- Email: [email protected]
Dosimetric comparison of volume-based and inverse planning simulated annealing–based dose optimizations for high–dose rate brachytherapy
Medical Dosimetry ] (2015) ]]]–]]]
Medical Dosimetry journal homepage: www.meddos.org
Dosimetric comparison of volume-based and inverse planning simulated annealing–based dose optimizations for high–dose rate brachytherapy Satish Pelagade, Ph.D.,* Harshavardhan Reddy Maddirala, M.Sc.,* Rahul Misra, M.D.,† U. Suryanarayan, M.D.,† and J.P. Neema, M.D.† *
Department of Medical Physics and †Department of Radiotherapy, Gujarat Cancer & Research Institute, Ahmedabad, India
A R T I C L E I N F O
A B S T R A C T
Article history: Received 13 May 2014 Received in revised form 12 January 2015 Accepted 12 January 2015
The aim of this study was to compare the clinical benefits of inverse planning simulated annealing (IPSA)–based optimization over volume-based optimization for high–dose rate (HDR) cervix interstitial implants. Overall, 10 patients of cervical carcinoma were considered for treatment with HDR interstitial brachytherapy. Oncentra Master Plan brachytherapy planning system was used for generating 3-dimensional HDR treatment planning for all patients. All patient treatments were planned using volume-based optimization and inverse planning optimization (IPSA). The parameters V100, V150, and V200 for the target; D2 cm3 of bladder, rectum, and sigmoid colon; and V80 and V100 for bladder, rectum, and sigmoid colon were compared using dose-volume histograms (DVHs). The conformity index (CI), relative dose homogeneity index, overdose volume index (ODI), and dose nonuniformity index (DNR) were computed from cumulative DVHs. Good target coverage for prescription dose was achieved with volume-based optimization as compared with IPSA-based dose optimization. Homogeneity was good with the IPSA-based technique as compared with the volume-based dose optimization technique. Volume-based optimization resulted in a higher CI (with a mean value of 0.87) compared with the IPSAbased optimization (with a mean value of 0.76). ODI and DNR are better for the IPSA-based plan as compared with the volume-based plan. Mean doses to the bladder, rectum, and sigmoid colon were least with IPSA. IPSA also spared the critical organs but with considerable target conformity as compared with the volume-based plan. IPSA significantly reduces overall treatment planning time with improved reduced doses to the organs at risk compared with the volume-based optimization treatment planning method. & 2015 American Association of Medical Dosimetrists.
Keywords: Brachytherapy Optimization Inverse planning
Introduction Brachytherapy has been a standard technique of radiation therapy for cervical cancer since the discovery of radium. Dose optimization in brachytherapy is not a new concept and has been studied for several decades because of technological advances both in computing power and in imaging technology (computed tomography [CT] and magnetic resonance imaging).1,2 High–dose rate (HDR) brachytherapy treatment planning often involves optimization methods to calculate the dwell times at dwell
Reprint requests to: Satish Pelagade, Ph.D., Department of Medical Physics, Gujarat Cancer & Research Institute, New Civil Hospital Campus, Asarwa, Ahmedabad 380 016, India. E-mail: [email protected] http://dx.doi.org/10.1016/j.meddos.2015.01.003 0958-3947/Copyright Ó 2015 American Association of Medical Dosimetrists
positions of the radioactive source along specified applicator paths. The goal of HDR planning is to produce an acceptable optimized plan within a reasonable time period, which meets the desired dose constraints. Optimization technique for interstitial implant aims at obtaining adequate target coverage with maximum sparing of critical structures. Morton et al.3 observed that anatomy-based inverse planning simulated annealing (IPSA) and graphical optimization for HDR prostate brachytherapy generated similar dose coverage. IPSAbased plans deliver lower dose to critical structures and greater dose homogeneity than graphical optimization plans do. Jamema et al.4 observed that anatomy-based inverse optimization followed by graphical optimization is much superior in conformity and sparing of critical structures than geometric optimization. Lessard et al.5 reported the clinical benefits of IPSA for the treatment
Pelagade et al / Medical Dosimetry ] (2015) ]]]–]]]
2
planning of prostate HDR brachytherapy. Skowronek et al.6 examined the influence of the dose optimization procedures on the value of radiation doses in organs at risk (OARs) and compared the value of doses measured in healthy tissues according to different chosen pulsed–dose rate brachytherapy (PDRBT) and HDR brachytherapy fractionation schedules. A new 3-dimensionl (3D) treatment planning system Oncentra Master Plan, v 4.3 (Nucletron) was recently installed in our radiotherapy department. HDR interstitial brachytherapy treatment planning was performed using volume-based dose optimization in routine practice. IPSA-based optimization was observed to be more superior in target coverage, dose homogeneity within the target, and minimizing volume of noncontoured normal tissues.7 The overdose volume was observed to be more in volume-based dose optimization. To see whether the overdose volume was being reduced with IPSA-based optimization, we planned to compare these 2 optimization techniques with various parameters. We did not come across any published article comparing the benefits of IPSA over volume-based dose optimization for HDR cervix interstitial implants. The dose distributions obtained using volume optimization and IPSA technique for HDR cervix interstitial implants were compared using various parameters.
