Foundation model for cancer imaging biomarkers.

Foundation models in deep learning are characterized by a single large-scale model trained on vast amounts of data serving as the foundation for various downstream tasks. Foundation models are generally trained using self-supervised learning and excel in reducing the demand for training samples in downstream applications. This is especially important in medicine, where large labelled datasets are often scarce. Here, we developed a foundation model for cancer imaging biomarker discovery by training a convolutional encoder through self-supervised learning using a comprehensive dataset of 11,467 radiographic lesions. The foundation model was evaluated in distinct and clinically relevant applications of cancer imaging-based biomarkers. We found that it facilitated better and more efficient learning of imaging biomarkers and yielded task-specific models that significantly outperformed conventional supervised and other state-of-the-art pretrained implementations on downstream tasks, especially when training dataset sizes were very limited. Furthermore, the foundation model was more stable to input variations and showed strong associations with underlying biology. Our results demonstrate the tremendous potential of foundation models in discovering new imaging biomarkers that may extend to other clinical use cases and can accelerate the widespread translation of imaging biomarkers into clinical settings.

Bypassing the build-up phase for oral immunotherapy in shrimp-allergic children.

Background: Oral immunotherapy is an effective treatment for food allergies; however, its use in clinical practice is limited by resources and lack of standardized protocols for foods other than peanut. Previous studies have suggested that shrimp has a higher threshold for reaction than other allergenic foods, suggesting it may be safe to directly administer maintenance doses of immunotherapy. Methods: Children aged 3-17 years who had 1) skin prick test ≥3 mm and/or specific IgE level ≥0.35 kU/L and convincing objective IgE-mediated reaction to shrimp, or 2) no ingestion history and specific IgE level ≥5 kU/L, underwent a low-dose oral food challenge to 300 mg shrimp protein, with the goal of continuing daily ingestion of the 300 mg maintenance dose as oral immunotherapy. Results: Between January 2020 and April 2023, 17 children completed the low-dose oral food challenge. Nine (53%) tolerated this amount with no reaction, and 8 (47%) had a mild reaction (isolated oral pruritis or redness on chin). Sixteen (94%) continued maintenance low-dose oral immunotherapy eating 300 mg shrimp protein daily. None of the patients developed anaphylaxis related to the immunotherapy. Conclusion: Our case series suggests that some shrimp allergic patients being considered for oral immunotherapy should be offered a low-dose oral food challenge, to potentially bypass the build-up phase of immunotherapy.

Safety and Effectiveness of Bypassing Oral Immunotherapy Buildup With an Initial Phase of Sublingual Immunotherapy for Higher-Risk Food Allergy.

BACKGROUND: Because of its favorable safety, sublingual immunotherapy (SLIT) for food allergy has been proposed as an alternative treatment for those in whom oral immunotherapy (OIT) is of higher risk-older children, adolescents, adults, and those with a history of severe reactions. Although safe, SLIT has been shown to be less effective than OIT. OBJECTIVE: To describe the safety of multifood SLIT in pediatric patients aged 4 to 18 years and the effectiveness of bypassing OIT buildup with an initial phase of SLIT. METHODS: Patients aged 4 to 18 years were offered (multi)food SLIT. Patients built up to 2 mg protein SLIT maintenance over the course of 3 to 5 visits under nurse supervision. After 1 to 2 years of daily SLIT maintenance, patients were offered a low-dose oral food challenge (OFC) (cumulative dose, 300 mg protein) with the goal of bypassing OIT buildup. RESULTS: Between summer 2020 and winter 2023, 188 patients were enrolled in SLIT (median age, 11 years). Four patients (2.10%) received epinephrine during buildup and went to the emergency department, but none experienced grade 4 (severe) reaction. A subset of 20 patients had 50 low-dose OFCs to 300 mg protein and 35 (70%) OFCs were successful, thereby bypassing OIT buildup. CONCLUSIONS: In combination with very favorable safety of SLIT, with no life-threatening reactions and few reactions requiring epinephrine, we propose that an initial phase of SLIT to bypass supervised OIT buildup be considered for children in whom OIT is considered to be of higher risk.

Edge roughness quantifies impact of physician variation on training and performance of deep learning auto-segmentation models for the esophagus.

