Democratizing Artificial Intelligence in Anatomic Pathology.

CONTEXT.­: Artificial intelligence is a transforming technology for anatomic pathology. Involvement within the workforce will foster support for algorithm development and implementation. OBJECTIVE.­: To develop a supportive ecosystem that enables pathologists with variable expertise in artificial intelligence to create algorithms in a development environment with seamless transition to a production environment. RESULTS.­: The development team considered internal development and vended solutions. Because of the extended timeline and resource requirements for internal development, a decision was made to use a vended solution. Vendor proposals were solicited and reviewed by pathologists, IT, and security groups. A vendor was selected and pipelines for development and production were established. Proposals for development were solicited from the pathology department. Eighty-four investigators were selected for the initial cohort, receiving training and access to dedicated subject matter experts. A total of 30 of 31 projects progressed through the model development process of annotating, training, and validation. Based on these projects, 15 abstracts were submitted to national meetings. CONCLUSIONS.­: Democratizing artificial intelligence by creating an ecosystem to support pathologists with varying levels of expertise can break down entry barriers, reduce overall cost of algorithm development, improve algorithm quality, and enhance the speed of adoption.

Phase I trial of adjuvant mature autologous dendritic cell/allogeneic tumor lysate vaccines in combination with temozolomide in newly diagnosed glioblastoma.

Background: Glioblastoma (GBM) has poor prognosis despite aggressive treatment. Dendritic cell (DC) vaccines are promising, but widespread clinical use has not been achieved, possibly reflecting manufacturing issues of antigen choice and DC potency. We previously optimized vaccine manufacture utilizing allogeneic human GBM tumor cell lysate and potent, mature autologous DCs. Here, we report a phase I study using this optimized DC vaccine in combination with standard therapy. Methods: Following surgical resection and radiation with concurrent temozolomide (TMZ), newly diagnosed adult GBM patients received intradermal DC vaccines plus TMZ. Primary endpoints were safety and feasibility. Immune and treatment responses were recorded. Results: Twenty-one patients were enrolled in this study. One progressed between leukapheresis and vaccine manufacture. Twenty patients received treatment per protocol. Vaccine doses (≥15) were generated following a single leukapheresis for each patient. No dose-limiting vaccine toxicities were encountered. One patient had symptomatic, histologically proven pseudoprogression. Median progression-free survival was 9.7 months. Median overall survival was 19 months. Overall survival was 25% at 2 years and 10% at 4 years. One patient remains progression-free 5 years after enrollment. Specific CD8 T-cell responses for the tumor-associated antigen gp100 were seen post-vaccination. Patients entered the trial with a leukocyte deficit compared to healthy donors which partly normalized over the course of therapy. Conclusions: This vaccine platform is safe and highly feasible in combination with standard therapy for newly diagnosed patients. Imaging, histological, survival, and immunological data suggest a positive biological response to therapy that warrants further investigation.

Tao brush endometrial cytology is a sensitive diagnostic tool for cancer and hyperplasia among women presenting to clinic with abnormal uterine bleeding.

BACKGROUND: Abnormal uterine bleeding requires the investigation of the endometrium. Histology is typically used but there remains room for the improvement and use of cytology. METHODS: Women presenting for clinically indicated office endometrial biopsy were prospectively enrolled. Tao endometrial brushing and office endometrial biopsy were performed, and surgical procedure if clinically indicated. Tao brush cytology specimens were blindly reviewed by up to three pathologists, consensus obtained, and scored as: benign, atypical (favor benign), suspicious, positive for malignancy, or non-diagnostic. Cytology and histology were compared to surgical pathology to determine sensitivity, specificity, positive, and negative predictive values to detect AH (atypical hyperplasia) or EC (endometrial cancer). RESULTS: Clinical indications of 197 enrolled patients included postmenopausal bleeding (90, 45.7%), abnormal uterine bleeding (94, 47.7%), and abnormal endometrium on ultrasound without bleeding (13, 6.6%). Of the 197 patients, 185 (93.9%) had cytology score consensus and a total of 196 (99.5%) had consensus regarding cytology positivity. Surgical pathology diagnoses (N = 85) were 13 (15.3%) FIGO grade 1 or 2 EC, 3 (3.5%) AH, and 69 (81.2%) benign endometrium. Sensitivity and specificity to detect EC or AH were 93.7% and 100%, respectively, via endometrial biopsy; 87.5% and 63.8%, respectively, via endometrial cytology when scores of malignancy, suspicious, or atypical were considered positive. CONCLUSIONS: In a high-risk population, Tao brush endometrial cytology showed high sensitivity to detect AH and EC comparable to biopsy histology when considering scores of malignancy, suspicious, atypical, and non-diagnostic. Revisiting the potential value of endometrial cytology in the contemporary era of endometrial diagnostic workup is warranted.

