==== Front Front Oncol Front Oncol Front. Oncol. Frontiers in Oncology 2234-943X Frontiers Media S.A. 10.3389/fonc.2023.1200676 Oncology Original Research Dose outside of the prostate is associated with improved outcomes for high-risk prostate cancer patients treated with brachytherapy boost Shortall Jane 1 * Vasquez Osorio Eliana 1 Green Andrew 1 McWilliam Alan 1 2 Elumalai Thriaviyam 2 Reeves Kimberley 1 Johnson-Hart Corinne 2 Beasley William 2 Hoskin Peter 1 2 Choudhury Ananya 1 2 van Herk Marcel 1 1 Division of Cancer Sciences, Faculty of Biology, Medicine and Health, The University of Manchester, Manchester, United Kingdom 2 Department of Radiotherapy Related Research, The Christie National Health Service (NHS) Foundation Trust, Manchester, United Kingdom Edited by: Nils H. Nicolay, University Hospital Leipzig, Germany Reviewed by: Constantinos Zamboglou, German Oncology Center, Cyprus; Cozzarini Cesare, San Raffaele Hospital (IRCCS), Italy *Correspondence: Jane Shortall, jane.shortall@manchester.ac.uk 15 6 2023 2023 13 120067605 4 2023 31 5 2023 Copyright © 2023 Shortall, Vasquez Osorio, Green, McWilliam, Elumalai, Reeves, Johnson-Hart, Beasley, Hoskin, Choudhury and van Herk 2023 Shortall, Vasquez Osorio, Green, McWilliam, Elumalai, Reeves, Johnson-Hart, Beasley, Hoskin, Choudhury and van Herk https://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Background One in three high-risk prostate cancer patients treated with radiotherapy recur. Detection of lymph node metastasis and microscopic disease spread using conventional imaging is poor, and many patients are under-treated due to suboptimal seminal vesicle or lymph node irradiation. We use Image Based Data Mining (IBDM) to investigate association between dose distributions, and prognostic variables and biochemical recurrence (BCR) in prostate cancer patients treated with radiotherapy. We further test whether including dose information in risk-stratification models improves performance. Method Planning CTs, dose distributions and clinical information were collected for 612 high-risk prostate cancer patients treated with conformal hypo-fractionated radiotherapy, intensity modulated radiotherapy (IMRT), or IMRT plus a single fraction high dose rate (HDR) brachytherapy boost. Dose distributions (including HDR boost) of all studied patients were mapped to a reference anatomy using the prostate delineations. Regions where dose distributions significantly differed between patients that did and did-not experience BCR were assessed voxel-wise using 1) a binary endpoint of BCR at four-years (dose only) and 2) Cox-IBDM (dose and prognostic variables). Regions where dose was associated with outcome were identified. Cox proportional-hazard models with and without region dose information were produced and the Akaike Information Criterion (AIC) was used to assess model performance. Results No significant regions were observed for patients treated with hypo-fractionated radiotherapy or IMRT. Regions outside the target where higher dose was associated with lower BCR were observed for patients treated with brachytherapy boost. Cox-IBDM revealed that dose response was influenced by age and T-stage. A region at the seminal vesicle tips was identified in binary- and Cox-IBDM. Including the mean dose in this region in a risk-stratification model (hazard ratio=0.84, p=0.005) significantly reduced AIC values (p=0.019), indicating superior performance, compared with prognostic variables only. The region dose was lower in the brachytherapy boost patients compared with the external beam cohorts supporting the occurrence of marginal misses. Conclusion Association was identified between BCR and dose outside of the target region in high-risk prostate cancer patients treated with IMRT plus brachytherapy boost. We show, for the first-time, that the importance of irradiating this region is linked to prognostic variables. radiation oncology image based data mining prostate outcomes radiotherapy Prostate Cancer UK 10.13039/501100000771 This work was supported by Prostate Cancer UK [grant number RIA15-ST2-031]. section-in-acceptanceRadiation Oncology ==== Body pmc1 Introduction Radiotherapy is the primary curative treatment for approximately 30% of prostate cancer patients in the UK, and has been attributed to high overall survival rates (1). However, prognosis remains poor for high-risk patients (2, 3), and 20-30% of patients experience biochemical recurrence (BCR) within five years of radiotherapy (4). There is currently no consensus for the optimal management of high-risk prostate cancer (2), and current standard of care radiotherapy is delivered to the prostate and seminal vesicles only, as opposed to lymph node and Whole Pelvis Radiotherapy (WPRT) which has shown some benefit for some high-risk patients (2, 5–10). Some patients may receive lymph node radiotherapy, however this is administered using risk-stratification