Article Text

Download PDFPDF

Use of an elastic stain to show venous invasion in colorectal carcinoma: a simple technique for detection of an important prognostic factor
  1. C J Howlett,
  2. E J Tweedie,
  3. D K Driman
  1. Department of Pathology, London Health Sciences Centre and University of Western Ontario, London, Ontario, Canada
  1. Correspondence to Dr D K Driman, Department of Pathology, London Health Sciences Centre, 339 Windermere Road, London, Ontario N6A 5A5, Canada; ddriman{at}uwo.ca

Abstract

Background: Venous invasion (VI) is an important prognostic factor in colorectal cancer; it is positively associated with visceral metastases and may affect the decision to treat with adjuvant therapy.

Aims: To evaluate whether an elastic tissue (Movat) stain facilitates identification of VI, the number of Movat-stained blocks needed to detect VI, and whether VI identified with a Movat stain is prognostically equivalent to VI identified on H&E-stained slides.

Methods: H&E-stained sections from colorectal carcinomas from the year 2000 (n = 92) were examined for VI and compared to Movat-stained slides. Clinical charts were reviewed to compare rates of metastases in VI-positive versus VI-negative patients.

Results: With the Movat stain, VI was identified in 44% of cases previously categorised as negative (p<0.001) on review of H&E slides alone. One Movat-stained section was often sufficient to identify VI, with a statistically significant benefit to performing multiple stains if necessary. In H&E sections, two clues helped identify VI: the “unaccompanied artery” sign, where large arteries were seen without an accompanying vein; and the “protruding tongue” sign, where smooth tongues of tumour extended into pericolic/rectal fat. Metastases were present in 61% of cases positive for VI compared to 35% in VI-negative cases (p = 0.03). 45% of cases positive for intramural VI only developed metastases (p = 0.39), while 65% of cases positive for extramural VI only developed metastases (p = 0.03).

Conclusions: Pathologists should look for morphological clues of VI in H&E stained sections; when VI is not apparent, an elastic tissue stain on all tumour blocks significantly improves identification of VI. Morphological clues include the “unaccompanied artery” and “protruding tongue” signs.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Colorectal cancer is the second leading cause of cancer-related death in the United States and Canada.1 2 In the post-surgical specimen, accurate assessment by the pathologist is critical to assess a variety of factors that are important in determining prognosis and the potential use of adjuvant therapy. Such factors include lymph node metastasis, venous or lymphatic invasion, serosal penetration, peritoneal involvement, and residual tumour following surgery with curative intent.3 4 This is especially important for stage II cancers (ie, tumour penetration through the muscularis propria (TNM tumour category pT3 or pT4, lymph node-negative, negative for distant metastasis), as documentation of adverse prognostic factors may induce oncologists to treat with adjuvant therapy.3 4 However, the pathological reporting of some prognostic factors is inconsistent and likely underreported, particularly venous invasion (VI).4 Therefore, many patients may be undertreated with potentially reduced survival.

Invasion of blood vessels by colorectal carcinoma as a prognostic factor has been studied at the gross and histological level since the late 1930s. Brown and Warren’s seminal study in 1938 first identified VI as a risk factor for metastatic disease.5 In their postmortem study on 170 rectal cancer patients, they showed the presence of visceral metastases in 71% of those cases positive for intravascular invasion, utilising smooth muscle-highlighting stains to facilitate identification of VI. No metastases were identified in cases without vascular invasion. Numerous subsequent studies have confirmed VI as an important, stage-independent prognostic marker with respect to risk of visceral metastasis and/or survival.4 6 7 8 9 10 11 12 13 14 15 However, the percentage of colorectal carcinoma cases positive for VI varies considerably among studies, with a range of 17–72%. This variability may partially reflect the routine use of special histochemical stains to facilitate identification of blood vessels in some studies,5 7 12 13 16 17 18 and not in others.6 10 11 14 15 In a recent review, Quirke and Morris4 suggest that extramural VI would be expected in approximately 30% of cases following careful pathological evaluation, whereas in routine practice it is usually around 10%.

The College of American Pathologists’ protocol for examination of colorectal cancer specimens recommends the documentation of intramural or extramural VI, but does not recommend the use of special stains or immunohistochemistry to help identify VI.19 Identification of VI may not be straightforward, as it can be mimicked by histological artefacts and obscured by pathological changes; our experience has been that the use of an elastic tissue stain is of considerable help in the identification of VI.

This retrospective study has four aims: (1) to establish whether the use of an elastic tissue stain can facilitate identification of both intramural and extramural VI in colorectal carcinomas versus H&E-stained slides only; (2) to establish the number of tissue blocks that need to be examined in order to detect VI; (3) to look for morphological clues of VI in H&E-stained sections; and (4) to compare clinical outcomes (ie, metastasis) of cases with intramural or extramural VI versus those without VI.

