Int. J. Cancer (Pred. Oncol.
): 95, 18 –22 (2001)
© 2001 Wiley-Liss, Inc.
TISSUE EXPRESSION OF HUMAN CHORIONIC GONADOTROPIN ␤ PREDICTS
OUTCOME IN COLORECTAL CANCER: A COMPARISON
WITH SERUM EXPRESSION
M. LUNDIN1, S. NORDLING2, J. LUNDIN1, H. ALFTHAN3, U.-H. STENMAN3 and C. HAGLUND1,2*
1
Department of Surgery, University of Helsinki, Helsinki, Finland
2
Department of Pathology, University of Helsinki, Helsinki, Finland
3
Department of Clinical Chemistry, University of Helsinki, Helsinki, Finland
Production of the glycoprotein hormone human chorionic
gonadotropin ␤ (hCG␤) has been associated with more ag-
gressive behavior in non-trophoblastic tumors. In this study,
the prognostic value of immunohistochemical hCG␤ expres-
sion was evaluated in 239 patients with colorectal cancer.
Paraffin-embedded, formalin-fixed specimens were stained
with hCG␤-specific monoclonal antibody, and the results
were compared with serum levels determined with an assay
based on the same antibody. hCG␤ immunoreactivity was
seen in 52 of 239 tumors (22%). The difference in survival
time between patients with histologically hCG␤-negative
(median survival 94 months) and -positive (median survival
27 months) tumors was statistically significant (p ⴝ 0.014).
The risk ratio during follow-up for patients with positive
hCG␤ tissue expression was 1.65 (95% CI 1.11–2.46). In a Cox
multivariate analysis, Dukes’ stage, hCG␤ and age remained
independent prognostic factors. There was moderate agree-
ment between immunohistochemical and serum expression
levels of hCG␤ (␬ ⴝ 0.30). Using a combination of histological
and serum levels of hCG␤, the difference between survival
rates was highly significant (p < 0.001). The accuracy when
predicting 5-year survival status with the combined results of
serum and tissue expression was 1.3% higher compared to
hCG␤ tissue expression alone. Our results show that hCG␤
expression in both tumor tissue and serum has prognostic
significance independent of other clinicopathological vari-
ables. Positive tumor staining does not always occur together
with elevated serum levels, and the prognostic accuracy can
slightly be increased by combining the results.
© 2001 Wiley-Liss, Inc.
Key words: hCG␤; colorectal cancer; prognosis; immunohistochem-
istry; tumor markers
Human chorionic gonadotropin (hCG) is a glycoprotein hor-
mone composed of 2 non-covalently linked polypeptide subunits,
hCG␣ and hCG␤.1 hCG has traditionally been used as a marker for
pregnancy, pregnancy-associated disorders and trophoblastic dis-
ease; but it is also secreted in non-trophoblastic malignancies.2,3
The presence of hCG in any of its forms is an in vivo phenotypic
characteristic of human cancer cells.4 Intact hCG appears to be the
main form of hCG immunoreactivity (hCG-IR) during growth and
local invasion of a tumor, whereas expression of the free ␤ subunit
(hCG␤) correlates more closely with a metastatic phenotype.4
hCG-IR has been detected in a wide variety of non-trophoblastic
neoplasms in both tumor cells5–7 and serum.9 –14 In colorectal
cancer, the reported rate of elevated serum hCG is between 2% and
41%2,3,12,14 –17 and the rate of immunohistochemical tissue expres-
sion is between 17% and 52 %.18 –25 In studies on the hCG␤
subunit, no expression has been found in non-neoplastic mucosa of
the rectosigmoid colon or in benign colorectal neoplasms.4,26,27
According to several reports, hCG production is associated with
more aggressive behavior in non-trophoblastic tumors.18,28,29
Other studies indicate that hCG expression is of limited prognostic
value.12,17,30,31 The usefulness of hCG as a prognostic marker in
colorectal cancer is also undetermined. Some immunohistochem-
ical studies suggest that it is,21 or may be,18,19,22,25 a prognostic
marker in colorectal cancer, while in 2 studies it was of no
prognostic value.20,24 Most studies have found no association
between elevated serum levels of hCG and aggressive tumor
Publication of the International Union Against Cancer
behavior in colorectal cancer.12,15,17 In most reports, a polyclonal
antibody reacting with hCG␤ but with high cross-reactivity to
intact hCG has been used.32
We have previously shown that serum hCG␤ is an independent
prognostic factor in colorectal cancer, using an immunofluoromet-
ric assay based on a ␤-specific monoclonal antibody (MAb).11 The
purpose of the present study was to investigate the incidence of
immunohistochemical tissue expression of hCG␤ in colorectal
tumors using the same MAb as in the serum assay and to correlate
its presence with clinicopathological parameters. We also aimed at
comparing tissue and serum expression levels of hCG␤ and as-
sessing the prognostic significance of these separately and in
combination.
