Automatic Planning of Whole Breast Radiation Therapy Using Machine Learning Models

PurposeBreast conserving surgery (BCS) followed by whole breast irradiation (WBI) is the standard of care for breast cancer patients. However, there is a risk of coronary events with WBI therapy. In this study, we compared the radiation dose in the left anterior descending artery (LAD) in patients receiving partial breast irradiation (PBI) with WBI.Material and methodsWe evaluated consecutive patients who underwent adjuvant radiotherapy after BCS between October 2008 and July 2014. Whole breast irradiation patients received 50 Gy in fractions of 2 Gy to the entire breast. Partial breast irradiation was performed using multicatheter brachytherapy at a dose of 32 Gy in eight fractions. The mean and maximal cumulative doses to LAD were calculated. The radiotherapeutic biologically effective dose of PBI was adjusted to WBI, and radiation techniques were compared.ResultsOf 379 consecutive patients with 383 lesions receiving radiotherapy (151 WBI and 232 PBI lesions), 82 WBI and 100 PBI patients were analyzed. In WBI patients, the mean and maximal cumulative doses for left-sided breast cancer (2.13 ± 0.11 and 8.19 ± 1.21 Gy, respectively) were significantly higher than those for right-sided (0.37 ± 0.02 and 0.56 ± 0.03 Gy, respectively; p < 0.0001). In PBI patients with left-sided breast cancer, the doses for tumors in inner quadrants or central location (2.54 ± 0.21 and 4.43 ± 0.38 Gy, respectively) were significantly elevated compared to outer quadrants (1.02 ± 0.17 and 2.10 ± 0.29 Gy, respectively; p < 0.0001). After the adjustment, the doses in PBI patients were significantly reduced in patients with tumors only in outer quadrants (1.12 ± 0.20 and 2.43 ± 0.37 Gy, respectively; p = 0.0001).ConclusionsTumor control and dose to LAD should be considered during treatment since PBI may reduce the risk of coronary artery disease especially in patients with lateral tumors in the left breast.

Purpose: To compare accelerated partial breast irradiation with whole breast irradiation in post breast conservation surgery (BCS) women with breast cancer. Material and methods: Women &amp;gt;35 years of age with invasive or noninvasive breast cancer ≤4 cm treated by BCS were randomized to 3D CRT APBI (34 Gy in 10 fractions given twice daily) or WBI (40 Gy in 16 fractions given once daily ± boost irradiation).The primary outcome was ipsilateral breast tumour recurrence and important secondary outcomes were adverse cosmetic outcome (fair or poor on a HARVARD/NSABP/RTOG breast cosmesis grading scale) and toxicity using the RTOG scores and LENT SOMA scale. Radiation toxicities and cosmesis was assessed directly by radiation oncologist during radiotherapy and in follow up.Patient and tumour characteristics, locoregional recurrence and distant metastases rates were compared using Fisher's exact tests.All statistical tests were two sided p values less than 0.05 were considered statistically significant. Results: Between June 2011 and December 2015, 133 women were randomized to 3D CRT APBI or WBI. Patient characteristics were balanced between two arms. There was an increased rate of acute grade 2 dermatitis in WBI arm(p=0.33). Median follow up was 60 months(range 9-84 months). Grades 1 and 2 late radiation toxicities were higher in the WBI arm compared to the APBI arm; pigmentation, fibrosis and breast shrinkage were significantly more in the WBI arm. Grades 3 and 4 late toxicity was not seen in any of the treatment arms. Adverse cosmesis at last follow up was significantly higher in patients treated with WBI, 32% as compared to 6% with APBI (p=&amp;lt;0.001). Local recurrence with APBI was 3% as compared to 1.5% with WBI and distant metastasis rate was high in WBI arm as compared with APBI, 5.9% vs 3% respectively; both not statisticaly significant. Table 1.Patient and tumor characteristicsCharacteristicsAPBI(n=65)WBI(n=68)p-valueMean age(range)50(36-75)(36-75)50(37-67)(37-67)0.13T- Stage T134(52)37(54) T228(43)28(41)0.96T33(5)3(4) N stage N057(88)59(87) N17(11)28(41)1.00N21(1)1(1) Grade 114(22)12(18) 237(56)40(59)0.5314(22)16(23) Surgical margins Negative60(92)61(90)0.79Positive5(8)7(10) LVI No56(86)55(81)0.49Yes9(14)13(19) ER Positive42(65)45(66) Negative19(29)21(31)1.00Unknown5(6)2(3) PR Positive38(58)39(57) Negative23(35)25(37)1.0Unknown4(7)4(6) Her2-neu Positive7(11)6(9)0.78Chemotherapy Yes35(54)52(76)0.78No30(46)16(24) Hormone therapy Yes49(75)49(72)0.78No16(25)19(28) Trastuzumab Yes3(43)3(50)1.00No4(57)3(50) Conclusion: In women with BCS, APBI was associated with better cosmetic outcome and late radiation toxicities compared to WBI. Local recurrence and distant metastasis was comparable in both the arms. Citation Format: Yadav BS, L S, Sharma SC, Singh R, Dahiya D. Comparison of accelerated partial breast irradiation (APBI) with whole breast irradiation (WBI) using 3D conformal external beam radiation therapy (3D CRT) [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-12-13.

