ALPHA CRADLE® BRAND IMMOBILIZATION DEVICE REDUCES POSITIONING ERRORS IN PROSTATE CANCER PATIENTS TREATED WITH SIX-FIELD CONFORMAL RADIATION THERAPY

Seth A. Rosenthal, MD
University of California
Sacramento, California

INTRODUCTION
Cancer of the prostate is the most common malignancy in American males (excluding skin cancers), with an estimated 132,000 cases diagnosed in 1992.2 Of these patients, a significant percentage will present with localized disease amenable to potentially curative therapy with the local treatment modalities of radiation therapy (RT) and/or surgery. As physicians choose which therapeutic modalities to recommend for their patients, they must consider the probability of tumor control as well as the potential for morbidity that each therapeutic option may entail.
The use of the dimensional conformal radiation therapy (3D-CRT) for patients with prostate cancer has been developed in recent years. 3D-CRT techniques enable the design of isodose surfaces which accurately conform to the target volume. This allows for reduction in the dose delivered to adjacent normal structures, permitting trials of dose-escalation to be initiated. 3,9,10 In addition, the use of 3D-CRT has been show to result in a reduction in acute morbidity associated with RT for prostate cancer,6,11 although longer-term follow-up will be necessary in order to determine if late effects will also be diminished.
At present, a six-field coplanar conformal technique, similar to that described by Ten Haken et al15 is being used in our department for the entire treatment course or as a boost following treatment of the pelvic lymph nodes. As target volumes include not only tumor volume, but also a margin of error for variability in patient positioning, we set about to determine the variability that occurs in patient positioning during 3D-CRT, both with and without immobilization, in order to ascertain the amount of variability to be expected during RT, and to determine if the routine use of immobilization reduced variability in patient positioning.
Inability to deliver the prescribed dose of RT to the tumor volume has resulted in reduced cure rates,5 and the uncertainty created by variability in patient positioning has been identified as a factor contributing to the overall spatial uncertainty.13 Immobilization devices may be useful in reducing the degree of spatial uncertainty secondary to positioning error. The use of immobilization may be especially important in the pelvis, where treatment to treatment variations may be larger than in other body sites. In a study examining variability in patient positioning for a variety of anatomic sites, a group from the Massachusetts General Hospital noted that 20% of the pelvic portals had positioning errors of > 1.19cm. This was higher than for any other body site studied.7 The relatively large uncertainties in pelvic positioning, perhaps not clinically significant when large pelvic fields were treated, are not appropriate for the treatment of prostatic carcinoma using tight margins and 3D-CRT.
Conformal techniques are inherently vulnerable to uncertainties in patient positioning. They are dependent on accurate definition of, and tight blocking around, the target volume so that a minimum of normal tissue will be irradiated to high dose. As part of a continued program to develop conformal radiation therapy for prostatic carcinoma, we undertook a study to determine the magnitude of patient positioning errors associated with six field conformal therapy for carcinoma of the prostate, and to assess the impact of ALPHA CRADLE® immobilization on these errors.

MATERIALS AND METHODS
The study was conducted at two of the treatment facilities within the Department of Radiation Oncology at the University of California, San Francisco (UCSF): the facility at UCSF and the facility at the University of California, Davis (UCD). From August 1991 to February 1992, 22 patients were treated using CT planned six field conformal therapy for carcinoma of the prostate, either for the entire course of treatment or as a boost following pelvic irradiation. At one facility (UCD), patients were routinely treated with immobilization, while at the other (UCSF) no rigid immobilization was employed. Ten patients were treated with immobilization, and 12 were treated without immobilization. ALPHA CRADLE® (Smithers Medical Products, North Canton, OH) Patient Repositioning Systems, similar to those described by Soffen et al,12 were used to immobilize patients at UCD, with patients undergoing simulation and treatment planning CT, as well as daily treatments, in their immobilization devices.
A total of 288 films were reviewed from the records of the 22 patients studied. For each film, bony landmarks were used as reference points, and a measurement of the deviation of the treated field from the simulated field (simulation-to-treatment variability) was made in both the Superior-Inferior (SI) and Anterior-Posterior (AP) Dimensions. Magnification effects were taken into account. For each portal, the AP and SI deviations were then calculated. This resulted in a single measurement which represented the deviation of the isocenter of the portal film from the isocenter of the simulation film, as shown in Equation 1:

Isocenter Deviation + [(AP Deviation)² + (SI Deviation)²]1/2

The isocenter deviation values were sorted by magnitude and grouped by percentile. Average and median values were obtained for the cohorts of immobilized and non-immobilized patients. Statistical comparisons were obtained using the SAS software package.
The treatment-to-treatment variability, that is, the deviation of each film from the average portal film position for an individual patient, was also determined. For each patient, the average SI and AP deviation of each film from the average deviation for each patient was then determined. The differences between these two measures of variability have been discussed previously,8 with the simulation-to-treatment variability representing the deviation of each film from the simulation film, and the treatment-to-treatment variability representing the deviation of the individual portals from the average deviation noted for that patient.

