Can large language models reason about medical questions?

Although large language models often produce impressive outputs, it remains unclear how they perform in real-world scenarios requiring strong reasoning skills and expert domain knowledge. We set out to investigate whether closed- and open-source models (GPT-3.5, Llama 2, etc.) can be applied to answer and reason about difficult real-world-based questions. We focus on three popular medical benchmarks (MedQA-US Medical Licensing Examination [USMLE], MedMCQA, and PubMedQA) and multiple prompting scenarios: chain of thought (CoT; think step by step), few shot, and retrieval augmentation. Based on an expert annotation of the generated CoTs, we found that InstructGPT can often read, reason, and recall expert knowledge. Last, by leveraging advances in prompt engineering (few-shot and ensemble methods), we demonstrated that GPT-3.5 not only yields calibrated predictive distributions but also reaches the passing score on three datasets: MedQA-USMLE (60.2%), MedMCQA (62.7%), and PubMedQA (78.2%). Open-source models are closing the gap: Llama 2 70B also passed the MedQA-USMLE with 62.5% accuracy.

Blood donor genotyping for prediction of blood group antigens: Results from 5 years' experience (2017-2022).

BACKGROUND AND OBJECTIVES: For 5 years, routine genotyping has been performed for selected blood groups of blood donors in the Copenhagen Capital Region, Denmark. The result is summarized in the following. MATERIALS AND METHODS: Genotyping was carried out by an external service provider using the competitive allele specific PCR (KASP) technology. The genotypes were returned to the blood bank and translated into phenotypes by a proprietary IT application. RESULTS: In total, 65 alleles from 16 blood group systems (ABO, MNS, Rh, Lutheran, Kell, Duffy, Kidd, Diego, Yt, Dombrock, Colton, Landsteiner-Wiener, Cromer, Knops, Vel, secretor status) and the HPA1, HPA5 and HPA15 antigens were interrogated. After translation, phenotypes were imported into the laboratory information management system of the blood bank. The results from 31,538 genotyped blood donors were used to calculate the allele frequencies for a Danish blood donor population. ABO genotyping was done for sample ID purposes. Determination of the 1061delC single nucleotide polymorphism (SNP) (NM_020469.2), most frequently characteristic of ABO*A2, was validated against a series of 1287 samples with Dolichos biflorus lectin determination of the A1 phenotype. CONCLUSION: We report allele frequencies and phenotype frequencies for 16 blood groups from a total of 31,538 genotyped blood donors. Blood products were supplied from a total of 64,312 active blood donors, and of these active blood donors 25,396 (39.5%) were genotyped. These donors represent a valuable resource for patient treatment. This genotyping has enabled the provision of rare genotyped donor blood for patients with alloantibodies and rare reagent cells for serology.

ThoughtSource: A central hub for large language model reasoning data.

Large language models (LLMs) such as GPT-4 have recently demonstrated impressive results across a wide range of tasks. LLMs are still limited, however, in that they frequently fail at complex reasoning, their reasoning processes are opaque, they are prone to 'hallucinate' facts, and there are concerns about their underlying biases. Letting models verbalize reasoning steps as natural language, a technique known as chain-of-thought prompting, has recently been proposed as a way to address some of these issues. Here we present ThoughtSource, a meta-dataset and software library for chain-of-thought (CoT) reasoning. The goal of ThoughtSource is to improve future artificial intelligence systems by facilitating qualitative understanding of CoTs, enabling empirical evaluations, and providing training data. This first release of ThoughtSource integrates seven scientific/medical, three general-domain and five math word question answering datasets.

Hyperhidrosis and human leucocyte antigens in the Danish Blood Donor Study.