Five-Year Outcomes, Cosmesis, and Toxicity With 3-Dimensional Conformal External Beam Radiation Therapy to Deliver Accelerated Partial Breast Irradiation

BackgroundWe conducted a randomized trial to evaluate the efficacy of heparinoid moisturization for radiation dermatitis. We report the time-course of sebum content after whole breast radiotherapy (WBRT) and the efficacy of heparinoid moisturizer.MethodsPatients receiving adjuvant breast RT were randomly assigned into three groups; prophylaxis, post-WBRT and control groups. Patients used moisturizer on the irradiated breast from the beginning of RT in the prophylaxis group, 2 weeks post-RT in the post-WBRT group, and no moisturizer in the control group. Sebum content of the irradiated and non-irradiated breast was measured to assess sebaceous gland damage. Sebum composition was also analyzed.ResultsA total of 76 patients were analyzed; 30 in the post-WBRT group, 32 in the control group, 14 in the prophylaxis group. The sebum content in the irradiated breast significantly decreased after WBRT in the post-WBRT and control groups. The decrease was sustained in the control group. In the non-irradiated breast, sebum content also decreased after WBRT in the post-WBRT and control groups. After moisturizer application, sebum content by sebumeter returned to pre-RT level in the post-WBRT group, while the decrease was sustained in the control group. Sebum content measured by evaporative light scattering detector and sebumeter was similar in the control group, but the dissociation was observed after moisturizer application in the post-WBRT group. The proportion of wax esters decreased in the irradiated breast after WBRT.ConclusionsRadiotherapy significantly reduced sebum content in both irradiated and non-irradiated breast, indicating that RT caused quantifiably persistent sebaceous gland damage in irradiated sites and the surrounding tissue. Combined with the results from our previous study, heparinoid moisturizer treatment effectively prevents water loss by retaining oil contents on the skin surface.Trial registrationUMIN, UMIN000005532. Registered 1 April 2011.

Ten-year outcomes of 3D-conformal accelerated partial vs. whole breast irradiation after breast-conserving surgery: A randomized study from India.

On the basis of low risk of local recurrence in elderly patients with breast cancer after conservative surgery followed by whole breast irradiation (WBI), the Danish Breast Cancer Group initiated the noninferiority external-beam partial breast irradiation (PBI) trial (ClinicalTrials.gov identifier: NCT00892814). We hypothesized that PBI was noninferior to WBI regarding breast induration. Patients operated with breast conservation for relatively low-risk breast cancer were randomly assigned to WBI versus PBI, and all had 40 Gy/15 fractions. The primary end point was 3-year grade 2-3 breast induration. In total, 865 evaluable patients (434 WBI and 431 PBI) were enrolled between 2009 and 2016. Median follow-up was 5.0 years (morbidity) and 7.6 years (locoregional recurrence). The 3-year rate of induration was 9.7% for WBI and 5.1% for PBI (P = .014). Large breast size was significantly associated with induration with a 3-year incidence of 13% (WBI) and 6% (PBI) for large-breasted patients versus 6% (WBI) and 5% (PBI) for small-breasted patients. PBI showed no increased risk of dyspigmentation, telangiectasia, edema, or pain, and patient satisfaction was high. Letrozole and smoking did not increase the risk of radiation-associated morbidity. Sixteen patients had a locoregional recurrence (six WBI and 10 PBI; P = .28), 20 patients had a contralateral breast cancer, and eight patients had distant failure (five WBI and three PBI). A nonbreast second cancer was detected in 73 patients (8.4%), and there was no difference between groups. External-beam PBI for patients with low-risk breast cancer was noninferior to WBI in terms of breast induration. Large breast size was a risk factor for radiation-associated induration. Few recurrences were detected and unrelated to PBI.

