Ordinary Colorectal Adenocarcinoma v s . Primary Colorectal Signet-Ring Cell Carcinoma Study Matched for Age, Gender, Grade, and Stage Dimitrios Psathakis, M.D., Thomas H. K. Schiedeck, M.D., Florian Krug, M.D., Elisabeth Oevermann, M.D., Peter Kujath, M.D., Hans-Peter Bruch, M.D.

From the Department of Surgery, University of Liibeck, L~beck, Germany

PURPOSE.. This study contributes to the characterization of primary colorectal signet-ring cell cancer in contrast to ordinary colorectal carcinoma. Primary colorectal signet- ring cell cancer is a rare but distinctive primary neoplasm of the large bowel with stilPcontroversial clinicopatho- logic features. METHODS: Clinicopathologic features and survival data are evaluated in comparison with those of the ordinary colorectal adenocarcinoma (nonsignet colo- rectal carcinoma) in a retrospective study matched for age, gender, grade, and stage. RESULTS: In a series of 1,600 consecutive colorectal cancer patients since 1979, 14 patients (0.88 percent) with a signet-ring cell cancer were identified. Gender ratio was balanced, and mean age was 67.5 years. The majority of patiens had an ad- vanced tumor stage at the time of diagnosis (57.1 percent Stage IV and 35.7 percent Stage liD. Median survival time was only 16 months. In a study matched for age, gender, grade, and stage, a lower survival rate was found for patients with signet-ring cell cancer, but the difference did not reach statistical significance. In contrast to non- signet colorectal carcinoma, signet-ring cell cancer was characterized by a significantly higher incidence of peri- toneal tumor spread (64.3 percent) and a lower inci- dence of hepatic metastases (14.3 percent) . CONCLU- SIONS: Signet-ring cell cancer represents a rare but distinctive primary neoplasm of the large bowel. It is frequently diagnosed in an advanced tumor stage, thus showing an overall poorer prognosis than nonsignet colo- rectal carcinoma. Usually only palliative surgery is possi- ble. A high incidence of peri toneal seeding and a low incidence of hepat ic metastasis is characteristic of signet- ring cell cancer. [Key words: Carcinoma; Signet-ring cell; Colorectal neoplasms]

Psathakis D, Schiedeck THK, Krug F, Oevermann E, Kujath P, Bruch H-P. Ordinary colorectal adenocarcinoma vs. pri- mary colorectal signet-ring cell carcinoma: study matched for age, gender, grade and stage. Dis Colon Rectum 1999; 42:1618-1625.

n 1951 Laufman a n d Saphir ~ we re the first t o de-

scr ibe colorecta l s ignet-r ing cell cancer as a dist inct

t ype of p r imary n e o p l a s m of the large bowel . A1-

Presented at the 24th Koloproktologie-Tage, Mfinchen, Germany, March 5 m 8, 1998, and at the XVIIth Biennial Congress of the International Society of University Colon and Rectal Surgeons, MalmO, Sweden, June 7 to 11, 1998. No reprints are available.

t hough a substant ia l n u m b e r of s tudies have b e e n

p u b l i s h e d s ince then, the r e p o r t e d c l in icopa tho log ic

data are variable. Repor t ed overal l survival rate shows

a w i d e range, f rom ex t r eme ly poor , "no 5-year survi-

vor", 2 to f ive-year survival rates of 49 percent , wi th the

genera l impres s ion that p rognos i s is p o o r e r than

p rognos i s o f o rd ina ry colorec ta l a d e n o c a r c i n o m a

(nons igne t colorecta l ca rc inoma) .~8 The ques t ion es-

pec ia l ly still l ack ing a: sat isfactory a n s w e r is w h e t h e r

the p rognos i s is genera l ly p o o r or the l ow overal l

survival of the p r imary colorecta l s ignet-r ing cell can-

cer in re la t ion to the nons igne t colorecta l ca rc inoma is

just a resul t of the grea ter p r o p o r t i o n of s ignet-r ing

cell cancer pat ients wi th an a d v a n c e d s tage at d iag-

nosis. Many studies p u b l i s h e d u sed n o n a t t a c h e d con-

trols or ones m a t c h e d on ly for age a n d gender . 3-6

O n l y two s tudies in the l i terature we re p e r f o r m e d

wi th a control g r o u p m a t c h e d for age, gender , and

stage.< 9 To ou r k n o w l e d g e no s tudy m a t c h e d for age,

gender , grade, a n d s tage has b e e n r e p o r t e d in the

l i terature. To further charac te r ize s ignet-r ing cell can-

cer in contras t to nons igne t cotorecta l ca rc inoma w e

p e r f o r m e d the fo l lowing re t rospec t ive s tudy m a t c h e d

for age, gender , grade , a n d stage.