Familial clustering of the skin disease primary hyperhidrosis suggests a genetic component to the disease. The human leucocyte antigen (HLA) is implicated in a range of diseases, including many comorbidities to hyperhidrosis. No study has investigated whether the HLA genes are involved in the pathogenesis of hyperhidrosis. We, therefore, compared HLA alleles in individuals with and without hyperhidrosis in this study of 65 000 blood donors. In this retrospective cohort study, we retrieved information on individuals with and without hyperhidrosis using self-reported questionnaires, the Danish National Patient Registry and the Danish National Prescription Registry on participants recruited to the Danish Blood Donor Study between 2010 and 2019. Association tests using logistic regression were conducted for each HLA allele corrected for sex, age, body mass index, smoking and principal components. Overall, 145 of 65 795 (0.2%) participants had hospital diagnosed hyperhidrosis. Similarly, 1379 of 15 530 (8.9%) participants had moderate-severe self-reported hyperhidrosis, of whom 447 (2.9%) had severe self-reported hyperhidrosis. Altogether, 28 participants had both hospital diagnosed and moderate-severe self-reported hyperhidrosis. Severe self-reported hyperhidrosis was associated with HLA-A*80:01 (adjusted odds ratio 26.97; 95% confidence interval 5.32-136.70; n = 7; P < .001). Moderate-severe self-reported hyperhidrosis and hospital diagnosed hyperhidrosis were not associated with any HLA. The association between hyperhidrosis and HLA-A*80:01 was based on a very small number of cases and not replicated in other patient subsets, and therefore likely a chance finding. Thus, this study suggests that genes other than the HLA are involved in the pathogenesis of hyperhidrosis.

Estimation of SARS-CoV-2 Infection Fatality Rate by Real-time Antibody Screening of Blood Donors.

BACKGROUND: The pandemic due to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has tremendous consequences for our societies. Knowledge of the seroprevalence of SARS-CoV-2 is needed to accurately monitor the spread of the epidemic and to calculate the infection fatality rate (IFR). These measures may help the authorities make informed decisions and adjust the current societal interventions. The objective was to perform nationwide real-time seroprevalence surveying among blood donors as a tool to estimate previous SARS-CoV-2 infections and the population-based IFR. METHODS: Danish blood donors aged 17-69 years giving blood 6 April to 3 May were tested for SARS-CoV-2 immunoglobulin M and G antibodies using a commercial lateral flow test. Antibody status was compared between geographical areas, and an estimate of the IFR was calculated. Seroprevalence was adjusted for assay sensitivity and specificity taking the uncertainties of the test validation into account when reporting the 95% confidence intervals (CIs). RESULTS: The first 20 640 blood donors were tested, and a combined adjusted seroprevalence of 1.9% (95% CI, .8-2.3) was calculated. The seroprevalence differed across areas. Using available data on fatalities and population numbers, a combined IFR in patients <70 years is estimated at 89 per 100 000 (95% CI, 72-211) infections. CONCLUSIONS: The IFR was estimated to be slightly lower than previously reported from other countries not using seroprevalence data. The IFR is likely severalfold lower than the current estimate. We have initiated real-time nationwide anti-SARS-CoV-2 seroprevalence surveying of blood donations as a tool in monitoring the epidemic.

Risk of COVID-19 in health-care workers in Denmark: an observational cohort study.

BACKGROUND: Health-care workers are thought to be highly exposed to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. We aimed to investigate the prevalence of antibodies against SARS-CoV-2 in health-care workers and the proportion of seroconverted health-care workers with previous symptoms of COVID-19. METHODS: In this observational cohort study, screening was offered to health-care workers in the Capital Region of Denmark, including medical, nursing, and other students who were associated with hospitals in the region. Screening included point-of-care tests for IgM and IgG antibodies against SARS-CoV-2. Test results and participant characteristics were recorded. Results were compared with findings in blood donors in the Capital Region in the study period. FINDINGS: Between April 15 and April 23, 2020, we screened 29 295 health-care workers, of whom 28 792 (98·28%) provided their test results. We identified 1163 (4·04% [95% CI 3·82-4·27]) seropositive health-care workers. Seroprevalence was higher in health-care workers than in blood donors (142 [3·04%] of 4672; risk ratio [RR] 1·33 [95% CI 1·12-1·58]; p<0·001). Seroprevalence was higher in male health-care workers (331 [5·45%] of 6077) than in female health-care workers (832 [3·66%] of 22 715; RR 1·49 [1·31-1·68]; p<0·001). Frontline health-care workers working in hospitals had a significantly higher seroprevalence (779 [4·55%] of 16 356) than health-care workers in other settings (384 [3·29%] of 11 657; RR 1·38 [1·22-1·56]; p<0·001). Health-care workers working on dedicated COVID-19 wards (95 [7·19%] of 1321) had a significantly higher seroprevalence than other frontline health-care workers working in hospitals (696 [4·35%] of 15 983; RR 1·65 [1·34-2·03]; p<0·001). 622 [53·5%] of 1163 seropositive participants reported symptoms attributable to SARS-CoV-2. Loss of taste or smell was the symptom that was most strongly associated with seropositivity (377 [32·39%] of 1164 participants with this symptom were seropositive vs 786 [2·84%] of 27 628 without this symptom; RR 11·38 [10·22-12·68]). The study is registered at ClinicalTrials.gov, NCT04346186. INTERPRETATION: The prevalence of health-care workers with antibodies against SARS-CoV-2 was low but higher than in blood donors. The risk of SARS-CoV-2 infection in health-care workers was related to exposure to infected patients. More than half of seropositive health-care workers reported symptoms attributable to COVID-19. FUNDING: Lundbeck Foundation.

