- Email: [email protected]
Normal Tissue Quantitative T1 and T2∗ MRI Relaxation Time Responses to Hypercapnic and Hyperoxic Gases
Normal Tissue Quantitative T1 and T2* MRI Relaxation Time Responses to Hypercapnic and Hyperoxic Gases Jeff D. Winter, PhD, Marvin Estrada, RVT, Hai-Ling Margaret Cheng, PhD Rationale and Objectives: Longitudinal (T1) and effective transverse (T2*) magnetic resonance (MR) relaxation times provide noninvasive measures of tissue oxygenation. The objective for this study was to quantify independent effects of inhaled O2 and CO2 on normal tissue T1 and T2* in rabbit liver, kidney, and paraspinal muscle. Materials and Methods: Three gas challenges (100% O2, 10% CO2 [balance air], and carbogen [90% O2 + 10% CO2]) were delivered to the rabbits in random order to isolate the effects of inspired O2 and CO2. During each challenge, quantitative T1 and T2* maps were collected on a 1.5 Tesla MR imaging. Mean changes in T1 (DT1) and T2* (DT2*) were calculated from regions of interest in each organ. Results: Greatest DT1 and DT2* changes were observed in liver for 10% CO2 and in kidney for 100% O2. DT1 and DT2* generally followed predicted patterns when transitioning from air breathing: lower T1/higher T2* with inspired O2, higher T1/lower T2* with inspired CO2, and variable T1/T2* changes in the presence of both (ie, carbogen). New observations also emerged: 1) between-gas-challenge transitions revealed the greatest significance in DT2* for the liver and kidney resulting from the isolation of independent O2 and CO2 effects; 2) DT2* provided the best sensitivity and detected both tissue oxygenation and blood volume modulation; and 3) DT1 sensitivity was restricted mainly to tissue oxygenation in the absence of counteracting vasodilatation. Conclusion: Robust use of MR relaxation times as noninvasive biomarkers requires an understanding of their relative sensitivity to organspecific physiological responses. Key Words: T1 relaxation time; T2* relaxation time; hypercapnia; hyperoxia; tissue oxygen. ªAUR, 2011
N
oninvasive magnetic resonance (MR) measures of longitudinal (T1) and effective transverse (T2*) relaxation times have demonstrated potential as indirect markers of oxygenation in the brain (1) and peripheral tissues (2–4). Typically, changes in MR relaxation times are measured during experimental manipulation of the fraction of inspired O2 (FiO2) and CO2 (FiCO2) using one or more gas transitions. The manipulation of inspired gases may be directed to improve tissue oxygenation (eg, augmenting tumor oxygen for cancer therapy) or may also be applied to alter blood flow and volume, so as to provide a means for evaluating vasoreactivity (eg, brain). These concomitant changes in tissue oxygen and perfusion are known to exert independent, possibly opposing influences on MR relaxation times. Therefore, understanding their
Acad Radiol 2011; 18:1159–1167 The Research Institute and Diagnostic Imaging, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada, M5G 1X8 (J.D.W., M.E., H.-L.M.C.); Department of Medical Biophysics, University of Toronto, Toronto, Canada (H.-L.M.C.). Received December 18, 2010; accepted April 29, 2011. Funding: SickKids Foundation/Institute for Human Development, Child and Youth Health—CIHR. Address correspondence to: H.-L.M.C. e-mail: [email protected] ªAUR, 2011 doi:10.1016/j.acra.2011.04.016
independent effects is critical for proper interpretation of MR measurements. For most extracranial tissues or tumors, our understanding of these physiological responses and related MR measurements is relatively poor compared to that in the brain. Investigations in normal tissue have been particularly limited (2,3) but are essential in their own right and for cancer applications, because physiological differences in the normal host organ response may provide insight into the inconsistent MR results reported in tumors to date. Oxygen-related T1 changes are postulated to stem primarily from molecular O2 dissolved in blood plasma and intra- and extracellular fluids (2,4). Increasing the FiO2 will progressively increase O2 dissolved in blood plasma, because the total amount of O2 bound to hemoglobin (ie, the arterial oxygen saturation [SaO2]) is normally close to 100% and will be relatively unaffected. As molecular O2 is weakly paramagnetic, its presence effectively shortens T1. The expected T1 reduction, however, has been consistently observed only in some tissues, such as spleen and myocardium (2–5). Other tissues, such as liver, kidney, and muscle, have shown smaller T1 reductions (2,3) or none at all (4,5). T2* changes originate from local field inhomogeneities generated by paramagnetic deoxy-hemoglobin (Hb). Increasing the FiO2 allows a greater fraction of O2 extraction 1159
WINTER ET AL
from blood plasma than from oxyhemoglobin (HbO2), which effectively increases the venous ratio of HbO2-to-Hb. Because HbO2 is diamagnetic, a greater venous fraction of HbO2 increases T2*. Increasing the FiCO2 will also affect T2* because the presence of CO2 in blood causes a ‘‘right shift’’ in the oxygen dissociation curve (Bohr effect), which increases the level of Hb and thus decreases T2*. In the brain, consistent T2*-weighted imaging signal changes during hyperoxia are well established (1). Only a few studies have reported T2* in peripheral organs, in which investigators found that T2* was unaffected by increasing FiO2 (3,6), but decreased with increasing FiCO2, as expected (3). The rationale for gas challenges with altered inspired O2 and CO2 levels, and our current understanding of the associated physiological changes, can be traced back to cancer treatment strategies. Increasing FiO2 was initially proposed to improve tumor oxygenation as a means to reduce the required radiation dose, since x- and g-ray radiation depends on the presence of intracellular molecular O2 to generate O2 free radicals (7). To mitigate an undesired vasoconstriction that O2 was believed to yield, researchers proposed increasing FiCO2 by employing carbogen (90–99% O2, 1–10% CO2) instead of 100% O2. For carbogen, the O2 component is expected to decrease T1 and increase T2*, as discussed earlier, whereas the CO2 component is expected to increase T1 and decrease T2*, owing to augmented blood flow and concomitantly increased blood volume. The latter is reinforced by both higher amounts of Hb in an imaging voxel and the Bohr effect. Obviously, when using manipulations that invoke concurrent perfusion changes, T1 and T2* relaxation times are difficult to interpret as biomarkers of strictly tissue oxygenation. Previous studies have examined changes in T1 and T2* during inhalation of carbogen and 100% O2 in normal abdominal tissues (2,3). To date, however, the organ-specific and independent effects of FiO2 and FiCO2 on MR relaxation times are still unclear. In this study, our aim was to assess T1 and T2* response to alterations in tissue oxygenation and blood flow invoked by different combinations of FiO2 and FiCO2 in normal abdominal tissues in the rabbit, including liver, kidney, and paraspinal muscle. Three gas challenges, namely, 100% O2, 10% CO2 (balance air), and carbogen (90% O2 + 10% CO2), were applied in a randomized order to isolate the independent effects of FiO2 and FiCO2. MATERIALS AND METHODS This study was approved by our institutional animal care committee (protocol #8784), and all procedures were conducted in accordance with the Canadian Council on Animal Care. MR imaging (MRI) was performed on five New Zealand white rabbits (4.0–4.5 kg) in six separate sessions, with imaging performed twice on one animal. Repeat imaging performed in the one rabbit was conducted to investigate the effects of different gas order delivery in a single rabbit, not to assess reproducibility. 1160
Academic Radiology, Vol 18, No 9, September 2011
Gas Challenges
The four inspired gases delivered to the rabbits were: room air (21% O2, balance N2), 10% CO2 (balance air), 100% O2, and carbogen (10% CO2 + 90% O2). Each inspired gas challenge was delivered for approximately 10 minutes. Unlike previous studies that employ a fixed order of gas transitions, the current study randomized the order to assess whether consistent changes in MR relaxation times were obtained from the same initial and end states (Fig 1). Two of the rabbits were studied on a separate day in which each of the inspired gas challenges was delivered in 10-minute blocks. Blood gas samples were extracted from a cannulated ear artery at the end of each inspired gas challenge and analyzed on a blood gas analyzer (Radiometer ABL 700 Series, Block Scientific, Inc., Bohemia, NY). The overall goal of altering FiO2 and FiCO2 was to study their organ-specific independent effects on measured changes in T1 and T2*, ultimately providing better insight into the physiological responses these MR measurements can robustly detect. Table 1 summarizes the hypothesized physiological changes between each of the gas challenges along with the expected direction of T1 and T2* changes. Animal Preparation
Rabbits were induced with 5% isoflurane delivered by a face mask, and a catheter was inserted into the ear vein to maintain hydration (4 mL/kg/h 0.9% NaCl). A laryngeal mask was inserted into the trachea to create a tight seal and minimize dead space for the MRI experiments. Anesthesia was maintained at 1%. The rabbits were spontaneously breathing room air (21% O2, balance N2) before the start of the gas challenges, and the heart rate and SaO2 were continuously monitored using a pulse oximeter. MRI All MR data were collected on a 1.5 T GE scanner (Signa EXCITE TwinSpeed; General Electric Healthcare, Milwaukee, WI) using a transmit/receive quadrature knee coil. Rabbits were scanned feet first and prone, with coronal imaging slices positioned to encompass the kidney, liver, and paraspinal muscle. Quantitative T1 values were generated using a series of three-dimensional fast spoiled gradient recalled (SPGR) echo scans with three different flip angles (FA) (8). Imaging parameters included: repetition time (TR) = 7.2 ms, echo time (TE) = 3.1 ms, FA = 2, 10 and 21 , readout bandwidth (BW) = 31.2 kHz, field-of-view (FOV) = 160 mm, matrix = 256 160, slice thickness (SLTH) = 3 mm, slice spacing (SLSP) = 0 mm, number of slices (NSL) = 10, number of averages (NAVG) = 4. Quantitative T2* measurements were generated using a two-dimensional multiecho gradient-echo sampling of the T2* decay. Imaging parameters included: TR = 100 ms, 16 equally spaced TEs = 2.1–47.1 ms, FA = 30 , BW = 83.3 kHz, FOV = 160 mm, matrix = 256 192, SLTH = 3 mm and SLSP = 1 mm, and NSL = 6, and NAVG = 4.