P A T I E N T S A N D M E T H O D S

The da ta f rom 1,600 pat ients o p e r a t e d on for co lo-

rectal cancer b e t w e e n Janua ry 1979 a n d D e c e m b e r

1997 at the D e p a r t m e n t of Surgery, Univers i ty of Lfi-

beck , G e r m a n y were rev iewed . Fou r t een pat ients

wi th a p r imary colorec ta l s ignet-r ing cell cancer we re

identif ied. In a c c o rda nc e wi th the l i terature, the tu-

mors we re classif ied as s ignet- r ing celt cance r w h e n

the major i ty o f t umor ceils ( > 50 pe rcen t ) we re signet-

r ing cells. 1, z, 4, 10--12 A s e c o n d a r y colorecta l s ignet- r ing

cell cancer was e x c l u d e d b y different m e a n s ( endos -

copy, CT scan, and in t raoperat ively) . For 11 pat ients

1618

Vol. 42, No. 12 COLORECTAL SIGNET-RING CELL CARCINOMA 1619

follow-up information was sufficient, for long-term survival considerations. Clinicopathologic features of the three other patients were included in our consid- erations where possible (e.g., presenting symptoms and tumor stage).

To characterize signet-ring cell cancer in compari- son with nonsignet colorectal carcinoma, 56 patients matched for age and gender, 56 patients matched for age, gender, and stage, and 45 patients matched for age, gender, grade, and stage with a nonsignet colo- rectal carcinoma, treated as close in time as possible at the same institution, served as control groups. No adjuvant therapy was performed in the group of pa- tients from whom the nonsignet-ring cell cancer con- trol patients were recruited. The same was true for the signet-ring cell cancer patients. For 49 of the 56 pa- tients matched for age, gender, and stage and for 45 patients matched for age, gender, stage, and grade, long-term follow-up data for survival calculations were available. The following parameters were ana- lyzed: age, gender, presenting symptoms, duration of symptoms at the time of diagnosis, location of the tumor, morphology, Dukes and TNM classification, grade and stage at the time of diagnosis, site and pattern of metastases, incidence of peritoneal dissem- ination, surgical resection, survival rates, and median survival time. To evaluate the role of advanced tumor stage at the time of diagnosis, survival of the signet- ring cell cancer-group was compared with a nonsig- net colorectal carcinoma group matched for age, gen- der, and stage and a nonsignet colorectal carcinoma group matched for age, gender, stage, and grade. A match for grade was performed to investigate whether there is an additional influence of grade in an already stage-matched study. For the other parameters the nonsignet colorectal carcinoma group matched for age, gender, and stage served as a control. Emphasis was placed on the prognosis-associated parameters such as stage at diagnosis, recurrence pattern, and survival. The median follow-up period was 16 months for the signet-ring cell cancer group (mean, 15.8; standard error, 3.8; range, 0-31 months), 22 months for the nonsignet colorectal carcinoma group matched for age, gender, and stage (mean, 16.1; stan- dard error, 1.2; range, 0-64 months), and 27 months for the nonsignet colorectal carcinoma group matched for age, gender, grade, and stage (mean, 30.3; standard error, 3.8; range, 0-193 months).

Curative resection was defined as complete one- step removal of all gross tumor with a tumor-free distance between surgical margin and tumor of more

than 5 cm or if less distance (curative resection of carcinomas of the lower rectum with preservation of the sphincter ani) with negative surgical margins on microscopic examination.

For statistical analysis the SAS computer program (version 5, SAS Institute Inc., Cary, NC) was used. The cancer-related median survival time and the cumula- tive survival rate were estimated using the Kaplan- Meier method. The significance of differences in sur- vival rates were calculated using the log-rank test. To deterlnine the statistical significance of difference of other parameters the chi-squared test and the Sokal and Rohlf test were used.

