Multi-center application of a convolutional neural network for preoperative detection of cavernous sinus invasion in pituitary adenomas.

OBJECTIVE: Cavernous sinus invasion (CSI) plays a pivotal role in determining management in pituitary adenomas. The study aimed to develop a Convolutional Neural Network (CNN) model to diagnose CSI in multiple centers. METHODS: A total of 729 cases were retrospectively obtained in five medical centers with (n = 543) or without CSI (n = 186) from January 2011 to December 2021. The CNN model was trained using T1-enhanced MRI from two pituitary centers of excellence (n = 647). The other three municipal centers (n = 82) as the external testing set were imported to evaluate the model performance. The area-under-the-receiver-operating-characteristic-curve values (AUC-ROC) analyses were employed to evaluate predicted performance. Gradient-weighted class activation mapping (Grad-CAM) was used to determine models' regions of interest. RESULTS: The CNN model achieved high diagnostic accuracy (0.89) in identifying CSI in the external testing set, with an AUC-ROC value of 0.92 (95% CI, 0.88-0.97), better than CSI clinical predictor of diameter (AUC-ROC: 0.75), length (AUC-ROC: 0.80), and the three kinds of dichotomizations of the Knosp grading system (AUC-ROC: 0.70-0.82). In cases with Knosp grade 3A (n = 24, CSI rate, 0.35), the accuracy the model accounted for 0.78, with sensitivity and specificity values of 0.72 and 0.78, respectively. According to the Grad-CAM results, the views of the model were confirmed around the sellar region with CSI. CONCLUSIONS: The deep learning model is capable of accurately identifying CSI and satisfactorily able to localize CSI in multicenters.

Perioperative fluctuation and overall evaluation of adenohypophyseal hormone secretion in patients with nonfunctioning pituitary adenoma.

OBJECTIVE: Perioperative adenohypophyseal hormone assessment can improve therapeutic strategies and be used to evaluate the prognosis of pituitary adenomas. An individual hormone level does not entirely reflect the pituitary gland. Thus, this study aimed to analyze perioperative hormonal changes and propose a normalized method to facilitate overall assessment of the adenohypophysis. METHODS: The authors retrospectively analyzed 89 male patients with nonfunctioning pituitary adenoma (NFPA) who underwent transsphenoidal surgery. Preoperative clinical data, imaging data, and perioperative hormone levels of the anterior pituitary gland were evaluated. Hormone values were rescaled using minimum-maximum normalization. The sum of the normalized hormone levels was defined as the total hormonal rate (THR). RESULTS: Preoperative findings indicated correlations among different adenohypophyseal hormones. Luteinizing hormone (p = 0.62) and adrenocorticotropic hormone (p = 0.89) showed no significant changes after surgery, but growth hormone levels increased (p < 0.001). On the contrary, the levels of thyroid-stimulating hormone (p < 0.001), follicle-stimulating hormone (p = 0.02), and prolactin (p < 0.001) decreased. THR indicated a significant postoperative reduction in adenohypophyseal function (p = 0.04). Patients with postoperative hypopituitarism had significantly lower THR than those without (p = 0.003), with an area under the curve of 0.66. For NFPAs that presented with normal preoperative hormone levels, THR was a good clinical predictor of immediate postoperative hypopituitarism, with an area under the curve of 0.74. CONCLUSIONS: The normalized synthesis index of hormones is a novel and clinically valuable method used to reflect adenohypophyseal secretion. Compared with individual hormones, these results indicated that THR can facilitate the analysis of general hormone levels despite various fluctuations in adenohypophyseal hormones. THR may also contribute to the effective prediction of short-term surgery-induced hypopituitarism.

Predictive factors for delayed hyponatremia after transsphenoidal surgery in patients with Rathke's cleft cysts.

