Bayesian inference for identifying tumour-specific cancer dependencies through integration of ex-vivo drug response assays and drug-protein profiling

The identification of tumor-specific molecular dependencies is essential for the development of effective cancer therapies. Genetic and chemical perturbations are powerful tools for discovering these dependencies. Even though chemical perturbations can be applied to primary cancer samples at large scale, the interpretation of experiment outcomes is often complicated by the fact that one chemical compound can affect multiple proteins. To overcome this challenge, Batzilla et al. (PLoS Comput Biol 18(8): e1010438, 2022) proposed DepInfeR, a regularized multi-response regression model designed to identify and estimate specific molecular dependencies of individual cancers from their ex-vivo drug sensitivity profiles. Inspired by their work, we propose a Bayesian extension to DepInfeR. Our proposed approach offers several advantages over DepInfeR, including e.g. the ability to handle missing values in both protein-drug affinity and drug sensitivity profiles without the need for data pre-processing steps such as imputation. Moreover, our approach uses Gaussian Processes to capture more complex molecular dependency structures, and provides probabilistic statements about whether a protein in the protein-drug affinity profiles is informative to the drug sensitivity profiles. Simulation studies demonstrate that our proposed approach achieves better prediction accuracy, and is able to discover unreported dependency structures.

Concordance of randomised controlled trials for artificial intelligence interventions with the CONSORT-AI reporting guidelines

The Consolidated Standards of Reporting Trials extension for Artificial Intelligence interventions (CONSORT-AI) was published in September 2020. Since its publication, several randomised controlled trials (RCTs) of AI interventions have been published but their completeness and transparency of reporting is unknown. This systematic review assesses the completeness of reporting of AI RCTs following publication of CONSORT-AI and provides a comprehensive summary of RCTs published in recent years. 65 RCTs were identified, mostly conducted in China (37%) and USA (18%). Median concordance with CONSORT-AI reporting was 90% (IQR 77-94%), although only 10 RCTs explicitly reported its use. Several items were consistently under-reported, including algorithm version, accessibility of the AI intervention or code, and references to a study protocol. Only 3 of 52 included journals explicitly endorsed or mandated CONSORT-AI. Despite a generally high concordance amongst recent AI RCTs, some AI-specific considerations remain systematically poorly reported. Further encouragement of CONSORT-AI adoption by journals and funders may enable more complete adoption of the full CONSORT-AI guidelines.

Association between pregnancy-related complications and development of type 2 diabetes and hypertension in women: an umbrella review

BACKGROUND

Despite many systematic reviews and meta-analyses examining the associations of pregnancy complications with risk of type 2 diabetes mellitus (T2DM) and hypertension, previous umbrella reviews have only examined a single pregnancy complication. Here we have synthesised evidence from systematic reviews and meta-analyses on the associations of a wide range of pregnancy-related complications with risk of developing T2DM and hypertension.

METHODS

Medline, Embase and Cochrane Database of Systematic Reviews were searched from inception until 26 September 2022 for systematic reviews and meta-analysis examining the association between pregnancy complications and risk of T2DM and hypertension. Screening of articles, data extraction and quality appraisal (AMSTAR2) were conducted independently by two reviewers using Covidence software. Data were extracted for studies that examined the risk of T2DM and hypertension in pregnant women with the pregnancy complication compared to pregnant women without the pregnancy complication. Summary estimates of each review were presented using tables, forest plots and narrative synthesis and reported following Preferred Reporting Items for Overviews of Reviews (PRIOR) guidelines.

RESULTS

Ten systematic reviews were included. Two pregnancy complications were identified. Gestational diabetes mellitus (GDM): One review showed GDM was associated with a 10-fold higher risk of T2DM at least 1 year after pregnancy (relative risk (RR) 9.51 (95% confidence interval (CI) 7.14 to 12.67) and although the association differed by ethnicity (white: RR 16.28 (95% CI 15.01 to 17.66), non-white: RR 10.38 (95% CI 4.61 to 23.39), mixed: RR 8.31 (95% CI 5.44 to 12.69)), the between subgroups difference were not statistically significant at 5% significance level. Another review showed GDM was associated with higher mean blood pressure at least 3 months postpartum (mean difference in systolic blood pressure: 2.57 (95% CI 1.74 to 3.40) mmHg and mean difference in diastolic blood pressure: 1.89 (95% CI 1.32 to 2.46) mmHg). Hypertensive disorders of pregnancy (HDP): Three reviews showed women with a history of HDP were 3 to 6 times more likely to develop hypertension at least 6 weeks after pregnancy compared to women without HDP (meta-analysis with largest number of studies: odds ratio (OR) 4.33 (3.51 to 5.33)) and one review reported a higher rate of T2DM after HDP (hazard ratio (HR) 2.24 (1.95 to 2.58)) at least a year after pregnancy. One of the three reviews and five other reviews reported women with a history of preeclampsia were 3 to 7 times more likely to develop hypertension at least 6 weeks postpartum (meta-analysis with the largest number of studies: OR 3.90 (3.16 to 4.82) with one of these reviews reporting the association was greatest in women from Asia (Asia: OR 7.54 (95% CI 2.49 to 22.81), Europe: OR 2.19 (95% CI 0.30 to 16.02), North and South America: OR 3.32 (95% CI 1.26 to 8.74)).

CONCLUSIONS

GDM and HDP are associated with a greater risk of developing T2DM and hypertension. Common confounders adjusted for across the included studies in the reviews were maternal age, body mass index (BMI), socioeconomic status, smoking status, pre-pregnancy and current BMI, parity, family history of T2DM or cardiovascular disease, ethnicity, and time of delivery. Further research is needed to evaluate the value of embedding these pregnancy complications as part of assessment for future risk of T2DM and chronic hypertension.

Rarity: discovering rare cell populations from single-cell imaging data

MOTIVATION

Cell type identification plays an important role in the analysis and interpretation of single-cell data and can be carried out via supervised or unsupervised clustering approaches. Supervised methods are best suited where we can list all cell types and their respective marker genes a priori, while unsupervised clustering algorithms look for groups of cells with similar expression properties. This property permits the identification of both known and unknown cell populations, making unsupervised methods suitable for discovery. Success is dependent on the relative strength of the expression signature of each group as well as the number of cells. Rare cell types therefore present a particular challenge that is magnified when they are defined by differentially expressing a small number of genes.

RESULTS

Typical unsupervised approaches fail to identify such rare subpopulations, and these cells tend to be absorbed into more prevalent cell types. In order to balance these competing demands, we have developed a novel statistical framework for unsupervised clustering, named Rarity, that enables the discovery process for rare cell types to be more robust, consistent, and interpretable. We achieve this by devising a novel clustering method based on a Bayesian latent variable model in which we assign cells to inferred latent binary on/off expression profiles. This lets us achieve increased sensitivity to rare cell populations while also allowing us to control and interpret potential false positive discoveries. We systematically study the challenges associated with rare cell type identification and demonstrate the utility of Rarity on various IMC datasets.

AVAILABILITY AND IMPLEMENTATION

Implementation of Rarity together with examples is available from the Github repository (https://github.com/kasparmartens/rarity).

Completing a genomic characterisation of microscopic tumour samples with copy number

BACKGROUND

Genomic insights in settings where tumour sample sizes are limited to just hundreds or even tens of cells hold great clinical potential, but also present significant technical challenges. We previously developed the DigiPico sequencing platform to accurately identify somatic mutations from such samples.

RESULTS

Here, we complete this genomic characterisation with copy number. We present a novel protocol, PicoCNV, to call allele-specific somatic copy number alterations from picogram quantities of tumour DNA. We find that PicoCNV provides exactly accurate copy number in 84% of the genome for even the smallest samples, and demonstrate its clinical potential in maintenance therapy.

CONCLUSIONS

PicoCNV complements our existing platform, allowing for accurate and comprehensive genomic characterisations of cancers in settings where only microscopic samples are available.

Sociodemographic characteristics and longitudinal progression of multimorbidity: A multistate modelling analysis of a large primary care records dataset in England

BACKGROUND

Multimorbidity, characterised by the coexistence of multiple chronic conditions in an individual, is a rising public health concern. While much of the existing research has focused on cross-sectional patterns of multimorbidity, there remains a need to better understand the longitudinal accumulation of diseases. This includes examining the associations between important sociodemographic characteristics and the rate of progression of chronic conditions.

METHODS AND FINDINGS

We utilised electronic primary care records from 13.48 million participants in England, drawn from the Clinical Practice Research Datalink (CPRD Aurum), spanning from 2005 to 2020 with a median follow-up of 4.71 years (IQR: 1.78, 11.28). The study focused on 5 important chronic conditions: cardiovascular disease (CVD), type 2 diabetes (T2D), chronic kidney disease (CKD), heart failure (HF), and mental health (MH) conditions. Key sociodemographic characteristics considered include ethnicity, social and material deprivation, gender, and age. We employed a flexible spline-based parametric multistate model to investigate the associations between these sociodemographic characteristics and the rate of different disease transitions throughout multimorbidity development. Our findings reveal distinct association patterns across different disease transition types. Deprivation, gender, and age generally demonstrated stronger associations with disease diagnosis compared to ethnic group differences. Notably, the impact of these factors tended to attenuate with an increase in the number of preexisting conditions, especially for deprivation, gender, and age. For example, the hazard ratio (HR) (95% CI; p-value) for the association of deprivation with T2D diagnosis (comparing the most deprived quintile to the least deprived) is 1.76 ([1.74, 1.78]; p < 0.001) for those with no preexisting conditions and decreases to 0.95 ([0.75, 1.21]; p = 0.69) with 4 preexisting conditions. Furthermore, the impact of deprivation, gender, and age was typically more pronounced when transitioning from an MH condition. For instance, the HR (95% CI; p-value) for the association of deprivation with T2D diagnosis when transitioning from MH is 2.03 ([1.95, 2.12], p < 0.001), compared to transitions from CVD 1.50 ([1.43, 1.58], p < 0.001), CKD 1.37 ([1.30, 1.44], p < 0.001), and HF 1.55 ([1.34, 1.79], p < 0.001). A primary limitation of our study is that potential diagnostic inaccuracies in primary care records, such as underdiagnosis, overdiagnosis, or ascertainment bias of chronic conditions, could influence our results.

CONCLUSIONS

Our results indicate that early phases of multimorbidity development could warrant increased attention. The potential importance of earlier detection and intervention of chronic conditions is underscored, particularly for MH conditions and higher-risk populations. These insights may have important implications for the management of multimorbidity.

