The Evolving Role of Radiotherapy for Pediatric Cancers With Advancements in Molecular Tumor Characterization and Targeted Therapies

Large language models for pretreatment education in pediatric radiation oncology: A comparative evaluation study

Background and purpose

Pediatric radiotherapy patients and their parents are usually aware of their need for radiotherapy early on, but they meet with a radiation oncologist later in their treatment. Consequently, they search for information online, often encountering unreliable sources. Large language models (LLMs) have the potential to serve as an educational pretreatment tool, providing reliable answers to their questions. We aimed to evaluate the responses provided by generative pre-trained transformers (GPT), the most popular subgroup of LLMs, to questions about pediatric radiation oncology.

Materials and methods

We collected pretreatment questions regarding radiotherapy from patients and parents. Responses were generated using GPT-3.5, GPT-4, and fine-tuned GPT-3.5, with fine-tuning based on pediatric radiotherapy guides from various institutions. Additionally, a radiation oncologist prepared answers to these questions. Finally, a multi-institutional group of nine pediatric radiotherapy experts conducted a blind review of responses, assessing reliability, concision, and comprehensibility.

Results

The radiation oncologist and GPT-4 provided the highest-quality responses, though GPT-4's answers were often excessively verbose. While fine-tuned GPT-3.5 generally outperformed basic GPT-3.5, it often provided overly simplistic answers. Inadequate responses were rare, occurring in 4% of GPT-generated responses across all models, primarily due to GPT-3.5 generating excessively long responses.

Conclusions

LLMs can be valuable tools for educating patients and their families before treatment in pediatric radiation oncology. Among them, only GPT-4 provides information of a quality comparable to that of a radiation oncologist, although it still occasionally generates poor-quality responses. GPT-3.5 models should be used cautiously, as they are more likely to produce inadequate answers to patient questions.

Risk-Stratified Radiotherapy in Pediatric Cancer

While the cure rate of cancer in children has markedly improved in the last few decades, late effects continue to be a problem in survivors. Radiotherapy, which is a major component of treatment in many cancers, is one of the major agents responsible for late toxicity. In the past decade, radiotherapy has been omitted in patients achieving excellent response to chemotherapy, such as in Hodgkin lymphoma and some Wilms tumors with lung metastases. Likewise, response to chemotherapy has been used to determine whether lower doses of radiation can be delivered in intracranial germinoma and pediatric nasopharyngeal carcinoma. Molecular subtyping in medulloblastoma is currently being employed, and in WNT-pathway M0 tumors, the reduction in radiotherapy dose to the craniospinal axis and tumor bed is currently being investigated. Finally, dose escalation was recently evaluated in patients with rhabdomyosarcoma > 5 cm who do not achieve a complete response to initial 9 weeks of chemotherapy as well as for unresectable Ewing sarcoma patients to improve local control.

Oral Manifestations: A Warning-Sign in Children with Hematological Disease Acute Lymphocytic Leukemia

Acute lymphocytic leukemia (ALL) is the most frequent form of all childhood leukemias, mostly affecting children between 2 and 4 years old. Oral symptoms, such as mouth ulcers, mucositis, xerostomia, Herpes or Candidiasis, gingival enlargement and bleeding, petechiae, erythema, mucosal pallor and atrophic glossitis, are very common symptoms of ALL and can be early signs of the disease. Secondary and tertiary complications, a direct effect of chemo and radiotherapy, are associated with more severe bleeding, higher susceptibility to infections, ulcerations, inflammation of the mucous membranes, osteoradionecrosis, xerostomia, taste alterations, trismus, carious lesions and dental abnormalities. Immunotherapy, though less toxic, causes oral dysesthesia and pain. Overall, the effects in the oral cavity are transient but there are long-term consequences like caries, periodontal disease and tooth loss that impair endodontic and orthodontic treatments. Also, dental abnormalities resulting from disturbed odontogenesis are known to affect a child's quality of life. The medical dentist should identify these complications and perform appropriate oral care in tandem with other health professionals. Thus, poor oral hygiene can lead to systemic ALL complications. The aim of this review is to describe the oral complications in children with ALL who are undergoing chemo, radio or immunotherapy.

Dosiomics-Based Prediction of Radiation-Induced Valvulopathy after Childhood Cancer

Valvular Heart Disease (VHD) is a known late complication of radiotherapy for childhood cancer (CC), and identifying high-risk survivors correctly remains a challenge. This paper focuses on the distribution of the radiation dose absorbed by heart tissues. We propose that a dosiomics signature could provide insight into the spatial characteristics of the heart dose associated with a VHD, beyond the already-established risk induced by high doses. We analyzed data from the 7670 survivors of the French Childhood Cancer Survivors' Study (FCCSS), 3902 of whom were treated with radiotherapy. In all, 63 (1.6%) survivors that had been treated with radiotherapy experienced a VHD, and 57 of them had heterogeneous heart doses. From the heart-dose distribution of each survivor, we extracted 93 first-order and spatial dosiomics features. We trained random forest algorithms adapted for imbalanced classification and evaluated their predictive performance compared to the performance of standard mean heart dose (MHD)-based models. Sensitivity analyses were also conducted for sub-populations of survivors with spatially heterogeneous heart doses. Our results suggest that MHD and dosiomics-based models performed equally well globally in our cohort and that, when considering the sub-population having received a spatially heterogeneous dose distribution, the predictive capability of the models is significantly improved by the use of the dosiomics features. If these findings are further validated, the dosiomics signature may be incorporated into machine learning algorithms for radiation-induced VHD risk assessment and, in turn, into the personalized refinement of follow-up guidelines.

