Vascular Invasion in Carcinoma of the Colon and Rectum FRANCIS J . BURNS, M.D. AND J O t l N P F A F F ,
From the Departments of Surgery and Patbology, St. Louis Unirersity School of Medicine, St. Louis, Missouri.
JR.,
M.D., S L
Louis, A ' I i s s o u r i
ASCUtAR dissemination of colonic and rectal carcinoma has been recognized for inany years. However, until recently, most investigators have directed their attention priinarily toward the lymphatic spread of such
Clinic two different groupsf1.7 utilizing venous radiography, report 38 per cent and 43 per cent vascular invasion respectively. Others, 8-~2 mainly employing histologic methods of examination, have indicated blood vessel invasion varying from 19 per cent to 6x per cent. Few observers have stressed the finding of malignant tumor ceils within small vascular
TABLE 1
TABLE 11
V
No• Casc:
Per cent
~
A n a t o m i c Site
Rectum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rectosigmold . . . . . . . . . . . . . . . . . . . . . . . .
Sigmoid. . . . . . . . . . . . . . . . . . . . . . . . . . . . Left colon. . . . . . . . . . . . . . . . . . . . . . . . . . Transverse colon. . . . . . . . . . . . . . . . . . . . Right colon. . . . . . . . . . . . . . . . . . . . . . . . Total . . . . . . . . . . . . . . . . . . . . . . . . . . . .
No~
of Total
of CaseS
~64 34 6t 19 19
48
41
12
338
I O0
Surgical t u m o r resection . . . . . . . . . . . . . . Autopsy ............................ Surgical rcscctlon with a u t o p s y . . . . . A u t o p s y only . . . . . . . . . . . . . . . . . . . . . Biopsy of tile liver only . . . . . . . . . . . . . .
IO
18 6 6
82 276 92 I 27 I1 38 16 54
8
2
channels in and about the primary neoplasm. This study stresses the means of identification and the invasion of these small vessels by malignant cells. In addition, an attempt is made to correlate the incidence of this invasion with subsequent embolic tumor dissemination.
malignaneids. The organ most commonly sccdcd by these tumors is tile liver and while lymph channel metastasis to this organ is possible, its occurrence is unlikely. Since almost the entire venous drainage of the large intestine passes through the liver via the portal system, local tumor invasion of blood vessels would seem of greater significance. Standard textbooks t-4 either discuss such vascular invasion brietIy, or indicate that it occurs only in a relatively small percentage of cases. Investigators particularly interested in this problem have reported wMabIe results. By gross examination primarily, Dukes and Busseys found venous tumor involvement in x7 per cent of rectal carcinomas. At the Mayo American Journal of Surgcrjr Volume 9:t, Norember, t956
Per
cent
METttODS AND MATEIUALS
During a ten 3'ear period from I94O through x949, there were 57I cases of clinically diagnosed colonic and rectal carcinoma at Firntin Desloge and St. Mary's Hospitals. Of these, 338 were suitable for purposes of this study. Criteria of case selection included surgically resected and/or autopsied malignant epithelial tumors on which there was adequate tissue for evaluation of vascular invasion. Cases of metastatic neoplasia proved by biopsy of the 704
Vascular Invasion in Carcinoma of Colon and R e c t u m
FIG. I. Easily identifiable vein containing blood and t u m o r cells; hematoxylin and eosin stain, X 75. T A B L E III
T A B L E IV
No
A n a t o m i c Site
No. IP e r
"
Rectum and rectosigmold. Colon . . . . . . . . . . . . . . . . . . Total . . . . . . . . . . . . . . . . .
cent
of" with I with "ases Venous I V e n o t l s
I_
338 '1
Microscopic v e n o u s invasion with vascular m e t a s t a s i s . . . . . . . . . . . . . . . . . . . Microscople v e n o u s invasion w i t h o u t vascular m e t a s t a s i s . . . . . . . . . . . . . . . .
93 I 46-9 75 I 53.6 168
PCF
of
cent
of
ascs Cases
Invasion
,98 I 14 ~ I
~O.
