Robert J. Marinelli

Systematic Analysis of Breast Cancer Morphology Uncovers Stromal Features Associated with Survival

Automated quantification of thousands of morphologic features in microscopic images of breast cancer allows the construction of a robust prognostic model. An Automated Pathologist Reads Cancer Biopsies How is a camera different from the human eye? Only the eye’s images undergo extensive secondary processing as they are interpreted by the human brain. But what if we could program a computer to do the secondary processing? A pathologist reading a cancer biopsy slide matches his or her brain’s memory of certain cancer-related features (tubules, atypical nuclei, and mitosis) against the tissue. This decades-old scoring system is still standard in most places for prognosis and treatment of cancer, despite its variability and often unreliability. Now, Beck et al. have created an automated pathologist by replacing the human brain with sophisticated image processing software and instructing it to find quantitative aspects of breast cancer tissue that predict prognosis. The software located a set of features that strongly predicted breast cancer outcome in both training and validation samples. With an image analysis protocol they termed C-Path, the authors set their program loose on a set of samples from patients in the Netherlands. From more than 6000 features, the software found a set that were associated with samples from patients who had died sooner. The key aspect of this analysis was that these features were not predefined by a pathologist as being relevant to cancer; instead, the software itself found the cancer-related features among the very large set of measurements of the image. Classifying the tissue as epithelial or stromal, an important part of cancer diagnosis, took a bit of extra work: The authors needed to provide the software with some hand-marked samples so it could learn the difference. The C-Path score yielded information above and beyond that from many other measures of cancer severity including pathology grade, estrogen receptor status, tumor size, and lymph node status. In another, completely independent group of women from Vancouver, the C-Path score was also associated with overall survival. An unexpected finding was that the features that were the best predictors of patient survival were not from the cancer itself but were from the adjacent stromal tissue. Women with worse outcomes tended to have thin cords of epithelial cells infiltrating the stroma, which resulted in high-risk stromal matrix variability scores. These patients also tended to have more inflammatory cells in the stroma (picked up as dark areas by the software). Replacing the human brain with an unbiased image processing system can extract more information from microcopy images and discover new biological aspects of cancer tissue. The morphological interpretation of histologic sections forms the basis of diagnosis and prognostication for cancer. In the diagnosis of carcinomas, pathologists perform a semiquantitative analysis of a small set of morphological features to determine the cancer’s histologic grade. Physicians use histologic grade to inform their assessment of a carcinoma’s aggressiveness and a patient’s prognosis. Nevertheless, the determination of grade in breast cancer examines only a small set of morphological features of breast cancer epithelial cells, which has been largely unchanged since the 1920s. A comprehensive analysis of automatically quantitated morphological features could identify characteristics of prognostic relevance and provide an accurate and reproducible means for assessing prognosis from microscopic image data. We developed the C-Path (Computational Pathologist) system to measure a rich quantitative feature set from the breast cancer epithelium and stroma (6642 features), including both standard morphometric descriptors of image objects and higher-level contextual, relational, and global image features. These measurements were used to construct a prognostic model. We applied the C-Path system to microscopic images from two independent cohorts of breast cancer patients [from the Netherlands Cancer Institute (NKI) cohort, n = 248, and the Vancouver General Hospital (VGH) cohort, n = 328]. The prognostic model score generated by our system was strongly associated with overall survival in both the NKI and the VGH cohorts (both log-rank P ≤ 0.001). This association was independent of clinical, pathological, and molecular factors. Three stromal features were significantly associated with survival, and this association was stronger than the association of survival with epithelial characteristics in the model. These findings implicate stromal morphologic structure as a previously unrecognized prognostic determinant for breast cancer.

Placental S100 (S100p) and GATA3 : Markers for transitional epithelium and urothelial carcinoma discovered by complementary DNA microarray

