Chanika Phornphutkul

Introduction: Artificial intelligence (AI) is increasingly prevalent. Patients and clinicians may use AI-based tools in many different languages. Objective: To investigate AI translation tools for descriptions of genetic conditions and how AI identification of genetic conditions is affected by translations. Materials and Methods: We used Neural machine translation (NMT) and large language-model (LLM) translation to translate descriptions of 40 genetic conditions into 191 and 93 languages, respectively. Excluding translations retaining English medical terms verbatim, we respectively focused on 139 and 70 languages. After assessing translations, we assessed the ability of 3 proprietary and 3 open-weight general LLMs to identify conditions in the translations. We analyzed how accuracy was affected by the conditions’ prevalence in the literature, and attributes of the languages (the script, language family, and prevalence of the language in training sources). We also investigated adaptive translation for select languages. Results: We found significant differences in condition identification based on the translation method, condition, language, and prediction model. The accuracy of some models was more affected than others by factors like the conditions’ literature prevalence, language script, family, and language prevalence. Adaptive translation for select languages did not improve translations or diagnostic accuracy with the 3 tested LLMs. However, further analysis with 1 language showed that this approach was more effective with smaller LLMs. Conclusions: AI-based translation has variable performance, which can affect the ability of AI models to recognize genetic conditions. These findings should inform safe medical AI use to support consistent performance in different languages.

We describe a 19-month-old male with GH deficiency, central hypothyroidism, and hypoprolactinemia due to a novel mutation of POU1F1. The patient presented with short stature and a compound heterozygous mutation of POU1F1: the pathogenic nonsense variant was inherited from his mother [c.514C > T (p.Arg172)] and a variant of unknown significance from his father [c.182G > T (p.Gly61Val)]. He had a recumbent length of 69 cm (Z score –5.27) and a weight of 7.94 kg (Z score –3.09) at 19 months. Further evaluation showed free T4 0.8 ng/dL (10.3 pmol/L) [reference range: 0.88-4.8 ng/dL (11.3-61.8 pmol/L)], TSH 3.270 µIU/mL (reference range: 0.82-5.91 µIU/mL), IGF binding protein 3 407 µg/L (reference range: 1289 −3634 µg/L), and prolactin 3.1 ng/mL [(7.75 nmol/L) (reference range: 5.3-28.9 ng/mL) (13.5-72.3 nmol/L)]. Levothyroxine, 3.1 µg/kg/day was started. A GH stimulation test using clonidine showed a peak GH level of 0.7 ng/mL (0.7 µg/L) [reference range: ≥ 10 ng/mL (≥ 10 µg/L)]. Magnetic resonance imaging of the brain showed anterior pituitary hypoplasia. He was started on GH 0.17 mg/kg/week with good response. To the best of our knowledge, the c.182G > T (p. Gly61Val) variant in POU1F1 is likely pathogenic, contributing to the pituitary hormone deficiencies in this child.

PRKAG2 encodes the γ₂-subunit isoform of 5′-AMP-activated protein kinase (AMPK), a heterotrimeric enzyme with major roles in the regulation of energy metabolism in response to cellular stress. Mutations in PRKAG2 have been implicated in a unique hypertrophic cardiomyopathy (HCM) characterized by cardiac glycogen overload, ventricular preexcitation, and hypertrophy. We identified a novel, de novo PRKAG2 mutation (K475E) in a neonate with prenatal onset of HCM. We aimed to investigate the cellular impact, signaling pathways involved, and therapeutic options for K475E mutation using cells stably expressing human wild-type (WT) or the K475E mutant. In human embryonic kidney-293 cells, the K475E mutation induced a marked increase in the basal phosphorylation of T172 and AMPK activity, reduced sensitivity to AMP in allosteric activation, and a loss of response to phenformin. In H9c2 cardiomyocytes, the K475E mutation induced inhibition of AMPK and reduced the response to phenformin and increases in the phosphorylation of p70S6 kinase (p70S6K) and eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1). Primary fibroblasts from the patient with the K475E mutation also showed marked increases in the phosphorylation of p70S6K and 4E-BP1 compared with those from age-matched, nondiseased controls. Moreover, overexpression of K475E induced hypertrophy in H9c2 cells, which was effectively reversed by treatment with rapamycin. Taken together, we have identified a novel, de novo infantile-onset PRKAG2 mutation causing HCM. Our study suggests the K475E mutation induces alteration in basal AMPK activity and results in a hypertrophy phenotype involving the mechanistic target of rapamycin signaling pathway, which can be reversed with rapamycin. NEW & NOTEWORTHY We identified a novel, de novo PRKAG2 mutation (K475E) in the cystathionine β-synthase 3 repeat, a region critical for AMP binding but with no previous reported mutation. Our data suggest the mutation affects AMP-activated protein kinase activity, activates cell growth pathways, and results in cardiac hypertrophy, which can be reversed with rapamycin.