MATERIAL AND METHODS
Patients
Surgical specimens of primary tumors were obtained from 239
patients with histologically verified colorectal cancer treated at the
Department of Surgery, Helsinki University Central Hospital, be-
tween 1984 and 1989. The study group consisted of 120 men
(median age 69, range 25– 85, years) and 119 women (median age
70, range 36 – 88, years). Tumors were classified according to the
modified Dukes classification.33 Thirty-six patients had Dukes’ A,
92 patients Dukes’ B, 57 Dukes’ C and 54 Dukes’ D colorectal
cancer. Colonic cancer was found in 138 and rectal cancer in 101
patients. There was no difference in oncological treatment routine
between patients with rectal and colonic cancers. Survival data to
the end of 1997 were obtained from patient records, the Finnish
Cancer Registry and the Population Registry. Median follow-up
time was 116 months (range 101–161 months). During follow-up,
124 (52%) patients died of colorectal cancer.
Tissue samples
Tissue samples taken at operation were fixed in 4% buffered
formaldehyde, processed, embedded in paraffin and stored in the
files of the Department of Pathology, University of Helsinki.
Hematoxylin and eosin–stained slides of all samples were re-
viewed by 1 pathologist (SN) and the most representative samples
chosen for immunohistochemical staining. The histological type of
carcinomas was evaluated according to established criteria.34
Grant sponsors: Finska Läkaresällskapet; Medicinska Understöds-
föreningen Liv och Hälsa.
*Correspondence to: Department of Surgery, Helsinki University Cen-
tral Hospital, P.O. Box 262, FIN-00029 Helsinki, Finland. Fax: ⫹358-9-
47176093. E-mail: caj.haglund@helsinki.fi
Received 15 February 2000; Revised 7 August 2000; Accepted 14
August 2000
Published online 27 January 2001
 hCG␤ IN COLORECTAL CANCER 19

Staining procedure TABLE I – DISTRIBUTION OF hCG␤ IMMUNOREACTIVITY ACCORDING TO

SERUM LEVEL OF hCG␤ AND PREOPERATIVE CHARACTERISTICS IN 239
Immunohistochemistry was performed with a MAb specific for PATIENTS WITH COLORECTAL CANCER
hCG␤.8 Cross-reaction with hCG was 0.1%. With hCG␣, lutein- hCG␤
izing hormone and follicle-stimulating hormone, the cross-reaction positive
␹2
was ⬍0.05%. Clinicopathological variable Patients patients p value
n %
Four-micrometer-thick paraffin sections were mounted on slides
coated with 3-aminopropyl-triethoxy-silane (Sigma, St. Louis, Serum hCB␤ level 21.15 ⬍0.001
MO) and dried for 12 to 24 hr at 37°C. Sections were deparaf- ⬍2 pmol/L 192 32 17
finized in xylene and rehydrated through graded concentrations of ⬎⫽2 pmol/L 40 20 50
ethanol and distilled water. Sections were treated with 0.5% tryp- Age 1.62 0.203
sin (Sigma) for 40 min at 37°C. All sections were then incubated ⬍69.5 years 120 23 19
in 0.5% hydrogen peroxide in methanol for 30 min, to block ⱖ69.5 years 119 31 26
endogenous peroxidase. The next step was incubation with a 1:67 Tumor location 0.99 0.319
dilution of non-immune mouse serum. This was followed by Colon 138 28 20
overnight incubation at room temperature with the antibody, di- Rectum 101 26 26
luted 1:1,000. Gender 2.29 0.131
Bound antibody was visualized by the avidin– biotin complex Female 120 32 27
(ABC) immunoperoxidase technique (Elite ABC Kit Vectastain; Male 119 22 18
Vector, Burlingame, CA) following the manufacturer’s instruc- Histologic type 1.89 0.585
tions. Sections were incubated with the biotinylated second-layer Well differentiated 4 2 50
antibody and peroxidase-labeled ABC for 30 min each. Moderately differentiated 141 30 21
All dilutions were made in PBS (pH 7.2) and all incubations in Poorly differentiated 47 11 23
the ABC method carried out in humid chambers at room temper- Mucinous carcinoma 47 11 23
ature. Between each step in the staining procedure, slides were Dukes stage 7.96 0.047
rinsed in 3 changes of PBS. Dukes A 36 2 6
Dukes B 92 21 23
Peroxidase staining was visualized with 3-amino-9-ethyl-carba- Dukes C 57 15 26
zole (Sigma, A-5754), 0.2 mg/ml in 0.05 M acetate buffer con- Dukes D 54 16 30
taining 0.03% perhydrol (pH 5.0) at room temperature for 15 min.
