Sensitivity of dose‐volume indices to computation settings in high‐dose‐rate prostate brachytherapy treatment plan evaluation

Marjolein C. van der Meer, Peter A. N. Bosman, Bradley R. Pieters, Yury Niatsetski, N. Van Wieringen, Tanja Alderliesten, Arjan BelJournal of Applied Clinical Medical Physics 2019 · DOI

TL;DR

This study reveals that seemingly minor differences in how treatment planning systems (TPS) compute dose-volume (DV) indices for prostate brachytherapy can lead to significant variations in results, particularly for organs like the bladder and seminal vesicles. Understanding these computational sensitivities is crucial for ensuring accurate treatment plan evaluation and consistent clinical decision-making across different centers.

What Problem Does This Paper Address?

In high-dose-rate (HDR) prostate brachytherapy, dose-volume (DV) indices are essential metrics for evaluating the effectiveness of treatment and the safety of surrounding organs. These indices quantify how much of a target organ receives a certain dose or how much of an organ-at-risk (OAR) is exposed to a specific dose level. However, the accuracy and comparability of these DV indices can be affected by the underlying computational settings within the treatment planning systems (TPS) used to generate them. Different TPSs, or even different settings within the same TPS, can lead to variations in DV index calculations. This lack of standardization poses a challenge for inter-institutional comparisons of treatment plans and could potentially influence clinical decisions. This paper aims to quantify the impact of these common computational setting variations on DV indices derived from actual clinical treatment plans.

Methods

The researchers developed an in-house software tool to systematically evaluate the impact of five key computational settings on DV indices derived from 26 patient treatment plans. These settings were:

  1. Number of dose-calculation points: How densely the dose is sampled within an organ.
  2. Radioactive source description: The specific model used to represent the radiation source (e.g., mHDR-v2, mHDR-v2r, mHDR-v2c).
  3. Contour interpolation: How the 3D shape of an organ is reconstructed from 2D slice contours.
  4. Organ intersection handling: Whether overlapping regions between organs (e.g., urethra and prostate) are assigned to one organ, the other, or both.
  5. Top and bottom slice thickness: Whether the outermost contours are assumed to represent the full slice thickness or just the surface of the organ.

They compared these variations against a "baseline" setting that closely mirrored the default configurations of their clinical TPS. For each setting, DV indices were recalculated, and the differences from the baseline were analyzed.

Datasets

The study utilized data from 26 consecutive patients who underwent prostate HDR brachytherapy at the authors' medical center between February 2015 and April 2017. Treatment planning was based on MRI scans with a 3.3 mm slice thickness. No new benchmarks or datasets were introduced; the focus was on analyzing existing clinical data under varied computational conditions.

Key Insights and Ideas

  • The devil is in the details: Even seemingly small differences in computational settings can lead to substantial variations in DV indices, particularly for organs with complex shapes or small volumes.
  • OARs are more sensitive: Organs-at-risk (OARs) like the bladder and urethra, and to some extent seminal vesicles, showed greater sensitivity to computational settings compared to the prostate target. This is likely due to their often smaller volumes and more complex anatomical relationships.
  • Sampling density matters, especially for OARs: While a moderate number of dose-calculation points suffice for targets, OARs require a significantly higher density to achieve a narrow confidence interval for DV indices, highlighting the need for careful sampling strategies.
  • Organ representation is critical: How the TPS reconstructs the 3D shape of organs from 2D slices and how it handles overlaps between organs can have a profound impact on DV index values. This is especially true for the bladder's caudal contour and the seminal vesicles.

Key Results

  • Dose-calculation points: To achieve a 95% confidence interval width of 1% or less for DV indices, approximately 32,000 points per target were needed, but a much larger 256,000 points per OAR were required.
  • Source models: Differences in DV indices due to variations in radioactive source models were minimal, with a maximum change of 0.94% for seminal vesicles.
  • Contour interpolation and intersection:
    • Urethra inclusion/exclusion significantly impacted prostate DV indices, with median changes of 1.0% and 1.2% for D90% and V150% respectively.
  • Full vs. partial slice thickness for top/bottom contours led to notable changes, especially for the bladder (up to 8.5% increase in D1cm3) and seminal vesicles (up to 9.8% decrease in V80%).
  • Overall impact: In extreme cases, DV indices could change by as much as 8.5% for the bladder and 9.8% for the seminal vesicles compared to the baseline setting, potentially influencing clinical decisions.

Limitations and Future Work

The authors acknowledge that this is a single-center study, and the findings are based on treatment plans optimized within their specific clinical TPS. Different TPSs might yield different results. Furthermore, the study was retrospective, and the influence of these computational settings on the treatment planning optimization process itself, and consequently on patient outcomes, was not directly assessed.

Future work could involve investigating the impact of these settings on treatment plan optimization algorithms and exploring robust optimization strategies that account for these uncertainties. Additionally, extending this analysis to other treatment sites beyond prostate brachytherapy would be valuable. The authors also suggest that using smaller slice thicknesses and delineating more contours could help improve the accuracy of DV index calculations.

Summary updated

Abstract

The values of DV indices used in prostate HDR brachytherapy treatment planning are influenced by the computation settings in a TPS, especially at the most caudal part of the bladder, as well as in the seminal vesicles.
BibTeX
@article{meer2019sensitivity,
  author = {Marjolein C. van der Meer and Peter A. N. Bosman and Bradley R. Pieters and Yury Niatsetski and N. Van Wieringen and Tanja Alderliesten and Arjan Bel},
  title = {{Sensitivity of dose‐volume indices to computation settings in high‐dose‐rate prostate brachytherapy treatment plan evaluation}},
  journal = {Journal of Applied Clinical Medical Physics},
  year = {2019},
  doi = {10.1002/acm2.12563},
}
Figure