•A system for automatic histological grading of prostate cancer was developed.•The classifier was trained based on detailed annotations of six pathologists.•The inter-observer variability was ...addressed by adapting a crowdsourcing approach.•Novel features based on spatial statistics of the nuclei were proposed.•The performance of the classifier was within agreement levels among pathologists.
Display omitted
Prostate cancer (PCa) is a heterogeneous disease that is manifested in a diverse range of histologic patterns and its grading is therefore associated with an inter-observer variability among pathologists, which may lead to an under- or over-treatment of patients. In this work, we develop a computer aided diagnosis system for automatic grading of PCa in digitized histopathology images using supervised learning methods. Our pipeline comprises extraction of multi-scale features that include glandular, cellular, and image-based features. A number of novel features are proposed based on intra- and inter-nuclei properties; these features are shown to be among the most important ones for classification. We train our classifiers on 333 tissue microarray (TMA) cores that were sampled from 231 radical prostatectomy patients and annotated in detail by six pathologists for different Gleason grades. We also demonstrate the TMA-trained classifier’s performance on additional 230 whole-mount slides of 56 patients, independent of the training dataset, by examining the automatic grading on manually marked lesions and randomly sampled 10% of the benign tissue. For the first time, we incorporate a probabilistic approach for supervised learning by multiple experts to account for the inter-observer grading variability. Through cross-validation experiments, the overall grading agreement of the classifier with the pathologists was found to be an unweighted kappa of 0.51, while the overall agreements between each pathologist and the others ranged from 0.45 to 0.62. These results suggest that our classifier’s performance is within the inter-observer grading variability levels across the pathologists in our study, which are also consistent with those reported in the literature.
Sessile serrated adenomas (SSAs) are associated with colorectal carcinomas (CRCs) that demonstrate high microsatellite instability (MSI-H). Currently, SSAs are managed clinically in a similar fashion ...to adenomatous polyps (APs). We studied the natural history of SSA by analyzing the outcome of previously undiagnosed SSAs and comparing it with that of hyperplastic polyps (HPs) and APs. All colorectal polyps diagnosed between 1980 and 2001 as HP were selected for study. Polyps identified as possible SSAs were reviewed by 3 pathologists, and the diagnosis was confirmed. Clinical follow-up was obtained for each SSA patient and matched with control HP and AP patients. In total, 1402 colorectal polyps diagnosed as HP were examined and 81 polyps in 55 patients (5.8%) were rediagnosed as SSA. Of these, 40 SSA patients had no previous history of either CRC or AP with high-grade dysplasia (HGD). Of these 40 patients, 5 developed subsequent CRCs and 1 developed AP with HGD. The incidence of subsequent CRCs was significantly higher in SSA patients than in control patients with HP (12.5% vs. 1.8%) and AP (12.5% vs. 1.8%). All of the subsequent CRCs or APs with HGD developed in the proximal colon. Four of the 5 CRCs demonstrated a high microsatellite instability phenotype. We conclude that SSAs are high-risk lesions, with 15% of the SSA patients developing subsequent CRCs or APs with HGD. This incidence is higher than that of the control HP and AP patients, and would support close endoscopic follow-up in patients harboring SSAs.
Akt/PKB is a serine/threonine kinase that promotes tumor cell growth by phosphorylating transcription factors and cell cycle proteins. There is particular interest in finding tumor-specific ...substrates for Akt to understand how this protein functions in cancer and to provide new avenues for therapeutic targeting. Our laboratory sought to identify novel Akt substrates that are expressed in breast cancer. In this study, we determined that activated Akt is positively correlated with the protein expression of the transcription/translation factor Y-box binding protein-1 (YB-1) in primary breast cancer by screening tumor tissue microarrays. We therefore questioned whether Akt and YB-1 might be functionally linked. Herein, we illustrate that activated Akt binds to and phosphorylates the YB-1 cold shock domain at Ser102. We then addressed the functional significance of disrupting Ser102 by mutating it to Ala102. Following the stable expression of Flag:YB-1 and Flag:YB-1 (Ala102) in MCF-7 cells, we observed that disruption of the Akt phosphorylation site on YB-1 suppressed tumor cell growth in soft agar and in monolayer. This correlated with an inhibition of nuclear translocation by the YB-1(Ala102) mutant. In conclusion, YB-1 is a new Akt substrate and disruption of this specific site inhibits tumor cell growth.
