Published online before print January 25, 2002, 10.1148/radiol.2223010639
(Radiology 2002;222:681-686.)
© RSNA, 2002
Conformal Therapy for Pancreatic Cancer: Variation of Organ Position due to Gastrointestinal Distention—Implications for Treatment Planning1
Eckehard Horst, MD,
Oliver Micke, MD,
Christos Moustakis, PhD,
Andreas Schuck, MD,
Ulrich Schäfer, MD and
Normann A. Willich, MD
1 From the Department of Radiation Oncology, University of Münster, Albert-Schweitzer-Strasse 33, 48129 Münster, Germany. From the 2000 RSNA scientific assembly. Received March 19, 2001; revision requested April 17; revision received July 5; accepted July 27. Address correspondence to E.H. (e-mail: horste@uni-muenster.de).
 |
ABSTRACT
|
|---|
PURPOSE: To quantify nonrespiratory organ motion in the pancreatic region and its effect on clinical target volume.
MATERIALS AND METHODS: Three-dimensional translations of the geometric centers of the volumes of interest—pancreatic head, body, and tail; left and right kidney; and the superior mesenteric artery—were measured in 20 patients by analyzing three spiral computed tomographic (CT) protocols performed at static exhalation and representing differential gastrointestinal distention. Wilcoxon test for paired differences was applied to determine statistical significance (P < .05). Spearman rank correlation coefficients were calculated between combinations of statistically significant translations. With the assumption that the organ positions were represented by a three-dimensional Gaussian distribution that occurs during treatment, clinical target volume expansions were calculated to account for organ motion and a typical setup error.
RESULTS: Significant translations of the volume of interest were observed. The most mobile parts of the target organs were the pancreatic tail (P = .001) and the superior mesenteric artery (P = .01). Larger variations from the mean in the planning CT protocol in which negative contrast material was used usually resulted in a slightly larger clinical target volume expansion.
CONCLUSION: Our data may provide a basis for further studies of organ motion and ways of modifying treatment margins.
© RSNA, 2002
Index terms: Pancreas, CT, 77.12111, 77.12112, 77.12115, 77.12117 • Pancreas, therapeutic radiology, 77.125 • Treatment planning
|
INTRODUCTION
|
|---|
In the treatment of patients with pancreatic cancer, advanced treatment techniques, such as those that use noncoplanar beams and three-dimensional conformal therapy, provide greater normal tissue protection and offer opportunities to escalate the radiation dose to improve tumor control (1–3). The accuracy of tumor and target volume delineation has substantial importance to realize these goals. Exact identification of the surrounding viscera is critical, because collapsed duodenal, small bowel, or stomach segments can simulate disease or obscure pathologic abnormalities. Therefore, treatment-planning computed tomography (CT) performed with contrast material administered orally and intravenously is usually recommended, although a standardized CT protocol has not been reported to our knowledge (2–4).
The validity and reliability of the patient model derived from the treatment-planning CT are influenced by physiologic organ motion (5). Different organ and tissue positions during treatment planning, simulation, and treatment delivery may make the calculated dose distribution a poor estimate of the actual dose applied and may increase the risk ofmarginal failure and treatment toxicity. To account for respiration, the range and time course of organ movement in the abdominal region have been quantified (6–8).
Organ translations owing to a variable filling of the adjacent hollow organs with fluid and air are an active area of investigation in the pelvic region, but sparse data are available for the upper abdomen (5,9,10). Apart from respiration, organ motion in the upper abdomen can be due to peristaltic motion, gaseous distention, and differential gastrointestinal filling. The purpose of our study was to quantify nonrespiratory organ motion in the pancreatic region and its effect on clinical target volume (CTV).
|
MATERIALS AND METHODS
|
|---|
Patient Selection
From January 1998 to October 2000, 24 patients who were referred to our department with histologically proven local advanced pancreatic cancer that was not resectable were selected. All patients underwent explorative laparotomy 10–14 days before referral. The selection criteria were tumor location in the pancreatic head or body and absence of gastroduodenal bypass. All patients subsequently underwent accelerated chemoradiation at 44.8 Gy (11). Four patients were excluded due to technical imaging problems; therefore, data of 20 patients were available for evaluation. The characteristics of patients and tumors are summarized in Table 1. The protocol was approved by our local institutional review board. The study procedure was explained in detail to each patient, and informed consent was obtained.