The morphologic distinction between prostate and urothelial carcinoma can be difficult. To identify novel diagnostic markers that may aid in the differential diagnosis of prostate versus urothelial carcinoma, we analyzed expression patterns in prostate and bladder cancer tissues using complementary DNA microarrays. Together with our prior studies on renal neoplasms and normal kidney, these studies suggested that the gene for placental S100 (S100P) is specifically expressed in benign and malignant urothelial cells. Using tissue microarrays, a polyclonal antiserum against S100P protein stained 86% of 295 urothelial carcinomas while only 3% of 260 prostatic adenocarcinomas and 1% of 133 renal cell carcinomas stained. A commercially available monoclonal antibody against S100P stained 78% of 300 urothelial carcinomas while only 2% of 256 prostatic adenocarcinomas and none of 137 renal cell carcinomas stained. A second gene, GATA3, also showed high level expression in urothelial tumors by cDNA array. A commercially available monoclonal antibody against GATA3 stained 67% of 308 urothelial carcinomas, but none of the prostate or renal carcinomas. For comparison, staining was also performed for p63 and cytokeratin 5/6. p63 stained 87% of urothelial carcinomas whereas CK5/6 stained 54%. Importantly, when S100P and p63 were combined 95% of urothelial carcinomas were labeled by one or both markers. We conclude that the detection of S100P and GATA3 protein expression may help distinguish urothelial carcinomas from other genitourinary neoplasms that enter into the differential diagnosis.

The Macrophage Colony-Stimulating Factor 1 Response Signature in Breast Carcinoma

Purpose: Macrophages play an important role in breast carcinogenesis. The pathways that mediate the macrophage contribution to breast cancer and the heterogeneity that exists within macrophages are incompletely understood. Macrophage colony-stimulating factor 1 (CSF1) is the primary regulator of tissue macrophages. The purpose of this study was to define a novel CSF1 response signature and to evaluate its clinical and biological significance in breast cancer. Experimental Design: We defined the CSF1 response signature by identifying genes overexpressed in tenosynovial giant cell tumor and pigmented villonodular synovitis (tumors composed predominantly of macrophages recruited in response to the overexpression of CSF1) compared with desmoid-type fibromatosis and solitary fibrous tumor. To characterize the CSF1 response signature in breast cancer, we analyzed the expression of CSF1 response signature genes in eight published breast cancer gene expression data sets (n = 982) and did immunohistochemistry and in situ hybridization for CSF1 response genes on a breast cancer tissue microarray (n = 283). Results: In both the gene microarray and tissue microarray analyses, a consistent subset (17-25%) of breast cancers shows the CSF1 response signature. The signature is associated with higher tumor grade, decreased expression of estrogen receptor, decreased expression of progesterone receptor, and increased TP53 mutations (P < 0.001). Conclusions: Our data show that the CSF1 response signature is consistently seen in a subset of breast carcinomas and correlates with biological features of the tumor. Our findings provide insight into macrophage biology and may facilitate the development of personalized therapy for patients most likely to benefit from CSF1-targeted treatments.

Prognostic significance of macrophage infiltration in leiomyosarcomas

Purpose: Macrophages are migratory cells that are frequently recruited to the site of tumors. Their presence is associated with poor clinical outcome in a variety of epithelial malignancies. The aim of this study is to examine the prognostic significance of tumor-associated macrophages in sarcomas. Experimental Design: Global gene expression profiling data of a series of soft tissue tumors were analyzed for macrophage-associated gene expression. Immunohistochemistry on tissue microarrays containing leiomyosarcoma cases with known clinical outcome was used to verify the presence of macrophages and to examine the relationship between tumor-associated macrophages and clinical outcome. Results: Gene expression profiling revealed high-level expression of several macrophage-associated genes such as CD163 and CD68 in a subset of leiomyosarcomas, indicating the presence of variable numbers of tumor-infiltrating macrophages. This was confirmed by CD68 and CD163 immunostaining of a tissue microarray containing 149 primary leiomyosarcomas. Kaplan-Meier survival analysis showed that high density of tumor-infiltrating macrophages as identified by CD163 or CD68 staining is associated with a significantly worse disease-specific survival in nongynecologic leiomyosarcomas, whereas leiomyosarcomas arising from the gynecologic tract showed no significant association between macrophage infiltration and survival. The presence of tumor necrosis did not correlate significantly with outcome. Conclusions: An increased density of CD163- or CD68-positive tumor-infiltrating macrophages is associated with poor outcome in nongynecologic leiomyosarcomas. This may help the clinical management of patients with leiomyosarcomas.

