fMRI of eloquent cortex is established as a potentially useful technique to predict the risk of surgical resection or endovascular obliteration and can help guide therapeutic approaches. Blood oxygen level–dependent (BOLD) fMRI has been used for presurgical mapping for over 3 decades, but the technique has significant limitations that contribute to making it an imperfect indicator of functional risks. Inherent in the indirect (vascular) nature of the BOLD contrast mechanism is limited reproducibility, such that concurrence ratios of normal reactivated regions can range from 55–75%. BOLD fMRI also suffers from constraints in paradigm contrast (task-control condition interplay), relatively low temporal resolution, and patient-dependent variability in task performance. A major limitation of BOLD fMRI is lesion-induced (or radiation-induced) neurovascular uncoupling, which can underestimate cortical activation. Lesion-induced neurovascular uncoupling may have particularly profound consequences in preoperative assessments where activation patterns cause the appearance of lesion-induced cortical reorganization, which is the topic of this discussion.^1,2 If normal or near-normal function is maintained but fMRI shows a shift in relative cortical activation away from a resectable lesion, then it might be assumed that the function of the affected cortex has been taken over by the homologous area in the unimpaired hemisphere or heterologous cortex in the ipsilesional hemisphere. However, this premise could be erroneous if the BOLD mechanism, rather than neuronal function, is selectively impaired — particularly if there are increased demands placed on supportive brain regions. On the other hand, genuine lesion-induced cortical reorganization may occur in patients with tumor and AVM and may be demonstrated by fMRI.

Our investigation “Lesion-Induced Pseudo-Dominance at Functional Magnetic Resonance Imaging: Implications for Preoperative Assessments” sought to understand the impact of neurovascular uncoupling in the estimation of homotopic cortical reorganization.¹ Activation patterns in 50 patients with 85 functional areas adjacent to surgically treatable lesions were studied. The results supported the role of fMRI in preoperative mapping, as 97% of the functional areas of interest near brain lesions in our series were localizable by fMRI. In 23 (27%) functional areas, reduced fMRI signal in perilesional eloquent cortex in conjunction with preserved or increased signal in homologous contralateral brain areas showed activation dominance opposite the side of the lesion. In 7 of these areas, however, fMRI data were inconsistent with behavioral evidence for functional localization. That is to say that subsequent injury to apparently nonfunctional cortex, suggested by fMRI, resulted in neurologic deficits. Lesion-induced pseudo-dominance was observed in language, supplementary motor area (SMA), and corticobulbar networks. Sixteen patients showed proven (1 patient), probable (2 patients), or possible (13 patients) but unproven lesion-induced homotopic cortical reorganization. Thus, the phenomenon of pseudo-reorganization of eloquent cortex at fMRI should be considered when atypical hemispheric dominance or ipsilesional hemispheric activation patterns are observed.

The limitations of BOLD fMRI and its benefits are not mutually exclusive. Indeed, BOLD fMRI data may be crucial in the preoperative estimation of lesion-induced cortical reorganization, but must be interpreted in the specific context surrounding each patient. In effect, fMRI data are contributors to the estimated probability of cortical reorganization, which may range from essentially no chance of reorganization to near-certainty. Any pattern of cortical reorganization suggested at fMRI must be integrated with other data. This includes the likelihood of eloquent cortex reorganization. For example, SMA/pre-SMA and corticobulbar motor cortex reorganize extremely efficiently after injury. Somatosensory cortical injury often produces minor or insignificant deficits due to redundant networks. Corticospinal motor and elementary visual cortex, on the other hand, rarely demonstrate significant reorganization, particularly in the adult patient. Heteromodal and cognitive cortical reorganization are variable, depending on the time duration since lesion onset, the inherent degree of network redundancy, the size of a lesion-induced insult, and individual variability.

An anatomic analysis of lesion location may be predictive in that the degree of eloquent network destruction influences the likelihood of cortical reorganization associated with functional recovery. The expected time course and character of functional recovery also supports the notion of reorganization, and must be weighed against the mechanism of the network injury. For example, it may take 3–6 weeks for reorganization of SMA/pre-SMA cortex (homotopic or heterotopic), compared with 8–12 weeks for corticobulbar motor cortex (generally homotopic), resulting in near-complete functional recovery of both systems. Corticospinal motor functional recovery is better for gross proximal limb movements than fine coordinated movements of the distal extremity, and it may take many months (and significant rehabilitation) for recovery to occur. Steroid-induced improvement in neurologic function supports perilesional edema as the culprit for partial deficits and reduced BOLD activation, rather than tumor-induced destruction and cortical reorganization remote from the lesion. Indolent lesion-induced reorganization is more likely than reorganization from more rapid-onset deficits. Developmental versus acquired lesions and the age of onset of a lesion-induced deficit will influence the likelihood of reorganization as well.

In short, the congruence of anatomic, functional, and behavioral data favors the likelihood of cortical reorganization, and BOLD fMRI often is crucial in this estimation. If used judiciously, fMRI can provide critical evidence for cortical reorganization.

References:

1. Ulmer JL, Hacein-Bey L, Mathews VP, et al. Lesion-induced pseudo-dominance at functional magnetic resonance imaging: implications for preoperative assessments. Neurosurgery 2004;55:569–81, 10.1227/01.NEU.0000134384.94749.B2
2. Ulmer JL, Krouwer HG, Mueller WM, et al. Pseudo-reorganization of language cortical function at FMR imaging: a consequence of tumor-induced neurovascular uncoupling. AJNR Am J Neuroradiol 2003;24:213–17

 

Read this article in Neurosurgery …