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

      Lee, Rebecca J; Wysocki, O.; Zhou, Cong; Calles, A.; Eastlake, L.; Ganatra, S.; Harrison, M.; Horsley, L.; Huddar, P.; Khan, K.; et al. (2021)
      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 athttps://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.
    • Development of a model to predict hospital admission and severe outcome in cancer patients with COVID-19

      Lee, Rebecca J; Zhou, Cong; Shotton, R; Tivey, Ann; Dickens, E.; Huddar, P.; McKenzie, H.; Boyce, H.; Maynard, A.; Rowe, M. P.; et al. (2020)
      Background: Patients (pts) with cancer are at increased risk of severe COVID-19 infection and death. Due to the heterogeneity of manifestations of COVID-19, accurate assessment of patients presenting to hospital is crucial. Early identification of pts who are likely to deteriorate allows timely discussions regarding escalation of care. It is equally important to identify pts who could be safely managed at home. To aid clinical decision making, we developed a model to determine which pts should be admitted vs. discharged at presentation to hospital. Methods: Consecutive pts with solid or haematological malignancies presenting with symptoms who tested positive for SARS-CoV-2 at 10 UK hospitals from March-May 2020 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. The associations between clinical features and outcomes were examined using ANOVA or Chi-squared tests. A logistic model was developed using clinical features with p<0.05 to predict patients who need hospital admission. Results: 52 pts were included (27 male, 25 female; median age 63). 80.5% pts had solid cancers, 19.5% haematological. Association analysis indicated that smoking status, prior cancer therapy and comorbidities had no significant association with outcomes. A number of other factors presented in the table had significant associations. A multivariate logistic regression model was generated to predict need for admission to hospital. Of note, age and male sex lost significance in the multivariate model (p>0.8). Using haematological cancer, NEWS2 score, dyspnoea, CRP and albumin, the model predicted requirement for admission with an area under the curve of 0.88. Conclusions: We have developed a model to predict which pts require hospital admission. Further refinement and validation in larger cohorts of pts will be presented
    • Longitudinal characterisation of haematological and biochemical parameters in cancer patients prior to and during COVID-19 reveals features associated with outcome

      Lee, Rebecca J; Wysocki, O; Bhogal, T.; Shotton, R; Tivey, Ann; Angelakas, A.; Aung, T.; Banfill, K.; Baxter, M.; Boyce, H.; et al. (2020)
      Background: Cancer patients are at increased risk of death from severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Cancer and its treatment affect many haematological and biochemical parameters, therefore we analysed these prior to and during coronavirus disease 2019 (COVID-19) and correlated them with outcome. Patients and methods: Consecutive patients with cancer testing positive for SARS-CoV-2 in centres throughout the United Kingdom were identified and entered into a database following local governance approval. Clinical and longitudinal laboratory data were extracted from patient records. Data were analysed using Mann-Whitney U test, Fisher's exact test, Wilcoxon signed rank test, logistic regression, or linear regression for outcomes. Hierarchical clustering of heatmaps was performed using Ward's method. Results: In total, 302 patients were included in three cohorts: Manchester (n = 67), Liverpool (n = 62), and UK (n = 173). In the entire cohort (N = 302), median age was 69 (range 19-93 years), including 163 males and 139 females; of these, 216 were diagnosed with a solid tumour and 86 with a haematological cancer. Preinfection lymphopaenia, neutropaenia and lactate dehydrogenase (LDH) were not associated with oxygen requirement (O2) or death. Lymphocyte count (P < 0.001), platelet count (P = 0.03), LDH (P < 0.0001) and albumin (P < 0.0001) significantly changed from preinfection to during infection. High rather than low neutrophils at day 0 (P = 0.007), higher maximal neutrophils during COVID-19 (P = 0.026) and higher neutrophil-to-lymphocyte ratio (NLR; P = 0.01) were associated with death. In multivariable analysis, age (P = 0.002), haematological cancer (P = 0.034), C-reactive protein (P = 0.004), NLR (P = 0.036) and albumin (P = 0.02) at day 0 were significant predictors of death. In the Manchester/Liverpool cohort 30 patients have restarted therapy following COVID-19, with no additional complications requiring readmission. Conclusion: Preinfection biochemical/haematological parameters were not associated with worse outcome in cancer patients. Restarting treatment following COVID-19 was not associated with additional complications. Neutropaenia due to cancer/treatment is not associated with COVID-19 mortality. Cancer therapy, particularly in patients with solid tumours, need not be delayed or omitted due to concerns that treatment itself increases COVID-19 severity.
    • Palbociclib combined with aromatase inhibitors (AIs) in women >= 75 years with oestrogen receptor positive (ER plus ve), human epidermal growth factor receptor 2 negative (HER2-ve) advanced breast cancer: A real-world multicentre UK study

