1. |
|
|
2. |
|
|
3. |
|
|
4. |
|
|
5. |
- Belavy, D. L., et al.
(author)
-
Characterization of Intervertebral Disc Changes in Asymptomatic Individuals with Distinct Physical Activity Histories Using Three Different Quantitative MRI Techniques
- 2020
-
In: Journal of Clinical Medicine. - : MDPI AG. - 2077-0383. ; 9:6
-
Journal article (peer-reviewed)abstract
- (1) Background: Assessments of intervertebral disc (IVD) changes, and IVD tissue adaptations due to physical activity, for example, remains challenging. Newer magnetic resonance imaging techniques can quantify detailed features of the IVD, where T2-mapping and T2-weighted (T2w) and Dixon imaging are potential candidates. Yet, their relative utility has not been examined. The performances of these techniques were investigated to characterize IVD differences in asymptomatic individuals with distinct physical activity histories. (2) Methods: In total, 101 participants (54 women) aged 25-35 years with distinct physical activity histories but without histories of spinal disease were included. T11/12 to L5/S1 IVDs were examined with sagittal T2-mapping, T2w and Dixon imaging. (3) Results: T2-mapping differentiated Pfirrmann grade-1 from all other grades (p< 0.001). Most importantly, T2-mapping was able to characterize IVD differences in individuals with different training histories (p< 0.005). Dixon displayed weak correlations with the Pfirrmann scale, but presented significantly higher water content in the IVDs of the long-distance runners (p< 0.005). (4) Conclusions: Findings suggested that T2-mapping best reflects IVD differences in asymptomatic individuals with distinct physical activity histories changes. Dixon characterized new aspects of IVD, probably associated with IVD hypertrophy. This complementary information may help us to better understand the biological function of the disc.
|
|
6. |
- Hebelka, Hanna, 1977, et al.
(author)
-
The importance of level stratification for quantitative MR studies of lumbar intervertebral discs: a cross-sectional analysis in 101 healthy adults
- 2019
-
In: European Spine Journal. - : Springer Science and Business Media LLC. - 0940-6719 .- 1432-0932. ; 28:9, s. 2153-2161
-
Journal article (peer-reviewed)abstract
- Purpose: To investigate whether quantitative T2-times depend on lumbar intervertebral disc (IVD) level. Methods: The lumbar spine (Th12/L1–L5/S1) of 101 participants (53.5% female, 30.0[± 3.6]years, 173.5[± 9.6]cm and 69.9[± 13.4]kg), without history of back pain, was examined on a 3T scanner with sagittal T2-mapping. All IVDs were stratified according to Pfirrmann grade and lumbar level, with mean T2-time determined for the entire IVD volume and in five subregions of interests. Results: Significant level-dependent T2-time differences were detected, both for the entire IVD volume and its subregions. For the entire IVD volume, Pfirrmann grade 2 IVDs displayed 9–18% higher T2-times in Th12/L1 IVDs compared to L2/L3–L5/S1 IVDs (0.001 > p < 0.004) and significantly different T2-times in L1/L2–L2/L3 IVDs compared to most of the IVDs in the lower lumbar spine. In Pfirrmann grades 1, 3 and 4 IVDs, no significant level-dependent T2-time differences were observed for the entire IVD. More pronounced results were observed when comparing IVD subregions, with significant level-dependent differences also within Pfirrmann grade 1 and grade 3 IVDs. For example, in posterior IVD subregions mean T2-time was 80–82% higher in Th12/L1 compared to L3/L4–L4/L5 Pfirrmann grade 1 IVDs (p < 0.05) and 10–14% higher in L5/S1 compared to L3/L4–L4/L5 Pfirrmann grade 3 IVDs (0.02 > p < 0.001). Discussion: Significant level-dependent T2-time differences within several Pfirrmann grades, both for the entire IVD volume and for multiple IVD subregions, were shown in this large cohort study. The T2-time differences between levels existed in both non-degenerated and degenerated IVDs. These findings show the importance of stratifying for lumbar level when quantitative IVD studies are performed using T2-mapping. Graphic abstract: These slides can be retrieved under Electronic Supplementary Material.[Figure not available: see fulltext.]. © 2019, The Author(s).
|
|
7. |
|
|
8. |
|
|
9. |
- Strom, Peter, et al.
