- Email: [email protected]
Influence of dose point and inverse optimization on interstitial cervical and oropharyngeal carcinoma brachytherapy
Radiotherapy and Oncology 73 (2004) 331–337 www.elsevier.com/locate/radonline
Influence of dose point and inverse optimization on interstitial cervical and oropharyngeal carcinoma brachytherapy Faten Ahmada,c,*, Pierre Alettia,c, Claire Charra-Brunaudb, Rob Van der Laarsed, Michel Lapeyreb, Sylvette Hoffstetterb, Didier Peiffertb, Alain No¨ela,c a
Medical Physics Department, Centre Alexis Vautrin, av. de Bourgogne, 54511 Vandoeuvre-les-Nancy Cedex, France Brachytherapy Department, Centre Alexis Vautrin, av. de Bourgogne, 54511 Vandoeuvre-les-Nancy Cedex, France c CRAN, CNRS UMR 7039, Institut National Polytechnique de Lorraine, av. de la foreˆt de Haye, 54516 Vandoeuvre-les-Nancy, France d Nucletron, Waardgelder 1, 3905 TH, Veenendaal, The Netherlands b
Received 9 September 2003; received in revised form 5 October 2004; accepted 14 October 2004
Abstract Background and purpose: Evaluation of the use of optimization methods in interstitial cervical and oropharyngeal brachytherapy; evaluation of the conformal index (COIN) and the natural dose ratio (NDR) to quantify the implant quality. Material and methods: CT-based dose distributions were obtained for seven implants according to the Paris system. CT-based implants were used to assess the dose point and inverse optimization methods. To compare the results of these planning methods, the coverage index (CI), normal tissue irradiation (NTI), and the protection of organs at risk (OARs) were evaluated using cumulative dose volume histograms (CDVH). Results: In regular cervical implants, a CI of 94 and 96%; a NTI of 35 and 28% resulted for non-optimized and optimized implants, respectively. In irregular cervical implants, a CI of 88, 96, and 90%; a NTI of 44, 37, and 44% resulted for non-optimized, dose point optimized, and inverse optimized implants, respectively. Compared to the non-optimized implants; both optimization methods resulted in better protection for the bladder wall. As for the protection of the rectal wall, only the inverse optimization gave a better result. In oropharyngeal implants, a better CI resulted after dose point optimization. Irradiation of the contralateral parotid were improved after both optimization methods. The maximum change in COIN that could have been achieved by optimization was 3%, as CI and NTI increased similarly. For the same value of COIN, an underdosage of PTV was avoided by the optimization methods as NDR increased from 0.86 to 1.01. Conclusion: CT-based optimized implant allows conformation of the dose distribution to the PTV while sparing normal tissue and organs at risk. COIN and NDR should be used together to evaluate both doses to normal tissue and organs at risk, and an under- or overdose inside the PTV. q 2004 Published by Elsevier Ireland Ltd. Keywords: Cervical implant; Oropharyngeal implant; Inverse optimization; Dose point optimization; COIN; NDR
1. Introduction Until now, the clinical use of imaging modalities such as CT images [8,22], does not play a big role in dose calculation for cervical and oropharyngeal implants. In our institution,
* Corresponding author. Address: Radiophysics Department, Centre Alexis Vautrin, Av. de Bourgogne, 54511 Vandoeuvre-les-Nancy Cedex, France. 0167-8140/$ - see front matter q 2004 Published by Elsevier Ireland Ltd. doi:10.1016/j.radonc.2004.10.008
cervical interstitial implants are used with pulsed dose rate (PDR) brachytherapy using a stepping 192Ir source. Oropharyngeal implants are used with low dose rate (LDR) brachytherapy using 192Ir wires. The clinical treatment planning is performed with orthogonal radiographs. LDR can be replaced by remote controlled PDR afterloading brachytherapy using a stepping source [4] with the same overall treatment time and average dose rate. Brachytherapy performed with PDR systems has a distinctive advantage
332
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
compared to LDR, because the dose distribution can be optimized by stepping the source at certain positions (dwell positions) at predefined times (dwell times) [1,6,10–12,18]. The aim of optimization is to produce a dose distribution that covers the planning target volume (PTV), spares the critical structures, and limits the areas of very high dose values within the PTV [3,11,12]. Optimization algorithms have been developed in stepping source dosimetry systems, such as dose point and geometric optimization [20]. The goal of these algorithms is to optimize the uniformity of the dose distribution inside the treated volume. These algorithms only take into account the implant geometry with no consideration for the PTV and the critical structures. Recently, two three-dimensional (3D) optimization methods have been introduced in the PlatoBPS treatment planning system: inverse optimization and graphical optimization. In addition to the requirement of homogeneous target coverage, inverse and graphical optimization attempt to spare critical organs which are near, or even inside, the implanted volume. This study provides an evaluation of the effect of dose point, inverse, and graphical optimization on the dose distribution in cervical and oropharyngeal implants. This study also investigates two quality indices to examine the implant quality: the conformal index (COIN) [3], and the natural dose ratio (NDR) [2,14].
with the same overall treatment time and an identical average dose rate. 2.2. PTV and organs at risk Post implant CT images were obtained for all implants. Slice thickness was 3.2 mm and the interslice separation was 3 mm. PTV definition was performed according to ICRU 58 recommendations [9]. In the cervical implants, the organs at risk (OARs) were the rectal wall and the bladder wall, both with a wall thickness of 5 mm. In the oropharyngeal implants the horizontal ipsilateral mandible and both lateral parotids were taken as OARs. 2.3. Implant geometry based on orthogonal radiographs and on CT images: active length considerations For clinical treatment, the dose distribution was calculated according to the Paris system using the implant geometry based on orthogonal radiographs. With the Paris system, the active lengths of the catheters extended beyond the PTV (Active length/Treated lengthZ0.7). The dwell positions were confined to the PTV for the CT-based implant. Using optimization methods based on CT images, only the catheter segments inside the PTV were loaded (Active length/Treated lengthZ1.0). 2.4. Dose point, inverse, and graphical optimization
2. Materials and methods The dose distributions and the dose volume histograms (DVHs) were calculated with the PLATO Brachytherapy Planning System (PLATO-BPS) v14.2. The TG43 dose calculation formalism was used by the PLATO system [15]. 2.1. Implant geometries 2.1.1. Cervical implants Four interstitial PDR cervical implants were used as a boost (20 Gy) for treatment of a tumor extending laterally from the cervix. Two implants with regular geometry were performed following the implantation rules of the Paris system [7]. In the other implants the geometry was irregular due to a large PTV close to the rectum and the bladder. A range of 4–9 catheters were inserted into the paravaginal and parametrial region. 2.1.2. Oropharyngeal implants Three oropharyngeal LDR implants were performed as a boost (25 Gy) for the treatment of laterally localized tumors. The implants were performed according to the technique described by Pernot [16,17] where two plastic catheters are passed in parallel through the anterior and posterior pillars on the side of the lesion, and a supplementary loop is implanted in the posterior portion of the tongue. The LDR oropharyngeal implants were replaced by PDR implants
In addition to the Paris-type implant, where uniform dwell times were used, dose point optimization, inverse optimization and graphical optimization were evaluated [20]. 2.4.1. Dose point optimization Dose point optimization on distance was performed in cervical and oropharyngeal implants using equidistant target dose points automatically generated by the planning system on the surface of PTV with a spacing of 5 mm. 2.4.2. Inverse optimization Inverse optimization was performed to fulfil dose constraints imposed on the PTV and on organs at risk. The inverse optimization algorithm identifies the vulnerability of each organ at risk as a weight to the factor in the object function that minimizes the maximum absorbed dose on the organ surface. The vulnerability of each organ at risk is correlated with the distance between the organ at risk and the implant. Farther from the implant the inverse square law is predominant and no optimization is possible. As a result, depending on the distance between an organ at risk and the implant, this organ is either taken into account by the optimization, or is ignored. In the interstitial cervical implant, the dose constraints imposed on the PTV was 100%. The 100% dose in the inverse optimization algorithm is the mean dose on the PTV surface. The dose constraints
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
for the bladder wall and rectal wall as a percentage of the mean dose on the PTV surface, were 87% (range: 40–110%) and 89% (range: 70–100%), respectively. In oropharyngeal implants, organs defined around the PTV were not taken into account as organs at risk by the inverse optimization algorithm, because of the large distance (O1 cm) between the implant and these organs, so the inverse optimization was performed with the dose constraint of 100% imposed on the PTV only. 2.4.3. Graphical optimization The impact of this method was studied in one cervical implant. In each slice the dose distribution was adjusted graphically to improve the PTV coverage and normal tissue irradiation after dose point optimization or after inverse optimization. 2.5. Dose prescription 2.5.1. Regular implants For the regular cervical implants, the dose distribution was always normalized in relation to the average dose in the basal dose points defined according to the Paris system. The mean central dose (MCD) was defined to be equal to the average dose in the basal dose points as it is recommended by ICRU 58 [9]. In non-optimized implants, the dose prescription was selected at 85% (range: 83–87%) of the MCD. In optimized implants the dose prescription was selected at 90% (range: 85–95%) of the MCD to avoid an increase in the treated volume [10,18]. 2.5.2. Irregular implants For the irregular cervical and oropharyngal implants, a method similar to the Paris system was used. A central plane passing through the geometric center of the implant was defined. In this central plane, triangles were reconstructed from the intersections with the catheters. The geometric centers of these triangles were used as reference points. These reference points where the minimal doses were located, could be regarded as basal dose points. The MCD was defined as the average dose over these points. In the non-optimized and in the optimized implant, the dose prescription was selected at 67% (range: 57–77%), and at 80% (range: 73–85%) of the MCD for the irregular cervical implants and for oropharyngeal implants, respectively.