Materials and methods

Ethics approval for this study was obtained from the Health Science Research Ethics Board at the University of Western Ontario. The pathology department archives at the London Health Sciences Centre (LHSC) were searched for 100 consecutive colorectal cancer cases from the year 2000, irrespective of stage. Eight cases were removed from the study due to missing tissue blocks, leaving 92 cases. H&E-stained slides from all tumour blocks were independently examined by three reviewers—a gastrointestinal pathologist, a general anatomical pathologist and an anatomical pathology resident—and categorised as positive, negative or equivocal for VI.

VI was defined as tumour present in a well defined vessel, or in a vessel-like structure surrounded by presumed elastic fibres. Each reviewer then selected one block from each case that was considered most likely to contain VI. A Movat pentachrome stain was then performed on all tumour blocks and evaluated as positive or negative for VI, specifying location as intramural or extramural. These results were analysed to determine: (1) if the Movat stain improved detection of VI; and (2) whether a Movat stain on all blocks increased the yield of VI detection compared to a Movat stain on the single selected block only. In addition to analysis of data from individual observers, data from the three individual committee members were grouped in order to generate overall results. Any discrepancy among reviewers was resolved by group consensus.

Patients’ clinical charts were reviewed for evidence of metastasis (non-lymph node) to compare rates of metastatic disease in cases positive for VI versus those that were negative. In cases where follow-up data was not available at LHSC, attempts to contact the patients’ family physicians were made to try to acquire the data. At least two years of follow-up data postoperatively was required for inclusion in this aspect of the study (in the absence of an earlier history of visceral metastatic disease) as an arbitrary cut-off to allow metastases to declare themselves.

All statistical analyses were performed using χ2 tests.

Results

Demographics

Tumour stage was assessed through a blind review of all cases by one investigator. Of the 92 colorectal cancers, 3 (3%) were pT2, 69 (75%) were pT3 and 20 were pT4 (22%). Seventy-seven cases (84%) were adenocarcinoma NOS, while 15 (16%) were mucinous adenocarcinoma. Forty of 92 cases had lymph node metastases. Fifty-two patients received adjuvant treatment (chemotherapy ± radiation therapy), 38 patients did not, and 2 cases had no follow-up information in this regard. Of those patients who did not receive adjuvant therapy, 4 died in the immediate postoperative period.

Identification of VI

Both grouped (table 1) and ungrouped (table 2) data are presented. Initial review of the H&E-stained slides by the three reviewers resulted in the categorisation of 50 cases as negative (“negative” group), 25 as equivocal (“equivocal” group), and 17 cases as positive (“positive” group) for VI (table 1). On review of Movat-stained tumour sections (average number of tumour blocks was 3.6, range 2–5), VI was identified in 22/50 (44%) of the “negative” group cases (p<0.001), 19/25 (76%) of the “equivocal” group cases, and 16/17 (94%) of the “positive” group cases (table 1). Overall 57/92 (62%) Movat-stained cases were unequivocally positive for VI, compared with 17/92 (18%) without the use of the Movat stain (p<0.001). Of the positive cases, 15/57 (26%) were intramural only, 32/57 (56%) were extramural only and 10/57 (18%) contained both intramural and extramural VI. It was apparent that morphological clues of VI were present in many cases; in particular, tongues of tumour protruding into the pericolic fat and the presence of well circumscribed nodules of tumour adjacent to arteries where no normal vein was readily identified (fig 1).

Figure 1

(A–C) Protruding tongue sign: round-oval smooth-bordered protrusion of tumour into pericolic fat with adjacent artery visible in B; delicate concentric elastic fibres surrounding tumour in C (arrows). (D–E) Unaccompanied artery sign: a thick-walled artery is visible with an adjacent oval-shaped focus of tumour but no normal vein; concentric elastic fibres surrounding tumour in (F). (A,D: H&E; B,C,E,F: Movat pentachrome stain)

Table 1

Detection of venous invasion with H&E stain alone vs Movat stain (overall grouped data)

Table 2

Detection of venous invasion with H&E stain alone vs Movat stain (individual observer results), with effect of additional Movat-stained blocks

Effect of additional Movat-stained blocks

As table 2 shows, individual results for each investigator on cases thought to be free of VI on the H&E stain showed a statistically significant increase in yield of venous invasion when all tumour blocks were Movat-stained versus staining of one block (investigator #1 from 14/71 to 37/71, p<0.001; investigator #2 from 16/61 to 31/61, p<0.001; investigator #3 from 19/63 to 30/63, p = 0.0127). Conversely, no significant increase in yield of VI was accrued in cases initially deemed equivocal on H&E stain when all tumour blocks were Movat-stained versus staining of one block (investigator #1 from 11/16 to 15/16, NS; investigator #2 from 16/23 to 19/23, NS; investigator #3 from 9/13 to 11/13, NS). Sixteen of 17 cases deemed positive on initial H&E slides were confirmed by Movat stains (table 2).

