Multi-center Dose Prediction Using Attention-aware Deep learning Algorithm Based on Transformers for Cervical Cancer Radiotherapy

AIMS

Accurate dose delivery is crucial for cervical cancer volumetric modulated arc therapy (VMAT). We aimed to develop a robust deep-learning (DL) algorithm for fast and accurate dose prediction of cervical cancer VMAT in multicenter datasets and then explore the feasibility of the DL algorithm to endometrial cancer VMAT with different prescriptions.

MATERIALS AND METHODS

We proposed the AtTranNet algorithm for three-dimensional dose prediction. A total of 367 cervical patients were enrolled in this study. Three hundred twenty-two cervical patients from 3 centers were randomly divided into 70%, 10%, and 20% as training, validation, and testing sets, respectively. Forty-five cervical patients from another center were selected for external testing. Moreover, 70 patients of endometrial cancer with different prescriptions were further selected to test the model. Prediction precision was evaluated by dosimetric difference, dose map, and dose-volume histogram metrics.

RESULTS

The prediction results were all clinically acceptable. The mean absolute error within the body in internal testing was 0.66 ± 0.63%. The maximum |δD| for planning target volume was observed in D98, which is 1.24 ± 2.73 Gy. The maximum |δD| for organs at risk was observed in Dmean of bladder, which is 4.79 ± 3.14 Gy. The maximum |δV| were observed in V40 of pelvic bones, which is 4.77 ± 4.48%.

CONCLUSION

AtTranNet showed the feasibility and reasonable accuracy in the dose prediction for cervical cancer in multiple centers. The model can also be generalized for endometrial cancer with different prescriptions without any transfer learning.

A novel strategy for predicting the efficacy of temozolomide treatment for metastatic pheochromocytomas/paragangliomas

BACKGROUND

There are few studies on the efficacy of temozolomide (TMZ) in the treatment of Metastatic pheochromocytoma / paraganglioma (MPP) patients. And it remains unclear which MPP patients may benefit from TMZ treatment.

METHODS

This was a prospective study. MPP patients were enrolled. Patients were treated with TMZ until disease progression or intolerable toxicities. The primary endpoints were disease control rate (DCR) and objective response rate (ORR). Secondary endpoints included biochemical response rate progression-free survival (PFS) and safety. We compared the difference between effective and ineffective groups, to explore which patients are more suitable for TMZ treatment.

RESULTS

62 patients with MPP were enrolled and tumor response were evaluated in 54 patients. The DCR was 83% (35/42), and the ORR was 24% (10/41) among the progressive patients. PFS was 25.2 ± 3.1 months. The most common adverse event was nausea (41/55). We found that 92.9% (13/14) of patients with MGMT methylation greater than 7% respond to treatment. For the patients with MGMT methylation less than 7%, Ki-67 index could be used to guide the use of TMZ in these patients. Among the patients with Ki-67 index less than 5%, 66% (8/12) patients showed respond to treatment, and only 33% (4/12) patients with Ki-67 index more than 5% showed respond to TMZ.

CONCLUSIONS

This study indicated that TMZ is a potential choice for the treatment of MPP with the high ability on disease control and well tolerability. We recommended to MGMT methylation analysis test and Ki-67 index to guide TMZ application.

Novel alternative tools for metastatic pheochromocytomas/paragangliomas prediction

OBJECTIVE

The existing prediction models for metastasis in pheochromocytomas/paragangliomas (PPGLs) showed high heterogeneity in different centers. Therefore, this study aimed to establish new prediction models integrating multiple variables based on different algorithms.

DESIGN AND METHODS

Data of patients with PPGLs undergoing surgical resection at the Peking Union Medical College Hospital from 2007 to 2022 were collected retrospectively. Patients were randomly divided into the training and testing sets in a ratio of 7:3. Subsequently, decision trees, random forest, and logistic models were constructed for metastasis prediction with the training set and Cox models for metastasis-free survival (MFS) prediction with the total population. Additionally, Ki-67 index and tumor size were transformed into categorical variables for adjusting models. The testing set was used to assess the discrimination and calibration of models and the optimal models were visualized as nomograms. Clinical characteristics and MFS were compared between patients with and without risk factors.

RESULTS

A total of 198 patients with 59 cases of metastasis were included and classified into the training set (n = 138) and testing set (n = 60). Among all models, the logistic regression model showed the best discrimination for metastasis prediction with an AUC of 0.891 (95% CI, 0.793-0.990), integrating SDHB germline mutations [OR: 96.72 (95% CI, 16.61-940.79)], S-100 (-) [OR: 11.22 (95% CI, 3.04-58.51)], ATRX (-) [OR: 8.42 (95% CI, 2.73-29.24)] and Ki-67 ≥ 3% [OR: 7.98 (95% CI, 2.27-32.24)] evaluated through immunohistochemistry (IHC), and tumor size ≥ 5 cm [OR: 4.59 (95% CI, 1.34-19.13)]. The multivariate Cox model including the above risk factors also showed a high C-index of 0.860 (95% CI, 0.810-0.911) in predicting MFS after surgery. Furthermore, patients with the above risk factors showed a significantly poorer MFS (P ≤ 0.001).

CONCLUSIONS

Models established in this study provided alternative and reliable tools for clinicians to predict PPGLs patients' metastasis and MFS. More importantly, this study revealed for the first time that IHC of ATRX could act as an independent predictor of metastasis in PPGLs.

