An evaluation of penalised survival methods for developing prognostic models with rare events

A Systematic Review of Artificial Intelligence Models for Time-to-Event Outcome Applied in Cardiovascular Disease Risk Prediction

Artificial intelligence (AI) based predictive models for early detection of cardiovascular disease (CVD) risk are increasingly being utilised. However, AI based risk prediction models that account for right-censored data have been overlooked. This systematic review (PROSPERO protocol CRD42023492655) includes 33 studies that utilised machine learning (ML) and deep learning (DL) models for survival outcome in CVD prediction. We provided details on the employed ML and DL models, eXplainable AI (XAI) techniques, and type of included variables, with a focus on social determinants of health (SDoH) and gender-stratification. Approximately half of the studies were published in 2023 with the majority from the United States. Random Survival Forest (RSF), Survival Gradient Boosting models, and Penalised Cox models were the most frequently employed ML models. DeepSurv was the most frequently employed DL model. DL models were better at predicting CVD outcomes than ML models. Permutation-based feature importance and Shapley values were the most utilised XAI methods for explaining AI models. Moreover, only one in five studies performed gender-stratification analysis and very few incorporate the wide range of SDoH factors in their prediction model. In conclusion, the evidence indicates that RSF and DeepSurv models are currently the optimal models for predicting CVD outcomes. This study also highlights the better predictive ability of DL survival models, compared to ML models. Future research should ensure the appropriate interpretation of AI models, accounting for SDoH, and gender stratification, as gender plays a significant role in CVD occurrence.

Development and Validation of Prognostic Models Using Radiomic Features from Pre-Treatment Positron Emission Tomography (PET) Images in Head and Neck Squamous Cell Carcinoma (HNSCC) Patients

High-dimensional radiomics features derived from pre-treatment positron emission tomography (PET) images offer prognostic insights for patients with head and neck squamous cell carcinoma (HNSCC). Using 124 PET radiomics features and clinical variables (age, sex, stage of cancer, site of cancer) from a cohort of 232 patients, we evaluated four survival models-penalized Cox model, random forest, gradient boosted model and support vector machine-to predict all-cause mortality (ACM), locoregional recurrence/residual disease (LR) and distant metastasis (DM) probability during 36, 24 and 24 months of follow-up, respectively. We developed models with five-fold cross-validation, selected the best-performing model for each outcome based on the concordance index (C-statistic) and the integrated Brier score (IBS) and validated them in an independent cohort of 102 patients. The penalized Cox model demonstrated better performance for ACM (C-statistic = 0.70, IBS = 0.12) and DM (C-statistic = 0.70, IBS = 0.08) while the random forest model displayed better performance for LR (C-statistic = 0.76, IBS = 0.07). We conclude that the ML-based prognostic model can aid clinicians in quantifying prognosis and determining effective treatment strategies, thereby improving favorable outcomes in HNSCC patients.

Bias reduction for semi-competing risks frailty model with rare events: application to a chronic kidney disease cohort study in South Korea

In a semi-competing risks model in which a terminal event censors a non-terminal event but not vice versa, the conventional method can predict clinical outcomes by maximizing likelihood estimation. However, this method can produce unreliable or biased estimators when the number of events in the datasets is small. Specifically, parameter estimates may converge to infinity, or their standard errors can be very large. Moreover, terminal and non-terminal event times may be correlated, which can account for the frailty term. Here, we adapt the penalized likelihood with Firth's correction method for gamma frailty models with semi-competing risks data to reduce the bias caused by rare events. The proposed method is evaluated in terms of relative bias, mean squared error, standard error, and standard deviation compared to the conventional methods through simulation studies. The results of the proposed method are stable and robust even when data contain only a few events with the misspecification of the baseline hazard function. We also illustrate a real example with a multi-centre, patient-based cohort study to identify risk factors for chronic kidney disease progression or adverse clinical outcomes. This study will provide a better understanding of semi-competing risk data in which the number of specific diseases or events of interest is rare.

Comparison of likelihood penalization and variance decomposition approaches for clinical prediction models: A simulation study

Logistic regression is one of the most commonly used approaches to develop clinical risk prediction models. Developers of such models often rely on approaches that aim to minimize the risk of overfitting and improve predictive performance of the logistic model, such as through likelihood penalization and variance decomposition techniques. We present an extensive simulation study that compares the out-of-sample predictive performance of risk prediction models derived using the elastic net, with Lasso and ridge as special cases, and variance decomposition techniques, namely, incomplete principal component regression and incomplete partial least squares regression. We varied the expected events per variable, event fraction, number of candidate predictors, presence of noise predictors, and the presence of sparse predictors in a full-factorial design. Predictive performance was compared on measures of discrimination, calibration, and prediction error. Simulation metamodels were derived to explain the performance differences within model derivation approaches. Our results indicate that, on average, prediction models developed using penalization and variance decomposition approaches outperform models developed using ordinary maximum likelihood estimation, with penalization approaches being consistently superior over the variance decomposition approaches. Differences in performance were most pronounced on the calibration of the model. Performance differences regarding prediction error and concordance statistic outcomes were often small between approaches. The use of likelihood penalization and variance decomposition techniques methods was illustrated in the context of peripheral arterial disease.

Risk-prediction models for intravenous immunoglobulin resistance in Kawasaki disease: Risk-of-Bias Assessment using PROBAST

BACKGROUND

The prediction model of intravenous immunoglobulin (IVIG) resistance in Kawasaki disease can calculate the probability of IVIG resistance and provide a basis for clinical decision-making. We aim to assess the quality of these models developed in the children with Kawasaki disease.

METHODS

Studies of prediction models for IVIG-resistant Kawasaki disease were identified through searches in the PubMed, Web of Science, and Embase databases. Two investigators independently performed literature screening, data extraction, quality evaluation, and discrepancies were settled by a statistician. The checklist for critical appraisal and data extraction for systematic reviews of prediction modeling studies (CHARMS) was used for data extraction, and the prediction models were evaluated using the Prediction Model Risk of Bias Assessment Tool (PROBAST).

RESULTS

Seventeen studies meeting the selection criteria were included in the qualitative analysis. The top three predictors were neutrophil measurements (peripheral neutrophil count and neutrophil %), serum albumin level, and C-reactive protein (CRP) level. The reported area under the curve (AUC) values for the developed models ranged from 0.672 (95% confidence interval [CI]: 0.631-0.712) to 0.891 (95% CI: 0.837-0.945); The studies showed a high risk of bias (ROB) for modeling techniques, yielding a high overall ROB.

CONCLUSION

IVIG resistance models for Kawasaki disease showed high ROB. An emphasis on improving their quality can provide high-quality evidence for clinical practice.

IMPACT STATEMENT

This study systematically evaluated the risk of bias (ROB) of existing prediction models for intravenous immunoglobulin (IVIG) resistance in Kawasaki disease to provide guidance for future model development meeting clinical expectations. This is the first study to systematically evaluate the ROB of IVIG resistance in Kawasaki disease by using PROBAST. ROB may reduce model performance in different populations. Future prediction models should account for this problem, and PROBAST can help improve the methodological quality and applicability of prediction model development.

Comparing predictions among competing risks models with rare events: application to KNOW-CKD atudy-a multicentre cohort study of chronic kidney disease

A prognostic model to determine an association between survival outcomes and clinical risk factors, such as the Cox model, has been developed over the past decades in the medical field. Although the data size containing subjects' information gradually increases, the number of events is often relatively low as medical technology develops. Accordingly, poor discrimination and low predicted ability may occur between low- and high-risk groups. The main goal of this study was to evaluate the predicted probabilities with three existing competing risks models in variation with censoring rates. Three methods were illustrated and compared in a longitudinal study of a nationwide prospective cohort of patients with chronic kidney disease in Korea. The prediction accuracy and discrimination ability of the three methods were compared in terms of the Concordance index (C-index), Integrated Brier Score (IBS), and Calibration slope. In addition, we find that these methods have different performances when the effects are linear or nonlinear under various censoring rates.