Finally, sections were lightly counterstained in Mayer⬘s hematox-
ylin and mounted in aqueous medium (Aquamount; BDH, Poole,
UK). (8%) and strong in 4 (2%) specimens. Adjacent colonic mucosa
did not express hCG␤.
A placental tissue section was included as a positive control in
every staining batch. Sections treated with non-immune mouse There was no statistically significant association between the
serum served as negative controls. level of hCG␤ IR and age, tumor location, histological differenti-
ation or gender (Table I). However, Dukes’ stage was significantly
Red cytoplasmic staining against blue counterstain of the nuclei associated with hCG␤ IR. The proportion of positive stainings in
was considered a positive result. Staining intensity was graded on the different Dukes’ stages were 6% (stage A), 23% (stage B),
a scale of 0 through 3 by scoring the percentage of positive cancer 26% (stage C) and 30% (stage D).
cells as absent (0), ⬍5%; weak (1), 5% to 20%; moderate (2), 21%
to 35%; or strong (3), ⬎35%. The intensity of the stain did not Serum levels and histological expression of hCG␤
influence the score.
There were 40 patients (17%) with a serum level above the
Serum samples and assays cut-off (2 pmol/l). Median serum levels of the immunohistochemi-
A total of 232 pre-operative serum samples were available for cally negative and positive patients were 0.73 and 1.45 pmol/l,
analysis. Serum levels of hCG␤ were measured by an immunoflu- respectively. The corresponding average serum levels were 1.47
orometric assay, as described previously.35 The detection limit was and 6.31 pmol/l. The difference was significant (p ⬍ 0.01). Median
0.5 pmol/l. The upper reference limit (2 pmol/l) was used in the serum levels of patients with weak, moderate and strong immu-
statistical analysis. nohistochemical stainings were 1.04, 2.19 and 6.28 pmol/l, respec-
tively.
Statistical analysis One hundred and sixty patients were both histologically and
The ␹2 test was used to test for association between factors. serologically negative for hCG␤, 32 were histologically positive
Agreement was assessed with ␬statistics. Life-tables were calcu- and serologically negative, 20 were histologically negative and
lated according to Kaplan-Meier. Deaths were deaths due to colo- serologically positive and 20 were hCG␤-positive by both histol-
rectal cancer. Deaths due to other causes were censored. The ogy and serum. The observed percent agreement was 77.6% and ␬
statistical significance of the difference in survival between groups 0.298, which can be interpreted as fair.
was calculated using the log-rank test. Multivariate survival anal-
ysis was performed with the Cox proportional hazards model36 in Expression of hCG␤ in tumor tissue and survival
a backward stepwise manner, entering the following covariates: Dukes’ stage showed the strongest association with survival in
age, gender, Dukes’ stage, tumor location and histological differ- univariate analysis, followed by the combination of hCG␤ expres-
entiation. Histological expression and serum levels of hCG␤ and a sion in tissue and serum, serum level alone, hCG␤ IR alone and
combination of these 2 (either positive) were entered into the age (Table II). Gender, tumor location and histological differenti-
model. A p value of 0.05 was adopted as the limit for inclusion or ation were not significantly associated with survival.
