Preclinical Development of CAR T Cells with Antigen-Inducible IL18 Enforcement to Treat GD2-Positive Solid Cancers.

PURPOSE: Cytokine-engineering of chimeric antigen receptor-redirected T cells (CAR T cells) is a promising principle to overcome the limited activity of canonical CAR T cells against solid cancers. EXPERIMENTAL DESIGN: We developed an investigational medicinal product, GD2IL18CART, consisting of CAR T cells directed against ganglioside GD2 with CAR-inducible IL18 to enhance their activation response and cytolytic effector functions in the tumor microenvironment. To allow stratification of patients according to tumor GD2 expression, we established and validated immunofluorescence detection of GD2 on paraffin-embedded tumor tissues. RESULTS: Lentiviral all-in-one vector engineering of human T cells with the GD2-specific CAR with and without inducible IL18 resulted in cell products with comparable proportions of CAR-expressing central memory T cells. Production of IL18 strictly depends on GD2 antigen engagement. GD2IL18CART respond to interaction with GD2-positive tumor cells with higher IFNγ and TNFα cytokine release and more effective target cytolysis compared with CAR T cells without inducible IL18. GD2IL18CART further have superior in vivo antitumor activity, with eradication of GD2-positive tumor xenografts. Finally, we established GMP-compliant manufacturing of GD2IL18CART and found it to be feasible and efficient at clinical scale. CONCLUSIONS: These results pave the way for clinical investigation of GD2IL18CART in pediatric and adult patients with neuroblastoma and other GD2-positive cancers (EU CT 2022-501725-21-00).

Flipped Classroom: Improved team performance during resuscitation training through interactive pre-course content - a cluster-randomised controlled study.

BACKGROUND: Resuscitation is a team effort, and it is increasingly acknowledged that team cooperation requires training. Staff shortages in many healthcare systems worldwide, as well as recent pandemic restrictions, limit opportunities for collaborative team training. To address this challenge, a learner-centred approach known as flipped learning has been successfully implemented. This model comprises self-directed, asynchronous pre-course learning, followed by knowledge application and skill training during in-class sessions. The existing evidence supports the effectiveness of this approach for the acquisition of cognitive skills, but it is uncertain whether the flipped classroom model is suitable for the acquisition of team skills. The objective of this study was to determine if a flipped classroom approach, with an online workshop prior to an instructor-led course could improve team performance and key resuscitation variables during classroom training. METHODS: A single-centre, cluster-randomised, rater-blinded study was conducted on 114 final year medical students at a University Hospital in Germany. The study randomly assigned students to either the intervention or control group using a computer script. Each team, regardless of group, performed two advanced life support (ALS) scenarios on a simulator. The two groups differed in the order in which they completed the flipped e-learning curriculum. The intervention group started with the e-learning component, and the control group started with an ALS scenario. Simulators were used for recording and analysing resuscitation performance indicators, while professionals assessed team performance as a primary outcome. RESULTS: The analysis was conducted on the data of 96 participants in 21 teams, comprising of 11 intervention groups and 10 control groups. The intervention teams achieved higher team performance ratings during the first scenario compared to the control teams (Estimated marginal mean of global rating: 7.5 vs 5.6, p < 0.01; performance score: 4.4 vs 3.8, p < 0.05; global score: 4.4 vs 3.7, p < 0.001). However, these differences were not observed in the second scenario, where both study groups had used the e-learning tool. CONCLUSION: Flipped classroom approaches using learner-paced e-learning prior to hands-on training can improve team performance. TRIAL REGISTRATION: German Clinical Trials Register ( https://drks.de/search/de/trial/DRKS00013096 ).

Efficacy of an Internet- and Mobile-Based Intervention for Subclinical Anxiety and Depression (ICare Prevent) with Two Guidance Formats: Results from a Three-Armed Randomized Controlled Trial.

