Evaluation of a machine-learning model based on laboratory parameters for the prediction of acute leukaemia subtypes: a multicentre model development and validation study in France.

BACKGROUND: Acute leukaemias are life-threatening haematological cancers characterised by the infiltration of transformed immature haematopoietic cells in the blood and bone marrow. Prompt and accurate diagnosis of the three main acute leukaemia subtypes (ie acute lymphocytic leukaemia [ALL], acute myeloid leukaemia [AML], and acute promyelocytic leukaemia [APL]) is of utmost importance to guide initial treatment and prevent early mortality but requires cytological expertise that is not always available. We aimed to benchmark different machine-learning strategies using a custom variable selection algorithm to propose an extreme gradient boosting model to predict leukaemia subtypes on the basis of routine laboratory parameters. METHODS: This multicentre model development and validation study was conducted with data from six independent French university hospital databases. Patients aged 18 years or older diagnosed with AML, APL, or ALL in any one of these six hospital databases between March 1, 2012, and Dec 31, 2021, were recruited. 22 routine parameters were collected at the time of initial disease evaluation; variables with more than 25% of missing values in two datasets were not used for model training, leading to the final inclusion of 19 parameters. The performances of the final model were evaluated on internal testing and external validation sets with area under the receiver operating characteristic curves (AUCs), and clinically relevant cutoffs were chosen to guide clinical decision making. The final tool, Artificial Intelligence Prediction of Acute Leukemia (AI-PAL), was developed from this model. FINDINGS: 1410 patients diagnosed with AML, APL, or ALL were included. Data quality control showed few missing values for each cohort, with the exception of uric acid and lactate dehydrogenase for the cohort from Hôpital Cochin. 679 patients from Hôpital Lyon Sud and Centre Hospitalier Universitaire de Clermont-Ferrand were split into the training (n=477) and internal testing (n=202) sets. 731 patients from the four other cohorts were used for external validation. Overall AUCs across all validation cohorts were 0·97 (95% CI 0·95-0·99) for APL, 0·90 (0·83-0·97) for ALL, and 0·89 (0·82-0·95) for AML. Cutoffs were then established on the overall cohort of 1410 patients to guide clinical decisions. Confident cutoffs showed two (0·14%) wrong predictions for ALL, four (0·28%) wrong predictions for APL, and three (0·21%) wrong predictions for AML. Use of the overall cutoff greatly reduced the number of missing predictions; diagnosis was proposed for 1375 (97·5%) of 1410 patients for each category, with only a slight increase in wrong predictions. The final model evaluation across both the internal testing and external validation sets showed accuracy of 99·5% for ALL diagnosis, 98·8% for AML diagnosis, and 99·7% for APL diagnosis in the confident model and accuracy of 87·9% for ALL diagnosis, 86·3% for AML diagnosis, and 96·1% for APL diagnosis in the overall model. INTERPRETATION: AI-PAL allowed for accurate diagnosis of the three main acute leukaemia subtypes. Based on ten simple laboratory parameters, its broad availability could help guide initial therapies in a context where cytological expertise is lacking, such as in low-income countries. FUNDING: None.

Late relapse after hematopoietic stem cell transplantation for acute leukemia: a retrospective study by SFGM-TC.

Late relapse (LR) after allogeneic hematopoietic stem cell transplantation (AHSCT) for acute leukemia is a rare event (nearly 4.5%) and raises the questions of prognosis and outcome after salvage therapy. We performed a retrospective multicentric study between January 1, 2010, and December 31, 2016, using data from the French national retrospective register ProMISe provided by the SFGM-TC (French Society for Bone Marrow Transplantation and Cellular Therapy). We included patients presenting with LR, defined as a relapse occurring at least 2 years after AHSCT. We used the Cox model to identify prognosis factors associated with LR. During the study period, a total of 7582 AHSCTs were performed in 29 centers, and 33.8% of patients relapsed. Among them, 319 (12.4%) were considered to have LR, representing an incidence of 4.2% for the entire cohort. The full dataset was available for 290 patients, including 250 (86.2%) with acute myeloid leukemia and 40 (13.8%) with acute lymphoid leukemia. The median interval from AHSCT to LR was 38.2 months (interquartile range [IQR], 29.2 to 49.7 months), and 27.2% of the patients had extramedullary involvement at LR (17.2% exclusively and 10% associated with medullary involvement). One-third of the patients had persistent full donor chimerism at LR. Median overall survival (OS) after LR was 19.9 months (IQR, 5.6 to 46.4 months). The most common salvage therapy was induction regimen (55.5%), with complete remission (CR) obtained in 50.7% of cases. Ninety-four patients (38.5%) underwent a second AHSCT, with a median OS of 20.4 months (IQR, 7.1 to 49.1 months). Nonrelapse mortality after second AHSCT was 18.2%. The Cox model identified the following factors as associated with delay of LR: disease status not in first CR at first HSCT (odds ratio [OR], 1.31; 95% confidence interval [CI], 1.04 to 1.64; P = .02) and the use of post-transplantation cyclophosphamide (OR, 2.23; 95% CI, 1.21 to 4.14; P = .01). Chronic GVHD appeared to be a protective factor (OR, .64; 95% CI, .42 to .96; P = .04). The prognosis of LR is better than in early relapse, with a median OS after LR of 19.9 months. Salvage therapy associated with a second AHSCT improves outcome and is feasible, without creating excess toxicity.