Targeted long-read sequencing for comprehensive detection of CYP21A2 mutations in patients with 21-hydroxylase deficiency

BACKGROUND

21-Hydroxylase deficiency (21-OHD) is caused by pathogenic CYP21A2 variations. CYP21A2 is arranged in tandem with its highly homologous pseudogene CYP21A1P; therefore, it is prone to mismatch and rearrangement, producing different types of complex variations. There were few reports on using only one method to detect different CYP21A2 variants simultaneously.

AIMS

Targeted long-read sequencing method was used to detect all types of CYP21A2 variants in a series of patients with 21-OHD.

METHODS

A total of 59 patients with 21-OHD were enrolled from Peking Union Medical College Hospital. Long-range locus-specific PCR and long-read sequencing (LRS) were performed to detect the pathogenic variants in CYP21A2.

RESULTS

Copy-number variants of CYP21A2 were found in 25.4% of patients, including 5.1% with 3 copies of CYP21A2, 16.9% with 1 copy of CYP21A2, and 3.4% with 0 copy of CYP21A2. The remaining 74.6% of patients had 2 copies of CYP21A2. Pathogenic variants were identified in all 121 alleles of 59 patients. Specifically, single-nucleotide variants and small insertions/deletions (< 50 bp) were detected in 79 alleles, of which conversed from CYP21A1P were detected in 63 alleles, and rare variants were found in the other 16 alleles. Large gene conversions (> 50 bp) from pseudogene were detected in 10 alleles, and different chimeric genes (CYP21A1P/CYP21A2 or TNXA/TNXB) formed by large deletions were detected in 32 alleles. Of all variants, p.I173N was the most common variant (19.0%).

CONCLUSIONS

Our study demonstrated that targeted long-read sequencing is a comprehensive method for detecting CYP21A2 variations, which is helpful for genetic diagnosis in 21-OHD patients.

[Clinicopathological and molecular characteristics of esophageal carcinoma with ductal differentiation: analysis of 17 cases]

Objective: To investigate the clinicopathological features and molecular genetic characteristics of esophageal carcinoma with ductal differentiation, and to summarize the experiences in its diagnosis and treatment. Methods: A total of 17 cases of esophageal carcinoma with ductal differentiation diagnosed in Ningbo Clinical Pathological Diagnosis Center, Ningbo, China from June 2011 to December 2022 were collected. The clinical information and pathological diagnosis was reviewed. The tumor histological features and immunohistochemical results were analyzed. The next-generation sequencing was performed to detect and analyze the gene mutations in tumor samples. Results: The 17 patients included in this study were 54-77 years old, with a median age of 66 years. There were 16 males and 1 female. Among them, 9 cases were mainly carcinoma with ductal differentiation. The squamous epithelium on the tumor's surface was accompanied by high-grade intraepithelial neoplasia. The tumor and atypical squamous epithelium were transitional, and the focus was accompanied by various proportions of squamous cell carcinoma component (less than 10%). The other 8 cases were mostly squamous cell carcinoma, basaloid squamous cell carcinoma or sarcomatoid carcinoma with various degrees of tumor specific differentiation and focal area of carcinoma with ductal differentiation (less than 10%). The tumor cells in the area with ductal differentiation were mainly arranged in small tubes, while the tubes showed a double-layer structure, including the inner cells and outer cells of the lumen. Immunohistochemical results showed that the outer cells of the tumorous tubules expressed p63, p40, CK5/6 and CK34βE12, while the inner cells expressed CK7. Compared with esophageal squamous cell carcinoma reported in the literature, the frequency of gene mutations such as MYC (P=0.002), TP63 (P=0.002), CDKN1C (P=0.002) and NFE2L2 (P=0.045) was significantly lower in this group of cases. At the signaling pathway level, the mutation frequency of NOTCH signaling pathway (P=0.041) was significantly higher, while the mutation frequencies of NRF2 pathway (P=0.013) and PI3K pathway (P=0.009) were significantly lower than that of esophageal squamous cell carcinoma. Conclusion: Esophageal carcinoma with ductal differentiation is a type of esophageal carcinoma with unique morphology, and its molecular changes are also significantly different from those of conventional esophageal squamous cell carcinoma.

Demonstrating Agreement between Radio and Fluorescence Measurements of the Depth of Maximum of Extensive Air Showers at the Pierre Auger Observatory

We show, for the first time, radio measurements of the depth of shower maximum (X_{max}) of air showers induced by cosmic rays that are compared to measurements of the established fluorescence method at the same location. Using measurements at the Pierre Auger Observatory we show full compatibility between our radio and the previously published fluorescence dataset, and between a subset of air showers observed simultaneously with both radio and fluorescence techniques, a measurement setup unique to the Pierre Auger Observatory. Furthermore, we show radio X_{max} resolution as a function of energy and demonstrate the ability to make competitive high-resolution X_{max} measurements with even a sparse radio array. With this, we show that the radio technique is capable of cosmic-ray mass composition studies, both at Auger and at other experiments.

