Low Concordance between Gene Expression Signatures in ER Positive HER2 Negative Breast Carcinoma Could Impair Their Clinical Application

Decision Theory versus Conventional Statistics for Personalized Therapy of Breast Cancer

Estrogen and progesterone receptors being present or not represents one of the most important biomarkers for therapy selection in breast cancer patients. Conventional measurement by immunohistochemistry (IHC) involves errors, and numerous attempts have been made to increase precision by additional information from gene expression. This raises the question of how to fuse information, in particular, if there is disagreement. It is the primary domain of Dempster–Shafer decision theory (DST) to deal with contradicting evidence on the same item (here: receptor status), obtained through different techniques. DST is widely used in technical settings, such as self-driving cars and aviation, and is also promising to deliver significant advantages in medicine. Using data from breast cancer patients already presented in previous work, we focus on comparing DST with classical statistics in this work, to pave the way for its application in medicine. First, we explain how DST not only considers probabilities (a single number per sample), but also incorporates uncertainty in a concept of ‘evidence’ (two numbers per sample). This allows for very powerful displays of patient data in so-called ternary plots, a novel and crucial advantage for medical interpretation. Results are obtained according to conventional statistics (ODDS) and, in parallel, according to DST. Agreement and differences are evaluated, and the particular merits of DST discussed. The presented application demonstrates how decision theory introduces new levels of confidence in diagnoses derived from medical data.

Decision theory for precision therapy of breast cancer

Correctly estimating the hormone receptor status for estrogen (ER) and progesterone (PGR) is crucial for precision therapy of breast cancer. It is known that conventional diagnostics (immunohistochemistry, IHC) yields a significant rate of wrongly diagnosed receptor status. Here we demonstrate how Dempster Shafer decision Theory (DST) enhances diagnostic precision by adding information from gene expression. We downloaded data of 3753 breast cancer patients from Gene Expression Omnibus. Information from IHC and gene expression was fused according to DST, and the clinical criterion for receptor positivity was re-modelled along DST. Receptor status predicted according to DST was compared with conventional assessment via IHC and gene-expression, and deviations were flagged as questionable. The survival of questionable cases turned out significantly worse (Kaplan Meier p < 1%) than for patients with receptor status confirmed by DST, indicating a substantial enhancement of diagnostic precision via DST. This study is not only relevant for precision medicine but also paves the way for introducing decision theory into OMICS data science.

Quadruple Negative Breast Cancers (QNBC) Demonstrate Subtype Consistency among Primary and Recurrent or Metastatic Breast Cancer

PURPOSE

Despite the availability of current standards of care treatments for triple negative breast cancer (TNBC), many patients still die from this disease. Quadruple negative tumors, which are TNBC tumors that lack androgen receptor (AR), represent a more aggressive subtype of TNBC; however, the molecular features are not well understood.

METHODS

Immunohistochemistry of estrogen receptor (ER), progesterone receptor (PR), HER2, and AR was determined in 244 primary and 630 recurrent/metastatic site biopsies. Expression was correlated with a panel of 25 cancer-related genes and proteins by IHC and in situ hybridization (ISH).

RESULTS

We observed that 80.2% (65 of 81) of primary TNBC tumors and 75.7% (159 of 210) of recurrent/metastatic TNBC tumors are QNBC. Bivariate fit analysis demonstrated that QNBC (n = 224) significantly (P < .03) correlated with younger aged patients at initial biopsy compared to AR positive TNBC patients (n = 51). In paired primary tissue samples and primary to recurrent/metastatic samples, at least 70% Luminal, HER2 enriched, and QNBC subtype did not change molecular profile. But, TNBC seems to be the "unstable" subtype. Within the total cohort, discordance in molecular profiles was identified in both synchronous (20%) and asynchronous (21%) intra-individual analyses. Irrespective of sample type, (Synchronous or Asynchronous), QNBC demonstrated higher concordant than TNBC. IHC and ISH results of the cancer related genes, demonstrated that gene/protein expression differ by molecular profile: TNBC (HR-/HER2-, AR+) and QNBC (HR-/HER2-, AR-). IHC in metastatic tumors, showed that the percentage of tumors positive of EGFR were higher, while PTEN and TLE3 were lower in QNBC compared to TNBC.

CONCLUSION

Standard treatment of Breast Cancer (BC) relies on reliable assessment by IHC analysis of ER, PR, and HER2. Our analyses suggest that the heterogeneity of TNBC is at least partially associated with the presence or absence of AR expression, suggesting that QNBC should be considered as a clinically relevant BC subtype. IHC analysis of AR appears to be a practical assay to determine the most aggressive TNBC subtypes and identifies tumors that could benefit from available targeted therapies.