Predicting Fear Extinction in Posttraumatic Stress Disorder

Fear extinction is the basis of exposure therapies for posttraumatic stress disorder (PTSD), but half of patients do not improve. Predicting fear extinction in individuals with PTSD may inform personalized exposure therapy development. The participants were 125 trauma-exposed adults (96 female) with a range of PTSD symptoms. Electromyography, electrocardiogram, and skin conductance were recorded at baseline, during dark-enhanced startle, and during fear conditioning and extinction. Using a cross-validated, hold-out sample prediction approach, three penalized regressions and conventional ordinary least squares were trained to predict fear-potentiated startle during extinction using 50 predictor variables (5 clinical, 24 self-reported, and 21 physiological). The predictors, selected by penalized regression algorithms, were included in multivariable regression analyses, while univariate regressions assessed individual predictors. All the penalized regressions outperformed OLS in prediction accuracy and generalizability, as indexed by the lower mean squared error in the training and holdout subsamples. During early extinction, the consistent predictors across all the modeling approaches included dark-enhanced startle, the depersonalization and derealization subscale of the dissociative experiences scale, and the PTSD hyperarousal symptom score. These findings offer novel insights into the modeling approaches and patient characteristics that may reliably predict fear extinction in PTSD. Penalized regression shows promise for identifying symptom-related variables to enhance the predictive modeling accuracy in clinical research.

Derivation and validation of a machine learning-based risk prediction model in patients with acute heart failure

BACKGROUND

Risk stratification is important in patients with acute heart failure (AHF), and a simple risk score that accurately predicts mortality is needed. The aim of this study is to develop a user-friendly risk-prediction model using a machine-learning method.

METHODS

A machine-learning-based risk model using least absolute shrinkage and selection operator (LASSO) regression was developed by identifying predictors of in-hospital mortality in the derivation cohort (REALITY-AHF), and its performance was externally validated in the validation cohort (NARA-HF) and compared with two pre-existing risk models: the Get With The Guidelines risk score incorporating brain natriuretic peptide and hypochloremia (GWTG-BNP-Cl-RS) and the acute decompensated heart failure national registry risk (ADHERE).

RESULTS

In-hospital deaths in the derivation and validation cohorts were 76 (5.1 %) and 61 (4.9 %), respectively. The risk score comprised four variables (systolic blood pressure, blood urea nitrogen, serum chloride, and C-reactive protein) and was developed according to the results of the LASSO regression weighting the coefficient for selected variables using a logistic regression model (4 V-RS). Even though 4 V-RS comprised fewer variables, in the validation cohort, it showed a higher area under the receiver operating characteristic curve (AUC) than the ADHERE risk model (AUC, 0.783 vs. 0.740; p = 0.059) and a significant improvement in net reclassification (0.359; 95 % CI, 0.10-0.67; p = 0.006). 4 V-RS performed similarly to GWTG-BNP-Cl-RS in terms of discrimination (AUC, 0.783 vs. 0.759; p = 0.426) and net reclassification (0.176; 95 % CI, -0.08-0.43; p = 0.178).

CONCLUSIONS

The 4 V-RS model comprising only four readily available data points at the time of admission performed similarly to the more complex pre-existing risk model in patients with AHF.

Development and validation of a new stage-specific nomogram model for predicting cancer-specific survival in patients in different stages of colon cancer: A SEER population-based study and external validation

Background

The effects of laterality of the primary tumor on survival in patients in different stages of colon cancer are contradictory. We still lack a strictly evaluated and validated survival prediction tool, considering the different roles of tumor laterality in different stages.

Methods

A total of 101,277 and 809 colon cancer cases were reviewed using the Surveillance, Epidemiology, and End Results database and the First Affiliated Hospital of Xi 'an Jiaotong University database, respectively. We established training sets, internal validation sets and external validation sets. We developed and evaluated stage-specific prediction models and unified prediction models to predict cancer-specific survival and compared the prediction abilities of these models.

Results

Compared with right-sided colon cancers, the risk of cancer-specific death of left-sided colon cancer patients was significantly higher in stage I/II but was markedly lower in stage III patients. We established stage-specific prediction models for stage I/II and stage III separately and established a unified prediction model for all stages. By evaluating and validating the validation sets, we reported high prediction ability and generalizability of the models. Furthermore, the stage-specific prediction models had better predictive power and efficiency than the unified model.

Conclusions

Right-sided colon cancer patients have better cancer-specific survival than left-sided colon cancer patients in stage I/II and worse cancer-specific survival in stage III. Using stage-specific prediction models can further improve the prediction of cancer-specific survival in colon cancer patients and guide clinical decisions.

Early versus Late initiation of direct oral Anticoagulants in post-ischaemic stroke patients with atrial fibrillatioN (ELAN): Protocol for an international, multicentre, randomised-controlled, two-arm, open, assessor-blinded trial

RATIONALE

Direct oral anticoagulants (DOAC) are highly effective in preventing ischaemic strokes in people with atrial fibrillation (AF). However, it is unclear how soon they should be started after acute ischaemic stroke (AIS). Early initiation may reduce early risk of recurrence but might increase the risk of haemorrhagic complications.

AIM

To estimate the safety and efficacy of early initiation of DOACs compared to late guideline-based initiation in people with AIS related to AF.

METHODS AND DESIGN

An international, multicentre, randomised (1:1) controlled, two-arm, open, assessor-blinded trial is being conducted. Early treatment is defined as DOAC initiation within 48 h of a minor or moderate stroke, or at day 6-7 following major stroke. Late treatment is defined as DOAC initiation after day 3-4 following minor stroke, after day 6-7 following moderate stroke and after day 12-14 following major stroke. Severity of stroke is defined according to imaging assessment of infarct size.

SAMPLE SIZE

ELAN will randomise 2000 participants 1:1 to early versus late initiation of DOACs. This assumes a risk difference of 0.5% favouring the early arm, allowing an upper limit of the 95% confidence interval up to 1.5% based on the Miettinen & Nurminen formula.

OUTCOMES

The primary outcome is a composite of symptomatic intracranial haemorrhage, major extracranial bleeding, recurrent ischaemic stroke, systemic embolism or vascular death at 30 ± 3 days after randomisation. Secondary outcomes include the individual components of the primary outcome at 30 ± 3 and 90 ± 7 days and functional status at 90 ± 7 days.

DISCUSSION

ELAN will estimate whether there is a clinically important difference in safety and efficacy outcomes following early anticoagulation with a DOAC compared to late guideline-based treatment in neuroimaging-selected people with an AIS due to AF.

Psychosocial markers of age at onset in bipolar disorder: a machine learning approach

BACKGROUND

Bipolar disorder is a chronic and severe mental health disorder. Early stratification of individuals into subgroups based on age at onset (AAO) has the potential to inform diagnosis and early intervention. Yet, the psychosocial predictors associated with AAO are unknown.

AIMS

We aim to identify psychosocial factors associated with bipolar disorder AAO.

METHOD

Using data from the Bipolar Disorder Research Network UK, we employed least absolute shrinkage and selection operator regression to identify psychosocial factors associated with bipolar disorder AAO. Twenty-eight factors were entered into our model, with AAO as our outcome measure.

RESULTS

We included 1022 participants with bipolar disorder (μ = 23.0, s.d. ± 9.86) in our model. Six variables predicted an earlier AAO: childhood abuse (β = -0.2855), regular cannabis use in the year before onset (β = -0.2765), death of a close family friend or relative in the 6 months before onset (β = -0.2435), family history of suicide (β = -0.1385), schizotypal personality traits (β = -0.1055) and irritable temperament (β = -0.0685). Five predicted a later AAO: the average number of alcohol units consumed per week in the year before onset (β = 0.1385); birth of a child in the 6 months before onset (β = 0.2755); death of parent, partner, child or sibling in the 6 months before onset (β = 0.3125); seeking work without success for 1 month or more in the 6 months before onset (β = 0.3505) and a major financial crisis in the 6 months before onset (β = 0.4575).

CONCLUSIONS

The identified predictor variables have the potential to help stratify high-risk individuals into likely AAO groups, to inform treatment provision and early intervention.

Development and Validation of Retinal Vasculature Nomogram in Suspected Angina Due to Coronary Artery Disease

AIMS

To develop and validate a nomogram using retinal vasculature features and clinical variables to predict coronary artery disease (CAD) in patients with suspected angina.

METHODS

The prediction model consisting of 795 participants was developed in a training set of 508 participants with suspected angina due to CAD, and data were collected from January 2018 to June 2019. The held-out validation was conducted with 287 consecutive patients from July 2019 to November 2019. All patients with suspected CAD received optical coherence tomography angiography (OCTA) examination before undergoing coronary CT angiography. LASSO regression model was used for data reduction and feature selection. Multivariable logistic regression analysis was used to develop the retinal vasculature model for predicting the probability of the presence of CAD.