Activating two-dimensional semiconductors for photocatalysis: a cross-dimensional strategy

The emerging two-dimensional (2D) semiconductors substantially extend materials bases for versatile applications such as semiconductor photocatalysis demanding semiconductive matrices and large surface areas. The dimensionality, while endowing 2D semiconductors the unique properties to host photocatalytic functionality of pollutant removal and hydrogen evolution, hurdles the activation paths to form heterogenous photocatalysts where the photochemical processes are normally superior over these on the mono-compositional counterparts. In this perspective, we present a cross-dimensional strategy to employ thenD (n= 0-2) clusters or nanomaterials as activation partners to boost the photocatalytic activities of the 2D semiconductors. The formation principles of heterogenous photocatalysts are illustrated specifically for the 2D matrices, followed by selection criteria of them among the vast 2D database. The computer investigations are illustrated in the density functional theory route and machine learning benefitted from the vast samples in the 2D library. Synthetic realizations and characterizations of the 2D heterogenous systems are introduced with an emphasis on chemical methods and advanced techniques to understand materials and mechanistic studies. The perspective outlooks cross-dimensional activation strategies of the 2D materials for other applications such as CO2removal, and materials matrices in other dimensions which may inspire incoming research within these fields.

[Progress in pathophysiology research and update on diagnosis and treatment strategies for fecal incontinence]

The traditional barrier theory believes that fecal incontinence is related to an imbalance of the recto-anal barrier and the characteristics of stool. However, in clinical practice this theory proves unable to explain all types of fecal incontinence. In recent years, research on the mechanisms related to fecal incontinence has shifted to a new integrative concept with the rectum and anus as functional units, and the central-peripheral nervous system and internal and external anal sphincters as a control loop. The diagnosis and treatment strategy of fecal incontinence, which is replaced by sacral neuromodulation, is undergoing a quiet change. With the progressively aging population in China, the need to improve the diagnosis and treatment of fecal incontinence has become increasingly urgent. This article explores the trends in diagnosis and treatment and mechanism research from the perspective of recent advances in pathophysiological research and updated diagnosis and treatment methods for fecal incontinence.

[A follow-up study on the pain changes trend and effects in patients diagnosed with herpes zoster in Beijing City]

Objective: To understand the changes in pain and its effects in patients with the diagnosis of herpes zoster. Methods: A total of 3 487 patients diagnosed with herpes zoster (HZ) for the first time at the outpatient department of Miyun District Hospital from January 1, 2017, to December 31, 2019, were included in the study. The information of patients was registered and issued with a record card. Patients were required to record the time of pain and rash by themselves. Telephone follow-up was conducted at 21, 90, 180 and 365 days after the onset of rashes, including hospitalization, location of rash and pain, and the time of start and end. The impact of pain on life was evaluated by the Zoster Brief Pain Inventory (ZBPI). Results: The age of 2 999 HZ patients included in the analysis were (53±16) years old, including 1 377 (45.91%) males and 1 903 (63.45%) patients aged 50 years and older. After 21 days of rash, mild, moderate and severe pain accounted for 20.87% (626 cases), 37.98% (1 139 cases) and 33.81% (1 014 cases), respectively. Only 5.07% (152 cases) had no pain or discomfort, and 2.27% (68 cases) had no pain but discomfort. Most of the pain sites were consistent with the rash sites. The chest and back and waist and abdomen were the most common, accounting for 35.58% (1 067 cases) and 29.18% (875 cases), respectively, followed by the limbs and face and neck, accounting for 16.74% (502 cases) and 16.40% (492 cases), respectively. The M (Q1, Q3) of pain days in the HZ patients was 14 (8, 20) days, and the incidence of post-herpetic neuralgia (PHN) was 6.63% (171/2 580) (excluding 419 patients who refused to visit or lost to visit on 90 days after the onset of rash). The pain score of HZ patients within 21 days after the rash was (5.19±2.73) points, and the pain score of PHN patients was (7.61±2.13) points, which was significantly higher than that of non-PHN patients [(5.04±2.69) points] (P<0.001). Daily activities, emotions, walking ability, work, social interaction, sleep and recreation were affected for 21 days after the rash in HZ patients, ranging from 60.79% to 83.83%, with sleep being the most affected (83.83%). The impact scores of pain and life dimensions in PHN patients ranged from 4.59 to 7.61 points on the ZBPI scale, which were higher than those in non-PHN patients (2.49-5.04) (t values ranged from 8.86 to 11.67, all P values <0.001). Conclusion: The proportion of pain in HZ patients after the diagnosis is high, and the pain is more obvious in patients with PHN and HZ patients aged 50 and older, which has a greater impact on their daily lives.

Polyglucosan body density in the aged mouse hippocampus is controlled by a novel modifier locus on chromosome 1

Aging can be associated with the accumulation of hypobranched glycogen molecules (polyglucosan bodies, PGBs), particularly in astrocytes of the hippocampus. While PGBs have a detrimental effect on cognition in diseases such as adult polyglucosan body disease and Lafora disease, the underlying mechanism and clinical relevance of age-related PGB accumulation remains unknown. Here, we have investigated the genetic basis and functional impact of age-related PGB accumulation in 32 fully sequenced BXD-type strains of mice which exhibit a 400-fold variation in PGB burden in 16-18 month old females. We mapped a major locus controlling PGB density in the hippocampus to chromosome 1 at 72-75 Mb (linkage of 4.9 -logP), which we defined as the Pgb1 locus. To identify potentially causal gene variants within Pgb1, we generated extensive hippocampal transcriptome datasets and identified two strong candidate genes for which mRNA correlates with PGB density-Smarcal1 and Usp37. In addition, both Smarcal1 and Usp37 contain non-synonymous allele variations likely to impact protein function. A phenome-wide association analysis highlighted a trans-regulatory effect of the Pgb1 locus on expression of Hp1bp3, a gene known to play a role in age-related changes in learning and memory. To investigate the potential impact of PGBs on cognition, we performed conditioned fear memory testing on strains displaying varying degrees of PGB burden, and a phenome-wide association scan of ~12,000 traits. Importantly, we did not find any evidence suggesting a negative impact of PGB burden on cognitive capacity. Taken together, we have identified a major modifier locus controlling PGB burden in the hippocampus and shed light on the genetic architecture and clinical relevance of this strikingly heterogeneous hippocampal phenotype.

