Advertisement

Improvement in plan quality after Implementation of clinical goals in a large network of cancer centers

Open AccessPublished:November 18, 2022DOI:https://doi.org/10.1016/j.meddos.2022.10.003

      Abstract

      Clinical Goals (CG) is a tool available in the Varian Eclipse planning system to objectively and visually evaluate the quality of treatment plans based upon user-defined dose-volume parameters. We defined a set of CG for Stereotactic Radiosurgery (SRS) and Intensity-Modulated Radiotherapy (IMRT) based on published data and guidelines and implemented this in a network of cancer centers in India (American Institute of Oncology). A dosimetric study was performed to compare brain SRS and breast IMRT plan quality before and after CG implementation.The CG defined for SRS plans were target V100% ≥ 98%, dose gradient measure (GM) ≤ 0.5 cm, conformity index (CI) 1.0 to 1.2. For breast IMRT plans, CG defined target V100% ≥ 97%, V95% ≥ 95%, V107% ≤ 2%, V105% ≤ 10%, and Dmax ≤ 2.4 Gy. Dose limits to organs-at-risk (OAR) were summarize in supplemental materials. Twenty brain SRS and 10 breast IMRT treatment plans that were previously delivered on patients were selected and re-planned using CG. The pre and postoptimized plan parameters were compared using student t-tests.For brain SRS plans, the V100, GM, and CI for the pre- and post-Clinical-Goals plans were 93.22% ± 7.2% vs 97.96% ± 0.29% (p = 0.009), 0.63 ± 0.16 vs 0.42 ± 0.05 (p < 0.001) and 1.07 ± 0.18 vs 1.06 ± 0.06 (p = 0.79), respectively. There were no differences in max dose to OARs. In breast IMRT plans, the target V107% for pre and postimplemented plans were 16.50% ± 10.98% vs 0.32% ± 0.32%, respectively (p = 0.001). The average target V105% were 44.00% ± 15.72% and 8.69% ± 4.53%, respectively (p < 0.001). No differences were found in the average target V100% (p = 0.128) and V95% (p = 0.205). The average target Dmax were 112.28% ± 1.59% and 109.14% ± 0.73%, respectively (p < 0.001). There were only minor differences in doses to OARs.The implementation of CG in Varian Eclipse significantly improved SRS and IMRT plan quality with enhanced coverage, dose GM, and CI without increased dose to OARs.

