Publications by Year: 2025

BACKGROUND: Lymphangioleiomyomatosis is an incurable cystic lung disease primarily affecting women whose pathophysiologic features are only partially elucidated. Vascular endothelial growth factor D (VEGF-D) is a diagnostic biomarker for lymphangioleiomyomatosis, but about one-third of patients have low levels, necessitating invasive lung biopsy.

RESEARCH QUESTION: Does high-throughput proteomics reveal new diagnostic biomarkers and disease mechanisms in lymphangioleiomyomatosis?

STUDY DESIGN AND METHODS: We used the high-throughput proteomic Somascan assay (Somalogic) to measure approximately 10,000 plasma proteins in 20 patients with lymphangioleiomyomatosis and 10 healthy control participants. Enzyme-linked immunosorbent assay (ELISA) confirmed key protein levels in a validation cohort for a total of 53 patients with lymphangioleiomyomatosis and 64 control participants. A least absolute shrinkage and selection operator logistic regression model was used to develop a multiprotein diagnostic score.

RESULTS: We identified 662 differentially expressed proteins. Matrix metalloproteinase 8 (MMP8) was the top upregulated protein in the plasma of patients with lymphangioleiomyomatosis, and neutrophil degranulation was the most enriched pathway. Fifty-eight neutrophil proteins were enriched in lymphangioleiomyomatosis plasma. Premelanosome protein (PMEL) was 2-fold higher in the plasma of patients with lymphangioleiomyomatosis (log2-fold change, 1.47; P = .006), and it was expressed selectively by lymphangioleiomyomatosis cells. ELISA analysis confirmed that plasma MMP8 levels were elevated significantly in lymphangioleiomyomatosis (662 pg/mL vs 255 pg/mL in control participants) and were correlated negatively with baseline FEV1 % predicted (P = .0014). A 6-protein diagnostic score (the LAMScore) achieved 100% specificity (area under the receiver operating characteristic curve, 0.997) in discriminating patients with lymphangioleiomyomatosis from control participants, including those with low VEGF-D.

INTERPRETATION: Our comprehensive proteomic analysis highlights the potential role of neutrophils in lymphangioleiomyomatosis pathogenesis, with MMP8 and other proteases as potential drivers of lung destruction. PMEL and the 6-protein LAMScore are promising biomarkers for lymphangioleiomyomatosis, especially in patients with low VEGF-D levels. The high precision and reproducibility, the wide array of analytes measured, and the lowering costs of proteomics platforms indicate that in the future, lymphangioleiomyomatosis diagnosis could rely on clinically approved custom analyte panels without lung biopsy.

In the past, many health reform efforts have been touted as a means to tangentially reduce healthcare disparities. Few have shown any demonstrable efficacy in this arena. There is reasonable concern that the machinations of Medicare's Transforming Episode Accountability Model (TEAM) may also exert unintended effects on health access and delivery, potentially worsening existent disparities for racial and ethnic minorities as well as other vulnerable populations. As TEAM has yet to be implemented, this review intends to prognosticate potential pitfalls and behaviors that may be motivated by this project that could otherwise lead to worsening healthcare disparities within spine fusion care. We present a narrative review with our prognostications regarding mechanisms and behaviors that may be influenced by TEAM and that could result in worsening healthcare disparities and/or reduced access to care for vulnerable populations. These are informed by published experiences with other health reform efforts including centers of excellence, bundled payment programs, Accountable Care Organizations and Comprehensive Care for Joint Replacement. Based on previous published experiences with similar health reform initiatives, we believe there are several areas in which TEAM may potentiate or worsen existing healthcare disparities. These include the areas of access to care, undertreatment and healthcare segregation, as well as adverse behaviors such as cherry picking, lemon dropping and asymmetric pressure on small hospitals and safety-net institutions. There remain several aspects of TEAM that could limit access to care and aggravate healthcare disparities. Some of these behaviors could result in implicit or explicit undertreatment, restricted access to care and worsened healthcare segregation with negative feedback loops that continue to syphon resources from smaller hospitals and safety-net hospitals resulting in deterioration in the quality of care and general health of the already vulnerable populations these facilities serve.

