Stefano Parodi

Differential diagnosis of pleural mesothelioma using Logic Learning Machine

BackgroundTumour markers are standard tools for the differential diagnosis of cancer. However, the occurrence of nonspecific symptoms and different malignancies involving the same cancer site may lead to a high proportion of misclassifications.Classification accuracy can be improved by combining information from different markers using standard data mining techniques, like Decision Tree (DT), Artificial Neural Network (ANN), and k-Nearest Neighbour (KNN) classifier. Unfortunately, each method suffers from some unavoidable limitations. DT, in general, tends to show a low classification performance, whereas ANN and KNN produce a black-box classification that does not provide biological information useful for clinical purposes.MethodsLogic Learning Machine (LLM) is an innovative method of supervised data analysis capable of building classifiers described by a set of intelligible rules including simple conditions in their antecedent part. It is essentially an efficient implementation of the Switching Neural Network model and reaches excellent classification accuracy while keeping low the computational demand.LLM was applied to data from a consecutive cohort of 169 patients admitted for diagnosis to two pulmonary departments in Northern Italy from 2009 to 2011. Patients included 52 malignant pleural mesotheliomas (MPM), 62 pleural metastases (MTX) from other tumours and 55 benign diseases (BD) associated with pleurisies. Concentration of three tumour markers (CEA, CYFRA 21-1 and SMRP) was measured in the pleural fluid of each patient and a cytological examination was also carried out.The performance of LLM and that of three competing methods (DT, KNN and ANN) was assessed by leave-one-out cross-validation.ResultsLLM outperformed all other considered methods. Global accuracy was 77.5% for LLM, 72.8% for DT, 54.4% for KNN, and 63.9% for ANN, respectively. In more details, LLM correctly classified 79% of MPM, 66% of MTX and 89% of BD. The corresponding figures for DT were: MPM = 83%, MTX = 55% and BD = 84%; for KNN: MPM = 58%, MTX = 45%, BD = 62%; for ANN: MPM = 71%, MTX = 47%, BD = 76%.Finally, LLM provided classification rules in a very good agreement with a priori knowledge about the biological role of the considered tumour markers.ConclusionsLLM is a new flexible tool potentially useful for the differential diagnosis of pleural mesothelioma.

Diagnostic value of mesothelin in pleural fluids: comparison with CYFRA 21-1 and CEA

CYFRA 21-1 and CEA have been applied for the differential diagnosis of malignant pleural mesothelioma (MPM). The soluble mesothelin-related peptide (SMRP) has been proposed as a specific marker for distinguishing MPM from benign diseases and other malignancies in pleural effusions (PEs). In this study, we evaluated the usefulness of SMRP in PEs in the detection of mesotheliomas by comparing it with that of CYFRA 21-1, CEA, and with cytological examination. One hundred and seventy-seven consecutive patients (57 MPM, 64 metastatic tumors, and 56 benign diseases) were evaluated using commercial tests. The performance of the markers was analyzed by standard ROC analysis methods, using the area under a ROC curve (AUC) as a measure of accuracy. CYFRA 21-1 better differentiated malignant from benign effusions. The corresponding area under the receiver operating characteristic curve was 0.87, while it was 0.74 for SMRP and 0.64 for CEA (pxa0<xa00.001). Conversely, SMRP differentiated MPM from all other PEs better than both CYFRA 21-1 and CEA (AUCxa0=xa00.84, 0.76, and 0.32, respectively, pxa0=xa00.003). Low levels of CEA were associated with a MPM diagnosis. The AUC for differentiating MPM from metastases was 0.81 for SMRP, 0.61 for CYFRA 21-1, and 0.20 for CEA (pxa0<xa00.001). In cases with negative or suspicious cytology, SMRP and CYFRA 21-1 identified 36/71 and 46/66 malignant PEs (29 and 31 MPM, respectively). Only 1 MPM showed a high CEA concentration. No single marker showed the best performance in any comparison. Results suggest that SMRP could improve CYFRA 21-1 and CEA accuracy in the differential diagnosis of MPM.