Fig. 1. Planning target volume (PTV) and critical organ (CO). Sources are located on a regular grid in 3 source planes. The reference points are either only at the surface of PTV or on a rigid grid. where gij denotes the dose rate at reference point i, if the source is located at dwell position number j, and tj represents the corresponding dwell time. For the PTV, a minimum target dose Dmin is prescribed, which results in the following inequalities: di Z Dmin for all reference points inside the PTV
ð2Þ
If Dk is the maximum tolerable dose inside of critical organ number k, then we get additional inequalities of the form: di r Dk for all reference point inside the critical organ k
Methods and Materials In this study, 10 patients with cervical carcinoma (Stage IIIB) considered for treatment with HDR interstitial brachytherapy using volume-based optimization were retrospectively analyzed for IPSA optimization. The interstitial implant was performed using Martinez Universal Perineal Interstitial Template. All patients received external beam radiotherapy of 50 Gy in 25 fractions using a 6-MV linear accelerator before they started brachytherapy. As a brachytherapy procedure, patients were examined under anesthesia to assess the residual disease at vault, parametrium, normal pelvic anatomy, vagina, and its relationship to the normal structures. The rectum was prepared for the patient by giving Peglec powder. After assessing the vaginal length, rectal suction catheter and Foley catheter were inserted followed by vaginal obturator as per our institutional protocol. To prevent rectal perforation, 18-gauge stainless steel needles with closed trocar tip were inserted with a finger in rectum. The depth of penetration was decided according to the cranial extent of the disease and the involvement of parametrium. The needles were guided by the holes of the template. If the needles were obstructed by bone, they were placed through oblique holes to circumvent the bones. Cystoscopy and proctoscopy were performed to detect rectal or bladder perforation by the needles. The template was secured to the perineum by 4 corner stitches. After tightening all the needles, the outer plate was then placed over the needles to secure their positions. Patients were then taken for imaging and planning. The CT image acquisition was done on the same day for treatment planning after the implant. Somatom Emotion CT scanner (Siemens Medical Systems, Germany) was used to take an axial CT scan of 5-mm slice thickness. The images were then transferred to Brachytherapy Treatment Planning System (Oncentra Master Plan, v. 4.3, Nucletron). The delineation of the planning target volume (PTV) and the OARs (rectum, bladder, and sigmoid colon) was performed by the radiation oncologist. All implant needles were reconstructed and active dwell positions were selected according to the target extent. Oncentra Master Plan brachytherapy planning system, version 4.3 (Nucletron) was used for generating 3D HDR treatment plan using volume-based optimization for all 10 patients. The desired goal was D95 of8 100% of prescription dose for PTV and D2 cm3 of the rectum, bladder, and sigmoid getting r80% of prescribed dose. Kneschaurek et al.9 described a volume-based optimization technique for brachytherapy dose distribution. For volume-based optimization method, the dose points were generated at 5 mm on the PTV surface and the dose was prescribed to these dose points. The dose at these reference dose points can be calculated if the positions of the sources are known (alternate dwell positions 1, 3, 5, …). The reference isodose is that isodose which conforms to the target adequately. This method is quite useful and allows 3D dose optimization for the target volume, where the prescribed dose is applied to the whole target. The radiation oncologist prescribes the minimum dose to the PTV and observes upper dose limit to the critical organs. Each volume of interest is represented by a number of discrete reference points inside or at the surface of this volume. All these points may be distributed randomly or on a rigid grid, as shown in Fig. 1. The dose di at reference point i is given by9 di ¼
X
gij nt j
ð1Þ
ð3Þ
These inequalities are linear with respect to the dwell times tj. Hence, a negative dwell time tj is not possible. Therefore, we get another set of inequalities as follows: tj Z 0 for all dwell times
ð4Þ
The set of dwell times tj are calculated, such that all the inequalities (2, 3, and 4) are fulfilled simultaneously. There exist many different solutions for the dwell times that satisfy all these inequalities. This is obviously the case if only the minimum dose in the PTV is prescribed and no critical organ constraints are defined. The prescribed dose is 16 Gy in 4 fractions (4 Gy/fraction), twice daily with a gap of 6 hours with full bladder to reduce the small intestine dose. The plan was then assessed for PTV coverage, homogeneity, and sparing of OARs. Treatment planning was done for all 10 patients using inverse planning optimization. IPSA is an inverse planning designed for 3D brachytherapy treatment planning. IPSA can consider multiple targets (PTV, clinical target volume, and boost) and multiple OARs (rectum, bladder, and sigmoid colon). The adjustment of weighting factors sets the relative importance of dose objectives for each organ and betweendose conformity and dose homogeneity. IPSA automatically selects continuous active dwell positions and optimizes the dwell times to fulfill the dose objectives. A simulated annealing optimization engine then identifies the best solution in less than a minute. Prescription for inverse planning is based on clinically identified volumes. It maximizes the target dose coverage while taking into account dose homogeneity and the protection of OARs. The dose objectives were setup to minimize the high dose to the rectum, bladder, and sigmoid colon and to deliver a prescribed dose to the PTV. The prescription is global. It covers all volumes and all objectives so that they can be optimized simultaneously. However, if a particular set of objectives generates the desired dose distribution, then the same set of objectives can be used for optimization for clinically similar cases without further adjustments. The specific dose constraints used in this study for all cases are listed in Table 1. To evaluate the consistency of the treatment plans generated for 2 different optimization techniques, the dose-volume histograms (DVHs) of target, bladder, rectum, and sigmoid colon and the dosimetric indices from 10 consecutive patients Table 1 Specific dose constraints used in this study ROI
Surface
Volume
Weight Min (Gy) PTV (reference 100 target) Rectum (organ) 0 Bladder (organ) 0 Sigmoid colon 0 (organ) ROI ¼ region of interest.
Max (Gy)
Weight Weight Min (Gy)
Max (Gy)
4.2
6
5
100
4.2
8
0 0 0
2.4 1.6 1.6
2 2 2
0 0 0
0 0 0
2.4 1.6 1.6
Weight
1 20 20 20
Pelagade et al / Medical Dosimetry ] (2015) ]]]–]]]
3
were analyzed. The following dosimetric data were collected: V100, V150, and V200 for the target; maximum dose to 2 cm3 of the bladder, rectum, and sigmoid colon; and V80 and V100 for the bladder, rectum, and sigmoid colon. The conformity index (CI), relative dose homogeneity index (HI), overdose volume index (ODI), and dose nonuniformity index (DNR) were computed from cumulative DVHs. The percentage of the target volume that receives at least the prescribed dose (V100) was used as coverage index, the percentage of the target volume that receives 1.5 times the prescribed dose (V150) was used as the high-dose volume, and the percentage of the target volume that receives twice the prescribed dose (V200) is used as overdose index. The dose that covers 2-cm3 volume of the bladder, rectum, and sigmoid colon was the parameter related to the critical organs for comparison of both the optimization techniques. Besides these parameters, there are other indices such as V80 and V100 that are based on the DVHs for the critical organs. CI2 is the ratio of the target volume covered by the prescription dose to the volume covered by the target. It is a measure of conformity of the plan. CI ¼
V100 Vtot
where V100 is the volume of PTV receiving 100% of the prescription dose, Vtot is the total target volume, HI10 measures the fraction of target volume receiving dose in the interval of 1.0 to 1.5 times the reference dose. It is a measure of homogeneity of the plan. HI ¼
ðV 100 V150 Þ Vtot
where V100 and V150 are the volumes of PTV receiving 100% and 150% of the prescription dose, respectively, and Vtot is the total target volume. ODI11 is the ratio of the target volume that receives a dose equal to or more than twice the reference dose. ODI ¼
V200 V100
DNR11-14 is defined as the ratio of high-dose volume relative to the reference volume. DNR is used to evaluate the quality of the plan. DNR ¼
Fig. 2. A plot of absolute dwell times from an 18-gauge catheter interstitial implant based on volume and IPSA optimizations. Few dwell positions are completely turned off in the IPSA plan. (Color version of figure is available online.)