RESULTS

Incidence

From 1,600 consecutive large-bowel carcinomas, 14 (0.88 percent) cases of primary colorectal signet-ring cell cancer were identified.

Age and Gender

In the signet-ring cell cancer group a balanced gender ratio was found. The patients had a mean age of 67.5 years, with a standard deviation of 16,9 years and a range from 21 to 90 years. Only one patient was younger than 40 years (21 years old).

Associated Diseases, Family History of Colorectal Cancer, and Symptoms

Two patients (14.3 percen0 with signet-ring cell cancer had a long history of ulcerative colitis (30 and 50 years). In the nonsignet colorectal carcinoma group one patient (1.8 percent) had such a history. No patient in either group suffered from Crohn's disease. Three patients in the nonsignet colorectal carcinoma group and none in the signet-ring cell cancer group had a positive family history" of colorectal cancer.

No patient with a signet-ring cell cancer was asymptomatic at the time of diagnosis. The most fre- quent presenting symptoms were a change in bowel habits (42.9 percent; nonsignet colorectal carcinoma, 44.6 percent), macroscopic rectal bleeding (42.9 per- cent; nonsignet colorectal carcinoma, 30.4 percent) and abdominal pain (42.9 percent; nonsignet colorec- tal carcinoma, 41 percent). The other symptoms were distributed as follows: Occult blood in feces (7.1 per- cent; nonsignet colorectal carcinoma, 8.9 percent), anemia (14.3 percent; nonsignet colorectal carci-

1620 PSATHAKIS ETAL Dis Colon Rectum, December 1999

noma, 0 percent), body weight loss (21.4 percent; nonsignet colorectal carcinoma, 32.1 percent), gen- eral weakness (0 percent; nonsignet colorectal carci- noma, 5.4 percent), subileus (14.3 percent; nonsignet colorectal carcinoma, 7.1 percent), iteus (28.6 per- cent; nonsignet colorectai carcinoma, 14.3 percent), vomiting (14.3 percent; nonsignet colorectal carci- noma, 8.9 percent).

Diagnosis and Emergency Procedures

colectomy, one sigmoid resection, one perineal resec- tion, and one intestinal bypass. Only in three cases (21.4 percent) were (potentially) curative procedures feasible.

Macroscopic Features

There was no significant difference in gross appear- ance between signet-ring cell cancer and nonsignet colorectal carcinoma in our study.

The essential diagnostic procedure at our institu- tion was colonoscopy. Only when total colonoscopy was not possible preoperatively, contrast enema or intraoperative colonoscopy was performed. Four pa- tients (28.6 percent) presented with signs and symp- toms of ileus, which led to laparotomy as an emer- gency procedure. In the nonsignet colorectal carcinoma group only three patients (5.4 percent) presented as emergency cases.

There was also no significant difference in opera- tion time. Despite the higher incidence of emergency procedures resulting from ileus in the signet-ring cell cancer group, the duration of symptoms was not sig- nificantly different in comparison with the nonsignet- ring cell group.

Site of Primary Tumor

The most common location of signet-ring cell can- cer was the right colon, with six cases (2 cecum signet-ring cell cancers, 2 signet-ring cell cancers in the ascending colon, and 2 signet-ring celt cancers in the hepatic flexure), followed by the left colon, with four cases (one case in the splenic flexure and three cases in sigmoid colon), and the rectum, with three cases. One tumor was located in the transverse colon. In the nonsignet colorectat carcinoma group the tu- mor site followed the well-established distribution of carcinomas in the large bowel, with the rectum (21 patients; 37.5 percent) and sigmoid colon (18 pa- tients; 32.1 percent) being the preferential tumor lo- cations. Twenty-five percent of the nonsignet colorec- tal carcinomas (14 patients) were located in the right hemicolon. Two of the three patients left had a car- cinoma in the splenic flexure and the remaining pa- tient had a carcinoma in the transverse colon.

Procedures Performed

The following procedures were performed: four right hemicolectomy, four anterior resection, three

Stage at Diagnosis (Table 1)

In the signet-ring cell cancer group no patient had Stage I disease (nonsignet colorectal carcinoma group, 11 patients; 19.6 percent). Only 1 patient (7.1 percent) had Stage II disease (nonsignet colorectal carcinoma-group, 15 patients; 26.8 percent), 5 pa- tients (35.7 percent) had Stage III disease (nonsignet colorectal carcinoma-group, 15 patients; 25 percent), and the majority of patients with signet-ring cell can- cer had Stage IV disease at the time of diagnosis (8 patients; 57.1 percent; nonsignet colorectal carcino- ma-group, 15 patients; 25 percent). Difference in stage between the signet-ring cell cancer group and the nonsignet colorectal carcinoma group was statis- tically significant (P < 0.05; Sokal and Rohlf test).