Discovery of molecular subtypes in leiomyosarcoma through integrative molecular profiling

Leiomyosarcoma (LMS) is a soft tissue tumor with a significant degree of morphologic and molecular heterogeneity. We used integrative molecular profiling to discover and characterize molecular subtypes of LMS. Gene expression profiling was performed on 51 LMS samples. Unsupervised clustering showed three reproducible LMS clusters. Array comparative genomic hybridization (aCGH) was performed on 20 LMS samples and showed that the molecular subtypes defined by gene expression showed distinct genomic changes. Tumors from the ‘muscle-enriched’ cluster showed significantly increased copy number changes (P=0.04). A majority of the muscle-enriched cases showed loss at 16q24, which contains Fanconi anemia, complementation group A, known to have an important role in DNA repair, and loss at 1p36, which contains PRDM16, of which loss promotes muscle differentiation. Immunohistochemistry (IHC) was performed on LMS tissue microarrays (n=377) for five markers with high levels of messenger RNA in the muscle-enriched cluster (ACTG2, CASQ2, SLMAP, CFL2 and MYLK) and showed significantly correlated expression of the five proteins (all pairwise P<0.005). Expression of the five markers was associated with improved disease-specific survival in a multivariate Cox regression analysis (P<0.04). In this analysis that combined gene expression profiling, aCGH and IHC, we characterized distinct molecular LMS subtypes, provided insight into their pathogenesis, and identified prognostic biomarkers.

The gene expression profile of extraskeletal myxoid chondrosarcoma.

Extraskeletal myxoid chondrosarcoma (EMC) is a soft tissue tumour that occurs primarily in the extremities and is characterized by a balanced translocation most commonly involving t(9;22) (q22;q12). The morphological spectrum of EMC is broad and thus a diagnosis based on histology alone can be difficult. Currently, no systemic therapy exists that improves survival in patients with EMC. In the present study, gene expression profiling has been performed to discover new diagnostic markers and potential therapeutic targets for this tumour type. Global gene expression profiling of ten EMCs and 26 other sarcomas using 42 000 spot cDNA microarrays revealed that the cases of EMC were closely related to each other and distinct from the other tumours profiled. Significance analysis of microarrays (SAM) identified 86 genes that distinguished EMC from the other sarcomas with 0.25% likelihood of false significance. NMB, DKK1, DNER, CLCN3, and DEF6 were the top five genes in this analysis. In situ hybridization for NMB gene expression on tissue microarrays (TMAs) containing a total of 1164 specimens representing 62 different sarcoma types and 15 different carcinoma types showed that NMB was highly expressed in 17 of 22 EMC cases and very rarely expressed in other tumours and thus could function as a novel diagnostic marker. High levels of expression of PPARG and the gene encoding its interacting protein, PPARGC1A, in most EMCs suggest activation of lipid metabolism pathways in this tumour. Small molecule inhibitors for PPARG exist and PPARG could be a potential therapeutic target for EMC. Copyright

Immunohistochemical characterization of nasal-type extranodal NK/T-cell lymphoma using a tissue microarray: an analysis of 84 cases.

Nasal-type extranodal natural killer (NK)/T-cell lymphoma is an uncommon malignancy. By using a tissue microarray, we characterized 84 cases of extranodal NK/T-cell lymphoma with regard to expression of 18 immunohistochemical markers and the presence of Epstein-Barr virus (EBV) RNA. In our series, CD2 was positive in 69 (93%) of 74 cases, CD3 in 68 (84%) of 81, CD5 in 22 (27%) of 81, CD20 in 0 (0%) of 82, CD29 in 75 (91%) of 82, CD30 in 29 (35%) of 84, CD43 in 81 (96%) of 84, CD54 in 58 (72%) of 81, CD56 in 46 (58%) of 79, CD62L in 23 (28%) of 83, CD183 in 66 (80%) of 83, BCL2 in 33 (39%) of 84, cutaneous lymphocyte antigen in 21 (25%) of 84, granzyme B in 70 (83%) of 84, Ki-67 in 59 (71%) of 83, linker for activation of T cells in 60 (71%) of 84, perforin in 66 (86%) of 77, TIA1 in 76 (90%) of 84, and EBV in 73 (87%) of 84. Hierarchical cluster analysis separated primary cutaneous cases from cases manifesting in other sites based on lower expression of the cell adhesion molecule CD54.

Coordinate expression of colony-stimulating factor-1 and colony-stimulating factor-1-related proteins is associated with poor prognosis in gynecological and nongynecological leiomyosarcoma.