      El Badri, S.; Tahir, B.; Balachandran, K.; Bezecny, P.; Britton, Fiona; DeSouza, K.; Hills, D.; Moe, M.; Pigott, T.; Proctor, A.; et al. (2021)
      Background Breast cancer accounts for 21% of all cancer diagnoses in women aged ?75 years. The older population is under-represented in clinical trials; thus, real-world data in this patient group is critical to guide management. In this large-scale UK-wide real-world study, we evaluated the tolerability and efficacy of palbociclib combined with an AI for first-line treatment of advanced ER+ve/HER2-ve breast cancer in elderly women. Methods 14 cancer centres participated in this national retrospective study. Patients aged ?75 years who received at least one cycle of palbociclib combined with an AI for first-line treatment of advanced ER+ve/HER2-ve breast cancer were eligible. Data included baseline demographics, co-morbidities, metastatic disease burden, toxicities, dose reductions and delays, response to treatment and in-patient secondary care burden. Multivariable Cox regression was used to assess independent predictors of progression-free survival (PFS). Results 276 patients met the eligibility criteria. The median age of patients was 78 (range 75-92) years. The PFS rates at 12 and 24 months were 75.9% and 64.9%, respectively. The best radiological response was complete response (2%), partial response (32.9%) and stable disease (54.9%) with a clinical benefit rate at 24 weeks of 87%. The most common toxicities were neutropenia, fatigue, anaemia and thrombocytopenia. 50.7% of patients required a dose reduction and 59.2% required at least one dose delay. 22 patients (9.6%) required hospital admission due to toxicity and 6 patients (2.2%) had febrile neutropenia. Multivariable analysis identified fewer dose delays, increasing ECOG performance status and age-adjusted Charlson co-morbidity index, and increasing number of metastatic sites to be independent adverse predictors of PFS. Conclusions This largest known dataset of Palbociclib tolerability and efficacy in women aged ?75 years shows that this is an effective therapy that is well tolerated and appropriately managed with dose delays/reductions resulting in very low levels of clinically significant toxicity requiring hospital admission.
    • Palbociclib in combination with aromatase inhibitors in patients ≥ 75 years with oestrogen receptor-positive, human epidermal growth factor receptor 2 negative advanced breast cancer: A real-world multicentre UK study

      El Badri, S.; Tahir, B.; Balachandran, K.; Bezecny, P.; Britton, Fiona; Davies, M.; Desouza, K.; Dixon, S.; Hills, D.; Moe, M.; et al. (2021)
      Background: Breast cancer incidence increases with age and real-world data is essential to guide prescribing practices in the older population. The aim of this study was to collect large scale real-world data on tolerability and efficacy of palbociclib + AI in the first line treatment of ER+/HER2-advanced breast cancer in those aged ≥75 years. Methods: 14 cancer centres participated in this national UK retrospective study. Patients aged ≥75 years treated with palbociclib + AI in the first line setting were identified. Data included baseline demographics, disease characteristics, toxicities, dose reductions and delays, treatment response and survival data. Multivariable Cox regression was used to assess independent predictors of PFS, OS and toxicities. Results: 276 patients met the eligibility criteria. The incidence of febrile neutropenia was low (2.2%). The clinical benefit rate was 87%. 50.7% of patients had dose reductions and 59.3% had dose delays. The 12- and 24- month PFS rates were 75.9% and 64.9%, respectively. The 12- and 24- month OS rates were 85.1% and 74.0%, respectively. Multivariable analysis identified PS, Age-adjusted Charlson Comorbidity Index (ACCI) and number of metastatic sites to be independent predictors of PFS. Dose reductions and delays were not associated with adverse survival outcomes. Baseline ACCI was an independent predictor of development and severity of neutropenia. Conclusion: Palbociclib is an effective therapy in the real-world older population and is well-tolerated with low levels of clinically significant toxicities. The use of geriatric and frailty assessments can help guide decision making in these patients.
    • A randomised phase IB/IIA study of CApecitabine plus Radium-223 in breast cancer patients with BONe metastases (CARBON) - Safety and preliminary efficacy findings