The Negative Impact of Night Shifts on Diet in Emergency Healthcare Workers.

Despite the consequences of night-shift work, the diet of night-shift workers has not been widely studied. To date, there are no studies related to food intake among emergency healthcare workers (HCWs). We performed a prospective observational study to assess the influence of night work on the diet of emergency HCWs. We monitored 24-h food intake during a day shift and the consecutive night, and during night work and the daytime beforehand. We analyzed 184 emergency HCWs' food intakes. Emergency HCWs had 14.7% lower (-206 kcal) of their 24-h energy intake during night shifts compared to their day-shift colleagues (1606.7 ± 748.2 vs. 1400.4 ± 708.3 kcal, p = 0.049) and a 16.7% decrease in water consumption (1451.4 ± 496.8 vs. 1208.3 ± 513.9 mL/day, p = 0.010). Compared to day shifts, night-shift had 8.7% lower carbohydrates, 17.6% proteins, and 18.7% lipids. During the night shift the proportion of emergency HCWs who did not drink for 4 h, 8 h and 12 h increased by 20.5%, 17.5%, and 9.1%, respectively. For those who did not eat for 4 h, 8 h and 12 h increased by 46.8%, 27.7%, and 17.7%, respectively. A night shift has a huge negative impact on both the amount and quality of nutrients consumed by emergency healthcare workers.

Impact of Air Transport on SpO2/FiO2 among Critical COVID-19 Patients during the First Pandemic Wave in France.

During the first wave of the COVID-19 pandemic, some French regions were more affected than others. To relieve those areas most affected, the French government organized transfers of critical patients, notably by plane or helicopter. Our objective was to investigate the impact of such transfers on the pulse oximetric saturation (SpO2)-to-inspired fraction of oxygen (FiO2) ratio among transferred critical patients with COVID-19. We conducted a retrospective study on medical and paramedical records. The primary endpoint was the change in SpO2/FiO2 during transfers. Thirty-eight patients were transferred between 28 March and 5 April 2020, with a mean age of 62.4 years and a mean body mass index of 29.8 kg/m2. The population was 69.7% male, and the leading medical history was hypertension (42.1%), diabetes (34.2%), and dyslipidemia (18.4%). Of 28 patients with full data, we found a decrease of 28.9 points in SpO2/FiO2 (95% confidence interval, 5.8 to 52.1, p = 0.01) between the starting and the arrival intensive care units (SpO2/FiO2, 187.3 ± 61.3 and 158.4 ± 62.8 mmHg, respectively). Air medical transfers organized to relieve intensive care unit teams under surging conditions during the first COVID wave were associated with significant decreases in arterial oxygenation.

Tranexamic acid in non-traumatic intracranial bleeding: a systematic review and meta-analysis.