models that consider known prognostic factors, but have limited consideration for the risk of microscopic spread of disease (2, 11–15). Advances in diagnostic imaging, such as prostate-specific membrane antigen positron emission tomography (PSMA-PET) imaging has revealed that microscopic disease spread to regions outside of the usual pelvic lymph nodes are more common than previously thought (16, 17). With radiotherapy dose distributions becoming increasingly more conformal around the prostate, many patients are potentially left with under- or un-treated lymph node or microscopic disease outside of the target volume, which could consequently lead to BCR (18–22). Clinical studies and trials comparing lymph node and WPRT with prostate only radiotherapy (PORT) have conflicting outcomes (2, 5–10). Controversy associated with WPRT is likely due to inability to detect those at risk and to adequately treat involved lymph nodes or regions of microscopic disease whilst remaining within dose tolerances. As a result, WPRT often results in large volumes being irradiated with low doses, which is sub-optimal for both disease control and normal tissue complications. Conversely, whilst radiotherapy techniques try to limit dose outside of the Planning Target Volume (PTV) to spare normal tissues, studies have shown that incidental dose outside of the PTV could be inadvertently treating subclinical disease (23). Witte et al. (18) and Chen et al. (24) used Image Based Data Mining technique (IBDM), where the local association between dose and outcome is explored on a per-voxel basis, and found that that incidental dose to the obturator was associated with improved outcomes (25), a region that had previously been associated with clinically relevant incidental dose (26). More recently, Witte et al. suggested that the relationship between incidental dose and BCR is influenced by fractionation when they identified a region in the in obturator internus muscles where incidental dose was beneficial for patients treated with standard fractionation radiotherapy, but not in patients treated with a hypo-fractionated schedule (27). These IBDM studies just consider the dose distribution, and do not include known prognostic factors that current stratification models use. Green et al. presented a Cox-IBDM methodology, whereby a Cox regression is performed in every voxel of the dose distribution, allowing dose-sensitive regions to be identified considering both dose and prognostic clinical variables simultaneously (28). Our Cox methodology works per voxel using permutation testing to correct for multiple testing. We, firstly, aim to explore the hypothesis that targeted dose outside of the prostate reduces BCR in a selection of high-risk prostate cancer patients (18). Secondly, we aim to investigate whether fractionation schedule influences this “extra-prostatic dose-effect relation” (27). Thirdly, we apply Cox-IBDM, accounting for both dose and prognostic variables to develop a predictive model to select high-risk prostate cancer patients that may benefit from adapted target volumes. 2 Method We refer the reader to Supplementary Material Appendix A for more detailed methodology. 2.1 Study design A total of 612 patients with high-risk prostate cancer (2) treated with radiotherapy between 2005 and 2013 at a single academic center were included. Institutional approval had been granted to use this data (research ethics committee reference: 17/NW/0060). Patients were treated with either conformal hypo-fractionated radiotherapy (50Gy in 16 fractions (equivalent dose in 2Gy fractions (EQD2)=66.07Gy, using alpha-beta ratio=1.5), n=258, prostate and 1-2cm proximal seminal vesicles irradiated, 2005-2011), hypo-fractionated Intensity Modulated Radiotherapy (IMRT) (57Gy or 60Gy in 19 or 20 fractions respectively (EQD2 = 73.29/77.14Gy), n=245, prostate and 1-2cm proximal seminal vesicles irradiated in line with CHHiP guidelines (29), 2008-2013), or IMRT (37.5 Gy to the prostate in 15 fractions) plus a 15Gy single fraction High Dose Rate (HDR) brachytherapy boost to the prostate and 1-2cm proximal seminal vesicles (n=109, (EQD2 = 113.57Gy), 2009-2013). Treatment fractionation schedule was assigned by consultant clinical oncologists according to local practice. No patients received elective nodal irradiation, and Androgen Deprivation Therapy (ADT) was the only systemic therapy used. Planning Computed Tomography (CT) scan and delineations, 3D planned dose distribution (Philips Pinnacle treatment planning system archive) and patient and tumor characteristics (age, T-stage, Gleason grade, Androgen Deprivation Therapy (ADT) duration and baseline PSA) were collected for all patients. Note that no other systemic therapies than ADT were used Brachytherapy boost was planned using ultrasound imaging and a single PTV to the prostate and delivered as an HDR treatment using an Ir-192 source. As dose distribution export is not supported by the used brachytherapy planning system, the planned brachytherapy dwell positions and catheter positions and times were collected and used to reconstruct the planned dose using the dose distribution of a nominal 10Ci Ir-192 source (30) and the recorded dwell position in the DICOM plan object (31). The dose distributions of each dwell position were then summed, accounting for contributions from all sources, to provide the total dose. See Supplementary Material Appendix A section 1.1 for more details on brachytherapy dose reconstruction. The end-point of the study was BCR (PSA nadir + 2ng/ml) (32), and the three cohorts were analyzed separately. 