RADIATION PHYSICS

Radiation Physics is responsible for the accurate calculation and reliable delivery of the prescribed radiation dose for patients receiving radiation treatment. Research in this group involves the development of new modes of radiation treatment, techniques for improving the calculation of the dose distribution, methods for improving the dose delivery, methods for assessing the accuracy with which the prescribed dose is realized, and the analysis of treatment results.

The overall objective of this study is to characterize fully the dose and volume response of gastrointestinal (GI) and genito-urinary (GU) morbidity from conformal therapy of cancer of the prostate. No effective cancer treatment is without risk, but characterizing the risk versus the benefit is essential to the treatment decision. The database in the Department of Radiation Oncology at Fox Chase contains extensive data on more than 1400 patients with cancer of the prostate. Using these data, we are establishing the dose and volume response, as well as the importance of comorbidities, for acute grade 2, as well as late grade 2 and 3, GI and GU complications of cancer of the prostate. The specific components of this project are as follows: 1) To determine the factors that are significantly related to the incidence of acute grade 2 GI and GU morbidity. Analyzing GI and GU morbidity separately, logistic regression is used to assess the statistical relationship of tumor factors, comorbidities, treatment factors, and factors in the medical history to these morbidities. 2) To determine the factors that are significantly related to the incidence of late grade 2 and grade 3 GI and GU morbidity. Time at risk is variable, as is the time to onset of symptoms. Therefore, the proportional hazards model, which accounts for censoring, is used to determine the statistical significance of factors (listed above) putatively related to these late effects. 3) To fit dose response models to the incidence data for late grade 2 and grade 3 GI and GU morbidity, accounting for the relative risk of factors determined by the proportional hazards analysis. The proportional hazards modeling can determine whether dose is significantly related to complications, but it does not produce a dose-response function. The dose-volume response function is obtained by using maximum likelihood estimation to fit various dose-volume response models to our data. 4) to determine whether these data suggest if any of the treatment techniques used in conformal radiation treatment of the prostate can be shown to be significantly superior. The results of dose-volume response modeling will be used to predict the incidence of complications for techniques used by the major cancer centers in the country to determine if any technique is clearly superior.

Dose distribution plays an important role in the outcome analysis of radiation treatment; however, optimum dose distribution cannot be achieved in a majority of patients due to technical difficulties. Traditionally, Co-60 photon beams had been used; however, presently 6 MV is preferred due to the wide availability of this energy in most institutions. A large subset of the patients treated in the United States is relatively obese with the chest wall separation in the range of 22 to 35 cm. Three-dimensional dosimetric analyses are presented with respect to choice of beam energy for an optimum radiation treatment of breast cancer.

Frequency distribution of chest wall separation in our breast database was performed for 1,695 patients. It was noted that chest wall separation varied from 14 to 35 cm. The distribution is a skewed Gaussian with a peak at 22 cm; nearly 60% patients had a chest wall separation greater than 22 cm. Such an observation is not universal; in some countries chest wall separation in breast cancer patients is extremely small. Fifty sequential patients referred for conservative radiation treatment having chest wall separation greater than 22 cm were selected in this study. Patients were computed tomography (CT) scanned with a 3 mm slice thickness and 3 mm step size for a better digitally reconstructed radiograph (DRR), which is used for the comparison of the treatment field. The CT images were transferred to the Voxel Q processor for virtual simulation. The planning target volume (PTV) was defined on each slice to cover the breast tissue adequately. Lung volumes were also delineated on each slice for hetero-geneity correction for all cases. Medial and tangential fields were chosen such that it covered breast tissue adequately with 2 to 3 cm lung tissue. The posterior beam edge was made coplanar and non-divergent using the Fox Chase technique. Treatment planning was performed with and without lung correction. Three possible photon beams (6, 10 and 18 MV) were chosen for treatment planning. The same beam energy was utilized in each planning, but optimized properly with wedge and beam weight. Dose normalization was performed as described by Das et al. (Radiother. Oncol. 44: 83, 1997). Analysis was performed with hot spot, chest wall separation, and breast volume with and without lung correction.

Results reconfirm our previous observation (Das et al., Radiother Oncol 44:83, 1997) that chest wall separation is a sole parameter that dictates a hot spot in breast tissue. Breast volume could not be correlated with the hot spot. Figure 1 shows the hot spot with respect to the chest wall separation for three beam energies. The choice of beam energy becomes critical if the hot spot in breast tissue is tolerated to a minimum of 10%. Based on our data, a 6 MV beam is ideal for a chest wall with separation less than 21 cm. A 10 MV beam is needed if the chest wall is less than 25 cm. A higher energy (18 MV) is desired for patients with large breast and chest wall separation, greater than 28 cm. This finding is extremely important for a suitable treatment of breast cancer. If higher energies are used, the surface doses are significantly reduced. To augment radiation dose to the superficial part of breast tissues, high-energy beams must be treated with a beam spoiler so that adequate surface dose is maintained for better cosmetic results.