333
covered by the prescription dose (VPTV,PD). The resulting volume was normalized to the treated volume: NTI Z ðV100 K VPTV;PD Þ=V100 The recommendations of ICRU 58 [9] were followed to calculate the MTD (minimum target dose) which is defined as the reference dose (85% of the MCD) in the Paris system, and the high dose volumes which are defined as the volumes encompassed by the isodose surface corresponding to 150% of the MCD. COIN was calculated from the cumulative DVH (CDVH) of the PTV and the implant [3,13]. The NDR was derived from the natural DVH (NDVH) of the implant and the CDVH of the PTV. It is defined as the ratio of the natural prescription dose (NPD) and the prescription dose (PD) (NDRZNPD/PD), where NPD, located at the base of the NDVH peak, is the optimal prescription dose according to the treated volume, and PD is the optimal prescription dose according to PTV coverage [14,20]. Dose to OARs was evaluated by the CDVH calculated for each organ. In the cervical implants the dose evaluation for the OARs was performed for 2 and 5 cm3 of the rectal wall and bladder wall. A volume of 2 cm3 of tissue in the highest dose region is considered clinically as a relevant volume which corresponds to the size of a fistula [22], and a volume of 5 cm3 was chosen to present the dose to a significant volume of tissue in the highest dose region. In the oropharyngeal implants, the part of volume receiving at least 8 Gy (V8 Gy) for the horizontal ipsilateral mandible was recorded, the choice of the dose value received by the organs at risk was based on a representative value in the highest dose region. The mean dose over both lateral parotids was evaluated using Evaluation System (PLATO-EVAL) v3.0. 2.7. Plan comparison Non-optimized implants based on the orthogonal radiographs were compared to non-optimized CT-based implants. The optimized plans were compared to the nonoptimized ones using only the CT-based implant. The impact of additional graphical optimization on the dose distribution was evaluated in one regular cervical implant, and the use of COIN and NDR to determine implant quality was evaluated in this implant.
2.6. Plan evaluation 3. Results The coverage index (CI) [13,19], defined by the fraction of the PTV receiving a dose equal to or greater than the prescription dose, was derived from the CDVH calculated for the PTV. The normal tissue irradiation (NTI) was measured using the difference between the treated volume (V100), defined as the volume encompassed by the prescription dose (PD) surface, and the volume of PTV
For cervical implants the MTD was 20 Gy for the nonoptimized regular implants, 19 Gy for the optimized regular ones and 26 Gy for the non-optimized and optimized irregular ones (Tables 1 and 2). For the regular cervical implants, when the CT-based implant geometry was used for the dose calculation,
334
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
Table 1 The mean values and the variation range of the evaluation parameters for the non-optimized and optimized dose distribution of the regular cervical implants
Non-optimized implant based on radiographs CT-based implant Without optimization Dose point or inverse optimizationb a b
Treated volume (cm3)
CIa (%)
NTIa (%)
High dose volume (cm3)
65 (range: 35– 95)
96 (range: 94–97)
49 (range: 37–60)
8.9 (range: 6.4–11)
46 (range: 35–56) 43 (range: 30–56)
94 (range: 90–97) 96 (range: 95–96)
35 (range: 33–37) 28 (range: 26–30)
7.2 (range: 6.4–8) 7 (range: 5.8–9)
See appendix for notation. The parameter values for the dose point optimization and the inverse one did not differ significantly so both optimization methods are combined.
the non-optimized plan resulted in a decrease of the NTI from 49 to 35%, and of the high dose volume from 8.9 to 7.2 cm3, compared to the non-optimized implant based on the orthogonal radiographs (Table 1). Using optimization, the target coverage improved to 96%, and the NTI decreased further to 28%. For the irregular cervical implants, there are no differences between the parameters for non-optimized implants based on radiographs and the ones based on CT images (Table 2). The target coverage was improved with both optimization methods. A better target coverage (CIZ96%) and a better normal tissue irradiation (NTIZ37%) resulted from the dose point optimization, compared to the inverse optimization where a CI Z90% and a NTIZ44% were obtained. The optimized CT-based implant resulted in a decrease in the bladder wall dose (Table 3). The best protection of the rectal wall resulted from the inverse optimization method. Since the dose constraints in the inverse optimization algorithm are calculated as percentages of the mean dose on the PTV surface, the resulting maximum organ at risk dose is also expressed as a percentage of the mean dose on the PTV surface. The mean dose on the PTV surface was 33.5 Gy (range: 28–43 Gy). The mean dose constraints imposed on the bladder wall and rectal wall were 87% (range: 40–110%) and 89% (range: 70–100%), respectively. The mean resulting maximum doses on the organs at risk are below the mean dose constraints imposed on them. A maximum dose of 78% (range: 25–107%) and 70% (range: 52–94%) were the results for the bladder wall and rectal wall, respectively. The CI could be increased to 99% with dose point or inverse optimization, followed by graphical optimization
(Table 4) at the cost of a small increase in normal tissue irradiation (C5.3%, range: 5–5.5%) and in high dose volume (C2.4 cm3, range: 2–2.8 cm3). The maximum change in COIN due to the addition of graphical optimization was 3%, from 0.61 to 0.58, as the CI and normal tissue dose similarly increased. However, graphical optimization allowed to avoid an underdosage of the PTV as the NDR increased from 0.86 to 1.01. For all oropharyngeal implants the MTD was 26.5 Gy for the non-optimized and optimized implant. When comparing the CT-based implant to the implant based on orthogonal radiographs, the CT-based implant allowed to obtain the same CI (94%) with a decrease in normal tissue irradiation of 53% (range: 44–61%) to 40% (range: 35–50%), and almost identical high dose volume of 8.3 cm3 (range: 4–12 cm3). After dose point optimization, a better target coverage (CIZ96%) resulted with the same normal tissue irradiation (NTIZ40%) and with a limited variation of NTI range (range: 36–42%) compared to non-optimized CT-based implants. This was also true when the dose point method was compared to the inverse optimization where a CIZ90% and a NTIZ40% (range: 33–44%) were obtained. The mean value of the high dose volume was 8.7 cm3 (range: 4–15 cm3) using both optimization methods. When comparing non-optimized CT-based oropharyngeal implant to implant based on orthogonal radiographs, the V8 Gy of the horizontal ipsilateral mandible decreased from 1.2 to 0.42 cm3, without any modification of the mean dose (3.1 Gy; range: 2–4.6 Gy) over ipsilateral parotid and of the mean dose (1.13 Gy; range: 0–1.74 Gy) over contralateral one. When comparing optimized to non-optimized CT-based implants, the mean dose over the contralateral parotid decreased from 1.13 to 0.6 Gy
Table 2 The mean values and the variation range of the evaluation parameters for the non-optimized and optimized dose distribution of the irregular cervical implants
Non-optimized implant based on radiographs or CT imagesb CT-based implant Dose point optimization Inverse optimization a
Treated volume (cm3)
CIa (%)
NTIa (%)
High dose volume (cm3)
119 (range: 70–169)
88 (range: 80–96)
44 (range: 32–56)
33 (range: 14–52)
104 (range: 87–121) 108 (range: 105–111)
96 (range: 92–99) 90 (range: 89–90)
37 (range: 37–37) 44 (range: 38–50)
30 (range: 23–37) 33 (range: 27–38)
See appendix for notation. The parameter values for non-optimized implant based on radiographs or on CT images did not differ significantly so both non-optimized implant modes are combined. b
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
335
Table 3 The mean dose and the dose range, as a percentage of the prescribed dose, received by 2 and 5 cm3 of the rectal and bladder walls Dose to bladder wall (%)
Non-optimized implant based on radiographs or CT imagesa CT-based implant Dose point optimization Inverse optimization
Dose to rectal wall (%) 5 cm
2 cm3
5 cm3
82 (range: 29–118)
53 (range: 19–91)
80 (range: 57–113)
59 (range: 42–95)
69.4 (range: 26–86) 71 (range: 22–92)
46 (range: 19–72) 47 (range: 16–48)
80 (range: 51–109) 68 (range: 51–76)
58 (range: 38–87) 47 (range: 38–57)
2 cm
3
3
a