Synchronous or Metachronous Malignancy, Site of Metastases, and Survival

(Fig. 1; Table 2)

One patient in the signet-ring cell cancer group (7.1 percent) had synchronous signet-ring cell carcinomas in the cecum and the splenic flexure (nonsignet colo- rectal carcinoma group, 1 patient; 1.8 percent), and one patient with signet-ring cell cancer (7.1 percent) had a synchronous malignancy of other origin (non- signet colorectal carcinoma group, 1 patient; 1.8 per-

Table 1. Stage of the Signet-Ring Cell Cancer Group and the

Age-Matched and Gender-Matched Ordinary Adenocarcinoma Group

Stage Signet-Ring Ordinary Cell Cancer Adenocarcinoma

(n = 14) (n = 56)

I 0 11 (19.6) II 1 (7.1) 15 (26.8) Ill 5 (35.7) 15 (25) IV 8 (57,1) 15 (25)

Figures are number of patients and (percent).

Vol. 42, No. 12 COLORECTAL SIGNET-RING CELL CARCINOMA 1621

Survival (%)

6O

40-

20

m

I - - - "1

I

. . . .

I t.-- I

L . _ ]

0 ! ! ! ! ! !

0 6 12 18 24 30 36

months

Figure 1. Cumulative survival rates of the primary colorectal signet-ring cell cancer group (light lines), ordinary colorectal adenocarcinoma the groups matched for age, gender, and stage (heavy lines) and age, gender, stage, and grade (dashed lines).

Table 2. Pattern of Tumor Dissemination of Primary Colorectal

Signet-Ring Cell Cancer and Ordinary Colorectal Adenocarcinoma

Site of Signet-Ring Ordinary Cell Cancer Adenocarcinoma

Dissemination (n = 14) (n = 56)

Peritoneal 9 (64.3) 8 (14.3) Hepatic 2 (14.3) 24 (42.9) Pulmonary 0 4 (7.1) Ovarian 1 (7.1) 1 Other 2 (14.3) 4 Figures are number of patients and (percentage).

cent). Three patients (21.4 percent) had metachro- nous malignancies, but not of colorectal origin (nonsignet colorectal carcinoma group, 2 patients; 3.6 percent).

Pattern of dissemination is summarized in Table 2. In the signet-ring cell cancer group peritoneal seeding was the most frequent way of dissemination (nine

patients, 64.3 percent). Only two signet-ring cell can-

cer patients (14.3 percent) had hepatic metastases. This pattern of tumor dissemination is in contrast to the nonsignet colorectal carcinoma group. In the non- signet colorectal carcinoma group only 8 patients

(14.3 percent) showed peritoneal dissemination, and in 24 patients (42.9 percent) hepatic metastases were found. The different incidence of peritoneal and he- patic dissemination in the two groups is statistically

significant (P < 0.05; Sokal and Rohlf test). Three patients with signet-ring cell cancer died

within 30 days after surgery of cardiopulmonary causes. The age range of these three patients was between 77 and 90 years; two patients had Stage IV disease and one patient, Stage III disease.

Comparing the signet-ring cell cancer group with the control group matched for age, gender, and stage, there was a trend toward lower survival rates of the signet-ring cell cancer group, which does not reach statistical significance. Comparing the signet-ring cell

1622 PSATHAKIS ETAL Dis Colon Rectum, December 1999

cancer group with the control group matched for age, gender, grade, and stage, this trend was also present (Fig, 1).

D I S C U S S I O N

Signet-ring cell cancer is a rare but distinctive ma- lignant disease with still-controversial clinicopatho- logic features. In certain respects there are differences from nonsignet colorectal carcinoma. In this study we have presented the clinicopathologic findings of sig- net-ring cell cancer as an entity and a characterization of signet-ring cell cancer in relation to nonsignet colo- rectal carcinoma in a retrospective study matched for age, gender, grade, and stage.

