The analytical validation of the Oncotype DX Recurrence Score assay

Emerging Landscape of Targeted Therapy of Breast Cancers With Low Human Epidermal Growth Factor Receptor 2 Protein Expression

CONTEXT.—

Human epidermal growth factor receptor 2 (HER2) status in breast cancer is currently classified as negative or positive for selecting patients for anti-HER2 targeted therapy. The evolution of the HER2 status has included a new HER2-low category defined as an HER2 immunohistochemistry score of 1+ or 2+ without gene amplification. This new category opens the door to a targetable HER2-low breast cancer population for which new treatments may be effective.

OBJECTIVE.—

To review the current literature on the emerging category of breast cancers with low HER2 protein expression, including the clinical, histopathologic, and molecular features, and outline the clinical trials and best practice recommendations for identifying HER2-low-expressing breast cancers by immunohistochemistry.

DATA SOURCES.—

We conducted a literature review based on peer-reviewed original articles, review articles, regulatory communications, ongoing and past clinical trials identified through ClinicalTrials.gov, and the authors' practice experience.

CONCLUSIONS.—

The availability of new targeted therapy potentially effective for patients with breast cancers with low HER2 protein expression requires multidisciplinary recognition. In particular, pathologists need to recognize and identify this category to allow the optimal selection of patients for targeted therapy.

Impact of reactive changes on multigene testing: histopathologic analysis of low-grade breast cancers with high-risk 21-gene recurrence scores

PURPOSE

The 21-gene recurrence score (RS) assay predicts the recurrence risk and magnitude of chemotherapy benefit in patients with invasive breast cancer (BC). This study examined low-grade tumors yielding a high-risk RS and their outcomes.Kindly check the edit made in the article titleOk  METHODS: We compared patients with grade 1 BC and a high-risk RS to those with low-risk RS. Histologic sections were reviewed and features reported to elevate the RS were noted, mainly biopsy cavity and reactive stromal changes (BXC).

RESULTS

A total of 54 patients had high-risk RS (median RS of 28, range 26-36). On review, BXC were seen in all cases. Thirty BCs in this group also had low to negative PR. Treatment regimens included: chemoendocrine therapy (63%), endocrine therapy alone (31%) and no adjuvant therapy (6%). There were no additional breast cancer events over a median follow-up of 54.0 months (range 6.2 to 145.3). A total of 108 patients had low-risk RS (median RS of 7, range 0-9). BXC were seen in 47% of cases and none were PR negative. One patient had a recurrence at 64.8 months while the rest had no additional events over a median of 68.1 months (2.4 to 100).

CONCLUSION

We provide further evidence that reactive stromal changes and/or low-PR scores enhance the elevation of the RS. A high-RS result in low grade, PR-positive BC may not reflect actual risk and any suspected discrepancies should be discussed with the management teams. Multigene testing results should be interpreted after correlation with pathologic findings to optimize patient care.

Evolution of Breast Cancer Recurrence Risk Prediction: A Systematic Review of Statistical and Machine Learning-Based Models

PURPOSE

Selection of appropriate adjuvant therapy to ultimately reduce the risk of breast cancer (BC) recurrence is a challenge for medical oncologists. Several automated risk prediction models have been developed using retrospective clinical data and have evolved significantly over the years in terms of predictors of recurrence, data usage, and predictive techniques (statistical/machine learning [ML]).

METHODS

Following PRISMA guidelines, we performed a systematic literature review of the aforementioned statistical and ML models published between January 2008 and December 2022 through searching five digital databases-PubMed, ScienceDirect, Scopus, Cochrane, and Web of Science. The comprehensive search yielded a total of 163 papers and after a screening process focusing on papers that dealt exclusively with statistical/ML methods, only 23 papers were deemed appropriate for further analysis. We benchmarked the studies on the basis of development, evaluation metrics, and validation strategy with an added emphasis on racial diversity of patients included in the studies.

RESULTS

In total, 30.4% of the included studies use statistical techniques, while 69.6% are ML-based. Among these, traditional ML models (support vector machines, decision tree, logistic regression, and naïve Bayes) are the most frequently used (26.1%) along with deep learning (26.1%). Deep learning and ensemble learning provide the most accurate predictions (AUC = 0.94 each).

