Clinicopathological analysis of diffuse large B-cell lymphoma using molecular biomarkers: a retrospective analysis from 7 Hungarian centers

Background

The clinical and genetic heterogeneity of diffuse large B-cell lymphoma (DLBCL) presents distinct challenges in predicting response to therapy and overall prognosis. The main objective of this study was to assess the application of the immunohistochemistry- and interphase fluorescence in situ hybridization (FISH)-based molecular markers in the diagnosis of DLBCL and its prognostic value in patients treated with rituximab-based immunochemotherapy.

Methods

This is a multicenter, retrospective study, which analyzed data from 7 Hungarian hematology centers. Eligible patients were adults, had a histologically confirmed diagnosis of DLBCL, were treated with rituximab-based immunochemotherapy in the first line, and had available clinicopathological data including International Prognostic Index (IPI). On the specimens, immunohistochemistry and FISH methods were performed. Germinal center B-cell like (GCB) and non-GCB subtypes were classified by the Hans algorithm. Outcomes included overall survival (OS), event-free survival (EFS), and EFS at 2 years (EFS24). For survival analysis, we used Kaplan-Meier curves with the log-rank test and multivariate Cox regression.

Results

A total of 247 DLBCL cases were included. Cases were positive for MYC, BCL2, BCL6, and MUM1 expression in 52.1%, 66.2%, 72.6%, and 77.8%, respectively. BCL6 translocation, BCL2 gene copy number (GCN) gain, IGH::MYC translocation, MYC GCN gain, IGH::BCL2 translocation, and BCL6 GCN gain were detected in 21.4%, 14.1%, 7.3%, 1.8%, 7.3%, and 0.9%, respectively. At a median follow-up of 52 months, 140 patients (56.7%) had disease progression or relapse. The Kaplan-Meier estimate for EFS24 was 56.2% (CI: 50.4-62.8%). In univariate analysis, only IPI and BCL6 expression were significant predictors of both OS and EFS, whereas MUM1 predicted EFS only. In multivariate analysis, the IPI score was a significant independent negative, whereas MIB-1 and BCL6 protein expressions were significant independent positive predictors of both OS and EFS.

Conclusion

In our study, we found that only IPI, BCL6 protein expression and MIB-1 protein expression are independent predictors of survival outcomes in DLBCL. We did not find any difference in survival by GCB vs. non-GCB subtypes. These findings may improve prognostication in DLBCL and can contribute to designing further research in the area.

Automated data preparation for in vivo tumor characterization with machine learning

Background

This study proposes machine learning-driven data preparation (MLDP) for optimal data preparation (DP) prior to building prediction models for cancer cohorts.

Methods

A collection of well-established DP methods were incorporated for building the DP pipelines for various clinical cohorts prior to machine learning. Evolutionary algorithm principles combined with hyperparameter optimization were employed to iteratively select the best fitting subset of data preparation algorithms for the given dataset. The proposed method was validated for glioma and prostate single center cohorts by 100-fold Monte Carlo (MC) cross-validation scheme with 80-20% training-validation split ratio. In addition, a dual-center diffuse large B-cell lymphoma (DLBCL) cohort was utilized with Center 1 as training and Center 2 as independent validation datasets to predict cohort-specific clinical endpoints. Five machine learning (ML) classifiers were employed for building prediction models across all analyzed cohorts. Predictive performance was estimated by confusion matrix analytics over the validation sets of each cohort. The performance of each model with and without MLDP, as well as with manually-defined DP were compared in each of the four cohorts.

Results

Sixteen of twenty established predictive models demonstrated area under the receiver operator characteristics curve (AUC) performance increase utilizing the MLDP. The MLDP resulted in the highest performance increase for random forest (RF) (+0.16 AUC) and support vector machine (SVM) (+0.13 AUC) model schemes for predicting 36-months survival in the glioma cohort. Single center cohorts resulted in complex (6-7 DP steps) DP pipelines, with a high occurrence of outlier detection, feature selection and synthetic majority oversampling technique (SMOTE). In contrast, the optimal DP pipeline for the dual-center DLBCL cohort only included outlier detection and SMOTE DP steps.

Conclusions

This study demonstrates that data preparation prior to ML prediction model building in cancer cohorts shall be ML-driven itself, yielding optimal prediction models in both single and multi-centric settings.

