Predicting recurrence of venous thromboembolism in anticoagulated cancer patients using real-world data and machine learning

e18742

Background: The clinical predictors of venous thromboembolism (VTE) recurrence in patients with cancer are not well known. Our objective was to develop a predictive model for risk of VTE recurrence in anticoagulant-treated cancer patients within the first 6 months following VTE diagnosis. Methods: Using the EHRead technology, based on Natural Language Processing (NLP) and machine learning (ML), the unstructured clinical data in EHRs from 9 Spanish hospitals between 2014 and 2018 was extracted and analyzed. The study population, comprising all adult anticoagulated cancer patients with VTE, was downsampled to prevent bias and class imbalance. A total of 94 patient characteristics were explored, and Random Forest (RF) feature selection was performed to identify the most relevant predictors for VTE recurrence. Multiple algorithms were used to train different prediction models, which were subsequently validated in a hold-out dataset. The model with the best performance metrics (i.e., ROC-AUC) was selected as the final model. Results: From a source population of 2,893,208 patients, 21,227 anticoagulant-treated patients with VTE and active cancer (53.9% male, median age of 70 years) were identified. Across the study period, yearly incidence of VTE remained relatively stable, ranging from 2.7 to 3.9%. The most common type of VTE was deep vein thrombosis (68.2% of patients), followed by pulmonary embolism (28.4%). Regarding primary cancer locations, the most frequent were colorectal (10.1%) and lung cancer (8.5%). Of all trained and validated models, the RF approach yielded the best performance, with a ROC-AUC = 0.72. The following predictors of VTE recurrence were identified: pulmonary embolism, deep vein thrombosis, metastasis, adenocarcinoma, hemoglobin values, serum creatinine values, platelet count, leukocyte count, family history of VTE, and patients’ age. Conclusions: Using NLP and ML, we were able to use the real-world data in EHRs to build a predictive model of VTE recurrence in cancer patients based on individual clinical features. These results may improve the clinical management of VTE recurrence in this population.

CORONET; COVID-19 in Oncology evaluatiON Tool: Use of machine learning to inform management of COVID-19 in patients with cancer

1505

Background: Patients (pts) with cancer are at increased risk of severe COVID-19 infection and death. Due to COVID-19 outcome heterogeneity, accurate assessment of pts is crucial. Early identification of pts who are likely to deteriorate allows timely discussions regarding escalation of care. Likewise, safe home management will reduce risk of nosocomial infection. To aid clinical decision-making, we developed a model to help determine which pts should be admitted vs. managed as an outpatient and which pts are likely to have severe COVID-19. Methods: Pts with active solid or haematological cancer presenting with symptoms/asymptomatic and testing positive for SARS-CoV-2 in Europe and USA were identified following institutional board approval. Clinical and laboratory data were extracted from pt records. Clinical outcome measures were discharge within 24 hours, requirement for oxygen at any stage during admission and death. Random Forest (RF) algorithm was used for model derivation as it compared favourably vs. lasso regression. Relevant clinical features were identified using recursive feature elimination based on SHAP. Internal validation (bootstrapping) with multiple imputations for missing data (maximum ≤2) were used for performance evaluation. Cost function determined cut-offs were defined for admission/death. The final CORONET model was trained on the entire cohort. Results: Model derivation set comprised 672 pts (393 male, 279 female, median age 71). 83% had solid cancers, 17% haematological. Predictive features were selected based on clinical relevance and data availability, supported by recursive feature elimination based on SHAP. RF model using haematological cancer, solid cancer stage, no of comorbidities, National Early Warning Score 2 (NEWS2), neutrophil:lymphocyte ratio, platelets, CRP and albumin achieved AUROC for admission 0.79 (+/-0.03) and death 0.75 (+/-0.02). RF explanation using SHAP revealed NEWS2 and C-reactive protein as the most important features predicting COVID-19 severity. In the entire cohort, CORONET recommended admission of 96% of patients requiring oxygen and 99% of patients who died. We then built a decision support tool using the model, which aids clinical decisions by presenting model predictions and explaining key contributing features. Conclusions: We have developed a model and tool available at https://coronet.manchester.ac.uk/ to predict which pts with cancer and COVID-19 require hospital admission and are likely to have a severe disease course. CORONET is being continuously refined and validated over time.

COVID-19 mortality in cancer patients in a Madrid hospital during the first 3 weeks of the epidemic

Background and objective

The Covid-19 pandemic especially affects cancer patients with higher incidence and mortality according to published series of original pandemic foci. The study aims to determine the mortality in our center due to covid-19 in cancer patients during the first 3 weeks of the epidemic.