INTRODUCTION: Limited research exists on intervention efficacy for comorbid subclinical anxiety and depressive disorders, despite their common co-occurrence. Internet- and mobile-based interventions (IMIs) are promising to reach individuals facing subclinical symptoms. OBJECTIVE: This study aimed to evaluate the efficacy of a transdiagnostic and self-tailored IMI in reducing subclinical anxiety and depressive symptom severity with either individualized (IG-IMI) or automated (AG-IMI) guidance compared to a waitlist control group with care-as-usual access (WLC). METHODS: Participants included 566 adults with subclinical anxiety (GAD-7 ≥ 5) and/or depressive (CES-D ≥16) symptoms, who did not meet criteria for a full-syndrome depressive or anxiety disorder. In a three-arm randomized clinical trial, participants were randomized to a cognitive behavioral 7-session IMI plus booster session with IG-IMI (n = 186) or AG-IMI (n = 189) or WLC (n = 191). Primary outcomes included observer-rated anxiety (HAM-A) and depressive (QIDS) symptom severity 8 weeks after randomization assessed by blinded raters via telephone. Follow-up outcomes at 6 and 12 months are reported. RESULTS: Symptom severity was significantly lower with small to medium effects in IG-IMI (anxiety: d = 0.45, depression: d = 0.43) and AG-IMI (anxiety: d = 0.31, depression: d = 0.32) compared to WLC. No significant differences emerged between guidance formats in primary outcomes. There was a significant effect in HAM-A after 6 months favoring AG-IMI. On average, participants completed 85.38% of IG-IMI and 77.38% of AG-IMI. CONCLUSIONS: A transdiagnostic, self-tailored IMI can reduce subclinical anxiety and depressive symptom severity, but 12-month long-term effects were absent. Automated guidance holds promise for enhancing the scalability of IMIs in broad prevention initiatives.

A statistical framework for planning and analysing test-retest studies of repeatability.

There is an increasing number of potential quantitative biomarkers that could allow for early assessment of treatment response or disease progression. However, measurements of such biomarkers are subject to random variability. Hence, differences of a biomarker in longitudinal measurements do not necessarily represent real change but might be caused by this random measurement variability. Before utilizing a quantitative biomarker in longitudinal studies, it is therefore essential to assess the measurement repeatability. Measurement repeatability obtained from test-retest studies can be quantified by the repeatability coefficient, which is then used in the subsequent longitudinal study to determine if a measured difference represents real change or is within the range of expected random measurement variability. The quality of the point estimate of the repeatability coefficient, therefore, directly governs the assessment quality of the longitudinal study. Repeatability coefficient estimation accuracy depends on the case number in the test-retest study, but despite its pivotal role, no comprehensive framework for sample size calculation of test-retest studies exists. To address this issue, we have established such a framework, which allows for flexible sample size calculation of test-retest studies, based upon newly introduced criteria concerning assessment quality in the longitudinal study. This also permits retrospective assessment of prior test-retest studies.

Using web-based, guided self-help to bridge the waiting time for face-to-face out-patient treatment for bulimic-spectrum disorders: randomised controlled trial.

BACKGROUND: Although effective treatments for bulimic-spectrum eating disorders exist, access is often delayed because of limited therapist availability and lengthy waiting lists. Web-based self-help interventions have the potential to bridge waiting times for face-to-face treatment and overcome existing treatment gaps. AIMS: This study aims to assess the effectiveness of a web-based guided self-help intervention (everyBody Plus) for patients with bulimia nervosa, binge eating disorder and other specified feeding and eating disorders who are waiting for out-patient treatment. METHOD: A randomised controlled trial was conducted in Germany and the UK. A total of 343 patients were randomly assigned to the intervention 'everyBody Plus' or a waitlist control condition. The primary outcome was the number of weeks after randomisation until a patient achieved a clinically relevant improvement in core symptoms for the first time. Secondary outcomes included eating disorder attitudes and behaviours, and general psychopathology. RESULTS: At 6- and 12-month follow-up, the probability of being abstinent from core symptoms was significantly larger for the intervention group compared with the control group (hazard ratio: 1.997, 95% CI 1.09-3.65; P = 0.0249). The intervention group also showed larger improvements in eating disorder attitudes and behaviours, general psychopathology, anxiety, depression and quality of life, compared with the control group at most assessment points. Working alliance ratings with the online therapist were high. CONCLUSIONS: The self-help intervention everyBody Plus, delivered with relatively standardised online guidance, can help bridge treatment gaps for patients with bulimic-spectrum eating disorders, and achieve faster and greater reductions in core symptoms.

