Chemical-shift selective saturation recovery (CSSR) MR spectroscopy using hyperpolarized xenon-129 (HXe) is developing into a promising modality for assessing lung function through quantification of pulmonary gas-exchange parameters^1-6. In healthy human subjects, CSSR uptake curves can be divided into delay-time (τ) regimes that are dominated by three fairly independent parameters: 1) The surface-to-volume ratio (S/V) (τ < 20 ms); 2) the septal wall thickness (τ = 20-200 ms); and 3) the blood flow rate (τ > 200 ms). Nonetheless, the entire data range is commonly analyzed collectively using theoretical multi-parametric models^3,7,8. While these models are very useful to characterize the saturation region of the uptake curve (τ = 20-200 ms), the constraint of fitting all delay time regimes simultaneously frequently results in mediocre fits for all of them. In this work we investigated for which delay times the simple, but suitable, Butler model¹ is applicable for determination of alveolar S/V through correlation with the apparent diffusion coefficients (ADCs) in chronic-obstructive pulmonary disease (COPD) patients, healthy smokers (HS) and age-matched normal (AMN) subjects.

Eight AMN (age 63 ± 10), 8 HS (age 60 ± 9), and 13 COPD (age 67 ± 8) subjects underwent HXe ADC MRI and CSSR MR spectroscopy. All MR studies were performed at 1.5T (Avanto; Siemens), using a flexible xenon-129 chest RF coil (Clinical MR Solutions), under a physician’s IND for HXe MRI. Informed consent was obtained in all cases and a physician supervised each study. Enriched xenon gas (87% xenon-129) was polarized using a prototype commercial system (XeBox-E10, Xemed). For CSSR data analysis the slope of a linear regression line fitted through the tissue-plasma (TP) and red-blood cell (RBC) peak areas as a function of the square root of the delay time τ between saturation of the dissolved-phase magnetization and RF excitation of a free induction decay was used to calculate S/V according to⁹:

$$F(\tau) = \lambda\frac{S_{A}}{V_{Gas}}\sqrt{\frac{4D_{diss}\tau}{\pi}}$$

F(τ) is the measured dissolved-phase (either TP or RBC) to gas-phase ratio, λ is the Oswald xenon solubility (0.1 for TP and 0.17 for RBC), S_A is the alveolar surface area, V_Gas is the gas volume and D_diss is the xenon diffusion constant in tissue (3.3 × 10^-6cm²/s)¹⁰. Up to five delay times (3, 5, 7.5, 10, and 15 ms) were included in the computation of S/V.

Figure 1 depicts the correlation between the TP SV calculated using the 5 shortest delay times of the CSSR acquisition, and the HXe ADC measurement averaged over the entire lung. Considering the conceptual similarity of these two quantities an R² value of 0.31 (p = 0.002) for the correlation was unexpectedly low. Interestingly, R² increased to 0.6 (p = 9×10^-7) when only the shortest TP delay time (3 ms) was employed in the S/V calculation and, maybe even more surprisingly, peaked at an R² of 0.65 (p = 1×10^-7) for the S/V based on the 3-ms delay time of the RBC uptake curve (see Fig. 2). While the link between a morphometric S/V and the CSSR-derived S/V is intuitively fairly clear for the TP measurement, the interpretation of the S/V for the RBC measurement is less obvious. The TP and RBC S/V were well correlated (R² = 0.81; p = 3×10^-11) within our subject cohort (Fig. 3) but we would argue that the former (TP S/V) is tied to structure while the latter (RBC S/V) is rather an effective S/V of functional nature. Thus, these two parameters are tied to different aspects of pulmonary gas exchange and one might expect their correlation to weaken for other disease patterns (e.g., fibrosis). Nevertheless, our results seem to indicate that it is instrumental to measure the dissolved-phase to gas-phase ratios at very short delay times (5 ms or less) for an accurate assessment of the pulmonary S/V.We demonstrated that for very short delay times following saturation of the HXe dissolved-phase magnetization in the lung, CSSR spectroscopy is highly correlated with the ADC measurement and, hence, the related pulmonary S/V. This correlation quickly diminishes for delay times larger than about 5 ms.