Material and methods

The cancer patients who died of covid-19 during the analysis period have been reviewed describing the oncological and the covid-19 infection characteristics and the treatments established.

Results

Confirmed cases covid-19: 1069 with 132 deaths (12.3%). With cancer 36 patients (3.4%), 15 deceased (41.6%). Of the deceased, only 6 patients (40%) were in active treatment. The most frequent associated tumor was lung (8/15 patients, 53.3%), 11 with metastatic disease (11/15, 73.3%). No specific treatment was established in 40 % (6/15) of the patients. The rest of them received treatments with the active protocols.

Conclusion

Covid-19 mortality in cancer patients is almost four times higher than that of the general population. Until we have effective treatments or an effective vaccine, the only possibility to protect our patients is to prevent the infection with the appropriate measures.

The systemic inflammation response index (SIRI) predicts oncological outcome and correlates with tumor burden in metastatic pancreatic cancer

e15537

Background: The SIRI, defined by neutrophil x monocyte/lymphocyte 109/L, has recently emerged as a prognostic factor for pancreatic cancer. However, the association between SIRI values after chemotherapy and tumor response has not been evaluated. Methods: 161 metastatic pancreatic cancer patients were retrospectively analyzed. Associations between overall survival (OS), chemotherapy and SIRI were analyzed. A larger number of patients with pre-treatment SIRI (pre-SIRI) were collected so, post-treatment SIRI (post-SIRI) evaluated after three cycles of chemotherapy, was adjusted for analysis. Results: Median age 66 years. 59 (36%) received gemcitabine + nab-paclitaxel, 40 (24%) gemcitabine, 22 (17%) mFOLFIRINOX, 13 (7%) other regimens. 27 (16%) had not received treatment. Pre-SIRI≥2.3×109/L was an independent, negative predictor of OS compared to pre-SIRI < 2.3×109/L [5 versus 16 months, Hazard Ratio (HR) 2.87, Confidence Interval (CI) 95% 2.02-4.07, P< 0.0001]. In the whole cohort, we observed SIRI values increased after treatment (median pre-SIRI: 1.6×109/L; post-SIRI: 2.3×109/L; P= 0.007). Thus, we analyzed the association between tumor response measured by RECIST and pre-SIRI and post- SIRI values. Patients with progressive disease (PD) showed a higher pre-SIRI than those who had a response to chemotherapy (2.7×109/L versus 1.2×109/L, respectively; P< 0.001). We also observed a statistically significant increase in post-SIRI values for PD compared to tumor response (3.2×109/L versus 1.7×109/L, respectively; P= 0.012). As observed for pre-SIRI, post-SIRI ≥2.3×109/L showed a shorter OS compared to post-SIRI < 2.3×109/L (8 versus 17 months, respectively; P= 0.016). Furthermore, patients with pre-SIRI ≥2.3×109/L were more likely to benefit from mFOLFIRINOX showing a median OS of 17 months compared to 6 and 4 months for gemcitabine + nab-paclitaxel and gemcitabine, respectively ( P< 0.001). Conversely, there was no difference for pre-SIRI < 2.3×109/L: 15.9 months versus 16.5 and 16, respectively. Conclusions: SIRI≥2.3×109/L was a prognostic factor for metastatic pancreatic cancer. An elevated post-SIRI showed an association with disease progression and a negative impact on survival. Therefore, an increase in SIRI could be related to high tumor burden and be useful to appropriately select patients who would benefit of a more intensive first-line chemotherapy regimen.

Identifying burnout in young oncologists: The sooner the better

11010

Background: Professional burnout is an important issue for the healthcare systems with potentially relevant consequences in the quality of patient care. Young oncologists are a special risk population, due to high workload, academic pressure and other specific factors related to cancer care. Work life and lifestyle factors are related with burnout levels and may define specific interventions to reduce and prevent burnout. Methods: A survey based on the validated Maslach Burnout Inventory Human Services Survey for Medical Personnel (MBI-HSS MP) was conducted. Additional questions to explore work life and lifestyle factors were included. We studied potential burnout amongst young Spanish oncologists (residents and oncologist in the first 5 years of professional performance). Statistical analyses, including linear regression, were carried out to test the relations between overall burnout score and work and lifestyle factors. Results: We obtained 243 responses to the survey. The sample was representative in terms of age, sex, geographic region, professional profile (residents and young oncologists) and a homogeneous distribution per year of residence (final participation rate was of 26.6%). Regarding 5 Burnout profiles, 32.1% of the participants were classified as Ineffective, 25.1% as Burnout, 21% as Engaged, 17.3% as Overextended and 4.5% as Disengaged. Percentage of Burnout profile is higher in medical oncology residents (28.24%) than in young oncologists (19.79%). The highest percentage of burnout profile was observed in the second-year residents (35,71%) and gradually decreases, in parallel to an increase in the engaged profile. The multivariable linear regression analysis showed a significant correlation between not having a good work life balance and adequate vacation time and the burnout score. Conclusions: Burnout affects a significant number of young oncologists with significantly different profiles and differences across regions. Adapted interventions to the most frequent profiles and at different stages of the training and professional career may be necessary particularly at the early beginning. Actions towards achieving a better work and personal life balance and stress management could be effective. These results are the basis for the prospective part of our study that aims to design an intervention and to assess its efficacy in this population.