Functional neurological disorders (FNDs) are neurological conditions resulting from altered brain network activity causing physical symptoms that are genuine but not explained by structural changes in the brain. FND results from abnormal connectivity in the limbic system and overlapping circuitry dysfunction in salience networks. The autonomic nervous system (ANS) refers to the part of the nervous system devoted to unconscious processes, the viscera and homeostasis. The ANS has afferent pathways, central nuclei and networks, and efferent pathways. Since unconscious neural processing and automatic behaviors are under the purview of the ANS, there is great interest in understanding the role of abnormal ANS activity in FND. To date, the overlap between ANS dysfunction and FND has been relatively underexplored. Here, we discuss the role of the ANS in FND and the overlap between autonomic dysfunction and FND.

Polygenic risk scores (PRSs) have generated considerable interest as a means of personalizing prostate cancer (PC) screening by stratifying individuals according to their genetic risk. The single-arm BARCODE1 study recently showed that PRS-based screening detected more PCs than prostate-specific antigen (PSA) testing or magnetic resonance imaging (MRI). The absence of a comparator arm limits the interpretability of this finding. We compared outcomes from BARCODE1 with those from two contemporaneous, large-scale screening trials-Göteborg-2 and ProScreen-that used MRI with or without a PC blood marker to risk-stratify men for biopsy. When standardized to 10 000 men tested, BARCODE1 biopsied more men (704 vs 386 and 338), diagnosed more low-grade PCs (126 vs 103 and 41), and detected fewer high-grade PCs (155 vs 178 and 165) versus Göteborg-2 and ProScreen. Combining PRS with PSA or MRI in BARCODE1 reduced the detection of high-grade PCs by 50-75% in comparison to Göteborg-2 and ProScreen. These findings reflect the limited risk discrimination of PRSs and their inability, unlike MRI and blood-based markers, to preferentially detect aggressive disease. PRS-based PC screening underperforms relative to current best practice and, on the basis of BARCODE1 data, should not be adopted in clinical practice.

Physical activity is important in the development of mental and physical well-being during childhood. We aimed to investigate physical activity levels among early pubertal youth presenting for gender-affirming care. Among 93 participants 8-14 years of age, who had physical activity level measured using the Physical Activity Questionnaire for Children (PAQ-C), participants designated male at birth had lower levels of physical activity than participants designated female at birth both before starting gonadotropin-releasing hormone analog treatment and during gender-affirming medical treatment (baseline PAQ-C scores 2.24±0.73 vs. 2.67±0.73, p=0.006).

INTRODUCTION: Accurate segmentation of brain regions in magnetic resonance imaging (MRI) is essential for diagnosing and managing neurological diseases. FreeSurfer is a widely used tool for brain MRI segmentation, but its limitations in speed and usability pose challenges in clinical practice. Neurophet AQUA, an advanced automated segmentation tool, aims to overcome these challenges by offering rapid and reliable segmentation. This study evaluates two segmentation pipelines: (1) a T1-based brain region segmentation pipeline, comparing the performance and reliability of Neurophet AQUA and FreeSurfer v7.3.2 using the standard recon-all pipeline in segmenting gray matter, white matter, and subcortical structures; and (2) a T2-FLAIR-based white matter lesion segmentation pipeline of Neurophet AQUA, assessing the detection of white matter hyperintensities (WMH).

METHODS: Four main datasets were used. For the T1-based segmentation pipeline, the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset was used to compare the segmentation results of Neurophet AQUA and FreeSurfer, with quality assessed by expert evaluation. The MarkVCID dataset was used to evaluate the scan-rescan repeatability and inter-scanner reproducibility of Neurophet AQUA. For the T2-FLAIR-based pipeline, WMH segmentation performance was assessed using 2D and 3D FLAIR sequences from the ADNI dataset by comparing the segmentations to ground truth (GT) labels and calculating Dice similarity coefficients (DSC).

RESULTS: Segmentation quality and reliability showed that Neurophet AQUA and FreeSurfer achieved comparable performance in most regions, with no significant differences. However, Neurophet AQUA had significantly faster processing time. In intracranial volume (ICV) measurements, Neurophet AQUA showed better repeatability than FreeSurfer in both rescans (ICC: 0.999 vs. 0.991) and inter-scanner settings (ICC: 0.983 vs. 0.866). AQUA also demonstrated consistent WMH segmentation across 2D and 3D FLAIR images.