      Keywords

      Introduction

      Over the past few decades, rapid advancements in radiotherapy delivery capabilities came at the cost of increasing complexity in treatment planning.
      • Van Dyk J.
      Quality assurance of radiation therapy planning systems: current status and remaining challenges.
      ,
      • Tol JP
      • Dahele M
      • Delaney AR
      • et al.
      Can knowledge-based DVH predictions be used for automated, individualized quality assurance of radiotherapy treatment plans?.
      The increasing need for radiation precision, homogeneity, conformity, and more individual organs-at-risk (OARs) have made it challenging to efficiently generate consistent, high quality radiotherapy treatment plans.
      • Das IJ
      • Cheng C-W
      • Chopra KL
      • et al.
      Intensity-modulated radiation therapy dose prescription, recording, and delivery: Patterns of variability among institutions and treatment planning systems.
      ,
      • Nelms BE
      • Robinson G
      • Markham J
      • et al.
      Variation in external beam treatment plan quality: An inter-institutional study of planners and planning systems.
      While auto planning software can certainly minimize human subjectivity and time consumption in this planning process, plan quality assurance (QA) is also an important step to ensure the optimization of individual plans to maximize clinical outcomes.
      • Tol JP
      • Dahele M
      • Delaney AR
      • et al.
      Can knowledge-based DVH predictions be used for automated, individualized quality assurance of radiotherapy treatment plans?.
      ,
      • Peters LJ
      • O'Sullivan B
      • Giralt J
      • et al.
      Critical impact of radiotherapy protocol compliance and quality in the treatment of advanced head and neck cancer: Results from TROG 02.02.
      ,
      • Liu H
      • Clark R
      • Magliari A
      • et al.
      RapidPlan hippocampal sparing whole brain model version 2-how far can we reduce the dose?.
      Treatment plan selection and evaluation can be a highly subjective process. Historically, a popular approach is a physician-based quantitative metrics which are used to assign a score to a dosimetric plan to ensure its adherence based on particular physical criteria.
      • Hodapp N.
      [The ICRU Report 83: Prescribing, recording and reporting photon-beam intensity-modulated radiation therapy (IMRT)].
      Overall the used indices, though somewhat useful, do not account for the overall dosimetric information provided by dose-volume histograms (DVH). Therefore, most treatment centers in the United States and around the world still rely heavily on DVH for evaluation of plan quality.
      • Alfonso JC L.
      • Herrero MA
      • Núñez L
      A dose-volume histogram based decision-support system for dosimetric comparison of radiotherapy treatment plans.
      However, due to the similarity of competing tentative plans to the naked eye, considerable skills are required to elicit subtle differences from a mere inspection of the DVHs and isodose curves.
      • Alfonso JC L.
      • Herrero MA
      • Núñez L
      A dose-volume histogram based decision-support system for dosimetric comparison of radiotherapy treatment plans.
      Furthermore, DHVs do not provide a clear measure of critical parameters such as target coverage, dose gradient measure (GM), conformity index, and dose to OARs. Therefore, a better tool is needed to for treatment plan evaluation that provides objectivity, flexibility, and ease of use.
      Clinical Goals (CG) is a tool introduced to Varian EclipseTM planning system Version 16.0 in 2020 that objectively and visually evaluate the quality of treatment plans based upon user-defined dose-volume parameters. We implemented CG for common disease sites based on published data and guidelines in our cancer network. For this study we used brain SRS and breast IMRT dosimetry to compare plan quality before and after CG implementation.

      Materials Methods

      Treatment plan selection

      For this study, 20 brain SRS plans and 10 breast IMRT plans previously used for clinical treatment on patients in 2021 and 2022 were selected from the American Oncology Institute, which includes a large network of 16 cancer centers in South Asia. All plans (SRS and IMRT), before and after CG implementation, were generated using Varian EclipseTM planning system Version 16 and Anisotropic Analytic Algorithm.

      CG parameters for SRS

      Based on published data from QUANTEC and AAPM-RSS, the Varian EclipseTM Version 16.1 CG tool defined for brain SRS included the following plan quality measures: target V100% ≥ 98%, dose gradient measure (GM) ≤ 0.5-cm, conformity index (CI) 1.0 to 1.2, mean dose to brain < 8 Gy, and maximal dose (for 3 types of fractionations) to OARs such as the optic chiasm, lens, optic nerve, brainstem, and cochlea is listed in Appendix A. Target volume was characterized as gross target volume with a 2-mm PTV margin.
      Treatment plans were generated using a VMAT technique for a TrueBeamTM STx equipped with a high definition multileaf collimators. A minimum of 5 arcs (3 non-coplanar, 90, 70, 50 degrees) and maximum of 6 arcs (3 non-coplanar) were used. For full arcs, gantry angles were 181 to 170 degrees. For non-coplanar arcs, half arcs, or partial arcs, gantry angles were 181 to 0 or 181 to 330 degrees. The typical collimation angles were 10, 15, 25, 350, 325, 95, 85, and 125 degrees, depending on the PTV shape and surrounding OARs. To maintain the CI and GM to the PTV, 3 concentric ring structures were used, each ring has a 5-mm thickness with 1-mm gap between each ring. Optimization priority for PTV was 150, first ring was 140, with second and third ring priorities greater than the PTV priority. For example, if PTV priority is 150, the priority for second and third ring are 160 and 175, respectively. The prescription isodose line takes from 70% to 80% for all plans. Dose and fractionations varied for each plan due to the initial physician's discretion. Twenty brain SRS treatment plans previously delivered in clinical practice were re-planned with the use of CG, with all plans done by the same dosimetrist. Along with dose to OARs, target V100, dose GM, and CI were calculated for all plans, including retrospectively for the pre-CG plans since they were not evaluated using these parameters. The parameter values of pre- and post-implementation were then compared using student t-tests and χ2 tests.