Thrombopoietin levels in sepsis and septic shock - a systematic review and meta-analysis.

OBJECTIVES: Sepsis is a life-threatening condition implicating an inadequate activation of the immune system. Platelets act as modulators and contributors to immune processes. Indeed, altered platelet turnover, thrombotic events, and changes in thrombopoietin levels in systemic inflammation have been reported, but thrombopoietin-levels in sepsis and septic-shock have not yet been systematically evaluated. We therefore performed a meta-analysis of thrombopoietin (TPO)-levels in patients with sepsis. METHODS: Two independent reviewers screened records and full-text articles for inclusion. Scientific databases were searched for studies examining thrombopoietin levels in adult sepsis and septic-shock patients until August 1st 2022. RESULTS: Of 95 items screened, six studies met the inclusion criteria, including 598 subjects. Both sepsis and severe sepsis were associated with increased levels of thrombopoietin (sepsis vs. control: standardized mean difference 3.06, 95 % CI 1.35-4.77; Z=3.50, p=0.0005) (sepsis vs. severe sepsis: standardized mean difference -1.67, 95 % CI -2.46 to -0.88; Z=4.14, p<0.0001). TPO-levels did not show significant differences between severe sepsis and septic shock patients but differed between sepsis and inflammation-associated non-septic controls. Overall, high heterogeneity and low sample size could be noted. CONCLUSIONS: Concluding, increased levels of thrombopoietin appear to be present both in sepsis and severe sepsis with high heterogeneity but thrombopoietin does not allow to differentiate between severe sepsis and septic-shock. TPO may potentially serve to differentiate sepsis from non-septic trauma and/or tissue damage related (systemic) inflammation. Usage of different assays and high heterogeneity demand standardization of methods and further large multicenter trials.

Adjusting Acute Kidney Injury Kidney Disease: Improving Global Outcomes Urine Output Criterion for Predicted Body Weight Improves Prediction of Hospital Mortality.

BACKGROUND: Based on the Kidney Disease: Improving Global Outcomes (KDIGO) definitions, urine output, serum creatinine, and need for kidney replacement therapy are used for staging acute kidney injury (AKI). Currently, AKI staging correlates strongly with mortality and can be used as a predictive tool. However, factors associated with the development of AKI may affect its predictive ability. We tested whether adjustment for predicted (versus actual) body weight improved the ability of AKI staging to predict hospital mortality. METHODS: A total of 3279 patients who had undergone cardiac surgery in a university hospital were retrospectively analyzed. AKI was staged according to KDIGO criteria (standard staging) and after adjustment for hourly urine output adjusted by predicted body weight for each patient and each day of their hospital stay. RESULTS: The incidence of AKI (all stages) was 43% (predicted body weight adjusted) and 50% (standard staging), respectively ( P < .001). In sensitivity-specificity analyses for predicting hospital mortality, the area under the curve was significantly higher after adjustment for predicted body weight than with standard staging ( P = .002). CONCLUSIONS: Compared to standard staging, adjustment of urine output for predicted body weight increases the specificity and improves prediction of hospital mortality in patients undergoing cardiac surgery.

Sex-associated differences in frequencies and prognostic impact of recurrent genetic alterations in adult acute myeloid leukemia (Alliance, AMLCG).

