Download Immunoreactive Secretory Component of IgA in Human Tissues and

Immunoreactive Secretory Component of IgA in Human
Tissues and Tumors
JOHN J. BROOKS, M.D. AND CAROLYN S. ERNST, M.D.
Secretory component (SC) of IgA is known to be produced by
many glandular epithelial cells. To assess the usefulness of
this antigen as a glandular tumor differentiation marker, the
authors evaluated its normal body distribution and tested a
variety of glandular neoplasms. The known normal distribution
of SC was confirmed and extended to include the prostate.
Immunoreactive SC was detected only in epithelial and glandular
tumors. Many types of well-differentiated adenocarcinomas,
(ovary, prostate, small bowel, pancreas, stomach, biliary) however, contained little or no immunoreactive SC. Therefore, the
authors conclude that immunoreactive SC cannot be used as a
general differentiation marker in tumors of glandular derivation.
A high frequency of antigenic expression was found in adenocarcinomas of the lung (4/10), breast (5/10), and colon (24/
27). No correlation between expression and tumor differentiation
was observed in pulmonary or mammary tumors. Only in
colonic neoplasia was such a relationship detected. SC expression in colonic tumors was not related to mucin content but
was associated with the presence of a visible brush border.
(Key words: Secretory component in human tumors) Am J
Clin Pathol 1984; 82: 660-665
Department of Pathology and Laboratory Medicine,
Hospital of the University of Pennsylvania,
Philadelphia, Pennsylvania
secretory immunoglobulin IgA found in external secretions.3'4,5,12 SC already has been identified in a number
of normal epithelial tissues 2 '" 1 3 1 6 ' 1 7 1 9 and in several
forms of epithelial malignancy. 8101416
We previously have shown that immunoreactive SC
is present in mesothelium and mesotheliomas7 and thus
cannot be used as a cellular marker to distinguish
between adenocarcinomas and mesothelioma. In this
study, we confirm and extend the bodily distribution of
this epithelial antigen; we also examine its expression in
a variety of adenocarcinomas and attempt to correlate
its expression with difTerentiation.
Materials and Methods
ATTEMPTS TO PREDICT the biologic behavior of
human epithelial malignancies, for the most part, have
been based upon histologic features defining "grades of
differentiation." With the advent of immunohistochemical technics, pathologists now can approach the question
of tumor biology in a different way. One now can ask:
to what extent do malignant epithelial cells recapitulate
the biochemical composition and function of the corresponding normal cells from which they arose? Further,
to what extent does this relate to metastatic potential?
The expression of blood group isoantigens in urothelial
tumors, apparently inversely related to biologic aggressiveness, is an important example of the application of
immunohistochemistry to predict neoplastic potential.1-6
In this study, our intention was to explore the expression of another epithelial antigen, the Secretory Component of IgA (SC), in a variety of human glandular
tumors. SC is an antigenically distinct portion of the
Received February 21, 1984; received revised manuscript and accepted
for publication April 9, 1984.
Dr. Brooks is the recipient of an American Cancer Society Junior
Faculty Clinical Fellowship.
Dr. Ernst's work was performed in part under an American Cancer
Society Clinical Fellowship.
Address for reprint requests to Dr. Ernst: Department of Pathology
and Laboratory Medicine, Hospital of the University of Pennsylvania,
34th and Spruce Streets, Philadelphia, Pennsylvania 19104.
From the Surgical Pathology files of the Hospital of
the University of Pennsylvania, examples of normal
tissues as well as adenocarcinomas of the lung, breast,
ovary, pancreas, prostate, liver, small bowel, stomach,
thyroid, and colon were collected. Some nonepithelial
or nonglandular tumors also were selected for testing.
Blocks were cut at 5 jim, and unstained slides were
prepared. The original tumors were reexamined, histologically classified, and graded. Patient names at no time
were identified, and the procedures followed were in
accord with the ethical standards of our institution.