RESULTS

Three potential OCTA parameters including vessel density of the nasal and temporal perifovea in the superficial capillary plexus and vessel density of the inferior parafovea in the deep capillary plexus were further selected as independent retinal vasculature predictors. Model clinical electrocardiogram (ECG) OCTA (clinical variables＋ECG＋OCTA) was presented as the individual prediction nomogram, with good discrimination (AUC of 0.942 [95% CI, 0.923-0.961] and 0.897 [95% CI, 0.861-0.933] in the training and held-out validation sets, respectively) and good calibration. Decision curve analysis indicated the clinical applicability of this retinal vasculature nomogram.

CONCLUSIONS

The presented retinal vasculature nomogram based on individual probability can accurately identify the presence of CAD, which could improve patient selection and diagnostic yield of aggressive testing before determining a diagnosis.

Comparison of standard and penalized logistic regression in risk model development

Objective

Regression models are ubiquitous in thoracic surgical research. We aimed to compare the value of standard logistic regression with the more complex but increasingly used penalized regression models using a recently published risk model as an example.

Methods

Using a standardized data set of clinical T1-3N0 esophageal cancer patients, we created models to predict the likelihood of unexpected pathologic nodal disease after surgical resection. Models were fitted using standard logistic regression or penalized regression (ridge, lasso, elastic net, and adaptive lasso). We compared the model performance (Brier score, calibration slope, C statistic, and overfitting) of standard regression with penalized regression models.

Results

Among 3206 patients with clinical T1-3N0 esophageal cancer, 668 (22%) had unexpected pathologic nodal disease. Of the 15 candidate variables considered in the models, the key predictors of nodal disease included clinical tumor stage, tumor size, grade, and presence of lymphovascular invasion. The standard regression model and all 4 penalized logistic regression models had virtually identical performance with Brier score ranging from 0.138 to 0.141, concordance index ranging from 0.775 to 0.788, and calibration slope from 0.965 to 1.05.

Conclusions

For predictive modeling in surgical outcomes research, when the data set is large and the outcome of interest is relatively frequent, standard regression models and the more complicated penalized models are very likely to have similar predictive performance. The choice of statistical methods for risk model development should be on the basis of the nature of the data at hand and good statistical practice, rather than the novelty or complexity of statistical models.

Predicting ecological footprint based on global macro indicators in G-20 countries using machine learning approaches

Paying attention to human activities in terms of land grazing infrastructure, crops, forest products, and carbon impact, the so-called ecological impact (EF) is one of the most important economic issues in the world. For the present study, global database data were used. The ability of the penalized regression (RR) approaches (including Ridge, Lasso and Elastic Net) and artificial neural network (ANN) to predict EF indices in the G-20 countries over the past two decades (1999-2018) was illustrated and compared. For this purpose, 10-fold cross-validation was used to evaluate the predictive performance and determine the penalty parameter for PR models. According to the results, the predictive performance compared to linear regression improved somewhat using the penalized methods. Using the elastic net model, more global macro indices were selected than Lasso. Although Lasso selected only a few indicators, it had better predictive performance among PR ns models. In addition to relative improvement in the predictive performance of PR methods, their interest in selecting a subset of indicators by shrinking coefficients and creating a parsimonious model was evident and significant. As a result, PR methods would be preferred, using variable selection and interpretive considerations to predictive performance alone. On the other hand, ANN models with higher determination coefficients (R²) and lower RMSE values performed significantly better than PR and OLS and showed that they were more accurate in predicting EF. Therefore, ANN could provide considerable and appropriate predictions for EF indicators in the G-20 countries.

Multivariate analysis of NMR-based metabolomic data

Nuclear magnetic resonance (NMR) spectroscopy allows for simultaneous detection of a wide range of metabolites and lipids. As metabolites act together in complex metabolic networks, they are often highly correlated, and optimal biological insight is achieved when using methods that take the correlation into account. For this reason, latent-variable-based methods, such as principal component analysis and partial least-squares discriminant analysis, are widely used in metabolomic studies. However, with increasing availability of larger population cohorts, and a shift from analysis of spectral data to using quantified metabolite levels, both more traditional statistical approaches and alternative machine learning methods have become more widely used. This review aims at providing an overview of the current state-of-the-art multivariate methods for the analysis of NMR-based metabolomic data as well as alternative methods, highlighting their strengths and limitations.

Developing clinical prediction models when adhering to minimum sample size recommendations: The importance of quantifying bootstrap variability in tuning parameters and predictive performance

Recent minimum sample size formula (Riley et al.) for developing clinical prediction models help ensure that development datasets are of sufficient size to minimise overfitting. While these criteria are known to avoid excessive overfitting on average, the extent of variability in overfitting at recommended sample sizes is unknown. We investigated this through a simulation study and empirical example to develop logistic regression clinical prediction models using unpenalised maximum likelihood estimation, and various post-estimation shrinkage or penalisation methods. While the mean calibration slope was close to the ideal value of one for all methods, penalisation further reduced the level of overfitting, on average, compared to unpenalised methods. This came at the cost of higher variability in predictive performance for penalisation methods in external data. We recommend that penalisation methods are used in data that meet, or surpass, minimum sample size requirements to further mitigate overfitting, and that the variability in predictive performance and any tuning parameters should always be examined as part of the model development process, since this provides additional information over average (optimism-adjusted) performance alone. Lower variability would give reassurance that the developed clinical prediction model will perform well in new individuals from the same population as was used for model development.

A novel cardiovascular magnetic resonance risk score for predicting mortality following surgical aortic valve replacement

The increasing prevalence of patients with aortic stenosis worldwide highlights a clinical need for improved and accurate prediction of clinical outcomes following surgery. We investigated patient demographic and cardiovascular magnetic resonance (CMR) characteristics to formulate a dedicated risk score estimating long-term survival following surgery. We recruited consecutive patients undergoing CMR with gadolinium administration prior to surgical aortic valve replacement from 2003 to 2016 in two UK centres. The outcome was overall mortality. A total of 250 patients were included (68 ± 12 years, male 185 (60%), with pre-operative mean aortic valve area 0.93 ± 0.32cm², LVEF 62 ± 17%) and followed for 6.0 ± 3.3 years. Sixty-one deaths occurred, with 10-year mortality of 23.6%. Multivariable analysis showed that increasing age (HR 1.04, P = 0.005), use of antiplatelet therapy (HR 0.54, P = 0.027), presence of infarction or midwall late gadolinium enhancement (HR 1.52 and HR 2.14 respectively, combined P = 0.12), higher indexed left ventricular stroke volume (HR 0.98, P = 0.043) and higher left atrial ejection fraction (HR 0.98, P = 0.083) associated with mortality and developed a risk score with good discrimination. This is the first dedicated risk prediction score for patients with aortic stenosis undergoing surgical aortic valve replacement providing an individualised estimate for overall mortality. This model can help clinicians individualising medical and surgical care.

Trial Registration ClinicalTrials.gov Identifier: NCT00930735 and ClinicalTrials.gov Identifier: NCT01755936.

Prediction of post-traumatic neuropathy following impacted mandibular third molar removal

OBJECTIVES

The extraction of impacted mandibular third molars is a common surgical procedure often associated with complications including post-traumatic neuropathy. Previous work has focused on identifying confounding factors, but a robust preoperative risk prediction model remains elusive.

METHODS

Using a dataset of 648 patients and 812 impacted mandibular third molars, we used least absolute shrinkage and selection operator (LASSO) to fit prediction models based on risk factors assessed at both the tooth and patient levels. In addition, we fitted multivariable logistic regression models with the Firth correction for generalized estimating equations (GEE).

RESULTS

The LASSO model for post-traumatic neuropathy identified distoangular impaction of ≥ 45° (odds ratio [OR] = 2.9), proximity to the inferior alveolar nerve of ≤ 3 mm (OR = 1.9), disadvantageous curving (OR = 1.4), and psychiatric conditions (OR = 2.1) as predictors [area under the receiving operator characteristic curve (AUC) = 0.75]. Among other complications analyzed, the LASSO model for bleeding identified deep embedding or full impaction (OR = 1.8), psychiatric conditions (OR = 1.3), and age (OR = 0.9) as predictors (AUC = 0.64). These associations between predictors and postoperative complications were fundamentally reinforced by the corresponding GEE models.

CONCLUSIONS

Our findings point to the predictability of post-traumatic neuropathy and bleeding based on tooth anatomy and patient characteristics, overall suggesting that preoperatively identifiable factors can predict the risk of adverse outcomes in the extraction of impacted mandibular third molars.

