RESUMO
BACKGROUND: It is unclear how cumulative multivariable effects of clinically relevant covariates impact response to pharmacological treatments for lower urinary tract symptoms (LUTS)/benign prostatic enlargement (BPE). OBJECTIVE: To develop models to predict treatment response in terms of International Prostate Symptom Score (IPSS) and the risk of acute urinary retention (AUR) or BPE-related surgery, based on large data sets and using as predictors baseline characteristics that commonly define the risk of disease progression. DESIGN, SETTING, AND PARTICIPANTS: A total of 9167 patients with LUTS/BPE at risk of progression in three placebo-controlled dutasteride trials and one comparing dutasteride, tamsulosin, and dutasteride + tamsulosin combination therapy (CT) were included in the analysis to predict response to placebo up to 24 mo and active treatment up to 48 mo. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: Predictors included age, IPSS, total prostate volume (PV), maximum urinary flow rate (Qmax), prostate-specific antigen, postvoid residual urine (PVR), α-blocker usage within 12 mo, and randomised treatment. A generalised least-squares model was developed for longitudinal IPSS and a Cox proportional-hazards model for time to first AUR/surgery. RESULTS AND LIMITATIONS: The vast majority of patients benefit from dutasteride or CT when compared with tamsulosin alone. The predicted IPSS improvement with dutasteride or CT increased with greater PV and severity of symptoms at baseline. The tamsulosin effect was lower with greater baseline PV and tended to decrease over time. Predicted AUR/surgery risk was greater with tamsulosin versus CT or dutasteride; this risk increased with larger PV, higher PVR, and lower Qmax (all at baseline). An educational interactive web-based tool facilitates visualisation of the results (www.bphtool.com). Limitations include: the placebo and active-treatment predictions are from different studies, the lack of similar studies for external validation, and the focus on a population at risk of progression from the 4-yr CombAT study. CONCLUSIONS: Predictive modelling based on large data sets and visualisation of the risk for individual profiles can improve our understanding of how risk factors for disease progression interact and affect response to different treatments, reinforcing the importance of an individualised approach for LUTS/BPE management. PATIENT SUMMARY: We used data from previous studies to develop statistical models for predicting how men with lower urinary tract symptoms or benign prostate enlargement and at risk of disease complications respond to certain treatments according to their individual characteristics.
Assuntos
Sintomas do Trato Urinário Inferior , Hiperplasia Prostática , Retenção Urinária , Masculino , Humanos , Dutasterida/uso terapêutico , Tansulosina/uso terapêutico , Azasteroides/uso terapêutico , Sulfonamidas/uso terapêutico , Resultado do Tratamento , Quimioterapia Combinada , Hiperplasia Prostática/complicações , Hiperplasia Prostática/tratamento farmacológico , Hiperplasia Prostática/cirurgia , Retenção Urinária/complicações , Sintomas do Trato Urinário Inferior/etiologia , Sintomas do Trato Urinário Inferior/complicações , Progressão da DoençaRESUMO
BACKGROUND: Clustering methods are becoming widely utilized in biomedical research where the volume and complexity of data is rapidly increasing. Unsupervised clustering of patient information can reveal distinct phenotype groups with different underlying mechanism, risk prognosis and treatment response. However, biological datasets are usually characterized by a combination of low sample number and very high dimensionality, something that is not adequately addressed by current algorithms. While the performance of the methods is satisfactory for low dimensional data, increasing number of features results in either deterioration of accuracy or inability to cluster. To tackle these challenges, new methodologies designed specifically for such data are needed. RESULTS: We present 2D-EM, a clustering algorithm approach designed for small sample size and high-dimensional datasets. To employ information corresponding to data distribution and facilitate visualization, the sample is folded into its two-dimension (2D) matrix form (or feature matrix). The maximum likelihood estimate is then estimated using a modified expectation-maximization (EM) algorithm. The 2D-EM methodology was benchmarked against several existing clustering methods using 6 medically-relevant transcriptome datasets. The percentage improvement of Rand score and adjusted Rand index compared to the best performing alternative method is up to 21.9% and 155.6%, respectively. To present the general utility of the 2D-EM method we also employed 2 methylome datasets, again showing superior performance relative to established methods. CONCLUSIONS: The 2D-EM algorithm was able to reproduce the groups in transcriptome and methylome data with high accuracy. This build confidence in the methods ability to uncover novel disease subtypes in new datasets. The design of 2D-EM algorithm enables it to handle a diverse set of challenging biomedical dataset and cluster with higher accuracy than established methods. MATLAB implementation of the tool can be freely accessed online ( http://www.riken.jp/en/research/labs/ims/med_sci_math or http://www.alok-ai-lab.com /).