Comparison of Toxicity and Cosmetic Outcomes After Accelerated Partial Breast Irradiation or Whole Breast Irradiation Using 3-Dimensional Conformal External Beam Radiation Therapy

Background: Whole breast irradiation (WBI) after breast conserving surgery for ductal carcinoma in-situ (DCIS) reduces local recurrence. We aimed to evaluate whether a tumor bed boost after WBI improved long-term outcomes, and examine radiation dose fractionation sensitivity for non-low-risk DCIS. Methods: The study was an international, randomized, unmasked, phase 3 trial involving 136participating centers from six clinical trials organizations in 11 countries. Eligible patients were women aged ≥18 years who had breast conserving surgery with ≥1 mm of clear radial resection margins for unilateral, histologically proven, non-low-risk DCIS defined as age &amp;lt;50 years or age ≥50 years plus at least one of the risk factors for local recurrence (palpable tumor, multifocal disease, tumor size ≥1.5cm, intermediate or high nuclear grade, central necrosis, comedo histology and/or surgical margin&amp;lt;10 mm). Patients were stratified by age (&amp;lt;50 years vs ≥50 years), planned endocrine therapy (yes or no) and treating center. They were randomized to one of four groups (1:1:1:1) of no boost versus boost after conventional versus hypofractionated WBI, or to one of two groups (1:1) of no boost versus boost after each center-prespecified conventional or hypofractionated WBI. The conventional WBI was 50 Gy in 25 fractions, and hypofractionated WBI was 42·5 Gy in 16 fractions. A boost dose of 16 Gy in eight fractions, if allocated, was delivered after WBI. The primary endpoint was time to local recurrence. Secondary endpoints were time-to-disease recurrence, overall survival and treatment toxicity. Cosmetic outcome and health-related quality of life will be reported separately. This trial is registered with ClinicalTrials.gov (NCT00470236). Results: Between June 25, 2007, and June 30, 2014, 1608 patients were randomly assigned to no tumor bed boost (805 patients) or boost (803 patients). Conventional WBI was given to 831 patients, and hypofractionated WBI was given to 777 patients. Endocrine therapy was used in 13% of patients. Median follow-up was 10.2 years. The 10-year free-from-local-recurrence rates were 87% (95% CI 84-89%) in the no-boost group and 93% (91-95%) in the boost group (hazard ratio 0.49; 0.34-0.69;p&amp;lt;0.001). Fifty-four percent and 58% of local recurrences were invasive in the no-boost and boost groups, respectively. There were no significant differences in the 10-year free-from-local recurrence rates between the conventional WBI and hypofractionated WBI groups in the 4-arm randomization category (90% vs. 89%, P=0.82) or in all randomized patients (89% vs. 91%, P=0.80). The test for interaction between boost and dose-fractionation was not significant in the 4-arm randomization category (P=0.39) or in all randomized patients (P=0.77). The 10-year free-from-disease recurrence rates were lower in the no-boost group (79%) than in the boost group (87%; hazard ratio, 0.67;0.52−0.86; p=0.002). There was no statistically significant difference in 10-year overall survival rates between the no-boost (94%) and boost (96%) groups (0·73; 0·47-1·14; p=0·17). The boost group had higher rates of grade 2 or higher breast induration (7% [5-9%] vs 17% [14-19%], p&amp;lt;0.001) and breast pain (11% [9-13%] vs 17% [14-20%], p=0·002), with no suggestion of interaction with WBI dose fractionation. Conclusions: Our results provide the first long-term randomized trial data to demonstrate the efficacy of tumor bed boost radiation after postoperative WBI on local recurrence in patients with non-low-risk DCIS with an associated increase in grade 2 or greater breast induration and pain. Within the limits of the trial design, moderately hypofractionated WBI was not associated with an increase in local recurrence or toxicity. Citation Format: B. H. Chua, E. K. Link, I. A. Olivotto, I. Kunkler, T. Whelan, P. Deseyne, G. Gruber, BIG 3-07/TROG 07.01 trial investigators.. Radiation doses and fractionation schedules in non-low-risk ductal carcinoma in situ in the breast (BIG 3-07/TROG 07.01): final 10-year analysis of a randomised, factorial, multicentre, open-label, phase 3 study [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr GS2-12.

