Risk Assessment of Non-Metastatic Breast Cancer in the Genomic Era

Underwriting breast cancer in 2017 will require an understanding of the current and growing role of genomics in the assessment of its mortality risk. This is not to lessen the importance of well-known prognosticators such as tumor/node/metastasis (TNM) staging, estrogen receptor/progesterone receptor (ER/PR) and human epidermal receptor growth factor 2 (HER2) receptor status, grade, or the mitotic activity index (MAI).

Rather, it is to keep abreast of additional genomics-based prognostic tools that can be used to further stratify breast cancer and its risks. 

The importance of genomics today is being seen in precision medicine: the increasing ability to predict the biological behavior of particular subclasses of diseases and then directly target treatments and hopefully influence favorable outcomes. Genomic-based treatments can also potentially reduce the use of ineffective therapies thereby avoiding the oft-used phrase “The treatment was a success, but the patient died.” 

This article will focus on current trends and advances in breast cancer genomics and prognostication and how they are enabling better assessments of individual mortality risk. Microarray studies, for example, that can distinguish breast cancer subtypes into luminal, basal and HER2-enriched, are enabling genetic profiling of breast cancers with validated recurrence scores using typing tests such as Oncotype DX and MammaPrint. The scores generated by these tests in combination with other prognosticators can be used to assess with greater precision the likelihood of long-term disease-free survival for these individuals. 

Introduction

Breast cancer is the most common cancer in women worldwide. Nearly 1.7 million new cases were diagnosed in 2016 according to World Cancer Research Fund International, representing 12% of all new cancer cases and 25% of all cancers in women¹. Breast cancer also ranks as the fifth most common cause of death from all cancers. While it is the most common cause of cancer death in women in less-developed regions, it is the second most common cause of cancer death in developed regions¹. 

According to Globoscan, a project of the International Agency for Research on Cancer (IARC) that provides estimates by cancer site and sex for 184 countries, the highest breast cancer rates are found in Belgium, Denmark, France and the Netherlands, and the lowest in Asia and Africa. 

With the emergence of genomics has come a surge of information that can help clinicians differentiate and predict the biological behavior of individual breast cancer types. Just as no two individuals have the exact combination of genes, no two breast cancer genomes are exactly alike. Making sense of these genetic variations to find usefulness in underwriting is the current goal.

Microarray studies and genome sequencing have yielded advances which are substantially altering how breast cancer subtypes are risk-stratified. Insurers, in turn, would be well advised to adjust underwriting risk guidelines as information evolves, providing additional biomarkers to assess mortality risk including response to treatment and assessment of recurrence. Essentially, given the widening variety of prognostic tools, breast cancer might eventually be underwritten on a case-by-case basis, for what might appear to be the same assessment on the basis of TNM and grade could, in actuality, be far different in terms of biological behavior when comparing the DNA of one cancer tissue to another.

Traditional Prognosticators

In order to appreciate the progress in research, it is prudent to review prognosticators used since the 1980s. Many are still valid today and remain the primary basis for the current guidelines used by insurers to assess mortality. Valid, in this usage, means having analytical and clinical validity as well as clinical utility.² 

These prognosticators fall into three distinct groups.

• Age. Older-age breast cancer patients have more estrogen/progesterone (ER/PR) positive hormone receptors than do younger patients. This correlates with better prognoses for postmenopausal women with early-stage breast cancer.³
• Pathologic factors. Valid factors include: tumor size, nodal involvement, metastasis, tumor morphology, histological grade, and degree of peritumoral lymphovascular invasion (PLVI). (Adverse pathologic factors associated with PLVI specifically for breast cancer include: large tumor size; high grade, positive nodal status; tubular, papillary, mucinous, medullary and adenoid cystic histology [as opposed to micropapillary and metaplastic carcinomas]; and estrogen-receptor negativity.)⁴
• Tissue markers. Receptor status, especially of hormone receptors such as estrogen and progesterone, as well as evidence of HER2 overexpression are important prognosticators.

Microarray studies of breast cancer subtypes show good correlation with these three groups, further supporting the validity of these prognosticators. These subtypes correlate most closely with hormone tissue markers and HER2 status. 