The value of circulating microRNAs for early diagnosis of B-cell lymphoma: A case-control study on historical samples.

MicroRNAs are small regulatory RNAs that are deregulated in a wide variety of human cancers, including different types of B-cell lymphoma. Nevertheless, the feasibility of circulating microRNA for early diagnosis of B-cell lymphoma has not been established. To address the possibility of detecting specific circulating microRNAs years before a B-cell lymphoma is diagnosed, we studied the plasma expression of microRNA first in pre-treatment samples from patients with diffuse large B-cell lymphoma and subsequently in repository samples from blood donors who later developed B-cell lymphomas. In addition, we studied the microRNA expression in the diagnostic lymphoma biopsy. The most strongly induced (miR-326) and suppressed (miR-375) plasma microRNA at diagnosis, when compared with healthy blood donors, were also substantially up- or down-regulated in plasma repository samples taken from several months to up to two years before the blood donors were diagnosed with B-cell lymphoma. Importantly, at these time points the donors had no signs of disease and felt healthy enough to donate blood. In conclusion, this first study of plasma microRNA profiles from apparently healthy individuals, taken several years before B-cell lymphoma diagnosis, suggests that plasma microRNA profiles may be predictive of lymphoma development.

Next Generation Sequencing-Based Fetal ABO Blood Group Prediction by Analysis of Cell-Free DNA from Maternal Plasma.

INTRODUCTION: ABO blood group incompatibility between a pregnant woman and her fetus as a cause of morbidity or mortality of the fetus or newborn remains an important, albeit rare, risk. When a pregnant woman has a high level of anti-A or anti-B IgG antibodies, the child may be at risk for hemolytic disease of the fetus and newborn (HDFN). Performing a direct prenatal determination of the fetal ABO blood group can provide valuable clinical information. OBJECTIVE: Here, we report a next generation sequencing (NGS)-based assay for predicting the prenatal ABO blood group. MATERIALS AND METHODS: A total of 26 plasma samples from 26 pregnant women were tested from gestational weeks 12 to 35. Of these samples, 20 were clinical samples and 6 were test samples. Extracted cell-free DNA was PCR-amplified using 2 primer sets followed by NGS. NGS data were analyzed by 2 different methods, FASTQ analysis and a grep search, to ensure robust results. The fetal ABO prediction was compared with the known serological infant ABO type, which was available for 19 samples. RESULTS: There was concordance for 19 of 19 predictable samples where the phenotype information was available and when the analysis was done by the 2 methods. For immunized pregnant women (n = 20), the risk of HDFN was predicted for 12 fetuses, and no risk was predicted for 7 fetuses; one result of the clinical samples was indeterminable. Cloning and sequencing revealed a novel variant harboring the same single nucleotide variations as ABO*O.01.24 with an additional c.220C>T substitution. An additional indeterminable result was found among the 6 test samples and was caused by maternal heterozygosity. The 2 indeterminable samples demonstrated limitations to the assay due to hybrid ABO genes or maternal heterozygosity. CONCLUSIONS: We pioneered an NGS-based fetal ABO prediction assay based on a cell-free DNA analysis from maternal plasma and demonstrated its application in a small number of samples. Based on the calculations of variant frequencies and ABO*O.01/ABO*O.02 heterozygote frequency, we estimate that we can assign a reliable fetal ABO type in approximately 95% of the forthcoming clinical samples of type O pregnant women. Despite the vast genetic variations underlying the ABO blood groups, many variants are rare, and prenatal ABO prediction is possible and adds valuable early information for the prevention of ABO HDFN.

