Positive predictive values of Breast Imaging Reporting and Data System (BI-RADS®) categories 3, 4 and 5 in breast lesions submitted to percutaneous biopsy

Objective To evaluate the BI-RADS as a predictive factor of suspicion for malignancy in breast lesions by correlating radiological with histological results and calculating the positive predictive value for categories 3, 4 and 5 in a breast cancer reference center in the city of São Paulo. Materials and Methods Retrospective, analytical and cross-sectional study including 725 patients with mammographic and/or sonographic findings classified as BI-RADS categories 3, 4 and 5 who were referred to the authors' institution to undergo percutaneous biopsy. The tests results were reviewed and the positive predictive value was calculated by means of a specific mathematical equation. Results Positive predictive values found for categories 3, 4 and 5 were respectively the following: 0.74%, 33.08% and 92.95%, for cases submitted to ultrasound-guided biopsy, and 0.00%, 14.90% and 100% for cases submitted to stereotactic biopsy. Conclusion The present study demonstrated high suspicion for malignancy in lesions classified as category 5 and low risk for category 3. As regards category 4, the need for systematic biopsies was observed.

A Preclinical Model for ERα-Positive Breast Cancer Points to the Epithelial Microenvironment as Determinant of Luminal Phenotype and Hormone Response.

Seventy-five percent of breast cancers are estrogen receptor α positive (ER(+)). Research on these tumors is hampered by lack of adequate in vivo models; cell line xenografts require non-physiological hormone supplements, and patient-derived xenografts (PDXs) are hard to establish. We show that the traditional grafting of ER(+) tumor cells into mammary fat pads induces TGFβ/SLUG signaling and basal differentiation when they require low SLUG levels to grow in vivo. Grafting into the milk ducts suppresses SLUG; ER(+) tumor cells develop, like their clinical counterparts, in the presence of physiological hormone levels. Intraductal ER(+) PDXs are retransplantable, predictive, and appear genomically stable. The model provides opportunities for translational research and the study of physiologically relevant hormone action in breast carcinogenesis.

Positive Self-Image in Tournaments

This article analyzes the implications of worker overestimation of productivity for firms in which incentives take the form of tournaments. Each worker overestimates his productivity but is aware of the bias in his opponent's self-assessment. The manager of the firm, on the other hand, correctly assesses workers' productivities and self-beliefs when setting tournament prizes. The article shows that, under a variety of circumstances, firms can benefit from worker positive self-image. The article also shows that worker positive self-image can improve welfare in tournaments. In contrast, workers' utility declines due to their own misguided choices.

Effects of continuous positive airway pressure treatment on coronary vasoreactivity measured by (82)Rb cardiac PET/CT in obstructive sleep apnea patients.

PURPOSE: Obstructive sleep apnea syndrome (OSA) increases the risk of cardiovascular disease. We aimed at evaluating the effect of continuous positive airway pressure (CPAP) treatment on coronary endothelium-dependent vasoreactivity in OSA patients by quantifying myocardial blood flow (MBF) response to cold pressure testing (CPT). METHODS: In the morning after polysomnography (PSG), all participants underwent a dynamic (82)Rb cardiac positron emitting tomography/computed tomography (PET/CT) scan at rest, during CPT and adenosine stress. PSG and PET/CT were repeated at least 6 weeks after initiating CPAP treatment. OSA patients were compared to controls and according to response to CPAP. Patients' characteristics and PSG parameters were used to determine predictors of CPT-MBF. RESULTS: Thirty-two untreated OSA patients (age 58 ± 13 years, 27 men) and 9 controls (age 62 ± 5 years, 4 men) were enrolled. At baseline, compared to controls (apnea-hypopnea index (AHI) = 5.3 ± 2.6/h), untreated OSA patients (AHI = 48.6 ± 19.7/h) tend to have a lower CPT-MBF (1.1 ± 0.2 mL/min/g vs. 1.3 ± 0.4 mL/min/g, p = 0.09). After initiating CPAP, CPT-MBF was not different between well-treated patients (AHI <10/h) and controls (1.3 ± 0.3 mL/min/g vs. 1.3 ± 0.4 mL/min/g, p = 0.83), but it was lower for insufficiently treated patients (AHI ≥10/h) (0.9 ± 0.2 mL/min/g vs. 1.3 ± 0.4 mL/min/g, p = 0.0045). CPT-MBF was also higher in well-treated than in insufficiently treated patients (1.3 ± 0.3 mL/min/g vs. 0.9 ± 0.2 mL/min/g, p = 0.001). Mean nocturnal oxygen saturation (β = -0.55, p = 0.02) and BMI (β = -0.58, p = 0.02) were independent predictors of CPT-MBF in OSA patients. CONCLUSIONS: Coronary endothelial vasoreactivity is impaired in insufficiently treated OSA patients compared to well-treated patients and controls, confirming the need for CPAP optimization.

