Impact of endometriosis on obstetric outcome after natural conception: a multicenter Italian study.

PURPOSE: To evaluate obstetric outcome in women with endometriosis who conceive naturally and receive standard obstetric care in Italy. METHODS: Cases were consecutive women with endometriosis managed in eleven Italian referral centers. Controls were women in whom endometriosis was excluded. All women filled in a questionnaire addressing previous natural pregnancies. Marginal logistic regression models were fitted to evaluate the impact of endometriosis on obstetric outcome. A post hoc analysis was performed within the endometriosis group comparing women with severe adenomyosis versus women with absent or mild adenomyosis. RESULTS: Three hundred and fifty-five pregnancies in endometriosis group and 741 pregnancies in control group were included. Women with endometriosis had a higher risk of preterm delivery < 34 weeks (6.4% vs 2.8%, OR 2.42, 95% CI 1.22-4.82), preterm delivery < 37 weeks (17.8% vs 9.7%, OR 1.98, 95% CI 1.23-3.19), and neonatal admission to Intensive Care Unit (14.1% vs 7.0%, OR 2.04, 95% CI 1.23-3.36). At post hoc analysis, women with endometriosis and severe adenomyosis had an increased risk of placenta previa (23.1% vs 1.8%, OR 16.68, 95% CI 3.49-79.71), cesarean delivery (84.6% vs 38.9%, OR 8.03, 95% CI 1.69-38.25) and preterm delivery < 34 weeks (23.1% vs 5.7%, OR 5.52, 95% CI 1.38-22.09). CONCLUSION: Women with endometriosis who conceive naturally have increased risk of preterm delivery and neonatal admission to intensive care unit. When severe adenomyosis is coexistent with endometriosis, women may be at increased risk of placenta previa and cesarean delivery. TRIAL REGISTRATION: Clinical trial registration number: NCT03354793.

Major adverse cardiovascular events associated with VEGF-targeted anticancer tyrosine kinase inhibitors: a real-life study and proposed algorithm for proactive management.

BACKGROUND: Vascular endothelial growth factor receptor (VEGFR)-targeted tyrosine kinase inhibitors (TKIs) are widely used in cancer treatment and burdened by cardiovascular toxicity. The majority of data come from clinical trials, thus in selected populations. The aim of our study is to evaluate the cardiotoxicity profile of VEGFR-targeted TKIs and the impact of cardiovascular risk factors in a real-life population. PATIENTS AND METHODS: In this cohort, population-based study, patients treated with VEGFR-targeted TKIs, bevacizumab and trastuzumab between 2009 and 2014 were analyzed. A multi-source strategy for data retrieval through hospital, pharmaceutical and administrative databases of the Lombardy region, Italy, has been adopted. The primary endpoint was to determine the incidence and type of major adverse cardiovascular events (MACEs) along with their temporal trend. The secondary endpoint was to define the impact of cardiovascular risk factors in the occurrence of MACEs. RESULTS: A total of 829 patients were treated with VEGFR-targeted TKIs. Eighty-one MACEs occurred in the first year of follow-up [crude cumulative incidence (CCI): 9.79%] mainly consisting of arterial thrombotic events (ATEs, 31 events, CCI: 3.99%), followed by rhythm disorders (22 events, CCI: 2.66%), pulmonary embolisms and heart failures (13 events each, CCI: 1.57%). While the incidence of most MACEs showed a plateau after 6 months, ATEs kept increasing along the year of follow-up. Hypertension and dyslipidemia were associated with an increase in risk of ATEs [relative risk difference (RRD) +209.8% and +156.2%, respectively], while the presence of previous MACEs correlated with a higher risk of all MACEs in multivariate analysis (RRD 151.1%, 95% confidence interval 53.6% to 310.3%, P < 0.001). CONCLUSIONS: MACEs occur in a clinically significant proportion of patients treated with VEGFR-targeted TKIs, with ATEs being predominant, mainly associated with hypertension and dyslipidemia. A clinical algorithm for effective proactive management of these patients is warranted.