Academic Radiology, Vol 18, No 9, September 2011
Figure 1. Inspired gas levels used for magnetic resonance imaging experiments. Each block represents 10 minutes within which there was a 4-minute period allotted for stabilization followed by quantitative T2* (3 minutes) and T1 (3 minutes) imaging.
All image postprocessing of the quantitative T1 and T2* data was performed on a pixel-wise basis using in-house scripts developed in Matlab (V.7.0, Mathworks Inc., Natick, MA). T1 parameter maps were generated using the signal equation for the SPGR steady state magnetization with analytical-based flip angle correction using B1 field maps acquired separately (8). T2* parameter maps were computed by fitting T2* signal intensity versus TE to a monoexponential signal decay function (9). Data Analysis
Regions of interest (ROI) were outlined in three major tissue types: paraspinal muscle (one to two regions), liver (two to three regions), and kidney cortex (one to three regions), as illustrated in Figures 2 and 3. The multiple ROIs outlined on different slices in each tissue type were averaged to provide a mean value for each tissue. Absolute T1 and T2* measurements across all animals were assessed using a Kruskal-Wallis nonparametric test for each separate tissue type, with the inspired gas mixture as the between-group factor. Post-hoc Tukey-Kramer tests were performed for each inspired gas mixture combination for tissues exhibiting significant differences in the omnibus Kruskal-Wallis test. Changes in relaxation times, DT1 and DT2*, in each animal were also calculated between different gas challenges (baseline air, 100% O2, 10% CO2, and carbogen). A series of Wilcoxon signed-rank tests were performed to assess whether DT1 and DT2* results were significantly different from the null hypothesis (zero).
T1 AND T2* RESPONSE TO O2 AND CO2
Wallis test only revealed a statistically significant effect between the different inspired gases for kidney T2*, mainly because of considerable scatter in baseline measurements that arose from inter-animal variations in basal physiology. Based on the post-hoc Tukey-Kramer tests, we found that this significant difference was attributed to differences between air and 100% O2 (P < .05) and between 100% O2 and 10% CO2 (P < .05). It is important to bear in mind the expected changes, namely, lower T1/higher T2* on breathing 100% O2 and higher T1/lower T2* on introducing CO2, when interpreting organ-specific responses in Figure 4. Changes in T1 and T2* in individual animals for different gas transitions were then assessed (Figs 5 and 6). Figure 5 considers transitions between air and 100% O2, 10% CO2, or carbogen. Expected trends in DT1 and DT2* were observed for all transitions, particularly in liver and kidney, including: negative DT1/positive DT2* for 100% O2, positive DT1/ negative DT2* for 10% CO2, and equivocal changes for carbogen. Note, in particular, that a significant effect was demonstrated on both DT1 and DT2* for 10% CO2 in liver and for 100% O2 in kidney. Figure 6 considers transitions between the three gas mixtures (ie, 10% CO2 to 100% O2, 10% CO2 to carbogen, and 100% O2 to carbogen). These transitions consider the extremes of expected blood volume and oxygenation changes (10% CO2 to 100% O2) and allow us to tease apart the independent effects of O2 and CO2 in carbogen. Trends observed were as expected: negative DT1/positive DT2* on 10% CO2 to either 100% O2 or carbogen, and positive DT1/negative DT2* on 100% O2 to carbogen. The sensitivity of both T1 and T2* to oxygenation is maximum when the accompanying change in blood volume is expected to be greatest (10% CO2 to 100% O2). As the blood volume change is moderated (10% CO2 to carbogen), only T2* remains slightly sensitive. For transitions in which only blood volume changes were expected to predominate (100% O2 to carbogen), T2* was again more sensitive than T1 and demonstrated significant changes in all tissues. Table 1 summarizes observed DT1/ DT2* for all gas transitions and compares them to expected changes.
DISCUSSION RESULTS Arterial blood sample results for each inspired gas challenge are provided in Table 2. Carbogen generated similar arterial partial pressure of CO2 (paCO2) levels to 10% CO2 and similar arterial partial pressure of O2 (paO2) levels to 100% O2. Representative T1 parametric images from a single subject for each gas challenge are displayed in Figure 2, with the ROIs used in the study overlaid. The corresponding T2* parametric images are provided in Figure 3. Absolute T1 and T2* measurements across all animals for different gas challenges are shown in Figure 4. The Kruskal-
To date, only a small number of investigations have characterized T1 or T2* changes in healthy peripheral organs, all of which employ 100% O2 or carbogen gas challenges applied in a fixed order (2–6,10). In the current study, we investigated T1 and T2* changes during different transitions between air, 100% O2, 10% CO2, and carbogen in abdominal organs of the rabbit. With this study design, we elucidated organ-specific independent effects of FiO2 and FiCO2 and determined the sensitivity of T1 and T2* changes to the different physiological response mechanisms. In most cases, T1 decreased with elevated FiO2 (resulting from higher PaO2) and increased with elevated FiCO2 (resulting from 1161
WINTER ET AL
Academic Radiology, Vol 18, No 9, September 2011
TABLE 1. Expected and Observed T1 and T2* Changes for Different Inspired Gas Transitions Expected Changes in MR Relaxation Times
Gas Transition
PaO2
BF
Hb/(HbO2 + Hb)
Air / 100% O2
[
Y
Y
Y from increased PaO2, possibly reinforced by decreased blood volume.
Air / 10% CO2
NC
[
[
[ from increased blood volume
Air / carbogen
[
[
?
Unclear because of competing effects of PaO2 and blood volume
10% CO2 / 100% O2
[
Y
Y
Y from increased PaO2, reinforced by decreased blood volume
10% CO2 / carbogen
[
Y
Y
Y from increased PaO2, possibly reinforced by decreased blood volume
100% O2 / carbogen
NC
[
[
[ from increased blood volume, with no change in PaO2.
