Evaluation of the accuracy and readability of ChatGPT-4 and Google Gemini in providing information on retinal detachment: a multicenter expert comparative study.

BACKGROUND: Large language models (LLMs) such as ChatGPT-4 and Google Gemini show potential for patient health education, but concerns about their accuracy require careful evaluation. This study evaluates the readability and accuracy of ChatGPT-4 and Google Gemini in answering questions about retinal detachment. METHODS: Comparative study analyzing responses from ChatGPT-4 and Google Gemini to 13 retinal detachment questions, categorized by difficulty levels (D1, D2, D3). Masked responses were reviewed by ten vitreoretinal specialists and rated on correctness, errors, thematic accuracy, coherence, and overall quality grading. Analysis included Flesch Readability Ease Score, word and sentence counts. RESULTS: Both Artificial Intelligence tools required college-level understanding for all difficulty levels. Google Gemini was easier to understand (p = 0.03), while ChatGPT-4 provided more correct answers for the more difficult questions (p = 0.0005) with fewer serious errors. ChatGPT-4 scored highest on most challenging questions, showing superior thematic accuracy (p = 0.003). ChatGPT-4 outperformed Google Gemini in 8 of 13 questions, with higher overall quality grades in the easiest (p = 0.03) and hardest levels (p = 0.0002), showing a lower grade as question difficulty increased. CONCLUSIONS: ChatGPT-4 and Google Gemini effectively address queries about retinal detachment, offering mostly accurate answers with few critical errors, though patients require higher education for comprehension. The implementation of AI tools may contribute to improving medical care by providing accurate and relevant healthcare information quickly.

Prognostic value of clinical and radiomic parameters in patients with liver metastases from uveal melanoma.

Approximately every second patient with uveal melanoma develops distant metastases, with the liver as the predominant target organ. While the median survival after diagnosis of distant metastases is limited to a year, yet-to-be-defined subgroups of patients experience a more favorable outcome. Therefore, prognostic biomarkers could help identify distinct risk groups to guide patient counseling, therapeutic decision-making, and stratification of study populations. To this end, we retrospectively analyzed a cohort of 101 patients with newly diagnosed hepatic metastases from uveal melanoma by using Cox-Lasso regression machine learning, adapted to a high-dimensional input parameter space. We show that substantial binary risk stratification can be performed, based on (i) clinical and laboratory parameters, (ii) measures of quantitative overall hepatic tumor burden, and (iii) radiomic parameters. Yet, combining two or all three domains failed to improve prognostic separation of patients. Additionally, we identified highly relevant clinical parameters (including lactate dehydrogenase, thrombocyte counts, aspartate transaminase, and the metastasis-free interval) at first diagnosis of metastatic disease as predictors for time-to-treatment failure and overall survival. Taken together, the risk stratification models, built by our machine-learning algorithm, identified a comparable and independent prognostic value of clinical, radiological, and radiomic parameters in uveal melanoma patients with hepatic metastases.

Electroporation with Calcium or Bleomycin: First Application in an In Vivo Uveal Melanoma Patient-Derived Xenograft Model.

Uveal melanoma (UM) represents a rare tumor of the uveal tract and is associated with a poor prognosis due to the high risk of metastasis. Despite advances in the treatment of UM, the mortality rate remains high, dictating an urgent need for novel therapeutic strategies. The current study introduces the first in vivo analysis of the therapeutic potential of calcium electroporation (CaEP) compared with electrochemotherapy (ECT) with bleomycin in a patient-derived xenograft (PDX) model based on the chorioallantoic membrane (CAM) assay. The experiments were conducted as monotherapy with either 5 or 10 mM calcium chloride or 1 or 2.5 µg/mL bleomycin in combination with EP or EP alone. CaEP and ECT induced a similar reduction in proliferative activity, neovascularization, and melanocytic expansion. A dose-dependent effect of CaEP triggered a significant induction of necrosis, whereas ECT application of 1 µg/mL bleomycin resulted in a significantly increased apoptotic response compared with untreated tumor grafts. Our results outline the prospective use of CaEP and ECT with bleomycin as an adjuvant treatment of UM, facilitating adequate local tumor control and potentially an improvement in metastatic and overall survival rates.

Treatment Outcomes of Patients with Orbital Inflammatory Diseases: Should Steroids Still Be the First Choice?

