Subjects with mild cognitive impairment who get abnormal readings on PET scans and do poorly on a word recall test are almost 12 times more likely to become afflicted with Alzheimer's disease, according to a new study.

The study, published online Wednesday in the journal Neurology, found that a battery of tests, including MRI brain scans and blood biomarkers, were able to help predict dementia risk.

But so-called FDG-PET scans coupled with poor performance on an episodic memory task were the best predictors of mild cognitive troubles progressing into full-blown Alzheimer's dementia.

"We find individually the [tests] were each for the most part useful for predicting conversion to Alzheimer's disease, but some were more useful than others," lead author Dr. Susan M. Landau, a researcher at University of California-Berkeley, told DOTmed News. "And the two most useful were the episodic memory and FDG-PET."

The study followed about 85 patients with mild cognitive problems, who were part of the broader Alzheimer's Disease Neuroimaging Initiative, for close to two years.

The patients were given a series of tests known to be predictive of dementia. Researchers checked for the APOE gene, associated with Alzheimer's, as well as the ratios of biochemical biomarkers derived from cerebral spinal fluid. These included the protein tau, implicated in the neurofibrillary tangles that develop in the brain in Alzheimer's patients, and amyloid beta, which forms plaque in patients' brains.

Patients also got MRI brain scans, to measure the volume of the hippocampus, a region of the brain involved in memory and known to shrink in Alzheimer's patients.

For PET, researchers scanned the patients' brains to measure 18-fluorodeoxyglucose uptake, as glucose metabolism is diminished in dementia. Patients also received a memory test to determine how many words they could recall from a list.

Although many studies have looked at each of these tests in isolation or compared a few against each other, this is the first report to directly compare them all to see which predicts both Alzheimer's and general cognitive decline, the researchers said.

"Each of these tests have independently shown promise in predicting disease progression, however, prior to the Alzheimer's Disease Neuroimaging Initiative, they had never been compared to one another in the same study before," Landau said in a statement.

When looking at all variables together, the researchers found that patients who had diminished glucose metabolism as revealed by the FDG-PET scan and performed poorly on the memory test were 11.7 times more likely to progress from mild cognitive impairment to Alzheimer's dementia.

In terms of general cognitive decline, just measuring downward change in cognition independent of Alzheimer's development, one of the cerebral spinal fluid markers - the ratio of total tau over amyloid beta - was the most useful, Landau said.

"It's kind of surprising we found this difference," she said.

When looked at individually, though, the humble pen-and-paper episodic memory test bested advanced imaging modalities as well as biochemical tests in forecasting dementia risk. Patients who performed poorly on the task were 4 times more likely to develop Alzheimer's, compared with abnormal PET scans (3 times more likely) or MRI scans (2 and a half times more likely), Landau said.

Landau suggests the reason for the slight predictive superiority of the memory test, called the Auditory Verbal Learning Test, is that it so closely resembles the sorts of tests used to actually diagnose Alzheimer's.

"Since AVLT is a paper-and-pencil cognitive test, and the diagnosis of Alzheimer's is based on the same questionnaire type of information, that could be one reason why they're related to one another," she said.

Alzheimer's is generally diagnosed by a neurologist after giving a patient a series of tests and interviewing the patient and people who know the patient well, Landau said.

Generally, a definite diagnosis can only be achieved during an autopsy, according to the National Institute on Aging.

In fact, one of the goals of Landau's research is to try to discover more objective ways to diagnose the disease while the patient is still alive.

"Officially giving someone an Alzheimer's diagnosis is kind of a subjective thing; it's not based on the kind of biological information we're looking at with the imaging scans," she said. "We'd like to have the diagnosis based on something concrete. It makes it hard to evaluate the predictive value of biomarkers when it's really hard to say this patient officially has Alzheimer's disease."

But the only problem with imaging exams, which might help for more objective diagnoses, is that they might not be cost-effective.

"From an economic point of view, the superiority of PET is challenging," write Dr. Carol Lippa, a neurologist at Drexel University, and Gael Chetelat in an editorial accompanying the article.

"It is hard to ponder health care reform in the context of biomarker studies as costly as PET," they continue. "The routine use of PET scans or other expensive diagnostics in order to obtain additional evidence to support predictions of progression or conversion may not be realistic."

While cognitive paper-based tests might be cheaper, Landau believes they're still hobbled by their subjectivity.

"The problem with cognitive tests is their variability," she said. "Is someone tired that day? Perhaps the person administering the test didn't speak clearly. With neuroimaging the hope is these markers would be more reliable and more objective."