Assembly of E-cadherin-based adherens junctions (AJ) is obligatory for establishment of polarized epithelia and plays a key role in repressing the invasiveness of many carcinomas. Here we show that ...type Igamma phosphatidylinositol phosphate kinase (PIPKIgamma) directly binds to E-cadherin and modulates E-cadherin trafficking. PIPKIgamma also interacts with the mu subunits of clathrin adaptor protein (AP) complexes and acts as a signalling scaffold that links AP complexes to E-cadherin. Depletion of PIPKIgamma or disruption of PIPKIgamma binding to either E-cadherin or AP complexes results in defects in E-cadherin transport and blocks AJ assembly. An E-cadherin germline mutation that loses PIPKIgamma binding and shows disrupted basolateral membrane targeting no longer forms AJs and leads to hereditary gastric cancers. These combined results reveal a novel mechanism where PIPKIgamma serves as both a scaffold, which links E-cadherin to AP complexes and the trafficking machinery, and a regulator of trafficking events via the spatial generation of phosphatidylinositol-4,5-bisphosphate.
Purpose: FOXA1, a forkhead family transcription factor, is essential for optimum expression of ∼50% of estrogen receptor α (ERα):estrogen
responsive genes. FOXA1 is expressed in breast cancer cells. ...It segregates with genes that characterize the luminal subtypes
in DNA microarray analyses. The utility of FOXA1 as a possible independent prognostic factor has not been determined in breast
cancers.
Materials and Methods: A tissue microarray comprising tumors from 438 patients with 15.4 years median follow-up was analyzed for FOXA1 expression
by immunohistochemistry. Interpretable FOXA1 expression obtained in 404 patients was analyzed along with other prognostic
factors like tumor grade, size, nodal status, ER, progesterone receptor (PR), and HER2/ neu .
Results: FOXA1 expression (score >3) was seen in 300 of 404 breast cancers and it correlated with ER ( P = 0.000001), PR ( P = 0.00001), and luminal A subtype ( P = 0.000001). Loss of expression was noted with worsening tumor grade ( P = 0.001). Univariate analysis showed nodal status ( P = 0.0000012), tumor size ( P = 0.00001), FOXA1 ( P = 0.0004), and ER ( P = 0.012) to be predictors of breast cancer–specific survival. Multivariate analysis showed only nodal status ( P = 0.001) and tumor size ( P = 0.039) to be significant prognostic factors, whereas FOXA1 ( P = 0.060) and ER ( P = 0.131) were not significant. In luminal subtype A patient subgroup, FOXA1 expression was associated with better cancer-specific
survival ( P = 0.024) and in ER-positive subgroup, it was better predictor of cancer-specific survival ( P = 0.009) than PR ( P = 0.213).
Conclusion: FOXA1 expression correlates with luminal subtype A breast cancer and it is significant predictor of cancer-specific survival
in patients with ER-positive tumors. Prognostic ability of FOXA1 in these low-risk breast cancers may prove to be useful in
clinical treatment decisions.
The loss of E-cadherin based cell-cell contacts and tumor cell migration to the vasculature and lymphatic system are hallmarks of metastasis of epithelial cancers. Type I gamma phosphatidylinositol ...phosphate kinase (PIPKIgamma), an enzyme that generates phosphatidylinositol 4,5-bisphosphate (PI4,5P2) a lipid messenger and precursor to many additional second messengers, was found to regulate E-cadherin cell-cell contacts and growth factor-stimulated directional cell migration, indicating that PIPKIgamma regulates key steps in metastasis. Here, we assess the expression of PIPKIgamma in breast cancers and have shown that expression correlated with disease progression and outcome.
Using a tissue microarray, we analyzed 438 breast carcinomas for the levels of PIPKIgamma and investigated the correlation of PIPKIgamma expression with patient survival via Kaplan-Meier survival analysis. Moreover, via knockdown of the expression of PIPKIgamma in cultured breast cancer cells with siRNA, the roles of PIPKIgamma in breast cancer migration, invasion, and proliferation were examined.
Tissue microarray data shows that approximately 18% of the cohort immunostained showed high expression of PIPKIgamma. The Kaplan-Meier survival analysis revealed a significant inverse correlation between strong PIPKIgamma expression and overall patient survival. Expression of PIPKIgamma correlated positively with epidermal growth factor receptor (EGFR) expression, which regulates breast cancer progression and metastasis. In cultured breast cancer cells, PIPKIgamma is required for growth factor stimulated migration, invasion, and proliferation of cells.