TMEM30A loss-of-function mutations drive lymphomagenesis and confer therapeutically exploitable vulnerability in B-cell lymphoma.

Transmembrane protein 30A (TMEM30A) maintains the asymmetric distribution of phosphatidylserine, an integral component of the cell membrane and 'eat-me' signal recognized by macrophages. Integrative genomic and transcriptomic analysis of diffuse large B-cell lymphoma (DLBCL) from the British Columbia population-based registry uncovered recurrent biallelic TMEM30A loss-of-function mutations, which were associated with a favorable outcome and uniquely observed in DLBCL. Using TMEM30A-knockout systems, increased accumulation of chemotherapy drugs was observed in TMEM30A-knockout cell lines and TMEM30A-mutated primary cells, explaining the improved treatment outcome. Furthermore, we found increased tumor-associated macrophages and an enhanced effect of anti-CD47 blockade limiting tumor growth in TMEM30A-knockout models. By contrast, we show that TMEM30A loss-of-function increases B-cell signaling following antigen stimulation-a mechanism conferring selective advantage during B-cell lymphoma development. Our data highlight a multifaceted role for TMEM30A in B-cell lymphomagenesis, and characterize intrinsic and extrinsic vulnerabilities of cancer cells that can be therapeutically exploited.

Epidemiology of hepatitis E virus infection in a cohort of 4023 immunocompromised patients.

OBJECTIVES: The prevalence of active, chronic, and former hepatitis E virus (HEV) infections was investigated in a cohort of immunocompromised patients. The association with transfusion transmitted HEV was evaluated, and the HEV seroprevalence was compared with that in healthy blood donors. STUDY DESIGN AND METHODS: Serum samples from 4023 immunocompromised patients at Rigshospitalet, Denmark were retrospectively tested for HEV RNA and anti-HEV IgG. HEV RNA-positive patients were followed up by HEV testing, clinical symptoms, and transfusion history. Factors associated with anti-HEV were explored by multivariable logistic regression analysis. Samples from 1226 blood donors were retrospectively tested for anti-HEV IgG. RESULTS: HEV RNA was detected in six patients (0.15%) with no indications of chronic HEV infection. HEV RNA prevalence rates among recipients of allogeneic haematopoietic stem cell transplantation (allo-HSCT) and solid organ transplantation (SOT) were 0.58% and 0.21%, respectively. Transfusion transmitted infections were refuted, and transfusion history was not associated with anti-HEV positivity. The difference in HEV seroprevalence between patients (22.0%) and blood donors (10.9%) decreased when adjusting for age and sex (odds ratio 1.20, 95% confidence interval 0.97-1.48). CONCLUSIONS: HEV viremia among allo-HSCT and SOT recipients suggests that clinicians should be aware of this diagnosis. The lack of association of blood transfusion with anti-HEV positivity supports food-borne transmission as the main transmission route of HEV common to both patients and blood donors.

Genetic profiling of MYC and BCL2 in diffuse large B-cell lymphoma determines cell-of-origin-specific clinical impact.

The clinical significance of MYC and BCL2 genetic alterations in diffuse large B-cell lymphoma (DLBCL), apart from translocations, has not been comprehensively investigated using high-resolution genetic assays. In this study, we profiled MYC and BCL2 genetic alterations using next-generation sequencing and high-resolution SNP array in 347 de novo DLBCL cases treated with R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone) at the British Columbia Cancer Agency. Cell-of-origin (COO) subtype was determined by Lymph2Cx digital gene expression profiling. We showed that the incidence of MYC/BCL2 genetic alterations and their clinical significance were largely dependent on COO subtypes. It is noteworthy that the presence of BCL2 gain/amplification is significantly associated with poor outcome in activated B-cell-like and BCL2 translocation with poor outcome in germinal center B-cell subtypes, respectively. Both have prognostic significance independent of MYC/BCL2 dual expression and the International Prognostic Index (IPI). Furthermore, the combination of BCL2 genetic alterations with IPI identifies markedly worse prognostic groups within individual COO subtypes. Thus, high-resolution genomic assays identify extremely poor prognostic groups within each COO subtype on the basis of BCL2 genetic status in this large, uniformly R-CHOP-treated population-based cohort of DLBCL. These results suggest COO subtype-specific biomarkers based on BCL2 genetic alterations can be used to risk-stratify patients with DLBCL treated with immunochemotherapy.