[Mediating effects of obesity and metabolic factors in hyperuricemia and prehypertension]

Objective: To assess the mediating effects of obesity and metabolic factors in the relationship between hyperuricemia (HUA) and prehypertension. Methods: A total of 9 399 individuals were selected using a multistage stratified whole-group random sampling method from 90 villages (neighborhood committees) in 30 towns (streets) of 5 districts (counties) in Fuzhou. A total of 4 754 study subjects were included. A linear regression model was used to analyze the association of HUA with obesity and metabolic factors. Single-factor and multi-factor logistic regression models were used to analyze the association of HUA, obesity, and metabolic factors with prehypertension. Mediating effects models were used to analyze the mediating effects of obesity and metabolic factors on the association between HUA and prehypertension. Results: After adjusting for confounders, the association between HUA and cholesterol, triglycerides, HDL-C, LDL-C, BMI, waist circumference, creatinine, and urea nitrogen were significantly correlated (P<0.001). HUA, waist circumference, BMI, and triglycerides were significantly associated with prehypertension (P<0.001). Waist circumference, BMI, and triglycerides mediated the relationship between HUA and prehypertension, with OR (95%CI) of 1.018 (1.007-1.027), 1.010 (1.002-1.018), and 1.010 (1.003-1.017) (P<0.001), with mediating proportions of 7.76%, 4.31%, and 4.31% respectively. No mediating effect of cholesterol, HDL-C, LDL-C, creatinine, and urea nitrogen was found on the relationship (P>0.05). Conclusions: Waist circumference, BMI, and triglycerides all had mediating effects in the association between HUA and prehypertension. For the general population, weight control, waist circumference, and a high-fat diet should be used to reduce the occurrence of prehypertension.

Deep Learning-Based Multi-Modality Segmentation of Primary Gross Tumor Volume in CT and MRI for Nasopharyngeal Carcinoma

PURPOSE/OBJECTIVE(S)

The delineation of primary gross tumor volume (GTV) of nasopharyngeal carcinoma (NPC) is an essential step for radiotherapy planning. In clinical practice, radiation oncologists manually delineate the GTV in planning CT with the help of diagnostic MRI. This is because NPC tumors are closely adjacent to many important anatomic structures, and CT and MRI provide complementary strength to accurately determine the tumor extension boundary. Manual delineation is time-consuming with the potential registration errors between MRI and CT decreasing the delineation accuracy. In this study, we propose a fully automated GTV segmentation method based on CT and MRI by first aligning MRI to CT, and then, segmenting the GTV using a multi-modality deep learning model.

MATERIALS/METHODS

We collected 104 nasopharyngeal carcinoma patients with both planning CT and diagnostic MRI scans (T1 & T2 phases). An experienced radiation oncologists manually delineated the GTV, which was further examined by another senior radiation oncologist. Then, a coarse to fine cross-modality registration from MRI to CT was conducted as follows: (1) A rigid transformation was performed on MRI to roughly align MRI to CT with similar anatomic position. (2) Then, the region of interest (RoI) on both CT and rigid-transformed MRI were cropped. (3) A leading cross-modality deformable registration algorithm, named DEEDS, was applied on the cropped MRI and CT RoIs to find an accurate local alignment. Next, using CT and registered MRI as the combined input, a multi-modality deep segmentation network based on nnUNet was trained to generate the GTV prediction. 20% patients were randomly selected as the unseen testing set to quantitatively evaluate the performance.

RESULTS

The quantitative NPC GTV segmentation performance is summarized in Table 1. The deep segmentation model using CT alone achieved reasonable high performance with 76.6% Dice score and 1.34mm average surface distance (ASD). When both CT and registered MRI were used, the segmentation model further improved the performance by 0.9% Dice score increase and 11% relative ASD error reduction, demonstrating the complementary strength of CT and MRI in determining NPC GTV. Notably, the achieved 77.5% Dice score and 1.19mm ASD by the multimodality model is among the top performing results reported in recent automatic NPC GTV segmentation using either CT or MRI modality.

CONCLUSION

We developed a fully automated multi-modal deep-learning model for NPC GTV segmentation. The developed model can segment the NPC GTV in high accuracy. With further optimization and validation, this automated model has potential to standardize the NPC GTV segmentation and significantly decrease the workload of radiation oncologists in clinical practice.

A Novel Strategy to Enhance Radiotherapy Efficacy: Results from the Prospective Phase I Clinical Trial of MR-Guided Focused Ultrasound-Stimulated Microbubbles (MRgFUS+MB) Treatment for Breast Cancer

PURPOSE/OBJECTIVE(S)

Preclinical in vitro and in vivo studies have demonstrated that tumor cell death can be enhanced 10- to 40-fold when radiotherapy (RT) is combined with focused-ultrasound (FUS)-stimulated microbubbles (MB) treatment. MBs are gas microspheres used as intravascular contrast agents. The acoustic exposure of MBs within the target volume causes bubbles cavitation that induces perturbation of tumor vasculature. This activates apoptotic pathways responsible for the ablative effect of stereotactic body radiotherapy, which would otherwise require high-dose radiotherapy (>8-10 Gy/fraction) to be activated. Subsequent irradiation of an MB-sensitized tumor causes increased anoxic tumor killing, which occurs in addition to canonical RT-induced DNA damage. Given the compelling results of preclinical data, we conducted a phase I clinical trial of magnetic resonance (MR)-guided FUS-stimulated MBs (MRgFUS+MB) treatment for breast cancer patients (pts). We report the safety and efficacy results of this new radio enhancement treatment.