colon, respectively). The CT images with dose distribution using both optimization techniques for same implant case are shown in Fig. 3. It can be observed that the bladder and the rectum are spared to the maximum extent with IPSA. It also spares the critical organs but with considerable target conformity as compared with volume-based plans.
V150 V100
For an ideal implant, CI should be 1, HI should be 1, ODI should be 0, and DNR should be 0.
Results Optimized plans were generated for each implant using volume-based and IPSA-based optimizations. For IPSA, the same dose constraints were used for all inverse planning calculations without manual adjustment of dwell weight. Similarly, no manual adjustment of dwell weight was used in volume optimization. Volume-based optimization showed a greater range of relative dwell times; some with very long dwell times and in others dwell positions were completely turned off, as shown in Fig. 2. In IPSA, only few dwell times were completely turned off. The total dwell time of volume-based plans was slightly longer. Dosimetric parameters considered were tabulated in Table 2 for volume-based and IPSA-based dose optimizations. Good target coverage for prescription dose was achieved with volume-based optimization as compared with IPSA-based dose optimization. The target volume receiving 150% and 200% of the target dose was 53.06% and 22.63% of the prescription dose for volume-based and 35.46% and 12.09% for IPSA-based dose optimization, respectively. Slightly lower target coverage was achieved with IPSA plans with a significant decrease in the tumor volume receiving a high dose (mean V150 and mean V200). The mean HI values are 0.37 and 0.45 for volume-based and IPSA-based optimizations, respectively. Homogeneity was good with the IPSA-based technique as compared with the volume-based dose optimization technique. Volume-based optimization resulted in a higher CI (with a mean value of 0.87) compared with the IPSA-based optimization (with a mean value of 0.76). ODI and DNR were better for IPSA as compared with volume-based plans. Mean doses to the bladder, rectum, and sigmoid colon were least with IPSA (mean V80 of 2.62%, 5.38%, and 0%; mean V100 of 0.22%, 0.79%, and 0%; and D2 cm3 of 77.41%, 83.23%, and 22.60% for the bladder, rectum, and sigmoid
Discussion HDR brachytherapy plays an important role in the management of carcinoma of the cervix. The available optimization techniques such as geometric and dose point optimization fail to use the anatomic information. These optimization techniques are based only on the location of the active dwell times. Hence, these methods necessarily result in an approximation of the shape of the anatomy. To maintain the complete target coverage and reduction of dose to normal OARs, the dose distribution should be as conformal as possible. It is useful to be reminded of the huge patient load that we receive in our hospital. One should also remember that inserting
Table 2 Dosimetric parameters for the volume- and IPSA-based optimization techniques Parameter
Volume based Mean (SD)
IPSA based Mean (SD)
Target V100 Target V150 Target V200 CI HI ODI DNR Bladder (D2 cm3 ) Bladder V80 Bladder V100 Rectum (D2 cm3 ) Rectum V80 Rectum V100 Sigmoid colon (D2 cm3 ) Sigmoid colon V80 Sigmoid colon V100
97.56 (1.33)% 53.06 (14.9)% 22.63 (18.88)% 0.87 (0.37) 0.37 (0.19) 0.23 (0.19) 0.54 (0.15) 103.28 (26.81)% 9.14 (7.11)% 3.86 (5.21)% 109.62 (32.33)% 14.96 (10.97)% 6.32 (7.21)% 31.80 (8.28)% 0% 0%
86.30 (1.41)% 35.46 (8.24)% 12.09 (3.78)% 0.76 (0.33) 0.45 (0.21) 0.14 (0.04) 0.40 (0.09) 77.41 (11.35)% 2.62 (2.17)% 0.22 (0.44)% 83.23 (15.46)% 5.38 (4.49)% 0.79 (1.56)% 22.60 (4.47)% 0% 0%
D2 cm3 ¼ dose received by 2 cm3 of volume; V100 ¼ volume receiving 100% of the prescription dose; SD ¼ standard deviation.
4
Pelagade et al / Medical Dosimetry ] (2015) ]]]–]]]
Fig. 3. Comparative slice views on the same CT slice and 3D views for (A) volume-based and (B) IPSA-based dose optimizations. (Color version of figure is available online.)
