Digital Health Innovation and Informatics
PD 14 - Digital Health Information & Informatics - Poster Discussion
1123 - Knowledge Based Quality Assurance and Improvement in Locally Advanced Lung Cancer Radiation Therapy in a Statewide Consortium of Academic and Community Practice Centers
Tuesday, October 23
5:21 PM - 5:27 PM
Location: Room 217 A/B
Martha Matuszak, PhD
University of Michigan
University of Michigan: Associate Professor: Employee; William Beaumont Hospital: Orthopaedic Surgeon: Employee
Varian Medical System: Research Funding and Consulting; Varian Medical Systems: Consultant
Knowledge Based Quality Assurance and Improvement in Locally Advanced Lung Cancer Radiation Therapy in a Statewide Consortium of Academic and Community Practice Centers
M. M. Matuszak1, M. Grubb1, R. Marsh1, K. M. Masi2, D. Lack3, D. A. Dryden4, M. Wilson5, D. Jarema1, D. Tatro1, E. P. Short6, T. J. Bichay7, J. M. Moran8, P. A. Paximadis9, M. M. Dominello10, J. D. Radawski11, L. L. Kestin5, L. J. Pierce1, S. Jolly1, J. A. Hayman1, and T. P. Boike12; 1Michigan Medicine, Ann Arbor, MI, 2Karmanos Cancer Institute, Detroit, MI, 3Beaumont Health System, Troy, MI, 4McLaren Northern Michigan, Petoskey, MI, 5Michigan Healthcare Professionals, 21st Century Oncology, Farmington Hills, MI, 6Department of Radiation Oncology, Providence Cancer Center, Novi, MI, 7Saint Mary's Health Care, Ann Arbor, MI, 8University of Michigan, Ann Arbor, MI, 9Lakeland Radiation Oncology, St. Joseph, MI, 10Department of Oncology, Wayne State University School of Medicine, Detroit, MI, 11West Michigan Cancer Center, Rockford, MI, 12Karmanos Cancer Network, Detroit, MI
Purpose/Objective(s): Treatment planning for locally advanced lung cancer (LALC) is complex due to large PTVs and their close proximity to multiple organs at risk (OAR). Given the difficult tradeoff decisions encountered, it can be challenging to judge plan quality. Here, a statewide consortium of academic and community practices collaborated to develop and validate a multi-institutional knowledge based planning (KBP) model of high quality lung plans for use as a quality assurance (QA) and improvement tool.
Materials/Methods: Fifty-six LALC definitive radiation therapy plans were collected from 9 institutions with different planning systems and techniques. Plans were scored by a group of physicians, physicists, and dosimetrists. High quality clinical (HQC) plans were selected to build a KBP model. After model building, removal of outliers, and refinement of the objective function template, the model was validated via automated KBP planning with 10 HQC cases not included in model building. Differences in dosimetric plan quality indicators (PTV D95%; lung: mean, V20Gy, V5Gy; esophagus: mean, D2cc; heart: mean, D0.03cc; cord: mean, D0.03cc) in HQC vs. KBP plans were compared to determine if the model could produce plans of similar quality. After validation, an additional 32 randomly selected, clinically treated cases from 7 consortium institutions were re-planned without normalization using model feedback to determine the utility of KBP for statewide QA and improvement in lung planning.
Results: Out of the 56 scored HQC lung plans, 43 were included in the final KBP model, 3 were excluded, and 10 were used for validation. In the 10 validation cases, there were no significant differences in any of the target or OAR metrics in the HQC vs. KBP plans, demonstrating that the model was able to predict and confirm HQC plans. In 32 randomly selected clinical plans, the KBP model identified potential improvements in PTV coverage (2.3 +/- 3.0 Gy average improvement in D95%) in 78% of cases. In 41% of cases, the KBP model identified improvement for ≥ 5 OAR metrics. For example, the average estimated KBP reduction in heart mean dose was 1.9 +/- 1.6 Gy. In 25% of cases, KBP identified simultaneous improvement in PTV coverage and ≥ 5 OAR metrics, with 3DCRT cases more likely see an improvement vs. intensity modulated cases.
Conclusion: A multi-institutional LALC KBP model was created from HQC plans in a statewide collaborative of community and academic practices. The model was validated and confirmed to predict high quality plans as scored by a multidisciplinary team. When using the model for QA in randomly selected clinical plans, 78% and 41% of cases were identified as having potential improvement in PTV coverage and ≥ 5 dosimetric OAR goals, respectively. Work is ongoing to develop a web-based interface to provide patient-specific feedback on plan quality to consortium institutions.
Author Disclosure: M.M. Matuszak: Employee; William Beaumont Hospital. Consultant; Varian Medical Systems. M. Grubb: None. K.M. Masi: None. D. Lack: None. D.A. Dryden: None. E.P. Short: None. J.M. Moran: Research Grant; Varian Medical Systems, Blue Cross Blue Shield of Michigan, NIH. We have a collaboration regarding the use of gel dosimetry. Modus Medical supplies gels for the research.; Modus Medical Devices. Consultant; Chartrounds, St. Jude Children's Research Hospital, VA National Center for Patient Safety. Travel Expenses; AAPM, St. Jude Children's Research Hospital. Chair; AAPM. P.A. Paximadis: None. J.D. Radawski: Executive Committee representative; Michigan Radiation Oncology Quality Consortium. L.L. Kestin: board member; Michigan Healthcare Professionals. National Director of Thoracic & Lung Care Services; 21st Century Oncology. L.J. Pierce: Royalty; UpToDate. J.A. Hayman: Research Grant; Blue Cross Blue Shield of Michigan. T.P. Boike: Partner; Petoskey Radiation Oncology. Honoraria; ASTRO APEx Reviewer. Member; NCI Prostate Cancer Task Force.