Immune responses to gut bacteria associated with time to diagnosis and clinical response to T cell-directed therapy for type 1 diabetes prevention

Immune-targeted therapies have efficacy for treatment of autoinflammatory diseases. For example, treatment with the T cell-specific anti-CD3 antibody teplizumab delayed disease onset in participants at high risk for type 1 diabetes (T1D) in the TrialNet 10 (TN-10) trial. However, heterogeneity in therapeutic responses in TN-10 and other immunotherapy trials identifies gaps in understanding disease progression and treatment responses. The intestinal microbiome is a potential source of biomarkers associated with future T1D diagnosis and responses to immunotherapy. We previously reported that antibody responses to gut commensal bacteria were associated with T1D diagnosis, suggesting that certain antimicrobial immune responses may help predict disease onset. Here, we investigated anticommensal antibody (ACAb) responses against a panel of taxonomically diverse intestinal bacteria species in sera from TN-10 participants before and after teplizumab or placebo treatment. We identified IgG2 responses to three species that were associated with time to T1D diagnosis and with teplizumab treatment responses that delayed disease onset. These antibody responses link human intestinal bacteria with T1D progression, adding predictive value to known T1D risk factors. ACAb analysis provides a new approach to elucidate heterogeneity in responses to immunotherapy and identify individuals who may benefit from teplizumab, recently approved by the U.S. Food and Drug Administration for delaying T1D onset.

Detection and evaluation of signals associated with exposure to individual and combination of medications in pregnancy: a signal detection study protocol

INTRODUCTION

Considering the high prevalence of polypharmacy in pregnant women and the knowledge gap in the risk-benefit safety profile of their often-complex treatment plan, more research is needed to optimise prescribing. In this study, we aim to detect adverse and protective effect signals of exposure to individual and pairwise combinations of medications during pregnancy.

METHODS AND ANALYSIS

Using a range of real-world data sources from the UK, we aim to conduct a pharmacovigilance study to assess the safety of medications prescribed during the preconception period (3 months prior to conception) and first trimester of pregnancy. Women aged between 15 and 49 years with a record of pregnancy within the Clinical Practice Research Datalink (CPRD) Pregnancy Register, the Welsh Secure Anonymised Information Linkage (SAIL), the Scottish Morbidity Record (SMR) data sets and the Northern Ireland Maternity System (NIMATS) will be included. A series of case control studies will be conducted to estimate measures of disproportionality, detecting signals of association between a range of pregnancy outcomes and exposure to individual and combinations of medications. A multidisciplinary expert team will be invited to a signal detection workshop. By employing a structured framework, signals will be transparently assessed by each member of the team using a questionnaire appraising the signals on aspects of temporality, selection, time and measurement-related biases and confounding by underlying disease or comedications. Through group discussion, the expert team will reach consensus on each of the medication exposure-outcome signal, thereby excluding spurious signals, leaving signals suggestive of causal associations for further evaluation.

ETHICS AND DISSEMINATION

Ethical approval has been obtained from the Independent Scientific Advisory Committee, SAIL Information Governance Review Panel, University of St. Andrews Teaching and Research Ethics Committee and Office for Research Ethics Committees Northern Ireland (ORECNI) for access and use of CPRD, SAIL, SMR and NIMATS data, respectively.

Identifying factors associated with user retention and outcomes of a digital intervention for substance use disorder: a retrospective analysis of real-world data

Objectives

Successful delivery of digital health interventions is affected by multiple real-world factors. These factors may be identified in routinely collected, ecologically valid data from these interventions. We propose ideas for exploring these data, focusing on interventions targeting complex, comorbid conditions.

Materials and Methods

This study retrospectively explores pre-post data collected between 2016 and 2019 from users of digital cognitive behavioral therapy (CBT)-containing psychoeducation and practical exercises-for substance use disorder (SUD) at UK addiction services. To identify factors associated with heterogenous user responses to the technology, we employed multivariable and multivariate regressions and random forest models of user-reported questionnaire data.

Results

The dataset contained information from 14 078 individuals of which 12 529 reported complete data at baseline and 2925 did so again after engagement with the CBT. Ninety-three percent screened positive for dependence on 1 of 43 substances at baseline, and 73% screened positive for anxiety or depression. Despite pre-post improvements independent of user sociodemographics, women reported more frequent and persistent symptoms of SUD, anxiety, and depression. Retention-minimum 2 use events recorded-was associated more with deployment environment than user characteristics. Prediction accuracy of post-engagement outcomes was acceptable (Area Under Curve [AUC]: 0.74-0.79), depending non-trivially on user characteristics.

Discussion

Traditionally, performance of digital health interventions is determined in controlled trials. Our analysis showcases multivariate models with which real-world data from these interventions can be explored and sources of user heterogeneity in retention and symptom reduction uncovered.

Conclusion

Real-world data from digital health interventions contain information on natural user-technology interactions which could enrich results from controlled trials.

The development of a core outcome set for studies of pregnant women with multimorbidity

BACKGROUND

Heterogeneity in reported outcomes can limit the synthesis of research evidence. A core outcome set informs what outcomes are important and should be measured as a minimum in all future studies. We report the development of a core outcome set applicable to observational and interventional studies of pregnant women with multimorbidity.

METHODS

We developed the core outcome set in four stages: (i) a systematic literature search, (ii) three focus groups with UK stakeholders, (iii) two rounds of Delphi surveys with international stakeholders and (iv) two international virtual consensus meetings. Stakeholders included women with multimorbidity and experience of pregnancy in the last 5 years, or are planning a pregnancy, their partners, health or social care professionals and researchers. Study adverts were shared through stakeholder charities and organisations.

RESULTS

Twenty-six studies were included in the systematic literature search (2017 to 2021) reporting 185 outcomes. Thematic analysis of the focus groups added a further 28 outcomes. Two hundred and nine stakeholders completed the first Delphi survey. One hundred and sixteen stakeholders completed the second Delphi survey where 45 outcomes reached Consensus In (≥70% of all participants rating an outcome as Critically Important). Thirteen stakeholders reviewed 15 Borderline outcomes in the first consensus meeting and included seven additional outcomes. Seventeen stakeholders reviewed these 52 outcomes in a second consensus meeting, the threshold was ≥80% of all participants voting for inclusion. The final core outcome set included 11 outcomes. The five maternal outcomes were as follows: maternal death, severe maternal morbidity, change in existing long-term conditions (physical and mental), quality and experience of care and development of new mental health conditions. The six child outcomes were as follows: survival of baby, gestational age at birth, neurodevelopmental conditions/impairment, quality of life, birth weight and separation of baby from mother for health care needs.

CONCLUSIONS

Multimorbidity in pregnancy is a new and complex clinical research area. Following a rigorous process, this complexity was meaningfully reduced to a core outcome set that balances the views of a diverse stakeholder group.

On the Difficulty of Predicting Engagement with Digital Health for Substance Use

Digital interventions can be an important instrument in treating substance use disorder. However, most digital mental health interventions suffer from early, frequent user dropout. Early prediction of engagement would allow identification of individuals whose engagement with digital interventions may be too limited to support behaviour change, and subsequently offering them support. To investigate this, we used machine learning models to predict different metrics of real-world engagement with a digital cognitive behavioural therapy intervention widely available in UK addiction services. Our predictor set consisted of baseline data from routinely-collected standardised psychometric measures. Areas under the ROC curve, and correlations between predicted and observed values indicated that baseline data do not contain sufficient information about individual patterns of engagement.

An atlas of genetic scores to predict multi-omic traits

The use of omic modalities to dissect the molecular underpinnings of common diseases and traits is becoming increasingly common. But multi-omic traits can be genetically predicted, which enables highly cost-effective and powerful analyses for studies that do not have multi-omics1. Here we examine a large cohort (the INTERVAL study2; n = 50,000 participants) with extensive multi-omic data for plasma proteomics (SomaScan, n = 3,175; Olink, n = 4,822), plasma metabolomics (Metabolon HD4, n = 8,153), serum metabolomics (Nightingale, n = 37,359) and whole-blood Illumina RNA sequencing (n = 4,136), and use machine learning to train genetic scores for 17,227 molecular traits, including 10,521 that reach Bonferroni-adjusted significance. We evaluate the performance of genetic scores through external validation across cohorts of individuals of European, Asian and African American ancestries. In addition, we show the utility of these multi-omic genetic scores by quantifying the genetic control of biological pathways and by generating a synthetic multi-omic dataset of the UK Biobank3 to identify disease associations using a phenome-wide scan. We highlight a series of biological insights with regard to genetic mechanisms in metabolism and canonical pathway associations with disease; for example, JAK-STAT signalling and coronary atherosclerosis. Finally, we develop a portal ( https://www.omicspred.org/ ) to facilitate public access to all genetic scores and validation results, as well as to serve as a platform for future extensions and enhancements of multi-omic genetic scores.

Maternal and child outcomes for pregnant women with pre-existing multiple long-term conditions: protocol for an observational study in the UK

INTRODUCTION

One in five pregnant women has multiple pre-existing long-term conditions in the UK. Studies have shown that maternal multiple long-term conditions are associated with adverse outcomes. This observational study aims to compare maternal and child outcomes for pregnant women with multiple long-term conditions to those without multiple long-term conditions (0 or 1 long-term conditions).

METHODS AND ANALYSIS

Pregnant women aged 15-49 years old with a conception date between 2000 and 2019 in the UK will be included with follow-up till 2019. The data source will be routine health records from all four UK nations (Clinical Practice Research Datalink (England), Secure Anonymised Information Linkage (Wales), Scotland routine health records and Northern Ireland Maternity System) and the Born in Bradford birth cohort. The exposure of two or more pre-existing, long-term physical or mental health conditions will be defined from a list of health conditions predetermined by women and clinicians. The association of maternal multiple long-term conditions with (a) antenatal, (b) peripartum, (c) postnatal and long-term and (d) mental health outcomes, for both women and their children will be examined. Outcomes of interest will be guided by a core outcome set. Comparisons will be made between pregnant women with and without multiple long-term conditions using modified Poisson and Cox regression. Generalised estimating equation will account for the clustering effect of women who had more than one pregnancy episode. Where appropriate, multiple imputation with chained equation will be used for missing data. Federated analysis will be conducted for each dataset and results will be pooled using random-effects meta-analyses.

ETHICS AND DISSEMINATION

Approval has been obtained from the respective data sources in each UK nation. Study findings will be submitted for publications in peer-reviewed journals and presented at key conferences.

Polypharmacy during pregnancy and associated risk factors: a retrospective analysis of 577 medication exposures among 1.5 million pregnancies in the UK, 2000-2019

BACKGROUND

The number of medications prescribed during pregnancy has increased over the past few decades. Few studies have described the prevalence of multiple medication use among pregnant women. This study aims to describe the overall prevalence over the last two decades among all pregnant women and those with multimorbidity and to identify risk factors for polypharmacy in pregnancy.

METHODS

A retrospective cohort study was conducted between 2000 and 2019 using the Clinical Practice Research Datalink (CPRD) pregnancy register. Prescription records for 577 medication categories were obtained. Prevalence estimates for polypharmacy (ranging from 2+ to 11+ medications) were presented along with the medications commonly prescribed individually and in pairs during the first trimester and the entire pregnancy period. Logistic regression models were performed to identify risk factors for polypharmacy.

