SurvIAE: Survival prediction with Interpretable Autoencoders from Diffuse Large B-Cells Lymphoma gene expression data.

BACKGROUND: In Diffuse Large B-Cell Lymphoma (DLBCL), several methodologies are emerging to derive novel biomarkers to be incorporated in the risk assessment. We realized a pipeline that relies on autoencoders (AE) and Explainable Artificial Intelligence (XAI) to stratify prognosis and derive a gene-based signature. METHODS: AE was exploited to learn an unsupervised representation of the gene expression (GE) from three publicly available datasets, each with its own technology. Multi-layer perceptron (MLP) was used to classify prognosis from latent representation. GE data were preprocessed as normalized, scaled, and standardized. Four different AE architectures (Large, Medium, Small and Extra Small) were compared to find the most suitable for GE data. The joint AE-MLP classified patients on six different outcomes: overall survival at 12, 36, 60 months and progression-free survival (PFS) at 12, 36, 60 months. XAI techniques were used to derive a gene-based signature aimed at refining the Revised International Prognostic Index (R-IPI) risk, which was validated in a fourth independent publicly available dataset. We named our tool SurvIAE: Survival prediction with Interpretable AE. RESULTS: From the latent space of AEs, we observed that scaled and standardized data reduced the batch effect. SurvIAE models outperformed R-IPI with Matthews Correlation Coefficient up to 0.42 vs. 0.18 for the validation-set (PFS36) and to 0.30 vs. 0.19 for the test-set (PFS60). We selected the SurvIAE-Small-PFS36 as the best model and, from its gene signature, we stratified patients in three risk groups: R-IPI Poor patients with High levels of GAB1, R-IPI Poor patients with Low levels of GAB1 or R-IPI Good/Very Good patients with Low levels of GPR132, and R-IPI Good/Very Good patients with High levels of GPR132. CONCLUSIONS: SurvIAE showed the potential to derive a gene signature with translational purpose in DLBCL. The pipeline was made publicly available and can be reused for other pathologies.

ARGO 2.0: a Hybrid NLP/ML Framework for Diagnosis Standardization.

A relevant problem in medicine is the standardization of the diagnosis associated with a clinical case. Although diagnosis formulation is an intrinsically subjective and uncertain process, its standardization may take benefit from digital solutions automating the routines at the basis of such a decision. In this work, we propose ARGO 2.0: a framework for the development of decision support systems for diagnosis formulation. The framework can read free-text reports and store their clinically relevant information as personalized electronic Case Report Forms. A hybrid strategy, exploiting the synergy of Natural Language Processing and Machine Learning techniques, is used to automatically suggest a diagnosis in a standardized fashion. ARGO 2.0 has been designed to be template-independent and easily tailored to specific medical fields. We here demonstrate its feasibility in hemo lympho-pathology, by detailing its implementation, object of an ongoing validation campaign in a standing medical institute. ARGO 2.0 achieved an average Accuracy of 95.07%, an average precision of 94.85%, an average Recall of 96.31% and a F-Score of 95.32% onto the test set, outperforming both its embedded components, based on Natural Language Processing and Machine Learning.

Candidate germline biomarkers of lenalidomide efficacy in mantle cell lymphoma: the Fondazione Italiana Linfomi MCL0208 trial.

In the Fondazione Italiana Linfomi MCL0208 phase 3 trial, lenalidomide maintenance (LEN) after autologous stem cell transplantation (ASCT) in mantle cell lymphoma (MCL) improved progression-free survival (PFS) vs observation (OBS). The host pharmacogenetic background was analyzed to decipher whether single-nucleotide polymorphisms (SNPs) of genes encoding transmembrane transporters, metabolic enzymes, or cell-surface receptors might predict drug efficacy. Genotypes were obtained via real-time polymerase chain reaction of the peripheral blood germ line DNA. Polymorphisms of ABCB1 and VEGF were found in 69% and 79% of 278 patients, respectively, and predicted favorable PFS vs homozygous wild-type (WT) in the LEN arm was 3-year PFS of 85% vs 70% (P < .05) and 85% vs 60% (P < .01), respectively. Patients carrying both ABCB1 and VEGF WT had the poorest 3-year PFS (46%) and overall survival (76%); in fact, in these patients, LEN did not improve PFS vs OBS (3-year PFS, 44% vs 60%; P = .62). Moreover, the CRBN polymorphism (n = 28) was associated with lenalidomide dose reduction or discontinuation. Finally, ABCB1, NCF4, and GSTP1 polymorphisms predicted lower hematological toxicity during induction, whereas ABCB1 and CRBN polymorphisms predicted lower risk of grade ≥3 infections. This study demonstrates that specific SNPs represent candidate predictive biomarkers of immunochemotherapy toxicity and LEN efficacy after ASCT in MCL.

