Reproducibility of Internal Target Positions for Breath-held Conformal External-beam Radiotherapy

W. G. O'Dell, C. R. Maurer, Jr., M. C. Schell, A. Sandhu, P. Okunieff
Department of Radiation Oncology, University of Rochester, 601 Elmwood Avenue, Box 670, Rochester, NY 14642, USA
Department of Neurosurgery, Stanford University, 300 Pasteur Drive, Room S-012, Stanford, CA 94305-5327, USA
I. Introduction
Recent advances in Stereotactic Radiosurgery/Conformal Radiotherapy have made it possible to deliver precise radiation therapy to small lesions while preserving function to surrounding structures. Unfortunately, the application of 3D conformal radiotherapy to mobile tumors in the lung and liver is geared toward slowing the progression of disease rather than obtaining a cure. Here, the traditional therapeutic approach is to measure the range over which the tumor moves during the respiratory cycle and to then irradiate a volume that encloses the entire tumor over its entire motion range. The oncologist’s dilemma is that prescribing a lethal radiation dose to the area would not only kill the tumor but also damage a sufficiently large volume of healthy tissue to cause significant clinical repercussions, including death. Our ultimate goal is to hit, with a very focused and high-dose radiation beam, moving targets within the body with such high precision that we will cure these cancer patients of their disease while sparing the surrounding healthy tissue.

II. Objectives
The fundamental physiologic questions relevant to this approach are:
  • What are the respiratory-derived motions typical of lesions in the lung and liver?
  • What is the reproducibility in lesion positioning using end-expiration breath-holding?
    and
  • Can we use passive breathing and/or breath-holding without additional complications for the treatment of lung and liver tumors?

III. Methods
1. Imaging
Three sets of chest MRI volume data were acquired during periods of breath-hold (BH) at relaxed end-expiration in a patient with five ~8--15mm diameter lung lesions. The patient was coached to perform 2 deep inspiration breathing cycles followed by a relaxed exhale and hold. This same procedure was performed during the planning CT and radiotherapy treatment sessions. Each MRI 3D dataset consisted of 168 overlapping sagittal slices with slice thickness 4 mm and slice center-to-center separation of 1 mm. Breath holding for the duration of acquisition (~32s) was well-tolerated by the patient. An MRI end-inspiration BH dataset was also acquired to compute total respiratory motion.

2. Position Variability
For each of the 5 lesions, the standard deviations [mm] about the mean position (assuming a Gaussian distribution) were computed over the 3 end-expiration breath-hold trials. These values are similar to those found previously[1,2] for end-expiration breath-hold reproducibility of diaphragm position.

Lesion 1Lesion 2Lesion 3Lesion 4Lesion 5Average
S-I 2.3 2.9 2.0 2.0 2.2 2.3
A-P 1.3 2.1 2.8 1.4 1.3 1.8
R-L 0.8 1.5 1.5 0.8 0.8 1.1
3D distance 2.9 3.9 3.7 2.5 2.6 3.1

IV. Radiotherapy Planning
1. Margin Specification: 7x7x10mm [APxRLxSI]
s 2s 3s
68% 95% 99.87%
The specified margin is > 3 s in each direction. Thus, there is > 99.87% certainty of enclosing the entire tumor on any given treatment fraction.
2. Radiation Dose
For each tumor, two arcs were prescribed, offset by 20 degrees (the yellow and blue arcs shown in the figure). Each fraction was administered during a separate ~30s breath-hold. The 2-arc protocol was repeated 10 times over a 2-week period to give a total tumor exposure of 50 Gy using 6 MV X-rays.
3. Patient Repositioning using ExacTrac
Automatic registration of 8 surface markers affixed to the patient’s chest and abdomen was performed using the ExacTrac system (BrainLAB, AG) incorporated into our Novalis treatment facility. The positions of all 8 skin markers were monitored in real-time in all three dimensions to determine the extent of variation in their position. A deviation of any surface marker position exceeding 2mm in any direction, using the initial planning scan locations for reference, signaled cessation of treatment until the patient repositioned himself through another relaxed exhalation.
4. Lesion Localization Verification
Repeat CT scans were obtained thrice during the 2-week treatment. These scans were utilized to perform “virtual” treatments whereby the isodose distributions from each plan were superimposed on each target for all time points. Figure: repeat-day tumor contours overlaid onto the planning CT for Lesion 3. The central yellow, purple and pink curves mark the day 1, day 3 and day 6 locations.

V. Treatment Results
At the 7-week follow-up 4 of the 5 tumors had completely disappeared and the volume of the radio-dense material at the site of the remaining lesion had decreased significantly [Graph: CT1 is at treatment day 1; CT2 is at day 3; CT3 at day 6; CT4 at day 9; and FU at 7-week Follow-Up]. There were no indications of clinical side-effects and pulmonary function was normal. At the 6-month follow-up, a radiologist determined that all tumors had been completely eradicated, however there was evidence of pneumonitis at the sites of lesions 1&3.
Lesion 3: Before 7-week Post 6-month Post

VI. Conclusions
The variability in target 1D position over repeated end-expiration breath-holds was found, on average, to be 1.1—2.3mm with the largest motion in the SI direction. Using these values as a guide, radiotherapy treatments were planned and administered to 5 lung lesions in a single patient. 50Gy doses were given to each lesion while maintaining a favorable dose-volume histogram for the surrounding lung tissue. We have shown the feasibility of using breath-holding and surface marker registration to perform fractionated external-beam radiotherapy of internal, mobile targets. Although a curative response of these tumors was likely achieved, our future work will focus on obtaining a larger number of trials and better statistical analysis of tumor position variability to enable us to further limit the margin size, decreasing the incidence of pneumonitis. Real-time motion modeling and beam gating are also being developed.

  1. Balter, J., et al.
    Improvement of CT-based treatment-planning models of abdominal targets using static exhale imaging.
    International Journal of Radiation Oncology, Biology, Physics, 1998. 41(4): p. 939-43
  2. Holland, A., J. Goldfarb, and R. Edelman
    Diaphragmatic and cardiac motion during suspended breathing: preliminary experience and implications for breath-hold MR imaging.
    Radiology, 1998. 209(2): p. 483-9

Acknowledgments
The Authors wish to acknowledge the support of BrainLAB GmbH, and the University of Rochester‘s James. P. Wilmot Cancer Center