CT Protocol
The patients underwent three spiral CT examinations (PQ5000; Marconi Medical Systems, Highland Heights, Ohio) in the treatment position, with standardized breath-hold unenhanced and oral negative and positive contrast material protocols. Alignment of the patients was performed with the external laser system (PatPos; LAP, Lueneburg, Germany). Sequential generation of scans with no contrast material and with negative contrast material on day one and with positive contrast material on the following day, each day after a fasting of at least 6 hours, was the standardized approach. For CT with positive contrast material, the patients were given 500 mL of dilute barium sulfate 30 minutes before the examination. For CT with negative contrast material, the patients were given 750–900 mL of water 15 minutes before the examination. The different oral contrast material protocols, which were associated with low (no contrast material), medium (positive contrast material), and high (negative contrast material) fluid content of the proximal gastrointestinal tract, were selected to modify the gastrointestinal distention in each patient.
To optimize depiction of the anatomy, peristaltic motion was reduced in the protocol in which negative contrast material was used with 20 mg of butylscopolamine bromide (Buscopan; Boehringer Ingelheim, Ingelheim, Germany) administered intravenously. After the patients were asked to practice breath holding at normal exhalation, scans were obtained every 5 mm with 3-mm reconstruction (pitch of 1.5) during a single breath hold at static exhalation. The scanning volume was the pancreatic and kidney region. The normal exhalation position was selected because of its reproducibility, which also has been considered suitable for respiratory gating technology. The reproducibility of the normal exhalation position was 0.9 mm (2 SDs), as opposed to 2.6 mm for the inhalation position (8). For oral contrast material–enhanced scans, 150 mL of nonionic contrast agent (Imeron 300 [iomeprol]; Bracco-Byk Gulden, Konstanz, Germany) was administered intravenously by using a mechanical injector (Envision CT; Medrad, Indianola, Pa) at a rate of 3 mL/sec. Portal phase images were obtained at approximately 60 seconds after the initiation of injection. Parts of the pancreatic tail were missed in three patients.
Organ Contour and Quantification of Organ Translations
The following volumes of interest (VOIs) were selected for evaluation: right and left kidney; the SMA at 15 and 30 mm from the origin; and the pancreatic head, body, and tail. The head of the pancreas was defined as the portion to the right of the horizontal part of the gland, which was usually located anterior to the second lumbar vertebrae. The body and the tail were defined by dividing the remaining pancreas into thirds, where the body was the proximal one-third and the tail was the distal two-thirds. Tumor volumes were included in the pancreatic VOI. In the treatment planning of pancreatic cancer, both kidneys were studied because they are important mobile organs with low tolerance to therapeutic irradiation. The SMA is a critical target, particularly in neoadjuvant strategies, because a residual tumor infiltration usually precludes the Whipple procedure (12). The anatomic course of the SMA perpendicular to the transverse plane allowed accurate and consistent delineation. The entirety of the organs and structures were contoured interactively on nonenhanced and contrast-enhanced scans by using semiautomatic and manual segmentation routines. In three patients, the pancreatic tail and body were not defined because of partial scanning of the tail.
All VOIs were contoured by two radiation oncologists (E.H., O.M.) whose interpretations were made by means of consensus. Substantial operator judgement was required in the interpretation of soft-tissue volumes, particularly at the inferior duodenal-pancreatic interface. To detect outlining divergence on the scans obtained in the same patient, superimposed three-dimensional projections of the organ contours and organ volume calculations were compared and judged for similarity. The comparison allowed categorization of the observed outlining divergence either as difficulty in anatomic interpretation or as an actual organ translation.