Monitoring the haemodynamic response to visual stimulation in glaucoma patients.

In this paper, we used time-domain functional near infrared spectroscopy (TD-fNIRS) to evaluate the haemodynamic response function (HRF) in the occipital cortex following visual stimulation in glaucomatous eyes as compared to healthy eyes. A total of 98 subjects were enrolled in the study and clinically classified as healthy subjects, glaucoma patients (primary open-angle glaucoma) and mixed subjects (i.e. with a different classification for the two eyes). After quality check data were used from HRF of 73 healthy and 62 glaucomatous eyes. The amplitudes of the oxygenated and deoxygenated haemoglobin concentrations, together with their latencies with respect to the stimulus onset, were estimated by fitting their time course with a canonical HRF. Statistical analysis showed that the amplitudes of both haemodynamic parameters show a significant association with the pathology and a significant discriminating ability, while no significant result was found for latencies. Overall, our findings together with the ease of use and noninvasiveness of TD-NIRS, make this technique a promising candidate as a supporting tool for a better evaluation of the glaucoma pathology.

Solitary and multiple simultaneous malignant epithelial mammary tumours in dogs: An explorative retrospective study.

Canine mammary tumours represent a hard-prognostic task for veterinary clinicians. TNM staging and grading systems refer to a single tumour. Significant limits come to light when these systems are applied to multiple mammary tumours due to the arbitrary criterion in determining which single tumour is representative of the patient's prognosis. This study explored some clinical features of 50 dogs affected by at least one malignant mammary tumour. Clinical features and staging, together with histological classification and grading, have been related to disease-free survival (DFS) with the purpose to evaluate their impact on prognosis. The prognosis was worse in 10-11-year-old dogs (P < 0.05), in dogs affected by complex carcinoma (P < 0.05), and in patients assigned to Peña grade I (P < 0.05). The bodyweight was not linearly related to DFS (P < 0.01), and patients with a low number of neoformations (n ≤ 2) showed a better prognosis than dogs with 3-5 tumours (P < 0.05). Both the average and the total size of malignant tumours were related to DFS (P < 0.05). Dogs assigned with stage I had the best DFS (P < 0.05). In conclusion, the Peña grade I alone would not seem to guarantee a favourable prognosis when applied to mammary tumours in dogs affected by multiple simultaneous presentations. Different characteristics, besides tumour grading, such as tumour immunophenotype and expression of hormonal receptors, could in the future, contribute to elucidate the clinical behaviour of multiple canine mammary tumours.

Evaluation of leukocyte counts and neutrophil-to-lymphocyte ratio as predictors of local recurrence of feline injection site sarcoma after curative intent surgery.

Local recurrence (LR) is the major concern in the treatment of feline injection-site sarcoma (FISS). Pretreatment leukocyte counts and ratios have been reported as diagnostic and/or prognostic markers in human and canine oncology. The aim of this retrospective study was to explore the prognostic impact on LR and overall survival time (OST) of pretreatment neutrophil-to-lymphocyte ratio (NLR), white blood cell count (WBCC), neutrophil count (NC) and lymphocyte count (LC) in cats with surgically excised FISS. Eighty-two cats with histologically confirmed FISS at first presentation, without distant metastases, and with available pretreatment haematological analyses were retrospectively enrolled. The correlation of NLR, WBCC, NC, LC with tumour variables and patient variables was explored. NLR was correlated with tumour size (P = .004), histological pattern of tumour growth (P = .024) and histotype (P = .029), while WBCC and NC were associated with ulceration (P = .007, P = .011) and pattern of growth (P = .028, P = .004). No significant relationships emerged between LC and any of the considered variables. The impact of NLR, WBCC, NC, LC on LR and OST was then estimated in univariate and multivariate analysis. In univariate analysis, NLR, WBCC and NC were significant prognostic factors for both LR and OST. NLR, WBCC and NC remained prognostic in multivariate analysis for LR but not for OST. When NLR, WBCC and NC were jointly analysed, WBCC was the marker with the greater impact on LR. Preoperative NLR, WBCC and NC may aid in identifying cats at higher risk of LR.

