Who's your data? Primary immune deficiency differential diagnosis prediction via machine learning and data mining of the USIDNET registry.

PURPOSE: There are currently more than 480 primary immune deficiency (PID) diseases and about 7000 rare diseases that together afflict around 1 in every 17 humans. Computational aids based on data mining and machine learning might facilitate the diagnostic task by extracting rules from large datasets and making predictions when faced with new problem cases. In a proof-of-concept data mining study, we aimed to predict PID diagnoses using a supervised machine learning algorithm based on classification tree boosting. METHODS: Through a data query at the USIDNET registry we obtained a database of 2396 patients with common diagnoses of PID, including their clinical and laboratory features. We kept 286 features and all 12 diagnoses to include in the model. We used the XGBoost package with parallel tree boosting for the supervised classification model, and SHAP for variable importance interpretation, on Python v3.7. The patient database was split into training and testing subsets, and after boosting through gradient descent, the predictive model provides measures of diagnostic prediction accuracy and individual feature importance. After a baseline performance test, we used the Class Weighting Hyperparameter, or scale_pos_weight to correct for imbalanced classification. RESULTS: The twelve PID diagnoses were CVID (1098 patients), DiGeorge syndrome, Chronic granulomatous disease, Congenital agammaglobulinemia, PID not otherwise classified, Specific antibody deficiency, Complement deficiency, Hyper-IgM, Leukocyte adhesion deficiency, ectodermal dysplasia with immune deficiency, Severe combined immune deficiency, and Wiskott-Aldrich syndrome. For CVID, the model found an accuracy on the train sample of 0.80, with an area under the ROC curve (AUC) of 0.80, and a Gini coefficient of 0.60. In the test subset, accuracy was 0.76, AUC 0.75, and Gini 0.51. The positive feature value to predict CVID was highest for upper respiratory infections, asthma, autoimmunity and hypogammaglobulinemia. Features with the highest negative predictive value were high IgE, growth delay, abscess, lymphopenia, and congenital heart disease. For the rest of the diagnoses, accuracy stayed between 0.75 and 0.99, AUC 0.46-0.87, Gini 0.07-0.75, and LogLoss 0.09-8.55. DISCUSSION: Clinicians should remember to consider the negative predictive features together with the positives. We are calling this a proof-of-concept study to continue with our explorations. A good performance is encouraging, and feature importance might aid feature selection for future endeavors. In the meantime, we can learn from the rules derived by the model and build a user-friendly decision tree to generate differential diagnoses.

Cutaneous barrier leakage and gut inflammation drive skin disease in Omenn syndrome.

BACKGROUND: Severe early-onset erythroderma and gut inflammation, with massive tissue infiltration of oligoclonal activated T cells are the hallmark of Omenn syndrome (OS). OBJECTIVE: The impact of altered gut homeostasis in the cutaneous manifestations of OS remains to be clarified. METHODS: We analyzed a cohort of 15 patients with OS and the 129Sv/C57BL/6 knock-in Rag2R229Q/R229Q (Rag2R229Q) mouse model. Homing phenotypes of circulating lymphocytes were analyzed by flow cytometry. Inflammatory cytokines and chemokines were examined in the sera by ELISA and in skin biopsies by immunohistochemistry and in situ RNA hybridization. Experimental colitis was induced in mice by dextran sulfate sodium salt. RESULTS: We show that memory/activated T cells from patients with OS and from the Rag2R229Q mouse model of OS abundantly express the skin homing receptors cutaneous lymphocyte associated antigen and CCR4 (Ccr4), associated with high levels of chemokine C-C motif ligands 17 and 22. Serum levels of LPS are also elevated. A broad Th1/Th2/Th17 inflammatory signature is detected in the periphery and in the skin. Increased Tlr4 expression in the skin of Rag2R229Q mice is associated with enhanced cutaneous inflammation on local and systemic administration of LPS. Likewise, boosting colitis in Rag2R229Q mice results in increased frequency of Ccr4+ splenic T cells and worsening of skin inflammation, as indicated by epidermal thickening, enhanced epithelial cell activation, and dermal infiltration by Th1 effector T cells. CONCLUSIONS: These results support the existence of an interplay between gut and skin that can sustain skin inflammation in OS.