Manual segmentation of tumors and organs-at-risk (OAR) in 3D imaging for radiation-therapy planning is time-consuming and subject to variation between different observers. Artificial intelligence (AI) can assist with segmentation, but challenges exist in ensuring high-quality segmentation, especially for small, variable structures, such as the esophagus. We investigated the effect of variation in segmentation quality and style of physicians for training deep-learning models for esophagus segmentation and proposed a new metric, edge roughness, for evaluating/quantifying slice-to-slice inconsistency. This study includes a real-world cohort of 394 patients who each received radiation therapy (mainly for lung cancer). Segmentation of the esophagus was performed by 8 physicians as part of routine clinical care. We evaluated manual segmentation by comparing the length and edge roughness of segmentations among physicians to analyze inconsistencies. We trained eight multiple- and individual-physician segmentation models in total, based on U-Net architectures and residual backbones. We used the volumetric Dice coefficient to measure the performance for each model. We proposed a metric, edge roughness, to quantify the shift of segmentation among adjacent slices by calculating the curvature of edges of the 2D sagittal- and coronal-view projections. The auto-segmentation model trained on multiple physicians (MD1-7) achieved the highest mean Dice of 73.7 ± 14.8%. The individual-physician model (MD7) with the highest edge roughness (mean ± SD: 0.106 ± 0.016) demonstrated significantly lower volumetric Dice for test cases compared with other individual models (MD7: 58.5 ± 15.8%, MD6: 67.1 ± 16.8%, p < 0.001). A multiple-physician model trained after removing the MD7 data resulted in fewer outliers (e.g., Dice ≤ 40%: 4 cases for MD1-6, 7 cases for MD1-7, Ntotal = 394). While we initially detected this pattern in a single clinician, we validated the edge roughness metric across the entire dataset. The model trained with the lowest-quantile edge roughness (MDER-Q1, Ntrain = 62) achieved significantly higher Dice (Ntest = 270) than the model trained with the highest-quantile ones (MDER-Q4, Ntrain = 62) (MDER-Q1: 67.8 ± 14.8%, MDER-Q4: 62.8 ± 15.7%, p < 0.001). This study demonstrates that there is significant variation in style and quality in manual segmentations in clinical care, and that training AI auto-segmentation algorithms from real-world, clinical datasets may result in unexpectedly under-performing algorithms with the inclusion of outliers. Importantly, this study provides a novel evaluation metric, edge roughness, to quantify physician variation in segmentation which will allow developers to filter clinical training data to optimize model performance.

Large language models to identify social determinants of health in electronic health records.

Social determinants of health (SDoH) play a critical role in patient outcomes, yet their documentation is often missing or incomplete in the structured data of electronic health records (EHRs). Large language models (LLMs) could enable high-throughput extraction of SDoH from the EHR to support research and clinical care. However, class imbalance and data limitations present challenges for this sparsely documented yet critical information. Here, we investigated the optimal methods for using LLMs to extract six SDoH categories from narrative text in the EHR: employment, housing, transportation, parental status, relationship, and social support. The best-performing models were fine-tuned Flan-T5 XL for any SDoH mentions (macro-F1 0.71), and Flan-T5 XXL for adverse SDoH mentions (macro-F1 0.70). Adding LLM-generated synthetic data to training varied across models and architecture, but improved the performance of smaller Flan-T5 models (delta F1 + 0.12 to +0.23). Our best-fine-tuned models outperformed zero- and few-shot performance of ChatGPT-family models in the zero- and few-shot setting, except GPT4 with 10-shot prompting for adverse SDoH. Fine-tuned models were less likely than ChatGPT to change their prediction when race/ethnicity and gender descriptors were added to the text, suggesting less algorithmic bias (p < 0.05). Our models identified 93.8% of patients with adverse SDoH, while ICD-10 codes captured 2.0%. These results demonstrate the potential of LLMs in improving real-world evidence on SDoH and assisting in identifying patients who could benefit from resource support.

Widening the therapeutic window for central and ultra-central thoracic oligometastatic disease with stereotactic MR-guided adaptive radiation therapy (SMART).