Ki-67 "hot spot" digital analysis is useful in the distinction of hepatic adenomas and well-differentiated hepatocellular carcinomas.

This study aims to investigate the utility of digital protocols for Ki-67 immunohistochemistry quantitative analysis ("hot spot" method) in the setting of well-differentiated hepatocellular neoplasms. Resection cases of typical hepatic adenomas (HAs) (n = 40), atypical HAs (n = 9), and well-differentiated hepatocellular carcinomas (WD HCCs) (n = 56) were selected. HAs were further classified by immunohistochemistry using antibodies against liver fatty acid binding protein, glutamine synthetase, B-catenin, hepatic serum amyloid A, and C-reactive protein. Ki-67 proliferative index by immunohistochemistry was evaluated in all cases by digital analysis using a modified neuroendocrine tumor "hot spot" protocol. The proliferative rate of HAs (typical, median 1.2% (range 0-7.4%) and atypical, median 1.0% (range 0.3-3%)) was significantly lower than that of WD HCCs (median 4.5%, range 0-49.8%) (P < 0.0001). Only a few (7.5%) of the adenomas (all inflammatory/telangiectatic type) had proliferative rates higher than 4%, compared to most (51%) of HCCs. Ki-67 is a potentially useful adjunct marker in the evaluation of WD hepatocellular neoplasms, as "hot spot" proliferative rates are consistently very low in HAs but vary significantly in WD HCCs.

Ki-67 Labeling Index in Pulmonary Carcinoid Tumors: Comparison Between Small Biopsy and Resection Using Tumor Tracing and Hot Spot Methods.

CONTEXT.­: Pulmonary carcinoids are classified as typical or atypical by assessing necrosis and mitoses, which usually cannot be adequately assessed on small biopsies. Ki-67 is not currently used to grade pulmonary carcinoids, but it may be helpful to determine preliminary grade in biopsies. However, the rate at which Ki-67 could underestimate or overestimate grade on small biopsies has not been well studied. OBJECTIVE.­: To compare Ki-67 labeling obtained on small biopsies to subsequent resection. DESIGN.­: Ki-67 was performed on paired biopsy and resection specimens from 55 patients. Slides were scanned using Aperio ScanScope. Labeling index was determined using automated hot spot and tumor tracing methods. RESULTS.­: The study included 41 typical and 14 atypical carcinoids. Atypical carcinoids were larger and had more distant metastases. Death from disease occurred in 3 patients (all had atypical carcinoids). Median hot spot Ki-67 labeling index was greater in resection compared with biopsy by 0.7% (P = .02). Median tumor tracing Ki-67 was lower in resection compared with biopsy by 0.5% (P < .001). Receiver-operating characteristic analysis showed similar hot spot Ki-67 cutoffs to predict atypical histology (3.5% for biopsy, 3.6% for resection; area under the curve [AUC], 0.75 and 0.74, respectively). Different optimal cutoffs were needed for tracing method based on biopsy (2.1%; AUC, 0.75) compared with resection (1.0%; AUC, 0.67). CONCLUSIONS.­: Hot spot Ki-67 tends to underestimate grade on small biopsies, whereas grade is overestimated by tumor tracing. Hot spot Ki-67 cutoff of 3.5% predicted atypical histology for both biopsy and resection. Different biopsy and resection cutoffs were necessary for tumor tracing, which would make clinical implementation more difficult.

Macrophage Density Predicts Facial Nerve Outcome and Tumor Growth after Subtotal Resection of Vestibular Schwannoma.

