MRI-guided Adaptive Radiation Therapy Research Articles

2129: Phase 2 Trial of Stereotactic MRI-guided Adaptive Radiation Therapy in One Fraction (SMART ONE)

2129: Phase 2 Trial of Stereotactic MRI-guided Adaptive Radiation Therapy in One Fraction (SMART ONE)

Structure-specific rigid dose accumulation dosimetric analysis of ablative stereotactic MRI-guided adaptive radiation therapy in ultracentral lung lesions.

Definitive local therapy with stereotactic ablative radiation therapy (SABR) for ultracentral lung lesions is associated with a high risk of toxicity, including treatment related death. Stereotactic MR-guided adaptive radiation therapy (SMART) can overcome many of the challenges associated with SABR treatment of ultracentral lesions. We retrospectively identified 14 consecutive patients who received SMART to ultracentral lung lesions from 10/2019 to 01/2021. Patients had a median distance from the proximal bronchial tree (PBT) of 0.38 cm. Tumors were most often lung primary (64.3%) and HILUS group A (85.7%). A structure-specific rigid registration approach was used for cumulative dose analysis. Kaplan-Meier log-rank analysis was used for clinical outcome data and the Wilcoxon Signed Rank test was used for dosimetric data. Here we show that SMART dosimetric improvements in favor of delivered plans over predicted non-adapted plans for PBT, with improvements in proximal bronchial tree DMax of 5.7 Gy (p = 0.002) and gross tumor 100% prescription coverage of 7.3% (p = 0.002). The mean estimated follow-up is 17.2 months and 2-year local control and local failure free survival rates are 92.9% and 85.7%, respectively. There are no grade ≥ 3 toxicities. SMART has dosimetric advantages and excellent clinical outcomes for ultracentral lung tumors. Daily plan adaptation reliably improves target coverage while simultaneously reducing doses to the proximal airways. These results further characterize the therapeutic window improvements for SMART. Structure-specific rigid dose accumulation dosimetric analysis provides insights that elucidate the dosimetric advantages of SMART more so than per fractional analysis alone.

Evaluating contouring accuracy and dosimetry impact of current MRI-guided adaptive radiation therapy for brain metastases: a retrospective study

BackgroundMagnetic resonance imaging (MRI) guided adaptive radiotherapy (MRgART) has gained increasing attention, showing clinical advantages over conventional radiotherapy. However, there are concerns regarding online target delineation and modification accuracy. In our study, we aimed to investigate the accuracy of brain metastases (BMs) contouring and its impact on dosimetry in 1.5 T MRI-guided online adaptive fractionated stereotactic radiotherapy (FSRT).MethodsEighteen patients with 64 BMs were retrospectively evaluated. Pre-treatment 3.0 T MRI scans (gadolinium contrast-enhanced T1w, T1c) and initial 1.5 T MR-Linac scans (non-enhanced online-T1, T2, and FLAIR) were used for gross target volume (GTV) contouring. Five radiation oncologists independently contoured GTVs on pre-treatment T1c and initial online-T1, T2, and FLAIR images. We assessed intra-observer and inter-observer variations and analysed the dosimetry impact through treatment planning based on GTVs generated by online MRI, simulating the current online adaptive radiotherapy practice.ResultsThe average Dice Similarity Coefficient (DSC) for inter-observer comparison were 0.79, 0.54, 0.59, and 0.64 for pre-treatment T1c, online-T1, T2, and FLAIR, respectively. Inter-observer variations were significantly smaller for the 3.0 T pre-treatment T1c than for the contrast-free online 1.5 T MR scans (P < 0.001). Compared to the T1c contours, the average DSC index of intra-observer contouring was 0.52‒0.55 for online MRIs. For BMs larger than 3 cm3, visible on all image sets, the average DSC indices were 0.69, 0.71 and 0.64 for online-T1, T2, and FLAIR, respectively, compared to the pre-treatment T1c contour. For BMs < 3 cm3, the average visibility rates were 22.3%, 41.3%, and 51.8% for online-T1, T2, and FLAIR, respectively. Simulated adaptive planning showed an average prescription dose coverage of 63.4‒66.9% when evaluated by ground truth planning target volumes (PTVs) generated on pre-treatment T1c, reducing it from over 99% coverage by PTVs generated on online MRIs.ConclusionsThe accuracy of online target contouring was unsatisfactory for the current MRI-guided online adaptive FSRT. Small lesions had poor visibility on 1.5 T non-contrast-enhanced MR-Linac images. Contour inaccuracies caused a one-third drop in prescription dose coverage for the target volume. Future studies should explore the feasibility of contrast agent administration during daily treatment in MRI-guided online adaptive FSRT procedures.