The parameter values for non-optimized implants based on radiographs or on CT images did not differ significantly so both non-optimized implant modes are combined.
optimization or geometric optimization [1,10,13,18]. A value of 90% is recommended for geometric optimization [1,18]. Our study used a value of 90% of the MCD for the dose point and the inverse optimization methods, which resulted in a good target coverage (CIZ96%), with almost the same treated volume and high dose volume, and with a significant decrease in normal tissue irradiation, compared to non-optimized CTbased implant. In the irregular implants there was no increase in the treated volume using optimization methods, so the dose prescription was kept unchanged between non-optimized and optimized implants. A similar result was shown in a publication on the optimization of an interstitial implant [10] where the dose specification in an irregular volume implant of the base of tongue was kept unchanged after geometric optimization, as the volume enclosed by the reference isodose did not increase. Currently available optimization methods, such as geometric and dose point optimization, are based only on the location of the active dwell positions. The objective of these methods is to improve the dose homogeneity over the treated volume. A publication on the relationship between ‘hot’ spots and complications following high dose rate brachytherapy [21] showed that the heterogeneity appeared to be minor with regard to complications, so the hot spots need not preclude optimization to ensure adequate dosage to all parts of the PTV. The introduction of three-dimensional images in brachytherapy planning should allow to optimize the dose distribution based on anatomic information. The objective
(range: 0–1 Gy) without any modification of the mean dose over the ipsilateral parotid.
4. Discussion 4.1. Role of CT images The dose distribution of an implant calculated using geometry based on orthogonal radiographs can be improved significantly in terms of conformation to the PTV when CT images are used. The definition of the PTV is not possible on orthogonal radiographs, so the conformality of the dose distribution to the PTV cannot be defined and the dose distribution is always calculated according to the implant geometry. 4.2. Impact of optimization methods When the dose distribution was optimized, the active length inside the catheters was adapted to the PTV length. With the dose point and geometric optimization methods, the increase in dwell times of source positions at the outer parts of the implant decreases the bending of the reference isodose between the ends of the catheters [10,13]. The inverse optimization algorithm is essentially based on the geometric optimization method, so the active length should also be adapted to the PTV length to avoid an increase in the treated volume. Several studies propose for regular implants to select a higher isodose than 85% of the MCD after dose point
Table 4 Impact of the dose point and inverse optimization followed by graphical optimization of a cervical implant
CT-based implant geometry without optimization Dose point optimization Without graphical With graphical Inverse optimization Without graphical With graphical a
See appendix for notation.
CIa (%)
NTIa (%)
High dose volume (cm3)
COINa
NDRa
90
43
9
0.52
0.98
91 99
38.5 44
9.2 12
0.56 0.55
0.86 0.99
95 99
36 41
9.3 11.3
0.61 0.58
0.86 1.01
336
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
of the optimization based on CT images can be expressed in target coverage, homogeneity of dose over the target, irradiation of normal tissue and organs at risk. The dose point optimization used in this study is always based on the active dwell positions, but the use of CT images allows us to keep the dwell positions inside the target volume and also to optimize the dose distribution using dose points placed on the target surface. The inverse optimization method optimizes the dose distribution to satisfy the dose constraints imposed on the target volume and the organs at risk. In the cervical implants, the benefit of the dose point method with dwell positions inside the target volume as defined by the CT images, was the improvement of the target coverage with a decrease in the normal tissue and bladder wall irradiation. The impact of inverse optimization was significant for rectal wall and bladder wall protection, but this reduction of irradiation on organs at risk came with a decrease in target coverage compared to dose point optimization in the irregular implants. However, target coverage, normal tissue irradiation and organs at risk protection obtained with the inverse optimization method, were clinically considered as an interesting improvement of the treatment plans. The dose constraints of the inverse optimization method are based on the implant geometry, the influence of these dose constraints on the resulting dose distribution is strongly related to the distance between the implant and each organ at risk. In the cervical implants used in this study, the rectal wall was located close to the implant, consequently, the dose constraints imposed were relatively high as a percentage of the mean target dose. The bladder wall was located near the implant in three implants, and it was located slightly farther in one implant (3 mm from the last active position longitudinally), consequently, the dose constraints imposed were again relatively high as a percentage of the mean target dose except in one implant. The resulting maximum doses in these organs satisfied the dose constraints imposed. When an implant was optimized without additional graphical optimization, the target was not completely covered by the prescription isodose. With graphical optimization the target coverage could be further improved: the CI increased, together with an increase of normal tissue irradiation. For the value of COIN, the increase in PTV coverage was compensated by the increase in normal tissue irradiation. Comparing two dose distributions where COIN has almost the same value (0.56–0.55), the NDR indicated that in one dose distribution the PTV coverage was realized at the expense of underdosage (NDRZ0.86), but in the other dose distribution this underdosage was completely avoided (NDRZ0.99). A value of NDR lower than 1.0 indicates that the prescription isodose surface lies (partly) outside the treated volume [14]. The manual adjustment of the dose distribution with graphical optimization in some CT slices or in other planes
through the implant, results in local modifications of the dose distribution of the whole implant, so a sufficient optimization is difficult to reach with this method. With graphical optimization, the whole dose distribution should be evaluated after each adjustment. Peiffert et al. [16] showed the feasibility of PDR brachytherapy for head and neck carcinomas and suggested that clinical optimization of dose distribution should be performed, taking into account the information of 3D imaging correlated with computed tomography scan slices. With this new information and the use of optimization, the occurrence of mandible osteonecrosis can be reduced. The results of our study on the CT-based oropharyngeal implants showed an improvement of the PTV coverage with a reduction of normal tissue irradiation using dose point optimization compared to non-optimized implants. Using inverse optimization, a bad PTV coverage was observed in one patient (76%) because only a small number of catheters were inside the PTV. There were no dose constraints for organs at risk in the oropharyngeal implants because of their distant location from the implant, resulting in a low dose gradient. Modification of the dose in organs distant from the target cannot be done by changing the relative dwell times of dwell positions, while keeping the prescription dose of the implant unchanged. The decrease in the horizontal ipsilateral mandible dose when using CT-based implant, compared to standard implant using orthogonal radiographs, showed the possibility of reducing the mandible osteonecrosis. Another advantage of the optimization method is the reduction of the mean dose to the contralateral parotid without increasing the dose to the ipsilateral parotid. A prospective study on salivary function sparing in patients with head and neck cancers receiving intensity-modulated or three-dimensional radiation therapy showed that the sparing of the parotid glands translates into improvement of both xerostomia and quality-of-life scores [5].
