Placenta accreta: an immunohistological study of trophoblast populations

Placenta (1987), 8, 273-282

Placenta Accreta: an Immunohistological Study of Trophoblast Populations

U. EARLa, J. N. BULMER’+ & A. BRIONE$ a Department of Pathology. University of Leeds, Leeds LS2 93T, UK b Department of Obstetrics and Gynaecologv, Universidad de Chile, Santiago, Chile

’ To whom correspondence should be addressed Paper accepted 2.10.19&.5

INTRODUCTION Placenta accreta is an uncommon complication of pregnancy, presenting clinically as ‘the abnormal adherence, either in whole or in part, of the afterbirth to the underlying uterine wall’ (Irving and Hertig, 1937). The histopathological diagnosis is made when chorionic villi implant directly onto, or penetrate through, the myometrium (Fox, 1972). The reported incidence of the condition varies from one case per IOOOO deliveries (Morison, 1978) to one case per 540 deliveries (Sumawong et al, 1966). Much of this discrepancy may be attributable to lack of histological confirmation of the diagnosis and reliance on clinical observations (Hutton, Yang and Bernstein, 1983). Recent immunohistological studies with trophoblast-reactive monoclonal antibodies (mAbs) have highlighted the phenotypic heterogeneity of extravillous trophoblast populations in both normal and ectopic human pregnancy (Bulmer, Billington and Johnson, 1984; Bulmer and Johnson, 1985; Earl, Wells and Bulmer, 1985). Furthermore, the localization of hormonal products within the diverse trophoblast populations may identify a trophoblast cell capable of vascular and tissue invasion both in normal pregnancy and in gestational trophoblastic neoplasia (Gosseye and Fox, 1984; Kurman, Main and Chen, 1984; Kurman et al, 1984). Most authorities consider that the underlying cause of placenta accreta is the total or partial absence of maternal decidua basalis (Irving and Hertig, 1937; Fox, 1972; Hutton, Yang and Bernstein, 1983). An increased capacity for trophoblast invasiveness is not thought to be a factor in the aetiology. Villous and extravillous trophoblast appears morphologically indistinguishable in placenta accreta and in normal term placenta. In the present report, trophoblast populations in placenta accreta have been characterized using antibodies directed against cell membrane antigens, placental hormonal products and low-molecular-weight cytokeratins. The results have been compared with those reported for the normal term placenta. 273

Placenta (19~97)~Vol. 8

a74

MATERIALS

AND METHODS

Tissues Four cases of placenta accreta presenting at the Department of Obstetrics and Gynaecology, University of Chile, Santiago were studied. Placental bed tissues and adherent myometrium obtained from hysterectomy specimens removed following parturition were fixed in IO per cent formalin and routinely embedded in paraffin wax. Six-micrometre serial sections were mounted on clean slides. At least one section from each block was stained with haematoxylin and eosin to allow morphological assessment. Two blocks from each specimen were selected for immunohistochemical studies. Antibodies (Table I) Four murine mAbs were employed. NDOGI was raised by immunization with syncytiotropohoblast membranes from human term placentae (Sunderland, Redman and Stirrat, 1981). NDOGI recognizes a high-molecular-weight proteoglycan complex thought to be a hyaluronic acid (Sunderland et al, 1985). CAM 5.2 is reactive with low-molecular-weight cytokeratins and labels simple epithelia, including normal and neoplastic trophoblast (Makin, Bobrow and Bodmer, 1984). HMFGr was raised against human milk fat globule protein and is directed against an epithelial membrane antigen (Burchell, Durbin and Taylor-Papadimitriou, 1983). Rabbit polyclonal antibodies to human placental lactogen (hPL), human chorionic gonadotrophin (hCG) and pregnancy-specific j,-glycoprotein (SP,) were purchased from Dakopatts A/S, Denmark. A rabbit polyclonal antibody raised by immunization with term human syncytiotrophoblast membranes (rabbit-anti-StMPM) was donated by Professor P. M. Johnson, University of Liverpool. rabbit-anti-mouse immunoglobulin, peroxidaseSwine-anti-rabbit immunoglobulin, conjugated rabbit-anti-mouse immunoglobulin and rabbit peroxidase-antiperoxidase complex were purchased from Dakopatts. Immunoperoxidase techniques Sections were dewaxed and then trypsinized in 0.1 per cent trypsin in Tris-buffered saline (TBS), pH 7.6, for 20 min at 37°C. After a short wash in TBS, endogenous peroxidase was blocked by incubation in I per cent hydrogen peroxidase in methanol for 30 min at room