Estimating extreme value cumulative distribution functions using bias-corrected kernel approaches

We propose a new kernel estimation of the cumulative distribution function based on transformation and on bias reducing techniques. We derive the optimal bandwidth that minimises the asymptotic integrated mean squared error. The simulation results show that our proposed kernel estimation improves alternative approaches when the variable has an extreme value distribution with heavy tail and the sample size is small.

Learning Preferences with Kernel-Based Methods

Learning of preference relations has recently received significant attention in machine learning community. It is closely related to the classification and regression analysis and can be reduced to these tasks. However, preference learning involves prediction of ordering of the data points rather than prediction of a single numerical value as in case of regression or a class label as in case of classification. Therefore, studying preference relations within a separate framework facilitates not only better theoretical understanding of the problem, but also motivates development of the efficient algorithms for the task. Preference learning has many applications in domains such as information retrieval, bioinformatics, natural language processing, etc. For example, algorithms that learn to rank are frequently used in search engines for ordering documents retrieved by the query. Preference learning methods have been also applied to collaborative filtering problems for predicting individual customer choices from the vast amount of user generated feedback. In this thesis we propose several algorithms for learning preference relations. These algorithms stem from well founded and robust class of regularized least-squares methods and have many attractive computational properties. In order to improve the performance of our methods, we introduce several non-linear kernel functions. Thus, contribution of this thesis is twofold: kernel functions for structured data that are used to take advantage of various non-vectorial data representations and the preference learning algorithms that are suitable for different tasks, namely efficient learning of preference relations, learning with large amount of training data, and semi-supervised preference learning. Proposed kernel-based algorithms and kernels are applied to the parse ranking task in natural language processing, document ranking in information retrieval, and remote homology detection in bioinformatics domain. Training of kernel-based ranking algorithms can be infeasible when the size of the training set is large. This problem is addressed by proposing a preference learning algorithm whose computation complexity scales linearly with the number of training data points. We also introduce sparse approximation of the algorithm that can be efficiently trained with large amount of data. For situations when small amount of labeled data but a large amount of unlabeled data is available, we propose a co-regularized preference learning algorithm. To conclude, the methods presented in this thesis address not only the problem of the efficient training of the algorithms but also fast regularization parameter selection, multiple output prediction, and cross-validation. Furthermore, proposed algorithms lead to notably better performance in many preference learning tasks considered.

Kernel-Based Ranking. Methods for Learning and Performance Estimation

Machine learning provides tools for automated construction of predictive models in data intensive areas of engineering and science. The family of regularized kernel methods have in the recent years become one of the mainstream approaches to machine learning, due to a number of advantages the methods share. The approach provides theoretically well-founded solutions to the problems of under- and overfitting, allows learning from structured data, and has been empirically demonstrated to yield high predictive performance on a wide range of application domains. Historically, the problems of classification and regression have gained the majority of attention in the field. In this thesis we focus on another type of learning problem, that of learning to rank. In learning to rank, the aim is from a set of past observations to learn a ranking function that can order new objects according to how well they match some underlying criterion of goodness. As an important special case of the setting, we can recover the bipartite ranking problem, corresponding to maximizing the area under the ROC curve (AUC) in binary classification. Ranking applications appear in a large variety of settings, examples encountered in this thesis include document retrieval in web search, recommender systems, information extraction and automated parsing of natural language. We consider the pairwise approach to learning to rank, where ranking models are learned by minimizing the expected probability of ranking any two randomly drawn test examples incorrectly. The development of computationally efficient kernel methods, based on this approach, has in the past proven to be challenging. Moreover, it is not clear what techniques for estimating the predictive performance of learned models are the most reliable in the ranking setting, and how the techniques can be implemented efficiently. The contributions of this thesis are as follows. First, we develop RankRLS, a computationally efficient kernel method for learning to rank, that is based on minimizing a regularized pairwise least-squares loss. In addition to training methods, we introduce a variety of algorithms for tasks such as model selection, multi-output learning, and cross-validation, based on computational shortcuts from matrix algebra. Second, we improve the fastest known training method for the linear version of the RankSVM algorithm, which is one of the most well established methods for learning to rank. Third, we study the combination of the empirical kernel map and reduced set approximation, which allows the large-scale training of kernel machines using linear solvers, and propose computationally efficient solutions to cross-validation when using the approach. Next, we explore the problem of reliable cross-validation when using AUC as a performance criterion, through an extensive simulation study. We demonstrate that the proposed leave-pair-out cross-validation approach leads to more reliable performance estimation than commonly used alternative approaches. Finally, we present a case study on applying machine learning to information extraction from biomedical literature, which combines several of the approaches considered in the thesis. The thesis is divided into two parts. Part I provides the background for the research work and summarizes the most central results, Part II consists of the five original research articles that are the main contribution of this thesis.