removal of a covariate. The nominally coded variables were The difference in survival time between patients with histolog-
grouped as in Table I. ically hCG␤-negative and -positive tumors was statistically signif-
icant (p ⫽ 0.014; Fig. 1, Table II). The risk ratio during follow-up
for patients with positive hCG␤ tissue expression was 1.65 [95%
RESULTS
confidence interval (CI) 1.11–2.46]. Median survival in the hCG␤-
Immunohistochemical tissue expression of hCG␤ negative group was 94 months compared to 27 months in the
hCG␤ IR (ⱖ5% positive cancer cells) was observed in 52 of 239 hCG␤-positive group. Cumulative 5-year survival rates of hCG␤-
tumors (22%). Staining was weak in 32 (13%), moderate in 18 negative and -positive patients were 57.3% and 33.8%, respec-
20 LUNDIN ET AL.

TABLE II – UNIVARIATE ANALYSIS OF THE RELATIONSHIP BETWEEN PREOPERATIVE CHARACTERISTICS AND SURVIVAL IN 239 PATIENTS WITH
COLORECTAL CANCER

5-year cumulative
Clinicopathological variable Patients 95% Cl ␹2 p value
survival, %

hCG␤ immunoreactivity 6.05 0.014

Negative 185 57.3 54.5–60.2
Positive 54 33.8 30.4–37.2
hCG␤ serum level 21.36 ⬍0.001
⬍2 pmol/L 192 58.6 56.8–60.4
⬎⫽2 pmol/L 40 23.1 19.5–26.8
Combination of hCG␤ immunoreactivity and serum level 24.49 ⬍0.001
Both negative 160 61.6 59.7–63.6
Hist. pos & serum neg 32 42.9 38.4–47.5
Hist. neg & serum pos 20 25.7 20.5–31.0
Both positive 20 20.6 15.7–25.6
Age 4.25 0.039
⬍69.5 years 120 57.3 55.0–59.6
ⱖ69.5 years 119 47.3 45.0–49.7
Tumor location 0.17 0.678
Colon 138 52.6 50.4–54.8
Rectum 101 51.8 49.3–54.4
Gender 0.11 0.740
Female 120 52.0 49.7–54.4
Male 119 52.6 50.3–55.0
Histologic type 1.22 0.747
Well differentiated 4 75.0 64.2–85.9
Moderately differentiated 141 54.2 52.1–56.4
Poorly differentiated 47 51.4 47.7–55.2
Mucinous carcinoma 47 45.6 41.9–49.3
Dukes stage 187.58 ⬍0.001
Dukes A 36 86.1 83.2–89.0
Dukes B 92 73.1 70.7–75.5
Dukes C 57 38.7 35.3–42.1
Dukes D 54 7.4 5.6–9.2
␹2 and corresponding p values were calculated using the log-rank test.

tively. Within the separate Dukes’ stage groups, survival of hCG␤-

positive patients was poorer than that of hCG␤-negative patients,
but the differences were not significant. In Dukes’ C cancer, the
difference approached significance (p ⫽ 0.072).

Serological and histological expression of hCG␤ and survival

The difference in survival time between patients with a serum
level below and above 2 pmol/l was highly significant (p ⬍ 0.001),
as described earlier11 (Table II). The risk ratio during follow-up for
patients with elevated serum hCG␤ was 2.62 (95% CI 1.71– 4.00).
Median survival in the group with a serum level ⬍2 pmol/l was
100 months compared to 13 months in the group with elevated
serum levels. Cumulative 5-year survival rates of patients with
serum levels below and above the cut-off were 59% and 23%,
respectively.
The difference between survival rates when using a combination
(either positive) of histological expression and serum levels of
hCG␤ was highly significant (p ⬍ 0.001), with a better separation FIGURE 1 – Kaplan-Meier survival curves for 239 colorectal cancer
of patients than when using tissue or serum expression alone (Fig. patients with positive (solid line) vs. negative (dashed line) hCG␤
2, Table II). Using these variables as predictors of 5-year cancer- immunohistochemical staining. The difference in survival time is
specific survival, histological expression of hCG␤ was correct in significant.
59.0%, serum level in 61.5% and the combination in 62.8% of
patients. Differences were not statistically significant.
