Search

Background: Histopathology assessments of cancers require highly skilled pathologists, are labor intensive and prone to errors without proper training or fatigue. Machine learning can assist pathologists by increasing efficiency and minimizing individual variability. This study adapted a deep learning model to reliably identify 3 cell types from triple negative breast cancers fixed on H&E slides and verified performance with an expert pathologist. Methods: We apply the U-Net architecture to analyse pathology slides fixed with triple negative breast cancer (TNBC) tissue. The public dataset NuCLS was used for training. Semantic segmentation was used to identify single cells of 3 types: tumor cells, fibroblasts, and tumor infiltrating lymphocyte (TILs). For validation, the pathologist annotated 8 random H&E tiles. The 3 cell types accuracy for NuCLS and the model was evaluated by our pathologist. Results: Overall, there was a 73% agreement with the pathologists (Pathologist vs. NuCLS). A set of 1,555 (90%) TNBC slides were used for training and 173 for validation (10%; unseen data). Table 1 outlines the accuracy metrics for each cell type and for each comparison. Compared to our pathologist, the model accurately identified TILs (62%), followed by fibroblasts (42%) and lastly tumor cells (26%). A significant source of discrepancy was variation in labeled single cell boundaries. The model was better at identifying TILs. The pathologist took 1.5 hours to annotate 8 tiles for the 3 cells and our model 644ms. Conclusion: It is possible to develop a deep learning model to identify breast cancer cells, fibroblasts and TILs from H&E stained slides, with similar accuracy levels as a trained pathologist. The model performed better than a pathologist in identifying TILs, but both struggled with fibroblasts. Accuracy of 71% overall and 87% for TILs, motivates expansion to further datasets and other cancer types. Table 1. - Accuracy metrics. Quality assessment8 H&E tiles Pathologist vs. NuCLS U-Net model vs. NuCLs U-Net model vs. Pathologist U-Net model vs. NuCLs (Full) Background 86% 71% 62% 73% Tumour 43% 35% 26% 71% Fibroblast 45% 39% 42% 31% TILs 34% 96% 62% 87% Overall 73% 74% 61% 71% Citation Format: Luiz Augusto Zillmann da Silva, Alistair R. Williams, Aidan Kubeyev, Andrea Giorni, Jordan Laurie, Prabu Sivasubramaniam, Matthew Foster, Matthew Griffiths, Uzma Asghar. A deep learning approach (AI) which accurately identifies breast tumor cells, tumor infiltrating lymphocytes (TILS) and fibroblasts from H&E slides. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5424.

Background: Accurate modelling of the impact of patient-specific features and cancer treatments on survival allows the assignment of targeted therapy. There has not been any effort to build a multi-source model for the survival analysis of breast cancer. We show in this study a prognostic model, which integrates genetic (DNA), clinical and therapy inputs to predict survival for early breast cancer (stages 1-3) for all breast cancer subtypes. Methods: We used a data-driven Random Survival Forest approach, a statistical non-parametric ensemble learning method, that incorporates censor and time-to-event data. The learning is performed by creating numerous decision trees and selecting the model based on the correct responses in unseen data. We used The Cancer Genome Atlas Breast Cancer (TCGA) dataset and observed improvements in the accuracy when more sources of data were used, in line with the previous research. Integrating the impact of non-silent somatic tumor mutations (whole exome) and gene copy number variation (CNV) were analyzed on all mutations and per particular mutation. Results: Data from 1096 women with stage 1-3 early breast cancer were inputs to the model n=437 ER+ve HER2-ve, n=123 HER2+ve ER+ve, n=40 HER2+ ER-ve and n=126 TNBC. Pathological stage 1, n=183; stage 2, n=620; and stage 3, n=249. The following chemotherapy and hormonal treatments were used in the analysis: anthracycline, taxanes, platinum, alkylating and anti-metabolite agents, anti-oestrogen, aromatase inhibitors, ovarian suppression and HER2 antibody treatment. The model accuracy for predicting survival for early breast cancer using only clinical data was 0.78 for Area Under Curve (AUC) and c-index. The predictive accuracy improved stepwise by adding hormone, genetic and treatment data to AUC of 0.86 and c-index to 0.85. We observed the same trend if the proportion of test data increased from 0.25 to 0.75. Changes in median genes FGFR2 and CDKN2A copy number were strongly prognostic with p=0.0001 and p=0.002, and weaker signals for CBFB p=0.05, HRAS p=0.06, AKT p=0.07. Conclusion: Using public datasets, we developed a predictive survival model for an individual with early breast cancer up to 5 years from diagnosis using multi-source and patient-specific data. We show that using this approach for survival analysis yields good accuracy. Citation Format: Aidan (Amanzhol) Kubeyev, Andrea Giorni, Prabu Siva, Luiz Silva, Jordan Laurie, Matthew Foster, Matthew Griffiths, Uzma Asghar. A prognostic machine learning model for early breast cancer which combines clinical and genetic data in patients treated with neo/adjuvant chemotherapy. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5696.