CONCLUSION

ML-based prediction models exhibit outstanding performance, yet their practical applicability might be hindered by limited interpretability and reduced generalization. Moreover, predictive models for BC recurrence often focus on limited variables related to tumor, treatment, molecular, and clinical features. Imbalanced classes and the lack of open-source data sets impede model development and validation. Furthermore, existing models predominantly overlook African and Middle Eastern populations, as they are trained and validated mainly on Caucasian and Asian patients.

Refining the definition of HER2-low class in invasive breast cancer

BACKGROUND

Emerging evidence indicates that breast cancer (BC) patients whose tumours express HER2 protein without HER2 gene amplification (HER2-low), can benefit from antibody-drug conjugates (ADC). However, the current definition of HER2-low BC remains incomplete with low rates of concordance. This study aims to refine HER2-low definition with emphasis on distinguishing HER2 score 0 from score 1+ to identify patients who are eligible for ADC.

METHODS

A BC cohort (n = 363) with HER2 IHC scores 0, 1+ and 2+ (without HER2 gene amplification) and available HER2 mRNA was included. HER2 staining intensity, pattern and subcellular localisation were reassessed. Artificial neural network analysis was applied to cluster the cohort and to distinguish HER2 score 0 from 1+. Reproducibility and reliability of the refined criteria were tested.

RESULTS

HER2 IHC score 1+ was refined as membranous staining in invasive cells as either: (1) faint intensity in ≥ 20% of cells regardless the circumferential completeness, (2) weak complete staining in ≤ 10%, (3) weak incomplete staining in > 10% and (4) moderate incomplete staining in ≤ 10%. Based on this, 63% of the HER2-negative cases were reclassified as positive (HER2-low). The refined score showed perfect observer agreement compared to the moderate agreement in the original clinical scores. Similar results were generated when the refined score was applied on the independent BC cohorts. A proposal to refine the definition of other HER2 classes is presented.

CONCLUSION

This study refined the definition of HER2-low BC based on correlation with HER2 mRNA and distinguished between HER2 IHC score 1+ and score 0 tumours.

MiRNA expression deregulation correlates with the Oncotype DX® DCIS score

Background

Current clinical criteria do not discriminate well between women who will or those who will not develop ipsilateral invasive breast cancer (IBC), or a DCIS recurrence after a ductal carcinoma in situ (DCIS) diagnosis. The 12-gene Oncotype DX® DCIS assay (RT qPCR gene-based scoring system) was established and shown to predict the risk of subsequent ipsilateral IBC or DCIS recurrence. Recent studies have shown that microRNA (miRNA) expression deregulation can contribute to the development of IBC, but very few have evaluated miRNA deregulation in DCIS lesions. In this study, we sought to determine whether specific miRNA expression changes may correlate with Oncotype DX® DCIS scores.

Methods

For this study, we used archived formalin-fixed, paraffin-embedded (FFPE) specimens from 41 women diagnosed with DCIS between 2012 and 2018. The DCIS lesions were stratified into low (n = 26), intermediate (n = 10), and high (n = 5) risk score groups using the Oncotype DX® DCIS assay. Total RNA was extracted from DCIS lesions by macro-dissection of unstained FFPE sections, and next-generation small-RNA sequencing was performed. We evaluated the correlation between miRNA expression data and Oncotype score, as well as patient age. RT-qPCR validations were performed to validate the topmost differentially expressed miRNAs identified between the different risk score groups.

Results

MiRNA sequencing of 32 FFPE DCIS specimens from the three different risk group scores identified a correlation between expression deregulation of 17 miRNAs and Oncotype scores. Our analyses also revealed a correlation between the expression deregulation of 9 miRNAs and the patient’s age. Based on these results, a total of 15 miRNAs were selected for RT-qPCR validation. Of these, miR-190b (p = 0.043), miR-135a (p = 0.05), miR-205 (p = 0.00056), miR-30c (p = 0.011), and miR-744 (p = 0.038) showed a decreased expression in the intermediate/high Oncotype group when compared to the low-risk score group. A composite risk score was established using these 5 miRNAs and indicated a significant association between miRNA expression deregulation and the Oncotype DX® DCIS Score (p < 0.0021), between high/intermediate and low risk groups.

