ESTRO 2024 - Abstract Book

S2495

Clinical - Urology

ESTRO 2024

partial volume effects obfuscating the effects of interest, all masks were eroded with a 2 mm radius spherical kernel prior to value extraction.

Polynomial fits were made to the extracted data, per patient individually, with statistical tests at the group level performed afterwards. Given the limited number of degrees of freedom (despite 20 data points (volumes), only 5 time points) we have limited the order of this polynomial to be as conservative as possible in our analysis. The prostate volume data clearly exhibited an inverted U-shape and therefore required at least a second order fit. The prostate intensity and ADC showed monotonic behaviour for which a linear fit was deemed sufficient.

Results:

Figure 1 shows the effects in T2w images. Thin lines show individual fits to patient measures, thick lines show group averages. Over the course of treatment, the prostate gland initially increases in size by 11% on average, peaking halfway through the treatment (blue curves, stat. significance: deviation from 100% at Fraction 3: p=2.8e-30). For reference: for a perfectly spherical prostate with our group’s average volume of 46 ml, this corresponds to a radius increase of ~1.2mm. When extrapolating the fits to ~3 days after the last treatment (i.e. “fraction 6”) the volume would return to just below baseline (97%). The T2w intensity steadily declines throughout the entire treatment (slope different from zero: p=1e-18).

The ADC values are plotted in Figure 2. Thin lines show individual data (not fitted), thick lines shows the average of individual fits. Here the GTV shows a linear increase (p=0.00036), whereas the prostate gland ADC (without GTV) stays constant (p = 0.28). The difference between GTV and prostate trends (p=9e-5) is also significant. (i.e. Bonferroni corrected, pair-wise t-test)