RESULTS

During the first trimester (812,354 pregnancies), the prevalence of polypharmacy ranged from 24.6% (2+ medications) to 0.1% (11+ medications). During the entire pregnancy period (774,247 pregnancies), the prevalence ranged from 58.7 to 1.4%. Broad-spectrum penicillin (6.6%), compound analgesics (4.5%) and treatment of candidiasis (4.3%) were commonly prescribed. Pairs of medication prescribed to manage different long-term conditions commonly included selective beta 2 agonists or selective serotonin re-uptake inhibitors (SSRIs). Risk factors for being prescribed 2+ medications during the first trimester of pregnancy include being overweight or obese [aOR: 1.16 (1.14-1.18) and 1.55 (1.53-1.57)], belonging to an ethnic minority group [aOR: 2.40 (2.33-2.47), 1.71 (1.65-1.76), 1.41 (1.35-1.47) and 1.39 (1.30-1.49) among women from South Asian, Black, other and mixed ethnicities compared to white women] and smoking or previously smoking [aOR: 1.19 (1.18-1.20) and 1.05 (1.03-1.06)]. Higher and lower age, higher gravidity, increasing number of comorbidities and increasing level of deprivation were also associated with increased odds of polypharmacy.

CONCLUSIONS

The prevalence of polypharmacy during pregnancy has increased over the past two decades and is particularly high in younger and older women; women with high BMI, smokers and ex-smokers; and women with multimorbidity, higher gravidity and higher levels of deprivation. Well-conducted pharmaco-epidemiological research is needed to understand the effects of multiple medication use on the developing foetus.

Protocol for development and validation of postpartum cardiovascular disease (CVD) risk prediction model incorporating reproductive and pregnancy-related candidate predictors

BACKGROUND

Cardiovascular disease (CVD) is a leading cause of death among women. CVD is associated with reduced quality of life, significant treatment and management costs, and lost productivity. Estimating the risk of CVD would help patients at a higher risk of CVD to initiate preventive measures to reduce risk of disease. The Framingham risk score and the QRISK® score are two risk prediction models used to evaluate future CVD risk in the UK. Although the algorithms perform well in the general population, they do not take into account pregnancy complications, which are well known risk factors for CVD in women and have been highlighted in a recent umbrella review. We plan to develop a robust CVD risk prediction model to assess the additional value of pregnancy risk factors in risk prediction of CVD in women postpartum.

METHODS

Using candidate predictors from QRISK®-3, the umbrella review identified from literature and from discussions with clinical experts and patient research partners, we will use time-to-event Cox proportional hazards models to develop and validate a 10-year risk prediction model for CVD postpartum using Clinical Practice Research Datalink (CPRD) primary care database for development and internal validation of the algorithm and the Secure Anonymised Information Linkage (SAIL) databank for external validation. We will then assess the value of additional candidate predictors to the QRISK®-3 in our internal and external validations.

DISCUSSION

The developed risk prediction model will incorporate pregnancy-related factors which have been shown to be associated with future risk of CVD but have not been taken into account in current risk prediction models. Our study will therefore highlight the importance of incorporating pregnancy-related risk factors into risk prediction modeling for CVD postpartum.

In simulated data and health records, latent class analysis was the optimum multimorbidity clustering algorithm

BACKGROUND AND OBJECTIVES

To investigate the reproducibility and validity of latent class analysis (LCA) and hierarchical cluster analysis (HCA), multiple correspondence analysis followed by k-means (MCA-kmeans) and k-means (kmeans) for multimorbidity clustering.

METHODS

We first investigated clustering algorithms in simulated datasets with 26 diseases of varying prevalence in predetermined clusters, comparing the derived clusters to known clusters using the adjusted Rand Index (aRI). We then them investigated the medical records of male patients, aged 65 to 84 years from 50 UK general practices, with 49 long-term health conditions. We compared within cluster morbidity profiles using the Pearson correlation coefficient and assessed cluster stability using in 400 bootstrap samples.

RESULTS

In the simulated datasets, the closest agreement (largest aRI) to known clusters was with LCA and then MCA-kmeans algorithms. In the medical records dataset, all four algorithms identified one cluster of 20-25% of the dataset with about 82% of the same patients across all four algorithms. LCA and MCA-kmeans both found a second cluster of 7% of the dataset. Other clusters were found by only one algorithm. LCA and MCA-kmeans clustering gave the most similar partitioning (aRI 0.54).

CONCLUSION

LCA achieved higher aRI than other clustering algorithms.

Genetically personalised organ-specific metabolic models in health and disease

Understanding how genetic variants influence disease risk and complex traits (variant-to-function) is one of the major challenges in human genetics. Here we present a model-driven framework to leverage human genome-scale metabolic networks to define how genetic variants affect biochemical reaction fluxes across major human tissues, including skeletal muscle, adipose, liver, brain and heart. As proof of concept, we build personalised organ-specific metabolic flux models for 524,615 individuals of the INTERVAL and UK Biobank cohorts and perform a fluxome-wide association study (FWAS) to identify 4312 associations between personalised flux values and the concentration of metabolites in blood. Furthermore, we apply FWAS to identify 92 metabolic fluxes associated with the risk of developing coronary artery disease, many of which are linked to processes previously described to play in role in the disease. Our work demonstrates that genetically personalised metabolic models can elucidate the downstream effects of genetic variants on biochemical reactions involved in common human diseases.

Patient Derived Organoids Confirm That PI3K/AKT Signalling Is an Escape Pathway for Radioresistance and a Target for Therapy in Rectal Cancer

Objectives

Partial or total resistance to preoperative chemoradiotherapy occurs in more than half of locally advanced rectal cancer patients. Several novel or repurposed drugs have been trialled to improve cancer cell sensitivity to radiotherapy, with limited success. We aimed to understand the mechanisms of resistance to chemoradiotherapy in rectal cancer using patient derived organoid models.

Design

To understand the mechanisms underlying this resistance, we compared the pre-treatment transcriptomes of patient-derived organoids (PDO) with measured radiotherapy sensitivity to identify biological pathways involved in radiation resistance coupled with single cell sequencing, genome wide CRISPR-Cas9 and targeted drug screens.

Results

RNA sequencing enrichment analysis revealed upregulation of PI3K/AKT/mTOR and epithelial mesenchymal transition pathway genes in radioresistant PDOs. Single-cell sequencing of pre & post-irradiation PDOs showed mTORC1 and PI3K/AKT upregulation, which was confirmed by a genome-wide CRSIPR-Cas9 knockout screen using irradiated colorectal cancer (CRC) cell lines. We then tested the efficiency of dual PI3K/mTOR inhibitors in improving cancer cell sensitivity to radiotherapy. After irradiation, significant AKT phosphorylation was detected (p=0.027) which was abrogated with dual PI3K/mTOR inhibitors and lead to significant radiosensitisation of the HCT116 cell line and radiation resistant PDO lines.

Conclusions

The PI3K/AKT/mTOR pathway upregulation contributes to radioresistance and its targeted pharmacological inhibition leads to significant radiosensitisation in CRC organoids, making it a potential target for clinical trials.

Epidemiology of pre-existing multimorbidity in pregnant women in the UK in 2018: a population-based cross-sectional study

BACKGROUND

Although maternal death is rare in the United Kingdom, 90% of these women had multiple health/social problems. This study aims to estimate the prevalence of pre-existing multimorbidity (two or more long-term physical or mental health conditions) in pregnant women in the United Kingdom (England, Northern Ireland, Wales and Scotland).

STUDY DESIGN

Pregnant women aged 15-49 years with a conception date 1/1/2018 to 31/12/2018 were included in this population-based cross-sectional study, using routine healthcare datasets from primary care: Clinical Practice Research Datalink (CPRD, United Kingdom, n = 37,641) and Secure Anonymized Information Linkage databank (SAIL, Wales, n = 27,782), and secondary care: Scottish Morbidity Records with linked community prescribing data (SMR, Tayside and Fife, n = 6099). Pre-existing multimorbidity preconception was defined from 79 long-term health conditions prioritised through a workshop with patient representatives and clinicians.

RESULTS

The prevalence of multimorbidity was 44.2% (95% CI 43.7-44.7%), 46.2% (45.6-46.8%) and 19.8% (18.8-20.8%) in CPRD, SAIL and SMR respectively. When limited to health conditions that were active in the year before pregnancy, the prevalence of multimorbidity was still high (24.2% [23.8-24.6%], 23.5% [23.0-24.0%] and 17.0% [16.0 to 17.9%] in the respective datasets). Mental health conditions were highly prevalent and involved 70% of multimorbidity CPRD: multimorbidity with ≥one mental health condition/s 31.3% [30.8-31.8%]). After adjusting for age, ethnicity, gravidity, index of multiple deprivation, body mass index and smoking, logistic regression showed that pregnant women with multimorbidity were more likely to be older (CPRD England, adjusted OR 1.81 [95% CI 1.04-3.17] 45-49 years vs 15-19 years), multigravid (1.68 [1.50-1.89] gravidity ≥ five vs one), have raised body mass index (1.59 [1.44-1.76], body mass index 30+ vs body mass index 18.5-24.9) and smoked preconception (1.61 [1.46-1.77) vs non-smoker).

CONCLUSION

Multimorbidity is prevalent in pregnant women in the United Kingdom, they are more likely to be older, multigravid, have raised body mass index and smoked preconception. Secondary care and community prescribing dataset may only capture the severe spectrum of health conditions. Research is needed urgently to quantify the consequences of maternal multimorbidity for both mothers and children.

CIDER: an interpretable meta-clustering framework for single-cell RNA-seq data integration and evaluation

Clustering of joint single-cell RNA-Seq (scRNA-Seq) data is often challenged by confounding factors, such as batch effects and biologically relevant variability. Existing batch effect removal methods typically require strong assumptions on the composition of cell populations being near identical across samples. Here, we present CIDER, a meta-clustering workflow based on inter-group similarity measures. We demonstrate that CIDER outperforms other scRNA-Seq clustering methods and integration approaches in both simulated and real datasets. Moreover, we show that CIDER can be used to assess the biological correctness of integration in real datasets, while it does not require the existence of prior cellular annotations.

Protocol for the development of a core outcome set for studies of pregnant women with pre-existing multimorbidity

INTRODUCTION

Increasingly more pregnant women are living with pre-existing multimorbidity (≥two long-term physical or mental health conditions). This may adversely affect maternal and offspring outcomes. This study aims to develop a core outcome set (COS) for maternal and offspring outcomes in pregnant women with pre-existing multimorbidity. It is intended for use in observational and interventional studies in all pregnancy settings.