DT1
Observed Changes in MR Relaxation Times
DT2*
DT1
DT2*
[ from decreased Hb related to PaO2, possibly reinforced by decreased blood volume Y from increased Hb related to Bohr effect, reinforced by increased blood volume Unclear because of competing effects of increased PaO2 and Bohr effect on HbO2 and Hb levels [ from decreased Hb related to PaO2 and Bohr effect, reinforced by decreased blood volume [ from decreased Hb related to PaO2, possibly reinforced by decreased blood volume Y from increased Hb related to Bohr effect, reinforced by increased blood volume
Y liver* Y kidney* Y muscle
[ liver [ kidney** [ muscle
[ liver * [ kidney [ muscle
Y liver ** Y kidney › muscle
Y liver [ kidney [ muscle
Y liver* Y kidney Y muscle *
Y liver* Y kidney* Y muscle
[ liver** [ kidney** [ muscle
Y liver Y kidney › muscle
[ livery [ kidneyy [ muscle
[ liver [ kidney* [ muscle
Y liver* Y kidney* Y muscle*
Significance in DT1 DT2*: yP < .10, *P < .05, **P < .01. Observations that did not follow predicted changes are highlighted in bold. BF, blood flow; Hb, deoxyhemoglobin; HbO2, oxyhemoglobin; NC, no change; PaO2, arterial partial pressure of O2. Gases: air (21% O2), 10% CO2 (balance air), carbogen (90% O2 + 10% CO2).
vasodilatation). T2* increased with elevated FiO2 (resulting from lower Hb levels) and decreased with elevated FiCO2 (resulting from higher Hb levels from Bohr effect and vasodilatation). The competing influences from elevated FiO2 and FiCO2 for the transition between air and carbogen generated the most variable results in all organs, which underscores the variable effects from breathing carbogen (11–13). Several new observations also emerged from this study, offering new insight into the sensitivity of T1 and T2* measurements to different physiological mechanisms. First, as biomarkers for tissue O2 and perfusion, changes in T1 and T2* between gas challenges were generally more useful than absolute values due to the large baseline variability between subjects. Second, sensitivity of DT1 to tissue oxygenation was demonstrated in some but not all cases, which corroborates both positive (2,3) and negative (4,5) literature findings of significant T1 changes in liver, kidney, or muscle. Specifically, the expected T1 reduction was significant when higher dissolved oxygen content was reinforced by no 1162
change or a decrease in blood volume (Figs 5 and 6: air or 10% CO2 /100% O2). However, in the presence of both augmented O2 and blood volume (air / carbogen), the T1 change was offset and variable. In cases in which blood volume changes predominated but PaO2 was stable (air / 10% CO2, 100% O2 / carbogen), an expected T1 increase was observed, but this increase was usually insignificant. This last point highlights the poor sensitivity of DT1 to blood volume changes, which is partly due to the insufficient blood flow response required for an appreciable T1 effect, and partly due to the nondistinction between dissolved oxygen in blood and in tissue. Third, in general, DT2* provided greater sensitivity than DT1 for assessing normal tissue responses. Because T2* is uniquely sensitive to the Hb level and not dissolved O2, T2* changes arising from modulations in blood volume would not have extravascular interferences, thus providing a more robust blood volume biomarker compared to DT1. The most significant T2* change was obtained when a lower Hb level was reinforced
Academic Radiology, Vol 18, No 9, September 2011
T1 AND T2* RESPONSE TO O2 AND CO2
Figure 2. Representative T1 parametric maps for a single rabbit at each of the four inspired gas challenges: (a) air (21% O2), (b) 100% O2, (c) 10% CO2 (balance air), and (d) carbogen (90% O2 + 10% CO2). In each frame, two T1 images from different spatial locations are displayed side by side and divided by a black line. The regions of interest used to analyze the magnetic resonance imaging data in the liver, kidney, and paraspinal muscle are overlaid.
by decreased blood volume (air or 10% CO2/ 100% O2) or vice versa (air / 10% CO2, 100% O2 / carbogen). A smaller but marginally significant DT2* persisted even with a smaller reinforcement from altered blood volume (10% CO2 / carbogen). The greatest DT2* variability existed when Hb level and blood volume changes counteracted (air / carbogen). In virtually all cases where DT2* was significant, we observed greater sensitivity than DT1, which suggests that even though DT1 may be more specific, DT2* based on Hb levels may provide a more robust indicator of tissue oxygen status. Different organs in the body are known to exert distinct responses in perfusion and oxygenation as a result of manipulating FiO2 and FiCO2. Their behavior is not as well studied as in the brain, where tight controls always maintain adequate perfusion and oxygenation and generate predictable patterns of T2* and blood flow response (14). The unique behavior of MR relaxation time changes in each organ is described in the following.
The liver, physiologically speaking, is the most unique of the three tissues studied, owing to high vascularity, dual input of the hepatic artery and portal vein and need to maintain homeostasis (similar to the brain). Inhalation of 10% CO2 generated the greatest changes in T2* compared to other gases and to other organs investigated. These large T2* changes likely reflect the highly vascularized nature of the liver. The significant T2* drop and T1 increase, is likely attributed to elevated perfusion measurements that reflect increased blood volume, as previously reported in rat liver (15). A potential contribution to increased Hb-to-HbO2 ratio may stem from the hepatic arterial buffer response, in which arterial hepatic blood flow is adjusted to offset variations in the portal venous blood flow to maintain control of total blood flow to the liver (16). During hypercapnia (paCO2 = 70 mm Hg), researchers previously found that the hepatic arterial flow decreased 43%, the portal venous flow decreased 15%, and total O2 consumption did not vary considerably (17). This lowered arterial-tovenous blood flow ratio likely contributes to an observed 1163
WINTER ET AL
Academic Radiology, Vol 18, No 9, September 2011
Figure 3. Representative T2* parametric maps for a single rabbit at each of the four inspired gas challenges: (a) air (21% O2), (b) 100% O2, (c) 10% CO2 (balance air), and (d) carbogen (90% O2 + 10% CO2). In each frame, two T2* images from different spatial locations are displayed side by side and divided by a black line. The regions of interest used to analyze the magnetic resonance imaging data in the liver, kidney, and paraspinal muscle are overlaid.
TABLE 2. Arterial Blood Gas Measurements Gas Challenge
tHb (g/dL)
pH
PaCO2 (mm Hg)
PaO2 (mm Hg)
SaO2 (%)
Air 10% CO2 100% O2 carbogen
12.0 1.0 12.5 0.6 11.8 0.1 11.7 0.4
7.440 0.001 7.214 0.014 7.401 0.011 7.215 0.030
34.4 2.9 70.4 3.1 40.2 4.5 69.3 6.9
71.7 2.3 86.5 6.3 454 12 445 31
99.7 2.1 98.2 1.2 105.0 2.8 104.6 2.8
Gases: air (21% O2), 10% CO2 (balance air), carbogen (90% O2 + 10% CO2). PaCO2, arterial partial pressure of CO2; PaO2, arterial partial pressure of O2; SaO2, arterial oxygen saturation; tHb, total hemoglobin. Mean SD (n = 2).
overall increase in the Hb-to-HbO2, as observed in the T2* decrease in the current study. Changes in T1, a measure of tissue pO2, were smaller compared to those the kidney. We also noted the unanticipated response to 100% O2, in which 1164
changes in T1, and T2* were consistently minimal (Fig 5). A likely source for this observation may again be the dual hepatic input: since the portal vein brings blood from the small intestine, stomach, pancreas, and spleen to the liver,
Academic Radiology, Vol 18, No 9, September 2011
Figure 4. Absolute magnetic resonance relaxation times (a) T1 and (b) T2* across all animals for each inspired gas challenge: air (21% O2) (closed diamonds), 100% O2 (open circles), 10% CO2 (balance air) (closed circles), and carbogen (90% O2 + 10% CO2) (gray circles). Dash symbol represents the median value for each inspired gas. Significant differences: *P < .05.
the extraction of dissolved O2 in these organs will partially dictate the overall O2 status of the liver. Discrepancy exists in literature reports on 100% O2 inhalation; for example, some human studies have observed T1 shortening and no change in T2* (2,3), whereas others have reported negligible T1 changes in humans (4) and increased hepatic T2* in the rabbit (18). The kidney also represents a unique organ in terms of the vasculature and O2 distribution, as the cortex exhibits considerably higher pO2 and blood flow than the medulla. In our study, we exclusively examined the renal cortex with quantitative MRI measurements. In contrast to the liver, kidney response was greatest to 100% O2 but not 10% CO2. This can be appreciated on both T1 and T2* changes (Fig 5). Again, there are several notable comparisons to previous literature reports. While some studies have observed changes in the blood oxygen level-dependent (BOLD) signal intensity for 100% O2 breathing (10), other studies have failed to observe quantitative T2* changes (3,6). In muscle, T2* changes were very small and T1 changes were variable. Literature reports have been similarly disparate, with some also failing to demonstrate T2* changes (3) or T1 shortening during 100% O2 breathing (4,5), whereas others found significant T1 shortening during hyperoxia (2,3,19).
