23 March 2005. With heart disease, high cholesterol levels prompt the doctor to prescribe statins and urge other preventive measures. A suspicious PSA screens warns the doctor that an aging prostate is misbehaving. Why is such a laboratory test not available for AD?
It’s not for lack of trying, say scientists. The search for biomarkers has been difficult. Early candidates have fallen by the wayside for lack of robustness, sensitivity, or specificity, even as the advent of new experimental therapies has made the need for a reliable test more pressing. Researchers believe that new therapies will show the most promise for the earliest stages of AD, which may precede overt symptoms by years. For this reason, that elusive, telltale giveaway that the underlying disease process is gnawing away at a person’s brain has become a coveted prize for both practicing clinicians and clinical trial researchers. The 7th AD/PD 2005 conference held March 9-13 in Sorrento offered the latest news on some of the ongoing efforts to finger such a marker for AD, and budding efforts in Parkinson disease (PD). Here are selected highlights.
John Growdon of Massachusetts General Hospital in Boston said that all too often, molecules that come recommended by cellular or epidemiological science sadly disappoint when tested for their potential as predictive biomarkers. For example, ApoE is a genetic risk factor that makes for a poor biomarker. The latest case in point is homocysteine, which boasts a host of data implicating it in age-related neurodegeneration in general, and in Parkinson disease in particular (see Alzforum live discussion). However, a human study at MGH showed that measuring its levels does not enable a clinician to distinguish AD from Mild Cognitive Impairment (MCI), PD, or normal controls. Levels did correlate with age, levodopa use, and cognitive impairment in PD, and they were low in people who took multivitamins. This further strengthens a growing literature suggesting that dietary folate supplements may be a helpful component of PD treatment and prevention, but it also means that homocysteine will not become a biomarker, Growdon said.
In Alzheimer's, the seemingly obvious approach of turning Aβ levels into a biomarker has been anything but straightforward. Doctors would most like to have a blood test, as patients more readily roll up a sleeve than curl up sideways for a spinal tap. But no one to date has found Aβ differences in serum that robustly predict the patient’s condition. Growdon added another downer to this hope by reporting that in a plasma Aβ40/42 study conducted at MGH, mean Aβ40/42 levels did not differ among MCI, AD, or control patients. Plasma Aβ levels did rise with age, but were not related to AD history, severity, or acetylcholinesterase medication. “For sporadic AD, plasma Aβ itself is not likely to be a useful biomarker,” Growdon said.
Even so, blood-based detection of amyloid-related markers is far from dead, at least when approached from a different angle. For example, researchers led by Andre Delacourte at INSERM in Lille, France, are trying to link to AD changes in certain C-terminal fragments of APP, as well as what they believe are pathological fragments of Aβ found in lymphocytes and platelets.
It’s All in the Spine
How about cerebrospinal fluid, then? Generally speaking, the invasive nature of a spinal tap has made American clinicians more reluctant to perform it than is the case in Europe, where CSF biomarker research has long been actively pursued. Kaj Blennow presented a summary of such work at the University of Goteborg in Sweden. Blennow noted that more than 45 biomarker studies have documented a large increase of CSF total tau (T-tau) in AD. However, the sensitivity (how many cases does it pick out?) and especially the specificity (how well does it distinguish between similar conditions?) of this measure limits its use as a diagnostic marker. CSF Aβ42 levels decrease with AD, but its specificity, too, is insufficient.
By contrast, phosphorylated tau (P-tau) discriminates better between related disorders and will supersede total tau as a marker, Blennow said. Helping put to rest a debate about which of tau’s phosphorylation epitopes are best, Blennow reported that in a recent study, three different tau methods yielded the same results when compared in material from the same patients.
The ability of the combination of T-tau, P-tau, and Aβ42 to predict who among a group of 52 MCI cases would progress to full-blown AD in two years reached a respectable 94 percent, Blennow reported. Furthermore, a study asking whether known CSF markers can predict who among a group of 54 healthy population-based people would develop AD found that both T-tau and Aβ42 were significantly changed as early as eight years prior to diagnosis; a community-based study yielded the same result. In a separate presentation, Mony de Leon, from New York University, reported similar results from his MCI study, concluding that the increase in CSF P-tau and decrease in Aβ42 can sharpen the diagnosis at early stages.
The main limitation of these markers is that they differentiate AD poorly from other forms of dementia, Blennow said.
Another important part of putting a putative biomarker through its paces involves testing whether it can identify the biochemical effect of a drug. Blennow reported that in AD patients who took the known acetylcholinesterase inhibitors donepezil or galantamine for 6 to 12 months, the researchers detected a marked increase in CSF acetylcholinesterase activity with donepezil, and a smaller increase with galantamine. Neither tau nor Aβ levels changed during this drug treatment, confirming again that these drugs improve symptoms but do not change the disease process.
What about experimental treatments? CSF tau and Aβ levels did not budge, either, after treatment with clioquinol, a drug that disrupts metal-Aβ complexes (see Alzforum live discussion ). Nor did they change after up to 12 months of treatment with simvastatin or atorvastatin, Blennow said. This latter finding was a bit disheartening, because a small previous study had been able to detect a dent in CSF levels of Aβ40 in people with mild AD. Yet even when Blennow’s group divided the subjects in his small study into mild and moderate cases, CSF Aβ and tau stayed flat. “We could not reproduce that earlier trend,” Blennow said (see Simons et al., 2002, Sjogren et al., 2003).