the Martinez Universal Perineal Interstitial Template itself is a time-consuming process, and it is also necessary that the treatment is started as soon as possible keeping in mind the pain and inconvenience that the patient goes through. Hence, the value of time in this procedure is unparalleled. Therefore, we require a procedure that does optimal planning in the shortest possible time. IPSA offers us just that. Not only does it significantly reduce the treatment time, but it also offers much better dose
homogeneity, which is the need of the hour. The reduction in doses to the bladder, rectum, and sigmoid colon would significantly reduce the long-term morbidity associated with the procedure that often deters physicians from using this procedure. In fact, just as intensity-modulated radiation therapy is preferred over 3D conformal radiotherapy in external radiotherapy because of its ability to model the treatment plan according to the desired objectives and constraints, similarly, we feel that IPSA is a much
Pelagade et al / Medical Dosimetry ] (2015) ]]]–]]]
better technique than volume-based planning with respect to planning time and dose homogeneity. Dose distribution can be manually obtained by adjusting relative dwell times until an acceptable solution is found. The computer is used only to calculate the dose distribution once the plan has been decided by the planner. This approach, in combination with geometric optimization, requires more time and skill. The existing commercial planning systems may not be truly anatomy based without a genuinely anatomy-based optimization. Procurement of anatomy-based optimization brings the planning process nearest to the real clinical issues. In this study, we described and compared 2 methods of 3D dose optimization (volume-based and inverse planning methods), by retrospectively analyzing dose distributions to the target and OARs using various dosimetric indices. The volume-based optimization method proved to provide superior dosimetric target coverage than IPSA-based treatment plan. However, this improvement in target coverage was associated with a slightly higher dose to the OARs. IPSA also decreased the volume of the hot spot within the target volume and dose to normal structures outside the implant. This was illustrated through quantitative comparisons of DVH parameters and dosimetric indices. IPSA accomplished this by utilizing anatomic information provided by 3D planning. Data from this study support the need to move to 3D brachytherapy treatment planning and the use of anatomy-based dwell time optimization for carcinoma of cervix HDR brachytherapy. There are no clearly defined rules for treatment plan optimization in brachytherapy, and the exact limits of variations between dwell times are not yet known. In this regard, software tools performing automatic optimization of the dose distribution should be used with caution. This is not the case with the inverse planning method, where constraints for the dwell times can be easily set, resulting in more homogeneous dwell times and making this method appear safer. Thus, inverse planning is better and reliable.
Conclusions (1) The article demonstrates the new IPSA-based optimization method for interstitial implants of the cervix. IPSA-based optimization produces clinically acceptable dose distributions if the constraints are chosen appropriately. With the use of IPSA, the dose coverage of the target decreases slightly; however, the dose homogeneity increases and the dose to OARs decreases, as compared with the use of volume-based optimization. (2) The high-dose regions and overdose regions were smaller and the dose distribution was more homogeneous with IPSAbased dose optimization than with volume-based dose optimization. (3) We proved that for interstitial implant of the cervix, IPSAbased optimization technique produces clinically acceptable dose distributions in the coverage and the homogeneity. Additionally, the dose to OARs can be kept well under acceptable limits. The dose coverage of the target with IPSA
5
optimization was a little lower than the volume-based optimization. The high-dose regions were larger and dose distribution was less homogenous with volume-based optimization. There was significant dose difference to OARs between both the techniques, but volume-based optimization resulted in larger OAR doses. (4) We demonstrated that the use of IPSA-based optimization can improve the dose distributions. Volume-based dose optimization performed on points placed on the surface of PTV can result in highly conformal dose distributions, but only at the cost of deterioration of dose homogeneity. (5) With IPSA optimization, a spectrum of alternative solutions is available and the treatment planner can select the solution that best satisfies the clinical constraints. The computerbased inverse planning method provides better dose distributions. The planner requires much larger time with volume-based optimization method to obtain a satisfactory solution requiring adjustments of the catheter positions than the automatic inverse planning method. This reduces the overall treatment time significantly, allowing the treatment of more patients.
References 1. Isohashi, F.; Yoshioka, Y.; Koizumi, M.; et al. High-dose-rate interstitial brachytherapy for previously untreated cervical carcinoma. Brachytherapy 8(2):234–9; 2009. 2. Bahaduri, Y.; Constantinescu, C.; Ezzat, M.; et al. 3D anatomy-based planning optimization for HDR brachytherapy of cervix cancer. Saudi J. Obstet. Gynecol. 11 (2):1430; 2009. 3. Morton, G.C.; Sankreacha, R.; Halina, P.; et al. A comparison of anatomy-based inverse planning with simulated annealing and graphical optimization for high-dose-rate prostate brachytherapy. Brachytherapy 7(1):12–6; 2008. 4. Jamema, S.V.; Saju, S.; Shetty, U.M.; et al. Dosimetric comparison of inverse optimization with geometric optimization in combination with graphical optimization for HDR prostate implants. J. Med. Phys. 31(2):89–94; 2006. 5. Lessard, E.; Hsu, I-Chow; Pouliot, J. Clinical benefits of inverse planning for high dose rate prostate brachytherapy, IFMBE proceedings, 14/3; 2006. 6. Skowronek, J.; Zwierzchowski, G.; Piotrowski, T.; et al. Optimization and it’s influence on value of doses in HDR and PDR brachytherapy. Neoplasma 57 (4):369–76; 2010. 7. Hsu, I-Chow J.; Lessard, E.; Weinberg, V.; et al. Comparison of inverse planning simulated annealing and geometrical optimization for prostate high-dose-rate brachytherapy. Brachytherapy 3:147–52; 2004. 8. Kim, H.; Beriwal, S.; Houser, C. IPSA based HDR interstitial brachytherapy planning for vaginal cancer: dosimetric comparison with manually optimized planning. Brachytherapy 10(1):S57–8; 2011. 9. Kneschaurek, P.; Schiessl, W.; Wehrmann, R. Volume-based dose optimization in brachytherapy. Int. J. Radiat. Oncol. Biol. Phys. 45(3):811–5; 1999. 10. Wu, A.; Ulin, K.; Sternick, E.S. A dose homogeneity index for evaluating Ir-192 interstitial breast implants. Med. Phys. 15:104–7; 1988. 11. Anbumani, S.; Anchineyan, P.; Narayanasamy, A.N.; et al. Treatment planning methods in high dose rate interstitial brachytherapy of carcinoma cervix: a dosimetric and radiobiological analysis. ISRN Oncol.; 20142014 Article ID 125020. 12. Saw, C.B.; Suntharalingam, N. Quantitative assessment of interstitial implants. Int. J. Radiat. Oncol. Biol. Phys. 20(1):135–9; 1991. 13. Chaswal, V.; Yoo, S.; Thomadsen, B.R.; et al. Multi-species prostate implant treatment plans incorporating 192Ir and 125I using a Greedy Heuristic based 3D optimization algorithm. Med. Phys. 34(2):436–44; 2007. 14. Shwetha, B.; Ravikumar, M.; Katke, A.; et al. Dosimetric comparison of various optimization techniques for high dose rate brachytherapy of interstitial cervix implants. J. Appl. Clin. Med. Phys. 11(3):225–30; 2010.