METHODS AND ANALYSIS

We propose a four stage study design: (1) systematic literature search, (2) focus groups, (3) Delphi surveys and (4) consensus group meeting. The study will be conducted from June 2021 to August 2022. First, an initial list of outcomes will be identified through a systematic literature search of reported outcomes in studies of pregnant women with multimorbidity. We will search the Cochrane library, Medline, EMBASE and CINAHL. This will be supplemented with relevant outcomes from published COS for pregnancies and childbirth in general, and multimorbidity. Second, focus groups will be conducted among (1) women with lived experience of managing pre-existing multimorbidity in pregnancy (and/or their partners) and (2) their healthcare/social care professionals to identify outcomes important to them. Third, these initial lists of outcomes will be prioritised through a three-round online Delphi survey using predefined score criteria for consensus. Participants will be invited to suggest additional outcomes that were not included in the initial list. Finally, a consensus meeting using the nominal group technique will be held to agree on the final COS. The stakeholders will include (1) women (and/or their partners) with lived experience of managing multimorbidity in pregnancy, (2) healthcare/social care professionals involved in their care and (3) researchers in this field.

ETHICS AND DISSEMINATION

This study has been approved by the University of Birmingham's ethical review committee. The final COS will be disseminated through peer-reviewed publication and conferences and to all stakeholders.

Adipocyte-like signature in ovarian cancer minimal residual disease identifies metabolic vulnerabilities of tumor-initiating cells

Similar to tumor-initiating cells (TICs), minimal residual disease (MRD) is capable of reinitiating tumors and causing recurrence. However, the molecular characteristics of solid tumor MRD cells and drivers of their survival have remained elusive. Here we performed dense multiregion transcriptomics analysis of paired biopsies from 17 ovarian cancer patients before and after chemotherapy. We reveal that while MRD cells share important molecular signatures with TICs, they are also characterized by an adipocyte-like gene expression signature and a portion of them had undergone epithelial-mesenchymal transition (EMT). In a cell culture MRD model, MRD-mimic cells showed the same phenotype and were dependent on fatty acid oxidation (FAO) for survival and resistance to cytotoxic agents. These findings identify EMT and FAO as attractive targets to eradicate MRD in ovarian cancer and make a compelling case for the further testing of FAO inhibitors in treating MRD.

Associations between baseline opioid use disorder severity, mental health and biopsychosocial functioning, with clinical responses to computer-assisted therapy treatment

Background: Increasing rates of opioid-related overdose have been identified globally. Treatment for opioid use disorders (OUD) includes medications for opioid use disorder (MOUD) alongside behavioral support. Novel approaches to behavioral support should be explored, including computer-assisted therapy (CAT) programs.Objectives: Examine differences between baseline and post-treatment measures of opioid use and biopsychosocial functioning for individuals with OUD engaging with the CAT program 'Breaking Free Online,' and the extent to which participant characteristics may be associated with post-treatment measures.Methods: 1107 individuals engaged with CAT and provided baseline and post-treatment data - 724 (65.4%) were male, 383 (34.6%) were female.Results: Significant differences between baseline and post-treatment measures were identified (all p <.0001, effect sizes range:15 -.50). Participant characteristics were associated with post-treatment measures of opioid use, opioid dependence, mental health issues, quality of life, and biopsychosocial impairment (all p <.0001). An aggregated consensus measure of clinical impairment was found to be associated with changes in opioid use and post-treatment biopsychosocial functioning measures, with those participants with greater baseline clinical impairment demonstrating a greater magnitude of improvement from baseline to post-treatment than those with lower clinical impairment.Conclusion: CAT may reduce opioid use and improve biopsychosocial functioning in individuals with OUD. CAT could therefore provide a solution to the global opioid crisis if delivered as combination behavioral support alongside MOUD. Findings also indicate that it may be important for treatment systems to identify individuals with psychosocial complexity who might require behavioral support and MOUD.

Promises and challenges of adoptive T-cell therapies for solid tumours

Cancer is a leading cause of death worldwide and, despite new targeted therapies and immunotherapies, many patients with advanced-stage- or high-risk cancers still die, owing to metastatic disease. Adoptive T-cell therapy, involving the autologous or allogeneic transplant of tumour-infiltrating lymphocytes or genetically modified T cells expressing novel T-cell receptors or chimeric antigen receptors, has shown promise in the treatment of cancer patients, leading to durable responses and, in some cases, cure. Technological advances in genomics, computational biology, immunology and cell manufacturing have brought the aspiration of individualised therapies for cancer patients closer to reality. This new era of cell-based individualised therapeutics challenges the traditional standards of therapeutic interventions and provides opportunities for a paradigm shift in our approach to cancer therapy. Invited speakers at a 2020 symposium discussed three areas-cancer genomics, cancer immunology and cell-therapy manufacturing-that are essential to the effective translation of T-cell therapies in the treatment of solid malignancies. Key advances have been made in understanding genetic intratumour heterogeneity, and strategies to accurately identify neoantigens, overcome T-cell exhaustion and circumvent tumour immunosuppression after cell-therapy infusion are being developed. Advances are being made in cell-manufacturing approaches that have the potential to establish cell-therapies as credible therapeutic options. T-cell therapies face many challenges but hold great promise for improving clinical outcomes for patients with solid tumours.

Predicted sensitivity to endocrine therapy for stage II-III hormone receptor-positive and HER2-negative (HR+/HER2-) breast cancer before chemo-endocrine therapy

BACKGROUND

We proposed that a test for sensitivity to the adjuvant endocrine therapy component of treatment for patients with stage II-III breast cancer (SET2,3) should measure transcription related to estrogen and progesterone receptors (SET_ER/PR index) adjusted for a baseline prognostic index (BPI) combining clinical tumor and nodal stage with molecular subtype by RNA4 (ESR1, PGR, ERBB2, and AURKA).

PATIENTS AND METHODS

Patients with clinically high-risk, hormone receptor-positive (HR+), human epidermal growth factor receptor 2 (HER2)-negative (HR+/HER2-) breast cancer received neoadjuvant taxane-anthracycline chemotherapy, surgery with measurement of residual cancer burden (RCB), and then adjuvant endocrine therapy. SET2,3 was measured from pre-treatment tumor biopsies, evaluated first in an MD Anderson Cancer Center (MDACC) cohort (n = 307, 11 years' follow-up, U133A microarrays), cut point was determined, and then independent, blinded evaluation was carried out in the I-SPY2 trial (n = 268, high-risk MammaPrint result, 3.8 years' follow-up, Agilent-44K microarrays, NCI Clinical Trials ID: NCT01042379). Primary outcome measure was distant relapse-free survival. Multivariate Cox regression models tested prognostic independence of SET2,3 relative to RCB and other molecular prognostic signatures, and whether other prognostic signatures could substitute for SET_ER/PR or RNA4 components of SET2,3.

RESULTS

SET2,3 added independent prognostic information to RCB in the MDACC cohort: SET2,3 [hazard ratio (HR) 0.23, P = 0.004] and RCB (HR 1.77, P < 0.001); and the I-SPY2 trial: SET2,3 (HR 0.27, P = 0.031) and RCB (HR 1.68, P = 0.008). SET2,3 provided similar prognostic information irrespective of whether RCB-II or RCB-III after chemotherapy, and in both luminal subtypes. Conversely, RCB was most strongly prognostic in cancers with low SET2,3 status (MDACC P < 0.001, I-SPY2 P < 0.001). Other molecular signatures were not independently prognostic; they could effectively substitute for RNA4 subtype within the BPI component of SET2,3, but they could not effectively substitute for SET_ER/PR index.

CONCLUSIONS

SET2,3 added independent prognostic information to chemotherapy response (RCB) and baseline prognostic score or subtype. Approximately 40% of patients with clinically high-risk HR+/HER2- disease had high SET2,3 and could be considered for clinical trials of neoadjuvant endocrine-based treatment.

Pan-cancer detection of driver genes at the single-patient resolution

BACKGROUND

Identifying the complete repertoire of genes that drive cancer in individual patients is crucial for precision oncology. Most established methods identify driver genes that are recurrently altered across patient cohorts. However, mapping these genes back to patients leaves a sizeable fraction with few or no drivers, hindering our understanding of cancer mechanisms and limiting the choice of therapeutic interventions.

RESULTS

We present sysSVM2, a machine learning software that integrates cancer genetic alterations with gene systems-level properties to predict drivers in individual patients. Using simulated pan-cancer data, we optimise sysSVM2 for application to any cancer type. We benchmark its performance on real cancer data and validate its applicability to a rare cancer type with few known driver genes. We show that drivers predicted by sysSVM2 have a low false-positive rate, are stable and disrupt well-known cancer-related pathways.

CONCLUSIONS

sysSVM2 can be used to identify driver alterations in patients lacking sufficient canonical drivers or belonging to rare cancer types for which assembling a large enough cohort is challenging, furthering the goals of precision oncology. As resources for the community, we provide the code to implement sysSVM2 and the pre-trained models in all TCGA cancer types ( https://github.com/ciccalab/sysSVM2 ).

The Oxford Classic Links Epithelial-to-Mesenchymal Transition to Immunosuppression in Poor Prognosis Ovarian Cancers

PURPOSE

Using RNA sequencing, we recently developed the 52-gene-based Oxford classifier of carcinoma of the ovary (Oxford Classic, OxC) for molecular stratification of serous ovarian cancers (SOCs) based on the molecular profiles of their cell of origin in the fallopian tube epithelium. Here, we developed a 52-gene NanoString panel for the OxC to test the robustness of the classifier.

EXPERIMENTAL DESIGN

We measured the expression of the 52 genes in an independent cohort of prospectively collected SOC samples (n = 150) from a homogenous cohort who were treated with maximal debulking surgery and chemotherapy. We performed data mining of published expression profiles of SOCs and validated the classifier results on tissue arrays comprising 137 SOCs.

RESULTS

We found evidence of profound nongenetic heterogeneity in SOCs. Approximately 20% of SOCs were classified as epithelial-to-mesenchymal transition-high (EMT-high) tumors, which were associated with poor survival. This was independent of established prognostic factors, such as tumor stage, tumor grade, and residual disease after surgery (HR, 3.3; P = 0.02). Mining expression data of 593 patients revealed a significant association between the EMT scores of tumors and the estimated fraction of alternatively activated macrophages (M2; P < 0.0001), suggesting a mechanistic link between immunosuppression and poor prognosis in EMT-high tumors.

CONCLUSIONS

The OxC-defined EMT-high SOCs carry particularly poor prognosis independent of established clinical parameters. These tumors are associated with high frequency of immunosuppressive macrophages, suggesting a potential therapeutic target to improve clinical outcome.

HIPEC after neoadjuvant chemotherapy and interval debulking is associated with development of platinum-refractory or -resistant disease

OBJECTIVE

To describe our single-institution oncologic outcomes of patients who received neoadjuvant chemotherapy (NACT) and interval debulking surgery (IDS) with or without hyperthermic intraperitoneal chemotherapy (HIPEC).