The histone lysine methyltransferase KMT2D sustains a gene expression program that represses B cell lymphoma development.

The gene encoding the lysine-specific histone methyltransferase KMT2D has emerged as one of the most frequently mutated genes in follicular lymphoma and diffuse large B cell lymphoma; however, the biological consequences of KMT2D mutations on lymphoma development are not known. Here we show that KMT2D functions as a bona fide tumor suppressor and that its genetic ablation in B cells promotes lymphoma development in mice. KMT2D deficiency also delays germinal center involution and impedes B cell differentiation and class switch recombination. Integrative genomic analyses indicate that KMT2D affects methylation of lysine 4 on histone H3 (H3K4) and expression of a set of genes, including those in the CD40, JAK-STAT, Toll-like receptor and B cell receptor signaling pathways. Notably, other KMT2D target genes include frequently mutated tumor suppressor genes such as TNFAIP3, SOCS3 and TNFRSF14. Therefore, KMT2D mutations may promote malignant outgrowth by perturbing the expression of tumor suppressor genes that control B cell-activating pathways.

Prognostic Significance of Diffuse Large B-Cell Lymphoma Cell of Origin Determined by Digital Gene Expression in Formalin-Fixed Paraffin-Embedded Tissue Biopsies.

PURPOSE: To evaluate the prognostic impact of cell-of-origin (COO) subgroups, assigned using the recently described gene expression-based Lymph2Cx assay in comparison with International Prognostic Index (IPI) score and MYC/BCL2 coexpression status (dual expressers). PATIENTS AND METHODS: Reproducibility of COO assignment using the Lymph2Cx assay was tested employing repeated sampling within tumor biopsies and changes in reagent lots. The assay was then applied to pretreatment formalin-fixed paraffin-embedded tissue (FFPET) biopsies from 344 patients with de novo diffuse large B-cell lymphoma (DLBCL) uniformly treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) at the British Columbia Cancer Agency. MYC and BCL2 protein expression was assessed using immunohistochemistry on tissue microarrays. RESULTS: The Lymph2Cx assay provided concordant COO calls in 96% of 49 repeatedly sampled tumor biopsies and in 100% of 83 FFPET biopsies tested across reagent lots. Critically, no frank misclassification (activated B-cell-like DLBCL to germinal center B-cell-like DLBCL or vice versa) was observed. Patients with activated B-cell-like DLBCL had significantly inferior outcomes compared with patients with germinal center B-cell-like DLBCL (log-rank P < .001 for time to progression, progression-free survival, disease-specific survival, and overall survival). In pairwise multivariable analyses, COO was associated with outcomes independent of IPI score and MYC/BCL2 immunohistochemistry. The prognostic significance of COO was particularly evident in patients with intermediate IPI scores and the non-MYC-positive/BCL2-positive subgroup (log-rank P < .001 for time to progression). CONCLUSION: Assignment of DLBCL COO by the Lymph2Cx assay using FFPET biopsies identifies patient groups with significantly different outcomes after R-CHOP, independent of IPI score and MYC/BCL2 dual expression.

Integration of gene mutations in risk prognostication for patients receiving first-line immunochemotherapy for follicular lymphoma: a retrospective analysis of a prospective clinical trial and validation in a population-based registry.