MATERIALS/METHODS

This is a single-center, single-arm, investigator-initiated phase 1 clinical trial (NCT04431674). We included pts with stage I-IV breast cancer with tumor in situ for whom breast or chest wall RT was deemed adequate by a multidisciplinary team. Pts were excluded if they had contraindications for contrast-enhanced MR or MB administration. Pts underwent 2-3 MRgFUS+MB treatments throughout the RT course. We used an MR-coupled FUS-device operating at 500 KHz and 540 kPa peak negative pressure to deliver the treatment. The FUS sonicated intravenously administrated MB within the MR-guided target volume. Pts were monitored for 30-min post-procedure and subsequently treated with RT. The primary outcome was acute toxicity per Common Terminology for Adverse Events V5.0. Secondary outcomes were radiological response at 3 months and local control (LC) at 1 year. Kaplan-Meier method was used to estimate local control. All pts signed a written consent form before study participation.

RESULTS

We enrolled 18 females with 20 primary breast cancer treated with MRgFUS+MB therapy. The median age was 60 years (range, 44-90). The molecular subtypes consisted of basal-like (n = 3/20), luminal (n = 11/20), and HER2-enriched (n = 6/20). The prescribed dose was 20 Gy/5 fractions (n = 8/20), 30-35 Gy/5 fractions (n = 7/20), 30-40 Gy/10 fractions (n = 3/20), and 66 Gy/33 fractions (n = 2/20). The median follow-up was 9 months (range, 0.3-29). All pts completed the planned MRgFUS+MB treatments. The only MRgFUS+MB treatment-related toxicity consisted of Grade 1 allergic reaction (mild cough) 30 minutes after the last MB injection. All worst acute toxicities were radiation dermatitis (Grade 3 = 2/20, Grade 2 = 2/20, Grade 1 = 14/20). At 3 months, 75% had partial (n = 6/20) or complete (n = 9/20) response, with a single progression. The LC rate at 1 year was 86%.

CONCLUSION

MRgFUS+MB was a safe and efficient treatment that provided durable responses.

Deep Learning for Automatic Prediction of Lymph Node Station Metastasis in Esophageal Cancer Patients from Contrast-Enhanced CT

PURPOSE/OBJECTIVE(S)

The diagnosis of lymph node (LN) metastasis in computed tomography (CT) is an essential yet challenging task in esophageal cancer staging and treatment planning. Although criteria (e.g., RECIST, morphological/texture features) are proposed to predict LN metastasis, the diagnostic accuracy remains low with sensitivity <50% and specificity <75%, as reported in previous studies. Deep learning (DL) has the potential to address this issue by learning from large-scale labeled data. However, due to the practical surgery procedure in lymph node dissection, it is difficult to pair the metastasis of individual LN reported in the pathology report to the LN instance found in the CT image. Hence, in this study, we first use pathology reports to determine the LNS metastasis, then develop a multiple instance deep learning (MIDL) model to predict lymph node station (LNS) metastasis.

MATERIALS/METHODS

We collected 1200 esophageal cancer patients with preoperative contrast-enhanced CT before surgery. A recently developed automatic mediastinal LNS segmentation model was first applied to segment LNS of 1 to 8 based on the IASLC protocol. For each LNS, the local CT region of interest (ROI) was cropped to generate a station-wise CT patch, where the LNS was labeled as metastatic if at least one metastatic LN was indicated in the pathology report. Using the station-wise CT patch and LNS label, we train a 3D MIDL model, MobileNetV3, to predict LNS metastasis. To better provide the LN position priors in MIDL, LN instances (with a short axis >4mm) were also segmented using an automatic LN detection algorithm and were added to the MIDL model as an auxiliary input. Five-fold cross-validation was conducted to evaluate the MIDL performance.

RESULTS

The MIDL model's performance is summarized in Table 1. The MIDL model incorporating an additional LN instance mask demonstrated a superior overall AUC of 0.7539, surpassing the model without the LN mask input by 2.93%. The specificity was evaluated at a threshold resulting in a recall of 0.7, and the best model outperformed the CT input model in terms of specificity by 2.11%. This highlights the value of including the LN position prior to the MIDL model. Notably, when a threshold was set to result in a specificity of 75%, the best MIDL model demonstrated a significantly higher recall compared to the previously reported clinical diagnostic recall (39.7% vs. 63.21%).

CONCLUSION

We developed a MIDL classification model to predict LNS metastasis using CT scans of 1200 patients. Our findings suggest that the MIDL model can substantially improve LNS metastasis prediction and has the potential to play an essential role in cancer staging, treatment planning, and prognostic analysis.

Anatomy-Guided Deep Learning Model for Accurate and Robust Gross Tumor Volume Segmentation in Lung Cancer Radiation Therapy

PURPOSE/OBJECTIVE(S)

In lung cancer radiation therapy, clinicians must outline the gross tumor volume (GTV) precisely on the planning computed tomography (pCT) for accurate radiation dose delivery. However, due to the limited contrast between tumor and normal tissues in lung parenchyma, accurate delineation of tumor boundaries is difficult leading to large inter-observer variation. In this study, we develop an anatomy-guided lung GTV deep segmentation model using a training cohort of multi-center datasets. The quantitative segmentation performance is evaluated on an independent dataset, where the inter-observer delineation variation is also assessed.

