Background: Large language models (LLMs) have raised both interest and concern in the academic community. They offer the potential for automating literature search and synthesis for systematic reviews but raise concerns regarding their reliability, as the tendency to generate unsupported (hallucinated) content persist. Objective: The aim of the study is to assess the performance of LLMs such as ChatGPT and Bard (subsequently rebranded Gemini) to produce references in the context of scientific writing. Methods: The performance of ChatGPT and Bard in replicating the results of human-conducted systematic reviews was assessed. Using systematic reviews pertaining to shoulder rotator cuff pathology, these LLMs were tested by providing the same inclusion criteria and comparing the results with original systematic review references, serving as gold standards. The study used 3 key performance metrics: recall, precision, and F1-score, alongside the hallucination rate. Papers were considered “hallucinated” if any 2 of the following information were wrong: title, first author, or year of publication. Results: In total, 11 systematic reviews across 4 fields yielded 33 prompts to LLMs (3 LLMs×11 reviews), with 471 references analyzed. Precision rates for GPT-3.5, GPT-4, and Bard were 9.4% (13/139), 13.4% (16/119), and 0% (0/104) respectively (P<.001). Recall rates were 11.9% (13/109) for GPT-3.5 and 13.7% (15/109) for GPT-4, with Bard failing to retrieve any relevant papers (P<.001). Hallucination rates stood at 39.6% (55/139) for GPT-3.5, 28.6% (34/119) for GPT-4, and 91.4% (95/104) for Bard (P<.001). Further analysis of nonhallucinated papers retrieved by GPT models revealed significant differences in identifying various criteria, such as randomized studies, participant criteria, and intervention criteria. The study also noted the geographical and open-access biases in the papers retrieved by the LLMs. Conclusions: Given their current performance, it is not recommended for LLMs to be deployed as the primary or exclusive tool for conducting systematic reviews. Any references generated by such models warrant thorough validation by researchers. The high occurrence of hallucinations in LLMs highlights the necessity for refining their training and functionality before confidently using them for rigorous academic purposes.

During progression of melanoma, malignant melanocytes can be reprogrammed into mesenchymal-like cells through a process similar to epithelial-mesenchymal transition (EMT), which is associated with downregulation of the junctional protein E-cadherin and acquisition of a migratory phenotype. Recent evidence supports a role for SLUG, a transcriptional repressor of E-cadherin, as a melanocyte lineage transcription factor that predisposes to melanoma metastasis. However, the signals responsible for SLUG expression in melanoma are unclear and its role in the invasive phenotype is not fully elucidated. Here, we report that SLUG expression and activation is driven by SPARC (also known as osteonectin), a secreted extracellular matrix-associated factor that promotes EMT-like changes. Ectopic expression or knockdown of SPARC resulted in increased or reduced expression of SLUG, respectively. SLUG increase occurred concomitantly with SPARC-mediated downregulation of E-cadherin and P-cadherin, and induction of mesenchymal traits in human melanocytes and melanoma cells. Pharmacological blockade of PI3 kinase/AKT signaling impeded SPARC-induced SLUG levels and cell migration, whereas adenoviral introduction of constitutively active AKT allowed rescue of SLUG and migratory capabilities of SPARC knockdown cells. We also observed that pharmacological inhibition of oncogenic BRAF^V600E using PLX4720 did not influence SLUG expression in melanoma cells harboring BRAF^V600E. Furthermore, SLUG is a bona fide transcriptional repressor of E-cadherin as well as a regulator of P-cadherin in melanoma cells and its knockdown attenuated invasive behavior and blocked SPARC-enhanced cell migration. Notably, inhibition of cell migration in SPARC-depleted cells was rescued by expression of a SLUG transgene. In freshly isolated metastatic melanoma cells, a positive association between SPARC and SLUG mRNA levels was also found. These findings reveal that autocrine SPARC maintains heightened SLUG expression in melanoma cells and indicate that SPARC may promote EMT-associated tumor invasion by supporting AKT-dependent upregulation of SLUG.

Disruption of the endothelial barrier by tumour-derived secreted factors is a critical step in cancer cell extravasation and metastasis. Here, by comparative proteomic analysis of melanoma secretomes, we identify the matricellular protein SPARC as a novel tumour-derived vascular permeability factor. SPARC deficiency abrogates tumour-initiated permeability of lung capillaries and prevents extravasation, whereas SPARC overexpression enhances vascular leakiness, extravasation and lung metastasis. SPARC-induced paracellular permeability is dependent on the endothelial VCAM1 receptor and p38 MAPK signalling. Blocking VCAM1 impedes melanoma-induced endothelial permeability and extravasation. The clinical relevance of our findings is highlighted by high levels of SPARC detected in tumour from human pulmonary melanoma lesions. Our study establishes tumour-produced SPARC and VCAM1 as regulators of cancer extravasation, revealing a novel targetable interaction for prevention of metastasis.