BACKGROUND/PURPOSE: Central/ultra-central thoracic tumors are challenging to treat with stereotactic radiotherapy due potential high-grade toxicity. Stereotactic MR-guided adaptive radiation therapy (SMART) may improve the therapeutic window through motion control with breath-hold gating and real-time MR-imaging as well as the option for daily online adaptive replanning to account for changes in target and/or organ-at-risk (OAR) location. MATERIALS/METHODS: 26 central (19 ultra-central) thoracic oligoprogressive/oligometastatic tumors treated with isotoxic (OAR constraints-driven) 5-fraction SMART (median 50 Gy, range 35-60) between 10/2019-10/2022 were reviewed. Central tumor was defined as tumor within or touching 2 cm around proximal tracheobronchial tree (PBT) or adjacent to mediastinal/pericardial pleura. Ultra-central was defined as tumor abutting the PBT, esophagus, or great vessel. Hard OAR constraints observed were ≤ 0.03 cc for PBT V40, great vessel V52.5, and esophagus V35. Local failure was defined as tumor progression/recurrence within the planning target volume. RESULTS: Tumor abutted the PBT in 31 %, esophagus in 31 %, great vessel in 65 %, and heart in 42 % of cases. 96 % of fractions were treated with reoptimized plan, necessary to meet OAR constraints (80 %) and/or target coverage (20 %). Median follow-up was 19 months (27 months among surviving patients). Local control (LC) was 96 % at 1-year and 90 % at 2-years (total 2/26 local failure). 23 % had G2 acute toxicities (esophagitis, dysphagia, anorexia, nausea) and one (4 %) had G3 acute radiation dermatitis. There were no G4-5 acute toxicities. There was no symptomatic pneumonitis and no G2 + late toxicities. CONCLUSION: Isotoxic 5-fraction SMART resulted in high rates of LC and minimal toxicity. This approach may widen the therapeutic window for high-risk oligoprogressive/oligometastatic thoracic tumors.

Stereotactic Magnetic Resonance-guided Adaptive Radiation Therapy for Localized Kidney Cancer: Early Outcomes from a Prospective Phase 1 Trial and Supplemental Cohort.

Stereotactic magnetic resonance (MR)-guided adaptive radiotherapy (SMART) for renal cell carcinoma may result in more precise treatment delivery through the capabilities for improved image quality, daily adaptive planning, and accounting for respiratory motion during treatment with real-time MR tracking. In this study, we aimed to characterize the safety and feasibility of SMART for localized kidney cancer. Twenty patients with localized kidney cancer (ten treated in a prospective phase 1 trial and ten in the supplemental cohort) were treated to 40 Gy in five fractions on a 0.35 T MR-guided linear accelerator with daily adaptive planning and a cine MR-guided inspiratory breath hold technique. The median follow-up time was 17 mo (interquartile range: 13-20 months). A single patient developed local failure at 30 mo. No grade ≥3 adverse events were reported. The mean decrease in estimated glomerular filtration rate was -1.8 ml/min/1.73 m2 (95% confidence interval or CI [-6.6 to 3.1 ml/min/1.73 m2]), and the mean decrease in tumor diameter was -0.20 cm (95% CI [-0.6 to 0.2 cm]) at the last follow-up. Anterior location and overlap of the 25 or 28 Gy isodose line with gastrointestinal organs at risk were predictive of the benefit from online adaptive planning. Kidney SMART is feasible and, at the early time point evaluated in this study, was well tolerated with minimal decline in renal function. More studies are warranted to further evaluate the safety and efficacy of this technique. PATIENT SUMMARY: For patients with localized renal cell carcinoma who are not surgical candidates, stereotactic magnetic resonance--guided adaptive radiotherapy is a feasible and safe noninvasive treatment option that results in minimal impact on kidney function.

Automated temporalis muscle quantification and growth charts for children through adulthood.

Lean muscle mass (LMM) is an important aspect of human health. Temporalis muscle thickness is a promising LMM marker but has had limited utility due to its unknown normal growth trajectory and reference ranges and lack of standardized measurement. Here, we develop an automated deep learning pipeline to accurately measure temporalis muscle thickness (iTMT) from routine brain magnetic resonance imaging (MRI). We apply iTMT to 23,876 MRIs of healthy subjects, ages 4 through 35, and generate sex-specific iTMT normal growth charts with percentiles. We find that iTMT was associated with specific physiologic traits, including caloric intake, physical activity, sex hormone levels, and presence of malignancy. We validate iTMT across multiple demographic groups and in children with brain tumors and demonstrate feasibility for individualized longitudinal monitoring. The iTMT pipeline provides unprecedented insights into temporalis muscle growth during human development and enables the use of LMM tracking to inform clinical decision-making.

