30 March 2011. Dementia with Lewy bodies (DLB) sits at the interface of several disorders. Sharing features with Parkinson’s, PD with dementia (PDD), and Alzheimer’s disease, DLB challenges the diagnostician mightily. The weapon needed to meet this challenge is an arsenal of good biomarkers, scientists agree, but until quite recently, the stockpile was limited. Part of the problem is that DLB was first recognized only about 20 years ago, and so research in this field has lagged behind that for AD and PD. Two years ago, a workshop in Kassel, Germany, held prior to the 9th International Conference on Alzheimer’s and Parkinson’s Diseases, described some of the early efforts to find DLB biomarkers (see ARF related news story). Scientists provided an update on progress in this area at a symposium at the 10th AD/PD Conference, held 9-13 March 2011 in Barcelona, Spain. They focused particularly on markers that will distinguish between DLB and AD. A repeated theme was that no one biomarker will do the job.
“If you rely on a single biomarker, you’re going to have a lot of people who are misclassified,” said James Galvin at New York University Langone Medical Center, New York City. This is because these dementias overlap in both pathology and clinical symptoms. In particular, many people have features of both AD and DLB. “Combining biomarkers allows you to get fairly strong classification of whether people have pure or mixed forms of dementia,” Galvin told ARF. At the symposium, he showed preliminary data demonstrating how this strategy might work to divide people with AD, DLB, and mixed pathology. The ability to define these groups “could have significant therapeutic implications as we develop more specific disease-modifying medicines,” Galvin said.
Speakers covered a range of biomarker types, including brain imaging, body fluids, and cognitive and neuropsychological testing. “Within each biomarker domain, I think we’re identifying what the best possible markers will be,” David Salmon at the University of California in San Diego told ARF. “We have biomarkers now that are more reliable than we had two to four years ago.” In the future, combining these markers synergistically will improve diagnostic confidence, Salmon predicted.
In a conversation with ARF, Galvin noted another advantage of the boom in biomarkers: It will allow scientists to do more hypothesis-driven Lewy body research, similar to how biomarkers are now driving the AD field.
A Spectrum of Dementias
DLB is one of the most common—and some say one of the worst—progressive dementias. It more often afflicts men than women. The disorder is currently diagnosed using a combination of cognitive tests, clinical features, interviews with family members, and positron emission tomography (PET) or single photon emission computed tomography (SPECT) to detect low dopamine transporter levels, said Ian McKeith at Newcastle University, Newcastle upon Tyne, U.K. However, DLB is underdiagnosed. Only about 4 to 5 percent of dementia cases are diagnosed as DLB, McKeith said, whereas 15 percent prove at autopsy to have been DLB. Another way to put this is that about three-quarters of people with DLB initially receive the wrong diagnosis. The disease is particularly tricky to diagnose in the prodromal phase, McKeith said, because early DLB has many non-specific symptoms seen in several neurodegenerative diseases and often shows up at the neurologist’s doorstep with atypical presentations. Better biomarkers are needed to pin down the disease earlier, he said.
DLB is a close cousin of PD and PDD. Its predominant pathology is α-synuclein, which accumulates in Lewy bodies, but amyloid pathology is very common as well. As in PD, dopaminergic neurons in the substantia nigra wither and die. The difference is that DLB brains at autopsy have lost fewer neurons than those with PD or PDD, said Dennis Dickson at the Mayo Clinic in Jacksonville, Florida. Also, in DLB the α-synuclein load in the striatum decreases as the disease progresses, unlike in PD and PDD, where it increases, Dickson said. The main clinical features of DLB include visual hallucinations, movement problems, and fluctuations in alertness, attention, and cognitive function. Sleep disturbances and depression are also common, Dickson said. Many of these features occur in PD as well, although in DLB, motor symptoms are usually milder and hallucinations are more prominent. Like PD, DLB is also marked by loss of the sense of smell, reduced sympathetic innervation of the heart, and low dopamine levels in the brain, Dickson said. The order in which dementia and movement symptoms appear further distinguishes DLB and PDD. When dementia develops several years after Parkinson’s symptoms, it is PDD; when dementia precedes parkinsonism, or arrives within the same year, it is DLB. This is because in Parkinson’s disease, pathology tends to progress from the bottom of the brain up, Dickson said, but in DLB the progression is more top-down, affecting the cortex first.