METHODS

We compared clinicopathologic information and outcomes for all patients with advanced stage, high-grade serous ovarian cancer who received NACT and IDS with (N = 20) or without (N = 48) HIPEC at our institution from 2010 to 2019 RESULTS: Mean age (62 years with HIPEC and 60 years without HIPEC) and proportion of stage 4 disease (40% for both) did not differ between cohorts. HIPEC patients had higher rates of complete cytoreduction (95% vs 50%), longer mean duration of surgery (530 vs. 216 min), more grade 3 or 4 postoperative complications (65% vs. 4%), and longer mean length of hospital stay (8 vs. 5 days). HIPEC patients had significantly higher risk for platinum-refractory progression or platinum-resistance recurrence (50% vs 23%; RR = 2.18; 95% CI 1.11, 4.30, p = 0.024). Median progression free survival (11.5 vs. 12 months) and all-cause mortality (19.1 vs. 30.5 months) in the HIPEC and non-HIPEC cohorts, respectively, did not differ CONCLUSIONS: HIPEC was associated with increased risk for platinum refractory or resistant disease. Higher surgical complexity may contribute to higher complication rates without improving oncologic outcomes in our patients. Further investigations and long-term follow-up are needed to assess the utility of HIPEC in primary treatment of advanced stage ovarian cancer.

The quality of life index: a pilot study integrating treatment efficacy and quality of life in oncology

The majority of women diagnosed with breast cancer will experience some form of drug-related toxicity and subsequent impairments in Health-related Quality of Life (HRQoL). Despite this, HRQoL is assessed inconsistently and there is no validated method to integrate HRQoL data into the assessment of therapeutic agents. This proof of concept study utilizes data from the neoadjuvant I-SPY 2 clinical trial to describe the development of the Quality of Life Index (QoLI) measure. The QoLI represents a single composite score that incorporates validated longitudinal measures of clinical efficacy and QoL and one that permits a more comprehensive, direct comparison of individual therapeutic agents. Preliminary data suggest the QoLI is able to distinguish between agents based on their efficacy and toxicity; with further validation, the QoLI has the potential to provide more patient-centered evaluations in clinical trials and help guide treatment decision making in breast cancer and other oncologic diseases.

A highly accurate platform for clone-specific mutation discovery enables the study of active mutational processes

Bulk whole genome sequencing (WGS) enables the analysis of tumor evolution but, because of depth limitations, can only identify old mutational events. The discovery of current mutational processes for predicting the tumor’s evolutionary trajectory requires dense sequencing of individual clones or single cells. Such studies, however, are inherently problematic because of the discovery of excessive false positive (FP) mutations when sequencing picogram quantities of DNA. Data pooling to increase the confidence in the discovered mutations, moves the discovery back in the past to a common ancestor. Here we report a robust WGS and analysis pipeline (DigiPico/MutLX) that virtually eliminates all F results while retaining an excellent proportion of true positives. Using our method, we identified, for the first time, a hyper-mutation (kataegis) event in a group of ∼30 cancer cells from a recurrent ovarian carcinoma. This was unidentifiable from the bulk WGS data. Overall, we propose DigiPico/MutLX method as a powerful framework for the identification of clone-specific variants at an unprecedented accuracy.

Association of HLA-dependent islet autoimmunity with systemic antibody responses to intestinal commensal bacteria in children

Microbiome sequence analyses have suggested that changes in gut bacterial composition are associated with autoimmune disease in humans and animal models. However, little is known of the mechanisms through which the gut microbiota influences autoimmune responses to distant tissues. Here, we evaluated systemic antibody responses against cultured human gut bacterial strains to determine whether observed patterns of anticommensal antibody (ACAb) responses are associated with type 1 diabetes (T1D) in two cohorts of pediatric study participants. In the first cohort, ACAb responses in sera collected from participants within 6 months of T1D diagnosis were compared with age-matched healthy controls and also with patients with recent onset Crohn's disease. ACAb responses against multiple bacterial species discriminated among these three groups. In the second cohort, we asked whether ACAb responses present before diagnosis were associated with later T1D development and with HLA genotype in participants who were discordant for subsequent progression to diabetes. Serum IgG2 antibodies against Roseburia faecis and against a bacterial consortium were associated with future T1D diagnosis in an HLA DR3/DR4 haplotype-dependent manner. These analyses reveal associations between antibody responses to intestinal microbes and HLA-DR genotype and islet autoantibody specificity and with a future diagnosis of T1D. Further, we present a platform to investigate antibacterial antibodies in biological fluids that is applicable to studies of autoimmune diseases and responses to therapeutic interventions.

The Repertoire of Serous Ovarian Cancer Non-genetic Heterogeneity Revealed by Single-Cell Sequencing of Normal Fallopian Tube Epithelial Cells

The inter-differentiation between cell states promotes cancer cell survival under stress and fosters non-genetic heterogeneity (NGH). NGH is, therefore, a surrogate of tumor resilience but its quantification is confounded by genetic heterogeneity. Here we show that NGH in serous ovarian cancer (SOC) can be accurately measured when informed by the molecular signatures of the normal fallopian tube epithelium (FTE) cells, the cells of origin of SOC. Surveying the transcriptomes of ∼6,000 FTE cells, predominantly from non-ovarian cancer patients, identified 6 FTE subtypes. We used subtype signatures to deconvolute SOC expression data and found substantial intra-tumor NGH. Importantly, NGH-based stratification of ∼1,700 tumors robustly correlated with survival. Our findings lay the foundation for accurate prognostic and therapeutic stratification of SOC.

Immunophenotypes of pancreatic ductal adenocarcinoma: Meta-analysis of transcriptional subtypes

Pancreatic ductal adenocarcinoma (PDAC) is the most common malignancy of the pancreas and has one of the highest mortality rates of any cancer type with a 5-year survival rate of <5%. Recent studies of PDAC have provided several transcriptomic classifications based on separate analyses of individual patient cohorts. There is a need to provide a unified transcriptomic PDAC classification driven by therapeutically relevant biologic rationale to inform future treatment strategies. Here, we used an integrative meta-analysis of 353 patients from four different studies to derive a PDAC classification based on immunologic parameters. This consensus clustering approach indicated transcriptomic signatures based on immune infiltrate classified as adaptive, innate and immune-exclusion subtypes. This reveals the existence of microenvironmental interpatient heterogeneity within PDAC and could serve to drive novel therapeutic strategies in PDAC including immune modulation approaches to treating this disease.

Comprehensive transcriptomic analysis of cell lines as models of primary tumors across 22 tumor types

Cancer cell lines are a cornerstone of cancer research but previous studies have shown that not all cell lines are equal in their ability to model primary tumors. Here we present a comprehensive pan-cancer analysis utilizing transcriptomic profiles from The Cancer Genome Atlas and the Cancer Cell Line Encyclopedia to evaluate cell lines as models of primary tumors across 22 tumor types. We perform correlation analysis and gene set enrichment analysis to understand the differences between cell lines and primary tumors. Additionally, we classify cell lines into tumor subtypes in 9 tumor types. We present our pancreatic cancer results as a case study and find that the commonly used cell line MIA PaCa-2 is transcriptionally unrepresentative of primary pancreatic adenocarcinomas. Lastly, we propose a new cell line panel, the TCGA-110-CL, for pan-cancer studies. This study provides a resource to help researchers select more representative cell line models.

Cell lines are used ubiquitously in cancer research but how well they represent the tumor type they were derived from is variable. Here, the authors compare transcriptomic profiles of 22 tumor types and cell lines and propose a new comprehensive cell line panel for pan-cancer studies.

Patient-specific cancer genes contribute to recurrently perturbed pathways and establish therapeutic vulnerabilities in esophageal adenocarcinoma

The identification of cancer-promoting genetic alterations is challenging particularly in highly unstable and heterogeneous cancers, such as esophageal adenocarcinoma (EAC). Here we describe a machine learning algorithm to identify cancer genes in individual patients considering all types of damaging alterations simultaneously. Analysing 261 EACs from the OCCAMS Consortium, we discover helper genes that, alongside well-known drivers, promote cancer. We confirm the robustness of our approach in 107 additional EACs. Unlike recurrent alterations of known drivers, these cancer helper genes are rare or patient-specific. However, they converge towards perturbations of well-known cancer processes. Recurrence of the same process perturbations, rather than individual genes, divides EACs into six clusters differing in their molecular and clinical features. Experimentally mimicking the alterations of predicted helper genes in cancer and pre-cancer cells validates their contribution to disease progression, while reverting their alterations reveals EAC acquired dependencies that can be exploited in therapy.

Abstract 467: Deep single-cell RNA-seq of the putative cell of origin revealed a novel molecular subtype of high-grade serous ovarian cancer with poor prognosis

Accurate molecular classification in cancer is essential for rationalized therapy. However, achieving stable subtyping is highly challenging due to the underlying genomic complexity of tumors. An example of the most successful classification is in breast cancer where it is possible to link individual tumor types, basal vs luminal, to individual cell types of origin. However, the generalization of such an approach has not been easy because of the lack of knowledge about subtypes of the putative cells of origin. For instance, it has been difficult to establish a stable molecular classification of high-grade serous ovarian cancer (HGSOC). More recent evidence strongly supports the notion that HGSOC originates from the secretory cells in fallopian tube epithelium (FTE). However, whether or not there are multiple subtypes of secretory cells and how such subtypes are linked to tumor molecular subtypes have remained elusive. To address this issue, we applied single-cell RNA sequencing (Smart-Seq2) on around 4000 FTE cells from patients with ovarian or endometrial cancers. We developed and employed an in-house differential-expression-based clustering method, termed ClinCluster, for clinical samples that takes into account inter-patient variability. Apart from the established secretory and ciliated cell types, we discovered four novel subtypes of secretory cells and a rare intermediate cell type that co-expresses secretory and ciliated markers. We termed these four novel secretory subtypes as cell cycle, progenitor, epithelial-mesenchymal transition (EMT) and immunoreactive subtypes. Based on fifty-three biomarkers of these subtypes, we designed a panel of transcriptomic signatures and exploited them to deconvolute two bulk expression datasets of ovarian cancer: the RNA-seq dataset of TCGA (The Cancer Genome Atlas) and the microarray dataset of AOCS (Australian Ovarian Cancer Study). The deconvolution analysis revealed a poor-prognostic EMThigh subtype of HGSOC, which was not driven by copy number aberrations. Cox regression analysis indicated that the EMThigh tumors were strongly associated with poor overall survival in the two independent datasets tested (hazard ratio = 2.7, 95% confidence interval = 1.4 - 5.1, P = 0.002, n = 307 for TCGA; hazard ratio = 3.2, 95% confidence interval = 1.7 - 6.1, P = 0.0002, n = 253 for AOCS). Importantly, we found that the PI3K-Akt signaling pathway is upregulated in the EMThigh tumors (FDR &lt; 0.002 for the two independent datasets) giving a strong potential for effective rationalized therapies. In summary, our work portrays a new model for the landscape of fallopian tube epithelium, the putative origin of HGSOC, and illustrates that the cellular subtypes in FTE are related to the HGSOC tumor types with important prognostic and therapeutic implications.