(author)
-
Artificial intelligence for diagnosis and grading of prostate cancer in biopsies : a population-based, diagnostic study
- 2020
-
In: The Lancet Oncology. - : Elsevier. - 1470-2045 .- 1474-5488. ; 21:2, s. 222-232
-
Journal article (peer-reviewed)abstract
- BackgroundAn increasing volume of prostate biopsies and a worldwide shortage of urological pathologists puts a strain on pathology departments. Additionally, the high intra-observer and inter-observer variability in grading can result in overtreatment and undertreatment of prostate cancer. To alleviate these problems, we aimed to develop an artificial intelligence (AI) system with clinically acceptable accuracy for prostate cancer detection, localisation, and Gleason grading.MethodsWe digitised 6682 slides from needle core biopsies from 976 randomly selected participants aged 50–69 in the Swedish prospective and population-based STHLM3 diagnostic study done between May 28, 2012, and Dec 30, 2014 (ISRCTN84445406), and another 271 from 93 men from outside the study. The resulting images were used to train deep neural networks for assessment of prostate biopsies. The networks were evaluated by predicting the presence, extent, and Gleason grade of malignant tissue for an independent test dataset comprising 1631 biopsies from 246 men from STHLM3 and an external validation dataset of 330 biopsies from 73 men. We also evaluated grading performance on 87 biopsies individually graded by 23 experienced urological pathologists from the International Society of Urological Pathology. We assessed discriminatory performance by receiver operating characteristics and tumour extent predictions by correlating predicted cancer length against measurements by the reporting pathologist. We quantified the concordance between grades assigned by the AI system and the expert urological pathologists using Cohen's kappa.FindingsThe AI achieved an area under the receiver operating characteristics curve of 0·997 (95% CI 0·994–0·999) for distinguishing between benign (n=910) and malignant (n=721) biopsy cores on the independent test dataset and 0·986 (0·972–0·996) on the external validation dataset (benign n=108, malignant n=222). The correlation between cancer length predicted by the AI and assigned by the reporting pathologist was 0·96 (95% CI 0·95–0·97) for the independent test dataset and 0·87 (0·84–0·90) for the external validation dataset. For assigning Gleason grades, the AI achieved a mean pairwise kappa of 0·62, which was within the range of the corresponding values for the expert pathologists (0·60–0·73).InterpretationAn AI system can be trained to detect and grade cancer in prostate needle biopsy samples at a ranking comparable to that of international experts in prostate pathology. Clinical application could reduce pathology workload by reducing the assessment of benign biopsies and by automating the task of measuring cancer length in positive biopsy cores. An AI system with expert-level grading performance might contribute a second opinion, aid in standardising grading, and provide pathology expertise in parts of the world where it does not exist.
|
|
10. |
- Wang, M., et al.
(author)
-
Validation of risk stratification models in acute myeloid leukemia using sequencing-based molecular profiling
- 2017
-
In: Leukemia. - : NATURE PUBLISHING GROUP. - 0887-6924 .- 1476-5551. ; 31:10, s. 2029-2036
-
Journal article (peer-reviewed)abstract
- Risk stratification of acute myeloid leukemia (AML) patients needs improvement. Several AML risk classification models based on somatic mutations or gene-expression profiling have been proposed. However, systematic and independent validation of these models is required for future clinical implementation. We performed whole-transcriptome RNA-sequencing and panel-based deep DNA sequencing of 23 genes in 274 intensively treated AML patients (Clinseq-AML). We also utilized the The Cancer Genome Atlas (TCGA)-AML study (N = 142) as a second validation cohort. We evaluated six previously proposed molecular-based models for AML risk stratification and two revised risk classification systems combining molecular-and clinical data. Risk groups stratified by five out of six models showed different overall survival in cytogenetic normal-AML patients in the Clinseq-AML cohort (P-value < 0.05; concordance index > 40.5). Risk classification systems integrating mutational or gene-expression data were found to add prognostic value to the current European Leukemia Net (ELN) risk classification. The prognostic value varied between models and across cohorts, highlighting the importance of independent validation to establish evidence of efficacy and general applicability. All but one model replicated in the Clinseq-AML cohort, indicating the potential for molecular-based AML risk models. Risk classification based on a combination of molecular and clinical data holds promise for improved AML patient stratification in the future.