RESULTS
A total of 288 portals were reviewed for 22 patients; 10 patients treated with immobilization, and 12 patients treated without immobilization. The average AP portal deviations were 0.3 cm for the immobilized patients, and 0.4 cm for those not immobilized. The corresponding deviations in the SI dimensions were 0.2 cm and 0.4 cm, respectively. There was no significant difference noted between the oblique fields and the lateral fields with respect to AP, SI, or isocenter deviation. The above values translated into an average difference of the isocenter from simulation-to-treatment of 0.4 cm for the immobilized patients, and 0.6 cm for those not immobilized, as shown in Table 1.

Table 1

The distribution of deviation from simulation was grouped into 0.25 cm intervals, and this data is presented in Table 2. For both the immobilized and non-immobilized cohorts, approximately one-quarter of the values were less than 0.25 cm. However, when the number of values less than 0.5, 0.75. and 1.0 cm were compared significant differences were noted with immobilization. 66% of immobilized values were <0.5 cm compared with 43% of the non-immobilized values. 98% of the immobilized portals were within 1.0 cm of the simulated isocenter, compared with 85% of the non-immobilized values.

Table 2

The percentile distribution of simulation-to-treatment variability is presented in Table 3. There is a 0.2 cm difference in variability between the immobilized and non-immobilized cohorts at the 50th percentile. A smaller difference noted at the 10th and 25th percentiles, and a greater difference for the 75th and 90th percentiles. If the 75th percentile is examined, there is a 0.3 cm difference noted with immobilization, and at the 90th percentile, a 0.4 cm difference. The increasing difference in variability without immobilization is demonstrated in Figure 1.

Table 3

Fig. 1

The treatment-to-treatment variability for the films of each patient was also determined (Table 4). For the immobilized patients, the average variability was 0.4 cm, while for the non-immobilized patients it was 0.6 cm. Furthermore, there was a significantly larger number of non-immobilized patients noted with treatment-to-treatment variability of greater than 0.5 cm (p=0.01). These results indicate that those patients treated with immobilization had significantly less variation from their average treatment position.

DISCUSSION
The impact of ALPHA CRADLE® immobilization on positioning error in prostatic carcinoma has previously been examined in patients treated using a conformal, four-field technique. Soffen et al noted an average unidimensional (SP or SI) error of 0.33 cm, which translates to an average deviation of the isocenter of 0.47 cm in those patients who were immobilized.12 These results are comparable to our findings of an average 0.4 cm deviation from simulation in those patients who were immobilized. The consistency of these results indicates that average errors in positioning of less than 0.5 cm from simulation can be expected in patients with prostate cancer if ALPHA CRADLE® immobilization is used.