49-7
6o
74
21
26
81
IOO
i
Total ..........................
liver were included only if the primary site of origin had been established. Adenomatous polyps showing only mucosal malignant change were excluded since controversy prevails conccrnlng the malignancy or benignancy of these lesions. Table I illustrates the anatomic.tumor distributlon which, when compared with previous statistical surveys, indicates a representative case group. Table H indicates case disposition. Some of the cases grouped under "surgical tumor resection" represent incomplete tumor excisions performcd as a palliative measure only. Histologic identification and differentiation of small veins, venules and lymph channels is in many instances extremely difficult. Special
histochemical technics, such as tile Masson trlchrome and the Weigcrt elastic tissue stains, were utilized when necessary to identify larger blood vessels. Routine hematoxylin and cosin preparations, however, were adequate in most instances. (Fig. I.) Other criteria were found to be useful in identifying vessels of smaller caliber. They are as follows: intraluminaI blood following tile endothelial contour; (Fig. 2) communication,with an identifiable vessel, usually of larger caliber; and number of endotlaelial cells, greater in venules than lymph channels. (Figs. 3 and 4.) Similar criteria were effective in evaluating tile tumor thromboembolic process. Neoplastic 705
Burns and Pfaff
Fro. 2. Vcnule s h o w i n g intraluminal blood which follows endothelial c o n t o u r and contains t u m o r cells; h e m a t o x y l i u and eosin stain, original magnification X 250.
TABLEV No. Cause of D e a t h
of C~.scs
Metastasis . . . . . . . . . . . . . . . . . . . . . . . . . Regional extension . . . . . . . . . . . . . . . . . . N e i t h e r m e t a s t a s i s nor regional extension . . . . . . . . . . '. . . . . . . . . . . . . . . . . . . Total . . . . . . . . . . . . . . . . . . . . . . . . . .
68
grade these tumors according to their relative invasive tendency, or to subdivide particnlar carcinoma variants (such as seirrhous, mucinous, and tile like). There were x58 (46.7 per cent) men and x8o (53.3 per cent) women. The average age of the patient at the time of initial examination of the tissue was fifty-nine years. Tile youngest was twenty-five and the oldest eighty-four years of age. Table m illustrates the frequency of venous invasion related to the tumor anatomic distribution. These results are based either upon the histologic findings of venous thromboemboli in the tissue specimeu or on the morphologic identificatioff of liver metastases. A large number of cases in which no evidence of gross visceral metastasis was observed at the time the tumor was resected revealed microscopic neoplastic invasion of blood vessels in the operative tissue Specimen. Only a relatively small number of these were able to be evaluated subsequently for metastatic vascular dissemination. Table Iv indicates the frequency of metastasis in these eases and is based upon postmortem examinations, gross findings of secondary surgical procedures and subsequent tissue biopsy specimens. The cause of death was determined in It9 (35 per cent) of the total cases studied. The
Pcr ccnt
of Cases
ii
57 9
4~
34
It9
ioo
ceils must either fill the vessel and follow an endothelial contour or be enmeshed within intraluminal blood. Careful interpretation is necessary in order that blood or tumor elements carried, by microtome sectioning, from adjacent stroma to an intralunfinal position are not mistaken for intravascular components. Histologlc evaluation of the resected tissue was done separately from and without knowledge of the clinical findings or subsequent autopsy examination. FINDINGS
Of these 338 cases, all were adenocarcinomas with the exception of one rectal squamous cell tumor. No attempt was made to classify or 7o6
Vascular Invasion in Carcinoma of Colon and Rectum
FIG. 3. Venule showing numerous endothelial cells and containing tumor cells; hema9 toxylin and eosln stain, Y z go.
d ....
L
....
.~--
Fie,. 4- Venulc showing numerous endothelial ceils and containing blood and tumor ceils; hematoxylin and eosin stain, X 250.
