- Email: [email protected]
Clinical experience of a compensator-based intensity modulated treatment technique
Proceedings of the 45th Annual ASTRO Meeting
the dosimetric analysis on the prostate and organs at risk (OARs). A comparison is made between the real and ideal cases based on the DVHs and differences (⌬ ⫽ X(3D) – X(I)) in relevant dosimetric quantities (⌬D10, ⌬D50, ⌬D90, ⌬V100 and ⌬V200) obtained. The dose values are given in [Gy] and the volumes in [%]. Results: The angle (measured from the axis of implantation on the coronal plane) has a narrow distribution centered on 0° with 30% of the sources within ⫾2,5° (⬍⬎ ⫽ 1,11°, ⫽ 22,9°). The q angle (measured from the axis of implantation on the sagittal plane) as on the other hand a large distribution centered on 0° (⬍⬎ ⫽ 1,11°, ⫽ 22,9°) in opposition to the hypothesis of a uniform distribution of sources used by other publications in the field. Cumulative DVHs or dosimetric quantities show no significant differences for the dose to the prostate (⬍⌬D90⬎ ⫽ 0,87, ⫽ 3,21; ⬍⌬V100⬎ ⫽ -0,25, ⫽ 1,26; ⬍⌬V200⬎ ⫽ 0,23, ⫽ 1,03 ). Differential DVHs show a difference between the real and ideal case: the volume receiving less than 100% of the prescribed dose is on average higher in the ideal situation (⬍⌬D90⬎ ⫽ – 0,25, ⫽ 1,27). When the OARs were considered, the dosimetric quantities to investigate were different than those of the prostate. The results show that there was a significant variation in some instances: rectum (⬍⌬D10⬎ ⫽ – 0,85, ⫽ 4,75), bladder (⬍⌬D10⬎ ⫽1,76, ⫽ 2,44) and penile bulb (⬍⌬D10⬎ ⫽ 1,75, ⫽ 3,40;⬍⌬D50⬎ ⫽ 1,88, ⫽ 2,64). Conclusions: The real distribution of 1625 individual sources was obtained revealing that the f angle distribution is strongly peaked while the q angle distribution is wider. It also shows that the difference between an ideally aligned situation and the real one does not affect significantly the cumulative or differential DVHs of the prostate or rectum. For the bladder and the penile bulb, the ideal situation gives a lower dose. That is attributed to the anisotropy of the Amersham 6711 seed model (which yields a lower dose at its tips than on the side) and the position of those organs relative to the implantation axis. This lower dose should be taken into account when reporting dose effects to these OARs in post-implant evaluation (compared to ideal source configuration or point source configuration).
2150
Impact of MLC Leaf Width Reduction on Normal Tissue Dose in the Treatment of Infratentorial Ependymoma
M.R. Sontag, R. Siochi, Y. Zhu, S.S. Samant, B. Crawford, X. Ying, T.E. Merchant St. Jude Children’s Research Hospital, Memphis, TN Purpose/Objective: The benefit of MLC leaf width diminution usually is based on inductive reasoning in which it is assumed that better conformity to the target volume will reduce irradiation of surrounding tissues. Decrease in beam edge scalloping as measured by reduction of the effective penumbral width has been demonstrated. Neither approach ascertains whether this decrease is clinically important. A study has been undertaken to determine the dosimetric benefit of two approaches to mitigate the effect of MLC width. Materials/Methods: 20 pediatric patients were selected from a group of 67 patients with infratentoral ependymoma that were treated at our institution with conformal 3DRTP on either a Siemens Primart (6MV x-rays) or Siemens Primus (6MV and/or 15MV x-rays), each having identical doubly focused 1cm leaf width MLCs. All patients were planned using CT and MR for target and normal tissue delineation. Dose was calculated using a 0.2 cm grid. Measurements were made using Kodak verification film to ensure correct dosimetric description of penumbra in the dose model. Patients received 54 Gy (8.76 ⫾ 2.86 beams) to the PTV followed by a 5.4 Gy (4.95 ⫾ 3.01 beams) off-cord boost. All beams were shaped using MLC and approximately half of the beams employed virtual wedges. Two additional plans were developed for each patient employing either a 0.43 cm leaf width Radionics mini MLC (mMLC) or the Siemens HD270 technique, which uses the 1 cm wide MLC and a set of 3 slightly shifted beams which produces smearing of the beam edge. The new plans differed from the original MLC plan only in actual beam shape. Beam energy, orientation and wedging were identical. For each plan, comparable PTV dose was obtained as ascertained by DVH analysis. Dose-volume histograms and integral dose were calculated for target volumes and pertinent critical structures for each of the three (MLC, mMLC and