In this series of 29 patients with BAVM who underwent surgical nidus removal, we found that nidus depth may be more useful than nidus size as a predictor of the surgical risk for postoperative surgery-related morbidity in eloquent BAVM, especially with BAVMs in the motor and sensory areas. As the maximum nidus depth from the brain surface increased, the morbidity rate elevated, suggesting that the morbidity was permanent.
Many grading scales and scores to indicate the surgical risk associated with BAVM removal, on behalf of S-M grading scale, have been previously reported [1,2,3,4, 11,12,13]. From these reports, it is well known that the surgical risk is higher with eloquent BAVM. Most of the grading scores include nidus size as a surgical risk factor, and its importance has been previously reported [1,2,3,4, 11,12,13]. There is no doubt that the surgical risk increases with increasing size of the BAVM nidus. However, regardless of the nidus size, there are reports of morbidity due to surgery even with a smaller nidus. Thus, in BAVMs in eloquent areas, the surgical risk is higher from the very beginning [1, 2]. Furthermore, there are no reports that have evaluated the assessment of eloquence in detail. Therefore, certain problems still remain, such as the lack of inclusion of white matter eloquent fibers in the grading systems, lack of definition of the width and depth of the eloquent area, and lack of quantitative variables to describe the relationship between the nidus and eloquence .
Following the findings of a randomized control trial (RCT) for unruptured BAVM , the treatment for eloquent BAVM is still considered challenging. However, there are also some cases wherein BAVM resulted in hemorrhage with high morbidity and presented with uncontrollable symptoms, suggesting that surgical treatment should be considered for eloquent BAVM. As was reported previously, although symptoms appear transiently after surgical removal, they often resolve during follow-up [5, 8, 16, 17]. It appears that there may be risk factors other than nidus size. If there is a factor that can predict whether eloquent BAVM is treatable or not, then the decision for surgical intervention can be better judged.
Nidus Depth as a New Indicator of Surgical Risk for Eloquent BAVM
In previous reports, the following variables have been reported as risk factors for surgical removal in BAVM: BAVM size, eloquence, deep venous drainage, diffuseness, perforating artery supply, and unruptured presentation [1,2,3,4, 11,12,13]. It is also known that relatively good outcomes are obtained with surgical treatment of BAVM with hemorrhagic onset [5, 18]. Particularly with regard to the nidus size, it is easy to predict that BAVM localization will dominate the critical area as it gets bigger, and it will lead to an increase in the risk. However, in measuring the size, the obtained values and implications vary depending on the direction used to measure on the image since the size can be measured in all directions. There are no uniform guidelines on assessing this measurement. Therefore, we focused on the nidus depth and performed the analyses for surgical morbidity for eloquent BAVM in this study. Although it seems like common sense that more extensive and deeper AVMs in eloquent locations will be associated with more surgical morbidity, to the best of our knowledge there are no reports in the literature that investigated and mentioned BAVM depth. Various factors that were previously reported to be significantly correlated with postoperative morbidity were not found to be significantly correlated in this study; the p values of the factors, though relatively low, were not significant. It is suggested that the maximum nidus depth may be more useful as a predictor in the surgical treatment of BAVM in the Rolandic area (Table 3 and Fig. 2).
Optimal Cutoff Value of Maximum Nidus Depth for Surgical Morbidity
To determine the optimal cutoff values for the maximum nidus depth and size in relationship with surgical morbidity, ROC curves were generated. The cutoff value for the maximum nidus depth was 36 mm for total morbidity and 41 mm for permanent morbidity (Table 4). Therefore, the deeper the maximum nidus depth, the higher the surgical risk. The cutoff value for the maximum nidus size for permanent morbidity was 30 mm. The AUC of the maximum depth (total morbidity: 0.74, permanent morbidity: 0.88) was larger than that of the maximum size (total morbidity: 0.58, permanent morbidity: 0.76); therefore, maximum nidus depth is a more useful predictor than the maximum nidus size (Table 4, Fig. 2).
