinal fluid biomarkers are used first line. When CSF is inconclusive, 18-fluorodeoxyglucose positron emission tomography (FDG-PET) is used second line.
When the clinical profile suggests frontotemporal lobar degeneration or motor tauopathies, FDG-PET is first line and CSF biomarkers second line in atypical metabolic patter cases. When the clinical profile suggests Lewy body disease, dopamine transporter SPECT is first line and cardia I23I-metaiodobenzylguanidine scintigraphy is second line.
Dr. Frisoni noted that the panel strongly recommends performing biomarker tests for patients younger than 70. For those aged 70-85 years, biomarker testing is only recommended for patients with specific clinical features. For patients older than 85, biomarker testing is recommended only in “exceptional circumstances.”
Dr. Frisoni noted that the consensus document has a number of limitations.
“First of all, we could not capture all the theoretical possible combinations” of potential diagnosis and relevant biomarker tests. “There are so many that it’s virtually impossible.”
He also noted that the agreement among the panel for the use of some markers was “relatively low” at “barely 50%,” while for others, the agreement was approximately 70%.
The consensus document also does not explicitly address patients with “mixed pathologies,” which are common. In addition, it does not include emerging biomarkers, such as neurofilament light polypeptide levels, an indicator of axonal compromise.
“Last, but not least,” Dr. Frisoni said, the consensus document requires validation.
“This is a paper and pencil exercise. We, as self-appointed experts, can recommend ... whatever we want, but we must check whether what we write is applicable, feasible.”
In other words, it must be determined whether the “real patient journey” fits with the “ideal patient journey” set out in the consensus document.
This kind of validation, Dr. Frisoni said, is “usually not done for this type of exercise,” but “we want to do it in this case.”
Pros and cons
Bogdan Draganski, MD, consultant in neurology at the department of clinical neurosciences and director of the neuroimaging research laboratory, University Hospital of Lausanne (Switzerland), who cochaired the session, told this news organization that he was “swaying between two extremes” when considering the usefulness of the consensus document.
On one hand, the “reductionist approach” of breaking down a “complex issue into an algorithm” via the Delphi method risks introducing subjective bias.
He said machine learning and artificial intelligence could answer some of the questions posed by clinicians and, by extension, the statements included in the Delphi process by assessing the available data in a more objective manner.
On the other hand, Dr. Draganski said that reducing the options available to clinicians when making a differential diagnosis into the current algorithm is, pragmatically speaking, a “good approach.”
From this standpoint, the danger of using machine learning to answer clinical questions is that it “doesn’t take the responsibility” for the final decision, which means “we’re closing the loop of subjective decision-making for an individual doctor.”
He also applauded the idea of trying to provide more uniform patient assessment across Europe, although he believes “we have a long way to go” before it can deliver on the promise of personalized medicine.
Like Dr. Frisoni, Dr. Draganski noted the fact that patients with potential neurocognitive disorders often have multiple pathologies, which can include cardiovascular problems, depression, and cancer and that that could affect the choice of diagnostic biomarkers.
The second issue, he said, concerns implementation of the consensus document, which is a political decision that centers around “how politicians will define ‘uniformity’ and equal access to technological or nontechnological platforms.”
Achieving uniformity will require a pan-regional collaboration, he noted.
The task force was supported by unrestricted grants from F. Hoffmann-La Roche, Biogen International GmbH, Eisai Europe Limited, Life Molecular Imaging GmbH, and OM Pharma Suisse SA. The authors have disclosed no relevant financial relationships.
A version of this article first appeared on Medscape.com.
A new European consensus statement offers expert guidance on which biomarkers to use for patients presenting with cognitive complaints.
Led by Giovanni B. Frisoni, MD, laboratory of neuroimaging of aging, University of Geneva, and director of the memory clinic at Geneva University Hospital, the multidisciplinary task force set out to define a patient-centered diagnostic workflow for the rational and cost-effective use of biomarkers in memory clinics.
The new algorithm is part of a consensus statement presented at the Congress of the European Academy of Neurology 2023. An interim update was published in June in Alzheimer’s and Dementia.
Which biomarker?
Many biomarkers can aid diagnosis, said Dr. Frisoni; the challenge is choosing which biomarker to use for an individual patient.
A literature-based search, he said, yields a number of recommendations, but the vast majority of these are either disease based or biomarker based. The task force notes that “in vivo biomarkers enable early etiological diagnosis of neurocognitive disorders. While they have good analytical validity, their clinical validity and utility are uncertain.”
“When you have a patient in front of you, you don’ t know whether they have Alzheimer’s disease,” Dr. Frisoni said.
“You have a differential diagnosis to make, and you have a number of biomarkers – a number of weapons in your armamentarium – you have to choose. You can’t use all of them – we would like to, but we cannot.”
He added that trying to determine from the literature which biomarker is most appropriate given individual clinical conditions and all of the potential combinations is impossible.
“You will not find evidence of the comparative diagnostic value and the added diagnostic value” of one test vs, another, he noted.
“Is CSF [cerebrospinal fluid] better than amyloid PET in a particular clinical situation? What do I gain in terms of positive and negative predictive value in all the possible clinical conditions that I encounter in my clinical practice?”
