- Email: [email protected]
The clinical value of fluid biomarkers for dementia diagnosis – Authors' reply
Correspondence
The clinical value of fluid biomarkers for dementia diagnosis We read with interest the Article by Bob Olsson and colleagues1 examining CSF and blood biomarkers for the diagnosis of Alzheimer’s disease. Olsson and colleagues recommend the use of the CSF biomarkers total tau, phosphorylated tau, amyloid β42, and neurofilament light protein in the diagnosis of Alzheimer’s disease and mild cognitive impairment.1 By contrast, Cochrane reviews2,3 based on similar studies have concluded that there is not sufficient evidence to support the routine use of CSF amyloid β or PET in the diagnosis of mild cognitive impairment due to Alzheimer’s disease and other dementias. This discrepancy in interpreting the evidence base for diagnostic technologies can only serve to create variations in practice and confusion for both doctors and their patients. Although the use of biomarkers as diagnostic criteria for Alzheimer’s disease might be valuable in research, concerns have been raised about the premature use of such criteria beyond the research setting.4 There is no doubt that many of the changes in CSF and blood biomarkers are associated with Alzheimer’s disease and mild cognitive impairment, but such associations by themselves might be inadequate to demonstrate clinical usefulness. Establishment of clinical usefulness requires the use of thresholds for positive and negative results and standard accuracy metrics such as sensitivity, specificity, and likelihood ratios.5 Alternative measures such as correlations or ratios of blood biomarker concentrations in patients and controls might be important to indicate the potential value of a marker in the research phase, but these studies do not provide information about how the test will perform in a typical clinical setting. In clinical practice, patients 1204
requiring further testing will not be those with established Alzheimer’s disease or cognitively healthy controls.5 The use of such widely differing groups in accuracy studies will result in spectrum bias and does not provide evidence of the performance of the test in a true clinical setting. More definitive evidence from studies done in consecutive series of presenting patients (even if this series includes people with a range of diagnoses) is needed before these biomarkers can be recommended for routine clinical use.5 The true value of the use of a diagnostic test in practice should relate to its benefit for the patient, in terms of outcomes such as quality of life. Even if biomarkers are proven to be highly accurate in diagnosing Alzheimer’s disease, the probable patient benefit should still be considered. The value of testing for a diagnosis of mild cognitive impairment, while important for research advances, can be questioned given the uncertain balance between the benefits and harms of a diagnosis while a treatment is not available. This might be particularly important in tests with impressive sensitivity but inadequate specificity, leading to a large number of false positive diagnoses in patients who might never develop dementia. Some researchers, enthusiastic about promising results for diagnostic tests in a common and debilitating disease, might be frustrated by the requirement for more definitive evidence of the accuracy of diagnostic tests before routine use. However, improvements in clinical practice will only come with more consistent interpretation of evidence before recommendations for adopting new technologies are made. LF is an Editor for The Cochrane Dementia and Cognitive Improvement Group. SD was previously employed conducting health technology assessments for the Medical Services Advisory Committee, who make recommendations for public funding of diagnostic technologies through Medicare in Australia. All authors were involved in development of clinical practice guidelines for dementia in Australia.
*Suzanne M Dyer, Leon Flicker, Kate Laver, Craig Whitehead, Robert Cumming [email protected] Department of Rehabilitation, Aged and Extended Care, School of Health Sciences, Flinders University, Adelaide, SA, Australia (SMD, KL, CW); NHMRC Partnership Centre in Cognitive and Functional Decline (SMD, KL, CW) and School of Public Health (RC), The University of Sydney, Sydney, NSW, Australia; and Centre for Medical Research, University of Western Australia, Crawley, WA, Australia (LF) 1
2
3
4
5
Olsson B, Lautner R, Andreasson U, et al. CSF and blood biomarkers for the diagnosis of Alzheimer’s disease: a systematic review and meta-analysis. Lancet Neurol 2016; 15: 673–84. Ritchie C, Smailagic N, Noel-Storr Anna H, et al. Plasma and cerebrospinal fluid amyloid beta for the diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2014; 10: CD008782. Zhang S, Smailagic N, Hyde C, et al. 11C-PIB-PET for the early diagnosis of Alzheimer?s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2014; 23: CD010386. Brayne C. A population perspective on the IWG-2 research diagnostic criteria for Alzheimer’s disease. Lancet Neurol 2014; 13: 532–34. Sackett DL, Haynes RB. The architecture of diagnostic research. BMJ 2002; 324: 539–41.
