Giovani L. Silva

Bayesian analysis of correlated misclassified binary data

A Bayesian analysis for a random effect binary logistic regression model in the presence of misclassified data is considered. The introduction of a random effect captures the possible correlation among the binary data in each covariate pattern and hence may provide a good alternative to standard models in terms of overall fit. Markov Chain Monte Carlo methods are applied to perform the computations needed to draw inferences and make model assessment, through an illustrative example involving a real medical data set.

Familial vs sporadic papillary thyroid carcinoma: a matched-case comparative study showing similar clinical/prognostic behaviour.

OBJECTIVE Familial non-medullary thyroid cancer has been proposed as an aggressive clinical entity. Our aim in this study is to investigate potential distinguishing features as well as the biological and clinical aggressiveness of familial vs sporadic papillary thyroid carcinoma (PTC). We assessed clinicopathological characteristics, outcome measures and DNA ploidy. DESIGN A matched-case comparative study. METHODS A series of patients with familial PTC (n=107) and two subgroups, one with three or more affected elements (n=32) and another including index cases only (n=61), were compared with patients with sporadic PTC (n=107), matched by age, gender, pTNM disease extension and approximate follow-up duration. Histological variant, extrathyroidal extension, vascular invasion, tumour multifocality and bilateral growth were evaluated. Ploidy pattern was analysed in available samples by DNA flow cytometry. The probabilities of disease-free survival (DFS) and overall survival (OS) were estimated according to the Kaplan-Meier (K-M) method. RESULTS No patient with familial PTC died of disease during follow-up (median, 72 months), contrarily to five patients (4.7%) (P=0.06) with sporadic PTC (median, 90 months). There was a significantly higher tumour multifocality in familial PTC (index cases subgroup) vs sporadic PTC (P=0.035), and a trend, in the familial PTC cohort with three or more affected elements, to show extrathyroidal extension (P=0.054) more frequently. No difference was observed in DNA ploidy status. The K-M analyses showed no significant differences between both entities in relation to DFS or OS. CONCLUSION Apart from multifocality, familial PTC appears to have similar clinical/prognostic behaviour when compared with sporadic forms of the disease.

Overall survival in advanced breast cancer: Relevance of progesterone receptor expression and DNA ploidy in fine-needle aspirates of 392 patients

Fine-needle aspiration cytology (FNAC) is essential for making a diagnosis in advanced breast cancer. The determination of hormone receptors in the material obtained is useful for predicting patient response to endocrine therapy, but the prognostic value of hormone receptor expression as well as the clinical utility of DNA flow cytometry are controversial. The aim of this prospective study with long-term follow-up (median: 81 months) was to evaluate these biomarkers in relation to overall survival in a series of 392 patients with advanced breast cancer (stage IIB, n=106; IIIA, n=66; IIIB, n=174; and IV, n=46) using FNAC. Estrogen and progesterone receptor expression was found in 65.1% and 46.1% of the tumors, respectively. Hormone receptors were not found to be associated with clinical staging. DNA aneuploidy was present in 70.9% of the cases and the median S-phase fraction (SPF) was 9.4%. There was a significant correlation of aneuploidy and high SPF with lack of hormone receptors. In univariate analysis, advanced disease stage, absence of hormone receptors, DNA aneuploidy and high SPF showed a statistically significant correlation with poor clinical outcome. In multivariate analysis, disease stage, progesterone receptors and DNA ploidy retained independent prognostic significance in relation to overall survival. These data indicate that progesterone receptor expression and DNA ploidy are independent prognostic factors in advanced breast cancer.

BCL-6 oncoprotein in breast cancer: loss of expression in disease progression.

