The changing role of pathology in breast cancer diagnosis and treatment.

Pathological examination has been the gold standard for diagnosis in cancer and its role has also included the elucidation of etiology, pathogenesis, clinicopathological correlation, and prognostication. The advent of newer technologies and the realization that breast cancer is heterogeneous has shifted the focus to prognostication, with increased attention being paid to the identification of morphological features and immunohistochemical markers of prognostic relevance. However, despite the massive efforts invested in the identification of immunohistochemical biomarkers in breast cancer the majority have not proven to be of value in multivariate analyses and only estrogen receptor, progesterone receptor, and Her2/neu expression have remained essential components of pathological examination. These 3 markers were initially employed for prognostication but their role in treatment also rendered them of predictive value. Newer molecular methods, especially high-throughput technologies, have shown that even morphologically similar subtypes of breast cancer can show molecular heterogeneity; moreover, infiltrating ductal carcinoma can be separated into at least 4 molecular subtypes designated luminal (ER+, PR+, and Her2/neu-), Her2 overexpressing (ER-, PR-, and Her2/neu+), basal-like (ER-, PR-, Her2/neu-, and CK5/6+, EGFR+), and normal breast-like (ER-, PR-, and Her2/neu-), each with different clinical outcomes. The importance of proliferative gene expression in these subtypes has been demonstrated and surrogate immunohistochemical markers include ER, PR, Her2/neu, and Ki67 for the more expensive molecular tests. Molecular technologies, importantly, have not only provided further insights into the heterogeneity of breast cancer but have also opened new avenues for treatment through the identification of signaling molecules important in the proliferation and survival of the neoplastic cells. The treatment of cancer thus shifts from the conventional approach of 'one size fits all' to one of personalized treatment tailored to the specific characteristics of the tumor. Pathologists continue to play their traditional role in diagnosis but, as purveyors of the excised tissue, pathologists now have the additional role of identifying biomarkers responsive to therapeutic manipulation, thus playing an inextricable role as diagnostic oncologists in the management of breast cancer.

Standardization in immunohistology.

The rapid acceptance of immunohistology as an invaluable adjunct to morphologic diagnosis has been possible because of the development of new and more sensitive antibodies and detection systems that allow its application to formalin-fixed, paraffin-embedded tissue (FFPT). More importantly, antigen-retrieval techniques have resulted in some degree of consistency allowing immunohistology to be used reliably as a diagnostic tool. The advent of prognostic and predictive biomarkers, and the desire for individualized therapy has resulted in mounting pressure to employ the immunohistological assay in a quantitative manner. While it was not a major issue when the technique was employed in a qualitative manner, the numerous variables in the preanalytical and analytical phases of the test procedure that influence the immunoexpression of proteins in FFPT become critical to standardization. Tissue fixation is pivotal to antigen preservation but exposure to fixative prior to accessioning by the laboratory is not controlled. Antigen retrieval, crucial in the analytical phase, continues to be employed in an empirical manner with the actual mechanism of action remaining elusive. There is great variation in reagents, methodology, and duration of tissue processing and immunostaining procedure, and the detection systems employed are not standardized between laboratories. While many of these variables are offset by the application of antigen retrieval, which enables the detection of a wide range of antigens in FFPT, the method itself is not standardized. This myriad of variables makes it inappropriate to provide meaningful comparisons of results obtained in different laboratories and even in the same laboratory, as in current practice, each specimen experiences different preanalytical variables. Furthermore, variables in interpretation exist and cutoff thresholds for positivity differ. Failure to recognize false-positive and false-negative stains leads to further errors of quantitative measurement. Many of the problems relating to the technology and interpretation of immunostaining originate from failure to recognize that this procedure is different from other histological stains and involves many more steps that cannot be monitored until the end result is attained. While several remedial measures can be suggested to address some of these problems, accurate and reproducible quantitative assessment of immunostains presently remains elusive as important variables that impact on antigen preservation in the paraffin-embedded biopsy -cannot be standardized.

