Search

Purpose: Neoadjuvant treatment (NAT) of early breast cancer is increasingly being used to downstage tumors, allowing improved chances of breast-conserving surgery.Here we combine data from multiple studies to identify pre-treatment and on-treatment biomarkers of response to NAT with the potential to lead to more efficient patient stratification. Methods: We pool and analyse 10 independent NAT studies that have publicly available gene expression data (1861 samples, 1020 patients). Differential gene expression analysis was conducted on the pooled samples to derive a NAT response signature (NRS) and two NAT response subtypes. The NRS was then used along with additional variables to train a NAT response classifier. We use an additional 4 studies (418 samples, 258 patients) to further evaluate the performance of our classifier. Results: We identified 166 deferentially expressed genes between responders and non-responders, which are mainly involved in cell cycle and DNA repair pathways. We derive two molecular subtypes associated with NAT response and other clinical predictors. Our NAT response classifier achieves an Area Under the Curve (AUC) of 0.79 on a held-out test set (134 samples). Remarkably, in one external validation study, AUC increased from 0.64 to 0.82 when focusing on the estrogen receptor positive (ER+) samples only. Conclusion: We report a set of markers that are differentially expressed in NAT responders and demonstrate how they can be used to predict response to NAT for ER+ patients with early stage breast cancer, which might allow for improved risk stratification, surveillance or different treatments.

Type 2 diabetes mellitus (T2D) is one of the most prevalent diseases in the world and presents a major health and economic burden, a notable proportion of which could be alleviated with improved early prediction and intervention. While standard risk factors including age, obesity, and hypertension have shown good predictive performance, we show that the use of CpG DNA methylation information leads to a significant improvement in the prediction of 10-year T2D incidence risk. Whilst previous studies have been largely constrained by linear assumptions and the use of CpGs one-at-the-time, we have adopted a more flexible approach based on a range of linear and tree-ensemble models for classification and time-to-event prediction. Using the Generation Scotland cohort (n=9,537) our best performing model (Area Under the Curve (AUC)=0.880, Precision Recall AUC (PRAUC)=0.539, McFaddens R2=0.316) used a LASSO Cox proportional-hazards predictor and showed notable improvement in onset prediction, above and beyond standard risk factors (AUC=0.860, PRAUC=0.444 R2=0.261). Replication of the main finding was observed in an external test dataset (the German-based KORA study, p=3.7x10-4). Tree-ensemble methods provided comparable performance and future improvements to these models are discussed. Finally, we introduce MethylPipeR, an R package with accompanying user interface, for systematic and reproducible development of complex trait and incident disease predictors. While MethylPipeR was applied to incident T2D prediction with DNA methylation in our experiments, the package is designed for generalised development of predictive models and is applicable to a wide range of omics data and target traits.

Tumour mutation burden and other exome-wide biomarkers are used to determine which patients will benefit from immunotherapy. However, the cost of whole exome sequencing limits the widespread use of such biomarkers. Here, we introduce a data-driven framework for the design of targeted gene panels for estimating a broad class of biomarkers including tumour mutation burden and tumour indel burden. Our first goal is to develop a generative model for the profile of mutation across the exome, which allows for gene- and variant type-dependent mutation rates. Based on this model, we then propose a procedure for constructing biomarker estimators. Our approach allows the practitioner to select a targeted gene panel of prespecified size and construct an estimator that only depends on the selected genes. Alternatively, our method may be applied to make predictions based on an existing gene panel, or to augment a gene panel to a given size. We demonstrate the excellent performance of our proposal using data from three non small-cell lung cancer studies, as well as data from six other cancer types.

