0443

A Model for Interpreting Hyperpolarized ¹²⁹Xe Gas Exchange MRI

Ziyi Wang¹, Leith Rankine¹, Aparna Swaminathan¹, Elianna A Bier¹, Matthew P Thorpe², Robert Tighe¹, Yuh Chin Huang¹, Sudarshan Rajagopal¹, and Bastiaan Driehuys¹
¹Duke University, Durham, NC, United States, ²Mayo Clinic, Rochester, MN, United States

Synopsis

Hyperpolarized ¹²⁹Xe gas exchange (GX) MR imaging of pulmonary ventilation, barrier uptake and red blood cell (RBC) transfer has shown sensitivity to a wide range of pathology. However, the physiological interpretation of regional RBC transfer defects is not yet fully established and its connection to conventional measures has yet to be studied in a broad range of pathology. Here we evaluate the extent to which ¹²⁹Xe RBC transfer reflects local perfusion, by testing its spatial correlation to ^99mTc scintigraphy and propose a generalized model connecting ¹²⁹Xe gas exchange metrics to the membrane and capillary blood volume components of DL_CO.

Introduction

Hyperpolarized ¹²⁹Xe MRI has emerged as a powerful means to image pulmonary gas exchange (GX), with its ability to provide 3D non-invasive mapping of ventilation, barrier uptake and red blood cell (RBC) transfer¹. Prior work has shown ¹²⁹Xe RBC transfer to be sensitive to a broad range of pathology². Initial application to healthy and IPF subjects suggested the ratio of RBC:barrier correlates strongly with the clinical measure of gas exchange — DL_CO³. However, the physiological interpretation of regional RBC transfer remains to be established. Moreover, the simple RBC-barrier ratio does not account for factors such as emphysematous loss of membrane conductance or accessible alveolar volume V_A. Thus, we sought to develop a comprehensive means to interpret ¹²⁹Xe MRI across a broad range of pathology. First, we evaluate the extent to which ¹²⁹Xe RBC transfer imaging reflects local perfusion by testing its spatial correlation to ^99mTc scintigraphy. Second, we propose a generalized model that uses ¹²⁹Xe GX MRI to reveal both the membrane and capillary blood volume conductance contributions to the transfer coefficient K_CO. We then use the ventilation images to estimate accessible alveolar volume V_A, and thereby calculate the diffusing capacity DL_CO.

Methods

We analyzed 17 patients who had ^99mTc scintigraphy within 9 months of ¹²⁹Xe MRI: 2 with chronic obstructive pulmonary disease (COPD), 6 chronic thromboembolic pulmonary hypertension (CTEPH), 5 pulmonary arterial hypertension (PAH), 1 left heart failure (LHF) and 3 with unknown dyspnea. ¹²⁹Xe MRI produced 3D maps of ventilation, barrier uptake and RBC transfer. RBC images were rendered into conventional planar projections used for ^99mTc perfusion, by applying attenuation from lung parenchyma and body tissues. The averaged coronal RBC projection was quantitatively compared to the geometric means of the anterior and posterior ^99mTc scans using 6-zone analysis. This framework reports the functional contributions of each zone and allows calculation of a Pearson correlation coefficient (R²) between modalities as well as Bland-Altman plots to evaluate the limits of agreement (95% confidence intervals) for each zone.

A generalized model to estimate DL_CO from ¹²⁹Xe MRI was developed based on 95 subjects who had DL_CO testing and high-quality ¹²⁹Xe MRI available. This included: 17 healthy, 21 COPD, 13 IPF, 29 nonspecific interstitial pneumonia (NSIP), 6 LHF, and 9 PAH. Using the framework of Roughton and Forster⁴ (Figure 1), we propose that image-derived mean values of ¹²⁹Xe barrier uptake and RBC transfer reflect the membrane conductance D_M and capillary blood volume V_c respectively. Therefore, K_CO can be represented as K_CO = (α*Bar_REL^-1+β*RBC_REL^-1)^-1 , where α and β are coefficients. Bar_REL and RBC_RELare calculated by dividing the patient’s mean value by that of the healthy reference cohort. Specifically, the inverse of Bar_REL is used when it exceeds unity, as in patients with interstitial lung diease³, allowing it to convey both alveolar surface area and membrane thickness aspects of D_M. Image-derived DL_CO is then calculated with V_A estimated from the ventilation volume VV using V_A = γ*VV.

Results

Figure 2 (left) shows ^99mTc and ¹²⁹Xe-RBC projections in a PAH patient and a CTEPH patient. The two modalities agreed well in the PAH patient. In contrast, the CTEPH patient showed absent lobes on the perfusion scintigraphy, that nonetheless exhibit signal on ¹²⁹Xe projections. Over the entire cohort, the two modalities demonstrated strong correlation (R²=0.84±0.14), while individual zones exhibited negligible bias between modalities (within 3%), and limits of agreement within 10%.

The DL_CO model was fit across the entire cohort (N=95) to determine the coefficients: α=11.2, β=14.6 mL/min/mmHg/L, and γ=1.43. Figure 3 shows a 3D surface of the image-estimated K_CO as a function of Bar_REL and RBC_REL, compared to the clinically measured values. The associated graphs show that image estimated V_A correlates well with the measured value (R²=0.66, p<0.001). Combining these to generate image estimated DL_CO shows a good correlation (R²=0.77, p<0.001) with the clinical measurements.