Immunodiffusion and Electrophoresis
Specificity of antiserum was demonstrated using immunodiffusion and immunoelectrophoresis. A single
line of reactivity was demonstrated with rabbit antihuman free secretory component (FSC) (Behring Diagnostics) against purified FSC (courtesy of Dr. Michael
Lamm, New York University) and colostrum. No reactivity was demonstrated with anti-FSC against serum
IgA or normal human serum.
Immunohistochemistry
660
The immunoperoxidase procedure was performed according to Sternberger18 using the chromagen reaction
661
SECRETORY COMPONENT IN HUMAN TUMORS
Vol. 82 • No. 6
of Graham and Karnovsky.9 Briefly, paraffin-embedded
formalin or Bouin's fixed tissues were deparaffinized
and incubated in methanol with 0.3% hydrogen peroxide
to eliminate endogenous peroxidase activity. After preincubation with 20% normal goat serum, rabbit antihuman FSC (Behring Diagnostics, 1/1,000 dilution) was
applied and incubated overnight at 4 °C. The goat antirabbit IgG (Cappel, 1/60 dilution) and rabbit peroxidaseantiperoxidase (Cappel, 1/100 dilution) were applied
sequentially. Following reaction with the chromagen
diaminobenzidine (Sigma), the slides were counterstained
with hematoxylin and mounted. Serum controls consisted of (1) substitution of anti-FSC with normal rabbit
serum; (2) absorption of anti-FSC with FSC; and (3)
absorption of anti-FSC with albumin. Absorption of the
anti-serum with purified FSC or with bovine serum
albumin was performed as follows: to a 1/1,000 dilution
of anti-serum was added a precalculated amount of
antigen (in excess of two moles antigen per mole specific
antibody). The mixture was incubated in rotation (two
hours, room temperature), centrifuged (110,000 g 4 °C,
30 minutes), and the supernatant stored at 4 °C until
use. Positive tissue controls also were included in each
experiment. Importantly, no immunoreactivity was observed after absorption of the antiserum with free secretory component; absorption with albumin did not affect
antibody staining.
The effects of fixation on the expression of immunoreactive secretory component had been tested prior to
the main study using both normal tissue (breast and
lung) and colon carcinomafixedin either neutral buffered
formalin or Bouin's solution. Immunoreactive SC was
easily detectable in both fixatives in normal and cancerous tissues. Neither fixative appeared superior over the
other. Concerning colon carcinoma, several cases were
tested in both fixatives (including one negative case),
and the results were identical.
Results
Normal Tissues
A wide variety of normal tissues were tested for the
presence of immunoreactive SC (Table 1). Among the
positive tissues, breast, colon, and small bowel demonstrated the strongest reaction in both the intensity and
in the frequency of positive cells. For example, essentially
all nongoblet cells in the small bowel and colon contained
positive staining. Although there was diffuse cytoplasmic
staining, the reaction product was frequently most
prominent on the apical surface (Fig. 1). Goblet cells
were uniformly negative. Strong reactivity was also
observed in normal breast lobules, ducts, and secretions.
Other positive tissues demonstrated either less intense
Table 1. Immunoreactive Secretory Component
in Normal Tissues
Positive
Negative
Colon
Small bowel
Breast
Salivary gland
Bronchial secretions
Prostate
Gallbladder
Renal Tubules
Pancreatic ducts
Liver
Thyroid
Pancreatic acini
Skin
Muscle
Testis
Cervix
Esophagus
or only focal staining. For example, only a subpopulation
of salivary ductal cells were positive (Fig. 2). Here, and
in gallbladder epithelium, prostate glands, and occasional
renal tubules, staining intensity was also variable. While
bronchial glandular secretions stained positively, the
epithelium itself appeared negative. Tissues that failed
to stain were largely nonsecretory or nonepithelial in
nature.