Table I. Primary antibodies

NDOGI CAM 5.2 HMFGI

Anti-StMPM Anti-hPL Anti-hCG Anti-SP,

Dilution

Reference/source

Antibody

Sunderland, Redman and Stirrat (1981); Sunderland et al (198.5) Makin, Bobrow and Bodmer (1984) (Becton Dickinson. CA. USA) burchell, Durbin and TaylorPapadimitriou (1983) (Unipath, Bedford) Professor P. M. Johnson, UK Dakopatts A/S, Denmark Dakopatts A/S, Denmark Dakopatts A/S, Denmark

CS = culture supematant;

PE = peritoneal exudate.

1:4 cs I:2 cs

r:scm 1:2m 1:

100

I

: 50

I : IM)

PE

h’arl, Bulmer. Briones: Trophoblast in Placenta Accreta

275

temperature. The tissues were then rinsed in TBS and incubated for 15 min in 20 per cent normal swine serum (NSS) in TBS to block non-specific binding. NDOGI and HMFGI were employed in a four-layer peroxidase-antiperoxidase (PAP) technique which has been described previously (Bulmer, Billington and Johnson, 1984). CAM 5.2 was used in a standard indirect immunoperoxidase method with a primary antibody incubation time of I h (Earl, Wells and Bulmer, 1985). Anti-SP,, anti-hCG, anti-hPL and antiStMPM were employed in a standard three-stage PAP technique with overnight incubation of the primary antibody. The colour reaction was developed in a 0.5 mg,/ml solution of j’,j’-diaminobenzidine tetrahydrochloride (Sigma Chemical Company, Poole) in TBS containing 0.04 per cent hydrogen peroxide. Sections were lightly counterstained with Harris’s haematoxylin, dehydrated and mounted in synthetic resin. Negative controls were performed for each method and for each tissue and included omission of the primary antibody with replacement by normal rabbit serum or irrelevant hybridoma supernatant as appropriate.

RESULTS Morphology Chorionic villi were present in all tissue sections and appeared morphologically normal; there was no evidence of trophoblastic hyperplasia. Most villi were separated from myometrium by a variably thick layer of fibrin but occasionally villi were seen to implant directly onto myometrium (Figure I). The myometrium in direct contact with villi showed hyalinization and

Figure I. Chorionicvilli (V) implanting onto myometrium.Decidua is absent. A small amount of fibrin (f) is present. Infiltrating extravillous cytotrophoblast is seen within myometrium (arrows). HE, x 185.

a76

Placenta (1987), Vol. 8

attenuation. A small area of decidua was identified in one section only. Infiltration of superficial myometrium by mononuclear and multinucleate extravillous trophoblast was seen in all specimens either as isolated single cells, or more extensively as cords and clumps of cells. Perivascular aggregates of extravillous trophoblast were a common finding. Endovascular trophoblast was not identified, although in one case perivascular trophoblast had penetrated though the wall of a vessel. Focal aggregates of mononuclear inflammatory cells were present in the myometrium often associated with implanted villi. No endometrial glands were identified. Trophoblast membrane antigens NDOGI and rabbit anti-StMPM showed similar linear reactivity with the apical membrane of villous syncytiotrophoblast. Neither antibody reacted with the villous stroma nor with any extravillous trophoblast population. At the villous implantation sites, NDOGI produced a dense granular extracellular reaction product (Figure 2). This reaction was not seen with antiStMPM. HMFGI was unreactive with all villous trophoblast and with most extravillous trophoblast. However, occasional mononuclear extravillous cytotrophoblast cells were reactive with HMFGI, these cells appearing randomly distributed within the myometrium. Low-molecular-weight cytokeratins In contrast to HMFGI, CAM 5.2 consistently labelled all extravillous trophoblast populations, producing a dense granular cytoplasmic reaction product. The reaction product with villous syncytiotrophoblast was identical in intensity and cytological distribution but was unevenly localized around the villus, some areas showing no reactivity (Figure 3).

