Ancient Loss of Catalytic Selenocysteine Spurred Convergent Adaptation in a Mammalian Oxidoreductase

Selenocysteine, the 21st amino acid specified by the genetic code, is a rare selenium-containing residue found in the catalytic site of selenoprotein oxidoreductases. Selenocysteine is analogous to the common cysteine amino acid, but its selenium atom offers physical-chemical properties not provided by the corresponding sulfur atom in cysteine. Catalytic sites with selenocysteine in selenoproteins of vertebrates are under strong purifying selection, but one enzyme, glutathione peroxidase 6 (GPX6), independently exchanged selenocysteine for cysteine <100 million years ago in several mammalian lineages. We reconstructed and assayed these ancient enzymes before and after selenocysteine was lost and up to today and found them to have lost their classic ability to reduce hydroperoxides using glutathione. This loss of function, however, was accompanied by additional amino acid changes in the catalytic domain, with protein sites concertedly changing under positive selection across distant lineages abandoning selenocysteine in glutathione peroxidase 6. This demonstrates a narrow evolutionary range in maintaining fitness when sulfur in cysteine impairs the catalytic activity of this protein, with pleiotropy and epistasis likely driving the observed convergent evolution. We propose that the mutations shared across distinct lineages may trigger enzymatic properties beyond those in classic glutathione peroxidases, rather than simply recovering catalytic rate. These findings are an unusual example of adaptive convergence across mammalian selenoproteins, with the evolutionary signatures possibly representing the evolution of novel oxidoreductase functions.

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.

Interpretable surface-based detection of focal cortical dysplasias: a Multi-centre Epilepsy Lesion Detection study

One outstanding challenge for machine learning in diagnostic biomedical imaging is algorithm interpretability. A key application is the identification of subtle epileptogenic focal cortical dysplasias (FCDs) from structural MRI. FCDs are difficult to visualize on structural MRI but are often amenable to surgical resection. We aimed to develop an open-source, interpretable, surface-based machine-learning algorithm to automatically identify FCDs on heterogeneous structural MRI data from epilepsy surgery centres worldwide. The Multi-centre Epilepsy Lesion Detection (MELD) Project collated and harmonized a retrospective MRI cohort of 1015 participants, 618 patients with focal FCD-related epilepsy and 397 controls, from 22 epilepsy centres worldwide. We created a neural network for FCD detection based on 33 surface-based features. The network was trained and cross-validated on 50% of the total cohort and tested on the remaining 50% as well as on 2 independent test sites. Multidimensional feature analysis and integrated gradient saliencies were used to interrogate network performance. Our pipeline outputs individual patient reports, which identify the location of predicted lesions, alongside their imaging features and relative saliency to the classifier. On a restricted 'gold-standard' subcohort of seizure-free patients with FCD type IIB who had T1 and fluid-attenuated inversion recovery MRI data, the MELD FCD surface-based algorithm had a sensitivity of 85%. Across the entire withheld test cohort the sensitivity was 59% and specificity was 54%. After including a border zone around lesions, to account for uncertainty around the borders of manually delineated lesion masks, the sensitivity was 67%. This multicentre, multinational study with open access protocols and code has developed a robust and interpretable machine-learning algorithm for automated detection of focal cortical dysplasias, giving physicians greater confidence in the identification of subtle MRI lesions in individuals with epilepsy.

It ain't what you do, it's the way that you do it: The pitfalls of using routine data to measure early infant HIV diagnosis in HIV-exposed infants

BACKGROUND

Early infant HIV diagnosis (EID) is critical to ensuring timely diagnosis of HIV-exposed infants, and treatment in those found to be infected. However estimates of coverage vary considerably, depending on data sources used. We used 4 methods to estimate coverage among a historical cohort of HIV-exposed infants in rural South Africa, between 2010-2016.

METHODS

We estimated the proportion of infants ever tested (methods 1-3) and tested by 7 weeks of age (1-4) as follows: (1) infants born to women identified as HIV-positive in demographic surveillance were linked to those with ≥1 EID result in routine laboratory surveillance; (2) the number of infants with ≥1 EID result in laboratory surveillance divided by the estimated number of HIV-exposed infants, calculated as total live births multiplied by antenatal HIV seroprevalence; (3) the number of infants with ≥1 EID result in routine laboratory surveillance, divided by the number of HIV-exposed infants as estimated by the district health service; (4) from documentation in infants' Road-to-Health-booklets.

RESULTS

The proportion ever tested was 43%, 88% and 138% for methods 1-3, and by 7 weeks of age was 25%, 49%, 86% and 46% for methods 1-4 respectively.

CONCLUSIONS

The four methods, applied to a range of routine data sources, resulted in estimates varying considerably, and the true coverage of EID remains unclear. Our findings highlight the importance of developing unique patient identifiers, improving training of healthcare providers using reporting systems, and ensuring the accuracy of healthcare records, to ensure the best possible health outcomes for HIV-exposed infants.