Table 4

In addition to reduction in simulation-to-treatment variability, the treatment-to-treatment variability was significantly reduced for those patients who were immobilized. Only 1/10 of the immobilized patients had variability of more than 0.5 cm from average position, compared with 8/12 of the non-immobilized patients (p=0.01). This indicates that once a satisfactory position has been achieved for immobilized patients, the radiation oncologist can be assured that comparable set ups will be achieved throughout the course of treatment.
The decrease in variability with patient positioning, although only 0.2 cm at the median value, is higher if one compares the 75th or 90th percentile values. This effect of immobilization in reducing larger magnitude errors more than smaller magnitude errors has been previously noted12 and has important implications. In choosing a margin to the tumor volume to account for uncertainties in patient positioning, the radiation oncologist may select a margin which would encompass the tumor volume despite positioning uncertainties, in the vast majority of cases, rather than a smaller margin which would include the entire volume only 50% of the time. Thus it may be at least as important to consider the reduction in variability in positioning at the 75th or 90th percentile as at the median value, and the decrease in variability with immobilization is most apparent when these levels are compared. The use of immobilization significantly reduced the proportion of higher magnitude positioning errors, especially those greater than 0.5 cm. Over 90% of the immobilized patients had set up errors of <0.75 cm, compared with only 68% of those not immobilized. Optimal selection of tumor margin may make possible reductions in treatment volume which may spare both acute and long term toxicity and allow for dose escalation.
An issue not addressed by this study is potential in vivo movement of the prostate relative to the bladder and rectum. Ten Haken et al noted an average difference in prostate position of 0.5cm relative to the full of empty status of bladder and rectum. This potential for in vivo mobility of target structures, along with variability in patient positioning, represents a potential limitation to the application of 3D-CRT techniques, and should be considered when designing target volumes. We are currently considering approaches to reduce the effect of this in vivo organ movement on tumor position, such as making efforts to consistently plan, simulate, and treat patient with bladder full and rectum empty. However, these are preliminary efforts, and further study of these factors will be important at 3D-CRT for prostate cancer continues to be refined.
Reduction in acute morbidity has been noted in prostate cancer patients treated with conformal techniques as compared to non-conformal techniques. Soffen, Hanks, et al noted that when patients who were treated to small fields with conformal techniques were compared to historical controls treated with non-conformal techniques, a significant reduction in the number of patients noting urinary or rectal symptoms requiring medication and/or unplanned treatment breaks was noted. This reduction in acute symptomatology was attributed to an average reduction of 14% in the volume of both bladder and rectum treated using conformal techniques and immobilization when compared to stage matched controls.11 Similar reduction in acute morbidity have been noted by Leibel et al.6
The reduction in treatment volume associated with conformal therapy may also allow for dose escalation. Early experience from the University of Michigan suggests that toxicity can be maintained at acceptable levels with escalation of doses beyond 70 Gy if conformal techniques are used.10 This may be clinically significant for patients with Stage C disease, where there is a suggestion of dose response relationships for local control above 70 Gy.4 Caution will need to be used in trials of dose escalation, as there is potential for unexpected sever late effects. In a cohort of patients who received a proton beam boost to doses of 75.6 cobalt-Gy-equivalent (CGE), severe rectal injury was noted in 34% of patients, with the likelihood of rectal injury correlating with treatment of >40% of the anterior rectal wall.1 This experience underscores the need to minimize treatment volumes if trials of dose escalation are to be successful.
To illustrate the implication of decreased variability in patient positioning, we examined the consequences of changing the margin added to the prostate volume in order to obtain a target volume for a representative patient with a Stage T2a (B1) prostate cancer (Table 5). Using the Scandiplan treatment planning system, we found that if a 1.5 cm margin was chosen, a target volume of 214 cc resulted, but that if a 1.3 cm or 1.1 cm margin around the prostate can have on the final target volume, and in a corresponding fashion, on the volume of adjacent normal tissue treated. Although the absolute reductions in positioning error with immobilization may seem small in magnitude, they translate into potentially large differences in volumes of normal tissue irradiated. As noted above, small increases in the volume of anterior rectal wall treated may result in significant added morbidity.

Table 5

Many factors must be considered when arriving at a target volume from a tumor volume. These include difficulties in the definition of tumor volume, potential microscopic tumor extension, type of beam arrangement used in the treatment plan, adjustment for build up of isodose lines near field edges, possible in vivo organ mobility, and variability in patient positioning. Recognizing that reduction of the variability in patient positioning corrects for only one of these factors, we have instituted the use of immobilization for all prostate patients treated using 3D-CRT in our department. However, if 3D-CRT is to achieve its full potential, all factors contributing to the spatial uncertainty in dose delivery will also need to be optimized.
The use of the ALPHA CRADLE® immobilization devices has been incorporated into our routine treatment for patients with prostate cancer. The device is made at the time of initial, pre-CT simulation. Manufacture of the 
ALPHA CRADLE® device takes approximately 15 minutes. The patient then undergoes a CT scan for treatment planning (in some cases for diagnostic purposes as well) in the cast, and a verification simulation once the treatment plan has been generated. The use of immobilization devices has been well accepted by both patients and staff and has not presented any significant problems in our practice.
Perceived difficulty in evaluating oblique portal films has been mentioned as a reason not to use 3D-CRT techniques. However, we have found that verification of the oblique fields used in the six field technique was not difficult once initial experience was obtained in reviewing simulation and portal films. The finding that there was no significant difference between the oblique and lateral films with respect to positioning error confirms our clinical impression that interpretation of the oblique films used in this six-field conformal 3D-CRT technique is not more difficult than the interpretation of portals obtained on conventional fields, and should not represent a limitation to the applicability of the six-field technique.
The use of immobilization results in reduced simulation-to-treatment variability and a significant reduction in the number of positioning errors greater than 0.5 cm. The decrease in variability in patient positioning noted with immobilization allows the clinician greater assurance that the improved isodose distributions obtained with 3D-CRT will treat the target volume on a daily basis. In addition the reduction in the number of higher magnitude positioning errors may allow for smaller target volumes by decreasing the margin which must be added to the tumor volume in order to account for variability in patient positioning. The use of immobilization devices can be an important, and easy to institute, component of a 3D-CRT program for patients with prostate cancer. Research will need to continue in this area, as well as into the other factors contributing to the uncertainty in radiation dose delivery to target structures, if 3D-CRT is to achieve its full potential for the treatment of prostate cancer.