Pulmonary Disease Burden in Primary Immune Deficiency Disorders: Data from USIDNET Registry.

PURPOSE: Pulmonary manifestations are common in patients with primary immunodeficiency disorders (PIDs) but the prevalence, specific diseases, and their patterns are not well characterized. METHODS: We conducted a retrospective analysis of pulmonary diseases reported in the database of the United States Immunodeficiency Network (USIDNET), a program of the Immune Deficiency Foundation. PIDs were categorized into 10 groups and their demographics, pulmonary diagnoses and procedures, infections, prophylaxis regimens, and laboratory findings were analyzed. RESULTS: A total of 1937 patients with various PIDs (39.3% of total patients, 49.6% male, average age 37.9 years (SD = 22.4 years)) were noted to have a pulmonary disease comorbidity. Pulmonary diseases were categorized into broad categories: airway (86.8%), parenchymal (18.5%), pleural (4.6%), vascular (4.3%), and other (13.9%) disorders. Common variable immune deficiency (CVID) accounted for almost half of PIDs associated with airway, parenchymal, and other pulmonary disorders. Pulmonary procedures performed in 392 patients were mostly diagnostic (77.3%) or therapeutic (16.3%). These patients were receiving a wide variety of treatments, which included immunoglobulin replacement (82.1%), immunosuppressive (32.2%), anti-inflammatory (12.7%), biologic (9.3%), and cytokine (7.6%)-based therapies. Prophylactic therapy was being given with antibiotics (18.1%), antifungal (3.3%), and antiviral (2.2%) medications, and 7.1% of patients were on long-term oxygen therapy due to advanced lung disease. CONCLUSIONS: Pulmonary manifestations are common in individuals with PID, but long-term pulmonary outcomes are not well known in this group of patients. Further longitudinal follow-up will help to define long-term prognosis of respiratory comorbidities and optimal treatment modalities.

Evolution of Gene Therapy, Historical Perspective.

The earliest conceptual history of gene therapy began with the recognition of DNA as the transforming substance capable of changing the phenotypic character of a bacterium and then as the carrier of the genomic code. Early studies of oncogenic viruses that could insert into the mammalian genome led to the concept that these same viruses might be engineered to carry new genetic material into mammalian cells, including human hematopoietic stem cells (HSC). In addition to properly engineered vectors capable of efficient safe transduction of HSC, successful gene therapy required the development of efficient materials, methods, and equipment to procure, purify, and culture HSC. Increased understanding of the preparative conditioning of patients was needed to optimize the engraftment of genetically modified HSC. Testing concepts in pivotal clinical trials to assess the efficacy and determine the cause of adverse events has advanced the efficiency and safety of gene therapy. This article is a historical overview of the separate threads of discovery that joined together to comprise our current state of gene therapy targeting HSC.

Eosinophilic gastrointestinal disorders in patients with inborn errors of immunity: Data from the USIDNET registry.

Rationale: Eosinophilic gastrointestinal disorders (EGID), including eosinophilic esophagitis (EoE), are inflammatory disorders of the gastrointestinal mucosa mediated by complex immune mechanisms. Although there have been initial reports of EGID in patients with inborn errors of immunity (IEI), little is known about the presentation of EGID in immunodeficient individuals. Methods: We queried the U.S. Immunodeficiency Network (USIDNET) for patient records including the terms eosinophilic esophagitis, gastritis, enteritis, or colitis. We analyzed 74 patient records from the database, including diagnoses, demographics, infectious history, laboratory findings, genetic studies, therapeutic interventions, and clinical outcomes. Results: We examined 74 patient records. A total of 61 patients had isolated EoE, and 13 had distal gastrointestinal involvement consistent with EGID. The most common IEI were common variable immunodeficiency (43.2%), some form of combined immunodeficiency (21.6%), chronic granulomatous disease (8.1%), hyper-IgE syndrome (6.8%), and autoimmune lymphoproliferative syndrome (6.8%). The median age at presentation with IEI was 0.5 years (IQR 1.725, max 39 years) and 56.76% were male. Approximately 20% of the patients in the cohort received a hematopoietic stem cell transplantation for treatment of IEI, but the timing of the HSCT in relationship to the EGID diagnosis was unknown. Conclusions: Here, we report EGID in a diverse cohort of IEI patients, suggesting that both non-EoE EGID and EoE can be seen as comorbid conditions with a variety of IEI. Our data suggests that EGID may be more common in patients with IEI than would be expected based on estimates of EGID in the general population.

