Machine Learning of Bone Marrow Histopathology Identifies Genetic and Clinical Determinants in Patients with MDS.

In myelodysplastic syndrome (MDS) and myeloproliferative neoplasm (MPN), bone marrow (BM) histopathology is assessed to identify dysplastic cellular morphology, cellularity, and blast excess. Yet, other morphologic findings may elude the human eye. We used convolutional neural networks to extract morphologic features from 236 MDS, 87 MDS/MPN, and 11 control BM biopsies. These features predicted genetic and cytogenetic aberrations, prognosis, age, and gender in multivariate regression models. Highest prediction accuracy was found for TET2 [area under the receiver operating curve (AUROC) = 0.94] and spliceosome mutations (0.89) and chromosome 7 monosomy (0.89). Mutation prediction probability correlated with variant allele frequency and number of affected genes per pathway, demonstrating the algorithms' ability to identify relevant morphologic patterns. By converting regression models to texture and cellular composition, we reproduced the classical del(5q) MDS morphology consisting of hypolobulated megakaryocytes. In summary, this study highlights the potential of linking deep BM histopathology with genetics and clinical variables. SIGNIFICANCE: Histopathology is elementary in the diagnostics of patients with MDS, but its high-dimensional data are underused. By elucidating the association of morphologic features with clinical variables and molecular genetics, this study highlights the vast potential of convolutional neural networks in understanding MDS pathology and how genetics is reflected in BM morphology. See related commentary by Elemento, p. 195.

Assessment of Lifestyle Factors Helps to Identify Liver Fibrosis Due to Non-Alcoholic Fatty Liver Disease in Obesity.

Only some individuals with obesity develop liver fibrosis due to non-alcoholic fatty liver disease (NAFLD-fibrosis). We determined whether detailed assessment of lifestyle factors in addition to physical, biochemical and genetic factors helps in identification of these patients. A total of 100 patients with obesity (mean BMI 40.0 ± 0.6 kg/m2) referred for bariatric surgery at the Helsinki University Hospital underwent a liver biopsy to evaluate liver histology. Physical activity was determined by accelerometer recordings and by the Modifiable Activity Questionnaire, diet by the FINRISK Food Frequency Questionnaire, and other lifestyle factors, such as sleep patterns and smoking, by face-to-face interviews. Physical and biochemical parameters and genetic risk score (GRS based on variants in PNPLA3, TM6SF2, MBOAT7 and HSD17B13) were measured. Of all participants 49% had NAFLD-fibrosis. Independent predictors of NAFLD-fibrosis were low moderate-to-vigorous physical activity, high red meat intake, low carbohydrate intake, smoking, HbA1c, triglycerides and GRS. A model including these factors (areas under the receiver operating characteristics curve (AUROC) 0.90 (95% CI 0.84-0.96)) identified NAFLD-fibrosis significantly more accurately than a model including all but lifestyle factors (AUROC 0.82 (95% CI 0.73-0.91)) or models including lifestyle, physical and biochemical, or genetic factors alone. Assessment of lifestyle parameters in addition to physical, biochemical and genetic factors helps to identify obese patients with NAFLD-fibrosis.

Natural Course of Nonalcoholic Fatty Liver Disease and Type 2 Diabetes in Patients With Human Immunodeficiency Virus With and Without Combination Antiretroviral Therapy-associated Lipodystrophy: A 16-Year Follow-up Study.

BACKGROUND: Abnormal glucose metabolism and nonalcoholic fatty liver disease (NAFLD) are common in patients with human immunodeficiency virus (HIV+ patients), but longitudinal data are lacking. We determined the natural course of NAFLD (liver fat [LFAT]) and type 2 diabetes mellitus (T2DM) in HIV+ patients with and without lipodystrophy (LD+ and LD-, respectively) during a 16-year longitudinal study. METHODS: LFAT (by proton magnetic resonance spectroscopy) and clinical characteristics were measured in 41 HIV+ patients at baseline and after 16 years. Liver fibrosis was estimated by measuring liver stiffness using transient elastography (TE) and magnetic resonance elastography (MRE) at 16 years. We also longitudinally studied 28 healthy subjects. RESULTS: During follow-up, the HIV+ patients gained more body fat (8.6% ± 0.7%) than the control patients (4.5% ± 0.6%, P < .001). Features of insulin resistance increased significantly in the HIV+ patients but not the control patients. A significant proportion (20%, P < .01 vs 0% at baseline) of the HIV+ but none of the control patients developed T2DM. LFAT was significantly higher at baseline in the LD+ (4.3 [1.9-11.8]) than the LD- (1.0 [0.5-1.5]; P < .001) HIV+ patients. LFAT remained stable during follow-up in all groups. At follow-up, liver stiffness measured with TE was similar among all HIV, LD+, LD-, and control patients and between the LD+ and LD- patients measured with MRE. Advanced fibrosis by MRE was observed in 3 of LD+ and none of LD- patients. CONCLUSIONS: During 16 years of follow-up, progression of NAFLD is rare compared to development of T2DM in HIV+ patients.

Human PNPLA3-I148M variant increases hepatic retention of polyunsaturated fatty acids.

The common patatin-like phospholipase domain-containing protein 3 (PNPLA3) variant I148M predisposes to nonalcoholic liver disease but not its metabolic sequelae. We compared the handling of labeled polyunsaturated fatty acids (PUFAs) and saturated fatty acids (SFA) in vivo in humans and in cells harboring different PNPLA3 genotypes. In 148M homozygous individuals, triglycerides (TGs) in very low-density lipoproteins (VLDL) were depleted of PUFAs both under fasting and postprandial conditions compared with 148I homozygotes, and the PUFA/SFA ratio in VLDL-TGs was lower relative to the chylomicron precursor pool. In human PNPLA3-148M and PNPLA3-KO cells, PUFA but not SFA incorporation into TGs was increased at the expense of phosphatidylcholines, and under lipolytic conditions, PUFA-containing diacylglycerols (DAGs) accumulated compared with PNPLA3-148I cells. Polyunsaturated TGs were increased, while phosphatidylcholines (PCs) were decreased in the human liver in 148M homozygous individuals as compared with 148I homozygotes. We conclude that human PNPLA3-I148M is a loss-of-function allele that remodels liver TGs in a polyunsaturated direction by impairing hydrolysis/transacylation of PUFAs from DAGs to feed phosphatidylcholine synthesis.