Conclusions

Our analyses identified a subset of 5 miRNAs able to discriminate between Oncotype DX® DCIS score subgroups. Together, our data suggest that miRNA expression analysis may add value to the predictive and prognostic evaluation of DCIS lesions.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13058-022-01558-4.

DAISM-DNNXMBD: Highly accurate cell type proportion estimation with in silico data augmentation and deep neural networks

Understanding the immune cell abundance of cancer and other disease-related tissues has an important role in guiding disease treatments. Computational cell type proportion estimation methods have been previously developed to derive such information from bulk RNA sequencing data. Unfortunately, our results show that the performance of these methods can be seriously plagued by the mismatch between training data and real-world data. To tackle this issue, we propose the DAISM-DNN^XMBD (XMBD: Xiamen Big Data, a biomedical open software initiative in the National Institute for Data Science in Health and Medicine, Xiamen University, China.) (denoted as DAISM-DNN) pipeline that trains a deep neural network (DNN) with dataset-specific training data populated from a certain amount of calibrated samples using DAISM, a novel data augmentation method with an in silico mixing strategy. The evaluation results demonstrate that the DAISM-DNN pipeline outperforms other existing methods consistently and substantially for all the cell types under evaluation in real-world datasets.

•

We propose a data augmentation method (DAISM) for DNN-based cell type deconvolution

•

DAISM-DNN enables accurate cell type deconvolution with dataset-specific training data

•

DAISM-DNN is robust to random errors in calibration samples

•

Trained DAISM-DNN model is reusable across biomedical experiments following same SOP

Computational cell type deconvolution methods were developed to understand the cellular heterogeneity in disease-related tissues from bulk RNA-seq data. Due to the presence of strong batch effects, the performance of existing methods could fluctuate greatly when applied to different datasets even with the latest development in batch normalization or platform-agnostic signature designs. To tackle this issue, we proposed a DNN-based cell abundance estimation method with dataset-specific training data populated from a certain number of calibrated samples from a target dataset using DAISM, a data augmentation method using an in silico mixing strategy. DAISM-DNN enables accurate cell type proportions prediction and is robust to random errors in the ground truth cell type proportions of calibration samples. Importantly, we showed that with strict SOPs, it is possible to create a “train once, reuse many times” DAISM-DNN model for multiple biomedical experiments without the need for retraining.

The clinician's perspective on the 21-gene assay in early breast cancer

Most patients with early HR+ and HER2- breast cancer receive a hormone therapy; the clinical question still open is how to identify patients who can really benefit from adjuvant chemotherapy. The accurate identification of these patients is essential to avoid an over-treatment, increasing the risk of an unnecessary toxicity; on the contrary, the omission of chemotherapy can deprive high risk patients of a potential life-saving treatment (under-treatment). Several multigene assays (MGAs), assessing the risk of relapse according to the biological characteristics of the tumor, have been developed. To date, the 21-gene assay (Oncotype DX Breast Recurrence Score^®) is the only test developed and validated to be actionable, i.e., able to predict the benefit of adjuvant chemotherapy. The different available tests can be classified according to their clinical utility based on their prognostic and predictive value. A prognostic test gives information about the outcome of the disease, regardless of the administered therapy. When the aim of the test is to drive the treatment decisions, the predictive component, and therefore the ability to accurately identify which patients could benefit from chemotherapy, is essential. This review summarizes the clinical evidences of the Oncotype DX^® test supporting its clinical utility.

Plasma Proteome Signature to Predict the Outcome of Breast Cancer Patients Receiving Neoadjuvant Chemotherapy