Citation Format: Zhiyuan Hu, Abdulkhaliq Alsaadi, Nina Wietek, Laura Santana González, Christopher Yau, Ahmed Ashour Ahmed. Deep single-cell RNA-seq of the putative cell of origin revealed a novel molecular subtype of high-grade serous ovarian cancer with poor prognosis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 467.

Abstract P5-15-01: The use of 18F-FDG PET/CT as an initial staging procedure for stage II-III breast cancer reduces false positives, costs, and time to treatment: A multicenter value analysis in the I-SPY2 trial

Introduction: Diagnostic metastatic staging imaging (SI) for asymptomatic stage I-II patients (pts) is not routinely recommended, but is warranted in stage II-III pts with high risk biological subtypes, where previous trials have shown up to a 15% rate of de novo metastatic disease. NCCN guidelines endorse CT CAP and bone scan (STD) for stage III pts, but not PET/CT, and PET/CT is not covered in most parts of the country. We present data on the performance and value of PET/CT.

Methods: Data were available for 799 high risk clinical stage II-III pts screened for I-SPY2 at UCSF, Uminn, UAB, and Georgetown. Of these, 564 pts ranging in age from 25-81 (median = 48) had complete records that were retrospectively reviewed for SI and potential false positives (FP), defined as incidental findings on SI proven benign by subsequent workup. Economic evaluation was conducted from the payer perspective using the mean national 2018 Medicare Physician Fee Schedule and representative costs from the UCSF billing department. The incremental cost effectiveness ratio (ICER) measured the cost of using PET/CT per percent patient (pt) who avoided a FP.

Results: The rate of de novo metastatic disease was 4.8% (38/799), range 3.6-6.4%. Of the 564 pts with complete records, diagnostic SI varied significantly among the four sites (p &lt; 0.0001). STD was used for most pts at UAB (92.8%, 141/152) and Georgetown (85.7%, 54/63), while PET/CT was used for most pts at UCSF (86.6%, 226/261) and Uminn (63.6%, 56/88). Chest X-ray was used for 29.5% (26/88) at Uminn. There were significantly more pts with FP in the group that received STD (22.1%, 51/231) vs. PET/CT (11.1%, 33/298) (p &lt; 0.05). Mean time between incidental finding on SI to determination of FP was 10.8 days. When controlling for institution, mean time from cancer diagnosis to initiation of neoadjuvant chemotherapy was significantly different between STD (44.3 days) and PET/CT (37.5 days) groups (p &lt; 0.05). When aggregating the four sites using mean costs from the 2018 Medicare Physician Fee Schedule, the mean cost/pt was $1132 for STD vs. $1477 for PET/CT. The mean increase in price from baseline SI costs due to FP workup was $216 (23.6%) for STD vs. $65 (4.6%) for PET/CT. The ICER was $31 per percent pt who avoided a FP. When analyzing UCSF pts alone using representative reimbursements from Medicare, the mean cost/pt was $1236 for STD vs. $1081 for PET/CT; using representative reimbursements from Anthem Blue Cross, the mean cost/pt was $3080 for STD vs. $1662 for PET/CT. The ICERs were -$10 and -$95 per percent pt who avoided a FP, respectively.

Conclusion: As compared to STD metastatic staging workup, PET/CT added value by decreasing FP two-fold. This reduced direct costs of FP workup procedures that took a mean time of 10.8 days to resolve. PET/CT also accelerated treatment start. Reducing the chance of FP workup for metastatic disease is of enormous value to pts. Our data establish the value of PET/CT for staging in our high risk clinical stage II-III trial population and highlight the need for alignment between hospital pricing strategies and payer coverage policies in order to deliver high value care to pts.

Citation Format: Hyland CJ, Varghese F, Yau C, Beckwith H, Khoury K, Varnado W, Hirst G, Chien J, Yee D, Isaacs C, Forero-Torres A, Esserman L, Melisko M, I-SPY2 Consortium. The use of 18F-FDG PET/CT as an initial staging procedure for stage II-III breast cancer reduces false positives, costs, and time to treatment: A multicenter value analysis in the I-SPY2 trial [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P5-15-01.

Abstract PD4-04: Role of breast MRI in predicting pathologically negative nodes after neoadjuvant chemotherapy in cN0 patients in the I-SPY2 trial

Background

In clinically node-negative (cN0) breast cancer patients with triple negative (TN) and HER2+ disease and breast pathological complete response (breast pCR), low rates of nodal positivity after neoadjuvant chemotherapy (NAC) have been demonstrated. In these patients, the omission of surgical axillary staging has been proposed. However, this information is not routinely known preoperatively. We aimed to validate the correlation between pathologic breast response and pathologic nodal status, and evaluate the relationship between response of the breast tumor on MRI and pathologic nodal status after NAC in cN0 patients in the I-SPY2 trial.

Methods

We identified all patients with cT1-4 cN0 breast cancer prior to NAC from graduated arms of the I-SPY2 trial, a prospective neoadjuvant chemotherapy trial. Absence of residual disease post-NAC was defined as longest diameter (LD) of 0 mm on MRI. Breast pCR was defined as the absence of invasive tumor in the breast at surgery. Associations between ypN0 and patient, MRI, and tumor characteristics were assessed using chi-square tests and univariate regression.

Results

Of 365 cT1-4 cN0 patients included, 128 had HR+/HER2- tumors (35%), 60 HR+/HER2+ tumors (16%), 34 HR-/HER2+ tumors (9%) and 143 TN tumors (39%). Overall, 283 patients (78%) were ypN0 after NAC and 152 patients (42%) had a breast pCR. ypN0 rate was higher in patients with a breast pCR than those with residual disease (93% vs 66%, p&lt;0.001). Patients with HR-/HER2+ and TN tumors were more likely to be ypN0 (97% and 87% respectively) than patients with HR+/HER2- and HR+/HER2+ disease (66% and 71% respectively, p&lt;0.001). Other characteristics associated with ypN0 were tumor grade (grade I 57%, grade II 66%, grade III 84%; p=0.002), MammaPrint Classification (High Risk 1 68% and High Risk 2 87%; p&lt;0.001) and absence of residual tumor in the breast on MRI (87% vs 72% in patients with evidence of tumor on MRI post-NAC/pre-surgery; p=0.003).

In patients with HR-/HER2+, HR+/HER2+, HR-/HER2+ or TN disease and a breast pCR, ypN0 rate was respectively 82%, 96%, 96% and 97% (table 1). In patients with HR+/HER2-, HR+/HER2+, HR-/HER2+ or TN disease and with no evidence of residual disease in the breast on MRI, rate of ypN0 was 71%, 80%, 94% and 96% respectively.

Conclusion

In cT1-4 cN0 breast cancer patients with HR+/HER2+, HR-/HER2+ and TN tumors and a breast pCR, ypN0 rates after NAC are extremely high. In patients with HR-/HER2+ and TN tumors with no residual breast disease on MRI after NAC and pre-surgery, ypN0 rates are high enough to consider omission of axillary surgery. In patients with HR+ tumors, MRI is unsufficiently predictive for pathological response and can therefore not be used to select ypN0 patients. Research on the prediction of ypN0 in cN+ I-SPY2 patients is ongoing.

Nodal status in patients with pCR and absence of residual disease on MRI Number of positive nodesBreast Cancer Subtype0123AllBreast pCR     HR+/HER2-27(82)2(6)4(12)033(100)HR+/HER2+24(96)01(4)025(100)HR-/HER2+24(96)1(4)0025(100)TN67(97)2(3)0069(100)Absence of residual disease on MRI     HR+/HER2-24(71)7(21)3(9)034(100)HR+/HER2+16(80)3(15)01(5)20(100)HR-/HER2+15(94)1(6)0016(100)TN54(96)2(4)0056(100)

Citation Format: van der Noordaa ME, Esserman L, Yau C, Mukhtar R, Price E, Hylton N, Abe H, Wolverton D, Crane EP, Ward KA, Nelson M, Niell BL, Oh K, Brandt KR, Bang DH, Ojeda-Fournier H, Eghtedari M, Sheth PA, Bernreuter WK, Umphrey H, Rosen MA, Dogan B, Yang W, Joe B, van 't Veer L, Hirst G, Lancaster R, Wallace A, Alvaredo M, Symmans F, Asare S, Boughey JC, I-SPY2 Consortium. Role of breast MRI in predicting pathologically negative nodes after neoadjuvant chemotherapy in cN0 patients in the I-SPY2 trial [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr PD4-04.

Abstract P2-14-01: The impact of local therapy on locoregional recurrence in women with high risk breast cancer in the neoadjuvant I-SPY2 TRIAL

Background: In women with breast cancer receiving neoadjuvant chemotherapy, residual cancer burden (RCB) predicts distant recurrence and survival. In those with high risk tumors, locoregional recurrence (LRR) remains a concern, and has been associated with type of local therapy received. We evaluated the impact of local therapy on LRR in the ISPY-2 TRIAL.

Methods: Data were analyzed in Stata 14.2, using Chi2 test, log rank test, and a Cox proportional hazards model. RCB was considered a categorical variable (0/1 versus 2/3), as described in prior publications. Breast surgery categories were lumpectomy +/- radiotherapy, or mastectomy +/- radiotherapy. Axillary surgery was defined as sentinel lymph node (SLN) surgery (≤6 nodes removed) or axillary dissection (&gt;6 nodes).

Results: Follow up data from the I-SPY2 TRIAL were available for 630 patients (median follow up 2.76 yrs, range 0.4-7.2). Type of local therapy was significantly associated with clinical stage at presentation, with stage III patients most frequently undergoing mastectomy + radiation (p&lt;0.001). Women with higher RCB were more likely to undergo mastectomy than those with lower RCB (61.3% vs 48.8% mastectomy rate, p=0.002), and more likely to receive adjuvant radiotherapy (62.0% vs 53.9%, p=0.048). There was no association between clinical stage, type of surgery, or radiotherapy and LRR (Table). Higher RCB was significantly associated with LRR, with 3 year locoregional recurrence free rate of 95.1% in RCB 0/1 versus 89.9% in RCB 2/3 (p=0.003).

In a Cox model adjusting for clinical stage, tumor subtype, surgical therapy, RCB status, nodal radiation, and age, significant predictors for LRR were tumor subtype and RCB status. Hazard ratio (HR) for LRR in those with RCB 0/1 was 0.39 compared to those with RCB 2/3 (95% CI 0.17-0.87, p=0.021). There was no difference in LRR between breast conservation and mastectomy; within the breast conservation group, those who had lumpectomy alone had higher hazard of LRR compared to those having lumpectomy + radiation (HR 3.1, 95% CI 1.1-9.2, p=0.043).

