External validation of a deep learning-based algorithm for detection of tall cells in papillary thyroid carcinoma: A multicenter study.

The tall cell subtype (TC-PTC) is an aggressive subtype of papillary thyroid carcinoma (PTC). The TC-PTC is defined as a PTC comprising at least 30% epithelial cells that are three times as tall as they are wide. In practice, this definition is difficult to adhere to, resulting in high inter-observer variability. In this multicenter study, we validated a previously trained deep learning (DL)-based algorithm for detection of tall cells on 160 externally collected hematoxylin and eosin (HE)-stained PTC whole-slide images. In a test set of 360 manual annotations of regions of interest from 18 separate tissue sections in the external dataset, the DL-based algorithm detected TCs with a sensitivity of 90.6% and a specificity of 88.5%. The DL algorithm detected non-TC areas with a sensitivity of 81.6% and a specificity of 92.9%. In the validation datasets, 20% and 30% TC thresholds correlated with a significantly shorter relapse-free survival. In conclusion, the DL algorithm detected TCs in unseen, external scanned HE tissue slides with high sensitivity and specificity without any retraining.

NGS of brush cytology samples improves the detection of high-grade dysplasia and cholangiocarcinoma in patients with primary sclerosing cholangitis: A retrospective and prospective study.

BACKGROUND: Biliary dysplasia, a precursor of cholangiocarcinoma (CCA), is a common complication of primary sclerosing cholangitis. Patients with high-grade dysplasia (HGD) or early CCA who have received oncological treatment are candidates for liver transplantation. The preoperative diagnosis of CCA or HGD is challenging, and the sensitivity of biliary brush cytology (BC) is limited. METHODS: By using next-generation sequencing (NGS), we retrospectively analyzed archived tissue samples (n=62) obtained from explanted liver tissue and CCA samples to identify oncogenic mutations that occur during primary sclerosing cholangitis carcinogenesis. BC samples were prospectively collected from patients with primary sclerosing cholangitis (n=97) referred for endoscopic retrograde cholangiography to measure the diagnostic utility of NGS combined with BC compared with traditional cytology alone. RESULTS: Mutations in KRAS, GNAS, FLT3, RNF43, TP53, ATRX, and SMAD4 were detected in archived CCA or HGD samples. KRAS, GNAS, TP53, CDKN2A, FBXW7, BRAF, and ATM mutations were detected in prospectively collected brush samples from patients with histologically verified CCA or HGD. One patient with low-grade dysplasia in the explanted liver had KRAS and GNAS mutations in brush sample. No mutations were observed in brush samples or archived tissues in liver transplantation cases without biliary neoplasia. While KRAS mutations are common in biliary neoplasms, they were also observed in patients without biliary neoplasia during surveillance. CONCLUSIONS: In summary, NGS of BC samples increased the sensitivity of detecting biliary neoplasia compared with traditional cytology. Performing NGS on BC samples may help diagnose HGD or early CCA, benefiting the timing of liver transplantation.

Protocol liver biopsy predicts graft survival after liver transplantation.

BACKGROUND: The use of protocol liver biopsy to monitor liver allograft status remains controversial. There is limited data from modern transplantation populations that includes protocol biopsies to evaluate its value in predicting clinical outcomes. METHODS: All protocol liver biopsies were identified from 875 patients who underwent liver transplantation at Helsinki University Hospital between 2000 and 2019. Each histologic component was analyzed for its ability to predict long-term outcomes, especially graft survival. We determined the frequency of significant biopsy findings based on the Banff working group definition. Liver function tests (LFTs) and clinical markers were evaluated for their ability to predict significant biopsy findings. RESULTS: In total, 867 protocol liver biopsies were analyzed. Significant findings were identified in 20.1% of the biopsies. In the first protocol biopsy, steatohepatitis (hazard ratio [HR] 3.504, p = .03) and moderate or severe congestion (HR 3.338, p = .04) predicted graft loss. The presence of cholangitis (HR 2.563, p = .04), necrosis (HR 7.635, p < .001), mild congestion (HR 4.291, p = .009), and significant biopsy finding (HR 2.540, p = .02) predicted inferior death-censored graft survival. While the degree of elevation of LFTs was positively associated with significant biopsy findings, the discrimination was poor (AUC .572-.622). Combined LFTs and clinical risk factors remained suboptimal for discriminating significant biopsy findings (AUC .696). CONCLUSIONS: Our findings support the use of protocol liver biopsies after liver transplantation since they frequently revealed changes associated with long-term outcomes, even when LFTs were normal.