Objective: To clarify the therapy response in orbital inflammatory diseases (OID), we analyzed the treatment effects of steroid therapy, the use of disease-modifying antirheumatic drugs (DMARDS), and biologicals in our tertiary referral center cohort. Methods: We collected the clinical and demographic data of all patients treated for non-specific orbital inflammation (NSOI) (n = 111) and IgG4-ROD (n = 13), respectively at our center from 2008 to 2020 and analyzed them with descriptive statistics. NSOI were sub-grouped according to the location into either idiopathic dacryoadenitis (DAs) (n = 78) or typical idiopathic orbital myositis (n = 32). Results: Mean age at first clinical manifestation was significantly different between subgroups (IOI: 49.5 ± 18, IgG4-ROD: 63.2 ± 14, p = 0.0171). Among all examined OID, 63 patients (50%) achieved full remission (FR) with corticosteroids (NSOI 53%/IgG4-ROD 31%). In contrast, classic myositis showed a significantly higher response (76%). Disease-modifying drugs (DMARDS) for myositis accomplished only 33% FR (NSOI 57%) and 66% did not respond sufficiently (NSOI 43%). The biologic agent (Rituximab) was significantly more efficient: 19 of 23 patients (82%) achieved full remission and only 4 (17%) did not respond fully and needed orbital irradiation or orbital decompressive surgery.

The Chick Chorioallantoic Membrane as a Xenograft Model for the Quantitative Analysis of Uveal Melanoma Metastasis in Multiple Organs.

Uveal melanoma (UM) is the most common intraocular tumor in adults, and nearly 50% of patients develop metastatic disease with a high mortality rate. Therefore, the development of relevant preclinical in vivo models that accurately recapitulate the metastatic cascade is crucial. We exploited the chick embryo chorioallantoic membrane (CAM) xenograft model to quantify both experimental and spontaneous metastasis by qPCR analysis. Our study found that the transplanted UM cells spread predominantly and early in the liver, reflecting the primary site of metastasis in patients. Visible signs of pigmented metastasis were observed in the eyes, liver, and distal CAM. Lung metastases occurred rarely and brain metastases progressed more slowly. However, UM cell types of different origins and genetic profiles caused an individual spectrum of organ metastases. Metastasis to multiple organs, including the liver, was often associated with risk factors such as high proliferation rate, hyperpigmentation, and epithelioid cell type. The severity of liver metastasis was related to the hepatic metastatic origin and chromosome 8 abnormalities rather than monosomy 3 and BAP1 deficiency. The presented CAM xenograft model may prove useful to study the metastatic potential of patients or to test individualized therapeutic options for metastasis in different organs.

Evidence in Focus: The Sparse Landscape of Randomized Trials on Retinoblastoma Treatment.

Background: Retinoblastoma, although rare, is one of the most common intraocular malignancies worldwide. Its prognosis has improved significantly in the past few decades, thanks to modern treatments, like systemic, intra-arterial, and intravitreal chemotherapy. However, regarding survival, there are significant differences between high- and low-income countries, eye salvage is still a challenge worldwide, and treatment-related toxicity needs to be carefully and sufficiently managed. Summary: To appraise the strength of supporting evidence, we performed a systematic review of randomized controlled trials investigating any therapeutic protocol for retinoblastoma. Four trials with 174 participants (188 eyes) were eligible, all pertaining to different intravenous chemotherapy regimens. Vincristine, etoposide, and carboplatin (VEC) appear superior to a 5-drug combination for stage III retinoblastoma. Moreover, etoposide and carboplatin as neoadjuvant chemotherapy followed by thermochemotherapy seem to offer better local control than vincristine and carboplatin. However, increasing carboplatin dose in the VEC protocol failed to improve treatment efficacy. Key Messages: Retinoblastoma is a success story of modern medicine. However, only intravenous chemotherapy has been studied through randomized trials, while evidence for the most novel retinoblastoma treatments has mainly stemmed from observational studies. International collaborations for multicenter randomized trials could overcome difficulties and increase certainty and precision in the field.

Prognostic factors for surgical treatment of radiation-induced scleral necrosis after brachytherapy for uveal melanoma.