Microwaves for chromogenic in situ hybridization.

In situ hybridization can be employed in formalin-fixed, paraffin-embedded tissue sections (FFPT) and allows direct visualization of amplified genes and chromosomes in individual cell nuclei. Fluorescence in situ hybridization (FISH) is the most widely employed method, but the fluorescence preparations suffer from the main disadvantages of fading over time and poor visualization, the latter making it difficult to accurately separate invasive from in situ cancer cells. Chromogenic in situ hybridization (CISH) is a viable alternative to FISH in FFPT as it employs a peroxidase reaction to visualize the chromogen thus allowing the convenience of bright field microscopy and the correlation of the visualized gene amplification with cytomorphology. It is relatively less expensive and allows a permanent record, with several studies attesting to its validity. As with FISH, heat pretreatment and enzyme digestion are two critical components of the protocol. We describe a protocol for CISH in which a microwave-induced target retrieval step is introduced as a replacement for heat pretreatment. The same procedure is performed following enzyme digestion to produce consistent signals in amplified and nonamplified cells that are both larger in size and numbers when compared with those produced by the conventional protocol.

Immunohistology--past, present, and future.

The rapid development of immunohistochemistry, a morphology-based technique, has come about through refinements in detection systems and an increasing range of sensitive and specific antibodies that have allowed application of the technique to formalin-fixed, paraffin-embedded tissues. The introduction of heat-induced antigen retrieval has been a significant milestone to compliment these developments so that the immunohistochemistry is firmly entrenched as an indispensable adjunct to morphologic diagnosis. Although this ancillary stain was initially used in a qualitative manner, problems surrounding the many variables that influence antigen preservation in formalin-fixed, paraffin-embedded tissues were not a major issue and laboratories strived to optimize their staining protocols to the material they accessioned and processed. The advent of personalized medicine and targeted cancer treatment has imposed the need to quantitate the stain reaction product and has resulted in calls to standardize the process of immunostaining. A closer examination of the variables that influence the ability to show antigens in formalin-fixed, paraffin-embedded tissues revealed many important variables, particularly in the preanalytical phase of the assay, that are beyond the control of the accessioning laboratory. Although analytical factors have the potential to be standardized, the actions of many pivotal procedures including fixation and antigen retrieval are not completely understood. Postanalytical processes including threshold and cut-off values require consensus and standardization and it is clear that some of these goals can be achieved through the direction of national and international organizations associated with cancer diagnosis and treatment. With the ability to serve as a surrogate marker of many genetic abnormalities, immunohistochemistry enters a new era and the need to better understand some of the mechanisms fundamental to the technique become more pressing and the development of true quantitative assays is imperative. There is also an increasing appreciation that the technique highlights patterns of staining that reflect exquisite localization to organelles and tissue structures that are not appreciable in routine stains, adding a further dimension to morphologic diagnosis.

Citraconic anhydride: a new antigen retrieval solution.

BACKGROUND: The introduction of heat-induced antigen retrieval has been a major milestone in diagnostic immunohistochemistry, enabling the application of many antibodies to fixed paraffin-embedded tissue sections. A number of important variables affect the preservation of tissue antigens, among which are analytical variables including the antigen retrieval methodology. Temperature of retrieval, duration of heating, source of heat, pH and nature of retrieval solution are among more important variables pivotal to results. Citrate buffer at pH 6.0 has become widely embraced as the universal fixative but some antibodies remain capricious and yield poor staining results. METHODS: This study examines the recent suggestion that citraconic anhydride may be a suitable universal retrieval reagent. Immunostaining of 65 commonly employed antibodies following microwave antigen retrieval in 0.05% citraconic anhydride for 10 minutes at 98 degrees C was compared with consecutive tissues sections subjected to antigen retrieval in citrate buffer at pH 6.0 at the same duration and temperature. RESULTS: Thirty-five of the 65 antibodies examined yielded more intense staining following antigen retrieval in citraconic anhydride, including some capricious antibodies such as MyoD1, myogenin, perforin, TIA-1, Tdt, RET and MiTF, confirming the efficacy of this retrieval solution. CONCLUSION: It is recommended that consideration be given to 0.05% citraconic anhydride as an antigen retrieval solution, particularly for antibodies that fail to work or stain weakly in fixed paraffin-embedded tissue sections.