Marked difference in liver fat measured by histology vs. magnetic resonance-proton density fat fraction: A meta-analysis.

Background & Aims: Pathologists quantify liver steatosis as the fraction of lipid droplet-containing hepatocytes out of all hepatocytes, whereas the magnetic resonance-determined proton density fat fraction (PDFF) reflects the tissue triacylglycerol concentration. We investigated the linearity, agreement, and correspondence thresholds between histological steatosis and PDFF across the full clinical spectrum of liver fat content associated with non-alcoholic fatty liver disease. Methods: Using individual patient-level measurements, we conducted a systematic review and meta-analysis of studies comparing histological steatosis with PDFF determined by magnetic resonance spectroscopy or imaging in adults with suspected non-alcoholic fatty liver disease. Linearity was assessed by meta-analysis of correlation coefficients and by linear mixed modelling of pooled data, agreement by Bland-Altman analysis, and thresholds by receiver operating characteristic analysis. To explain observed differences between the methods, we used RNA-seq to determine the fraction of hepatocytes in human liver biopsies. Results: Eligible studies numbered 9 (N = 597). The relationship between PDFF and histology was predominantly linear (r = 0.85 [95% CI, 0.80-0.89]), and their values approximately coincided at 5% steatosis. Above 5% and towards higher levels of steatosis, absolute values of the methods diverged markedly, with histology exceeding PDFF by up to 3.4-fold. On average, 100% histological steatosis corresponded to a PDFF of 33.0% (29.5-36.7%). Targeting at a specificity of 90%, optimal PDFF thresholds to predict histological steatosis grades were ≥5.75% for ≥S1, ≥15.50% for ≥S2, and ≥21.35% for S3. Hepatocytes comprised 58 ± 5% of liver cells, which may partly explain the lower values of PDFF vs. histology. Conclusions: Histological steatosis and PDFF have non-perfect linearity and fundamentally different scales of measurement. Liver fat values obtained using these methods may be rendered comparable by conversion equations or threshold values. Impact and implications: Magnetic resonance-proton density fat fraction (PDFF) is increasingly being used to measure liver fat in place of the invasive liver biopsy. Understanding the relationship between PDFF and histological steatosis fraction is important for preventing misjudgement of clinical status or treatment effects in patient care. Our analysis revealed that histological steatosis fraction is often significantly higher than PDFF, and their association varies across the spectrum of fatty liver severity. These findings are particularly important for physicians and clinical researchers, who may use these data to interpret PDFF measurements in the context of histologically evaluated liver fat content.

Deep learning quantification reveals a fundamental prognostic role for ductular reaction in biliary atresia.

