Novel diagnostic and prognostic tools for patients with chronic idiopathic neutropenia: Data on a cohort of 266 patients

Abstract Background: Chronic Idiopathic Neutropenia (CIN) is a neutrophil disorder characterized by the persistent and unexplained reduction in the number of peripheral blood (PB) neutrophils. It is an exclusion diagnosis that can be established after a thorough clinical/laboratory investigation according to the recently published guidelines (doi: 10.1097/HS9.0000000000000872). Aims: The aim of the study is to assess the definitive diagnosis and natural history of neutropenia in a group of patients initially characterized as CIN, following the application of molecular, genetic and machine learning (ML) tools. Methods: Clinical and laboratory tests for the diagnostic work-up include cell blood counts (CBC), serum biochemistry, immunoglobulin levels, antibodies to viral and bacterial pathogens, immunophenotype of PB and bone marrow (BM), BM morphology, trephine biopsy, karyotype, data from anti-neutrophil antibody tests following granulocyte agglutination (GAT) and immunofluorescence (GIFT), real-time PCR for the detection of DARC/ACKR1 polymorphisms, data from myeloid panel of Next Generation Sequencing (NGS) for clonal hematopoiesis (CH) and Whole Exome Sequencing (WES) for unexplained cases. JADBio's auto-ML platform was employed to test classification models and extract predictive biomarkers with model performances being reported as area under the receiver operating characteristic (AUROC). Results: The neutropenia cohort consists of 266 well characterized adult neutropenia patients (212 females, 54 males, aged 19-92 years (median 61 years)). Mean absolute neutrophil counts -ANCs- were 1386±444/μL; median 1530, range 100-1700/μL. Median follow-up time was 84 months (range 12-360 months). After completion of the diagnostic work-up, the patients were characterized as CIN in 66.5% (n=177) , as clonal cytopenia with undetermined significance (CCUS) in 12.8% (n=34), as primary autoimmune (pAIN) in 6.0% (n=16), 6.8% as familial with undefined genetic defect (n=18), 1.1% as ADAN (previously known as ethnic neutropenia) (n=3), 4.5% as secondary neutropenia (n=12), and 2.2% as severe congenital neutropenia (n=6). The congenital neutropenia cases included a male with cyclic neutropenia with an ELANE mutation, a male and one female with a double G6PC3 mutation, one male with GFI1 mutation and two females with a homozygous and a compound heterozygous CXCR2 mutation, respectively. Moreover, two female patients with homozygous and heterozygous pathogenic variants in ADA2, respectively, were also identified. The frequency of transformation to a hematological malignancy was low in the total group of patients (6/266 CIN patients i.e. 2.25%). However, of the transformed patients, 5 belonged to the clonal (5/34 clonal patients, i.e. 14.7%) and one to the non-clonal (1/232 non-clonal patients, i.e. 0.43%) group (P<0.0001). The spectrum of hematological neoplasms included chronic myelomonocytic leukemia (CMML) (n=2), unclassifiable MDS/myeloproliferative neoplasm (MDS/MPN) (n=1), acute (myelomonocytic) myeloid leukemia (AML) (n=1), MDS with multilineage dysplasia (n=1; non-clonal patient) and T-acute lymphoblastic leukemia (T-ALL) (n=1). Binary classification analysis showed that the CCUS subtype could be efficiently predicted using all available features (AUROC=0.80). Feature selection identified 1 signature comprising 7 out of 84 features in the original dataset (AUROC=0.75). More specifically, lower levels of IgG4, higher age, presence of myoskeletal pains, and cardiac disease were associated with higher probability of a patient being classified as CCUS.Conclusions: The follow-up and close investigation of such a rare and heterogenous disorder with a wide spectrum of natural history is important. Notably, this is the first time that a large number of patients who were initially classified as unexplained have been investigated. Interestingly, a large percentage of the patients were characterized as congenital after the genetic testing, indicating that a constitutional/congenital background exists in some CIN cases. Moreover, we detected the presence of 3 ADAN cases in our adult CIN cohort which was of Caucasian ethnicity. The transformation rate to malignancy was overall low, however CIN patients with CH have increased propensity to develop overt MDS/AML. Preliminary data from the application of ML in the diagnostic work-up of neutropenia provide valuable insights into the potential of ML to aid in the classification of neutropenia subtypes.