Pancreatic cancer (pancreatic ductal adenocarcinoma; PDAC) remains a lethal malignancy with a 5-year survival rate of only 11%.1 Completion of multimodality therapy including systemic chemotherapy, radiation therapy (RT), and surgical resection provides the best chance for long-term survival.2, 3, 4 Nevertheless, only 20% of patients are eligible for resection, as approximately 40% of patients are diagnosed with locally advanced pancreatic cancer (LAPC), and 40% present with metastasized disease.5 In addition, some patients with a resectable tumor are not fit enough to undergo complex pancreatic surgery or are unwilling to do so, particularly among frail elderly.6,7 Of patients who undergo surgery, almost all develop disease recurrence within a few years.8,9 In one-third of these patients this comprises isolated local recurrence in the pancreatic remnant or surgical bed.10

Although PDAC is recognized for its aggressive systemic spread, highlighting the importance of systemic therapy, local destructive tumor growth is known to cause substantial morbidity.11 This includes pain and gastrointestinal (GI) obstruction, leading to malnutrition, and degrading the patients’ quality of life (QoL). Also, local tumor growth has been shown to be the cause of death in 30% of PDAC patients.12 This emphasizes the need for effective local therapies in patients not amenable to surgical resection. Ideally, such therapies would have a low incidence of toxicity and be minimally-invasive. If such therapies could improve local PDAC control, this could potentially prolong survival and preserve QoL. Local tumor treatments have become more important over the last decade as more effective systemic therapies have developed.13 This trend will continue in the upcoming years as therapies continue to improve.

RT is a locally ablative anticancer therapy that has shown its effectiveness in many cancer types.14 A principal limitation of effective RT for PDAC has been poor visibility of pancreatic tumors and surrounding organs at risk (OARs) using conventional computed tomography (CT) guided techniques. Moreover, these targets move significantly due to peristalsis and breathing.15,16 The toxicity risk that comes with these issues has impeded the delivery of ablative irradiation doses to pancreatic malignancies that match a critical biologically equivalent dose (BED10) of at least 100 Gy.17 The clinical introduction of online adaptive magnetic resonance (MR) guided systems in 2017 has substantial implications for the application of RT in patients with pancreatic tumors.18,19 These systems have allowed dose escalation towards an effective BED, theoretically improving clinical efficacy while minimizing the risk of serious toxicity.20,21

This review aims to provide a comprehensive overview of the current literature on the use and clinical outcomes of MR-guided RT for treatment of PDAC.

A total of 16 papers were included in this review (Table 1).22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37 These were mainly single-center (n = 13) studies of retrospective character (n = 10), in which the number of patients treated for PDAC ranged from 3 to 63. Four articles reported different clinical outcomes of an overlapping patient cohort.24,28,31,32 Two other studies also described an overlapping study population.23,29 Eight studies included patients with localized (nonmetastatic) primary PDAC only, including LAPC, borderline resectable, and resectable PDAC.22,24,25,27,28,32,33,36 Two studies additionally included patients with metastatic PDAC.23,26 The remaining 6 studies described a more heterogeneous study population, also including other primary tumor types, such as neuroendocrine or periampullary tumors, pancreatic metastasis, (upper) abdominal and/or pelvic tumors in general, or all patients treated on a 1.5T MR-Linac.29, 30, 31,34,35,37 All but 1 study reported on patients who were RT naïve. The remaining study included patients with lesions in the abdomen or pelvis treated with prior RT to a median of 50 (range 30-59) Gy in 25 (range 5-28) fractions, after a median of 26.8 (range 7.6-59.0) months.31

One study consisted of a phase I dose-escalation trial to determine the maximum tolerated dose of ablative hypofractionated radiation with full-dose gemcitabine/nab-paclitaxel in patients (Table 1).27 Nine studies investigated ablative RT with a median of 50 Gy in 5 fractions (BED10=100 Gy),22, 23, 24, 25, 26,28,32,33,37 whilst 40 Gy in 5 fractions was applied in 1 study.31 Four studies described various dose and fractionation schedules,29,30,35,36 and the median dose given to pancreas lesions was unknown in 1 study.34 Most studies (n = 13) reported that the majority of patients received chemotherapy (range 20%-100%), most frequently FOLFIRINOX (range 12%-73%), gemcitabine/nab-paclitaxel (range 2%-61%) or gemcitabine (range 2%-11%).22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33,36 In 1 study, all patients underwent post-RT surgery,25 and a small part of patients in 9 other studies as well.22, 23, 24,27, 28, 29,32,33,36