Purpose: We aimed to assess factors influencing the occurrence of delayed hyponatremia after transsphenoidal surgery in patients with Rathke's cleft cysts (RCCs). Methods: We retrospectively collected the clinical data of patients who underwent transsphenoidal surgery for RCCs from January 2014 to January 2022. Univariate and multivariate analyses were used to determine the factors influencing the occurrence of postoperative delayed hyponatremia. Results: Of the 78 microscopic transsphenoidal surgery recipients with RCCs, 15 experienced postoperative delayed hyponatremia. There were 35 men and 43 women, and mean age was 43.75 ± 14.95 years. The clinical manifestations of RCCs were headache (62 cases, 79.5%), visual dysfunction (35 cases, 44.9%), endocrine dysfunction symptoms (12 cases, 15.4%). After transsphenoidal surgery, 93.5% (58/62) had improvements in headache, and 97.1% (34/35) had improved or resolved compressive visual symptoms. Delayed hyponatremia occurred on average on day 6.46 and lasted on average for 4.40 days. Logistic regression analysis showed that the independent influencing factor of delayed hyponatremia after transsphenoidal surgery in patients with RCCs was postoperative diaphragma sellae height. Conclusion: Postoperative diaphragma sellae height was identified as an independent influencing factor for delayed hyponatremia after transsphenoidal surgery in patients with RCCs.

A Convolutional Neural Network Model for Detecting Sellar Floor Destruction of Pituitary Adenoma on Magnetic Resonance Imaging Scans.

Objective: Convolutional neural network (CNN) is designed for image classification and recognition with a multi-layer neural network. This study aimed to accurately assess sellar floor invasion (SFI) of pituitary adenoma (PA) using CNN. Methods: A total of 1413 coronal and sagittal magnetic resonance images were collected from 695 patients with PAs. The enrolled images were divided into the invasive group (n = 530) and the non-invasive group (n = 883) according to the surgical observation of SFI. Before model training, 100 images were randomly selected for the external testing set. The remaining 1313 cases were randomly divided into the training and validation sets at a ratio of 80:20 for model training. Finally, the testing set was imported to evaluate the model performance. Results: A CNN model with a 10-layer structure (6-layer convolution and 4-layer fully connected neural network) was constructed. After 1000 epoch of training, the model achieved high accuracy in identifying SFI (97.0 and 94.6% in the training and testing sets, respectively). The testing set presented excellent performance, with a model prediction accuracy of 96%, a sensitivity of 0.964, a specificity of 0.958, and an area under the receptor operator curve (AUC-ROC) value of 0.98. Four images in the testing set were misdiagnosed. Three images were misread with SFI (one with conchal type sphenoid sinus), and one image with a relatively intact sellar floor was not identified with SFI. Conclusion: This study highlights the potential of the CNN model for the efficient assessment of PA invasion.

Metabolic Characteristics and M2 Macrophage Infiltrates in Invasive Nonfunctioning Pituitary Adenomas.

Objective: The aim of this study was to investigate the metabolic differences between invasive and non-invasive nonfunctioning pituitary adenomas (NFPAs), determine the expression of an M2 macrophage marker in NFPAs, and analyze the effects of metabolic changes in invasive NFPAs on M2 macrophage infiltrates. Methods: Tissue samples of NFPAs from patients who underwent transsphenoidal or craniotomy surgery from January 2021 to August 2021 were collected. NFPA tissues were analyzed based on a gas chromatography-mass spectrometry non-targeted metabolomics platform, and immunohistochemical staining for M2 macrophage marker CD206 was performed. Results: We evaluated 15 invasive and 21 non-invasive NFPAs. A total of 22 metabolites were identified through non-targeted metabolomics analysis. Among them, the expression of 1-octadecanol, inosine 5'-monophosphate, adenosine 5'-monophosphate, guanosine 5'-monophosphate, creatinine, desmosterol, taurine, hypotaurine, lactic acid, and succinic acid was upregulated in invasive NFPAs, while that of 1-oleoylglycerol, arachidonic acid, cis-11-eicosenoic acid, docosahexaenoic acid, glyceric acid, hypoxanthine, linoleic acid, lysine, oleic acid, uracil, valine, and xanthine was downregulated. Immunohistochemical analysis suggested that the number of CD206-positive cells was higher in invasive NFPAs than in non-invasive NFPAs. Conclusion: Invasive and non-invasive NFPAs showed distinct metabolite profiles. The levels of succinic acid and lactic acid were higher in invasive NFPAs, and the high expression of the M2 macrophage marker was verified in invasive NFPAs.

Microscopic Transsphenoidal Resection of Giant Pituitary Adenomas: Analysis of the Factors Limiting the Degree of Resection in 73 Cases.

