Commercially available artificial intelligence tools for fracture detection: the evidence

Missed fractures are a costly healthcare issue, not only negatively impacting patient lives, leading to potential long-term disability and time off work, but also responsible for high medicolegal disbursements that could otherwise be used to improve other healthcare services. When fractures are overlooked in children, they are particularly concerning as opportunities for safeguarding may be missed. Assistance from artificial intelligence (AI) in interpreting medical images may offer a possible solution for improving patient care, and several commercial AI tools are now available for radiology workflow implementation. However, information regarding their development, evidence for performance and validation as well as the intended target population is not always clear, but vital when evaluating a potential AI solution for implementation. In this article, we review the range of available products utilizing AI for fracture detection (in both adults and children) and summarize the evidence, or lack thereof, behind their performance. This will allow others to make better informed decisions when deciding which product to procure for their specific clinical requirements.

Body weight-based iodinated contrast immersion timing for human fetal postmortem microfocus computed tomography

Objectives

The aim of this study was to evaluate the length of time required to achieve full iodination using potassium tri-iodide as a contrast agent, prior to human fetal postmortem microfocus computed tomography (micro-CT) imaging.

Methods

Prospective assessment of optimal contrast iodination was conducted across 157 human fetuses (postmortem weight range 2-298 g; gestational age range 12-37 weeks), following micro-CT imaging. Simple linear regression was conducted to analyse which fetal demographic factors could produce the most accurate estimate for optimal iodination time.

Results

Postmortem body weight (r2 = 0.6435) was better correlated with iodination time than gestational age (r2 = 0.1384), producing a line of best fit, y = [0.0304 × body weight (g)] - 2.2103. This can be simplified for clinical use whereby immersion time (days) = [0.03 × body weight (g)] - 2.2. Using this formula, for example, a 100-g fetus would take 5.2 days to reach optimal contrast enhancement.

Conclusions

The simplified equation can now be used to provide estimation times for fetal contrast preparation time prior to micro-CT imaging and can be used to manage service throughput and parental expectation for return of their fetus.

Advances in knowledge

A simple equation from empirical data can now be used to estimate preparation time for human fetal postmortem micro-CT imaging.

Performance of the PRIMaCY sudden death risk prediction model for childhood hypertrophic cardiomyopathy: implications for implantable cardioverter-defibrillator decision-making

AIMS

The validated HCM Risk-Kids model provides accurate individualized estimates of sudden cardiac death risk in children with hypertrophic cardiomyopathy (HCM). A second validated model, PRIMaCY, also provides individualized estimates of risk, but its performance and clinical impact has not been independently investigated. The aim of this study was to investigate the clinical impact of using the PRIMaCY sudden cardiac death (SCD) risk model in childhood HCM.

METHODS AND RESULTS

The estimated 5-year SCD risk was calculated for children meeting diagnostic criteria for HCM in a large single-centre cohort using PRIMaCY (clinical and genetic) and HCM Risk-Kids model, and model performance was assessed. Three hundred one patients [median age 10 (interquartile range 4-14)] were followed up for an average of 4.9 (±3.8) years, during which 30 (10.0%) reached the SCD or equivalent event endpoint. Harrell's C-statistic for the clinical and genetic models was 0.66 [95% confidence interval (CI) 0.52-0.8] and 0.66 (95% CI 0.54-0.80) with a calibration slope of 0.19 (95% CI 0.04-0.54) and 0.26 (95% CI -0.03-0.62), respectively. The number needed to treat to potentially treat one life-threatening arrhythmia for the PRIMaCY clinical, PRIMaCY genetic, and HCM Risk-Kids models was 13.7, 14.5, and 9.4, respectively.

CONCLUSION

Although PRIMaCY has a similar discriminatory ability to that reported for HCM Risk-Kids, estimated risk estimates did not correlate well with observed risk. A higher proportion of patients met implantable cardioverter-defibrillator thresholds using PRIMaCY model compared with HCM Risk-Kids. This has important clinical implications as these patients will be exposed to a lifetime risk of complications and inappropriate therapies.

Startle responses in Duchenne muscular dystrophy: a novel biomarker of brain dystrophin deficiency