NR1H3 (LXRα) is associated with pro-inflammatory macrophages, predicts survival and suggests potential therapeutic rationales in diffuse large b-cell lymphoma.

The role of macrophages (Mo) and their prognostic impact in diffuse large B-cell lymphomas (DLBCL) remain controversial. By regulating the lipid metabolism, Liver-X-Receptors (LXRs) control Mo polarization/inflammatory response, and their pharmacological modulation is under clinical investigation to treat human cancers, including lymphomas. Herein, we surveyed the role of LXRs in DLBCL for prognostic purposes. Comparing bulk tumors with purified malignant and normal B-cells, we found an intriguing association of NR1H3, encoding for the LXR-α isoform, with the tumor microenvironment (TME). CIBERSORTx-based purification on large DLBCL datasets revealed a high expression of the receptor transcript in M1-like pro-inflammatory Mo. By determining an expression cut-off of NR1H3, we used digital measurement to validate its prognostic capacity on two large independent on-trial and real-world cohorts. Independently of classical prognosticators, NR1H3high patients displayed longer survival compared with NR1H3low cases and a high-resolution Mo GEP dissection suggested a remarkable transcriptional divergence between subgroups. Overall, our findings indicate NR1H3 as a Mo-related biomarker identifying patients at higher risk and prompt future preclinical studies investigating its mouldability for therapeutic purposes.

Punctual and kinetic MRD analysis from the Fondazione Italiana Linfomi MCL0208 phase 3 trial in mantle cell lymphoma.

Minimal residual disease (MRD) analysis is a known predictive tool in mantle cell lymphoma (MCL). We describe MRD results from the Fondazione Italiana Linfomi phase 3 MCL0208 prospective clinical trial assessing lenalidomide (LEN) maintenance vs observation after autologous stem cell transplantation (ASCT) in the first prospective comprehensive analysis of different techniques, molecular markers, and tissues (peripheral blood [PB] and bone marrow [BM]), taken at well-defined time points. Among the 300 patients enrolled, a molecular marker was identified in 250 (83%), allowing us to analyze 234 patients and 4351 analytical findings from 10 time points. ASCT induced high rates of molecular remission (91% in PB and 83% in BM, by quantitative real-time polymerase chain reaction [RQ-PCR]). Nevertheless, the number of patients with persistent clinical and molecular remission decreased over time in both arms (up to 30% after 36 months). MRD predicted early progression and long-term outcome, particularly from 6 months after ASCT (6-month time to progression [TTP] hazard ratio [HR], 3.83; P < .001). In single-timepoint analysis, BM outperformed PB, and RQ-PCR was more reliable, while nested PCR appeared applicable to a larger number of patients (234 vs 176). To improve MRD performance, we developed a time-varying kinetic model based on regularly updated MRD results and the MIPI (Mantle Cell Lymphoma International Prognostic Index), showing an area under the ROC (Receiver Operating Characteristic) curve (AUROC) of up to 0.87 using BM. Most notably, PB reached an AUROC of up to 0.81; with kinetic analysis, it was comparable to BM in performance. MRD is a powerful predictor over the entire natural history of MCL and is suitable for models with a continuous adaptation of patient risk. The study can be found in EudraCT N. 2009-012807-25 (https://eudract.ema.europa.eu/).

Adenosine pathway inhibitors: novel investigational agents for the treatment of metastatic breast cancer.

