Comparison of Large Language Models in Answering Immuno-Oncology Questions: A Cross-Sectional Study.

BACKGROUND: The capability of large language models (LLMs) to understand and generate human-readable text has prompted the investigation of their potential as educational and management tools for patients with cancer and healthcare providers. MATERIALS AND METHODS: We conducted a cross-sectional study aimed at evaluating the ability of ChatGPT-4, ChatGPT-3.5, and Google Bard to answer questions related to 4 domains of immuno-oncology (Mechanisms, Indications, Toxicities, and Prognosis). We generated 60 open-ended questions (15 for each section). Questions were manually submitted to LLMs, and responses were collected on June 30, 2023. Two reviewers evaluated the answers independently. RESULTS: ChatGPT-4 and ChatGPT-3.5 answered all questions, whereas Google Bard answered only 53.3% (P < .0001). The number of questions with reproducible answers was higher for ChatGPT-4 (95%) and ChatGPT3.5 (88.3%) than for Google Bard (50%) (P < .0001). In terms of accuracy, the number of answers deemed fully correct were 75.4%, 58.5%, and 43.8% for ChatGPT-4, ChatGPT-3.5, and Google Bard, respectively (P = .03). Furthermore, the number of responses deemed highly relevant was 71.9%, 77.4%, and 43.8% for ChatGPT-4, ChatGPT-3.5, and Google Bard, respectively (P = .04). Regarding readability, the number of highly readable was higher for ChatGPT-4 and ChatGPT-3.5 (98.1%) and (100%) compared to Google Bard (87.5%) (P = .02). CONCLUSION: ChatGPT-4 and ChatGPT-3.5 are potentially powerful tools in immuno-oncology, whereas Google Bard demonstrated relatively poorer performance. However, the risk of inaccuracy or incompleteness in the responses was evident in all 3 LLMs, highlighting the importance of expert-driven verification of the outputs returned by these technologies.

Utility of Large Language Models for Health Care Professionals and Patients in Navigating Hematopoietic Stem Cell Transplantation: Comparison of the Performance of ChatGPT-3.5, ChatGPT-4, and Bard.

BACKGROUND: Artificial intelligence is increasingly being applied to many workflows. Large language models (LLMs) are publicly accessible platforms trained to understand, interact with, and produce human-readable text; their ability to deliver relevant and reliable information is also of particular interest for the health care providers and the patients. Hematopoietic stem cell transplantation (HSCT) is a complex medical field requiring extensive knowledge, background, and training to practice successfully and can be challenging for the nonspecialist audience to comprehend. OBJECTIVE: We aimed to test the applicability of 3 prominent LLMs, namely ChatGPT-3.5 (OpenAI), ChatGPT-4 (OpenAI), and Bard (Google AI), in guiding nonspecialist health care professionals and advising patients seeking information regarding HSCT. METHODS: We submitted 72 open-ended HSCT-related questions of variable difficulty to the LLMs and rated their responses based on consistency-defined as replicability of the response-response veracity, language comprehensibility, specificity to the topic, and the presence of hallucinations. We then rechallenged the 2 best performing chatbots by resubmitting the most difficult questions and prompting to respond as if communicating with either a health care professional or a patient and to provide verifiable sources of information. Responses were then rerated with the additional criterion of language appropriateness, defined as language adaptation for the intended audience. RESULTS: ChatGPT-4 outperformed both ChatGPT-3.5 and Bard in terms of response consistency (66/72, 92%; 54/72, 75%; and 63/69, 91%, respectively; P=.007), response veracity (58/66, 88%; 40/54, 74%; and 16/63, 25%, respectively; P<.001), and specificity to the topic (60/66, 91%; 43/54, 80%; and 27/63, 43%, respectively; P<.001). Both ChatGPT-4 and ChatGPT-3.5 outperformed Bard in terms of language comprehensibility (64/66, 97%; 53/54, 98%; and 52/63, 83%, respectively; P=.002). All displayed episodes of hallucinations. ChatGPT-3.5 and ChatGPT-4 were then rechallenged with a prompt to adapt their language to the audience and to provide source of information, and responses were rated. ChatGPT-3.5 showed better ability to adapt its language to nonmedical audience than ChatGPT-4 (17/21, 81% and 10/22, 46%, respectively; P=.03); however, both failed to consistently provide correct and up-to-date information resources, reporting either out-of-date materials, incorrect URLs, or unfocused references, making their output not verifiable by the reader. CONCLUSIONS: In conclusion, despite LLMs' potential capability in confronting challenging medical topics such as HSCT, the presence of mistakes and lack of clear references make them not yet appropriate for routine, unsupervised clinical use, or patient counseling. Implementation of LLMs' ability to access and to reference current and updated websites and research papers, as well as development of LLMs trained in specialized domain knowledge data sets, may offer potential solutions for their future clinical application.