BACKGROUND: We aimed to quantify ductular reaction (DR) in biliary atresia using a neural network in relation to underlying pathophysiology and prognosis. METHODS: Image-processing neural network model was applied to 259 cytokeratin-7-stained native liver biopsies of patients with biliary atresia and 43 controls. The model quantified total proportional DR (DR%) composed of portal biliary epithelium (BE%) and parenchymal intermediate hepatocytes (PIH%). The results were related to clinical data, Sirius Red-quantified liver fibrosis, serum biomarkers, and bile acids. RESULTS: In total, 2 biliary atresia biopsies were obtained preoperatively, 116 at Kasai portoenterostomy (KPE) and 141 during post-KPE follow-up. DR% (8.3% vs. 5.9%, p=0.045) and PIH% (1.3% vs. 0.6%, p=0.004) were increased at KPE in patients remaining cholestatic postoperatively. After KPE, patients with subsequent liver transplantation or death showed an increase in DR% (7.9%-9.9%, p = 0.04) and PIH% (1.6%-2.4%, p = 0.009), whereas patients with native liver survival (NLS) showed decreasing BE% (5.5%-3.0%, p = 0.03) and persistently low PIH% (0.9% vs. 1.3%, p = 0.11). In Cox regression, high DR predicted inferior NLS both at KPE [DR% (HR = 1.05, p = 0.01), BE% (HR = 1.05, p = 0.03), and PIH% (HR = 1.13, p = 0.005)] and during follow-up [DR% (HR = 1.08, p<0.0001), BE% (HR = 1.58, p = 0.001), and PIH% (HR = 1.04, p = 0.008)]. DR% correlated with Sirius red-quantified liver fibrosis at KPE (R = 0.47, p<0.0001) and follow-up (R = 0.27, p = 0.004). A close association between DR% and serum bile acids was observed at follow-up (R = 0.61, p<0.001). Liver fibrosis was not prognostic for NLS at KPE (HR = 1.00, p = 0.96) or follow-up (HR = 1.01, p = 0.29). CONCLUSIONS: DR predicted NLS in different disease stages before transplantation while associating with serum bile acids after KPE.

Immunohistochemical Glucagon-like Peptide-1 Receptor Expression in Human Insulinomas.

Insulinomas are rare functional pancreatic neuroendocrine tumours, which metastasize in 10% of cases. As predicting the prognosis can be challenging, there is a need for the determination of clinicopathological factors associated with metastatic potential. The aim of this study is to evaluate the glucagon-like peptide-1 receptor (GLP-1R) expression in insulinomas and to analyse its association with clinicopathological features and patient outcome. This retrospective study involves pancreatic tumour tissue samples from fifty-two insulinoma patients. After histological re-evaluation, formalin-fixed paraffin-embedded tissue samples were processed into tissue microarrays and stained immunohistochemically with a monoclonal GLP-1R antibody. Forty-eight of the forty-nine (98%) non-metastatic tumours expressed GLP-1R, while one non-metastatic, multiple endocrine neoplasia type 1 (MEN1)-related tumour and all three of the metastatic tumours lacked GLP-1R expression. The lack of GLP-1R expression was associated with impaired overall survival, larger tumour diameter, higher Ki-67 PI and weaker insulin staining. Somatostatin receptor 1-5 expression did not differ between GLP-1R-positive and GLP-1R-negative insulinomas. In conclusion, the lack of GLP-1R expression is associated with metastatic disease and impaired survival in insulinoma patients. Thus, GLP-1R expression could be a useful biomarker in estimating the metastatic potential of the tumour and the prognosis of surgically treated patients.

HSP90 expression is associated with outcome in pulmonary carcinoid tumor patients.

Background: Pulmonary carcinoids (PCs) are rare tumors that account for <2% of all lung cancer cases. Patients who undergo resection for PC tumors generally have a favorable prognosis, but there is a risk for late recurrence and distant metastasis. The objective of this study was to identify biomarkers for PC tumors using RNA sequencing and immunohistochemistry. Methods: A total of 128 formalin-fixed, paraffin-embedded PC tumor samples from patients surgically treated at Helsinki University Hospital between 1990 and 2013 were analyzed in the study. RNA sequencing was first used to detect genes with higher expression in specific histological subtypes and metastatic and nonmetastatic tumors than in adjacent lung tissue. The diagnostic potential of the biomarkers was assessed using immunohistochemistry. Results: Through gene expression analysis, HSP90AB1 expression was found to be significantly elevated in metastatic PC tumors (P<0.0001). The paralog of the gene, HSP90AA1, was also overexpressed, but the finding was not statistically significant. Through immunohistochemical analysis, HSP90 protein expression was found to be associated with shorter disease-specific survival (DSS) (P=0.009) and increased risk of disease-specific death [hazard ratio (HR) 6.4, 95% confidence interval (CI): 1.3-31.8]. Conclusions: This study confirms that HSP90 has a prognostic role in PC tumors and that inhibition of HSP90 may possess therapeutic potential in the management of PC tumor patients in the future.