Background: Predicting tissue of origin (ToO) using clinical and molecular data improves diagnostic accuracy up to 95% in patients with Cancer of Unknown Primary (CUP). It is hypothesized that better treatment stratification of CUP patients using omics and machine learning (ML) classifiers may improve prognosis. Methods: We used publicly available whole exome somatic mutation data from 4733 primary solid tissue samples, across 11 tumor types from the TCGA database, and employed a ML classifier to predict their ToO. We used 5 sets of modeling features: 1) Non-silent somatic mutation burden of 230 cancer-related genes 2) Frequency of SNP substitution type 3) Trinucleotide mutation frequency 4) Copy number variation of the 230 cancer-related genes 5) Presence of hotspot mutations. We trained a Support Vector Machine on a training subset (80% of samples) and tuned the hyperparameters maximizing a 5-fold cross-validation F1-score. We then tested the model performance on a validation subset (20% of samples) and on a limited (n=6) dataset of metastatic samples present in the TCGA database. Results: On the primary tumor validation set, we achieved an average AUC of 0.98(std: 0.02) and top 1, top 2 and top 3 accuracies of 80%(std: 0.11), 90%(std: 0.08) and 95%(std: 0.04) respectively, across 11 tumor types. The classification accuracy plateaus after ~300 samples, suggesting further data collection may benefit low performing tumor types. The 2 worst performers: esophageal and stomach cancers were mostly misclassified with colorectal cancers, reflecting their relative similarity. On metastatic samples (n=6) the model achieved a 67% accuracy, this is work in progress. Conclusion: Our study confirms the potential for a DNA-based machine learning approach to improve prognosis in CUP patients by aiding diagnosis of ToO. To this end, we plan to take this study further by applying this approach to large, independent datasets derived from metastatic samples and liquid biopsies from CUP patient cohorts. Table 1. Top_1_acc Top_2_acc Top_3_acc Precision Recall F1_score Training_size Breast 0.88 0.97 0.99 0.88 0.84 0.86 756 Colorectal 0.85 0.97 0.98 0.83 0.83 0.83 460 Oesophagus 0.5 0.75 0.89 0.51 0.5 0.51 144 Liver 0.83 0.92 0.96 0.82 0.89 0.85 288 Lung 0.86 0.91 0.94 0.98 0.81 0.88 396 Ovary 0.87 0.94 0.97 0.77 0.9 0.83 312 Pancreas 0.73 0.79 0.85 0.62 0.7 0.66 132 Prostate 0.88 0.94 0.98 0.75 0.88 0.81 380 Sarcoma 0.78 0.84 0.96 0.8 0.82 0.81 180 Stomach 0.73 0.91 0.95 0.68 0.58 0.62 340 Endometrial 0.9 0.97 0.99 0.86 0.87 0.87 400 Mean 80.09% 90.09% 95.09% 0.77 0.78 0.78 344 Top_1_acc_n Top_2_acc_n Top_3_acc_n n samples Metastatic breast 2 2 2 2 Metastatic prostate 0 0 0 1 Metastatic pancreas 0 0 0 1 Metastatic sarcoma 1 1 1 1 Metastatic oesophagus 1 1 1 1 Mean accuracy 67% 67% 67% Citation Format: Andrea Giorni, Prabu Sivasubramiam, Aidan Kubeyev, Jordan Laurie, Luiz Silva, Matthew Foster, Uzma Asghar, Matthew Griffiths. Using machine learning to predict tissue of origin from somatic mutation features. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5429.