The plasma proteome of 51 non-metastatic breast cancer patients receiving neoadjuvant chemotherapy (NCT) was prospectively analyzed by high-resolution mass spectrometry coupled with nano-flow liquid chromatography using blood drawn at the time of diagnosis. Plasma proteins were identified as potential biomarkers, and their correlation with clinicopathological variables and survival outcomes was analyzed. Of 51 patients, 20 (39.2%) were HR+/HER2-, five (9.8%) were HR+/HER2+, five (9.8%) were HER2+, and 21 (41.2%) were triple-negative subtype. During a median follow-up of 52.0 months, there were 15 relapses (29.4%) and eight deaths (15.7%). Four potential biomarkers were identified among differentially expressed proteins: APOC3 had higher plasma concentrations in the pathological complete response (pCR) group, whereas MBL2, ENG, and P4HB were higher in the non-pCR group. Proteins statistically significantly associated with survival and capable of differentiating low- and high-risk groups were MBL2 and P4HB for disease-free survival, P4HB for overall survival, and MBL2 for distant metastasis-free survival (DMFS). In the multivariate analysis, only MBL2 was a consistent risk factor for DMFS (HR: 9.65, 95% CI 2.10-44.31). The results demonstrate that the proteomes from non-invasive sampling correlate with pCR and survival in breast cancer patients receiving NCT. Further investigation may clarify the role of these proteins in predicting prognosis and thus their therapeutic potential for the prevention of recurrence.

Validation of an Acoustic-Based Framework of Speech Motor Control: Assessing Criterion and Construct Validity Using Kinematic and Perceptual Measures

PURPOSE

This study investigated the criterion (analytical and clinical) and construct (divergent) validity of a novel, acoustic-based framework composed of five key components of motor control: Coordination, Consistency, Speed, Precision, and Rate.

METHOD

Acoustic and kinematic analyses were performed on audio recordings from 22 subjects with amyotrophic lateral sclerosis during a sequential motion rate task. Perceptual analyses were completed by two licensed speech-language pathologists, who rated each subject's speech on the five framework components and their overall severity. Analytical and clinical validity were assessed by comparing performance on the acoustic features to their kinematic correlates and to clinician ratings of the five components, respectively. Divergent validity of the acoustic-based framework was then assessed by comparing performance on each pair of acoustic features to determine whether the features represent distinct articulatory constructs. Bivariate correlations and partial correlations with severity as a covariate were conducted for each comparison.

RESULTS

Results revealed moderate-to-strong analytical validity for every acoustic feature, both with and without controlling for severity, and moderate-to-strong clinical validity for all acoustic features except Coordination, without controlling for severity. When severity was included as a covariate, the strong associations for Speed and Precision became weak. Divergent validity was supported by weak-to-moderate pairwise associations between all acoustic features except Speed (second-formant [F2] slope of consonant transition) and Precision (between-consonant variability in F2 slope).

CONCLUSIONS

This study demonstrated that the acoustic-based framework has potential as an objective, valid, and clinically useful tool for profiling articulatory deficits in individuals with speech motor disorders. The findings also suggest that compared to clinician ratings, instrumental measures are more sensitive to subtle differences in articulatory function. With further research, this framework could provide more accurate and reliable characterizations of articulatory impairment, which may eventually increase clinical confidence in the diagnosis and treatment of patients with different articulatory phenotypes.

Development and Performance of a CD8 Gene Signature for Characterizing Inflammation in the Tumor Microenvironment across Multiple Tumor Types

Across multiple tumor types, immune checkpoint inhibitors (ICIs) have demonstrated clinical benefit to patients with cancer, yet there is a need to identify predictive biomarkers of response to these therapies. A multiparameter gene expression profiling-based tumor inflammation assay may offer robust characterization of the tumor microenvironment, thereby extending the utility of single-gene analysis or immunohistochemistry (IHC) in predicting response to ICIs. The authors interrogated 1778 commercially procured, formalin-fixed, paraffin-embedded samples using gene expression profiling and pathology-assisted digital CD8 IHC. A machine-learning approach was used to develop gene expression signatures that predicted CD8⁺ immune cell abundance as surrogates for tumor inflammation in melanoma and squamous cell carcinoma of the head and neck samples. An assay for a 16-gene CD8 signature was developed and analytically validated across 12 tumor types. CD8 signature scores correlated with CD8 IHC in a platform-independent manner, and inflammation prevalence was similar between assay methods for all tumor types except prostate cancer and small cell lung cancer. In retrospective analyses, CD8 signature scores were associated with progression-free survival and overall survival with nivolumab in patients with urothelial carcinoma from CheckMate 275. This study demonstrated that the CD8 signature assay can be used to accurately quantify CD8⁺ immune cell abundance in the tumor microenvironment and has potential clinical utility for determining patients with cancer likely to respond to ICIs.

