(Radiology. 1999;213:839-844.)
© RSNA, 1999
Permanent Implantation of 125I Sources in the Prostate: Radical Limits of Simplicity1
Patrick W. McLaughlin, MD,
Vrinda Narayana, PhD,
Marc T. Fields, MD,
Michael E. Dworzanin, CMD,
Raymond J. Winfield, MD and
Peter L. Roberson, PhD
1 From the Department of Radiation Oncology, University of Michigan Medical Center, Ann Arbor (P.W.M., M.T.F., P.L.R.), and the Providence Cancer Center, 22301 Foster Winter Dr, Southfield, MI 48075 (P.W.M., V.N., M.E.D., R.J.W., P.L.R.). From the 1997 RSNA scientific assembly. Received December 24, 1997; revision requested February 18, 1998; revision received February 1, 1999; accepted June 7. Address reprint requests to P.W.M.
 |
Abstract
|
|---|
PURPOSE: To determine the effect of reducing the number of sources per implantation on the dose coverage of the prostate volume.
MATERIALS AND METHODS: Idealized source distributions were planned for four, eight, 16, 24, 32, and 48 sources. The peripheral loading technique was used to plan a uniform, conformal dose distribution to the target volume, which was the prostate volume as visualized at ultrasonography. Source-placement error was estimated by using measured error magnitudes and was expressed with systematic and random components. The relative sensitivities of the plans to the source-placement error were studied.
RESULTS: Idealized planned target coverage can be adequately achieved with comparable dose distributions with eight or more sources. The sensitivity to source-placement error is comparable for plans with 16 or more sources.
CONCLUSION: It is theoretically possible to radically simplify implantation without compromising target coverage or error tolerance.
Index terms: Dosimetry, 844.1299 • Prostate, neoplasms, 844.32 • Prostate, therapeutic radiology, 844.1299 • Treatment planning, 844.1299
|
Introduction
|
|---|
Ultrasonographic (US) guidance in the placement of permanent implants in the prostate has improved the dose distributions, and impressive clinical results with 10 years of follow-up have made this an attractive alternative to other modalities (1–3). The treatment-planning nomograms employed at the outset of the modern era called for a certain number of millicuries of activity, usually distributed between 40 to over 100 sources (4,5). Although the clinical results and prostate-specific antigen, or PSA, levels with this general planning nomogram are encouraging, alternative methods have emerged. Several institutions use plans that emphasize peripheral loading as opposed to homogeneous loading (6–8). While both homogeneous and peripheral loading have led to promising clinical results, the actual implantation dosimetry achieved suggests that further improvements are possible (9,10).
One desirable approach is the simplification of implantation through the use of fewer but higher-activity sources. This strategy requires consideration of accuracy of placement and tolerance of error. We hypothesize that simpler implants are possible and remain tolerant to the common placement differences of sources from ideal preimplantation planned positions to postimplantation actual positions. While this study is a theoretic exercise, the practical implications are profound. Since the cost of high-activity sources is approximately equal to the cost of low-activity sources, a reduction in the number of sources could reduce the cost of implantation. Reduction in the number of sources could also reduce the time required to perform implantation. Most important, reduction in the number of sources could reduce the time required for postimplantation dosimetry and also could make online dosimetry during the procedure a realistic possibility.
We performed this study to determine the effect of reducing the number of sources per implantation on dose coverage of the prostate volume.
|
MATERIALS AND METHODS
|
|---|
Distributions were planned for a representative (27-cm3) prostate volume by using four, eight, 16, 24, 32, and 48 iodine 125 sources. The computer simulation was performed by using University of Michigan treatment-planning software, or UMPLAN (11). The peripheral loading technique was used to plan a uniform, conformal dose distribution to the target volume, which was the prostate volume as visualized at US. Source placement was determined by using geometric symmetry, particularly for the plans with a low number of sources. The high-source-number plans could be designed to be more conformal to the target. All sources were assigned the same activity for each plan. Activity-per-source levels were adjusted to provide 160-Gy coverage to greater than 99% of the target volume. Source and plan activity levels are given in the Table. No target coverage margin was included to help determine the sensitivity of each plan to source-placement error.