DISCUSSION
When adjusting for all covariates in a backward stepwise Cox
multivariate analysis, Dukes’ stage, immunohistochemical tissue Our study shows that hCG␤, measured immunohistochemically
expression of hCG␤ and age emerged as independent prognostic in tumor tissue, is an independent prognostic factor in colorectal
factors (Table III). Also, if the histological level of hCG␤ was cancer. There was a moderate correlation between expression in
substituted with the serum level or the combination of serum level tissue and serum, and the prognostic accuracy was increased,
and histological expression, the same variables were retained in the though not significantly, by using a combination of tissue and
model. serum expression.
 hCG␤ IN COLORECTAL CANCER
FIGURE 2 – Kaplan-Meier survival curves for 239 colorectal cancer
patients according to hCG␤ expression in tissue and serum. Tissue-
and serum-negative (solid line), tissue-positive and serum-negative
(short dashed line), tissue-negative and serum-positive (dash-dot-dash
line), tissue- and serum-positive (long dashed line).
TABLE III – STEPWISE MULTIVARIATE ANALYSIS OF PROGNOSTIC
COVARIATES OF SURVIVAL IN 239 PATIENTS WITH COLORECTAL
CANCER
Covariate RH p value
Dukes stage ⬍0.001
Dukes A
Dukes B 1.176 0.700
Dukes C 4.087 ⬍0.001
Dukes D 19.050 ⬍0.001
Age 1.024 0.013
hCG␤ immunoreactivity 1.299 0.047
Gender NS
Histologic type NS
Tumor location NS
RH ⫽ relative hazard. NS ⫽ not significant. Age was entered as a
continuous and hCG␤ immunoreactivity as an ordinal variable, all
other variables were nominally coded.
With 239 patients, this is, to our knowledge, the largest report on
immunohistochemical tissue expression of hCG in colorectal can-
cer. We also compared hCG expression in serum and tumor tissue
as prognostic factors in the same patient series using the same
antibody for immunohistochemistry and determination of serum
levels.
Our specimens were stained with hCG␤-specific MAb, which
we previously used in the determination of serum levels.8,9,11 In
the hCG␤ assay, cross-reactivity with intact hCG was 0.1%, while
a cross-reactivity of 60% with intact hCG has been reported for the
most widely used polyclonal antibody.32
Expression of hCG␤ has been associated with a metastatic
phenotype of cancer cells,4,21,26,37 but the biological mechanisms
behind this association remain unclear. In vitro, hCG␤ increases
growth in cancer cell lines38,39 and an autocrine/paracrine stimu-
lation of tumor growth by endogenously produced hCG␤ has been
suggested.38 It has also been suggested that hCG has an immuno-
suppressive effect by changing the cell-mediated and humoral
immune response to the stimulus of cancer antigens40 – 42 and that
hCG on the tumor surface suppresses the action of T cells, thereby
favoring high proliferative activity and invasive capacity.7 Also,
hCG-positive tumors have a multifaceted resemblance to tropho-
blastic tissue, which has been suggested to be a contributory factor
to the more aggressive behavior.22 The most widely adopted hy-
pothesis on the histological origin of hCG␤-producing cells is that
of trophoblastic metaplasia within the carcinomatous tissues.43– 45
21
In the present study, the proportion of positive hCG␤ tumors
was between 6% and 30% in the different Dukes’ stages (total 22%
positive), which is consistent with previous reports (17% to
52%).18 –25 The strong association between hCG␤ and stage of
disease suggests that the variations in proportion of positive tu-
mors between different studies could at least partly be explained
by differing stages of the included patients. We did not observe a
correlation between hCG␤ and histological differentiation, which
was reported in a previous study.19
In our study, immunohistochemical expression of hCG␤ was
compared with previously reported serum levels of the same
patients. hCG␤ in tumor tissue, like in serum, emerged as a
prognostic factor in both uni- and multivariate analyses. Within the
Dukes’ stages, hCG-positive tumors carried a worse prognosis but
the differences in survival were not statistically significant. In
other reports, expression of hCG␤ has been associated directly
with reduced survival21,25 and indirectly through a correlation with
poor histological differentiation,19 tumor invasiveness18,19,25 and a
higher incidence of metastases.22 However, no correlation between
hCG␤ staining and survival was found in 2 studies.20,24
The proportion of patients with positive hCG␤ expression was
higher in tumor tissue (52/239, 22%) than in serum (40/239, 17%).
