Development of a predictive model of venous thromboembolism recurrence in anticoagulated cancer patients using machine learning.

INTRODUCTION: Patients with cancer and venous thromboembolism (VTE) show a high risk of VTE recurrence during anticoagulant treatment. This study aimed to develop a predictive model to assess the risk of VTE recurrence within 6 months at the moment of primary VTE diagnosis in these patients. MATERIALS AND METHODS: Using the EHRead® technology, based on Natural Language Processing (NLP) and machine learning (ML), the unstructured data in electronic health records from 9 Spanish hospitals between 2014 and 2018 were extracted. Both clinically- and ML-driven feature selection were performed to identify predictors for VTE recurrence. Logistic regression (LR), decision tree (DT), and random forest (RF) algorithms were used to train different prediction models, which were subsequently validated in a hold-out data set. RESULTS: A total of 16,407 anticoagulated cancer patients with diagnosis of VTE were identified (54.4 % male and median age 70). Deep vein thrombosis, pulmonary embolism and metastases were observed in 67.2 %, 26.6 %, and 47.7 % of the patients, respectively. During the study follow-up, 11.4 % of the patients developed a recurrent VTE, being more frequent in patients with lung cancer. Feature selection and model training based on ML identified primary pulmonary embolism, deep vein thrombosis, metastasis, adenocarcinoma, hemoglobin and serum creatinine levels, platelet and leukocyte count, family history of VTE, and patients' age as predictors of VTE recurrence within 6 months of VTE diagnosis. The LR model had an AUC-ROC (95 % CI) of 0.66 (0.61, 0.70), the DT of 0.69 (0.65, 0.72) and the RF of 0.68 (0.63, 0.72). CONCLUSIONS: This is the first ML-based predictive model designed to predict 6-months VTE recurrence in patients with cancer. These results hold great potential to assist clinicians to identify the high-risk patients and improve their clinical management.

Polygenic risk score across distinct colorectal cancer screening outcomes: from premalignant polyps to colorectal cancer.

BACKGROUND: Different risk-based colorectal cancer (CRC) screening strategies, such as the use of polygenic risk scores (PRS), have been evaluated to improve effectiveness of these programs. However, few studies have previously assessed its usefulness in a fecal immunochemical test (FIT)-based screening study. METHODS: A PRS of 133 single nucleotide polymorphisms was assessed for 3619 participants: population controls, screening controls, low-risk lesions (LRL), intermediate-risk (IRL), high-risk (HRL), CRC screening program cases, and clinically diagnosed CRC cases. The PRS was compared between the subset of cases (n = 648; IRL+HRL+CRC) and controls (n = 956; controls+LRL) recruited within a FIT-based screening program. Positive predictive values (PPV), negative predictive values (NPV), and the area under the receiver operating characteristic curve (aROC) were estimated using cross-validation. RESULTS: The overall PRS range was 110-156. PRS values increased along the CRC tumorigenesis pathway (Mann-Kendall P value 0.007). Within the screening subset, the PRS ranged 110-151 and was associated with higher risk-lesions and CRC risk (ORD10vsD1 1.92, 95% CI 1.22-3.03). The cross-validated aROC of the PRS for cases and controls was 0.56 (95% CI 0.53-0.59). Discrimination was equal when restricted to positive FIT (aROC 0.56), but lower among negative FIT (aROC 0.55). The overall PPV among positive FIT was 0.48. PPV were dependent on the number of risk alleles for positive FIT (PPVp10-p90 0.48-0.57). CONCLUSIONS: PRS plays an important role along the CRC tumorigenesis pathway; however, in practice, its utility to stratify the general population or as a second test after a FIT positive result is still doubtful. Currently, PRS is not able to safely stratify the general population since the improvement on PPV values is scarce.

Tumour characteristics and survivorship in a cohort of breast cancer: the MCC-Spain study.

PURPOSE: The objective of this study is to analyse the relative survival with breast cancer in women diagnosed after new treatments were generalised and to ascertain the current effect that tumour characteristics such as grade, stage or subtype have on survival as well as the new AJCC-pathological prognostic score. METHODS: The breast cancer MCC-Spain follow-up study is a prospective cohort study of 1685 incident breast cancer cases. Women between 20 and 85 years old were recruited between the years 2008 and 2013 in 18 hospitals located in 10 Spanish provinces and they have been followed until 2017/2018. Relative survival was estimated after 3, 5 and 8 years of follow-up using Ederer II method. In addition, Weibull regression adjusted by age, hospital, grade and stage was used to investigate prognosis factors. RESULTS: Among components of TNM staging system, tumour size greater than 50 mm (i.e. T3 or T4) more than doubled the risk of dying, while N3 nodal involvement and presence of metastasis had a huge effect on mortality. The AJCC pathological prognostic score strongly correlated with survival; thus, hazard ratios increased as the score rose, being 2.31, 4.00, 4.94, 7.92, 2.26, 14.9 and 58.9 for scores IB, IIA, IIB, IIIA, IIIB, IIIC and IV, respectively. CONCLUSION: Both TNM staging and histological/molecular biomarkers are associated with overall survival in Spanish women with breast cancer; when both are combined in the AJCC pathological prognosis score, the prognostic value improved with risk indices that increased rapidly as the pathological prognosis score increased.

Factors Associated with the Selection of First-line Bevacizumab plus Chemotherapy and Clinical Response in HER2-negative Metastatic Breast Cancer: ONCOSUR AVALOX Study.