HD270) techniques. Results: Measurement by film finds that the 80-20% penumbral width of the Siemens MLC and the Radionics mMLC leaf edge differs ⬍1 mm. Replacement of the MLC with mMLC resulted in integral dose reductions for total brain (5.1 ⫾ 3.7%; mean ⫾ SD), left temporal lobe (6.3 ⫾ 5.5%) and right temporal lobe (5.8 ⫾ 5.5%). Greater integral dose reductions were observed for left cochlea (13.7 ⫾ 8.9%), right cochlea (14.1 ⫾ 12.2%), chiasm (20.8 ⫾ 21.4%), hypothalamus (18.7 ⫾ 25.2%) and pituitary (18.2 ⫾ 21.4%). Replacement of MLC with the HD270 technique resulted in integral dose reductions in total brain (0.5 ⫾ 1.0%), left temporal lobe (1.2 ⫾ 0.9%), right temporal lobe (1.0 ⫾ 1.0%), chiasm (2.8 ⫾ 3.9%), hypothalamus (3.7 ⫾ 5.1%) and pituitary (0.8 ⫾ 3.4%). Integral dose increases were found for left cochlea (0.2 ⫾ 3.3%) and right cochlea (0.4 ⫾ 5.9%). Conclusions: The closer distance to the patient of the mMLC with its non divergent leaf edge offsets the advantage of the divergent leaf edge of the MLC, resulting in similar penumbras. The 5-6% reduction in brain dose obtained utilizing mMLC instead of MLC is attributable to reduction in beam scalloping obtained with the former’s smaller leaf width. More considerable dose reductions using the mMLC occurred for the chiasm, auditory and hypothalamic-pituitary unit because these structures are adjacent to the PTV and benefit from the more precise shaping afforded by the narrower mMLC leaf width. There is some modest additional reduction due to the diminution of beam scalloping. The large standard deviation in the reported dose reductions resulted from inter-patient variability of organ location relative to the PTV. In the HD270 technique, only the leaves of the original beam were placed based the location of critical structures as well as the PTV. Since shifted beam leaf positions were based only on the leaf positions of the original beam, critical structures lying adjacent to PTV were randomly further shielded or further exposed to the shifted beams. The small systematic dose reductions found were due to decrease in beam scalloping alone. While use of the mMLC resulted in meaningful dose reduction, this was not found for HD270.
2151
Clinical Experience of a Compensator-based Intensity Modulated Treatment Technique
S. Chang, T. Cullip, K. Deschesne, J. Rosenman Radiation Oncology, University of North Carolina at Chapel Hill, Chapel Hill, NC Purpose/Objective: Compensators offer an alternative means to deliver intensity-modulated radiotherapy (IMRT). Compared to the multi-leaf collimator (MLC)-based IMRT delivery techniques the drawback of compensator-based IMRT techniques is the lack of automation. The advantages of the compensator-based IMRT techniques, however, are less understood. We have implemented a compensator-based IMRT delivery technique in our clinic since 1996 and have treated more than seven hundred patients to date. The
S431
S432
I. J. Radiation Oncology
● Biology ● Physics
Volume 57, Number 2, Supplement, 2003
purpose of this presentation is to report our recent retrospective analysis on treatment dosimetric quality and treatment delivery efficiency of the compensator-based IMRT technique, in comparison with the segmental MLC-based IMRT treatments. Materials/Methods: The ideal intensity modulated treatment plans are generated by the in-house treatment planning system (TPS) PLanUNC using the index-dose gradient minimization algorithm. The optimization goals are user-defined via dose volume histogram (DVH) for each structure of interest. The resultant continuous intensity maps are converted into a compensator thickness file for compensator mold fabrication. Styrofoam compensator molds are milled by the Par Scientific milling machine, filled with tin granules, and inserted into durable plastic boxes for easy use by therapists in daily treatment. We have implemented the segmental MLC-IMRT technique for clinical use since 1998 and have been used both IMRT techniques routinely. The segmental MLC-IMRT treatment is created by IMFAST segmentation software of Siemens using discrete intensity maps truncated from the original continuous intensity maps. We compared the compensator-based and the segmental MLC-based IMRT treatments in terms of dosimetric quality and the treatment delivery efficiency. Exemplary clinical cases will be shown to demonstrate the dosimetric quality differences among the ideal treatment directly generated from the dose optimization, the compensator- based IMRT treatment, and the segmental MLC-based IMRT