[COVID-19 mortality in cancer patients in a Madrid hospital during the first 3 weeks of the epidemic]

Antecedentes y objetivo

La pandemia por covid-19 afecta especialmente a pacientes con cáncer, con mayor incidencia y mortalidad según series publicadas de focos originales de pandemia. El estudio pretende conocer la mortalidad en nuestro centro por covid-19 en pacientes con cáncer durante las primeras 3 semanas de epidemia.

Material y métodos

Se han revisado los pacientes con cáncer fallecidos por covid-19 durante el periodo de análisis, describiendo las características oncológicas de la infección por covid-19 y los tratamientos instaurados.

Resultados

Casos confirmados por covid-19: 1.069 con 132 fallecimientos (12,3%). Con cáncer 36 pacientes (3,4%), 15 fallecidos (41,6%). De los fallecidos solo 6 pacientes (40%) se encontraban en tratamiento activo. El tumor más frecuente asociado fue pulmón (8/15 pacientes, 53,3%), 11 con enfermedad metastásica (11/15, 73,3%). El 40% (6/15) no recibió tratamiento específico contra covid-19, el resto fue tratado con los protocolos activos.

Conclusión

La mortalidad por covid-19 en pacientes con cáncer casi cuadriplica la de la población general. Hasta disponer de tratamientos eficaces o una vacuna efectiva la única posibilidad de proteger a nuestros pacientes es impedir el contagio con las medidas adecuadas.

Association of efficacy of anti-PD-1 immunotherapy in patients with advanced cancer with immune-related adverse events (irAEs)

e14064

Background: Cancer immune therapy has shown benefit in many cancer types, although it may initiate autoimmune-related disorders in some patients. We have attempted to establish whether the incidence of irAEs after the use of nivolumab or pembrolizumab in advanced malignancies is associated with anti-PD-1 treatment efficacy. Methods: We retrospectively reviewed all clinical records of patients treated with single-agent nivolumab or pembrolizumab for advanced cancer from january 2016 until september 2018. IrAEs were identified clinically and graded as per the CTCAE v.4.0. Efficacy was evaluated with objective response rate (ORR, iRECIST criteria), progression-free survival (PFS) and overall survival (OS). Uni and multivariate tests were performed to determine the association between irAEs and ORR, PFS or OS. Results: We identified 110 patients. Primary diagnoses were lung cancer (n = 78), melanoma (n = 10), head and neck carcinoma (n = 7), renal carcinoma (n = 5), Hodgkin's lymphoma (n = 3), urothelial carcinoma (n = 3) and gallbladder adenocarcinoma, hepatocellular carcinoma, gastric adenocarcinoma and Merkel cell carcinoma (n = 1 each). IrAEs were observed in 40 patients (36.4%). The most frequent irAEs were hypothyroidism (n = 15), nephritis (n = 5) and hyperthyroidism (n = 4). Objective response was observed in 45 patients (40.9%), median PFS was 5 months (0.5-32 months) and median overall survival was 24.5 months (3-110 months). Thirty-three of the 40 patients with irAEs had objective response (82.5%) in contrast with 12 of the 70 cases without irAEs (17%) (OR 22.78, 95% CI 5.9-87.0, P < 0.000001). PFS in patients with irAEs was 10 months and 3 months in those without irAEs (HR 2.4, P = 0.004). OS in patients with irAEs was 32 months and 22 in those without irAEs, without statistically significant differences. Conclusions: In advanced cancer treated with nivolumab or pembrolizumab, patients who developed irAEs showed a significantly improved efficacy (ORR of 82.5% and PFS of 10 months) over patients without irAEs (ORR of 17% and PFS of 3 months).