Safety and effectiveness of the Canadian food ladders for children with IgE-mediated food allergies to cow's milk and/or egg.

BACKGROUND: Food ladders are tools designed to facilitate home-based dietary advancement in children with food allergies through stepwise exposures to increasingly allergenic forms of milk and egg. Several studies have now documented safety and efficacy of food ladders. In 2021, we published a Canadian adaptation of the previously existing milk and egg ladders originating in Europe using foods more readily available/consumed in Canada. Our study adds to the growing body of evidence supporting food ladder use and provides safety and effectiveness data for our Canadian adaptation of the milk and egg ladders. METHODS: Surveys were distributed to families of children using the Canadian Milk Ladder and/or the Canadian Egg Ladder at baseline, with follow up surveys at 3 months, 6 months, and 12 months. Data were analyzed using REDCap and descriptive and inferential statistics are presented. RESULTS: One hundred and nine participants were started on milk/egg ladders between September 2020 and June 2022. 53 participants responded to follow up surveys. Only 2 of 53 (3.8%) participants reported receiving epinephrine during the study. Severe grade 4 reactions (defined according to the modified World Allergy Organization grading system) were not reported by any participants. Minor cutaneous adverse reactions were common, with about 71% (n = 10/14) of respondents reporting cutaneous adverse reactions by 1 year of food ladder use. An increasing proportion of participants could tolerate most foods from steps 2-4 foods after 3, 6, and 12 months of the food ladder compared to baseline. CONCLUSION: The Canadian food ladders are safe tools for children with cow's milk and/or egg allergies, and participants tolerated a larger range of foods with food ladder use compared to baseline.

Risk factors associated with safety of preschool peanut oral immunotherapy.

Background: An understanding of how patient characteristics such as age, baseline peanut-specific IgE, and atopic comorbidities may influence potential safety outcomes during peanut oral immunotherapy (P-OIT) could aid in shared decision making between clinicians and patient families. Objective: This study explored the relationship between baseline patient characteristics and reactions during P-OIT using a large sample size to better understand potential risk factors influencing P-OIT safety. Methods: Data were obtained from the Food Allergy Immunotherapy (FAIT) registry, which collects real-world OIT data from community and academic allergy clinics across Canada. Multivariable logistic regression modeling was performed to examine the relationship between baseline patient characteristics and reactions during P-OIT. Multiple imputation was applied to reduce potential bias caused by missingness and to maximize the use of available information to preserve statistical power. Results: Between April 2017 and June 2021, a total of 653 eligible patients initiated P-OIT. Multivariable regression analysis showed pre-OIT grade 2+ initial reaction (odds ratio [OR] = 1.33, 95% confidence interval [CI] 1.10, 1.61), allergic rhinitis (OR = 1.60, 95% CI 1.08, 2.38), older age (OR = 1.01, 95% CI 1.00, 1.02), and higher baseline peanut-specific IgE (OR = 1.02, 95% CI 1.02, 1.03) were associated with grade 2+ reaction during P-OIT after adjusting for potential risk factors. Conclusion: Our study identified several clinically important risk factors for grade 2+ reactions during P-OIT: pre-OIT grade 2+ initial reaction, allergic rhinitis, older age, and higher baseline peanut-specific IgE. These results highlight the need for individualized risk stratification for OIT.

Decoding biological age from face photographs using deep learning.