707
Burns a n d Pfaff majority (77 per cent) were evaluated by autopsy examination. In the remaining cases (23 per cent), roentgenograms, visceral organ biopsies or patient symptomatology indicated a predominant lethal factor. Table v represents this evaluation. Tumor invasion of lymph channels was noted in 290 (85.8 per cent) of the 338 cases. This finding would sccm of significance as regards complete evaluation in a study of tumor dissemination. It bears no direct relationship to the primary purpose of this investigation however, and will therefore not be included in subsequent discussion.
more of several factors may have been implicated. It is possible that all tumor tissue was removed at operation prior to tumor emboliza. tion, or that the patient died before metastases were recognizable. One may speculate that in other cases tumor thrombi would never become emboli,, or that if disseminated the emboli would remain dormant or die. Of tile I19 cases evaluated terminally as regards the cause of death, a majority (57.2 per cent) revealed visceral metastasis as the naost significant lethal factor. A small nmnber (9.3 per cent) comprised cases of regional tumor extension. These invaded locally, causing death by interference with adjacent vital structures (such as, uretcral obstruction with pyelonephritis and terminal uremia). This small proportion indicates that the growth pattern of colonic and rectal carcinomas is not one primarily of local extension. Approximately one third (33.5 pcr cent) of cases revealed no evidence of either metastasis or regional exten. sion of significant import to be considered other than minor contributing factors in the cause of death. Many were aged individuals who did not tolerate extensive surgical procedures. Pulmonary embolism, pneumonia and congestive heart failure were major causes of death in this group.
CO.MMEN~FS
The total percentage of blood vessel tumor invasion obtained in this study (49.7 per cent) is significantly higher than that previously reported by many investigators. This may be due in part to the exclusion of preinvaslve mucosal carcinomas from the group studied. The inclusion of ten such cases, none of which revealed vascnlar tumor invasion, would have lowered the percentage to 48.2. The major factor responslbIe for this higher percentage however, would appear to be the emphasis placed upon tumor invasion of minute vascular channels. There are several factors which probabIy indicate an even greater percentage of vascular invasion than this. study reveaIs. Material utilized was that obtained from routine pathologlc examination. Certainly a higher iueidence would be expected if all tissue were sectioned from a standpoint of detcrnlining such invasion. Further, some equivocal examples of tumor thromboembolisnl unquestionably could be proved by the examination of additional tissue. In addition, many tumors considered inoperable because of widespread visceral metastases (obvious vaseuIar dissemination) were excluded since no tissue was removed. Others 6a.9 in Smaller series of cases have previously reported a higher incidence of vascular invasion from rectal lesions than from those in the colon. Results of this study as shown in "Fable m do not indicate a significant difference. The results as indicated in Table w tend to emphasize the prognostic significance of tumor invasion of small venous channels. In a large majority of cases (74 per cent), subsequent "vascular dissemination occurred. In 26 per cent of cases with no evidence of metastasis, one or
CONCLUSION
From clinico-pathoIogic anaIysis of 338 eases selected from a total of 57I cases of colonic and rectal carcinoma reviewed, the following findings appear significant: I. Carcinomas of the colon and rectum invade blood vesseIs with more frequency than has previously been reported by many investigators. 2. Carcinomas of the rectum and rectosigmoid reveal approximately an equal tendency to vascular invasion when compared with those in the colon. 3. A majority (74 per cent) of patients with histologic vascular tumor invasion at the time of surgical carcinoma excision ultimately show neoplastic metastases. 4. A majority of patients with colon and rectal carcinoma die primarily as a result of visceral metastasis; relatively few of regional tumor extension; and approximately one third of non-associated complicating factors. REFERENCES t. ANDERSON, W. A. D. Pathology. St. Louis, t953. C. V. Mosby Co.