The diagnosis of primary colorectal signet-ring cell cancer is based on the microscopic appearance of the specimen and the exclusion of secondary colorectal signet-ring ceil cancer of other origin. The histologic appearance of the tumor is characterized by cells with abundant intracytoplasmic mucin and peripherally placed nuclei, arranged solely or in loose clusters and spreading diffusely through the bowel wall (Figs. 2 and 3). Mucin lakes containing small, primitive, and abortive gland structures may be present. 9, 2 An over-

lap of signet-ring cell cancer and mucinous colorectal cancer is generally accepted, 1, 4, 10, n but when the

majority of cells are signet-ring ceils and a diffusely infiltrating growth pattern is present, the diagnosis can be made. 9, ,3 Because the majority of signet-ring

cell carcinomas (>96 percent) arise in the stomach, with a further small proportion being of other primary origin, including colorectum, gallbladder, pancreas, urinary bladder, and breast, one has to exclude sec- ondary infiltration of or metastasis to the cotorectum before definite diagnosis of primary colorectal signet- ring cell cancer can be established.< 14

In our series of 1,600 colorectal carcinomas, signet- ring cell cancer accounts for 0.88 percent (n = 14). Reported incidence in western countries ranges from 0.8 to 2.6 percent. 3, 4, 8, 3_0, 14, 15 The wide range of re-

ported incidence may reflect the fact that there are sometimes difficulties in establishing the histologic diagnosis, particularly in contrast to mucinous adeno- carcinoma of the large bowel, which is characterized by the majority of mucin accumulated being extracel- lular. 1° In addition, the frequent use of such gross descriptive terms as linitis plastica or lymphangiosis- type carcinoma instead of the histologically based term signet-ring cell cancer may have contributed to the variance in the literature, i, < 7, 10 In Jordan, Leba-

Figure 2. Primary colorectal signet-ring cell cancer (he- matoxylin and eosin; ×200). Normal colon crypts sur- rounded by diffusely infiltrating cancer cells. (S. Kr0ger, M.D., Institute of Pathology, University of L0beck, Ger- many.)

i !

Figure 3. Primary colorectal signet-ring cell cancer (he- matoxylin and eosin; ×400). Clusters of carcinoma ceils, many of them with signet-ring cell morphology. (S. KrLiger, M.D., Institute of Pathology, University of L0beck, Germany.)

non, and Zaire, countries with a low- overall preva- lence of colorectal cancer, a much higher incidence of mucinous and signet-ring cell carcinomas (14-31 per- cent) of the large bowel is reported. ~6-1s This phe-

nomenon might be related to causative and hereditary factors, dietary habits, environmental factors, or other unknown reasons. 16-1s

There are experimental data that suggest some pos- sible role of environmental factors in the pathogenesis of signet-ring cell cancer. The chemical carcinogen azoxymethane induces colorectal carcinomas in rats. Initially this tumor model was designed to study pathogenesis and behavior of the ordinary colorectal adenocarcinoma. However, many of the resulting tu- mors are signet-ring cell carcinomas rather than ade- nocarcinomas.9, 19

Vol. 42, No. 12 COLORECTAL SIGNET-RING CELL CARCINOMA 1623

The early form of signet-ring cell cancer is still

unclear and there is little information about it in the

literature. It is unknown whether signet-ring cell can-

cer usually develops from a pre-existing adenomatous

polyp or as a so-called d e n o v o carcinoma, which is

postulated not to have an adenomatous polyp as a

precursor. Only two cases of direct association of

signet-ring cell cancer and adenomatous polyps are

reported: a case of a six-yea>old male with a signet-

ring cell carcinoma in a polyp of the colon and a case

of a rectal adenoma with multiple foci of signet-ring cell carcinoma are described in the literature. 2°, 21 In

our study four patients had coexisting colorectal pol-

yps, but without foci of signet-ring cell carcinoma.

Thus, in our study it remains unclear whether signet-

ring cell cancer had an adenomatous polyp as a pre-

cursor. Controlled studies concerning the association of

adenomas and signet-ring cell carcinomas are lacking.

In the literature an association of inflammatory

bowel disease and signet-ring cell cancer has been

suggested, and figures up to 14 percent are report- ed.10, 11, 22, 23 Similar results are found in our study.

