In this study, evaluating patients with MMD treated with revascularization surgery and normal controls, parametric changes before surgery were most frequently observed in the parameters representing extracellular free water rather than free water-eliminated neuronal parameters. These results suggest that altered parenchymal free water may largely contribute to alterations in conventional diffusion parameters rather than the change in neurites in patients with MMD, as well as other neurological diseases[4, 8, 7]. This result is consistent with a previous study using NODDI and showed increased extracellular free water (Viso) in patients with MMD[14]. Previous studies reported increased perivascular spaces in patients with moyamoya disease[23, 24] and suggested that the increased free water may result from an impaired glymphatic system that uses arterial input as the driving force[25, 26]. Disruption of the glymphatic system is also suggested in some studies evaluating acute ischemic stroke[27]. Another possible reason for the increased free water in MMD is the disruption of the blood–brain barrier induced by chronic hemodynamic impairment[3]. Previous studies using fluorescent dyes and serum biomarkers revealed blood–brain barrier dysfunction in patients with MMD[28–31], and chronic hemodynamic impairment itself is known to cause blood–brain barrier dysfunction[3]. A study evaluating acute ischemic stroke patients using MRI also suggested blood–brain barrier dysfunction under cerebral ischemia[32]. Dysfunction of the blood–brain barrier may lead to the leakage of plasma proteins and thus pathologically accumulated extracellular free water in patients with MMD[24]. However, these assumptions are just hypotheses, and the mechanism underlying free water accumulation and decrease after revascularization surgery in MMD should be clarified in the future by evaluating the blood–brain barrier and glymphatic system in this disease population.
Our results also suggest that increased parenchymal free water may be decreased after revascularization surgery in patients with MMD and may explain, at least in part, the postoperative increase in FA. Previous studies reported the regression of WM hyperintensity after revascularization surgery[33, 34] and a postoperative decrease in parenchymal volume[8] in patients with MMD; these phenomena might be explained by the decrease in parenchymal free water after surgery. The postoperative decrease in free water was not accompanied by an increased number of neurites (i.e., increased Vic or FAt), suggesting that increased parenchymal free water is not a mere reflection of the decreased number of neurites accompanying chronic ischemic damage[3, 14]. A postoperative decrease in free water may suggest improvement of the disrupted glymphatic system and/or blood–brain barrier; however again, this assumption must be evaluated in the future. Widespread changes throughout the vascular territories might be obtained because the effect of indirect revascularization surgery sometimes extends outside the bone windows and improves hemodynamic disturbance in the entire hemisphere [35]; this point should be clarified in the future by concomitantly evaluating perfusion studies with reliable quantitative parameters such as 15O-gas positron emission tomography[22].
We found no direct relationship between postoperative changes in conventional FA and postoperative cognitive improvement, as in previous studies[8, 4, 7], and found no direct relationship between the postoperative reduction in free water and cognitive improvement. It is reasonable to assume that the postoperative cognitive improvement resulting from neuronal alterations causes functional and metabolic improvement[36, 8, 22] and that cognition is unrelated to the amount of extracellular free water[14]. The lack of a direct relationship between conventional diffusion parameters and cognitive performance may be explained by extracellular free water affecting conventional diffusion parameters. Additionally, we observed a significant difference between preoperative and postoperative FAt in GM regions that were not observed for conventional FA. This finding might be explained by the improved test-retest reproducibility for free water elimination[12] and improved sensitivity of gray matter abnormality[19, 37] with respect to conventional DTI. However, the clinical significance of this finding remains unclear regarding the lack of correlation between cognitive performance and FAt in GM regions.
Among our patients, the postoperative decrease in the OD value in the WM of the ACA was significantly correlated with the postoperative improvement in the working memory index, likely reflecting frontal lobe function. This finding was somewhat unexpected because our surgical treatment did not directly target the ACA area. Revascularization surgery may affect remote regions via WM fibers, such as the superior longitudinal fasciculus and superior and inferior fronto-occipital fasciculus [38], as opposed to the ischemic damage affecting remote regions[39, 40]. Previous studies evaluating patients with MMD at a single point found no significant correlation between OD values and the Working Memory Index[14, 13], which is attributed to the relatively uniform decrease in the Working Memory Index in this disease population[14]. These studies have also reported that the correlation between the hemodynamic parameters and difference between normal controls of the OD values in the WM are not as prominent as other parameters (Vic and Viso). Although these studies reported a decrease in the OD values in the WM in patients with MMD, increased OD values have been reported under mild pathological conditions, such as preclinical Huntington disease[41] and normal aging[42]. We believe the change in the OD values in the WM may be more complex than the other parameters. The tortuous axons and disrupted networks may cause a high dispersion, while the reduced number of axons caused by neuronal death and loss of fibers could simultaneously decrease the number of axonal networks, leading to low dispersion. The postoperative increase in the OD value is in accordance with the increase in FA reported in a previous study[8], suggesting modification of tortuous axons and disrupted networks that lead to cognitive improvement. When evaluating all the patients, the OD values were not significantly different before and after surgery, indicating that decreased OD values after the surgery may identify patients with postoperative cognitive improvement. These results must be interpreted cautiously because of the mixed laterality and surgical procedure, small sample size and patients with preexisting brain lesions. However, free water elimination analysis may detect neurite changes related to postoperative cognitive improvement, unlike conventional diffusion MRI.
The limitations of this study include the small sample size, heterogeneous patient backgrounds and surgical procedures. To improve the reliability of the results, we recruited age- and sex-matched normal controls, performed longitudinal analysis of the same patients, and used P values corrected for multiple comparisons, but substantial bias still exists. We did not perform quantitative analysis of perfusion study (ASL) together because quantitative value of ASL acquired in a clinical scanner is not always reliable, especially in patients with moyamoya disease whose arterial transit time is elongated[43]. We are planning to perform quantitative analysis between diffusion MRI and quantitative parametric values of 15O-gas positron emission tomography in the future after the accumulation of cases. As we stated above, the results regarding cognitive improvement should be considered preliminary because of the heterogeneity of revascularization procedures, laterality, and patients with preexisting brain lesions. Although a study reported histological correlations with neurite parameters of NODDI[44], histological validation does not exist regarding bitensor DTI and extracellular free water in any imaging modality[3], likely because evaluating free water components in fixed samples cannot be performed[45]. Despite these limitations, this study is the first to indicate altered extracellular free water before and after revascularization surgery in adult patients with MMD. A larger sample size is warranted to elucidate the microstructural parameters related to postoperative cognitive improvement in this disease population.