Benign Tumors
Some benign tumors arising in positive normal tissue
types also were subjected to analysis. In the breast, the
epithelial cells in fibroadenomas and fibrocytic disease
contained immunoreactive SC, but the most intense
reaction was observed in the lactating adenoma (Fig. 3).
Here, the great majority of mammary acinar cells were
positive. Colonic adenomas ("adenomatous polyps")
mimicked their normal counterpart by exhibiting diffuse,
cytoplasmic staining in essentially all cells. For the most
part, surface concentration of staining was also observed
as it was in the normal colon.
Malignant Tumors
Over 80 malignant tumors were tested for the presence
of SC. As expected, the nonepithelial tumors such as
rhabdomyosarcoma (0/1) and hemangiopericytoma (0/
2) lacked expression of this antigen. Further, neoplasms
arising from negative normal tissues were themselves
negative: cutaneous melanoma (0/2), esophageal squamous carcinoma (0/2), and thyroid carcinoma (0/2). In
fact, all nonglandular neoplasms including squamous
(0/2) and oat-cell carcinoma (0/1) of the lung failed to
show immunoreactivity.
Because reactivity appeared linked to gland-forming
tumors, these were investigated in greater numbers. Of
all such tumors of various primary sites, only a fraction
(36/80 or 45%) displayed positivity with anti-SC antiserum (Table 2).
Immunoreactivity of SC in human adenocarcinomas
was very much site dependant. No detectable staining
662
BROOKS AND ERNST
AJ.C.P. • December 1984
FIG. 1 (upper, left). Normal colon. Immunoreactive SC (dark)
was found throughout the cytoplasm but was concentrated at the
apical or luminal surface. Hematoxylin counterstain (X400).
FIG. 2 (upper, right). Normal salivary gland. Scattered ductal
epithelial cells (dark) display variable staining intensity. Hematoxylin
counterstain (X400).
FIG. 3 (lower). Lactating adenoma of the breast. Note the
immunoreactivity both in glandular secretions and in many epithelial
cells (dark). Hematoxylin counterstain (X400).
was observed in gland-forming tumors of the ovary,
pancreas, stomach, or endometrium. Despite prominent
SC reactivity in corresponding normal tissues, only rare
Table 2. Immunoreactive Secretory Component
in Adenocarcinomas
Site
No. Positive/Total
Percent Positive
Colon
Lung
Breast
Prostate
Biliary
Small bowel
Ovary
Pancreas
Stomach
Endometrium
24/27
4/10
5/10
1/5
1/5
1/5
0/5
0/5
0/5
0/1
89
40
50
20
20
20
0
0
0
0
cases of prostatic, biliary, or small intestinal adenocarcinomas were positive. The majority of the positive
tumors were located in either the breast, lung, or colorectal area.
Ten breast carcinomas were evaluated, and SC was
present in half of the cases (Table 2). Four of seven (4/
7) infiltrating duct carcinomas were positive. Here, focal
discrete islands of tumor were positive; the stroma and
most of the infiltrating cells were negative. One case of
adenoid-cystic carcinoma also exhibited reactivity both
in tumor cells and in the surrounding edematous stroma.
In contrast to these positive ductal lesions (5/8 or 62%),
two invasive lobular carcinomas were negative.
Predominantly the bronchoalveolar type of pulmonary
adenocarcinoma was tested to evaluate the frequency of
positivity in this commonly well-differentiated tumor.
Vol. 82 • No. 6
SECRETORY COMPONENT IN HUMAN TUMORS
663
ft.
» *
•"
% ^
».
FIG. 4. Colonic adenocarcinoma. In some cases, only surface or brush border immunoreactivity was observed (A) (left).
Still other cases exhibited only cytoplasmic staining (B) (right)- Hematoxylin counterstain (X400).
However, only one-third of these tumors were positive.
In the positive cases, the tumor cell cytoplasm stained
with variable intensity but diffusely. Intraluminal secretions were also immunoreactive. The one poorly differentiated lung tumor tested showed much less positivity
with only scattered focal cells staining.