Preoperative magnetic resonance imaging and especially parenchymal sequences (breath-hold fat-suppressed turbo spin echo T1) aided in the delineation of pancreatic organ contours in each patient. Image registration was performed among all images with respect to the bone anatomy by using fusion software (ACQ-SIM; Marconi Medical Systems) (13). The matrix resolution was 0.06 cm/pixel. The point-based registration method was applied by using internal bone landmarks (14). A minimum of five noncoplanar point pairs were defined to account for the translation and rotation. A least squares fit on the point pairs was computed to remark or exclude poorly defined points. After image registration, the alignment of the two scans was visually inspected. The numeric accuracy of the image registration was evaluated by measuring the alignment of the geometric centers of the three registered vertebral body pairs. The registrations were modified until the border uncertainty of bone alignment was measured to be less than or equal to 1.0 mm in transaxial and transverse directions.
After the procedure, a three-dimensional matrix was generated and used to describe VOI translations between the two CT data sets. The geometric center of each VOI was calculated by using the organ contours. To quantify the effect of the states of gastrointestinal distention, the translations of the geometric centers were measured between the nonenhanced and the oral contrast-enhanced scans. The CTV expansion was calculated after the translations of the geometric centers had been measured between a planning scan (oral contrast-enhanced scan) and each of the three treatment scans. Each scan represented VOI positions during treatment. The directions of the translations were right to left, anteroposterior, and craniocaudal. Transverse translations were obtained for the SMA. The measurement uncertainty in the craniocaudal direction was estimated to be one-half of the section separation, or 1.5 mm.
CTV Expansion for Organ Translations and Setup Error
On the basis of the assumption that the target organ positions on the CT scans are but one sample of a random distribution that occurs during treatment, the pooled data of 60 CT scans were used to calculate the margins that should be added to the CTV to account for a variable gastrointestinal distention and a typical setup error. The organ translations were further divided into systematic and random components. The systematic translation was the difference between the organ position on the planning scan and the average position as calculated from the treatment positions. Both oral contrast-enhanced protocols were evaluated as planning scans. The random translation was the difference of the organ position on each treatment scan from its average position. The International Commission on Radiation Units and Measurements Report 50 (15) expresses the combined SD of treatment uncertainties as 
combined = (2organ + 2setup)0.5, where organ is the combined systematic and random organ translation and setup is the setup error. A setup error of 5 mm was used. On the basis of the CTV expansion model of Austin-Seymour et al (16), the combined SD is multiplied by an expansion factor associated with statistics, which incorporates a certain nominal probability. By assuming a normal distribution associated with a nominal probability of 95% or 90% in each of the three dimensions, the expansion factor is 1.96 or 1.64, respectively, setting the CTV expansion margin = 1.96 or 1.64 x combined. The selected nominal probabilities correspond to total (three-dimensional) probabilities of 85.7% (0.953) and 72.9% (0.93). An evaluation of the CTV margins performed during the development of the model revealed that margins applied in various tumor sites and patient conditions generally corresponded to a nominal probability of 75%.
Data Analysis
Organ translation measurements are expressed as median, interquartile range (q25–q75), and range. The Wilcoxon test for paired differences was applied to determine statistical significance (P < .05). Spearman rank correlation coefficients were calculated with a computer software program (SPSS for Windows, version 10.0.7; SPSS, Chicago, Ill) between combinations of statistically significant translations.
|
RESULTS
|
|---|
Organ Translation
Translations of the geometric centers of the VOIs measured in three dimensions are summarized in Table 2. The analysis of the kidney position in the CT study with positive contrast material (medium fluid volume) demonstrated a significant shift of the left kidney in the caudal direction (P = .001) and a slight shift of the right kidney in the cranial direction (P = .01). The evaluation of parts of the target volume revealed significant (P = .01) positional changes of the pancreatic tail and body in the caudal direction. A displacement to the right (P = .08) was found for the SMA at 15 and 30 mm from its origin. The median displacements clustered around relatively small 3–4-mm changes, but the magnitude of organ translations varied between patients. Maximum target organ translations of 15.6–17.6 mm occurred.