Because humans age at different rates, a person's physical appearance may yield insights into their biological age and physiological health more reliably than their chronological age. In medicine, however, appearance is incorporated into medical judgments in a subjective and non-standardized fashion. In this study, we developed and validated FaceAge, a deep learning system to estimate biological age from easily obtainable and low-cost face photographs. FaceAge was trained on data from 58,851 healthy individuals, and clinical utility was evaluated on data from 6,196 patients with cancer diagnoses from two institutions in the United States and The Netherlands. To assess the prognostic relevance of FaceAge estimation, we performed Kaplan Meier survival analysis. To test a relevant clinical application of FaceAge, we assessed the performance of FaceAge in end-of-life patients with metastatic cancer who received palliative treatment by incorporating FaceAge into clinical prediction models. We found that, on average, cancer patients look older than their chronological age, and looking older is correlated with worse overall survival. FaceAge demonstrated significant independent prognostic performance in a range of cancer types and stages. We found that FaceAge can improve physicians' survival predictions in incurable patients receiving palliative treatments, highlighting the clinical utility of the algorithm to support end-of-life decision-making. FaceAge was also significantly associated with molecular mechanisms of senescence through gene analysis, while age was not. These findings may extend to diseases beyond cancer, motivating using deep learning algorithms to translate a patient's visual appearance into objective, quantitative, and clinically useful measures.

Pneumonitis in Patients Receiving Thoracic Radiotherapy and Osimertinib: A Multi-Institutional Study.

Introduction: Thoracic radiotherapy (TRT) is increasingly used in patients receiving osimertinib for advanced NSCLC, and the risk of pneumonitis is not established. We investigated the risk of pneumonitis and potential risk factors in this population. Methods: We performed a multi-institutional retrospective analysis of patients under active treatment with osimertinib who received TRT between April 2016 and July 2022 at two institutions. Clinical characteristics, including whether osimertinib was held during TRT and pneumonitis incidence and grade (Common Terminology Criteria for Adverse Events version 5.0) were documented. Logistic regression analysis was performed to identify risk factors associated with grade 2 or higher (2+) pneumonitis. Results: The median follow-up was 10.2 months (range: 1.9-53.2). Of 102 patients, 14 (13.7%) developed grade 2+ pneumonitis, with a median time to pneumonitis of 3.2 months (range: 1.5-6.3). Pneumonitis risk was not significantly increased in patients who continued osimertinib during TRT compared with patients who held osimertinib during TRT (9.1% versus 15.0%, p = 0.729). Three patients (2.9%) had grade 3 pneumonitis, none had grade 4, and two patients had grade 5 events (2.0%, diagnosed 3.2 mo and 4.4 mo post-TRT). Mean lung dose was associated with the development of grade 2+ pneumonitis in multivariate analysis (OR = 1.19, p = 0.021). Conclusions: Although the overall rate of pneumonitis in patients receiving TRT and osimertinib was relatively low, there was a small risk of severe toxicity. The mean lung dose was associated with an increased risk of developing pneumonitis. These findings inform decision-making for patients and providers.

Foundation Models for Quantitative Biomarker Discovery in Cancer Imaging.

Foundation models represent a recent paradigm shift in deep learning, where a single large-scale model trained on vast amounts of data can serve as the foundation for various downstream tasks. Foundation models are generally trained using self-supervised learning and excel in reducing the demand for training samples in downstream applications. This is especially important in medicine, where large labeled datasets are often scarce. Here, we developed a foundation model for imaging biomarker discovery by training a convolutional encoder through self-supervised learning using a comprehensive dataset of 11,467 radiographic lesions. The foundation model was evaluated in distinct and clinically relevant applications of imaging-based biomarkers. We found that they facilitated better and more efficient learning of imaging biomarkers and yielded task-specific models that significantly outperformed their conventional supervised counterparts on downstream tasks. The performance gain was most prominent when training dataset sizes were very limited. Furthermore, foundation models were more stable to input and inter-reader variations and showed stronger associations with underlying biology. Our results demonstrate the tremendous potential of foundation models in discovering novel imaging biomarkers that may extend to other clinical use cases and can accelerate the widespread translation of imaging biomarkers into clinical settings.

Use of Artificial Intelligence Chatbots for Cancer Treatment Information.

This survey study examines the performance of a large language model chatbot in providing cancer treatment recommendations that are concordant with National Comprehensive Cancer Network guidelines.

Development and Validation of an Automated Image-Based Deep Learning Platform for Sarcopenia Assessment in Head and Neck Cancer.