Clinical outcome of patients with acute myeloid leukemia (AML) is associated with demographic and genetic features. Although the associations of acquired genetic alterations with patients' sex have been recently analyzed, their impact on outcome of female and male patients has not yet been comprehensively assessed. We performed mutational profiling, cytogenetic and outcome analyses in 1726 adults with AML (749 female and 977 male) treated on frontline Alliance for Clinical Trials in Oncology protocols. A validation cohort comprised 465 women and 489 men treated on frontline protocols of the German AML Cooperative Group. Compared with men, women more often had normal karyotype, FLT3-ITD, DNMT3A, NPM1 and WT1 mutations and less often complex karyotype, ASXL1, SRSF2, U2AF1, RUNX1, or KIT mutations. More women were in the 2022 European LeukemiaNet intermediate-risk group and more men in adverse-risk group. We found sex differences in co-occurring mutation patterns and prognostic impact of select genetic alterations. The mutation-associated splicing events and gene-expression profiles also differed between sexes. In patients aged <60 years, SF3B1 mutations were male-specific adverse outcome prognosticators. We conclude that sex differences in AML-associated genetic alterations and mutation-specific differential splicing events highlight the importance of patients' sex in analyses of AML biology and prognostication.

Donor Proteinuria and Allograft Function in Kidney Transplantation: Short- and Long-Term Results From a Retrospective Cohort Study.

Donor proteinuria (DP) is a common but rarely evaluated aspect of today's kidney transplant allocation process. While proteinuria after kidney transplantation is a risk factor for impaired graft function and survival, the long-term effects of DP in kidney transplantation have not yet been evaluated. Therefore, this study aims to investigate the impact of DP on the long-term outcome after kidney transplantation. A total of 587 patients were found to be eligible and were stratified into two groups: (1) those receiving a graft from a donor without proteinuria (DP-) and (2) those receiving a graft from a donor with proteinuria (DP+). At 36 months, there was no difference in the primary composite endpoint including graft loss and patient survival (log-rank test, p = 0.377). However, the analysis of DP+ subgroups showed a significant decrease in overall patient survival in the group with high DP (p = 0.017). DP did not adversely affect patient or graft survival over 36 months. Nevertheless, this work revealed a trend towards decreased overall survival of patients with severe proteinuria in the subgroup analysis. Therefore, the underlying results suggest caution in allocating kidneys from donors with high levels of proteinuria.

WATCH-BPM-Comparison of a WATCH-Type Blood Pressure Monitor with a Conventional Ambulatory Blood Pressure Monitor and Auscultatory Sphygmomanometry.

BACKGROUND: Smart devices that are able to measure blood pressure (BP) are valuable for hypertension or heart failure management using digital technology. Data regarding their diagnostic accuracy in comparison to standard noninvasive measurement in accordance to Riva-Rocci are sparse. This study compared a wearable watch-type oscillometric BP monitor (Omron HeartGuide), a wearable watch-type infrared BP monitor (Smart Wear), a conventional ambulatory BP monitor, and auscultatory sphygmomanometry. METHODS: Therefore, 159 consecutive patients (84 male, 75 female, mean age 64.33 ± 16.14 years) performed observed single measurements with the smart device compared to auscultatory sphygmomanometry (n = 109) or multiple measurements during 24 h compared to a conventional ambulatory BP monitor on the upper arm (n = 50). The two BP monitoring devices were simultaneously worn on the same arm throughout the monitoring period. In a subgroup of 50 patients, single measurements were also performed with an additional infrared smart device. RESULTS: The intraclass correlation coefficient (ICC) between the difference and the mean of the oscillometric Omron HeartGuide and the conventional method for the single measurement was calculated for both systole (0.765) and diastole (0.732). This is exactly how the ICC was calculated for the individual mean values calculated over the 24 h long-term measurement of the individual patients for both systole (0.880) and diastole (0.829). The ICC between the infrared device and the conventional method was "bad" for SBP (0.329) and DBP (0.025). Therefore, no further long-term measurements were performed with the infrared device. CONCLUSION: The Omron HeartGuide device provided comparable BP values to the standard devices for single and long-term measurements. The infrared smart device failed to acquire valid measurement data.

Diffusion-Weighted MRI for Treatment Response Assessment in Osteoblastic Metastases-A Repeatability Study.