      CG parameters for IMRT

      Based on NRG and Patel et al., the Varian EclipseTM Version 16.1 CG tool defined for breast IMRT included the following plan quality measures: Target V100% ≥ 97%, V95% ≥ 95%, V107% ≤ 2%, V105% ≤ 10%, Dmax ≤ 107%; opposite breast Dmax ≤ 2.4 Gy; heart V16% ≤ 5%, and Dmean ≤ 1.6 Gy; ipsilateral lung V16% ≤ 15%, V8% ≤ 35%, and V4% ≤ 50%; contralateral lung V4% ≤ 10%.
      Treatment plans were generated with 9 fields of fixed IMRT technique with multiple energies (6X and 15X). Typical gantry angles were 298, 315, 330, 135, and 155 degrees for breast PTV; 335, 355, 20, and 175 degrees for supraclavicular nodes. Out of 9 fields, 5 beams were used for breast or chest wall and 4 beams were used for supraclavicular fossa PTV. Higher energies (15X) were used, where the depth is more than 10-cm to achieve PTV conformity, especially to meet the goals of V105% and V107%. The prescription dose was 42.56Gy in 16 fractions. Two rings were used, Ring 1 for maintaining conformity for PTV and Ring 2 minimizing dose spillage outside the PTV. Similar to brain SRS plans, ten breast IMRT plans previously used in clinical practice were re-planned with the use of CG, and again, with all plans done by the same dosimetrist. The above parameter values were calculated for all plans before and after CG implementation and compared using student t-tests.

      Results

      The impact of the CG tool on SRS treatment plan quality

      The quality of brain SRS treatment plans improved significantly with the use of CG for certain parameters (Table 1, full data presented in Appendix A). The average target V100% (≥98%) in pre and postimplemented plans were 93.22% ± 7.2% and 97.96% ± 0.29%, respectively (p = 0.009). For dose GM (≤0.5), the average were 0.63 ± 0.16 cm and 0.42 ± 0.05 cm for pre and post plans, respectively (p < 0.001). No differences were found between the pre and post plans for average CI, which were 1.07 ± 0.18 and 1.06 ± 0.06, respectively (p = 0.79).
      Table 1Average brain SRS plan parameters before and after clinical goals implementation
      ParameterBefore clinical goalsAfter clinical goalsp-value
      Target V100% (%)93.22 ± 7.2097.96 ± 0.290.009
      Dose GM (cm)0.63 ± 0.160.42 ± 0.05<0.001
      Conformity index1.07 ± 0.181.06 ± 0.060.794
      Brainstem Dmax (Gy)5.85 ± 7.545.32 ± 7.470.270
      Left optic nerve Dmax (Gy)0.55 ± 0.930.64 ± 0.700.549
      Right optic nerve Dmax (Gy)1.06 ± 1.841.24 ± 2.210.312
      Optic chiasm Dmax (Gy)1.50 ± 2.731.89 ± 2.830.044
      Brain Dmean (Gy)0.91 ± 0.690.91 ± 0.611.000
      Left cochlea Dmax (Gy)1.46 ± 3.871.17 ± 2.570.359
      Right cochlea Dmax (Gy)3.26 ± 6.213.27 ± 6.370.681
      Left lens Dmax (Gy)0.25 ± 0.520.30 ± 0.310.667
      Right lens Dmax (Gy)0.35 ± 0.500.29 ± 0.380.623
      Only 14 of 20 preimplemented plans met a V100 of ≥98%, 2 of 20 met a GM ≤0.5 cm, and 12 of 20 met a CI between 1.0 and 1.2, compared to 100% met in all postimplemented plans (p < 0.001, p < 0.001, p = 0.005, respectively) (Table 1). Dose to OARs did not change significantly with the use of CG except for 1 case of left cochlea.
      In Plan 2 and Plan 4 (Appendix A), significant differences in V100% between the pre-CG and post-CG plans were observed (Plan 2: 85.00% vs 98.00%; Plan 4: 69.60% vs 98.00%).