90Y-ibritumomab Tiuxetan in B-cell Non-Hodgkin Lymphomas: Real-world Data From the United Arab Emirates

Purpose

B-cell non-Hodgkin lymphomas (NHLs) are significant contributors to cancer-related mortality. In this single-arm, retrospective cohort study, we aimed to examine the outcomes of a radioimmunotherapeutic modality, ⁹⁰Y-labeled ibritumomab tiuxetan (⁹⁰YIT) in B-cell NHLs.

Methods and Materials

We conducted this study based on data from the United Arab Emirates lymphoma registry. All patients with NHL subjected to ⁹⁰YIT were eligible for inclusion. The country of research lacked a national autologous stem cell transplantation (ASCT) center, but many ASCT-eligible patients received ⁹⁰YIT. We investigated overall survival (OS) and event-free survival (EFS), as well as safety outcomes.

Results

Between 2004 and 2008, 54 of 111 patients with B-cell NHL received radioimmunotherapy. The therapy was applied as first-line treatment in 18 cases (33.3%) and second- or later-line treatment in 36 cases (66.7%). All patients were evaluable for response. The first-line group consisted mainly of follicular lymphoma cases, and 3 of 18 patients died (16.7%) during the follow-up (range, 22-67 months). Median OS was not reached. No progression occurred after treatment (median EFS, 36.5 months [Q₁-Q₃ range, 30.5-44 months]). The second- or later-line group consisted mainly of diffuse large B-cell lymphoma cases, and 3 of 36 patients died (8.3%) during the follow-up (range, 4-68 months). Median OS was not reached. One case of progression was registered (median EFS: 33 months [Q₁-Q₃ range, 30.5-44 months]). ⁹⁰YIT had acceptable short- and long-term safety profiles.

Conclusions

The findings suggest that patients with NHL may benefit from ⁹⁰YIT as salvage treatment if ASCT is not available; however, this should be validated in randomized studies.

Two-Year Event-Free Survival Prediction in DLBCL Patients Based on In Vivo Radiomics and Clinical Parameters

Purpose

For the identification of high-risk patients in diffuse large B-cell lymphoma (DLBCL), we investigated the prognostic significance of in vivo radiomics derived from baseline [¹⁸F]FDG PET/CT and clinical parameters.

Methods

Pre-treatment [¹⁸F]FDG PET/CT scans of 85 patients diagnosed with DLBCL were assessed. The scans were carried out in two clinical centers. Two-year event-free survival (EFS) was defined. After delineation of lymphoma lesions, conventional PET parameters and in vivo radiomics were extracted. For 2-year EFS prognosis assessment, the Center 1 dataset was utilized as the training set and underwent automated machine learning analysis. The dataset of Center 2 was utilized as an independent test set to validate the established predictive model built by the dataset of Center 1.

Results

The automated machine learning analysis of the Center 1 dataset revealed that the most important features for building 2-year EFS are as follows: max diameter, neighbor gray tone difference matrix (NGTDM) busyness, total lesion glycolysis, total metabolic tumor volume, and NGTDM coarseness. The predictive model built on the Center 1 dataset yielded 79% sensitivity, 83% specificity, 69% positive predictive value, 89% negative predictive value, and 0.85 AUC by evaluating the Center 2 dataset.

Conclusion

Based on our dual-center retrospective analysis, predicting 2-year EFS built on imaging features is feasible by utilizing high-performance automated machine learning.

Bone structure and density via HR-pQCT in 60d bed-rest, 2-years recovery with and without countermeasures

We examined the effects of bed-rest, recovery and exercise countermeasures on bone density and structure at the distal tibia and radius as measured via high-resolution peripheral computed tomography. 24 subjects underwent 60-days of head-down tilt bed-rest and performed either resistive vibration exercise (RVE; n = 7), resistive exercise only (RE; n = 8) or no exercise (n = 9; 2nd Berlin BedRest Study; BBR2-2). Measurements were performed regularly during and up to 2-years after 60d bed-rest. At the distal tibia marked reductions in cortical area, cortical thickness and bone density but increases in periosteal perimeter and trabecular area were seen (p all<0.001). Recovery of most parameters occurred within 180d after bed-rest. At the distal radius, persistent increases in cortical area, cortical thickness, cortical density and total density and decreases in trabecular area were seen (p all ≤ 0.005). A significant effect of RVE (p = 0.003), but not RE, was seen on cortical area at the distal tibia, with few effects of the countermeasures observed on the remaining parameters. The current study represents the first implementation of high-resolution peripheral computed tomography in bed-rest in male subjects and helps to understand the patterns of bone remodeling due to bed-rest and recovery.