AIM: To evaluate factors associated with the selection of first-line bevacizumab plus chemotherapy and clinical response in HER2-negative metastatic breast cancer (MBC) in clinical practice in Spain. PATIENTS AND METHODS: All consecutive adult female patients with HER2-negative MBC who had received first-line bevacizumab plus chemotherapy for at least 3 months were enrolled in the present study. RESULTS: A total of 292 evaluable patients were included; 25% had triple-negative breast cancer (TNBC) and 75% had hormone receptor-positive breast cancer (HRPBC). Nearly 40% of patients had ≥3 metastatic sites, mainly located in the bone (48%) and liver (40%). Bevacizumab was mostly combined with paclitaxel (67.1%). ER-positive tumors were only identified as an independent factor associated with the choice of treatment (odds ratio (OR): 0.538; p=0.02). The overall response rate (ORR) was 63.7% (TNBC: 57.5%; HRPBC: 65.9%). Patients aged 36-50 years (OR: 3.03; p=0.028) and those with metastases at sites other than the bone (OR: 0.38; p=0.001) and ≥3 metastatic sites (OR: 1.41; p=0.018) were more likely to achieve objective responses. CONCLUSION: First-line bevacizumab plus chemotherapy, mainly paclitaxel, is an effective and well-tolerated treatment option for HER2-negative MBC, particularly in more aggressive disease.

Final results of an international retrospective observational study in patients with advanced breast cancer treated with oral vinorelbine-based chemotherapy.

BACKGROUND: Full oral chemotherapy (CT) is an active and convenient therapeutic option for patients with advanced breast cancer (ABC). In this retrospective analysis, we reviewed the characteristics and the outcome of patients treated by oral vinorelbine either as a single-agent or in combination with capecitabine as first- or second-line CT in the metastatic setting. PATIENTS AND METHODS: We analysed 216 patients with ABC who started treatment with a full oral CT at 13 centers and seven countries between 2006 and 2008. To be eligible, patients must have received either as a first-(56%) or second-line (44%) therapy oral vinorelbine as a single-agent (54%) or in combination with capecitabine (46%). RESULTS: Main patients' characteristics in the full population (n=216): median age (range): 61 (32-87) years; categories of age: <50 years: 18%, 50-65 years: 44%, ≥ 65 years: 38%; hormone receptor-positive: 63%; ≥ 2 metastatic sites: 58%; visceral metastases: 49%; prior CT: 86%; prior CT for ABC: 44%; prior anthracycline treatment: 69%; prior taxane treatment: 43%, prior anthracycline plus taxane: 38%; prior endocrine therapy: 63%. Median number of cycles: 6 (range=1-54); 48% of patients received more than 6 cycles. G3/4 toxicities: neutropenia 8%, anaemia 2%, thrombocytopenia 1%, febrile neutropenia/neutropenic infection 2%, nausea 6%, vomiting 4%, diarrhea 6%, fatigue 6%, hand-foot syndrome 14% (combination with capecitabine), neuropathy 1%, alopecia (grade 2) 1%. EFFICACY: disease control was achieved in 77% of patients [95% confidence interval=71-83%], 74% as single-agent, 81% in combination, 82% in first-line, 71% in second-line. Median progression-free survival was 9.7 months [95% confidence interval=8.2-12.6 months] in first-line and 6.6 months [95% confidence interval=5.5-8.5 months] in second-line therapy. Caregivers described these oral regimens as convenient (81%), well-tolerated (84%) and with a good compliance by patients (76%). CONCLUSION: These data from everyday practice confirm, as shown in different clinical trials, that oral vinorelbine is an active and well-tolerated CT for ABC, either as a first- or second-line in patients pre-treated with anthracyclines or taxanes. The convenience of its oral administration in association with its good tolerance profile, allows for continuation of treatment until disease progression without a pre-planned maximum of cycles.

Use of biosimilar epoetin to increase hemoglobin levels in patients with chemotherapy-induced anemia: real-life clinical experience.

AIM: To evaluate the effectiveness of a biosimilar erythropoiesis-stimulating agent (Binocrit) for the treatment of patients with cancer and chemotherapy-induced anemia in real-life clinical practice. MATERIALS & METHODS: Data were collected retrospectively from patients at five European centers (in France, Italy, The Netherlands, Romania and Spain) who received treatment with Binocrit. Hemoglobin (Hb) levels were recorded at regular intervals during Binocrit therapy for up to 26 weeks. Hb response (an increase of ≥ 1 g/dl in 4 weeks or a Hb level in the range 10-12 g/dl during the study) was assessed in patients with a Hb level ≥ 8.5 g/dl at the start of therapy who received treatment for at least 6 weeks. Hb response rates in patients who did and did not receive intravenous (iv.) iron were also assessed, and data on any serious unexpected adverse events were collected. RESULTS: Among evaluable patients (n = 113), 79% achieved a Hb response. Response rates were similar among evaluable patients who received an initial Binocrit dose of 30,000 or 40,000 IU/week (81 vs 78%; p = not significant). The Hb response rate was significantly greater in patients who received iv. iron than in patients who did not receive iv. iron (93 vs 77%; p < 0.05). No serious unexpected adverse events were reported. CONCLUSION: Use of the biosimilar erythropoiesis-stimulating agent Binocrit is effective and safe for the treatment of patients with cancer and chemotherapy-induced anemia. Supplementation with iv. iron increases the response rate compared with oral or no iron supplementation.