Dr. Frisoni said the reality is that clinicians in memory clinics end up using biomarkers that are “based on clinical opportunities.”
For instance, “if you have a proficient nuclear medic, you use PET a lot.” In contrast, “if you have a proficient laboratory medic,” CSF markers will be favored – a situation that he said is “not ideal” and has resulted in large discrepancies in diagnostic approaches across Europe.
Harmonizing clinical practice
In a bid to harmonize clinical practice, 22 European experts from 11 European scientific societies and the executive director of Alzheimer Europe set out to develop a multidisciplinary consensus algorithm for the biomarker-based diagnosis of neurocognitive disorders in general, rather than specific neurocognitive disorders.
They used the Delphi method, in which a systematic literature review of the literature was followed by the drafting of a series of clinical statements by an executive board. These were then presented to the expert panel. If a majority consensus was reached on a given statement, it was considered closed. Questions for which there was no consensus were revised and presented to the panel again. The process was repeated until a consensus was reached.
A total of 56 statements underwent six rounds of discussion. A final online meeting led to the development of a diagnostic algorithm for patients who attend memory clinics for cognitive complaints.
The algorithm features three potential assessment waves. Wave 1 defines 11 clinical profiles that are based on the results of clinical and neuropsychological assessments, blood exams, brain imaging, and, in specific cases, electroencephalography. Wave 2 defines first-line biomarkers based on Wave 1 clinical profiles, and Wave 3 defines the second-line biomarker based on Wave 2 biomarker results.
When a patient’s clinical profile suggests Alzheimer’s disease and, in undefined cases, cerebrospinal fluid biomarkers are used first line. When CSF is inconclusive, 18-fluorodeoxyglucose positron emission tomography (FDG-PET) is used second line.
When the clinical profile suggests frontotemporal lobar degeneration or motor tauopathies, FDG-PET is first line and CSF biomarkers second line in atypical metabolic patter cases. When the clinical profile suggests Lewy body disease, dopamine transporter SPECT is first line and cardia I23I-metaiodobenzylguanidine scintigraphy is second line.
Dr. Frisoni noted that the panel strongly recommends performing biomarker tests for patients younger than 70. For those aged 70-85 years, biomarker testing is only recommended for patients with specific clinical features. For patients older than 85, biomarker testing is recommended only in “exceptional circumstances.”
Dr. Frisoni noted that the consensus document has a number of limitations.
“First of all, we could not capture all the theoretical possible combinations” of potential diagnosis and relevant biomarker tests. “There are so many that it’s virtually impossible.”
He also noted that the agreement among the panel for the use of some markers was “relatively low” at “barely 50%,” while for others, the agreement was approximately 70%.
The consensus document also does not explicitly address patients with “mixed pathologies,” which are common. In addition, it does not include emerging biomarkers, such as neurofilament light polypeptide levels, an indicator of axonal compromise.
“Last, but not least,” Dr. Frisoni said, the consensus document requires validation.
“This is a paper and pencil exercise. We, as self-appointed experts, can recommend ... whatever we want, but we must check whether what we write is applicable, feasible.”
In other words, it must be determined whether the “real patient journey” fits with the “ideal patient journey” set out in the consensus document.
This kind of validation, Dr. Frisoni said, is “usually not done for this type of exercise,” but “we want to do it in this case.”
Pros and cons
Bogdan Draganski, MD, consultant in neurology at the department of clinical neurosciences and director of the neuroimaging research laboratory, University Hospital of Lausanne (Switzerland), who cochaired the session, told this news organization that he was “swaying between two extremes” when considering the usefulness of the consensus document.
On one hand, the “reductionist approach” of breaking down a “complex issue into an algorithm” via the Delphi method risks introducing subjective bias.
He said machine learning and artificial intelligence could answer some of the questions posed by clinicians and, by extension, the statements included in the Delphi process by assessing the available data in a more objective manner.
On the other hand, Dr. Draganski said that reducing the options available to clinicians when making a differential diagnosis into the current algorithm is, pragmatically speaking, a “good approach.”
From this standpoint, the danger of using machine learning to answer clinical questions is that it “doesn’t take the responsibility” for the final decision, which means “we’re closing the loop of subjective decision-making for an individual doctor.”
He also applauded the idea of trying to provide more uniform patient assessment across Europe, although he believes “we have a long way to go” before it can deliver on the promise of personalized medicine.
Like Dr. Frisoni, Dr. Draganski noted the fact that patients with potential neurocognitive disorders often have multiple pathologies, which can include cardiovascular problems, depression, and cancer and that that could affect the choice of diagnostic biomarkers.
The second issue, he said, concerns implementation of the consensus document, which is a political decision that centers around “how politicians will define ‘uniformity’ and equal access to technological or nontechnological platforms.”
Achieving uniformity will require a pan-regional collaboration, he noted.
The task force was supported by unrestricted grants from F. Hoffmann-La Roche, Biogen International GmbH, Eisai Europe Limited, Life Molecular Imaging GmbH, and OM Pharma Suisse SA. The authors have disclosed no relevant financial relationships.
A version of this article first appeared on Medscape.com.