The apparent diffusion coefficient (ADC) is a candidate marker of treatment response in osteoblastic metastases that are not evaluable by morphologic imaging. However, it is unclear whether the ADC meets the basic requirement for reliable treatment response evaluation, namely a low variance of repeated measurements in relation to the differences found between viable and nonviable metastases. The present study addresses this question by analyzing repeated in vivo ADCmedian measurements of 65 osteoblastic metastases in nine patients, as well as phantom measurements. PSMA-PET served as a surrogate for bone metastasis viability. Measures quantifying repeatability were calculated and differences in mean ADC values according to PSMA-PET status were examined. The relative repeatability coefficient %RC of ADCmedian measurements was 5.8% and 12.9% for phantom and in vivo measurements, respectively. ADCmedian values of bone metastases ranged from 595×10-6mm2/s to 2090×10-6mm2/s with an average of 63% higher values in nonviable metastases compared with viable metastases (p < 0.001). ADC shows a small repeatability coefficient in relation to the difference in ADC values between viable and nonviable metastases. Therefore, ADC measurements fulfill the technical prerequisite for reliable treatment response evaluation in osteoblastic metastases.

Which risk factors significantly influence the outcome of traumatic brain injured patients with alcohol use disorder?

PURPOSE: Every year, approximately 10 million people worldwide suffer a traumatic brain injury that leads to hospitalization or mortality. Chronic and acute alcohol intoxication increase the risk of developing traumatic brain injury. Alcohol use disorder (AUD) as a predictor of outcome in neurosurgical patients and the definition of risk factors have been sparsely addressed so far. This study aims to improve the understanding of the effects of alcohol use disorder in the context of neurosurgical therapy. METHODS: This study included patients admitted to Münster University Hospital with a traumatic brain injury and alcohol use disorder from January 1, 2010, to December 31, 2018. Univariate and multivariate analyses were performed to identify risk factors for a poorer outcome, assessed by the Glasgow Outcome Score. RESULTS: Of the 197 patients included, 156 (79%) were male, and 41 (21%) were female, with a median age of 49 years (IQR 38-58 years). In multivariate analyses, age (p < 0.001), the occurrence of a new neurologic deficit (p < 0.001), the development of hydrocephalus (p = 0.005), and CT-graphic midline shift due to intracerebral hemorrhage (p = 0.008) emerged as significant predictors of a worse outcome (GOS 1-3). In addition, the level of blood alcohol concentration correlated significantly with the occurrence of seizures (p = 0.009). CONCLUSIONS: Early identification of risk factors in patients with alcohol use disorder and traumatic brain injury is crucial to improve the outcome. In this regard, the occurrence of hydrocephalus or seizures during the inpatient stay should be considered as cause of neurological deterioration in this patient group.

Bronchoscopist-Directed Continuous Flow Propofol Based Analgosedation during Flexible Interventional Bronchoscopy and EBUS.

Sedation techniques in interventional flexible bronchoscopy and endobronchial ultrasound-guided transbronchial-needle aspiration (EBUS-TBNA) are inconsistent and the evidence for required general anesthesia under full anesthesiologic involvement is scarce. Moreover, we faced the challenge of providing bronchoscopic care with limited personnel. Hence, we retrospectively identified 513 patients that underwent flexible interventional bronchoscopy and/or EBUS-TBNA out of our institution between January 2020 and August 2022 to evaluate our deep analgosedation approach based on pethidine/meperidine bolus plus continuous flow adjusted propofol, the bronchoscopist-directed continuous flow propofol based analgosedation (BDcfP) in a two-personnel setting. Consequently, 502 out of 513 patients received BDcfP for analgosedation. We identified cardiovascular comorbidities, chronic obstructive pulmonary disease, and arterial hypertension as risk factors for periprocedural hypotension. Propofol flow rate did not correlate with hypotension. Theodrenaline and cafedrine might be used to treat periprocedural hypotension. Moreover, midazolam might be used to support the sedative effect. In conclusion, BDcfP is a safe and feasible sedative approach during interventional flexible bronchoscopy and EBUS-TBNA. In general, after the implementation of safety measures, EBUS-TBNA and interventional flexible bronchoscopy via BDcfP might safely be performed even with limited personnel.

Acute megakaryoblastic leukaemia shows high frequency of chromosome 1q aberrations and dismal outcome.