The impact of extirpation of non-palpable/normal-sized regional lymph nodes on staging of canine cutaneous mast cell tumours: A multicentric retrospective study.

Metastasis to regional lymph nodes (RLNs) in dogs with cutaneous mast cell tumour (cMCT) has been correlated with shortened survival time and higher risk of spread to distant sites. In the present study, extirpation of non-palpable or normal-sized RLNs was included in the surgical management of cMCT in dogs. Correlations between histological nodal status (HN0-3) and tumour variables were analysed. Ninety-three dogs with single cMCT without distant metastasis that underwent wide surgical excision of the primary tumour and extirpation of non-palpable or normal-sized RLN were included. The association between HN (HN0 vs HN > 0; HN0-1 vs HN2-3) and tumour variables (site, longest diameter, ulceration, 3-tier and 2-tier histological grades) was analysed by a generalized linear model with multinomial error. Then, 33 (35.5%) RLNs were HN0, 14 (15%) were HN1, 26 (28%) were HN2 and 20 (21.5%) were HN3. The presence of positive (HN > 0) RLN was significantly associated with cMCT larger than 3 cm. No other association was statistically significant. Non-palpable/normal-sized RLN in dogs with cMCT can harbour histologically detectable metastatic disease in nearly half of the cases. Extirpation of the RLN should always perfomed to obtain a correct staging of the disease, even in the absence of clinical suspicion of metastasis. Further studies should evaluate the possible therapeutical effect of the tumour burden reduction obtained by exrtipartion of a positive RLN.

Clinical and computed tomography tumour dimension assessments for planning wide excision of injection site sarcomas in cats: how strong is the agreement?

In injection site sarcoma (ISS) in cats lateral as well as deep margins should be correctly planned for a successful surgical outcome. The discrepancy between clinical and computed tomography (CT) measurements of dimension in resectable tumour has led to possible bias that affects the subsequent surgical dose. The aim of this study was to prospectively investigate the agreement between clinical and CT measurements of dimension in newly diagnosed ISS in cats. Fifty-three client-owned cats that underwent both clinical and CT measurements of the length and width of ISS were included. CT measurements showed a tendency towards being larger than clinical dimensions, and this difference increased with increasing tumour size. Based on our results, in further studies focusing on ISS in cats, the kind of assessment used to define tumour dimensions (CT versus clinic) should be declared and specified to properly consider surgical results and prognostic impact of this variable.

Growth Factors and COX2 Expression in Canine Perivascular Wall Tumors.

Canine perivascular wall tumors (PWTs) are a group of subcutaneous soft tissue sarcomas developing from vascular mural cells. Mural cells are involved in angiogenesis through a complex crosstalk with endothelial cells mediated by several growth factors and their receptors. The evaluation of their expression may have relevance since they may represent a therapeutic target in the control of canine PWTs. The expression of vascular endothelial growth factor (VEGF) and receptors VEGFR-I/II, basic fibroblast growth factor (bFGF) and receptor Flg, platelet-derived growth factor B (PDGFB) and receptor PDGFRß, transforming growth factor ß1 (TGFß1) and receptors TGFßR-I/II, and cyclooxygenase 2 (COX2) was evaluated on frozen sections of 40 PWTs by immunohistochemistry and semiquantitatively scored to identify their potential role in PWT development. Statistical analysis was performed to analyze possible correlations between Ki67 labeling index and the expression of each molecule. Proteins of the VEGF-, PDGFB-, and bFGF-mediated pathways were highly expressed in 27 (67.5%), 30 (75%), and 19 (47.5%) of 40 PWTs, respectively. Proteins of the TGFß1- and COX2-mediated pathways were highly expressed in 4 (10%) and 14 (35%) of 40 cases. Statistical analysis identified an association between VEGF and VEGFR-I/II (P = .015 and .003, respectively), bFGF and Flg (P = .038), bFGF and PDGFRß (P = .003), and between TGFß1 and COX2 (P = .006). These findings were consistent with the mechanisms that have been reported to play a role in angiogenesis and in tumor development. No association with Ki67 labeling index was found. VEGF-, PDGFB-, and bFGF-mediated pathways seem to have a key role in PWT development and growth. Blockade of tyrosine kinase receptors after surgery could represent a promising therapy with the aim to reduce the PWT relapse rate and prolong the time to relapse.