Figure 2. Villous syncytiotrophoblast membrane showing linear reactivity with NDOGI (largearrow).Note granular extracellular reactivity at the villous implantation site (small arrows). Haematoxylin counterstain, x 286.

Earl, Bulmer, Brrones: Trophoblast m Placenta Accreta

217

(b) Figure 3. (a) Viilous syncytiotrophoblast labelled with CAM 5.2. Haematoxylin counterstain, extravillous cytotrophoblast labelled with CAM 5.2. Haematoxylin counterstain, x 218.

x 218. (b) Interstitial

Placenta (I&),

278

Vol. 8

Secretory products All three secretory products (hCG, hPL and SP,) were localized in the cytoplasm of villous syncytiotrophoblast, although differences in the density of reaction product were observed. hPL was the most abundant secretory product demonstrated, being present consistently around the entire villous circumference (Figure 4a). In contrast, anti-SP, and anti-hCG both produced a patchy positivity in villous syncytiotrophoblast. SP, was demonstrated more frequently than hCG, but often less than half the villous circumference labelled with anti-SP, . hPL was demonstrated in a moderate proportion (40 to 60 per cent) of the interstitial extravillous trophoblast; both mononuclear and multinucleate trophoblastic cells were reactive with anti-hPL (Figure 4b). Perivascular extravillous trophoblast was labelled with anti-hPL in the same proportion and intensity as interstitial cytotrophoblast. The reaction product with extravillous trophoblast was invariably weaker in intensity than that seen with villous syncytiotrophoblast in the same section. Anti-SP, labelled a small proportion of extravillous trophoblast but less than 5 per cent of the total interstitial trophoblastic cells were SP,-positive. SP,-reactive cells appeared randomly distributed and there was no assoication with hPL localization. hCG was not demonstrated in any extravillous trophoblast population. None of the three antibodies were reactive with villous stroma or maternal tissues.

DISCUSSION Fetal trophoblast invasion of maternal uterine decidua, myometrium and spiral arteries is an essential component of normal human placental development (Boyd and Hamilton, 1970). Infiltrating trophoblast in normal pregnancy takes the form of mononuclear cytotrophoblast which may later fuse to form syncytial giant cells. Although infiltration by fetal trophoblast is particularly prominent in the decidua basalis, large numbers of trophoblastderived multinucleate giant cells are found in the myometrium at term (Pijnenborg et al, 1980). Immunohistochemical studies of trophoblast membrane antigens, placental secretory products and major histocompatibility complex (MHC) antigen expression by trophoblast have further highlighted the heterogeneity of human trophoblast populations (Bulmer, Billington and Johnson, 1984; Kurman, Main and Chen, 1984; Bulmer and Johnson, 1985). In four cases of placenta accreta examined in this study, chorionic villi implanted directly on to myometrium and interstitial cytotrophoblast was present deep within myometrium. Both villous and extravillous populations appeared morphologically normal, some extravillous populations presumably corresponding to the basal plate trophoblast of the normal term placenta. The distribution of human placental lactogen (hPL), human chorionic gonadotrophin (hCG) and pregnancy-specific /I,-glycoprotein (SP,) within the trophoblast in placenta accreta appeared identical to that reported in normal term placenta (Gosseye and Fox, 1984; Kurman, Main and Chen, 1984). Abnormal localization of these secretory products is found in disorders such as choriocarcinoma and placental site trophoblastic tumour, where trophoblast proliferation is abnormal and there is widespread infiltration of maternal tissues (Kurman et al, 1984). Although not directly implicated in the mechanisms of trophoblast proliferation and tissue infiltration, the production of peptide hormones may indicate the degree of cellular differentiation achieved by trophoblast tissue (Hoshina et al, 1985). Additionally, peptide hormones can act as substrate for the epidermal growth factor (EGF) receptor and thereby provide autonomous growth regulation of tissues (Baldwin et al, 1983). EGF receptors have been demonstrated on syncytiotrophoblast, as have receptors for other cellular growth factors such as transferrin and insulin (Richards et al, 1983; Johnson, 1984; Magid et al, 1985).

