Show simple item record

dc.contributor.authorShortall, J.
dc.contributor.authorOsorio, E. V.
dc.contributor.authorMcWilliam, A.
dc.contributor.authorGreen, A.
dc.contributor.authorChoudhury, A.
dc.contributor.authorHoskin, P.
dc.contributor.authorWest, C.
dc.contributor.authorLane, B.
dc.contributor.authorElumalai, T.
dc.contributor.authorThiruthaneeswaran, N.
dc.contributor.authorBibby, B.
dc.contributor.authorPereira, R. R.
dc.contributor.authorvan Herk, M.
dc.date.accessioned2022-01-11T11:59:57Z
dc.date.available2022-01-11T11:59:57Z
dc.date.issued2021en
dc.identifier.citationShortall J, Osorio EV, McWilliam A, Green A, Choudhury A, Hoskin P, et al. Towards personalised treatment for prostate cancer: biology improves image-based data mining models. Radiotherapy and Oncology. 2021;161:S504-S6.en
dc.identifier.urihttp://hdl.handle.net/10541/624922
dc.description.abstractPurpose or Objective Radiotherapy (RT) is a curative treatment for prostate cancer (PCa). However, with 1 in 3 men recurring, better stratification to identify patients benefiting nodal RT is needed. We aimed to test for the first time whether adding biology to an Image Based Data Mining (IBDM) model improved performance. Material and Methods Planning CTs, dose distributions and clinical information were collected for 354 high-risk PCa patients treated with 57Gy/60Gy in 19/20 fractions (n=254) or 37.5Gy in 15 fractions +15Gy HDR brachytherapy boost to the prostate (n=109). Hypoxia scores (HS) and radiosensitivity index (RSI) were calculated for a subset of 145 patients: 90 treated in 19/20 fractions, and 55 with brachytherapy boost. CTs were deformably registered to an arbitrarily chosen reference patient and dose distributions mapped to the same anatomy. Registration uncertainty was estimated using standard deviations of landmarks set at the seminal vesicles (SV) and apex, and dose distributions blurred with a Gaussian filter of corresponding width (≤0.9cm). Mean dose distributions for patients that did and did not fail were calculated, and dose differences between the two groups assessed using IBDM and permutation testing (4-year BioChemical Recurrence (BCR). Iso-T levels indicating significance (p≤0.05) were plotted on the observed T-map to identify regions of significant dose differences between patients who did and did not fail. Mean dose in each iso-T region was calculated for each patient and included in Cox analysis (BCR during follow-up). Backwards selection was applied to derive final models with and without biology. Akaike Information Criterion (AIC) was compared to assess model performance. Results No iso-T region was observed in the 19/20 fractions cohort. However, two regions were observed for the HDR boost patients, one in the SV tips and one in the apex (Fig1). Lower dose in these regions was associated with higher BCR. The population of brachytherapy patients with gene signatures was well balanced with the entire population of brachytherapy patients (p≥0.31 for all characteristics included in analysis). Mean dose to SV tips, age and baseline PSA were selected in final recurrence models for brachytherapy patients with and without gene signatures (Table1). T-stage and RSI were also selected in the gene signature model. AIC values were lower (i.e. better model performance) when including HS and RSI (Table1). Conclusion We show for the first time that adding biological information (RSI) improves performance of patient stratification models for high-risk PCa. Our model suggests older patients with lower T-stage, baseline PSA and RSI who receive excess dose to SV tips have lower BCR. Admittedly RSI was not significant and had unlikely Hazard Ratios and confidence intervals which will become clearer in larger cohorts. These findings and differences between fractionation schedules suggest links between fractionation and biology that should be explored in larger cohorts.en
dc.titleTowards personalised treatment for prostate cancer: biology improves image-based data mining modelsen
dc.typeMeetings and Proceedingsen
dc.contributor.departmentThe University of Manchester, Faculty of Biology, Medicine and Health, Division of Cancer Sciences, Manchester, United Kingdomen
dc.identifier.journalRadiotherapy and Oncologyen
dc.description.noteen]


This item appears in the following Collection(s)

Show simple item record