Conclusions: Extent of surgical therapy was not associated with local tumor control, regardless of advanced tumor stage at presentation. Rather, tumor biology and response to therapy were the best predictors of LRR. These data highlight the opportunity to minimize the morbidity of extensive surgical therapy for patients with excellent response to systemic therapy.

LRR rates by clinical features and treatment status FrequencyLRR RateP valueClinical Stage  0.5I240 (47.5%)5.8% II185 (36.6%)8.7% III80 (15.8%)6.3% Tumor Subtype  0.014ER+PR+Her2-161 (26.4%)3.1% ER+PR-Her2-56 (9.2%)3.6% Her2+176 (28.9%)6.3% Triple negative216 (35.5%)11.1% Local therapy  0.169Lumpectomy85 (13.5%)11.8% Lumpectomy with radiation198 (31.4%)5.6% Mastectomy173 (27.5%)5.2% Mastectomy with radiation174 (27.6%)8.6% Axillary surgery  0.23None5 (0.8%)20% SLN329 (52.2%)5.8% ALND296 (47%)8.5% Axillary radiation  0.535Yes42 (6.7%)9.5% No588 (93.3%)7.0% Axillary management  0.2No surgery or radiation5 (0.8%)20.0% SLN312 (50%)5.3% SLN+Axillary radiation17 (2.7%)8.3% ALND271 (43%)10.3% ALND+Axillary radiation25 (4%)5.4% RCB  0.0020/1293 (50.1%)3.8% 2/3292 (49.9%)10.3% 

Citation Format: Silverstein J, Suleiman L, Yau C, Price ER, Singhrao R, Yee D, DeMichele A, Isaacs C, Albain KS, Chien AJ, Forero-Torres A, Wallace AM, Pusztai L, Ellis ED, Elias AD, Lang JE, Lu J, Han HS, Clark AS, Korde L, Nanda R, Northfelt DW, Khan QJ, Viscusi RK, Euhus DM, Edmiston KK, Chui SY, Kemmer K, Wood WC, Park JW, Liu MC, Olopade O, Leyland-Jones B, Tripathy D, Moulder SL, Rugo HS, Schwab R, Lo S, Helsten T, Beckwith H, I-SPY 2 TRIAL Consortium, Berry DA, Asare SM, Esserman LJ, Boughey JC, Mukhtar RA. The impact of local therapy on locoregional recurrence in women with high risk breast cancer in the neoadjuvant I-SPY2 TRIAL [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-14-01.

Abstract P2-07-03: Refining neoadjuvant predictors of three year distant metastasis free survival: Integrating volume change as measured by MRI with residual cancer burden

Background: Patients achieving a pathologic complete response (pCR) following neoadjuvant therapy have significantly improved event-free survival relative to those who do not; and pCR is an FDA-accepted endpoint to support accelerated approval of novel agents/combinations in the neoadjuvant treatment of high risk early stage breast cancer. Previous studies have shown that recurrence risk increased with increasing burden of residual disease (as assessed by the RCB index). As well, these studies suggest that patients with minimum residual disease (RCB-I class) also have favorable outcomes (comparable to those achieving a pCR) within high risk tumor subtypes. In this study, we assess whether integrating RCB with MRI functional tumor volume (FTV), which in itself is prognostic, can improve prediction of distant recurrence free survival (DRFS); and identify a subset of patients with minimal residual disease with comparable DRFS as those who achieved a pCR. Imaging tools can then be used to identify the subset that will do well early and guide the timing of surgical therapy.

Method: We performed a pooled analysis of 596 patients from the I-SPY2 TRIAL with RCB, pre-surgical MRI FTV data and known follow-up (median 2.5 years). We first assessed whether FTV predicts residual disease (pCR or pCR/RCB-I) using ROC analysis. We applied a power transformation to normalize the pre-surgical FTV distribution; and assessed its association with DRFS using a bi-variate Cox proportional hazard model adjusting for HR/HER2 subtype. We also fitted a bivariate Cox model of RCB index adjusting for subtype; and assessed whether adding pre-surgical FTV to this model further improves association with DRFS using a likelihood ratio (LR) test. For the Cox modeling, penalized splines approximation of the transformed FTV and RCB index with 2 degrees of freedom was used to allow for non-linear effects of FTV and RCB on DRFS.

Result: Pre-surgical MRI FTV is significantly associated with DRFS (Wald p&lt;0.00001), and more effective at predicting pCR/RCB-I than predicting pCR alone (AUC: 0.72 vs. 0.65). Larger pre-surgical FTV remains associated with worse DRFS adjusting for subtype (Wald p &lt;0.00001). The RCB index is also significantly associated with DRFS adjusting for subtype (Wald p&lt;0.00001). Adding FTV to a model containing RCB and subtype further improves association with DRFS (LR p=0.0007). RCB-I patients have excellent DRFS (94% at 3 years compared to 95% in the pCR group). Efforts are underway to identify an optimal threshold for dichotomizing pre-surgical FTV and FTV change measures for use in combination with pCR/RCB-I class to generate integrated RCB (iRCB) groups as a composite predictor of DRFS.

Conclusion: Pre-surgical MRI FTV is effective at predicting minimal residual disease (RCB0/I) in the I-SPY 2 TRIAL. Despite the association between FTV and RCB, FTV appears to provide independent added prognostic value (to RCB and subtype), suggesting that integrating MRI volume measures and RCB into a composite predictor may improve DRFS prediction.

Citation Format: Hylton NM, Symmans WF, Yau C, Li W, Hatzis C, Isaacs C, Albain KS, Chen Y-Y, Krings G, Wei S, Harada S, Datnow B, Fadare O, Klein M, Pambuccian S, Chen B, Adamson K, Sams S, Mhawech-Fauceglia P, Magliocco A, Feldman M, Rendi M, Sattar H, Zeck J, Ocal I, Tawfik O, Grasso LeBeau L, Sahoo S, Vinh T, Yang S, Adams A, Chien AJ, Ferero-Torres A, Stringer-Reasor E, Wallace A, Boughey JC, Ellis ED, Elias AD, Lang JE, Lu J, Han HS, Clark AS, Korde L, Nanda R, Northfelt DW, Khan QJ, Viscusi RK, Euhus DM, Edmiston KK, Chui SY, Kemmer K, Wood WC, Park JW, Liu MC, Olopade O, Tripathy D, Moulder SL, Rugo HS, Schwab R, Lo S, Helsten T, Beckwith H, Haugen PK, van't Veer LJ, Perlmutter J, Melisko ME, Wilson A, Peterson G, Asare AL, Buxton MB, Paoloni M, Clennell JL, Hirst GL, Singhrao R, Steeg K, Matthews JB, Sanil A, Berry SM, Abe H, Wolverton D, Crane EP, Ward KA, Nelson M, Niell BL, Oh K, Brandt KR, Bang DH, Ojeda-Fournier H, Eghtedari M, Sheth PA, Bernreuter WK, Umphrey H, Rosen MA, Dogan B, Yang W, Joe B, I-SPY 2 TRIAL Consortium, Yee D, Pusztai L, DeMichele A, Asare SM, Berry DA, Esserman LJ. Refining neoadjuvant predictors of three year distant metastasis free survival: Integrating volume change as measured by MRI with residual cancer burden [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-07-03.

Abstract PD4-03: MRI detection of residual disease following neoadjuvant chemotherapy (NAC) in the I-SPY 2 TRIAL

Background: Detecting residual disease accurately using MRI after NAC to identify both responders and non-responders is essential for de-escalating therapy or redirecting patients to more effective treatment. The purpose of this study is to determine if the combination of longest diameter (LD) and functional tumor volume (FTV) from dynamic contrast enhanced (DCE-) MRI is superior to FTV alone or LD alone for assessing treatment response after neoadjuvant therapy in breast cancer patients.

Methods: Data from patients in the graduated drug arms of the I-SPY 2 trial were included in the analysis. Both LD and FTV were assessed using DCE-MRI after neoadjuvant therapy. LD was measured by the site radiologist as the longest dimension of the enhanced area on early post-contrast images. Functional tumor volume (FTV) was assessed as the sum of voxels with enhancement above specific thresholds within the pre-defined region-of-interest (ROI). A linearized variable was derived to represent the combination of FTV and LD. The area under the receiver operating characteristic curve (AUC) was used to evaluate the assessment of treatment response, pathologic complete response (pCR), defined as no invasive disease in the breast and lymph nodes, and in-breast pCR, defined as no invasive disease in the breast only. The analysis was performed in the full cohort and in breast cancer subtype defined by hormone receptor status and HER2 status.

Results: Among the patient cohort of N=675 with FTV and LD, 247 (37%) did and 428 (41%) did not achieve pCR after neoadjuvant therapy. pCR rates varied among HR/HER2 subtypes (HR+/HER2-: 19%; HR+/HER2+: 38%; HR-/HER2+: 71%; HR-/HER2- (triple negative, TN): 43%). In-breast pathologic complete response rates were slightly higher in each group (full: 41%; HR+/HER2-: 23%; HR+/HER2+: 43%; HR-/HER2+: 72%; HR-/HER2-: 49%). Table 1 shows AUCs for assessing pCR using FTV alone, LD alone, and the variable combining FTV and LD. Higher AUCs were observed in all patient groups using the combined variable. AUC of 0.79 (95% CI: 0.77, 0.81) was observed for the combined variable to assess pCR in the full cohort. AUCs varied from 0.69 to 0.86 among HR/HER2 subgroups (HR+/HER2-: 0.69; HR+/HER2+: 0.74; HR-/HER2+: 0.86; HR-/HER2-: 0.80), with no difference in assessing pCR or in-breast pCR. The performance is best for the HR- subtypes.

Conclusions: Both FTV and LD can be used in the assessment of invasive disease residual after neoadjuvant therapy. The combined variable of FTV and LD achieved highest AUCs, compared to using individual variable alone. Tools to improve performance in the HR+ subsets are underway.

AUCs of MR measurements for identifying pCR  FTV alone (95% CI)LD alone (95% CI)Combined (95% CI)FullWith subtype adj.0.73 (0.71, 0.75)0.77 (0.74, 0.79)0.79 (0.77, 0.81)FullWithout subtype adj0.69 (0.65, 0.73)0.72 (0.68, 0.76)0.75 (0.71, 0.79)HR+/HER2- 0.68 (0.60, 0.77)0.68 (0.59, 0.77)0.69 (0.61, 0.77)HR+/HER2+ 0.65 (0.56, 0.75)0.72 (0.64, 0.80)0.74 (0.66, 0.82)HR-/HER2+ 0.69 (0.55, 0.83)0.82 (0.71, 0.92)0.86 (0.77, 0.95)HR-/HER2- (TN) 0.72 (0.66, 0.79)0.73 (0.67, 0.80)0.80 (0.74, 0.85)

Citation Format: Li W, Newitt D, Yun BL, Kornak J, Joe B, Yau C, Abe H, Wolverton D, Crane E, Ward K, Nelson M, Niell B, Drukteinis J, Oh K, Brandt K, Bang DH, Ojeda H, Eghtedari M, Sheth P, Bernreuter W, Umphrey H, Rosen M, Dogan B, Yang W, Esserman L, Hylton N. MRI detection of residual disease following neoadjuvant chemotherapy (NAC) in the I-SPY 2 TRIAL [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr PD4-03.

