Characterization of Pancreatic Tumor Motion Using 4D MRI: Surrogates for Tumor Position Should be Used With Caution

Reviewer: Charles B. Simone, II, MD
Abramson Cancer Center of the University of Pennsylvania
Last Modified: October 30, 2007

Presenter: Mary Feng, MD
Presenter's Affiliation: University of Michigan Medical Center
Type of Session: Plenary


  • Pancreatic cancer is associated with poor outcomes and limited long-term survival, and there are numerous challenges that treating physicians encounter. Less than 20% of patients are operable at the time of diagnosis. Additionally, the delivery of radiation doses high enough to improve local control or survival is often difficult to achieve due to treatment toxicity to nearby organs at risk.   
  • Due to organ motion, there is significant uncertainty as to the pancreatic tumor location during irradiation. This has historically led to large PTV expansions, which has resulted in increased normal tissue toxicity.
  • The authors in this trial used four-dimensional (4D) MRI, which employs three-dimensional imaging at a sequence of respiratory phases to allow for spatial and temporal information on shape and mobility to be acquired synchronously, to evaluate pancreatic tumor margin motion and the relationship of the tumor margins to the positions of the abdominal wall and diaphragm.
  • The purpose of this investigation was to individualize PTV margins and assess the accuracy of using surrogates for pancreatic tumor position, such as the abdominal wall or diaphragm.

Materials and Methods

  • Seventeen patients with unresectable pancreatic cancer were examined in this prospective study.
  • Participants were imaged free-breathing using non-contrast dynamic MRI on a 3T scanner.  Images were taken at a rate of three images per second for 60 seconds in single coronal and sagittal planes. 
  • Images were obtained before and during therapy with chemoradiation. 
  • Tumor volumes were contoured in collaboration with a radiation oncologist and an abdominal MRI radiologist.  
  • Tumor margin motion in the superior (sup), inferior (inf), anterior (ant), posterior (post), and lateral directions were quantified using an in-house functional imaging analysis tools program.
  • Investigators determined the average amplitudes and standard deviations of tumor margin motion, evaluated correlations between motion of different tumor margins, diaphragmatic motion and abdominal excursion, and formulated a time-weighted distribution of tumor position that was relative to end-exhalation.


  • During inspiration, pancreatic tumors moved inf and ant. Mean margin motion in coronal slices was 13 mm sup and 14 mm inf, while mean motion in sagittal slices was 14 mm sup, 13 mm inf, 6 mm ant, 4 mm post. Lateral motion for both was not significant.  
  • The average motion differences between sequential scans for a single patient ranged from two to six millimeters.  
  • Only sup and inf margin movement was correlated with consecutive slices on a single slice orientation (p < 0.0001 for both), confirming pancreatic deformation during breathing.
  • Diaphragmatic motion and abdominal wall excursion were both found to poorly correlate with all tumor margin motions. Good correlation was obtained with the diaphragm in only 15% of cases, while the abdominal wall allowed for good correlation in only 18% of cases.
  • The inf margin was found to be more than 10 mm from end-exhale position 10–12% of the time but was greater than 15 mm only 2–3% of the time.  Similarly, the sup margin was more than 10 mm from end-exhale position 9–11% of the time but greater than 15 mm only 2% of the time.

Author's Conclusions

  • Pancreatic tumors margin motion is highly variable between patients and is generally larger than traditionally accounted for when determining target volumes.
  • Breathing was shown to cause substantial deformation of the pancreas, while motion margin was shown not to correlate well with abdominal wall or diaphragmatic motion.
  • Therefore, radiation oncologists should use caution when designing treatment plans that rely on the motion of a single fiducial or surrogates such as the abdominal wall or diaphragm for tumor tracking.
  • Treatment designs should instead use time-weighted tumor position in dose calculation since tumors were found to spend only a small proportion of time >10 mm from end exhale.

Clinical/Scientific Implications

This study demonstrates that the margins of pancreatic tumors are significantly influence by organ motion with breathing. To account for this, radiation oncologists have historically employed large PTV expansions in an attempt to ensure treatment of the entire tumor. However, according to the findings in this study, a traditional 1 cm margin is not adequate since the superior and inferior margins were found to spend a significant amount of time greater than 10 mm from end-exhale position. This may contribute to underdosing of the tumor volume and the high local failure that is seen among patients with pancreatic cancer. Additionally, current tracking or gaiting strategies should be used with caution, as they were shown to correlate poorly with tumor margin motion. The increased use of 4D treatment planning could help to minimize underdosing of tumor volumes and overdosing of normal tissues, thus limiting treatment toxicity. It should be noted that larger tumors will have less motion than smaller malignancies, and further investigation will be needed to determine the differences in motion between tumors at the head versus the body or tail of the pancreas.  

Partially funded by an unrestricted educational grant from Bristol-Myers Squibb.