These findings raise the question of the underlying mechanism that can explain how the nidus depth affects the surgical risk. Frequently this is due to small arterioles arriving at the deep part of the AVM nidus, or due to small draining venules, which are bleeding notoriously and are difficult to coagulate at the same time. It is also already known that the deeper a BAVM is, the more the nidus involves the CST  and is closer to the perforating supply area. In previous reports, lesion-to-CST distance measured by DTI [6, 7] and LAD measured by fMRI  have been reported as surgical risk factors for vascular malformations of the eloquent area. Furthermore, the plasticity and cortical reorganization of the eloquent cortical area were also reported [5, 16, 17], and it was suggested, based on fMRI, that significant activation did not lie within the nidus of BAVM . The anatomic location alone of a BAVM may not provide any information on the functional reorganization. In fact, we have encountered cases where translocation of the eloquent area was found on intraoperative direct stimulation. Thus, the plasticity may contribute to the transient deterioration in the cortical regions after the surgery. On the other hand, impairment of CST and deep white fibers may be more critical and irreversible, resulting in permanent deterioration. Therefore, the nidus depth can be a risk factor as it reflects the association between a BAVM and the deep white fibers.
Although DTI and fMRI are useful for the prediction of the surgical outcomes, they are often complicated. It is often difficult to perform imaging accurately because of methodological limitations and technical difficulties like different stepwise procedures between institutions, false-positives and false-negatives, and hemorrhage and edema that can affect the results . On the other hand, nidus depth, as proposed in this study, is an easily measurable parameter that does not require special images. It can be measured with routine images that do not vary significantly between institutions.
ARUBA trial was the only RCT in unruptured BAVMs with a negative stance on interventions , and it suggested that interventions for unruptured BAVM should be cautious; however, there have been objections to that report [19, 20]. In this study, asymptomatic presentation was a significant risk factor in univariate analysis but not in multivariate analysis. Maximum nidus depth was the only significant factor in multivariate analysis. Its cutoff value obtained in this study does not strongly recommend the surgical removal of eloquent BAVMs; however, it can be one of the indicators of the surgical risk in cases that require treatment. This study does not deny the importance of the size of the nidus as a risk factor since there were cases wherein the measured maximum size was in the same direction as the depth. Therefore, the direction in which the maximum size is measured is also important.
Many outstanding questions and issues pertinent to this study remain. The results obtained in this study are from limited locations of eloquent BAVMs, specifically the Rolandic area. However, it is unclear as to which area is better—the motor or sensory area. The significance of the depth in other eloquent areas like the language area remains unknown. Furthermore, we do not discuss the relations between DTI tractography and closeness of the nidus. Although the effectiveness of GKS and multimodal treatment for high-grade BAVM has been reported in recent years [21,22,23], the current study is not a comparative one between the modalities of treatment. The question remains whether surgical removal is superior to GKS or another multimodal strategy. As another limitation, this study was a retrospective one and the sample size was small. Further analyses and experiments are needed in this regard. This small subset of patients may overestimate the value of nidus depth compared with other variables that are well known to be factors in development of perioperative morbidity, including hemorrhagic status and size, primarily, and others. However, this is an interesting case series of an uncommon entity: surgically operated motorsensory cortex. We think that it is a meaningful report derived from rare cases.
From our results, the radicality of surgical removal was extremely high and the requirement for surgery cannot be completely ruled out. In the future, downgrading using multimodal treatment may also be promising as a new strategy to prevent surgical morbidities in such challenging entities. Thus, critical deep lesions or eloquent locations should first be eliminated with relatively noninvasive treatment modalities with higher priority , resulting in downgrading by decreasing the maximum nidus depth and affected eloquent area followed by surgical treatment. It is necessary to consider new treatment strategies for safe and highly radical BAVM treatment. Maximum nidus depth is more likely to be associated with surgical morbidity in BAVMs in the eloquent Rolandic areas. Although surgical treatment should be carefully considered, the maximum nidus depth is a simpler and stronger predictor of the outcome than maximum nidus size in patients with this challenging entity.