Colorectal Tumors: By far the highest frequency of
positivity was noted in tumors of this type (24/27, 89%).
Two types of staining were observed: surface staining
concentrated at the luminal edge of the cells and diffuse
cytoplasmic staining (Fig. 4). While these two patterns
could be seen within the same tumor, many tumors
segregated into one or the other pattern. Therefore, we
divided the cases into those with predominantly surface
or cytoplasmic staining patterns and attempted to correlate this with (1) the histologic grade of differentiation;
(2) the amount of mucin production; and (3) the presence
of a visible brush border on hematoxylin and eosin
stained sections. Immunohistochemical evaluation was
performed independently of the other hematoxylin and
eosin characteristics.
The grade of differentiation, based primarily on architecture, was defined as follows: grade I—composed
of large well-defined glands, with tall columnar cells
frequently adenomatous in appearance; grade II—composed of smaller glands with more cuboidal cells; grade
III—composed of poorly formed and rare small glands;
grade IV—composed of a sheet-like arrangement of cells
with virtually no gland formation. The degree of mucin
production was scored as follows: 1+ = only rare and
intracellular mucin, little or no goblet-like cells present;
2+ = both intracellular and extracellular mucin, gobletlike cells present; 3+ = pools of mucin visible as well
as numerous goblet/mucin cells; 4+ = colloid carcinoma
(pools of extracellular mucin with scanty cellular elements). The presence of a brush border on the cells of
a tumor, seen as increased eosinophilia at the luminal
surface, arbitrarily was designated as absent, present on
scattered cells, or prominant throughout the tumor.
The relationship between the pattern of staining for
immunoreactive SC and tumor grade, mucin production,
and brush border differentiation is given in Table 3. As
can be seen, immunoreactive SC was noted to be related
inversely to the histologic grade of differentiation. Thus,
there appeared to be a progressive loss of SC reactivity
with worsening grade. This relationship was more dramatic for the surface staining pattern as compared with
the cytoplasmic staining pattern. For example, most
664
BROOKS AND ERNST
Table 3. Immunoreactive SC and Differentiation
in Colorectal Carcinoma
Immunoreactive Pattern*
Histologic Feature
A. Histologic grade
1
II
III
IV
Mucin content
1+
2+
3+
4+
Brush border
Absent
Scatttered
Prominent
Surface
Cytoplasmic
8/12
3/9
0/4
0/2
(67)
(33)
(0)
(0)
12/12(100)
8/9 (89)
3/4 (75)
1/2 (50)
2/6
4/4
5/14
0/3
(33)
(100)
(36)
(0)
4/6 (67)
4/4 (100)
14/14 (100)
2/3 (67)
3/10(30)
1/8 (13)
7/9 (78)
8/10 (80)
7/8 (88)
9/9 (100)
• Number positive over number tested: percent positive cases in parentheses.
grade I but fewer grade II and no grade HI or IV tumors
exhibited this surface staining pattern. This may reflect
the retention of biochemically functional microvilli in
better differentiated lesions. The detection of SC did not
correlate, however, with mucin production. Interestingly,
the presence of a visible brush border did seem to
predict to some degree the surface SC staining pattern.
Discussion
In this study, our aims were to (1) confirm the normal
body distribution of immunoreactive Secretory Component; (2) determine whether or not this antigen could
be used as a general differentiation marker for tumors
of glandular derivation; and (3) ascertain the relationship
between the expression of this antigen and tumor differentiation.
The normal body distribution of SC, which has been
reported previously with the use of immunofluorescent
m e t h o d s 2 " ' 3 1 6 1 7 1 9 was confirmed here by the immunoperoxidase technic. Even the focal occurrence of SC
in renal tubules seen here had been observed before by
Tourville and associates.19 The localization of SC in
pancreatic acini, also reported by Tourville and associates
el a/.19 was not demonstrated convincingly in our experiments. Importantly, we found the same type of
surface and cytoplasmic staining patterns in the small
and large bowel mucosa seen by others.14"16 However,
to our knowledge this is the first demonstration of SC
in prostatic glandular cells, a site not tested previously
by other investigators.