Artificial thymic organoids represent a reliable tool to study T-cell differentiation in patients with severe T-cell lymphopenia.

The study of early T-cell development in humans is challenging because of limited availability of thymic samples and the limitations of in vitro T-cell differentiation assays. We used an artificial thymic organoid (ATO) platform generated by aggregating a DLL4-expressing stromal cell line (MS5-hDLL4) with CD34+ cells isolated from bone marrow or mobilized peripheral blood to study T-cell development from CD34+ cells of patients carrying hematopoietic intrinsic or thymic defects that cause T-cell lymphopenia. We found that AK2 deficiency is associated with decreased cell viability and an early block in T-cell development. We observed a similar defect in a patient carrying a null IL2RG mutation. In contrast, CD34+ cells from a patient carrying a missense IL2RG mutation reached full T-cell maturation, although cell numbers were significantly lower than in controls. CD34+ cells from patients carrying RAG mutations were able to differentiate to CD4+CD8+ cells, but not to CD3+TCRαß+ cells. Finally, normal T-cell differentiation was observed in a patient with complete DiGeorge syndrome, consistent with the extra-hematopoietic nature of the defect. The ATO system may help determine whether T-cell deficiency reflects hematopoietic or thymic intrinsic abnormalities and define the exact stage at which T-cell differentiation is blocked.

Noninfectious Manifestations and Complications of Chronic Granulomatous Disease.

Chronic granulomatous disease (CGD), a primary immunodeficiency characterized by a deficient neutrophil oxidative burst and the inadequate killing of microbes, is well known to cause a significantly increased risk of invasive infection. However, infectious complications are not the sole manifestations of CGD; substantial additional morbidity is driven by noninfectious complications also. These complications can include, for example, a wide range of inflammatory diseases that affect the gastrointestinal tract, lung, skin, and genitourinary tract and overt autoimmune disease. These diseases can occur at any age and are especially problematic in adolescents and adults with CGD. Many of these noninfectious complications present a highly challenging therapeutic conundrum, wherein immunosuppression must be balanced against an already markedly increased risk of invasive fungal and bacterial infections. In this review, the myriad noninfectious complications of CGD are discussed, as are important gaps in our understanding of these processes, which warrant further investigation.

Second-site mutation in the Wiskott-Aldrich syndrome (WAS) protein gene causes somatic mosaicism in two WAS siblings.

Revertant mosaicism due to true back mutations or second-site mutations has been identified in several inherited disorders. The occurrence of revertants is considered rare, and the underlying genetic mechanisms remain mostly unknown. Here we describe somatic mosaicism in two brothers affected with Wiskott-Aldrich syndrome (WAS). The original mutation causing disease in this family is a single base insertion (1305insG) in the WAS protein (WASP) gene, which results in frameshift and abrogates protein expression. Both patients, however, showed expression of WASP in a fraction of their T cells that were demonstrated to carry a second-site mutation causing the deletion of 19 nucleotides from nucleotide 1299 to 1316. This deletion abrogated the effects of the original mutation and restored the WASP reading frame. In vitro expression studies indicated that mutant protein encoded by the second-site mutation was expressed and functional, since it was able to bind to cellular partners and mediate T cell receptor/CD3 downregulation. These observations were consistent with evidence of in vivo selective advantage of WASP-expressing lymphocytes. Molecular analysis revealed that the sequence surrounding the deletion contained two 4-bp direct repeats and that a hairpin structure could be formed by five GC pairs within the deleted fragment. These findings strongly suggest that slipped mispairing was the cause of this second-site mutation and that selective accumulation of WASP-expressing T lymphocytes led to revertant mosaicism in these patients.

Analysis of T-cell repertoire diversity in Wiskott-Aldrich syndrome.