Medical Dosimetry journal homepage: www.meddos.org
Dosimetric comparison of volume-based and inverse planning simulated annealing–based dose optimizations for high–dose rate brachytherapy Satish Pelagade, Ph.D.,* Harshavardhan Reddy Maddirala, M.Sc.,* Rahul Misra, M.D.,† U. Suryanarayan, M.D.,† and J.P. Neema, M.D.† *
Department of Medical Physics and †Department of Radiotherapy, Gujarat Cancer & Research Institute, Ahmedabad, India
A R T I C L E I N F O
A B S T R A C T
Article history: Received 13 May 2014 Received in revised form 12 January 2015 Accepted 12 January 2015
The aim of this study was to compare the clinical benefits of inverse planning simulated annealing (IPSA)–based optimization over volume-based optimization for high–dose rate (HDR) cervix interstitial implants. Overall, 10 patients of cervical carcinoma were considered for treatment with HDR interstitial brachytherapy. Oncentra Master Plan brachytherapy planning system was used for generating 3-dimensional HDR treatment planning for all patients. All patient treatments were planned using volume-based optimization and inverse planning optimization (IPSA). The parameters V100, V150, and V200 for the target; D2 cm3 of bladder, rectum, and sigmoid colon; and V80 and V100 for bladder, rectum, and sigmoid colon were compared using dose-volume histograms (DVHs). The conformity index (CI), relative dose homogeneity index, overdose volume index (ODI), and dose nonuniformity index (DNR) were computed from cumulative DVHs. Good target coverage for prescription dose was achieved with volume-based optimization as compared with IPSA-based dose optimization. Homogeneity was good with the IPSA-based technique as compared with the volume-based dose optimization technique. Volume-based optimization resulted in a higher CI (with a mean value of 0.87) compared with the IPSAbased optimization (with a mean value of 0.76). ODI and DNR are better for the IPSA-based plan as compared with the volume-based plan. Mean doses to the bladder, rectum, and sigmoid colon were least with IPSA. IPSA also spared the critical organs but with considerable target conformity as compared with the volume-based plan. IPSA significantly reduces overall treatment planning time with improved reduced doses to the organs at risk compared with the volume-based optimization treatment planning method. & 2015 American Association of Medical Dosimetrists.
Keywords: Brachytherapy Optimization Inverse planning
Introduction Brachytherapy has been a standard technique of radiation therapy for cervical cancer since the discovery of radium. Dose optimization in brachytherapy is not a new concept and has been studied for several decades because of technological advances both in computing power and in imaging technology (computed tomography [CT] and magnetic resonance imaging).1,2 High–dose rate (HDR) brachytherapy treatment planning often involves optimization methods to calculate the dwell times at dwell
Reprint requests to: Satish Pelagade, Ph.D., Department of Medical Physics, Gujarat Cancer & Research Institute, New Civil Hospital Campus, Asarwa, Ahmedabad 380 016, India. E-mail: [email protected] http://dx.doi.org/10.1016/j.meddos.2015.01.003 0958-3947/Copyright Ó 2015 American Association of Medical Dosimetrists
positions of the radioactive source along specified applicator paths. The goal of HDR planning is to produce an acceptable optimized plan within a reasonable time period, which meets the desired dose constraints. Optimization technique for interstitial implant aims at obtaining adequate target coverage with maximum sparing of critical structures. Morton et al.3 observed that anatomy-based inverse planning simulated annealing (IPSA) and graphical optimization for HDR prostate brachytherapy generated similar dose coverage. IPSAbased plans deliver lower dose to critical structures and greater dose homogeneity than graphical optimization plans do. Jamema et al.4 observed that anatomy-based inverse optimization followed by graphical optimization is much superior in conformity and sparing of critical structures than geometric optimization. Lessard et al.5 reported the clinical benefits of IPSA for the treatment
Pelagade et al / Medical Dosimetry ] (2015) ]]]–]]]
2
planning of prostate HDR brachytherapy. Skowronek et al.6 examined the influence of the dose optimization procedures on the value of radiation doses in organs at risk (OARs) and compared the value of doses measured in healthy tissues according to different chosen pulsed–dose rate brachytherapy (PDRBT) and HDR brachytherapy fractionation schedules. A new 3-dimensionl (3D) treatment planning system Oncentra Master Plan, v 4.3 (Nucletron) was recently installed in our radiotherapy department. HDR interstitial brachytherapy treatment planning was performed using volume-based dose optimization in routine practice. IPSA-based optimization was observed to be more superior in target coverage, dose homogeneity within the target, and minimizing volume of noncontoured normal tissues.7 The overdose volume was observed to be more in volume-based dose optimization. To see whether the overdose volume was being reduced with IPSA-based optimization, we planned to compare these 2 optimization techniques with various parameters. We did not come across any published article comparing the benefits of IPSA over volume-based dose optimization for HDR cervix interstitial implants. The dose distributions obtained using volume optimization and IPSA technique for HDR cervix interstitial implants were compared using various parameters.
