Appropriate investigation of the etiology and simultaneous management in children with hypoglycemia is paramount to prevent (irreversible) brain injury or even death (1), although controversy remains on the definition (e.g., diagnostic plasma glucose threshold, definition of at–risk neonates) of childhood hypoglycemia (2–4). From a pathophysiological perspective, hypoglycemia is caused by an imbalance between the amount of glucose molecules entering the blood compartment and the glucose clearance. A continuous supply of glucose is crucial for the function and development of the brain, because of the absence of sufficient cerebral energy stores in the form of glycogen, protein or fat. Glucose oxidation is responsible for ~ 70% of the energy production in the brain, whereas lactate and ketone bodies can function as important alternative fuels. Hypoglycemia causes energy deficiency in brain cells ultimately leading to brain cell death due to secondary molecular mechanisms, such as activation of neuronal glutamate receptors, oxidative stress, neuronal zinc release and activation of poly-ADP-ribose polymerase-1 (5, 6). Clinically, untreated hypoglycemia can lead to serious neurologic morbidity (encephalopathy, convulsions, coma, developmental delay), and even mortality (7). In individuals with a possible, underlying disorder, unrecognized hypoglycemia may accelerate and aggravate the metabolic decompensation resulting in poor outcomes.

Multiple metabolic pathways cooperate to maintain normal blood glucose concentrations in the fed and fasted state, such as glycogenolysis, gluconeogenesis, mitochondrial fatty acid oxidation, ketogenesis and ketolysis, ensuring energy homeostasis. These pathways are tightly regulated by hormonal (e.g., insulin, glucagon, cortisol, and growth hormone) and autonomic (e.g., catecholamines) responses (8) to maintain normoglycemia. Besides acquired conditions (e.g., insufficient supply/increased loss of nutrients, dysmaturity, drugs, hypothermia, sepsis), hypoglycemia may be explained by a genetic defect of these mechanisms that normally guarantee glucose homeostasis during feeding and fasting. Hypoglycemia is an important feature of several inherited metabolic disorders (IMDs) characterized by a history of childhood fasting intolerance (9).

Compared to adults, children are at an increased risk of developing hypoglycemia, as they have smaller hepatic glycogen and muscle protein stores, together with an increased energy expenditure and higher brain/body weight ratio (10). Consequently, most young children are not able to maintain normoglycemia during more than 24 hours of fasting (11–16). Studies with stable isotope techniques have quantified endogenous glucose production (EGP) in newborns (17) and older children (18). Huidekoper et al. designed a non-linear regression model for EGP from infancy to adulthood (19). Table 1 illustrates that glucose homeostasis is tightly controlled in an age-dependent manner, with a small glucose pool in the blood compartment (row D) and high glucose turnover rate (row E).

The most used diagnostic approach to childhood hypoglycemia consists of pattern recognition by combining a variety of information, including medical history (e.g., age at onset, relation with food), physical examination (e.g., growth charts, liver size), and laboratory tests. If critical samples (i.e. samples collected during metabolic stress) of plasma and urine can be obtained, static or single-point concentrations of biomarkers (e.g., blood glucose, lactate, gases, free fatty acids, ketones, acylcarnitines, amino acids, insulin, cortisol, growth hormone, and urine organic acids) can provide crucial information to establish an etiological diagnosis (21). When critical samples are not available or inconclusive, next-generation sequencing may further direct the diagnostic process. Several diagnostic algorithms exist for the clinical problem of childhood hypoglycemia (1, 3), but these schemes are often not sufficiently specific. Moreover, they underestimate the complexity of the swiftly changing clinical signs and concentrations of different fuels in the blood compartment. Hence, many hypoglycemic children do not fit these textbook schemes, and additional investigation is often required. In the end, an important group of individuals is diagnosed with idiopathic ketotic hypoglycemia (22, 23).

In this narrative review, we illustrate several dynamic methods which may be considered in the diagnostic work-up and monitoring of children with hypoglycemia. Three methods are presented: 1) functional in vivo tests; 2) in vivo metabolic profiling by continuous glucose monitoring (CGM); 3) stable isotope tracing, focusing on glycogen storage disease type I (GSD I). Future applications and benefits of dynamic methods in children with hypoglycemia are also discussed.

For decades, functional in vivo tests have been used for the etiological clarification of fasting intolerance, unexplained hypoglycemia, and to diagnose specific IMDs ( Table 2 ). In principle, these tests were particularly relevant as biochemical abnormalities might only appear intermittently in many IMDs. In times when diagnostic enzymatic studies and DNA studies were not routinely available, assessment of blood lactate curves after hexose (e.g. glucose, fructose and galactose) loading was used as a screening procedure in patients suspected of hepatic GSDs (24–26). The results of these tests combined with the assessment of response of blood glucose to glucagon injection provided major clues to identify the hepatic GSD subtype. As an example, the combination of increased blood lactate levels observed after oral glucose administration together with no response of blood glucose to intramuscular glucagon injection pointed to GSD III. Subsequently, the abnormally increasing lactate curves after ingestion of both fructose and galactose in GSD I patients has become one of the pillars, on which the current dietary recommendations are based. If performed, functional in vivo tests are conducted strictly adhering to a defined protocol to maximize the reliability and minimize the risks for patients. Functional in vivo tests should only be performed in experienced centers and under close medical supervision. Protocols for several functional in vivo (provocation and/or loading) tests can be found elsewhere (9, 27).

