| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (M.A.B., T.M., V.D., C.M.) and Gastroenterology (J.D.), Erasme Hospital, and IRIBHN, Statistical Unit (V.D.M.), Université Libre de Bruxelles, Route de Lennik, 808, 1070 Brussels, Belgium; and Philips Medical Systems Benelux, Brussels, Belgium (V.D.). Received April 8, 2007; revision requested June 12; revision received August 29; final version accepted September 27. Address correspondence to M.A.B. (e-mail: mbali{at}ulb.ac.be).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEIMPLICATION FOR PATIENT CAREReferences |
|---|
Materials and Methods: The study had institutional review board approval, and written informed consent was obtained. Ten healthy volunteers (five men, five women; mean age, 24.7 years ± 1.9 [standard deviation]; range, 22–29 years) underwent MR imaging pancreatic perfusion studies performed twice without secretin and twice during secretin stimulation. Dynamic contrast-enhanced MR imaging consisted of saturation-recovery T1-weighted turbo-field-echo imaging with peripheral pulse triggering and respiratory tracking. A dose of 0.05 mmol gadodiamide per kilogram of body weight was injected at a rate of 3.5 mL/sec. Regional perfusion parameters were fitted with a one-compartment model. The analysis of variance test for repeated measurements was used to assess differences in pancreatic perfusion without and that with secretin administration.
Results: Significant differences in perfusion parameters between the three pancreatic regions were observed (P < .05). During secretin stimulation, a significant difference was observed only between the body and the tail of the pancreas (P = .02). A significant increase (P = .003) in pancreatic perfusion was observed after secretin administration. Mean pancreatic perfusion was 184 mL/min/100 g of tissue ± 71, 207 mL/min/100 g ± 77, and 230 mL/min/100 g ± 87 without secretin and 342 mL/min/100 g ± 154, 338 mL/min/100 g ± 156, and 373 mL/min/100 g ± 176 after secretin stimulation in the head, body, and tail of the pancreas, respectively. Intraindividual variability was 21% without secretin stimulation and 46% with secretin stimulation.
Conclusion: Dynamic contrast-enhanced MR imaging enables noninvasive quantification of regional pancreatic perfusion in resting conditions and demonstrates the increase in pancreatic perfusion during secretin stimulation in healthy subjects.
© RSNA, 2008
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEIMPLICATION FOR PATIENT CAREReferences |
|---|
Miles at al (4) used contrast-enhanced dynamic computed tomography (CT) to measure regional pancreatic perfusion by using a first-pass analysis of extravascular contrast agent. This method has been used to evaluate patients with acute pancreatitis (5). Despite the promising results of that preliminary study, the technique is limited by the necessity for prolonged breath holding and requires intravenous administration of iodinated contrast agent, with its debated potential risks in patients with acute pancreatitis (6–8). Previous investigators of dynamic contrast-enhanced magnetic resonance (MR) imaging have assessed pancreatic perfusion by performing a semiquantitative analysis of gadolinium enhancement parameters (9,10).
Quantitative analysis of regional blood perfusion by using dynamic contrast-enhanced MR imaging is based on the pharmacokinetics of tracer transport between compartments. Several investigations estimating kinetic parameters from dynamic contrast-enhanced T1-weighted MR imaging data have been undertaken, and standardized definitions of parameters have been provided (11). Materne et al (12) have demonstrated, in an experimental animal study, the feasibility and validity of MR imaging for quantifying hepatic perfusion with a one-compartment model that takes into account the dual hepatic arterial and portal venous input of the liver.
Because, unlike in animal studies, no reference standard is available, pancreatic perfusion quantification by using dynamic contrast-enhanced MR imaging has not been fully validated in humans. Moreover, its reproducibility has not been assessed. To address these issues, we suggest an alternative approach of stimulating the exocrine pancreas with exogenous secretin, which is responsible in normal conditions for an increase in pancreatic perfusion.