Fig. 1. Planning target volume (PTV) and critical organ (CO). Sources are located on a regular grid in 3 source planes. The reference points are either only at the surface of PTV or on a rigid grid. where gij denotes the dose rate at reference point i, if the source is located at dwell position number j, and tj represents the corresponding dwell time. For the PTV, a minimum target dose Dmin is prescribed, which results in the following inequalities: di Z Dmin for all reference points inside the PTV
ð2Þ
If Dk is the maximum tolerable dose inside of critical organ number k, then we get additional inequalities of the form: di r Dk for all reference point inside the critical organ k
Methods and Materials In this study, 10 patients with cervical carcinoma (Stage IIIB) considered for treatment with HDR interstitial brachytherapy using volume-based optimization were retrospectively analyzed for IPSA optimization. The interstitial implant was performed using Martinez Universal Perineal Interstitial Template. All patients received external beam radiotherapy of 50 Gy in 25 fractions using a 6-MV linear accelerator before they started brachytherapy. As a brachytherapy procedure, patients were examined under anesthesia to assess the residual disease at vault, parametrium, normal pelvic anatomy, vagina, and its relationship to the normal structures. The rectum was prepared for the patient by giving Peglec powder. After assessing the vaginal length, rectal suction catheter and Foley catheter were inserted followed by vaginal obturator as per our institutional protocol. To prevent rectal perforation, 18-gauge stainless steel needles with closed trocar tip were inserted with a finger in rectum. The depth of penetration was decided according to the cranial extent of the disease and the involvement of parametrium. The needles were guided by the holes of the template. If the needles were obstructed by bone, they were placed through oblique holes to circumvent the bones. Cystoscopy and proctoscopy were performed to detect rectal or bladder perforation by the needles. The template was secured to the perineum by 4 corner stitches. After tightening all the needles, the outer plate was then placed over the needles to secure their positions. Patients were then taken for imaging and planning. The CT image acquisition was done on the same day for treatment planning after the implant. Somatom Emotion CT scanner (Siemens Medical Systems, Germany) was used to take an axial CT scan of 5-mm slice thickness. The images were then transferred to Brachytherapy Treatment Planning System (Oncentra Master Plan, v. 4.3, Nucletron). The delineation of the planning target volume (PTV) and the OARs (rectum, bladder, and sigmoid colon) was performed by the radiation oncologist. All implant needles were reconstructed and active dwell positions were selected according to the target extent. Oncentra Master Plan brachytherapy planning system, version 4.3 (Nucletron) was used for generating 3D HDR treatment plan using volume-based optimization for all 10 patients. The desired goal was D95 of8 100% of prescription dose for PTV and D2 cm3 of the rectum, bladder, and sigmoid getting r80% of prescribed dose. Kneschaurek et al.9 described a volume-based optimization technique for brachytherapy dose distribution. For volume-based optimization method, the dose points were generated at 5 mm on the PTV surface and the dose was prescribed to these dose points. The dose at these reference dose points can be calculated if the positions of the sources are known (alternate dwell positions 1, 3, 5, …). The reference isodose is that isodose which conforms to the target adequately. This method is quite useful and allows 3D dose optimization for the target volume, where the prescribed dose is applied to the whole target. The radiation oncologist prescribes the minimum dose to the PTV and observes upper dose limit to the critical organs. Each volume of interest is represented by a number of discrete reference points inside or at the surface of this volume. All these points may be distributed randomly or on a rigid grid, as shown in Fig. 1. The dose di at reference point i is given by9 di ¼
X
gij nt j
ð1Þ
ð3Þ
These inequalities are linear with respect to the dwell times tj. Hence, a negative dwell time tj is not possible. Therefore, we get another set of inequalities as follows: tj Z 0 for all dwell times
ð4Þ
The set of dwell times tj are calculated, such that all the inequalities (2, 3, and 4) are fulfilled simultaneously. There exist many different solutions for the dwell times that satisfy all these inequalities. This is obviously the case if only the minimum dose in the PTV is prescribed and no critical organ constraints are defined. The prescribed dose is 16 Gy in 4 fractions (4 Gy/fraction), twice daily with a gap of 6 hours with full bladder to reduce the small intestine dose. The plan was then assessed for PTV coverage, homogeneity, and sparing of OARs. Treatment planning was done for all 10 patients using inverse planning optimization. IPSA is an inverse planning designed for 3D brachytherapy treatment planning. IPSA can consider multiple targets (PTV, clinical target volume, and boost) and multiple OARs (rectum, bladder, and sigmoid colon). The adjustment of weighting factors sets the relative importance of dose objectives for each organ and betweendose conformity and dose homogeneity. IPSA automatically selects continuous active dwell positions and optimizes the dwell times to fulfill the dose objectives. A simulated annealing optimization engine then identifies the best solution in less than a minute. Prescription for inverse planning is based on clinically identified volumes. It maximizes the target dose coverage while taking into account dose homogeneity and the protection of OARs. The dose objectives were setup to minimize the high dose to the rectum, bladder, and sigmoid colon and to deliver a prescribed dose to the PTV. The prescription is global. It covers all volumes and all objectives so that they can be optimized simultaneously. However, if a particular set of objectives generates the desired dose distribution, then the same set of objectives can be used for optimization for clinically similar cases without further adjustments. The specific dose constraints used in this study for all cases are listed in Table 1. To evaluate the consistency of the treatment plans generated for 2 different optimization techniques, the dose-volume histograms (DVHs) of target, bladder, rectum, and sigmoid colon and the dosimetric indices from 10 consecutive patients Table 1 Specific dose constraints used in this study ROI
Surface
Volume
Weight Min (Gy) PTV (reference 100 target) Rectum (organ) 0 Bladder (organ) 0 Sigmoid colon 0 (organ) ROI ¼ region of interest.
Max (Gy)
Weight Weight Min (Gy)
Max (Gy)
4.2
6
5
100
4.2
8
0 0 0
2.4 1.6 1.6
2 2 2
0 0 0
0 0 0
2.4 1.6 1.6
Weight
1 20 20 20
Pelagade et al / Medical Dosimetry ] (2015) ]]]–]]]
3
were analyzed. The following dosimetric data were collected: V100, V150, and V200 for the target; maximum dose to 2 cm3 of the bladder, rectum, and sigmoid colon; and V80 and V100 for the bladder, rectum, and sigmoid colon. The conformity index (CI), relative dose homogeneity index (HI), overdose volume index (ODI), and dose nonuniformity index (DNR) were computed from cumulative DVHs. The percentage of the target volume that receives at least the prescribed dose (V100) was used as coverage index, the percentage of the target volume that receives 1.5 times the prescribed dose (V150) was used as the high-dose volume, and the percentage of the target volume that receives twice the prescribed dose (V200) is used as overdose index. The dose that covers 2-cm3 volume of the bladder, rectum, and sigmoid colon was the parameter related to the critical organs for comparison of both the optimization techniques. Besides these parameters, there are other indices such as V80 and V100 that are based on the DVHs for the critical organs. CI2 is the ratio of the target volume covered by the prescription dose to the volume covered by the target. It is a measure of conformity of the plan. CI ¼
V100 Vtot
where V100 is the volume of PTV receiving 100% of the prescription dose, Vtot is the total target volume, HI10 measures the fraction of target volume receiving dose in the interval of 1.0 to 1.5 times the reference dose. It is a measure of homogeneity of the plan. HI ¼
ðV 100 V150 Þ Vtot
where V100 and V150 are the volumes of PTV receiving 100% and 150% of the prescription dose, respectively, and Vtot is the total target volume. ODI11 is the ratio of the target volume that receives a dose equal to or more than twice the reference dose. ODI ¼
V200 V100
DNR11-14 is defined as the ratio of high-dose volume relative to the reference volume. DNR is used to evaluate the quality of the plan. DNR ¼
Fig. 2. A plot of absolute dwell times from an 18-gauge catheter interstitial implant based on volume and IPSA optimizations. Few dwell positions are completely turned off in the IPSA plan. (Color version of figure is available online.)
