Machine Learning Approaches to Determine Feature Importance for Predicting Infant Autopsy Outcome.

INTRODUCTION: Sudden unexpected death in infancy (SUDI) represents the commonest presentation of postneonatal death. We explored whether machine learning could be used to derive data driven insights for prediction of infant autopsy outcome. METHODS: A paediatric autopsy database containing >7,000 cases, with >300 variables, was analysed by examination stage and autopsy outcome classified as 'explained (medical cause of death identified)' or 'unexplained'. Decision tree, random forest, and gradient boosting models were iteratively trained and evaluated. RESULTS: Data from 3,100 infant and young child (<2 years) autopsies were included. Naïve decision tree using external examination data had performance of 68% for predicting an explained death. Core data items were identified using model feature importance. The most effective model was XG Boost, with overall predictive performance of 80%, demonstrating age at death, and cardiovascular and respiratory histological findings as the most important variables associated with determining medical cause of death. CONCLUSION: This study demonstrates feasibility of using machine-learning to evaluate component importance of complex medical procedures (paediatric autopsy) and highlights value of collecting routine clinical data according to defined standards. This approach can be applied to a range of clinical and operational healthcare scenarios.

Use of Whole-Genome Sequencing of Adenovirus in Immunocompromised Pediatric Patients to Identify Nosocomial Transmission and Mixed-Genotype Infection.

Background: Adenoviruses are significant pathogens for the immunocompromised, arising from primary infection or reinfection. Serotyping is insufficient to support nosocomial transmission investigations. We investigate whether whole-genome sequencing (WGS) provides clinically relevant information on transmission among patients in a pediatric tertiary hospital. Methods: We developed a target-enriched adenovirus WGS technique for clinical samples and retrospectively sequenced 107 adenovirus-positive residual diagnostic samples, including viremias (>5 × 104 copies/mL), from 37 patients collected January 2011-March 2016. Whole-genome sequencing was used to determine genotype and for phylogenetic analysis. Results: Adenovirus sequences were recovered from 105 of 107 samples. Full genome sequences were recovered from all 20 nonspecies C samples and from 36 of 85 species C viruses, with partial genome sequences recovered from the rest. Whole-genome phylogenetic analysis suggested linkage of 3 genotype A31 cases and uncovered an unsuspected epidemiological link to an A31 infection first detected on the same ward 4 years earlier. In 9 samples from 1 patient who died, we identified a mixed genotype adenovirus infection. Conclusions: Adenovirus WGS from clinical samples is possible and useful for genotyping and molecular epidemiology. Whole-genome sequencing identified likely nosocomial transmission with greater resolution than conventional genotyping and distinguished between adenovirus disease due to single or multiple genotypes.

Metabolite and thymocyte development defects in ADA-SCID mice receiving enzyme replacement therapy.

Deficiency of adenosine deaminase (ADA, EC3.5.4.4), a housekeeping enzyme intrinsic to the purine salvage pathway, leads to severe combined immunodeficiency (SCID) both in humans and mice. Lack of ADA results in the intracellular accumulation of toxic metabolites which have effects on T cell development and function. While untreated ADA-SCID is a fatal disorder, there are different therapeutic options available to restore ADA activity and reconstitute a functioning immune system, including enzyme replacement therapy (ERT). Administration of ERT in the form of pegylated bovine ADA (PEG-ADA) has proved a life-saving though non-curative treatment for ADA-SCID patients. However, in many patients treated with PEG-ADA, there is suboptimal immune recovery with low T and B cell numbers. Here, we show reduced thymus cellularity in ADA-SCID mice despite weekly PEG-ADA treatment. This was associated with lack of effective adenosine (Ado) detoxification in the thymus. We also show that thymocyte development in ADA-deficient thymi is arrested at the DN3-to-DN4 stage transition with thymocytes undergoing dATP-induced apoptosis rather than defective TCRß rearrangement or ß-selection. Our studies demonstrate at a detailed level that exogenous once-a-week enzyme replacement does not fully correct intra-thymic metabolic or immunological abnormalities associated with ADA deficiency.

Plasticity of the Immune System in Children Following Treatment Interruption in HIV-1 Infection.