Source-placement error was estimated by using error magnitudes previously reported (10)(Roberson PL, Narayana V, McLaughlin PW, Winfield RJ, unpublished data, 1999). Errors were separated into five categories: source-string length, source spacing, needle depth, needle position in the plane perpendicular to the needle axis, and source-string splaying. Each category was expressed as a mean and SD and represented both systematic and random errors. Measured values representative of sources held in a string with absorbable suture material (Rapid Strand; Medi-Physics, Arlington Heights, Ill) were used.
Source and needle errors were calculated by using a random-number generator to represent the random error component (SD), as previously reported (10). Needle- and source-placement errors were combined to yield a source distribution representative of clinical experience. Twenty error calculations were performed for each plan to characterize the effect of the range of source-placement error. Source activities were increased by a nominal 15% to help account for source-placement error. The dosimetric results with source-placement error introduced were less than clinically acceptable because no target margin was included and no allowance for additional sources to compensate for poor source positioning was included (our practice is to evaluate source placement during the implantation procedure by using fluoroscopy and to correct obvious deviations immediately).
The relative dependence of the plans on source-placement errors was studied by using dose-volume histogram comparison.
|
RESULTS
|
|---|
Idealized plans were developed by using symmetry of source placement, with a concentration of sources on the target periphery. Cumulative dose-volume histograms for the idealized plans (Fig 1) have similar gradients and high-dose volumes, except for the four-source plan. Small differences are observed due to the particular asymmetries of the target volume and the source-placement symmetry set chosen. Idealized planned target coverage can be adequately achieved with comparable dose distributions with eight or more sources. As the number of sources was increased, the activity per source decreased, and the total activity tended to decrease (Table). Comparable total activities were achieved for 16 or more sources.
View larger version (115K):
[in this window]
[in a new window]
|
Figure 1. Dose-volume plots for idealized plans with four, eight, 16, 24, 32, and 48 sources. Note the comparable dose-volume plots for all but the four-source plan.
|
|
The volumes (internal and external to the target) included in each isosurface (160-Gy dose envelope) for the idealized plans tended to decrease with an increasing number of sources (Fig 2). Again, results are similar for eight and for 16 or more sources. The 160-Gy isosurface volumes, in particular, were very close for 16–48 sources. The slight increase for the 24-source case was due to the source-placement strategy, which allowed an additional dose outside of the target volume and thus increased the dose margin on the target slightly.
View larger version (13K):
[in this window]
[in a new window]
|
Figure 2. Graph shows the total volume (in cubic centimeters) receiving 160, 240, or 320 Gy versus the number of sources. Note the higher volumes for the four- and eight-source plans but the comparable volumes for the 16-48-source plans.
|
|
Error sensitivity was determined by recording the decrease in target coverage with error (Fig 3). Shown in Figure 3 are the variations in target coverage that resulted for each of the 20 error plans compared with each idealized plan. Also shown are the results for the error plan with the median value of coverage. The four- and eight-source plans are the least sensitive to error but also included the largest volume within the prescribed isosurface (Fig 2). Results for 16–48 sources were comparable.
View larger version (10K):
[in this window]
[in a new window]
|
Figure 3. Graph shows the percentage of the prostate volume receiving the full dose versus the number of sources (indicated by the number above each set of gray dots) for 20 error-simulated plans. The median value shown as the black circles indicates similar coverage for the 16-48-source plans. Gray dots represent results of individual error simulations.
|
|
Although these data suggest that coverage to the target is not compromised by simplified plans, another critical concern in clinical practice is the effect on normal tissues. To evaluate this, reference points were placed beyond the prostate volume. The rectal reference point was placed 1 cm from the medial posterior prostate volume. The bladder reference points were placed 1 cm from the base of the prostate at a 45° angle. The effect of error on the rectal and bladder reference points was calculated for all 20 runs for each plan.