If both CT protocols are compared, displacement measurements were often larger in the negative contrast material protocol (high-fluid volume), but statistically significant differences were found only for the posterior (P = .002) and caudal (P = .04) translation of the left kidney. In the negative contrast material protocol, the right-sided shifts of the SMA were significant and reached 16 (P = .01) and 20.2 (P = .03) mm. Figure 1 indicates the patient-to-patient variation of the SMA position at 15 mm from the origin in the right-left and anteroposterior directions, which resulted in interquartile ranges (q25–q75) of -10.9 to -1.2 mm and -3.1 to 3.2 mm, respectively. A significant translation to the right (P = .02) was found for the pancreatic head, which is located in proximity to the SMA. Caudal displacement was significant for the pancreatic tail (P = .001). For the left kidney, a slight posterior shift (P = .001) was noted. Figures 2 and 3 show examples of registered breath-hold noncontrast and negative contrast material scans and illustrate the caudal shift of the pancreatic tail and the right-sided displacement of the SMA at 15 mm from the origin and the pancreatic head.
View larger version (19K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Graph depicts the anteroposterior (a-p) and right-left (r-l) translations of the SMA at 15 mm from origin in the negative contrast material protocol for each patient. The intersection of the thick solid lines indicates the median translations and their lengths indicate interquartile ranges (q25-q75) in the two dimensions.
|
|
View larger version (82K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2. CT scan to CT scan registration. Kidney contours (yellow, light green, light blue) and the target contour (violet) on the transverse noncontrast CT scan (right) project on the transverse negative contrast material (high fluid content of proximal gastrointestinal tract) CT scan (left). The significant caudal translation of the pancreatic tail (light green) and parts of the pancreatic body (light yellow) owing to differential gastrointestinal distention are identified.
|
|
View larger version (79K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3. CT scan to CT scan registration. Kidney contours (yellow, light green, light blue), pancreatic head contours (red), and the target contour (violet) on the transverse negative contrast material (high fluid content of proximal gastrointestinal tract CT scan (left) project on the transverse noncontrast CT scan (right). The right-sided translation (11.2 mm) of the SMA at 15 mm from the origin (arrow) is illustrated. Note the displacement of the high-attenuation stent as a marker of a positional change of the pancreatic head. The CT registration shows a motion of the superior mesenteric vein parallel to the SMA and also indicates the caudal translation of the left kidney.
|
|
Correlation between Organ Translations
In the negative and positive contrast material CT protocols, 42 and 30 correlations, respectively, were calculated between combinations of statistically significant translations (the right-sided shift of the SMA in the positive contrast material CT protocol was included). A minority of combinations were significant (five of 42 and four of 30, respectively). Only these relations are listed in Table 3. In the negative contrast material protocol, a correlation between the translations of the pancreatic head and the SMA occurred, which are located in close anatomic proximity. In the positive contrast material protocol, anatomy-related correlations are obvious between the pancreatic tail and the body.
CTV Expansion
Table 4 summarizes the CTV expansions required for the organ motion evaluated in this study and a typical setup error () of 5 mm. If the negative contrast material CT scan was used as a planning scan, larger variations from the average organ position usually resulted in larger CTV expansions. The differences were slight and varied between 0% and 9.4%.
|
DISCUSSION
|
|---|
We observed statistically significant nonrespiratory translations of the pancreatic head, body, and tail; the SMA; and both kidneys on spiral CT scans in response to differential gastrointestinal distention. Larger displacements were measured in the CT protocol with negative contrast material and high fluid volume, but statistically significant differences between both protocols were noted only for the left kidney. However, the results of the present study are limited by the small sample.
The most mobile parts of the target organs examined on both scans were the pancreatic tail and the SMA. The proximity of the organs to the proximal gastrointestinal tract can explain the measured translations. This conclusion is supported by anatomy-related correlations among translations, although one must consider that the organ system under evaluation is complex. An individual interaction of factors such as organ morphology, peristaltic motion, gaseous distention, tumor location, tumor size, and fixation to different adjacent structures (as shown in Table 1) contribute to the variable magnitude of translations. A distinction between these factors is difficult, presumably impossible, to perform, which will impair the predictability of organ translations in a specific patient.