Somatostatin receptor expression in Merkel cell carcinoma: correlation with clinical data.

BACKGROUND: Merkel cell carcinoma (MCC) is a rare, high-grade neuroendocrine neoplasm (NEN) of the skin. Somatostatin receptors (SSTRs) are G protein-linked receptors that regulate cell proliferation and growth. SSTRs are expressed in many NENs; however, scant information is available on their expression in MCCs or their association with clinical parameters and patient outcomes. MATERIAL AND METHODS: This retrospective study was conducted at Helsinki University Hospital and the University of Helsinki. Using a tissue microarray, we investigated SSTR1-5 expression by immunohistochemistry in 99 MCC tissue samples. Samples were collected between 1983 and 2017 and coupled with the patients' clinical data. RESULTS: SSTR2-SSTR5 were detected in 69%, 6%, 4%, and 1% of the tumours, respectively. However, SSTR1 expression was not observed. Cytoplasmic SSTR2 positivity was associated with metastatic disease at the time of diagnosis (p = 0.009), but it did not correlate with disease-specificity or overall survival. CONCLUSION: SSTR2-5 expression was observed in MCCs. In particular, SSTR2 expression is clinically valid because it is associated with metastatic disease at the time of diagnosis and can thus serve as a prognostic marker. Moreover, SSTR2 overexpression provides a molecular basis for tumour imaging and treatment with somatostatin analogues.

The role of serology, liver function tests and imaging in screening of primary sclerosing cholangitis: the HelPSCreen score.

OBJECIVES: At present, no sensitive or specific screening test exists for primary sclerosing cholangitis (PSC). PSC screening is mainly based on elevated alkaline phosphatase (ALP) in patients with inflammatory bowel disease (IBD). We aimed to produce a screening score based on laboratory tests to predict the likelihood of PSC. Moreover, we evaluated the additional roles of liver histology and magnetic resonance cholangiopancreatography (MRCP) in the diagnosis of PSC. MATERIALS AND METHODS: The data of 385 patients who came for their first endoscopic retrograde cholangiography (ERC) to confirm PSC diagnosis were retrieved from the PSC registry of the Helsinki University Hospital. Overall, 69 patients referred for ERC with suspected PSC, in whom PSC was excluded by ERC or liver biopsy and MRCP, served as controls. We included patients' demographics and 13 laboratory test results in the analysis. Variables with significant odds ratios were selected for multivariate logistic regression, which was used to create a novel scoring system for PSC. The presence of IBD, serum perinuclear anti-neutrophil cytoplasmic antibodies, and ALP levels demonstrated the highest predictive value for PSC. A score was assigned for each statistically significant predictor. RESULTS: The optimal cut-off point for the score was ≥3, with an AUC of 0.83 (95%CI: 0.78-0.88). The addition of liver histology or MRCP findings to the score did not add a predictive value. CONCUSIONS: In conclusion, we created a novel, simple scoring system to screen the probability of PSC. The HelPSCreen-score may help to assess the disease prevalence and to target further investigations in patients suspected of PSC.

Histopathological Helsinki score of colorectal liver metastases predicts survival after liver resection.