Genomic Profiles

Microarray studies were first introduced in 1983 as a means to assess antibodies. It was not, however, until 1995 that microarrays were first used to assess DNA. Since then, the utility of microarray studies has grown significantly, paving the way for genomic profiling. 

Microarray studies enabled the categorization of breast cancer tissue into four distinct subtypes.⁵

• Luminal A. This subtype comprises 40% of all breast cancers. It is associated with high expression of ER/PR positive receptors, low expression of HER2, and low proliferation clusters. It carries the best prognosis of all breast cancer subtypes. Luminal A expressed genes are associated with epithelial cells of normal breast tissue. They are characterized by the expression of luminal cytokeratins (8 and 18).
• Luminal B. This subtype is associated with low expression of ER/PR, variable expressions of HER2, and with high-proliferation gene clusters, comprising 25% to 35% of breast cancers. The prognosis for the B subtype is not as favorable as for the A subtype and those individuals will have higher Oncotype DX and MammaPrint recurrence scores.
• HER2-enriched. This subtype, 10% to 15% of breast cancers, is characterized by
  high expression of HER2, low expression of ER/PR, and low expression of luminal and basal clusters. Most are ER/PR negative and HER2 positive, but about 30% of HER2-enriched subtype are HER2 negative. This discordance likely represents HER2 mutations, producing a similar expression phenotype without the HER2 amplification or protein overexpression. 
• ER-negative. These subtypes include multiple basal-like carcinoma subtypes. Basal-like types have similar gene expression to that of basal epithelial cells and make up 15% to 20% of breast cancers. These tumors are characterized by low expression of ER/PR and HER2 and are known as “triple-negative breast cancers” (TNBC). TNBC is associated with the poorest prognosis and highest recurrence rates. Most of these tumors are infiltrating ductal tumors and are characterized by high nuclear grade, presence of central necrosis, and high mitotic activity indices. These tend to have aggressive clinical behavior and high rates of metastasis to brain and lung⁵.

Although an applicant may be ER/PR positive and HER2 negative, that does not confirm the best prognosis, nor does being TNBC automatically equate to a grim prognosis. Additional information on the subtype classifications can provide much more accurate prediction of mortality risk and if available should be used in the risk assessment. 

Gene Expression Profiles

The emergence of DNA microarray studies enabled the simultaneous measuring of the expression of thousands of genes in order to identify biologically based prognostic profiles. 

Breast cancer is a heterogeneous and phenotypically diverse disease with a range of genetic profiles and biological behaviors that respond uniquely to different forms of treatment. Since genetic profiles can predict breast cancer outcomes and responses to treatment, they can also be used to guide decision-making about adjuvant therapy.⁶ 

Morbidity and mortality caused by adverse effects of adjuvant treatments are the primary reason all treatment arsenals cannot be implemented on each cancer type. With genetic profiling, adjuvant treatment can be spared for favorable breast cancer gene profiles and encouraged for those associated with aggressive behaviors and high recurrence rates. Thus, underwriting must take into account all information regarding prognostics, not just on TNM and hormone receptors, but on genomic and gene expression profiles as well. 

The most commonly used commercial providers of gene expression profiles are Oncotype DX, MammaPrint and PAM50.

• Oncotype DX. Oncotype DX’s 21-gene recurrence score (RS) uses multianalyte reverse transcription-PCR genomic tests to predict the likelihood of breast cancer recurrence in early stage, node negative, ER-positive breast cancers. Measurement of gene expression from fixed formalin paraffin-embedded (FFPE) tissues yielded concordance in archival breast cancer specimens dating from 1985 to 2001 in all specimens when comparing ER/PR and HER 2 receptor status by Cronin et al.⁷
  
  Gene recurrence assay scores for breast cancer range from 1 to 100. The assay not only predicts the likelihood of tumor recurrence, but can predict the magnitude of chemotherapy benefit. High RS scores (≥31) predict benefit of chemotherapy (a decrease in 10-year distant recurrence risk) by 28%, whereas those with low RS scores (<31) derive minimal if any benefit from chemotherapy. In short, gene recurrence scores provide a benefit in adjuvant decision-making for node negative disease, but do not as yet have any role for hormone-negative cancer types and limited value for HER2 positive cancer types⁹. Underwriting may eventually incorporate the RS scores in these scenarios as a prognostic tool, especially given their incorporation in adjuvant decision guidelines per the American Society of Clinical Oncology (ASCO), the National Comprehensive Cancer Network (NCCN), National Institute for Health and Care Excellence (NICE)¹⁰, Arbeitsgemeinschaft Gynäkologische Onkologie (AGO)¹¹ and the St Gallen International Expert Consensus¹².