Figure 4 is a plot of the dose to the rectal point as a function of the number of sources used in the plan. The rectal point is profoundly influenced by the source configuration, especially when the source of high activity is close to the reference point. A range of reference point doses resulted from each plan, with the 48-source plan emerging with the least spread in dose. One of the 20 plans for both the 16- and the 24-source configurations resulted in a rectal reference point dose increase greater than 1.5 times the ideal dose.
View larger version (10K):
[in this window]
[in a new window]
|
Figure 4. Graph shows the dose to the rectal reference point (1 cm from the posterior prostate volume) for 20 error-simulated plans.  = dose to the rectal reference point for the ideal plan. Gray dots represent results of individual error simulations.
|
|
The bladder reference dose deviation (Fig 5) was more profound with all plans. But, if one of 20 plans is excluded from the 16–32–source group, the error tolerance is comparable. With consideration of the limitations of the reference point data relative to the volumetric data, the results do not suggest a profound detrimental effect of plan simplification on normal tissues, with the exception of extremely errant plans.
View larger version (9K):
[in this window]
[in a new window]
|
Figure 5. Graph shows the dose to the bladder reference point (1 cm from the base of the prostate at a 45° angle) for 20 error-simulated plans. = dose to the bladder reference point for the ideal plan. One simulation resulted in a 503-Gy point (off the scale). Gray dots represent results of individual error simulations.
|
|
Isodose color wash displays for axial ideal and error-simulation plans are displayed in Figure 6. The plans with median target-volume coverage were chosen for display. The initial volume enclosed by the four-source plan was large compared to the volume in the 16- and 48-source plans. This greater volume accounts, in part, for the greater tolerance to error. The 16- and 48-source plans showed comparable target coverage. A larger number of sources did not have a target coverage advantage with error because each source influenced a smaller volume. Patient treatment plans had target volumes that included the prostate plus a margin to reduce their sensitivity to source-placement error.
View larger version (80K):
[in this window]
[in a new window]
|
Figure 6. Transverse (upper panels), sagittal (middle panels), and coronal (lower panels) US images show midprostate dose distributions for idealized and error-simulated implantations with four, 16, and 48 sources. Blue = 80-160 Gy, purple = 160-240 Gy, pink = 240-360 Gy, and red =  360 Gy. Contours for the prostate (red), the prostate plus a 5-mm margin (yellow), and the urethra (white) are shown. Error-simulated plans have a median percentage of prostate coverage.
|
|
|
DISCUSSION
|
|---|
Historically, in the precomputer era, three major systems (Manchester, Paris, and Quimby) were developed to allow planning of brachytherapy. Each helped solve the inevitable problems of placement in interstitial brachytherapy in a different way. Those problems include central implant overdose; uncrossed needle ends; length effects or, in Paris terms, "escargot" (longer sources of equal activity have greater dose delivery lateral to the sources); and volume implants. Modern brachytherapy techniques borrow, at times unconsciously, from solutions offered by these systems but depend heavily on computer planning, which allows rapid dose calculation. In the beginning of the modern prostate-implant era, a nomographic approach with equal loading throughout the implant was employed (4). Over time, there has been a movement to more peripheral loading. In modern terms, there is equal spacing; unequal (central overdose) dosing; and unequal spacing (peripheral placement), with less central overdose and with urethral sparing (12). Excellent clinical results have been reported with both methods (1–3).
In this study, we sought to determine if implantation could be radically simplified by reducing the number of sources. If this were possible, a fourfold advantage would be apparent. There would be ease of planning, shortening of operating room time, ease of postimplantation dosimetry, and decreased cost.