CLINICAL SIGNIFICANCE

Mandibular third molar extraction is both a routine procedure and a leading cause of trigeminal neuropathy. Prevention of post-traumatic neuropathy, aided by individualized preoperative risk prediction, is of high clinical relevance.

Expression Profile and Prognostic Value of Wnt Signaling Pathway Molecules in Colorectal Cancer

Colorectal cancer (CRC) is a heterogeneous disease with changes in the genetic and epigenetic levels of various genes. The molecular assessment of CRC is gaining increasing attention, and furthermore, there is an increase in biomarker use for disease prognostication. Therefore, the identification of different gene biomarkers through messenger RNA (mRNA) abundance levels may be useful for capturing the complex effects of CRC. In this study, we demonstrate that the high mRNA levels of 10 upregulated genes (DPEP1, KRT80, FABP6, NKD2, FOXQ1, CEMIP, ETV4, TESC, FUT1, and GAS2) are observed in CRC cell lines and public CRC datasets. Moreover, we find that a high mRNA expression of DPEP1, NKD2, CEMIP, ETV4, TESC, or FUT1 is significantly correlated with a worse prognosis in CRC patients. Further investigation reveals that CTNNB1 is the key factor in the interaction of the canonical Wnt signaling pathway with 10 upregulated CRC-associated genes. In particular, we identify NKD2, FOXQ1, and CEMIP as three CTNNB1-regulated genes. Moreover, individual inhibition of the expression of three CTNNB1-regulated genes can cause the growth inhibition of CRC cells. This study reveals efficient biomarkers for the prognosis of CRC and provides a new molecular interaction network for CRC.

Prediction of acute kidney injury after cardiac surgery from preoperative N-terminal pro-B-type natriuretic peptide

BACKGROUND

Acute kidney injury (AKI) is common after cardiac surgery and is difficult to predict. N-terminal pro-B-type natriuretic peptide (NT-proBNP) is highly predictive for perioperative cardiovascular complications and may also predict renal injury. We therefore tested the hypothesis that preoperative NT-proBNP concentration is associated with renal injury after major cardiac surgery.

METHODS

We included 35 337 patients who had cardiac surgery and measurements of preoperative NT-proBNP and postoperative creatinine. The primary outcome was Kidney Disease: Improving Global Outcomes Stages 1-3 AKI. We also separately considered severe AKI, including Stage 2, Stage 3, and new-onset dialysis.

RESULTS

Postoperative AKI occurred in 11 999 (34.0%) patients. Stage 2 AKI occurred in 1200 (3.4%) patients, Stage 3 AKI in 474 (1.3%) patients, and new-onset dialysis was required in 241 (0.7%) patients. The NT-proBNP concentrations (considered continuously or in quartiles) were significantly correlated with any-stage AKI and severe AKI (all adjusted P<0.01). Including NT-proBNP significantly improved AKI prediction (net reclassification improvement: 0.24 [0.22-0.27]; P<0.001) beyond basic models derived from other baseline factors in the overall population. Reclassification was especially improved for higher grades of renal injury: 0.30 (0.25-0.36) for Stage 2, 0.46 (0.37-0.55) for Stage 3, and 0.47 (0.35-0.60) for dialysis.

CONCLUSIONS

Increased preoperative NT-proBNP concentrations were associated with postoperative AKI in patients having cardiac surgery. Including NT-proBNP substantially improves AKI predictions based on other preoperative factors.

The effects of cancer therapies on physical fitness before oesophagogastric cancer surgery: a prospective, blinded, multi-centre, observational, cohort study

Background

Neoadjuvant cancer treatment is associated with improved survival following major oesophagogastric cancer surgery. The impact of neoadjuvant chemo/chemoradiotherapy on physical fitness and operative outcomes is however unclear. This study aims to investigate the impact of neoadjuvant chemo/chemoradiotherapy on fitness and post-operative mortality.

Methods

Patients with oesophagogastric cancer scheduled for chemo/chemoradiotherapy and surgery were recruited to a prospective, blinded, multi-centre, observational cohort study. Primary outcomes were changes in fitness with chemo/chemoradiotherapy, measured using cardiopulmonary exercise testing and its association with mortality one-year after surgery. Patients were followed up for re-admission at 30-days, in-hospital morbidity and quality of life (exploratory outcomes).

Results

In total, 384 patients were screened, 217 met the inclusion criteria, 160 consented and 159 were included (72% male, mean age 65 years). A total of 132 patients (83%) underwent chemo/chemoradiotherapy, 109 (71%) underwent chemo/chemoradiotherapy and two exercise tests, 100 (63%) completed surgery and follow-up. A significant decline in oxygen uptake at anaerobic threshold and oxygen uptake peak was observed following chemo/chemoradiotherapy: -1.25ml.kg -1.min -1 (-1.80 to -0.69) and -3.02ml.kg -1.min -1 (-3.85 to -2.20); p<0.0001). Baseline chemo/chemoradiotherapy anaerobic threshold and peak were associated with one-year mortality (HR=0.72, 95%CI 0.59 to 0.88; p=0.001 and HR=0.85, 0.76 to 0.95; p=0.005). The change in physical fitness was not associated with one-year mortality.

Conclusions

Chemo/chemoradiotherapy prior to oesophagogastric cancer surgery reduced physical fitness. Lower baseline fitness was associated with reduced overall survival at one-year. Careful consideration of fitness prior to chemo/chemoradiotherapy and surgery is urgently needed.

Penalization and shrinkage methods produced unreliable clinical prediction models especially when sample size was small

OBJECTIVES

When developing a clinical prediction model, penalization techniques are recommended to address overfitting, as they shrink predictor effect estimates toward the null and reduce mean-square prediction error in new individuals. However, shrinkage and penalty terms ('tuning parameters') are estimated with uncertainty from the development data set. We examined the magnitude of this uncertainty and the subsequent impact on prediction model performance.

STUDY DESIGN AND SETTING

This study comprises applied examples and a simulation study of the following methods: uniform shrinkage (estimated via a closed-form solution or bootstrapping), ridge regression, the lasso, and elastic net.

RESULTS

In a particular model development data set, penalization methods can be unreliable because tuning parameters are estimated with large uncertainty. This is of most concern when development data sets have a small effective sample size and the model's Cox-Snell R2 is low. The problem can lead to considerable miscalibration of model predictions in new individuals.

CONCLUSION

Penalization methods are not a 'carte blanche'; they do not guarantee a reliable prediction model is developed. They are more unreliable when needed most (i.e., when overfitting may be large). We recommend they are best applied with large effective sample sizes, as identified from recent sample size calculations that aim to minimize the potential for model overfitting and precisely estimate key parameters.

Regression of posterior uveal melanoma following iodine-125 plaque radiotherapy based on pre-treatment tumor apical height

PURPOSE

The aim of this study was to analyze regression rates and local control of uveal melanoma patients treated with iodine-125 ( 125I) brachytherapy based on initial tumor apical height.

MATERIAL AND METHODS

Patients treated in a single institution from January 1st, 1996 to 2019 with 125I plaques (ROPES and COMS) for uveal melanoma were included in this study. Patients treated with brachytherapy for iris and those treated with transpupillary thermotherapy prior to brachytherapy were excluded. The sample was classified into 4 categories depending on initial apical tumor height (h0), i.e., h0 ≤ 2.5 (small), 2.5 < h0 ≤ 6.25 (small-medium), 6.25 < h0 ≤ 10 (medium-large), and h0 > 10 mm (large). Percentage of original tumor apical height (Δh) was collected during follow-ups. Patterns of regression were evaluated using linear least squares adjustments. Multivariable Cox regression were performed.

RESULTS

In total, 305 patients met the inclusion criteria, and 27, 166, 100, and 13 were considered for small, small-medium, medium-large, and large categories, respectively. Median follow-up was 82.4, 56.8, 76.1, 89.1, and 100.1 months for the entire cohort and each sub-group, respectively. Pattern of decrease when h0 ≤ 2.5 mm was not detectable. For the rest sub-groups, changes in height could be fitted using functional form: Δh (T) = ae-bT + c, R 2 ≥ 0.97. Multivariate Cox analysis factors predictive of local control failure revealed a hazard ratio (HR) of 6.1 (95% CI: 0.7-58.2%, p = 0.05) for patients who remained similar sized after treatment for small-medium tumors. For the rest sub-groups, Cox analysis did not indicate statistical significance in any single variable.