2.2 Dose mapping For brachytherapy boost, the reconstructed dose distribution was spatially aligned to the external beam planned dose distribution and the total treatment dose summed using EQD2 dose (33) ( Supplementary Material Appendix A section 2). Note that, as analysis explored relative differences between dose distributions of the same treatment technique and not absolute dose, the alpha-beta ratio is not critical in the range of relevant alpha-betas (34). Dose reconstruction and alignment was visually checked for 10 arbitrary patients to ensure the 15Gy line aligned with the prostate contour. Sensitivity for uncertainties in the dose summation were simulated by randomly shifting the brachytherapy dose by up to 1cm in any cardinal direction (as to simulate a worst-case scenario) and repeating analysis. Next, planning CTs were spatially registered, and dose distributions mapped to an arbitrarily chosen reference patient. A region of interest forming a sphere of approximately 10cm radius, centered on the centre of the reference prostate, was chosen for analysis to remove regions where spurious results could occur due to differences in radiotherapy plans depending on patient anatomy. See Supplementary Material Appendix A section 2 for details of algorithm used. Prior to dose mapping, dose distributions of all patients were flipped in the left-right direction and included in analysis twice. This method, which has been commonly used in other IBDM studies (35), assumes that the likelihood of microscopic disease is symmetric, avoids spurious laterality biased results caused by small asymmetries in the dose distributions, and improves statistical power. To assess and account for the accuracy of the dose mapping, Target Registration Error analysis was performed (36) and dose distributions blurred using a 3D Gaussian filter of width according to the standard deviation in each cardinal direction of manually made landmarks on the seminal vesicle tips and apex for a selection of patients (LR: 0.49, AP: 0.65, SI: 0.91cm, Supplementary Material Figure S1 ). 2.3 Voxel-based analysis Binary-IBDM and Cox-IBDM were performed to assess differences in the dose distributions of patients who did and did-not recur (in-house software (37)). For this, mapped doses were grouped depending on BCR status (PSA failure free survival (bNED)=0, fail=1). Binary-IBDM: Mapped doses for each voxel of the bNED and fail at four-years post radiotherapy (18) groups were compared using a Students T-test. T-maps containing the observed t-values in each voxel were created. A positive t-value indicates that excess dose is associated with reduced BCR. Cox-IBDM: For Cox-IBDM, a Cox proportional-hazards model was constructed for each voxel, including mapped doses to that voxel, age (continuous), T-stage (≥T3 vs 18 months 75 114 59 Mean baseline PSA (ng/ml) 40.04 32.05 26.67 Median baseline PSA (ng/ml) 24.00 21.00 22.00 Number of BCR during follow-up 124 77 33 Number of 4-year BCR 49 49 24 Number of 5-year BCR 69 60 27 Mean recurrence time (years) 5.57 5.11 4.55 Median recurrence time (years) 5.24 5 4.64 Tables 2A, B show univariable and multivariable analysis for prognostic variables. No variable was significant across all cohorts. No variables were significant for brachytherapy patients in univariable analysis, but T-stage was significant (p=0.048, HR: 0.46 (0.22, 0.99)) in multivariable analysis. For interest, we performed multivariable analysis dichotomizing Gleason class ≥3 vs ≤2, i.e. ≥4 + 3 vs ≤3 + 4, as according to The International Society of Urological Pathology (ISUP), as opposed to ≥8 vs <8, i.e. ISUP Class 4-5 vs ≤3 (43) ( Supplementary Material Table S1 ). Results were not different for any cohort. Table 2 Univariable (panel A) and multivariable (panel B) Cox proportional-hazards analysis for clinical prognostic covariates included in the study (age, T-stage, Gleason grade, ADT duration, baseline PSA). A Univariable Conformal hypo-fractionated (50Gy in 16#) IMRT (57/60Gy in 19/20#) EBRT + Brachytherapy (37.5Gy in 15# + 15Gy HDR boost) HR (95% CI) p-value HR (95% CI) p-value HR (95% CI) p-value Age (continuous) 0.98 (0.95 - 1.00) 0.100 0.97 (0.94 - 1.01) 0.130 0.96 (0.91 - 1.01) 0.110 T-stage (≥T3 reference) – – – – – –