PURPOSE: Radiation-induced scleral necrosis (RISN) is a less frequent complication of brachytherapy for uveal melanoma, and may require surgical treatment in selected cases. We aimed to identify the prognostic factors for RISN treatment. METHODS: All patients with brachytherapy for uveal melanoma treated at our institution between 01/1999 and 12/2016 who developed RISN were followed until 02/2021. Various parameters were evaluated through univariable and multivariable Cox regression analysis. The surgical intervention due to RISN was the principal outcome event of this study. RESULTS: Of 115 patients in the final cohort, 51 individuals (44%) underwent RISN treatment (conjunctival revision [n = 2], patching [n = 46] or enucleation [n = 3]) at median 1.80 months after RISN occurrence. Significant RISN characteristics were summarized into a novel RISN severity scale - Grade I: largest diameter ≤ 5 mm and no progression; Grade II: largest diameter > 5 mm or any progression during the follow-up; Grade III: presence of uveal prolapse; and Grade IV: leakage through open eyewall perforation. In the multivariable analysis, the RISN severity scale (aHR = 2.37 per grade increase, p = 0.01) and the time between brachytherapy and RISN occurrence (<15 months, aHR = 6.33, p < 0.0001) were independently associated with the study endpoint. The RISN severity scale showed high diagnostic accuracy for prediction of RISN treatment (AUC = 0.869). CONCLUSIONS: In our series, about the half of RISN cases underwent surgical treatment. The presented novel severity scale for RISN might become a helpful tool for clinical management of individuals with RISN. We recommend external validation of the diagnostic accuracy of the presented scale.

Wolfram-silicone implants as effective radiation shielding for ocular brachytherapy: dosimetric features and in vivo animal study on biocompatibility.

OBJECTIVE: To evaluate wolfram as a photon and beta absorber in the management of uveal melanoma with radiotherapy, examining its potential ocular adverse effects and physiologic tolerance using an in vivo rabbit ocular model. METHODS: A method of manufacturing implants from mixtures of wolfram and silicone was developed. Their shielding effect on the radiation of sources used in ocular brachytherapy was investigated by dosimetric measurement in an eye phantom as well as numerical simulations. Different wolfram implantation techniques, such as extraocular fixation of a wolfram-silicone implant (n = 1), vitrectomy with silicone oil and intravitreal injection of a wolfram-silicone oil suspension (n = 2), and concurrent attachment of a wolfram implant onto the sclera (n = 2), were tested to investigate the long-term effects of wolfram. A vitrectomy with silicone oil without wolfram implantation was carried out in 2 rabbits (n = 2), constituting the control group. The eyes were enucleated after 3 months for histologic analysis. RESULTS: Wolfram-silicone mixtures have been dosimetrically proven to be very effective radiation absorbers for use in ocular brachytherapy. Severe complications, such as endophthalmitis, secondary glaucoma, cornea decompensation, and vessel occlusion, were not documented in the tested rabbit eyes after the application of wolfram. Histologic examination of the bulbi after enucleation showed epiretinal gliosis without further pathologic findings in all eyes after vitrectomy. CONCLUSIONS: The results of this study show that wolfram and wolfram-silicone implants constitute a promising candidate as potential radiation shielding substrates.

Uncertainties in ocular proton planning and their impact on required margins.

PURPOSE: To review required margins in ocular proton therapy (OPT) based on an uncertainty estimation and to compare them with widely used values. Further, uncertainties when using registered funduscopy images in the 3D model is investigated. METHODS: An uncertainty budget in planning and delivery was defined to determine required aperture and range margins. Setup uncertainties were considered for a cohort of treated patients and tested in a worst-case estimation. Other uncertainties were based on a best-guess and knowledge of institutional specifics, e.g. range reproducibility. Margins for funduscopy registration were defined resulting from scaling, rotation and translation of the image. Image formation for a wide-field fundus camera was reviewed and compared to the projection employed in treatment planning systems. RESULTS: Values for aperture and range with margins of 2.5 mm as reported in literature could be determined. Aperture margins appear appropriate for setup uncertainties below 0.5 mm, but depend on lateral penumbra. Range margins depend on depth and associated density uncertainty in tissue. Registration of funduscopy images may require margins of >2 mm, increasing towards the equator. Difference in the projection may lead to discrepancies of several mm. CONCLUSIONS: The commonly used 2.5 mm aperture margin was validated as an appropriate choice, while range margins could be reduced for lower ranges. Margins may however not include uncertainties in contouring and possible microscopic spread. If a target base is contoured on registered funduscopy images care must be taken as they are subject to larger uncertainties. Multimodal imaging approach in OPT remains advisable.