Clinical significance

Follow-up data beginning in the year 2002 were available for 83 of 92 patients. Of these, 11 were removed from this portion of the study for the following reasons: 6 patients died within 2 years postoperatively with no known history of metastases, and 5 patients had inconclusive evidence of metastases (“possible” metastases in follow-up imaging studies). Of the 72 remaining patients, 37 had visceral metastases, while 35 had no evidence of metastatic disease.

VI in cases with metastases

A total of 28/37 patients with metastases had VI (17 extramural only, 5 intramural only, 6 both intramural and extramural), compared with 19/35 patients without metastases who had VI (9 extramural only, 7 intramural only, 3 both intramural and extramural) (p = 0.02).

Metastases in cases with VI

The presence of metastases in cases negative for VI (9/26) was compared with the presence of metastases in each of four groups (table 3), as follows: intramural VI only (5/11, p = 0.39), extramural VI only (17/26, p = 0.03), all cases with extramural VI (23/35, p = 0.02) and all cases with VI in any location (28/46, p = 0.03).

Table 3

Comparison of metastases in cases positive for venous invasion (VI) (all blocks Movat-stained) with cases negative for venous invasion

Metastases in cases of venous invasion diagnosed on H&E versus Movat stain

For the purposes of comparing venous invasion diagnosed on H&E alone versus cases that required additional sections stained with Movat, the cases that had been deemed equivocal on H&E stain (ie, features suggestive of, but not diagnostic for, venous invasion) were grouped with cases that were deemed negative on H&E. This allowed for evaluation of the clinical significance (ie, visceral metastases) of venous invasion diagnosed on H&E alone versus VI diagnosed on Movat stain (table 4). Thirteen cases of VI were diagnosed with H&E slides alone. Ten of these patients developed visceral metastases (77%). Of the remaining 59 cases that did not have diagnostic features of VI on H&E, 33 were shown to have VI with Movat-stained sections. Of these, 18 patients developed visceral metastases (55%).

Table 4

Comparison of visceral metastases in cases with venous invasion (VI) diagnosed on H&E stain versus Movat stain

Effect of chemotherapy

Of the 72 cases with outcome data, 43 had chemotherapy ± radiation therapy, while 26 did not have adjuvant therapy. In 2 cases, patient status regarding adjuvant therapy was unknown. Of patients who received chemotherapy, 30 patients had venous invasion, and of these 19 had metastases. Of patients who did not receive adjuvant treatment, 14 had venous invasion, and of these 6 had metastases. Statistical comparison of the number of metastases in patients who had adjuvant therapy versus those who did not was non-significant (p = 0.2).

Discussion

It is clear from our study and others that VI is an important negative prognostic factor in colorectal cancer. Reporting the presence of VI may have treatment implications for colorectal cancer patients, and may ultimately affect survival. Despite this, current guidelines do not recommend using special stains to facilitate identification of VI. In this study we have shown that the use of an elastic (Movat pentachrome) stain on at least one tissue block improves detection of both intramural and extramural VI over H&E staining alone.

Two previous studies that have compared elastic-stained to H&E-stained sections have also shown that the elastic stain is superior to H&E stain for the detection of VI.20 21 In the first study, Sternberg et al found that the use of an elastic stain (Weigert’s stain) resulted in the identification of an additional 19% of cases above that of H&E stain alone in a study of 81 stage IV (positive for distant metastasis) colorectal carcinomas (14). Interestingly they found that only intramural veins were involved in 32% of cases, supporting the significance of intramural VI as a negative prognostic factor. In the second study, Vass et al compared Miller’s elastic stained sections to H&E-stained sections in a retrospective analysis of 75 colorectal carcinomas (18). They reported increases of 29% and 19% in detection of intramural and extramural VI, respectively, when an elastic stain was utilised. Their increase in detection of VI was 30% overall when intramural and extramural results were combined.