Breast-conserving therapy with partial or whole breast irradiation: Ten-year results of the Budapest randomized trial

Breast conserving therapy remains one of the most important advances in cancer management in the past century, with multiple trials demonstrating its equivalence to mastectomy with respect to long term cancer outcomes. Unfortunately, studies have documented that adjuvant radiation therapy, a critical component of breast conserving therapy, remains under utilized. Data suggest that treatment duration (5–6.5 weeks) can frequently be attributed to this finding. In light of this and secondary to the increasing costs associated with breast cancer radiation therapy, shortened courses of radiation therapy have been explored including both hypofractionated whole breast irradiation (WBI) and accelerated partial breast irradiation (APBI) that shorten treatment duration to three and one week, respectively. Prospective randomized trials evaluating these techniques have been promising, demonstrating equivalence to traditional WBI with respect to clinical outcomes, toxicities, and breast cosmesis. Intraoperative radiation therapy (IORT) represents an alternative to these techniques that delivers treatment at the time of surgery as a single fraction of radiation (in the majority of cases). Increasingly, IORT is being offered to women with early-stage breast cancer as the definitive radiation therapy modality following breast conserving surgery. What is concerning is that these patients are being offered this, as yet, unproven therapy off-protocol with insufficient data to support its safety and efficacy compared with WBI or alternative techniques (APBI or hypofractionated WBI). This is unfortunate as radiation therapy in breast cancer represents an area of research that has undergone progress in the past five decades through systematic evidence-based treatment paradigm evolution; breast conserving therapy was verified via multiple randomized trials demonstrating comparable outcomes to mastectomy, and postmastectomy radiation therapy similarly demonstrated a survival benefit in several trials before wide-scale incorporation into treatment paradigms. Therefore, the rapid introduction of IORT is concerning as these steps have not been taken prior to wide-scale utilization, leading to a potential for higher rates of local recurrence or increased toxicity. It should be noted that two large randomized trials have been performed to evaluate the role of IORT in early-stage breast cancer compared with WBI, the current standard of care. The targeted intraoperative radiotherapy (TARGIT) trial was a randomized noninferiority trial that included 3,451 women enrolled in 10 countries between 2000 and 2012. Patients were randomized to IORT (delivered at the same time as surgery or as a separate procedure) or WBI; however, 21 % of the prepathology IORT patients received WBI because of predefined factors including lobular carcinoma in situ, extensive intraductal component, lymphovascular space invasion, node positivity, or other factors specific to each institution. IORT was delivered using a 50-kv X-ray source to deliver 20 Gy to the surface of the excision cavity (5 Gy at 1 cm). While initial data with short follow-up (median follow-up \2 years) and limited events (n = 11) was promising, a recent update found that IORT, even with WBI supplementation in one-fifth of the Society of Surgical Oncology 2013

Purpose: To improve the robustness of a knowledge based automatic lung IMRT planning method and to further validate the reliability of this algorithm by utilizing for the planning of clinical cases with non-coplanar beams. Methods: A lung IMRT planning method which automatically determines both plan optimization objectives and beam configurations with non-coplanar beams has been reported previously. A beam efficiency index map is constructed to guide beam angle selection in this algorithm. This index takes into account both the dose contributions from individual beams and the combined effect of multiple beams which is represented by a beam separation score. We studied the effect of this beam separation score on plan quality and determined the optimal weight for this score.14 clinical plans were re-planned with the knowledge-based algorithm. Significant dosimetric metrics for the PTV and OARs in the automatic plans are compared with those in the clinical plans by the two-sample t-test. In addition, a composite dosimetric quality index was defined to obtain the relationship between the plan quality and the beam separation score. Results: On average, we observed more than 15% reduction on conformity index and homogeneity index for PTV and V40, V60 for heart while an 8% and 3% increase on V5, V20 for lungs, respectively. The variation curve of the composite index as a function of angle spread score shows that 0.6 is the best value for the weight of the beam separation score. Conclusion: Optimal value for beam angle spread score in automatic lung IMRT planning is obtained. With this value, model can result in statistically the “best” achievable plans. This method can potentially improve the quality and planning efficiency for IMRT plans with no-coplanar angles.