CONCLUSIONS

Height changes can be modeled by a negative exponential function for the first 7 years after treatment depending on the initial height, except for those less than 2.5 mm. Non-responding small-medium tumors multiply by 6 the probability of failure in local control.

Predicting reoperation after operative treatment of proximal humerus fractures

PURPOSE

The current understanding of the factors associated with a second surgery or loss of alignment after operative treatment of a proximal humerus fracture has relied on small sample studies with stepwise regression analysis. In this study, we used a powerful regression analysis over a large sample and with many variables to test the null hypothesis that there are no factors associated with a revision surgery or loss of alignment after operative treatment of proximal humerus fractures.

METHODS

A retrospective review of all surgically treated proximal humerus fractures from January 1, 2000, to December 31, 2015, was performed at a tertiary level hospital. We extracted longitudinal medical records for all patients, and the data were organized into two categories of predictors: fracture/operative characteristics and patient characteristics.

RESULTS

During the study period, 423 patients met the inclusion criteria. Three hundred and fourteen of the fractures underwent Open Reduction Internal Fixation (ORIF) and 109 underwent Hemiarthroplasty. Thirty-three patients underwent revision surgery (8%). Seventy-nine patients treated with ORIF had loss of alignment (25%). Across the entire cohort, the least absolute shrinkage selection operator (LASSO) analysis found that patients between 40 and 60 years of age had a higher odds of revision surgery (OR = 1.6). In patients treated with ORIF, the LASSO regression found an unreduced calcar to be the strongest predictor of loss of alignment (OR = 5.5), followed by osteoporosis (OR = 1.3), prior radiation treatment (OR = 1.3), unreduced greater tuberosity (OR = 1.2) and age over 80 years (OR = 1.2).

CONCLUSION

Reoperation after proximal humerus surgery is infrequent even though loss of alignment is common. In our cohort, not all patients who had a loss of alignment underwent revision surgery; consequently, obtaining the best possible reduction at the index surgery is paramount.

A Risk Prediction Model for Mortality Among Smokers in the COPDGene® Study

BACKGROUND

Risk factor identification is a proven strategy in advancing treatments and preventive therapy for many chronic conditions. Quantifying the impact of those risk factors on health outcomes can consolidate and focus efforts on individuals with specific high-risk profiles. Using multiple risk factors and longitudinal outcomes in 2 independent cohorts, we developed and validated a risk score model to predict mortality in current and former cigarette smokers.

METHODS

We obtained extensive data on current and former smokers from the COPD Genetic Epidemiology (COPDGene®) study at enrollment. Based on physician input and model goodness-of-fit measures, a subset of variables was selected to fit final Weibull survival models separately for men and women. Coefficients and predictors were translated into a point system, allowing for easy computation of mortality risk scores and probabilities. We then used the SubPopulations and InteRmediate Outcome Measures In COPD Study (SPIROMICS) cohort for external validation of our model.

RESULTS

Of 9867 COPDGene participants with standard baseline data, 17.6% died over 10 years of follow-up, and 9074 of these participants had the full set of baseline predictors (standard plus 6-minute walk distance and computed tomography variables) available for full model fits. The average age of participants in the cohort was 60 for both men and women, and the average predicted 10-year mortality risk was 18% for women and 25% for men. Model time-integrated area under the receiver operating characteristic curve statistics demonstrated good predictive model accuracy (0.797 average), validated in the external cohort (0.756 average). Risk of mortality was impacted most by 6-minute walk distance, forced expiratory volume in 1 second and age, for both men and women.

CONCLUSIONS

Current and former smokers exhibited a wide range of mortality risk over a 10- year period. Our models can identify higher risk individuals who can be targeted for interventions to reduce risk of mortality, for participants with or without chronic obstructive pulmonary disease (COPD) using current Global initiative for obstructive Lung Disease (GOLD) criteria.

Risk factors for child abuse, neglect and exposure to intimate partner violence in early childhood: Findings in a representative cross-sectional sample in Germany

BACKGROUND

The KiD 0-3 national main study is a cross-sectional study on adversity in early childhood and parental access to support services, conducted as part of a long-term policy program for early intervention services in Germany.

OBJECTIVE

To identify risk factors for child abuse, neglect and exposure to intimate partner violence (IPV) and investigate if parental use of early intervention programs or contact to child welfare services was associated with reported child maltreatment.

PARTICIPANTS AND SETTING

8063 families with infants and toddlers participated in the survey. Parents answered a written questionnaire during mandatory health checks for the child. The sampling was based on a regionally clustered model of pediatricians' practices.

METHODS

An automatic variable selection process was used to test risk factors and logistic regression models were employed for each outcome.

RESULTS

Significant risk factors (p < 0.05) for child abuse (1.91 %) were child age, IPV and parental stress. Neglect (0.83 %) was associated with couple distress, adverse childhood experiences, young maternal age, cramped housing, and migration history. IPV (2.98 %) was associated with child age, couple distress, depression/anxiety, harsh punishment, adverse childhood experiences, young maternal age, and poverty. Parents were more likely to use selective prevention programs in cases of child abuse and exposure to IPV.

CONCLUSION

Child abuse is mainly associated with proximal risk factors and neglect with distal factors. Exposure to IPV violence is associated with child abuse as well as with an accumulation of adversities. The association between service use and child maltreatment is discussed.

An advanced prediction model for postoperative complications and early implant failure

OBJECTIVES

Risk prediction in implant dentistry presents specific challenges including the dependence of observations from patients with multiple implants and rare outcome events. The aim of this study was to use advanced statistical methods based on penalized regression to assess risk factors in implant dentistry.

MATERIAL AND METHODS

We conducted a retrospective study from January 2016 to November 2018 recording postoperative complications including bleeding, hematoma, local infection, and nerve damage, as well as early implant failure. We further assessed patient- and implant-related risk factors including smoking and diabetes, as well as treatment parameters including types of gaps and surgical procedures. Univariable and multivariable generalized estimating equation (GEE) models were estimated to assess predictor effects, and a prediction model was fitted using L1 penalized estimation (lasso).

RESULTS

In a total of 1,132 patients (mean age: 50.6 ± 16.5 years, 55.4% female) and 2,413 implants, postoperative complications occurred in 71 patients. Sixteen implants were lost prior to loading. Multivariable GEE models showed a higher risk of any complication for diabetes mellitus (p = .006) and bone augmentation (p = .039). The models further revealed a higher risk of local infection for bone augmentation (p = .003), and a higher risk of hematoma formation for diabetes mellitus (p = .007) and edentulous jaws (p = .024). The lasso model did not select any risk factors into the prediction model.

CONCLUSIONS

Using novel methodology well-suited to tackle the specific challenges of risk prediction in implant dentistry, we were able to reliably estimate associations of risk factors with outcomes.

Glucose-6-phosphate dehydrogenase deficiency and stroke outcomes

OBJECTIVE

To assess the risk of glucose-6-phosphate dehydrogenase (G6PD) on stroke prognosis, we compared outcomes between patients with stroke with and without G6PD deficiency.

METHODS

The study recruited 1,251 patients with acute ischemic stroke. Patients were individually categorized into G6PD-deficiency and non-G6PD-deficiency groups according to G6PD activity upon admission. The primary endpoint was poor outcome at 3 months defined by a modified Rankin Scale (mRS) score ≥2 (including disability and death). Secondary outcomes included the overall mRS score at 3 months and in-hospital death and all death within 3 months. Logistic regression and Cox models, adjusted for potential confounders, were fitted to estimate the association of G6PD deficiency with the outcomes.

RESULTS

Among 1,251 patients, 150 (12.0%) were G6PD-deficient. Patients with G6PD deficiency had higher proportions of large-artery atherosclerosis (odds ratio [OR] 1.53, 95% confidence interval [CI] 1.09-2.17) and stroke history (OR 1.93, 95% CI 1.26-2.90) compared to the non-G6PD-deficient group. The 2 groups differed significantly in the overall mRS score distribution (adjusted common OR 1.57, 95% CI 1.14-2.17). Patients with G6PD deficiency had higher rates of poor outcome at 3 months (adjusted OR 1.73, 95% CI 1.08-2.76; adjusted absolute risk increase 13.0%, 95% CI 2.4%-23.6%). The hazard ratio of in-hospital death for patients with G6PD-deficiency was 1.46 (95% CI 1.37-1.84).