Glomeruloid peritoneal implants in ovarian serous borderline tumours--distinction between invasive and non-invasive implants and pathogenesis.

AIMS: To determine whether or not the glomeruloid implants (GI) composed of papillary cores within clear spaces lined by mesothelial cells or tumour cells located in superficial or deep peritoneal tissue in ovarian serous borderline tumours (SBTs) are invasive. METHODS AND RESULTS: We examined the differences in incidence, histological and immunohistochemical findings among three groups: 100 GI with mesothelial cells lining clear space (type I), 100 GI with tumour cells lining clear space (type II), and 100 invasive implants with clefts but no lining cells from 30 cases of SBT with peritoneal implants. The type I lesion had characteristics of non-invasive implants with a tendency for smooth contours (100/100), superficial location (71/100), absence of desmoplasia (100/100) and absence of surrounding destructive invasion (100/100), In contrast, type II GI had irregular contours (67/100), deep location (93/100), presence of desmoplastic reaction (100/100) and presence of destructive invasion (12/100). Immunohistological studies suggested intermediate forms between the two types of lesions. CONCLUSIONS: Type I GI are non-invasive implants, whereas type II GI are invasive implants and it is important to evaluate the presence and nature of cells lining the clear space in determining whether implants associated with ovarian SBTs are invasive or not.

p16, cyclin D1, Ki-67, and AMACR as markers for dysplasia in Barrett esophagus.

Barrett esophagus (BE) is an established precursor of esophageal adenocarcinoma (AdenoCa). One hundred and one cases of BE diagnosed by esophageal biopsy and resections were examined morphologically for dysplasia. These were categorized as BE without dysplasia (n=25), indefinite for dysplasia (IND, n=17), low-grade dysplasia (LGD, n=18), high-grade dysplasia (HGD, n=15), and AdenoCa (n=26). Immunostaining for p16 (INK4A/CDKN2A), Cyclin D1 (CCND1), Ki-67, and alpha-methylacyl-CoA racemase (AMACR) was employed to assess their potential as diagnostic discriminators. Abnormal p16 expression (negative, cytoplasmic, or combined cytoplasmic and nuclear staining) was present in all categories, rising from 68% in BE without dysplasia to 100% in AdenoCa, with cytoplasmic staining only showing a significant correlation with the severity of dysplasia. Cyclin D1 expression was present in almost all cases, but high expression (>50% cells positive) was displayed mostly in HGD and AdenoCa (46.7% and 42.3%, respectively). Ki-67 index increased with the severity of dysplasia and labeling extended from the lower third of the crypts to the superficial epithelium. The frequency of AMACR-positivity was 12% in BE, 47.1% in IND, 44.4% in LGD, 93.3% in HGD, and 96.2% in AdenoCa. The intensity and extent of AMACR staining also increased with the severity of dysplasia. Aberrant p16 and high-cyclin D1 expression may reflect early genetic events during the progression of Barrett-associated carcinogenesis. Cytoplasmic staining of p16 is specific. It may represent a different pathway of p16 dysfunction. The pattern and extent of Ki-67 staining and AMACR overexpression are useful additional parameters for identifying dysplasia in BE.

CDX2 and villin are useful markers of intestinal metaplasia in the diagnosis of Barrett esophagus.