To explore this question, a fixed prostate volume was planned with a range of four to 48 sources. The activity was adjusted to allow 99% coverage of the gland with the nominal dose of 160 Gy. Although inhomogeneity was expected to be greatest for the lowest number of sources and to improve with increasing the number of sources, one remarkable finding was that the dose-volume plot for eight sources virtually overlay the dose-volume plot for 48 sources. One possible explanation is that, with a lower number of sources, there are overdose volumes immediately around the sources, but the sources are far enough apart that the overdose volumes do not combine as readily and remain relatively discrete. The net total volume of such overdose areas remains relatively small compared with the volume of the implant. With 48 sources, although the activity of each source is lower, the homogeneous distribution results in an overdose region near the target periphery, with a total volume comparable to the volume seen in the eight- or 16-source implant. The difference is that the volume of overdose in the 48-source implant is more contiguous than that in the plans with the lower number of sources.
Another major concern regarding the reduction in the number of sources and the simplification of implantation is the effect of error. It is obvious that it has been well recognized that prostate motion, source migration, or even embolization may occur during implantation, which makes source-placement error an important factor in postimplantation dosimetric evaluation. The argument for security in numbers intuitively seems sound, and the initial impression of drastically reducing the number of sources is that it is a hazardous solution. To explore the effect of error in such implants, a program (10) was employed in which error was simulated. This program has been previously employed and validated (Roberson PL, Narayana V, McLaughlin PW, Winfield RJ, unpublished data, 1999). The program combines random migration and systematic error and allows theoretic prediction of the dosimetry of actual implants. Again, the results were counterintuitive. The 48-source plan had similar error tolerance as the 16-source plan.
The relative tolerance to source-placement error is related to the decrease in dose from a single source as a function of the inverse distance squared. When the sources are spaced farther apart, the dose overlap from different sources is in the low-gradient, midsource region. With wider spacing in this region, there is less change in dose than there is with closer spacing of sources, with which a rapid decrease or increase in dose may occur with a change in source position. If the source-placement error were entirely random, placement error would favor the plans with the larger number of sources, with the final dose distribution self-averaging (random errors would cancel each other out and tend to an average deviation less than seen here). However, the systematic error is substantial, and findings of previous studies have demonstrated a greater likelihood of error at the base of the prostate, with splaying of needles and greater spacing of the sources (Roberson PL, Narayana V, McLaughlin PW, Winfield RJ, unpublished data, 1999) (12). A standard loading in an ideal location allows excellent coverage of the base, but minimal error results in underdosing. This explains, in part, the relative lack of increased error tolerance of the 48-source plan compared with the 16-source plan.
Although reduction in the number of sources with error allowed excellent coverage of the prostate, the use of high-activity sources posed a potential risk to normal tissues. Even if erroneous placement results in excellent target coverage, a misplaced 1.5- or 2-mCi (55.5- or 74-MBq) source has greater potential for serious complications due to damage of normal tissues. The use of high-activity sources requires a highly experienced team and a different planning method that calls for the distribution of sources as far from normal tissues as possible. Recently, our treatment planning has emphasized peripheral and lateral prostate zones away from the bladder and rectum. Even in the worse-case scenario (greater source shift), the greater risk is neutralized. Predictable placement and stability of sources outside the prostate were improved by using absorbable suture material, which reduces source clustering and migration and prevents embolization (10).
Another great clinical concern after the dose to the target is the dose to normal tissues. The reference point data presented suggest that simple implants with high-activity sources and extreme error may result in a dose to the reference points. However, this does not prove that there is a high dose to a volume. Even with the bias that reference point dose comparisons result in overestimation of the error effect in simple implants owing to the proximity of the points to the single high-activity sources, if extremely errant plans are excluded, the resultant deviations in dose with source-placement error are comparable.
Several advantages to implantation simplification make this an attractive alternative. First, planning in optimization programs could be drastically simplified. Second, the time investment for implantation could also be shortened. Third, the most important phase of implantation evaluation, postimplantation dosimetry, could be made more efficient. All of these would decrease the cost of the implantation by decreasing planning time, operation time, and postimplantation dosimetry time. Finally, since the cost of a source is not directly related to the strength, a lower number of sources with higher activity could potentially cut the cost of the sources per implantation in half, with no reduction in quality. Conservative, stepwise reduction in the planned sources per implantation will allow validation of these theoretic advantages in the clinic.
