Multiparametric MRI-Based Interpretable Radiomics Machine Learning Model Differentiates Medulloblastoma and Ependymoma in Children: A Two-Center Study

RATIONALE AND OBJECTIVES

Medulloblastoma (MB) and Ependymoma (EM) in children, share similarities in age group, tumor location, and clinical presentation. Distinguishing between them through clinical diagnosis is challenging. This study aims to explore the effectiveness of using radiomics and machine learning on multiparametric magnetic resonance imaging (MRI) to differentiate between MB and EM and validate its diagnostic ability with an external set.

MATERIALS AND METHODS

Axial T2 weighted image (T2WI) and contrast-enhanced T1weighted image (CE-T1WI) MRI sequences of 135 patients from two centers were collected as train/test sets. Volume of interest (VOI) was manually delineated by an experienced neuroradiologist, supervised by a senior. Feature selection analysis and the least absolute shrinkage and selection operator (LASSO) algorithm identified valuable features, and Shapley additive explanations (SHAP) evaluated their significance. Five machine-learning classifiers-extreme gradient boosting (XGBoost), Bernoulli naive Bayes (Bernoulli NB), Logistic Regression (LR), support vector machine (SVM), linear support vector machine (Linear SVC) classifiers were built based on T2WI (T2 model), CE-T1WI (T1 model), and T1 + T2WI (T1 + T2 model). A human expert diagnosis was developed and corrected by senior radiologists. External validation was performed at Sun Yat-Sen University Cancer Center.

RESULTS

31 valuable features were extracted from T2WI and CE-T1WI. XGBoost demonstrated the highest performance with an area under the curve (AUC) of 0.92 on the test set and maintained an AUC of 0.80 during external validation. For the T1 model, XGBoost achieved the highest AUC of 0.85 on the test set and the highest accuracy of 0.71 on the external validation set. In the T2 model, XGBoost achieved the highest AUC of 0.86 on the test set and the highest accuracy of 0.82 on the external validation set. The human expert diagnosis had an AUC of 0.66 on the test set and 0.69 on the external validation set. The integrated T1 + T2 model achieved an AUC of 0.92 on the test set, 0.80 on the external validation set, achieved the best performance. Overall, XGBoost consistently outperformed in different classification models.

CONCLUSION

The combination of radiomics and machine learning on multiparametric MRI effectively distinguishes between MB and EM in childhood, surpassing human expert diagnosis in training and testing sets.

Prostate cancer epidemiology and prognostic factors in the United States

Objective

Using the latest cohort study of prostate cancer patients, explore the epidemiological trend and prognostic factors, and develop a new nomogram to predict the specific survival rate of prostate cancer patients.

Methods

Patients with prostate cancer diagnosed from January 1, 1975 to December 31, 2019 in the Surveillance, Epidemiology, and End Results Program (SEER) database were extracted by SEER stat software for epidemiological trend analysis. General clinical information and follow-up data were also collected from 105 135 patients with pathologically diagnosed prostate cancer from January 1, 2010 to December 1, 2019. The factors affecting patient-specific survival were analyzed by Cox regression, and the factors with the greatest influence on specific survival were selected by stepwise regression method, and nomogram was constructed. The model was evaluated by calibration plots, ROC curves, Decision Curve Analysis and C-index.

Results

There was no significant change in the age-adjusted incidence of prostate cancer from 1975 to 2019, with an average annual percentage change (AAPC) of 0.45 (95% CI:-0.87~1.80). Among the tumor grade, the most significant increase in the incidence of G2 prostate cancer was observed, with an AAPC of 2.99 (95% CI:1.47~4.54); the most significant decrease in the incidence of G4 prostate cancer was observed, with an AAPC of -10.39 (95% CI:-13.86~-6.77). Among the different tumor stages, the most significant reduction in the incidence of localized prostate cancer was observed with an AAPC of -1.83 (95% CI:-2.76~-0.90). Among different races, the incidence of prostate cancer was significantly reduced in American Indian or Alaska Native and Asian or Pacific Islander, with an AAPC of -3.40 (95% CI:-3.97~-2.82) and -2.74 (95% CI:-4.14~-1.32), respectively. Among the different age groups, the incidence rate was significantly increased in 15-54 and 55-64 age groups with AAPC of 4.03 (95% CI:2.73~5.34) and 2.50 (95% CI:0.96~4.05), respectively, and significantly decreased in ≥85 age group with AAPC of -2.50 (95% CI:-3.43~-1.57). In addition, age, tumor stage, race, PSA and gleason score were found to be independent risk factors affecting prostate cancer patient-specific survival. Age, tumor stage, PSA and gleason score were most strongly associated with prostate cancer patient-specific survival by stepwise regression screening, and nomogram prediction model was constructed using these factors. The Concordance indexes are 0.845 (95% CI:0.818~0.872) and 0.835 (95% CI:0.798~0.872) for the training and validation sets, respectively, and the area under the ROC curves (AUC) at 3, 6, and 9 years was 0.7 or more for both the training and validation set samples. The calibration plots indicated a good agreement between the predicted and actual values of the model.