T1 AND T2* RESPONSE TO O2 AND CO2
Figure 5. Changes in magnetic resonance relaxation times (a) DT1 and (b) DT2* in individual animals for inspired gas transitions between air (21% O2) and: 100% O2 (open circles), 10% CO2 (balance air) (closed circles), and carbogen (90% O2 + 10% CO2) (gray circles). Dash symbol represents the median value for each gas transition. Significant differences: * P < .05, ** P < .01.
Previous work has shown insignificant blood flow changes in normal rabbit muscle during hypercapnia, as assessed using computed tomography perfusion (20). A small blood flow response in muscle is most likely related to low baseline blood flow levels compared with other abdominal organs, such that increases in PaCO2 have little effect. This low baseline compromises the ability of even T2* to detect flow changes. A previous study showed that even when blood flow was increased 10-fold during hyperemia induced by ischemia-reperfusion in human muscle, T2* increases were very small (<4 ms) (21). Use of MR relaxation times is one of numerous approaches for assessing pO2 (22). An alternative method is positron emission tomography (PET) using 18F-labeled fluoromisonidazole (18F-FMISO), a label that freely diffuses throughout cells under normal conditions but binds with intracellular macromolecules under hypoxia. Recent work has characterized the pharmacokinetics of 18F-FMISO PET in head and neck tumors (23). However, the parameters generated are not confirmed surrogates for hypoxia, and this method involves radiation. Yet another approach to quantify pO2 is 19F MRI, which uses the reporter molecular hexafluorobenzene to provide absolute values of tissue pO2 (24,25). In tumors, 1165
WINTER ET AL
Academic Radiology, Vol 18, No 9, September 2011
MRI data acquisitions to further reduce the effects of motion on the quantitative T1 and T2* estimates. In conclusion, changes in T2* and T1 are useful for monitoring organ-specific tissue pO2 and perfusion responses to gas challenges. Changes in T2* provided the best sensitivity, being able to detect both tissue oxygenation and blood volume modulation. The sensitivity of T1 was restricted mainly to tissue oxygenation and only in the absence of counteractive effects from vasodilatation. Although T2* and T1 changes can be predicted based on the gas transition, robust use of MR relaxation times as noninvasive biomarkers requires an understanding of their relative sensitivity to organ-specific physiological responses. Future studies focused on tissuespecific T1 and T2* responses to gas transitions with larger sample sizes are needed to confirm observations made in this study. REFERENCES
Figure 6. Changes in magnetic resonance relaxation times (a) DT1 and (b) DT2* in individual animals for inspired gas transitions: 10% CO2 (balance air) to 100% O2 (open circles), 10% CO2 to carbogen (90% O2 + 10% CO2) (gray circles), and 100% O2 to carbogen (closed circles). DT2* and DT1 were negated for gas transitions in the opposite direction. Dash symbol represents the median value for each gas transition. Significant differences: y P < .10, * P < .05, ** P < .01.
19
F MRI estimates were correlated to BOLD signal changes but not measures of T2* (25). Despite the key advantage of 19 F MRI, namely, providing absolute pO2 measurements, this method is limited by the need to directly inject reporter molecules into tissue, limited availability of the resources, and limited applications outside of animal models. For clinical applications, quantitative MR relaxation times remains an appealing surrogate measure for assessing tissue pO2 because of the wide availability of clinical MRI scanners, no need for injection of exogenous agents, and the ability to perform repeated measures. One limitation of the current study was the potential effects of anesthetic on the blood flow response to FiCO2 manipulation. Although we applied a low dose of isoflurane, its vasodilatory properties may modulate potential vascular responses to changes in the FiCO2. To provide greater relative perfusion response to FiCO2 increases, we selected a FiCO2 of 10% CO2, which is greater than the 5% CO2 typically used. Imaging abdominal organs also presents a challenge in terms of motion artifacts, which may degrade estimates of the relaxation time measurements. In the current study, we optimized the rabbit positioning and support to minimize respiratory motion and we also employed multiple averages for the 1166
1. Prisman E, Slessarev M, Han J, et al. Comparison of the effects of independently-controlled end-tidal PCO(2) and PO(2) on blood oxygen level-dependent (BOLD) MRI. J Magn Reson Imaging 2008; 27: 185–191. 2. O’Connor JP, Jackson A, Buonaccorsi GA, et al. Organ-specific effects of oxygen and carbogen gas inhalation on tissue longitudinal relaxation times. Magn Reson Med 2007; 58:490–496. 3. O’Connor JP, Naish JH, Jackson A, et al. Comparison of normal tissue R1 and R*2 modulation by oxygen and carbogen. Magn Reson Med 2009; 61: 75–83. 4. Tadamura E, Hatabu H, Li W, et al. Effect of oxygen inhalation on relaxation times in various tissues. J Magn Reson Imaging 1997; 7:220–225. 5. Noseworthy MD, Kim JK, Stainsby JA, et al. Tracking oxygen effects on MR signal in blood and skeletal muscle during hyperoxia exposure. J Magn Reson Imaging 1999; 9:814–820. 6. Jones RA, Ries M, Moonen CT, et al. Imaging the changes in renal T1 induced by the inhalation of pure oxygen: a feasibility study. Magn Reson Med 2002; 47:728–735. 7. Kaanders JH, Bussink J, van der Kogel AJ. ARCON: a novel biology-based approach in radiotherapy. Lancet Oncol 2002; 3:728–737. 8. Cheng HL, Wright GA. Rapid high-resolution T(1) mapping by variable flip angles: accurate and precise measurements in the presence of radiofrequency field inhomogeneity. Magn Reson Med 2006; 55: 566–574. 9. Beaumont M, Odame I, Babyn PS, et al. Accurate liver T2 measurement of iron overload: a simulations investigation and in vivo study. J Magn Reson Imaging 2009; 30:313–320. 10. Boss A, Martirosian P, Jehs MC, et al. Influence of oxygen and carbogen breathing on renal oxygenation measured by T2*-weighted imaging at 3.0 T. NMR Biomed 2009; 22:638–645. 11. Lanzen JL, Braun RD, Ong AL, et al. Variability in blood flow and pO2 in tumors in response to carbogen breathing. Int J Radiat Oncol Biol Phys 1998; 42:855–859. 12. Thews O, Kelleher DK, Vaupel P. Dynamics of tumor oxygenation and red blood cell flux in response to inspiratory hyperoxia combined with different levels of inspiratory hypercapnia. Radiother Oncol 2002; 62:77–85. 13. Neeman M, Dafni H, Bukhari O, et al. In vivo BOLD contrast MRI mapping of subcutaneous vascular function and maturation: validation by intravital microscopy. Magn Reson Med 2001; 45:887–898. 14. Lu J, Dai G, Egi Y, et al. Characterization of cerebrovascular responses to hyperoxia and hypercapnia using MRI in rat. Neuroimage 2009; 45: 1126–1134. 15. Barash H, Gross E, Matot I, et al. Functional MR imaging during hypercapnia and hyperoxia: noninvasive tool for monitoring changes in liver perfusion and hemodynamics in a rat model. Radiology 2007; 243: 727–735. 16. Lautt WW. Mechanism and role of intrinsic regulation of hepatic arterial blood flow: hepatic arterial buffer response. Am J Physiol 1985; 249(5 Pt 1):G549–G556.
Academic Radiology, Vol 18, No 9, September 2011
17. Chang C, Uchiyama A, Ma L, et al. A comparison of the effects on respiratory carbon dioxide response, arterial blood pressure, and heart rate of dexmedetomidine, propofol, and midazolam in sevoflurane-anesthetized rabbits. Anesth Analg 2009; 109:84–89. 18. Rhee TK, Larson AC, Prasad PV, et al. Feasibility of blood oxygenation level-dependent MR imaging to monitor hepatic transcatheter arterial embolization in rabbits. J Vasc Interv Radiol 2005; 16:1523–1528. 19. McGrath DM, Naish JH, O’Connor JP, et al. Oxygen-induced changes in longitudinal relaxation times in skeletal muscle. Magn Reson Imaging 2008; 26:221–227. 20. Purdie TG, Lee TY. Carbon dioxide reactivity of computed tomography functional parameters in rabbit VX2 soft tissue tumour. Phys Med Biol 2003; 48:849–860. 21. Lebon V, Carlier PG, Brillault-Salvat C, et al. Simultaneous measurement of perfusion and oxygenation changes using a multiple gradient-echo
T1 AND T2* RESPONSE TO O2 AND CO2
22.
