Extramural venous invasion in rectal cancer Dr Gina Brown Royal Marsden Hospital UK gina.brown@rmh.nhs.uk

Vascular Invasion historic data Brown and Warren Surg Obstet Gynaecol1938 170 rectal cancer post mortem examinations majority palliative colostomy/ no surgery/ immediate postoperative death. histological evidence of tumour invasion of veins in 61% of 165 rectal adenocarcinomas 67 of the 100 patients with vascular invasion were found to have visceral metastases, mostly liver. Only one case of metastasis in the absence of any vascular invasion was found

Detection of venous invasion The search for vessel invasion as recommended by Brown and Warren. At least three sections of the tumor were taken in each case and stained with Masson's aniline blue trichromestain to emphasize the smooth muscle wall of the small veins.

Venous invasion important “as far as the prediction of visceral metastases in rectal carcinoma from the local growth and nodes is concerned, the presence of intravascular tumour means as much from the prognostic standpoint as neoplastic nodes, and their absence means much more” Brown and Warren 1938

Do you think this is a venous deposit or a Lymph node? Tumour along the course of a vessel Classifed as N1c disease – ie extranodal disease

Lymph node or venous deposit?

Poor interobserver agreement for EMVI Large variations in reporting rates 10% -50% - underreporting widespread Lack of agreement of definitions

EMVI detection by MRI EMVI is Present in 30%-40% of rectal cancer patients MRI enables pre-operative detection of EMVI.  Sometimes the vascular invasion is not continuous with the main tumour, such as in this example. This is a sagittal scan of a female pelvis, with an upper rectal tumour  highlighted by the red arrow and a separate nodule of tumour  within the superior rectal vein shown by the green arrow.  An oblique axial section from the same patient again shows the rectal primary,  and the tumour nodule. The corresponding histological section  clearly demonstrates tumour invasion of several small branches, and  complete filling of the lumen of a larger vessel. This example shows good correlation between radiology and histology, however, obvious examples of vascular invasion seen on an MRI scan are sometimes not identified by the pathologist because the normal vessel architecture has been obliterated by tumour. Histology of ‘nodule’ shows some microscopic EMVI (black arrows) and tumour filling lumen of larger vessel Upper rectal tumour (red arrow) + separate ‘nodule’ in superior rectal vein

Characteristic features of EMVI Expansion of extramural vessels by tumour Serpiginous / tubular extension of tumour signal MRI for detection of extramural vascular invasion in rectal cancer. AJR Am J Roentgenol 191(5): 1517-1522.

Grinnell – mapping of nodes along lymphovascular channels

Gross tubular extension along the course of lateral rectal vein

Gross lateral vein invasion

Venous invasion is associated with pelvic sidewall nodal spread

Histological EMVI status & Outcome n=135. Median follow-up=3·12 (0·9-5·7) years. Histological EMVI- 100 Histological EMVI+ In order to assess the prognostic significance of EMVI further, Kaplan-Meier survival curves were constructed which illustrate the differences in outcome between patients with and without histological vascular invasion. The median length of recorded follow-up was just over 3 years, therefore Relapse-free survival was chosen as the outcome measure. Obviously, the 13 patients with synchronous metastatic disease were excluded from the outcome analysis. The  relapse-free survival rate for patients who did not have EMVI was 73% at 3 years. This compared  to only 28% for those patients with histological EMVI. The difference was statistically highly significant.  80 73% % Relapse-free 60 p < 0·00001 40 28% 20 1 2 3 4 5 6 Time since operation (Years)

MRI-EMVI score & Outcome Smith et al. “Prognostic significance of MRI-detected Extramural Vascular Invasion." BJS. 2008 MRI-EMVI score & Outcome n=135. Median follow-up=3·12 (0·9-5·7) years. MRI-EMVI score= 0-2 100 MRI-EMVI score= 3-4 A similar outcome analysis was performed on the same cohort, this time comparing patients according to the MRI-EMVI status. We would not have expected these results to be as good because of the limitations of resolution and our inability to detect microscopic invasion. But As you can see, the separation of outcome curves was virtually identical to that obtained by histology. Relapse-free survival rates were  71% at 3 years for the negative group, compared to  only 32% for the positive group. Again the difference was  highly significant.  80 71% 60 % Relapse-free p = 0·0015 40 32% 20 1 2 3 4 5 6 Time since operation (Years)

MRI detected more persistent EMVI post CRT than pathology Chand M, Evans J, Swift RI, et al. Prognostic Significance of Postchemoradiotherapy High-Resolution MRI and Histopathology Detected Extramural Venous Invasion in Rectal Cancer. Ann Surg. 2014.