Conclusions

Although there was no significant change in the overall incidence of prostate cancer in this study, significant changes occurred in the incidence of prostate cancer with different characteristics. In addition, the nomogram prediction model of prostate cancer-specific survival rate constructed based on four factors has a high reference value, which helps physicians to correctly assess the patient-specific survival rate and provides a reference basis for patient diagnosis and prognosis evaluation.

Preliminary study on miRNA in prostate cancer

OBJECTIVE

To screen for miRNAs differentially expressed in prostate cancer and prostate hyperplasia tissues and to validate their association with prostate cancer.

METHODS

Patients diagnosed by pathology in the Department of Urology of the First Affiliated Hospital of Xinjiang Medical University from October 2021 to June 2022 were selected, and their general clinical information, blood samples, and prostate tissue samples were collected. miRNA microarray technology was performed to obtain differentially expressed miRNAs in prostate cancer and hyperplasia tissues, and miRNAs to be studied were screened by microarray results and review of relevant literature. The detection of miRNA expression in the patients' blood and prostate tissue samples was measured. The miRNA-222-mimics were transfected into PC3 cells, and cell biology experiments such as CCK8, scratch, Transwell, and flow cytometry were performed to detect the effects of overexpressed miRNA-222 on the growth and proliferation, invasive ability, apoptotic ability, and metastatic ability of prostate cancer cells.

RESULTS

The results of the miRNA microarray showed that there were many differentially expressed miRNAs in prostate cancer and hyperplasia tissues, and four miRNAs, miRNA-144, miRNA-222, miRNA-1248, and miRNA-3651 were finally selected as the subjects by reviewing relevant literature. The results showed that the expression of miRNA-222 in prostate cancer tissues was lower than that in prostate hyperplasia tissues (P < 0.05). The expression of miRNA-222, miRNA-1248, and miRNA-3651 in blood samples of prostate cancer patients was lower than that in prostate hyperplasia patients (P < 0.05). The analysis results indicated that the f/t ratio and the relative expression of miRNA-222 and miRNA-1248 were independent influences of prostate cancer (P < 0.05), in which overexpression of miRNA-222 decreased the proliferative, invasive, and metastatic abilities of PC3 cells and enhanced the level of apoptosis of cancer cells.

CONCLUSIONS

Although there was no significant change in the overall incidence of prostate cancer in this study, significant changes occurred in the incidence of prostate cancer with different characteristics. In addition, the nomogram prediction model of prostate cancer-specific survival rate constructed based on four factors has a high reference value, which helps physicians to correctly assess the patient-specific survival rate and provides a reference basis for patient diagnosis and prognosis evaluation.

Effect of interaction between occupational stress and polymorphisms of MTHFR gene and SELE gene on hypertension

BACKGROUND

Gene-environment interaction is related to the prevalence of hypertension, but the impact of genetic polymorphisms on hypertension may vary due to different geography and population.

OBJECTIVE

To explore the impact of the interaction among occupational stress and MTHFR gene and SELE gene polymorphism on the prevalence of hypertension in Xinjiang oil workers.

METHODS

A case-control study was conducted on 310 oil workers. In an oilfield base in Karamay City, Xinjiang, 155 hypertensive patients aged 18~60 years old with more than one year of service were selected as the case group, and 155 oil workers without hypertension were selected as the control group according to the 1:1 matching principle (matching conditions: the gender and shift were the same. The age is around 2 years old). The Occupational Stress Scale was used to evaluate the degree of occupational stress, PCR technique was used to detect MTHFR and SELE gene polymorphism, Logistic regression analysis was used to analyze the effects of gene and occupational stress on hypertension, and gene-gene and gene-environment interactions were analyzed by generalized multi-factor dimension reduction method.

RESULTS

The G98T polymorphism of SELE gene (χ² = 6.776, P = 0.034), the C677T (χ² = 7.130, P = 0.028) and A1298C (χ² = 12.036, P = 0.002) loci of MTHFR gene and the degree of occupational stress (χ² = 11.921, P = 0.003) were significantly different between the case group and the control group. The genotypes GT at the G98T polymorphism of the SELE gene (OR = 2.151, 95% CI [1.227-3.375]), and the dominant model (AC/CC vs AA, OR = 1.925, 95% CI [1.613-3.816]); AC and CC at the A1298C polymorphism of the MTHFR gene (OR _AC = 1.917, 95% CI [1.064-3.453]; OR _CC = 2.233, 95% CI [1.082-4.609]), the additive model (CC vs AA, OR = 2.497, 95% CI [1.277-4.883]) and the dominant model (AC/CC vs AA, OR = 2.012, 95% CI [1.200-3.373]); at the C677T polymorphism of the MTHFR gene CT and TT (OR _CT = 1.913, 95% CI [1.085-3.375]; OR _TT = 3.117, 95% CI [1.430-6.795]), the additive model (CC vs AA, OR = 1.913, 95% CI [1.085-3.375]) and the dominant model (AC/CC vs AA, OR = 2.012, 95% CI [1.200-3.373]), which could increase hypertension risk (P < 0.05). The gene-gene interaction showed that there was a positive interaction between the A1298C and C677T sites of the MTHFR gene, and the gene-occupational stress interaction showed that there was a positive interaction between the A1298C and C677T sites of the MTHFR gene and the occupational stress.

CONCLUSION

The interaction of gene mutation and occupational stress in Xinjiang oil workers maybe increase the risk of hypertension.