Molecular characterization of breast cancer needle core biopsy specimens by the 21-gene Breast Recurrence Score test

BACKGROUND AND OBJECTIVE

Recent COVID-19 pandemic guidelines recommend genomic assessment of core biopsies to help guide treatment decisions in estrogen receptor (ER)-positive early-stage breast cancer. Herein we characterize biopsy and excisional breast cancer specimens submitted for 21-gene testing.

METHODS

US samples submitted to Genomic Health for 21-gene testing (01/2004-04/2020) were assessed by pathologists and analyzed by a standardized quantitative reverse transcription-polymerase chain reaction. Predefined cutoffs were: ESR1 (positive ≥6.5), PGR (positive ≥5.5), and ERBB2 (negative <10.7). ER status by immunohistochemistry (IHC) and lymph node status were determined locally. Median and interquartile range were reported for continuous variables, and total and percent for categorical variables. Distributions were assessed overall, by age, and by nodal involvement.

RESULTS

Of 919 701 samples analyzed, 13% were biopsies and 87% were excisions. Initial assay success rates were 94.5% (biopsies) and 97.3% (excisions). ER IHC concordance with central ESR1 was 96.8% (biopsies) and 97.6% (excisions). Biopsy and excisional medians were: Recurrence Score results 16 (each); ESR1 10.2 (each); PGR 7.7 and 7.6; ERBB2 9.4 and 9.2, respectively.

CONCLUSIONS

Biopsy submissions for 21-gene testing are common and consistently generate results that are very similar to the experience with excisions. The 21-gene test can be performed reliably on core biopsies.

Advances in Precision Health and Emerging Diagnostics for Women

During the Dutch winter famine of 1944–1945, an interesting observation was made about the offspring born during this time—They had an increased risk of developing metabolic syndrome and other chronic diseases. Subsequent research has confirmed this finding as well as noting that health outcomes for many diseases are different, and often worse, for women. These findings, combined with the lack of enrollment of women in clinical trials and/or analysis of sex-specific differences are important factors which need to be addressed. In fact, Women’s health research and sex differences have historically been overlooked or lumped together and assumed equivalent to those of men. Hence, a focus on women’s health and disease prevention is critical to improve the lives of women in the 21st Century. In this review, we point out the critical differences biologically and socially that present both challenges and opportunities for development of novel platforms for precision health. The technologic and scientific advances specific to women’s precision health have the potential to improve the health and wellbeing for all females across the world.

A plasma microRNA biomarker of melanoma as a personalised assessment of treatment response

New tools for monitoring response to primary melanoma treatment are needed to reduce recurrence rates and patient anxiety. A previously developed plasma-based microRNA signature (MEL38) was measured in four melanoma patient samples obtained before and 12-14 days after treatment (i.e. surgical excision), as well as in two nonmelanoma controls. The value of the MEL38 score and selected individual genes were compared between the time points. The MEL38 scores of the four patients with melanoma became more 'normal like' after tumour excision, with a statistically significant 15% mean reduction. MicroRNAs involved in tumour suppression were upregulated in the postexcision samples and those involved in facilitating treatment resistance and tumour invasion were downregulated. Based on these limited preliminary data, the MEL38 signature may have clinical utility in assessing an individual patient's response to the most common form of melanoma treatment. Additional studies are needed on larger, clinically diverse patient cohorts, sampled over longer periods of time.

Next generation sequencing of PD-L1 for predicting response to immune checkpoint inhibitors

BACKGROUND

PD-L1 immunohistochemistry (IHC) has been traditionally used for predicting clinical responses to immune checkpoint inhibitors (ICIs). However, there are at least 4 different assays and antibodies used for PD-L1 IHC, each developed with a different ICI. We set to test if next generation RNA sequencing (RNA-seq) is a robust method to determine PD-L1 mRNA expression levels and furthermore, efficacy of predicting response to ICIs as compared to routinely used, standardized IHC procedures.