colon, respectively). The CT images with dose distribution using both optimization techniques for same implant case are shown in Fig. 3. It can be observed that the bladder and the rectum are spared to the maximum extent with IPSA. It also spares the critical organs but with considerable target conformity as compared with volume-based plans.
V150 V100
For an ideal implant, CI should be 1, HI should be 1, ODI should be 0, and DNR should be 0.
Results Optimized plans were generated for each implant using volume-based and IPSA-based optimizations. For IPSA, the same dose constraints were used for all inverse planning calculations without manual adjustment of dwell weight. Similarly, no manual adjustment of dwell weight was used in volume optimization. Volume-based optimization showed a greater range of relative dwell times; some with very long dwell times and in others dwell positions were completely turned off, as shown in Fig. 2. In IPSA, only few dwell times were completely turned off. The total dwell time of volume-based plans was slightly longer. Dosimetric parameters considered were tabulated in Table 2 for volume-based and IPSA-based dose optimizations. Good target coverage for prescription dose was achieved with volume-based optimization as compared with IPSA-based dose optimization. The target volume receiving 150% and 200% of the target dose was 53.06% and 22.63% of the prescription dose for volume-based and 35.46% and 12.09% for IPSA-based dose optimization, respectively. Slightly lower target coverage was achieved with IPSA plans with a significant decrease in the tumor volume receiving a high dose (mean V150 and mean V200). The mean HI values are 0.37 and 0.45 for volume-based and IPSA-based optimizations, respectively. Homogeneity was good with the IPSA-based technique as compared with the volume-based dose optimization technique. Volume-based optimization resulted in a higher CI (with a mean value of 0.87) compared with the IPSA-based optimization (with a mean value of 0.76). ODI and DNR were better for IPSA as compared with volume-based plans. Mean doses to the bladder, rectum, and sigmoid colon were least with IPSA (mean V80 of 2.62%, 5.38%, and 0%; mean V100 of 0.22%, 0.79%, and 0%; and D2 cm3 of 77.41%, 83.23%, and 22.60% for the bladder, rectum, and sigmoid
Discussion HDR brachytherapy plays an important role in the management of carcinoma of the cervix. The available optimization techniques such as geometric and dose point optimization fail to use the anatomic information. These optimization techniques are based only on the location of the active dwell times. Hence, these methods necessarily result in an approximation of the shape of the anatomy. To maintain the complete target coverage and reduction of dose to normal OARs, the dose distribution should be as conformal as possible. It is useful to be reminded of the huge patient load that we receive in our hospital. One should also remember that inserting
Table 2 Dosimetric parameters for the volume- and IPSA-based optimization techniques Parameter
Volume based Mean (SD)
IPSA based Mean (SD)
Target V100 Target V150 Target V200 CI HI ODI DNR Bladder (D2 cm3 ) Bladder V80 Bladder V100 Rectum (D2 cm3 ) Rectum V80 Rectum V100 Sigmoid colon (D2 cm3 ) Sigmoid colon V80 Sigmoid colon V100
97.56 (1.33)% 53.06 (14.9)% 22.63 (18.88)% 0.87 (0.37) 0.37 (0.19) 0.23 (0.19) 0.54 (0.15) 103.28 (26.81)% 9.14 (7.11)% 3.86 (5.21)% 109.62 (32.33)% 14.96 (10.97)% 6.32 (7.21)% 31.80 (8.28)% 0% 0%
86.30 (1.41)% 35.46 (8.24)% 12.09 (3.78)% 0.76 (0.33) 0.45 (0.21) 0.14 (0.04) 0.40 (0.09) 77.41 (11.35)% 2.62 (2.17)% 0.22 (0.44)% 83.23 (15.46)% 5.38 (4.49)% 0.79 (1.56)% 22.60 (4.47)% 0% 0%
D2 cm3 ¼ dose received by 2 cm3 of volume; V100 ¼ volume receiving 100% of the prescription dose; SD ¼ standard deviation.
4
Pelagade et al / Medical Dosimetry ] (2015) ]]]–]]]
Fig. 3. Comparative slice views on the same CT slice and 3D views for (A) volume-based and (B) IPSA-based dose optimizations. (Color version of figure is available online.)
