Background: Natural language processing models such as ChatGPT can generate text-based content and are poised to become a major information source in medicine and beyond. The accuracy and completeness of ChatGPT for medical queries is not known. Methods: Thirty-three physicians across 17 specialties generated 284 medical questions that they subjectively classified as easy, medium, or hard with either binary (yes/no) or descriptive answers. The physicians then graded ChatGPT-generated answers to these questions for accuracy (6-point Likert scale; range 1 – completely incorrect to 6 – completely correct) and completeness (3-point Likert scale; range 1 – incomplete to 3 - complete plus additional context). Scores were summarized with descriptive statistics and compared using Mann-Whitney U or Kruskal-Wallis testing. Results: Across all questions (n=284), median accuracy score was 5.5 (between almost completely and completely correct) with mean score of 4.8 (between mostly and almost completely correct). Median completeness score was 3 (complete and comprehensive) with mean score of 2.5. For questions rated easy, medium, and hard, median accuracy scores were 6, 5.5, and 5 (mean 5.0, 4.7, and 4.6; p=0.05). Accuracy scores for binary and descriptive questions were similar (median 6 vs. 5; mean 4.9 vs. 4.7; p=0.07). Of 36 questions with scores of 1-2, 34 were re-queried/re-graded 8-17 days later with substantial improvement (median 2 vs. 4; p<0.01). Conclusions: ChatGPT generated largely accurate information to diverse medical queries as judged by academic physician specialists although with important limitations. Further research and model development are needed to correct inaccuracies and for validation.

Lower health literacy levels are independently associated with longer index hospitalization lengths of stay for patients who are undergoing major abdominal surgery. The role of health literacy needs to be further evaluated within surgical practices to improve health care outcomes and use.

IMPORTANCE Pancreatic mucinous cystic neoplasms (MCNs) harbor malignant potential, and current guidelines recommend resection. However, data are limited on preoperative risk factors for malignancy (adenocarcinoma or high-grade dysplasia) occurring in the setting of an MCN. OBJECTIVES To examine the preoperative risk factors for malignancy in resected MCNs and to assess outcomes of MCN-associated adenocarcinoma. DESIGN, SETTING, AND PARTICIPANTS Patients who underwent pancreatic resection of MCNs at the 8 academic centers of the Central Pancreas Consortium from January 1, 2000, through December 31, 2014, were retrospectively identified. Preoperative factors of patients with and without malignant tumors were compared. Survival analyses were conducted for patients with adenocarcinoma. MAIN OUTCOMES AND MEASURES Binary logistic regression models were used to determine the association of preoperative factors with the presence of MCN-associated malignancy. RESULTS A total of 1667 patients underwent resection of pancreatic cystic lesions, and 349 (20.9%) had an MCN (310 women [88.8%]; mean (SD) age, 53.3 [14.7] years). Male sex (odds ratio [OR], 3.72; 95% CI, 1.21–11.44; P = .02), pancreatic head and neck location (OR, 3.93; 95% CI, 1.43–10.81; P = .01), increased radiographic size of the MCN (OR, 1.17; 95% CI, 1.08–1.27; P < .001), presence of a solid component or mural nodule (OR, 4.54; 95% CI, 1.95–10.57; P < .001), and duct dilation (OR, 4.17; 95% CI, 1.63–10.64; P = .003) were independently associated with malignancy. Malignancy was not associated with presence of radiographic septations or preoperative cyst fluid analysis (carcinoembryonic antigen, amylase, or mucin presence). The median serum CA19-9 level for patients with malignant neoplasms was 210 vs 15 U/mL for those without (P = .001). In the 44 patients with adenocarcinoma, 41 (93.2%) had lymph nodes harvested, with nodal metastases in only 14 (34.1%). Median follow-up for patients with adenocarcinoma was 27 months. Adenocarcinoma recurred in 11 patients (25%), with a 64% recurrence-free survival and 59% overall survival at 3 years. CONCLUSIONS AND RELEVANCE Adenocarcinoma or high-grade dysplasia is present in 14.9% of resected pancreatic MCNs for which risks include male sex, pancreatic head and neck location, larger MCN, solid component or mural nodule, and duct dilation. Mucinous cystic neoplasm-associated adenocarcinoma appears to have decreased nodal involvement at the time of resection and increased survival compared with typical pancreatic ductal adenocarcinoma. Indications for resection of MCNs should be revisited.

Purpose Aberrant promoter methylation and genomic instability occur frequently during colorectal cancer (CRC) development. CpG island methylator phenotype (CIMP) has been shown to associate with microsatellite instability, BRAF mutation and often found in the right-side colon. Nevertheless, the relative importance of CIMP and chromosomal instability (CIN) for tumorigenesis has yet to be thoroughly investigated in sporadic CRCs. Experimental Design We determined CIMP in 161 primary CRCs and 66 matched normal mucosae using a quantitative bisulfite/PCR/LDR/Universal Array assay. The validity of CIMP was confirmed in a subset of 60 primary tumors using MethyLight assay and five independent markers. In parallel, chromosomal instability was analyzed in the same study cohort using Affymetrix 50K Human Mapping arrays. Results The identified CIMP-positive cancers correlate with microsatellite instability (p=0.075) and the BRAF mutation V600E (p=0.00005). The array-based high-resolution analysis of chromosomal aberrations indicated that the degree of aneuploidy is spread over a wide spectrum among analyzed CRCs. Whether CIN was defined by copy number variations in selected microsatellite loci (criterion 1) or considered as a continuous variable (criterion 2), CIMP-positive samples showed a strong correlation with low-degree chromosomal aberrations (p=0.075 and 0.012, respectively). Similar correlations were observed when CIMP was determined using MethyLight assay (p=0.001 and 0.013, respectively). Conclusion CIMP-positive tumors generally possess lower chromosomal aberrations, which may only be revealed using a genome-wide approach. The significant difference in the degree of chromosomal aberrations between CIMP-positive and the remainder samples suggests that epigenetic (CIMP) and genetic (CIN) abnormalities may arise from independent molecular mechanisms of tumor progression.

Introduction Colorectal cancer (CRC) diagnosed before age 30 is a fatal disease whose biology remains poorly understood. To understand its pathogenesis, we compared molecular and clinical data in surgically treated early-age onset and adult onset patients. Materials and methods Clinical data and tumor tissue were collected retrospectively for 94 patients with early-age onset CRC (≤age 30) and compared to 275 adult CRC patients (≥age 50). Tumor morphology, microsatellite instability (MSI) and stability (MSS), KRAS and BRAF mutations, and mismatch repair (MMR) expression (MSH2, MLH1, MSH6, PMS2) were assessed. Results Early-age CRC was distinguished from adult CRC by advanced stage presentation (P<0.001), frequent high grade cancers (P<0.001), and poor prognosis (P<0.001). MSI was associated with favorable survival and MMR loss in both groups. Compared to adults, MSI in early-onset CRC was more prevalent (P<0.01), not tightly linked to MLH1/PMS2 loss, and never associated with BRAFV600E mutations (P<0.01). MSS/BRAFV600E genotype had poor prognosis and was more prevalent in early-age CRC (9% vs. 3%). Discussion Specific genetic subtypes are found at different frequencies in early-age onset and adult onset CRC. Complete absence of the indolent MSI/BRAFV600E genotype and enrichment in the unfavorable MSS/BRAFV600E genotype help explain the poor prognosis of early onset CRC.