Loss of PTEN expression and AKT activation in HER2-positive breast carcinomas

PURPOSE: To examine the expression of AKT and PTEN in a series of HER2-positive primary invasive breast tumors using immunohistochemistry, and to associate these expression profiles with classic pathologic features such as tumor grade, hormone receptor expression, lymphatic vascular invasion, and proliferation.METHODS: A total of 104 HER2-positive breast carcinoma specimens were prepared in tissue microarrays blocks for immunohistochemical detection of PTEN and phosphorylated AKT (pAKT). Original histologic sections were reviewed to assess pathological features, including HER2 status and Ki-67 index values. The associations between categorical and numeric variables were identified using Pearson's chi-square test and the Mann-Whitney, respectively.RESULTS: Co-expression of pAKT and PTEN was presented in 59 (56.7%) cases. Reduced levels of PTEN expression were detected in 20 (19.2%) cases, and these 20 tumors had a lower Ki-67 index value. In contrast, tumors positive for pAKT expression [71 (68.3%)] were associated with a higher Ki-67 index value.CONCLUSION: A role for AKT in the proliferation of HER2-positive breast cancers was confirmed. However, immunohistochemical detection of PTEN expression did not correlate with an inhibition of cellular proliferation or control of AKT phosphorylation, suggesting other pathways in these mechanisms of control.

Pulse response based control for positive definite systems

Pulse Response Based Control (PRBC) is a recently developed minimum time control method for flexible structures. The flexible behavior of the structure is represented through a set of discrete time sequences, which are the responses of the structure due to rectangular force pulses. The rectangular force pulses are given by the actuators that control the structure. The set of pulse responses, desired outputs, and force bounds form a numerical optimization problem. The solution of the optimization problem is a minimum time piecewise constant control sequence for driving the system to a desired final state. The method was developed for driving positive semi-definite systems. In case the system is positive definite, some final states of the system may not be reachable. Necessary conditions for reachability of the final states are derived for systems with a finite number of degrees of freedom. Numerical results are presented that confirm the derived analytical conditions. Numerical simulations of maneuvers of distributed parameter systems have shown a relationship between the error in the estimated minimum control time and sampling interval

The effects of positive end-expiratory pressure on respiratory system mechanics and hemodynamics in postoperative cardiac surgery patients

We prospectively evaluated the effects of positive end-expiratory pressure (PEEP) on the respiratory mechanical properties and hemodynamics of 10 postoperative adult cardiac patients undergoing mechanical ventilation while still anesthetized and paralyzed. The respiratory mechanics was evaluated by the inflation inspiratory occlusion method and hemodynamics by conventional methods. Each patient was randomized to a different level of PEEP (5, 10 and 15 cmH2O), while zero end-expiratory pressure (ZEEP) was established as control. PEEP of 15-min duration was applied at 20-min intervals. The frequency dependence of resistance and the viscoelastic properties and elastance of the respiratory system were evaluated together with hemodynamic and respiratory indexes. We observed a significant decrease in total airway resistance (13.12 ± 0.79 cmH2O l-1 s-1 at ZEEP, 11.94 ± 0.55 cmH2O l-1 s-1 (P<0.0197) at 5 cmH2O of PEEP, 11.42 ± 0.71 cmH2O l-1 s-1 (P<0.0255) at 10 cmH2O of PEEP, and 10.32 ± 0.57 cmH2O l-1 s-1 (P<0.0002) at 15 cmH2O of PEEP). The elastance (Ers; cmH2O/l) was not significantly modified by PEEP from zero (23.49 ± 1.21) to 5 cmH2O (21.89 ± 0.70). However, a significant decrease (P<0.0003) at 10 cmH2O PEEP (18.86 ± 1.13), as well as (P<0.0001) at 15 cmH2O (18.41 ± 0.82) was observed after PEEP application. Volume dependence of viscoelastic properties showed a slight but not significant tendency to increase with PEEP. The significant decreases in cardiac index (l min-1 m-2) due to PEEP increments (3.90 ± 0.22 at ZEEP, 3.43 ± 0.17 (P<0.0260) at 5 cmH2O of PEEP, 3.31 ± 0.22 (P<0.0260) at 10 cmH2O of PEEP, and 3.10 ± 0.22 (P<0.0113) at 15 cmH2O of PEEP) were compensated for by an increase in arterial oxygen content owing to shunt fraction reduction (%) from 22.26 ± 2.28 at ZEEP to 11.66 ± 1.24 at PEEP of 15 cmH2O (P<0.0007). We conclude that increments in PEEP resulted in a reduction of both airway resistance and respiratory elastance. These results could reflect improvement in respiratory mechanics. However, due to possible hemodynamic instability, PEEP should be carefully applied to postoperative cardiac patients.