the Martinez Universal Perineal Interstitial Template itself is a time-consuming process, and it is also necessary that the treatment is started as soon as possible keeping in mind the pain and inconvenience that the patient goes through. Hence, the value of time in this procedure is unparalleled. Therefore, we require a procedure that does optimal planning in the shortest possible time. IPSA offers us just that. Not only does it significantly reduce the treatment time, but it also offers much better dose
homogeneity, which is the need of the hour. The reduction in doses to the bladder, rectum, and sigmoid colon would significantly reduce the long-term morbidity associated with the procedure that often deters physicians from using this procedure. In fact, just as intensity-modulated radiation therapy is preferred over 3D conformal radiotherapy in external radiotherapy because of its ability to model the treatment plan according to the desired objectives and constraints, similarly, we feel that IPSA is a much
Pelagade et al / Medical Dosimetry ] (2015) ]]]–]]]
better technique than volume-based planning with respect to planning time and dose homogeneity. Dose distribution can be manually obtained by adjusting relative dwell times until an acceptable solution is found. The computer is used only to calculate the dose distribution once the plan has been decided by the planner. This approach, in combination with geometric optimization, requires more time and skill. The existing commercial planning systems may not be truly anatomy based without a genuinely anatomy-based optimization. Procurement of anatomy-based optimization brings the planning process nearest to the real clinical issues. In this study, we described and compared 2 methods of 3D dose optimization (volume-based and inverse planning methods), by retrospectively analyzing dose distributions to the target and OARs using various dosimetric indices. The volume-based optimization method proved to provide superior dosimetric target coverage than IPSA-based treatment plan. However, this improvement in target coverage was associated with a slightly higher dose to the OARs. IPSA also decreased the volume of the hot spot within the target volume and dose to normal structures outside the implant. This was illustrated through quantitative comparisons of DVH parameters and dosimetric indices. IPSA accomplished this by utilizing anatomic information provided by 3D planning. Data from this study support the need to move to 3D brachytherapy treatment planning and the use of anatomy-based dwell time optimization for carcinoma of cervix HDR brachytherapy. There are no clearly defined rules for treatment plan optimization in brachytherapy, and the exact limits of variations between dwell times are not yet known. In this regard, software tools performing automatic optimization of the dose distribution should be used with caution. This is not the case with the inverse planning method, where constraints for the dwell times can be easily set, resulting in more homogeneous dwell times and making this method appear safer. Thus, inverse planning is better and reliable.
Conclusions (1) The article demonstrates the new IPSA-based optimization method for interstitial implants of the cervix. IPSA-based optimization produces clinically acceptable dose distributions if the constraints are chosen appropriately. With the use of IPSA, the dose coverage of the target decreases slightly; however, the dose homogeneity increases and the dose to OARs decreases, as compared with the use of volume-based optimization. (2) The high-dose regions and overdose regions were smaller and the dose distribution was more homogeneous with IPSAbased dose optimization than with volume-based dose optimization. (3) We proved that for interstitial implant of the cervix, IPSAbased optimization technique produces clinically acceptable dose distributions in the coverage and the homogeneity. Additionally, the dose to OARs can be kept well under acceptable limits. The dose coverage of the target with IPSA
5
optimization was a little lower than the volume-based optimization. The high-dose regions were larger and dose distribution was less homogenous with volume-based optimization. There was significant dose difference to OARs between both the techniques, but volume-based optimization resulted in larger OAR doses. (4) We demonstrated that the use of IPSA-based optimization can improve the dose distributions. Volume-based dose optimization performed on points placed on the surface of PTV can result in highly conformal dose distributions, but only at the cost of deterioration of dose homogeneity. (5) With IPSA optimization, a spectrum of alternative solutions is available and the treatment planner can select the solution that best satisfies the clinical constraints. The computerbased inverse planning method provides better dose distributions. The planner requires much larger time with volume-based optimization method to obtain a satisfactory solution requiring adjustments of the catheter positions than the automatic inverse planning method. This reduces the overall treatment time significantly, allowing the treatment of more patients.
References 1. Isohashi, F.; Yoshioka, Y.; Koizumi, M.; et al. High-dose-rate interstitial brachytherapy for previously untreated cervical carcinoma. Brachytherapy 8(2):234–9; 2009. 2. Bahaduri, Y.; Constantinescu, C.; Ezzat, M.; et al. 3D anatomy-based planning optimization for HDR brachytherapy of cervix cancer. Saudi J. Obstet. Gynecol. 11 (2):1430; 2009. 3. Morton, G.C.; Sankreacha, R.; Halina, P.; et al. A comparison of anatomy-based inverse planning with simulated annealing and graphical optimization for high-dose-rate prostate brachytherapy. Brachytherapy 7(1):12–6; 2008. 4. Jamema, S.V.; Saju, S.; Shetty, U.M.; et al. Dosimetric comparison of inverse optimization with geometric optimization in combination with graphical optimization for HDR prostate implants. J. Med. Phys. 31(2):89–94; 2006. 5. Lessard, E.; Hsu, I-Chow; Pouliot, J. Clinical benefits of inverse planning for high dose rate prostate brachytherapy, IFMBE proceedings, 14/3; 2006. 6. Skowronek, J.; Zwierzchowski, G.; Piotrowski, T.; et al. Optimization and it’s influence on value of doses in HDR and PDR brachytherapy. Neoplasma 57 (4):369–76; 2010. 7. Hsu, I-Chow J.; Lessard, E.; Weinberg, V.; et al. Comparison of inverse planning simulated annealing and geometrical optimization for prostate high-dose-rate brachytherapy. Brachytherapy 3:147–52; 2004. 8. Kim, H.; Beriwal, S.; Houser, C. IPSA based HDR interstitial brachytherapy planning for vaginal cancer: dosimetric comparison with manually optimized planning. Brachytherapy 10(1):S57–8; 2011. 9. Kneschaurek, P.; Schiessl, W.; Wehrmann, R. Volume-based dose optimization in brachytherapy. Int. J. Radiat. Oncol. Biol. Phys. 45(3):811–5; 1999. 10. Wu, A.; Ulin, K.; Sternick, E.S. A dose homogeneity index for evaluating Ir-192 interstitial breast implants. Med. Phys. 15:104–7; 1988. 11. Anbumani, S.; Anchineyan, P.; Narayanasamy, A.N.; et al. Treatment planning methods in high dose rate interstitial brachytherapy of carcinoma cervix: a dosimetric and radiobiological analysis. ISRN Oncol.; 20142014 Article ID 125020. 12. Saw, C.B.; Suntharalingam, N. Quantitative assessment of interstitial implants. Int. J. Radiat. Oncol. Biol. Phys. 20(1):135–9; 1991. 13. Chaswal, V.; Yoo, S.; Thomadsen, B.R.; et al. Multi-species prostate implant treatment plans incorporating 192Ir and 125I using a Greedy Heuristic based 3D optimization algorithm. Med. Phys. 34(2):436–44; 2007. 14. Shwetha, B.; Ravikumar, M.; Katke, A.; et al. Dosimetric comparison of various optimization techniques for high dose rate brachytherapy of interstitial cervix implants. J. Appl. Clin. Med. Phys. 11(3):225–30; 2010.