Do brain tumours allow valid conclusions on the localisation of human brain functions?

In their objections to the paper by Shallice et al. (2010), Karnath and Steinbach (2011, this issue) called into question the use of brain tumours to draw conclusions on the localisation of human brain functions. However, their argumentation is very questionnable. Firstly, they write that “it is incorrect to consider brain areas of patients with tumours, in particular gliomas, intact that appear normal in MR imaging”. It is true. However, they forget to mention that in stroke, which seems to be an “ideal” model for them, it is also possible to have functional disturbances due to remote dysfunction in areas which appear normal in MR imaging (e.g., diaschisis). As a consequence, if brain tumours do not represent a good model to identify the “critical lesion sites”, the same conclusions can be made for stroke. Secondly, they insist on the fact that neural function can be well preserved in areas invaded by the tumour. Nonetheless, it is only true for low-grade gliomas. In their series, Shallice and colleagues analyzed 76 patients: among them, 25 had a high-grade glioma, 3 had an anterio-venous malformation, 15 had a meningioma, and 5 had a metastasis e thus in essence with no residual function in at least 63% of cases. Moreover, it was demonstrated in the recent literature that even in low-grade gliomas, neural function can be preserved in the tumour in only approximately 15e20% of the cases (Duffau, 2009). As a consequence, it is likely that more than 90% of patients in the series by Shallice and colleagues had no more function within their tumour. Even if it could be a possible counfounding factor, it cannot alter the main results and conclusions of the study.