CONCLUSION: Neurophet AQUA demonstrated high segmentation accuracy and excellent repeatability in rescanned measurements, as well as exploratory evidence of inter-scanner reproducibility on T1-weighted MRI, showing comparable performance to established tools such as FreeSurfer. It also showed consistent WMH segmentation across FLAIR types. Neurophet AQUA is highly suitable for clinical applications that require accurate analysis, high repeatability and reproducibility, and rapid brain MRI processing, making it particularly well-suited for multicenter research studies.

INTRODUCTION: Alzheimer's disease (AD) is the most prevalent form of dementia and a major public health challenge. In the absence of a cure, accurate and innovative early diagnostic methods are essential for proactive life and healthcare planning. Speech metrics have shown promising potential for identifying individuals with mild cognitive impairment (MCI) and AD, prompting investigation into whether speech motor features can detect elevated risk even prior to cognitive decline. This preliminary study examined whether speech kinematic features measured during a color-word interference task could distinguish cognitively normal APOE-ε4 carriers (ε4+) from non-carriers (ε4-).

METHODS: Sixteen cognitively normal older adults (n = 9 ε4+, n = 7 ε4-) completed a sentence-based color-word interference task while three-dimensional electromagnetic articulography recorded lower-lip movements. Lip movement duration (s), average speed (mm/s), and range of movement (mm³) were extracted for three sentence segments: pre-interference, during-interference, and post-interference. Difference measures (ΔDuring-Pre, ΔDuring-Post) were computed to quantify task-related modulation. Descriptive statistics and independent t-tests were used to examine group-level trends. For classification, a support vector machine (SVM) with a degree-2 polynomial kernel and leave-one-out cross-validation evaluated all feature combinations derived from the 15 kinematic measures.

RESULTS: Although no group differences reached statistical significance after accounting for multiple testing, several features showed moderate effect sizes. The optimal SVM model achieved 87.5% cross-validated accuracy (precision 88.9%, sensitivity 88.9%, specificity 85.7%) using three features: (1) lip movement duration during the pre-interference segment, (2) average lip speed during interference, and (3) the change in lip movement range from pre- to during-interference segments (ΔDuring-Pre).

DISCUSSION: These findings suggest that lip kinematic responses to mild cognitive-motor interference may capture subtle neuromotor differences associated with APOE-ε4 status in cognitively intact older adults. The identified features point to potential alterations in anticipatory motor planning, interference susceptibility, and articulatory adaptability in ε4+ individuals. However, the small sample size, risk of overfitting, and sex imbalance limit interpretability. Thus, these results should be viewed as hypothesis-generating. Larger, sex-balanced, and longitudinal studies are needed to validate these candidate markers and clarify their role in multimodal early AD risk stratification.

Purpose: To compare the long-term outcomes of scleral-sutured intraocular lens (IOL) fixation using expanded polytetrafluoroethylene vs polypropylene sutures, with a focus on suture-related complications. Methods: A retrospective comparative review was conducted of 102 patients who underwent scleral-sutured IOL fixation between 2015 and 2019. Forty-eight eyes of 48 patients received 4-point fixation with expanded polytetrafluoroethylene sutures, and 55 eyes of 54 patients received 2-point fixation with polypropylene sutures. Postoperative complications, visual outcomes, and risk factors for suture-related complications were analyzed over a minimum follow-up of 5 years. Results: No instances of suture breakage were observed in either group over an average follow-up of 6.7 years. Suture exposure rates were 10.4% (5/48) for expanded polytetrafluoroethylene and 18.2% (10/55) for polypropylene (P = .27). The mean time to suture exposure was 2.7 ± 3.3 years in the expanded polytetrafluoroethylene group and 2.9 ± 2.5 years in the polypropylene group (P = .88). Mean logMAR visual acuity at final follow-up was 1.0 ± 1.2 in the expanded polytetrafluoroethylene group and 1.5 ± 1.2 in the polypropylene group (P = .06). The only significant risk factor for suture exposure was concurrent corneal transplantation or glaucoma surgery (odds ratio [OR], 9.3; P = .003). Surgical correction was required in all cases of exposure with expanded polytetrafluoroethylene sutures and in 2 cases with polypropylene sutures. One case of suture-associated infectious scleritis with endophthalmitis occurred in the expanded polytetrafluoroethylene group, and 1 case of endophthalmitis related to an explanted corneal graft occurred in the polypropylene group. Conclusions: Both expanded polytetrafluoroethylene and polypropylene sutures demonstrated durable outcomes with similar complication rates. Suture breakage was not observed in either group, and there were no differences in suture exposure rates between the 2 groups. Risk factors for suture exposure included the performance of a concurrent procedure.