treatment. Treatment efficiency for 94 patients is retrospectively analyzed using the actual patient treatment delivery information from the LANTIS record and verify system. The statistics of the compensator-IMRT QA results for 340 patients will also be presented. Results: We find the DVHs of the compensator-based IMRT treatments were consistently closer to those of the ideal IMRT treatment comparing to the corresponding segmental MLC-based IMRT treatments. We attribute this finding to the better intensity modulation resolution of the compensator technique. However, the medium effective density tin granule compensators have the drawback of not being able to provide the magnitude of intensity modulation needed for certain cases, thus deteriorating the dosimetry. In fact, lack of intensity modulation magnitude is the major reason why the MLC technique is chosen over the compensator technique in our clinical application. The retrospective analysis of more than one hundred patients who received IMRT treatments of 2-7 fields showed the treatment delivery time for the compensator treatments is considerably less than for the MLC technique. Conclusions: Our years of experience in routine clinical application demonstrated that the compensator-based IMRT technique is a valuable alternative for IMRT treatment delivery. The lack of automation did not lead to decreased treatment efficiency for the compensator-based technique. We have shown that for many clinical cases the compensator-based IMRT technique offered a better treatment dosimetry and better treatment efficiency comapred to the segmental MLC-based IMRT technique. To achieve the large intensity modulation magnitude required by many clinical cases high-density compensator materials should be used.
2152
SIB-II: Improved Parotid Gland Sparing Using A Two-phase Planning Strategy for Head and Neck Intensity Modulated Radiotherapy (IMRT)
A.J. Crimaldi, Y. Wu, O. Abayomi, M. Morris, A. Lauve, J. Zhou, Q. Wu, R. Schmidt-Ullrich Radiation Oncology, Virginia Commonwealth University-MCV Hospital, Richmond, VA Purpose/Objective: To demonstrate improved parotid gland sparing using a two-phase IMRT plan to treat advanced head and neck squamous cell carcinoma (HNSCC). Background: An IMRT Simultaneous Integrated Boost (SIB) technique is used to treat HNSCC at our institution. Gross Tumor Volume (GTV), Clinical Target Volume (CTV) and Electively Treated nodal Volume (ETV) receive 70.8 Gy, 60 Gy and 54 Gy, respectively, in 30 equal fractions over 40 days. This technique effectively reduces the dose and the dose per fraction to organs at risk, including parotid glands. However, clinical xerostomia is still observed, and sparing of parotid glands is often limited by the need to deliver adequate doses to nearby electively treated lymph nodes. Materials/Methods: Ten patients treated using SIB-IMRT on a clinical protocol were re-planned using a two-phase (SIB II) approach. In the first 25 fractions, Phase 1, the ETV received 50 Gy in 2 Gy fractions (biologically equivalent to 54 Gy in 30 fractions). In the remaining 5 fractions, Phase 2, the ETV was maximally spared. Prescription doses to GTV and CTV were unchanged, as were dosimetric guidelines calling for 99%, 95%, and 90% coverage for GTV, CTV and ETV, respectively, and constraints of 45 Gy for cord, 55 Gy for brainstem, and ⬍10% for GTV dose homogeneity index were retained. A composite SIB-II plan was derived and compared to the original plan for each patient, including biologically equivalent dose (BED) volume histograms, mean doses (Dmean), equivalent uniform doses (EUD), and Normal Tissue Complication Probabilities (NTCP). Results: Composite SIB-II dose distributions for GTV and CTV were comparable to those of the SIB plan. ETV doses were biologically equivalent. Doses to other critical structures were variably improved. Although the fractional dose to the parotids was higher in the first phase of SIB-II than in SIB, the composite total dose and the BEDs were lower in SIB-II. The BED mean dose to the total and distant parotid glands averaged 5.3% and 7.4% lower, respectively. Because these doses are on the steep portion of the dose response curve, this modest mean dose reduction translates into a NTCP reduction of 26.1% and 40.0%, respectively. Conclusions: Two-phase IMRT planning significantly improves the sparing of parotid glands over the current single-plan SIB technique used in the treatment of HNSCC. The resulting NTCP reduction suggests a potential decrease in xerostomia using this treatment method.