Breast-Conserving Treatment With Partial or Whole Breast Irradiation for Low-Risk Invasive Breast Carcinoma—5-Year Results of a Randomized Trial

The purpose of this study was to report the efficacy and the safety profile on the subset of selected early breast cancer (BC) patients aged 70 years or older from a single-center phase 3 trial comparing whole breast irradiation (WBI) to accelerated partial breast irradiation (APBI) using intensity-modulated radiation therapy technique. Between 2005 and 2013, 520 patients aged more than 40 years old were enrolled and randomly assigned to receive either WBI or APBI in a 1:1 ratio. Eligible patients were women with early BC (maximum diameter 2.5 cm) suitable for breast conserving surgery. This study is registered with ClinicalTrials.gov, NCT02104895. A total of 117 patients aged 70 years or more were analyzed (58 in the WBI arm, 59 in the APBI arm). At a median follow-up of 5-years (range 3.4-7.0), the ipsilateral breast tumor recurrence (IBTR) rate was 1.9 % in both groups. No significant difference between the two groups was identified (log-rank test p = 0.96). The 5-year disease-free survival (DFS) rates in the WBI group and APBI group were 6.1 and 1.9 %, respectively (p = 0.33). The APBI group presented significantly better results in terms of acute skin toxicity, considering both any grade (p = 0.0001) and grade 2 or higher (p = 0.0001). Our subgroup analyses showed a very low rate and no significant difference in terms of IBTR, using both WBI and APBI. A significant impact on patients compliance in terms of acute and early late toxicity was shown, which could translate in a consistent improvement of overall quality of life.

Automatic radiotherapy planning based on deep learning-based dose prediction shows promise in improving planning efficiency, consistency, and quality. However, most existing studies have focused on scenario-specific dose prediction models, thereby limiting the widespread clinical application of such algorithms. This study aims to develop a generalizable dose prediction model capable of predicting dose distributions across various tumor sites, delivery techniques, and prescription doses, with high accuracy. Furthermore, we seek to demonstrate the feasibility of universal automatic planning guided by this generalizable dose prediction model. Model development and evaluation were conducted using the GDP-HMM dataset, comprising 3234 plans encompassing different tumor sites (head and neck cancer, lung cancer), delivery techniques (IMRT, VMAT), and prescription doses. The proposed dose prediction model, based on the 3D MedNeXt architecture, was trained on a subset of 2878 plans. Physics-informed priors were incorporated by designing model inputs such as normalized distance-aware beam plates and mass density maps to enhance performance. Large-kernel convolutions, combined with the UpKern initialization strategy, were employed to expand the model's receptive field and improve prediction accuracy. Model generalizability and accuracy were assessed on the remaining 356 plans. Ablation studies were performed to evaluate the impact of physics-informed priors and large-kernel convolutions. To demonstrate the feasibility of universal automatic planning, predicted dose distributions for four representative test cases were converted into deliverable plans using fallback planning process in the RayStationTM (RaySearch Laboratories, Stockholm, Sweden) treatment planning system. The proposed model accurately predicted dose distributions for plans with varying tumor sites, delivery techniques, and prescription doses, achieving median prediction errors within 2 percentage points (pp) for all assessed dose-volume histogram (DVH) metrics. The mean absolute prediction errors across all DVH metrics were 2.13 pp for head and neck plans and 1.65 pp for lung plans. Ablation studies revealed that incorporating mass density maps and normalized distance-aware beam plates as model inputs significantly reduced body mean absolute errors (MAE) by 2.1% (p<0.001) and 5.8% (p<0.001), respectively. Increasing the convolution kernel size from 3 to 5 with the UpKern initialization strategy further reduced body MAE by 2.1% (p<0.001). The model achieved second place in the AAPM GDP-HMM Challenge. Deliverable plans generated from predicted dose distributions for four representative cases of various clinical scenarios demonstrated quality comparable to reference plans. We have developed a generalizable and accurate dose prediction model capable of predicting dose distributions for diverse tumor sites, delivery techniques, and prescription doses. The integration of physics-informed priors and large-kernel convolutions significantly enhanced prediction accuracy. The feasibility of universal automatic planning based on the generalizable dose prediction model has been demonstrated, highlighting its potential to facilitate broader clinical application of automatic planning and to improve planning efficiency, consistency, and quality. The code and trained models are publicly available at https://github.com/tyxiong123/RTPDosePred.

A Dosimetric Evaluation of the Impact of Seroma Volume During Breast Boost Radiotherapy