Co-expressed genes enhance precision of receptor status identification in breast cancer patients

PURPOSE

Therapeutic decisions in breast cancer patients crucially depend on the status of estrogen receptor, progesterone receptor and HER2, obtained by immunohistochemistry (IHC). These are known to be inaccurate sometimes, and we demonstrate how to use gene-expression to increase precision of receptor status.

METHODS

We downloaded data from 3241 breast cancer patients out of 36 clinical studies. For each receptor, we modelled the mRNA expression of the receptor gene and a co-gene by logistic regression. For each patient, predictions from logistic regression were merged with information from IHC on a probabilistic basis to arrive at a fused prediction result.

RESULTS

We introduce Sankey diagrams to visualize the step by step increase of precision as information is added from gene expression: IHC-estimates are qualified as 'confirmed', 'rejected' or 'corrected'. Additionally, we introduce the category 'inconclusive' to spot those patients in need for additional assessments so as to increase diagnostic precision and safety.

CONCLUSIONS

We demonstrate a sound mathematical basis for the fusion of information, even if partly contradictive. The concept is extendable to more than three sources of information, as particularly important for OMICS data. The overall number of undecidable cases is reduced as well as those assessed falsely. We outline how decision rules may be extended to also weigh consequences, being different in severity for false-positive and false-negative assessments, respectively. The possible benefit is demonstrated by comparing the disease free survival between patients whose IHC could be confirmed versus those for which it was corrected.

Gene expression information improves reliability of receptor status in breast cancer patients

Immunohistochemical (IHC) determination of receptor status in breast cancer patients is frequently inaccurate. Since it directs the choice of systemic therapy, it is essential to increase its reliability.

We increase the validity of IHC receptor expression by additionally considering gene expression (GE) measurements. Crisp therapeutic decisions are based on IHC estimates, even if they are borderline reliable. We further improve decision quality by a responsibility function, defining a critical domain for gene expression. Refined normalization is devised to file any newly diagnosed patient into existing data bases. Our approach renders receptor estimates more reliable by identifying patients with questionable receptor status. The approach is also more efficient since the rate of conclusive samples is increased. We have curated and evaluated gene expression data, together with clinical information, from 2880 breast cancer patients. Combining IHC with gene expression information yields a method more reliable and also more efficient as compared to common practice up to now.

Several types of possibly suboptimal treatment allocations, based on IHC receptor status alone, are enumerated. A ‘therapy allocation check’ identifies patients possibly miss-classified. Estrogen: false negative 8%, false positive 6%. Progesterone: false negative 14%, false positive 11%. HER2: false negative 2%, false positive 50%. Possible implications are discussed.

We propose an ‘expression look-up-plot’, allowing for a significant potential to improve the quality of precision medicine.

Methods are developed and exemplified here for breast cancer patients, but they may readily be transferred to diagnostic data relevant for therapeutic decisions in other fields of oncology.

Extracting replicable associations across multiple studies: Empirical Bayes algorithms for controlling the false discovery rate

In almost every field in genomics, large-scale biomedical datasets are used to report associations. Extracting associations that recur across multiple studies while controlling the false discovery rate is a fundamental challenge. Here, we propose a new method to allow joint analysis of multiple studies. Given a set of p-values obtained from each study, the goal is to identify associations that recur in at least k > 1 studies while controlling the false discovery rate. We propose several new algorithms that differ in how the study dependencies are modeled, and compare them and extant methods under various simulated scenarios. The top algorithm, SCREEN (Scalable Cluster-based REplicability ENhancement), is our new algorithm that works in three stages: (1) clustering an estimated correlation network of the studies, (2) learning replicability (e.g., of genes) within clusters, and (3) merging the results across the clusters. When we applied SCREEN to two real datasets it greatly outperformed the results obtained via standard meta-analysis. First, on a collection of 29 case-control gene expression cancer studies, we detected a large set of consistently up-regulated genes related to proliferation and cell cycle regulation. These genes are both consistently up-regulated across many cancer studies, and are well connected in known gene networks. Second, on a recent pan-cancer study that examined the expression profiles of patients with and without mutations in the HLA complex, we detected a large active module of up-regulated genes that are both related to immune responses and are well connected in known gene networks. This module covers thrice more genes as compared to the original study at a similar false discovery rate, demonstrating the high power of SCREEN. An implementation of SCREEN is available in the supplement.

When analyzing results from multiple studies, extracting replicated associations is the first step towards making new discoveries. The standard approach for this task is to use meta-analysis methods, which usually make an underlying null hypothesis that a gene has no effect in all studies. On the other hand, in replicability analysis we explicitly require that the gene will manifest a recurring pattern of effects. In this study we develop new algorithms for replicability analysis that are both scalable (i.e., can handle many studies) and allow controlling the false discovery rate. We show that our main algorithm called SCREEN (Scalable Cluster-based REplicability ENhancement) outperforms the other methods in simulated scenarios. Moreover, when applied to real datasets, SCREEN greatly extended the results of the meta-analysis, and can even facilitate detection of new biological results.