Abstract A005: A multi-institutional randomized phase III trial of chemotherapy alone versus chemotherapy and ablative 5-fraction MRI-guided adaptive radiation therapy for locally advanced pancreatic cancer (LAP-ABLATE)

Abstract Background: Prospective studies demonstrate that chemotherapy and non-ablative radiation therapy (RT) achieve superior locoregional control (LRC) although have not improved overall survival (OS) relative to chemotherapy alone for patients with locally advanced pancreatic cancer (LAPC). Emerging data suggest that significant radiation dose escalation may further enhance LRC, and it has been hypothesized that in a subset of patients this may improve OS. However, delivering ablative radiation dose in 5 or fewer fractions with computerized tomography (CT) guidance is associated with unacceptable rates of grade 3-5 toxicity. Safe dose escalation can be facilitated by radiation delivery systems that utilize magnetic resonance image (MRI) guidance and on-table adaptive replanning. A growing body of literature describes promising efficacy and safety of ablative 5-fraction stereotactic MRI-guided adaptive radiation therapy (SMART) for LAPC compared to historical non-ablative outcomes. Based on these encouraging results, the LAP-ABLATE trial has been designed to test the hypothesis that ablative SMART improves OS. Methods: The LAP-ABLATE trial is a multi-institutional group sequential randomized phase III study for patients with LAPC. Patients are eligible after first receiving 4-6 months of (modified)FOLFIRINOX or gemcitabine/nab-paclitaxel with either stable or responding disease. Other key eligibility criteria include CA 19-9 &amp;lt;180 U/mL and no contraindication to undergoing MRI scans. Patients are randomized (1:2) to receiving no ablative SMART (arm A) versus ablative SMART (arm B), with subsequent systemic therapy or surgery being optional in both arms. Ablative SMART will be delivered using the MRIdian Linac, a 0.35 Tesla MRI-guided linear accelerator. The prescribed dose is 50 Gy to gross disease and 33 Gy to elective regions in 5 fractions using a simultaneous integrated boost (SIB). The primary objective is to compare 2-year OS with secondary objectives including assessment of LRC, distant metastasis free survival (DMFS), progression-free survival (PFS), quality of life (QOL), and toxicity. Randomization will be stratified by participating center, baseline CA 19-9 level and induction chemotherapy regimen. Target accrual is 267 patients (arm A: 89, arm B: 178). Discussion: The LAP-ABLATE trial is the first prospective randomized phase III study designed to compare OS in LAPC patients following induction systemic therapy randomized to with or without ablative 5-fraction RT. Citation Format: Michael D. Chuong, Jessica M. Frakes, Kathryn E. Mittauer, James Good, Josh Kim, Daniel Chang, Dae Won Kim, Stefanie Corradini, Luca Boldrini, Mokenge Malafa, Eileen O'Reilly. A multi-institutional randomized phase III trial of chemotherapy alone versus chemotherapy and ablative 5-fraction MRI-guided adaptive radiation therapy for locally advanced pancreatic cancer (LAP-ABLATE) [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Pancreatic Cancer; 2023 Sep 27-30; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(2 Suppl):Abstract nr A005.

Automated deep learning auto-segmentation of air volumes for MRI-guided online adaptive radiation therapy of abdominal tumors

Objective. In the current MR-Linac online adaptive workflow, air regions on the MR images need to be manually delineated for abdominal targets, and then overridden by air density for dose calculation. Auto-delineation of these regions is desirable for speed purposes, but poses a challenge, since unlike computed tomography, they do not occupy all dark regions on the image. The purpose of this study is to develop an automated method to segment the air regions on MRI-guided adaptive radiation therapy (MRgART) of abdominal tumors. Approach. A modified ResUNet3D deep learning (DL)-based auto air delineation model was trained using 102 patients’ MR images. The MR images were acquired by a dedicated in-house sequence named ‘Air-Scan’, which is designed to generate air regions that are especially dark and accentuated. The air volumes generated by the newly developed DL model were compared with the manual air contours using geometric similarity (Dice Similarity Coefficient (DSC)), and dosimetric equivalence using Gamma index and dose-volume parameters. Main results. The average DSC agreement between the DL generated and manual air contours is 99% ± 1%. The gamma index between the dose calculations with overriding the DL versus manual air volumes with density of 0.01 is 97% ± 2% for a local gamma calculation with a tolerance of 2% and 2 mm. The dosimetric parameters from planning target volume—PTV and organs at risk—OARs were all within 1% between when DL versus manual contours were overridden by air density. The model runs in less than five seconds on a PC with 28 Core processor and NVIDIA Quadro® P2000 GPU. Significance: a DL based automated segmentation method was developed to generate air volumes on specialized abdominal MR images and generate results that are practically equivalent to the manual contouring of air volumes.

Local Therapies for Hepatocellular Carcinoma and Role of MRI-Guided Adaptive Radiation Therapy.

Hepatocellular carcinoma (HCC) is the most common liver tumor, with a continually rising incidence. The curative treatment for HCC is surgical resection or liver transplantation; however, only a small portion of patients are eligible due to local tumor burden or underlying liver dysfunction. Most HCC patients receive nonsurgical liver-directed therapies (LDTs), including thermal ablation, transarterial chemoembolization (TACE), transarterial radioembolization (TARE), and external beam radiation therapy (EBRT). Stereotactic ablative body radiation (SABR) is a specific type of EBRT that can precisely deliver a high dose of radiation to ablate tumor cells using a small number of treatments (or fractions, typically 5 or less). With onboard MRI imaging, MRI-guided SABR can improve therapeutic dose while minimizing normal tissue exposure. In the current review, we discuss different LDTs and compare them with EBRT, specifically SABR. The emerging MRI-guided adaptive radiation therapy has been reviewed, highlighting its advantages and potential role in HCC management.

