Background: Neoadjuvant therapy response correlates with survival in multiple gastrointestinal malignancies. To potentially augment neoadjuvant response for pancreas adenocarcinoma, we intensified treatment with stereotactic body radiotherapy (SBRT) following multi-agent chemotherapy. Using this regimen, we analyzed whether the College of American Pathology (CAP) tumor regression grade (TRG) at pancreatectomy correlated with established response biomarkers and survival. Materials and methods: We identified borderline resectable (BRPC) and locally advanced (LAPC) pancreatic cancer patients treated according to our institutional clinical pathway who underwent surgical resection with reported TRG (n = 81, median follow-up after surgery 24.2 months). Patients had baseline CA19-9, computed tomography (CT), endoscopic ultrasound, and FDG positron emission tomography (PET)/CT then underwent multi-agent chemotherapy (79% with three cycles of gemcitabine, docetaxel and capecitabine) followed by 5-fraction SBRT. They then underwent restaging CT, PET/CT and CA19-9. Overall (OS) and progression-free (PFS) survival were estimated and compared by Kaplan–Meier and log-rank methods. Univariate ordinal logistic regression correlated TRG with baseline, restaging and change in CA19-9 and the PET maximum standardized uptake value (SUVmax). Results: Restaging level and decrease in CA19-9 correlated with improved TRG (p = .02 for both) as did restaging SUVmax (p < .01), yet there was no TRG correlation with decrease in SUVmax (p = .10) or CT response (p = .30). The TRG groups had similar OS and PFS except the TRG 0 (complete response) group. Compared to partial response levels (TRG 1-3, median OS 33.9 months, median PFS 13.0 months), the six (7%) patients with TRG 0 had no deaths (p = .05) and only one progression (p = .03). A group of 10 (12%) TRG 1 patients with only residual isolated tumor cells had similar outcomes to the other TRG 1-3 patients. Conclusion: Pre-operative PET-CT and CA19-9 response correlate with histopathologic tumor regression. Patients with complete pathologic response have superior outcomes, suggesting a rationale for intensification and personalization of neoadjuvant therapy in BRPC and LAPC.

Background: Sarcopenia is an independent predictor of clinical outcomes in multiple gastrointestinal cancers. Total psoas area (TPA), as measured on a single cross-sectional CT image at the L4 vertebral body level, has been correlated with sarcopenia. We sought to evaluate whether TPA was predictive of acute grade ≥3 toxicity, pathologic response, and overall survival in patients with locally advanced esophageal cancer receiving tri-modality therapy. Methods: An institutional database of esophageal cancer patients treated with neoadjuvant chemoradiation followed by surgery was queried. Of 77 patients treated from 2008 to 2012 with intensity modulated radiation therapy (IMRT) and image guided radiation therapy (IGRT), 56 patients were eligible based on having CT imaging that included the L4 vertebral body. The L4 vertebra was identified on axial CT and the psoas muscle was manually contoured bilaterally to determine the skeletal muscle index. Sarcopenia was defined by the presence of the psoas area less than the median of the cohort. Acute toxicity was defined as within 3 months of radiotherapy based on Common Terminology Criteria for Adverse Events. ROC curve, logistic regression, and Kaplan Meier estimates were used when appropriate. Results: Sarcopenia was associated with increased acute grade ≥3 toxicity from chemoradiation by ROC analysis using a cut off of 841.5 mm²/m² (P=0.003, AUC 0.709, sensitivity 60.9%, specificity 78.8%) and logistic regression (P=0.002). Patients with TPA <841.5 mm²/m² were 5.78 times more likely to develop grade 3 or higher toxicity (P=0.004). Sarcopenia did not predict a difference in overall survival (P=0.217) and was not significant for pathologic complete response or favorable pathologic response (TRG 0/1). Conclusions: In our cohort of patients, sarcopenia was associated with a significant increase in acute grade ≥3 toxicity with chemoradiation, suggesting a potential role for neoadjuvant patient selection strategies. There was no difference in pathologic response or overall survival.