Several perioperative scoring systems have been created to predict outcomes in metastatic colorectal cancer; however, these rarely include histological parameters. We evaluated histological factors used for patients with liver metastases operated between 2000 and 2019 and compared the results with the Fong score. Many scoring models for overall disease-free survival (DFS) were established and compared using multivariate Cox proportional hazard models. Statistically significant predictors at a 5% level in the univariate analysis were included in the multivariate models using the backward and forward selection methods. Per these models, we established a score of eight histological factors. We defined low-, intermediate-, and high-risk groups and compared them using the Kaplan-Meier survival and receiver operating characteristics (ROC) analyses. The histological score's accuracy was compared with the modified Fong clinical risk score. The following factors constituted the Helsinki score: advanced pT stage, node-positive primary, ≥2 metastases, size >50 mm, vitality >30%, margin <5 mm, vascular invasion and biliary invasion. The high-risk group had significantly worse DFS and overall survival. In ROC analyses, the Helsinki score was slightly better than the modified Fong clinical risk score. Helsinki score challenges physicians to acknowledge histological factors as important outcome measures.

Immunohistochemical somatostatin receptor expression in insulinomas.

Insulinomas are rare pancreatic neuroendocrine tumours. Most patients can be cured with surgery, but patients with a metastatic disease show impaired survival. The aim of this study was to evaluate somatostatin receptor (SSTR) 1-5 expression in insulinomas and to correlate the expression profile with clinicopathological variables and with patient outcome. This retrospective study involved 52 insulinoma patients. After histological re-evaluation, formalin-fixed paraffin-embedded tissue samples were processed into tissue microarrays and stained immunohistochemically with monoclonal SSTR1-5 antibodies. All the 52 tumours (49 non-metastatic, 3 metastatic) expressed at least one SSTR subtype. SSTR2 was expressed most frequently (71%), followed by SSTR3 (33%), SSTR1 (27%), SSTR5 (6%) and SSTR4 (0%). SSTR3 expression was associated with a larger tumour size (median diameter 19 mm vs. 13 mm, p = 0.043), and SSTR3 and SSTR5 expression were associated with impaired overall survival [HR 3.532 (95% CI 1.106-11,277), p = 0.033, and HR 6.805 (95% CI 1.364-33.955), p = 0.019 respectively]. Most insulinomas express SSTR2, which may be utilized in diagnostic imaging, and in planning individualized treatment strategies for insulinoma patients. Further studies are needed to clarify the association between SSTR profile and overall survival.

Inhibition of HSD17B13 protects against liver fibrosis by inhibition of pyrimidine catabolism in nonalcoholic steatohepatitis.

Nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease, in which prognosis is determined by liver fibrosis. A common variant in hydroxysteroid 17-beta dehydrogenase 13 (HSD17B13, rs72613567-A) is associated with a reduced risk of fibrosis in NAFLD, but the underlying mechanism(s) remains unclear. We investigated the effects of this variant in the human liver and in Hsd17b13 knockdown in mice by using a state-of-the-art metabolomics approach. We demonstrate that protection against liver fibrosis conferred by the HSD17B13 rs72613567-A variant in humans and by the Hsd17b13 knockdown in mice is associated with decreased pyrimidine catabolism at the level of dihydropyrimidine dehydrogenase. Furthermore, we show that hepatic pyrimidines are depleted in two distinct mouse models of NAFLD and that inhibition of pyrimidine catabolism by gimeracil phenocopies the HSD17B13-induced protection against liver fibrosis. Our data suggest pyrimidine catabolism as a therapeutic target against the development of liver fibrosis in NAFLD.

Impact on follow-up strategies in patients with primary sclerosing cholangitis.