Importance: Sarcopenia is an established prognostic factor in patients with head and neck squamous cell carcinoma (HNSCC); the quantification of sarcopenia assessed by imaging is typically achieved through the skeletal muscle index (SMI), which can be derived from cervical skeletal muscle segmentation and cross-sectional area. However, manual muscle segmentation is labor intensive, prone to interobserver variability, and impractical for large-scale clinical use. Objective: To develop and externally validate a fully automated image-based deep learning platform for cervical vertebral muscle segmentation and SMI calculation and evaluate associations with survival and treatment toxicity outcomes. Design, Setting, and Participants: For this prognostic study, a model development data set was curated from publicly available and deidentified data from patients with HNSCC treated at MD Anderson Cancer Center between January 1, 2003, and December 31, 2013. A total of 899 patients undergoing primary radiation for HNSCC with abdominal computed tomography scans and complete clinical information were selected. An external validation data set was retrospectively collected from patients undergoing primary radiation therapy between January 1, 1996, and December 31, 2013, at Brigham and Women's Hospital. The data analysis was performed between May 1, 2022, and March 31, 2023. Exposure: C3 vertebral skeletal muscle segmentation during radiation therapy for HNSCC. Main Outcomes and Measures: Overall survival and treatment toxicity outcomes of HNSCC. Results: The total patient cohort comprised 899 patients with HNSCC (median [range] age, 58 [24-90] years; 140 female [15.6%] and 755 male [84.0%]). Dice similarity coefficients for the validation set (n = 96) and internal test set (n = 48) were 0.90 (95% CI, 0.90-0.91) and 0.90 (95% CI, 0.89-0.91), respectively, with a mean 96.2% acceptable rate between 2 reviewers on external clinical testing (n = 377). Estimated cross-sectional area and SMI values were associated with manually annotated values (Pearson r = 0.99; P < .001) across data sets. On multivariable Cox proportional hazards regression, SMI-derived sarcopenia was associated with worse overall survival (hazard ratio, 2.05; 95% CI, 1.04-4.04; P = .04) and longer feeding tube duration (median [range], 162 [6-1477] vs 134 [15-1255] days; hazard ratio, 0.66; 95% CI, 0.48-0.89; P = .006) than no sarcopenia. Conclusions and Relevance: This prognostic study's findings show external validation of a fully automated deep learning pipeline to accurately measure sarcopenia in HNSCC and an association with important disease outcomes. The pipeline could enable the integration of sarcopenia assessment into clinical decision making for individuals with HNSCC.

Natural Language Processing to Automatically Extract the Presence and Severity of Esophagitis in Notes of Patients Undergoing Radiotherapy.

PURPOSE: Radiotherapy (RT) toxicities can impair survival and quality of life, yet remain understudied. Real-world evidence holds potential to improve our understanding of toxicities, but toxicity information is often only in clinical notes. We developed natural language processing (NLP) models to identify the presence and severity of esophagitis from notes of patients treated with thoracic RT. METHODS: Our corpus consisted of a gold-labeled data set of 1,524 clinical notes from 124 patients with lung cancer treated with RT, manually annotated for Common Terminology Criteria for Adverse Events (CTCAE) v5.0 esophagitis grade, and a silver-labeled data set of 2,420 notes from 1,832 patients from whom toxicity grades had been collected as structured data during clinical care. We fine-tuned statistical and pretrained Bidirectional Encoder Representations from Transformers-based models for three esophagitis classification tasks: task 1, no esophagitis versus grade 1-3; task 2, grade ≤1 versus >1; and task 3, no esophagitis versus grade 1 versus grade 2-3. Transferability was tested on 345 notes from patients with esophageal cancer undergoing RT. RESULTS: Fine-tuning of PubMedBERT yielded the best performance. The best macro-F1 was 0.92, 0.82, and 0.74 for tasks 1, 2, and 3, respectively. Selecting the most informative note sections during fine-tuning improved macro-F1 by ≥2% for all tasks. Silver-labeled data improved the macro-F1 by ≥3% across all tasks. For the esophageal cancer notes, the best macro-F1 was 0.73, 0.74, and 0.65 for tasks 1, 2, and 3, respectively, without additional fine-tuning. CONCLUSION: To our knowledge, this is the first effort to automatically extract esophagitis toxicity severity according to CTCAE guidelines from clinical notes. This provides proof of concept for NLP-based automated detailed toxicity monitoring in expanded domains.