It is possible to both have and not have Alzheimer’s disease. Contradictory as this statement is, a study reported from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) supports it.

In a paper published in the October issue of the Annals of Neurology investigators reported the results of biomarker studies of 53 patients with dementia caused by Alzheimer’s disease. They found a notable proportion of these patients lacked one of the signature pathologies: brain amyloid. This result has notable scientific and policy implications.

Since 2004, ADNI, a longitudinal, NIA-industry funded study, has meticulously followed a cohort of persons with normal cognition, mild cognitive impairment, and dementia caused by Alzheimer’s disease. Key measures are biomarkers, a term that describes a measure that captures a complex pathologic process, such as how low density liproprotein (the LDL or “bad” cholesterol) captures the myriad of events leading to heart disease, and if abnormal, prompts a clinician to prescribe treatment.

In the case of Alzheimer’s disease, one of the most provocative biomarker measures is PET amyloid imaging, a measure of the presence of amyloid plaques in the brain. Amyloid is one of Alzheimer’s disease two pathologic signatures, the other being tangles of dysfunctional tau protein. An amyloid scan of the brain of a person with dementia caused by Alzheimer’s disease is a kind of living biopsy of the brain. An expert clinician, such as those studying the patients in ADNI, would expect to find that a patient with Alzheimer’s disease would have a “positive amyloid scan.”

Which is why the study is so revolutionary. Twelve of the 53 patients that an expert clinician diagnosed as demented because of Alzheimer’s disease had a negative amyloid scan. In short, their virtual brain biopsy was negative.

For clinicians, this result from a small sample of subjects is not enough to change clinical practice for the thousands of older adults diagnosed every year with Alzheimer’s disease dementia. But this may change.

In the Spring of 2012, the Food and Drug Administration approved the use of florbetapir for the imaging of brain amyloid, and Medicare will soon decide what kinds of indications for florbetapir, owned by Lilly, will receive reimbursement. Their decision will, in turn, create precedent for what private insurers will cover. If more studies show the same result as this study, they will challenge whether to expand what is expected to be a fairly narrow set of reimbursable uses.

For researchers and clinicians, this result raises an important question. If one in five of carefully examined older adults with Alzheimer’s disease in fact do not have one of its signature pathologies, what disease do they have? ADNI researchers will undoubtedly hurry to explore the study’s vast treasure chest of psychometric, imaging and spinal fluid data answer this question. They may find that what clinicians call Alzheimer’s disease is, in fact, many diseases.

Pharmaceutical companies and clinical trialists should also consider this result. The history of drug development in Alzheimer’s disease has been a serial disappointment. Just this summer, the giants Pfizer and Lilly reported negative results for their drugs—bapineuzamab and solaneuzemab, respectively—that targeted amyloid in persons with Alzheimer’s disease.

The ADNI results suggest that the drugs may not have been the problem. Instead, the studies may have failed because some of the patients, perhaps as many as one in five, did not in fact have the very pathology the drugs were designed to attack, namely amyloid. Pfizer may want to revisit its decision to shut down a large portion of its once vast Alzheimer’s drug development operation.

For this day forward, it is reasonable to argue that all patients with Alzheimer’s disease recruited for a clinical trial that targets amyloid should have a PET scan for amyloid and that only those who are amyloid positive be enrolled in the trial. The clinical implication of such a design, if it shows the drug is successful, is that prior to initiating treatment, a patient with Alzheimer’s disease will need a PET scan. The future price tag for treating the millions of patients just went up by a few thousand dollars.

This price tag is just one reason why health care policymakers should ponder the implications of this research. One of the compelling reasons why Alzheimer’s disease is called a “tsunami” is that millions of people have it, and, over the next two decades, millions more. And yet, if perhaps twenty percent of them have some other disease, this prevalence count is unstable. This instability suggests that the value of measuring the size of the problem is the number of persons with dementia, regardless of the cause.

Jason Karlawish is Professor of Medicine, Medical Ethics and Health Policy, University of Pennsylvania.  He blogs at his self-titled site, Jason Karlawish.