23.
24.
25.
sequence: application to human muscle study. Magn Reson Imaging 1998; 16:721–729. Tatum JL, Kelloff GJ, Gillies RJ, et al. Hypoxia: importance in tumor biology, noninvasive measurement by imaging, and value of its measurement in the management of cancer therapy. Int J Radiat Biol 2006; 82: 699–757. Wang W, Lee NY, Georgi JC, et al. Pharmacokinetic analysis of hypoxia (18)F-fluoromisonidazole dynamic PET in head and neck cancer. J Nucl Med 2010; 51:37–45. Jordan BF, Cron GO, Gallez B. Rapid monitoring of oxygenation by 19F magnetic resonance imaging: simultaneous comparison with fluorescence quenching. Magn Reson Med 2009; 61:634–638. Zhao D, Jiang L, Hahn EW, et al. Comparison of 1H blood oxygen leveldependent (BOLD) and 19F MRI to investigate tumor oxygenation. Magn Reson Med 2009; 62:357–364.
1167
N
oninvasive magnetic resonance (MR) measures of longitudinal (T1) and effective transverse (T2*) relaxation times have demonstrated potential as indirect markers of oxygenation in the brain (1) and peripheral tissues (2–4). Typically, changes in MR relaxation times are measured during experimental manipulation of the fraction of inspired O2 (FiO2) and CO2 (FiCO2) using one or more gas transitions. The manipulation of inspired gases may be directed to improve tissue oxygenation (eg, augmenting tumor oxygen for cancer therapy) or may also be applied to alter blood flow and volume, so as to provide a means for evaluating vasoreactivity (eg, brain). These concomitant changes in tissue oxygen and perfusion are known to exert independent, possibly opposing influences on MR relaxation times. Therefore, understanding their
Acad Radiol 2011; 18:1159–1167 The Research Institute and Diagnostic Imaging, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada, M5G 1X8 (J.D.W., M.E., H.-L.M.C.); Department of Medical Biophysics, University of Toronto, Toronto, Canada (H.-L.M.C.). Received December 18, 2010; accepted April 29, 2011. Funding: SickKids Foundation/Institute for Human Development, Child and Youth Health—CIHR. Address correspondence to: H.-L.M.C. e-mail: [email protected] ªAUR, 2011 doi:10.1016/j.acra.2011.04.016
independent effects is critical for proper interpretation of MR measurements. For most extracranial tissues or tumors, our understanding of these physiological responses and related MR measurements is relatively poor compared to that in the brain. Investigations in normal tissue have been particularly limited (2,3) but are essential in their own right and for cancer applications, because physiological differences in the normal host organ response may provide insight into the inconsistent MR results reported in tumors to date. Oxygen-related T1 changes are postulated to stem primarily from molecular O2 dissolved in blood plasma and intra- and extracellular fluids (2,4). Increasing the FiO2 will progressively increase O2 dissolved in blood plasma, because the total amount of O2 bound to hemoglobin (ie, the arterial oxygen saturation [SaO2]) is normally close to 100% and will be relatively unaffected. As molecular O2 is weakly paramagnetic, its presence effectively shortens T1. The expected T1 reduction, however, has been consistently observed only in some tissues, such as spleen and myocardium (2–5). Other tissues, such as liver, kidney, and muscle, have shown smaller T1 reductions (2,3) or none at all (4,5). T2* changes originate from local field inhomogeneities generated by paramagnetic deoxy-hemoglobin (Hb). Increasing the FiO2 allows a greater fraction of O2 extraction 1159
WINTER ET AL
from blood plasma than from oxyhemoglobin (HbO2), which effectively increases the venous ratio of HbO2-to-Hb. Because HbO2 is diamagnetic, a greater venous fraction of HbO2 increases T2*. Increasing the FiCO2 will also affect T2* because the presence of CO2 in blood causes a ‘‘right shift’’ in the oxygen dissociation curve (Bohr effect), which increases the level of Hb and thus decreases T2*. In the brain, consistent T2*-weighted imaging signal changes during hyperoxia are well established (1). Only a few studies have reported T2* in peripheral organs, in which investigators found that T2* was unaffected by increasing FiO2 (3,6), but decreased with increasing FiCO2, as expected (3). The rationale for gas challenges with altered inspired O2 and CO2 levels, and our current understanding of the associated physiological changes, can be traced back to cancer treatment strategies. Increasing FiO2 was initially proposed to improve tumor oxygenation as a means to reduce the required radiation dose, since x- and g-ray radiation depends on the presence of intracellular molecular O2 to generate O2 free radicals (7). To mitigate an undesired vasoconstriction that O2 was believed to yield, researchers proposed increasing FiCO2 by employing carbogen (90–99% O2, 1–10% CO2) instead of 100% O2. For carbogen, the O2 component is expected to decrease T1 and increase T2*, as discussed earlier, whereas the CO2 component is expected to increase T1 and decrease T2*, owing to augmented blood flow and concomitantly increased blood volume. The latter is reinforced by both higher amounts of Hb in an imaging voxel and the Bohr effect. Obviously, when using manipulations that invoke concurrent perfusion changes, T1 and T2* relaxation times are difficult to interpret as biomarkers of strictly tissue oxygenation. Previous studies have examined changes in T1 and T2* during inhalation of carbogen and 100% O2 in normal abdominal tissues (2,3). To date, however, the organ-specific and independent effects of FiO2 and FiCO2 on MR relaxation times are still unclear. In this study, our aim was to assess T1 and T2* response to alterations in tissue oxygenation and blood flow invoked by different combinations of FiO2 and FiCO2 in normal abdominal tissues in the rabbit, including liver, kidney, and paraspinal muscle. Three gas challenges, namely, 100% O2, 10% CO2 (balance air), and carbogen (90% O2 + 10% CO2), were applied in a randomized order to isolate the independent effects of FiO2 and FiCO2. MATERIALS AND METHODS This study was approved by our institutional animal care committee (protocol #8784), and all procedures were conducted in accordance with the Canadian Council on Animal Care. MR imaging (MRI) was performed on five New Zealand white rabbits (4.0–4.5 kg) in six separate sessions, with imaging performed twice on one animal. Repeat imaging performed in the one rabbit was conducted to investigate the effects of different gas order delivery in a single rabbit, not to assess reproducibility. 1160
Academic Radiology, Vol 18, No 9, September 2011
Gas Challenges
The four inspired gases delivered to the rabbits were: room air (21% O2, balance N2), 10% CO2 (balance air), 100% O2, and carbogen (10% CO2 + 90% O2). Each inspired gas challenge was delivered for approximately 10 minutes. Unlike previous studies that employ a fixed order of gas transitions, the current study randomized the order to assess whether consistent changes in MR relaxation times were obtained from the same initial and end states (Fig 1). Two of the rabbits were studied on a separate day in which each of the inspired gas challenges was delivered in 10-minute blocks. Blood gas samples were extracted from a cannulated ear artery at the end of each inspired gas challenge and analyzed on a blood gas analyzer (Radiometer ABL 700 Series, Block Scientific, Inc., Bohemia, NY). The overall goal of altering FiO2 and FiCO2 was to study their organ-specific independent effects on measured changes in T1 and T2*, ultimately providing better insight into the physiological responses these MR measurements can robustly detect. Table 1 summarizes the hypothesized physiological changes between each of the gas challenges along with the expected direction of T1 and T2* changes. Animal Preparation
Rabbits were induced with 5% isoflurane delivered by a face mask, and a catheter was inserted into the ear vein to maintain hydration (4 mL/kg/h 0.9% NaCl). A laryngeal mask was inserted into the trachea to create a tight seal and minimize dead space for the MRI experiments. Anesthesia was maintained at 1%. The rabbits were spontaneously breathing room air (21% O2, balance N2) before the start of the gas challenges, and the heart rate and SaO2 were continuously monitored using a pulse oximeter. MRI All MR data were collected on a 1.5 T GE scanner (Signa EXCITE TwinSpeed; General Electric Healthcare, Milwaukee, WI) using a transmit/receive quadrature knee coil. Rabbits were scanned feet first and prone, with coronal imaging slices positioned to encompass the kidney, liver, and paraspinal muscle. Quantitative T1 values were generated using a series of three-dimensional fast spoiled gradient recalled (SPGR) echo scans with three different flip angles (FA) (8). Imaging parameters included: repetition time (TR) = 7.2 ms, echo time (TE) = 3.1 ms, FA = 2, 10 and 21 , readout bandwidth (BW) = 31.2 kHz, field-of-view (FOV) = 160 mm, matrix = 256 160, slice thickness (SLTH) = 3 mm, slice spacing (SLSP) = 0 mm, number of slices (NSL) = 10, number of averages (NAVG) = 4. Quantitative T2* measurements were generated using a two-dimensional multiecho gradient-echo sampling of the T2* decay. Imaging parameters included: TR = 100 ms, 16 equally spaced TEs = 2.1–47.1 ms, FA = 30 , BW = 83.3 kHz, FOV = 160 mm, matrix = 256 192, SLTH = 3 mm and SLSP = 1 mm, and NSL = 6, and NAVG = 4.
