Traditional Methods with Modern Outcomes

Molecular biology is an exciting and fast moving area, which has provided pathologists with a wealth of new information about health and disease. Despite this, I don’t believe molecular analysis can tell the whole story. Conventional analytical techniques are at risk of being overlooked, and I feel this would be a serious mistake.

In my work in breast carcinoma (BC), histopathological methods to assess disease and provide prognosis are, and will continue to be, completely essential.

Traditional, not out of touch

While more conventional approaches to analyzing disease may not be seen as cutting-edge, they still hold a lot of value. They are proven and consistent – tried, tested and improved upon over many years of use. Classifying BC is not a simple process, and I am very disappointed when we focus only on microscopic tumor images. I am even more disappointed when we focus exclusively on biomarkers. Despite the emphasis now placed on molecular phenotyping, it cannot replace classic morphological examination, which provides answers molecular tests can’t – how large is the tumor? How extensive is the disease? Is it multifocal or diffuse?

Instead of focusing only on some sources of information, providing a prognosis must be a multimodal process, and in collaboration with our clinical and radiologist colleagues. With such a complex disease, the more information we can get, the better.

To highlight the importance of combining prognostic techniques, I have focused on BC as this is the area that I have studied most widely, but I believe these principles could be applied to a range of other cancer types and indeed the pathology of other conditions.

It goes without saying that BC has a complex pathology and outcomes can vary greatly. Despite this, there are many methods for assessing tumors, and these can be split into two main groups: subgross and molecular (Table 1). Both take in many different factors and are far from straightforward, but both have clear links to patient survival (1, 2, 3).

Morphology meets molecular

I believe there is currently a problem in the way in which we study subgross morphological parameters (SMPs) and molecular phenotypes – separately. We now have defined molecular phenotypes in BC, and we know the ways in which SMPs can aid in characterizing the disease; but we are not giving enough attention to the relationship between these two areas.

My colleagues and I at Falun County Hospital, Sweden, use both in our everyday practice. We use large-section histology to judge disease parameters, such as extent and focality, and we gain further insight from radiology results – we discuss every case at multidisciplinary meetings in which the radiologists demonstrate their findings, and the pathologists project their slides parallel with the radiological images, in order to correlate all our information. At the same time, we use a panel of biomarkers to define molecular phenotype (different classifications exist, but within our lab we use the St. Gallen 2013 system, as modified by the Swedish Breast Cancer Group).

So how exactly do these morphological and molecular parameters relate to each other? In a review of over 1,000 cases at my hospital I noted many examples (4):

Size

• The number of luminal A cases decreased as tumor size increased, and the opposite was true for triple negative cases (Table 2).
• The likelihood that an invasive BC less than 10 mm in size was HER2 positive, triple negative or ER negative was very low.
• HER2 expression appeared to begin during the early stages of some BC cases (though overall number of HER2 cases was low in this study; 3 percent of the total, or 32 cases).

Shape

• The radiological shape of the tumor (i.e. spherical or stellate) increased the correct prediction of molecular phenotype (Figure 1).
• The majority of HER2 and triple negative tumors appeared to be spherical (Figure 1).
• Both luminal A and B types were more likely to be stellate.
• 95% of stellate tumors expressed ER.

Focality

• In 60 percent of cases, the invasive component of the tumor was unifocal, and multifocal in roughly one-third.
• In HER2-type cancers, over half of the cases were multifocal.
• Diffuse cases were very rare, and in 92 percent of cases were ER positive. This was of particular note as diffuse, invasive cases have an unfavorable prognosis, in contrast with their luminal phenotype, which confers a more favorable prognosis.

Microcalcifications

• HER2 positive tumors were most likely to be calcified, and high HER2 expression is associated with high grade tumors.

Key recommendations

Based on my experience, when describing malignant breast lesions, regardless of the imaging method used, I would highly recommend that the following morphological parameters are always assessed as standard:

• The distribution of the lesions (unifocal, multifocal or diffuse) – separately for invasive and in situ lesions.
• The extent of the disease (including all invasive and in situ tumor structures).
• The size of the tumor – this corresponds to the largest dimension of the largest individual invasive tumor.
• Evidence of intratumoral or intertumoral heterogeneity.

I believe that our research has clearly shown correlations between molecular phenotype and subgross morphology. The most important take home messages, in my opinion, are: HER2 type tumors are more likely to be multifocal and have a high-grade in situ component; luminal A types are commonly small and stellate; and diffusive invasive BC, while rare, has an unfavorable prognosis, which contrasts with its luminal phenotype. I am sure that as this area becomes more widely studied, we will discover more such associations.

As well as providing us with a more comprehensive understanding of the carcinomas we are studying, this approach may also improve patient outcomes. For example, work done by my colleagues at the Carillion Center, USA, is now beginning to show that the use of MRI and large section histology may influence survival rates in multifocal cases – although unfortunately, their results are not yet published.

Histological pitfalls and missing guidelines

I believe that of all the methods we have available for assessing SMPs, histology is by far the most accurate, but as with all of the techniques I have described, there is room for error: inadequate sampling, for example, including only part of the tumor in your sections; inadequate interpretation, such as mistakenly including invasive extensions when sizing a tumor.

There is also the risk of failing to properly correlate radiological and pathological information available to you. For example, it is common for pathologists to have no information on radiological findings, or only a text report, meaning they may not know how many foci were identified within the breast and therefore how many should be found.

The guidelines and quality control measures available in this area are another issue; I believe they are simply not sufficient, and require revision. The WHO classification of tumors of the breast (5), a basic reference for pathology departments all over the world, has 240 pages but does not mention disease extent as a parameter. Multifocality is mentioned just once, in connection to rates of lymph node metastasis. Also, the Tumor Node Metastasis (TNM) classification of breast tumors was produced a long time ago, and although repeatedly revised, is still oriented towards naked-eye examination and specimen sampling, and does not appreciate the importance of detailed radiological-pathological correlation.

On a positive note, there is an International Agency for Research on Cancer (IARC)/ WHO initiative for producing an additional document, focusing on the above mentioned subgross parameters (the Eurocan Platform), but the work is progressing very slowly. The lack of international guidelines which properly interpret these parameters concerns me, as the pathology community is mostly guideline-governed, and currently this information is not fully standardized and available to everyone.

Better together

To summarize, it’s only if we assess all of the parameters discussed here, and we correlate all of the information we receive, that we can provide our colleagues and our patients with the best possible guidance. While both molecular and morphological analytical techniques undoubtedly have much to offer, we can improve them by combination. It is for this reason that I believe a new prognostic index is required. Molecular classification is, of course, a powerful tool – but it’s even more powerful when combined with conventional parameters. Surely this should be our goal.