Objective: To investigate the biological role of BCL-6 oncoprotein in breast cancer disease progression (recurrence and metastasis). Methods: The series consisted of 93 consecutive female patients with primary breast cancer and median follow-up of 10 years. BCL-6 expression was assessed in vivo by immunohistochemistry and real-time PCR. Breast cancer cell lines and some metastasis-related genes (CXCR4, Itgβ-3 and FLT-1) were also analysed by molecular techniques. Prognostic evaluation was performed by fitting a multivariate Cox regression model. Results: BCL-6 immunoexpression was positive in 22 (23.7%) tumours and negative in 71 (76.3%). All axillary lymph node metastases of 47 node-positive patients were negative, including 12 cases showing BCL-6-positive primary tumours. Likewise, in 9 recurrence cases, BCL-6 expression was similar or decreased compared with primary tumours. No correlation between immunoexpression and gene expression of BCL-6 was observed. BCL-6 was significantly reduced both in derived metastases of a breast cancer cell line (M435) and when the latter was treated with a demethylation agent (5-azacytidine). However, BCL-6-transfected breast cancer cell lines expressed significantly higher levels of CXCR4, Itgβ-3 and FLT-1. Co-expression of the 4 genes was found in 4 of 17 tumours evaluated, but lacking prognostic significance. BCL-6 oncoprotein revealed no significant influence on outcome. Conclusion: The results strongly suggest the loss of BCL-6 expression in breast cancer progression, which might be related with methylation status alterations of still unknown partner gene(s).

DNA Flow Cytometry but Not Telomerase Activity as Predictor of Disease-Free Survival in pT1–2/N0/G2 Breast Cancer

Objective: In the subgroup of patients with node-negative (N0) moderately differentiated (G2) breast cancer, the clinical decision of giving adjuvant therapy is critical. The aim of this study was to investigate the prognostic value of biomarkers (DNA flow cytometry and telomerase activity in correlation with routinely used estrogen receptors (ER) and HER oncoprotein) in pT1–2/N0/G2 breast cancer, for improving therapeutic management. Methods: The series involved 135 patients with pT1–2/N0/G2 breast cancer and median follow-up of 58.5 months. DNA ploidy and S-phase fraction (SPF) (≤5%; 5–10%; >10%) were assessed on frozen samples. Telomerase activity, ER and c-erbB-2 expression were analyzed by standardized immunohistochemistry techniques. A Cox regression analysis was performed for prognostic evaluation. Results: Aneuploidy significantly correlated with high SPF and lack of ER, while high SPF showed significant correlations with high telomerase activity, c-erbB-2 overexpression and absence of ER. Kaplan-Meier curves showed significant differences for ploidy and SPF in relation with disease-free survival (DFS) and overall survival (OS), and a statistical trend for ER. By Cox regression analysis, DNA aneuploidy (RR = 16.7; p = 0.007) and high SPF (RR = 23.1; p = 0.004) revealed significant correlations with worse DFS. Among patients with diploid (n = 76) and low/intermediate SPF (n = 85) tumors, only one had recurrence of the disease. No association between telomerase activity and clinical outcome was observed. Conclusion: In pT1–2/N0/G2 breast cancer patients, DNA ploidy and SPF are relevant prognostic biomarkers that should be considered as additional tools in the therapeutic planning.

Bayesian Analysis of an Additive Survival Model with Frailty

Abstract There has been some interest in survival models with additive hazard function (e.g., the Aalens linear model), which is considered an alternative or supplementary model to the most common Cox model in analysis of censored data. One advantage of the Aalens model is to allow the influence of each covariate to vary separately over time. In this article, an additive survival model, based on counting processes, is presented introducing additive frailties into the intensity in order to account for the heterogeneity of the data due to unobserved risk factors or dependence among individuals in study. The additive frailty model is analyzed under a Bayesian point of view and illustrated by two well-known data sets.

Aneuploidy identifies subsets of patients with poor clinical outcome in grade 1 and grade 2 breast cancer.

OBJECTIVE Histological grade is a well-established prognostic/predictive factor in breast cancer. However, mainly within intermediate categories, patients may have unpredictable outcome. We hypothesised whether ploidy status can distinguish different prognostic groups among breast cancer patients with similar tumour grade. MATERIAL AND METHODS The study involved 684 patients with invasive breast carcinoma, and median follow-up of 134.5 months. Pathological staging was evaluated according to WHO classification. Tumour differentiation was assessed using the Nottingham grading system. Ploidy was determined prospectively by DNA flow cytometry. Disease-free survival (DFS) and overall survival (OS) were estimated by the Kaplan-Meier method. RESULTS There were 179 (26.2%) deaths and 239 (33.3%) disease recurrences. For grading, tumours were classified as follows: 163 (23.8%) G1, 356 (52.1%) G2 and 165 (24.1%) G3, while 389 (56.9%) tumours presented aneuploidy. Ploidy and grading are strongly associated (P < 0.001). Patients with aneuploid G2 tumours showed worse DFS (P = 0.001) and OS (P < 0.001), as well as those with aneuploid G1 tumours in relation to OS (P = 0.013). When a subset analysis was performed in early breast cancer patients (n = 451) with Stage I/IIA of disease, it remained the same significant associations of aneuploid G1 (to OS) and G2 tumours (to DFS and OS) with unfavourable prognosis. CONCLUSIONS Aneuploidy identifies subsets of breast cancer patients with G1 and G2 tumours who showed poor clinical outcome. The finding has therapeutic implications, as these patients are potential candidates to risk-adapted adjuvant therapy.