Authors’ reply We would like to thank Suzanne Dyer and colleagues for their comments about our Article.1 We would also like to clarify that the Cochrane Review2 excluded most of the studies that were included in our metaanalysis, which reported a highly consistent pattern of Alzheimer’s disease-related biomarker changes (increased total tau, phosphorylated tau, and neurofilament light protein, and decreased amyloid β42 concentrations in CSF), that not only manifest in Alzheimer’s disease but also in patients with mild cognitive impairment. These biomarkers are backed by solid evidence regarding the types of pathologies they reflect; they were not derived from random screens (eg, proteomics or multiplex protein panels) and have been validated with both neuropathology and amyloid PET findings. 3 Thus, there is, in our view, no risk that the consistent findings in the scientific
www.thelancet.com/neurology Vol 15 November 2016
Correspondence
literature are spurious to some kind of confounder that would be present in some patients seeking medical advice at memory clinics. However, we agree with Dyer and colleagues that, in the clinical setting, CSF biomarkers have to be interpreted together with data from a full medical assessment of the patient. The results of our meta-analysis show that there are biomarker-negative patients with Alzheimer’s disease and biomarker-positive patients and controls that do not have Alzheimer’s disease. The overlap in pathology between Alzheimer’s disease and other neurodegenerative disorders, and the high proportion of cognitively normal elderly patients with Alzheimer’s disease-like changes (ie, plaques, tangles, or neurodegeneration) preclude CSF biomarkers from achieving a specificity of 100%. The context of the clinical biomarker use we advocate is identical to what is recommended in the International Working Group-2 research diagnostic criteria.4 The performance of CSF biomarkers in a clinical setting has been assessed in several studies.5,6 At present, each laboratory has to establish their own reference limits and cutpoints, and ascertain longitudinal stability and minimise random variation in their measurements, which is (or should be) standard practice and results in high performance over time.7 An external global quality control programme to assist in this process is in place, and is supported by the Alzheimer’s Association.8 Finally, what ultimately determines the true value of a diagnostic test is whether clinicians request it or not. CSF Alzheimer’s disease biomarkers have been used in clinical practice and reimbursed by the health authorities in several countries (eg, Sweden and Germany) for more than 10 years, and the demand from the clinicians for these tests is steadily increasing, despite only symptomatic treatment for Alzheimer’s disease being available at present. We have to prepare for
the advent of disease-modifying drugs. We foresee a high demand from patients (and their relatives) to undergo a clinical assessment to learn if their cognitive problems are due to Alzheimer’s disease pathology and, if so, to start treatment with these drugs. However, since these drugs might have side-effects, and will most likely be expensive, clinicians will be required to determine whether Alzheimer’s disease-like phenotypes in the clinic are likely to be due to the pathology against which the drug candidate is directed. It is our duty as clinicians and clinical researchers to do everything we can to meet this need. HZ and KB reports that they are co-founders of Brain Biomarker Solutions in Gothenburg AB, a GU Venture-based platform company at the University of Gothenburg. KB also reports personal fees from Eli Lilly, IBL International, Roche Diagnostics, Fujirebio Europe, and Novartis. BO has nothing to disclose.
*Bob Olsson, Kaj Blennow, Henrik Zetterberg [email protected] Department of Psychiatry and Neurochemistry, Sahlgrenska Academy at the University of Gothenburg, Mölndal, Sweden 1
2
3
4
5
6
7
8
www.thelancet.com/neurology Vol 15 November 2016
Olsson B, Lautner R, Andreasson U, et al. CSF and blood biomarkers for the diagnosis of Alzheimer’s disease: a systematic review and meta-analysis. Lancet Neurol 2016; 15: 673–84. Ritchie C, Smailagic N, Noel-Storr Anna H, et al. Plasma and cerebrospinal fluid amyloid beta for the diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2014; 10: CD008782. Blennow K, Mattsson N, Scholl M, Hansson O, Zetterberg H. Amyloid biomarkers in Alzheimer’s disease. Trends Pharmacol Sci 2015; 36: 297–309. Dubois B, Feldman HH, Jacova C, et al. Advancing research diagnostic criteria for Alzheimer’s disease: the IWG-2 criteria. Lancet Neurol 2014; 13: 614–29. Bocchetta M, Galluzzi S, Kehoe PG, et al. The use of biomarkers for the etiologic diagnosis of MCI in Europe: an EADC survey. Alzheimers Dement 2015; 11: 195–206. Lehmann S, Dumurgier J, Schraen S et al. A diagnostic scale for Alzheimer’s disease based on cerebrospinal fluid biomarker profiles. Alzheimers Res Ther 2014; 6: 38. Palmqvist S, Zetterberg H, Mattsson N et al. Detailed comparison of amyloid PET and CSF biomarkers for identifying early Alzheimer disease. Neurology 2015; 85: 1240–49. Mattsson N, Andreasson U, Persson S et al. CSF biomarker variability in the Alzheimer’s Association quality control program. Alzheimers Dement 2013; 9: 251–61.