BACKGROUND & AIMS: Evidence for the benefit of scheduled imaging for early detection of hepatobiliary malignancies in primary sclerosing cholangitis (PSC) is limited. We aimed to compare different follow-up strategies in PSC with the hypothesis that regular imaging improves survival. METHODS: We collected retrospective data from 2975 PSC patients from 27 centres. Patients were followed from the start of scheduled imaging or in case of clinical follow-up from 1 January 2000, until death or last clinical follow-up alive. The primary endpoint was all-cause mortality. RESULTS: A broad variety of different follow-up strategies were reported. All except one centre used regular imaging, ultrasound (US) and/or magnetic resonance imaging (MRI). Two centres used scheduled endoscopic retrograde cholangiopancreatography (ERCP) in addition to imaging for surveillance purposes. The overall HR (CI95%) for death, adjusted for sex, age and start year of follow-up, was 0.61 (0.47-0.80) for scheduled imaging with and without ERCP; 0.64 (0.48-0.86) for US/MRI and 0.53 (0.37-0.75) for follow-up strategies including scheduled ERCP. The lower risk of death remained for scheduled imaging with and without ERCP after adjustment for cholangiocarcinoma (CCA) or high-grade dysplasia as a time-dependent covariate, HR 0.57 (0.44-0.75). Hepatobiliary malignancy was diagnosed in 175 (5.9%) of the patients at 7.9 years of follow-up. Asymptomatic patients (25%) with CCA had better survival if scheduled imaging had been performed. CONCLUSIONS: Follow-up strategies vary considerably across centres. Scheduled imaging was associated with improved survival. Multiple factors may contribute to this result including early tumour detection and increased endoscopic treatment of asymptomatic benign biliary strictures.

Gamma-glutamyltransferase predicts macrovesicular liver graft steatosis - an analysis of discarded liver allografts in Finland.

OBJECTIVE: Liver-transplantation activity is limited by the shortage of grafts. Donor-liver macrovesicular steatosis predisposes to ischemia-reperfusion injury and is associated with reduced graft survival. The increasing prevalence of fatty-liver disease underlines the importance of identifying macrovesicular steatosis in potential donor livers. We analyzed liver grafts discarded for transplantation, and particularly the role of gamma-glutamyltransferase (GGT) in predicting graft steatosis. METHODS: One-hundred sixty rejected cadaveric-donor liver grafts were studied. Donor selection was based on clinical data, and macroscopic graft inspection. Discarded grafts were biopsied at procurement of non-liver organs. RESULTS: The most common reasons for discarding the graft were abnormal liver tests, ultrasound-verified steatosis and history of harmful alcohol use. GGT correlated moderately with macrovesicular steatosis (r = 0.52, p < 0.001), but poorly with microvesicular steatosis (r = 0.36, p < 0.001). Increased correlation between GGT and macrovesicular steatosis was observed among alcohol abusers (r = 0.67, p < 0.001). Area under the curve (AUC) of GGT for predicting >30% macrovesicular steatosis was 0.79 (95% CI 0.71-0.88), and for >60% steatosis, 0.79 (95% CI 0.68-0.90). The optimal GGT-cut off for detecting >30% and >60% macrovesicular steatosis were, respectively, 66 U/L (sensitivity 76% and specificity 68%) and 142 U/L (sensitivity 66% and specificity 83%). Among alcohol users, a GGT value >90 U/L showed 100% sensitivity for >60% macrovesicular steatosis. AUC for GGT in predicting fibrosis Stages 2-4 was 0.82 (95% CI 0.71-0.92, p < 0.001, optimal cut off 68, sensitivity 92%, specificity 61%). CONCLUSIONS: Abnormal liver values, steatosis and harmful alcohol use were the main reasons for discarding liver-graft offers in Finland. GGT proved useful in predicting moderate and severe liver graft macrovesicular steatosis.

Automated image analysis of keratin 7 staining can predict disease outcome in primary sclerosing cholangitis.

