Radiation oncology always seeks to strike the right balance between cancer control and the risk of treatment toxicity. As our understanding and technological abilities have evolved, so too has the need for accurate ways to design treatments that will achieve the best balance to help patients.
For more conventional radiation therapy used in the 1.8-2 Gray daily dose, we have QUANTEC, published in 2010.1 With the evolution of safe and highly conformal techniques to give much higher doses per fraction, we need different evidence-based parameters to guide clinicians and treatment team to provide high quality radiation treatment.
HyTEC (‘Hy’ Dose per Fraction, Hypofractionated Treatment Effects in Clinic) seeks to expand our knowledge base in an essential, timely way. HyTEC provides evidence-based dose-volume parameters not only for normal tissue complication probability (NTCP) but also for estimates of tumor control probability (TCP).
For our June journal club with #radonc #jc, we will discuss an overview of this project and review how we arrive at some of these estimates. We have selected the two introductory articles of HyTEC to highlight the progress we have made, the caveats of modeling, and some of the work we still have before us.
A Primer on Dose-Response Modeling in Radiation Therapy. Moiseenko V, Marks LB, Grimm J, Jackson A, Milano MT, Hattangadi-Gluth JA, Huynh-Le M, Pettersson N, Yorke E, El Naqa I. Int J Radiat Oncol Biol Phys 2021;110(1):11-203
We are grateful to have Drs. Jimm Grimm, Søren Bentzen, Ronald Chen, Laura Dawson, Jona Hattangadi-Gluth, Brian Kavanagh, Percy Lee, Michael Milano, and other contributors join us to discuss this project. Our conversation will begin Saturday Jun 19th at 8 AM Central Standard Time with plans to discuss globally through the weekend. We’ll end at approximately 2 PM CST on Sunday June 20th.
We usually have a series of questions through the weekend ending with a live discussion hour. Because Father’s Day may interfere with everyone’s ability to participate, we will discuss more questions at a gradual pace without the live hour.
Thank you also to Dr. Sue Yom and the Red Journal for making the articles temporarily free to read June 10-20 for the discussion!
T1. HyTEC is an essential update to guide us regarding effective treatment planning for high dose radiation therapy. When and how did this project originate?
T2. Three-dimensional CT-based treatment planning lets us evaluate dose-volume relationships for clinically relevant outcomes. QUANTEC covered normal tissue tolerance for conventionally fractionated radiation therapy. What did HyTEC want to do differently?
T3. Targeting radiation more accurately and precisely requires reliable estimates of how non-uniform dose distributions may affect the normal tissue and organs in the human body if we’re going to give higher doses. How do we get reliable data?
T4. 50% of cancer patients received radiation therapy at some point in their cancer experience. What have been major barriers to sharing large volumes of data to generate better estimates of cancer control and treatment toxicity?
T5. AAPM TG 263 standardizes how we identify tumor targets and normal tissue.4 CTCAE standardizes definitions of treatment toxicity.5 We need data commonality to pool it. Do we have the right standards? How to we improve adoption?
T6. What are dataset articles for treatment planning, and why do we need them? How can they help us describe and estimate radiation dose-volume relationships more effectively?
T7. Let’s discuss how we achieve robust modeling. First, endpoint definition. How hard is it to get investigators to report treatment toxicity and tumor control in ways we can use for reliable estimates?
T8. Scatter plots and receiver operating characteristic curves can help identify potential clinically relevant associations. Assuming we have enough data, how do we best use them to identify useful cut-points?
T9. QUANTEC didn’t use the Youden index for selecting optimal cut-points, but HyTEC did. Do we need to re-analyze conventional radiation DVH parameters more robustly in the future?
T10. Generating sigmoid dose-response curves with maximum likelihood estimation lessens noise from data for outliers. Let’s take a look at Figure 2. For clinicians looking at this graph, what are caveats for whether to apply it to patients?
T11. We often use simple cut-off points, but there may be interplay. Mean lung dose (MLD) cutoff of X Gy may seem OK, but maybe not for all V20 Gy % values. Do we need more sophisticated use of DVH parameters to build a better banana?
T12. Do we have confidence in our confidence intervals? If the data are pooled, where do you find a validation data set?
T13. How are we doing in terms of using data sets with diverse patient cohorts? Are there ways we can do better, so that the predictive models we design are more representative of a broader patient experience?
T14. Speaking of patient experience, HyTEC highlighted patient empowerment as part of an idealized future state. In what ways can we make patient-centered predictive models? Should we see patients as partners?
T15. HyTEC shares estimates for both tumor control and toxicity. But it highlights a noticeable omission in stereotactic radiation – we leave cancer biology that we routinely include in conventional radiation planning. Do we need a BioTEC project to better include CTVs?
Some tips to participate: