Classification and Regression Tree Analysis for the Prediction of Aggressive Prostate Cancer on Biopsy

Efficacy of Postoperative Unilateral Neck Irradiation in Patients with Buccal Mucosa Squamous Carcinoma with Extranodal Extension: A Propensity Score Analysis

Unilateral radiotherapy (RT) as a postoperative treatment for multiple ipsilateral lymph node (LN) metastases remains controversial. We investigated the efficacy of postoperative unilateral RT for buccal mucosa squamous cell carcinoma (BMSCC) with extranodal extensions (ENEs). We retrospectively reviewed the clinical records of 186 patients with ENE+ BMSCC who received postoperative RT during 1997-2016. Propensity score matching was used to establish comparable cohorts. The endpoints were contralateral nodal control (CLNC), overall survival (OS), disease-free survival (DFS), distant metastasis-free survival (DMFS), local control (LC), and regional control (RC). After matching, 123 patients were selected for analysis; 45 (36.6%) and 78 (63.4%) patients underwent unilateral and bilateral RT, respectively. The median follow-up was 36.27 months. The survival outcomes in the unilateral and bilateral RT groups were similar: 3-year CLNC (85.6% vs. 89.1%, p = 0.748), OS (53.2% vs. 57.4%, p = 0.229), DFS (46.5% vs. 48.6%, p = 0.515), DMFS (70.7% vs. 72.0%, p = 0.499), LC (78.0% vs. 75.6%, p = 0.692), and RC (79.9% vs. 76.2%, p = 0.465). On multivariable Cox regression analysis, unilateral and bilateral RT showed comparable outcomes; the number of ENEs ≥ 4 was the only significant prognostic factor for all clinical outcomes. Using decision tree analysis, we classified our patients to have a low, intermediate, or high risk of contralateral failure based on three factors: number of ENEs, margin status, and tumor stage. In conclusion, postoperative unilateral RT did not worsen outcomes in patients with ENE+ BMSCC in this cohort. The decision tree model may assist physicians in optimizing and tailoring radiation fields.

Decision Tree Analysis for Prostate Cancer Prediction in Patients with Serum PSA 10 ng/ml or Less

Serum prostate-specific antigen (PSA) testing increases the number of persons who undergo prostate biopsy. However, the best possible strategy for selecting patients for prostate biopsy has not yet been defined. The aim of this study was to develop a classification and regression tree (CART) decision model that can be used to predict significant prostate cancer (PCa) in the course of prostate biopsy for patients with serum PSA levels of 10 ng/ml or less.

The following clinicopathological characteristics of patients who had undergone ultrasound-guided transrectal prostate biopsy were collected: age, PSA, digital rectal examination, volume of the prostate, and PSA density (PSAD). CART analysis was carried out by using all predictors. Different aspects of the predictive performances of the prediction model were assessed.

In this retrospective study, significant PCa values were detected in 26 (26.8%) of a total of 97 patients. The CART model had three branching levels based on PSAD as the most decisive variable and age. The model sensitivity was 73.1%, the specificity was 80.3% and the accuracy was 78.3%. Our model showed an area under the receiver operating characteristic curve of 82.6%. The model was well calibrated.

In conclusion, CART analysis determined that PSAD was the key parameter for the identification of patients with a minimal risk for positive biopsies. The model showed a good discrimination capacity that surpassed individual predictors. However, before recommending its use in clinical practice, an evaluation of a larger and more complete database is necessary for the prediction of significant PCa.

RERT: A Novel Regression Tree Approach to Predict Extrauterine Disease in Endometrial Carcinoma Patients

Some aspects of endometrial cancer (EC) preoperative work-up are still controversial, and debatable are the roles played by lymphadenectomy and radical surgery. Proper preoperative EC staging can help design a tailored surgical treatment, and this study aims to propose a new algorithm able to predict extrauterine disease diffusion. 293 EC patients were consecutively enrolled, and age, BMI, children’s number, menopausal status, contraception, hormone replacement therapy, hypertension, histological grading, clinical stage, and serum HE4 and CA125 values were preoperatively evaluated. In order to identify before surgery the most important variables able to classify EC patients based on FIGO stage, we adopted a new statistical approach consisting of two-steps: 1) Random Forest with its relative variable importance; 2) a novel algorithm able to select the most representative Regression Tree (RERT) from an ensemble method. RERT, built on the above mentioned variables, provided a sensitivity, specificity, NPV and PPV of 90%, 76%, 94% and 65% respectively, in predicting FIGO stage > I. Notably, RERT outperformed the prediction ability of HE4, CA125, Logistic Regression and single cross-validated Regression Tree. Such algorithm has great potential, since it better identifies the true early-stage patients, thus providing concrete support in the decisional process about therapeutic options to be performed.

Prostate tissue characterization/classification in 144 patient population using wavelet and higher order spectra features from transrectal ultrasound images

In this work, we have proposed an on-line computer-aided diagnostic system called "UroImage" that classifies a Transrectal Ultrasound (TRUS) image into cancerous or non-cancerous with the help of non-linear Higher Order Spectra (HOS) features and Discrete Wavelet Transform (DWT) coefficients. The UroImage system consists of an on-line system where five significant features (one DWT-based feature and four HOS-based features) are extracted from the test image. These on-line features are transformed by the classifier parameters obtained using the training dataset to determine the class. We trained and tested six classifiers. The dataset used for evaluation had 144 TRUS images which were split into training and testing sets. Three-fold and ten-fold cross-validation protocols were adopted for training and estimating the accuracy of the classifiers. The ground truth used for training was obtained using the biopsy results. Among the six classifiers, using 10-fold cross-validation technique, Support Vector Machine and Fuzzy Sugeno classifiers presented the best classification accuracy of 97.9% with equally high values for sensitivity, specificity and positive predictive value. Our proposed automated system, which achieved more than 95% values for all the performance measures, can be an adjunct tool to provide an initial diagnosis for the identification of patients with prostate cancer. The technique, however, is limited by the limitations of 2D ultrasound guided biopsy, and we intend to improve our technique by using 3D TRUS images in the future.

Brain metastases during follow-up of children and adolescents with extracranial malignant germ cell tumors: risk adapted management decision tree analysis based on data of the MAHO/MAKEI-registry

BACKGROUND

The overall risk for brain metastases among children and adolescents with extracranial malignant germ cell tumors (mGCT) is low but may vary between subgroups. Early identification of subgroups with an increased risk for brain metastasis is therefore important.

PROCEDURE

We analyzed 900/2,160 patients from the German MAHO/MAKEI registry on children and adolescents with malignant extracranial GCT (pure teratomas (grade 0-3) were not included). For follow-up evaluation, patients with brain metastases at diagnosis and those with a follow-up shorter than 32 months after diagnosis (longest interval to brain metastases in our cohort) were excluded. Patients were censored at detection of brain metastases or death due to other causes. A decision tree analysis considering age, gender, site of primary tumor, and presence of other metastases at diagnosis as risk factors for brain metastases was performed.

RESULTS

Among 838 eligible patients, 9 acquired brain metastases during follow-up, accounting for death in 5. There were no brain metastases in absence of extracranial metastases at diagnosis. If extracranial metastases were detected in absence of mediastinal mGCT the risk for brain metastases was 1.2% (95% CI: 0.2-3.5.%). In contrast, risk was increased to 37.5 (95% CI 15.2-64.6%) in patients with mediastinal GCTs and extracranial metastases.

CONCLUSION

A high-risk subgroup is detected with a decision tree analysis approach. These patients may benefit from an intensified chemotherapy. Close surveillance for CNS-metastases is warranted in this high-risk group while less close monitoring in low-risk patients is justified. Pediatr Blood Cancer 2013;60:217-223. © 2012 Wiley Periodicals, Inc.

Routine laboratory tests can predict in-hospital mortality in acute exacerbations of COPD