BACKGROUND AND AIMS: Primary sclerosing cholangitis (PSC) is a chronic cholestatic liver disease that obstructs the bile ducts and causes liver cirrhosis and cholangiocarcinoma. Efficient surrogate markers are required to measure disease progression. The cytokeratin 7 (K7) load in a liver specimen is an independent prognostic indicator that can be measured from digitalized slides using artificial intelligence (AI)-based models. METHODS: A K7-AI model 2.0 was built to measure the hepatocellular K7 load area of the parenchyma, portal tracts, and biliary epithelium. K7-stained PSC liver biopsy specimens (n = 295) were analyzed. A compound endpoint (liver transplantation, liver-related death, and cholangiocarcinoma) was applied in Kaplan-Meier survival analysis to measure AUC values and positive likelihood ratios for each histological variable detected by the model. RESULTS: The K7-AI model 2.0 was a better prognostic tool than plasma alkaline phosphatase, the fibrosis stage evaluated by Nakanuma classification, or K7 score evaluated by a pathologist based on the AUC values of measured variables. A combination of parameters, such as portal tract volume and area of K7-positive hepatocytes analyzed by the model, produced an AUC of 0.81 for predicting the compound endpoint. Portal tract volume measured by the model correlated with the histological fibrosis stage. CONCLUSIONS: The K7 staining of histological liver specimens in PSC provides significant information on disease outcomes through objective and reproducible data, including variables that cannot be measured by a human pathologist. The K7-AI model 2.0 could serve as a prognostic tool for clinical endpoints and as a surrogate marker in drug trials.

PNPLA3 allele frequency has no impact on biliary bile acid composition or disease course in patients with primary sclerosing cholangitis.

BACKGROUND AND AIMS: Primary sclerosing cholangitis (PSC) is a chronic inflammatory disease that leads to bile duct strictures, cholestasis, and biliary cirrhosis. PNPLA3 (patatin-like phospholipase domain containing 3), regulates cellular lipid synthesis by converting lysophosphatidic acid into phosphatidic acid. Isoleucine mutation to methionine at position 148 (I148M) causes a loss of this function. Only two studies, with contradictory results, have evaluated the role of PNPLA3 in PSC. The rs738409(G) variant of PNPLA3 has been associated with an increased risk for transplantation in male patients with dominant strictures (DS). The study aimed to evaluate the PNPLA3 allele frequency effect on the clinical outcomes, progression, and prognosis of PSC. Furthermore, we analyzed the impact of PNPLA3 on phospholipid and bile acid composition to evaluate the effect of the PNPLA3 status on UDCA response. PATIENTS AND METHODS: We recruited 560 patients prospectively and collected clinical and laboratory data as well as liver histology and imaging findings. PNPLA3 (CC, CG, GG) alleles were analyzed with TaqManTM. We also analyzed bile acids (BA), cholesterol and phospholipids and individual BA from a sample aspirated during endoscopic retrograde cholangiography (ERC). RESULTS: Among the recruited patients, 58.4%, 35.7% and 5.9% had the wild (CC), heterozygous (CG) and homozygous (GG) alleles, respectively. The PNPLA3 haplotype did not impact bile composition or individual BA. In addition, we found no differences in age at diagnosis, disease progression, liver fibrosis or survival between the cohorts. CONCLUSIONS: The PNPLA3 I148M variant had no significant impact on on bile composition, including UDCA content, clinical outcomes, progression of liver fibrosis, hepatobiliary cancer risk, liver transplantation, or overall survival.

A deep learning-based algorithm for tall cell detection in papillary thyroid carcinoma.

INTRODUCTION: According to the World Health Organization, the tall cell variant (TCV) is an aggressive subtype of papillary thyroid carcinoma (PTC) comprising at least 30% epithelial cells two to three times as tall as they are wide. In practice, applying this definition is difficult causing substantial interobserver variability. We aimed to train a deep learning algorithm to detect and quantify the proportion of tall cells (TCs) in PTC. METHODS: We trained the deep learning algorithm using supervised learning, testing it on an independent dataset, and further validating it on an independent set of 90 PTC samples from patients treated at the Hospital District of Helsinki and Uusimaa between 2003 and 2013. We compared the algorithm-based TC percentage to the independent scoring by a human investigator and how those scorings associated with disease outcomes. Additionally, we assessed the TC score in 71 local and distant tumor relapse samples from patients with aggressive disease. RESULTS: In the test set, the deep learning algorithm detected TCs with a sensitivity of 93.7% and a specificity of 94.5%, whereas the sensitivity fell to 90.9% and specificity to 94.1% for non-TC areas. In the validation set, the deep learning algorithm TC scores correlated with a diminished relapse-free survival using cutoff points of 10% (p = 0.044), 20% (p < 0.01), and 30% (p = 0.036). The visually assessed TC score did not statistically significantly predict survival at any of the analyzed cutoff points. We observed no statistically significant difference in the TC score between primary tumors and relapse tumors determined by the deep learning algorithm or visually. CONCLUSIONS: We present a novel deep learning-based algorithm to detect tall cells, showing that a high deep learning-based TC score represents a statistically significant predictor of less favorable relapse-free survival in PTC.