Duchenne muscular dystrophy (DMD) is characterized by loss of dystrophin in muscle, however patients also have variable degree of intellectual disability and neurobehavioural co-morbidities. In contrast to muscle, in which a single full-length dystrophin isoform (Dp427) is produced, multiple isoforms are produced in the brain, and their deficiency accounts for the variability of CNS manifestations, with increased risk of comorbidities in patients carrying mutations affecting the 3' end of the gene, which disrupt expression of shorter Dp140 and Dp71 isoforms. A mouse model (mdx mouse) lacks Dp427 in muscle and CNS and exhibits exaggerated startle responses to threat, linked to the deficiency of dystrophin in limbic structures such as the amygdala, which normalize with postnatal brain dystrophin-restoration therapies. A pathological startle response is not a recognized feature of DMD, and its characterization has implications for improved clinical management and translational research. To investigate startle responses in DMD, we used a novel fear-conditioning task in an observational study of 56 males aged 7-12 years (31 affected boys, mean age 9.7 ± 1.8 years; 25 controls, mean age 9.6 ± 1.4 years). Trials of two neutral visual stimuli were presented to participants: one 'safe' cue presented alone; one 'threat' cue paired with an aversive noise to enable conditioning of physiological startle responses (skin conductance response and heart rate). Retention of conditioned physiological responses was subsequently tested by presenting both cues without the aversive noise in an 'Extinction' phase. Primary outcomes were the initial unconditioned skin conductance and change in heart rate responses to the aversive 'threat' and acquisition and retention of conditioned responses after conditioning. Secondary and exploratory outcomes were neuropsychological measures and genotype associations. The mean unconditioned skin conductance response was greater in the DMD group than controls [mean difference 3.0 µS (1.0, 5.1); P = 0.004], associated with a significant threat-induced bradycardia only in the patient group [mean difference -8.7 bpm (-16.9, -0.51); P = 0.04]. Participants with DMD found the task more aversive than controls, with increased early termination rates during the Extinction phase (26% of DMD group versus 0% of controls; P = 0.007). This study provides the first evidence that boys with DMD show similar increased unconditioned startle responses to threat to the mdx mouse, which in the mouse respond to brain dystrophin restoration. Our study provides new insights into the neurobiology underlying the complex neuropsychiatric co-morbidities in DMD and defines an objective measure of this CNS phenotype, which will be valuable for future CNS-targeted dystrophin-restoration studies.

Infant excitation/inhibition balance interacts with executive attention to predict autistic traits in childhood

BACKGROUND

Autism is proposed to be characterised by an atypical balance of cortical excitation and inhibition (E/I). However, most studies have examined E/I alterations in older autistic individuals, meaning that findings could in part reflect homeostatic compensation. To assess the directionality of effects, it is necessary to examine alterations in E/I balance early in the lifespan before symptom emergence. Recent explanatory frameworks have argued that it is also necessary to consider how early risk features interact with later developing modifier factors to predict autism outcomes.

METHOD

We indexed E/I balance in early infancy by extracting the aperiodic exponent of the slope of the electroencephalogram (EEG) power spectrum ('1/f'). To validate our index of E/I balance, we tested for differences in the aperiodic exponent in 10-month-old infants with (n = 22) and without (n = 27) neurofibromatosis type 1 (NF1), a condition thought to be characterised by alterations to cortical inhibition. We then tested for E/I alterations in a larger heterogeneous longitudinal cohort of infants with and without a family history of neurodevelopmental conditions (n = 150) who had been followed to early childhood. We tested the relevance of alterations in E/I balance and our proposed modifier, executive attention, by assessing whether associations between 10-month aperiodic slope and 36-month neurodevelopmental traits were moderated by 24-month executive attention. Analyses adjusted for age at EEG assessment, sex and number of EEG trials.

RESULTS

Infants with NF1 were characterised by a higher aperiodic exponent, indicative of greater inhibition, supporting our infant measure of E/I. Longitudinal analyses showed a significant interaction between aperiodic slope and executive attention, such that higher aperiodic exponents predicted greater autistic traits in childhood, but only in infants who also had weaker executive functioning abilities.

LIMITATIONS

The current study relied on parent report of infant executive functioning-type abilities; future work is required to replicate effects with objective measures of cognition.

CONCLUSIONS

Results suggest alterations in E/I balance are on the developmental pathway to autism outcomes, and that higher executive functioning abilities may buffer the impact of early cortical atypicalities, consistent with proposals that stronger executive functioning abilities may modify the impact of a wide range of risk factors.

ZSWIM7 Is Associated With Human Female Meiosis and Familial Primary Ovarian Insufficiency

BACKGROUND

Primary ovarian insufficiency (POI) affects 1% of women and is associated with significant medical consequences. A genetic cause for POI can be found in up to 30% of women, elucidating key roles for these genes in human ovary development.

OBJECTIVE

We aimed to identify the genetic mechanism underlying early-onset POI in 2 sisters from a consanguineous pedigree.

METHODS

Genome sequencing and variant filtering using an autosomal recessive model was performed in the 2 affected sisters and their unaffected family members. Quantitative reverse transcriptase PCR (qRT-PCR) and RNA sequencing were used to study the expression of key genes at critical stages of human fetal gonad development (Carnegie Stage 22/23, 9 weeks post conception (wpc), 11 wpc, 15/16 wpc, 19/20 wpc) and in adult tissue.

RESULTS

Only 1 homozygous variant cosegregating with the POI phenotype was found: a single nucleotide substitution in zinc finger SWIM-type containing 7 (ZSWIM7), NM_001042697.2: c.173C > G; resulting in predicted loss-of-function p.(Ser58*). qRT-PCR demonstrated higher expression of ZSWIM7 in the 15/16 wpc ovary compared with testis, corresponding to peak meiosis in the fetal ovary. RNA sequencing of fetal gonad samples showed that ZSWIM7 has a similar temporal expression profile in the developing ovary to other homologous recombination genes.

MAIN CONCLUSIONS

Disruption of ZSWIM7 is associated with POI in humans. ZSWIM7 is likely to be important for human homologous recombination; these findings expand the range of genes associated with POI in women.

