SU-D-BRB-04: Nomogram for Prediction, Comparison, and Evaluation of Dose to Normal Tissue in SRS Planning




Radiosurgery of brain metastases is a unique treatment planning subset in which the treatment approach, high conformality and steep dose gradients are relatively invariant with varying location, size and shape of the target. Prescription doses are generally limited by the size of tumor or amount of surrounding normal brain receiving intermediate doses. We tabulated our treatment plans to establish a nomogram which provides planning expectations, useful for patient scheduling, planning efficiency, and training purposes.


The clinical treatment plans of 247 individual metastases (0.3–100cc, 1–5 fractions) were processed, after exclusion of targets qualitatively close to a sensitive brain structure. Plans were designed with 3–5 dynamic conformal arcs using BrainLab-Iplan or 1–2 VMAT arcs using Eclipse. The effective prescription (D95) was calculated for each plan, and the total volumes of tissue receiving 10%, 20%, 30%, etc. of that dose were calculated (V10, V20, V30 etc.). These data were fitted against the target equivalent diameter with a 3rd degree polynomial minimizing the least-squares difference, from which the upper and lower quartile fits were generated. Plans were then separated by qualitative concavity of the target, and by MLC type; 0.5cm central leaves versus 0.25cm for Millenium and HD MLCs, respectively.


The polynomial fitting all data predicted the treated V50 with an r2 value of 0.97 and a mean error of 3.0 cc. The lower-to-upper-quartile V50 range in cc was 12.2–16.2, 14.6–18.1, and 17.2–20.8 for convex, intermediate, and concave targets. The 0.5cm leaves exhibited 1.5cc higher V50 for the smallest targets over 0.25cm leaves.


Planning performance based on institutional knowledge data offers insights into the attainable results before the planning process begins, and provides support during plan evaluation. Our compiled database will be used to compare treatment approaches and to triage patients.