We introduce a novel data-driven framework for the design of targeted gene panels for estimating exome-wide biomarkers in cancer immunotherapy. Our first goal is to develop a generative model for the profile of mutation across the exome, which allows for gene- and variant type-dependent mutation rates. Based on this model, we then propose a new procedure for estimating biomarkers such as tumour mutation burden and tumour indel nurden. Our approach allows the practitioner to select a targeted gene panel of a prespecified size, and then construct an estimator that only depends on the selected genes. Alternatively, the practitioner may apply our method to make predictions based on an existing gene panel, or to augment a gene panel to a given size. We demonstrate the excellent performance of our proposal using data from three non-small cell lung cancer studies, as well as data from six other cancer types., Comment: 21 pages, 10 figures

We derive a new asymptotic expansion for the global excess risk of a local-$k$-nearest neighbour classifier, where the choice of $k$ may depend upon the test point. This expansion elucidates conditions under which the dominant contribution to the excess risk comes from the decision boundary of the optimal Bayes classifier, but we also show that if these conditions are not satisfied, then the dominant contribution may arise from the tails of the marginal distribution of the features. Moreover, we prove that, provided the $d$-dimensional marginal distribution of the features has a finite $\rho$th moment for some $\rho > 4$ (as well as other regularity conditions), a local choice of $k$ can yield a rate of convergence of the excess risk of $O(n^{-4/(d+4)})$, where $n$ is the sample size, whereas for the standard $k$-nearest neighbour classifier, our theory would require $d \geq 5$ and $\rho > 4d/(d-4)$ finite moments to achieve this rate. These results motivate a new $k$-nearest neighbour classifier for semi-supervised learning problems, where the unlabelled data are used to obtain an estimate of the marginal feature density, and fewer neighbours are used for classification when this density estimate is small. Our worst-case rates are complemented by a minimax lower bound, which reveals that the local, semi-supervised $k$-nearest neighbour classifier attains the minimax optimal rate over our classes for the excess risk, up to a subpolynomial factor in $n$. These theoretical improvements over the standard $k$-nearest neighbour classifier are also illustrated through a simulation study., Comment: 60 pages

mRNA transcriptomic markers have become a key tool in the classification of breast cancer and deciding treatment regimens for patients. However, limited studies have evaluated markers that might be predictive of response to neoadjuvant treatment (chemotherapy and endocrine treatment). Defining such key markers that are associated with treatment response could offer new insights on the relative molecular mechanisms and provide more tailored treatment regimens to target pathways that might be over-represented. We aimed to identify mRNA expression markers associated with breast cancer patients’ response to neoadjuvant treatment from a pool of studies in which tumor samples were collected at pre-treatment and on-treatment timepoints. We collated 1194 pre- and on-treatment samples (721 unique patients) from 9 publicly available gene expression datasets that met our inclusion criteria. The standardized gene expression values from each study were merged in a global matrix of 1551 genes and 1194 samples. Differential Gene Expression Analysis adjusted for timepoint, pam50 subtype (as predicted by the genefu package in R), treatment and batch effects was conducted. 14 significantly (FDR = 0.05) differentially expressed were identified (CCNA2, RFC3, VRK1, PSMB2, ALDH7A1, RRM2, ADRM1, TSPAN4, ZNF473, RNH1, HDAC1, CDK1, SMC1A and TOR1AIP1) when responders and non-responders were compared, some of which are known drug targets (e.g. PSMB2/carfilzomib). Using the set of identified genes we performed Monte-Carlo consensus clustering on the full set (optimal number of clusters: 6). Clusters were significantly associated with response (p = 2·10-8), timepoint (p < 10-15) and pam50 subtype (p < 10-15), but not treatment (p = 0.35). We then split our data into a training, a validation and a test set (776, 299, 119 samples respectively) and used the 14 genes (alone or in combination with metadata) as predictors to fit three types of machine learning models (lasso-regularized Logistic Regression, Decision Trees and Support Vector Machines). Support Vector Machines demonstrated the best classification performance on the validation set (75% classification accuracy) and achieved accuracy of 80% on the test set. Pathway analysis based on the identified genes revealed enriched nuclear membrane organization, protein deubiquitination, DNA replication and histone modification pathways. Prediction of response to treatment at baseline or mid-treatment can aid in patient stratification in the neoadjuvant setting and separate patients who would benefit from treatment the most and could undergo a less extensive surgical operation from patients who are unlikely to respond and should be scheduled for surgery sooner. This kind of early intervention has the potential to lead to improved patient outcomes and reduced side effects from unnecessary treatment administration. Citation Format: Aristeidis Sionakidis, Jonine D. Figueroa, Timothy I. Cannings. A novel 14-gene signature to predict response to neoadjuvant chemotherapy and endocrine treatment in breast cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1233.