MATERIALS/METHODS

We collected and curated four publicly available lung datasets with GTV annotations (Lung-PET-CT-Dx, LIDC-IDRI, NSCLC-Radiogenomics and RIDER-CT) for deep learning model development. A total of 871 CT scans of patients, who were diagnosed with T1-T4 NSCLC, were available for training after data curation. The GTV annotations of primary tumor were examined and edited by two experienced radiation oncologists following the RTOG 1106 protocol. An anatomy-guided deep learning model was proposed, which consisted two deep networks. The first deep network used CT scan as input and segmented 4 anatomic organs (airway, heart, pulmonary artery and pulmonary vein), while the second deep network took both CT scan and these pre-segmented 4 organs as input and segmented the lung GTV. With the help of anatomic priors from 4 pre-segmented organs, the second deep network could more easily locate the GTV. We used nnUNet as the deep segmentation network. For evaluation, we used NSCLC-Radiomics as the testing dataset, which contains 20 CT scans each annotated by 5 radiation oncologists. The auto-segmented GTV were compared against each of the manual GTV reference. Inter-observer variation was also assessed using the 5 manual GTV references.

RESULTS

The proposed anatomic-guided lung GTV segmentation model achieved a mean Dice score of 82.4% and 95% Hausdorff distance (HD95) of 6.9mm when averaged cross 20 patients and 5 GTV references (Table 1), which outperformed the basic deep GTV segmentation model by markedly reducing 19.4% HD95 error. The performance of proposed model was also comparable to the inter-observer variation (Dice score: 82.4% vs. 81.9%, HD95 6.9 vs. 6.4mm), indicating that our model had similar reproducibility as human observers.

CONCLUSION

We developed and tested an anatomy-guided deep learning model for segmenting GTV in NSCLC patients. The model achieves high quantitative segmentation performance, which is comparable to the human observer variation. It can be potentially used in radiotherapy practice to improve GTV delineation consistency and reduce workloads of radiation oncologists.

A naphthalimide-based fluorescent probe for rapid detection of nitrite and its application in food quality monitoring

Nitrite (NO₂^-) is a widely used food additive and long-term aging of cooked leftovers may also contribute to the formation of NO₂^-, excessive consumption of NO₂^- is harmful to human health. Developing an effective sensing strategy for on-site monitoring of NO₂^- has attracted considerable attention. Herein, a novel colorimetric and fluorometric probe ND-1 based on photoinduced electron transfer effect (PET) was designed for highly selective and sensitive detection of nitrite (NO₂^-) in foods. The probe ND-1 was strategically constructed by employing naphthalimide as the fluorophore and o-phenylendiamine as the specific recognition site for NO₂^-. The triazole derivative ND-1-NO₂^- could be produced exclusively by reacting with NO₂^-, leading to a visible colorimetric change from yellow to colorless accompanied by a significantly enhanced fluorescence intensity at 440 nm. The probe ND-1 exhibited promising sensing performances towards NO₂^- including high selectivity, rapid response time (within 7 min), low detection limit (47.15 nM) and wide quantitative detection range (0-35 μM). In addition, probe ND-1 was capable of quantitative detecting of NO₂^- in real food samples (including pickled vegetables and cured meat products) with satisfactory recovery rates (97.61%-103.08%). Moreover, the paper device loaded by probe ND-1 could be utilized for visual monitoring of NO₂^- levels variation of stir-fried greens. This study provided a feasible method for the accurate, traceable and rapid on-site monitoring NO₂^- in foods.

Phase I study of MSB2311, a novel pH-dependent anti-PD-L1 monoclonal antibody, treating patients with advanced solid tumors and lymphoma

MSB2311 is a novel pH-dependent humanized anti-programmed death-ligand 1 (PD-L1) monoclonal antibody. This phase I study primarily aimed to determine the maximum tolerated dose (MTD)/recommended phase 2 dose level (RP2D) of MSB2311 in patients with advanced solid tumors or lymphoma. MSB2311 was intravenously administered at 3, 10, and 20 mg/kg every 3 weeks (Q3W) and 10 mg/kg every 2 weeks (Q2W) using 3 + 3 design. During expansion phase, eligible patients with either PD-L1 overexpression, Epstein-Barr Virus positive, microsatellite instability high/mismatch repair deficient, or high tumor mutation burden tumors were treated at RP2D. A total of 37 Chinese patients were treated, including 31 with solid tumors and 6 lymphoma. No dose limiting toxicity was reported and MTD was not reached. The trial was expanded at 20 mg/kg Q3W or 10 mg/kg Q2W, both of which were determined as RP2D. Most common drug-related treatment-emergent adverse events were anemia (43.2%), aspartate aminotransferase increase (27.0%), proteinuria (21.6%), alanine aminotransferase increase and hypothyroidism (18.9% each), thyroid stimulating hormone increased and hyperglycemia (16.2% each). Out of 20 efficacy evaluable patients with biomarker positive solid tumors, 6 achieved confirmed partial response with the median duration of response of 11.0 months (95% CI 7.0-11.4) and 4 had stable disease, resulting an objective response rate of 30.0% (95% CI 11.9, 54.3) and disease control rate of 50.0% (95% CI 27.2, 72.8). One partial response was also observed among 6 patients with lymphoma. MSB2311 demonstrated a manageable safety profile and promising antitumor activity in patients with advanced solid tumors and lymphomas.