The identification of intestinal metaplasia (IM) in the esophagus is necessary for the selection of patients with Barrett esophagus (BE) for surveillance. We studied 108 esophageal biopsy and resection specimens, clinically diagnosed as BE, and stained them for CDX2, villin, HepPar-1, and cytokeratin (CK) 7 to investigate sensitivity for identifying IM. H&E-stained sections showed definite goblet cells in 94 cases. CDX2 and villin were positive in all 94 cases. Of 38 cardia- and 9 fundic-type mucosa samples associated with BE, 13 (34%) and 0 (0%) displayed CDX2 positivity and 21 (55%) and 1 (11%) displayed villin positivity, respectively. HepPar-1 was positive in 54 (57%) of 94 cases with IM and negative in the associated cardia- and fundic-type mucosa. A full-thickness CK7 staining pattern was present in 90 (96%) of samples with IM and 22 (58%) and 0 (0%) of the associated cardia- and fundic-type mucosa, respectively. None of 20 control samples of morphologically normal gastric mucosa stained for CDX2 or villin. CDX2 and villin are sensitive markers for early-stage IM and can supplement the histologic identification of this premalignant condition in the esophagus.

Vascular and myometrial changes in the human uterus at term.

At term, amniotic fluid contents may mediate the onset of labor through the activation of amniotic fluid macrophages and their migration into the myometrium. To test this concept, the authors examine the histological changes that occur in myometrial biopsies at term prior to (n = 53) and during (n = 15) labor. Biopsies were stained with an antimacrophage antibody, anti-CD34 (endothelial cells), and anti-caspase 3 (apoptotic cells). The samples showed a variable inflammatory infiltrate of neutrophils and macrophages, with a greater infiltrate in the samples obtained during labor (P < .001, Fisher exact test). Prior to labor, there were prominent changes in the myometrial fibers that reflected shearing, shrinkage, edema, and particularly apoptosis; endothelial cells of thin-walled vessels prominent in the biopsies displayed marked nuclear biotinylation, and the vascular lumen contained fibrin and platelet thrombi, microparticles, desquamated endothelial cells, amniotic squamous cells, and mucoid material. These changes were also present in samples obtained during labor. In an additional 10 patients in labor with male fetuses, myometrial samples were examined for the presence of macrophages carrying a Y chromosome indicative of a fetal origin, but none were observed. These findings suggest that endothelial cell damage and amniotic fluid embolism are very common at term prior to clinical labor and provide a mechanism by which surfactant protein A and phospholipids present in the amniotic fluid may access myometrial cells and provoke the inflammatory response that occurs during parturition. The authors' studies give no support to the suggestion that fetal macrophages might invade the human myometrium at term.

The pathology of dengue hemorrhagic fever.

An estimated 2.5 billion people are at risk of dengue infection, and of the 100 million cases of dengue fever per year, up to 500,000 develop dengue hemorrhagic fever (DHF) or dengue shock syndrome (DSS), the life-threatening forms of the infection. The large majority of DHF/DSS occurs as the result of a secondary infection with a different serotype of the virus. While not completely understood, there is evidence that the target cells include dendritic reticulum cells, monocytes, lymphocytes, hepatocytes, and vascular endothelial cells. Viral replication appears to occur in dendritic cells, monocytes, and possibly circulating lymphoid cells, and damage to these and other target cells occurs through immune-mediated mechanisms related to cross-reacting antibodies and cytokines released by dendritic cells, monocytes, and vascular endothelium. There is evidence of a concomitant cellular activation as well as immune suppression during the infection. The activation of memory T cells results in cascades of inflammatory cytokines, including tumor necrosis factor-alpha, interleukins (IL-2, IL-6, and IL-8), and other chemical mediators that increase vascular endothelial permeability or trigger death of target cells through apoptosis. Pathological studies in humans are uncommon, and a suitable animal model of DHF/DSS does not exist. The current treatment of DHF/DSS is symptomatic, and prevention is through vector control. As such, there is a great impetus for the development of vaccines and novel therapeutic molecules to impede viral replication in infected cells or counteract the effects of specific inflammatory mediators on target cells. The role of genetics in relation to resistance to DHF/DSS also requires clarification.