Thus, the purpose of our study was to prospectively quantify pancreatic regional perfusion with dynamic contrast-enhanced MR imaging by using a one-compartment model and to assess perfusion changes during secretin stimulation in healthy volunteers.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEIMPLICATION FOR PATIENT CAREReferences |
|---|
Study Participants
Ten healthy volunteers (five men, five women; mean age, 24.7 years ± 1.9 [standard deviation]; age range, 22–29 years) who had no history of pancreatic disease and who were not taking any medication were included. Each volunteer underwent four dynamic contrast-enhanced MR imaging studies after having fasted for 6 hours. MR imaging studies were performed twice without secretin administration and twice during secretin stimulation. The interval between each MR imaging study was 1 week.
MR Imaging Protocol
All pancreatic perfusion MR imaging studies were performed with a 1.5-T superconducting magnet (Gyroscan Intera; Philips Medical Systems, Best, the Netherlands) equipped with a phased-array surface coil. All individuals were placed in the magnet in the supine position.
The perfusion sequence consisted of a dynamic T1-weighted turbo-field-echo sequence with a nonselective saturation prepulse followed by a spoiler gradient just before the train of excitations. A linear phase-encoding order was used, and the saturation delay was 273 milliseconds. The imaging parameters were as follows: two 6-mm-thick sections; intersection gap, 6 mm; repetition time msec/echo time msec, 5/1.29; flip angle, 15°; field of view, 40 cm rectangular (percentage of acquisition, 65%); matrix, 128 x 256; and use of peripheral pulse triggering in the diastolic phase and free-breathing respiratory tracking with the navigator positioned at the lung-liver interface. One hundred fifty dynamic phases were acquired for each investigation. Each dynamic phase consisted of two sections acquired at two consecutive heartbeats. The imaging time was dependent on the individual heart rate. The mean imaging time for the perfusion sequence was 5.05 minutes (range, 3.23–7.43 minutes).
The sections were acquired in an oblique transverse plane to cover the pancreatic head, body, and tail and the external reference. The external reference consisted of a standard Philips phantom—that is, an aqueous solution of copper sulfate placed on the right side of the individual and included in the field of view. Signal inhomogeneities resulting from the use of the phased-array surface coil were corrected by using the coil sensitivity profiles provided by a coil reference scan (13). After the first 10 dynamic acquisitions, gadodiamide (Omniscan; GE Amersham, Oslo, Norway) was intravenously injected at a rate of 3.5 mL/sec by using a mechanical MR imaging injector at a dose of 0.05 mmol per kilogram of body weight. The pancreatic perfusion studies performed with secretin involved intravenous administration of a bolus of secretin (Secrelux; Sanochemia, Neuss, Germany) at a dose of 1 U/kg followed, after 1 minute, by gadodiamide administration and the previously described perfusion sequence.
All MR imaging examinations were completed and well tolerated by all volunteers. No side effects caused by gadodiamide or secretin administration were recorded in any of the volunteers. Regular cardiac synchronization was obtained in all subjects.
Data Analysis
All images were transferred for analysis to an independent diagnostic workstation (Philips Medical Systems) after the name of the participant and the date of the examination were eliminated. All measurements were performed by one experienced radiologist (M.A.B.).
Regions of interest (ROIs) were manually drawn in the aorta; in the head, body, and tail of the pancreas; and in an external reference to obtain mean signal intensity measurements. The largest possible ROI was placed over the aorta. The pancreatic ROIs were placed on the section that best depicted the desired portion of the pancreas. On this section, the largest possible ROI was drawn, with careful exclusion of the surrounding vessels. Mean ROI size was 200 mm2 (range, 109–312 mm2), 238 mm2 (range, 129–437 mm2), and 224 mm2 (range, 115–314 mm2) for the head, body, and tail of the pancreas, respectively (Fig 1). For the aorta and the external reference, ROIs were drawn on both sections. For the aorta, the ROI mean signal intensity curve with the lowest peak amplitude was chosen, and, for the external reference, the most stable mean signal intensity curve over time was selected. In the case of residual respiratory motion artifacts, the ROIs were manually repositioned from section to section.