Adamantinomatous craniopharyngioma as a model to understand paracrine and senescence-induced tumourigenesis

Cellular senescence is a process that can prevent tumour development in a cell autonomous manner by imposing a stable cell cycle arrest after oncogene activation. Paradoxically, senescence can also promote tumour growth cell non-autonomously by creating a permissive tumour microenvironment that fuels tumour initiation, progression to malignancy and metastasis. In a pituitary tumour known as adamantinomatous craniopharyngioma (ACP), cells that carry oncogenic β-catenin mutations and overactivate the WNT signalling pathway form cell clusters that become senescent and activate a senescence-associated secretory phenotype (SASP). Research in mouse models of ACP has provided insights into the function of the senescent cell clusters and revealed a critical role for SASP-mediated activities in paracrine tumour initiation. In this review, we first discuss this research on ACP and subsequently explore the theme of paracrine tumourigenesis in other tumour models available in the literature. Evidence is accumulating supporting the notion that paracrine signalling brought about by senescent cells may underlie tumourigenesis across different tumours and cancer models.

Loss-of-Function Variants in TBC1D32 Underlie Syndromic Hypopituitarism

CONTEXT

Congenital pituitary hormone deficiencies with syndromic phenotypes and/or familial occurrence suggest genetic hypopituitarism; however, in many such patients the underlying molecular basis of the disease remains unknown.

OBJECTIVE

To describe patients with syndromic hypopituitarism due to biallelic loss-of-function variants in TBC1D32, a gene implicated in Sonic Hedgehog (Shh) signaling.

SETTING

Referral center.

PATIENTS

A Finnish family of 2 siblings with panhypopituitarism, absent anterior pituitary, and mild craniofacial dysmorphism, and a Pakistani family with a proband with growth hormone deficiency, anterior pituitary hypoplasia, and developmental delay.

INTERVENTIONS

The patients were investigated by whole genome sequencing. Expression profiling of TBC1D32 in human fetal brain was performed through in situ hybridization. Stable and dynamic protein-protein interaction partners of TBC1D32 were investigated in HEK cells followed by mass spectrometry analyses.

MAIN OUTCOME MEASURES

Genetic and phenotypic features of patients with biallelic loss-of-function mutations in TBC1D32.

RESULTS

The Finnish patients harboured compound heterozygous loss-of-function variants (c.1165_1166dup p.(Gln390Phefs*32) and c.2151del p.(Lys717Asnfs*29)) in TBC1D32; the Pakistani proband carried a known pathogenic homozygous TBC1D32 splice-site variant c.1372 + 1G > A p.(Arg411_Gly458del), as did a fetus with a cleft lip and partial intestinal malrotation from a terminated pregnancy within the same pedigree. TBC1D32 was expressed in the developing hypothalamus, Rathke's pouch, and areas of the hindbrain. TBC1D32 interacted with proteins implicated in cilium assembly, Shh signaling, and brain development.

CONCLUSIONS

Biallelic TBC1D32 variants underlie syndromic hypopituitarism, and the underlying mechanism may be via disrupted Shh signaling.

Prognostic significance of age in 5631 patients with Wilms tumour prospectively registered in International Society of Paediatric Oncology (SIOP) 93-01 and 2001

BACKGROUND

To enhance risk stratification for Wilms tumour (WT) in a pre-operative chemotherapy setting, we explored the prognostic significance and optimal age cutoffs in patients treated according to International Society of Paediatric Oncology Renal Tumour Study Group (SIOP-RTSG) protocols.

METHODS

Patients(6 months-18 years) with unilateral WT were selected from prospective SIOP 93-01 and 2001 studies(1993-2016). Martingale residual analysis was used to explore optimal age cutoffs. Outcome according to age was analyzed by uni- and multivariable analysis, adjusted for sex, biopsy(yes/no), stage, histology and tumour volume at surgery.

RESULTS

5631 patients were included; median age was 3.4 years(IQR: 2-5.1). Estimated 5-year event-free survival (EFS) and overall survival (OS) were 85%(95%CI 83.5-85.5) and 93%(95%CI 92.0-93.4). Martingale residual plots detected no optimal age cutoffs. Multivariable analysis showed lower EFS with increasing age(linear trend P<0.001). Using previously described age categories, EFS was lower for patients aged 2-4(HR 1.34, P = 0.02), 4-10(HR 1.83, P<0.0001) and 10-18 years(HR 1.74, P = 0.01) as compared to patients aged 6 months-2 years. OS was lower for patients 4-10 years(HR 1.67, P = 0.01) and 10-18 years(HR 1.87, P = 0.04), but not for 2-4 years(HR 1.29, P = 0.23). Higher stage, histological risk group and tumour volume were independent adverse prognostic factors.

CONCLUSION

Although optimal age cutoffs could not be identified, we demonstrated the prognostic significance of age as well as previously described cutoffs for EFS (2 and 4 years) and OS (4 years) in children with WT treated with pre-operative chemotherapy. These findings encourage the consideration of age in the design of future SIOP-RTSG protocols.