METHODS

A total of 209 cancer patients treated on-label by FDA-approved ICIs, with evaluable responses were assessed for PD-L1 expression by RNA-seq and IHC, based on tumor proportion score (TPS) and immune cell staining (ICS). A subset of serially diluted cases was evaluated for RNA-seq assay performance across a broad range of PD-L1 expression levels.

RESULTS

Assessment of PD-L1 mRNA levels by RNA-seq demonstrated robust linearity across high and low expression ranges. PD-L1 mRNA levels assessed by RNA-seq and IHC (TPS and ICS) were highly correlated (p < 2e-16). Sub-analyses showed sustained correlation when IHC results were classified as high or low by clinically accepted cut-offs (p < 0.01), and results did not differ by tumor type or anti-PD-L1 antibody used. Overall, a combined positive PD-L1 result (≥1% IHC TPS and high PD-L1 expression by RNA-Seq) was associated with a 2-to-5-fold higher overall response rate (ORR) compared to a double negative result. Standard assessments of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) showed that a PD-L1 positive assessment for melanoma samples by RNA-seq had the lowest sensitivity (25%) but the highest PPV (72.7%). Among the three tumor types analyzed in this study, the only non-overlapping confidence interval for predicting response was for "RNA-seq low vs high" in melanoma.

CONCLUSIONS

Measurement of PD-L1 mRNA expression by RNA-seq is comparable to PD-L1 expression by IHC both analytically and clinically in predicting ICI response. RNA-seq has the added advantages of being amenable to standardization and avoidance of interpretation bias. PD-L1 by RNA-seq needs to be validated in future prospective ICI clinical studies across multiple histologies.

Multiple-Swarm Ensembles: Improving the Predictive Power and Robustness of Predictive Models and Its Use in Computational Biology

Machine learning is an integral part of computational biology, and has already shown its use in various applications, such as prognostic tests. In the last few years in the non-biological machine learning community, ensembling techniques have shown their power in data mining competitions such as the Netflix challenge; however, such methods have not found wide use in computational biology. In this work, we endeavor to show how ensembling techniques can be applied to practical problems, including problems in the field of bioinformatics, and how they often outperform other machine learning techniques in both predictive power and robustness. Furthermore, we develop a methodology of ensembling, Multi-Swarm Ensemble (MSWE) by using multiple particle swarm optimizations and demonstrate its ability to further enhance the performance of ensembles.

An international reproducibility study validating quantitative determination of ERBB2, ESR1, PGR, and MKI67 mRNA in breast cancer using MammaTyper®

Background

Accurate determination of the predictive markers human epidermal growth factor receptor 2 (HER2/ERBB2), estrogen receptor (ER/ESR1), progesterone receptor (PgR/PGR), and marker of proliferation Ki67 (MKI67) is indispensable for therapeutic decision making in early breast cancer. In this multicenter prospective study, we addressed the issue of inter- and intrasite reproducibility using the recently developed reverse transcription-quantitative real-time polymerase chain reaction-based MammaTyper® test.

Methods

Ten international pathology institutions participated in this study and determined messenger RNA expression levels of ERBB2, ESR1, PGR, and MKI67 in both centrally and locally extracted RNA from formalin-fixed, paraffin-embedded breast cancer specimens with the MammaTyper® test. Samples were measured repeatedly on different days within the local laboratories, and reproducibility was assessed by means of variance component analysis, Fleiss’ kappa statistics, and interclass correlation coefficients (ICCs).

Results

Total variations in measurements of centrally and locally prepared RNA extracts were comparable; therefore, statistical analyses were performed on the complete dataset. Intersite reproducibility showed total SDs between 0.21 and 0.44 for the quantitative single-marker assessments, resulting in ICC values of 0.980–0.998, demonstrating excellent agreement of quantitative measurements. Also, the reproducibility of binary single-marker results (positive/negative), as well as the molecular subtype agreement, was almost perfect with kappa values ranging from 0.90 to 1.00.

Conclusions

On the basis of these data, the MammaTyper® has the potential to substantially improve the current standards of breast cancer diagnostics by providing a highly precise and reproducible quantitative assessment of the established breast cancer biomarkers and molecular subtypes in a decentralized workup.

Electronic supplementary material

The online version of this article (doi:10.1186/s13058-017-0848-z) contains supplementary material, which is available to authorized users.