INTRODUCTION: The adenosine pathway has been suggested to play a key role in several carcinogenetic processes, with the metabolism of adenosine-5'-triphosphate (ATP) and its derivatives reported to be dysregulated in breast cancer. Preclinical evidence has supported the role of adenosine in the pathogenesis of this malignancy as well as the development of selective adenosine pathway inhibitors. AREAS COVERED: The paper overviews the evidence regarding the use of adenosine pathway inhibitors in breast cancer; a literature search was conducted in January 2022 of Pubmed/Medline, Cochrane library, and Scopus databases. EXPERT OPINION: The adenosine pathway regulates inflammation, apoptosis, metastasis, and cell proliferation in breast cancer cells, and adenosine pathway inhibitors have yielded encouraging results in early-phase clinical trials. Well-designed, multicenter studies focused on monotherapies and combination therapies (which include immune checkpoint inhibitors) are warranted in this setting.

Lean Perspectives in an Organizational Change in a Scientific Direction of an Italian Research Institute: Experience of the Cancer Institute of Bari.

Lean management is a relatively new organizational vision transferred from the automotive industry to the healthcare and administrative sector based on analyzing a production process to emphasize value and reduce waste. This approach is particularly interesting in a historical moment of cuts and scarcity of economic resources and could represent a low-cost organizational solution in many production companies. In this work, we analyzed the presentation and the initial management of current ministerial research projects up to the approval by the Scientific Directorate of an Italian research institute. Furthermore, the initial mode in 2021 ("as is") and the potential mode ("to be") according to a Lean model are studied, according to the current barriers highlighted by the final users of the process and carrying out some perspective analyses with some reference indicators.

Electronic case report forms generation from pathology reports by ARGO, automatic record generator for onco-hematology.

The unstructured nature of Real-World (RW) data from onco-hematological patients and the scarce accessibility to integrated systems restrain the use of RW information for research purposes. Natural Language Processing (NLP) might help in transposing unstructured reports into standardized electronic health records. We exploited NLP to develop an automated tool, named ARGO (Automatic Record Generator for Onco-hematology) to recognize information from pathology reports and populate electronic case report forms (eCRFs) pre-implemented by REDCap. ARGO was applied to hemo-lymphopathology reports of diffuse large B-cell, follicular, and mantle cell lymphomas, and assessed for accuracy (A), precision (P), recall (R) and F1-score (F) on internal (n = 239) and external (n = 93) report series. 326 (98.2%) reports were converted into corresponding eCRFs. Overall, ARGO showed high performance in capturing (1) identification report number (all metrics > 90%), (2) biopsy date (all metrics > 90% in both series), (3) specimen type (86.6% and 91.4% of A, 98.5% and 100.0% of P, 92.5% and 95.5% of F, and 87.2% and 91.4% of R for internal and external series, respectively), (4) diagnosis (100% of P with A, R and F of 90% in both series). We developed and validated a generalizable tool that generates structured eCRFs from real-life pathology reports.

Primary, Bilateral and Diffuse Renal Non-Hodgkin's Lymphoma in a Young Woman Suffering from Turner Syndrome.

Primary renal lymphoma (PRL) is a rare form of non-Hodgkin's lymphoma (NHL) restricted to and primarily involving one or both kidneys, with no lymph node extension. It accounts for <1% of extranodal lymphomas, and descriptions in the literature are limited. Here, we describe an unprecedented case of bilateral PRL in a 44-year-old woman with Turner syndrome and discuss both diagnostic and therapeutic issues in the light of the available literature in the field. A personalized approach to this rare disease is necessary.

Development and Validation of a Simplified Score to Predict Early Relapse in Newly Diagnosed Multiple Myeloma in a Pooled Dataset of 2,190 Patients.