Distinguishing DLB from AD, especially at early stages, is difficult. People with DLB often suffer from memory problems and cognitive impairments, frequently confusing clinicians. Sometimes the diagnosis depends on whether the patient happens to first see a clinician who specializes in movement disorder or in dementia. McKeith noted that atypical DLB cases, where pathology predominates in the brainstem, pose the greatest puzzle for diagnosis because the clinical presentation is most similar to AD. Amyloid imaging is of little help, as amyloid plaques appear in more than three-quarters of DLB brains, said Galvin. Likewise, about one-third of AD brains contain Lewy bodies, Galvin said, further conflating the two. DLB brains also display neurofibrillary tangles of tau protein, although they are less common than in AD, Dickson said, roughly the equivalent of Braak stage IV. People with DLB lose both dopaminergic neurons as in PD and cholinergic neurons as in AD, Dickson said.
What to do when the clinicopathological relationships are such a mess? Scientists are looking at numerous possible biomarkers for distinguishing these disorders. One is cerebrospinal fluid (CSF) α-synuclein, currently the best-validated CSF marker for DLB. Brit Mollenhauer at the Paracelsus-Elena-Klinik, Kassel, Germany, described recently published findings by her team showing that α-synuclein levels are lower in the CSF in DLB and PD patients compared to controls and people with AD (see ARF related news story on Mollenhauer et al., 2011). In a validation cohort of people with parkinsonism, the combination of CSF-tau and α-synuclein measurements allowed researchers to distinguish between synuclein diseases and other disorders with great accuracy, Mollenhauer said. This paper puts to rest prior debate about the strength of CSF findings for α-synuclein.
Other CSF molecules are also being investigated. A poster presented by Malin Wennström and colleagues at Lund University, Malmö, Sweden, reported that people with DLB have lower CSF levels of orexin/hypocretin, a hormone involved in the regulation of hunger and sleep, than healthy people. Orexin-producing neurons in the hypothalamus have been shown to die off in PD, and people with DLB often are excessively sleepy during the day. The authors found that orexin levels correlated with CSF α-synuclein levels, but not with age or cognitive function. CSF orexin has also been studied in other neurologic conditions, including narcolepsy and post-traumatic stress disorder, hence, is most likely not specific to DLB.
One promising way to differentiate DLB from AD, said Salmon, is through visuospatial tasks, as impairments in this domain are one of the signature features of DLB. Salmon described a retrospective study that compared neuropsychological test scores of people confirmed at autopsy to have had either DLB or AD. One of the most robust findings was that people with DLB do worse on visuospatial tasks (see Tiraboschi et al., 2006). In a second retrospective study, the researchers found that poor clock-drawing and Block Design performance (both are visuospatial tasks) correlated with faster cognitive decline in DLB patients over a two-year period, but had no predictive value in AD patients (see Hamilton et al., 2008). In addition, more than 60 percent of people with severe visuospatial defects at baseline later suffered from visual hallucinations, compared with about 10 percent of those with mild visuospatial problems, Salmon said.
In light of these findings, Salmon’s team investigated whether they could use a visuospatial task to tell apart people with DLB from those with AD. Visuospatial defects affect the ability to detect both motion and brightness. The scientists used the motion coherence paradigm, in which people look at a screen full of dots in motion. If a large enough fraction of the dots are moving in the same direction, normal people can see the motion. People with AD are as good at detecting the direction of the moving dots as healthy people, but people with DLB just cannot do it, Salmon said. Additionally, normal people are better at this task if the dots moving in the same direction are either more or less bright than the randomly moving dots. This is an example of sensory integration, Salmon said. When the researchers added this luminance clue to the test, both controls and AD patients got better at the task, while people with DLB remained the same. In one small sample, this task provided 100 percent discrimination between people with AD and DLB, Salmon said.
Galvin noted that this test is currently only a research paradigm, but in the future it might be quite useful clinically for differentiating early-stage dementias. Researchers will have to translate the test into a software package that clinicians could use, Galvin said.