Observation of high-energy neutrinos from the Galactic plane

The origin of high-energy cosmic rays, atomic nuclei that continuously impact Earth's atmosphere, is unknown. Because of deflection by interstellar magnetic fields, cosmic rays produced within the Milky Way arrive at Earth from random directions. However, cosmic rays interact with matter near their sources and during propagation, which produces high-energy neutrinos. We searched for neutrino emission using machine learning techniques applied to 10 years of data from the IceCube Neutrino Observatory. By comparing diffuse emission models to a background-only hypothesis, we identified neutrino emission from the Galactic plane at the 4.5σ level of significance. The signal is consistent with diffuse emission of neutrinos from the Milky Way but could also arise from a population of unresolved point sources.

[Exploration of clinical pathway-oriented optimal management diagnosis and treatment model for rare diseases]

This study takes Cushing's syndrome, a rare disease, as a model, and adopts the path of "Plan, Do, Check, Action" (PDCA) to explore new methods to optimize the clinical path, can improve the quality and efficiency of diagnosis and treatment of rare diseases. After sorting out the problems existing in the previous diagnosis and treatment mode, our team optimizes the path in various ways and establishes a standard operation procedure (SOP) for the new path. In the evaluation of the optimized mode, 55 patients with Cushing's syndrome were admitted to the Department of Endocrinology, Peking Union Medical College Hospital, including 19 males and 36 females, aged (41.8±14.4) years (6-68 years). The pathway group (28 cases) and the control group (27 cases) were divided according to whether they were included in the new path management at the time of admission, and the effect of path optimization was assessed in terms of time, efficacy, safety and cost. The results showed that compared with the control group, the pathway group had a shorter time of hospitalization in the Department of Endocrinology and critical tests, such as blood cortisol rhythm, low-dose dexamethasone inhibition test, and bilateral inferior petrosal sinus sampling (all P<0.05). There was no significant differences in the decrease of total cortisol after operation, the incidence of postoperative complications, and hospitalization expenses (all P>0.05). The optimized path improves the medical efficiency while ensuring medical quality, safety and no increase in cost. This study proposes PDCA path optimization for complex diseases and establishes SOP process, which provides experience in management optimization for the patient-centered and clinical path-oriented diagnosis and treatment mode of rare diseases.

[The past, present and future of sleep medicine in China]

Sleep is a highly conserved phenomenon in endotherms, and has a universal physiological function across all species. In mammals, sleep can be divided into two stages: rapid eye movement (REM) sleep and non-REM (NREM) sleep, which alternate in a cyclic manner. Humans spend about one-third of their lives asleep. Sufficient sleep is necessary for humans to sustain everyday functioning. Sleep plays an important role in regulating energy metabolism, immune defense, endocrine function, and the consolidation of memory process. With the development of social economy and the change of life style, sleep duration of the residents has gradually decreased and the incidence of sleep disturbances has increased. Sleep disturbances can lead to severe mental disorders, such as depression, anxiety disorders, dementia, and other mental diseases, and may increase the risk of physical diseases, such as chronic inflammation, heart disease, diabetes, hypertension, atherosclerosis and others. Maintaining good sleep is of great significance for developing social productive forces, promoting sustainable development of economic society, and is a necessary condition for carrying out the "Healthy China Strategy". The sleep research in China started in 1950s. After decades of development, researchers have made great progress in the molecular mechanisms of sleep and wakefulness, the pathogenesis of sleep disorders and the development of new therapies. With the advancement of science and technology and the public's attention to sleep, the level of clinical diagnosis and therapy of sleep disorders in China is gradually brought in line with international standards. The publication of diagnosis and treatment guidelines in the field of sleep medicine will promote the standardization of the construction. In the future, it is still necessary to promote the development of sleep medicine in the following aspects: Strengthening the professional training and discipline construction, improving the cooperation of sleep research, promoting the intelligent diagnosis and treatment of sleep disorders, and developing the new intervention strategies. Therefore, this review will comprehensively summarize the origin, current situation, and future expectations of sleep medicine in China, including discipline construction of sleep medicine, the number of sleep project grants, research findings, the status and progress of diagnosis and treatment of sleep disorders, and the development direction of sleep medicine.

[A region-level contrastive learning-based deep model for glomerular ultrastructure segmentation on electron microscope images]

OBJECTIVE

We propose a novel region- level self-supervised contrastive learning method USRegCon (ultrastructural region contrast) based on the semantic similarity of ultrastructures to improve the performance of the model for glomerular ultrastructure segmentation on electron microscope images.

METHODS

USRegCon used a large amount of unlabeled data for pre- training of the model in 3 steps: (1) The model encoded and decoded the ultrastructural information in the image and adaptively divided the image into multiple regions based on the semantic similarity of the ultrastructures; (2) Based on the divided regions, the first-order grayscale region representations and deep semantic region representations of each region were extracted by region pooling operation; (3) For the first-order grayscale region representations, a grayscale loss function was proposed to minimize the grayscale difference within regions and maximize the difference between regions. For deep semantic region representations, a semantic loss function was introduced to maximize the similarity of positive region pairs and the difference of negative region pairs in the representation space. These two loss functions were jointly used for pre-training of the model.