ADAPTeR is a prospective, phase II study of nivolumab (anti-PD-1) in 15 treatment-naive patients (115 multiregion tumor samples) with metastatic clear cell renal cell carcinoma (ccRCC) aiming to understand the mechanism underpinning therapeutic response. Genomic analyses show no correlation between tumor molecular features and response, whereas ccRCC-specific human endogenous retrovirus expression indirectly correlates with clinical response. T cell receptor (TCR) analysis reveals a significantly higher number of expanded TCR clones pre-treatment in responders suggesting pre-existing immunity. Maintenance of highly similar clusters of TCRs post-treatment predict response, suggesting ongoing antigen engagement and survival of families of T cells likely recognizing the same antigens. In responders, nivolumab-bound CD8+ T cells are expanded and express GZMK/B. Our data suggest nivolumab drives both maintenance and replacement of previously expanded T cell clones, but only maintenance correlates with response. We hypothesize that maintenance and boosting of a pre-existing response is a key element of anti-PD-1 mode of action.

BackgroundBAL101553 (lisavanbulin), the lysine prodrug of BAL27862 (avanbulin), exhibits broad anti-proliferative activity in human cancer models refractory to clinically relevant microtubule-targeting agents.MethodsThis two-part, open-label, phase 1/2a study aimed to determine the maximum tolerated dose (MTD) and dose-limiting toxicities (DLTs) of 2-h infusion of BAL101553 in adults with advanced or recurrent solid tumours. The MTD was determined using a modified accelerated titration design in phase I. Patients received BAL101553 at the MTD and at lower doses in the phase 2a expansion to characterise safety and efficacy and to determine the recommended phase 2 dose (RP2D).ResultsSeventy-three patients received BAL101553 at doses of 15–80 mg/m2(phase 1,n = 24; phase 2a,n = 49). The MTD was 60 mg/m2; DLTs observed at doses ≥60 mg/m2were reversible Grade 2–3 gait disturbance with Grade 2 peripheral sensory neuropathy. In phase 2a, asymptomatic myocardial injury was observed at doses ≥45 mg/m2. The RP2D for 2-h intravenous infusion was 30 mg/m2. The overall disease control rate was 26.3% in the efficacy population.ConclusionsThe RP2D for 2-h infusion of BAL101553 was well tolerated. Dose-limiting neurological and myocardial side effects were consistent with the agent’s vascular-disrupting properties.Clinical trial registrationEudraCT: 2010-024237-23.

Purpose: There is no standard second-line therapy for patients with oesophagogastric cancer who progress following first-line chemotherapy for advanced disease or relapse following radical multi-modality therapy. The aim of this retrospective study was to evaluate survival following rechallenge with platinum plus fluoropyrimidine (PF) +/– epirubicin. Methods: Patients treated with PF +/– epirubicin for oesophagogastric cancer at our institution were identified from the electronic prescribing database. Patients rechallenged with PF +/– epirubicin >3 months after completing initial chemotherapy were eligible. Primary endpoint was survival, calculated from day 1 of rechallenge treatment to date of death or last follow-up. Secondary endpoints were progression-free survival and response rate to PF-based re-challenge. Results: Between 2000 and 2008, 950 patients treated with PF +/– epirubicin for oesophagogastric cancer were identified. 298 patients progressed or relapsed >3 months after completing chemotherapy, of whom 106 patients were rechallenged with PF-based chemotherapy. Median progression-free survival and overall survival were 5.1 and 10 months, respectively, from date of rechallenge for patients treated with initial radical intent and 3.9 and 6.6 months, respectively, in patients treated with palliative intent from diagnosis. In a survival analysis, no significant prognostic factors were identified. Conclusion: Selected patients with oesophagogastric cancer who relapse or progress >3 months after initial treatment with PF +/– epirubicin may benefit from re-introduction of PF-based chemotherapy.

SummaryHepatocellular cancer is a significant global health problem yet the prognosis for the majority of patients has not changed significantly over the past few decades. For patients with advanced disease, sorafenib is currently the standard of care providing a survival advantage of 2–3months in selected patients. Cytotoxic chemotherapy has been used for over 30years but definite evidence that it prolongs survival has been lacking. Resistance remains a significant barrier for both targeted and cytotoxic agents and an understanding of the underlying mechanisms is critical if outcomes are to be improved. Here, we summarise the past and current data that constitute the evidence base for chemotherapy in HCC, review the causes of chemoresistance and suggest strategies to overcome these barriers.