279

Earl, Bulmer, Brimes: Trophoblart in Placenta Accreta

(b) Figure 4. (a) Localization of hPL in villous syncytiotrophoblast and faintly in extravillous trophoblast (arrowed). Haematoxylin counterstain, x 218. (b) Localization of hPL in extravillous trophoblast within myometrium. Note variable density of reaction product. Haematoxylin counterstain, x 336.

280

Placenta (1987), Vol. 8

The mAb NDOGI appears to be directed against hyaluronic acid and reacts with villous syncytiotrophoblast throughout pregnancy. NGODI also shows a granular extracellular reactivity within the cytotrophoblast cell columns and at the villousdecidual junction in normal and ectopic early human pregnancy (Sunderland et al, 1985). Similar extracellular reactivity is seen at term in very small amounts. It has been suggested that the production of hyaluronic acid may facilitate cytotrophoblast invasion of maternal tissues by opening the extracellular matrix of decidual stroma (Sunderland et al, 1985). Abnormal production of NDOGI has been demonstrated at the invading edges of choriocarcinomas and at a rare ectopic implantation site, the cornua of the uterus (Earl, Bulmer and Wells, 1985; Sunderland et al, 1985). In this present study labelling of the villous-maternal junction by NDOGI was found to be more prominant than expected, showing that synthesis of hyaluronic acid by the residual cytotrophoblast shell is retained in placenta accreta, even at term. This capacity appears to arise in response to implantation deep within the myometrim. It suggests that trophoblast can respond to, and regulate its own growth within, an unsuitable intramyometrial location. It is not indicative of an inherent abnormality of the trophoblast. HMFGI and CAM 5.2 produced patterns of reactivity with trophoblast populations in placenta accreta similar to those reported in normal term placenta. HMFGI has been shown to label scattered extravillous trophoblast cells within maternal decidua and myometrium in normal third-trimester basal plate (Bulmer and Johnson, 1985). The immunohistological localization of low-molecular-weight cytokeratins, membrane antigens and secretory products in trophoblast populations in placenta accreta appeared to be essentially identical to that reported in the normal third-trimester placenta. The invasive capacity of trophoblast was not excessive and appeared related to the depth of implantation. However, decidua was absent at the implantation site in all the cases studied with the exception of a small area of decidua in one tissue section, and the importance of decidua in the regulation of blastocyst implantation and placental growth must be considered. Soluble suppressor factors have been demonstrated in explants of first-trimester human decidua, and isolated suspensions of de&dual lymphocytes suppress concanavalin A-induced proliferation of peripheral blood lymphocytes (Golander et al, 1981; Daya et al, 1985). Successful development of the semi-allogeneic fetoplacental unit appears to depend on the interactions of maternal local suppressor mechanisms, the sharing of common surface antigens by maternal lymphocytes and trophoblast (TLX), and the absence of MHC antigens on villous trophoblast (Clark, 1985). However, extravillous trophoblast populations in both normal uterine pregnancy and at ectopic implantation sites express some form of class I MHC antigen (Redman et al, 1984; Earl, Wells and Bulmer, 1985). Recognition of fetal MHC antigens or trophoblastspecific antigens carried on interstitial extravillous trophoblast by maternal immunocompetent cells in decidua may regulate the local tissue infiltration by trophoblast in early implantation. Studies of mouse decidua have demonstrated a population of immune effector cells with natural killer cell (NK)activity, which show maximal lytic activity of target cells during the first two weeks of gestation; it has been suggested that these NK cells may limit early trophoblast invasion by direct lysis of trophoblastic cells (Croy et al, 1985; Gambel et al, 1985). No similar cells have been described in human pregnancy. The absence of decidua, through whatever mechanism, appears to play a fundamental role in the aetiology of placenta accreta, but studies of decidual cell populations are limited by lack both of fresh unfixed material and of placenta accreta tissues from earlier in gestation. Placental tissues provide a fascinating model of cellular growth, differentiation and tissue invasion. Studies of immune effector cells and their soluble mediators within decidua will enhance understanding of the controlling mechanisms involved in both normal and abnormal pregnancy.