Abstract P3-01-02: Detection of circulating tumor cells (CTC) in blood and disseminated tumor cells (DTC) in bone marrow at surgery identifies breast cancer patients (pts) with long-term risk of distant recurrence and breast cancer-specific death

We examined the prognostic impact of CTCs and DTCs detected at the time of definitive surgery in pts diagnosed with early breast cancer (EBC).

Methods: Blood and bone marrow samples from 742 treatment-naïve EBC pts, not eligible for neoadjuvant therapy, were collected immediately prior to surgery. 87% were hormone receptor (HR)-positive, and 71% were node-negative. DTCs (n=584) were enumerated using an EPCAM-based method involving immunomagnetic enrichment and flow cytometry (IE/FC). CTCs were enumerated either by IE/FC (n=288) or CellSearch (n=380). Optimal cutoffs for CTC-/DTC-positivity were selected using Monte-Carlo cross validation. Multivariate Cox regression analysis was performed to determine correlation between levels of CTCs/DTCs vs. distant recurrence-free survival (DRFS) and breast cancer-specific survival (BCSS). The overall median follow-up was 7.1 years for DRFS and and 9.1 years for BCSS, but extended up to 13.3 years in subset analyses (Table 1).

Results: CTC-positivity by CellSearch was associated with HER2-positivity (Fisher p=0.01). Using optimized cutoffs in multivariate analyses, we found that CTC-positive pts by CellSearch had a statistically significant increased risk of distant recurrence (HR 4.93, p=0.0067). Moreover, pts who were CTC-positive by IE/FC had a statistically significant increased risk of breast cancer-specific death (HR=3.54, p=0.0138). DTC status, by itself, was not prognostic; however, when combined with CTC status by IE/FC (n=273), positive detection for both (CTC+DTC+) was significantly associated with increased risk of distant recurrence (HR=3.09, p=0.0270) and breast cancer-specific death (HR=4.55, p=0.0205).

Table 1.Multivariate analysis to determine the prognostic significance of CTCs and DTCs detected at the time of surgery in treatment naive early breast cancer patients. Adjusted for age at diagnosis, tumor size, pathologic stage, HR and HER2 status, node status and grade.  DRFS  BCSS  Variable and Method% positiveHR [95% CI]Wald p-valueMedian f/u [range] Years*HR [95% CI]Wald p-valueMedian f/u [range] Years*CTC+ vs. CTC- by CellSearch94.93[1.56-15.6]0.00676.4 [0.16-13.8]4.50[0.76-26.5]0.09627.5 [0.71-15.0]CTC+ vs. CTC- by IE/FC401.92[0.93-3.95]0.07599.8 [0.09-18.5]3.54[1.29-9.72]0.013813.3 [1.93-18.5]DTC+ vs. DTC- by IE/FC181.46[0.75-2.81]0.26317.5 [0.09-18.5]1.48[0.64-3.42]0.35429.8 [1.55-18.5]CTC+DTC+ vs. CTC-DTC- by IE/FC8**3.09[1.14-8.40]0.02709.8 [0.09-18.5]4.55[1.26-16.39]0.020513.3 [1.93-18.5]*f/u - follow-up; **double positive

Conclusions: We demonstrate the impact of quantitative evaluation of CTCs and DTCs by IE/FC. Our large single institution dataset, in which CTCs and DTCs have been contemporaneously quantitated, has the longest patient follow-up. Simultaneous detection of CTCs and DTCs at the time of definitive surgery in treatment naïve EBC pts is an independent prognostic factor associated with increased long-term risk of distant recurrence and death due to breast cancer. Given the lack of early endpoints for low-risk patients, liquid biopsy may be an important consideration for future studies.

Citation Format: Magbanua MJM, Yau C, Wolf D, Lee JS, Chattopadhyay A, Scott JH, Yoder E, Hwang S, Alvarado M, Ewing CA, Delson AL, van't Veer L, Esserman L, Park JW. Detection of circulating tumor cells (CTC) in blood and disseminated tumor cells (DTC) in bone marrow at surgery identifies breast cancer patients (pts) with long-term risk of distant recurrence and breast cancer-specific death [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-01-02.

Abstract PD7-06: Molecular subtypes of invasive lobular breast cancer in the I-SPY2 TRIAL

Background: Invasive lobular carcinoma (ILC) of the breast has distinct histological and molecular variations compared to invasive ductal carcinoma (IDC), including absence of the adhesion protein E-cadherin. Recently, molecular subtypes within ILC have been described, with an analysis from The Cancer Genome Atlas (Ciriello et al) identifying three distinct groups within ILC based on gene expression—reactive-like, immune-related, and proliferative. In this study, we applied this 60-gene classifier to a locally advanced cohort of ILC and mixed ILC/IDC cases from patients screening for the I-SPY 2 neoadjuvant chemotherapy trial.

Methods: The I-SPY 2 TRIAL is open to women with more locally advanced, clinically/molecularly (as assessed by MammaPrint) high risk breast cancer. HR+HER2- MammaPrint Low risk patients ineligible for I-SPY 2 randomization are invited to join a MP Low risk registry. 131 ILC and mixed ILC/IDC tumors from these cohorts (I-SPY 2: n=80; low risk registry: n=51) with pre-treatment Agilent microarrays were available for analysis. We used the Classification to Nearest Centroid technique to assign TCGA subtype to our cohort. We assessed association between TCGA subtype, clinical covariates and response to therapy using a chi-square test. We also evaluated the Euclidean distance between each sample and the three subtype centroids. In an exploratory analysis, we used consensus clustering based on the 1000 most varying genes within the HR+HER2- I-SPY ILC cases to generate new unsupervised groupings, and assessed the concordance with the TCGA reactive-like, immune-related and proliferative subtype assignments.

Results: Of the 131 patients included, most (79%) were HR+HER2-, 11% were HR+HER2+, 2% were HR-HER2+ and 8% were HR-HER2- for a total of 10% HR-. 66 were pure ILC, while 65 were mixed ILC/IDC. Upon applying the TCGA 60-gene classifier, the distribution of ILC subtypes was as follows: 33 (25%) were classified as reactive-like, 50 (38%) were immune-related, and 48 (37%) were proliferative. 64% of triple negative cases were reactive-like; while the HR+HER2- and HER2+ cases were more likely to be in the proliferative or immune-related subtype (p=0.037). Among the 80 I-SPY 2 cases, the overall pathologic complete response rate was low (16%) but equivalent to the overall HR+HER2- I-SPY2 population (16%). This did not differ across the groups defined by the TCGA ILC subtypes (p=0.79).

Interestingly, a subset of cases assigned as reactive-like and immune-related were of similar distance to the proliferative subtype centroid as patients assigned to the proliferative subtype. When we used consensus clustering to identify new subsets within our locally advanced ILC cohort, our unsupervised groupings had only 32% concordance with the TCGA ILC subtype assignments.

Conclusion: The low concordance between our consensus cluster groupings and the TCGA subtype groupings may reflect underlying differences within a locally advanced cohort of ILC cases, like I-SPY, that may not be captured in the 60-gene classifier developed from the overall lower stage TCGA cohort. These findings suggest that considerable molecular heterogeneity exists in lobular cancers, which merits further investigation.

Citation Format: Zhu Z, Yau C, Chien AJ, Haddad T, Esserman LJ, van't Veer L, Mukhtar RA, I-SPY2 Consortium. Molecular subtypes of invasive lobular breast cancer in the I-SPY2 TRIAL [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr PD7-06.

Abstract P4-01-12: Low peripheral blood CD4/CD8 ratio at the time of surgery is a negative long-term prognostic factor in women with early stage breast cancer

It is hypothesized that cancer prognosis may be related to the functional status of the immune system. We examined the correlation between peripheral blood CD4/CD8 ratio measured at the time of surgery and clinical outcome in patients diagnosed with early stage breast cancer.

Patient and Methods

Peripheral blood from 57 treatment-naïve early breast cancer patients, not eligible for neoadjuvant chemotherapy, was collected on the day of definitive surgery. CD4+ and CD8+ T cells were enumerated using flow cytometry and the ratio between the two immune cell populations was calculated. Cox regression analyses were performed to determine the relationship between CD4/CD8 ratio vs. distant disease-free survival (DRFS), breast cancer-specific survival (BCSS) and overall survival (OS). The median follow-up times were 10.1 years (range: 0.4-17.5) and 15.0 (range: 1.0-18.5) for DRFS and BCSS/OS, respectively.

Results

The patients' mean age at diagnosis was 54 years old (range: 31-78). 82% were hormone receptor-positive, 21% HER2-positive, and 61% node-negative. The median CD4/CD8 ratio was 2; and a ratio ≤ 2 was considered low. CD4/CD8 ratio was not associated with any of the clinicopathologic variable examined. Multivariate analysis using a survival model that adjusted for potential confounding factors (age, tumor size, grade, stage, hormone receptor, HER2, lymph-node status) revealed that patients with low CD4/CD8 ratio have statistically significant increased risk of distant recurrence (DRFS HR 5.3, Wald p=0.0381) and death (OS HR 3.8 Wald p=0.0271).

Conclusions

Immune dysfunction at the time surgery is correlated with long-term increased risk for metastatic recurrence and death. Larger clinical studies are warranted to confirm the results of this study.

Citation Format: Magbanua MJM, Yau C, Scott JH, van't Veer L, Park JW, Esserman L, Campbell M. Low peripheral blood CD4/CD8 ratio at the time of surgery is a negative long-term prognostic factor in women with early stage breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P4-01-12.

Abstract P1-15-02: Withdrawn

This abstract was withdrawn by the authors.

Citation Format: Schwab R, Clark A, Yau C, Wolf D, Chien AJ, Majure M, Ewing C, Wallace A, Roesch E, Helsten T, Forero A, Stringer-Reasor E, Vaklavas C, Nanda R, Jaskowiak N, Boughey J, Haddad T, Han H, Lee C, Albain K, Isaacs C, Elias A, Ellis E, Shah P, Lang J, Lu J, Tripathy D, Kemmer K, Yee D, Haley B, Korde L, Edmiston K, Northfelt D, Viscusi R, Khan Q, I-SPY 2 Consortium, Symmans WF, Perlmutter J, Hylton N, Rugo H, Melisko M, Wilson A, Singhrao R, Asare S, van't Veer L, DeMichele A, Berry D, Esserman L. Withdrawn [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P1-15-02.