Deep-Transfer-Learning-Based Natural Language Processing of Serial Free-Text Computed Tomography Reports for Predicting Survival of Patients With Pancreatic Cancer.

PURPOSE: To explore the predictive potential of serial computed tomography (CT) radiology reports for pancreatic cancer survival using natural language processing (NLP). METHODS: Deep-transfer-learning-based NLP models were retrospectively trained and tested with serial, free-text CT reports, and survival information of consecutive patients diagnosed with pancreatic cancer in a Korean tertiary hospital was extracted. Randomly selected patients with pancreatic cancer and their serial CT reports from an independent tertiary hospital in the United States were included in the external testing data set. The concordance index (c-index) of predicted survival and actual survival, and area under the receiver operating characteristic curve (AUROC) for predicting 1-year survival were calculated. RESULTS: Between January 2004 and June 2021, 2,677 patients with 12,255 CT reports and 670 patients with 3,058 CT reports were allocated to training and internal testing data sets, respectively. ClinicalBERT (Bidirectional Encoder Representations from Transformers) model trained on the single, first CT reports showed a c-index of 0.653 and AUROC of 0.722 in predicting the overall survival of patients with pancreatic cancer. ClinicalBERT trained on up to 15 consecutive reports from the initial report showed an improved c-index of 0.811 and AUROC of 0.911. On the external testing set with 273 patients with 1,947 CT reports, the AUROC was 0.888, indicating the generalizability of our model. Further analyses showed our model's contextual interpretation beyond specific phrases. CONCLUSION: Deep-transfer-learning-based NLP model of serial CT reports can predict the survival of patients with pancreatic cancer. Clinical decisions can be supported by the developed model, with survival information extracted solely from serial radiology reports.

A single arm phase 2 clinical trial of YIV-906 with neoadjuvant concurrent chemo-radiation therapy in patients with locally advanced rectal cancer.

Background: Pre-operative chemoradiation for rectal cancer is often associated with severe gastrointestinal (GI) toxicity which can interrupt, delay, and/or lead to termination of treatment. In this study, we evaluated whether the addition of YIV-906, a novel herbal medicine proven to reduce GI toxicity associated with chemotherapy could also reduce GI side effects during standard pre-operative capecitabine and pelvic radiation therapy (RT) in the neoadjuvant setting for the treatment of locally advanced rectal cancer. Methods: This single arm clinical study enrolled 24 patients between Dec 23, 2014-Sep 17, 2018 at Smilow Cancer Hospital, a comprehensive cancer center at Yale New Haven Hospital. All patients were age ≥18 years, Eastern Cooperative Oncology Group 0-1 and with histologically confirmed T3-T4 and N0-N2, M0 adenocarcinoma of the rectum. Median follow-up was 61.9 months. All patients received concurrent pelvic external beam RT (50.4 Gy in 28 fractions), YIV-906 (taken orally 800 mg twice daily on days 1-4 of RT each week), and oral capecitabine delivered in a neo-adjuvant fashion, followed by definitive surgery. Toxicity was assessed weekly during radiation and until acute symptoms resolved and then at 28 days, 4 months, 7 months and 10 months. Toxicities were graded in accordance with Common Terminology Criteria for Adverse Events (CTCAE) version 4.0. Results: At the time of surgery, 4 patients (16.7%) had a complete or near-complete response. At a median follow-up of 61.9 months, the mean overall survival (OS) of our patient cohort was 74.9 months [95% confidence interval (CI): 67.3-82.5]. The estimated 5-year OS was 82.0%. We observed 0% acute grade 4 toxicities, and only two cases of acute grade 3 diarrhea (8.3%). Conclusions: The addition of YIV-906 to capecitabine based chemoradiation for locally advanced rectal cancer led to reduced rates of GI toxicity compared to historical controls, in particular grade 3 or greater diarrhea. These findings suggest YIV-906 should be evaluated in a randomized clinical trial to further assess potential reductions in the toxicity profile of chemoradiation for GI cancers.