BACKGROUND: Follicular lymphoma is a clinically and genetically heterogeneous disease, but the prognostic value of somatic mutations has not been systematically assessed. We aimed to improve risk stratification of patients receiving first-line immunochemotherapy by integrating gene mutations into a prognostic model. METHODS: We did DNA deep sequencing to retrospectively analyse the mutation status of 74 genes in 151 follicular lymphoma biopsy specimens that were obtained from patients within 1 year before beginning immunochemotherapy consisting of rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP). These patients were recruited between May 4, 2000, and Oct 20, 2010, as part of a phase 3 trial (GLSG2000). Eligible patients had symptomatic, advanced stage follicular lymphoma and were previously untreated. The primary endpoints were failure-free survival (defined as less than a partial remission at the end of induction, relapse, progression, or death) and overall survival calculated from date of treatment initiation. Median follow-up was 7·7 years (IQR 5·5-9·3). Mutations and clinical factors were incorporated into a risk model for failure-free survival using multivariable L1-penalised Cox regression. We validated the risk model in an independent population-based cohort of 107 patients with symptomatic follicular lymphoma considered ineligible for curative irradiation. Pretreatment biopsies were taken between Feb 24, 2004, and Nov 24, 2009, within 1 year before beginning first-line immunochemotherapy consisting of rituximab, cyclophosphamide, vincristine, and prednisone (R-CVP). Median follow-up was 6·7 years (IQR 5·7-7·6). FINDINGS: We established a clinicogenetic risk model (termed m7-FLIPI) that included the mutation status of seven genes (EZH2, ARID1A, MEF2B, EP300, FOXO1, CREBBP, and CARD11), the Follicular Lymphoma International Prognostic Index (FLIPI), and Eastern Cooperative Oncology Group (ECOG) performance status. In the training cohort, m7-FLIPI defined a high-risk group (28%, 43/151) with 5-year failure-free survival of 38·29% (95% CI 25·31-57·95) versus 77·21% (95% CI 69·21-86·14) for the low-risk group (hazard ratio [HR] 4·14, 95% CI 2·47-6·93; p<0·0001; bootstrap-corrected HR 2·02), and outperformed a prognostic model of only gene mutations (HR 3·76, 95% CI 2·10-6·74; p<0·0001; bootstrap-corrected HR 1·57). The positive predictive value and negative predictive value for 5-year failure-free survival were 64% and 78%, respectively, with a C-index of 0·80 (95% CI 0·71-0·89). In the validation cohort, m7-FLIPI again defined a high-risk group (22%, 24/107) with 5-year failure-free survival of 25·00% (95% CI 12·50-49·99) versus 68·24% (58·84-79·15) in the low-risk group (HR 3·58, 95% CI 2·00-6·42; p<0.0001). The positive predictive value for 5-year failure-free survival was 72% and 68% for negative predictive value, with a C-index of 0·79 (95% CI 0·69-0·89). In the validation cohort, risk stratification by m7-FLIPI outperformed FLIPI alone (HR 2·18, 95% CI 1·21-3·92), and FLIPI combined with ECOG performance status (HR 2·03, 95% CI 1·12-3·67). INTERPRETATION: Integration of the mutational status of seven genes with clinical risk factors improves prognostication for patients with follicular lymphoma receiving first-line immunochemotherapy and is a promising approach to identify the subset at highest risk of treatment failure. FUNDING: Deutsche Krebshilfe, Terry Fox Research Institute.

Development and blind clinical validation of a microRNA based predictor of response to treatment with R-CHO(E)P in DLBCL.

MicroRNAs (miRNA) are a group of short noncoding RNAs that regulate gene expression at the posttranscriptional level. It has been shown that microRNAs are independent predictors of outcome in patients with diffuse large B-cell lymphoma (DLBCL) treated with the drug combination R-CHOP. Based on the measured growth inhibition of 60 human cancer cell lines (NCI60) in the presence of doxorubicine, cyclophosphamide, vincristine and etoposide as well as the baseline microRNA expression of the 60 cell lines, a microRNA based response predictor to CHOP was developed. The response predictor consisting of 20 microRNAs was blindly validated in a cohort of 116 de novo DLBCL patients treated with R-CHOP or R-CHOEP as first line treatment. The predicted sensitivity based on diagnostic FFPE samples matched the clinical response, with decreasing sensitivity in poor responders (P = 0.03). When the International Prognostic Index (IPI) was included in the prediction analysis, the separation between responders and non-responders improved (P = 0.001). Thirteen patients developed relapse, and five patients predicted sensitive to their second and third line treatment survived a median 1194 days, while eight patients predicted not sensitive to their second and third line treatment survived a median 187 days (90% CI: 485 days versus 227 days). Among the latter group it was predicted that four would have been sensitive to another second line treatment than the one they received. The predictions were almost the same when diagnostic biopsies were used as when relapse biopsies were used. These preliminary findings warrant testing in a larger cohort of relapse patients to confirm whether the miRNA based predictor can select the optimal second line treatment and increase survival.