Feline upper respiratory tract lymphoma: site, cyto-histology, phenotype, FeLV expression, and prognosis.

Lymphoma is the most common feline upper respiratory tract (URT) tumor. Primary nasal and nasopharyngeal lymphomas have been evaluated as distinct pathological entities; however, data on their differing clinical behavior are missing. A total of 164 endoscopic- guided URT pinch biopsies were formalin fixed and routinely processed. Imprint cytological specimens were stained with May Grünwald-Giemsa. Immunohistochemistry for anti-CD20, CD3, FeLVp27, and FeLVgp70 was performed. Prognostic significance of clinicopathological variables was investigated by univariate and multivariate analysis. Lymphoma was diagnosed in 39 cats (24%). Most cats with lymphoma were domestic shorthair (32 [82%]), were male (F/M = 0.56), and had a mean age of 10.3 years (range, 1-16 years). Lymphomas were primary nasal in 26 cats (67%), nasopharyngeal in 6 (15%), and in both locations (combined lymphomas) in 7 cats (18%). Neoplastic growth pattern was diffuse in 35 cases (90%) and nodular in 4 (10%). Epitheliotropism was observed in 10 cases (26%). Tumor cells were large in 15 cases, were small and medium in 11 cases each, and 2 had mixed cell size. Submucosal lymphoplasmacytic inflammation was observed in 23 cases (59%). Cytology was diagnostic for lymphoma in 12 of 25 cases (48%). A B-cell origin prevailed (34 [87%]). Feline leukemia virus (FeLV) p27 or gp70 antigen was detected in 21 lymphomas (54%). URT lymphomas were aggressive, with survival varying from 0 to 301 days (mean, 53 days). Epitheliotropism in 8 B-cell lymphomas (80%) and in 2 T-cell lymphomas (20%) correlated with prolonged survival. Age younger or older than 10 years had a negative prognostic value. Lymphoplasmacytic inflammation and FeLV infection may represent favoring factors for URT lymphoma development.

Drug survival and reasons for discontinuation of the first course of biological therapy in 301 juvenile idiopathic arthritis patients.

The objective of this study was to determine long-term effectiveness and safety of 1st biological treatment (BT) in a cohort of 301 juvenile idiopathic arthritis (JIA) patients (pts), non-responders to disease-modifying antirheumatic drugs (DMARDs), in terms of drug survival (continuation rate on therapy) and to identify the baseline predictors of treatment discontinuation. Each JIA pt enrolled in BT is prospectively assessed at the start of treatment and then every 2 months for the evaluation of safety and efficacy according to ACR-Pedi30 criteria. All clinical charts of pts who started a BT between November 1999 and July 2010 have been reviewed. Survival analysis methods suitable for competing risks were used to study time to drug discontinuation due to disease control (defined according to Wallace criteria) or failure [adverse event (AE), lack of efficacy (LaE) or loss of efficacy (LoE) according ACR-Pedi30]. A number of 301 JIA pts, non-responders or intolerant to DMARDs and treated with one or more cycles of BT, were identified. Median disease duration, from onset to the beginning of 1st BT, was 7.8 years (interquartil range 2.21-15.1). In total, there were 294 1st corses with anti-TNF agents, 5 with abatacept and 2 with anakinra. A number of 298 pts were included in the analysis for drug discontinuation (3 pts with no follow-up data after their first dose of BT were excluded). The median follow-up on treatment, before discontinuation due to every cause, was 53.7 months (range 0.45-124.45). One hundred and sixty-five pts discontinued BT: 27 due to disease control, 135 because of failure (78 AEs, 12 LaE and 45 LoE), 3 pts temporarily stopped for pregnancy. Among 135 pts who discontinued for failure, 117 switched to a 2nd BT. Among 27 pts who discontinued due to remission, 13 pts restarted on BT for relapse of disease activity (10 pts restarted with the same BT, 3 switched to a different one). Predictors of discontinuation due to AE were female gender (P=0.01) and longer disease duration (P=0.02). Predictors of discontinuation due to LaE or LoE were systemic onset and polyarthritis FR positive (vs other JIA subtypes) (P<0.05) and use of mAb-anti-TNF (vs sTNFR) (P=0.02). Predictors of discontinuation due to inactive disease were male gender and shorter disease duration (P<0.05). Although only few pts discontinued BT due to a complete and persistent disease control, the majority of them remained on BT for a long time, suggesting that in our cohort of JIA pts, affected by a severe long lasting refractory disease, BT was globally well tolerate and efficacious in controlling the disease.