Microwave enhancement of CISH for HER2 oncogene.

We describe a modification to the prescribed procedure for the Zymed Spot-Light HER2 chromogenic in situ hybridization kit (84-0146, Zymed Laboratories, San Francisco, CA) by substituting the heat pretreatment step with MW irradiation in citrate buffer 10 mmol/L at pH 6.0 at 98 degrees C for 10 minutes and repeating the procedure afterenzyme digestion with time and temperature controlled in the Mega T/ T oven (Milestone s.r.l., Sorisole, Italy). The subsequent procedure leading up to hybridized was as per manufacturer's instructions. Invasive breast carcinoma previously scored by immunohistochemistry for HER2, comprising 18 cases of 3+, 18 cases of 2+, and 12 cases of 1+, were examined by chromogenic in situ hybridization using this modified procedure, with a parallel set of cases examined by the prescribed Zymed method. The introduction of the "MW retrieval" steps resulted in consistently a greater number of hybridization signals in amplified tumor cells with benign epithelial cells and lymphocytes displaying 2 clear dots compared with the weaker and less consistent signals obtained with the standard procedure. MW exposed sections showed larger numbers of large and small clusters that often allowed identification of amplified tumors without having to count single dots with crisp staining and absence of background precipitation.

How does antigen retrieval work?

The introduction of antigen retrieval has enabled immunohistology to become an integral component of morphologic diagnosis, routinely employed in cancer diagnosis, and for the identification of therapeutic and prognostic markers. The mechanism of antigen retrieval, however, remains speculative with the key to our understanding embedded in the actions of formaldehyde on proteins. One commonly held concept is that heat primarily breaks down protein cross-linkages that occur with aldehyde fixation, thus "unmasking" protein epitopes of interest. Enzymatic pretreatment is also thought to have a similar action whereas such "breakages" are the result of extremely rapid molecular movement induced by microwaves and ultrasound. The formation of rigid cagelike calcium complexes during formaldehyde fixation is another suggested mechanism of antigen masking requiring chelating agents for reversal. A more recent suggestion for the antigen retrieval phenomenon has evoked the Mannich reaction, which occurs with the cross-linking of some proteins. Such cross-linkages can be hydrolyzed by heat or alkalis so that the process of antigen retrieval may be the simple removal of such cross-linked proteins that are sterically interfering with the binding of antibodies to linear protein epitopes in the tissue section. We are clearly not yet in possession of all the answers to the problem.

The morphological spectrum of lymphadenopathy in HIV infected patients.

AIMS: To study the histological spectrum of lymphadenopathy in human immunodeficiency virus (HIV) infected Thai patients. METHODS: Lymph nodes from 55 HIV infected patients were accessioned over a 19 month period in two pathology laboratories in Bangkok, Thailand. These were examined with H&E, Ziehl-Neelsen, periodic acid-Schiff (PAS), PAS with diastase (PAS/D), Gram and methenamine stains. RESULTS: Six reaction patterns were observed: (1) classic necrotising granulomas (30 cases); (2) extensive necrosis with minimal granulomatous response (5 cases); (3) sarcoid-like non-necrotising granulomas (5 cases); (4) foamy macrophage or pseudo-Gaucher cell response (5 cases); (5) inflammatory pseudotumour-like proliferation (3 cases); and (6) non-specific lymphoid hyperplasia (7 cases). Myriads of intracellular, long, slender acid-fast bacilli were found in those cases with the pseudo-Gaucher cell and inflammatory pseudotumour-like response, while variable numbers of bacilli were identified in those cases with non-necrotising sarcoid-like granulomas. Few scattered acid-fast bacilli were found in five cases with necrotising granulomas. In one case, yeast-like organisms in keeping with Cryptococcus were identified. No organisms were identified in the cases showing lymphoid hyperplasia, extensive necrosis and minimal granulomatous response, and in the remaining cases of classic necrotising granulomas. CONCLUSIONS: The wide spectrum of histological changes in HIV-associated lymphadenomegaly requires recognition, particularly as the majority were associated with acid-fast organisms, mostly in keeping with the morphological features of Mycobacterium avium-M. intracellulare complex that was distinctively stained by Grocott methenamine-silver, Gram and PAS stains. The histological changes mimic those of infarction and other infective lymphadenitis, sarcoidosis, Whipple's disease, inflammatory pseudotumour and spindle cell neoplasms.