CONCLUSIONS

G6PD deficiency is associated with the risk of poor outcome at 3 months after ischemic stroke and may increase the risk of in-hospital death. These findings suggest the rationality of G6PD screening in patients with stroke.

Regression shrinkage methods for clinical prediction models do not guarantee improved performance: Simulation study

When developing risk prediction models on datasets with limited sample size, shrinkage methods are recommended. Earlier studies showed that shrinkage results in better predictive performance on average. This simulation study aimed to investigate the variability of regression shrinkage on predictive performance for a binary outcome. We compared standard maximum likelihood with the following shrinkage methods: uniform shrinkage (likelihood-based and bootstrap-based), penalized maximum likelihood (ridge) methods, LASSO logistic regression, adaptive LASSO, and Firth's correction. In the simulation study, we varied the number of predictors and their strength, the correlation between predictors, the event rate of the outcome, and the events per variable. In terms of results, we focused on the calibration slope. The slope indicates whether risk predictions are too extreme (slope < 1) or not extreme enough (slope > 1). The results can be summarized into three main findings. First, shrinkage improved calibration slopes on average. Second, the between-sample variability of calibration slopes was often increased relative to maximum likelihood. In contrast to other shrinkage approaches, Firth's correction had a small shrinkage effect but showed low variability. Third, the correlation between the estimated shrinkage and the optimal shrinkage to remove overfitting was typically negative, with Firth's correction as the exception. We conclude that, despite improved performance on average, shrinkage often worked poorly in individual datasets, in particular when it was most needed. The results imply that shrinkage methods do not solve problems associated with small sample size or low number of events per variable.

Pulse Wave Velocity and Machine Learning to Predict Cardiovascular Outcomes in Prediabetic and Diabetic Populations

Few studies have addressed the predictive value of arterial stiffness determined by pulse wave velocity (PWV) in a high-risk population with no prevalent cardiovascular disease and with obesity, hypertension, hyperglycemia, and preserved kidney function. This longitudinal, retrospective study enrolled 88 high-risk patients and had a follow-up time of 12.4 years. We collected clinical and laboratory data, as well as information on arterial stiffness parameters using arterial tonometry and measurements from ambulatory blood pressure monitoring. We considered nonfatal, incident cardiovascular events as the primary outcome. Given the small size of our dataset, we used survival analysis (i.e., Cox proportional hazards model) combined with a machine learning-based algorithm/penalization method to evaluate the data. Our predictive model, calculated with Cox regression and least absolute shrinkage and selection operator (LASSO), included body mass index, diabetes mellitus, gender (male), and PWV. We recorded 16 nonfatal cardiovascular events (5 myocardial infarctions, 5 episodes of heart failure, and 6 strokes). The adjusted hazard ratio for PWV was 1.199 (95% confidence interval: 1.09-1.37, p < 0.001). Arterial stiffness was a predictor of cardiovascular disease development, as determined by PWV in a high-risk population. Thus, in obese, hypertensive, hyperglycemic patients with preserved kidney function, PWV can serve as a prognostic factor for major adverse cardiac events.

Methodological standards for the development and evaluation of clinical prediction rules: a review of the literature

Clinical prediction rules (CPRs) that predict the absolute risk of a clinical condition or future outcome for individual patients are abundant in the medical literature; however, systematic reviews have demonstrated shortcomings in the methodological quality and reporting of prediction studies. To maximise the potential and clinical usefulness of CPRs, they must be rigorously developed and validated, and their impact on clinical practice and patient outcomes must be evaluated. This review aims to present a comprehensive overview of the stages involved in the development, validation and evaluation of CPRs, and to describe in detail the methodological standards required at each stage, illustrated with examples where appropriate. Important features of the study design, statistical analysis, modelling strategy, data collection, performance assessment, CPR presentation and reporting are discussed, in addition to other, often overlooked aspects such as the acceptability, cost-effectiveness and longer-term implementation of CPRs, and their comparison with clinical judgement. Although the development and evaluation of a robust, clinically useful CPR is anything but straightforward, adherence to the plethora of methodological standards, recommendations and frameworks at each stage will assist in the development of a rigorous CPR that has the potential to contribute usefully to clinical practice and decision-making and have a positive impact on patient care.

The Expression Profile and Prognostic Significance of Metallothionein Genes in Colorectal Cancer

Colorectal cancer (CRC) is a heterogeneous disease resulting from the combined influence of many genetic factors. This complexity has caused the molecular characterization of CRC to remain uncharacterized, with a lack of clear gene markers associated with CRC and the prognosis of this disease. Thus, highly sensitive tumor markers for the detection of CRC are the most essential determinants of survival. In this study, we examined the simultaneous downregulation of the mRNA levels of six metallothionein (MT) genes in CRC cell lines and public CRC datasets for the first time. In addition, we detected downregulation of these six MT mRNAs’ levels in 30 pairs of tumor (T) and adjacent non-tumor (N) CRC specimens. In order to understand the potential prognostic relevance of these six MT genes and CRC, we presented a four-gene signature to evaluate the prognosis of CRC patients. Further discovery suggested that the four-gene signature (MT1F, MT1G, MT1L, and MT1X) predicted survival better than any combination of two-, three-, four-, five-, or six-gene models. In conclusion, this study is the first to report that simultaneous downregulation of six MT mRNAs’ levels in CRC patients, and their aberrant expression together, accurately predicted CRC patients’ outcomes.

Identification of clinically relevant features in hypertensive patients using penalized regression: a case study of cardiovascular events

Appropriate management of hypertensive patients relies on the accurate identification of clinically relevant features. However, traditional statistical methods may ignore important information in datasets or overlook possible interactions among features. Machine learning may improve the prediction accuracy and interpretability of regression models by identifying the most relevant features in hypertensive patients. We sought the most relevant features for prediction of cardiovascular (CV) events in a hypertensive population. We used the penalized regression models least absolute shrinkage and selection operator (LASSO) and elastic net (EN) to obtain the most parsimonious and accurate models. The clinical parameters and laboratory biomarkers were collected from the clinical records of 1,471 patients receiving care at Mostoles University Hospital. The outcome was the development of major adverse CV events. Cox proportional hazards regression was performed alone and with penalized regression analyses (LASSO and EN), producing three models. The modeling was performed using 10-fold cross-validation to fit the penalized models. The three predictive models were compared and statistically analyzed to assess their classification accuracy, sensitivity, specificity, discriminative power, and calibration accuracy. The standard Cox model identified five relevant features, while LASSO and EN identified only three (age, LDL cholesterol, and kidney function). The accuracies of the models (prediction vs. observation) were 0.767 (Cox model), 0.754 (LASSO), and 0.764 (EN), and the areas under the curve were 0.694, 0.670, and 0.673, respectively. However, pairwise comparison of performance yielded no statistically significant differences. All three calibration curves showed close agreement between the predicted and observed probabilities of the development of a CV event. Although the performance was similar for all three models, both penalized regression analyses produced models with good fit and fewer features than the Cox regression predictive model but with the same accuracy. This case study of predictive models using penalized regression analyses shows that penalized regularization techniques can provide predictive models for CV risk assessment that are parsimonious, highly interpretable, and generalizable and that have good fit. For clinicians, a parsimonious model can be useful where available data are limited, as such a model can offer a simple but efficient way to model the impact of the different features on the prediction of CV events. Management of these features may lower the risk for a CV event. Graphical Abstract In a clinical setting, with numerous biological and laboratory features and incomplete datasets, traditional statistical methods may ignore important information and overlook possible interactions among features. Our aim was to identify the most relevant features to predict cardiovascular events in a hypertensive population, using three different regression approaches for feature selection, to improve the prediction accuracy and interpretability of regression models by identifying the relevant features in these patients.

Glycidamide Promotes the Growth and Migratory Ability of Prostate Cancer Cells by Changing the Protein Expression of Cell Cycle Regulators and Epithelial-to-Mesenchymal Transition (EMT)-Associated Proteins with Prognostic Relevance

Acrylamide (AA) and glycidamide (GA) can be produced in carbohydrate-rich food when heated at a high temperature, which can induce a malignant transformation. It has been demonstrated that GA is more mutagenic than AA. It has been shown that the proliferation rate of some cancer cells are increased by treatment with GA; however, the exact genes that are induced by GA in most cancer cells are not clear. In the present study, we demonstrated that GA promotes the growth of prostate cancer cells through induced protein expression of the cell cycle regulator. In addition, we also found that GA promoted the migratory ability of prostate cancer cells through induced epithelial-to-mesenchymal transition (EMT)-associated protein expression. In order to understand the potential prognostic relevance of GA-mediated regulators of the cell cycle and EMT, we present a three-gene signature to evaluate the prognosis of prostate cancer patients. Further investigations suggested that the three-gene signature (CDK4, TWIST1 and SNAI2) predicted the chances of survival better than any of the three genes alone for the first time. In conclusion, we suggested that the three-gene signature model can act as marker of GA exposure. Hence, this multi-gene panel may serve as a promising outcome predictor and potential therapeutic target in prostate cancer patients.