Acute megakaryoblastic leukaemia (AMKL) is associated with poor prognosis. Limited information is available on its cytogenetics, molecular genetics and clinical outcome. We performed genetic analyses, evaluated prognostic factors and the value of allogeneic haematopoietic stem cell transplantation (allo-HSCT) in a homogenous adult AMKL patient cohort. We retrospectively analysed 38 adult patients with AMKL (median age: 58 years, range: 21-80). Most received intensive treatment in AML Cooperative Group (AMLCG) trials between 2001 and 2016. Cytogenetic data showed an accumulation of adverse risk markers according to ELN 2017 and an unexpected high frequency of structural aberrations on chromosome arm 1q (33%). Most frequently, mutations occurred in TET2 (23%), TP53 (23%), JAK2 (19%), PTPN11 (19%) and RUNX1 (15%). Complete remission rate in 33 patients receiving intensive chemotherapy was 33% and median overall survival (OS) was 33 weeks (95% CI: 21-45). Patients undergoing allo-HSCT (n = 14) had a superior median OS (68 weeks; 95% CI: 11-126) and relapse-free survival (RFS) of 27 weeks (95% CI: 4-50), although cumulative incidence of relapse after allo-HSCT was high (62%). The prognosis of AMKL is determined by adverse genetic risk factors and therapy resistance. So far allo-HSCT is the only potentially curative treatment option in this dismal AML subgroup.

CAR T cells as micropharmacies against solid cancers: Combining effector T-cell mediated cell death with vascular targeting in a one-step engineering process.

To enhance the potency of chimeric antigen receptor (CAR) engineered T cells in solid cancers, we designed a novel cell-based combination strategy with an additional therapeutic mode of action. CAR T cells are used as micropharmacies to produce a targeted pro-coagulatory fusion protein, truncated tissue factor (tTF)-NGR, which exerts pro-coagulatory activity and hypoxia upon relocalization to the vascular endothelial cells that invade tumor tissues. Delivery by CAR T cells aimed to induce locoregional tumor vascular infarction for combined immune-mediated and hypoxic tumor cell death. Human T cells that were one-vector gene-modified to express a GD2-specific CAR along with CAR-inducible tTF-NGR exerted potent GD2-specific effector functions while secreting tTF-NGR that activates the extrinsic coagulation pathway in a strictly GD2-dependent manner. In murine models, the CAR T cells infiltrated GD2-positive tumor xenografts, secreted tTF-NGR into the tumor microenvironment and showed a trend towards superior therapeutic activity compared with control cells producing functionally inactive tTF-NGR. In vitro evidence supports a mechanism of hypoxia-mediated enhancement of T cell cytolytic activity. We conclude that combined CAR T cell targeting with an additional mechanism of antitumor action in a one-vector engineering strategy is a promising approach to be further developed for targeted treatment of solid cancers.

Radiomics of Tumor Heterogeneity in 18F-FDG-PET-CT for Predicting Response to Immune Checkpoint Inhibition in Therapy-Naïve Patients with Advanced Non-Small-Cell Lung Cancer.

We aimed to evaluate the predictive and prognostic value of baseline 18F-FDG-PET-CT (PET-CT) radiomic features (RFs) for immune checkpoint-inhibitor (CKI)-based first-line therapy in advanced non-small-cell lung cancer (NSCLC) patients. In this retrospective study 44 patients were included. Patients were treated with either CKI-monotherapy or combined CKI-based immunotherapy-chemotherapy as first-line treatment. Treatment response was assessed by the Response Evaluation Criteria in Solid Tumors (RECIST). After a median follow-up of 6.4 months patients were stratified into "responder" (n = 33) and "non-responder" (n = 11). RFs were extracted from baseline PET and CT data after segmenting PET-positive tumor volume of all lesions. A Radiomics-based model was developed based on a Radiomics signature consisting of reliable RFs that allow classification of response and overall progression using multivariate logistic regression. These RF were additionally tested for their prognostic value in all patients by applying a model-derived threshold. Two independent PET-based RFs differentiated well between responders and non-responders. For predicting response, the area under the curve (AUC) was 0.69 for "PET-Skewness" and 0.75 predicting overall progression for "PET-Median". In terms of progression-free survival analysis, patients with a lower value of PET-Skewness (threshold < 0.2014; hazard ratio (HR) 0.17, 95% CI 0.06-0.46; p < 0.001) and higher value of PET-Median (threshold > 0.5233; HR 0.23, 95% CI 0.11-0.49; p < 0.001) had a significantly lower probability of disease progression or death. Our Radiomics-based model might be able to predict response in advanced NSCLC patients treated with CKI-based first-line therapy.