Background: The Child-Pugh (CP) class is a commonly used scoring system to measure liver function in patients with hepatocellular carcinoma (HCC). We correlate the Albumin-Bilirubin (ALBI) grading system and CP to overall survival in our HCC patients receiving radioembolization. Methods: We retrospectively evaluated patients who received radioembolization for HCC between the years 2009–2014. We evaluated the albumin and bilirubin levels in our patients prior to receiving their first (n=124) radioembolization. The ALBI grades were calculated from these data with the formula (log₁₀ bilirubin ×0.66) + (albumin × −0.085) and correlated to outcomes using Mantel-Cox Log analysis. These statistical comparisons were duplicated with CP classes. Results: Median survival differences between CP class A and B and between ALBI grade 1 and 2 were 4.7 and 9.9 months, respectively. A subset of ALBI grades 1 and 2 were identified within our CP class A patients with a median survival difference of 9.9 months. Conclusions: ALBI is a more sensitive marker of liver function than CP in the setting of mild dysfunction. Using ALBI, we identified a subset of patients that have significantly better outcomes from Y-90 radioembolization than previously identified with CP.

Background: Total psoas area (TPA), a marker of sarcopenia, has been used as an independent predictor of clinical outcomes in gastrointestinal (GI) cancers as a proxy for frailty and nutritional status. Our study aimed to evaluate whether TPA, in contrast to traditional measurements of nutrition like body mass index (BMI) and body surface area (BSA), was predictive of outcomes in borderline resectable pancreatic cancer (BRPC) and locally advanced pancreatic cancer (LAPC) patients receiving stereotactic body radiation therapy (SBRT). Methods: Retrospective analysis of an institutional review board approved database of 222 BRPC and LAPC treated with SBRT from 2009–2016 yielded 183 patients that met our selection criteria of pre-SBRT computed tomography (CT) imaging with an identifiable L4 vertebra. Once the L4 vertebral level was identified, the bilateral psoas muscles were manually contoured. This area was normalized by patient height, with units described in mm²/m². Receiver operating characteristic (ROC) curves were generated for TPA, BMI, and BSA to elicit clinically relevant cutoffs. Regression and Kaplan-Meier analyses were used to correlate toxicity with survival functions. Results: Low TPA (OR =1.903, P=0.036) was predictive of acute toxicities, and only TPA was predictive of Grade 3 or higher acute toxicities (OR =10.24, P=0.007). Both findings were independent of tumor resectability. Pain (P=0.003), fatigue (P=0.040), and nausea (P=0.039) were significantly associated with low TPA. No association was identified between any measurement of nutritional status and the development of late toxicities, overall survival, local progression or local recurrence. However, BRPC patients survived longer (median =21.98 months) than their LAPC (median =16.2 months) counterparts (P=0.002), independent of nutritional status. Conclusions: TPA measurement is readily available and more specific than BMI or BSA as a predictor of acute radiotoxic complications following SBRT in BRPC/LAPC patients. A TPA of <500 mm²/m² is a clinically relevant cutoff that can direct physicians to address expected complications of pain, fatigue, and nausea. However, tumor resectability remains as the only predictor of overall survival in this cohort.

Real-time magnetic resonance image guided stereotactic ablative radiotherapy (MRgSBRT) is used to treat abdominal tumors. Longitudinal data is generated from daily setup images. Our study aimed to identify delta radiomic texture features extracted from these images to predict for local control in patients with liver tumors treated with MRgSBRT. Retrospective analysis of an IRB-approved database identified patients treated with MRgSBRT for primary liver and secondary metastasis histologies. Daily low field strength (0.35 T) images were retrieved, and the gross tumor volume was identified on each image. Next, images’ gray levels were equalized, and 39 s-order texture features were extracted. Delta-radiomics were calculated as the difference between feature values on the initial scan and after delivered biological effective doses (BED, α/β = 10) of 20 Gy and 40 Gy. Then, features were ranked by the Gini Index during training of a random forest model. Finally, the area under the receiver operating characteristic curve (AUC) was estimated using a bootstrapped logistic regression with the top two features. We identified 22 patients for analysis. The median dose delivered was 50 Gy in 5 fractions. The top two features identified after delivery of BED 20 Gy were gray level co-occurrence matrix features energy and gray level size zone matrix based large zone emphasis. The model generated an AUC = 0.9011 (0.752–1.0) during bootstrapped logistic regression. The same two features were selected after delivery of a BED 40 Gy, with an AUC = 0.716 (0.600–0.786). Delta-radiomic features after a single fraction of SBRT predicted local control in this exploratory cohort. If confirmed in larger studies, these features may identify patients with radioresistant disease and provide an opportunity for physicians to alter management much sooner than standard restaging after 3 months. Expansion of the patient database is warranted for further analysis of delta-radiomic features.