Academic Radiology, Vol 18, No 9, September 2011
Figure 1. Inspired gas levels used for magnetic resonance imaging experiments. Each block represents 10 minutes within which there was a 4-minute period allotted for stabilization followed by quantitative T2* (3 minutes) and T1 (3 minutes) imaging.
All image postprocessing of the quantitative T1 and T2* data was performed on a pixel-wise basis using in-house scripts developed in Matlab (V.7.0, Mathworks Inc., Natick, MA). T1 parameter maps were generated using the signal equation for the SPGR steady state magnetization with analytical-based flip angle correction using B1 field maps acquired separately (8). T2* parameter maps were computed by fitting T2* signal intensity versus TE to a monoexponential signal decay function (9). Data Analysis
Regions of interest (ROI) were outlined in three major tissue types: paraspinal muscle (one to two regions), liver (two to three regions), and kidney cortex (one to three regions), as illustrated in Figures 2 and 3. The multiple ROIs outlined on different slices in each tissue type were averaged to provide a mean value for each tissue. Absolute T1 and T2* measurements across all animals were assessed using a Kruskal-Wallis nonparametric test for each separate tissue type, with the inspired gas mixture as the between-group factor. Post-hoc Tukey-Kramer tests were performed for each inspired gas mixture combination for tissues exhibiting significant differences in the omnibus Kruskal-Wallis test. Changes in relaxation times, DT1 and DT2*, in each animal were also calculated between different gas challenges (baseline air, 100% O2, 10% CO2, and carbogen). A series of Wilcoxon signed-rank tests were performed to assess whether DT1 and DT2* results were significantly different from the null hypothesis (zero).
T1 AND T2* RESPONSE TO O2 AND CO2
Wallis test only revealed a statistically significant effect between the different inspired gases for kidney T2*, mainly because of considerable scatter in baseline measurements that arose from inter-animal variations in basal physiology. Based on the post-hoc Tukey-Kramer tests, we found that this significant difference was attributed to differences between air and 100% O2 (P < .05) and between 100% O2 and 10% CO2 (P < .05). It is important to bear in mind the expected changes, namely, lower T1/higher T2* on breathing 100% O2 and higher T1/lower T2* on introducing CO2, when interpreting organ-specific responses in Figure 4. Changes in T1 and T2* in individual animals for different gas transitions were then assessed (Figs 5 and 6). Figure 5 considers transitions between air and 100% O2, 10% CO2, or carbogen. Expected trends in DT1 and DT2* were observed for all transitions, particularly in liver and kidney, including: negative DT1/positive DT2* for 100% O2, positive DT1/ negative DT2* for 10% CO2, and equivocal changes for carbogen. Note, in particular, that a significant effect was demonstrated on both DT1 and DT2* for 10% CO2 in liver and for 100% O2 in kidney. Figure 6 considers transitions between the three gas mixtures (ie, 10% CO2 to 100% O2, 10% CO2 to carbogen, and 100% O2 to carbogen). These transitions consider the extremes of expected blood volume and oxygenation changes (10% CO2 to 100% O2) and allow us to tease apart the independent effects of O2 and CO2 in carbogen. Trends observed were as expected: negative DT1/positive DT2* on 10% CO2 to either 100% O2 or carbogen, and positive DT1/negative DT2* on 100% O2 to carbogen. The sensitivity of both T1 and T2* to oxygenation is maximum when the accompanying change in blood volume is expected to be greatest (10% CO2 to 100% O2). As the blood volume change is moderated (10% CO2 to carbogen), only T2* remains slightly sensitive. For transitions in which only blood volume changes were expected to predominate (100% O2 to carbogen), T2* was again more sensitive than T1 and demonstrated significant changes in all tissues. Table 1 summarizes observed DT1/ DT2* for all gas transitions and compares them to expected changes.
DISCUSSION RESULTS Arterial blood sample results for each inspired gas challenge are provided in Table 2. Carbogen generated similar arterial partial pressure of CO2 (paCO2) levels to 10% CO2 and similar arterial partial pressure of O2 (paO2) levels to 100% O2. Representative T1 parametric images from a single subject for each gas challenge are displayed in Figure 2, with the ROIs used in the study overlaid. The corresponding T2* parametric images are provided in Figure 3. Absolute T1 and T2* measurements across all animals for different gas challenges are shown in Figure 4. The Kruskal-
To date, only a small number of investigations have characterized T1 or T2* changes in healthy peripheral organs, all of which employ 100% O2 or carbogen gas challenges applied in a fixed order (2–6,10). In the current study, we investigated T1 and T2* changes during different transitions between air, 100% O2, 10% CO2, and carbogen in abdominal organs of the rabbit. With this study design, we elucidated organ-specific independent effects of FiO2 and FiCO2 and determined the sensitivity of T1 and T2* changes to the different physiological response mechanisms. In most cases, T1 decreased with elevated FiO2 (resulting from higher PaO2) and increased with elevated FiCO2 (resulting from 1161
WINTER ET AL
Academic Radiology, Vol 18, No 9, September 2011
TABLE 1. Expected and Observed T1 and T2* Changes for Different Inspired Gas Transitions Expected Changes in MR Relaxation Times
Gas Transition
PaO2
BF
Hb/(HbO2 + Hb)
Air / 100% O2
[
Y
Y
Y from increased PaO2, possibly reinforced by decreased blood volume.
Air / 10% CO2
NC
[
[
[ from increased blood volume
Air / carbogen
[
[
?
Unclear because of competing effects of PaO2 and blood volume
10% CO2 / 100% O2
[
Y
Y
Y from increased PaO2, reinforced by decreased blood volume
10% CO2 / carbogen
[
Y
Y
Y from increased PaO2, possibly reinforced by decreased blood volume
100% O2 / carbogen
NC
[
[
[ from increased blood volume, with no change in PaO2.
DT1
Observed Changes in MR Relaxation Times
DT2*
DT1
DT2*
[ from decreased Hb related to PaO2, possibly reinforced by decreased blood volume Y from increased Hb related to Bohr effect, reinforced by increased blood volume Unclear because of competing effects of increased PaO2 and Bohr effect on HbO2 and Hb levels [ from decreased Hb related to PaO2 and Bohr effect, reinforced by decreased blood volume [ from decreased Hb related to PaO2, possibly reinforced by decreased blood volume Y from increased Hb related to Bohr effect, reinforced by increased blood volume
Y liver* Y kidney* Y muscle
[ liver [ kidney** [ muscle
[ liver * [ kidney [ muscle
Y liver ** Y kidney › muscle
Y liver [ kidney [ muscle
Y liver* Y kidney Y muscle *
Y liver* Y kidney* Y muscle
[ liver** [ kidney** [ muscle
Y liver Y kidney › muscle
[ livery [ kidneyy [ muscle
[ liver [ kidney* [ muscle
Y liver* Y kidney* Y muscle*
Significance in DT1 DT2*: yP < .10, *P < .05, **P < .01. Observations that did not follow predicted changes are highlighted in bold. BF, blood flow; Hb, deoxyhemoglobin; HbO2, oxyhemoglobin; NC, no change; PaO2, arterial partial pressure of O2. Gases: air (21% O2), 10% CO2 (balance air), carbogen (90% O2 + 10% CO2).