Refinement of immunohistologic parameters for Her2/neu scoring validation by FISH and CISH.

The conventional method of scoring Her2/neu immunostaining is recognized to result in a high false-positive rate among 2+ cases when compared with results obtained with fluorescence in situ hybridization (FISH); however, costs and convenience dictates that immunohistochemistry remains the screening test for Her2/neu status in patients with breast cancer. We describe refined criteria for scoring of Her2/neu on the basis of anatomic localization rather than the subjective assessment of intensity. The presence of a circumferential tram track pattern that results from the staining of apposing cell membranes in >25% of the tumor cells was necessary for a 3+ score (Her2/neu overexpressed) and the presence of the tram track pattern in <25% was scored 2+ (equivocal); granular and fragmented membrane staining was scored 1+ (negative). The tram track pattern of Her2/neu overexpression showed 100% concordance with gene amplification. FISH and CISH testing in selected cases from the other categories validated the revised scoring method. These criteria reduced the numbers of equivocal staining cases that required FISH testing.

Controversies in the assessment of HER-2: more questions than answers.

Human epidermal growth factor receptor 2 (HER-2) is over-expressed in 15% to 30% of breast cancers and is a poor prognostic marker in node-positive patients. HER-2 expression is an indicator of greater sensitivity to anthracycline-based chemotherapy and is the major criterion for selection for treatment with the anti-HER-2 antibody trastuzumab (Herceptin). Fluorescence in situ hybridization and immunohistochemistry (IHC) are the 2 most commonly used methods for detection of the gene and protein, respectively. Criticisms have been levied at the IHC method of identifying HER-2 overexpression but convenience and costs of this technique cannot be overlooked. Modifications to the IHC technique and scoring accommodate for many of the problems that derive from variables in preanalytical and analytic factors that influence results but standardization is currently impossible to attain. Deficiencies in fluorescence in situ hybridization assay also exist and alternative molecular methods of assay are explored in this review.

Cytokeratin 20 and Ki-67 to distinguish carcinoma in situ from flat non-neoplastic urothelium.

Urothelial carcinoma in situ (CIS) is a high-grade neoplasm and an indicator of recurrence and progression that requires specific treatment. The distinction of CIS from flat non-neoplastic urothelium, in particular dysplasia, on the basis of histologic features is often difficult, and this study aims to validate cytokeratin 20 (CK20) and Ki-67 as discriminatory markers for this purpose. Immunostaining of these markers was applied to 26 cases of CIS, 14 atypia of unknown significance, 4 dysplasia, 6 normal, and 9 hyperplastic urothelium. CIS showed CK20 staining of deep urothelial cells in 23/26 CIS compared with restricted staining in surface cells in all non-neoplastic lesions. CIS had significantly increased Ki-67 index with a mean of 53.37% compared with that of non-neoplastic urothelium, which was <10% (P<0.0001). The proliferating cells were distributed randomly in CIS, whereas in non-neoplastic urothelium, staining was confined to the basal layer. Among the cases of atypia, 3/14 displayed deep staining for CK20 and 6/14 had elevated Ki-67 counts. In dysplasia similar findings were present in 1/4 and 2/4 cases, respectively. These findings suggest that CK20 and Ki-67 are objective markers to distinguish CIS from non-neoplastic urothelium. In cases of "atypia of unknown significance" and "dysplasia," positivity for both markers should raise the possibility of CIS or preneoplastic change and identify those cases for follow-up.

