“The journey is more important than the destination“; a phrase we all know well, but in cancer research the journey has been neglected in favour of studying the first part of a cancer cell’s journey and where it ends up in its journey to the metastatic site. Every pathologist knows that cancer cells look different inside vessels. That is part of the reason why jobbing histopathologists are able to recognise lymphovascular invasion (LVI) , sometimes even with a low power microscope objective.
There are lots of things pathologists see with our microscopes that we take for granted, and many reflect a snapshot of a fundamental biological process.
Lymphovascular invasion in carcinoma is one of those things. That is the reason why my (now retired) colleague Mark Smith and I investigated something we were very familiar with seeing on a regular basis. Despite the fundamental nature of metastasis in the clinical progression and subsequent death of patients with cancer, intravascular carcinoma has been rarely studied by histopathologists (those who see it most). A lot is known about the starting point and destination of cancer cells, but it is difficult to study cancer cells within blood vessels or lymphatics as they constitute such a small relative proportion of the total tumour bulk in a specimen.
Even in disseminated cancer, most carcinoma still resides outside of blood vessels, and not surprisingly this makes intravascular carcinoma difficult to study, despite being critical to carcinogenesis and metastasis. In University Hospital Plymouth, being a specialist upper GI cancer centre, we have access to almost 100 cancer resection specimens per year of oesophagus and stomach cancer surgical resections. Unfortunately (for the patient) the majority of these contain lymph node metastases, and of those, a high proportion also contain intravascular carcinoma, in and around the primary tumour and the lymph node deposits. Gastric carcinomas were an ideal group of tumours to study the intravascular changes in cohesion, differentiation and adhesion molecule expression we had seen in other cancers.
Most GI pathologists encounter colorectal adenocarcinoma with greatest frequency in their daily work and we have all seen cases with tumour cell budding at the invasive front.
This can be so prominent that the budding cells (for example Figure 2) at the advancing edge of the tumour can resemble spindle cells of a sarcoma or sarcomatoid carcinoma. This morphological change occurs in many types of cancer and is known as the epithelial mesenchymal transition (EMT). This is the process, hijacked by cancer cells from a developmental program allowing epithelial cells, which normally remain anchored to each other and basement membrane, to escape their neighbours and acquire more motile infiltrative, motile characteristics. One of these characteristics is undoubtedly the ability to infiltrate (intravasate) blood vessels and lymphatic spaces.
Gastric diffuse type adenocarcinoma is the archetypal cancer showing EMT in that its cells (often showing signet ring morphology) lose their connections early in their journey from the mucosal primary site. One of the most feared cancers by histopathology trainees and consultants due to their subtle infiltration through the tissues, they can be almost impossible to see in frozen sections, and surgeons can find their margins positive when they seemed widely clear at the time of resection. We noticed LVI in these tumours was particularly eye-catching as the intravascular tumour was remarkably more cohesive and better differentiated, compared with the stromal invasive cancer cells being so dyscohesive and infiltrative having undergone EMT.
We also showed the intravascular tumour to have higher expression of cell adhesion molecules and became better differentiated within the blood vessels compared with tumour outside. We hypothesise that the EMT process is reversed during LVI and therefore represents an intravascular “mesenchymal epithelial transition” (MET).
Given that many more tumour cells are in the intravascular compartment in lymph nodes and distant metastases compared with the primary site, this may (at least in part) explain the findings of other investigators that cell adhesion molecule is re-expressed at the metastatic site in cadherin defective signet ring carcinoma. The re-expression of cadherin at the metastatic site implies that the mechanism must be epigenetic rather than mutational. We think it is likely that the phenotypic plasticity resulting from epigenetic changes in these intravascular morphological changes we have observed are of fundamental importance, not just in this peculiar subtype of gastric cancer, but in almost every type of cancer showing a propensity for vascular invasion. We did not correlate these changes with prognosis but since we selected for cases with florid LVI and the phenomenon was seen most frequently in advanced cancers with lymph node metastasis it is likely to be a mechanism that potentiates metastatic dissemination.
We have noticed similar changes in breast and colorectal carcinoma, and also reversed intravascular glandular polarity, particularly in micropapillary carcinoma where the epithelial membrane conspicuously folds onto the outside of tumour clusters inside vessels. This is even occasionally seen in cancers where the predominant pattern outside of vessels is not papillary, and the epithelial membrane is enclosed within the stromal invasive tumour cell cluster as is usual for adenocarcinoma comprised of tubules.
Potential theories for these changes include potentiating attachment of antigens to endothelium and preventing programmed cell death caused by detachment of epithelial cells from local mesenchyme or removal or tumour cells from their normal nourishing local growth factors. Regardless of underlying biological cause of these changes they are so common and noticeable that we hypothesise that they must have profound biological importance in the process of local and distant metastatic dissemination. We would therefore like to bring other pathologists and researchers attention to these findings to encourage further research and correlate intravascular carcinoma phenotypic changes (the “Cowley-Smith phenomenon) with prognosis. In addition, given that lymphovascular invasion is such a well known prognostic factor for many cancer types, recognition and quantification of the intravascular Cowley-Smith both by pathologist and potentially by trained AI algorithms may increase the efficiency of LVI detection. Furthermore, a greater understanding of the phenotypic and epigenetic changes associated with this fundamental phenomenon may lead to targeted treatments to control and manage metastatic spread in different cancer types.
References with download links below
The original Cowley-Smith paper from 1995 is at this link