There were 20 patients in whom the immunofluorometric serum
assay showed circulating hCG␤, while immunohistochemical
staining could not verify tumor origin. This could be due to a lower
detection threshold of the sensitive serum assay. Positive serum
levels despite negative tissue stainings raise the question about
possible destruction of negatively charged sialoglycoproteins due
to the formaldehyde fixation method. Therefore, a small test series
of corresponding formalin-fixed, paraffin-embedded and frozen
sections from the same tumors was stained for hCG␤. There were
no notable staining differences between fixed and frozen speci-
mens (data not shown). Vice versa, there were 32 patients with
histologically positive tumors but serum levels below the cut-off.
The serum level of a certain antigen does not always reflect its
expression and production in tumor tissue, and there are several
factors affecting the serum level. Shedding of tumor-produced
antigen depends on factors like blood and lymphatic vessel inva-
sion, degradation of basal membranes and changes in the surround-
ing matrix. Because most patients were negative for hCG␤ in both
serum and tissue, the proportion of patients with concordant results
was high (78%). However, as measured by the ␬ test, the agree-
ment between serum and tissue expression was only moderate. In
advanced colorectal cancer, Webb et al.24 observed a correlation
between serum and tissue expression of hCG␤, though the sensi-
tivity and specificity were low. However, they used different
antibodies for the serum and tissue measurements, and tissue
expression lacked prognostic significance.
In the present study, the 20 patients who were both histologi-
cally and serologically positive for hCG␤ had a 5-year cancer-
specific cumulative survival of only 21%, while the 5-year survival
of the largest group with no expression in serum or tissue was
62%. hCG␤ serum level was a stronger prognostic factor than
tissue expression in both uni- and multivariate analyses, and the
combination of the 2 was the strongest prognostic factor, after
Dukes’ stage. However, the accuracy of predicting 5-year survival
status with the combined results of serum and tissue detection was
only 1.3% higher than hCG␤ serum levels alone.
In conclusion, our results show that hCG␤ expression in both
tumor tissue and serum and has prognostic significance indepen-
dent of other clinicopathological variables. Positive tumor staining
does not always occur together with elevated serum levels, and the
prognostic accuracy can slightly be increased by combining the
results. Further studies are required to clarify whether hCG␤ can
be used when selecting patients for adjuvant therapy or intensive
follow-up.
22 LUNDIN ET AL.
REFERENCES
1. Pierce JG, Parsons TF. Glycoprotein hormones: structure and func-
tion. Annu Rev Biochem 1981;50:465–95.
2. Braunstein GD, Vaitukaitis JL, Carbone PP, Ross GT. Ectopic pro-
duction of human chorionic gonadotrophin by neoplasms. Ann Intern
Med 1973;78:39 – 45.
3. Goldstein DP, Kosasa TS, Skarim AT. The clinical application of a
specific radioimmunoassay for human chorionic gonadotropin in tro-
phoblastic and non-trophoblastic tumors. Surg Gynecol Obstet 1974;
138:747–51.
4. Acevedo HF, Hartsock RJ. Metastatic phenotype correlates with high
expression of membrane-associated complete beta-human chorionic
gonadotropin in vivo. Cancer 1996;78:2388 –99.
5. Bhalang K, Kafrawy AH, Miles DA. Immunohistochemical study of
the expression of human chorionic gonadotropin-beta in oral squa-
mous cell carcinoma. Cancer 1999;85:757– 62.
6. Kuida CA, Braunstein GD, Shintaku P, Said JW. Human chorionic
gonadotropin expression in lung, breast, and renal carcinomas. Arch
Pathol Lab Med 1988;112:282–5.
7. McManus LM, Naughton MA, Martinez-Hernandez A. Human chorionic
gonadotropin in human neoplastic cells. Cancer Res 1976;36:3476 – 81.
8. Alfthan H, Haglund C, Dabek J, Stenman UH. Concentrations of
human choriogonadotropin, its beta-subunit, and the core fragment of
the beta-subunit in serum and urine of men and nonpregnant women.
Clin Chem 1992;38:1981–7.
9. Alfthan H, Haglund C, Roberts P, Stenman UH. Elevation of free beta
subunit of human choriogonadotropin and core beta fragment of human
choriogonadotropin in the serum and urine of patients with malignant
pancreatic and biliary disease. Cancer Res 1992;52:4628 –33.