5. Conclusions This work shows that the optimized cervical and oropharyngeal CT-based implants offer interesting possibilities to further adapt the dose distribution to the target volume while sparing normal tissue and organs at risk. Inverse optimization can be used in clinical practice for sparing the organs at risk, without the problem of placement of dose points as in dose point optimization. The evaluation indices, COIN and NDR, should be used together because the prescription dose can always be selected to cover the PTV but it is also essential to evaluate the dose to normal tissue and organs at risk, and the matching of the treated volume with the PTV.
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
Appendix A. Abbreviations and notation CDVH: the cumulative dose–volume histograms are representations of the dose–volume relation for each organ; COIN: the conformal index takes into consideration the coverage of the PTV by the reference dose and also the unwanted irradiation of normal tissue outside the PTV [3]; CI: the coverage index is the fraction of the PTV receiving a dose equal to or greater than the prescribed dose [13]; MCD: the mean central dose is taken to be the arithmetic mean of the local minimum doses between sources, in the central plane [9]; MTD: the minimum target dose is the reference dose (85% of the MCD) under the Paris system [9]; NTI: the normal tissue irradiation is the fraction of the normal tissue outside the PTV receiving the prescribed dose; NDVH: the natural dose volume histogram is a representation of the dose–volume relation for the dose distribution with suppression of the inverse square law effect [2]; NPD: the natural prescription dose, located at the base of the NDVH peak, is the optimal prescription dose according to the treated volume [14]; NDR: the natural dose ratio is derived from the NDVH of the implant and the CDVH of the PTV. It is defined as the ratio of the NPD and the PD (NDRZNPD/ PD) [14]; PD: the prescription dose; V100: the treated volume is the volume of the prescribed dose; VPTV,PD: the PTV volume covered by the prescribed dose; V8 Gy: The volume receiving at least 8 Gy.
References [1] Anacak Y, Esassolak M, Aydin A, Aras A, Olacak I, Haydaroglu A. Effect of geometrical optimization on the treatment volumes and the dose homogeneity of biplane interstitial brachytherapy implants. Radiother Oncol 1997;45:71–6. [2] Anderson LL. A ‘natural’ volume–dose histogram for brachytherapy. Med Phys 1986;13:898–903. [3] Baltas D, Kolotas C, Geramani K, et al. A conformal index (COIN) to evaluate implant quality and dose specification in brachytherapy. Int J Radiat Oncol Biol Phys 1998;40:515–24. [4] Brenner DJ, Hall EJ. Conditions for the equivalence of continuous to pulsed low dose rate brachytherapy. Int J Radiat Oncol Biol Phys 1991;20:181–90. [5] Chao KS, Deasy JO, Markman J, et al. A prospective study of salivary function sparing in patients with head-and-neck cancers receiving intensity-modulated or three-dimensional radiation therapy: initial results. Int J Radiat Oncol Biol Phys 2001;49:907–16. [6] Charra-Brunaud C, Hsu IC, Weinberg V, Pouliot J. Analysis of interaction between number of implant catheters and dose–volume histograms in prostate high- dose-rate brachytherapy using a computer model. Int J Radiat Oncol Biol Phys 2003;56:586–91.
337
[7] Dutreix A, Marinello G. The Paris system. In: Pierquin B, Wilson JF, Chassagne D, editors. Modern brachytherapy. New York: Masson; 1987. p. 25–42. [8] Fellner C, Po¨tter R, Knocke TH, Wambersie A. Comparison of radiography- and computed tomography-based treatment planning in cervix cancer in brachytherapy with specific attention to some quality assurance aspects. Radiother Oncol 2001;58:53–62. [9] ICRU. Dose and volume specification for reporting interstitial therapy. ICRU Report 58. Bethesda, USA: ICRU; 1997. [10] Kolkman-Deurloo IK, Visser AG, Niel CG, Driver N, Levendag PC. Optimization of interstitial volume implants. Radiother Oncol 1994; 31:229–39. [11] Lahanas M, Baltas D, Zamboglou N. Anatomy-based three-dimensional dose optimization in brachytherapy using multiobjective genetic algorithms. Med Phys 1999;26:1904–18. [12] Lessard E, Hsu IC, Pouliot J. Inverse planning for interstitial gynecologic template brachytherapy: truly anatomy-based planning. Int J Radiat Oncol Biol Phys 2002;54:1243–51. [13] Major T, Polgar C, Fodor J, Somogyi A, Nemeth G. Conformality and homogeneity of dose distributions in interstitial implants at idealized target volumes: a comparison between the Paris and dose-point optimized systems. Radiother Oncol 2002;62:103–11. [14] Moerland MA, Van der Laarse R, Luthmann RW, Wijrdeman HK, Battermann JJ. The combined use of the natural and the cumulative dose–volume histograms in planning and evaluation of permanent prostatic seed implants. Radiother Oncol 2000;57:279–84. [15] Nath R, Anderson LL, Luxton G, Weaver KA, Williamson JF, Meigooni AS. Dosimetry of interstitial brachytherapy sources: recommendations of the AAPM Radiation Therapy Committee Task Group No. 43, American Association of Physicists in Medicine. Med Phys 1995;22:209–34. [16] Peiffert D, Castelain B, Thomas L, Ardiet JM, Baillet F, Mazeron JJ. Pulsed dose rate brachytherapy in head and neck cancers. Feasibility study of a French cooperative group. Radiother Oncol 2001;58:71–5. [17] Pernot M, Luporsi E, Hoffstetter S, et al. Complications following definitive irradiation for cancers of the oral cavity and the oropharynx (in a series of 1134 patients). Int J Radiat Oncol Biol Phys 1997;37: 577–85. [18] Pieters BR, Saarnak AE, Steggerda MJ, Borger JH. A method to improve the dose distribution of interstitial breast implants using geometrically optimized stepping source techniques and dose normalization. Radiother Oncol 2001;58:63–70. [19] Saw CB, Suntharalingam N. Quantitative assessment of interstitial implants. Int J Radiat Oncol Biol Phys 1991;20:135–9. [20] Van der Laarse R, Luthmann RW. Computers in brachytherapy dosimetry. In: Joslin CAF, Flynn A, Hall E, editors. Principles and practice of brachytherapy using afterloading systems. London: Arnold; 2001. p. 49–80. [21] Vikram B, Deore S, Beitler JJ, et al. The relationship between dose heterogeneity (‘hot’ spots) and complications following high-dose rate brachytherapy. Int J Radiat Oncol Biol Phys 1999;43:983–7. [22] Wachter-Gerstner N, Wachter S, Reinstadler E, et al. Bladder and rectum dose defined from MRI based treatment planning for cervix cancer brachytherapy: comparison of dose–volume histograms for organ contours and organ wall, comparison with ICRU rectum and bladder reference point. Radiother Oncol 2003;68:269–76.
