Physics contribution
Monte Carlo–based dosimetry of head-and-neck patients treated with SIB-IMRT

https://doi.org/10.1016/j.ijrobp.2005.09.049Get rights and content

Purpose: To evaluate the accuracy of previously reported superposition/convolution (SC) dosimetric results by comparing with Monte Carlo (MC) dose calculations for head-and-neck intensity-modulated radiation therapy (IMRT) patients treated with the simultaneous integrated boost technique.

Methods and Materials: Thirty-one plans from 24 patients previously treated on a phase I/II head-and-neck squamous cell carcinoma simultaneous integrated boost IMRT protocol were used. Clinical dose distributions, computed with an SC algorithm, were recomputed using an EGS4-based MC algorithm. Phantom-based dosimetry quantified the fluence prediction accuracy of each algorithm. Dose–volume indices were used to compare patient dose distributions.

Results and Discussion: The MC algorithm predicts flat-phantom measurements better than the SC algorithm. Average patient dose indices agreed within 2.5% of the local dose for targets; 5.0% for parotids; and 1.9% for cord and brainstem. However, only 1 of 31 plans agreed within 3% for all indices; 4 of 31 agreed within 5%. In terms of the prescription dose, 4 of 31 plans agreed within 3% for all indices, whereas 28 of 31 agreed within 5%.

Conclusions: Average SC-computed doses agreed with MC results in the patient geometry; however deviations >5% were common. The fluence modulation prediction is likely the major source of the dose discrepancy. The observed dose deviations can impact dose escalation protocols, because they would result in shifting patients to higher dose levels.

Introduction

Intensity-modulated radiation therapy (IMRT) has taken on a significant role in radiotherapy because of its ability to deliver radiation conformally around structures in a highly heterogeneous medium. Head-and-neck (HN) cancer patients are well suited to benefit from IMRT’s conformality, because the tumor volumes are often irregularly shaped and are directly adjacent to many critical structures. Studies have shown that compared with 3D conformal planning, IMRT is not only more conformal, but it results in a better dose homogeneity and a sharper dose gradient at the boundaries (1, 2, 3, 4, 5, 6). Furthermore, IMRT planning techniques such as simultaneous integrated boost (SIB) can increase the conformality of treatment, shorten the overall treatment course, and limit the growth of aggressive diseases such as HN squamous cell carcinoma (6, 7, 8, 9, 10, 11).

Several studies document the dosimetric benefits of IMRT combined with SIB techniques (12, 13). Previously, our institution reported on the dosimetric analysis of a group of HN patients treated with SIB-IMRT (11, 13). The dosimetric calculations in that report were based upon superposition/convolution (SC) algorithms with the intensity modulation incorporated into the dose calculation via a fluence transmission matrix. This method of including intensity modulation is similar to that implemented in commercial treatment planning systems (TPS) (14, 15, 16, 17), although, instead of SC dose computation algorithms, pencil beam–based dose computation algorithms are more commonly used (18, 19, 20). Although beam-by-beam film-based quality assurance in a homogeneous phantom agreed with predictions using commonly accepted criteria in our previous study, discrepancies between the planned and delivered dose distributions were observed. However, the clinical impacts of these dose differences were not evaluated.

Intensity-modulated radiation therapy treatment fields, and SIB-IMRT treatment fields in particular, often have large intensity fluctuations that result in complex multileaf collimator (MLC) patterns and present challenges to dose calculation algorithms, because of the effects of radiation transmitted through and scattered from the MLC (21). A possible way to improve dose calculation accuracy is to use an algorithm such as Monte Carlo (MC), which can explicitly account for MLC scatter and leakage radiation during the dose calculation simulation. When compared to measurements, MC dose algorithms provide more accurate estimate of a dose delivered to a patient and, consequently, have been used to evaluate dose calculation accuracy (22, 23, 24, 25, 26, 27, 28, 29).

Wang et al. (29) investigated the use of MC to evaluate dosimetric effects of inhomogeneities for 5 clinical HN IMRT plans and 5 lung IMRT plans. A measurement-based pencil beam algorithm with an equivalent path length inhomogeneity correction was compared with MC, with the fluence modulation incorporated into the MC using the treatment planning system’s effective MLC transmission matrix. Although the majority of the dose indices in this study agreed well, 2 of the 10 plans showed dose indices with deviations >5%.

Leal et al. (24) studied the use of MC for IMRT verification. Step-and-shoot IMRT plans were calculated using the Plato treatment planning system (Veenendall, Netherlands) and compared with EGS4-based MC calculations and film dosimetry for 3 cases. Major differences were found in heterogeneous situations such as head and neck.