Random projections offer an appealing and flexible approach to a wide range of large-scale statistical problems. They are particularly useful in high-dimensional settings, where we have many covariates recorded for each observation. In classification problems there are two general techniques using random projections. The first involves many projections in an ensemble -- the idea here is to aggregate the results after applying different random projections, with the aim of achieving superior statistical accuracy. The second class of methods include hashing and sketching techniques, which are straightforward ways to reduce the complexity of a problem, perhaps therefore with a huge computational saving, while approximately preserving the statistical efficiency., 24 pages, 4 figures

Taking a capability approach perspective, our paper aims at advancing our understanding of poverty in Afghanistan, and at identifying the most deprived, including persons with disabilities, in order to address the first Sustainable Development Goal to eradicate poverty in all its forms. We used data from a national survey carried out in Afghanistan in 2005. We calculated one index using two weights structures, the adjusted headcount ratio, part of the multidimensional poverty measures. Following a participatory process, we identified and validated 13 indicators clustered in seven dimensions of poverty, including three usually neglected dimensions. Findings suggest that exploring various domains of deprivation would better inform poverty eradication policies than an approach focused only on income. Our results also demonstrate that nearly all Afghan adults are deprived in at least one dimension and those residing in rural areas, from minority ethnic groups, women, elderly people and persons disable...

We study the effect of imperfect training data labels on the performance of classification methods. In a general setting, where the probability that an observation in the training dataset is mislabelled may depend on both the feature vector and the true label, we bound the excess risk of an arbitrary classifier trained with imperfect labels in terms of its excess risk for predicting a noisy label. This reveals conditions under which a classifier trained with imperfect labels remains consistent for classifying uncorrupted test data points. Furthermore, under stronger conditions, we derive detailed asymptotic properties for the popular $k$-nearest neighbour ($k$nn), support vector machine (SVM) and linear discriminant analysis (LDA) classifiers. One consequence of these results is that the knn and SVM classifiers are robust to imperfect training labels, in the sense that the rate of convergence of the excess risks of these classifiers remains unchanged; in fact, our theoretical and empirical results even show that in some cases, imperfect labels may improve the performance of these methods. On the other hand, the LDA classifier is shown to be typically inconsistent in the presence of label noise unless the prior probabilities of each class are equal. Our theoretical results are supported by a simulation study., Comment: 44 pages, 7 figures

We introduce a very general method for high-dimensional classification, based on careful combination of the results of applying an arbitrary base classifier to random projections of the feature vectors into a lower-dimensional space. In one special case that we study in detail, the random projections are divided into disjoint groups, and within each group we select the projection yielding the smallest estimate of the test error. Our random projection ensemble classifier then aggregates the results of applying the base classifier on the selected projections, with a data-driven voting threshold to determine the final assignment. Our theoretical results elucidate the effect on performance of increasing the number of projections. Moreover, under a boundary condition implied by the sufficient dimension reduction assumption, we show that the test excess risk of the random projection ensemble classifier can be controlled by terms that do not depend on the original data dimension and a term that becomes negligible as the number of projections increases. The classifier is also compared empirically with several other popular high-dimensional classifiers via an extensive simulation study, which reveals its excellent finite-sample performance., 49 pages, 8 figures

Summary The conflict in Darfur represents a complex emergency situation. Challenges to humanitarian intervention are multiplied by inadequate information on local living conditions. This study applies the Alkire and Foster multidimensional poverty index to a set of dimensions of deprivation identified through a household survey, including dimensions usually missing such as economic exploitation, physical safety, empowerment, and psychological wellbeing. Our findings show that children with disabilities, particularly girls, are more often multidimensionally poor. We argue that multiple domains of deprivation complicate efforts to tackle poverty among children in Western Darfur and require a more comprehensive and coordinated approach by relief agencies.