Imaging markers are another key tool for distinguishing DLB and AD, said John O’Brien at Newcastle University, U.K. So far, the best approach is to use SPECT imaging with the 123I-FP-CIT tracer to measure dopamine transporters in the striatum, O’Brien said. Dopamine levels are normal in people with AD but reduced in DLB. This is the only “category A” biomarker for differentiating these disorders currently recognized by the European Federation of Neurological Societies. It can separate DLB from other dementias with 90 percent specificity and 78 percent sensitivity, O’Brien said (see McKeith et al., 2007). In addition, in diagnostically uncertain cases, an abnormal dopamine scan was highly predictive of DLB (see O’Brien et al., 2009).
Structural MRI also shows promise for distinguishing AD and DLB, O’Brien said. People with AD lose more volume in the hippocampus and medial temporal lobe than do people with either DLB or vascular cognitive impairment. In a study of about 50 people confirmed at autopsy to have had AD, DLB, or vascular cognitive impairment, medial temporal lobe volume diagnosed AD with 90 percent specificity and sensitivity (see Burton et al., 2009). However, another study failed to replicate this finding (see ARF related news story). Galvin pointed out that because people with DLB also have some shrinkage in these brain regions, this measure may be more useful for research than as a diagnostic test.
Another potential approach is to home in on hippocampal subfields using high field strength (3 Tesla) MRI, O’Brien said. Many hospitals now use 3 T magnets, making this feasible for many clinicians. Subiculum and CA1 regions are smaller in AD than in DLB, providing about 80 percent discrimination between the disorders, O’Brien said.
The Combinatorial Approach
The speakers emphasized that no one biomarker will adequately distinguish these dementias, and the best approach will be to use several of them together. This is no surprise because all these tests and markers partly overlap, as do the symptoms and the pathology, after all. Galvin discussed data from several small proof-of-concept studies showing how combos might work. In one study of 45 people, Galvin’s team combined PET imaging with Pittsburgh Compound B, which quantifies amyloid load, along with both cognitive and clinical ratings. For the cognitive assessment, the scientists used composite scores of episodic, semantic, and working memory, as well as visuospatial abilities and global cognitive scores. For the clinical score, Galvin’s team combined four measures—the Unified Parkinson’s Disease Rating Scale, the Mayo Sleep Questionnaire, Mayo Fluctuation Questionnaire, and the Neuropsychiatric Inventory—to create a Lewy Body Risk Score. This biomarker combination divided people into four groups: those predicted to have AD pathology, DLB pathology, mixed pathology, and no pathology. Agreement with the clinical diagnoses was almost 90 percent, Galvin said. All participants are still living, so no pathological confirmations could be made in this study.
In another study of 33 people, Galvin’s team combined two fluid biomarkers, Aβ42 and α-synuclein, and again saw four distinct groups. Four people in this study later died, and autopsy results showed that the biomarkers correctly categorized two people with AD and one with DLB. The biomarkers improved on the clinical diagnosis for one man who was listed as “possible DLB.” CSF results showed normal levels of Aβ and synuclein but elevated tau, and predicted that he had neither AD nor DLB, which turned out to be the case.
Galvin also discussed functional MRI studies that found abnormal connectivity in DLB brains. In a study of 85 people who had DLB, AD, or no dementia, people with DLB had a reversal of the normal connectivity between the default-mode network and other brain regions (Neurology, 2011, in press). The default-mode network reflects the resting activity of the brain and is also perturbed in AD (see ARF related news story). In a separate study, Galvin’s team investigated the basis of the iconic cognitive fluctuations of DLB. “This is an incredibly difficult symptom to get your hands around,” Galvin notes. They picked people who had early dementia with cognitive fluctuations but no other symptoms of DLB and compared the functional connectivity of their brains to that of people who had dementia without fluctuations. The fluctuators had an imbalance between their default and attention networks, Galvin said, suggesting that the “switch” that helps people change from states of daydreaming to attention and back might be faulty in these people (paper in review). “That might lead to people having some odd waxing and waning in their cognitive status,” Galvin speculated to ARF.—Madolyn Bowman Rogers.