Glymphatic system has been demonstrated as a lymphatic-like circulating system in brain tissue and was predicted existing in paravascular pathway. A recent study has confirmed this hypothesis by injecting fluorescent tracers into the cisterna magna and thus located their distribution in murine ¹. As a freely diffusible tracer, deuterium oxide (D₂O) may be an alternative in discovering glymphatic system through non-invasive tail-vein injection method. In addition to the quantification of hemodynamics ², D₂O may also trace the water passage in glymphatic system because of its water-close characteristic. Wang et al. have used D₂O as a negative contrast agent by ¹H-MRI imaging and showed improved signal-to-noise ratio comparing to direct measurement of D₂O ³. Therefore, we used this new strategy to acquire D­₂O perfusion images in this study, and extracted both blood and glymphatic dynamics simultaneously by applying two-compartment parallel model (2CPM). The traditional one-compartment Tofts model (1CTM) was also applied to quantify cerebral blood flow (CBF) for comparison ⁴.

Six adult male Spaugue-Dawley rats were anesthetized by 1.5 % isoflurane and injected with D₂O (2ml/100g) via tail vein at 7T Bruker Clinscan MRI scanner. 80 measurements were acquired by the following parameters: turbo-spin-echo (TSE) with TR/TE = 2000/14ms, matrix size = 128x256, FOV = 35mm, turbo factor = 3, slice number = 3, slice thickness = 1.5mm, distant factor = 40%, sampling interval = 34s. Then, the relative concentration-time curves were estimated by calculating signal change from the pre-injection baseline level. The arterial input function (AIF) was extracted by averaging three to six selected voxels in the area of cerebral artery and reconstructed by a₀*t for wash-in phase and a₁*exp(-b*t)+c for wash-out. The two pharmacokinetic models, 1CTM and 2CPM, were utilized as the following equation (1) and (2), respectively.

(1) $$$C_{t}(t)=AIF(t)\otimes[F_{0}\exp(-F_{0}t/v_{0})]$$$

(2) $$$C_{t}(t)=AIF(t)\otimes[F_{1}\exp(-F_{1}t/v_{1})+F_{2}\exp(-F_{2}t/v_{2})]$$$

where C_t: concentration-time curve, F_i: flow, v_i: distributed volume, where i=0,1,2. Parameter constraint: v₀<1, F₁≥F₂, v₁+v₂=1. Then we quantified F_i as flow with a unit of ml/min/100g.

Figure 1 shows the spatial maps of F₀ quantified from 1CTM (1a) and F₁ from 2CPM (1b) of six rats (R1-6), respectively. These two maps both show nice contrast of brain tissues and appear to be very similar to each other. The F₂ maps from 2CPM are presented in Figure 2. Only regions at some specific structures show significant flow values. In Figure 3, the F₂ values of Rat 5 higher than an empirical threshold (in this case, F₂ > 3 ml/min/100g) are included in the regions of interest (ROIs), and superimposed on the anatomical images. These locations of ROIs are near several arteries, including (I) middle cerebral artery, (II) azygos pericallosal artery, (III) supracollicular arterial network, and (IV) posterior communicating artery. Some regions surrounding on the cortical surface of brain might be surface and penetrating arteries. Figure 4 demonstrates the averaged concentration-time curves of whole brain and ROIs in Rat 5 with a curve of the corresponding AIF for reference. Comparing to the whole-brain concentration-time curve, the ROIs curve shows a higher magnitude at first-bolus arrival and slower clearance rate at the end of dynamic scans. Figure 5 summarizes flow values of all rats and their coefficients of determination, where R₁² is F₀ versus F₁ and R₂² is F₀ versus F₂, respectively.This study demonstrated that the multi-compartment analyses on D₂O perfusion by using the new imaging strategy could be useful to quantify the water dynamics with adequate spatial and temporal information. F₁ was highly coordinated with F₀ despite of an insignificant augment, thus F₁ could be speculated as the CBF. On the other hand, the F₂ value of a parallel flow was much irrelevant with the blood flow, and the spatial distributions of the parallel flow were adjacent to the location of cerebral arteries, which matched with the recent report of the paravascular pathway of cerebral-spinal fluid (CSF) ¹. Therefore, according to the blood irrelevant flow values and the spatial matched mapping, we have demonstrated that using 2CPM for tracing D₂O might noninvasively reveal the information of CSF-dynamics which is regulated by glymphatic system. Further investigations and applications should be conducted to connect D₂O tracer analysis of 2CPM with the comprehensive water passage in rat brain.