Non-traumatic intracranial bleeding (NTIB), comprising subarachnoid hemorrhage (SAH) and intra-cranial bleeding (ICH) is a significant public health concern. Tranexamic acid (TXA) is a promising treatment with benefits yet to be fully demonstrated. We conducted a systematic review and meta-analysis on the impact of TXA on mortality in NTIB. We searched the PubMed, Cochrane Library, Google Scholar and ScienceDirect databases for studies reporting mortality data following the use of TXA in NTIB for comparisons with a control group. We computed random-effect meta-analysis on estimates of risk and sensitivity analyses. We computed meta-regression to examine the putative effects of the severity of NTIB, sociodemographic data (age, sex), and publication date. Among potentially 10,008 articles, we included 15 studies representing a total of 4883 patients: 2455 receiving TXA and 2428 controls; 1110 died (23%) during the follow-up. The meta-analysis demonstrated a potential of 22% decrease in mortality for patients treated by TXA (RR = 0.78, 95%CI 0.58-0.98, p = 0.002). Meta-regression did not demonstrate any influence of the severity of NTIB, age, sex, length of treatment or date of publication. Sensitivity analyses confirmed benefits of TXA on mortality. TXA appears to be a therapeutic option to reduce non-traumatic intracranial bleeding mortality, particularly in patients with SAH.

[Richter Syndrome: Diagnostic and Therapeutic Management]. / Syndrome de Richter : prise en charge diagnostique et thérapeutique.

Richter syndrome (RS) is defined as the occurrence of an aggressive lymphoma, most commonly diffuse large B-cell lymphoma (DLBCL) and rarely Hodgkin lymphoma (HL), in a patient with prior or concomitant chronic lymphocytic leukemia (CLL). RS is estimated to occur in 0.5-1 % per year and is associated with adverse outcome. In the vast majority of patients (80 %), RS is clonally related to the prior CLL. Those with unrelated RS appear to have better outcome. The therapeutic approach is based on those of de novo DLBCL or HL. However, even with modern immunochemotherapy regimens, response rate remains low. In eligible patients with related RS, a consolidation by autologous or allogeneic stem-cell transplantation must be proposed. Combinations including therapies targeting BCR or BCL2 and effective in CLL are currently being evaluated in RS. Novels immunotherapies could be promising approaches based on preliminary results.

First-line treatment of double-hit and triple-hit lymphomas: Survival and tolerance data from a retrospective multicenter French study.

Historically, double or triple hit lymphoma (DHL and THL) have poor outcomes with conventional chemotherapy, but there is currently no guideline. We report the French experience in managing DHL and THL in first line using collective data on both survival and tolerance. All consecutive patients with newly diagnosis of large B-cell lymphoma with MYC, BCL2, and/or BCL6 rearrangements, as determined by FISH between January 2013 and April 2019 were included. Based on the eligibility criteria, 160 patients were selected among the 184 patients identified. With a median follow-up of 32 months, 2- and 4-year progression free survival (PFS) rates were 40% and 28% with R-CHOP compared with 57% and 52% with intensive chemotherapy (P = .063). There was no difference in overall survival (OS). For advanced stages, PFS was significantly longer with intensive chemotherapy than with R-CHOP (P = .029). There was no impact of autologous stem cell transplantation among patient in remission. For patients with central nervous system (CNS) involvement, the 2-year PFS and OS rate was 21% and 39%, vs 57% and 75% without CNS disease (P = .007 and P < .001). By multivariate analysis, elevated IPI score and CNS disease were strongly and independently associated with a poorer survival, whereas treatment was not significantly associated with OS. This is the largest series reporting the treatment of DHL and THL in Europe. The PFS was significantly longer with an intensive regimen for advanced stage, but no difference in OS, supporting the need for a prospective randomized trial.

[Daratumumab for multiple myeloma]. / Utilisation du daratumumab dans le myélome multiple.

Daratumumab is a monoclonal antibody that targets CD38. It has an anti-tumor action on the myeloma cell and an immunomodulatory action. For relapsing and/or refractory myeloma patients, daratumumab is effective both alone and in combination and significantly improves progression-free survival. Daratumumab has been approved in 2016 in France for treatment of relapses or refractory multiple myeloma.

[The CAR-T cells are here]. / Les CAR-T cells sont là !

Due to immunotherapy, a new era of treatment is opening up for the treatment of solid cancers and hematological tumors. T cells genetically modified with a chimeric antigen receptor (CAR-T cells) have recently been proved efficient in hematological malignancies expressing CD19. On June 20th 2018, the European Medicines Agency gave a favorable opinion on two types of anti-CD19 CAR-T cells - tisagenlecleucel and axicabtagène ciloleucel - in the management of malignant hemopathies B after two lines of treatment. Their remarkable efficacy has been demonstrated in the first clinical trials conducted in the United States, but they present new and potentially serious side effects (cytokine release syndrome, neurotoxicity, among others). However, this new class of gene therapy raises practical problems regarding its production circuit, its delivery conditions and its high cost.