|
The signal intensity curves were converted into gadolinium concentration–versus-time curves by using a described method (12). The intensity curves were normalized with respect to the external reference signal intensity and were converted to 1/T1 values by using the following equation: 1/T1(sec–1) = –0.61 sec–1 + 5.21 sec–1(S/Sref) – 3.7 sec–1(S/Sref)2 + 1.54 sec–1(S/Sref)3, where S represents the tissue or aortic signal intensity at a particular time point and Sref is the average signal intensity measured in the external reference. This relationship was obtained by our team previously by imaging tubes filled with 12 solutions with different concentrations of gadodiamide, ranging from 0 to 3.6 mmol/L with a fixed increment of 0.3 mmol/L, together with the external reference. S and Sref were calculated by means of ROIs placed over the tubes and the external reference on images acquired with the perfusion sequence. The value of the T1 relaxation time was also measured for each tube by using a mixed inversion-recovery spin-echo sequence (14). A third-order polynomial was subsequently fitted to the S/Sref-versus-1/T1 curve.
The 1/T1 values were converted to gadolinium concentrations (GCs) by using the relationship GC = [(1/T1)post – (1/T1)pre]/r1, where post is after gadodiamide adminstration, pre is before gadodiamide administration, and r1 = 3.9 sec–1/mmol/L. This value was determined on the basis of the calibration experiment as the slope of the curve of 1/T1 versus GC.
A one-compartment model, derived from the general two-compartment model of Brix et al (15), was used to calculate the perfusion parameters. The Brix et al model consists of a plasma compartment with arterial inflow and venous outflow and an interstitial compartment (extracellular and extravascular space) exchanging gadolinium with plasma through the capillary membranes (11,15). In our study, we made the hypothesis of high capillary wall permeability; the tissue concentration CT is then given by the following:
 |
is the distribution fraction—that is, the volume fraction of the tissue that is accessible to the contrast agent, which corresponds to the plasma and the interstitial space. This is equivalent to the flow-limited Kety model presented in an earlier publication by Tofts et al (11) (Eq [4] in that publication), with the exception that in the Tofts et al equation, there was further restriction of the accessible space to extracellular and extravascular space; hence, Tofts et al used ve, the volume fraction of the extracellular and extravascular space, instead of , as in our model. The solution of this equation is given as
 |
A is the transit time between the artery and the tissue, and Hct is the hematocrit, understood to be 0.45 (16).
In practice, given the arterial input and values of the three unknown parameters F/VT (pancreatic perfusion reported in milliliters per minute per 100 g of tissue), (the distribution fraction), and A (transit time), the theoretical tissue concentration curve is calculated by numerically integrating the latter equation. These three parameters—pancreatic perfusion, distribution fraction, and transit time—were obtained with nonlinear least-squares fitting of the experimental pancreatic gadolinium concentration curve by using the fminsearch built-in function of Matlab as a minimization algorithm.
Statistical Analysis
Quantitative variables are reported as means ± standard deviations. Statistical analysis was performed by using statistical software (SPSS, version 14.0; SPSS, Chicago, Ill). The analysis of variance test for repeated measures was performed to assess differences between MR imaging–derived pancreatic perfusion parameters obtained from (a) the two MR imaging sessions performed without secretin, (b) the two MR imaging sessions performed during secretin administration, and (c) the MR imaging sessions performed without and with secretin administration.
The same test was used to assess differences between pancreatic perfusion parameter values obtained in the three pancreatic regions (the head, body, and tail) without and with secretin. P < .05 was considered to indicate a significant difference.