Influence of dose point and inverse optimization on interstitial cervical and oropharyngeal carcinoma brachytherapy Faten Ahmada,c,*, Pierre Alettia,c, Claire Charra-Brunaudb, Rob Van der Laarsed, Michel Lapeyreb, Sylvette Hoffstetterb, Didier Peiffertb, Alain No¨ela,c a
Medical Physics Department, Centre Alexis Vautrin, av. de Bourgogne, 54511 Vandoeuvre-les-Nancy Cedex, France Brachytherapy Department, Centre Alexis Vautrin, av. de Bourgogne, 54511 Vandoeuvre-les-Nancy Cedex, France c CRAN, CNRS UMR 7039, Institut National Polytechnique de Lorraine, av. de la foreˆt de Haye, 54516 Vandoeuvre-les-Nancy, France d Nucletron, Waardgelder 1, 3905 TH, Veenendaal, The Netherlands b
Received 9 September 2003; received in revised form 5 October 2004; accepted 14 October 2004
Abstract Background and purpose: Evaluation of the use of optimization methods in interstitial cervical and oropharyngeal brachytherapy; evaluation of the conformal index (COIN) and the natural dose ratio (NDR) to quantify the implant quality. Material and methods: CT-based dose distributions were obtained for seven implants according to the Paris system. CT-based implants were used to assess the dose point and inverse optimization methods. To compare the results of these planning methods, the coverage index (CI), normal tissue irradiation (NTI), and the protection of organs at risk (OARs) were evaluated using cumulative dose volume histograms (CDVH). Results: In regular cervical implants, a CI of 94 and 96%; a NTI of 35 and 28% resulted for non-optimized and optimized implants, respectively. In irregular cervical implants, a CI of 88, 96, and 90%; a NTI of 44, 37, and 44% resulted for non-optimized, dose point optimized, and inverse optimized implants, respectively. Compared to the non-optimized implants; both optimization methods resulted in better protection for the bladder wall. As for the protection of the rectal wall, only the inverse optimization gave a better result. In oropharyngeal implants, a better CI resulted after dose point optimization. Irradiation of the contralateral parotid were improved after both optimization methods. The maximum change in COIN that could have been achieved by optimization was 3%, as CI and NTI increased similarly. For the same value of COIN, an underdosage of PTV was avoided by the optimization methods as NDR increased from 0.86 to 1.01. Conclusion: CT-based optimized implant allows conformation of the dose distribution to the PTV while sparing normal tissue and organs at risk. COIN and NDR should be used together to evaluate both doses to normal tissue and organs at risk, and an under- or overdose inside the PTV. q 2004 Published by Elsevier Ireland Ltd. Keywords: Cervical implant; Oropharyngeal implant; Inverse optimization; Dose point optimization; COIN; NDR
1. Introduction Until now, the clinical use of imaging modalities such as CT images [8,22], does not play a big role in dose calculation for cervical and oropharyngeal implants. In our institution,
* Corresponding author. Address: Radiophysics Department, Centre Alexis Vautrin, Av. de Bourgogne, 54511 Vandoeuvre-les-Nancy Cedex, France. 0167-8140/$ - see front matter q 2004 Published by Elsevier Ireland Ltd. doi:10.1016/j.radonc.2004.10.008
cervical interstitial implants are used with pulsed dose rate (PDR) brachytherapy using a stepping 192Ir source. Oropharyngeal implants are used with low dose rate (LDR) brachytherapy using 192Ir wires. The clinical treatment planning is performed with orthogonal radiographs. LDR can be replaced by remote controlled PDR afterloading brachytherapy using a stepping source [4] with the same overall treatment time and average dose rate. Brachytherapy performed with PDR systems has a distinctive advantage
332
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
compared to LDR, because the dose distribution can be optimized by stepping the source at certain positions (dwell positions) at predefined times (dwell times) [1,6,10–12,18]. The aim of optimization is to produce a dose distribution that covers the planning target volume (PTV), spares the critical structures, and limits the areas of very high dose values within the PTV [3,11,12]. Optimization algorithms have been developed in stepping source dosimetry systems, such as dose point and geometric optimization [20]. The goal of these algorithms is to optimize the uniformity of the dose distribution inside the treated volume. These algorithms only take into account the implant geometry with no consideration for the PTV and the critical structures. Recently, two three-dimensional (3D) optimization methods have been introduced in the PlatoBPS treatment planning system: inverse optimization and graphical optimization. In addition to the requirement of homogeneous target coverage, inverse and graphical optimization attempt to spare critical organs which are near, or even inside, the implanted volume. This study provides an evaluation of the effect of dose point, inverse, and graphical optimization on the dose distribution in cervical and oropharyngeal implants. This study also investigates two quality indices to examine the implant quality: the conformal index (COIN) [3], and the natural dose ratio (NDR) [2,14].
with the same overall treatment time and an identical average dose rate. 2.2. PTV and organs at risk Post implant CT images were obtained for all implants. Slice thickness was 3.2 mm and the interslice separation was 3 mm. PTV definition was performed according to ICRU 58 recommendations [9]. In the cervical implants, the organs at risk (OARs) were the rectal wall and the bladder wall, both with a wall thickness of 5 mm. In the oropharyngeal implants the horizontal ipsilateral mandible and both lateral parotids were taken as OARs. 2.3. Implant geometry based on orthogonal radiographs and on CT images: active length considerations For clinical treatment, the dose distribution was calculated according to the Paris system using the implant geometry based on orthogonal radiographs. With the Paris system, the active lengths of the catheters extended beyond the PTV (Active length/Treated lengthZ0.7). The dwell positions were confined to the PTV for the CT-based implant. Using optimization methods based on CT images, only the catheter segments inside the PTV were loaded (Active length/Treated lengthZ1.0). 2.4. Dose point, inverse, and graphical optimization
2. Materials and methods The dose distributions and the dose volume histograms (DVHs) were calculated with the PLATO Brachytherapy Planning System (PLATO-BPS) v14.2. The TG43 dose calculation formalism was used by the PLATO system [15]. 2.1. Implant geometries 2.1.1. Cervical implants Four interstitial PDR cervical implants were used as a boost (20 Gy) for treatment of a tumor extending laterally from the cervix. Two implants with regular geometry were performed following the implantation rules of the Paris system [7]. In the other implants the geometry was irregular due to a large PTV close to the rectum and the bladder. A range of 4–9 catheters were inserted into the paravaginal and parametrial region. 2.1.2. Oropharyngeal implants Three oropharyngeal LDR implants were performed as a boost (25 Gy) for the treatment of laterally localized tumors. The implants were performed according to the technique described by Pernot [16,17] where two plastic catheters are passed in parallel through the anterior and posterior pillars on the side of the lesion, and a supplementary loop is implanted in the posterior portion of the tongue. The LDR oropharyngeal implants were replaced by PDR implants
In addition to the Paris-type implant, where uniform dwell times were used, dose point optimization, inverse optimization and graphical optimization were evaluated [20]. 2.4.1. Dose point optimization Dose point optimization on distance was performed in cervical and oropharyngeal implants using equidistant target dose points automatically generated by the planning system on the surface of PTV with a spacing of 5 mm. 2.4.2. Inverse optimization Inverse optimization was performed to fulfil dose constraints imposed on the PTV and on organs at risk. The inverse optimization algorithm identifies the vulnerability of each organ at risk as a weight to the factor in the object function that minimizes the maximum absorbed dose on the organ surface. The vulnerability of each organ at risk is correlated with the distance between the organ at risk and the implant. Farther from the implant the inverse square law is predominant and no optimization is possible. As a result, depending on the distance between an organ at risk and the implant, this organ is either taken into account by the optimization, or is ignored. In the interstitial cervical implant, the dose constraints imposed on the PTV was 100%. The 100% dose in the inverse optimization algorithm is the mean dose on the PTV surface. The dose constraints
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
for the bladder wall and rectal wall as a percentage of the mean dose on the PTV surface, were 87% (range: 40–110%) and 89% (range: 70–100%), respectively. In oropharyngeal implants, organs defined around the PTV were not taken into account as organs at risk by the inverse optimization algorithm, because of the large distance (O1 cm) between the implant and these organs, so the inverse optimization was performed with the dose constraint of 100% imposed on the PTV only. 2.4.3. Graphical optimization The impact of this method was studied in one cervical implant. In each slice the dose distribution was adjusted graphically to improve the PTV coverage and normal tissue irradiation after dose point optimization or after inverse optimization. 2.5. Dose prescription 2.5.1. Regular implants For the regular cervical implants, the dose distribution was always normalized in relation to the average dose in the basal dose points defined according to the Paris system. The mean central dose (MCD) was defined to be equal to the average dose in the basal dose points as it is recommended by ICRU 58 [9]. In non-optimized implants, the dose prescription was selected at 85% (range: 83–87%) of the MCD. In optimized implants the dose prescription was selected at 90% (range: 85–95%) of the MCD to avoid an increase in the treated volume [10,18]. 2.5.2. Irregular implants For the irregular cervical and oropharyngal implants, a method similar to the Paris system was used. A central plane passing through the geometric center of the implant was defined. In this central plane, triangles were reconstructed from the intersections with the catheters. The geometric centers of these triangles were used as reference points. These reference points where the minimal doses were located, could be regarded as basal dose points. The MCD was defined as the average dose over these points. In the non-optimized and in the optimized implant, the dose prescription was selected at 67% (range: 57–77%), and at 80% (range: 73–85%) of the MCD for the irregular cervical implants and for oropharyngeal implants, respectively.