PURPOSE: Despite the improvement of therapeutic regimens, several patients with multiple myeloma (MM) still experience early relapse (ER). This subset of patients currently represents an unmet medical need. EXPERIMENTAL DESIGN: We pooled data from seven European multicenter phase II/III clinical trials enrolling 2,190 patients with newly diagnosed MM from 2003 to 2017. Baseline patient evaluation included 14 clinically relevant features. Patients with complete data (n = 1,218) were split into training (n = 844) and validation sets (n = 374). In the training set, a univariate analysis and a multivariate logistic regression model on ER within 18 months (ER18) were made. The most accurate model was selected on the validation set. We also developed a dynamic version of the score by including response to treatment. RESULTS: The Simplified Early Relapse in Multiple Myeloma (S-ERMM) score was modeled on six features weighted by a score: 5 points for high lactate dehydrogenase or t(4;14); 3 for del17p, abnormal albumin, or bone marrow plasma cells >60%; and 2 for λ free light chain. The S-ERMM identified three patient groups with different risks of ER18: Intermediate (Int) versus Low (OR = 2.39, P < 0.001) and High versus Low (OR = 5.59, P < 0.001). S-ERMM High/Int patients had significantly shorter overall survival (High vs. Low: HR = 3.24, P < 0.001; Int vs. Low: HR = 1.86, P < 0.001) and progression-free survival-2 (High vs. Low: HR = 2.89, P < 0.001; Int vs. Low: HR = 1.76, P < 0.001) than S-ERMM Low. The Dynamic S-ERMM (DS-ERMM) modulated the prognostic power of the S-ERMM. CONCLUSIONS: On the basis of simple, widely available baseline features, the S-ERMM and DS-ERMM properly identified patients with different risks of ER and survival outcomes.

Targeted locus amplification to detect molecular markers in mantle cell and follicular lymphoma.

Minimal residual disease (MRD) monitoring by PCR methods is a strong and standardized predictor of clinical outcome in mantle cell lymphoma (MCL) and follicular lymphoma (FL). However, about 20% of MCL and 40% of FL patients lack a reliable molecular marker, being thus not eligible for MRD studies. Recently, targeted locus amplification (TLA), a next-generation sequencing (NGS) method based on the physical proximity of DNA sequences for target selection, identified novel gene rearrangements in leukemia. The aim of this study was to test TLA in MCL and FL diagnostic samples lacking a classical, PCR-detectable, t(11; 14) MTC (BCL1/IGH), or t(14; 18) major breakpoint region and minor cluster region (BCL2/IGH) rearrangements. Overall, TLA was performed on 20 MCL bone marrow (BM) or peripheral blood (PB) primary samples and on 20 FL BM, identifying a novel BCL1 or BCL2/IGH breakpoint in 16 MCL and 8 FL patients (80% and 40%, respectively). These new breakpoints (named BCL1-TLA and BCL2-TLA) were validated by ASO primers design and compared as MRD markers to classical IGH rearrangements in eight MCL: overall, MRD results by BCL1-TLA were superimposable (R Pearson = 0.76) to the standardized IGH-based approach. Moreover, MRD by BCL2-TLA reached good sensitivity levels also in FL and was predictive of a primary refractory case. In conclusion, this study offers the proof of principle that TLA is a promising and reliable NGS-based technology for the identification of novel molecular markers, suitable for further MRD analysis in previously not traceable MCL and FL patients.

6q deletion in Waldenström macroglobulinaemia negatively affects time to transformation and survival.

Deletion of the long arm of chromosome 6 (del6q) is the most frequent cytogenetic abnormality in Waldenström macroglobulinaemia (WM), occurring in approximately 50% of patients. Its effect on patient outcome has not been completely established. We used fluorescence in situ hybridisation to analyse the prevalence of del6q in selected CD19+ bone marrow cells of 225 patients with newly diagnosed immunoglobulin M (IgM) monoclonal gammopathies. Del6q was identified in one of 27 (4%) cases of IgM-monoclonal gammopathy of undetermined significance, nine of 105 (9%) of asymptomatic WM (aWM), and 28/93 (30%) of symptomatic WM (sWM), and was associated with adverse prognostic features and higher International Prognostic Scoring System for WM (IPSSWM) score. Asymptomatic patients with del6q ultimately required therapy more often and had a shorter time to transformation (TT) to symptomatic disease (median TT, 30 months vs. 199 months, respectively, P < 0·001). When treatment was required, 6q-deleted patients had shorter progression-free survival (median 20 vs. 47 months, P < 0·001). The presence of del6q translated into shorter overall survival (OS), irrespective of the initial diagnosis, with a median OS of 90 compared with 131 months in non-del6q patients (P = 0·01). In summary, our study shows that del6q in IgM gammopathy is associated with symptomatic disease, need for treatment and poorer clinical outcomes.

