By Louise M. Klebanoff, MD

Associate Professor of Neurology, Weill Cornell Medical College

Dr. Klebanoff reports no financial relationships relevant to this field of study.

SYNOPSIS: In patients with clinically diagnosed dementia, the CSF biomarker profile of low CSF amyloid-ß1-42, high total tau, and high phosphorylated tau was seen in the majority of patients with clinically diagnosed Alzheimer’s disease. Substantial proportions of patients with non-Alzheimer’s dementia were also found to have the Alzheimer’s disease pathological profile. The value of CSF biomarker measurements in clinical practice is uncertain.

SOURCE: Skillback T, et al. Cerebrospinal fluid tau and amyloid-ß1-42 in patients with dementia. Brain 2015;138:2716-2731.

As the global population ages, Alzheimer’s disease, the most common cause of dementia affecting more than 20 million people worldwide, is of increasingly critical importance to world health. The pathological signs of the disease, including neuronal and synaptic/axonal degeneration with resultant brain atrophy, the accumulation of amyloid plaques, and intranuclear neurofibrillary tangles of phosphorylated tau, have been recognized for decades. In more recent years, cerebrospinal fluid (CSF) biomarkers for Alzheimer’s disease have been developed. These biomarkers include low levels of CSF amyloid-ß1-42, which correlates with greater plaque load; high levels of tau, which correlate with greater intensity of neuronal degeneration; and high levels of phosphorylated tau, which correlate with neurofibrillary tangles. In the context of a clinical presentation consistent with Alzheimer’s disease, a CSF profile combining low levels of amyloid-ß1-42, high levels of total tau, and high levels of phosphorylated tau support the diagnosis with 80-93% sensitivity and 82-90% specificity when compared to controls. Other common dementing disorders, however, can overlap with Alzheimer’s disease in terms of clinical presentation and CSF biomarkers. In addition, mixed syndromes are not uncommon.

Skillback et al examined CSF amyloid-ß1-42, total tau, and phosphorylated tau in patients with clinically diagnosed dementia in an effort to identify concomitant Alzheimer’s disease pathology in patients without a clinical diagnosis of Alzheimer’s disease. They also performed a biomarker-driven cluster analysis to test for the natural classification of study patients.

The investigators combined two sources of information for the study. They accessed a complete set of archived data of CSF biomarkers (amyloid-ß1-42, total tau, and phosphorylated tau) made in clinical practice at the Molndal site of Sahlgrenska University Hospital in Sweden that were collected between January 2004 and June 2012; this laboratory handles CSF biomarker measurements for all of Sweden. They also had clinical data from SveDen, the Swedish Dementia Registry, which covers 95% of all memory clinics and 70% of all primary care units in Sweden. When a patient is diagnosed with dementia in clinical practice, the clinical diagnosis, date of diagnosis, and Mini-Mental State Examination score is entered into the registry. The diagnosis is made on clinical grounds and ß1-42 is entered as one of the following nine categories: early-onset Alzheimer’s disease (onset < 65 years of age), late-onset Alzheimer’s disease (onset > 65 years of age), fronto-temporal dementia, dementia with Lewy bodies, Parkinson’s disease dementia, vascular dementia, mixed Alzheimer’s disease and vascular dementia, dementia not otherwise specified, or other dementias. More than 70% of the patients in the study were diagnosed by dementia specialists.

The investigators analyzed 5676 patients with complete CSF biomarker measurements obtained within 3 years of diagnosis. The combination of low levels of amyloid-ß1-42 with a pathological level of either total tau or phosphorylated tau was considered an Alzheimer’s disease-like pathological profile. The investigators found that amyloid-ß1-42 was lowest in the early-onset and late-onset Alzheimer’s groups and in the mixed Alzheimer’s and vascular dementia group and was highest in the frontotemporal dementia group. Total tau was highest in the early-onset Alzheimer’s group, followed by the late-onset group and then the mixed Alzheimer’s disease and vascular dementia group; patients with Lewy body dementia and Parkinson’s disease dementia had the lowest levels. Levels of phosphorylated tau and the amyloid-ß1-42:phosphorylated tau ratio showed a similar picture, with the highest levels in early onset Alzheimer’s followed by late onset Alzheimer’s and mixed Alzheimer’s disease and vascular dementia. Patients with frontotemporal dementia had a CSF biomarker profile most distinct from patients with Alzheimer’s disease.