If both CT protocols are compared as treatment-planning protocols, the negative contrast material protocol could offer certain advantages. In comparison to agents like dilute barium, water does not distort dose calculations on the basis of CT attenuation, and the negative contrast material in distended gastrointestinal walls can improve visualization of the posterior wall of the stomach and the medial duodenal wall (17,18). This may be important in planning a boost treatment, because these structures cannot be excluded from the high-dose region. However, the magnitude of the target organ translations measured in this study could have implications on treatment accuracy when appropriate margins for the CTV in three-dimensional conformal therapy are designed. A systematic deviation from the average organ position during treatment indicates that the target organs are not within the expected position relative to the beam edges, which affects each treatment delivered. The median of the significant target organ translations varied between 1.9 and 7.1 mm, with maximum shifts greater than 15 mm in both contrast-enhanced protocols.
A comparison of the kidney motion measured in this study with that in the literature data of respiration-induced movement showed that the medians of the left and right kidney translations along the craniocaudal axis reached almost one-fourth and one-half of the medians, respectively, reported for respiration-induced excursion (6). The dosimetric consequence will be overshadowed by the respiratory effect in conventional treatments. However, nonrespiratory kidney translations may be of importance for respiratory gating technology, where a relevant decrease or increase in dose might occur depending on the average kidney position relative to the isodose projections.
Determination of the average organ and tissue position over the course of treatment is important for designing accurate margins for the CTV. It is unclear whether the pooled data of the present study are representative, since the distention states were somewhat artificial. A complete analysis of physiologic organ translations on the accuracy of treatment planning would require the measurement of organ positions for each planning CT protocol and for each treatment. Obtaining such measurements is obviously not practical. The sampling procedure performed in our study, in which three CT protocols were used in each patient, resulted in a considerable range of organ translations. Therefore, the implications on the appropriate CTV expansion warranted further investigation.
Optimal margins for the CTV in conventional treatments must account for physiologic organ motion and setup errors, as well as the volume of normal tissues within the high-dose regions. The control of organ motion is a more complex challenge than setup errors, owing to the difficulty in establishing of their position on a megavoltage portal image and the variety of influencing factors. Larger margins reduce the risk of underdosage but lead to a higher volume of irradiated normal tissues. CTV margins for nonrespiratory effects have been estimated from anecdotal information. Although the results of the present study need to be interpreted with caution, the data may help to assess the adequacy of the margins currently in use and imply the necessity for additional studies.
Both nonrespiratory and respiratory organ motion is associated with the patient at treatment. In conventional irradiation techniques, margins around the CTV are enlarged to consider the types of uncertainties. Quantification of organ motion owing to respiration may easily be obtained with the two-phase CT planning technique, which allows the delineation of organ contours by using the spatial information of static exhalation and inhalation positions intersected in one CT data set (19). Alternative strategies currently under investigation are ventilatory gating of abdominal irradiation and inverse planning techniques that incorporate spatial probability distribution of targets and structures in the dose distribution (8,20).
In conclusion, significant organ translations occur in response to a differential gastrointestinal distention. The translations can introduce a systematic error, because the oral contrast-enhanced CT protocols represent the special setting under which a planning CT may be used in pancreatic cancer. With the assumption that the target organ positions are represented by a three-dimensional Gaussian distribution, CTV expansions were calculated to account for nonrespiratory organ motion and a typical setup error. These results may provide a basis for further studies on organ motion in the abdominal region and ways of modifying treatment margins.
|
FOOTNOTES
|
|---|
Abbreviations: CTV = clinical target volume,
SMA = superior mesenteric artery,
VOI = volume of interest
Author contributions: Guarantors of integrity of entire study, E.H., O.M., N.A.W.; study concepts, E.H., O.M.; study design, E.H., O.M., U.S.; literature research, E.H., O.M.; clinical studies, E.H., A.S., U.S.; data acquisition, E.H., A.S., C.M.; data analysis/interpretation, E.H., O.M., C.M., A.S.; statistical analysis, E.H., O.M., C.M.; manuscript preparation, E.H.; manuscript definition of intellectual content, E.H., O.M., C.M.; manuscript editing, E.H., A.S., N.A.W.; manuscript revision/review, E.H., O.M., A.S.; manuscript final version approval, N.A.W., U.S., O.M., E.H.
|
REFERENCES
|
|---|
-
Higgins PD, Sohn JW, Fine RM, Schell MC. Three-dimensional conformal pancreas treatment: comparison of four- to six-field techniques. Int J Radiat Oncol Biol Phys 1995; 31:605-609.
