Introduction

Preclinical studies demonstrate that lymphatics regulate tissue sodium levels via lymphangiogenesis^1,2,3. However, it is unclear whether lymphangiogenesis can assist with the clearance of excess tissue sodium storage, which has been observed in the skin and muscle by multi-nuclear sodium ²³Na-MRI and is a hallmark of patients with hypertension⁴. To study this physiology, we are developing sodium and lymphatic imaging strategies in a mouse model of lymphangiogenesis in adipose tissue (Adipo-VD)⁵, through inducible vascular endothelial growth factor-D (VEGF-D) expression. In this initial study, we measured sodium ²³Na relaxation and tissue sodium content (TSC) measurements at 15.2T to test the hypothesis that TSC is reduced in animals with lymphangiogenesis.

Methods

Animal Model. The Adipo-VD mouse model is capable of controlled de novo adipose tissue lymphangiogenesis from doxycycline-activated expression of murine VEGF-D via a tetracycline response element⁵. Three functionally Adipo-VD+ and three Adipo-VD- mice (N=6) were maintained in accordance with institutional guidelines and fed a 600 mg/kg doxycycline diet with standard 0.24% sodium content (Bio-Serv, Flemington, NJ, USA).

Image Acquisitions. ²³Na-MRI data were acquired with a 15.2T Bruker BioSpec (Bruker Biospin, Billerica, MA, USA) with a custom-built dual-tuned ¹H/²³Na volume coil. Animals were positioned with samples of 30 mmol/L and 100 mmol/L NaCl in deionized water in the field-of-view (FOV). Following localization of the area of interest consisting of abdominal fat including the inguinal lymph nodes (Figure 1A,1B), images were acquired with a 2D ultrashort echo time (UTE) radial center-out sequence with the following parameters: TR/TE=150/0.25 ms, number of projections=202, flip-angle=85 degrees, bandwidth=100,000 Hz, FOV=40x40x10 mm³, matrix=64x64, number of signal acquisitions (NA)=38. For T₁ measurements, 2D UTE was repeated with identical parameters except NA=6, TE=0.128 ms, and six different TR = 30, 50, 100, 150, 200, 250 ms. For T₂^* measurements, 2D UTE was repeated with TR=150 ms and TE = 0.13, 0.25, 0.50, 1.0, 1.5, 2.0, 2.5, 5.0, 10.0, 20.0, 30.0, 50.0 ms. Images were reconstructed offline with Kaiser-Bessel gridding using a custom MATLAB routine (Mathworks, Natick, MA, USA). Tissue sodium maps were calculated based on a linear calibration of ²³Na signal intensity in the standard samples (Figure 1C).

Relaxometry.
Naturally abundant ²³Na is a spin = 3/2 system⁶ and interactions of the quadrupole moment with the external magnetic field and tissue microenvironment produce characteristic relaxation time constants (T₁, T₂^*_short, T₂^*_long) which were quantified in two regions of interest (ROIs): skin and muscle. Relaxometry maps were generated by calculating the exponential time constants voxel-wise, and the mean time constants measured in each ROI. All fittings were performed in MATLAB using the non-linear trust-region-reflective fitting algorithm. Rician noise bias was corrected for using the method described by Gudbjartsson and Patz⁷ where noise was measured from a uniform region of thermal noise.

The ²³Na longitudinal relaxation time constant (T₁) was estimated using a model of mono-exponential recovery, as the short component is considered <20% of the signal even in non-aqueous environments⁸. A conventional Bloch equation model describing T₁ exponential recovery was fit for three parameters:
$$S=A\bigg(1-B e^{\frac{-TR}{T_{1}}}\bigg)$$
where A is constrained by the equilibrium signal intensity, and B is a term to correct for B₁ perturbation. The ²³Na transverse relaxation time constants corresponding to the short (T₂^*_short) and long (T₂^*_long) components were estimated using a model of bi-exponential decay⁹ fit for four parameters:
$$S=A\bigg(f_{short}e^{\frac{-TE}{{T_{2}^*}_{short}}} + (1-f_{short})e^{\frac{-TE}{{T_{2}^*}_{long}}}\bigg)$$
where A is constrained by the equilibrium signal intensity, and f_short is the relative fraction of the short component of the signal.

Results

In this preliminary cohort of Adipo-VD mice imaged at 15.2T, skin TSC was observed to be lower in Adipo-VD+ vs. Adipo-VD- animals (18.30±1.68 mmol/L vs. 28.11±5.89 mmol/L, Figure 1D). The measured ²³Na relaxation times and regional TSC are summarized in Figure 2. Relaxation times did not differ significantly between Adipo-VD+ and Adipo-VD- animals and were found to be T₁=44.25±5.80 ms in the skin and T₁=48.80±6.03 ms in the muscle (Figure 3). Bi-exponential T₂^* relaxometry in the skin was T₂^*_short=2.07±0.21 ms and T₂^*_long=21.21±0.33 ms, and in the muscle, T₂^*_short=1.86±0.27 ms and T₂^*_long=24.16±2.31 ms (Figure 4).

Discussion

Our results demonstrate preliminary differences in ²³Na-MRI tissue sodium content in a pre-clinical model of lymphangiogenesis, which may be useful for developing noninvasive imaging strategies to study sodium and lymphatic physiology. Sodium relaxometry in the rodent brain¹⁰ and human knee¹¹ at high field strengths have been previously reported, but to our knowledge sodium relaxation rates at 15.2T have not been reported in preclinical models. Observed trends of lower tissue sodium in animals with lymphangiogenesis may be related to enhanced lymphatic functional clearance networks, which will be explored in further studies.

Conclusion

We acquired sodium ²³Na UTE MRI at 15.2T and used non-linear fitting techniques to estimate sodium ²³Na-relaxation times and tissue sodium content in a unique animal model of controlled lymphatic growth. Animals with lymphangiogenesis demonstrated lower skin sodium content, while relaxation times in the skin and muscle were similar in this preliminary study.

Acknowledgements

Funding was provided by the American Heart Association Innovative Project Award (AHA IPA 19IPLOI34760518). All imaging experiments were performed at the Vanderbilt University Institute of Imaging Science Center for Small Animal Imaging.