The paper by Wahlund et al in this issue (pp630–635)1 tries to shed light on the usefulness of biological markers in the routine diagnosis of Alzheimer's disease. Tissue loss in the medial temporal lobe regions as detected by MRI is one of the most feasible and accurate markers developed to date and investigating its usefulness in a clinical setting is potentially relevant.

The paper has two main merits. Firstly, it highlights that the usefulness of diagnostic examinations needs to be judged within the specific diagnostic pathway.2 For any given patient coming to medical attention, physical signs and instrumental examinations are aimed to increment the prior probability of one (or few) of various possible diagnoses until a threshold of probability is passed, and the patient is diagnosed as having a given disease. Obviously, those examinations most likely to enhance the diagnostic probability are privileged. In the case of the patient suspected of having Alzheimer's disease, a clinical assessment—including the mini mental state examination (MMSE)—is carried out first, usually followed by instrumental examinations, among which are CT or MRI. Wahlundet al show that, although low performance on the MMSE alone is able itself to detect a large proportion (80%) of patients with Alzheimer's disease, the information on medial temporal lobe atrophy obtained from imaging leads to a significant increase of the detected proportion (95%). The bottom line is that imaging has significant added diagnostic value and is therefore a clinically useful procedure.

The second merit is the comparison of the added diagnostic value of a complex technique with that of one that is more applicable to the clinical routine. The authors have measured medial temporal lobe atrophy both with computed volumetric analysis on digital MRI and visual rating on MR films. Whereas the first, a precious research tool, requires specific software and hardware machinery as well as trained personnel, the second relies solely on rater's expertise and conventional films. Importantly, the findings indicate that in a clinical setting the simpler technique is at least as accurate as the more complex.

Although the authors' attempt to investigate the relevance of common clinical procedures should be commended, some issues are open to further investigation. The separation of Alzheimer's from non-Alzheimer's dementias on the basis of the protocol of Wahlundet al was not satisfactory (accuracy of 71%), suggesting that when it comes to separate different forms of dementia, atrophy in a single structure is not a good marker. Some findings indicate that the pattern of atrophy in some brain structures might be a more accurate differential marker.3 Secondly, the age of patients in the study of Wahlund et al was about 65 years, which is certainly not representative of most of the patients with Alzheimer's disease. Lastly, the value of imaging in addition to a complete neurogeriatric assessment4 was not considered by the authors, and more data are clearly needed on this issue.