Purpose: To analyze the risk factors affecting the gross-total resection of giant pituitary adenomas using a transsphenoidal approach under a microscope to provide a reference basis for formulating an appropriate surgical strategy. Methods: The clinical data of patients who underwent microscopic transsphenoidal resection of giant pituitary adenomas in a single center from January 2011 to December 2020 were retrospectively analyzed. Based on magnetic resonance imaging and surgical records, the predictive factors affecting the gross-total resection of giant pituitary adenomas under microscopy were determined through univariate and multivariate analyses. Results: A total of 73 patients with giant pituitary adenomas underwent transsphenoidal microsurgery. Gross-total resection was performed in 19 cases (26%), subtotal resection in 31 cases (42%), partial resection in 21 cases (29%), and the degree of resection was <50% in only two cases (3%). After binary logistic analysis, it was found that it was more difficult to completely remove giant pituitary adenomas with a Knosp grade 3-4 [odds ratio (OR) = 0.214, 95% confidence interval (CI): 0.05-0.917; P = 0.038], greater proportion of tumor suprasellar volume (odds ratio = 0.937, 95% confidence interval: 0.898-0.978; P = 0.003), and intraoperative evidence of invasion of the cavernous sinus (odds ratio = 0.187, 95% CI: 0.039-0.898; P = 0.036). Conclusion: It is difficult to remove a giant pituitary adenoma invading the cavernous sinus completely with a higher degree of invasion of the suprasellar region using microscopic transsphenoidal surgery. The combined application of multiple surgical methods can help to improve the degree of resection during a single operation.

Application of Convolutional Neural Network in the Diagnosis of Cavernous Sinus Invasion in Pituitary Adenoma.

Objectives: Convolutional neural network (CNN) is a deep-learning method for image classification and recognition based on a multi-layer NN. In this study, CNN was used to accurately assess cavernous sinus invasion (CSI) in pituitary adenoma (PA). Methods: A total of 371 patients with PA were enrolled in the retrospective study. The cohort was divided into the invasive (n = 102) and non-invasive groups (n = 269) based on surgically confirmed CSI. Images were selected on the T1-enhanced imaging on MR scans. The cohort underwent a fivefold division of randomized datasets for cross-validation. Then, a tenfold augmented dataset (horizontal flip and rotation) of the training set was enrolled in the pre-trained Resnet50 model for transfer learning. The testing set was imported into the trained model for evaluation. Gradient-weighted class activation mapping (Grad-CAM) was used to obtain the occlusion map. The diagnostic values were compared with different dichotomizations of the Knosp grading system (grades 0-1/2-4, 0-2/3a-4, and 0-3a/3b-4). Results: Based on Knosp grades, 20 cases of grade 0, 107 cases of grade 1, 82 cases of grade 2, 104 cases of grade 3a, 22 cases of grade 3b, and 36 cases of grade 4 were recorded. The CSI rates were 0%, 3.7%, 18.3%, 37.5%, 54.5%, and 88.9%. The predicted accuracies of the three dichotomies were 60%, 74%, and 81%. The area under the receiver operating characteristic (AUC-ROC) of Knosp grade for CSI prediction was 0.84; the cutoff was 2.5 with a Youden value of 0.62. The accuracies of the CNN model ranged from 0.80 to 0.96, with AUC-ROC values ranging from 0.89 to 0.98. The Grad-CAM saliency maps confirmed that the region of interest of the model was around the sellar region. Conclusions: We constructed a CNN model with a high proficiency at CSI diagnosis. A more accurate CSI identification was achieved with the constructed CNN than the Knosp grading system.

Predictive factors for delayed hyponatremia after transsphenoidal surgery in patients with pituitary adenomas.

Objective: The aim of this study was to evaluate the incidence and duration of delayed hyponatremia and to assess the factors influencing the development of delayed hyponatremia after transsphenoidal surgery (TSS) in pituitary adenomas. Methods: We retrospectively analyzed the clinical data of patients with pituitary adenoma who underwent TSS. Univariable and multivariable statistics were carried out to identify factors independently associated with the occurrence of delayed hyponatremia. Results: Of the 285 patients with pituitary adenoma who underwent microscopic TSS, 44 (15.4%) developed postoperative-delayed hyponatremia and 241 (84.6%) did not. The onset of delayed hyponatremia occurred an average of 5.84 days post-surgery and persisted for an average of 5.36 days. Logistic regression analysis showed the highest risk of delayed hyponatremia in patients with significant change in tumor cavity height (odds ratio (OR), 1.158; 95% CI, 1.062, 1.262; P = 0.001), preoperative hypothalamus-pituitary-thyroid axis hypofunction (OR, 3.112; 95% CI, 1.481, 6.539; P = 0.003), and significant difference in blood sodium levels before and 2 days after TSS (OR, 1.101; 95% CI, 1.005, 1.206; P = 0.039). Conclusions: Preoperative hypothyroidism, difference in blood sodium levels before and 2 days after TSS, and the change in tumor cavity height after TSS played important roles in predicting postoperative-delayed hyponatremia onset in patients with pituitary adenomas.