In 4 studies, a 1.5T-MR Linac was used,26,30,34,35 whilst the remaining 12 studies used a 0.35T-MR Linac system. Technical details, including GTV and PTV description and bowel constraints, are summarized in Table 2. A range of immobilization methods including breath hold techniques and abdominal compression were reported in all but 3 studies. On-table adaptive replanning was performed in 27%-100% of treatment fractions.22, 23, 24, 25, 26,28, 29, 30, 31, 32,36,37

Toxicity outcomes were described in 15 studies (Table 3).22,23,25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37 Acute or late grade 3 or higher toxicity was reported in 6 and 7 of the cohort studies, in 3%-8% and 3%-10% of patients, respectively.22,23,27,28,32, 33, 34, 35 The primary outcome in the dose-escalation trial was the maximum tolerated dose of concurrent gemcitabine/nab-paclitaxel and radiation.27 One patient experienced a dose limiting toxicity (DLT) of acute cholecystitis (unlikely related to treatment) at 32.4 of 45 Gy in 25 fractions (dose level 2) with nonadaptive MR-guided RT, at which point radiation was discontinued, resulting in a DLT rate of 14% (3%-25%). Acute grade 3 or higher GI toxicity was reported in 5/26 patients (19%) and late grade 3 or higher toxicities in 4/26 patients (15%), all in organs that abutted the primary tumor.27 In the remaining studies, no high-grade toxicity occurred.

Pathological outcomes were reported in 9 studies, including resection margin status in 9 studies (range 67%-100%) and lymph node status in 4 studies (range 80%-100%) (Table 3).22,23,25,27, 28, 29,32,33,36 Histopathologic tumor response was documented in 8 studies: the average pathologic effect ranged from 64% to 100% and pathological complete response occurred in 0%-33% of patients.22,23,25,27, 28, 29,32,33 Bryant et al. performed a multivariable ordinal logistic regression including the time from A-SMART to surgery (continuous), administration of neoadjuvant therapy (categorical) and RT dose (continuous), and showed that increased time interval from A-SMART was significantly associated with decreased tumor regression grade (P = 0.007).25

Other outcomes included causes of death after RT, feasibility and safety of post-RT PDAC resection, and presence of pain.

Chuong et al.24 presented causes of death 42/62 patients (68%) after a median follow-up time of 18.6 months from diagnosis and 11.0 months from A-SMART. Distant metastasis was the most common cause of PDAC-related death (n = 19; 45%), particularly intra-abdominal metastasis of the liver and peritoneum (84%). Other causes of death included locoregional progression in 3 patients (7%), of whom 2 also had distant metastasis at time of death, cancer-related cachexia/malnutrition in 5 patients (12%) and complications related to surgery or irreversible electroporation (IRE) in 4 patients (10%). Other reasons not related to cancer progression attributed to patient death in 5 patients (12%) and the cause of death was unknown for the remaining 6 patients (14%).

Bryant et al.25 focused on the feasibility and safety of A-SMART followed by PDAC resection, including 26 patients who underwent definitive surgery with curative intent (69% Whipple, 23% distal, 8% total pancreatectomy) after a median of 50 (range 37-115) days. Of 128 patients with pancreatic cancer treated with A-SMART, none were prevented from surgery due to RT-related toxicities. The median intraoperative time was 7h10 (range 3h57-12h12) and the median postoperative hospital stay was 8 (range 4-13) days. No perioperative deaths occurred within 90 days from resection (0%). Postoperative complications were grade 2-3 infection/abscess formation in 4 patients (15%), grade 1-2 chyle leak in 4 patients (15%), grade 1 pancreatic anastomosis leakage in 2 patients (8%), and grade 2 delayed gastric emptying in one patient (4%). Hemorrhage occurred in two patients (8%), which both were grade 4. Daamen et al.30 found that in 13 patients who reported pain prior to treatment, 11 patients (85%) experienced pain alleviated post-RT.