Multi-Institutional Outcomes of Patients Aged 75 years and Older With Pancreatic Ductal Adenocarcinoma Treated With 5-Fraction Ablative Stereotactic Magnetic Resonance Image-Guided Adaptive Radiation Therapy (A-SMART).

Treatment options for pancreatic ductal adenocarcinoma (PDAC) are commonly limited for patients with advanced age due to medical comorbidities and/or poor performance status. These patients may not be candidates for more aggressive chemotherapy regimens and/or surgical resection leaving few, if any, other effective treatments. Ablative stereotactic MRI-guided adaptive radiation therapy (A-SMART) is both efficacious and safe for PDAC and can achieve excellent long-term local control, however, the appropriateness of A-SMART for elderly patients with inoperable PDAC is not well understood. A retrospective analysis was performed of inoperable non-metastatic PDAC patients aged 75years or older treated on the MRIdian Linac at 2 institutions. Clinical outcomes of interest included overall survival (OS), progression-free survival (PFS), distant metastasis-free survival (DMFS), and locoregional (LRC). Toxicity was graded according to Common Terminology Criteria for Adverse Events (CTCAE, v5). A total of 49 patients were evaluated with a median age of 81years (range, 75-91) and a median follow-up of 14 months from diagnosis. PDAC was classified as locally advanced (46.9%), borderline resectable (36.7%), or medically inoperable (16.3%). Neoadjuvant chemotherapy was delivered to 84% of patients and all received A-SMART to a median 50Gy (range, 40-50Gy) in 5 fractions. 1Year LRC, PFS, and OS were 88.9%, 53.8%, and 78.9%, respectively. Nine patients (18%) had resection after A-SMART and benefited from PFS improvement (26 vs 6months, P = .01). ECOG PS <2 was the only predictor of improved OS on multivariate analysis. Acute and late grade 3 + toxicity rates were 8.2% and 4.1%, respectively. A-SMART is associated with encouraging LRC and OS in elderly patients with initially inoperable PDAC. This novel non-invasive treatment strategy appears to be well-tolerated in patients with advanced age and should be considered in this population that has limited treatment options.

Auto-detection of necessity for MRI-guided online adaptive replanning using a machine learning classifier.

MRI-guided adaptive radiation therapy (MRgART), particularly daily online adaptive replanning (OLAR) can substantially improve radiation therapy delivery, however, it can be labor-intensive and time-consuming. Currently, the decision to perform OLAR for a treatment fraction is determined subjectively. In this work, we develop a machine learning algorithm based on structural similarity index measure (SSIM) and change in entropy to quickly and objectively determine whether OLAR is necessary for a daily MRI set. A total of 109 daily MRI sets acquired on a 1.5T MR-Linac during MRgART for 22 pancreatic cancer patients each treated with five fractions were retrospectively analyzed. For each daily MRI set, OLAR and reposition (No-OLAR) plans were created and the superior plan with the daily fraction determined per clinical dose-volume criteria. SSIM and entropy maps were extracted from each daily MRI set, with respect to its reference (e.g., dry-run) MRI in the region enclosed by 50-100% isodose surfaces. A total of six common features were extracted from SSIM maps. Pearson's rank correlation coefficient was utilized to rule out redundant SSIM features. A t-test was used to determine significant SSIM features which were combined with the change in entropy to develop anensemble machine classifier with fivefold cross validation. The performance of the classifier was evaluated using the area under the curve (AUC) of the receiver operating characteristic curve. A machine learning classifier model using two SSIM features (mean and full width at half maximum) and change in entropy was determined to be able to significantly discriminate between No-OLAR and OLAR groups. The obtained machine learning ensemble classifier can predict OLAR necessity with a cross validated AUC of 0.93. Misclassification was found primarily for No-OLAR cases with dosimetric plan quality closely comparable to the corresponding OLAR plans, thus, are not a major practical concern. A machine learning classifier based on simple first-order image features, that is, SSIM features and change in entropy, was developed to determine when OLAR is necessary for a daily MRI set with practical acceptable prediction accuracy. This classifier may be implemented in the MRgART process to automatically and objectively determine if OLAR is required following daily MRI.