Differential expression of miR-155 and miR-21 in tumor and stroma cells in diffuse large B-cell lymphoma.

OncomiRs miR-21 and miR-155 have been linked to lymphomagenesis, but information on their implication in diffuse large B-cell lymphoma (DLBCL) is limited. Here, we used locked nucleic acid-based in situ hybridization (ISH) detection techniques on formalin-fixed paraffin-embedded DLBCL tissue samples to identify miR-155 and miR-21 at the cellular level in 56 patients diagnosed with DLBCL, and compared them to miR array data. miR-155 was observed in tumor cells in 19/56 (33.9%) of the samples evaluated by ISH. miR-21 was localized to the stromal compartment in 41/56 (73.2%). A subset of these, 16/56 (28.6%), also showed labeling in tumor cells. When comparing ISH-scores and miR array data, miR-155 in tumor cells, identified by ISH, was associated with miR-155 expression in miR array data (P=0.030). Equally, miR-21 expression by miR array data were highly associated with miR-21 ISH-scores in the stromal cells (P=0.002), whereas no association between miR array data and ISH of miR-21 in tumor cells was observed (P=0.673). We found no association of miR-155 and miR-21 with overall survival or germinal center B-cell-like (GCB) versus non-GCB-like subtypes of DLBCL. In conclusion, miR-ISH added to the biological interpretation of miR expression in DLBCL compared with miR array data, but miR-155 and miR-21 ISH did not add prognostic information in this series.

Diffuse large B-cell lymphoma with combined TP53 mutation and MIR34A methylation: Another "double hit" lymphoma with very poor outcome?

MiR34A, B and C have been implicated in lymphomagenesis, but information on their role in normal CD19+ B-cells (PBL-B) and de novo diffuse large B-cell lymphoma (DLBCL) is limited. We show that in normal and activated B-cells miR34A-5p plays a dominant role compared to other miR34 family members. Only miR34A-5p is expressed in PBL-B, and significantly induced in activated B-cells and reactive lymph nodes. In PBL-B, the MIR34A and MIR34B/C promoters are unmethylated, but the latter shows enrichment for the H3K4me3/H3K27me3 silencing mark. Nine de novo DLBCL cases (n=150) carry both TP53 mutation and MIR34A methylation ("double hit") and these patients have an exceedingly poor prognosis with a median survival of 9.4 months (P<0.0001), while neither TP53 mutation, MIR34A or MIR34B/C promoter methylation alone ("single hit") influence on survival. The TP53/MIR34A "double-hit" is an independent negative prognostic factor for survival (P=0.0002). In 2 DLBCL-cell lines with both TP53 mutation and promoter methylation of MIR34A, miR34A-5p is upregulated by 5-aza-2'deoxycytidine. Thus, the TP53/MIR34A "double hit" characterizes a very aggressive subgroup of DLBCL, which may be treatable with epigenetic therapy prior to or in combination with conventional immunochemotherapy.

Genome-wide profiling identifies a DNA methylation signature that associates with TET2 mutations in diffuse large B-cell lymphoma.