The results reveal a significant correlation between PIPKIgamma expression and the progression of breast cancer. This is consistent with PIPKIgamma 's role in breast cancer cell migration, invasion, and proliferation.
Purpose
We have previously demonstrated in a pilot study of 348 invasive breast cancers that mast cell (MC) infiltrates within primary breast cancers are associated with a good prognosis. Our aim was ...to verify this finding in a larger cohort of invasive breast cancer patients and examine the relationship between the presence of MCs and other clinical and pathological features.
Experimental design
Clinically annotated tissue microarrays (TMAs) containing 4,444 cases were constructed and stained with c-Kit (CD-117) using standard immunoperoxidase techniques to identify and quantify MCs. For statistical analysis, we applied a split-sample validation technique. Breast cancer specific survival was analyzed by Kaplan–Meier KM method and log rank test was used to compare survival curves.
Results
Survival analysis by KM method showed that the presence of stromal MCs was a favourable prognostic factor in the training set (
P =
0.001), and the validation set group (
P =
0.006). X-tile plot generated to define the optimal number of MCs showed that the presence of any number of stromal MCs predicted good prognosis. Multivariate analysis showed that the MC effect in the training set (Hazard ratio HR = 0.804, 95% Confidence interval CI, 0.653–0.991,
P =
0.041) and validation set analysis (HR = 0.846, 95% CI, 0.683–1.049,
P =
0.128) was independent of age, tumor grade, tumor size, lymph node, ER and Her2 status.
Conclusions
This study concludes that stromal MC infiltration in invasive breast cancer is an independent good prognostic marker and reiterates the critical role of local inflammatory responses in breast cancer progression.
Proper evaluation of the performance of artificial intelligence techniques in the analysis of digitized medical images is paramount for the adoption of such techniques by the medical community and ...regulatory agencies.
To compare several cross-validation (CV) approaches to evaluate the performance of a classifier for automatic grading of prostate cancer in digitized histopathologic images and compare the performance of the classifier when trained using data from 1 expert and multiple experts.
This quality improvement study used tissue microarray data (333 cores) from 231 patients who underwent radical prostatectomy at the Vancouver General Hospital between June 27, 1997, and June 7, 2011. Digitized images of tissue cores were annotated by 6 pathologists for 4 classes (benign and Gleason grades 3, 4, and 5) between December 12, 2016, and October 5, 2017. Patches of 192 µm2 were extracted from these images. There was no overlap between patches. A deep learning classifier based on convolutional neural networks was trained to predict a class label from among the 4 classes (benign and Gleason grades 3, 4, and 5) for each image patch. The classification performance was evaluated in leave-patches-out CV, leave-cores-out CV, and leave-patients-out 20-fold CV. The analysis was performed between November 15, 2018, and January 1, 2019.
The classifier performance was evaluated by its accuracy, sensitivity, and specificity in detection of cancer (benign vs cancer) and in low-grade vs high-grade differentiation (Gleason grade 3 vs grades 4-5). The statistical significance analysis was performed using the McNemar test. The agreement level between pathologists and the classifier was quantified using a quadratic-weighted κ statistic.
On 333 tissue microarray cores from 231 participants with prostate cancer (mean SD age, 63.2 6.3 years), 20-fold leave-patches-out CV resulted in mean (SD) accuracy of 97.8% (1.2%), sensitivity of 98.5% (1.0%), and specificity of 97.5% (1.2%) for classifying benign patches vs cancerous patches. By contrast, 20-fold leave-patients-out CV resulted in mean (SD) accuracy of 85.8% (4.3%), sensitivity of 86.3% (4.1%), and specificity of 85.5% (7.2%). Similarly, 20-fold leave-cores-out CV resulted in mean (SD) accuracy of 86.7% (3.7%), sensitivity of 87.2% (4.0%), and specificity of 87.7% (5.5%). Results of McNemar tests showed that the leave-patches-out CV accuracy, sensitivity, and specificity were significantly higher than those for both leave-patients-out CV and leave-cores-out CV. Similar results were observed for classifying low-grade cancer vs high-grade cancer. When trained on a single expert, the overall agreement in grading between pathologists and the classifier ranged from 0.38 to 0.58; when trained using the majority vote among all experts, it was 0.60.