Cognitive status after intracerebral haemorrhage We read with great interest the remarkable Article by Solene Moulin and colleagues;1 the study highlights the importance of close collaboration between neurocognitive departments and stroke units to recognise patients at high risk of developing dementia. Research done in the field of coagulation disorders such as haemophilia or thrombotic thrombocytopenic purpura might be able to contribute to exploring neurocognitive impairments arising as a consequence of spontaneous intracerebral haemorrhages, in both the short and long term. Neurocognitive impairments have been identified in patients with coagulation disorders.2–4 For example, mild cognitive impairment was reported in 73% of 49 patients with haemophilia irrespective of disease severity. Additionally, in this cohort, detection of cerebral microbleeds with MRI was associated with lower cognitive performance (R=–0·32, p<0·05).2 Data obtained from cohorts of patients with haematological disease might offer an opportunity to undertake a longitudinal study as intracerebral haemorrhages can occur in paediatric patients with these diseases, with a number of neurocognitive sequelae. 5 While many questions remain for future research, neurological, psychological, and haematological practices should consider routine collaboration to ensure early detection of altered cognitive status. We declare no competing interests.
*Silvia Riva, Gabriella Pravettoni [email protected] Department of Oncology and Hemato-Oncology, University of Milan, Italy (SR, GP); and Applied Research Division for Cognitive and Psychological Sciences, European Institute of Oncology (IEO), Milan, Italy (SR, GP)
1205
The clinical value of fluid biomarkers for dementia diagnosis We read with interest the Article by Bob Olsson and colleagues1 examining CSF and blood biomarkers for the diagnosis of Alzheimer’s disease. Olsson and colleagues recommend the use of the CSF biomarkers total tau, phosphorylated tau, amyloid β42, and neurofilament light protein in the diagnosis of Alzheimer’s disease and mild cognitive impairment.1 By contrast, Cochrane reviews2,3 based on similar studies have concluded that there is not sufficient evidence to support the routine use of CSF amyloid β or PET in the diagnosis of mild cognitive impairment due to Alzheimer’s disease and other dementias. This discrepancy in interpreting the evidence base for diagnostic technologies can only serve to create variations in practice and confusion for both doctors and their patients. Although the use of biomarkers as diagnostic criteria for Alzheimer’s disease might be valuable in research, concerns have been raised about the premature use of such criteria beyond the research setting.4 There is no doubt that many of the changes in CSF and blood biomarkers are associated with Alzheimer’s disease and mild cognitive impairment, but such associations by themselves might be inadequate to demonstrate clinical usefulness. Establishment of clinical usefulness requires the use of thresholds for positive and negative results and standard accuracy metrics such as sensitivity, specificity, and likelihood ratios.5 Alternative measures such as correlations or ratios of blood biomarker concentrations in patients and controls might be important to indicate the potential value of a marker in the research phase, but these studies do not provide information about how the test will perform in a typical clinical setting. In clinical practice, patients 1204
requiring further testing will not be those with established Alzheimer’s disease or cognitively healthy controls.5 The use of such widely differing groups in accuracy studies will result in spectrum bias and does not provide evidence of the performance of the test in a true clinical setting. More definitive evidence from studies done in consecutive series of presenting patients (even if this series includes people with a range of diagnoses) is needed before these biomarkers can be recommended for routine clinical use.5 The true value of the use of a diagnostic test in practice should relate to its benefit for the patient, in terms of outcomes such as quality of life. Even if biomarkers are proven to be highly accurate in diagnosing Alzheimer’s disease, the probable patient benefit should still be considered. The value of testing for a diagnosis of mild cognitive impairment, while important for research advances, can be questioned given the uncertain balance between the benefits and harms of a diagnosis while a treatment is not available. This might be particularly important in tests with impressive sensitivity but inadequate specificity, leading to a large number of false positive diagnoses in patients who might never develop dementia. Some researchers, enthusiastic about promising results for diagnostic tests in a common and debilitating disease, might be frustrated by the requirement for more definitive evidence of the accuracy of diagnostic tests before routine use. However, improvements in clinical practice will only come with more consistent interpretation of evidence before recommendations for adopting new technologies are made. LF is an Editor for The Cochrane Dementia and Cognitive Improvement Group. SD was previously employed conducting health technology assessments for the Medical Services Advisory Committee, who make recommendations for public funding of diagnostic technologies through Medicare in Australia. All authors were involved in development of clinical practice guidelines for dementia in Australia.