Machine-Learning Prediction of Postoperative Pituitary Hormonal Outcomes in Nonfunctioning Pituitary Adenomas: A Multicenter Study.

Objective: No accurate predictive models were identified for hormonal prognosis in non-functioning pituitary adenoma (NFPA). This study aimed to develop machine learning (ML) models to facilitate the prognostic assessment of pituitary hormonal outcomes after surgery. Methods: A total of 215 male patients with NFPA, who underwent surgery in four medical centers from 2015 to 2021, were retrospectively reviewed. The data were pooled after heterogeneity assessment, and they were randomly divided into training and testing sets (172:43). Six ML models and logistic regression models were developed using six anterior pituitary hormones. Results: Only thyroid-stimulating hormone (p < 0.001), follicle-stimulating hormone (p < 0.001), and prolactin (PRL; p < 0.001) decreased significantly following surgery, whereas growth hormone (GH) (p < 0.001) increased significantly. The postoperative GH (p = 0.07) levels were slightly higher in patients with gross total resection, but the PRL (p = 0.03) level was significantly lower than that in patients with subtotal resection. The optimal model achieved area-under-the-receiver-operating-characteristic-curve values of 0.82, 0.74, and 0.85 in predicting hormonal hypofunction, new deficiency, and hormonal recovery following surgery, respectively. According to feature importance analyses, the preoperative levels of the same type and other hormones were all important in predicting postoperative individual hormonal hypofunction. Conclusion: Fluctuation in anterior pituitary hormones varies with increases and decreases because of transsphenoidal surgery. The ML models could accurately predict postoperative pituitary outcomes based on preoperative anterior pituitary hormones in NFPA.

The Difference Between Preoperative and Postoperative Pituitary Stalk Deviation Angles Can Predict Delayed Hyponatremia After Transsphenoidal Surgery.

OBJECTIVE: Our aim was to assess the factors influencing the development of delayed hyponatremia after transsphenoidal surgery (TSS) for pituitary adenomas and analyze the effect of the difference between preoperative and postoperative pituitary stalk deviation angles on delayed hyponatremia. METHODS: A retrospective study was performed on the clinical data of patients with pituitary adenomas who were treated with TSS at a single institution. On the basis of the observation of indicators such as pituitary stalk deviation angle and length of "measurable pituitary stalk" on magnetic resonance imaging, we determined the predictors of postoperative delayed hyponatremia through univariate and multivariate analyses. RESULTS: Microscopic TSS was performed in 422 patients with pituitary adenoma, of whom 66 experienced postoperative delayed hyponatremia. Logistic regression analysis showed that the risk of delayed hyponatremia was greater for patients with a large difference between preoperative and postoperative pituitary stalk deviation angle (odds ratio = 1.040, 95% confidence interval: 1.018-1.051; P < 0.001) and a large difference in the "measurable pituitary stalk" (odds ratio = 1.128, 95% confidence interval: 1.011-1.258; P = 0.032), and patients with high blood sodium on the second day after surgery have a lower probability of developing delayed hyponatremia. CONCLUSIONS: This study is the first to suggest the important role of the difference between preoperative and postoperative pituitary stalk deviation angles in predicting the development of delayed hyponatremia after TSS for pituitary adenomas.

Diaphragma sellae sinking can predict the onset of hyponatremia after transsphenoidal surgery for pituitary adenomas.