When using established cutoffs for the measured biomarkers, patients were classified according to the presence or absence of Alzheimer’s disease pathology. The biomarker profile was found significantly more in the patients with clinically diagnosed Alzheimer’s disease, either late-onset, early-onset, or mixed Alzheimer’s disease and vascular dementia, with more than 80% of these patients showing low levels of amyloid-ß1-42, high total tau, and high phosphorylated tau levels. However, a substantial proportion of patients without a clinical diagnosis of Alzheimer’s disease also had abnormal biomarkers consistent with Alzheimer’s disease pathology. More than 50% of patients clinically diagnosed with vascular dementia, dementia with Lewy bodies, and Parkinson’s disease dementia had low levels of amyloid-ß1-42. More than 40% of patients diagnosed with vascular dementia and frontotemporal dementia had abnormal total tau measurements, and more than 20% of patients with vascular dementia and dementia with Lewy bodies had abnormal phosphorylated tau measurements. Therefore, CSF biomarker evidence of Alzheimer’s ß1-42 disease-like pathology was found in a considerable proportion of patients without a clinical diagnosis of Alzheimer’s disease.

The investigators applied cluster analysis using log (total tau) and log (amyloid-ß1-42:phosphorylated tau) to identify a division separating the subjects into two groups. The first cluster (n = 2851) contained more than 90% of the patients diagnosed with vascular dementia, frontotemporal dementia, Parkinson’s disease dementia, and dementia with Lewy bodies. The second cluster (n = 2825) contained the majority of the patients with a clinical diagnosis of early-onset Alzheimer’s disease (75%) and late-onset Alzheimer’s disease (73%).


As the world’s population ages and dementia becomes more prevalent, the ability to accurately diagnose dementia gains increasing importance. Once treatment interventions have been developed to alter the course of neurodegenerative conditions with associated dementia, the ability to diagnose these conditions in the preclinical stage will be essential. However, at this time, such treatment interventions are lacking.

The study by Skillback et al shows that CSF biomarkers can be obtained in a large population of patients with dementia. The study compared biomarker profiles obtained in clinical practice with patients who were clinically diagnosed to have one of nine dementing illnesses. Although certain profiles were expected, such as low amyloid-ß1-42, higher total tau, and high phosphorylated tau seen in a higher proportion of patients with clinically diagnosed Alzheimer’s disease, there were still a substantial proportion of patients clinically diagnosed with Alzheimer’s disease who did not have the expected biomarker profile. In addition, the Alzheimer’s disease biomarker profile was seen in a substantial proportion of patients who were diagnosed with dementia not attributed to Alzheimer’s disease. Since the biomarkers were obtained as part of clinical care, it is possible that the results influenced the clinical diagnosis, circular logic that the investigators readily acknowledged. Neuro-imaging and neuropathology were not included as part of the study data, also weakening the study.

At this time, CSF biomarkers can support a clinical diagnosis of Alzheimer’s disease, but cannot definitively confirm the diagnosis or exclude other neurodegenerative diseases as the cause of a specific patient’s dementia. CSF biomarkers may weakly correlate with disease severity, but do not provide prognostic information. It is not clear if CSF markers will predict the onset of Alzheimer’s disease and other dementing illnesses in the preclinical stages; this is obviously an area for further research. However, without providing definitive diagnosis of the type of dementia, the lack of prognostic information and in the absence of proven interventions for disease prevention, should CSF biomarkers be able to diagnose dementia in the preclinical stages, CSF biomarker measurement adds little to the clinical diagnosis and management of patients with dementia. While further research on CSF biomarkers in dementia is needed, currently, obtaining CSF biomarkers should not be the part of the standard of clinical care for patients presenting with dementia.