Previously, we showed that the presence of high numbers of macrophages correlates with poor prognosis in nongynecological leiomyosarcoma (LMS). In gynecological LMS, a similar trend was noted but did not reach statistical significance. Colony-stimulating factor-1 (CSF1) is a major chemoattractant for macrophages. Here we show that in a subset of LMS cases, CSF1 is expressed by the malignant cells. Previously, we found that CSF1 is translocated and highly expressed in tenosynovial giant cell tumors (TGCTs), and this observation allowed us to identify genes that showed a coordinate expression with CSF1. Here, we evaluated the expression of CSF1 and TGCT-associated proteins in 149 cases of LMS. The coordinate expression of CSF1 and three TGCT-associated proteins (CD163, FCGR3a, and CTSL1) identified cases with poor prognosis in both gynecological LMS (P = 0.00006) and nongynecological LMS (P = 0.03). In gynecological LMS, the coordinate expression of these four markers was the only independent prognosticator in multivariate analysis (hazard ratio, 4.2; 95% CI, 1.12 to 16; P = 0.03). Our findings indicate that CSF1 may play an important role in the clinical behavior of LMS that may open a window for new therapeutic reagents.

The Stanford Tissue Microarray Database

The Stanford Tissue Microarray Database (TMAD; http://tma.stanford.edu) is a public resource for disseminating annotated tissue images and associated expression data. Stanford University pathologists, researchers and their collaborators worldwide use TMAD for designing, viewing, scoring and analyzing their tissue microarrays. The use of tissue microarrays allows hundreds of human tissue cores to be simultaneously probed by antibodies to detect protein abundance (Immunohistochemistry; IHC), or by labeled nucleic acids (in situ hybridization; ISH) to detect transcript abundance. TMAD archives multi-wavelength fluorescence and bright-field images of tissue microarrays for scoring and analysis. As of July 2007, TMAD contained 205 161 images archiving 349 distinct probes on 1488 tissue microarray slides. Of these, 31 306 images for 68 probes on 125 slides have been released to the public. To date, 12 publications have been based on these raw public data. TMAD incorporates the NCI Thesaurus ontology for searching tissues in the cancer domain. Image processing researchers can extract images and scores for training and testing classification algorithms. The production server uses the Apache HTTP Server, Oracle Database and Perl application code. Source code is available to interested researchers under a no-cost license.

Diffuse myogenin expression by immunohistochemistry is an independent marker of poor survival in pediatric rhabdomyosarcoma: a tissue microarray study of 71 primary tumors including correlation with molecular phenotype.

The pathologic classification of rhabdomyosarcoma (RMS) into embryonal or alveolar subtype is an important prognostic factor guiding the therapeutic protocol chosen for an individual patient. Unfortunately, this classification is not always straightforward, and the diagnostic criteria are controversial in a subset of cases. Ancillary studies are used to aid in the classification, but their potential use as independent prognostic factors is rarely studied. The aim of this study is to identify immunohistochemical markers of potential prognostic significance in pediatric RMS and to correlate their expression with PAX-3/FKHR and PAX-7/FKHR fusion status. A single tissue microarray containing 71 paraffin-embedded pediatric RMSs was immunostained with antibodies against p53, bcl-2, Ki-67, CD44, myogenin, and MyoD1. The tissue microarray and whole paraffin blocks were studied for PAX-3/FKHR and PAX-7/FKHR gene fusions by fluorescence in situ hybridization and reverse transcription-polymerase chain reaction. Clinical follow-up data were available for each patient. Immunohistochemical staining results and translocation status were correlated with recurrence-free interval (RFI) and overall survival (OS) using the Kaplan-Meier method, the log-rank test, and Cox proportional hazard regression. The minimum clinical follow-up interval was 24 months (median follow-up=57 mo). On univariable analysis, immunohistochemical expression of myogenin, bcl-2, and identification of a gene fusion were associated with decreased 5-year RFI and 10-year OS (myogenin RFI P=0.0028, OS P=0.0021; bcl-2 RFI P=0.037, OS P=0.032; gene fusion RFI P=0.0001, OS P=0.0058). After adjustment for Intergroup Rhabdomyosarcoma Study-TNM stage, tumor site, age, tumor histology, and translocation status by multivariable analysis, only myogenin retained an independent association with RFI (P=0.034) and OS (P=0.0069). In this retrospective analysis, diffuse immunohistochemical reactivity for myogenin in RMS correlates with decreased RFI and OS, independent of histologic subtype, translocation status, tumor site, or stage.