Improving Outcomes Prediction Using a Combination of MRI Wavelet Delta-Radiomic Textures and Clinical Biomarkers for MR-Guided Adaptive Radiation Therapy of Pancreatic Cancer

<h3>Purpose/Objective(s)</h3> We have previously reported that wavelet delta radiomic features (WDRF) extracted from daily MRIs can predict treatment response of MRI-guided adaptive radiation therapy (MRgART) for pancreatic cancer. This study aims to investigate whether combining available clinical biomarkers with the WDRF would improve outcome prediction for pancreatic cancer treated with pre-operative MRgART. <h3>Materials/Methods</h3> This method was developed and demonstrated using 90 daily motion averaged MRI sets derived from 4D MRIs acquired during MRgART with concurrent chemotherapy for 18 pancreatic cancer patients treated in 5 SBRT fractions on a 1.5T MR-Linac, along with their CA19-9, CEA, metastatic status, and follow up data. For each daily MRI set, wavelet decompositions were applied a programming environment to provide a robust way for feature extraction and to overcome challenges with inter- and intra-patient MRI intensity variations. 438 WDRF were calculated from different decomposition levels. Good or bad response groups were defined based on pathological response or distant metastasis free progression, and 50% reduction in CA19-9. The time between the end of treatment and end points of death or presence of distant metastasis within 2 years were used to build progression free survival (PFS) model. Spearman correlations were applied to rule out redundant features. T-test and regression models were used to identify independent WDRFs with significant changes that correlate to response and to examine the effect of combining CA19-9, CEA, and WDRFs on response prediction. Concordance statistics was used to measure model effectiveness. A Cox analysis was performed to identify predictors of PFS and test the association of the combinations on PFS correlations. The effect of a decrease in CA19-9 or CEA levels during treatment versus failure of CA19-9 or CEA levels to normalize on PFS was examined. <h3>Results</h3> Spearman correlation showed that 67 WDRFs from different wavelet decomposition were not redundant and can be used for the analysis. 30 WDRF demonstrated opposite trends correlating to response. Of these, 8 WDRFs passed the t-test showing significant changes correlating to response by the 3rd or 4th fraction. CA199 and CEA were each correlated to 2 WDRFs. Increasing CA19-9, CEA levels during treatment were correlated to bad response. Incorporating CEA, CA19-9 with the significant WDRFs increased the concordance statistics from 0.65 to 0.96. The Cox multivariate analysis showed that a treatment related decrease in CA19-9 levels (p=0.0017), and 3 WDRFs (p=0.004-0.013) were independent predictors of PFS. The hazard ratio was reduced from 0.57, p=0.03 for using CA19-9 only to 0.163, p=0.004 for the combination of WDRFs and CA19-9. <h3>Conclusion</h3> Combining wavelet delta-radiomics features extracted from daily treatment MRIs with common clinical biomarkers improves outcome prediction for chemoradiation therapy of pancreatic cancer and may ultimately enable intensification of additional treatment to improve outcomes.

Primary Tumor-Directed MRI-Guided Ablative Radiation Therapy for Patients with Metastatic Pancreatic Adenocarcinoma

<h3>Purpose/Objective(s)</h3> Survival remains poor for patients with metastatic pancreatic cancer and local progression is commonly associated with high rates of morbidity and mortality. MRI-guided adaptive radiation therapy (MRgART) with ablative dose may offer local disease control and improve clinical outcomes for selected metastatic patients with chemotherapy-responsive disease. <h3>Materials/Methods</h3> We performed a multi-institutional retrospective analysis of de novo metastatic pancreatic cancer patients who received high biological effective dose MRgART to their primary disease on a 0.35T MR-Linac in 5 fractions. Local and distant disease recurrence was evaluated by Response Evaluation Criteria in Solid Tumors (RECIST 1.1) or by radiographic evidence of progression with initiation or change in systemic therapy. Toxicity was measured by Common Terminology Criteria for Adverse Events (v5) at routine follow up appointments. Primary clinical outcomes of interest were overall survival (OS) censored at last clinical follow up and freedom from local failure (FFLF) censored at last dedicated abdominal imaging. <h3>Results</h3> From December 2019 until April 2021 a total of 22 metastatic patients with a median age of 66 years and a slight male predominance (59%) were treated with induction FOLFIRINOX (64%, median 13 cycles, 224 days) or Gemcitabine/nab-Paclitaxel (36%, median 7 cycles, 220 days) followed by ablative dose MRgART to a median dose of 50 Gy (range 40-50). Median follow up from MRgART was 13.0 months. The median OS from diagnosis and from radiation were 23.9 months (95% CI 18.1-30.0) and 11.6 months (95% CI 9.6-13.6), respectively. The FFLF was 77% and 2 patients (9%) were documented with death due to local disease progression. At last follow up, 19 patients (86%) had experienced distant progression and 13 (69%) had died. A single patient underwent a margin negative distal pancreatectomy and metastasectomy of a treated retroperitoneal tumor implant and is without recurrence at 16 months post-diagnosis. Two patients experienced grade 3+ toxicity, one patient with duodenal involvement at diagnosis and on anti-coagulation during treatment experienced a gastrointestinal bleed at 69 days post-MRgART and portal vein pseudoaneurysm 168 days post-MRgART. The second patient developed gastric outlet obstruction 430 days post-MRgART. There was no grade 5 toxicity. <h3>Conclusion</h3> Some metastatic pancreatic cancer patients with stable or chemotherapy-responsive disease appear to benefit from aggressive local therapy which may mitigate morbidity of primary tumor progression. Further investigation into the optimal patient selection for this treatment is warranted.