A Clinical Prognostic Model Based on Machine Learning from the Fondazione Italiana Linfomi (FIL) MCL0208 Phase III Trial.

BACKGROUND: Multicenter clinical trials are producing growing amounts of clinical data. Machine Learning (ML) might facilitate the discovery of novel tools for prognostication and disease-stratification. Taking advantage of a systematic collection of multiple variables, we developed a model derived from data collected on 300 patients with mantle cell lymphoma (MCL) from the Fondazione Italiana Linfomi-MCL0208 phase III trial (NCT02354313). METHODS: We developed a score with a clustering algorithm applied to clinical variables. The candidate score was correlated to overall survival (OS) and validated in two independent data series from the European MCL Network (NCT00209222, NCT00209209); Results: Three groups of patients were significantly discriminated: Low, Intermediate (Int), and High risk (High). Seven discriminants were identified by a feature reduction approach: albumin, Ki-67, lactate dehydrogenase, lymphocytes, platelets, bone marrow infiltration, and B-symptoms. Accordingly, patients in the Int and High groups had shorter OS rates than those in the Low and Int groups, respectively (Int→Low, HR: 3.1, 95% CI: 1.0-9.6; High→Int, HR: 2.3, 95% CI: 1.5-4.7). Based on the 7 markers, we defined the engineered MCL international prognostic index (eMIPI), which was validated and confirmed in two independent cohorts; Conclusions: We developed and validated a ML-based prognostic model for MCL. Even when currently limited to baseline predictors, our approach has high scalability potential.

Early Relapse Risk in Patients with Newly Diagnosed Multiple Myeloma Characterized by Next-generation Sequencing.

PURPOSE: Duration of first remission is important for the survival of patients with multiple myeloma. EXPERIMENTAL DESIGN: From the CoMMpass study (NCT01454297), 926 patients with newly diagnosed multiple myeloma, characterized by next-generation sequencing, were analyzed to evaluate those who experienced early progressive disease (PD; time to progression, TTP ≤18 months). RESULTS: After a median follow-up of 39 months, early PD was detected in 191/926 (20.6%) patients, 228/926 (24.6%) patients had late PD (TTP >18 months), while 507/926 (54.8%) did not have PD at the current follow-up. Compared with patients with late PD, patients with early PD had a lower at least very good partial response rate (47% vs. 82%, P < 0.001) and more frequently acquired double refractoriness to immunomodulatory drugs (IMiD) and proteasome inhibitors (PI; 21% vs. 8%, P < 0.001). Patients with early PD were at higher risk of death compared with patients with late PD and no PD (HR, 3.65; 95% CI, 2.7-4.93; P < 0.001), showing a dismal median overall survival (32.8 months). In a multivariate logistic regression model, independent factors increasing the early PD risk were TP53 mutation (OR, 3.78, P < 0.001), high lactate dehydrogenase levels (OR, 3.15, P = 0.006), λ-chain translocation (OR, 2.25, P = 0.033), and IGLL5 mutation (OR, 2.15, P = 0.007). Carfilzomib-based induction (OR, 0.15, P = 0.014), autologous stem-cell transplantation (OR, 0.27, P < 0.001), and continuous therapy with PIs and IMiDs (OR, 0.34, P = 0.024) mitigated the risk of early PD. CONCLUSIONS: Early PD identifies a high-risk multiple myeloma population. Further research is needed to better identify baseline features predicting early PD and the optimal treatment approaches for patients at risk.

Applying Data Warehousing to a Phase III Clinical Trial From the Fondazione Italiana Linfomi Ensures Superior Data Quality and Improved Assessment of Clinical Outcomes.