the dosimetric analysis on the prostate and organs at risk (OARs). A comparison is made between the real and ideal cases based on the DVHs and differences (⌬ ⫽ X(3D) – X(I)) in relevant dosimetric quantities (⌬D10, ⌬D50, ⌬D90, ⌬V100 and ⌬V200) obtained. The dose values are given in [Gy] and the volumes in [%]. Results: The angle (measured from the axis of implantation on the coronal plane) has a narrow distribution centered on 0° with 30% of the sources within ⫾2,5° (⬍⬎ ⫽ 1,11°, ⫽ 22,9°). The q angle (measured from the axis of implantation on the sagittal plane) as on the other hand a large distribution centered on 0° (⬍⬎ ⫽ 1,11°, ⫽ 22,9°) in opposition to the hypothesis of a uniform distribution of sources used by other publications in the field. Cumulative DVHs or dosimetric quantities show no significant differences for the dose to the prostate (⬍⌬D90⬎ ⫽ 0,87, ⫽ 3,21; ⬍⌬V100⬎ ⫽ -0,25, ⫽ 1,26; ⬍⌬V200⬎ ⫽ 0,23, ⫽ 1,03 ). Differential DVHs show a difference between the real and ideal case: the volume receiving less than 100% of the prescribed dose is on average higher in the ideal situation (⬍⌬D90⬎ ⫽ – 0,25, ⫽ 1,27). When the OARs were considered, the dosimetric quantities to investigate were different than those of the prostate. The results show that there was a significant variation in some instances: rectum (⬍⌬D10⬎ ⫽ – 0,85, ⫽ 4,75), bladder (⬍⌬D10⬎ ⫽1,76, ⫽ 2,44) and penile bulb (⬍⌬D10⬎ ⫽ 1,75, ⫽ 3,40;⬍⌬D50⬎ ⫽ 1,88, ⫽ 2,64). Conclusions: The real distribution of 1625 individual sources was obtained revealing that the f angle distribution is strongly peaked while the q angle distribution is wider. It also shows that the difference between an ideally aligned situation and the real one does not affect significantly the cumulative or differential DVHs of the prostate or rectum. For the bladder and the penile bulb, the ideal situation gives a lower dose. That is attributed to the anisotropy of the Amersham 6711 seed model (which yields a lower dose at its tips than on the side) and the position of those organs relative to the implantation axis. This lower dose should be taken into account when reporting dose effects to these OARs in post-implant evaluation (compared to ideal source configuration or point source configuration).
2150
Impact of MLC Leaf Width Reduction on Normal Tissue Dose in the Treatment of Infratentorial Ependymoma
M.R. Sontag, R. Siochi, Y. Zhu, S.S. Samant, B. Crawford, X. Ying, T.E. Merchant St. Jude Children’s Research Hospital, Memphis, TN Purpose/Objective: The benefit of MLC leaf width diminution usually is based on inductive reasoning in which it is assumed that better conformity to the target volume will reduce irradiation of surrounding tissues. Decrease in beam edge scalloping as measured by reduction of the effective penumbral width has been demonstrated. Neither approach ascertains whether this decrease is clinically important. A study has been undertaken to determine the dosimetric benefit of two approaches to mitigate the effect of MLC width. Materials/Methods: 20 pediatric patients were selected from a group of 67 patients with infratentoral ependymoma that were treated at our institution with conformal 3DRTP on either a Siemens Primart (6MV x-rays) or Siemens Primus (6MV and/or 15MV x-rays), each having identical doubly focused 1cm leaf width MLCs. All patients were planned using CT and MR for target and normal tissue delineation. Dose was calculated using a 0.2 cm grid. Measurements were made using Kodak verification film to ensure correct dosimetric description of penumbra in the dose model. Patients received 54 Gy (8.76 ⫾ 2.86 beams) to the PTV followed by a 5.4 Gy (4.95 ⫾ 3.01 beams) off-cord boost. All beams were shaped using MLC and approximately half of the beams employed virtual wedges. Two additional plans were developed for each patient employing either a 0.43 cm leaf width Radionics mini MLC (mMLC) or the Siemens HD270 technique, which uses the 1 cm wide MLC and a set of 3 slightly shifted beams which produces smearing of the beam edge. The new plans differed from the original MLC plan only in actual beam shape. Beam energy, orientation and wedging were identical. For each plan, comparable PTV dose was obtained as ascertained by DVH analysis. Dose-volume histograms and integral dose were calculated for target volumes and pertinent critical structures for each of the three (MLC, mMLC and HD270) techniques. Results: Measurement by film finds that the 80-20% penumbral width of the Siemens MLC