Survival outcomes were presented in 13 articles (Table 4).22, 23, 24, 25, 26, 27, 28, 29,31, 32, 33,36,37 Due to substantial heterogeneity in study populations, applied pre- and post-RT therapies, presented survival estimates and defined time intervals to calculate survival outcomes, results could not be pooled. Median LC after RT in 5 studies ranged from 7.4 months to not reached within a median follow-up of 11.0 months (IQR 1.5-36.0 months).22,26,28,31,32 One- and 2-year LC rates after treatment ranged between 79%-98% in 9 studies and between 57% and 85% in 4 studies, respectively.22,23,25, 26, 27, 28, 29,31,32,36 Median PFS after treatment ranged from 7.0 to 21.0 months in 6 studies, and 1- and 2-year PFS ranged between 32% and 52% in 6 studies and between 21%-25% in 2 studies, respectively.22,26, 27, 28,31,32 Median DMFS from RT was 3.0-10.5 months in 4 studies, with 1-year DMFS after treatment ranging from 24% to 74% in 6 studies.22,23,27,29,32,36 Median OS from treatment was described in 8 studies, ranging from 9.8 to 18.0 months.22,23,26, 27, 28, 29,31,32 One-year OS after treatment ranged between 54% and 80%, and 2-year OS between 28% and 57%.22,23,26, 27, 28, 29,31,32,36 Chuong et al.32 also reported cause-specific survival (CSS), with a median of 9.8 months from RT and a 1-year CSS of 78% (95% CI 73%-85%).

Tringale et al.26 additionally calculated the cumulative incidence of local failure (LF) and distant metastases (DM) at 6-months and 1-year from the first RT fraction, with death as a competing risk. The cumulative incidence of LF and DM was 7% (95% CI 1%-20%) and 30% (95% CI 15%-47%) at 6-months, and 19% (95% CI 7%-37%) and 47% (95% CI 27%-66%) at 1-year, respectively.

Rudra et al.36 found that high-dose radiation (BED10 >70) and duration of induction CHT were correlated with OS in univariate analysis (HR 0.44 [95% CI 0.21-0.94]; P = 0.03 and HR 0.84 [95% CI 0.72-0.98]; P = 0.03, respectively), although multivariable analysis did not confirm this (HR 0.56 [95% CI 0.25-1.26]; P = 0.16 and HR 0.87 [95% CI 0.74-1.03]; P = 0.11). Bryant et al. found no significant association between RT dose (50 Gy vs <50 Gy) and OS (HR 2.81 [95% CI 0.79-9.98]; P = 0.11) or PFS (HR 3.30 [95% CI 0.97-11.23]; P = 0.06) in univariate analysis.22 Kim et al.27 showed no difference in local progression-free survival (P = 0.5), DMFS (P = 0.7), PFS (P = 0.7) and OS (P > 0.05) for patients treated in dose level 5, as compared with dose levels 1-4. In addition, they reported a median OS of 34.1 (95% CI 13.6-54.1) and 43.0 (95% CI 8.0-not reached) for patients with Eastern Collaborative Oncology Group (ECOG) 0 and carbohydrate antigen (CA) 19-9 <90 U/mL, respectively.27

Two studies mentioned QoL outcomes, measured using European Organization for Research and Treatment of Cancer Core Quality of Life Questionnaire (EORTC) C30 and PAN26.27,37 One of these studies showed a significant decrease in physical functioning at 3 month follow-up, as compared with baseline (P = 0.02).27 This study also showed that hepatic symptoms improved at 3 months follow-up (P < 0.01), although bowel symptoms and treatment-related symptoms decreased as compared with baseline (P = 0.03 and P < 0.01, respectively).27 No other differences in QoL during treatment and the acute post-therapy window were found.27,37