Immunohistological localisation of human FAT1 (hFAT) protein in 326 breast cancers. Does this adhesion molecule have a role in pathogenesis?

AIMS: To examine the immunohistological expression in human breast cancers of human FAT1 (hFAT) protein, a recently described member of the cadherin superfamily, and its correlation with histological type and grade. METHODS: A total of 326 cases of invasive and in situ breast cancer representing a broad spectrum of histological subtypes were immunostained with affinity-purified rabbit antibodies produced to the cytoplasmic region of hFAT using a standard avidin-biotin system. Staining intensity was arbitrarily graded on a scale of 0 to 3. RESULTS: All tumours showed diffuse staining for hFAT. Immunoexpression of the protein was generally strong in both lobular (LCIS, n = 2) and ductal in situ carcinoma (DCIS, n = 55). hFAT was also strongly immunoexpressed in all types of invasive carcinoma. Grade 3 DCIS displayed the highest hFAT intensity compared with lower grade tumours, with significant differences between grade 1 and 3 (p = 0.015) and grade 2 and 3 (p = 0.047). With invasive ductal carcinomas (n = 128) the difference was not as clear-cut, as most tumours showed moderate (n = 63) or strong staining (n = 49), although grade 3 IDC revealed significantly decreased immunoexpression compared with grade 1 IDC (p = 0.03). CONCLUSIONS: The results illustrate that hFAT1 does not display the pattern of expression seen with the E-cadherin-ss-catenin adhesion complex; however, its over-expression and diffuse expression in both in situ and invasive carcinoma strongly suggests a role in carcinogenesis. From the known functions of FAT1 it is suggested that the concurrent loss of classical cadherins from cell-cell junctions accompanied by increased FAT1 expression contributes to loss of duct formation, and increased cell migration and invasion.

Calretinin, CD34, and alpha-smooth muscle actin in the identification of peritoneal invasive implants of serous borderline tumors of the ovary.

The correct identification of invasive implants in the peritoneum in serous borderline tumors (SBTs) of the ovary is an important determinant of diagnosis, treatment, and prognosis. Although the histologic criteria to distinguish noninvasive from invasive implants have been defined, the distinction can still be difficult. We examined the presence and distribution of mesothelial cells, stromal fibrocytes, and myofibroblasts in invasive and noninvasive peritoneal implants in 100 noninvasive, 100 invasive, and 100 metastatic nests/foci from 20 cases of SBTs with peritoneal implants, 10 serous carcinomas with peritoneal metastasis, and 10 cases of endosalpingiosis by immunostaining for calretinin, CD34, and alpha-SMA. All 100 invasive nests from seven SBTs and all 100 metastatic nests from the cases of serous carcinoma showed loss of calretinin+ mesothelial cells and stromal CD34+ fibrocytes around the nests. In contrast, 72/100 noninvasive nests displayed the presence of mesothelial cells around the nests and 68 displayed preservation of surrounding stromal fibrocytes. Alpha-smooth muscle actin positive myofibroblasts were present as a stromal response in 100/100 metastatic nests, 100/100 invasive nests and 54/100 noninvasive nests. The loss of mesothelial cells and stromal fibrocytes surrounding invasive nests together with a proliferation of myofibroblasts as demonstrated by immunostaining proved to be a sensitive and specific tool to separate invasive from noninvasive implants and represents an important adjunct to morphologic diagnosis. Combined sensitivity and specificity of the three antibodies was 100 and 81%, respectively. These methods, however, may not be helpful for small biopsies of noninvasive desmoplastic implants. The distribution of these cells provides some insights into the histogenesis of invasive and noninvasive implants in SBTs.