Two (14.3 percent) of 14 patients had a long history of

ulcerative colitis. Anthony e t al . 11 reported two cases

of Crohn's disease that developed signet-ring cell can-

cer. None of our patients with signet-ring cell cancer

had pre-existing Crohn's disease. The low incidence

of pre-existing Crohn's disease is consistent with the

generally accepted lower incidence of malignancy

comparing patients with Crohn's disease with those

having ulcerative colitis.

A positive family history is an established risk factor

for ordinary colorectal cancer. In accordance with

other studies, we have the impression that a positive

family history is not a predictive factor (at least not a

strong one) for developing signet-ring cell cancer. 4 In

our study no patient with signet-ring cell cancer had a

family history of colorectal cancer. This finding can be

related to the relatively small number of cases or may

reflect an aspect in the pathogenesis and variability of

signet-ring cell cancer. Some authors report that the mean age of patients

with signet-ring cell cancer is lower (23-53 years)

than that of patients with nonsignet colorectal carci- noma.6, 8, 9, 12, 24, 25 In our study we did not find this

phenomenon . Our patients had a mean age of 67.5

years, but there was one very young patient aged 21 years at the time of diagnosis. Tung e t al . 4 reported a

mean age of 39.6 years and more than 50 percent of patients being younger than 40 years. Anthony e t al . 11

found seven of 29 patients with signet-ring cell cancer

being younger than 40 years of age at the time of

diagnosis. In contrast to these studies and in accor-

dance with our findings, Messerini e t a l . 3 observed a

mean age of 63.5, and Secco e t a l . 26 observed a

median age of 60 years. What seems appropriate to

conclude is that the proportion of signet-ring cell

cancer in the group of young patients with colorectal

cancer is high, but the overall age range and median

age of patients with signet-ring cell cancer is similar to

that of patients with nonsignet colorectal carcinoma.

Our impression is that prognosis is particularly bad in

young patients.

In the literature data concerning the gender ratio of

signet-ring cell cancer are not consistent. A female predominance is r e p o r t e d , 11,12, I4, w, 26 but also a

slightly increased prevalence of signet-ring cell cancer in men. 7, 12, 27 In our study we found a balanced gen-

der ratio. Taking both our own data and the data

reported in the literature into account, there seems

not to be a clear gender preference of this tumor type.

The differences in data in the literature probably rep-

resent the statistical variation in studies with small

numbers of patients.

No statistical significance was found concerning the

initial presenting symptoms. Abdominal pain, a

change of bowel habits, .and macroscopic rectal b lood

loss were the most frequent initial complaints in both

groups. No patient in the signet-ring cell cancer group

was asymptomatic at the time of diagnosis. The inci-

dence of emergency procedures was significantly

higher in the signet-ring cell cancer group (28.6 vs. 5.4

percent), which is consistent with the higher propor-

tion of patients with advanced tumor stage.

In contrast to our results, other authors found the

majority of signet-ring cell cancer arising in the left colonT, 11, 14, 27 Almagr012 even pointed out that the

location in the left colon is an important distinguish-

ing feature between this tumor type and a metastatic

(secondary) signet-ring cancer of gastric origin; the

latter usually occurs in the transverse colon, being the

result of hematogenous spread through the gastro-

colic ligament. In our study six tumors were located in

the right colon, four tumors in the left colon, and

three tumors in the rectum. In accordance with our

findings, Messerini e t a l . 3 documented that the major-

ity of tumors were located in the right colon. Halve>

son and Seim 15 documented a strong right-sided pre-

dominance, with 9 of 11 signet-ring cell carcinomas

being in the right colon. Tung e t al . 4 and Anthony

e t al . 11 reported an even distribution of tumor sites.

1624 PSATHAKIS ETAL Dis Colon Rectum, December 1999

Summarizing, the location of a colorectat tumor seems not to allow a reasonable differential diagnosis.