Risk of neuroblastoma, maternal characteristics and perinatal exposures: The SETIL study

PURPOSEnNeuroblastoma (NB) is the most common extra-cranial paediatric solid tumour. Incidence peaks in infancy, suggesting a role of in-utero and neonatal exposures but its aetiology is largely unknown. The aim of the present study is to evaluate the association between maternal characteristics and perinatal factors with the risk of NB, using data from the SETIL database.nnnMETHODSnSETIL is a large Italian population-based case-control study established to evaluate several potential cancer risk factors in 0-10 year olds. Information about maternal characteristics, reproductive history, environmental and occupational exposures during pregnancy, as well as newborns characteristics were obtained using a structured questionnaire. Extremely low frequency magnetic field (ELF-MF) home exposure was measured. The study included 1044 healthy controls and 153 NB cases, diagnosed between 1998 and 2001.nnnRESULTSnA twofold risk was associated to exposure in pregnancy to chemical products for domestic work and to hair dye. The risk associated with the latter was higher among 0-17 month old children (OR = 5.5, 95%CI: 1.0-29.3). Risk was increased for children whose mothers had suffered work related exposure in the preconception period to solvents (OR = 2.0 95%CI: 1.0-4.1) and in particular to aromatic hydrocarbons (OR = 9.2, 95%CI: 2.4-34.3). No association was observed with ELF-MF exposure. A higher risk was found among children with congenital malformations (OR = 4.9, 95%CI: 1.8-13.6) or neurofibromatosis (2 cases and 0 controls, p = 0.016).nnnCONCLUSIONSnOur study suggests maternal exposure to hair dyes and aromatic hydrocarbons plays a role and deserves further investigation. The association with congenital malformations might also be explained by over-diagnosis. External exposure, in particular during and before pregnancy might contribute to NB occurrence.

Use of Attribute Driven Incremental Discretization and Logic Learning Machine to build a prognostic classifier for neuroblastoma patients

BackgroundCancer patients outcome is written, in part, in the gene expression profile of the tumor. We previously identified a 62-probe sets signature (NB-hypo) to identify tissue hypoxia in neuroblastoma tumors and showed that NB-hypo stratified neuroblastoma patients in good and poor outcome [1]. It was important to develop a prognostic classifier to cluster patients into risk groups benefiting of defined therapeutic approaches. Novel classification and data discretization approaches can be instrumental for the generation of accurate predictors and robust tools for clinical decision support. We explored the application to gene expression data of Rulex, a novel software suite including the Attribute Driven Incremental Discretization technique for transforming continuous variables into simplified discrete ones and the Logic Learning Machine model for intelligible rule generation.ResultsWe applied Rulex components to the problem of predicting the outcome of neuroblastoma patients on the bases of 62 probe sets NB-hypo gene expression signature. The resulting classifier consisted in 9 rules utilizing mainly two conditions of the relative expression of 11 probe sets. These rules were very effective predictors, as shown in an independent validation set, demonstrating the validity of the LLM algorithm applied to microarray data and patients classification. The LLM performed as efficiently as Prediction Analysis of Microarray and Support Vector Machine, and outperformed other learning algorithms such as C4.5. Rulex carried out a feature selection by selecting a new signature (NB-hypo-II) of 11 probe sets that turned out to be the most relevant in predicting outcome among the 62 of the NB-hypo signature. Rules are easily interpretable as they involve only few conditions.Furthermore, we demonstrate that the application of a weighted classification associated with the rules improves the classification of poorly represented classes.ConclusionsOur findings provided evidence that the application of Rulex to the expression values of NB-hypo signature created a set of accurate, high quality, consistent and interpretable rules for the prediction of neuroblastoma patients outcome. We identified the Rulex weighted classification as a flexible tool that can support clinical decisions. For these reasons, we consider Rulex to be a useful tool for cancer classification from microarray gene expression data.

Lifestyle factors and risk of leukemia and non-Hodgkin's lymphoma: a case-control study.