Axillary dissection versus no axillary dissection in older T1N0 breast cancer patients: 15-year results of trial and out-trial patients.

BACKGROUND: Our randomized trial found no survival advantage for axillary dissection (AD) compared observation only (no AD) in older patients with early breast cancer and a clinically negative axilla, indicating that AD is unnecessary. We compared characteristics and outcomes in out-trial patients with those in trial patients to provide indications as to whether AD can be safely omitted outside the trial setting. METHODS: The trial started in 1996, recruiting 238 patients age 65-80 years with cT1cN0 breast cancer, randomized to conservative surgery with or without AD. Over the recruitment period, 109 eligible patients who refused to participate in the trial, also received conservative breast surgery with or without AD depending on patient preference/surgeon opinion. Trial and out-trial patients received conventionally-fractioned whole breast radiation and tamoxifen for five years. Endpoints were breast cancer mortality, overall survival, and cumulative incidence of axillary disease in patients not receiving AD. RESULTS: After 15 years of follow-up, breast cancer mortality and overall survival did not differ between the AD and no AD arms, in either the trial or out-trial cohorts. The 15-year cumulative incidence of axillary relapse was 6% in the no AD arm of the trial group, and zero in the no AD arm of the out-trial group. CONCLUSIONS: Outside the trial setting, older patients with T1N0 breast cancer can be safely treated by conservative surgery, postoperative radiotherapy and tamoxifen for five years (if ER-positive). Axillary surgery is appropriate only for the small proportion of patients who develop overt axillary disease during follow-up.

Canine perivascular wall tumors: high prognostic impact of site, depth, and completeness of margins.

Canine perivascular wall tumors (cPWTs) arise from vascular mural cells and are included among soft tissue sarcomas (STSs). Most prognostic studies are performed on canine STSs as a general group and regardless of their specific histotype. The aim of this study was to identify pathological parameters and profiles with prognostic impact for cutaneous/subcutaneous cPWTs. Anatomical location, type of growth, surgical margins, and size and depth of the tumor were collected in 56 cPWTs. The association between each pair of variables was evaluated by χ(2) test. Multiple correspondence analysis (MCA) was performed to describe the multivariate association of variables and was followed by cluster analysis to identify specific pathological profiles. The prognostic impact of variables and profiles was assessed by Cox regression model. Size and depth were significantly associated with increased relapse probability. Cases with complete surgical margins did not recur. Other single variables were not significantly associated with relapse. Cluster analysis on MCA considering site, depth, margins, and type of growth identified 3 pathological profiles associated with PWT relapse and having a high prognostic impact. Major prognostic factors for cPWTs were tumor size, depth of growth, and pathological profiles.

Canine cutaneous perivascular wall tumors at first presentation: clinical behavior and prognostic factors in 55 cases.