BACKGROUNDS: Ovarian cancer represents a deadly gynecological malignancy, with surgical treatment being a key component of its management. We sought to integrate clinical characteristics and ascites immune microenvironment features into a deep learning model to predict postoperative residual tumor status and assist in surgical decision-making.

METHODS: 118 FIGO III/IV high-grade serous ovarian cancer (HGSOC) patients treated at Peking University Third Hospital (2019-2024) were enrolled. Clinical characteristics, surgical methods, and postoperative residual tumor status were collected. Ascites samples were processed via density gradient centrifugation and flow cytometry. Deep learning model was built by fusing clinical and immune data, and its performance was validated under a gradient of feature quantities (5-45 features) to optimize feature selection. Model performance was comprehensively evaluated on a test set (20% of the dataset) using metrics including accuracy, precision, recall, and F1 score, and compared with traditional machine learning models (random forest, XGBoost, et al). Confusion matrices and probability heatmaps were used for visual analysis. For model interpretability, we presented feature importance and results from SHAP analysis.

RESULTS: Our model achieved 70.83% accuracy, 71.21% precision, 70.83% recall, and 70.89% F1 score on the test set, outperforming traditional machine learning models: random forest (accuracy: 64.6%, precision: 65.1%, recall: 64.6%, F1 score: 66.4%), XGBoost (accuracy: 66.7%, precision: 67.0%, recall: 66.7%, F1 score: 66.6%), and logistic regression (accuracy: 58.3%, precision: 59.0%, recall: 58.4%, F1 score: 58.2%). It demonstrated strong performance in identifying high-risk R2 cases but showed limitations in differentiating between R0 and R1 statuses. Probability heatmaps visualized the distribution of R0, R1, and R2 probabilities under different surgical methods, facilitating intuitive clinical reference. Interpretability analysis via permutation feature importance and SHAP highlighted the critical role of surgical methods and specific immune microenvironment features in predictive outcomes.

CONCLUSION: This study developed a novel deep learning-based model to predict postoperative residual tumor probability, integrating clinical and immune microenvironment data. While the model excelled in identifying high-risk cases (e.g., R2), further optimization is needed to improve R0 and R1 differentiation. Future research should expand datasets and integrate multi-omics data to enhance predictive accuracy and clinical applicability.

Diabetic cardiomyopathy (DCM) has long been considered as a left ventricular (LV) disease with diastolic dysfunction preceding systolic dysfunction in diabetes. However, it is increasingly recognized that the right ventricle (RV) is also affected by diabetes and may be independently responsible for adverse outcomes in diabetic patients with or without LV failure. Yu et al conducted a 30-week longitudinal evaluation of biventricular function and pathology in OVE26 diabetic mice and revealed early diastolic dysfunction preceding systolic decline, suggesting that early LV diastolic impairment precedes the later onset of systolic dysfunction. With age, the animals developed fibrosis, hypertrophy, and pulmonary arterial hypertension in the RV. The purpose of this editorial is to contextualize these findings within the existing literature by highlighting the interplay between cardiac chambers and the vasculature. We also seek to reiterate that DCM is a condition extending beyond left ventricular dysfunction. As the authors note, the right side of the heart may remain "the forgotten ventricle" in diabetic patients. We hope that the mechanisms discussed in this paper will help researchers to understand the pathogenesis of cardiovascular disease in this context and encourage clinicians to be more attentive to the associated clinical symptoms.