AbstractPurposeRisk factors for leukemia and lymphomas in adults are largely unknown. This study was aimed at evaluating the association between lifestyle factors and the risk of hematological malignancies in an adult population.nMethodsData were drawn from a population-based case–control study carried out in Italy and included 294 cases (199 lymphoid and 95 myeloid) and 279 controls. Analyses were performed using standard multivariable logistic regression.ResultsHair dye use for at least 15xa0years was associated with a higher risk of lymphoid malignancies among females (OR 2.3, 95xa0% CI 1.0–4.9, pxa0=xa00.036, test for trend). Furthermore, a protective effect of a moderate to heavy tea consumption on the risk of myeloid malignancies was observed (OR 0.4, 95xa0% CI 0.2–0.9, pxa0=xa00.017). No association was found for the use of alcoholic beverages and tobacco smoking.ConclusionsOur results confirm the potential carcinogenic effect of prolonged hair dye use observed in previous investigations. The excess risk could be explained by exposure to a higher concentration of toxic compounds in hair products used in the past. The protective effect of regular tea consumption observed in an area with a very high prevalence of black tea consumers deserves further investigation.

Seasonal variations of date of diagnosis and birth for neuroblastoma patients in Italy

BACKGROUNDnAnalysis of seasonal variation of diagnosis or birth of childhood cancers may provide useful insight about possible aetiological risk factors, such as infectious agents and environmental exposures, but studies on neuroblastoma are lacking.nnnPROCEDUREnTwo thousand seven hundred fifty-six cases of neuroblastoma, diagnosed between 1980 and 2010, registered in the Italian Neuroblastoma Registry, were included in the study. Subgroup analyses were carried out by age, gender and stage at diagnosis. Seasonal trend was assessed by a harmonic function in a Poisson regression model, adjusted for the number of live births.nnnRESULTSnNo trend in the date of diagnosis was found either in the entire cohort or in the various sub-groups. Similarly, a seasonal trend of birth was not observed in the whole cohort. Conversely, in the subgroup of infants with stage 4S, a significant peak of July births was found (23.6% increment from the average, p=0.042). The summer peak was confirmed after stratifying 4S patients by gender and period of diagnosis.nnnCONCLUSIONSnA major effect of risk factors related to seasonality does not appear to affect the risk of developing neuroblastoma. However, the time pattern of birth observed by stage at diagnosis is consistent with the hypothesis that Stage 4S is a distinct disease with probably a different aetiology, as indicated by investigations on its metastatic pattern and its peculiar gene expression. An aetiological role of seasonally related factors, e.g., favouring the survival of defective neural crest stem cells, remains speculative and need confirmation by independent studies.

Coffee and tea consumption and risk of leukaemia in an adult population: A reanalysis of the Italian multicentre case-control study

BACKGROUNDnCoffee and tea are the most frequently consumed beverages in the world. Their potential effect on the risk of developing different types of malignancies has been largely investigated, but studies on leukaemia in adults are scarce.nnnMETHODSnThe present investigation is aimed at evaluating the potential role of regular coffee and tea intake on the risk of adult leukaemia by reanalysing a large population based case-control study carried out in Italy, a country with a high coffee consumption and a low use of green tea. Interviewed subjects, recruited between 1990 and 1993 in 11 Italian areas, included 1771 controls and 651 leukaemia cases. Association between Acute Myeloid Leukaemia (AML), Acute Lymphoid Leukaemia, Chronic Myeloid Leukaemia, Chronic Lymphoid Leukaemia, and use of coffee and tea was evaluated by standard logistic regression. Odds Ratios (OR) were estimated adjusting for the following potential confounders: gender, age, residence area, smoking habit, educational level, previous chemotherapy treatment, alcohol consumption and exposure to electromagnetic fields, radiation, pesticides and aromatic hydrocarbons.nnnRESULTSnNo association was observed between regular use of coffee and any type of leukaemia. A small protective effect of tea intake was found among myeloid malignancies, which was more evident among AML (OR=0.68, 95%CI: 0.49-0.94). However, no clear dose-response relation was found.nnnCONCLUSIONnThe lower risk of leukaemia among regular coffee consumers, reported by a few of previous small studies, was not confirmed. The protective effect of tea on the AML risk is only partly consistent with results from other investigations.