|
|
11. |
|
|
12. |
|
|
13. |
|
|
14. |
|
|
15. |
|
|
16. |
- Bylesjö, Max, et al.
(author)
-
Normalization and Closure
- 2009
-
In: Comprehensive Chemometrics. - AMSTERDAM : Elsevier. - 9780444527028 ; , s. A109-A127
-
Book chapter (other academic/artistic)
|
|
17. |
|
|
18. |
- Drong, Alexander W, et al.
(author)
-
The presence of methylation quantitative trait loci indicates a direct genetic influence on the level of DNA methylation in adipose tissue.
- 2013
-
In: PLOS ONE. - : Public Library of Science (PLoS). - 1932-6203. ; 8:2
-
Journal article (peer-reviewed)abstract
- Genetic variants that associate with DNA methylation at CpG sites (methylation quantitative trait loci, meQTLs) offer a potential biological mechanism of action for disease associated SNPs. We investigated whether meQTLs exist in abdominal subcutaneous adipose tissue (SAT) and if CpG methylation associates with metabolic syndrome (MetSyn) phenotypes. We profiled 27,718 genomic regions in abdominal SAT samples of 38 unrelated individuals using differential methylation hybridization (DMH) together with genotypes at 5,227,243 SNPs and expression of 17,209 mRNA transcripts. Validation and replication of significant meQTLs was pursued in an independent cohort of 181 female twins. We find that, at 5% false discovery rate, methylation levels of 149 DMH regions associate with at least one SNP in a ±500 kilobase cis-region in our primary study. We sought to validate 19 of these in the replication study and find that five of these significantly associate with the corresponding meQTL SNPs from the primary study. We find that none of the 149 meQTL top SNPs is a significant expression quantitative trait locus in our expression data, but we observed association between expression levels of two mRNA transcripts and cis-methylation status. Our results indicate that DNA CpG methylation in abdominal SAT is partly under genetic control. This study provides a starting point for future investigations of DNA methylation in adipose tissue.
|
|
19. |
- Fritz, Jesper, et al.
(author)
-
Influence of a School-based Physical Activity Intervention on Cortical Bone Mass Distribution : A 7-year Intervention Study
- 2016
-
In: Calcified Tissue International. - : Springer Science and Business Media LLC. - 0171-967X .- 1432-0827. ; 99:5, s. 443-453
-
Journal article (peer-reviewed)abstract
- Cortical bone mass and density varies across a bones length and cross section, and may be influenced by physical activity. This study evaluated the long-term effects of a pediatric school-based physical activity intervention on tibial cortical bone mass distribution. A total of 170 children (72 girls and 98 boys) from one school were provided with 200 min of physical education per week. Three other schools (44 girls and 47 boys) continued with the standard 60 min per week. Tibial total and cortical area, cortical density, polar stress–strain index (SSI), and the mass and density distribution around the center of mass (polar distribution, mg) and through the bones cortex (radial distribution subdivided into endo-, mid-, and pericortical volumetric BMD: mg/cm3) at three sites (14, 38, and 66 %) were assessed using peripheral quantitative computed tomography after 7 years. Girls in the intervention group had 2.5 % greater cortical thickness and 6.9 % greater SSI at the 66 % tibia, which was accompanied by significantly greater pericortical volumetric BMD compared to controls (all P <0.05). Region-specific differences in cortical mass were also detected in the anterior, medial, and lateral sectors at the 38 and 66 % tibial sites. There were no group differences at the 14 % tibia site in girls, and no group differences in any of the bone parameters in boys. Additional school-based physical education over seven years was associated with greater tibial structure, strength, and region-specific adaptations in cortical bone mass and density distribution in girls, but not in boys.
|
|
20. |
|
|
21. |
- Kato, Bernet S., et al.