Association of personalized and tumor-informed ctDNA with patient survival outcomes in pancreatic adenocarcinoma.

INTRODUCTION: Personalized and tumor-informed circulating tumor DNA (ctDNA) testing is feasible and allows for molecular residual disease (MRD) identification in patients with pancreatic ductal adenocarcinoma (PDAC). METHODS: In this retrospective analysis of commercial cases from multiple US institutions, personalized, tumor-informed, whole-exome sequenced, and germline-controlled ctDNA levels were quantified and analyzed in patients with PDAC. Plasma samples (n = 1329) from 299 clinically validated patients were collected at diagnosis, perioperatively (MRD-window; within 2-12 weeks after surgery, before therapy), and during surveillance (>12 weeks post-surgery if no ACT or starting 4 weeks post-ACT) from November 2019 to March 2023. RESULTS: Of the initially diagnosed patients with stages I-III PDAC who went for resection, the median follow-up time from surgery was 13 months (range 0.1-214). Positive ctDNA detection rates were 29% (29/100) and 29.6% (45/152) during the MRD and surveillance windows, respectively. Positive ctDNA detection was significantly associated with shorter DFS within the MRD window (median DFS of 6.37 months for ctDNA-positive vs 33.31 months for ctDNA-negative patients; HR: 5.45, P < .0001) as well as during the surveillance period (median DFS: 11.40 months for ctDNA-positive vs NR for ctDNA-negative; HR: 12.38, P < .0001). Additionally, DFS was significantly better with KRAS wildtype status followed by KRASG12R (HR: 0.99, P = .97), KRASG12D (HR: 1.42, P = .194), and worse with KRASG12V (HR: 2.19, P = .002) status. In multivariate analysis, ctDNA detection at surveillance was found to be the most significant prognostic factor for recurrence (HR: 24.28, P < .001). CONCLUSIONS: Perioperative tumor-informed ctDNA detection in PDAC is feasible across all stages and is associated with patient survival outcomes.

Perioperative Modified FOLFIRINOX for Resectable Pancreatic Cancer: A Nonrandomized Controlled Trial.

Importance: Pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignant tumor, and durable disease control is rare with the current standard of care, even for patients who undergo surgical resection. Objective: To assess whether neoadjuvant modified 5-fluorouracil, leucovorin, oxaliplatin, and irinotecan (mFOLFIRINOX) leads to early control of micrometastasis and improves survival. Design, Setting, and Participants: This open-label, single-arm, phase 2 nonrandomized controlled trial for resectable PDAC was conducted at the Yale Smilow Cancer Hospital from April 3, 2014, to August 16, 2021. Pancreatic protocol computed tomography was performed at diagnosis to assess surgical candidacy. Data were analyzed from January to July 2023. Interventions: Patients received 6 cycles of neoadjuvant mFOLFIRINOX before surgery and 6 cycles of adjuvant mFOLFIRINOX. Whole blood was collected and processed to stored plasma for analysis of circulating tumor DNA (ctDNA) levels. Tumors were evaluated for treatment response and keratin 17 (K17) expression. Main Outcomes and Measures: The primary end point was 12-month progression-free survival (PFS) rate. Additional end points included overall survival (OS), ctDNA level, tumor molecular features, and K17 tumor levels. Survival curves were summarized using Kaplan-Meier estimator. Results: Of 46 patients who received mFOLFIRINOX, 31 (67%) were male, and the median (range) age was 65 (46-80) years. A total of 37 (80%) completed 6 preoperative cycles and 33 (72%) underwent surgery. A total of 27 patients (59%) underwent resection per protocol (25 with R0 disease and 2 with R1 disease); metastatic or unresectable disease was identified in 6 patients during exploration. Ten patients underwent surgery off protocol. The 12-month PFS was 67% (90% CI, 56.9-100); the median PFS and OS were 16.6 months (95% CI, 13.3-40.6) and 37.2 months (95% CI, 17.5-not reached), respectively. Baseline ctDNA levels were detected in 16 of 22 patients (73%) and in 3 of 17 (18%) after 6 cycles of mFOLFIRINOX. Those with detectable ctDNA levels 4 weeks postresection had worse PFS (hazard ratio [HR], 34.0; 95% CI, 2.6-4758.6; P = .006) and OS (HR, 11.7; 95% CI, 1.5-129.9; P = .02) compared with those with undetectable levels. Patients with high K17 expression had nonsignificantly worse PFS (HR, 2.7; 95% CI, 0.7-10.9; P = .09) and OS (HR, 3.2; 95% CI, 0.8-13.6; P = .07). Conclusions and Relevance: This nonrandomized controlled trial met its primary end point, and perioperative mFOLFIRINOX warrants further evaluation in randomized clinical trials. Postoperative ctDNA positivity was strongly associated with recurrence. K17 and ctDNA are promising biomarkers that require additional validation in future prospective studies. Trial Registration: ClinicalTrials.gov Identifier: NCT02047474.