10. Blackman MR, Weintraub BD, Rosen SW, Kourides IA, Steinwas-
cher K, Gail MH. Human placental and pituitary glycoprotein hor-
mones and their subunits as tumor markers: a quantitative assessment.
J Natl Cancer Inst 1980;65:81–93.
11. Carpelan-Holmstrom M, Haglund C, Lundin J, Alfthan H, Stenman UH,
Roberts PJ. Independent prognostic value of preoperative serum markers
CA 242, specific tissue polypeptide antigen and human chorionic gona-
dotrophin beta, but not of carcinoembryonic antigen or tissue polypeptide
antigen in colorectal cancer. Br J Cancer 1996;74:925–9.
12. Gailani S, Chu TM, Nussbaum A, Ostrander M, Christoff N. Human
chorionic gonadotrophins (hCG) in non-trophoblastic neoplasms. As-
sessment of abnormalities of hCG and CEA in bronchogenic and
digestive neoplasms. Cancer 1976;38:1684 – 6.
13. Hedstrom J, Grenman R, Ramsay H, Finne P, Lundin J, Haglund C,
et al. Concentration of free hCGbeta subunit in serum as a prognostic
marker for squamous-cell carcinoma of the oral cavity and orophar-
ynx. Int J Cancer 1999;84:525– 8.
14. Marcillac I, Troalen F, Bidart JM, Ghillani P, Ribrag V, Escudier B,
et al. Free human chorionic gonadotropin beta subunit in gonadal and
nongonadal neoplasms. Cancer Res 1992;52:3901–7.
15. Birkenfeld S, Noiman G, Krispin M, Schwartz S, Zakut H. The
incidence and significance of serum hCG and CEA in patients with
gastrointestinal malignant tumors. Eur J Surg Oncol 1989;15:103– 8.
16. Mercer DW, Talamo TS. Multiple markers of malignancy in sera of
patients with colorectal carcinoma: preliminary clinical studies. Clin
Chem 1985;31:1824 – 8.
17. Szymendera JJ, Kaminska JA, Nowacki MP, Szawlowski AW, Gadek
A. The serum levels of human alpha-fetoprotein, AFP, choriogona-
dotropin, hCG, placental lactogen, hPL, and pregnancy-specific beta
1-glycoprotein, SP1, are of no clinical significance in colorectal
carcinoma. Eur J Cancer Clin Oncol 1981;17:1047–52.
18. Buckley CH, Fox H. An immunohistochemical study of the signifi-
cance of hCG secretion by large bowel adenocarcinomata. J Clin
Pathol 1979;32:368 –72.
19. Campo E, Palacin A, Benasco C, Quesada E, Cardesa A. Human
chorionic gonadotropin in colorectal carcinoma. An immunohisto-
chemical study. Cancer 1987;59:1611– 6.
20. Connelly JH, Johnston DA, Bruner JM. The prognostic value of
human chorionic gonadotropin expression in colorectal adenocarcino-
mas. An immunohistochemical study of 102 stage B2 and C2 non-
mucinous adenocarcinomas. Arch Pathol Lab Med 1993;117:824 – 6.
21. Kido A, Mori M, Adachi Y, Yukaya H, Ishida T, Sugimachi K.
Immunohistochemical expression of beta-human chorionic gonado-
tropin in colorectal carcinoma. Surg Today 1996;26:966 –70.
22. Shousha S, Chappell R, Matthews J, Cooke T. Human chorionic
gonadotrophin expression in colorectal adenocarcinoma. Dis Colon
Rectum 1986;29:558 – 60.
23. Skinner JM, Whitehead R. Tumor-associated antigens in polyps and
carcinoma of the human large bowel. Cancer 1981;47:1241–5.
24. Webb A, Scott-Mackie P, Cunningham D, Norman A, Andreyev J,
O’Brien M, et al. The prognostic value of CEA, beta HCG, AFP,
CA125, CA19-9 and C-erb B- 2, beta HCG immunohistochemistry in
advanced colorectal cancer. Ann Oncol 1995;6:581–7.
25. Yamaguchi A, Ishida T, Nishimura G, Kumaki T, Katoh M, Kosaka
T, et al. Human chorionic gonadotropin in colorectal cancer and its
relationship to prognosis. Br J Cancer 1989;60:382– 4.