To assess the reproducibility of the quantification method, the absolute differences between data from the two MR imaging sessions were computed for each pancreatic region in each individual. The absolute differences for the three regions were averaged over all individuals. Additionally, the ratio between the absolute difference and the average of data from the two MR imaging sessions was calculated for the three pancreatic regions in each individual, and these ratios were averaged over all regions and individuals to yield the percentage intraindividual variability.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEIMPLICATION FOR PATIENT CAREReferences |
|---|
|
|
|
Regarding the mean pancreatic perfusion values obtained without and during secretin stimulation, a significant increase in perfusion was observed in the three pancreatic regions after secretin administration (P = .003) (Table 3).
|
The percentage intraindividual variability of pancreatic perfusion was 21% without secretin and 46% during secretin stimulation.
Distribution Fraction
The mean distribution fractions calculated in the head, body, and tail of the pancreas, respectively, were 0.15 ± 0.037, 0.16 ± 0.034, and 0.17 ± 0.035 without secretin stimulation and 0.17 ± 0.042, 0.17 ± 0.045, and 0.17 ± 0.047 during secretin stimulation. In the two MR imaging sessions performed without secretin, significant differences were found for distribution fraction between the head and the body (P = .004) and between the head and the tail (P < .001), while no significant difference was observed between the body and the tail (P = .18) of the pancreas. The distribution fraction calculated for the two MR imaging sessions performed after secretin administration was not significantly different for the three pancreatic regions (P = .30 between the head and body and body and tail and P = .13 between the head and tail). When we compared the distribution fraction obtained without secretin with that obtained with secretin, a significant difference was found only for the head of the pancreas (P = .02).
The reproducibility of the quantification method was characterized by a mean absolute intraindividual difference for the head, body, and tail of the pancreas, respectively, of 0.014, 0.018, and 0.016 when MR imaging was performed without secretin stimulation and 0.032, 0.028, and 0.023 when MR imaging was performed during secretin stimulation.
The percentage intraindividual variability of the distribution fraction was 10% without secretin and 16% during secretin stimulation.
The transit time was less than 2 seconds in all subjects, and any possible difference could not be assessed given the temporal resolution of the dynamic acquisition (two cardiac beats).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEIMPLICATION FOR PATIENT CAREReferences |
|---|
Pancreatic perfusion measurements obtained without secretin administration showed statistically significant regional differences, with regional perfusion values increasing progressively from the head to the body and the tail of the pancreas. These regional differences can probably be explained by the difference in regional arterial supply; the superior mesenteric artery supplies the head, and the celiac trunk supplies the body and the tail, through pancreatic arterial branches. Concomitantly, we observed that the distribution fraction was smaller in the head than in the body or in the tail of the pancreas. This observation could partially be explained by a different partial-volume effect due to the caliber of the main pancreatic duct, which is larger in the head than in the body and tail of the pancreas.
Results of previous animal and human studies performed by using more invasive techniques have demonstrated that secretin administration increases pancreatic perfusion (17–21). Secretin stimulation was achieved by using a dose of 1 U/kg because this dose is known to be responsible for a maximal response of the exocrine pancreas (22). We systematically used a delay of 1 minute between secretin administration and the injection of gadodiamide. This short delay was chosen on the basis of observations from previous experimental studies (17) whose results showed that a major change in pancreatic perfusion could be expected within a minute in healthy subjects.
Indeed, a statistically significant increase in pancreatic perfusion during secretin stimulation was observed in all three pancreatic regions. Interestingly, the increase in perfusion was more pronounced in the head of the pancreas, where perfusion values were lower at resting conditions, than in the body and tail. Similar behavior has previously been reported by Aune and Semb in their animal experiments (23), where the increase in pancreatic perfusion under secretin stimulation was more pronounced in animals with low perfusion values in resting conditions. According to our observations, this finding seems to be confirmed regionally in healthy human subjects. The regional differences in pancreatic perfusion observed at resting conditions became less pronounced after secretin stimulation. Furthermore, our measurements showed a statistically significant increase in distribution fraction after secretin administration only in the head of the pancreas. Our study results confirm the stimulating effect of secretin on pancreatic perfusion. However, the quantification method used allowed only one measurement of the perfusion values, during the first pass after a bolus injection of gadodiamide, while the distribution fraction was measurable for only a few minutes; this represents one of the limitations of our technique.