Francescon et al. (22) compared step-and-shoot IMRT dose distributions calculated using the Pinnacle3 TPS (Philips Medical Systems, Milpitas, CA) collapsed cone convolution algorithm with EGS4-based MC calculations for two plans. In the MC algorithms, particles were simulated through the MLC using the BEAM (30) MLC component module, which simplifies the MLC leaf geometry by ignoring MLC interleaf transmission. Isocenter point dose differences of 2.1% for a prostate plan and 2.9% for a head-and-neck plan were observed, and dose–volume histogram (DVH) data comparisons showed deviations of up to 6% for doses below 85% of the prescribed dose and higher deviations for doses above 85% of the prescription dose.

A study by Ma et al. (26) compared IMRT plans from Corvus TPS (Corvus, Nomos Corp., Sewickley, PA) with MC for two patient plans. In the MC simulation, an independently developed effective MLC fluence transmission matrix was used to incorporate intensity modulation into the MC dose calculation. Leaf leakage radiation was included as an empiric correction term to the model. Corvus’s finite-size pencil beam algorithm–calculated doses were found to have discrepancies of >5% for target structures and over 20% for critical structures, compared with the MC results.

The goal of this study was to evaluate the accuracy of previously reported SIB-IMRT dose distributions (11, 13) that were originally computed using SC by comparing results with detailed MC dose computations. To ensure accurate characterization of the radiation transmitted through the MLC during IMRT delivery, the intensity modulation was incorporated into the MC dose calculation using an algorithm that tracked particles through the moving MLC leaves (27). As a result, the MC fully simulated the effect of the MLC on treatment delivery and on radiation transport through the patient. The results of this paper will allow quantitative assessment of the dosimetric accuracy of previously reported SIB-IMRT treatments.

Section snippets

Methods and materials

Thirty-one SIB-IMRT plans from 24 different patients who participated in our institutional review board–approved locally advanced phase I/II head-and-neck squamous cell carcinoma protocol were used in this study (11). Details of the plans used for treatment are covered in detail elsewhere (11, 13), but are summarized here for completeness. All plans were created with 9 6-MV equally spaced coplanar beams, with the exception of Plan 28 (Patient 21), in which the location of the target structure

Results

Film measurements from 7 patient plans were analyzed and are summarized in Fig. 1. The numbers given in the figure are the percentage of points with gamma ≤1 with a 2% max dose and 2 mm distance-to-agreement distance criteria. The arithmetic average of all 7 plans with percentage of points passing gamma test is 76.6% with a range of 65.9–95.8% for SC and 95.4% with a range of 91.9–97.2% for the MC. This result demonstrates that our MC implementation better predicts in-phantom doses and can be

Discussion

This study shows that MC predicts higher doses than the SC for the dose calculation methods used for this patient cohort. Film dosimetry results confirm that the MC results can be considered to be a benchmark standard. The correlation between phantom dosimetry results and patient results is demonstrated in Fig. 7, which shows the average dose difference with respect to film dosimetry (MC − SC) observed for points with doses >10% of Dmax dose for the phantom calculations and the local percent

Conclusions

Monte Carlo was used to recompute dose distributions for 31 SIB-IMRT plans based on 24 different patients. On phantom, measurements demonstrated that MC dose calculations agree better with film than the SC algorithm as implemented using our in-house IMRT system, justifying its use as reference standard for patient-based dose computations. In the patient geometry, the SC algorithm results from the VCU-IMRT system agreed, on average, with the MC-based algorithm. Over all plans, the arithmetic

Acknowledgment

The authors would like to thank Devon Murphy-Stein for her meticulous editing of this manuscript.

References (50)

  • A. Leal et al.

    Routine IMRT verification by means of an automated Monte Carlo simulation system

    Int J Radiat Oncol Biol Phys

    (2003)
  • L. Wang et al.

    Monte Carlo evaluation of tissue inhomogeneity effects in the treatment of the head and neck

    Int J Radiat Oncol Biol Phys

    (2001)
  • M.A. Manning et al.

    The effect of setup uncertainty on normal tissue sparing with IMRT for head-and-neck cancer

    Int J Radiat Oncol Biol Phys

    (2001)
  • L. Cozzi et al.

    IMRT in the treatment of head and neck canceris the present already the future?

    Expert Rev Anticancer Ther

    (2002)
  • A. Eisbruch et al.

    Conformal and intensity modulated irradiation of head and neck cancerThe potential for improved target irradiation, salivary gland function, and quality of life

    Acta Otorhinolaryngol Belg

    (1999)
  • R.B. Kuppersmith et al.