26. Acevedo HF, Tong JY, Hartsock RJ. Human chorionic gonadotropin-
beta subunit gene expression in cultured human fetal and cancer cells of
different types and origins [see comments]. Cancer 1995;76:1467–75.
27. Fukayama M, Hayashi Y, Koike M. Human chorionic gonadotropin in
the rectosigmoid colon. Immunohistochemical study on unbalanced
distribution of subunits. Am J Pathol 1987;127:83–9.
28. Kahn CR, Rosen SW, Weintraub BD, Fajans SS, Gorden P. Ectopic
production of chorionic gonadotropin and its subunits by islet-cell tu-
mors. A specific marker for malignancy. N Engl J Med 1977;297:565–9.
29. Tormey DC, Waalkes TP, Simon RM. Biological markers in breast
carcinoma. II. Clinical correlations with human chorionic gonadotro-
phin. Cancer 1977;39:2391– 6.
30. Monteiro JC, Barker G, Ferguson KM, Wiltshaw E, Neville AM.
Ectopic production of human chorionic gonadotrophin (hCG) and
human placental lactogen (hPL) by ovarian carcinoma. Eur J Cancer
Clin Oncol 1983;19:173– 8.
31. Monteiro JC, Ferguson KM, McKinna JA, Greening WP, Neville AM.
Ectopic production of human chorionic gonadotrophin-like material
by breast cancer. Cancer 1984;53:957– 62.
32. Morrish DW, Marusyk H, Siy O. Demonstration of specific secretory
granules for human chorionic gonadotropin in placenta. J Histochem
Cytochem 1987;35:93–101.
33. Turnbull RB Jr, Kyle K, Watson FR, Spratt J. Cancer of the colon: the
influence of the no-touch isolation technic on survival rates. Ann Surg
1967;166:420 –7.
34. Jass JR, Sobin LH. World Health Organization international histological
classification of tumors. 2nd edition. New York: Springer-Verlag, 1989.
35. Alfthan H, Schroder J, Fraser R, Koskimies A, Halila H, Stenman UH.
Choriogonadotropin and its beta subunit separated by hydrophobic-inter-
action chromatography and quantified in serum during pregnancy by
time-resolved immunofluorometric assays. Clin Chem 1988;34:1758–62.
36. Cox DR. Regression models and life tables. J R Statist Soc 1972;B34:
187–220.
37. Regelson W. Have we found the “definitive cancer biomarker”? The
diagnostic and therapeutic implications of human chorionic gonado-
tropin-beta expression as a key to malignancy [editorial, comment].
Cancer 1995;76:1299 –301.
38. Gillott DJ, Iles RK, Chard T. The effects of beta-human chorionic
gonadotrophin on the in vitro growth of bladder cancer cell lines. Br J
Cancer 1996;73:323– 6.
39. Melmed S, Braunstein GD. Human chorionic gonadotropin stimulates
proliferation of Nb 2 rat lymphoma cells. J Clin Endocrinol Metab
1983;56:1068 –70.
40. Contractor SF, Davies H. Effect of human chorionic somatomammotro-
phin and human chorionic gonadotrophin on phytohaemagglutinin-in-
duced lymphocyte transformation. Nat New Biol 1973;243:284 – 86.
41. Fabris N, Piantanelli L, Muzzioli M. Differential effect of pregnancy
or gestagens on humoral and cell-mediated immunity. Clin Exp Im-
munol 1977;28:306 –14.
42. Strelkauskas AJ, Wilson BS, Dray D, Dodson M. Inversion of levels
of human T and B cells in early pregnancy. Nature 1975;258:331–32.
43. Campo E, Algaba F, Palacin A, Germa R, Sole-Balcells FJ, Cardesa A.
Placental proteins in high-grade urothelial neoplasms. An immunohisto-
chemical study of human chorionic gonadotropin, human placental lactogen,
and pregnancy-specific beta-1-glycoprotein. Cancer 1989;63:2497–504.
44. Kubosawa H, Nagao K, Kondo Y, Ishige H, Inaba N. Coexistence of
adenocarcinoma and choriocarcinoma in the sigmoid colon. Cancer
1984;54:866 – 68.
45. Park CH, Reid JD. Adenocarcinoma of the colon with choriocarci-
noma in its metastases. Cancer 1980;46:570 –75.