      Winter, M.; Kendall, J.; Brown, S.; Rathbone, E.; Wilson, C.; Howell, Sacha J; Twelves, C.; Palmieri, C.; Anand, A.; MacPherson, I.; et al. (2021)
      Background: Bone metastases (BMs) occur in approximately 70% of patients (pts) with metastatic breast cancer (MBC). Despite significant advances in the management of BMs with bone-targeted agents and the associated reduction in skeletal-related events, there remains an unmet need for further treatment options to improve median overall survival beyond 2-3 years. Radium-223 [R] dichloride is an alpha-emitting radiopharmaceutical that is avidly taken up, like calcium, into the bone where it emits high-energy, short-range alpha-particles resulting in a targeted anti-tumour effect on BMs. Combining R with current systemic therapy could potentially enhance efficacy in MBC with BMs. Methods: CARBON is a UK, open-label, multi-centre phase IB/IIA study evaluating the combination of capecitabine [C] (1000mg/m2 bd days 4-17, 12x21 day cycles) with R 55kBq/kg day 1 given on a 6-weekly schedule in pts with BMs from MBC (+/- other sites of disease) with ≥2 bone lesions on radionuclide bone scan and/or ≥1 lesion confirmed on plain radiographs, CT or MRI. Other eligibility criteria included ECOG PS 0-2, ≤ 2 lines of chemotherapy for MBC and current use of a bisphosphonate / denosumab for ≥6 weeks. To establish the feasibility and safety of C+R the phase IB opened in August 2016 registering 6 pts; the primary endpoint was dose-limiting toxicities (DLTs), defined as ≥grade 3 gastrointestinal toxicity lasting >48 hours or ≥grade 4 haematological toxicity lasting >7 days. Subsequently, between April 2017 and March 2019 28 pts were randomised (2:1) to C+R vs C in phase IIa to further characterise the safety profile, with frequency of CTC grade 3-4 toxicities and diarrhoea as primary endpoints. Preliminary evaluation of efficacy through assessment of bone turnover marker changes from baseline to end of cycle 5 and time to progression in bone and overall was made. Results: Baseline clinico-pathologic characteristics and prior treatments were well balanced between the arms; 13 C+R and 9 C pts had visceral metastases. There were 0 DLTs in the 6 phase IB pts, therefore the same C+R dose and schedule was studied in phase IIA. 2 pts randomised to C+R received C alone and are included in the C arm. The safety population consists of 34 pts (23 C+R, 11 C). Median number of cycles received was 8.5 (range 3-12) in C+R arm and 12 (range 1-12) in C arm. 38/307 (12%) treatment cycles were delayed (25 [13%] C+R arm, 13 [12%] C arm). 11 (48%) C+R and 6 (55%) C pts had a permanent C dose reduction. 94/95 (99%) prescribed R cycles were administered. 9 (39%) C+R and 9 (82%) C pts completed all 12 cycles. Other reasons for discontinuation were: progressive disease in 12 (52%) C+R and 0 in C pts; toxicity in 1 (4%) C+R and 1 (9%) C pt; clinician decision in 1 (9%) C pt; progressive disease and toxicity in 1 (4%) C+R pt. Only 25/575 (4%) reported AEs were grade 3-4 (n=21 in 11 [48%] C+R pts, n=4 in 4 [36%] C pts) with 0 episodes of grade 3-4 diarrhoea. Table 1 shows maximum grades of diarrhoea and haematological AEs experienced by arm. 18 SAEs occurred (n=11 in 8 C+R pts, n=7 in 2 C pts). 8 (44%) SAEs were grade 3 (C+R: 6, C: 2); none were related to diarrhoea. There were 0 SUSARs. Conclusion: In the first completed trial evaluating R with chemotherapy in MBC pts, the combination of C+R is safe and well-tolerated. Preliminary efficacy analyses including bone markers are ongoing and will be presented at the meeting. The creation of the data was supported in part by Bayer Plc and Yorkshire Cancer Research.