Clinical and prognostic impact of chronotropic incompetence in patients with hypertrophic cardiomyopathy

BACKGROUND

A blunted heart rate (HR) response is associated with an impaired peak oxygen uptake (pVO₂), a powerful outcome predictor in hypertrophic cardiomyopathy (HCM). The present multicenter study sought to determine the prognostic role for exercise-induced HR response in HCM.

METHODS

A total of 681 consecutive HCM outpatients on optimized treatment were recruited. The heart failure (HF) end-point was death due to HF, cardiac transplantation, NYHA III-IV class progression, HF worsening leading to hospitalization and severe functional deterioration leading to septal reduction. The sudden cardiac death (SCD) end-point included SCD, aborted SCD and appropriate implantable cardioverter defibrillator discharges.

RESULTS

During a median follow-up of 4.2 years (25-75th centile: 3.9-5.2), 81 patients reached the HF and 23 the SCD end-point. Covariates with independent effects on the HF end-point were left atrial diameter, left ventricular ejection fraction, maximal left ventricular outflow tract gradient and exercise cardiac power (ECP = pVO₂∗systolic blood pressure) (C-Index = 0.807) whereas the HCM Risk-SCD score and the ECP remained associated with the SCD end-point (C-Index = 0.674). When the VO₂-derived variables were not pursued, peak HR (pHR) re-entered in the multivariate HF model (C-Index = 0.777) and, marginally, in the SCD model (C-index = 0.656). A pHR = 70% of the maximum predicted resulted as the best cut-off value in predicting the HF-related events.

CONCLUSIONS

The cardiopulmonary exercise test is pivotal in the HCM management, however the pHR remains a meaningful alternative parameter. A pHR < 70% identified a HCM population at high risk of HF-related events, thus calling for a reappraisal of the chronotropic incompetence threshold in HCM.

Radiobiological doses, tumor, and treatment features influence on outcomes after epiescleral brachytherapy. A 20-year retrospective analysis from a single-institution: part II

PURPOSE

To assess the influence of the radiobiological doses, tumor, and treatment features on retinopathy, cataracts, retinal detachment, optic neuropathy, vitreous hemorrhage, and neovascular glaucoma at the authors' institution after brachytherapy for posterior uveal melanoma.

MATERIAL AND METHODS

Medical records of 243 eyes with uveal melanoma, treated by iodine brachytherapy between 1996 and 2016 at a single center were analyzed. Clinical and radiotherapy data were extracted from a dedicated database. Biologically effective dose (BED) was included in survival analysis performed using Kaplan-Meier and Cox regressions. Relative survival rates were estimated, and univariate/multivariate regression models were constructed for predictive factors of each item. Hazard ratio and confidence interval at 95% were determined. Variables statistically significant were analyzed and compared by log-rank tests.

RESULTS

The median follow-up was 73.9 months (range, 3-202 months). Cumulative probabilities of survival by Kaplan-Meier analysis at 3 and 5 years, respectively, were: 59% and 48% for retinopathy; 71% and 55% for cataracts; 63% and 57% for retinal detachment; 88% and 79% for optic neuropathy; 87% and 83% for vitreous hemorrhage; 92% and 89% for neovascular glaucoma, respectively. Using multivariate analysis, statistically significant risk factors were: age, tumor apical height, dose to foveola, and location of anterior border for retinopathy; age, dose to lens, type of plaque, and tumor shape, for cataracts; age, tumor apical height, and size of the plaque for retinal detachment; age, plaque shape, longest basal dimension, and BED to optic nerve for optic neuropathy; age, tumor apical height, and tumor shape for vitreous hemorrhage; tumor apical height and BED to foveola for neovascular glaucoma.

CONCLUSIONS

Tumor factors in addition to radiation treatment may contribute to secondary effects. Enhanced clinical optimization should evaluate radiobiological doses delivered to the tumor volume and surrounding normal ocular structures.

Radiobiological doses, tumor, and treatment features influence on local control, enucleation rates, and survival after epiescleral brachytherapy. A 20-year retrospective analysis from a single-institution: part I

PURPOSE

To assess influence of the radiobiological doses, tumor, and treatment features on local control, enucleation rates, overall and disease-specific survival rates after brachytherapy for posterior uveal melanoma.

MATERIAL AND METHODS

Local control, enucleation, overall and disease-specific survival rates were evaluated on the base of 243 patients from 1996 through 2016, using plaques loaded with iodine sources. Clinical and radiotherapy data were extracted from a dedicated prospective database. Biologically effective dose (BED) was included in survival analysis using Kaplan-Meier and Cox regressions. The 3-, 5-, 10-, and 15-year relative survival rates were estimated, and univariate/multivariate regression models were constructed for predictive factors of each item. Hazard ratio (HR) and confidence interval at 95% (CI) were determined.

RESULTS

The median follow-up was 73.9 months (range, 3-202 months). Cumulative probabilities of survival by Kaplan-Meier analysis at 3, 5, 10 and 15 years were respectively: 96%, 94%, 93%, and 87%, for local control; 93%, 88%, 81%, and 73% for globe preservation; 98%, 93%, 84%, and 73% for overall survival, and 98%, 96%, 92%, and 87% for disease-specific survival. By multivariate analysis, we concluded variables as significant: for local control failure - the longest basal diameter and the juxtapapillary location; for globe preservation failure - the longest basal dimension, the mushroom shape, the location in ciliary body, and the dose to the foveola; for disease-specific survival - the longest basal dimension. Some radiobiological doses were significant in univariate models but not in multivariate ones for the items studied.

CONCLUSIONS

The results show as predictive factors of local control, enucleation, and disease-specific survival rates those related with the features of the tumor, specifically the longest basal dimension. There is no clear relation between radiobiological doses or treatment parameters in patients after brachytherapy.

Impact of cholinesterase inhibitors or memantine on survival in adults with Down syndrome and dementia: clinical cohort study

BACKGROUND

There is little evidence to guide pharmacological treatment in adults with Down syndrome and Alzheimer's disease. Aims To investigate the effect of cholinesterase inhibitors or memantine on survival and function in adults with Down syndrome and Alzheimer's disease.

METHOD

This was a naturalistic longitudinal follow-up of a clinical cohort of 310 people with Down syndrome diagnosed with Alzheimer's disease collected from specialist community services in England.

RESULTS

Median survival time (5.59 years, 95% CI 4.67-6.67) for those on medication (n = 145, mainly cholinesterase inhibitors) was significantly greater than for those not prescribed medication (n = 165) (3.45 years, 95% CI 2.91-4.13, log-rank test P<0.001). Sequential assessments demonstrated an early effect in maintaining cognitive function.

CONCLUSIONS

Cholinesterase inhibitors appear to offer benefit for people with Down syndrome and Alzheimer's disease that is comparable with sporadic Alzheimer's disease; a trial to test the effect of earlier treatment (prodromal Alzheimer's disease) in Down syndrome may be indicated. Declaration of interest A.S. has undertaken consulting for Ono Pharmaceuticals, outside the submitted work. Z.W. has received a consultancy fee and grant from GE Healthcare, outside the submitted work.

Development and Validation of a Predictive Radiomics Model for Clinical Outcomes in Stage I Non-small Cell Lung Cancer

PURPOSE

To develop and validate a radiomics signature that can predict the clinical outcomes for patients with stage I non-small cell lung cancer (NSCLC).

METHODS AND MATERIALS

We retrospectively analyzed contrast-enhanced computed tomography images of patients from a training cohort (n = 147) treated with surgery and an independent validation cohort (n = 295) treated with stereotactic ablative radiation therapy. Twelve radiomics features with established strategies for filtering and preprocessing were extracted. The random survival forests (RSF) method was used to build models from subsets of the 12 candidate features based on their survival relevance and generate a mortality risk index for each observation in the training set. An optimal model was selected, and its ability to predict clinical outcomes was evaluated in the validation set using predicted mortality risk indexes.