333
covered by the prescription dose (VPTV,PD). The resulting volume was normalized to the treated volume: NTI Z ðV100 K VPTV;PD Þ=V100 The recommendations of ICRU 58 [9] were followed to calculate the MTD (minimum target dose) which is defined as the reference dose (85% of the MCD) in the Paris system, and the high dose volumes which are defined as the volumes encompassed by the isodose surface corresponding to 150% of the MCD. COIN was calculated from the cumulative DVH (CDVH) of the PTV and the implant [3,13]. The NDR was derived from the natural DVH (NDVH) of the implant and the CDVH of the PTV. It is defined as the ratio of the natural prescription dose (NPD) and the prescription dose (PD) (NDRZNPD/PD), where NPD, located at the base of the NDVH peak, is the optimal prescription dose according to the treated volume, and PD is the optimal prescription dose according to PTV coverage [14,20]. Dose to OARs was evaluated by the CDVH calculated for each organ. In the cervical implants the dose evaluation for the OARs was performed for 2 and 5 cm3 of the rectal wall and bladder wall. A volume of 2 cm3 of tissue in the highest dose region is considered clinically as a relevant volume which corresponds to the size of a fistula [22], and a volume of 5 cm3 was chosen to present the dose to a significant volume of tissue in the highest dose region. In the oropharyngeal implants, the part of volume receiving at least 8 Gy (V8 Gy) for the horizontal ipsilateral mandible was recorded, the choice of the dose value received by the organs at risk was based on a representative value in the highest dose region. The mean dose over both lateral parotids was evaluated using Evaluation System (PLATO-EVAL) v3.0. 2.7. Plan comparison Non-optimized implants based on the orthogonal radiographs were compared to non-optimized CT-based implants. The optimized plans were compared to the nonoptimized ones using only the CT-based implant. The impact of additional graphical optimization on the dose distribution was evaluated in one regular cervical implant, and the use of COIN and NDR to determine implant quality was evaluated in this implant.
2.6. Plan evaluation 3. Results The coverage index (CI) [13,19], defined by the fraction of the PTV receiving a dose equal to or greater than the prescription dose, was derived from the CDVH calculated for the PTV. The normal tissue irradiation (NTI) was measured using the difference between the treated volume (V100), defined as the volume encompassed by the prescription dose (PD) surface, and the volume of PTV
For cervical implants the MTD was 20 Gy for the nonoptimized regular implants, 19 Gy for the optimized regular ones and 26 Gy for the non-optimized and optimized irregular ones (Tables 1 and 2). For the regular cervical implants, when the CT-based implant geometry was used for the dose calculation,
334
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
Table 1 The mean values and the variation range of the evaluation parameters for the non-optimized and optimized dose distribution of the regular cervical implants
Non-optimized implant based on radiographs CT-based implant Without optimization Dose point or inverse optimizationb a b
Treated volume (cm3)
CIa (%)
NTIa (%)
High dose volume (cm3)
65 (range: 35– 95)
96 (range: 94–97)
49 (range: 37–60)
8.9 (range: 6.4–11)
46 (range: 35–56) 43 (range: 30–56)
94 (range: 90–97) 96 (range: 95–96)
35 (range: 33–37) 28 (range: 26–30)
7.2 (range: 6.4–8) 7 (range: 5.8–9)
See appendix for notation. The parameter values for the dose point optimization and the inverse one did not differ significantly so both optimization methods are combined.
the non-optimized plan resulted in a decrease of the NTI from 49 to 35%, and of the high dose volume from 8.9 to 7.2 cm3, compared to the non-optimized implant based on the orthogonal radiographs (Table 1). Using optimization, the target coverage improved to 96%, and the NTI decreased further to 28%. For the irregular cervical implants, there are no differences between the parameters for non-optimized implants based on radiographs and the ones based on CT images (Table 2). The target coverage was improved with both optimization methods. A better target coverage (CIZ96%) and a better normal tissue irradiation (NTIZ37%) resulted from the dose point optimization, compared to the inverse optimization where a CI Z90% and a NTIZ44% were obtained. The optimized CT-based implant resulted in a decrease in the bladder wall dose (Table 3). The best protection of the rectal wall resulted from the inverse optimization method. Since the dose constraints in the inverse optimization algorithm are calculated as percentages of the mean dose on the PTV surface, the resulting maximum organ at risk dose is also expressed as a percentage of the mean dose on the PTV surface. The mean dose on the PTV surface was 33.5 Gy (range: 28–43 Gy). The mean dose constraints imposed on the bladder wall and rectal wall were 87% (range: 40–110%) and 89% (range: 70–100%), respectively. The mean resulting maximum doses on the organs at risk are below the mean dose constraints imposed on them. A maximum dose of 78% (range: 25–107%) and 70% (range: 52–94%) were the results for the bladder wall and rectal wall, respectively. The CI could be increased to 99% with dose point or inverse optimization, followed by graphical optimization
(Table 4) at the cost of a small increase in normal tissue irradiation (C5.3%, range: 5–5.5%) and in high dose volume (C2.4 cm3, range: 2–2.8 cm3). The maximum change in COIN due to the addition of graphical optimization was 3%, from 0.61 to 0.58, as the CI and normal tissue dose similarly increased. However, graphical optimization allowed to avoid an underdosage of the PTV as the NDR increased from 0.86 to 1.01. For all oropharyngeal implants the MTD was 26.5 Gy for the non-optimized and optimized implant. When comparing the CT-based implant to the implant based on orthogonal radiographs, the CT-based implant allowed to obtain the same CI (94%) with a decrease in normal tissue irradiation of 53% (range: 44–61%) to 40% (range: 35–50%), and almost identical high dose volume of 8.3 cm3 (range: 4–12 cm3). After dose point optimization, a better target coverage (CIZ96%) resulted with the same normal tissue irradiation (NTIZ40%) and with a limited variation of NTI range (range: 36–42%) compared to non-optimized CT-based implants. This was also true when the dose point method was compared to the inverse optimization where a CIZ90% and a NTIZ40% (range: 33–44%) were obtained. The mean value of the high dose volume was 8.7 cm3 (range: 4–15 cm3) using both optimization methods. When comparing non-optimized CT-based oropharyngeal implant to implant based on orthogonal radiographs, the V8 Gy of the horizontal ipsilateral mandible decreased from 1.2 to 0.42 cm3, without any modification of the mean dose (3.1 Gy; range: 2–4.6 Gy) over ipsilateral parotid and of the mean dose (1.13 Gy; range: 0–1.74 Gy) over contralateral one. When comparing optimized to non-optimized CT-based implants, the mean dose over the contralateral parotid decreased from 1.13 to 0.6 Gy
Table 2 The mean values and the variation range of the evaluation parameters for the non-optimized and optimized dose distribution of the irregular cervical implants
Non-optimized implant based on radiographs or CT imagesb CT-based implant Dose point optimization Inverse optimization a
Treated volume (cm3)
CIa (%)
NTIa (%)
High dose volume (cm3)
119 (range: 70–169)
88 (range: 80–96)
44 (range: 32–56)
33 (range: 14–52)
104 (range: 87–121) 108 (range: 105–111)
96 (range: 92–99) 90 (range: 89–90)
37 (range: 37–37) 44 (range: 38–50)
30 (range: 23–37) 33 (range: 27–38)
See appendix for notation. The parameter values for non-optimized implant based on radiographs or on CT images did not differ significantly so both non-optimized implant modes are combined. b
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
335
Table 3 The mean dose and the dose range, as a percentage of the prescribed dose, received by 2 and 5 cm3 of the rectal and bladder walls Dose to bladder wall (%)
Non-optimized implant based on radiographs or CT imagesa CT-based implant Dose point optimization Inverse optimization
Dose to rectal wall (%) 5 cm
2 cm3
5 cm3
82 (range: 29–118)
53 (range: 19–91)
80 (range: 57–113)
59 (range: 42–95)
69.4 (range: 26–86) 71 (range: 22–92)
46 (range: 19–72) 47 (range: 16–48)
80 (range: 51–109) 68 (range: 51–76)
58 (range: 38–87) 47 (range: 38–57)
2 cm
3
3
a
The parameter values for non-optimized implants based on radiographs or on CT images did not differ significantly so both non-optimized implant modes are combined.