    Intensity-modulated radiotherapyFirst results with this new technology on neoplasms of the head and neck

    Ear Nose Throat J

    (1999)
  • B.S. Teh et al.

    Intensity modulated radiation therapy (IMRT)A new promising technology in radiation oncology

    Oncologist

    (1999)
  • W.H. Fu et al.

    Comparison of conformal and intensity-modulated techniques for simultaneous integrated boost radiotherapy of upper esophageal carcinoma

    World J Gastroenterol

    (2004)
  • Intensity-modulated radiotherapycurrent status and issues of interest

    Int J Radiat Oncol Biol Phys

    (2001)
  • E. Martin et al.

    Validation of intensity modulation on a commercial treatment planning system

    Med Phys

    (2003)
  • C.P. Ceberg et al.

    Experimental determination of the dose kernel in high-energy x-ray beams

    Med Phys

    (1996)
  • P. Storchi et al.

    Calculation of the absorbed dose distribution due to irregularly shaped photon beams using pencil beam kernels derived from basic beam data

    Phys Med Biol

    (1996)
  • J.D. Bourland et al.

    A finite-size pencil beam model for photon dose calculations in three dimensions

    Med Phys

    (1992)
  • R. Mohan et al.

    The impact of fluctuations in intensity patterns on the number of monitor units and the quality and accuracy of intensity modulated radiotherapy

    Med Phys

    (2000)
  • P. Francescon et al.

    Dose verification of an IMRT treatment planning system with the BEAM EGS4-based Monte Carlo code

    Med Phys

    (2003)
  • Cited by (37)

    • Measurement-based Monte Carlo simulation of high definition dose evaluation for nasopharyngeal cancer patients treated by using intensity modulated radiation therapy

      2014, Radiation Measurements
      Citation Excerpt :

      Sterpin et al. (2007) has demonstrated that MC results show better agreement with measurements than AAA in a heterogeneous multilayer phantom and IMRT clinical treatments. Sakthi et al. (2006) studied 31 SIBIMRT plans and showed that MC calculations predicted higher doses than the TPS computations. Also, measured doses agreed better with MC results than TPS data.

    • Application of the measurement-based Monte Carlo method in nasopharyngeal cancer patients for intensity modulated radiation therapy

      2014, Radiation Physics and Chemistry
      Citation Excerpt :

      Table 1 shows that absorbed dose to the middle ears obtained from MBMC is, on average, higher than that from TPS. Sakthi et al. (2006) studied 31 SIB IMRT plans and showed that MC calculations predicted higher doses than the TPS computations. Also, measured doses agreed better with MC results than TPS data.

    • Accuracy requirements for head and neck intensity-modulated radiation therapy based on observed dose response of the major salivary glands

      2009, Radiotherapy and Oncology
      Citation Excerpt :

      The quality of basic beam data defining baseline for the accuracy of dose calculations [33] and the existence of the accelerator output (Gy/MU) within the predefined limits of ±2% [34] have been confirmed for our practice in other reports. The clinically used dose calculation algorithms have been reported to provide Dmean values for the parotids differing up to 5% [35–37]. In the current study, all Dmeans were calculated using the same pencil beam algorithm.

    • Computational Challenges for Image-Guided Radiation Therapy: Framework and Current Research

      2007, Seminars in Radiation Oncology
      Citation Excerpt :

      Because RAIR reduces setup error and margins around structures decreases, the importance of dose-calculation accuracy increases because dose errors, which often occur in the periphery of an irradiated volume, will now reside more closely to the clinical target volume as it moves through the course of therapy. One reason for using MC for RAIR is that MC dose calculation is the benchmark standard by which other dose computation algorithms are measured.94-102 For intensity-modulated radiation therapy (IMRT) and RAIR, MC can accurately account for both details of the fluence delivery (via direct transport of particles through moving MLC segments) and patient tissue heterogeneities.103,104

    • Dosimetric and Monte Carlo verification of jaws-only IMRT plans calculated by the Collapsed Cone Convolution algorithm for head and neck cancers

      2019, Reports of Practical Oncology and Radiotherapy
      Citation Excerpt :

      MC predicts about a 4.2% higher mean dose to the parotid glands as compared to TPS. This also accords with Sakthi's (2006) observations.12 As shown in Fig. 6, a good agreement between the experimental measurements, the TPS and Monte Carlo simulation of PDD curves along the central beam axis was achieved.

    View all citing articles on Scopus

    This work was supported by Grant CA98524 from the National Cancer Institute.

    Rupert Schmidt-Ullrich, M.D., passed away in December 2004.

    View full text