To the best of our knowledge, ours is the first study that assessed the reproducibility of a noninvasive quantification technique for pancreatic perfusion. In our series, we observed that absolute intraindividual differences were of the same order of magnitude as interindividual variations. However, at MR imaging sessions performed during secretin stimulation, the intraindividual differences were larger than at sessions performed without secretin. A possible explanation could be found in the variable response of the exocrine pancreas to secretin stimulation within the delay of 1 minute after the secretin bolus injection. When we compared the intraindividual variability of pancreatic perfusion and distribution fractions, we observed a higher variability for pancreatic perfusion, which was observed without and during secretin stimulation.
The lack of a reference standard for quantitative pancreatic perfusion measurements in humans represents one of the limitations of our study. It prevented us from considering the calculated pancreatic perfusion parameters as absolute values and from assessing whether the observed intraindividual differences were due to physiologic variability and/or to a limited accuracy of our method.
The method used is prone to several sources of error. The relationship between gadolinium concentration and signal intensity is nonlinear, exhibiting a signal saturation at high concentration values that can account for the lack of accuracy of aortic concentration measurements. A dose of 0.05 mmol/kg of gadodiamide injected at a rate of 3.5 mL/sec was chosen to limit high aortic signal saturation effects in the curve relating signal intensity and 1/T1. These parameters were previously chosen by Annet et al (24) and were expected to still allow good sensitivity for tissue 1/T1 determination. Moreover, the aortic signal is also dependent on inflow velocity effects, which can lead to a potential underestimation of the perfusion values (25). A small flip angle of 15° and diastolic cardiac triggering were used to reduce inflow effects; however, no aortic velocity measurements were undertaken in our study. When the aortic signal intensity curves were different from one section to another, we systematically chose the lower aortic signal peak curve for our calculations.
Good spatial resolution is required for the adequate placement of ROIs in different pancreatic regions to avoid partial-volume effects with surrounding vessels. The investigation of perfusion over the head, body, and tail of the pancreas necessitated the use of two sections in combination with a phased-array coil to obtain higher spatial resolution and signal intensity compared with those obtained with the quadrature body coil. With surface coils, signal inhomogeneity can occur. This was corrected by using the coil's sensitivity profiles (13).
The relatively young age of the study participants could represent another limitation of our study. Indeed, a patient population would be considerably older, and thus pancreatic perfusion measurements might differ from those obtained in our study.
In conclusion, dynamic contrast-enhanced MR imaging enables quantitative measurements of regional pancreatic perfusion both in resting conditions and during exogenous secretin stimulation. Further clinical investigations should confirm these preliminary results and the usefulness of this technique in the diagnostic assessment of non-necrotizing acute pancreatitis and chronic pancreatitis. The aim could to be to identify prognostic factors concerning the evolution of the inflammatory process and to monitor the effects of therapy on regional pancreatic perfusion.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEIMPLICATION FOR PATIENT CAREReferences |
|---|
|
IMPLICATION FOR PATIENT CARE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEIMPLICATION FOR PATIENT CAREReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: ROI = region of interest
Author contributions: Guarantors of integrity of entire study, M.A.B., T.M., C.M.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, M.A.B., T.M., V.D., J.D.; clinical studies, M.A.B., T.M., C.M.; experimental studies, V.D.; statistical analysis, V.D.M.; and manuscript editing, M.A.B., T.M., V.D., C.M.
Authors stated no financial relationship to disclose.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEIMPLICATION FOR PATIENT CAREReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