Risk of leukaemia and residential exposure to air pollution in an industrial area in Northern Italy: a case-control study

Leukaemia risk in adult populations exposed to environmental air pollution is poorly investigated. We have carried out a population-based case-control study in an area that included a fossil fuel power plant, a coke oven and two big chemical industries. Information on residential history and several risk factors for leukaemia was obtained from 164 cases, diagnosed between 2002 and 2005, and 279 controls. A higher risk for subjects residing in polluted areas was observed, but statistical significance was not reached (adjusted OR = 1.11 and 1.56 for subjects living in moderately and in heavily polluted zones, respectively, p = 0.190). Results suggest a possible aetiological role of residential air pollution from industrial sites on the risk of developing leukaemia in adult populations. However, the proportion of eligible subjects excluded from the study and the lack of any measure of air pollution prevent definitive conclusions from being drawn.

Chronic diseases, medical history and familial cancer, and risk of leukemia and non-Hodgkin’s lymphoma in an adult population: a case–control study

PurposeThis investigation was aimed at evaluating the association between chronic diseases, medical history and familial cancer, and the risk of developing hematological malignancies.MethodsData were drawn from a population-based case–control study carried out to assess the risk of non-Hodgkin’s lymphoma and leukemia in an adult population exposed to environmental air pollution in Northern Italy. Each case was classified according to the WHO ICD-O-3 classification. Statistical analyses were performed by multivariable unconditional logistic regression in 573 interviewed subjects (199 lymphoid cases, 95 myeloid cases, and 279 healthy controls).ResultsLymphoid malignancies were associated with a history of gastroduodenal ulcer (OR 2.1, 95xa0% CI 1.2–3.6), rheumatoid arthritis (OR 4.4, 95xa0% CI 1.3–19.0), anemia (OR 3.3, 95xa0% CI 1.2–9.3), cholecystectomy (OR 2.9, 95xa0% CI 1.0–8.0), heavy diagnostic X-ray exposure (OR 2.1, 95xa0% CI 1.3–3.7), and a familial risk of non-Hodgkin’s lymphoma (OR 10.1, 95xa0% CI 1.3–458). Myeloid malignancies were associated with non-neoplastic thyroid diseases (OR 6.2, 95xa0% CI 1.7–35.6) and anemia (OR 6.8, 95xa0% CI 2.0–23.1). Subgroup analysis highlighted an excess risk of MALT in patients with gastroduodenal ulcer (OR 5.3, 95xa0% CI 1.04–23.7) and of AML in patients with rheumatoid arthritis (OR 6.9, 95xa0% CI 1.2–38.1), and of MDS in subjects exposed to heavy diagnostic X-ray (OR 3.4, 95xa0% CI 1.03–11.2) when the analysis was restricted to irradiation of pelvis, abdomen, or thorax.ConclusionsMost observed associations confirm results from previous studies. The higher risk of lymphoid malignancies among patients with a history of cholecystectomy needs further investigations.

Logic Learning Machine and standard supervised methods for Hodgkin's lymphoma prognosis using gene expression data and clinical variables.

This study evaluates the performance of a set of machine learning techniques in predicting the prognosis of Hodgkin’s lymphoma using clinical factors and gene expression data. Analysed samples from 130 Hodgkin’s lymphoma patients included a small set of clinical variables and more than 54,000 gene features. Machine learning classifiers included three black-box algorithms (k-nearest neighbour, Artificial Neural Network, and Support Vector Machine) and two methods based on intelligible rules (Decision Tree and the innovative Logic Learning Machine method). Support Vector Machine clearly outperformed any of the other methods. Among the two rule-based algorithms, Logic Learning Machine performed better and identified a set of simple intelligible rules based on a combination of clinical variables and gene expressions. Decision Tree identified a non-coding gene (XIST) involved in the early phases of X chromosome inactivation that was overexpressed in females and in non-relapsed patients. XIST expression might be responsible for the better prognosis of female Hodgkin’s lymphoma patients.