• MammaPrint. The Amsterdam 70-gene prognostic profile MammaPrint classifies tumors as low- and high-risk for recurrence using frozen tissue specimens obtained within an hour of surgery and sent for DNA microarray analysis. Reliable results can also be obtained with FFPE tissue. Unlike gene recurrence assays, gene prognostic profiles can prognosticate breast cancer patients regardless of hormone status and those whose cancers are HER2-positive as well.
  
  Results from the Microarray in Node-Negative Disease May Avoid Chemotherapy Trial (MINDACT), an international randomized trial, suggest that certain genetic profiles may identify those whose cancers are at low risk of metastasis despite high-risk clinical features^13,14,15. Observational studies are suggesting that the MammaPrint profiles may identify those with low chance of recurrence independent of nodal status, tumor grade, hormone receptor or HER2 status.¹⁶

• PAM50 (Prosigna). A third prognostic genetic test for breast cancer is the Predictor Analysis of Microarray 50 (PAM50), a 50-gene test that characterizes an individual tumor by intrinsic subtype. Results can, with high degree of analytical validity, stratify patients who are ER-positive into high-, medium- and low-risk subsets.
  
  In two separate trials – an Arimidex, Tamoxifen, Alone or in Combination (ATAC) trial and an Adjuvant Treatment in Patients with Hormone Receptor-Positive Breast Cancer with Good to Moderate Differentiation (ABCSG-8) trial – the risk of recurrence (ROR) score added prognostic information beyond what could be assessed by clinical factors in both node-negative and node-positive disease. The PAM50 ROR score also identified more patients with HER2 negative/node-negative tumors in the high-risk group and fewer in the intermediate group compared with the Oncotype DX RS gene recurrent test^17,18,19.
  
  Overall, the analysis of 1,478 postmenopausal patients who participated in the ABCSG-8 trial showed that the estimated 10-year distant relapse-free survival rates were 96.7%, 91.3% and 79.9% in the low-, intermediate- and high-risk groups respectively, based on the ROR score, regardless of whether pathological node involvement was present or not²⁰.

Other prognostic biomarkers

The protein Ki-67 has been shown in large meta-analyses to be an independent prognosticator associated with higher risk of relapse in node-positive and node-negative disease. There are, however, inconsistencies with retrospective studies and thus many medical societies such as ASCO and IMPAKT (Improving Care and Knowledge through Translational Research) do not recommend using proliferative markers such as Ki-67 for prognostic evaluation²¹.

Cancers grow and metastasize through angiogenesis. NOTCH1 is a human gene found to be implicated in metastasis and maintenance of cancer cells. Recent studies indicate that NOTCH1 is closely associated with TNBC and high recurrence rates and is an independent predictor of disease-free survival. 

Conclusion

The debilitating effects of chemotherapy, surgical complications, and/or radiation therapy are less likely to render morbidity and mortality concerns, given the added benefits genomics has provided in stratifying risk with a more direct, targeted approach.

Microarray studies that have molecularly divided breast cancers into subtypes such as luminal, basal, and HER2-enriched have advanced predictions of biological behavior of individual breast cancers that TNM staging, ER/PR, and HER2 status could not predict for long-term cancer-free survivals. 

Genomics has provided precision medicine with a tool to enable more effective predicting of the behavior of cancers to the extent that genomic assessment may supersede traditional staging, allowing improved risk assessment of individual breast cancer patients.

Moving forward, insurance medicine needs to keep pace with advances in breast cancer genomics, thus allowing for more appropriate and equitable risk assessment.