RESULTS

The optimal RSF model, consisting of 2 predictive features, kurtosis and the gray level co-occurrence matrix feature homogeneity2, allowed for significant risk stratification (log-rank P < .0001) and remained an independent predictor of overall survival after adjusting for age, tumor volume and histologic type, and Karnofsky performance status (hazard ratio [HR] 1.27; P < 2e-16) in the training set. The resultant mortality risk indexes were significantly associated with overall survival in the validation set (log-rank P = .0173; HR 1.02, P = .0438). They were also significant for distant metastasis (log-rank P < .05; HR 1.04, P = .0407) and were borderline significant for regional recurrence on univariate analysis (log-rank P < .05; HR 1.04, P = .0617).

CONCLUSIONS

Our radiomics model accurately predicted several clinical outcomes and allowed pretreatment risk stratification in stage I NSCLC, allowing the choice of treatment to be tailored to each patient's individual risk profile.

Feature selection through validation and un-censoring of endovascular repair survival data for predicting the risk of re-intervention

Background

Feature selection (FS) process is essential in the medical area as it reduces the effort and time needed for physicians to measure unnecessary features. Choosing useful variables is a difficult task with the presence of censoring which is the unique characteristic in survival analysis. Most survival FS methods depend on Cox’s proportional hazard model; however, machine learning techniques (MLT) are preferred but not commonly used due to censoring. Techniques that have been proposed to adopt MLT to perform FS with survival data cannot be used with the high level of censoring. The researcher’s previous publications proposed a technique to deal with the high level of censoring. It also used existing FS techniques to reduce dataset dimension. However, in this paper a new FS technique was proposed and combined with feature transformation and the proposed uncensoring approaches to select a reduced set of features and produce a stable predictive model.

Methods

In this paper, a FS technique based on artificial neural network (ANN) MLT is proposed to deal with highly censored Endovascular Aortic Repair (EVAR). Survival data EVAR datasets were collected during 2004 to 2010 from two vascular centers in order to produce a final stable model. They contain almost 91% of censored patients. The proposed approach used a wrapper FS method with ANN to select a reduced subset of features that predict the risk of EVAR re-intervention after 5 years to patients from two different centers located in the United Kingdom, to allow it to be potentially applied to cross-centers predictions. The proposed model is compared with the two popular FS techniques; Akaike and Bayesian information criteria (AIC, BIC) that are used with Cox’s model.

Results

The final model outperforms other methods in distinguishing the high and low risk groups; as they both have concordance index and estimated AUC better than the Cox’s model based on AIC, BIC, Lasso, and SCAD approaches. These models have p-values lower than 0.05, meaning that patients with different risk groups can be separated significantly and those who would need re-intervention can be correctly predicted.

Conclusion

The proposed approach will save time and effort made by physicians to collect unnecessary variables. The final reduced model was able to predict the long-term risk of aortic complications after EVAR. This predictive model can help clinicians decide patients’ future observation plan.

Electronic supplementary material

The online version of this article (doi:10.1186/s12911-017-0508-3) contains supplementary material, which is available to authorized users.

Toll-like receptor (TLR) diversity influences mycobacterial growth in African buffalo

Understanding the role of wildlife in the maintenance or spread of emerging infectious diseases is a growing priority across the world. Bovine tuberculosis (BTB) is a chronic, infectious disease caused by Mycobacterium bovis (M. bovis). BTB is widespread within game reserves in southern Africa, and within these ecosystems the primary wildlife host of this disease is the African buffalo. We used a modified bacterial killing assay for mycobacteria to investigate the effect of Toll-like receptor (TLR) genetic diversity and demographic parameters on the ability of African buffalo to restrict mycobacterial growth. Eosinophil count, time delay, bovine PPD response and avian PPD response were negatively correlated with mycobacterial growth. TLR6 diversity and the interaction of age group and sex were positively correlated with mycobacterial growth. Our results suggest that both demographic and individual immune parameters influence the ability to control mycobacterial infection in African buffalo. TLR6 diversity is particularly interesting as this locus has also shown associations with BTB in cattle, suggesting that further research into the effects, selection and role of TLR6 variants in bovine tuberculosis will be productive.

Performance of Firth-and logF-type penalized methods in risk prediction for small or sparse binary data

BACKGROUND

When developing risk models for binary data with small or sparse data sets, the standard maximum likelihood estimation (MLE) based logistic regression faces several problems including biased or infinite estimate of the regression coefficient and frequent convergence failure of the likelihood due to separation. The problem of separation occurs commonly even if sample size is large but there is sufficient number of strong predictors. In the presence of separation, even if one develops the model, it produces overfitted model with poor predictive performance. Firth-and logF-type penalized regression methods are popular alternative to MLE, particularly for solving separation-problem. Despite the attractive advantages, their use in risk prediction is very limited. This paper evaluated these methods in risk prediction in comparison with MLE and other commonly used penalized methods such as ridge.

METHODS

The predictive performance of the methods was evaluated through assessing calibration, discrimination and overall predictive performance using an extensive simulation study. Further an illustration of the methods were provided using a real data example with low prevalence of outcome.

RESULTS

The MLE showed poor performance in risk prediction in small or sparse data sets. All penalized methods offered some improvements in calibration, discrimination and overall predictive performance. Although the Firth-and logF-type methods showed almost equal amount of improvement, Firth-type penalization produces some bias in the average predicted probability, and the amount of bias is even larger than that produced by MLE. Of the logF(1,1) and logF(2,2) penalization, logF(2,2) provides slight bias in the estimate of regression coefficient of binary predictor and logF(1,1) performed better in all aspects. Similarly, ridge performed well in discrimination and overall predictive performance but it often produces underfitted model and has high rate of convergence failure (even the rate is higher than that for MLE), probably due to the separation problem.

CONCLUSIONS

The logF-type penalized method, particularly logF(1,1) could be used in practice when developing risk model for small or sparse data sets.

A METHYLATION-TO-EXPRESSION FEATURE MODEL FOR GENERATING ACCURATE PROGNOSTIC RISK SCORES AND IDENTIFYING DISEASE TARGETS IN CLEAR CELL KIDNEY CANCER

Many researchers now have available multiple high-dimensional molecular and clinical datasets when studying a disease. As we enter this multi-omic era of data analysis, new approaches that combine different levels of data (e.g. at the genomic and epigenomic levels) are required to fully capitalize on this opportunity. In this work, we outline a new approach to multi-omic data integration, which combines molecular and clinical predictors as part of a single analysis to create a prognostic risk score for clear cell renal cell carcinoma. The approach integrates data in multiple ways and yet creates models that are relatively straightforward to interpret and with a high level of performance. Furthermore, the proposed process of data integration captures relationships in the data that represent highly disease-relevant functions.

Prediction of the Risk of a Metachronous Advanced Colorectal Neoplasm Using a Novel Scoring System

BACKGROUND AND AIM

This study aimed to develop and validate a risk score model to estimate the probability of a metachronous advanced colorectal neoplasm (ACRN) at surveillance colonoscopy.

METHODS

A retrospective analysis of a prospectively obtained database of 11,042 asymptomatic subjects who underwent surveillance colonoscopy after a screening colonoscopy was conducted. Subjects were randomly divided into derivation (n = 7730) and validation sets (n = 3312). From the derivation cohort, risk factors for a metachronous ACRN were identified by a multivariable analysis. Risk points were allocated to each risk factor based on the hazard ratio to develop the Metachronous Advanced colorectal neoplasm Prediction Scoring (MAPS) model, the performance of which was assessed in the validation cohort.

RESULTS

In the derivation cohort, age, male, sessile serrated adenoma/polyp, and a high-risk CRN (ACRN or ≥3 adenomas) at screening colonoscopy were independent risk factors for a metachronous ACRN. These variables were incorporated into the MAPS model, and the risk score ranged 0-17 (high MAPS risk arbitrarily defined as 10-17). At the 3-year surveillance colonoscopy, ACRN was found in 5.1 % of the high MAPS risk group versus 3.9 % of the high-risk CRN group. The colonoscopy number needed to detect one metachronous ACRN at the 3-year surveillance was 19.5 (95 % CI 11.7-33.2) for the high MAPS risk group versus 25.8 (95 % CI 15.4-44.0) for the high-risk CRN group. These findings were similarly confirmed in the validation cohort.

CONCLUSIONS

Our MAPS model based on clinical and colonoscopic parameters effectively predicts the risk of a metachronous ACRN.