      The impact of the CG tool on IMRT treatment plan quality

      Similar to brain SRS plans, certain breast IMRT treatment plan parameters also improved after CG use (Table 2, full data presented in Appendix B). The average target V107% (≤2.0% as specified by guidelines) were 16.50% ± 10.98% and 0.32% ± 0.32% for pre and post implemented plans, respectively (p = 0.001). The average target V105% (≤10.0%) were 44.00% ± 15.72% and 8.69% ± 4.53% for the pre and post plans, respectively (p < 0.001). No differences were found between pre and post plans for average target V100% (≥97.0%, p = 0.128) and V95% (≥95.0%, p = 0.205). Finally, the average target Dmax (≤107.0%) for the pre and post plans were 112.28% ± 1.59% and 109.14% ± 0.73%, respectively (p < 0.001).
      Table 2Average breast IMRT plan parameters before and after clinical goals implementation
      ParameterBefore clinical goalsAfter clinical goalsp-value
      PTVV100% (%)93.90 ± 2.0895.00 ± 0.000.128
      V95% (%)98.70 ± 1.0899.25 ± 0.370.205
      V107% (%)16.50 ± 10.980.32 ± 0.320.001
      V105% (%)44.00 ± 15.728.69 ± 4.53<0.001
      Dmax (%)112.38 ± 1.59109.14 ± 0.73<0.001
      Opposite breastDmax (Gy)19.59 ± 9.8516.76 ± 11.900.552
      HeartV16% (%)8.70 ± 11.874.28 ± 2.680.302
      Dmean (Gy)4.22 ± 1.012.92 ± 0.920.002
      Ipsilateral LungV16% (%)15.17 ± 2.5516.29 ± 3.860.239
      V8% (%)35.64 ± 8.1827.26 ± 5.850.025
      V4% (%)72.23 ± 14.5245.37 ± 11.70<0.001
      Contralateral lungV4% (%)22.06 ± 17.630.65 ± 1.560.005
      For OARs, there were no differences between pre- and post-implemented plans for the average opposite breast Dmax (≤2.4 Gy, p = 0.552) and average heart V16% (≤5.0%, p = 0.302), but the average heart Dmean (≤1.6%) for the pre and post plans were 4.22 ± 1.01 Gy and 2.92 ± 0.92 Gy, respectively (p = 0.002).
      For the ipsilateral lung, which is also an OAR, the average V8% (≤35.0%) were 35.64% ± 8.18% and 27.26% ± 5.85% for pre and postimplemented plans, respectively (p = 0.025). The average V4% (≤50.0%) was 72.23% ± 14.52% and 45.37% ± 11.70% for pre and post plans, respectively (p < 0.005). There were no differences between the pre and post plans for average V16% (≤15.0%, p = 0.239). For the contralateral lung, the average V4% (≤10.0%) for the pre and post plans were 22.06% ± 17.63% and 0.65% ± 1.56%, respectively (p = 0.005).
      In Appendix B, there were significant differences observed in V105% (44.00 ± 15.72 vs 8.69 ± 4.53, p < 0.001) and V107% (16.50 ± 10.98 vs 0.32 ± 0.32, p = 0.001) between the pre-CG and post-CG plans.