optimization or geometric optimization [1,10,13,18]. A value of 90% is recommended for geometric optimization [1,18]. Our study used a value of 90% of the MCD for the dose point and the inverse optimization methods, which resulted in a good target coverage (CIZ96%), with almost the same treated volume and high dose volume, and with a significant decrease in normal tissue irradiation, compared to non-optimized CTbased implant. In the irregular implants there was no increase in the treated volume using optimization methods, so the dose prescription was kept unchanged between non-optimized and optimized implants. A similar result was shown in a publication on the optimization of an interstitial implant [10] where the dose specification in an irregular volume implant of the base of tongue was kept unchanged after geometric optimization, as the volume enclosed by the reference isodose did not increase. Currently available optimization methods, such as geometric and dose point optimization, are based only on the location of the active dwell positions. The objective of these methods is to improve the dose homogeneity over the treated volume. A publication on the relationship between ‘hot’ spots and complications following high dose rate brachytherapy [21] showed that the heterogeneity appeared to be minor with regard to complications, so the hot spots need not preclude optimization to ensure adequate dosage to all parts of the PTV. The introduction of three-dimensional images in brachytherapy planning should allow to optimize the dose distribution based on anatomic information. The objective
(range: 0–1 Gy) without any modification of the mean dose over the ipsilateral parotid.
4. Discussion 4.1. Role of CT images The dose distribution of an implant calculated using geometry based on orthogonal radiographs can be improved significantly in terms of conformation to the PTV when CT images are used. The definition of the PTV is not possible on orthogonal radiographs, so the conformality of the dose distribution to the PTV cannot be defined and the dose distribution is always calculated according to the implant geometry. 4.2. Impact of optimization methods When the dose distribution was optimized, the active length inside the catheters was adapted to the PTV length. With the dose point and geometric optimization methods, the increase in dwell times of source positions at the outer parts of the implant decreases the bending of the reference isodose between the ends of the catheters [10,13]. The inverse optimization algorithm is essentially based on the geometric optimization method, so the active length should also be adapted to the PTV length to avoid an increase in the treated volume. Several studies propose for regular implants to select a higher isodose than 85% of the MCD after dose point
Table 4 Impact of the dose point and inverse optimization followed by graphical optimization of a cervical implant
CT-based implant geometry without optimization Dose point optimization Without graphical With graphical Inverse optimization Without graphical With graphical a
See appendix for notation.
CIa (%)
NTIa (%)
High dose volume (cm3)
COINa
NDRa
90
43
9
0.52
0.98
91 99
38.5 44
9.2 12
0.56 0.55
0.86 0.99
95 99
36 41
9.3 11.3
0.61 0.58
0.86 1.01
336
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
of the optimization based on CT images can be expressed in target coverage, homogeneity of dose over the target, irradiation of normal tissue and organs at risk. The dose point optimization used in this study is always based on the active dwell positions, but the use of CT images allows us to keep the dwell positions inside the target volume and also to optimize the dose distribution using dose points placed on the target surface. The inverse optimization method optimizes the dose distribution to satisfy the dose constraints imposed on the target volume and the organs at risk. In the cervical implants, the benefit of the dose point method with dwell positions inside the target volume as defined by the CT images, was the improvement of the target coverage with a decrease in the normal tissue and bladder wall irradiation. The impact of inverse optimization was significant for rectal wall and bladder wall protection, but this reduction of irradiation on organs at risk came with a decrease in target coverage compared to dose point optimization in the irregular implants. However, target coverage, normal tissue irradiation and organs at risk protection obtained with the inverse optimization method, were clinically considered as an interesting improvement of the treatment plans. The dose constraints of the inverse optimization method are based on the implant geometry, the influence of these dose constraints on the resulting dose distribution is strongly related to the distance between the implant and each organ at risk. In the cervical implants used in this study, the rectal wall was located close to the implant, consequently, the dose constraints imposed were relatively high as a percentage of the mean target dose. The bladder wall was located near the implant in three implants, and it was located slightly farther in one implant (3 mm from the last active position longitudinally), consequently, the dose constraints imposed were again relatively high as a percentage of the mean target dose except in one implant. The resulting maximum doses in these organs satisfied the dose constraints imposed. When an implant was optimized without additional graphical optimization, the target was not completely covered by the prescription isodose. With graphical optimization the target coverage could be further improved: the CI increased, together with an increase of normal tissue irradiation. For the value of COIN, the increase in PTV coverage was compensated by the increase in normal tissue irradiation. Comparing two dose distributions where COIN has almost the same value (0.56–0.55), the NDR indicated that in one dose distribution the PTV coverage was realized at the expense of underdosage (NDRZ0.86), but in the other dose distribution this underdosage was completely avoided (NDRZ0.99). A value of NDR lower than 1.0 indicates that the prescription isodose surface lies (partly) outside the treated volume [14]. The manual adjustment of the dose distribution with graphical optimization in some CT slices or in other planes
through the implant, results in local modifications of the dose distribution of the whole implant, so a sufficient optimization is difficult to reach with this method. With graphical optimization, the whole dose distribution should be evaluated after each adjustment. Peiffert et al. [16] showed the feasibility of PDR brachytherapy for head and neck carcinomas and suggested that clinical optimization of dose distribution should be performed, taking into account the information of 3D imaging correlated with computed tomography scan slices. With this new information and the use of optimization, the occurrence of mandible osteonecrosis can be reduced. The results of our study on the CT-based oropharyngeal implants showed an improvement of the PTV coverage with a reduction of normal tissue irradiation using dose point optimization compared to non-optimized implants. Using inverse optimization, a bad PTV coverage was observed in one patient (76%) because only a small number of catheters were inside the PTV. There were no dose constraints for organs at risk in the oropharyngeal implants because of their distant location from the implant, resulting in a low dose gradient. Modification of the dose in organs distant from the target cannot be done by changing the relative dwell times of dwell positions, while keeping the prescription dose of the implant unchanged. The decrease in the horizontal ipsilateral mandible dose when using CT-based implant, compared to standard implant using orthogonal radiographs, showed the possibility of reducing the mandible osteonecrosis. Another advantage of the optimization method is the reduction of the mean dose to the contralateral parotid without increasing the dose to the ipsilateral parotid. A prospective study on salivary function sparing in patients with head and neck cancers receiving intensity-modulated or three-dimensional radiation therapy showed that the sparing of the parotid glands translates into improvement of both xerostomia and quality-of-life scores [5].