In our study 42.9 percent of patients (n = 24) with a nonsignet colorectat carcinoma showed hepatic me- tastases and 14.3 percent (n = 8) developed perito- neal dissemination. In contrast to this distribution, peritoneal seeding was the most frequent way of dissemination in the signet-ring cell cancer group. A total of 64.3 percent of signet-ring cell cancers (n = 9) spread through the peritoneal cavity, and only 14.3

percent of patients (n = 2) showed hepatic metasta- ses (Table 2). In agreement with the literature, the completely different pattern of tumor dissemination,

with a high incidence of peritoneal metastases and relatively low incidence of hepatic metastases, is re- garded as a characteristic feature distinguishing colo-

rectal signet-ring cell cancer from nonsignet colorec- tal carcinoma. 3, 4, 7-11 The high incidence of peritoneal

dissemination may implicate consideration of intra- peritoneal chemotherapy in the palliative treatment of signet-ring cell cancer, 1° but in our own experience patients do not benefit from this palliative treatment. A significant incidence of ovarian metastases is de- scribed for signet-ring cell cancer, i, i0, 27, a, Anthony

et al. 11 reported that in 25 percent of patients with signet-ring cell cancer, ovarian involvement occurred. In our study one patient with signet-ring cell cancer had ovarian metastases.

A frequently advanced tumor stage of signet-ring cell cancer at the time of diagnosis is confirmed by our own and many other studies. 3,4,6-1~, ~5, 26 This

phenomenon is associated with a low overall survival and corresponds with a higher incidence of emer- gency procedures. The reported overall five-year sur- vival rates range from 9.1 to 50 percent, with the majority being between 20 and 40 per- cent.3, 4, 6, 7, ,0, 1~, 26 The reported median survival time

ranges from 15 to 24 months. 3, 6, 26 The overall sur- vival time of our own patients is extremely poor, with no three-year survivor and a median survival time of 16 months (Fig, 1). An interesting point is the question of whether the prognosis of signet-ring cell cancer and nonsignet colorectal carcinoma is also different when comparing patients matched for age, gender, grade, and stage. Survival rates and median survival time are still poorer for the signet-ring cell cancer group, but the difference is smaller and does not reach statistical significance (Fig. 1). This trend is also confirmed by others. 8, a4,25 Obviously, the overall poor prognosis of patients with signet-ring cell cancer is a result of the majority of patients presenting in an

advanced stage of disease, and the difference in the course of tumor disease is smaller than the first im- pression suggests. One reason for the high proportion of patients presenting with advanced tumor disease is a delay in diagnosis. This is probably a result of the fact that clinical symptoms tend to occur late in the

course of signet-ring cell cancer. The vast majority of signet-ring cell cancer presents as a diffusely infiltrat- ing carcinoma, which therefore is described by many authors as tinitis plastica and lymphangiosis-type colorectal carcinoma.~, 2, 6, 7, 27, 28 The relatively long

period of intramural growth without penetrating the mucosa may be responsible for the absence of rectal blood loss and narrowing of the bowel lumen. This might be one explanation for the obvious delay in diagnosis and frequently advanced stage at diagnosis of signet-ring cell cancer.

Summarizing, primary colorectal signet-ring cell cancer represents a relatively rare subgroup of pri- mary colorectal carcinoma. It is frequently diagnosed in an advanced stage of disease, thus showing an overall poorer prognosis than ordinary colorectal ad- enocarcinoma. This correlates with the fact that usu- ally only palliative surgery is performed. A character- istic feature of primary colorectal signet-ring cell cancer in contrast to nonsignet colorectal carcinoma is the high incidence of peritoneal dissemination and the low incidence of hepatic metastases. The rare incidence of this primary colorectal tumor type limits statistical considerations.

A C K N O W L E D G M E N T S

The authors thank C, Killaitis for assistance with the statistical analysis of results.

REFERENCES

1. Laufman H, Saphir O. Primary linitis plastica type car- cinoma of the colon. Arch Surg 1951;62:79-91.

2. Rao TR, Hambrick E, Abcarian H, Salgia K, Recant WM. Colorectal linitis plastica. Dis Colon Rectum 1982;25: 239-44.

3. Messerini L, Palomba A, Zampi G. Primary signet-ring call carcinoma of the colon and rectum, Dis Colon Rectum 1995;38:1189-92.

4. Tung S, Wu C, Chen P. Primary signet-ring cell carci- noma of colorectum: an age- and gender-matched con- trolled study. Am J Gastroenterol 1996;91:2195-9.

5. Umpleby HC, Ranson DL, Williamson RC. Peculiarities of mucinous colorectal carcinoma. BrJ Surg 1985;72:715-8.

Vol. 42, No. 12 COLORECTAL SIGNET-RING CELL CARCINOMa, 1625

6. Nakahara H, Ishikawa T, Itabashi M, Hirota T. Diffusely infiltrating primary colorectal carcinoma of linitis plas- tica and lymphangiosis types. Cancer 1992;69:901-6.