(author)
-
Variance decomposition of protein profiles from antibody arrays using a longitudinal twin model
- 2011
-
In: Proteome Science. - : Springer Science and Business Media LLC. - 1477-5956. ; 9:1, s. 73-
-
Journal article (peer-reviewed)abstract
- The advent of affinity-based proteomics technologies for global protein profiling provides the prospect of finding new molecular biomarkers for common, multifactorial disorders. The molecular phenotypes obtained from studies on such platforms are driven by multiple sources, including genetic, environmental, and experimental. In characterizing the contribution of different sources of variation to the measured phenotypes, the aim is to facilitate the design and interpretation of future biomedical studies employing exploratory and multiplexed technologies. Thus, biometrical genetic modelling of twin or other family data can be used to decompose the variation underlying a phenotype into biological and experimental components. RESULTS: Using antibody suspension bead arrays and antibodies from the Human Protein Atlas, we study unfractionated serum from a longitudinal study on 154 twins. In this study, we provide a detailed description of how the variation in a molecular phenotype in terms of protein profile can be decomposed into familial i.e. genetic and common environmental; individual environmental, short-term biological and experimental components. The results show that across 69 antibodies analyzed in the study, the median proportion of the total variation explained by familial sources is 12% (IQR 1-22%), and the median proportion of the total variation attributable to experimental sources is 63% (IQR 53-72%). CONCLUSION: The variability analysis of antibody arrays highlights the importance to consider variability components and their relative contributions when designing and evaluating studies for biomarker discover with exploratory, high-throughput and multiplexed methods.
|
|
22. |
|
|
23. |
- Mer, Arvind Singh, et al.
(author)
-
Biological and therapeutic implications of a unique subtype of NPM1 mutated AML
- 2021
-
In: Nature Communications. - : Springer Nature. - 2041-1723. ; 12:1
-
Journal article (peer-reviewed)abstract
- In acute myeloid leukemia (AML), molecular heterogeneity across patients constitutes a major challenge for prognosis and therapy. AML with NPM1 mutation is a distinct genetic entity in the revised World Health Organization classification. However, differing patterns of co-mutation and response to therapy within this group necessitate further stratification. Here we report two distinct subtypes within NPM1 mutated AML patients, which we label as primitive and committed based on the respective presence or absence of a stem cell signature. Using gene expression (RNA-seq), epigenomic (ATAC-seq) and immunophenotyping (CyToF) analysis, we associate each subtype with specific molecular characteristics, disease differentiation state and patient survival. Using ex vivo drug sensitivity profiling, we show a differential drug response of the subtypes to specific kinase inhibitors, irrespective of the FLT3-ITD status. Differential drug responses of the primitive and committed subtype are validated in an independent AML cohort. Our results highlight heterogeneity among NPM1 mutated AML patient samples based on stemness and suggest that the addition of kinase inhibitors to the treatment of cases with the primitive signature, lacking FLT3-ITD, could have therapeutic benefit. Molecular heterogeneity of acute myeloid leukaemia (AML) across patients is a major challenge for prognosis and therapy. Here, the authors show that NPM1 mutated AML is a heterogeneous class, consisting of two subtypes which exhibit distinct molecular characteristics, differentiation state, patient survival and drug response.
|
|
24. |
- Parts, Leopold, et al.
(author)
-
Extent, causes, and consequences of small RNA expression variation in human adipose tissue.