BACKGROUND: Canine cutaneous perivascular wall tumors (c-PWT) are soft tissue sarcomas recently identified when hemangiopericytomas were reclassified. No previous clinical data are available for c-PWT. HYPOTHESIS/OBJECTIVES: To define the clinical behavior and prognostic role of clinical and pathological variables in a homogeneous population of c-PWT. ANIMALS: Fifty-five c-PWT in 53 client-owned dogs at first presentation undergoing surgery. METHODS: Retrospective case series. The endpoint was the relapse of tumor (local and/or distant). The prognostic values of clinical (age, sex, weight, site and tumor size, adjuvant therapy) and pathological (status of surgical margins, histological grade, mitosis, percentage of tumor necrosis) variables were investigated by univariate and bivariate analyses (P < .05). The pattern of associations between variables was explored by multivariate correspondence analysis (MCA). RESULTS: Twelve dogs had a relapse. Ten dogs had local recurrence, 1 had metastatic disease, and 1 had both. The estimated probability of local recurrence was 0.02, 0.08, 0.20, and 0.24 at 6 months, 1, 2, and 3 years, respectively. Size of the tumor was a significant prognostic factor while status of margins had only a clinically relevant hazard ratio. In MCA evaluation, young age, tumor size (< 5 cm), grade I, and location in the extremities were associated. Association was also observed for older age, tumor size (> 5 cm), grade II, and other location. CONCLUSION AND CLINICAL IMPORTANCE: C-PWT tend to locally recur a long time after surgery. An early diagnosis of c-PWT associated with small tumor size (< 5 cm) and clean surgical margins ensures a good prognosis independently of histological grade.

An immunohistochemically positive E-cadherin status is not always predictive for a good prognosis in human breast cancer.

BACKGROUND: in primary breast cancers dichotomic classification of E-cadherin expression, according to an arbitrary cutoff, may be inadequate and lead to loss of prognostic significance or contrasting prognostic indications. We aimed to assess the prognostic value of high and low E-cadherin levels in a consecutive case series (204 cases) of unilateral node-negative non-lobular breast cancer patients with a 8-year median follow-up and that did not receive any adjuvant therapy after surgery. METHODS: expression of E-cadherin was investigated by immunohistochemistry and assessed according to conventional score (0, 1+, 2+, 3+). Multiple correspondence analysis was used to visualise associations of both categorical and continuous variables. The impact of E-cadherin expression on patients outcome was evaluated in terms of event-free survival curves by the Kaplan-Meier method and proportional hazard Cox model. RESULTS: respect to intermediate E-cadherin expression values (2+), high (3+) or low (0 to 1+) E-cadherin expression levels had a negative prognostic impact. In fact, both patients with a low-to-nil (score 0 to 1+) expression level of E-cadherin and patients with a high E-cadherin expression level (score 3+) demonstrated an increased risk of failure (respectively, hazard ratio (HR)=1.71, confidence interval (CI)=0.72-4.06 and HR=4.22, CI=1.406-12.66) and an interesting association with young age. CONCLUSIONS: the findings support the evidence that high expression values of E-cadherin are not predictive for a good prognosis and may help to explain conflicting evidence on the prognostic impact of E-cadherin in breast cancer when assessed on dichotomic basis.

Hand impairment in systemic sclerosis: association of different hand indices with organ involvement.

OBJECTIVE: To evaluate the association between the assessment tools used to quantify hand impairment and organ involvement in patients with systemic sclerosis (SSc). METHODS: Eighty consecutive SSc patients were assessed for hand impairment using the Hand Anatomic Index (HAI), finger-to-palm distance in flexion (FTP), and the Hand Mobility in Scleroderma (HAMIS) test. Cluster analysis was used to identify patients having similar characteristics on the basis of the pattern of organ involvement in order to create clinically homogeneous groups, and to correlate these clusters with the measures of hand involvement. Finally, we evaluated the discriminating ability of the indices to identify the patients whose clinical condition was more severe. RESULTS: Two major clusters were identified by cluster analysis on the basis of organ involvement. The first (cluster A) included 61 patients and the second (cluster B) 19 patients characterized by minor and major extent of organ involvement, respectively. The extent of organ involvement and the hand impairment were related. The scores of hand indices were lower in cluster B. The area under the receiver operating characteristic (ROC) curve (C-index) for the logistic model including all three indices was 0.85 (95% confidence interval 0.74­0.95). CONCLUSION: The seriousness of hand involvement as measured by the three indices was associated with the extent of organ involvement. Further studies of hand impairment scales are needed to provide validated guidance as meaningful clinical measures.