Figure 4 shows the application of this model to images of representative subjects. Compared to the healthy subject, all patients exhibit diminished DL_CO resulting from different combination of reduced V_A (IPF, LHF, PAH), decreased barrier uptake due to emphysema, increased barrier uptake (IPF, LHF), and diminished RBC transfer.

Discussion and Conclusion

While ^99mTc and ¹²⁹Xe-RBC agree well in most cases, an explanation for the poor agreement in CTEPH patients is proposed (Figure 2). In these patients, occlusion of the larger pulmonary arteries reduces flow, and therefore limits delivery of ^99mTc to the capillary bed. However, if the capillary bed remains viable, it will remain filled with blood and thereby allow ¹²⁹Xe RBC transfer through interstitial diffusion⁵. Thus, while well-perfused regions will show good RBC transfer, and destruction of capillary bed will decrease both, partial occlusion of large vessels can lead to significant mismatch between ^99mTc and ¹²⁹Xe-RBC imaging. (It is notable that DL_CO can be normal in CTEPH patients⁵).

This interpretation is further supported by our DL_CO model, which connects to all aspects of ¹²⁹Xe MR imaging. Importantly, it suggests that ¹²⁹Xe imaging permits decomposing both the membrane conductance and capillary blood volume aspects of gas exchange, as well as resolving it 3-dimensionally.

This effort towards establishing a framework for interpreting ¹²⁹Xe RBC transfer imaging will benefit from further verification using well-controlled animal studies.

Acknowledgements

This work was supported by the NIH (R01-HL-105643, R01-HL-126771 and HHSN268201700001C).

References

1. Kaushik, S. S.; Robertson, S. H.; Freeman, M. S.; He, M.; Kelly, K. T.; Roos, J. E.; Rackley, C. R.; Foster, W. M.; McAdams, H. P.; Driehuys, B., Single-Breath Clinical Imaging of Hyperpolarized (129)Xe in the Airspaces, Barrier, and Red Blood Cells Using an Interleaved 3d Radial 1-Point Dixon Acquisition. Magn Reson Med 2016, 75, 1434-43.

2. Wang, Z.; Bier, E. A.; Swaminathan, A.; Parikh, K.; Nouls, J.; He, M.; Mammarappallil, J. G.; Luo, S.; Driehuys, B.; Rajagopal, S., Diverse Cardiopulmonary Diseases Are Associated with Distinct Xenon Mri Signatures. Eur Respir J 2019.

3. Wang, J. M., et al., Using Hyperpolarized (129)Xe Mri to Quantify Regional Gas Transfer in Idiopathic Pulmonary Fibrosis. Thorax 2018, 73, 21-28.

4. Roughton, F. J.; Forster, R. E., Relative Importance of Diffusion and Chemical Reaction Rates in Determining Rate of Exchange of Gases in the Human Lung, with Special Reference to True Diffusing Capacity of Pulmonary Membrane and Volume of Blood in the Lung Capillaries. J Appl Physiol 1957, 11, 290-302.

5. Zoia, M. C.; D'Armini, A. M.; Beccaria, M.; Corsico, A.; Fulgoni, P.; Klersy, C.; Piovella, F.; Vigano, M.; Cerveri, I.; Pavia Thromboendarterectomy, G., Mid Term Effects of Pulmonary Thromboendarterectomy on Clinical and Cardiopulmonary Function Status. Thorax 2002, 57, 608-12.

Figures

Using the equation from Roughton and Forster (top right), we propose that the mean value of ¹²⁹Xe barrier uptake and RBC transfer (normalized by reference values from healthy subjects) reflect membrane conductance (D_M, proportional to GX surface area and interstitial tissue thickness) and capillary blood volume (V_c) respectively (left). This allows us to use ¹²⁹Xe measurements to estimate V_A, K_CO, and subsequently DL_CO (bottom right).

^99mTc perfusion projections and ¹²⁹Xe RBC images from a PAH (top left) and a CTEPH patient (bottom left). Spatial agreement (green arrow) and mismatch (red arrow) of the 2 image modalities are marked. The 6-zone analysis is illustrated on the anterior view of the PAH patient. The table (top right) summarizes limits of agreement of the functional contribution from each lung region, calculated using 6-zone analysis. The diagram (bottom right) proposes a model to potentially explain the regional mismatch on the CTEPH patient.

Correlation of image estimated GX values vs. clinical measurements. The 3D surface (left) represents the image estimated K_CO as it depends on Bar_REL and RBC_REL, while the dots show the clinically measured values. The red dashed line denotes underestimated values while the green dashed line shows overestimated values. The graphs (right) show correlation between image estimated V_A and DL_CO values, demonstrating a strong correlation across a wide range of pathology.

GX images of representative subjects and their measured/estimated DL_CO values. The estimated DL_CO was calculated from ventilated volume VV, relative mean barrier percentage (Bar_REL), and RBC percentage (RBC_REL). A combination of reduced VV, diminished barrier uptake, escalated barrier uptake, and decreased RBC transfer resulted in the reduced DL_CO in all patients on the panel.

Proc. Intl. Soc. Mag. Reson. Med. 28 (2020)

0443