The magnitude of measured dose perturbation for high Z inhomogeneity is significant (15 to 20 fold) for kilovoltage Xrays, depending upon beam energy and the distance from the interface (Das and Chopra, Med. Phys. 22:767, 1995; Das, Med. Phys. 24:1781, 1997). Regulla et al. (Radiat. Res. 150:92, 1998) recently showed that the magnitude of dose enhancement could be as high as 100-fold at interface, based on thermally stimulated exoelectron emission (TSEE). Dosimetry of low energy is often difficult due to technical error, as well as uncertainty in the radiological parameters (stopping power, ionization potential, energy absorption coefficients, etc.) To overcome the discrepancies, the Monte Carlo approach is used.

Initial simulations were performed with EGS4, MCNP4B and Penelope Monte Carlo codes. These simulations significantly differed with the measured data recently published by Das (Med. Phys. 24:1781, 1997). It was realized that the differences could be due to the initial spectra used for kilovoltage beams. Dr. Moskovin used a low energy cut off Monte Carlo code and simulated the interface effect for monoenergetic beams of Cs-137 gamma rays. Figure 2 shows the interface effect of Cs-137 gamma rays in a homogeneous water medium with Pb interface. Photon trajectories for 300,000 beams were simulated. The statistical variance of the results in calculated dose values is not more than ±5%. Trajectories of electrons traced up to cut off energy E_min of 1 keV. The secondary electron spectra for homogenous and heterogeneous media are compared, which show significant differences at lower energy. This could explain the discrepancies in measurements and simulations. Monte Carlo simulation is critical in understanding the physical dose that relates to the biological effects of low energy and track-end electrons produced from high Z interfaces, which is under investigation.

MODULATION OF MULTI LEAF COLLIMATOR DOSE UNDULATION. DAS, MITRA

The effect of dosimetric undulation due to beam edge angle was investigated in this study for a double-focused Siemens multileaf collimator (MLC) whose characteristics were recently reported by Das et al. (1). The jagged beam edge of an MLC field visible radiographically at any depth can be quantified by undulation index (UI). In this study, UI is defined as the ratio of the width of the 50% undulation to the 20 to 80% penumbra width.

The undulations could be eliminated by changing the MLC pattern or by translating the treatment table. This portion of the study was conducted to investigate the feasibility of modulating the undulation to approximate the block edge dose distribution. Using a maximum undulation at 45 ° leaves, thereby producing a triangular field, the treatment table was incremented from its original position to a maximum of 8 mm in the x and y directions. In each case, CEA films (Cheng and Das, Med. Phys. 23:1225, 1996) at 5 cm depth were exposed with a millimeter table increment perpendicular to the MLC beam edge, and scanned with a laser film densitometer for isodose patterns.

Results show that table translations in the x and y directions do not provide improvements in beam edge dose distribution with MLC; however, when the table is increased perpendicular at step, the UI changed significantly. The pattern of UI with respect to the table increment also follows similar to a parabola with a minimum UI at 5 mm step. For an MLC width of 10 mm, such an observation is very analogous to the Nyquist frequency concept. Figures 3a and 3b show the approach of modulation of undulation. The dose distributions for a 45° beam edge with MLC leaves with maximum undulation is noted in Figure 3a. With a 5 mm perpendicular translational step, the dose distribution is noted in Figure 3b indicating no undulation and dose distribution very close to the blocked field. The success of modulation of undulation pattern through this process is visible from Figure 3. Such a dramatic improvement in visualization can be achieved with little effort by changing the table position. For patient treatment, a given dose can be delivered by several segments, one with original table position and the other with the translation of table perpendicular to the MLC leaves. Such smearing improves the dose distribution and maintains nearly the same penumbra for the 20 to 80% isodose lines; however, there is slight increase in 10 to 90% isodose penumbra with this method. The table translation can be easily achieved with a modern ZXT table with digital interface, and record and verification systems. The manufacturer may decide to provide the translational option in their software such that the MLC jaggedness can be totally eliminated. The translational approach opens a new dimension in the use and acceptance of MLC fields for routine clinical use.

BAT: ULTRASOUND GUIDED PATIENT POSITIONING FOR PROSTATE CANCER. McNEELEY

A new ultrasound-based patient-positioning device, BAT (Nomos Corporation, Sewickley, PA), has undergone initial testing and is now in clinical use at Fox Chase by the Department of Radiation Oncology. BAT, initially designed to be used in the treatment of prostate cancer, uses CT data in conjunction with real time transabdominal ultrasound imaging to three dimensionally locate the prostate and direct external beam radio-therapy treatments. Knowing the absolute position of the prostate at the time of treatment will allow for a smaller volume of normal tissue to be irradiated, leading to a lower complication rate for patients. Fox Chase was one of only two institutions in the world to do initial testing and validation of this device.