5. Conclusions This work shows that the optimized cervical and oropharyngeal CT-based implants offer interesting possibilities to further adapt the dose distribution to the target volume while sparing normal tissue and organs at risk. Inverse optimization can be used in clinical practice for sparing the organs at risk, without the problem of placement of dose points as in dose point optimization. The evaluation indices, COIN and NDR, should be used together because the prescription dose can always be selected to cover the PTV but it is also essential to evaluate the dose to normal tissue and organs at risk, and the matching of the treated volume with the PTV.
F. Ahmad et al. / Radiotherapy and Oncology 73 (2004) 331–337
Appendix A. Abbreviations and notation CDVH: the cumulative dose–volume histograms are representations of the dose–volume relation for each organ; COIN: the conformal index takes into consideration the coverage of the PTV by the reference dose and also the unwanted irradiation of normal tissue outside the PTV [3]; CI: the coverage index is the fraction of the PTV receiving a dose equal to or greater than the prescribed dose [13]; MCD: the mean central dose is taken to be the arithmetic mean of the local minimum doses between sources, in the central plane [9]; MTD: the minimum target dose is the reference dose (85% of the MCD) under the Paris system [9]; NTI: the normal tissue irradiation is the fraction of the normal tissue outside the PTV receiving the prescribed dose; NDVH: the natural dose volume histogram is a representation of the dose–volume relation for the dose distribution with suppression of the inverse square law effect [2]; NPD: the natural prescription dose, located at the base of the NDVH peak, is the optimal prescription dose according to the treated volume [14]; NDR: the natural dose ratio is derived from the NDVH of the implant and the CDVH of the PTV. It is defined as the ratio of the NPD and the PD (NDRZNPD/ PD) [14]; PD: the prescription dose; V100: the treated volume is the volume of the prescribed dose; VPTV,PD: the PTV volume covered by the prescribed dose; V8 Gy: The volume receiving at least 8 Gy.
References [1] Anacak Y, Esassolak M, Aydin A, Aras A, Olacak I, Haydaroglu A. Effect of geometrical optimization on the treatment volumes and the dose homogeneity of biplane interstitial brachytherapy implants. Radiother Oncol 1997;45:71–6. [2] Anderson LL. A ‘natural’ volume–dose histogram for brachytherapy. Med Phys 1986;13:898–903. [3] Baltas D, Kolotas C, Geramani K, et al. A conformal index (COIN) to evaluate implant quality and dose specification in brachytherapy. Int J Radiat Oncol Biol Phys 1998;40:515–24. [4] Brenner DJ, Hall EJ. Conditions for the equivalence of continuous to pulsed low dose rate brachytherapy. Int J Radiat Oncol Biol Phys 1991;20:181–90. [5] Chao KS, Deasy JO, Markman J, et al. A prospective study of salivary function sparing in patients with head-and-neck cancers receiving intensity-modulated or three-dimensional radiation therapy: initial results. Int J Radiat Oncol Biol Phys 2001;49:907–16. [6] Charra-Brunaud C, Hsu IC, Weinberg V, Pouliot J. Analysis of interaction between number of implant catheters and dose–volume histograms in prostate high- dose-rate brachytherapy using a computer model. Int J Radiat Oncol Biol Phys 2003;56:586–91.
337
[7] Dutreix A, Marinello G. The Paris system. In: Pierquin B, Wilson JF, Chassagne D, editors. Modern brachytherapy. New York: Masson; 1987. p. 25–42. [8] Fellner C, Po¨tter R, Knocke TH, Wambersie A. Comparison of radiography- and computed tomography-based treatment planning in cervix cancer in brachytherapy with specific attention to some quality assurance aspects. Radiother Oncol 2001;58:53–62. [9] ICRU. Dose and volume specification for reporting interstitial therapy. ICRU Report 58. Bethesda, USA: ICRU; 1997. [10] Kolkman-Deurloo IK, Visser AG, Niel CG, Driver N, Levendag PC. Optimization of interstitial volume implants. Radiother Oncol 1994; 31:229–39. [11] Lahanas M, Baltas D, Zamboglou N. Anatomy-based three-dimensional dose optimization in brachytherapy using multiobjective genetic algorithms. Med Phys 1999;26:1904–18. [12] Lessard E, Hsu IC, Pouliot J. Inverse planning for interstitial gynecologic template brachytherapy: truly anatomy-based planning. Int J Radiat Oncol Biol Phys 2002;54:1243–51. [13] Major T, Polgar C, Fodor J, Somogyi A, Nemeth G. Conformality and homogeneity of dose distributions in interstitial implants at idealized target volumes: a comparison between the Paris and dose-point optimized systems. Radiother Oncol 2002;62:103–11. [14] Moerland MA, Van der Laarse R, Luthmann RW, Wijrdeman HK, Battermann JJ. The combined use of the natural and the cumulative dose–volume histograms in planning and evaluation of permanent prostatic seed implants. Radiother Oncol 2000;57:279–84. [15] Nath R, Anderson LL, Luxton G, Weaver KA, Williamson JF, Meigooni AS. Dosimetry of interstitial brachytherapy sources: recommendations of the AAPM Radiation Therapy Committee Task Group No. 43, American Association of Physicists in Medicine. Med Phys 1995;22:209–34. [16] Peiffert D, Castelain B, Thomas L, Ardiet JM, Baillet F, Mazeron JJ. Pulsed dose rate brachytherapy in head and neck cancers. Feasibility study of a French cooperative group. Radiother Oncol 2001;58:71–5. [17] Pernot M, Luporsi E, Hoffstetter S, et al. Complications following definitive irradiation for cancers of the oral cavity and the oropharynx (in a series of 1134 patients). Int J Radiat Oncol Biol Phys 1997;37: 577–85. [18] Pieters BR, Saarnak AE, Steggerda MJ, Borger JH. A method to improve the dose distribution of interstitial breast implants using geometrically optimized stepping source techniques and dose normalization. Radiother Oncol 2001;58:63–70. [19] Saw CB, Suntharalingam N. Quantitative assessment of interstitial implants. Int J Radiat Oncol Biol Phys 1991;20:135–9. [20] Van der Laarse R, Luthmann RW. Computers in brachytherapy dosimetry. In: Joslin CAF, Flynn A, Hall E, editors. Principles and practice of brachytherapy using afterloading systems. London: Arnold; 2001. p. 49–80. [21] Vikram B, Deore S, Beitler JJ, et al. The relationship between dose heterogeneity (‘hot’ spots) and complications following high-dose rate brachytherapy. Int J Radiat Oncol Biol Phys 1999;43:983–7. [22] Wachter-Gerstner N, Wachter S, Reinstadler E, et al. Bladder and rectum dose defined from MRI based treatment planning for cervix cancer brachytherapy: comparison of dose–volume histograms for organ contours and organ wall, comparison with ICRU rectum and bladder reference point. Radiother Oncol 2003;68:269–76.