LAUSR

In a recent paper, Strand et al. [4] investigated the incidences and characteristics of strokes in a large series of 539 glioma surgery cases. They found an overall incidence rate of 44%, with a median volume of 1.7 cm3. They identified several risk factors, including known cerebrovascular disease for sector-shaped infarctions, and tumor volume, perioperative bleeding, and age for larger volume infarctions. Finally, they also described the topography of these mini-strokes. While this study added valuable information to our knowledge on this topic, it would have been interesting to put these results in perspective with those of two previous studies dealing with the exact same question. First, Loit et al. [2] reported their results on a series of 115 glioma surgery cases. It is striking to note that the incidences and volumes were very similar (50% for lower-grade glioma, 41% for glioblastoma, median volume of 1.6 cm3). Moreover, the topographical distribution is almost the same in the two studies (compare Fig. 1 in Loit et al. and Fig. 3 in Strand et al.). Loit et al. proposed to explain this topography as territories corresponding to perforating arteries, suggesting that such strokes could not be avoided whenever the fundus of a sulcus had to be resected for oncological reasons. Contrarily to Strand et al., Loit et al. could not identify any risks factors of stroke occurrence (as also reported by other authors [3]). Admittedly, the preexistence of a cerebrovascular disease was not assessed nor the role of tumor volume or perioperative bleeding in this previous study. But age was not found to correlate with higher incidence of postoperative strokes. Second, Berger et al. [1] also recently reported their results on a series of 82 patients. Lower incidence (23%) and larger volumes (mean of 14 cm3) suggest that these authors used a different criteria for the diagnosis of strokes. Importantly, the only risk factors found in their study were insular location and recurrent tumor, but not age. On the contrary, mean arterial pressure at the beginning of the surgery was lower in the infarcts group. One possibility for explaining the differences regarding risks factors in the three studies could be related to the larger size of the study of Strand et al., likely increasing statistical power. But this would have deserved a discussion. Last but not least, the two aforementioned studies also investigated the cognitive impact of these strokes. Remarkably, these two studies were in good agreement, demonstrating that the difference in terms of cognitive outcome was subtle albeit statistically significant (for semantic fluency in [2] and for verbal rhyming in [1]). It is disappointing that no information about cognitive impact of the strokes was provided by Strand et al. More generally, the fact that both authors and reviewers missed to discuss these two important papers demonstrates the difficulty to stay up-to-date with the literature, considering the exponentially increasing number of publications. This raises the question of implementing algorithms in the review process to automatically detect papers that should be included in the discussion, in order to progressively build a knowledge whose robustness would be testified by its reproducibility among different teams around the world.