and the Radionics mMLC leaf edge differs ⬍1 mm. Replacement of the MLC with mMLC resulted in integral dose reductions for total brain (5.1 ⫾ 3.7%; mean ⫾ SD), left temporal lobe (6.3 ⫾ 5.5%) and right temporal lobe (5.8 ⫾ 5.5%). Greater integral dose reductions were observed for left cochlea (13.7 ⫾ 8.9%), right cochlea (14.1 ⫾ 12.2%), chiasm (20.8 ⫾ 21.4%), hypothalamus (18.7 ⫾ 25.2%) and pituitary (18.2 ⫾ 21.4%). Replacement of MLC with the HD270 technique resulted in integral dose reductions in total brain (0.5 ⫾ 1.0%), left temporal lobe (1.2 ⫾ 0.9%), right temporal lobe (1.0 ⫾ 1.0%), chiasm (2.8 ⫾ 3.9%), hypothalamus (3.7 ⫾ 5.1%) and pituitary (0.8 ⫾ 3.4%). Integral dose increases were found for left cochlea (0.2 ⫾ 3.3%) and right cochlea (0.4 ⫾ 5.9%). Conclusions: The closer distance to the patient of the mMLC with its non divergent leaf edge offsets the advantage of the divergent leaf edge of the MLC, resulting in similar penumbras. The 5-6% reduction in brain dose obtained utilizing mMLC instead of MLC is attributable to reduction in beam scalloping obtained with the former’s smaller leaf width. More considerable dose reductions using the mMLC occurred for the chiasm, auditory and hypothalamic-pituitary unit because these structures are adjacent to the PTV and benefit from the more precise shaping afforded by the narrower mMLC leaf width. There is some modest additional reduction due to the diminution of beam scalloping. The large standard deviation in the reported dose reductions resulted from inter-patient variability of organ location relative to the PTV. In the HD270 technique, only the leaves of the original beam were placed based the location of critical structures as well as the PTV. Since shifted beam leaf positions were based only on the leaf positions of the original beam, critical structures lying adjacent to PTV were randomly further shielded or further exposed to the shifted beams. The small systematic dose reductions found were due to decrease in beam scalloping alone. While use of the mMLC resulted in meaningful dose reduction, this was not found for HD270.
2151
Clinical Experience of a Compensator-based Intensity Modulated Treatment Technique
S. Chang, T. Cullip, K. Deschesne, J. Rosenman Radiation Oncology, University of North Carolina at Chapel Hill, Chapel Hill, NC Purpose/Objective: Compensators offer an alternative means to deliver intensity-modulated radiotherapy (IMRT). Compared to the multi-leaf collimator (MLC)-based IMRT delivery techniques the drawback of compensator-based IMRT techniques is the lack of automation. The advantages of the compensator-based IMRT techniques, however, are less understood. We have implemented a compensator-based IMRT delivery technique in our clinic since 1996 and have treated more than seven hundred patients to date. The
S431
S432
I. J. Radiation Oncology
● Biology ● Physics
Volume 57, Number 2, Supplement, 2003
purpose of this presentation is to report our recent retrospective analysis on treatment dosimetric quality and treatment delivery efficiency of the compensator-based IMRT technique, in comparison with the segmental MLC-based IMRT treatments. Materials/Methods: The ideal intensity modulated treatment plans are generated by the in-house treatment planning system (TPS) PLanUNC using the index-dose gradient minimization algorithm. The optimization goals are user-defined via dose volume histogram (DVH) for each structure of interest. The resultant continuous intensity maps are converted into a compensator thickness file for compensator mold fabrication. Styrofoam compensator molds are milled by the Par Scientific milling machine, filled with tin granules, and inserted into durable plastic boxes for easy use by therapists in daily treatment. We have implemented the segmental MLC-IMRT technique for clinical use since 1998 and have been used both IMRT techniques routinely. The segmental MLC-IMRT treatment is created by IMFAST segmentation software of Siemens using discrete intensity maps truncated from the original continuous intensity maps. We compared the compensator-based and the segmental MLC-based IMRT treatments in terms of dosimetric quality and the treatment delivery efficiency. Exemplary clinical cases will be shown to demonstrate the dosimetric quality differences among the ideal treatment directly generated from the dose optimization, the compensator- based IMRT treatment, and the segmental MLC-based IMRT treatment. Treatment efficiency for 94 patients is