Current status of National Institutes of Health funding for thoracic surgeons in the United States: Beacon of hope or candle in the wind?

OBJECTIVE: Increasing forces threaten the viability of thoracic surgeon-initiated research, a core component of our academic mission. National Institutes of Health funding is a benchmark of research productivity and innovation. This study examined the current status of National Institutes of Health funding for thoracic surgeons. METHODS: Thoracic surgeon principal investigators on National Institutes of Health-funded grants during June 2010, June 2015, and June 2020 were identified using National Institutes of Health iSearchGrants (version 2.4). American Association of Medical Colleges data were used to identify all surgeons in the United States. Types and total costs of National Institutes of Health-funded grants were compared relative to other surgical specialties. RESULTS: A total of 61 of 4681 (1.3%), 63 of 4484 (1.4%), and 60 of 4497 (1.3%) thoracic surgeons were principal investigators on 79, 76, and 87 National Institutes of Health-funded grants in 2010, 2015, and 2020, respectively; these rates were higher than those for most other surgical specialties (P ≤ .0001). Total National Institutes of Health costs for Thoracic Surgeon-initiated grants increased 57% from 2010 to 2020, outpacing the 33% increase in total National Institutes of Health budget. Numbers and types of grants varied among cardiovascular, transplant, and oncology subgroups. Although the majority of grants and costs were cardiovascular related, increased National Institutes of Health expenditures primarily were due to funding for transplant and oncology grants. Per-capita costs were highest for transplant-related grants during both years. Percentages of R01-to-total costs were constant at 55%. Rates and levels of funding for female versus male thoracic surgeons were comparable. Awards to 5 surgeons accounted for 33% of National Institutes of Health costs for thoracic surgeon principal investigators in 2020; a similar phenomenon was observed for 2010 and 2015. CONCLUSIONS: Long-term structural changes must be implemented to more effectively nurture the next generation of thoracic surgeon scientists.

Pulmonary Metastasectomy for Adrenocortical Carcinoma-Not If, but When.

BACKGROUND: Adrenocortical carcinoma (ACC) commonly metastasizes to the lungs, and pulmonary metastasectomy (PM) is utilized due to limited systemic options. METHODS: All ACC patients with initially only lung metastases (LM) from a single institution constituted this observational case series. Kaplan-Meier and Cox proportional hazard analyses evaluated the association with potential prognostic factors and outcomes. Overall survival (OS) was calculated from the date of the PM or, in those patients who did not undergo surgery, from the development of LM. RESULTS: A total of 75 ACC patients over a 45-year period met the criteria; 52 underwent PM, and 23 did not. The patients undergoing PM had a median OS of 3.1 years (95% CI: 2.4, 4.7 years) with the 5- and 10-year OS being 35.5% and 32.8%, respectively. The total resected LM did not impact the OS nor the DFS. The patients who developed LM after 11 months from the initial ACC resection had an improved OS (4.2 years; 95% CI: 3.2, NR; p = 0.0096) compared to those developing metastases earlier (2.4 years; 95% CI: 1.6, 2.8). Patients who underwent PM within 11 months of adrenalectomy demonstrated a reduced OS (2.2 years; 95% CI: 1.0, 2.7) compared to those after 11 months (3.6 years, 95% CI: 2.6, NR; p = 0.0045). PM may provide benefit to those patients with LM at presentation (HR: 0.5; p = 0.2827), with the time to first PM as a time-varying covariate. CONCLUSIONS: PM appears to have a role in ACC patients. The number of nodules should not be an exclusion factor. Patients developing LM within a year of primary tumor resection may benefit from waiting before further surgeries, which may provide additional insight into who may benefit from PM.