One of the major goals in treating cancer with radiotherapy is to maximize radiation dose to the target, while sparing the surrounding normal tissues. To minimize the normal tissue irradiated, each beam of radiation in external beam radiotherapy is shaped to match the profile of the target. Unfortunately, this task is complicated by the fact that the target can move day-to-day within the body. Also, patient treatment positioning can vary slightly during the course of treatment. The net effect of these motions is a treatment portal that is larger than the actual tumor size. A larger portal size guarantees the target is hit everyday, but at the cost of extra normal tissues receiving high dosages. The use of BAT allows the absolute three-dimensional position of the prostate to be known at the time of treatment. Since the problem of setup error and organ motion is now minimized, smaller radiation portals can now be effectively used and lower patient complication rates realized.

Waterman et al. (Int. J. Radiat. Oncol. Biol. Phys. 41:1069, 1998) has shown that the prostate gland swells 15-20% soon after an I-125 brachytherapy procedure. The edema was shown to fall exponentially to a normal level within a week. One could, therefore, expect that the planned radiation dose might not be delivered to the actual prostate because of the changes in volume. To address this issue, we undertook a study to investigate the effect of needle placement in the prostate gland. Our initial observations were substantially different from those published by Waterman et al.

Selected patients had a baseline CT study and then underwent an ultrasound (US) guided process to estimate the volume. The I-125 implants were planned to deliver a minimum of 145 Gy to the PTV based on the recommendations of AAPM TG-43 (Nath et al., Med. Phys. 22:209, 1995). The PTV was defined per recommendations of the RTOG 98-05 (Ultrasound Guided Implant of the Prostate) to be expanded by 2 to 3 mm in the later dimension for each trans-rectal ultrasound (TRUS) image, while maintaining the same posterior border of the prostate. On any given plane, seeds are implanted in needles that are 10 mm apart. The treatment planning was performed on planes that are 5 mm apart and a peripheral loading technique was employed.

Post-implant CT studies were obtained and evaluated weekly. The original CT scan and the implant scans were fused using image registration software. Each subsequent scan was fused to estimate the differences in volume to measure possible edema.

The results showed that using our implant procedure, there is minimal edema in the lateral dimension (Figure 4). These images are the first and four week post implant CT images fused together using Voxel Q software. A volume reduction in the base area of the prostate is clearly visible. The prostatic edema is observed minimally in the first 12 to 15 mm of the prostate from the base. At more inferior levels, very little or no changes are observed over a time period. Image fusion allows the capabilities of overlaying prostate volume in each CT slice. Post implant dosimetry was performed in correlation with variable angle radiographs for determining the 3-dimensional distribution of seeds within the prostate volume.

Our results are significantly different from those referenced above. We hypothesize that these differences could be attributed to using preloaded needles and the peripheral loading technique. The pre-loaded needle method of implanting the prostate produces less trauma than the Mick Applicator method because needles are inserted less deeply and reinsertion is rare. Implanting a larger volume, as in the peripheral loading technique, allows a greater spacing between needles. Reduced post-implant edema should result in more accurate dosimetry. Image registration techniques allow us to localize these temporal variations in volume and evaluate the contribution of these perturbations on the dose to the prostate over the life of the implant. Long-term dosimetry is currently being studied for each case.

A major advancement for the treatment of cancer in recent years has been the development of conformal radiation therapy. The goal of conformal therapy is to achieve a dose distribution that conforms to or follows the shape of the target and spares nearby normal tissues. This is typically accomplished through a combination of beam directions and shapes. Beam directions are chosen to minimize the volume of normal organs intercepted by the beams and the beams are shaped to fit the shape of the target.

The newest and most advanced form of conformal therapy is known as intensity-modulated radiation therapy (IMRT). With IMRT, the beam is divided into many segments and the intensity of each segment is variable. Adjusting the intensity of every segment of every beam would be exceedingly complex and time-consuming using conventional treatment planning methods. For this reason, we use a technique known as "inverse planning," which allows us to describe the desired dose distribution first. An optimization algorithm then determines the segmentation and the resulting intensity of each beam segment that achieves as nearly as possible this distribution.

This technology offers the potential for improved local tumor control through increased dose conformity and dose escalation; however, the techniques, as well as the software and hardware, utilized are atypical compared with those used for conventional radiation therapy. Validation of the resultant dose distributions must be thoroughly tested and a viable quality assurance program put in place before this system can be clinical implemented. At this time, major efforts are being directed towards the development of this quality assurance program and the commissioning of the software.