      Discussion

      Currently, many radiotherapy treatment centers in the United States and other countries do not have standardized measures for plan quality evaluation. The plans in this study were previously used clinically before CG implementation, allowing for the assessment of CG's real-world applications. The use of Varian EclipseTM Version 16.1 CG tool offers a scorecard-based indication for optimization, leading to enhanced overall plan quality based on guideline-defined limitations without any inherent changes to the software planning process. However, CG is not an independent optimization tool and therefore cannot automatically adjust the constraints of the failed structures during optimization.
      In brain SRS plans, the target conformity and OAR sparing is similar before and after the implementation of CG, but target coverage and dose GM improved significantly. The CG tool provides direct visualization of such parameters, many of which cannot be evaluated directly on DVHs without further calculations. Therefore, this tool can easily identify the deficiencies of a plan and help achieve the specified target coverage, gradient, and conformity without compromising OAR sparing. For breast IMRT plans, target homogeneity, mean heart dose, as well as lung doses significantly improved with CG implementation. Given the variation in expertise of physicians and dosimetrists in the treatment centers around the world, CG can provide a boost in plan quality through assessment of not only DVHs, but also finer indices that DVHs do not account for.
      The profound differences found in SRS Plan 2 and 4 (Appendix A) may be due an oversight by the physician when reviewing the plan based on DVH parameters, leading to more focus on OAR dose constraints. However, this further exemplifies that CG can identify deficiencies in a plan that can sometimes be overlooked by the treating physician. Before the use of CG, dosimetrists generally were not rigid on dose homogeneity, thus leading to less optimized plans. The implementation of CG allows for appropriate optimization of parameters such as homogeneity, explaining the large discrepancy in PTV V100%, V105%, and V107% in IMRT plans (Appendix B).
      There are several advantages of using CG for plan quality evaluation over manual reliance on DVHs and isodose curves, which are highly subjective. This tool uses color schemes to determine whether specified plan goals were met (green = meets guidelines, yellow = variation acceptable, red = did not meet guidelines) (Fig. 1). In addition, the customizability of CG allows for guideline adjustments based on institutional experience and evolving data during the treatment planning and evaluation process.
      Fig 1
      Fig. 1An example of the Varian EclipseTM Version 16.1 Clinical Goals tool.
      Although this study did not provide outcome analysis in which the clinical significance of CG implementation is evaluated, many studies have shown the impact of deviated target delineation and failure of rigorous quality assurance (QA) can result in inferior clinical outcomes.
      • Li XA
      • Moughan J
      • White JR
      • et al.
      Patterns of failure observed in the 2-step institution credentialing process for NRG oncology/radiation therapy oncology group 1005 (NCT01349322) and lessons learned.
      • Chen X
      • LeCompte MC
      • Gui C
      • et al.
      Deviation from consensus contouring guidelines predicts inferior local control after spine stereotactic body radiotherapy.
      • Ohri N
      • Shen X
      • Dicker AP
      • et al.
      Radiotherapy protocol deviations and clinical outcomes: A meta-analysis of cooperative group clinical trials.
      For example, Chen et al. reported that deviation from contouring guidelines was associated with suboptimal local control 1 year post spine stereotactic body radiotherapy (SBRT) (63.0% vs 85.5%, p < 0.001).
      • Chen X
      • LeCompte MC
      • Gui C
      • et al.
      Deviation from consensus contouring guidelines predicts inferior local control after spine stereotactic body radiotherapy.
      Improvement in dose gradient with SRS can help reduce dose to uninvolved brain which can reduce risk of necrosis. Similarly dose homogeneity improvement in breast planning leads to better cosmetic outcome. In conclusion, the results of this study showed that CG's user defined acceptance criteria set as clinical goals for plan evaluation can improve overall plan quality. Institutions around the world could implement the use of the Varian EclipseTM Version 16.1 CG for all disease sites for its objectivity, customizability, and ease of use in plan quality evaluation.