*Suzanne M Dyer, Leon Flicker, Kate Laver, Craig Whitehead, Robert Cumming [email protected] Department of Rehabilitation, Aged and Extended Care, School of Health Sciences, Flinders University, Adelaide, SA, Australia (SMD, KL, CW); NHMRC Partnership Centre in Cognitive and Functional Decline (SMD, KL, CW) and School of Public Health (RC), The University of Sydney, Sydney, NSW, Australia; and Centre for Medical Research, University of Western Australia, Crawley, WA, Australia (LF) 1
2
3
4
5
Olsson B, Lautner R, Andreasson U, et al. CSF and blood biomarkers for the diagnosis of Alzheimer’s disease: a systematic review and meta-analysis. Lancet Neurol 2016; 15: 673–84. Ritchie C, Smailagic N, Noel-Storr Anna H, et al. Plasma and cerebrospinal fluid amyloid beta for the diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2014; 10: CD008782. Zhang S, Smailagic N, Hyde C, et al. 11C-PIB-PET for the early diagnosis of Alzheimer?s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2014; 23: CD010386. Brayne C. A population perspective on the IWG-2 research diagnostic criteria for Alzheimer’s disease. Lancet Neurol 2014; 13: 532–34. Sackett DL, Haynes RB. The architecture of diagnostic research. BMJ 2002; 324: 539–41.
Authors’ reply We would like to thank Suzanne Dyer and colleagues for their comments about our Article.1 We would also like to clarify that the Cochrane Review2 excluded most of the studies that were included in our metaanalysis, which reported a highly consistent pattern of Alzheimer’s disease-related biomarker changes (increased total tau, phosphorylated tau, and neurofilament light protein, and decreased amyloid β42 concentrations in CSF), that not only manifest in Alzheimer’s disease but also in patients with mild cognitive impairment. These biomarkers are backed by solid evidence regarding the types of pathologies they reflect; they were not derived from random screens (eg, proteomics or multiplex protein panels) and have been validated with both neuropathology and amyloid PET findings. 3 Thus, there is, in our view, no risk that the consistent findings in the scientific
www.thelancet.com/neurology Vol 15 November 2016
Correspondence
literature are spurious to some kind of confounder that would be present in some patients seeking medical advice at memory clinics. However, we agree with Dyer and colleagues that, in the clinical setting, CSF biomarkers have to be interpreted together with data from a full medical assessment of the patient. The results of our meta-analysis show that there are biomarker-negative patients with Alzheimer’s disease and biomarker-positive patients and controls that do not have Alzheimer’s disease. The overlap in pathology between Alzheimer’s disease and other neurodegenerative disorders, and the high proportion of cognitively normal elderly patients with Alzheimer’s disease-like changes (ie, plaques, tangles, or neurodegeneration) preclude CSF biomarkers from achieving a specificity of 100%. The context of the clinical biomarker use we advocate is identical to what is recommended in the International Working Group-2 research diagnostic criteria.4 The performance of CSF biomarkers in a clinical setting has been assessed in several studies.5,6 At present, each laboratory has to establish their own reference limits and cutpoints, and ascertain longitudinal stability and minimise random variation in their measurements, which is (or should be) standard practice and results in high performance over time.7 An external global quality control programme to assist in this process is in place, and is supported by the Alzheimer’s Association.8 Finally, what ultimately determines the true value of a diagnostic test is whether clinicians request it or not. CSF Alzheimer’s disease biomarkers have been used in clinical practice and reimbursed by the health authorities in several countries (eg, Sweden and Germany) for more than 10 years, and the demand from the clinicians for these tests is steadily increasing, despite only symptomatic treatment for Alzheimer’s disease being available at present. We have to prepare for
the advent of disease-modifying drugs. We foresee a high demand from patients (and their relatives) to undergo a clinical assessment to learn if their cognitive problems are due to Alzheimer’s disease pathology and, if so, to start treatment with these drugs. However, since these drugs might have side-effects, and will most likely be expensive, clinicians will be required to determine whether Alzheimer’s disease-like phenotypes in the clinic are likely to be due to the pathology against which the drug candidate is directed. It is our duty as clinicians and clinical researchers to do everything we can to meet this need. HZ and KB reports that they are co-founders of Brain Biomarker Solutions in Gothenburg AB, a GU Venture-based platform company at the University of Gothenburg. KB also reports personal fees from Eli Lilly, IBL International, Roche Diagnostics, Fujirebio Europe, and Novartis. BO has nothing to disclose.