It is possible to radically simplify implantation without compromising target coverage or error tolerance. Advantages of decreasing the source number include decreased planning time, decreased time in the operating room, decreased source cost per implantation, and rapid postimplantation dosimetric evaluations. Great caution and a change in planning assumptions are necessary to ensure safe application and validation of these theoretic findings in the clinic.
|
Footnotes
|
|---|
Author contributions: Guarantor of integrity of entire study, P.W.M.; study concepts and design, P.W.M., V.N., P.L.R.; definition of intellectual content, M.T.F., P.W.M., V.N., R.J.W., P.L.R.; literature research, P.W.M., V.N., P.L.R.; data acquisition and analysis, V.N., M.E.D., P.L.R.; manuscript preparation and editing, M.T.F., P.W.M., P.L.R.; manuscript review, M.T.F., P.W.M., V.N., P.L.R.
|
References
|
|---|
-
Blasko JC, Grimm PD, Ragde H. Brachytherapy and organ preservation in the management of carcinoma of the prostate. Semin Radiat Oncol 1993; 3:240-249.[Medline]
-
Blasko JC, Ragde H, Grimm PD, Cavanagh W, Kenny G. Transperineal ultrasound guided palladium-103 brachytherapy for prostate carcinoma (abstr). J Urol 1995; 153(suppl):385.[Medline]
-
Blasko JC, Wallner K, Grimm PD, Ragde H. PSA-based disease control following ultrasound guided I-125 implantation for stage T1/T2 prostate carcinoma. J Urol 1995; 154:1096-1099.[Medline]
-
Anderson LL. Brachytherapy planning and evaluation. Endocurietherapy Hyperthermia Oncol 1991; 7:139-146.
-
Hilaris BS, Nori D, Anderson LL. Atlas of brachytherapy New York, NY: Macmillian, 1988; 70-95.
-
Stock RG, Stone NN, Wesson MF, DeWyngaert JK. A modified technique allowing interactive ultrasound guided three dimensional transperineal prostate implantation. Int J Radiat Oncol Biol Phys 1995; 32:219-225.[Medline]
-
Wallner K, Chiu-Tsao S, Roy J, et al. An improved method for computerized tomography-planned transperineal iodine-125 implants. J Urol 1991; 146:90-95.[Medline]
-
Narayana V, Roberson PL, Pu AT, Sandler H, Winfield RJ, McLaughlin PW. Impact of differences in ultrasound and CT volumes on treatment planning of permanent prostate implants. Int J Radiat Oncol Biol Phys 1997; 37:1181-1185.[Medline]
-
Willins J, Wallner K. CT-based dosimetry for transperineal I-125 prostate brachytherapy. Int J Radiat Oncol Biol Phys 1997; 39:347-353.[Medline]
-
Roberson PL, Narayana V, McShan DL, Winfield RJ, McLaughlin PW. Source placement error for permanent implant of the prostate. Med Phys 1997; 24:251-257.[Medline]
-
Fraass BA, McStran DL. 3-D treatment planning I. Overview of a clinical planning system. In: Bruinvis IAD, ed. Use of computers in radiation therapy. Amsterdam, the Netherlands: Elsevier Science, 1987; 273-276.
-
Narayana V, Roberson PL, Winfield RJ, Kessler ML, McLaughlin PW. Optimal placement of radioisotopes for permanent prostate implants. Radiology 1996; 199:457-460.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
M. L. Thakur, R. Coss, R. Howell, D. Vassileva-Belnikolovska, J. Liu, S. P. Rao, G. Spana, P. Wachsberger, and D. L. Leeper
Role of Lipid-Soluble Complexes in Targeted Tumor Therapy
J. Nucl. Med.,
August 1, 2003;
44(8):
1293 - 1300.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
Abstract
Introduction