vasodilatation). T2* increased with elevated FiO2 (resulting from lower Hb levels) and decreased with elevated FiCO2 (resulting from higher Hb levels from Bohr effect and vasodilatation). The competing influences from elevated FiO2 and FiCO2 for the transition between air and carbogen generated the most variable results in all organs, which underscores the variable effects from breathing carbogen (11–13). Several new observations also emerged from this study, offering new insight into the sensitivity of T1 and T2* measurements to different physiological mechanisms. First, as biomarkers for tissue O2 and perfusion, changes in T1 and T2* between gas challenges were generally more useful than absolute values due to the large baseline variability between subjects. Second, sensitivity of DT1 to tissue oxygenation was demonstrated in some but not all cases, which corroborates both positive (2,3) and negative (4,5) literature findings of significant T1 changes in liver, kidney, or muscle. Specifically, the expected T1 reduction was significant when higher dissolved oxygen content was reinforced by no 1162
change or a decrease in blood volume (Figs 5 and 6: air or 10% CO2 /100% O2). However, in the presence of both augmented O2 and blood volume (air / carbogen), the T1 change was offset and variable. In cases in which blood volume changes predominated but PaO2 was stable (air / 10% CO2, 100% O2 / carbogen), an expected T1 increase was observed, but this increase was usually insignificant. This last point highlights the poor sensitivity of DT1 to blood volume changes, which is partly due to the insufficient blood flow response required for an appreciable T1 effect, and partly due to the nondistinction between dissolved oxygen in blood and in tissue. Third, in general, DT2* provided greater sensitivity than DT1 for assessing normal tissue responses. Because T2* is uniquely sensitive to the Hb level and not dissolved O2, T2* changes arising from modulations in blood volume would not have extravascular interferences, thus providing a more robust blood volume biomarker compared to DT1. The most significant T2* change was obtained when a lower Hb level was reinforced
Academic Radiology, Vol 18, No 9, September 2011
T1 AND T2* RESPONSE TO O2 AND CO2
Figure 2. Representative T1 parametric maps for a single rabbit at each of the four inspired gas challenges: (a) air (21% O2), (b) 100% O2, (c) 10% CO2 (balance air), and (d) carbogen (90% O2 + 10% CO2). In each frame, two T1 images from different spatial locations are displayed side by side and divided by a black line. The regions of interest used to analyze the magnetic resonance imaging data in the liver, kidney, and paraspinal muscle are overlaid.
by decreased blood volume (air or 10% CO2/ 100% O2) or vice versa (air / 10% CO2, 100% O2 / carbogen). A smaller but marginally significant DT2* persisted even with a smaller reinforcement from altered blood volume (10% CO2 / carbogen). The greatest DT2* variability existed when Hb level and blood volume changes counteracted (air / carbogen). In virtually all cases where DT2* was significant, we observed greater sensitivity than DT1, which suggests that even though DT1 may be more specific, DT2* based on Hb levels may provide a more robust indicator of tissue oxygen status. Different organs in the body are known to exert distinct responses in perfusion and oxygenation as a result of manipulating FiO2 and FiCO2. Their behavior is not as well studied as in the brain, where tight controls always maintain adequate perfusion and oxygenation and generate predictable patterns of T2* and blood flow response (14). The unique behavior of MR relaxation time changes in each organ is described in the following.
The liver, physiologically speaking, is the most unique of the three tissues studied, owing to high vascularity, dual input of the hepatic artery and portal vein and need to maintain homeostasis (similar to the brain). Inhalation of 10% CO2 generated the greatest changes in T2* compared to other gases and to other organs investigated. These large T2* changes likely reflect the highly vascularized nature of the liver. The significant T2* drop and T1 increase, is likely attributed to elevated perfusion measurements that reflect increased blood volume, as previously reported in rat liver (15). A potential contribution to increased Hb-to-HbO2 ratio may stem from the hepatic arterial buffer response, in which arterial hepatic blood flow is adjusted to offset variations in the portal venous blood flow to maintain control of total blood flow to the liver (16). During hypercapnia (paCO2 = 70 mm Hg), researchers previously found that the hepatic arterial flow decreased 43%, the portal venous flow decreased 15%, and total O2 consumption did not vary considerably (17). This lowered arterial-tovenous blood flow ratio likely contributes to an observed 1163
WINTER ET AL
Academic Radiology, Vol 18, No 9, September 2011
Figure 3. Representative T2* parametric maps for a single rabbit at each of the four inspired gas challenges: (a) air (21% O2), (b) 100% O2, (c) 10% CO2 (balance air), and (d) carbogen (90% O2 + 10% CO2). In each frame, two T2* images from different spatial locations are displayed side by side and divided by a black line. The regions of interest used to analyze the magnetic resonance imaging data in the liver, kidney, and paraspinal muscle are overlaid.
TABLE 2. Arterial Blood Gas Measurements Gas Challenge
tHb (g/dL)
pH
PaCO2 (mm Hg)
PaO2 (mm Hg)
SaO2 (%)
Air 10% CO2 100% O2 carbogen
12.0 1.0 12.5 0.6 11.8 0.1 11.7 0.4
7.440 0.001 7.214 0.014 7.401 0.011 7.215 0.030
34.4 2.9 70.4 3.1 40.2 4.5 69.3 6.9
71.7 2.3 86.5 6.3 454 12 445 31
99.7 2.1 98.2 1.2 105.0 2.8 104.6 2.8
Gases: air (21% O2), 10% CO2 (balance air), carbogen (90% O2 + 10% CO2). PaCO2, arterial partial pressure of CO2; PaO2, arterial partial pressure of O2; SaO2, arterial oxygen saturation; tHb, total hemoglobin. Mean SD (n = 2).
overall increase in the Hb-to-HbO2, as observed in the T2* decrease in the current study. Changes in T1, a measure of tissue pO2, were smaller compared to those the kidney. We also noted the unanticipated response to 100% O2, in which 1164
changes in T1, and T2* were consistently minimal (Fig 5). A likely source for this observation may again be the dual hepatic input: since the portal vein brings blood from the small intestine, stomach, pancreas, and spleen to the liver,
Academic Radiology, Vol 18, No 9, September 2011
Figure 4. Absolute magnetic resonance relaxation times (a) T1 and (b) T2* across all animals for each inspired gas challenge: air (21% O2) (closed diamonds), 100% O2 (open circles), 10% CO2 (balance air) (closed circles), and carbogen (90% O2 + 10% CO2) (gray circles). Dash symbol represents the median value for each inspired gas. Significant differences: *P < .05.
the extraction of dissolved O2 in these organs will partially dictate the overall O2 status of the liver. Discrepancy exists in literature reports on 100% O2 inhalation; for example, some human studies have observed T1 shortening and no change in T2* (2,3), whereas others have reported negligible T1 changes in humans (4) and increased hepatic T2* in the rabbit (18). The kidney also represents a unique organ in terms of the vasculature and O2 distribution, as the cortex exhibits considerably higher pO2 and blood flow than the medulla. In our study, we exclusively examined the renal cortex with quantitative MRI measurements. In contrast to the liver, kidney response was greatest to 100% O2 but not 10% CO2. This can be appreciated on both T1 and T2* changes (Fig 5). Again, there are several notable comparisons to previous literature reports. While some studies have observed changes in the blood oxygen level-dependent (BOLD) signal intensity for 100% O2 breathing (10), other studies have failed to observe quantitative T2* changes (3,6). In muscle, T2* changes were very small and T1 changes were variable. Literature reports have been similarly disparate, with some also failing to demonstrate T2* changes (3) or T1 shortening during 100% O2 breathing (4,5), whereas others found significant T1 shortening during hyperoxia (2,3,19).
T1 AND T2* RESPONSE TO O2 AND CO2
Figure 5. Changes in magnetic resonance relaxation times (a) DT1 and (b) DT2* in individual animals for inspired gas transitions between air (21% O2) and: 100% O2 (open circles), 10% CO2 (balance air) (closed circles), and carbogen (90% O2 + 10% CO2) (gray circles). Dash symbol represents the median value for each gas transition. Significant differences: * P < .05, ** P < .01.