*Bob Olsson, Kaj Blennow, Henrik Zetterberg [email protected] Department of Psychiatry and Neurochemistry, Sahlgrenska Academy at the University of Gothenburg, Mölndal, Sweden 1
2
3
4
5
6
7
8
www.thelancet.com/neurology Vol 15 November 2016
Olsson B, Lautner R, Andreasson U, et al. CSF and blood biomarkers for the diagnosis of Alzheimer’s disease: a systematic review and meta-analysis. Lancet Neurol 2016; 15: 673–84. Ritchie C, Smailagic N, Noel-Storr Anna H, et al. Plasma and cerebrospinal fluid amyloid beta for the diagnosis of Alzheimer’s disease dementia and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev 2014; 10: CD008782. Blennow K, Mattsson N, Scholl M, Hansson O, Zetterberg H. Amyloid biomarkers in Alzheimer’s disease. Trends Pharmacol Sci 2015; 36: 297–309. Dubois B, Feldman HH, Jacova C, et al. Advancing research diagnostic criteria for Alzheimer’s disease: the IWG-2 criteria. Lancet Neurol 2014; 13: 614–29. Bocchetta M, Galluzzi S, Kehoe PG, et al. The use of biomarkers for the etiologic diagnosis of MCI in Europe: an EADC survey. Alzheimers Dement 2015; 11: 195–206. Lehmann S, Dumurgier J, Schraen S et al. A diagnostic scale for Alzheimer’s disease based on cerebrospinal fluid biomarker profiles. Alzheimers Res Ther 2014; 6: 38. Palmqvist S, Zetterberg H, Mattsson N et al. Detailed comparison of amyloid PET and CSF biomarkers for identifying early Alzheimer disease. Neurology 2015; 85: 1240–49. Mattsson N, Andreasson U, Persson S et al. CSF biomarker variability in the Alzheimer’s Association quality control program. Alzheimers Dement 2013; 9: 251–61.
Cognitive status after intracerebral haemorrhage We read with great interest the remarkable Article by Solene Moulin and colleagues;1 the study highlights the importance of close collaboration between neurocognitive departments and stroke units to recognise patients at high risk of developing dementia. Research done in the field of coagulation disorders such as haemophilia or thrombotic thrombocytopenic purpura might be able to contribute to exploring neurocognitive impairments arising as a consequence of spontaneous intracerebral haemorrhages, in both the short and long term. Neurocognitive impairments have been identified in patients with coagulation disorders.2–4 For example, mild cognitive impairment was reported in 73% of 49 patients with haemophilia irrespective of disease severity. Additionally, in this cohort, detection of cerebral microbleeds with MRI was associated with lower cognitive performance (R=–0·32, p<0·05).2 Data obtained from cohorts of patients with haematological disease might offer an opportunity to undertake a longitudinal study as intracerebral haemorrhages can occur in paediatric patients with these diseases, with a number of neurocognitive sequelae. 5 While many questions remain for future research, neurological, psychological, and haematological practices should consider routine collaboration to ensure early detection of altered cognitive status. We declare no competing interests.
*Silvia Riva, Gabriella Pravettoni [email protected] Department of Oncology and Hemato-Oncology, University of Milan, Italy (SR, GP); and Applied Research Division for Cognitive and Psychological Sciences, European Institute of Oncology (IEO), Milan, Italy (SR, GP)
1205