Survival curves – 3-year DFS mrVein invasion neg mrVein converted pos to neg mrVein remains pos after Rx

Which came first the spread into the vessels or spread into the lymph nodes?

Irresectable liver metastases developed after 1 year The maximum depth of invasion of the tumour is >15mm laterally on the right (growing into a vein, arrow) from the R lateral wall of the rectum Irresectable liver metastases developed after 1 year

McGill University Health Centre 230 patients with all imaging available 6 patients (2.5%) imaging unavailable for review 236 patients enrolled T Vuong, A Garant, G Artho R Lisbona McGill University Health Centre 94 low risk 136 high risk Whole group: 33/230 (14.3%) distant mets on PET/CT Odds Ratio 4.6 (95% CI 2.9-14.4) P=0.001 5/94 (5.3%) distant mets on PET/CT 28/136 (20.6%) distant mets on PET/CT Same mets PET/CT and CT 2/94 (2.1%) CT mets & more mets on PET/CT 2/94 (2.1%) Mets only on PET/CT 1/94 (1.1%) Same mets PET/CT and CT 10/136 (7.4%) CT Mets & more mets on PET/CT 8/136 (5.9%) Mets only on PET/CT 10/136 (7.4%) Odds ratio 4.6 (95% CI 1.3-16.2) P=0.01 Any mets on PET/CT not CT 18/136 (13.2%) Any mets on PET/CT not CT 3/94 (3.2%)

Serenade trial Phase II study : in patients with high metastatic risk colorectal cancer (vein invasion visible on MRI,T3>5mm) primary objective : find early liver spread diagnosed by Liver diffusion weighted MRI when CT scan is negative for metastatic disease.

Endpoint phase II The primary endpoint will be to show a >5% increase in the detection of unsuspected spread to liver detected in patients at high risk by DW-MRI when standard CT is negative or not able to confirm the presence of metastatic disease.

What do we hope to achieve with the Serenade trial? Improve survival by treating patients with very early spread to liver earlier and when spread is more susceptible to chemotherapy/surgery

MARVEL NCRN Study Tissue banking of rectal cancers Examining clinical behaviour in EMVI+ positive tumours following CRT Radiological and molecular change Multi-centre Tissue banking of rectal cancers Microarray analysis of tumour profile Predict behaviour

Hypothesis mrEMVI positive rectal cancer has worse relapse rates than EMVI negative rectal cancer following CRT CLINICAL endpoint Where mrEMVI positive rectal cancer changes to mrEMVI negative following CRT, it is associated with an improvement in time to relapse. IMAGING PREDICTIVE BIOMARKER mrEMVI positive rectal cancer is associated with worse response rates following CRT. IMAGING PREDICTIVE BIOMARKER EMVI positive rectal cancer exhibits a distinct molecular/genetic profile compared to EMVI negative rectal cancer. MECHANISM AND THERAPEUTIC PATHWAYS

The maximum depth of invasion of the tumour is >15mm laterally on the right (growing into a vein, arrow) from the R lateral wall of the rectum

Grade 3 EMVI

Grade 3 EMVI

Grade 4: EMVI – manifest as a discontinuous deposit

EMVI grade 4

Identification of high risk, predicted margin safe patients MRI Tumour spread >5mm or Extramural venous invasion Look for metastases at diagnosis and surveillance (SERENADE trial)

The future of MR EMVI A poor prognostic group 30-40% of patients with significantly worse DFS than EMVI negative Node positive patients EMVI strongly associated with nodal spread and is underreported by pathologists – deeper sections and elastin stains for MRI histopathology discordance Current and future trials will be able to assess impact of neoadjuvant chemotherapy in improving DFS for imaging identified high risk Training and support for radiologists to seek and document EMVI – close assessment and surveillance metastatic disease: the MARVEL trial