Quantitative bile and serum proteomics for the screening and differential diagnosis of primary sclerosing cholangitis.

BACKGROUND: Primary sclerosing cholangitis (PSC) is a chronic liver disease characterized by biliary strictures, cholestasis, and a markedly increased risk of cholangiocarcinoma. New markers for the screening and differential diagnosis of PSC are needed. In this pilot study, we have analyzed both the bile and serum proteomic profiles of 80 PSC patients and non-PSC controls (n = 6 for bile and n = 18 for serum). AIM: The aim of this study was to discover candidates for new biomarkers for the differential diagnosis of PSC. METHODS: Bile and serum samples were processed and subsequently analyzed using ultra performance liquid chromatography-ultra definition mass spectrometry (UPLC-UDMSE). Further analysis included statistical analyses such as receiver operating characteristic curve analysis as well as pathway analysis using Ingenuity Pathway Analysis. RESULTS AND CONCLUSIONS: In bile, we discovered 64 proteins with significantly different levels between the groups, with fold changes of up to 129. In serum, we discovered 112 proteins with significantly different levels. Receiver operating characteristic curve analysis found multiple proteins with high area under the curve values, up to 0.942, indicating that these serum proteins are of value as new non-invasive classifiers of PSC. Pathway analysis revealed multiple canonical pathways that were enriched in the dataset, which have roles in bile homeostasis and metabolism. We present several serum proteins that could serve as new blood-based markers for the diagnosis of PSC after further validation. The measurement of serum levels of these proteins could be of use in the screening of patients with suspected PSC.

Can dysplasia surveillance be better targeted in ulcerative colitis by using faecal calprotectin?

Background: In the inflammatory bowel diseases, chronic inflammation predisposes to dysplasia and colorectal carcinoma, leading to the need of surveillance colonoscopies. The most-used marker of colonic inflammation is faecal calprotectin. Its correlation with endoscopic and histological findings is well-documented. In this study, we evaluated the role of sequential faecal calprotectin measurements in predicting colorectal dysplasia, to identify patients with increased risk of dysplasia or colonic malignancy in ulcerative colitis.Methods: We collected the faecal calprotectin measurements and colorectal histology reports of patients with ulcerative colitis treated in Helsinki University Hospital (Helsinki, Finland) between 2007 and 2017, with a focus on IBD-associated neoplasia, inflammatory activity, and sporadic adenomas. Using the time-weighted AUC of faecal calprotectin as a marker of inflammatory burden, we tested the performance of faecal calprotectin to predict the risk for colorectal neoplasia.Results: In total, 982 patients with ulcerative colitis were included. Of them, 845 had pancolitis and 127 concomitant primary sclerosing cholangitis. Forty-one patients (4%) had IBD-associated colorectal dysplasia and seven (0.7%) developed adenocarcinoma. In patients with constantly elevated faecal calprotectin level (>500 µg/g), colorectal neoplasia was more frequent compared to those with low (<200 µg/g) calprotectin (13% and 4%, p < 0.05). Histological inflammatory activity was also related to more frequent dysplastic changes.Conclusions: Colon dysplasia and adenocarcinoma are more common among ulcerative colitis patients with constantly elevated faecal calprotectin than in patients in remission. The role of inflammatory activity in inducing neoplastic changes in colon is further supported by histology, as histological inflammatory activity correlates with dysplasia.