PURPOSE: Sinking of the diaphragma sellae (DS) may stretch the pituitary stalk, which in turn impairs neurohypophyseal function; thus, it may play a role in the development of postoperative hyponatremia. We aimed to assess the factors influencing the development of hyponatremia after transsphenoidal surgery (TSS) for pituitary adenomas and analyze the effect of DS sinking on hyponatremia. METHODS: After applying the inclusion and exclusion criteria, we retrospectively analyzed the clinical data of patients with pituitary adenoma who underwent TSS. The pituitary gland was scanned using a 3.0-T magnetic resonance imaging, and sagittal and coronal images were acquired. We evaluated the following: preoperative and postoperative hypothalamus‒pituitary‒thyroid axis function, hypothalamus‒pituitary‒adrenal axis function, intra-operative cerebrospinal fluid leaks, diabetes insipidus, hyponatremia, time from the day of surgery to the day of discharge, and time of hyponatremia onset. RESULTS: Of the 460 patients who had microscopic TSS for pituitary adenoma, 83 experienced postoperative hyponatremia. Hyponatremia occurred approximately 5.25 days postoperatively and persisted for 5.54 days. The lowest average blood sodium level was 123.9 mEq/L, which occurred at 7.49 days after surgery. Logistic regression analysis showed that the risk of hyponatremia was greater for patients with a significant DS sinking depth, a large pituitary stalk deviation angle difference, and a longer postoperative "measurable pituitary stalk". The difference in blood sodium levels between pre-TSS and 2 days post-TSS was also an independent predictor of postoperative hyponatremia onset. CONCLUSION: DS sinking plays an important role in predicting hyponatremia onset after TSS for pituitary adenomas.

Outcomes of Preinjury Use of Statins in Patients with Traumatic Brain Injury: A Systematic Review and Meta-analysis.

BACKGROUND: No completely effective pharmacotherapies have been developed to improve the outcomes of traumatic brain injury (TBI). Given the reporting of cohort studies suggesting that preinjury statin use may reduce TBI-associated mortality, this study aimed to evaluate the effects of statin use in patients with TBI. METHODS: This study was performed according to the PRISMA guidelines. The PubMed, Embase, Cochrane Central, Web of Science, and China National Knowledge Infrastructure databases were searched from inception until April 13, 2021, using a search strategy that included 2 main terms: "statins" and "traumatic brain injury." The outcomes were mortality, hospital length of stay, and intensive care unit length of stay, which were evaluated using a random-effects model and represented by the pooled risk ratio with 95% confidence intervals. RESULTS: The search results identified 7 eligible studies, with a total of 111,935 patients with brain injury. Preinjury statin use in patients with TBI was associated with a significantly decreased risk of mortality compared with that in nonusers (risk ratio, 0.75; 95% confidence interval, 0.59-0.94; I2 = 53%). Subgroup analysis showed that statin withdrawal might increase mortality. Sensitivity analysis showed that the results were stable and robust. CONCLUSIONS: Preinjury statin use may contribute to mortality reduction in patients with TBI, whereas statin withdrawal might increase mortality. In clinical management, statin use should not be discontinued after TBI.

Diagnostic value of Knosp grade and modified Knosp grade for cavernous sinus invasion in pituitary adenomas: a systematic review and meta-analysis.

OBJECTIVE: This study aimed to examine the reliability of existing Knosp grade (KG) and modified KG for cavernous sinus invasion (CSI). METHODS: A thorough search of literature from 1993 to 2020 in six databases was performed. Studies that reported the intraoperative confirmation of CSI using KG and/or modified KG were included. The pooled estimates were calculated by meta-analysis with a bivariate mixed-effect model and the assessment of heterogeneity with I2 statistic. RESULTS: The final search yielded 12 eligible studies, which enrolled 3006 patients assessed with KG and 1315 patients assessed with modified KG. The results of the area under the receiver operating characteristic curve showed the good discriminative abilities of grades 2-4 (0.90), 3-4 (0.86) and 3B-4 (0.91) in predicting CSI. Grade 3A showed a remarkably lower CSI rate (44% versus 81%) and incomplete resection rate (26% versus 52%) than grade 3B. Grades 0 and 1 showed a low CSI rate. CSI and incomplete resection rates in grade 2 (30% and 21%, respectively) were close to those in grade 3A. CONCLUSIONS: Modifying the KG improved its prognostic role in CSI and gross total resection. However, these grading systems cannot be used as the group standard for invasive and non-invasive pituitary adenomas (PAs) because of the weak reliability of the scale's middle grades (grades 2 and 3A). Authors of future PA studies should consider reporting KG as high (grades 3B and 4), medium (grades 2 and 3A) and low (grades 0 and 1) to optimise the application of the scale.