Safety and Efficacy of Stereotactic MRI-Guided Adaptive Radiation Therapy for Localized Kidney Cancer

<h3>Purpose/Objective(s)</h3> To assess the safety, feasibility as well as short term oncologic and renal functional outcomes of stereotactic MRI guided adaptive radiotherapy (SMART) for localized kidney cancer. <h3>Materials/Methods</h3> Single institution experience of patients undergoing MRI guided radiation therapy with daily plan adaptation for localized kidney cancer from 2020-2021. Patients were treated to a dose of 4000cGy over 5 fractions utilizing a 0.35T MR-Linac. Prior to each fraction, a pre-treatment MRI was obtained to allow for re-contouring of tumor and organs-at-risk. Patients were treated using an MR guided gated breath hold technique. Adverse events were classified according to the <i>Common Terminology Criteria for Adverse Events</i> (CTCAE) guideline version 5.0. <h3>Results</h3> Our cohort includes 12 patients with median age 75 years, median tumor diameter 4.75 cm, and median RENAL Nephrometry score 7. In terms of staging, 5 were cT1a, 5 were cT1b, 1 was cT2a, and 1 was cT3a. In total, 67.7% of patients had baseline Stage III CKD or worse renal function, and median Charlson Comorbidity index was 8. Histologic subtypes were 58.3% RCC and 25% papillary RCC. There was a total of 11 adverse events (AE) within 90 days: 10 were grade 1 for fatigue, nausea, vomiting, and diarrhea; there was one grade 2 AE for nausea (Table 1). No grade 3 or higher events were observed. The median baseline eGFR was 56.5 mL/min/1.73m², with median change in eGFR of -10 mL/min/1.73m² and the median change in tumor diameter was -0.5 cm at 7.5 months of follow-up, with no local or distant progression. No SMART-related hospital admissions occurred (Figure 1). <h3>Conclusion</h3> SMART is a treatment option for localized RCC that is associated with minimal toxicity, limited decline in renal function, and high rates of local tumor control. For patients who are not surgical candidates, SMART may be an effective therapy for localized kidney cancer.

Evaluation of Practice Patterns and Outcomes after Implementing SMART for Pancreatic Cancer

<h3>Purpose/Objective(s)</h3> Stereotactic MRI-guided adaptive radiation therapy (SMART) enables safe dose escalation for locally advanced, borderline resectable, and medically inoperable pancreatic cancer and has shown favorable toxicity and survival outcomes. In late 2018, our institution commissioned SMART for these patients, making it available to all patient referrals. We wanted to review changes in our patient population and differences in clinical outcomes for patients before and after the implementation of SMART. <h3>Materials/Methods</h3> In this IRB approved analysis, we retrospectively reviewed 167 consecutive patients from 2015-2021 with locally advanced, borderline resectable, or medically unresectable pancreatic cancer who were treated with radiation therapy. Chemoradiation (chemoRT) was defined as any 28-30 fraction radiation regimen that included concurrent chemotherapy. SMART was defined as 50 Gy over 5 consecutive daily fractions without concurrent chemotherapy. Baseline patient characteristics were compared between groups. Overall survival (OS) was evaluated by Kaplan-Meier (KM) and log-rank test. Univariate (UVA) and multivariate analysis (MVA) were also performed on multiple treatment variables. <h3>Results</h3> Of the patients included, 58 received chemoRT (2015-2018) and 109 received SMART (2018-2021). Median follow up from time of diagnosis for the chemoRT and SMART cohorts were 53.7 months and 29.2 months respectively. Cohorts did not have significant differences in age, gender, race, T or N staging, rates of surgery or surgical margin status. Patients receiving SMART had overall worse performance (p = 0.011) including a lower percentage of PS 0 patients (22.9% vs 44.8%) and a higher percentage of PS 2+ patients (34% vs 15.5%). Similarly, the SMART patients did numerically more often have locally advanced (50% vs 43%) and medically inoperable (26% vs 21%) disease (p = 0.294). The SMART cohort did have longer neoadjuvant chemotherapy with mean of 3.5 months vs 2.3 months in the chemoRT cohort (p = 0.002). There was no OS difference between each group when measured from diagnosis (p=0.79) or from first day of radiation (p=0.2). Median survival in the chemoRT and SMART groups was 18.7 vs 17.4 months from diagnosis. When including only PS 0-1 patients, the median survival in the chemoRT and SMART groups was 18.8 vs 22.3 months (p=0.37). There was also no difference in locoregional control, distant control, or progression free survival using KM. On MVA positive prognostic factors for OS from diagnosis included ECOG <2 (HR 0.54, p=0.015), increasing months of neoadjuvant chemo (HR 0.88, p=0.004) and pancreatectomy (HR 0.14, p <0.001). <h3>Conclusion</h3> Despite the fact that the patient cohort receiving radiation therapy per the SMART approach had poorer performance status compared with chemoRT, OS was not significantly different. The multidisciplinary team was highly supportive of SMART with increased patients being treated.