[Nailfold capillaroscopic report: qualitative and quantitative methods]. / La refertazione dell'esame capillaroscopico: metodi di valutazione qualitativi e quantitativi.

Nailfold capillaroscopy (NVC) is a simple and non-invasive method used for the assessment of patients with Raynaud's phenomenon (RP) and in the differential diagnosis of various connective tissue diseases. The scleroderma pattern abnormalities (giant capillaries, haemorrages and/or avascular areas) have a positive predictive value for the development of scleroderma spectrum disorders. Thus, an analytical approach to nailfold capillaroscopy can be useful in quantitatively and reproducibly recording various parameters. We developed a new method to assess patients with RP that is capable of predicting the 5-year transition from isolated RP to RP secondary to scleroderma spectrum disorders. This model is a weighted combination of different capillaroscopic parameters (giant capillaries, microhaemorrages, number of capillaries) that allows physicians to stratify RP patients easily using a relatively simple diagram to deduce prognosis.

A prediction model for breast cancer recurrence after adjuvant hormone therapy.

Hormone therapy with tamoxifen has long been the established adjuvant treatment for node-positive, estrogen-receptor-positive breast cancer in postmenopausal women. Since 30-40% of these patients fail to respond, reliableoutcome prediction is necessary for successful treatment allocation. Using pathobiological variables (available in mostclinical records: tumor size, nodal involvement, estrogen and progesterone receptor content) from 596 patients recruitedat a comprehensive cancer center, we developed a prediction model which we validated in an independent cohort of 175patients recruited at a general hospital. Calculated at 3 and 4 years of follow-up, the discrimination indices were 0.716[confidence limits (CL) 0.641, 0.752] and 0.714 (CL 0.650, 0.750) for the training data, and 0.726 (CL 0.591, 0.769) and0.677 (CL 0.580, 0.745) for the testing data. Waiting for more effective approaches from genomic and proteomic studies, amodel based on consolidated pathobiological variables routinely assessed at relatively low costs may be considered as thereference for assessing the gain of new markers over traditional ones, thus substantially improving the conventional use ofprognostic criteria.

Double-blind evaluation and benchmarking of survival models in a multi-centre study.

Accurate modelling of time-to-event data is of particular importance for both exploratory and predictive analysis in cancer, and can have a direct impact on clinical care. This study presents a detailed double-blind evaluation of the accuracy in out-of-sample prediction of mortality from two generic non-linear models, using artificial neural networks benchmarked against a partial logistic spline, log-normal and COX regression models. A data set containing 2880 samples was shared over the Internet using a purpose-built secure environment called GEOCONDA (www.geoconda.com). The evaluation was carried out in three parts. The first was a comparison between the predicted survival estimates for each of the four survival groups defined by the TNM staging system, against the empirical estimates derived by the Kaplan-Meier method. The second approach focused on the accurate prediction of survival over time, quantified with the time dependent C index (C(td)). Finally, calibration plots were obtained over the range of follow-up and tested using a generalization of the Hosmer-Lemeshow test. All models showed satisfactory performance, with values of C(td) of about 0.7. None of the models showed a systematic tendency towards over/under estimation of the observed survival at tau=3 and 5 years. At tau=10 years, all models underestimated the observed survival, except for COX regression which returned an overestimate. The study presents a robust and unbiased benchmarking methodology using a bespoke web facility. It was concluded that powerful, recent flexible modelling algorithms show a comparative predictive performance to that of more established methods from the medical and biological literature, for the reference data set.