It is intriguing that, unlike adults with HIV-1, children with HIV-1 reach a greater CD4+ T cell recovery following planned treatment cessation. The reasons for the better outcomes in children remain unknown but may be related to increased thymic output and diversity of T cell receptor repertoires. HIV-1 infected children from the PENTA 11 trial tolerated planned treatment interruption without adverse long-term clinical, virological, or immunological consequences, once antiretroviral therapy was re-introduced. This contrasts to treatment interruption trials of HIV-1 infected adults, who had rapid changes in T cells and slow recovery when antiretroviral therapy was restarted. How children can develop such effective immune responses to planned treatment interruption may be critical for future studies. PENTA 11 was a randomized, phase II trial of planned treatment interruptions in HIV-1-infected children (ISRCTN 36694210). In this sub-study, eight patients in long-term follow-up were chosen with CD4+ count>500/ml, viral load <50c/ml at baseline: four patients on treatment interruption and four on continuous treatment. Together with measurements of thymic output, we used high-throughput next generation sequencing and bioinformatics to systematically organize memory CD8+ and naïve CD4+ T cell receptors according to diversity, clonal expansions, sequence sharing, antigen specificity, and T cell receptor similarities following treatment interruption compared to continuous treatment. We observed an increase in thymic output following treatment interruption compared to continuous treatment. This was accompanied by an increase in T cell receptor clonal expansions, increased T cell receptor sharing, and higher sequence similarities between patients, suggesting a more focused T cell receptor repertoire. The low numbers of patients included is a limitation and the data should be interpreted with caution. Nonetheless, the high levels of thymic output and the high diversity of the T cell receptor repertoire in children may be sufficient to reconstitute the T cell immune repertoire and reverse the impact of interruption of antiretroviral therapy. Importantly, the effective T cell receptor repertoires following treatment interruption may inform novel therapeutic strategies in children infected with HIV-1.

Using Whole Genome Sequences to Investigate Adenovirus Outbreaks in a Hematopoietic Stem Cell Transplant Unit.

A recent surge in human mastadenovirus (HAdV) cases, including five deaths, amongst a haematopoietic stem cell transplant population led us to use whole genome sequencing (WGS) to investigate. We compared sequences from 37 patients collected over a 20-month period with sequences from GenBank and our own database of HAdVs. Maximum likelihood trees and pairwise differences were used to evaluate genotypic relationships, paired with the epidemiological data from routine infection prevention and control (IPC) records and hospital activity data. During this time period, two formal outbreaks had been declared by IPC, while WGS detected nine monophyletic clusters, seven were corroborated by epidemiological evidence and by comparison of single-nucleotide polymorphisms. One of the formal outbreaks was confirmed, and the other was not. Of the five HAdV-associated deaths, three were unlinked and the remaining two considered the source of transmission. Mixed infection was frequent (10%), providing a sentinel source of recombination and superinfection. Immunosuppressed patients harboring a high rate of HAdV positivity require comprehensive surveillance. As a consequence of these findings, HAdV WGS is being incorporated routinely into clinical practice to influence IPC policy contemporaneously.

Fourth ventricle tumors in children: complications and influence of surgical approach.

OBJECTIVES: The goal of this study was to characterize the complications and morbidity related to the surgical management of pediatric fourth ventricle tumors. METHODS: All patients referred to the authors' institution with posterior fossa tumors from 2002 to 2018 inclusive were screened to include only true fourth ventricle tumors. Preoperative imaging and clinical notes were reviewed to extract data on presenting symptoms; surgical episodes, techniques, and adjuncts; tumor histology; and postoperative complications. RESULTS: Three hundred fifty-four children with posterior fossa tumors were treated during the study period; of these, 185 tumors were in the fourth ventricle, and 167 fourth ventricle tumors with full data sets were included in this analysis. One hundred patients were male (mean age ± SD, 5.98 ± 4.12 years). The most common presenting symptom was vomiting (63.5%). The most common tumor types, in order, were medulloblastoma (94 cases) > pilocytic astrocytoma (30 cases) > ependymoma (30 cases) > choroid plexus neoplasms (5 cases) > atypical teratoid/rhabdoid tumor (4 cases), with 4 miscellaneous lesions. Of the 67.1% of patients who presented with hydrocephalus, 45.5% had an external ventricular drain inserted (66.7% of these prior to tumor surgery, 56.9% frontal); these patients were more likely to undergo ventriculoperitoneal shunt (VPS) placement at a later date (p = 0.00673). Twenty-two had an endoscopic third ventriculostomy, of whom 8 later underwent VPS placement. Overall, 19.7% of patients had a VPS sited during treatment.Across the whole series, the transvermian approach was more frequent than the telovelar approach (64.1% vs 33.0%); however, the telovelar approach was significantly more common in the latter half of the series (p < 0.001). Gross-total resection was achieved in 70.7%. The most common postoperative deficit was cerebellar mutism syndrome (CMS; 28.7%), followed by new weakness (24.0%), cranial neuropathy (18.0%), and new gait abnormality/ataxia (12.6%). Use of intraoperative ultrasonography significantly reduced the incidence of CMS (p = 0.0365). There was no significant difference in the rate of CMS between telovelar or transvermian approaches (p = 0.745), and multivariate logistic regression modeling did not reveal any statistically significant relationships between CMS and surgical approach. CONCLUSIONS: Surgical management of pediatric fourth ventricle tumors continues to evolve, and resection is increasingly performed through the telovelar route. CMS is enduringly the major postoperative complication in this patient population.