Predicting necessity of daily online adaptive replanning based on wavelet image features for MRI guided adaptive radiation therapy

PurposeOnline adaptive replanning (OLAR) is generally labor-intensive and time-consuming during MRI-guided adaptive radiation therapy (MRgART). This work aims to develop a method to determine OLAR necessity during MRgART. MethodsA machine learning classifier was developed to predict OLAR necessity based on wavelet multiscale texture features extracted from daily MRIs and was trained and tested with data from 119 daily MRI datasets acquired during MRgART for 24 pancreatic cancer patients treated on a 1.5 T MR-Linac. Spearman correlations, interclass correlation (ICC), coefficient of variance (COV), t-test (p < 0.05), self-organized map (SOM) and maximum stable extremal region (MSER) algorithm were used to determine candidate features, which were used to build the prediction models using Bayesian classifiers. The model performance was judged using the AUC of the ROC curve. ResultsSpearman correlation identified 123 features that were not redundant (r < 0.9). Of them 82 showed high ICC for repositioning > 0.6, 67 had a COV greater than 9% for OLAR. Among the 38 features passed the t-test, 25 passed the SOM and 12 passed the MSER. These final 12 features were used to build the classifier model. The combination of 2–3 features at a time was used to build the classifier models. The best performing model was a 3-feature combination, which can predict OLAR necessity with a CV-AUC of 0.98. ConclusionsA machine learning classifier model based on the wavelet features extracted from daily MRI for pancreatic cancer was developed to automatically and objectively determine if OLAR is necessary for a treatment fraction avoiding unnecessary effort during MRgART.

Evolution of Radiation Therapy in Pancreas Cancer Management toward MRI-Guided Adaptive Radiation Therapy

Pancreas cancer has a poor prognosis despite aggressive treatment and is the fourth leading cause of cancer death in the United States. At diagnosis, most patients have either metastatic or locally advanced disease. In this article, we review the evolution of treatments in locally advanced pancreas cancer (LAPC) and discuss the various radiation therapy fractionation schemes. Furthermore, we examine the data supporting dose escalation and the delivery of ablative biologically effective doses in the setting of LAPC. Finally, we review the role of MRI-guided radiation therapy in escalating dose while sparing organs at risk in the era of stereotactic magnetic resonance-guided adaptive radiation therapy.

High-dose MRI-Guided Adaptive Radiation Therapy for Locally Recurrent Pancreatic Cancer

Introduction: The prognosis of recurrent pancreatic cancer is poor even after curative resection. There have been no reports of MRI-guided radiation therapy for locally recurrent pancreatic cancer after curative resection and chemotherapy. Case Presentation: A 66-year-old man with pancreatic cancer was referred to our institution for local recurrence after failure of surgical resection and second-line chemotherapy. He did not want to undergo further chemotherapy, so high-dose MRI-guided adaptive radiation therapy was performed in daily doses of 2.5 Gy to a total dose of 70 Gy over a period of 5.5 weeks. Three months after radiation therapy, the recurrent tumors disappeared and his CA19-9 level was within normal range without chemotherapy. There were no adverse events during treatment and three months of follow-up. Conclusions: High-dose MRI-guided adaptive radiation therapy may be safe and useful for locally recurrent pancreatic cancer.

Development and implementation of an automatic air delineation technique for MRI-guided adaptive radiation therapy

Objective. Auto-delineation of air regions on daily MRI for MR-guided online adaptive radiotherapy (MRgOART) of abdominal tumors is challenging since the air packets occur randomly and their MR intensities can be similar to some other tissue types. This work reports a new method to auto-delineate air regions on MRI. Approach. The proposed method (named DIFF method) consists of (1) generating a combined volume V comb , which is a union of the air-containing organs on a reference MR image offline, (2) transferring V comb from the reference MR to a daily MR via DIR, (3) combining the transferred V comb with a region of high DIR inaccuracy, and (4) applying a threshold to the obtained final combined volume to generate the air volumes. The high DIR inaccuracy region was calculated from the absolute difference between the deformed daily and the reference images. This method was tested on 36 abdominal daily MRI sets acquired from 7 patients on a 1.5 T MR-Linac. The performance of DIFF was compared with alternative auto-air generation methods that (1) does not account for DIR inaccuracies, and (2) uses rigid registration instead of DIR. Main results. The results show that the proposed DIFF method can be fully automated and can be executed within 25 s. The Dice similarity coefficient of manual and DIFF auto-generated air contours was >92% for all cases, while it was 90% for the alternative auto-delineation methods. Dosimetrically, the auto-generated air regions using DIFF resulted in practically identical DVHs as those generated by using manual air contours. Significance. The DIFF method is robust and accurate and can be implemented to automatically consider the inter- and intra- fractional air volume variations during MRgOART for abdominal tumors. The use of DIFF method improves dosimetric accuracy as compared to other methods, especially beneficial for the patients with large daily abdominal air volume variations.