Elucidating the neurological mechanism of the FLASH effect in juvenile mice exposed to hypofractionated radiotherapy

BACKGROUND

Ultrahigh dose-rate radiotherapy (FLASH-RT) affords improvements in the therapeutic index by minimizing normal tissue toxicities without compromising antitumor efficacy compared to conventional dose-rate radiotherapy (CONV-RT). To investigate the translational potential of FLASH-RT to a human pediatric medulloblastoma brain tumor, we used a radiosensitive juvenile mouse model to assess adverse long-term neurological outcomes.

METHODS

Cohorts of 3-week-old male and female C57Bl/6 mice exposed to hypofractionated (2 × 10 Gy, FLASH-RT or CONV-RT) whole brain irradiation and unirradiated controls underwent behavioral testing to ascertain cognitive status four months posttreatment. Animals were sacrificed 6 months post-irradiation and tissues were analyzed for neurological and cerebrovascular decrements.

RESULTS

The neurological impact of FLASH-RT was analyzed over a 6-month follow-up. FLASH-RT ameliorated neurocognitive decrements induced by CONV-RT and preserved synaptic plasticity and integrity at the electrophysiological (long-term potentiation), molecular (synaptophysin), and structural (Bassoon/Homer-1 bouton) levels in multiple brain regions. The benefits of FLASH-RT were also linked to reduced neuroinflammation (activated microglia) and the preservation of the cerebrovascular structure, by maintaining aquaporin-4 levels and minimizing microglia colocalized to vessels.

CONCLUSIONS

Hypofractionated FLASH-RT affords significant and long-term normal tissue protection in the radiosensitive juvenile mouse brain when compared to CONV-RT. The capability of FLASH-RT to preserve critical cognitive outcomes and electrophysiological properties over 6-months is noteworthy and highlights its potential for resolving long-standing complications faced by pediatric brain tumor survivors. While care must be exercised before clinical translation is realized, present findings document the marked benefits of FLASH-RT that extend from synapse to cognition and the microvasculature.

Developing H3K27M mutant selective radiosensitization strategies in diffuse intrinsic pontine glioma

Diffuse intrinsic pontine glioma (DIPG) is a rare but highly lethal pediatric and adolescent tumor located in the pons of the brainstem. DIPGs harbor unique and specific pathological and molecular alterations, such as the hallmark lysine 27-to-methionine (H3K27M) mutation in histone H3, which lead to global changes in the epigenetic landscape and drive tumorigenesis. While fractionated radiotherapy, the current standard of care, improves symptoms and delays tumor progression, DIPGs inevitably recur, and despite extensive efforts chemotherapy-driven radiosensitization strategies have failed to improve survival. Advances in our understanding of the role of epigenetics in the cellular response to radiation-induced DNA damage, however, offer new opportunities to develop combinational therapeutic strategies selective for DIPGs expressing H3K27M. In this review, we provide an overview of preclinical studies that explore potential radiosensitization strategies targeting the unique epigenetic landscape of H3K27M mutant DIPG. We further discuss opportunities to selectively radiosensitize DIPG through strategic inhibition of the radiation-induced DNA damage response. Finally, we discuss the potential for using radiation to induce anti-tumor immune responses that may be potentiated in DIPG by radiosensitizing-therapeutic strategies.

[Medulloblastoma: The latest major advances]

Medulloblastoma (MB) is a malignant brain tumor that mainly affects children. It is rarely found in adults. Among the four groups of MB defined today according to molecular characteristics, group 3 is the least favorable with an overall survival rate of 50 %. Current treatments, based on surgery, radiotherapy, and chemotherapy, are not sufficiently adapted to the different characteristics of the four MB groups. However, the use of new cellular and animal models has opened new doors to interesting therapeutic avenues. In this review, we detail recent advances in MB research, with a focus on the genes and pathways that drive tumorigenesis, with particular emphasis on the animal models that have been developed to study tumor biology, as well as advances in new targeted therapies.

Artificial Intelligence Algorithm-Based Intraoperative Magnetic Resonance Navigation for Glioma Resection

The study aimed to analyze the application value of artificial intelligence algorithm-based intraoperative magnetic resonance imaging (iMRI) in neurosurgical glioma resection. 108 patients with glioma in a hospital were selected and divided into the experimental group (intraoperative magnetic resonance assisted glioma resection) and the control group (conventional surgical experience resection), with 54 patients in each group. After the resection, the tumor resection rate, NIHSS (National Institute of Health Stroke Scale) score, Karnofsky score, and postoperative intracranial infection were calculated in the two groups. The results revealed that the average tumor resection rate in the experimental group was significantly higher than that in the control group (P < 0.05). There was no significant difference in Karnofsky score before and after the operation in the experimental group (P > 0.05). There was no significant difference in NIHSS score between the experimental group and the control group after resection (P > 0.05). The number of patients with postoperative neurological deficits in the experimental group was smaller than that in the control group. In addition, there was no significant difference in infection rates between the two groups after glioma resection (P > 0.05). In summary, intraoperative magnetic resonance navigation on the basis of a segmentation dictionary learning algorithm has great clinical value in neurosurgical glioma resection. It can maximize the removal of tumors and ensure the integrity of neurological function while avoiding an increased risk of postoperative infection, which is of great significance for the treatment of glioma.