1.   REGULLA, D.F., HIEBER, L.B., SEIDENBUSCH, M. Physical and biological interface dose effects in tissue due to X-ray induced release of secondary radiation from metallic gold surfaces. Radiat. Res. 150:92-100, 1998.

DAS, I.J., DESOBRY, G.E., MCNEELEY, S.W., CHENG, E.C., SCHULTHEISS, T.E. Beam characteristics of a retrofitted double-focused multileaf collimator. Med. Phys. 25:1676-1684, 1998.

CHAPMAN, J.D., STOBBE, C.C., GALES, T., DAS, I.J., ZELLMER, D.L., BIADE, S., MATSUMOTO, Y. Condensed chromatin and cell inactivation by single-hit kinetics. Radiat. Res. (in press).

CHENG, C.W., DAS, I.J. Treatment plan evaluation using dose-volume histograms (DVH) and spatial dose volume histogram (zDVH). Int. J. Radiat.Oncol. Biol. Phys. (in press).

DAS, I.J., AKBER, S.F. Ion recombination and polarity effect of ionization chambers in X-ray exposure measurements. Med. Phys. 25:1751-1757, 1998.

DAS, I.J., CHENG, E.C., FREEDMAN, G., FOWBLE, B. Lung and heart volume analyses with CT simulator in tangential field irradiation of breast cancer. Int. J. Radiat. Oncol. Biol. Phys. 42:11-19, 1998.

DAS, I.J., DESOBRY, G.E., McNEELEY, S.W., CHENG, E.C., SCHULTHEISS, T.E. Beam characteristics of a retrofitted Siemens multileaf collimator. Med. Phys. 25(9):1676-1684, 1998.

DAS, I.J., MCNEELEY, S.W., CHENG, C.-W. Ionization chamber shift correction and surface dose measurements in electron beams. Phys. Med. Biol. 43:3419-3424. 1998.

DAS, I.J., CHENG, C.W., AKBER, S.F. Low monitor units: Significance in special therapy. Phys. Med. Biol. (in press).

HANKS, G.E., HANLON, A.L., SCHULTHEISS, T.E., PINOVER, W.H., MOVSAS, B., EPSTEIN, B.E., HUNT, M.A. Dose escalation with 3D conformal treatment: Five year outcomes, treatment optimization, and future directions. Int. J. Radiat. Oncol. Biol. Phys. 41(3):501-510, 1998.

LATTANZI, J., MCNEELEY, S., HANLON, A., DAS, I.J., SCHULTHEISS, T.E., HANKS, G.E. Daily CT localization for correcting portal errors in treatment of prostate cancer. Int. J. Radiat.Oncol. Biol. Phys. 41:1079-1086, 1998.

ORTON, C.G., CHUNGBIN, S., KLEIN, E.E., GILLIN, M.T., SCHULTHEISS, T.E., SAUSE, W.T. Study of lung density corrections in a clinical trial (RTOG 88-08). Int. J. Radiat. Oncol. Biol. Phys. 41(4):787-794, 1998.

VERHAGEN, F., DAS, I.J., PALMANS, H. Monte Carlo dosimetry study of a 6 MV stereotactic radiosurgery unit. Phys. Med. Biol. 43(10):2755-2768, 1998.

WATERMAN, F.M., YUE, N., CORN, B.W., DICKER, A.P. Edema associated with I-125 or Pd-103 prostate brachytherapy and its impact on post-implant dosimetry: An analysis based on serial CT acquisition. Int. J. Radiat. Oncol. Biol. Phys. 41:1069-1077, 1998.

DAS, I.J. Broad beam attenuation of kilovoltage photon beams: effect of ion chambers. Br. J. Radiol. 71:68-73, 1998.

DESOBRY, G.E., WALDRON, T.J., DAS, I.J. Validation of a new virtual wedge model. Med. Phys. 25(1):71-72, 1998.

ZELLMER, D.L., CHAPMAN, J.D., STOBBE, C.C., XU, F., DAS, I.J. Radiation fields backscattered from material interfaces: I. Biological effectiveness. Radiat. Res. 150(4):406-415, 1998.

^§   Fox Chase researcher

^a   G.E. Desobry: Present address--Department of Physics, University of Pennsylvania, Philadelphia, PA 19104

^b   N. Shikama: Shinshu University Medical Center, Nagano, Japan

^c   C.W. Cheng: University Of Arizona, Tucson, AZ 85721

^d   F. Verhaegen: Gent University, Gent, Belgium

^e   V. Moskovin: Kharkov State University, Karhkov, Ukraine

Illustrations or unpublished data in these reports should not be used without permission of the author.

Fox Chase Cancer Center

Scientific Report 1998