Previous work has shown insignificant blood flow changes in normal rabbit muscle during hypercapnia, as assessed using computed tomography perfusion (20). A small blood flow response in muscle is most likely related to low baseline blood flow levels compared with other abdominal organs, such that increases in PaCO2 have little effect. This low baseline compromises the ability of even T2* to detect flow changes. A previous study showed that even when blood flow was increased 10-fold during hyperemia induced by ischemia-reperfusion in human muscle, T2* increases were very small (<4 ms) (21). Use of MR relaxation times is one of numerous approaches for assessing pO2 (22). An alternative method is positron emission tomography (PET) using 18F-labeled fluoromisonidazole (18F-FMISO), a label that freely diffuses throughout cells under normal conditions but binds with intracellular macromolecules under hypoxia. Recent work has characterized the pharmacokinetics of 18F-FMISO PET in head and neck tumors (23). However, the parameters generated are not confirmed surrogates for hypoxia, and this method involves radiation. Yet another approach to quantify pO2 is 19F MRI, which uses the reporter molecular hexafluorobenzene to provide absolute values of tissue pO2 (24,25). In tumors, 1165
WINTER ET AL
Academic Radiology, Vol 18, No 9, September 2011
MRI data acquisitions to further reduce the effects of motion on the quantitative T1 and T2* estimates. In conclusion, changes in T2* and T1 are useful for monitoring organ-specific tissue pO2 and perfusion responses to gas challenges. Changes in T2* provided the best sensitivity, being able to detect both tissue oxygenation and blood volume modulation. The sensitivity of T1 was restricted mainly to tissue oxygenation and only in the absence of counteractive effects from vasodilatation. Although T2* and T1 changes can be predicted based on the gas transition, robust use of MR relaxation times as noninvasive biomarkers requires an understanding of their relative sensitivity to organ-specific physiological responses. Future studies focused on tissuespecific T1 and T2* responses to gas transitions with larger sample sizes are needed to confirm observations made in this study. REFERENCES
Figure 6. Changes in magnetic resonance relaxation times (a) DT1 and (b) DT2* in individual animals for inspired gas transitions: 10% CO2 (balance air) to 100% O2 (open circles), 10% CO2 to carbogen (90% O2 + 10% CO2) (gray circles), and 100% O2 to carbogen (closed circles). DT2* and DT1 were negated for gas transitions in the opposite direction. Dash symbol represents the median value for each gas transition. Significant differences: y P < .10, * P < .05, ** P < .01.
19
F MRI estimates were correlated to BOLD signal changes but not measures of T2* (25). Despite the key advantage of 19 F MRI, namely, providing absolute pO2 measurements, this method is limited by the need to directly inject reporter molecules into tissue, limited availability of the resources, and limited applications outside of animal models. For clinical applications, quantitative MR relaxation times remains an appealing surrogate measure for assessing tissue pO2 because of the wide availability of clinical MRI scanners, no need for injection of exogenous agents, and the ability to perform repeated measures. One limitation of the current study was the potential effects of anesthetic on the blood flow response to FiCO2 manipulation. Although we applied a low dose of isoflurane, its vasodilatory properties may modulate potential vascular responses to changes in the FiCO2. To provide greater relative perfusion response to FiCO2 increases, we selected a FiCO2 of 10% CO2, which is greater than the 5% CO2 typically used. Imaging abdominal organs also presents a challenge in terms of motion artifacts, which may degrade estimates of the relaxation time measurements. In the current study, we optimized the rabbit positioning and support to minimize respiratory motion and we also employed multiple averages for the 1166
1. Prisman E, Slessarev M, Han J, et al. Comparison of the effects of independently-controlled end-tidal PCO(2) and PO(2) on blood oxygen level-dependent (BOLD) MRI. J Magn Reson Imaging 2008; 27: 185–191. 2. O’Connor JP, Jackson A, Buonaccorsi GA, et al. Organ-specific effects of oxygen and carbogen gas inhalation on tissue longitudinal relaxation times. Magn Reson Med 2007; 58:490–496. 3. O’Connor JP, Naish JH, Jackson A, et al. Comparison of normal tissue R1 and R*2 modulation by oxygen and carbogen. Magn Reson Med 2009; 61: 75–83. 4. Tadamura E, Hatabu H, Li W, et al. Effect of oxygen inhalation on relaxation times in various tissues. J Magn Reson Imaging 1997; 7:220–225. 5. Noseworthy MD, Kim JK, Stainsby JA, et al. Tracking oxygen effects on MR signal in blood and skeletal muscle during hyperoxia exposure. J Magn Reson Imaging 1999; 9:814–820. 6. Jones RA, Ries M, Moonen CT, et al. Imaging the changes in renal T1 induced by the inhalation of pure oxygen: a feasibility study. Magn Reson Med 2002; 47:728–735. 7. Kaanders JH, Bussink J, van der Kogel AJ. ARCON: a novel biology-based approach in radiotherapy. Lancet Oncol 2002; 3:728–737. 8. Cheng HL, Wright GA. Rapid high-resolution T(1) mapping by variable flip angles: accurate and precise measurements in the presence of radiofrequency field inhomogeneity. Magn Reson Med 2006; 55: 566–574. 9. Beaumont M, Odame I, Babyn PS, et al. Accurate liver T2 measurement of iron overload: a simulations investigation and in vivo study. J Magn Reson Imaging 2009; 30:313–320. 10. Boss A, Martirosian P, Jehs MC, et al. Influence of oxygen and carbogen breathing on renal oxygenation measured by T2*-weighted imaging at 3.0 T. NMR Biomed 2009; 22:638–645. 11. Lanzen JL, Braun RD, Ong AL, et al. Variability in blood flow and pO2 in tumors in response to carbogen breathing. Int J Radiat Oncol Biol Phys 1998; 42:855–859. 12. Thews O, Kelleher DK, Vaupel P. Dynamics of tumor oxygenation and red blood cell flux in response to inspiratory hyperoxia combined with different levels of inspiratory hypercapnia. Radiother Oncol 2002; 62:77–85. 13. Neeman M, Dafni H, Bukhari O, et al. In vivo BOLD contrast MRI mapping of subcutaneous vascular function and maturation: validation by intravital microscopy. Magn Reson Med 2001; 45:887–898. 14. Lu J, Dai G, Egi Y, et al. Characterization of cerebrovascular responses to hyperoxia and hypercapnia using MRI in rat. Neuroimage 2009; 45: 1126–1134. 15. Barash H, Gross E, Matot I, et al. Functional MR imaging during hypercapnia and hyperoxia: noninvasive tool for monitoring changes in liver perfusion and hemodynamics in a rat model. Radiology 2007; 243: 727–735. 16. Lautt WW. Mechanism and role of intrinsic regulation of hepatic arterial blood flow: hepatic arterial buffer response. Am J Physiol 1985; 249(5 Pt 1):G549–G556.
Academic Radiology, Vol 18, No 9, September 2011
17. Chang C, Uchiyama A, Ma L, et al. A comparison of the effects on respiratory carbon dioxide response, arterial blood pressure, and heart rate of dexmedetomidine, propofol, and midazolam in sevoflurane-anesthetized rabbits. Anesth Analg 2009; 109:84–89. 18. Rhee TK, Larson AC, Prasad PV, et al. Feasibility of blood oxygenation level-dependent MR imaging to monitor hepatic transcatheter arterial embolization in rabbits. J Vasc Interv Radiol 2005; 16:1523–1528. 19. McGrath DM, Naish JH, O’Connor JP, et al. Oxygen-induced changes in longitudinal relaxation times in skeletal muscle. Magn Reson Imaging 2008; 26:221–227. 20. Purdie TG, Lee TY. Carbon dioxide reactivity of computed tomography functional parameters in rabbit VX2 soft tissue tumour. Phys Med Biol 2003; 48:849–860. 21. Lebon V, Carlier PG, Brillault-Salvat C, et al. Simultaneous measurement of perfusion and oxygenation changes using a multiple gradient-echo
T1 AND T2* RESPONSE TO O2 AND CO2
22.
23.
24.
25.
sequence: application to human muscle study. Magn Reson Imaging 1998; 16:721–729. Tatum JL, Kelloff GJ, Gillies RJ, et al. Hypoxia: importance in tumor biology, noninvasive measurement by imaging, and value of its measurement in the management of cancer therapy. Int J Radiat Biol 2006; 82: 699–757. Wang W, Lee NY, Georgi JC, et al. Pharmacokinetic analysis of hypoxia (18)F-fluoromisonidazole dynamic PET in head and neck cancer. J Nucl Med 2010; 51:37–45. Jordan BF, Cron GO, Gallez B. Rapid monitoring of oxygenation by 19F magnetic resonance imaging: simultaneous comparison with fluorescence quenching. Magn Reson Med 2009; 61:634–638. Zhao D, Jiang L, Hahn EW, et al. Comparison of 1H blood oxygen leveldependent (BOLD) and 19F MRI to investigate tumor oxygenation. Magn Reson Med 2009; 62:357–364.
1167