PURPOSE: Data collection in clinical trials is becoming complex, with a huge number of variables that need to be recorded, verified, and analyzed to effectively measure clinical outcomes. In this study, we used data warehouse (DW) concepts to achieve this goal. A DW was developed to accommodate data from a large clinical trial, including all the characteristics collected. We present the results related to baseline variables with the following objectives: developing a data quality (DQ) control strategy and improving outcome analysis according to the clinical trial primary end points. METHODS: Data were retrieved from the electronic case reporting forms (eCRFs) of the phase III, multicenter MCL0208 trial (ClinicalTrials.gov identifier: NCT02354313) of the Fondazione Italiana Linfomi for younger patients with untreated mantle cell lymphoma (MCL). The DW was created with a relational database management system. Recommended DQ dimensions were observed to monitor the activity of each site to handle DQ management during patient follow-up. The DQ management was applied to clinically relevant parameters that predicted progression-free survival to assess its impact. RESULTS: The DW encompassed 16 tables, which included 226 variables for 300 patients and 199,500 items of data. The tool allowed cross-comparison analysis and detected some incongruities in eCRFs, prompting queries to clinical centers. This had an impact on clinical end points, as the DQ control strategy was able to improve the prognostic stratification according to single parameters, such as tumor infiltration by flow cytometry, and even using established prognosticators, such as the MCL International Prognostic Index. CONCLUSION: The DW is a powerful tool to organize results from large phase III clinical trials and to effectively improve DQ through the application of effective engineered tools.

Highly sensitive MYD88L265P mutation detection by droplet digital polymerase chain reaction in Waldenström macroglobulinemia.

We here describe a novel method for MYD88L265P mutation detection and minimal residual disease monitoring in Waldenström macroglobulinemia, by droplet digital polymerase chain reaction, in bone marrow and peripheral blood cells, as well as in circulating cell-free DNA. Our method shows a sensitivity of 5.00×10-5, which is far superior to the widely used allele-specific polymerase chain reaction (1.00×10-3). Overall, 291 unsorted samples from 148 patients (133 with Waldenström macroglobulinemia, 11 with IgG lymphoplasmacytic lymphoma and 4 with IgM monoclonal gammopathy of undetermined significance) were analyzed: 194 were baseline samples and 97 were followup samples. One hundred and twenty-two of 128 (95.3%) bone marrow and 47/66 (71.2%) baseline peripheral blood samples scored positive for MYD88L265P To investigate whether MYD88L265P detection by droplet digital polymerase chain reaction could be used for minimal residual disease monitoring, mutation levels were compared with IGH-based minimal residual disease analysis in 10 patients, and was found to be as informative as the classical, standardized, but not yet validated in Waldenström macroglobulinemia, IGH-based minimal residual disease assay (r2=0.64). Finally, MYD88L265P detection by droplet digital polymerase chain reaction on plasma circulating tumor DNA from 60 patients showed a good correlation with bone marrow findings (bone marrow median mutational value 1.92×10-2, plasma circulating tumor DNA value: 1.4×10-2, peripheral blood value: 1.03×10-3). This study indicates that droplet digital polymerase chain reaction assay of MYD88L265P is a feasible and sensitive tool for mutation screening and minimal residual disease monitoring in Waldenström macroglobulinemia. Both unsorted bone marrow and peripheral blood samples can be reliably tested, as can circulating tumor DNA, which represents an attractive, less invasive alternative to bone marrow for MYD88L265P detection.

Data quality improvement of a multicenter clinical trial dataset.

Medical datasets are usually affected by several problems, such as missing values, inconsistencies, redundancies, that can influence the data mining process and the extraction of useful knowledge. For these reasons, a preprocessing phase should be performed for improving the overall quality of data and, consequently, of the information that may be discovered from them. In this study we applied five steps of data preprocessing to improve the quality of a large dataset derived from a multicenter clinical trial. Our dataset included 298 patients enrolled in a prospective, multicenter, clinical trial, characterized by 22 input variables and one class variable (MIPI value). In particular, data coming from different medical centers were firstly integrated to obtain a homogeneous dataset. The latter was normalized to scale all variables into smaller and similar intervals. Then, all missing values were estimated by means of an imputation step. The complete dataset was finally discretized and reduced to remove redundant variables and decrease the amount of data to be managed. The improvement of data quality after each step was evaluated by means of the patients' classification accuracy using the KNN classifier. Our results showed that the proposed pipeline produced an increment of more than 20% of the classification performances. Moreover, the highest growth of accuracy was obtained after missing value imputation, whereas the discretization and feature selection steps allowed for a significant reduction of variables to be managed, without any deterioration of the information contained in data.