Introduction Vestibular schwannoma (VS) behavior following subtotal resection (STR) is highly variable. Overall progression rates have been reported as high as 44%, and optimal treatment is controversial. Correspondingly, identification of a reliable clinical or pathologic marker associated with progression after STR would help guide decision-making. Methods A prospectively maintained institutional VS registry from 1999 to 2014 was retrospectively reviewed for sporadic VS patients who underwent primary STR without preceding stereotactic radiosurgery (SRS) by a single neurosurgery-neurotology team. Primary endpoints included tumor progression and postoperative facial nerve function. Pathologic specimens were stained for Ki67, CD68, S100, and SOX10 and were quantitated by digital imaging analysis. Macrophage density was defined as the ratio of CD68 + macrophages to S100 + macrophages and Schwannian tumor cells. Clinical outcomes were correlated with pathologic markers. Results Forty-six patients met the study inclusion criteria. Thirteen (28%) progressed during a mean 57 months of follow-up (range 15-149). Favorable postoperative facial nerve function (House-Brackmann I-II) was achieved in 37 (80%). CD68 + cells were present at significantly higher concentrations in tumors that progressed ( p = 0.03). Higher macrophage density was significantly associated with both tumor progression ( p = 0.02) and unfavorable facial nerve function ( p = 0.02). Ki67 percent positivity was not significantly associated with either primary endpoint ( p = 0.83; p = 0.58). Conclusions Macrophage density may provide an important marker for individuals at the highest risk for progression of VS after STR, potentially prompting closer surveillance or consideration for upfront SRS following STR. This finding supports preceding conclusions that an intratumoral macrophage-predominant inflammatory response may be a marker for tumor growth and a potential therapeutic target.

Comparison of Three Ki-67 Index Quantification Methods and Clinical Significance in Pancreatic Neuroendocrine Tumors.

The Ki-67 index is essential in the pathological reports for pancreatic neuroendocrine tumors. There are three methods to determine the Ki-67 index including eyeball estimation, manual counting, or automated digital imaging analysis. The goal of this study was to compare the three quantification methods with the clinical outcome to determine the best method for clinical practice. Ki-67 immunostaining was performed on 97 resected pancreatic neuroendocrine tumors. The three methods of quantification were employed: (1) an average of eyeball estimation by three pathologists; (2) manual counting of at least 500 tumor cells; and (3) digital imaging analysis quantitation by selecting 8-10 hot spot regions. All tumors were graded according to the 2010 WHO grading system. The three quantification methods for the Ki-67 index had almost perfect agreement. The concordance between manual counting and digital imaging analysis and between manual counting and average eyeball estimation were 0.97 and 0.88, respectively. The concordance among the three pathologists' eyeball estimation was 0.86. All three methods correlated with patients' survival using the 2010 WHO grading system. Eyeball estimation scores were significantly less than those of the other two methods and tended to downgrade more tumors to grade 1, but they had higher predictive ability for survival and recurrence. The WHO system using the mitotic rate could also separate patients with different survival and even downgraded more tumors to grade 1. The results suggest the necessity of a consensus among pathologists for the method to determine the Ki-67 index and proper cutoff of the Ki-67 index for better clinical correlation.

Detection of endometrial cancer via molecular analysis of DNA collected with vaginal tampons.

OBJECTIVE: We demonstrate the feasibility of detecting EC by combining minimally-invasive specimen collection techniques with sensitive molecular testing. METHODS: Prior to hysterectomy for EC or benign indications, women collected vaginal pool samples with intravaginal tampons and underwent endometrial brushing. Specimens underwent pyrosequencing for DNA methylation of genes reported to be hypermethylated in gynecologic cancers and recently identified markers discovered by profiling over 200 ECs. Methylation was evaluated individually across CpGs and averaged across genes. Differences between EC and benign endometrium (BE) were assessed using two-sample t-tests and area under the curve (AUC). RESULTS: Thirty-eight ECs and 28 BEs were included. We evaluated 97 CpGs within 12 genes, including previously reported markers (RASSF1, HSP2A, HOXA9, CDH13, HAAO, and GTF2A1) and those identified in discovery work (ASCL2, HTR1B, NPY, HS3ST2, MME, ADCYAP1, and additional CDH13 CpG sites). Mean methylation was higher in tampon specimens from EC v. BE for 9 of 12 genes (ADCYAP1, ASCL2, CDH13, HS3ST2, HTR1B, MME, HAAO, HOXA9, and RASSF1) (all p<0.05). Among these genes, relative hypermethylation was observed in EC v. BE across CpGs. Endometrial brush and tampon results were similar. Within tampon specimens, AUC was highest for HTR1B (0.82), RASSF1 (0.75), and HOXA9 (0.74). This is the first report of HOXA9 hypermethylation in EC. CONCLUSION: DNA hypermethylation in EC tissues can also be identified in vaginal pool DNA collected via intravaginal tampon. Identification of additional EC biomarkers and refined collection methods are needed to develop an early detection tool for EC.