-
Steadham AM, Liu HH, Crane CH, Janjan NA, Rosen II. Optimization of beam orientations and weights for coplanar conformal beams in treating pancreatic cancer. Med Dosim 1999; 24:265-271.
-
Chu JCH, Lawrence JS, Chin CH, Kessler H, Hanks GE. Three-dimensional dosimetric comparison of radiation therapy treatment planning of the pancreas. Med Dosim 1993; 17:199-203.
-
Gunderson LL, Willett CG. Pancreas and hepatobiliary tract In: Perez CA, Brady LW, eds. Principles and practice of radiation oncology. Philadelphia, Pa: Lippincott-Raven, 1997.
-
Kutcher GJ, Mageras GS, Leibel SA. Control, correction, and modeling of setup errors and organ motion. Semin Radiat Oncol 1996; 5:134-145.
-
Ahmad NR, Huq MS, Corn BW. Respiration induced motion of the kidneys in whole abdominal radiotherapy: implications for treatment planning and late toxicity. Radiother Oncol 1997; 42:78-90.
-
Suramo I, Paivansalo M, Myllyla V. Craniocaudal movements of the liver, pancreas and kidneys in respiration. Acta Radiol Diagn 1984; 25:129-131.
-
Balter JM, Lam KL, McGinn CJ, Lawrence TS, Ten Haken RK. Improvement of CT-based treatment planning models of abdominal targets using static exhale imaging. Int J Radiat Oncol Biol Phys 1998; 41:939-943.
-
Ten Haken RK, Forman JD, Heimburger DK, et al. Treatment planning issues related to prostate movement in response to differential filling of the rectum and bladder. Int J Radiat Oncol Biol Phys 1991; 20:1317-1325.
-
Zelefsky MJ, Crean D, Mageras GS, et al. Quantification and predictors of prostate position variability in 50 patients evaluated with multiple CT scans during conformal radiotherapy. Radiother Oncol 1999; 50:225-234.
-
Prott FJ, Schonekaes K, Preusser P, et al. Combined modality treatment with accelerated radiotherapy and chemotherapy in patients with locally advanced inoperable carcinoma of the pancreas: results of a feasibility study. Br J Cancer 1997; 75:597-601.
-
Fuhrman GM, Charnsangavej C, Abbruzzese JL, et al. Thin-section contrast-enhanced computed tomography accurately predicts the resectability of malignant pancreatic neoplasms. Am J Surg 1994; 167:104-111.
-
Lattanzi JP, Fein DA, McNeely SW, Shaer A, Movsas B, Hanks GE. Computed tomography-magnetic resonance image fusion: a clinical evaluation of an innovative approach for improved tumor localization in primary central nervous system lesions. Radiat Oncol Invest 1997; 5:195-205.
-
Hill DLG, Hawkes DJ, Gleeson MJ, et al. Accurate frameless registration of MR and CT images of the head: applications in planning surgery and radiation therapy. Radiology 1994; 191:447-453.
-
International Commission on Radiation Units and Measurements. ICRU Report 50: prescribing, recording, and reporting photon beam therapy Washington, DC: International Commission on Radiation Units and Measurements, 1993.
-
Austin-Seymour M, Kalet I, McDonald J, et al. Three dimensional planning target volumes: a model and a software tool. Int J Radiat Oncol Biol Phys 1995; 33:1073-1080.
-
Winter TC, Ager JD, Nghiem HV, Hill RS, Harrison SD, Freeny PC. Upper gastrointestinal tract and abdomen: water as an orally administered contrast agent for helical CT. Radiology 1996; 201:365-370.
-
Horton KM, Eng J, Fishman EK. Normal enhancement of the small bowel: evaluation with spiral CT. J Comput Assist Tomogr 2000; 24:67-71.
-
Aruga T, Itami J, Aruga M, et al. Target volume definition for upper abdominal irradiation using CT scans obtained during inhale and exhale phases. Int J Radiat Oncol Biol Phys 2000; 48:465-469.
-
Li JG, Xing L. Inverse planning incorporating organ motion. Med Phys 2000; 27:1573-1578.
TOP
ABSTRACT
INTRODUCTION