Unsupervised meta-clustering identifies risk clusters in acute myeloid leukemia based on clinical and genetic profiles.

BACKGROUND: Increasingly large and complex biomedical data sets challenge conventional hypothesis-driven analytical approaches, however, data-driven unsupervised learning can detect inherent patterns in such data sets. METHODS: While unsupervised analysis in the medical literature commonly only utilizes a single clustering algorithm for a given data set, we developed a large-scale model with 605 different combinations of target dimensionalities as well as transformation and clustering algorithms and subsequent meta-clustering of individual results. With this model, we investigated a large cohort of 1383 patients from 59 centers in Germany with newly diagnosed acute myeloid leukemia for whom 212 clinical, laboratory, cytogenetic and molecular genetic parameters were available. RESULTS: Unsupervised learning identifies four distinct patient clusters, and statistical analysis shows significant differences in rate of complete remissions, event-free, relapse-free and overall survival between the four clusters. In comparison to the standard-of-care hypothesis-driven European Leukemia Net (ELN2017) risk stratification model, we find all three ELN2017 risk categories being represented in all four clusters in varying proportions indicating unappreciated complexity of AML biology in current established risk stratification models. Further, by using assigned clusters as labels we subsequently train a supervised model to validate cluster assignments on a large external multicenter cohort of 664 intensively treated AML patients. CONCLUSIONS: Dynamic data-driven models are likely more suitable for risk stratification in the context of increasingly complex medical data than rigid hypothesis-driven models to allow for a more personalized treatment allocation and gain novel insights into disease biology.

There are various ways in which clinicians can predict the risk of disease progression in patients with leukemia, helping them to treat the patients accordingly. However, these approaches are usually designed by human experts and might not fully capture the complexity of a patient's disease. Here, with a large cohort of patients with acute myeloid leukemia, we design an unsupervised machine learning model ­ a type of computer model that learns from patterns in data without human input­to separate these patients into subgroups according to risk. We identify four distinct groups which differ with regards to patient genetics, laboratory values, and clinical characteristics. These groups have differences in response to treatment and patient survival, and we validate our findings in another dataset. Our approach might help clinicians to better predict outcomes in patients with leukemia and make decisions on treatment.

Validation and refinement of the 2022 European LeukemiaNet genetic risk stratification of acute myeloid leukemia.

The revised 2022 European LeukemiaNet (ELN) AML risk stratification system requires validation in large, homogeneously treated cohorts. We studied 1118 newly diagnosed AML patients (median age, 58 years; range, 18-86 years) who received cytarabine-based induction chemotherapy between 1999 and 2012 and compared ELN-2022 to the previous ELN-2017 risk classification. Key findings were validated in a cohort of 1160 mostly younger patients. ELN-2022 reclassified 15% of patients, 3% into more favorable, and 12% into more adverse risk groups. This was mainly driven by patients reclassified from intermediate- to adverse-risk based on additional myelodysplasia-related mutations being included as adverse-risk markers. These patients (n = 79) had significantly better outcomes than patients with other adverse-risk genotypes (5-year OS, 26% vs. 12%) and resembled the remaining intermediate-risk group. Overall, time-dependent ROC curves and Harrel's C-index controlling for age, sex, and AML type (de novo vs. sAML/tAML) show slightly worse prognostic discrimination of ELN-2022 compared to ELN-2017 for OS. Further refinement of ELN-2022 without including additional genetic markers is possible, in particular by recognizing TP53-mutated patients with complex karyotypes as "very adverse". In summary, the ELN-2022 risk classification identifies a larger group of adverse-risk patients at the cost of slightly reduced prognostic accuracy compared to ELN-2017.