- 2012
-
In: PLOS Genetics. - : Public Library of Science (PLoS). - 1553-7390 .- 1553-7404. ; 8:5
-
Journal article (peer-reviewed)abstract
- Small RNAs are functional molecules that modulate mRNA transcripts and have been implicated in the aetiology of several common diseases. However, little is known about the extent of their variability within the human population. Here, we characterise the extent, causes, and effects of naturally occurring variation in expression and sequence of small RNAs from adipose tissue in relation to genotype, gene expression, and metabolic traits in the MuTHER reference cohort. We profiled the expression of 15 to 30 base pair RNA molecules in subcutaneous adipose tissue from 131 individuals using high-throughput sequencing, and quantified levels of 591 microRNAs and small nucleolar RNAs. We identified three genetic variants and three RNA editing events. Highly expressed small RNAs are more conserved within mammals than average, as are those with highly variable expression. We identified 14 genetic loci significantly associated with nearby small RNA expression levels, seven of which also regulate an mRNA transcript level in the same region. In addition, these loci are enriched for variants significant in genome-wide association studies for body mass index. Contrary to expectation, we found no evidence for negative correlation between expression level of a microRNA and its target mRNAs. Trunk fat mass, body mass index, and fasting insulin were associated with more than twenty small RNA expression levels each, while fasting glucose had no significant associations. This study highlights the similar genetic complexity and shared genetic control of small RNA and mRNA transcripts, and gives a quantitative picture of small RNA expression variation in the human population.
|
|
25. |
|
|
26. |
- Rantalainen, M
(author)
-
Combining metabonomics and other -omics data
- 2015
-
In: Methods in molecular biology (Clifton, N.J.). - New York, NY : Springer New York. - 1940-6029. ; 1277, s. 147-59
-
Journal article (peer-reviewed)
|
|
27. |
|
|
28. |
|
|
29. |
- Rantalainen, Mattias, et al.
(author)
-
Piecewise multivariate modelling of sequential metabolic profiling data
- 2008
-
In: BMC Bioinformatics. - : EMBO. - 1471-2105. ; 9, s. 105-
-
Journal article (peer-reviewed)abstract
- Background: Modelling the time-related behaviour of biological systems is essential for understanding their dynamic responses to perturbations. In metabolic profiling studies, the sampling rate and number of sampling points are often restricted due to experimental and biological constraints. Results: A supervised multivariate modelling approach with the objective to model the time-related variation in the data for short and sparsely sampled time-series is described. A set of piecewise Orthogonal Projections to Latent Structures (OPLS) models are estimated, describing changes between successive time points. The individual OPLS models are linear, but the piecewise combination of several models accommodates modelling and prediction of changes which are non-linear with respect to the time course. We demonstrate the method on both simulated and metabolic profiling data, illustrating how time related changes are successfully modelled and predicted. Conclusion: The proposed method is effective for modelling and prediction of short and multivariate time series data. A key advantage of the method is model transparency, allowing easy interpretation of time-related variation in the data. The method provides a competitive complement to commonly applied multivariate methods such as OPLS and Principal Component Analysis (PCA) for modelling and analysis of short time-series data.
|
|
30. |
- Rantalainen, Mattias, et al.
(author)
-
Robust linear models for cis-eQTL analysis
- 2015
-
In: PLoS One. - Stockholm : Karolinska Institutet, Dept of Medical Epidemiology and Biostatistics. - 1932-6203.