Concerning the investigation of SC in human tumors,
only colonic 1416 and mammary tumors 10 have been
examined thus far; these studies will be compared with
our own findings below. Since data concerning SC in
other tumors was lacking, we sampled many types of
A.J.C.P. • December 1984
adenocarcinomas to test the hypothesis that SC might
be a widespread marker present in many glandular
tumors enabling correlation with histologic grade of
differentiation as a general phenomenon. While it was
true that we found immunoreactive SC to be specific
for tumors of glandular derivation (i.e., it was not
observed in nonglandular malignancies), we were disappointed in its lack of sensitivity. Many glandular
tumor types either lacked expression or displayed a low
frequency of expression of this antigen, despite its presence within the corresponding normal organ. Our samples, though small, of different organ tumors deliberately
were biased toward well-differentiated tumors in which
expression would be most likely. These data appear to
exclude the use of SC as a general differentiation marker
in all glandular tumors.
In addition, immunoreactive SC did not seem useful
as a differentiation marker in breast tumors. Certainly,
our data confirm the presence of SC in normal breast
and in benign breast tumors, as demonstrated initially
by Harris and colleagues.10 However, no clear pattern
relating SC to mammary tumor grade was discerned
either by us or by Harris and colleagues.10 In that study,
all 20 breast carcinomas were positive, regardless of
subtype or tumor grade. It should be mentioned that
that study tested only ductal derived tumors; no lobular
carcinomas, negative in our study, were included. This
may account, in part, for the differences in frequency
of SC immunoreactivity between the two studies. Interestingly enough, the serum of many patients with breast
cancer contained elevated SC as determined by Puleo
and associates.15
The same situation apparently applies to pulmonary
adenocarcinomas. Again, the well-differentiated tumor,
bronchoalveolar carcinoma, purposely was selected for
study. Nonetheless, only one-third of these cases were
positive with anti-SC antiserum. Further, the positivity
in these three cases was quite variable, with large areas
of tumor being negative. The only other information
available on SC in lung tumors is from two reports, one
documenting its presence in two cases10 and one describing large scale SC production by an adenocarcinoma. 8
In that instance,8 it is interesting to note that the figure
is consistent with a bronchoalveolar carcinoma.
Thus, colonic neoplasia appeared to be the only
example of a clear relationship between SC expression
and tumor differentiation. Our results in this regard are
similar to those of Poger and Lamm. 14 First of all, we
confirmed their data on tubular adenomas in which the
vast majority of cells in all cases expressed SC. Secondly,
we found, as they did, a significant percentage of colonic
adenocarcinomas also expressed SC. Thirdly, we also
were able to demonstrate the progressive loss of this
antigen with decreasing tumor differentiation initially
SECRETORY COMPONENT IN HUMAN TUMORS
Vol. 82 • No. 6
described by them. Our data differ from theirs because
of different methods of scoring. For example, although
they state that only one-half of the 36 cases tested were
positive, they included as negative cases showing 0-20%
tumor cells positive. In contrast, we found 89% of cases
positive and scored any immunoreactivity as a positive
case. We did observe (as they noted) fewer cells positive
in the more poorly differentiated tumors. Although
Poger and Lamm described a direct relationship between
immunoreactive SC and mucin production,14 we were
not able to demonstrate this.
One interesting observation we made was the ability
of a morphologic marker to predict the presence of an
antigenic differentiation marker in colonic neoplasia.