RESULTS

In the segmentation task for 3 ultrastructures of the glomerular filtration barrier based on the private dataset GlomEM, USRegCon achieved promising segmentation results for basement membrane, endothelial cells, and podocytes, with Dice coefficients of (85.69 ± 0.13)%, (74.59 ± 0.13)%, and (78.57 ± 0.16)%, respectively, demonstrating a good performance of the model superior to many existing image-level, pixel-level, and region-level self-supervised contrastive learning methods and close to the fully- supervised pre-training method based on the large- scale labeled dataset ImageNet.

CONCLUSION

USRegCon facilitates the model to learn beneficial region representations from large amounts of unlabeled data to overcome the scarcity of labeled data and improves the deep model performance for glomerular ultrastructure recognition and boundary segmentation.

[A descriptive analysis of hyperlipidemia in adult twins in China]

Objective: To describe the distribution characteristics of hyperlipidemia in adult twins in the Chinese National Twin Registry (CNTR) and explore the effect of genetic and environmental factors on hyperlipidemia. Methods: Twins recruited from the CNTR in 11 project areas across China were included in the study. A total of 69 130 (34 565 pairs) of adult twins with complete information on hyperlipidemia were selected for analysis. The random effect model was used to characterize the population and regional distribution of hyperlipidemia among twins. The concordance rates of hyperlipidemia were calculated in monozygotic twins (MZ) and dizygotic twins (DZ), respectively, to estimate the heritability. Results: The age of all participants was (34.2±12.4) years. This study's prevalence of hyperlipidemia was 1.3% (895/69 130). Twin pairs who were men, older, living in urban areas, married,had junior college degree or above, overweight, obese, insufficient physical activity, current smokers, ex-smokers, current drinkers, and ex-drinkers had a higher prevalence of hyperlipidemia (P<0.05). In within-pair analysis, the concordance rate of hyperlipidemia was 29.1% (118/405) in MZ and 18.1% (57/315) in DZ, and the difference was statistically significant (P<0.05). Stratified by gender, age, and region, the concordance rate of hyperlipidemia in MZ was still higher than that in DZ. Further, in within-same-sex twin pair analyses, the heritability of hyperlipidemia was 13.04% (95%CI: 2.61%-23.47%) in the northern group and 18.59% (95%CI: 4.43%-32.74%) in the female group, respectively. Conclusions: Adult twins were included in this study and were found to have a lower prevalence of hyperlipidemia than in the general population study, with population and regional differences. Genetic factors influence hyperlipidemia, but the genetic effect may vary with gender and area.

[A descriptive analysis on hypertension in adult twins in China]

Objective: To describe the distribution characteristics of hypertension among adult twins in the Chinese National Twin Registry (CNTR) and to provide clues for exploring the role of genetic and environmental factors on hypertension. Methods: A total of 69 220 (34 610 pairs) of twins aged 18 and above with hypertension information were selected from CNTR registered from 2010 to 2018. Random effect models were used to describe the population and regional distribution of hypertension in twins. To estimate the heritability, the concordance rates of hypertension were calculated and compared between monozygotic twins (MZ) and dizygotic twins (DZ). Results: The age of all participants was (34.1±12.4) years. The overall self-reported prevalence of hypertension was 3.8%(2 610/69 220). Twin pairs who were older, living in urban areas, married, overweight or obese, current smokers or ex-smokers, and current drinkers or abstainers had a higher self-reported prevalence of hypertension (P<0.05). Analysis within the same-sex twin pairs found that the concordance rate of hypertension was 43.2% in MZ and 27.0% in DZ, and the difference was statistically significant (P<0.001). The heritability of hypertension was 22.1% (95%CI: 16.3%- 28.0%). Stratified by gender, age, and region, the concordance rate of hypertension in MZ was still higher than that in DZ. The heritability of hypertension was higher in female participants. Conclusions: There were differences in the distribution of hypertension among twins with different demographic and regional characteristics. It is indicated that genetic factors play a crucial role in hypertension in different genders, ages, and regions, while the magnitude of genetic effects may vary.

Radiomics model and clinical scale for the preoperative diagnosis of silent corticotroph adenomas

OBJECTIVE

Silent corticotroph adenomas (SCAs) are a subtype of nonfunctioning pituitary adenomas that exhibit more aggressive behavior. However, rapid and accurate preoperative diagnostic methods are currently lacking.

DESIGN

The purpose of this study was to examine the differences between SCA and non-SCA features and to establish radiomics models and a clinical scale for rapid and accurate prediction.

METHODS

A total of 260 patients (72 SCAs vs. 188 NSCAs) with nonfunctioning adenomas from Peking Union Medical College Hospital were enrolled in the study as the internal dataset. Thirty-five patients (6 SCAs vs. 29 NSCAs) from Fuzhou General Hospital were enrolled as the external dataset. Radiomics models and an SCA scale to preoperatively diagnose SCAs were established based on MR images and clinical features.

RESULTS

There were more female patients (internal dataset: p < 0.001; external dataset: p = 0.028) and more multiple microcystic changes (internal dataset: p < 0.001; external dataset: p = 0.012) in the SCA group. MRI showed more invasiveness (higher Knosp grades, p ≤ 0.001). The radiomics model achieved AUCs of 0.931 and 0.937 in the internal and external datasets, respectively. The clinical scale achieved an AUC of 0.877 and a sensitivity of 0.952 in the internal dataset and an AUC of 0.899 and a sensitivity of 1.0 in the external dataset.

CONCLUSIONS

Based on clinical information and imaging characteristics, the constructed radiomics model achieved high preoperative diagnostic ability. The SCA scale achieved the purpose of rapidity and practicality while ensuring sensitivity, which is conducive to simplifying clinical work.