-
Journal article (peer-reviewed)abstract
- Expression Quantitative Trait Loci (eQTL) analysis enables characterisation of functional genetic variation influencing expression levels of individual genes. In outbread populations, including humans, eQTLs are commonly analysed using the conventional linear model, adjusting for relevant covariates, assuming an allelic dosage model and a Gaussian error term. However, gene expression data generally have noise that induces heavy-tailed errors relative to the Gaussian distribution and often include atypical observations, or outliers. Such departures from modelling assumptions can lead to an increased rate of type II errors (false negatives), and to some extent also type I errors (false positives). Careful model checking can reduce the risk of type-I errors but often not type II errors, since it is generally too time-consuming to carefully check all models with a non-significant effect in large-scale and genome-wide studies. Here we propose the application of a robust linear model for eQTL analysis to reduce adverse effects of deviations from the assumption of Gaussian residuals. We present results from a simulation study as well as results from the analysis of real eQTL data sets. Our findings suggest that in many situations robust models have the potential to provide more reliable eQTL results compared to conventional linear models, particularly in respect to reducing type II errors due to non-Gaussian noise. Post-genomic data, such as that generated in genome-wide eQTL studies, are often noisy and frequently contain atypical observations. Robust statistical models have the potential to provide more reliable results and increased statistical power under non-Gaussian conditions. The results presented here suggest that robust models should be considered routinely alongside other commonly used methodologies for eQTL analysis.
|
|
31. |
- Robertson, S, et al.
(author)
-
Prognostic value of Ki67 analysed by cytology or histology in primary breast cancer
- 2018
-
In: Journal of clinical pathology. - : BMJ. - 1472-4146 .- 0021-9746. ; 71:9, s. 787-794
-
Journal article (peer-reviewed)abstract
- The accuracy of biomarker assessment in breast pathology is vital for therapy decisions. The therapy predictive and prognostic biomarkers oestrogen receptor (ER), progesterone receptor, HER2 and Ki67 may act as surrogates to gene expression profiling of breast cancer. The aims of this study were to investigate the concordance of consecutive biomarker assessment by immunocytochemistry on preoperative fine-needle aspiration cytology versus immunohistochemistry (IHC) on the corresponding resected breast tumours. Further, to investigate the concordance with molecular subtype and correlation to stage and outcome.MethodsTwo retrospective cohorts comprising 385 breast tumours with clinicopathological data including gene expression-based subtype and up to 10-year overall survival data were evaluated.ResultsIn both cohorts, we identified a substantial variation in Ki67 index between cytology and histology and a switch between low and high proliferation within the same tumour in 121/360 cases. ER evaluations were discordant in only 1.5% of the tumours. From cohort 2, gene expression data with PAM50 subtype were used to correlate surrogate subtypes. IHC-based surrogate classification could identify the correct molecular subtype in 60% and 64% of patients by cytology (n=63) and surgical resections (n=73), respectively. Furthermore, high Ki67 in surgical resections but not in cytology was associated with poor overall survival and higher probability for axillary lymph node metastasis.ConclusionsThis study shows considerable differences in the prognostic value of Ki67 but not ER in breast cancer depending on the diagnostic method. Furthermore, our findings show that both methods are insufficient in predicting true molecular subtypes.
|
|
32. |
|
|
33. |
|
|
34. |
- Sun, Y., et al.
(author)
-
An Integrated Bioinformatics Approach for Identifying Genetic Markers that Predict Cerebrospinal Fluid Biomarker p-tau(181)/A beta(1-42) Ratio in ApoE4-Negative Mild Cognitive Impairment Patients
- 2015
-
In: Journal of Alzheimers Disease. - : IOS Press. - 1387-2877 .- 1875-8908. ; 45:4, s. 1061-1076
-
Journal article (peer-reviewed)abstract
- Alzheimer's disease (AD) is the most common form of dementia, with no disease-modifying treatment yet available. Early detection of patients at risk of developing AD is of central importance. Blood-based genetic signatures can serve as early detection and as population-based screening tools. In this study, we aimed to identify genetic markers and gene signatures associated with cerebrospinal fluid (CSF) biomarkers levels of t-tau, p-tau(181), and with the two ratios t-tau/ A beta(1-42) and p-tau(181)/A beta(1-42) in the context of progression from mild cognitive impairment (MCI) to AD, and to identify a panel of genetic markers that can predict CSF biomarker p-tau(181)/A beta(1-42) ratio with consideration of APOE epsilon 4 stratification. We analyzed genome-wide the Alzheimer's Disease Neuroimaging Initiative dataset with up to 48 months follow-up. In the first part of the analysis, the main effect of single nucleotide polymorphisms (SNPs) under an additive genetic model was assessed for each of the four CSF biomarkers. In the second part of the analysis, we performed an integrated analysis of genome-wide association study results with pathway enrichment analysis, predictive modeling and network analysis in the subgroup of ApoE4-negative subjects. We identified a panel of five SNPs, rs6766238, rs1143960, rs1249963, rs11975968, and rs4836493, that are predictive for
|
|
35. |
|
|
36. |
- Trac, Quang Thinh, et al.