Sequential treatment with gemcitabine/nab-paclitaxel (GA) and FOLFIRINOX (FFX) followed by stereotactic MRI-guided adaptive radiation therapy (SMART) in patients with locally advanced pancreatic cancer (LAPC): GABRINOX-ART phase 2, multicenter trial.

TPS4191 Background: LAPC represents a major challenge with no standardized chemotherapy (CT) and radiotherapy (RT) treatment. Phase 2 studies (LAPACT/NEOLAP) indicated efficacy of FFX and GA, although addition of conventionally fractionated RT remains controversial. Phase 3 LAP07 trial obtained a reduction of progression free survival (PFS), albeit with no overall survival (OS) advantage. Since in metastatic pancreatic cancer we recently attained with GA/FFX sequential combination (GABRINOX) high objective response rate and promising OS with acceptable toxicity and no limiting neurotoxicity, we proposed in LAPC to complement GABRINOX with SMART, recently recognized beneficial in pancreatic tumors by a retrospective multicenter study (OS at 2 years) and our prospective registry study (dosimetric benefit of adaptation). In a first step (SEQ1), we will evaluate GABRINOX efficacy and select patients without progression for a second step (SEQ2), to evaluate feasibility and tolerance in patients without disease progression after SEQ1. Secondarily we will evaluate CT tolerance (SEQ1), acute toxicities and dosimetric results (SEQ2) and for both SEQ1+2, late toxicities, response to treatment, PFS, OS and quality of life (QoL). Methods: Naive patients with confirmed non-metastatic unresectable adenocarcinoma by centralized reading (WHO 0/1) and adequate organ function will receive in SEQ1 two cycles of GABRINOX, GA (1000 mg/m2, 125 mg/m2) on days 1, 8, and 15 followed by FFX on day 29 and 43. After 3-4 weeks, patients without progression or unacceptable toxicity will benefit from SMART (5 fractions of 10 Gy/day over 5 consecutive days). Specific dummy-run, contouring quality assurance and dosimetric plans will precede post-treatment monitoring every 6 weeks for 6 months for non-progressive patients and then every 2 months until progression: radiological assessment, biological markers (circulating tumor DNA) and QoL evaluation. Co-primary endpoints include success of SEQ1 (non-progression at 4 months, RECIST v1.1) and that of SEQ2 as absence of acute digestive non-toxicity rate &gt; grade 3 (NCI-CTCAE v5.0) within 90 days. Based on Fleming design with maximal inefficacy (p0) of 70% and 90% (α = 2.5% and β = 5%) we need 98 and 70 patients (SEQ1 and SEQ2), and total of 103 cases considering those entering in SEQ 2 (70%) and non-evaluable patients. Success rate, toxicities (by treatment sequences) and safety (System Organ Class) by patient and cycle will be considered while dosimetry will be correlated with gastro-intestinal toxicities. Median follow-up, OS and PFS will be expressed as medians and rates with 95% CI while QoL will be explored by QLQ-C30 and QLQ-PAN26 analyses using the time to definitive deterioration. From 2021, we included 5 patients (NCT04570943). Clinical trial information: NCT04570943.

Automatic Contour Refinement for Deep Learning Auto-segmentation of Complex Organs in MRI-guided Adaptive Radiation Therapy

PurposeFast and accurate auto-segmentation on daily images is essential for magnetic resonance imaging (MRI)–guided adaptive radiation therapy (ART). However, the state-of-the-art auto-segmentation based on deep learning still has limited success, particularly for complex structures in the abdomen. This study aimed to develop an automatic contour refinement (ACR) process to quickly correct for unacceptable auto-segmented contours. Methods and MaterialsAn improved level set–based active contour model (ACM) was implemented for the ACR process and was tested on the deep learning–based auto-segmentation of 80 abdominal MRI sets along with their ground truth contours. The performance of the ACR process was evaluated using 4 contour accuracy metrics: the Dice similarity coefficient (DSC), mean distance to agreement (MDA), surface DSC, and added path length (APL) on the auto-segmented contours of the small bowel, large bowel, combined bowels, pancreas, duodenum, and stomach. ResultsA portion (3%-39%) of the corrected contours became practically acceptable per the American Association of Physicists in Medicine Task Group 132 (TG-132) recommendation (DSC >0.8 and MDA <3 mm). The best correction performance was seen in the combined bowels, where for the contours with major errors (initial DSC <0.5 or MDA >8 mm), the mean DSC increased from 0.34 to 0.59, the mean MDA decreased from 7.02 mm to 5.23 mm, and the APL reduced by almost 20 mm, whereas for the contours with minor errors, the mean DSC increased from 0.72 to 0.79, the mean MDA decreased from 3.35 mm to 3.29 mm, and more than one-third (39%) of the ACR contours became clinically acceptable. The execution time for the ACR process on one subregion was less than 2 seconds using an NVIDIA GTX 1060 GPU. ConclusionsThe ACR process implemented based on the ACM was able to quickly correct for some inaccurate contours produced from MRI-based deep learning auto-segmentation of complex abdominal anatomy. The ACR method may be integrated into the auto-segmentation step to accelerate the process of MRI-guided ART.