Earl. Bulmer, Briones: Trophoblast in Placenta Accreta

281

SUMMARY Trophoblast populations in four cases of placenta accreta were characterized using antibodies directed against cell membrane antigens, placental hormonal products and low-molecularweight cytokeratins in standard immunoperoxidase techniques. The results obtained with antibody to syncytiotrophoblast membrane (rabbit anti-StMPM), antibody to an epithelial membrane antigen (HMFGI) and a cytokeratin marker (CAM 5.2) appeared identical to those reported for normal term placental tissues. Similarly the localization of human placental lactogen (hPL), human chorionic gonadotrophin (hCG) and pregnancy-specific p,-glycoprotein (SP,) within trophoblast populations in placenta accreta was identical to their reported distribution in term placenta. However, increased reactivity at the villous-maternal junction was demonstrated with NDOGI, an antibody raised against term syncytiotrophoblast membrane and directed against hyaluronic acid. NDOGI reactivity at this site is normally maximal during early placental development and is virtually absent in the third trimester. The results suggest that placenta accreta does not arise through excessive trophoblast invasiveness or proliferation and the absence of decidua is of more importance in the pathogenesis. Trophoblast may regulate its development at an unfavourable intramyometrial implantation site by the production of hyaluronic acid.

ACKNOWLEDGEMENTS We thank Dr A. Dabancens from the Service of Cytopathology and Cancer Control, School of Medicine, University of Chile for providing tissue blocks and Dr. C. A. Sunderland and Professor P. M. Johnson for their gifts of monoclonal antibodies.

REFERENCES Baldwin, G. S., Grcgo, B., He-am,M. T. W. et al (1983) Phosphorylation of human growth hormone by the epidermal growth factor-stimulated tyrosine kinase. Proceedings of rhe National Academy of Sciences of rhe USA, 80,527~5280. Boyd, J. D. & Hamilton, W. J. (1970) The Human Placenta. Cambridge, MA: Heffer. 365 pp. Buhner, J. N.&Johnson, P. M. (1985) Antigen expression by trophoblast populations in the human placenta and their possible immunok&a~ relevance. Placcnra, 6, 127-140. Buhner, J. N., Billington, W. D. & Johnson, P. M. (1984) Immunohistologicidentification of trophoblast populations in human pregnancy with the use of monocloml antibodies. American 3oburnaIof Obstefrw and Gynecology, 148, 19-26. Burchell, J., Durbin, H. & Taylor-Papadimitriou, J. (1983)Complexityof expressionof antigen determinants recognizedby monoclonal antibodies HMFGI and HMFG2 in normal and malignant human mammary epithelial cells. 30~rn.51of Immunology, 131, 508-513. Clark, D. A. (1985) Mawrno-fetal relations. Immunology Letters, 9, 239-247. Croy, B. A., Gambel, P., Rossan t, J. & Wegman, T. G. (1985) Characterisation of mlirine decidual natural killer (NK) cells and their relevance to the success of pregnancy. Cellular Immunology, 93, 315-326. Daya, S., Clark, D. A., Devlin, C. et al (1985) Supressor cells in human desidua. AmericanJournal of Obsrerrics and Gynecology, ISI, 267-270. Earl, U. M., Bulmer, J. N. & Wells, M. (1985) Immunohistological characterisation of antigen expression by trophobkw populations in ectopic comual implantation. Case Report. BririshJournal of Obstetrics and Gynaecology, 929843-846. Earl, U., Wells, M. & Bulmer, J. N. (1985) The expression of major histocompatibility complex antigens by trophoblast in ectopic Nbd pnzgnancy.3ournal of Reproductive Immunology, 6, 127-140. Earl, U., Wells, M. & Bulmer, J. N. (1986) Immunohistochemical characterisation of trophoblast antigens and secretory products in ectopic Nbpl pregnancy. Interna~ional3ournaI of Gynecological Pathology, 5, 132-142. Fox, H. (1972) PkXent? accreta, 1g.#5-1g6g. Obsretrical and Gynaecologicai Survey, 27, 475-490. Gambel, P., Croy, B. A., Moore, W. D. et al (1985)Characterisation of immune effector cells present in early murine decidua. Cellular Immunology, 93,303-314. Golander, G., Zakuth, V., Shechter,Y. & Spirer, Z. (1981)Suppression of lymphocyte reactivity in vitro by a soluble factor secreted by explants of human decidua. European J’ournal of Immunology, I I, 84@5 I.