retrospectively analyzed using the actual patient treatment delivery information from the LANTIS record and verify system. The statistics of the compensator-IMRT QA results for 340 patients will also be presented. Results: We find the DVHs of the compensator-based IMRT treatments were consistently closer to those of the ideal IMRT treatment comparing to the corresponding segmental MLC-based IMRT treatments. We attribute this finding to the better intensity modulation resolution of the compensator technique. However, the medium effective density tin granule compensators have the drawback of not being able to provide the magnitude of intensity modulation needed for certain cases, thus deteriorating the dosimetry. In fact, lack of intensity modulation magnitude is the major reason why the MLC technique is chosen over the compensator technique in our clinical application. The retrospective analysis of more than one hundred patients who received IMRT treatments of 2-7 fields showed the treatment delivery time for the compensator treatments is considerably less than for the MLC technique. Conclusions: Our years of experience in routine clinical application demonstrated that the compensator-based IMRT technique is a valuable alternative for IMRT treatment delivery. The lack of automation did not lead to decreased treatment efficiency for the compensator-based technique. We have shown that for many clinical cases the compensator-based IMRT technique offered a better treatment dosimetry and better treatment efficiency comapred to the segmental MLC-based IMRT technique. To achieve the large intensity modulation magnitude required by many clinical cases high-density compensator materials should be used.
2152
SIB-II: Improved Parotid Gland Sparing Using A Two-phase Planning Strategy for Head and Neck Intensity Modulated Radiotherapy (IMRT)
A.J. Crimaldi, Y. Wu, O. Abayomi, M. Morris, A. Lauve, J. Zhou, Q. Wu, R. Schmidt-Ullrich Radiation Oncology, Virginia Commonwealth University-MCV Hospital, Richmond, VA Purpose/Objective: To demonstrate improved parotid gland sparing using a two-phase IMRT plan to treat advanced head and neck squamous cell carcinoma (HNSCC). Background: An IMRT Simultaneous Integrated Boost (SIB) technique is used to treat HNSCC at our institution. Gross Tumor Volume (GTV), Clinical Target Volume (CTV) and Electively Treated nodal Volume (ETV) receive 70.8 Gy, 60 Gy and 54 Gy, respectively, in 30 equal fractions over 40 days. This technique effectively reduces the dose and the dose per fraction to organs at risk, including parotid glands. However, clinical xerostomia is still observed, and sparing of parotid glands is often limited by the need to deliver adequate doses to nearby electively treated lymph nodes. Materials/Methods: Ten patients treated using SIB-IMRT on a clinical protocol were re-planned using a two-phase (SIB II) approach. In the first 25 fractions, Phase 1, the ETV received 50 Gy in 2 Gy fractions (biologically equivalent to 54 Gy in 30 fractions). In the remaining 5 fractions, Phase 2, the ETV was maximally spared. Prescription doses to GTV and CTV were unchanged, as were dosimetric guidelines calling for 99%, 95%, and 90% coverage for GTV, CTV and ETV, respectively, and constraints of 45 Gy for cord, 55 Gy for brainstem, and ⬍10% for GTV dose homogeneity index were retained. A composite SIB-II plan was derived and compared to the original plan for each patient, including biologically equivalent dose (BED) volume histograms, mean doses (Dmean), equivalent uniform doses (EUD), and Normal Tissue Complication Probabilities (NTCP). Results: Composite SIB-II dose distributions for GTV and CTV were comparable to those of the SIB plan. ETV doses were biologically equivalent. Doses to other critical structures were variably improved. Although the fractional dose to the parotids was higher in the first phase of SIB-II than in SIB, the composite total dose and the BEDs were lower in SIB-II. The BED mean dose to the total and distant parotid glands averaged 5.3% and 7.4% lower, respectively. Because these doses are on the steep portion of the dose response curve, this modest mean dose reduction translates into a NTCP reduction of 26.1% and 40.0%, respectively. Conclusions: Two-phase IMRT planning significantly improves the sparing of parotid glands over the current single-plan SIB technique used in the treatment of HNSCC. The resulting NTCP reduction suggests a potential decrease in xerostomia using this treatment method.