Faced with these stubbornly flat levels of CSF Aβ and tau after all these different treatments, Blennow asked if anything could nudge them up or down. Was the problem in the method? Apparently not, because in 15 people who sustained a head trauma, Aβ rose dramatically after four days and fell down again after 10 days. “So we can detect these changes but have not yet found a treatment that affects Aβ CSF levels,” Blennow concluded.
Diagnosis by Committee: Biomarker Panels
Surely, there must be other biomarkers that perhaps would nail a diagnosis when combined with P-tau or Aβ in a larger panel? One such candidate might be found around oxidative enzyme reactions in the CSF, reported Domenico Pratico of the University of Pennsylvania in Philadelphia. Practico introduced 12/15 lipoxygenase as a member of an enzyme family that inserts molecular oxygen into polyunsaturated fatty acids and therefore represents an enzymatic source of oxidative products, rather than the free radicals commonly blamed for oxidative stress. After showing that relevant, amyloid-bearing brain regions of postmortem AD tissue had an increase in protein level and activity of this enzyme (Pratico et al., 2004), the researchers began exploring its potential as a future biomarker in living people with AD. In a study of 15 people with AD, 10 with MCI, and 20 controls, CSF levels of a biochemical 12/15 LOX readout were lowest in controls, higher in MCI cases, and yet higher in AD. This new data correlated with more established measures of oxidation, such as isoprostanes, and also with the expected Aβ42 decrease and tau increase in CSF, Pratico said (for review, see Pratico, 2005).
Blennow described his group’s attempt at finding new members for a panel of biomarkers as focusing, in part, on a multiplex immunoassay called Luminex xMAP. Researchers coat microspheres with antibodies to several different putative biomarkers, and then use laser excitation and fluorescence to quantify the signal. Testing the method, Blennow’s group covered microspheres with antibodies to tau, Aβ42, and P-tau, and found that results in a first human study resembled ELISA in sensitivity and specificity. His group is now using the technique to explore marker cocktails including N-terminal Aβ fragments, Aβ oligomers, synaptic proteins, cytokines, growth factors, and other proteins.
Several groups are performing CSF proteomics using surface-enhanced laser desorption ionization time-of-flight (Seldi-Tof) mass spectrometry to look for marker patterns. Blennow noted that a collaboration of Swedish, Danish, and Finnish groups has identified some 30 components in peaks of increased and decreased proteins among 98 early AD patients, 31 FTD patients, and 49 controls. The component proteins include transthyretin, cystatin C (which has been implicated in AD since the 1980s, most recently by Cathcart et al., 2005), secretogranin, chromogranin, neuroendocrine protein, β 2 microglobulin, various oxidative stress proteins, and others. Individually, none of these predicts AD better than Aβ or P-tau, but in the right combination they may well, Blennow said.
Gene expression microarrays and proteomics also serve to search for a genetic fingerprint of PD. Growdon and Peter Riederer at the University of Wuerzburg, Germany, both summarized such studies. Growdon mentioned initial data of an ongoing blood-based study of 100 idiopathic PD patients, which has to date identified genes of cellular quality control pathways, mitochondrial genes, and others. Likewise, Riederer’s group has begun a proteomics study with PD patients who donate a blood sample every six months for three years.
In addition, Riederer summarized a microarray study performed with Moussa Youdim at Technion in Haifa, Israel, in which the scientists used substantia nigra tissue samples from sporadic PD patients to establish two groups of about 70 genes whose expression went either up or down in that brain region. Standing out from among the genes whose expression dropped was SKP1A, which is part of an E3 ubiquitin ligase complex in substantia nigra and could help explain proteasome dysfunction in PD. Another gene whose expression went down (and which also comes with a biological rationale), is aldehyde dehydrogenase. It functions in dopamine metabolism. These and other genes are being investigated in blood samples of PD patients and healthy people to see whether any of them could pass muster as future biomarkers, said Riederer (see also Grunblatt et al., 2004).—Gabrielle Strobel.
Q&A with Kaj Blennow
Q: I would expect CSF AChE to decrease after treatment with AChE inhibitors?
A: This increase is a mystery. When we performed our study, our hypothesis was to find a decrease, but we found the reverse. We do not have a good explanation, but this finding has been replicated by other groups (Parnetti et al., 2002). The best guess today is that it is a compensatory upregulation.
Q: With such unexpected results, what ‘s the upshot of your CSF drug studies?
A: The main conclusion from our study is that using CSF,
1) it is possible to identify and monitor a biochemical effect of a drug;
2) it is possible to identify differences between drugs with different modes of action (donepezil vs. galantamine);
3) the changes in CSF are dose-dependent;
4) the changes correlate with the clinical effect of the drug.
Q: What’s the future of CSF analysis?
A: We hope that CSF analyses can be of value to identify and monitor the biochemical effect of new types of treatment, such as BACE1 inhibitors or Aβ vaccination regimes. We also think that there may be two types of CSF biomarkers for BACE1 inhibitors and Aβ vaccination:
1) "Primary" or "Specific" biomarkers based on the mode of action, i.e., Aβ42, Aβ40, α-sAPP and β-sAPP to directly monitor Aβ metabolism, but also
2) "Secondary" or "downstream" CSF biomarkers, e.g., CSF T-tau (or other markers for neuronal degeneration such as NSE), to monitor if these drugs also will slow down the degenerative process.