Clinical relevance of the reappraisal of negative hormone receptor expression in breast cancer

BackgroundAccurate assessment of estrogen (ER) and progesterone (PR) receptors is critical in predicting the response to endocrine therapies in breast cancer.Material and methodsFrom a series of 360 patients with breast invasive carcinoma assessed for hormone receptors by immunohistochemistry (IHC) in the 90’s, we re-analysed, on the same tumour material, the cases considered negative (n = 164), i.e., ER-/PR- (n = 95), ER+/PR- (n = 63) and ER-/PR+ (n=6), and 16 of 196 ER+/PR+ tumours with unfavourable outcome. Concordance between the previous IHC (Streptavidin-Biotin-Peroxidase) method and the current one (Peroxidase-Indirect-Polymer) was determined by the McNemar’s test. Relapse-free (RFS) and overall survival (OS) were estimated by the Kaplan-Meier method.ResultsFrom 101 ER- and 158 PR- cases, 38 (37.6%) and 58 (36.7%) became positive, increasing ER and PR expression from 71.9% and 56.1% to 82.5% and 72.2%, respectively (P<0.001). All 16 ER+/PR+ cases maintained their co-positivity, while all ER-/PR+ tumours changed to ER positive. Kaplan-Meier survival curves showed significant differences related to RFS and OS for PR, either in the whole series or in the subset (n = 151) submitted to hormonal treatment. The patients’ subgroup with ER+/PR- tumours exhibited the worst prognosis.ConclusionThe current IHC method improves the clinical usefulness of ER/PR assessment by decreasing the rate of false negative results.

S-phase fraction and ploidy as predictive markers in primary disease and recurrence of papillary thyroid carcinoma

To investigate the prognostic influence of DNA ploidy and S‐phase fraction (SPF) on disease‐free (DFS) and overall survival (OS) of patients with primary disease and loco‐regional lymph node recurrence of papillary thyroid carcinoma (PTC).

The influence of the winter North Atlantic Oscillation index on hospital admissions through diseases of the circulatory system in Lisbon, Portugal

The aim of this paper is to analyze the relationship between North Atlantic Oscillation (NAO), meteorological variables, air pollutants, and hospital admissions due to diseases of circulatory systems in Lisbon (Portugal) during winter months (2003–2012). This paper is one of the few studies analyzing the impact of NAO on health through its influence on thermal stress and air pollution and is the first to be conducted in Lisbon. This study uses meteorological data (synthetized into a thermal comfort index), air pollutant metrics, and the NAO index (all clustered in 10-day cycles to overcome daily variability of the NAO index). The relationship between morbidity, thermal comfort index, NAO index, and air pollutants was explored through several linear models adjusted to seasonality through a periodic function. The possible indirect effect between the NAO index and hospital admissions was tested, assuming that NAO (independent variable) is affecting hospital admissions (outcome variable) through thermal discomfort and/or pollution levels (tested as individual mediators). This test was conducted through causal mediation analysis and adjusted for seasonal variation. The results from this study suggest a possible indirect relationship between NAO index and hospital admissions. Although NAO is not significantly associated with hospital admissions, it is significantly associated with CO, PM2.5, NO, and SO2 levels, which in turn increase the probability of hospitalization. The discomfort index (built with temperature and relative humidity) is significantly associated with hospital admissions, but its variability is not explained by the NAO index. This study highlights the impacts of the atmospheric circulation patterns on health. Furthermore, understanding the influence of the atmospheric circulation patterns can support the improvement of the existing contingency plans.