The discovery that the Ten-Eleven Translocation (TET) hydroxylases cause DNA demethylation has fundamentally changed the notion of how DNA methylation is regulated. Clonal analysis of the hematopoetic stem cell compartment suggests that TET2 mutations can be early events in hematologic cancers and recent investigations have shown TET2 mutations in diffuse large B-cell lymphoma. However, the detection rates and the types of TET2 mutations vary, and the relation to global methylation patterns has not been investigated. Here, we show TET2 mutations in 12 of 100 diffuse large B-cell lymphomas with 7% carrying loss-of-function and 5% carrying missense mutations. Genome-wide methylation profiling using 450K Illumina arrays identified 315 differentially methylated genes between TET2 mutated and TET2 wild-type cases. TET2 mutations are primarily associated with hypermethylation within CpG islands (70%; P<0.0001), and at CpG-rich promoters (60%; P<0.0001) of genes involved in hematopoietic differentiation and cellular development. Hypermethylated loci in TET2 mutated samples overlap with the bivalent (H3K27me3/H3K4me3) silencing mark in human embryonic stem cells (P=1.5×10(-30)). Surprisingly, gene expression profiling showed that only 11% of the hypermethylated genes were down-regulated, among which there were several genes previously suggested to be tumor suppressors. A meta-analysis suggested that the 35 hypermethylated and down-regulated genes are associated with the activated B-cell-like type of diffuse large B-cell lymphoma in other studies. In conclusion, our data suggest that TET2 mutations may cause aberrant methylation mainly of genes involved in hematopoietic development, which are silenced but poised for activation in human embryonic stem cells.

MicroRNA profiling in ocular adnexal lymphoma: a role for MYC and NFKB1 mediated dysregulation of microRNA expression in aggressive disease.

PURPOSE: Ocular adnexal lymphoma (i.e., lymphoma with involvement of the orbit, eyelids, conjunctiva, lacrimal gland, and lacrimal sac), although rare, is common among malignant tumors involving the ocular adnexal region. The main subtypes are low-grade extranodal marginal zone lymphoma (EMZL) and aggressive diffuse large B-cell lymphoma (DLBCL). In rare cases, low-grade EMZL are reported to transform to DLBCL. It is unclear, however, which genetic events distinguish low-grade disease from aggressive, potentially fatal disease. METHODS: Using LNA-based arrays from Exiqon, we performed global microRNA (miRNA) expression profiling of 18 EMZLs and 25 DLBCLs involving ocular adnexal sites to investigate changes in the miRNA expression in low- versus high-grade disease. Findings were confirmed by real-time quantitative PCR (RTq-PCR). RESULTS: Our analysis revealed 43 miRNAs with altered expression profiles in DLBCL compared to EMZL. Seven of the miRNAs down-regulated in DLBCL relative to EMZL showed enrichment for a direct transcriptional repression by the oncoprotein MYC. We also report a possible loss-of-regulation of NFKB1 and its downstream miRNAs. In addition, our analysis identified a group of DLBCLs whose expression profiles resembled that of EMZL. Although transformation of EMZL to DLBCL in the ocular adnexal region is rare, we hypothesize that the intermediate group potentially may derive from transformation of EMZL that was not recognized by histology. CONCLUSIONS: We conclude that fundamental differences in miRNA expression exist between ocular adnexal EMZL and DLBCL, mainly due to differences in MYC and NF-ĸB regulatory pathways.

MicroRNA profiling in human neutrophils during bone marrow granulopoiesis and in vivo exudation.

The purpose of this study was to describe the microRNA (miRNA) expression profiles of neutrophils and their precursors from the initiation of granulopoiesis in the bone marrow to extravasation and accumulation in skin windows. We analyzed three different cell populations from human bone marrow, polymorphonuclear neutrophil (PMNs) from peripheral blood, and extravasated PMNs from skin windows using the Affymetrix 2.0 platform. Our data reveal 135 miRNAs differentially regulated during bone marrow granulopoiesis. The majority is differentially regulated between the myeloblast/promyelocyte (MB/PM) and myelocyte/metamyelocyte (MC/MM) stages of development. These 135 miRNAs were divided into six clusters according to the pattern of their expression. Several miRNAs demonstrate a pronounced increase or reduction at the transition between MB/PM and MC/MM, which is associated with cell cycle arrest and the initiation of terminal differentiation. Seven miRNAs are differentially up-regulated between peripheral blood PMNs and extravasated PMNs and only one of these (miR-132) is also differentially regulated during granulopoiesis. The study indicates that several different miRNAs participate in the regulation of normal granulopoiesis and that miRNAs might also regulate activities of extravasated neutrophils. The data present the miRNA profiles during the development and activation of the neutrophil granulocyte in healthy humans and thus serves as a reference for further research of normal and malignant granulocytic development.