7o8
Vascular
I n v a s i o n in C a r c i n o m a
2. ACKERMAN, L. V. Surgical Pathology. St. Louis, I953. C. V. Mosby Co. 3. BOYD,W. A Text-Book of Pathology. Philadelphia, 1947. Lea & Febiger. 4. MOORE, R. A. Text-Book of Patholog2r Philadelphia, London, 1951. XV. B. Saunders Co. 5. DUKES, C. E. and BUSSEV, II. J. R. Venous spread in rectal cancer. Proc. Roy. Soc. Mad., 34: 571574, 194t6. BAP,RINGER, P. S., DOCKERTY, 1"~I. B., XVAoGII, J. M. and BAItGEN,J. A. Carcinoma of the large intestine. Surg., Gynec. O" Obst., 98: 62-73, 1954. 7. MADISOX, M. S., DocK~RIv, M. B. and Wxuon, J. M. Venous invasion in carcinoma of the rectum as evidenced by venous radiography. Surg., Gynec. er Obst., 99: 17o-178, 19548. Du,x.xt.xG, E. J., JOXES, T. E. and HAzaltt~, J. B. Carcinoma of the rectum. Ann. Surg., 133: t66-173, 1951. 9. Gm~.XELL, R. S. The spread of carcinoma of the co,on and rectum. Cancer, 3:641-652, t95oio. SEt-rEeD, P. H. and BAI~GV.Y,J. A. The spread of carcinoma of the rectum: invasion of lymphatics, veins antl nerves. Ann. Surg., 118: 76-90, x943. t i. Suxovrtt.A,~r~, D. A. The significance of vein invasion by cancer of the rectum and sigmoid: a microscopic study of 2Io cases. Cancer, 2: 429437, t94912. BltowN, C. E. and WAnnEN, S. Visceral metastasis from rectal carcinoma. Surg., Gynec. ey Obst., 66: 6t t-62I, t938. DISCUSSION A. F. C,xswno (Washington, D. C.): The paper just presented by Drs. Burns and Pfaff is indicative of the modern, more intensive and extensive study of cancer. T h e way in which cancer spreads is just one facet of this field, but a most important one if we are to prevent it. As the authors pointed out, we ha(.e known for many years t h a t cancer can spread via the vascular routes, but it has not been until recent times t h a t studies have been successfully geared to the uncquivocal dcmonstratlon of how this happens. The authors' paper t o d a y is one of several important articles on vascular invasion of cancer of the colon and rectum which have appeared in medical journals in the last decade or so. Before 1952, the average l~ercentage of venous involvement was 28 per cent. I t is important to note here t h a t the first good study was t h a t of Brown and Warren, and they reportcd 6t per cent. Subsequent to that, the percentage found by other investigators has dropped considerably. Dr. Dukes' cases were all on gross examination only, so I imagine his percentage should be increased at least twice as much as this if he would include in these cases the ones studied microscopically. Several others have since been reported. Barringer, Dockerty, Waugh and Bargcn demonstrated gross venous invasion b y cancer by means of radiography. Turnbull and Fisher stated t h a t in 7o9
of Colon and Rectum
32 per cent of the twenty-five cases they studied, tumor ceils were recovered from the blood of the m~jor mesenterle venous channels. The studies of this mode of spread of cancer of the colon and rectum have been consciously or subconsciously instigated b y the realization that patients with apparently non-invasive and therefore curable lesions of the colonic mucosa could and did d~e a few months later with widespread metastases to the liver, lungs, brain and sometimes superficially to the skin. In Table Iv of Dr. Burns' paper there was no reference to cases in which vascular metastases developed later on without apparent pathologic evidence of venous invasion. The explanatlon of this fact is not supplied by lymphatic spread because no pathologle evidence of lymphatic invasion is present. If ~vc stop and think of the enormous number of capillaries, arterioles and venules situated in I sq. cm. of bowel wall, and also think of the average size of a cancer of the colon or rectum when surgery is performed, it is surprising that these investigations have not been able to demonstrate vascular invasion in more than 50 per cent, as Dr. Burns and Dr. Pfaff's figures demonstrate. It seems pertinent to state t h a t the larger the lesion, the greater the chance of venous invasion. I agree with the authors t h a t there are many factors playing a p a r t in this t y p e of spread. For instance, the s t u d y of a cancer for the presence of venous involvement cannot be accurate until numerous sections are made of the entire tumor, or until venous involvement is demonstrated either by penetration of the wall, seen in microscopic section, or b y the presence of cancer cells in the venous blood of the specimen. Is it possible t h a t cancer cells are constantly carried to the liver in all cases of malignancy of the colon and rectum, b u t t h a t the liver is capable of destroying them in the great maiorlty of cases? Are all these cells capable of establishing a pathologic metastasis? These and other factors t h a t still remain unanswered are intimately associated with the modus operandi of the cancer ceils and the subsequent death of the pathologic cell. The paper presented is well prepared, and I am sure it is only the beginning of further investigation by the authors along these lines. FRm;cts J. BuaNs (closing): I agree with Dr. Castro t h a t the percentage probably should be hlghcr than is recorded. It probably would be, if more sections were taken, but this presents quite a problem for the laboratory. A practical implication to be drawn f r o m an investigation of this sort is t h a t the operator should do a high llgatlon of the blood vessels and should do it before there is very much if any manipulation of the primary growth.