282

Placenta (1987), Vol. 8

comparison of the secretory capacity of villous and extravillous Gosseye, S. & Fox, H. (1984) An immunohistokqiul trophoblast in the human placenta. Placenta, 5, 349-370. Hoahina, M., Boothlay, M., Husaa, R. et al (1985) Linkage of human chorionic gonadotrophin and placental lactogen biosynthesis to trophoblast differentiation and tumorigenesis. Placenta, 6, 163-172. Hutton, L., Yang, S. S. & Bernstein, J. (1983) Placenta accreta-a 26-year clinicopathological review (19561981). New York Stare Journal of Medicine, 83, 857-866. Irving, F. C. & He&g, A. T. (1937) A study of placenta accreta. Surgery, GynaecologV and Obrtetncs, 64, 178-200. Johnson, P. M. (1984) Immunobiology of human trophoblast. In Immunological Aspects of Reproducrion in Mammals (Ed.) Crighton, D. B. pp. 109-131. London: Butterworth Press. Kurman, R. J., Main, C. S. & Chen. C-H. (1984) Intermediate trophoblast: a distinctive form of trophoblast with specific morphological and functional features. Placenra, 5, 34~370. localisation of placental lactogen and KUITIIWI,R. J., Young, R. H., Norris, H. I. et al (1984) Immunocytochemical chorionic gonadotrophin in the normal placenta and trophoblastic turnours, with emphasis on intermediate trophoblast and the placental site trophoblastic tumour. Internarional~ownal of Gynaeco/ogical Pathology,3,101-121. Magid, M., Nanney, L. B., Stotacheck, C. M. & King, L. E. (1985) Epidermal growth factor binding and receptor distribution in term human placenta. Placenta, 6, 519-526. Makin. C. A.. Bobrow. L. G. & Bodmer. W. F. (1984) Monoclonal antibodies to cvtokeratin for use in routine histopathol&. 3oumch of Clinical Pathol&y, 37, 97;<83. Morison, J. E. (1978) Placenta accreta-a clinicopathologic review of 67 cases. In Obstetrics and Gynecology Annual (Ed.) Wynn, R. M., Volume 7, pp. 107-123. New York: Appleton-Century-Crofts. Pijnenborg, R., Dixon, G., Robertson, W. B. & Broaena, I. (1980) Trophoblastic invasion of human decidua from 8-18 weeks of pregnancy. Placenta, I, 3-19. Redman, C. W. G., McMichael, A. J., Stirrat, G. M. et al (1984) Class I major histocompatibility complex antigens on human extra-villous trophoblast. Immunology, 52, 457-468. Richards, R. C., Beardmore, J. M., Brown, P. J. et al (1983) Epidermal growth factor receptors in isolated placental syncytiotrophoblast plasma membrane. Placenta, 4, 133-138. Sumawong, V., Nandastu, A., Thanapath, S. & Budthimedhee, V. (1966) Placenta accreta: a review of the literature and a summary of ten cases. Obstetricsand Gynecology, 27, 511-516. Sunderland, C. A., Redman, C. W. G. & Stirrat, G. M. (1981) Monoclonal antibodies to human syncytiotrophoblast. Immunology, 43, 541-546. Sunderland. C. A.. Bulmer. J. N.. Luscombe. M. et al (1985) _ _. Immunohistological and biochemical evidence for a role for hialuronic acid in~be &wth and d&elopment of the placenta. Jo&al of Reproductive Immunology, 8, 197-212.