Whenever a prominent brush border was visualized in
a tumor on hematoxylin and eosin section, it contained
immunoreactive secretory component, usually in the
surface staining pattern localized to the region of the
brush border. This relationship seemed logical, since
numerous terminal microvilli in the form of a brush
border themselves would be a morphologic indicator of
differentiation. We apparently have shown that such a
border, when present, is biochemically functional in
addition.
We conclude that the SC antigen is found in essentially
all glandular organs, that it can not be used as a general
marker to evaluate differentiation in all glandular tumors
(as was hoped), and that its presence specifically in
colonic neoplasia is related to tumor differentiation.
References
1. Bergman S, Javadpour N: The cell surface antigen A, B, or O (H)
as an indicator of malignant potential in Stage A bladder
carcinoma. J Urol 1978; 119:49-51
2. Brandtzaeg P: Mucosal and glandular distribution of immunoglobulin Components-immunohistochemistry with a cold ethanolfixation technique. Immunology (1974) 26:1101-1114
665
3. Brandtzaeg P: Intestinal secretion of IgA and IgM: A hypothetical
model. Ciba Foundation Symposium 1977; 46:77-113
4. Brown W, Isobe K, Nakane P, Pacini B: Studies on translocation
of immunoglobulin across intestinal epithelium. IV. Evidence
for binding of IgA and IgM to secretory component in intestinal
epithelium. Gastroenterology 1977; 73:1333-1339
5. Clancy R, Bienenstock J: Secretion immunoglobulins. Clin Gastroenterol 1976; 5:229-248
6. Decenzo J, Howard P, Irish C: Antigenic deletion and prognosis
of patients with Stage A transitional cell bladder carcinoma. J
Urol 1975; 114:874-878
7. Ernst C, Brooks J: Immunoperoxidase localization of secretory
component in reactive mesothelium and mesothelioma. J Histochem Cytochem 1981; 29:1102-1104
8. Gotoh T, Takishita Y, Doi H, Tsubura E: Secretory-component
producing lung cancer with hypergamma globulinemia of secretory IgA. Cancer 1981; 48:1776-1782
9. Graham R, Karnovsky M: The early stages of absorption of
injected horseradish peroxidase in the proximal tubules of
mouse kidney: Ultrastructural cytochemistry by a new technique.
J Histochem Cytochem 1966; 14:291-302
10. Harris J, Caleb M, South M: Secretory component in human
mammary carcinoma. Can Res 1975; 35:1861-1864
11. Hurlimann J, Dayal R, Gloor E: Immunoglobulin and secretory
component in endometrium and cervix. Virchows Arch [Pathol
Anat] 1978;377:211-215
12. Lamm M: Cellular aspects of immunoglobulin A. Adv Immunol
1976; 22:223-285
13. Poger M, Lamm M: Localization of free and bound secretory
component in human intestinal epithelial cells. A model for
assembly of secretory IgA. J Exp Med 1974; 139:629-641
14. Poger M, Hirsch B, Lamm M: Synthesis of secretory component
by colonic neoplasms. Am J Pathol 1976; 82:327-338
15. Puleo E, Haagensen D, Dawson J, Gall S: Study of secretory
component in patients with metastatic breast cancer. Am J
Obstet Gynecol 1979; .134:899-903
16. Rogrum T, Brandtzaeg P, Oriasaeler H, Fausa O: Immunohistochemistry of epithelial cell markers in normal and pathological
colon mucosa. Histochemistry 1980; 67:7-21
17. Rossen R, Morgan C, Hsu K, Butler W, Rose H: Localization of
lis external secretory IgA by immunoflourescence in tissues
lining the oral and respiratory passages in man. J Immunol
1968; 100:706-717
18. Sternberger L: Immunocytochemistry. Second edition. New York,
John Wiley and Sons, 1979
19. Tourville D, Adler R, Bienenstock J, Tomasi T: The human
secretory immunoglobulin systems—immunohistological localization of A, secretory piece, and lactoferrin in normal human
tissues. J Exp Med 1969; 129:411-428