Induction of interleukin-10 by HIV antigens in peripheral mononuclear cells of health care workers after occupational exposure to HIV-1-positive blood

Evaluation of HIV-induced IL-2 production by peripheral blood mononuclear cells (PBMC) and HIV-specific T helper and cytotoxic T lymphocyte (CTL) responses in health care workers (HCW) occupationally exposed to HIV reveals a high rate of response to HIV among non-seroconverters. IL-10 is also known to interfere with HIV infection in vitro. To evaluate the induction of IL-10 by HIV antigens in HCW occupationally exposed to HIV, 18 HCW with percutaneous injury were enrolled in this study, 9 of them exposed to HIV-contaminated blood, and 9 exposed to HIV-negative blood. PBMC were incubated on plates coated with HIV-1 antigens, and IL-10 was measured in supernatants by ELISA. Five of nine HCW exposed to HIV-contaminated blood presented HIV-induced IL-10. Two of nine HCW exposed to HIV-negative source patients also had detectable levels of HIV-induced IL-10, one of them in the sample obtained on the day of accidental exposure. There was a relationship between the type of device involved in injury and IL-10 production. Individuals exposed to hollow needles or scalpels presented HIV-induced IL-10, whereas those exposed to solid needles and to digital puncture did not, suggesting a relationship between infectious load and IL-10. Although occupational exposure to HIV leads to a low rate of seroconversion, these individuals can develop an antigen-specific immune response characterized in our study by induction of IL-10 in PBMC in vitro.

Vertical growth phase and positive sentinel node in thin melanoma

Sentinel node (SN) status is the most important prognostic factor for localized melanoma. Usually, patients with Breslow thickness of less than 1.0 mm are not included in SN protocols. However, the literature presents a rate ranging from 3 to 7% of nodal recurrence in thin melanoma. Ulceration, regression and high mitotic rate have been considered to be indications for an SN biopsy. The metastatic potential of the vertical growth phase is uncertain. To correlate pathological features in thin melanoma with SN metastasis, we reviewed 358 patients submitted to SN biopsy. Seventy-seven patients with lesions of 1 mm or smaller were included in the study group. Histological evaluation of the primary tumor included thickness, Clark level, mitotic rate, ulceration, regression, and growth phase. Lymphoscintigraphy was performed on all patients. Lymphatic mapping and gamma probe detection were both used for SN biopsy. Histological examination of SN consisted of hematoxylin-eosin and immunohistochemical staining. Median follow-up was 37 months. Six patients had micrometastases. Statistical analysis by the Fisher test showed that ulceration (P = 0.019), high mitotic rate (P = 0.008) and vertical growth phase (P = 0.002) were positively correlated with micrometastases. If other studies confirm these results, more melanoma patients must be submitted to SN biopsy.

Positive correlation of serum leptin with estradiol levels in patients with polycystic ovary syndrome

Patients with polycystic ovary syndrome (PCOS) usually are obese, insulin resistant and hyperinsulinemic. The known association between leptin, obesity andinsulin action suggests that leptin may have a role in PCOS but this has only been addressed peripherally. This study was designed to assess the relationship between serum leptin and the anthropometric, metabolic and endocrine variables of obese (body mass index, BMI ³30 kg/m²) and non-obese (BMI <30 kg/m²) PCOS patients. Twenty-eight PCOS patients and 24 control women subdivided into obese and non-obese groups were evaluated. Leptin, androgens, lipids, gonadotrophins and insulin-glucose response to the oral glucose tolerance test were measured by radioimmunoassay in all participants. The assays were done all in one time. The areas under the insulin curve (AUC-I) and the glycemia curve were calculated to identify patients with insulin resistance. Mean leptin levels were not significantly higher in patients with PCOS compared to the control group (21.2 ± 10.2 vs 27.3 ± 12.4 ng/ml). Leptin levels were found to be significantly higher in the obese subgroups both in patients with PCOS (26.9 ± 9.3 vs 14.1 ± 7.0 ng/ml) and in the control group (37.3 ± 15.5 vs 12.9 ± 5.8 ng/ml). The leptin of the PCOS group was correlated with BMI (r = 0.74; P < 0.0001) and estradiol (r = 0.48; P < 0.008) and tended to be correlated with the AUC-I (r = 0.36; P = 0.05). Of the parameters which showed a correlation with leptin in PCOS, only estradiol and probably insulinemia (AUC-I) did not show a significant correlation with BMI, suggesting that the other parameters were correlated with leptin due to their correlation with BMI. Estradiol correlated with leptin in PCOS patients regardless of their weight.