(author)
-
Prediction model for drug response of acute myeloid leukemia patients
- 2023
-
In: npj Precision Oncology. - : Springer Nature. - 2397-768X. ; 7
-
Journal article (peer-reviewed)abstract
- Despite some encouraging successes, predicting the therapy response of acute myeloid leukemia (AML) patients remains highly challenging due to tumor heterogeneity. Here we aim to develop and validate MDREAM, a robust ensemble-based prediction model for drug response in AML based on an integration of omics data, including mutations and gene expression, and large-scale drug testing. Briefly, MDREAM is first trained in the BeatAML cohort (n = 278), and then validated in the BeatAML (n = 183) and two external cohorts, including a Swedish AML cohort (n = 45) and a relapsed/refractory acute leukemia cohort (n = 12). The final prediction is based on 122 ensemble models, each corresponding to a drug. A confidence score metric is used to convey the uncertainty of predictions; among predictions with a confidence score >0.75, the validated proportion of good responders is 77%. The Spearman correlations between the predicted and the observed drug response are 0.68 (95% CI: [0.64, 0.68]) in the BeatAML validation set, -0.49 (95% CI: [-0.53, -0.44]) in the Swedish cohort and 0.59 (95% CI: [0.51, 0.67]) in the relapsed/refractory cohort. A web-based implementation of MDREAM is publicly available at https://www.meb.ki.se/shiny/truvu/MDREAM/.
|
|
37. |
|
|
38. |
- Wang, Y., et al.
(author)
-
Improved breast cancer histological grading using deep learning
- 2022
-
In: Annals of Oncology. - : Elsevier. - 0923-7534 .- 1569-8041. ; 33:1, s. 89-98
-
Journal article (peer-reviewed)abstract
- Background: The Nottingham histological grade (NHG) is a well-established prognostic factor for breast cancer that is broadly used in clinical decision making. However, similar to 50% of patients are classified as grade 2, an intermediate risk group with low clinical value. To improve risk stratification of NHG 2 breast cancer patients, we developed and validated a novel histological grade model (DeepGrade) based on digital whole-slide histopathology images (WSIs) and deep learning.Patients and methods: In this observational retrospective study, routine WSIs stained with haematoxylin and eosin from 1567 patients were utilised for model optimisation and validation. Model generalisability was further evaluated in an external test set with 1262 patients. NHG 2 cases were stratified into two groups, DG2-high and DG2-low, and the prognostic value was assessed. The main outcome was recurrence-free survival.Results: DeepGrade provides independent prognostic information for stratification of NHG 2 cases in the internal test set, where DG2-high showed an increased risk for recurrence (hazard ratio [HR] 2.94, 95% confidence interval [CI] 1.24-6.97, P = 0.015) compared with the DG2-low group after adjusting for established risk factors (independent test data). DG2-low also shared phenotypic similarities with NHG 1, and DG2-high with NHG 3, suggesting that the model identifies morphological patterns in NHG 2 that are associated with more aggressive tumours. The prognostic value of DeepGrade was further assessed in the external test set, confirming an increased risk for recurrence in DG2-high (HR 1.91, 95% CI 1.11-3.29, P = 0.019).Conclusions: The proposed model-based stratification of patients with NHG 2 tumours is prognostic and adds clinically relevant information over routine histological grading. The methodology offers a cost-effective alternative to molecular profiling to extract information relevant for clinical decisions.
|
|