Wiskott-Aldrich syndrome (WAS) is an X-linked immunodeficiency characterized by thrombocytopenia, eczema, and variable degrees of impaired cellular and humoral immunity. Age-dependent T-cell lymphopenia has been described in WAS, however, the diversity of the T-cell compartment over time in these patients has not been characterized. We have used complementarity-determining region 3 (CDR3) size distribution analysis to assess T-cell receptor (TCR) Vbeta repertoire in 13 patients with WAS. Diverse CDR3 size pattern was demonstrated in patients under 15 years of age regardless of the levels of WAS protein (WASP) expression. In contrast, older patients showed significantly higher skewing of TCRVbeta repertoire as compared with healthy adults. We did not find correlation between clinical score and complexity of TCRVbeta repertoire. These findings suggest that WASP deficiency does not limit thymic generation of a normal TCR and indicate that T-cell oligoclonality may contribute to the immunodeficiency in older patients with WAS.

Differential contribution of Wiskott-Aldrich syndrome protein to selective advantage in T- and B-cell lineages.

Somatic mosaicism because of in vivo reversion has been recently reported in a small number of patients affected with Wiskott-Aldrich syndrome (WAS). Flow cytometry analysis of WAS protein (WASP) expression has shown that these patients carried revertant cells only among T lymphocytes. Here, we have used high-resolution capillary electrophoresis to analyze genomic DNA from highly purified cells of one of these patients and detected revertant sequences also within the B-cell fraction. The demonstration of revertant cells among both T and B lymphocytes in this patient is consistent with the reversion event having occurred in a common lymphoid progenitor. However, although WASP-expressing T cells showed selective advantage and were readily detectable in the periphery of the mosaic patient, revertant B lymphocytes remained below the detection threshold of flow cytometry. These findings suggest that, contrary to T cells, differentiation and survival of B lymphocytes is minimally dependent on WASP.

Multiple patients with revertant mosaicism in a single Wiskott-Aldrich syndrome family.

We previously reported on a 43-year-old patient with Wiskott-Aldrich syndrome (WAS) who experienced progressive clinical improvement and revertant T-cell mosaicism. Deletion of the disease-causing 6-bp insertion was hypothesized to have occurred by DNA polymerase slippage. We now describe 2 additional patients from the same family who also had revertant T lymphocytes that showed selective in vivo advantage. Somatic mosaicism was demonstrated on leukocytes cryopreserved in the first patient when he was 22 years old, 11 years before his death from kidney failure. The second patient is now 16 years old, has a moderate clinical phenotype, and developed revertant cells after the age of 14 years. These results support DNA polymerase slippage as a common underlying mechanism, and they indicate that T-cell mosaicism may have different clinical effects in WAS.

Persistence and expression of the adenosine deaminase gene for 12 years and immune reaction to gene transfer components: long-term results of the first clinical gene therapy trial.

The first human gene therapy experiment begun in September 1990 used a retroviral vector containing the human adenosine deaminase (ADA) cDNA to transduce mature peripheral blood lymphocytes from patients with ADA deficiency, an inherited disorder of immunity. Two patients who had been treated with intramuscular injections of pegylated bovine ADA (PEG-ADA) for 2 to 4 years were enrolled in this trial and each received a total of approximately 10(11) cells in 11 or 12 infusions over a period of about 2 years. No adverse events were observed. During and after treatment, the patients continued to receive PEG-ADA, although at a reduced dose. Ten years after the last cell infusion, approximately 20% of the first patient's lymphocytes still carry and express the retroviral gene, indicating that the effects of gene transfer can be remarkably long lasting. On the contrary, the persistence of gene-marked cells is very low (< 0.1%), and no expression of the transgene is detectable in lymphocytes from the second patient who developed persisting antibodies to components of the gene transfer system. Data collected from these original patients have provided novel information about the longevity of T lymphocytes in humans and persistence of gene expression in vivo from vectors driven by the Moloney murine leukemia virus long-terminal repeat (LTR) promoter. This long-term follow-up has also provided unique evidence supporting the safety of retroviral-mediated gene transfer and illustrates clear examples of both the potential and the pitfalls of gene therapy in humans.