Automatically Determining Necessity of Online Adaptive Replanning Based on MRI Wavelet Multiscale Texture Features for MRI-Guided Adaptive Radiation Therapy

MRI-guided adaptive radiation therapy (MRgART), particularly daily online adaptive replanning (OLAR) as enabled with MR-LINAC, can substantially improve radiation therapy delivery. However, OLAR is labor-intensive and time-consuming. OLAR is unnecessary for all treatment fractions and being able to objectively and automatically determine when it is needed is desirable. In this study, we investigate whether wavelet multiscale texture features extracted from daily MRI can be used to automatically and objectively identify when OLAR is needed using a machine learning approach.The method was developed and demonstrated using 70 daily motion average MRI sets acquired on a 1.5T MR-LINAC using a balanced turbo field echo sequence during MRgART for 14 pancreatic cancer patients each treated with 5 fractions. For each daily MRI set, OLAR and repositioning (i.e., adapt to position) plans were created and judged per clinical dose-volume constrains by experienced MRL physicists. Multiscale texture features were extracted from the region enclosed by 50 and 100% reference isodose surfaces for different wavelet decomposition levels. Spearman correlations were utilized to rule-out redundant features. Inter-class correlation (ICC), coefficient of variance (COV), and t-test (P < 0.05) were used to determine significant features. Mahalonobis distance classifier with a leave one out cross-validation method was used to develop prediction models based on the identified features and their combinations. The performance of the model to predict whether OLAR is necessary for a given daily MRI set was measured using the AUC of the ROC curve.Spearman correlation identified 123 features from different wavelet decomposition levels that were not redundant (r < 0.9). Of these features, 82 showed high ICC for repositioning, 67 had a COV greater than 9% for OLAR and lower than 5% for repositioning. In addition, 38 of the 67 features passed the t-test with P value < 0.05. Best performing model was a 3-feature combination that can predict OLAR necessity with an AUC of 0.95. It was found that misclassification was attributed to some cases with very close dosimetric gain between repositioning and OLAR and hence a lower discrimination ability for this subgroup.The necessity of OLAR can be automatically determined using a machine learning model based on the wavelet multiscale texture features from the daily MRI immediately after its acquisition, avoiding unnecessary effort during MRgART.

Initial Outcomes And Treatment-Associated Toxicities Using Stereotactic MRI-Guided Adaptive Radiation Therapy For Pancreatic Cancer Patients Treated At A Single Institution

High dose stereotactic body radiotherapy (SBRT) may offer improved outcomes for pancreatic cancer patients, but treatment delivery remains a challenge due to tumor proximity to organs-at-risk (OAR) and daily anatomic variability. Stereotactic MRI-guided adaptive radiation therapy (SMART) with real-time tumor tracking allows for improved dose delivery and OAR sparing by accounting for anatomic changes with each treatment. This study reports initial outcomes, treatment-associated toxicities, and dosimetric outcomes in locally advanced, metastatic, and/or recurrent pancreatic cancer patients treated using SMART. We performed a retrospective analysis of locally advanced, metastatic, and/or recurrent pancreatic cancer patients who underwent adaptive planning with MRI-guided radiation treatments. Charts were reviewed for follow-up, imaging, and laboratory values, including CA19-9 and absolute lymphocyte count. A plan was generated for each patient after initial CT and MR-simulation. At each treatment, MR imaging was obtained for the daily anatomy and the initial (unadapted) plan was compared to obtain a predicted dose to target volumes and OARs using a prescribed dose ranging from 30-50 Gy in 3-5 fractions. If constraints were not met or coverage was inadequate, an adaptive plan was generated based on the anatomy imaged on the day of treatment. From June 2018 to February 2020, a total of 23 patients were analyzed. With a median follow up of 5 months, we observed 15/19 (79%) with local control, 12/19 (63%) distal metastatic control, and a median overall survival of 19 months from cancer diagnosis. We observed no grade 3 or higher treatment related toxicities. Two patients were found to have an absolute lymphocyte count less than 0.5 x 103 cells/μl within six months of radiation, and 11/17 (65%) had lower CA19-9 six months after radiation treatment compared to before treatment. Of the 23 patients treated, there were 71 adaptive plans available for analysis. Compared to unadapted plans, adaptive fractions had improved coverage of PTV (84.94% vs. 73.36% at 100% of prescribed dose) and CTV (94.62% vs 88.08% at 95% of prescribed dose). Unadapted plans were more likely to exceed OAR constraints with 63 of 71 (89%) unadapted fractions predicted to violate prescribed duodenal, stomach, or small bowel constraints. Moreover, 34/71 (48%) of unadapted fractions violated constraints for 2 of 3 abdominal OARs and 10/71 (14%) violated constraints for all 3 compared to none for adapted fractions. Stereotactic adaptive MRI-guided radiotherapy demonstrated minimal treatment-related toxicities, enhanced tumor volume coverage, and improved adherence to OAR constraints in pancreatic cancer patients. Additional follow up is needed to further assess outcomes of these patients.