Chronic obstructive pulmonary disease (COPD) has a rising global incidence and acute exacerbation of COPD (AECOPD) carries a high health-care economic burden. Classification and regression tree (CART) analysis is able to create decision trees to classify risk groups. We analysed routinely collected laboratory data to identify prognostic factors for inpatient mortality with AECOPD from our large district hospital. Data from 5,985 patients with 9,915 admissions for AECOPD over a 7-year period were examined. Randomly allocated training (n = 4,986) or validation (n = 4,929) data sets were developed and CART analysis was used to model the risk of all-cause death during admission. Inpatient mortality was 15.5%, mean age was 71.5 (±11.5) years, 56.2% were male, and mean length of stay was 9.2 (±12.2) days. Of 29 variables used, CART analysis identified three (serum albumin, urea, and arterial pCO(2)) to predict in-hospital mortality in five risk groups, with mortality ranging from 3.0 to 23.4%. C statistic indices were 0.734 and 0.701 on the training and validation sets, respectively, indicating good model performance. The highest-risk group (23.4% mortality) had serum urea >7.35 mmol/l, arterial pCO(2) >6.45 kPa, and normal serum albumin (>36.5 g/l). It is possible to develop clinically useful risk prediction models for mortality using laboratory data from the first 24 h of admission in AECOPD.

Staging of prostate cancer using automatic feature selection, sampling and Dempster-Shafer fusion

A novel technique of automatically selecting the best pairs of features and sampling techniques to predict the stage of prostate cancer is proposed in this study. The problem of class imbalance, which is prominent in most medical data sets is also addressed here. Three feature subsets obtained by the use of principal components analysis (PCA), genetic algorithm (GA) and rough sets (RS) based approaches were also used in the study. The performance of under-sampling, synthetic minority over-sampling technique (SMOTE) and a combination of the two were also investigated and the performance of the obtained models was compared. To combine the classifier outputs, we used the Dempster-Shafer (DS) theory, whereas the actual choice of combined models was made using a GA. We found that the best performance for the overall system resulted from the use of under sampled data combined with rough sets based features modeled as a support vector machine (SVM).

Validity of temporomandibular disorder examination procedures for assessment of temporomandibular joint status

INTRODUCTION

This hypothesis-generating study was performed to determine which items in the Research Diagnostic Criteria for Temporomandibular Disorders (RDC/TMD) and additional diagnostic tests have the best predictive accuracy for joint-related diagnoses.

METHODS

One hundred forty-nine TMD patients and 43 symptom-free subjects were examined in clinical examinations and with magnetic resonance imaging (MRI). The importance of each variable of the clinical examination for correct joint-related diagnosis was assessed by using MRI diagnoses. For this purpose, "random forest" statistical software (based on classification trees) was used.

RESULTS

Maximum unassisted jaw opening, maximum assisted jaw opening, history of locked jaw, joint sound with and without compression, joint pain, facial pain, pain on palpation of the lateral pterygoid area, and overjet proved suitable for distinguishing between subtypes of joint-related TMD. Measurement of excursion, protrusion, and midline deviation were less important.

CONCLUSIONS

The validity of clinical TMD examination procedures can be enhanced by using the 16 variables of greatest importance identified in this study. In addition to other variables, maximum unassisted and assisted opening and a history of locked jaw were important when assessing the status of the TMJ.

Identification of patients with low-risk for aneuploidy: comparative discriminatory models using linear and machine-learning classifiers in prostate cancer

BACKGROUND

Prostate needle biopsy (PNB) ploidy status has proven utility to predict adverse outcomes after prostatectomy. We sought to develop models to predict ploidy status using clinicopathologic variables.

METHODS

We identified a cohort of 169 patients with a diagnosis of prostatic adenocarcinoma on PNB, and estimated ploidy status (determined using Feulgen stained biopsy tissue) using four predictors, including age, prebiopsy PSA, highest Gleason score (GS), and the percentage of involvement by carcinoma at the biopsy site with the highest GS (PCARBX). Logistic regression (LR), Neural Network (NN), and CART classifiers were constructed.

RESULTS

Univariate analyses revealed all four predictors to be significantly associated with ploidy status. On multivariable analyses, LR identified a 2-parameter model, including GS and PCARBX that had a significant ability to predict ploidy status with a 74% and 75% correct classification rate (CCR), respectively. Using the same variables, CART and NN yielded similar CCRs of 70.4%. Within GS = 6 cohort, the CART model classified over 90% of biopsies as diploid when patients had a PCARBX < 55% and a log(PSA) < 1.7.

CONCLUSIONS

Our study demonstrates that models using GS and PCARBX are able to predict PNB ploidy status with acceptable accuracy. While machine learning classifier-derived models yield similar accuracy as LR-derived models, the latter methodology has the distinct advantage of being applicable in future datasets to estimate case-specific predictions. This information may be useful in identifying potentially aneuploid patients, who can then be targeted for more aggressive therapy.