Diagnosing vascular invasion in colorectal carcinomas: improving reproducibility and potential pitfalls
- 1Department of Pathology, Bács-Kiskun County Teaching Hospital, Kecskemét, Hungary
- 2Department of Pathology, University of Szeged, Szeged, Hungary
- Correspondence to Dr Gábor Cserni, Department of Pathology, Bács-Kiskun County Teaching Hospital, Nyiri ut 38, Kecskemét H-6000, Hungary;
- Received 27 February 2013
- Accepted 1 March 2013
- Published Online First 16 April 2013
Blood and lymphatic vessel invasion (BLVI) has been long recognised as a category I prognostic factor in colorectal carcinomas (CRC) along with the pT and pN categories of the Tumour-Node-Metastasis system, preoperative carcinoembryonal antigen levels and the residual tumour (R) classification.1
Small vessel involvement includes the invasion of capillary-type vessels (figure 1A), where a reliable distinction between lymphatic and blood capillaries cannot be made on H&E-stained or general endothelial marker (eg, CD31, CD34) immunostained sections. Theoretically, the two types of capillary invasion lead to different consequences: lymphatic vessel invasion (LI) precedes and may be predictive of lymph node metastasis, whereas blood capillaries may be the source of systemic dissemination.2–4 The development of lymphatic endothelial markers like podoplanin (D2-40) or lymphatic endothelial hyaluronan receptor (LYVE-1) enables a distinction between lymphatic and blood capillaries: lymphatics can be defined as D2-40, LYVE-1, CD31 and CD34 positive capillaries, whereas blood capillaries are D2-40 and LYVE-1 negative but can be highlighted by CD31 or CD34 antibodies.4–7 Without such a distinction, small vessel invasion is best recorded as (lympho-)vascular invasion (LVI) or angiolymphatic invasion.1 The impact of LI or LVI is mainly seen in patients with conservative removal (eg, polypectomy) of invasive carcinomas without a resection allowing the histopathologic assessment of the lymph nodes.
By contrast, large vessel invasion includes venous invasion (VI) (figure 1B,C) which is associated with synchronous or metachronous distant, especially liver metastases,8–11 and rarely arterial invasion (figure 1D) which can be associated with anaemic tumour necrosis. It is recommended to report small and large vessel invasion separately.1
Inconsistencies in reporting vascular invasion in general, or distinguishing between its subtypes, are also well known.1 In their recent paper, Kojima et al12 report on a possible way of improving reproducibility of diagnosing vascular invasion in CRC. After documenting a substantial difference in the proportion of CRC cases with BLVI, LI and VI between different centres, they establish a framework allowing a better diagnosis of this phenomenon. Importantly, the improvement is not only due to technical issues like the general use of D2-40 immunohistochemistry or special elastica stains. Although a recent evaluation of the introduction of an elastica stain on the reproducibility of recognising extramural VI has shown an improvement in interobserver agreement (κ 0.23 shifting to 0.41),13 the pure introduction of more sensitive methods was insufficient to increase the consistency of the diagnoses in the study by Kojima et al.12 Common interpretation issues needed to be agreed upon to allow better reproducibility. As a result, the diagnosis of both LI and VI became more consistent with κ values above 0.6. The authors have to be complimented for their efforts in improving these diagnoses. However, better reproducibility does not necessarily lead to better diagnosing of a phenomenon, and care should be taken before making a uniform application of the criteria for VI set by Kojima et al.12
As an obvious difference between cohort 1 (H&E-stained slides without agreed diagnostic criteria) and cohort 6 (elastica stained slides with agreed diagnostic criteria) in the cited study by members of the Pathology Working Group of the Japanese Society for Cancer of the Colon and Rectum,12 there is a substantial rise in the proportion of CRC cases with VI from 28% to 76%. Such an increase has been often documented with the use of elastica stains.11 ,13–19 However, owing to the rise in VI positive cases, proper attention should be used to avoid overdiagnosis of VI and recognise possible pseudo-VI.
Although the presence of an accompanying artery helps, and is also acknowledged by the diagnostic criteria agreed for VI, the final statement on diagnosing VI includes the following: ‘When the presence of elastica-stained internal elastic membrane covering more than half of circumference surrounding tumour cluster is identified, the lesion should be diagnosed as blood vessel invasion, even in the absence of accompanying artery, vascular structure or space.’12 There is no specification of the elastic stain used, but these stains are rather equivalent in their ability to identify elastic fibres. Orcein is the monochromatic stain,20 which is routinely used in CRC specimens in our setting.11 The following illustrations and descriptions highlight a few situations when an elastica-stained layer reminiscent of an internal elastic lamina forms a circular or semicircular pattern, which may mimic a vein whenever tumour cells are present inside, and can result in interpreting ‘pseudo-VI’ as VI. Many of these situations could fit within the criterion of tumour cell clusters (completely or with more than half their perimeter) surrounded by an elastic lamina, but the phenomena are inconsistent with VI.