      Declaration of Competing Interest

      Mr Hefei Liu had recent employment with Varian Medical Systems in the last 12 months. Dr Sushil Beriwal has a leadership role as the Vice President of Medical Affairs at Varian Medical Systems, reports grant as an Elsevier consultant, and reports participation in advisory board at Xoft DSMB. Dr Deepak Khuntia has a leadership role as the Senior Vice President and Chief Medical Officer at Varian Medical Systems. Mr Praveen Nuksani, Dr Malolan Rajagopalan, Dr Mangesh Patil, Dr Krishna Komanduri, and Mr Brent Murphy have nothing to disclose.

      Funding Statement

      This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

      Data sharing statement

      Research data is stored in Varian repository and will be shared upon request to the corresponding author.

      References

        • Van Dyk J.
        Quality assurance of radiation therapy planning systems: current status and remaining challenges.
        Int J Radiat Oncol Biol Phys. 2008; 71: S23-S27https://doi.org/10.1016/j.ijrobp.2007.04.095
        • Tol JP
        • Dahele M
        • Delaney AR
        • et al.
        Can knowledge-based DVH predictions be used for automated, individualized quality assurance of radiotherapy treatment plans?.
        Radiat Oncol. 2015; 10: 234https://doi.org/10.1186/s13014-015-0542-1
        • Das IJ
        • Cheng C-W
        • Chopra KL
        • et al.
        Intensity-modulated radiation therapy dose prescription, recording, and delivery: Patterns of variability among institutions and treatment planning systems.
        J Natl Cancer Inst. 2008; 100: 300-307https://doi.org/10.1093/jnci/djn020
        • Nelms BE
        • Robinson G
        • Markham J
        • et al.
        Variation in external beam treatment plan quality: An inter-institutional study of planners and planning systems.
        Pract Radiat Oncol. 2012; 2: 296-305https://doi.org/10.1016/j.prro.2011.11.012
        • Peters LJ
        • O'Sullivan B
        • Giralt J
        • et al.
        Critical impact of radiotherapy protocol compliance and quality in the treatment of advanced head and neck cancer: Results from TROG 02.02.
        J Clin Oncol Off J Am Soc Clin Oncol. 2010; 28: 2996-3001https://doi.org/10.1200/JCO.2009.27.4498
        • Liu H
        • Clark R
        • Magliari A
        • et al.
        RapidPlan hippocampal sparing whole brain model version 2-how far can we reduce the dose?.
        Med Dosim Off J Am Assoc Med Dosim. 2022; (Published online)https://doi.org/10.1016/j.meddos.2022.04.003
        • Hodapp N.
        [The ICRU Report 83: Prescribing, recording and reporting photon-beam intensity-modulated radiation therapy (IMRT)].
        Strahlenther und Onkol Organ der Dtsch Rontgengesellschaff. 2012; 188: 97-99https://doi.org/10.1007/s00066-011-0015-x
        • Alfonso JC L.
        • Herrero MA
        • Núñez L
        A dose-volume histogram based decision-support system for dosimetric comparison of radiotherapy treatment plans.
        Radiat Oncol. 2015; 10: 263https://doi.org/10.1186/s13014-015-0569-3
        • Li XA
        • Moughan J
        • White JR
        • et al.
        Patterns of failure observed in the 2-step institution credentialing process for NRG oncology/radiation therapy oncology group 1005 (NCT01349322) and lessons learned.
        Pract Radiat Oncol. 2020; 10: 265-273https://doi.org/10.1016/j.prro.2019.11.007
        • Chen X
        • LeCompte MC
        • Gui C
        • et al.
        Deviation from consensus contouring guidelines predicts inferior local control after spine stereotactic body radiotherapy.
        Radiother Oncol. 2022; 173: 215-222https://doi.org/10.1016/j.radonc.2022.05.035
        • Ohri N
        • Shen X
        • Dicker AP
        • et al.
        Radiotherapy protocol deviations and clinical outcomes: A meta-analysis of cooperative group clinical trials.
        J Natl Cancer Inst. 2013; 105: 387-393https://doi.org/10.1093/jnci/djt001