While the prognostic significance of extramural VI has been firmly established across several studies, the significance of intramural VI (within the submucosa/muscularis propria) is more controversial. In three studies where intramural VI was documented, VI was limited to intramural veins in 9–34%, compared to 13–36% of cases were it was limited to extramural veins.8 10 15 Two of these three studies did not routinely use special stains to help identify VI.10 15 In the most recent of these studies, Petersen et al showed that both intramural and extramural VI are independent prognostic factors in Dukes’ stage B cancers.10 Their resulting prognostic index for Dukes’ B colorectal cancers scores extramural and intramural VI equally. The prognostic significance of intramural VI may be lost when all stages of cancers are combined, as shown by Talbot et al in rectal carcinomas.15 Our present study did not show intramural invasion as a significant adverse prognostic factor. This may have been due to the inclusion of colorectal carcinomas of various stages (T2, T3 and T4). Alternatively, the relatively small number of cases harbouring intramural invasion only in our study may not have allowed the results to reach statistical significance. Additional studies utilising elastic stains to specifically examine intramural VI in a stage-dependent manner may resolve this issue.

The effect of adjuvant therapy was also addressed. More than half of the patients had chemotherapy ± radiotherapy. The statistical analysis comparing VI and metastases in patients with and without adjuvant therapy was not significant. This may be secondary to the small patient numbers in this study, and the fact that many patients were of high tumour stage (T3 or T4). Future studies separating patients based on adjuvant therapy status on a stage-specific basis may resolve this issue.

In this study we have shown that there is a statistically significant benefit to staining a section from at least one tumour block, when VI is not identified in the H&E-stained tumour sections. It has been suggested in the literature that the variability in detection of VI across studies, especially where a high detection rate has been achieved without the use of an elastic stain, may result partially from extensive sectioning and/or pathologist expertise.21 While there will always be some variability in detection of VI from pathologist to pathologist, we feel that in routine practice, this variability will be decreased by the use of an elastic stain, and would recommend staining all blocks in cases that are negative, or at least one block in cases that are equivocal on review of H&E stained sections. Importantly, we have also shown that VI identified via an elastic stain has similar prognostic implications (with respect to metastases) to VI identified on H&E-stained sections alone. In other words, the implications of identifying VI with an elastic stain are equivalent to identifying VI on the H&E stain.

This is the first study to address the issue of how many tissue blocks need to be stained with an elastic stain to detect VI. Our results show that a statistically significant benefit accrues if more than one block is selected for an elastic stain on cases initially thought to be negative with H&E stain only. However, in cases that were initially deemed equivocal on H&E stained sections, there was no statistically significant improvement in yield of venous invasion when all blocks were stained versus one block. These cases were initially deemed “equivocal” for venous invasion due to the presence of morphological clues to the presence of VI. These clues include: (1) the presence of tongues of tumour protruding beyond the lower edge of the advancing tumour into the pericolic fat; and (2) the presence of relatively well circumscribed tumour nodules adjacent to arteries where no vein is readily apparent (fig 1). We did not observe any morphological clues which were helpful for identification of intramural VI, which was more difficult to identify on H&E stain.

Although our analyses of patient outcome (presence or absence of visceral metastases) groups the data generated from initial H&E review into VI-positive and VI-negative categories, we felt that it was important to include an “equivocal for VI” category when initially reviewing the H&E stained slides. This category reflected cases that showed features suggestive of, but not diagnostic for, VI. Equivocal features will often prompt a pathologist to perform ancillary studies such as an elastic stain. Additional Movat stained sections confirmed VI in the majority of cases categorised as equivocal for all three investigators (table 2).

In summary, the most time-efficient approach when evaluating a case for VI would be to stain all tumour blocks up-front with an elastic stain, and this would result in the highest pick-up of VI. Alternatively, one may search for morphological clues on the H&E-stained slides, ie the “protruding tongue” and “unaccompanied artery” signs, and select one block for sectioning and staining with an elastic stain. Staining additional blocks will increase the yield of VI if the initial elastic stain is negative. If no clues are present, we recommend staining all tumour blocks for maximum pick-up of VI.

Take-home messages

  • It is important to look for tumour invasion of large veins in colorectal cancer specimens, as the presence of venous invasion is positively associated with systemic (especially hepatic) metastases.

  • The use of an elastic stain on all tumour blocks results in the highest pick-up of venous invasion; alternatively, morphological clues of venous invasion should be sought and blocks assessed with an elastic stain, as appropriate.

  • Morphologic clues on H&E-stained slides to venous invasion include the “protruding tongue” and “unaccompanied artery” signs. The former is when there are tongues of tumour protruding beyond the lower edge of the advancing tumour into the pericolic fat, while the latter is when there is a relatively well circumscribed tumour nodule adjacent to an artery, where no vein is readily apparent.

REFERENCES

Footnotes

  • Competing interests None.

  • Ethics approval Ethics approval was obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.