As highlighted in figure 1B,C, veins can have a substantial thickening of their internal elastic lamina, and therefore, distinction from arterial invasion must be borne in mind. Arteries have a more regular, sharp, wavy elastic lamina of rather even thickness localised at the luminal surface of the muscular media layer (figure 1D), whereas veins may have a usually thinner internal elastic lamina-like layer in the intima, which is irregular, wavy, but of uneven thickness, and often multiplied rather than single. Of note is the fact that the muscular media layer of arteries has sparse elastic fibres, by contrast with veins which have abundant elastic fibres at this location.21 Although these basic principles may be helpful, they may be altered in diseased arteries, and it is not exceptional to see multiplication of the internal elastic lamina and/or the accumulation of elastic fibres in the muscular layer.
The presence of a vascular structure and tumour cells or clusters within a circular structure may sometimes represent vascular wall invasion rather than real large vessel invasion with tumour cells in the lumen (figure 2). These features can be mistaken for VI on H&E-stained slides. The identification of the internal elastic lamina surrounded by tumour cell clusters rather than the internal elastic lamina surrounding tumour cells may be a clue to identify these lesions as vascular wall invasion. The presence of a single endothelium-lined lumen in the central unaffected part of the lesion may further substantiate this.
Some layers of the bowel wall may be bordered by elastic fibres. Although these are not consistently present everywhere, and may range from virtually lacking to prominent, their prominence being possibly related to previous injury (eg, neoadjuvant treatment). Their presence may sometimes take a circular form and therefore mimic the internal elastic lamina of a vein. Of these elastic fibre networks, the peritoneal, or subserosal elastic lamina, is probably the most important. It can be very prominent at some segments of the colon,22 and is particularly prone to form circular shapes in some planes of section (figure 3A,B). Elastic fibres may also occur beneath the muscularis mucosae, where diverticular protrusions of the mucosa into the submucosa may also give ground to pseudo-VI if such structures are involved by neoplastic glands (figure 3C,D). The presence of these elastic fibre networks at specific anatomic/histologic layers should always be considered when dealing with elastica stain-detected vascular invasion, although they represent differential diagnostic problems only when the tumour involves the tissues around them, and there are misleading planes of section. To avoid misdiagnosis, it must be remembered that relatively large vessels are not expected to be present at very close proximity to the serosa, and especially, they do not run perpendicular to its surface.
The intermuscular elastic layer is generally not prominent, but can be accentuated around the ganglia of the myenteric plexus (figure 4), which is not uncommonly involved by CRC infiltrating the muscularis propria layer.
The use of neoadjuvant radiotherapy or chemoradiotherapy in rectal cancer may lead to elastosis, the accumulation of elastic fibres at uncommon locations. In such conditions, periglandular fibres obviously create the situation of rounded structures that may mimic an internal elastic lamina (figure 5A–D). Muscle bundles may also be surrounded by elastic fibres in such situations, and this may also create a background for a pseudo-VI pattern if the tumour infiltrates these bundles (figure 5E,F). Although more common after neoadjuvant irradiation, similar elastosis may develop in other noxious settings, for instance, hypoxia or tears resulting from herniation.
Intraneural spread with perineural elastic fibres, very uncommonly perinodal fibres (lymph nodes capsules are generally devoid of elastic fibres), and in anorectal carcinomas the perifollicular elastic sheet may give ground to the morphology of tumour cells surrounded by elastica positive layers, that is, pseudo-VI (figure 6).
As a conclusion, it is evident that the rate of VI in CRC is subject to significant interobserver and interinstitutional variation, and this was also demonstrated by Kojima et al.12 ,23 The routine use of elastica stains substantially increases the detection rate of VI. Although extramural VI is generally recognised as a prognosticator, it is likely that the localisation of VI is less important than its presence, and intramural VI may also result in metastases.18 ,24 Since this detection rate correlates with the occurrence of synchronous or metachronous haematogeneous metastases,8–11 and patients without VI have better cancer-specific or metastasis-free survival rates if the lack of VI is determined with an elastica stain,25–27 the routine use of elastica stains may afford a relatively cheap alternative for identifying patients with risk of systemic spread. On the other hand, the rate of elastica stain detected VI being around or above 60–70% is higher than the proportion of CRC cases developing distant metastasis, therefore VI has a false positive rate as a predictor of developing stage IV disease. The simple introduction of elastica stains may improve consistency of diagnosing VI as reported by some authors,13 but as the interpretation of the special stains has a learning phase, the agreement on diagnostic criteria, as exemplified by the work of Kojima et al is also an important issue in achieving more consistent reporting of this important prognostic parameter. VI is generally underdiagnosed, and elastic stained sections help to identify this phenomenon more precisely (eg, distinguishing vessel wall invasion from vessel invasion) and at higher rate, but the present article highlights a few situations which may predispose to overdiagnosing this phenomenon with special stains: arterial invasion, tangentionally sectioned subserosal elastic lamina, mucosal protrusion into the submucosa, periganglionic, perineural, perinodal and perifollicular elastic fibres or periglandular and perimuscular elastosis.
Contributors GC drafted the first version of the manuscript, made the photos during a period of several years. Diagnoses, collection and discussion of cases during this period, finalisation of the manuscript done by all authors: GC, IS and RB.
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.