Clinical T Cell Receptor Repertoire Deep Sequencing and Analysis: An Application to Monitor Immune Reconstitution Following Cord Blood Transplantation.

Spectratyping assays are well recognized as the clinical gold standard for assessing the T cell receptor (TCR) repertoire in haematopoietic stem cell transplant (HSCT) recipients. These assays use length distributions of the hyper variable complementarity-determining region 3 (CDR3) to characterize a patient's T cell immune reconstitution post-transplant. However, whilst useful, TCR spectratyping is notably limited by its resolution, with the technique unable to provide data on the individual clonotypes present in a sample. High-resolution clonotype data are necessary to provide quantitative clinical TCR assessments and to better understand clonotype dynamics during clinically relevant events such as viral infections or GvHD. In this study we developed and applied a CDR3 Next Generation Sequencing (NGS) methodology to assess the TCR repertoire in cord blood transplant (CBT) recipients. Using this, we obtained comprehensive TCR data from 16 CBT patients and 5 control cord samples at Great Ormond Street Hospital (GOSH). These were analyzed to provide a quantitative measurement of the TCR repertoire and its constituents in patients post-CBT. We were able to both recreate and quantify inferences typically drawn from spectratyping data. Additionally, we demonstrate that an NGS approach to TCR assessment can provide novel insights into the recovery of the immune system in these patients. We show that NGS can be used to accurately quantify TCR repertoire diversity and to provide valuable inference on clonotypes detected in a sample. We serially assessed the progress of T cell immune reconstitution demonstrating that there is dramatic variation in TCR diversity immediately following transplantation and that the dynamics of T cell immune reconstitution is perturbed by the presence of GvHD. These findings provide a proof of concept for the adoption of NGS TCR sequencing in clinical practice.

Detection of Low Frequency Multi-Drug Resistance and Novel Putative Maribavir Resistance in Immunocompromised Pediatric Patients with Cytomegalovirus.

Human cytomegalovirus (HCMV) is a significant pathogen in immunocompromised individuals, with the potential to cause fatal pneumonitis and colitis, as well as increasing the risk of organ rejection in transplant patients. With the advent of new anti-HCMV drugs there is therefore considerable interest in using virus sequence data to monitor emerging resistance to antiviral drugs in HCMV viraemia and disease, including the identification of putative new mutations. We used target-enrichment to deep sequence HCMV DNA from 11 immunosuppressed pediatric patients receiving single or combination anti-HCMV treatment, serially sampled over 1-27 weeks. Changes in consensus sequence and resistance mutations were analyzed for three ORFs targeted by anti-HCMV drugs and the frequencies of drug resistance mutations monitored. Targeted-enriched sequencing of clinical material detected mutations occurring at frequencies of 2%. Seven patients showed no evidence of drug resistance mutations. Four patients developed drug resistance mutations a mean of 16 weeks after starting treatment. In two patients, multiple resistance mutations accumulated at frequencies of 20% or less, including putative maribavir and ganciclovir resistance mutations P522Q (UL54) and C480F (UL97). In one patient, resistance was detected 14 days earlier than by PCR. Phylogenetic analysis suggested recombination or superinfection in one patient. Deep sequencing of HCMV enriched from clinical samples excluded resistance in 7 of 11 subjects and identified resistance mutations earlier than conventional PCR-based resistance testing in 2 patients. Detection of multiple low level resistance mutations was associated with poor outcome.