Results of this study suggest that in prostate cancer classification from histopathologic images, patch-wise CV and single-expert training and evaluation may lead to a biased estimation of classifier's performance. To allow reproducibility and facilitate comparison between automatic classification methods, studies in the field should evaluate their performance using patient-based CV and multiexpert data. Some of these conclusions may be generalizable to other histopathologic applications and to other applications of machine learning in medicine.
Leiomyosarcomas are malignant smooth muscle tumors that occur most commonly in the gynecologic tract and soft tissue. There are different diagnostic criteria of malignancy for smooth muscle tumors ...arising at gynecologic and soft tissue sites and they may be managed differently but determining the primary site of a smooth muscle tumor can be difficult in some cases. In addition, the distinction between malignant and benign gynecologic tract smooth muscle tumors on morphologic grounds can be challenging. Using a series of tissue microarrays that contain 245 cases of leiomyosarcomas (102 gynecologic) with survival data, and 49 cases of uterine leiomyoma, we examined the ability of selected immune-markers (estrogen receptor (ER) and WT1) to distinguish between leiomyosarcomas of gynecologic and nongynecologic origin. In addition, we examined whether immunostains for p16, p53 and Ki-67 could distinguish between malignant and benign gynecologic smooth muscle tumors. ER nuclear positivity was observed in 3 and 50% of the nongynecologic and gynecologic leiomyosarcomas, respectively (P<0.001). Nuclear WT1 positivity was seen in 0 and 8% of the nongynecologic and gynecologic leiomyosarcomas, respectively (P<0.001). 87% of primary gynecologic leiomyosarcomas and 2% of uterine leiomyomas showed diffuse (>or=50% of cells) p16 staining (P<0.001). 23% of gynecologic leiomyosarcomas showed p53 immunopositivity (>or=50% of cells) whereas none of the leiomyomas were positive for p53 (P<0.001). 65% of the gynecologic leiomyosarcomas and 0% of the leiomyomas exhibited >10% Ki-67 proliferation index (P<0.001). Diffuse p16 and p53 immunopositivity and high Ki-67 proliferation index, singly or in combination, yielded an overall sensitivity of 92% and specificity of 98% for distinguishing between gynecologic leiomyosarcomas and leiomyomas and can be used as indicators of malignancy for gynecologic smooth muscle tumors. Although ER positivity can be used to support the gynecologic origin of a leiomyosarcomas, nuclear WT1 immunostaining is of little use.
Nuclear beta-catenin in mesenchymal tumors Ng, Tony L; Gown, Allen M; Barry, Todd S ...
Modern pathology,
January 2005, 2005-01-00, 2005-Jan, 20050101, Letnik:
18, Številka:
1
Journal Article
Recenzirano
Odprti dostop
β-Catenin is a crucial part of the Wnt and E-cadherin signalling pathways, which are involved in tumorigenesis. Dysregulation of these pathways allow β-catenin to accumulate and translocate to the ...nucleus, where it may activate oncogenes. Such nuclear accumulation can be detected by immunohistochemistry, which may be useful in diagnosis. Although the role of β-catenin has been established in various types of carcinomas, relatively little is known about its status in mesenchymal tumors. A number of studies suggest that β-catenin dysregulation is important in desmoid-type fibromatosis, as well as in synovial sarcoma. We wished to determine whether nuclear β-catenin expression is specific to and sensitive for particular bone and soft-tissue tumors, including sporadic desmoid-type fibromatosis. We studied the nuclear expression of β-catenin using tissue microarrays in a comprehensive range of bone and soft-tissue tumor types. A total of 549 cases were included in our panel. Nuclear immunohistochemical staining was determined to be either high level (>25% of cells), low level (0–25%) or none. High-level nuclear β-catenin staining was seen in a very limited subset of tumor types, including desmoid-type fibromatosis (71% of cases), solitary fibrous tumor (40%), endometrial stromal sarcoma (40%) and synovial sarcoma (28%). Although occasional cases of fibrosarcoma, clear cell sarcoma and carcinosarcoma had high-level staining, no high-level nuclear β-catenin expression was seen in any of 381 fibrohistocytic, muscular, adipocytic, chondroid or osseous tumor cases representing 42 diagnostic categories. All primary immunostain tissue microarray images are made publicly accessible in a searchable database. High-level nuclear β-catenin staining serves as a useful diagnostic tool, as it is specific to a small subset of mesenchymal tumors.