Altruistic feeding and cell-cell signaling during bacterial differentiation actively enhance phenotypic heterogeneity.

Starvation triggers bacterial spore formation, a committed differentiation program that transforms a vegetative cell into a dormant spore. Cells in a population enter sporulation non-uniformly to secure against the possibility that favorable growth conditions, which puts sporulation-committed cells at a disadvantage, may resume. This heterogeneous behavior is initiated by a passive mechanism: stochastic activation of a master transcriptional regulator. Here, we identify a cell-cell communication pathway that actively promotes phenotypic heterogeneity, wherein Bacillus subtilis cells that start sporulating early utilize a calcineurin-like phosphoesterase to release glycerol, which simultaneously acts as a signaling molecule and a nutrient to delay non-sporulating cells from entering sporulation. This produced a more diverse population that was better poised to exploit a sudden influx of nutrients compared to those generating heterogeneity via stochastic gene expression alone. Although conflict systems are prevalent among microbes, genetically encoded cooperative behavior in unicellular organisms can evidently also boost inclusive fitness.

Preclinical Evaluation of a Novel Series of Polyfluorinated Thalidomide Analogs in Drug-Resistant Multiple Myeloma.

Immunomodulatory imide drugs (IMiDs) play a crucial role in the treatment landscape across various stages of multiple myeloma. Despite their evident efficacy, some patients may exhibit primary resistance to IMiD therapy, and acquired resistance commonly arises over time leading to inevitable relapse. It is critical to develop novel therapeutic options to add to the treatment arsenal to overcome IMiD resistance. We designed, synthesized, and screened a new class of polyfluorinated thalidomide analogs and investigated their anti-cancer, anti-angiogenic, and anti-inflammatory activity using in vitro and ex vivo biological assays. We identified four lead compounds that exhibit potent anti-myeloma, anti-angiogenic, anti-inflammatory properties using three-dimensional tumor spheroid models, in vitro tube formation, and ex vivo human saphenous vein angiogenesis assays, as well as the THP-1 inflammatory assay. Western blot analyses investigating the expression of proteins downstream of cereblon (CRBN) reveal that Gu1215, our primary lead candidate, exerts its activity through a CRBN-independent mechanism. Our findings demonstrate that the lead compound Gu1215 is a promising candidate for further preclinical development to overcome intrinsic and acquired IMiD resistance in multiple myeloma.

Tofacitinib to prevent anti-drug antibody formation against LMB-100 immunotoxin in patients with advanced mesothelin-expressing cancers.

Background: LMB-100 is a mesothelin (MSLN)-targeting recombinant immunotoxin (iTox) carrying a Pseudomonas exotoxin A payload that has shown promise against solid tumors, however, efficacy is limited by the development of neutralizing anti-drug antibodies (ADAs). Tofacitinib is an oral Janus Kinase (JAK) inhibitor that prevented ADA formation against iTox in preclinical studies. Methods: A phase 1 trial testing LMB-100 and tofacitinib in patients with MSLN-expressing cancers (pancreatic adenocarcinoma, n=13; cholangiocarcinoma, n=1; appendiceal carcinoma, n=1; cystadenocarcinoma, n=1) was performed to assess safety and to determine if tofacitinib impacted ADA formation. Participants were treated for up to 3 cycles with LMB-100 as a 30-minute infusion on days 4, 6, and 8 at two dose levels (100 and 140 µg/kg) while oral tofacitinib was administered for the first 10 days of the cycle (10 mg BID). Peripheral blood was collected for analysis of ADA levels, serum cytokines and circulating immune subsets. Results: The study was closed early due to occurrence of drug-induced pericarditis in 2 patients. Pericarditis with the combination was not reproducible in a transgenic murine model containing human MSLN. Two of 4 patients receiving all 3 cycles of treatment maintained effective LMB-100 levels, an unusual occurrence. Sustained increases in systemic IL-10 and TNF-α were seen, a phenomenon not observed in prior LMB-100 studies. A decrease in activated T cell subsets and an increase in circulating immunosuppressive myeloid populations occurred. No radiologic decreases in tumor volume were observed. Discussion: Further testing of tofacitinib to prevent ADA formation is recommended in applicable non-malignant disease settings. Clinical trial registration: https://www.clinicaltrials.gov/study/NCT04034238.