Mutational signature-based identification of DNA repair deficient gastroesophageal adenocarcinomas for therapeutic targeting.

Homologous recombination (HR) and nucleotide excision repair (NER) are the two most frequently disabled DNA repair pathways in cancer. HR-deficient breast, ovarian, pancreatic and prostate cancers respond well to platinum chemotherapy and PARP inhibitors. However, the frequency of HR deficiency in gastric and esophageal adenocarcinoma (GEA) still lacks diagnostic and functional validation. Using whole exome and genome sequencing data, we found that a significant subset of GEA, but very few colorectal adenocarcinomas, show evidence of HR deficiency by mutational signature analysis (HRD score). High HRD gastric cancer cell lines demonstrated functional HR deficiency by RAD51 foci assay and increased sensitivity to platinum chemotherapy and PARP inhibitors. Of clinical relevance, analysis of three different GEA patient cohorts demonstrated that platinum treated HR deficient cancers had better outcomes. A gastric cancer cell line with strong sensitivity to cisplatin showed HR proficiency but exhibited NER deficiency by two photoproduct repair assays. Single-cell RNA-sequencing revealed that, in addition to inducing apoptosis, cisplatin treatment triggered ferroptosis in a NER-deficient gastric cancer, validated by intracellular GSH assay. Overall, our study provides preclinical evidence that a subset of GEAs harbor genomic features of HR and NER deficiency and may therefore benefit from platinum chemotherapy and PARP inhibitors.

NCI10066: a Phase 1/2 study of olaparib in combination with ramucirumab in previously treated metastatic gastric and gastroesophageal junction adenocarcinoma.

BACKGROUND: Our preclinical work revealed tumour hypoxia induces homologous recombination deficiency (HRD), increasing sensitivity to Poly (ADP-ribose) polymerase inhibitors. We aimed to induce tumour hypoxia with ramucirumab thereby sensitising tumours to olaparib. PATIENTS AND METHODS: This multi-institution single-arm Phase 1/2 trial enrolled patients with metastatic gastroesophageal adenocarcinoma refractory to ≥1 systemic treatment. In dose escalation, olaparib was evaluated at escalating dose levels with ramucirumab 8 mg/kg day 1 in 14-day cycles. The primary endpoint of Phase 1 was the recommended Phase 2 dose (RP2D), and in Phase 2 the primary endpoint was the overall response rate (ORR). RESULTS: Fifty-one patients received ramucirumab and olaparib. The RP2D was olaparib 300 mg twice daily with ramucirumab 8 mg/kg. In evaluable patients at the RP2D the ORR was 6/43 (14%) (95% CI 4.7-25.6). The median progression-free survival (PFS) was 2.8 months (95% CI 2.3-4.2) and median overall survival (OS) was 7.3 months (95% CI 5.7-13.0). Non-statistically significant improvements in PFS and OS were observed for patients with tumours with mutations in HRD genes. CONCLUSIONS: Olaparib and ramucirumab is well-tolerated with efficacy that exceeds historical controls with ramucirumab single agent for gastric cancer in a heavily pre-treated patient population.