7. Shirouzu K, Isomoto H, Morodomi T, Ogata Y, Akagi Y, Kakegawa T. Primary tinitis plastica carcinoma of the colon and rectum. Cancer 1994;74:1863-8.

8. Connelly JH, Robey-Cafferty SS, E1-Naggar AK, Cleary KR. Exophytic signet-ring cell carcinoma of the colorec- turn. Arch Pathol Lab Med 1991;115:134-6.

9. Giacchero U, Aste H, Baracchini P, et al. Primary signet- ring carcinoma of the large bowel: report of nine cases. Cancer 1985;56:2723-6.

10. Kirkham N. Colorectal signet ring cell carcinoma in young people. J Pathol 1988;155:93-4.

11. Anthony T, George R, Rodriguez-Bigas M, Petrelli NJ. Primary signet-ring ceil carcinoma of the colon and rectum. Ann Surg Oncol 1996;3:344-8.

12. Almagro UA. Primary signet-ring carcinoma of the co- lon. Cancer 1983;52:1453-7.

13. Sasaki O, Atkin WS, Jass JR. Mucinous carcinoma of the rectum. Histopathology 1987;11:259-72.

14. Chowdhury JR, Das K, Das KM. Primary linitis plastica of the colon: report of a case and review of the litera- ture. Dis Colon Rectum 1975;18:332-8.

15. Halvorsen TB, Seim E. Influence of mucinous compo- nents on survival in colorectal adenocarcinoma: a mul- tivariate analysis. J Clin Pathol 1988;41:1068-72.

16. Ibrahim NK, Abdul-Karim FW. Colorectal adenocarci- noma in young Lebanese adults: The American Univer- sity of Beirut-Medical center experience with 32 pa- tients. Cancer 1986;58:816-20

17. Dajani YF, Zayid I, Malatjalian DA, Kamal MF, Colorec- tal cancer in Jordan and Nova Scotia: a comparative epidemiologic and histopathologic study. Cancer 1980; 46:420-6.

18. Ngala Kenda JF. Cancer of the large bowel in Africa. A 15-year survey at Kinshasa University Hospital, Zaire. Br J Surg 1976;63:966-8.

19. Shamsuddin AK, Trump BF. Colon epithelium. II. In- vivo studies of colon carcinogenesis. Light microscopic, histochemical, and ultrastructural studies of histogene- sis of azoxymethane-induced colon carcinoma in Fisher 344 rats. J Natl Cancer Inst 1981;66:389-401.

20, Hamazaki N, Kono S, Mimaya J, et al. A case of a six year old boy with a signet-ring celt carcinoma in a polyp of the colon. Acta Pathol Jpn 1987;37:1679-84.

21. Nakamura T, Nakano G, Sakamoto K. Adenoma of the rectum with multiple foci of signet-ring cell carcinoma: report of a case. Dis Colon Rectum 1983;26:529-32.

22. Ojeda vJ, Mitchell KM, Walters NI, Gibson MJ. Primary colorectal linitis plastica type of carcinoma: report of two cases and review of the literature. Pathology 1982; 14:181-9.

23. Kallie NR, Fisher GF, Chan CY. Linitis plastica of the cecum. Can J Surg 1980;23:282-4.

24. Lui IO, Kung IT, Lee JM, Boey JH. Primary colorectal signet-ring celt carcinoma in young patients: report of three cases. Pathology 1985;17:31-5.

25. Bonello JC, Sternberg SS, Quan SH. The significance of the signet-cell variety adenocarcinoma of the rectum. Dis Colon Rectum 1980;23:180-3.

26. Secco GB, Fardelli R, Campora E, et al. Primary muci- nous adenocarcinomas and signet-ring cell carcinoma of colon and rectum. Oncology 1994;51:30-4.

27. Amorn Y, Knight WA. Primary linitis plastica of the colon: reports of two cases and review of the literature. Cancer 1978;41:2420-5.

28. Nadel L, Mori K, Shinya H. Primary linitis plastica of the colon and rectum: report of two cases. Dis Colon Rec- tum 1983;26:736-40.