In theory, evaluation of metabolic changes in response to an explorative or extended controlled fasting test may be helpful to reach the etiological diagnosis in children with hypoglycemia. Fasting induces several hormonal and metabolic responses to ensure an endogenous supply of energy after cessation of exogenous intake (28). Due to the complexity of the (patho)physiological response to fasting, complete hormonal and metabolic investigations should be performed before and at the end of the monitored fasting. Spare serum/urine samples should also be stored for any additional tests to be performed afterwards. Bedside monitoring of these laboratory parameters, and in particular during the extended controlled fasting test, vital parameter monitoring and CGM should also be performed. Under these circumstances, the maximal duration of the fasting is based on the clinical history (expected fasting tolerance) and on the child’s age (usually 12-16 hours at 6-12 months, 18 hours at 1-2 years, 20 hours at 2-7 years, 24 hours in children > 7 years). The fasting should be stopped at any time if the glucose concentration is below 2.6 mmol/L or the patient develops symptoms of hypoglycemia (21).

Even before the era of expanded newborn screening and next generation sequencing, the diagnostic yield of fasting studies has been limited. Morris et al. confirmed an endocrine disorder (n=15) or an IMD (n=15) in a group of 138 patients undergoing fasting studies (29). Bappal and Mula-Abed diagnosed an endocrine disorder (n=21) or an IMDs (n=31) in 96 patients subjected to diagnostic fasting tests (30). Both studies did not provide detailed information on confirmatory diagnostic testing in their patients. More recently Graves et al. showed that 4/91 fasting challenges performed in children with suspected or documented hypoglycemia resulted in an etiological diagnosis (i.e., hepatic GSD, partial ACTH deficiency, panhypopituitarism, isolated central hypothyroidism) (31).

Although in vivo functional tests have the advantage of triggering metabolic stress in a supervised setting, they are burdensome and inconvenient for young patients and their parents. If IMDs are not yet excluded, these tests can cause serious complications in unexperienced hands (e.g., worsening of hypoglycemia, accumulation of toxic metabolites). Therefore, psychological burden due to unexpected hypoglycemia for patients and their families should be balanced against the stress and risks associated with in vivo functional testing. Additionally, in vivo functional tests pose a significant organizational and logistical burden (e.g., need for a fully equipped laboratory and a dedicated medical team) by including several samples to be collected at specific time points (before and after the challenge), under strictly supervised conditions, and requiring hospital admission.

Based on these considerations and because indications have become more restricted, most of the in vivo functional tests have fallen into disuse. A few tests are nowadays even abandoned and considered obsolete (9). Historical indications for functional in vivo tests have been shifted with the increased awareness for rare diseases, the development of diagnostic algorithms and refined basal metabolic investigations (such as organic acid analysis, plasma acylcarnitines, specific enzyme analysis, in vitro metabolic flux analysis in cultured skin fibroblasts), introduction of expanded population newborn screening, and implementation of next-generation sequencing. Still, some functional in vivo tests -with slight modifications - may be useful to provide phenotypic information to support genetic variant classification, or to define outcome in clinical trials. In such situations, these tests remain the major (monitoring) methods allowing the assessment of metabolic changes in vivo in response to external factors (e.g., metabolic stress, treatment). In addition, controlled fasting test may guide the diagnostic process in suspected congenital hyperinsulinism when critical samples are not available (e.g., molecular genetic testing, ¹⁸F-DOPA positron emission tomography) (32, 33).

Continuous glucose monitoring (CGM) constitutes another dynamic method in the diagnostic work-up and monitoring of children with hypoglycemia. Progress has been made since CGM was firstly described in 1979 for a patient with insulinoma (34). Modern CGM devices consist of a subcutaneous sensor and a receiver. A wearable sensor is attached to the skin and measures glucose (with variable frequency) through a subcutaneous needle inserted in the interstitial fluid. Via an enzymatic technology (glucose oxidase), the sensor generates an electric current proportional to the glucose concentration in the interstitium. The electric current flows through a continuously shifting algorithm to generate a glucose value which is eventually transmitted to the receiver (receiving unit or compatible smartphone). The algorithm is finetuned via calibration by patient or factory to ensure accuracy over the recording periods (35). Major advantages of CGM are the availability of real-time data, continuous data (density), alarming function, at home use, and data-sharing with caregivers and health care providers, and it being a relatively non-invasive technique (no or fewer finger sticks).

Currently, CGM devices are only licensed for patients with diabetes mellitus (DM), for whom CGM-related outcome parameters (e.g., time-in-range and time-above-range) are defined to guide dietary changes and/or medication adjustments (36). Additionally, CGM has been used in clinical research to assess glucose metabolism in various conditions, including acute coronary syndrome (37) and individuals following an intermittent fasting dietary regimen (38). Evidence on its benefit is also accumulating in endocrine disorders (39, 40) and IMDs associated with hypoglycemia (41), particularly in hepatic GSDs ( Table 3 ).

A significant correlation between CGM values and capillary glucose values has been found in patients with GSD I (42, 46). CGM can unveil both unrecognized (nocturnal) hypoglycemia and hyperglycemia and hence, provide information on the effect of the dietary regimen on glucose homeostasis in patients with hepatic GSDs (43). By providing visual glucose trends, CGM also has the advantage of generating dynamic, daily data compared to traditional single-point data. Thus, it can improve the home site monitoring (constituting an easier tool for patients compared to traditional methods) (44, 47) and the day-to-day healthcare for patients with hepatic GSDs (45, 46). CGM has also proven a reliable tool to assess the effect of novel/optimized dietary and/or medical treatments in patients with hepatic GSDs (48, 49). Additionally, it can assist the physicians in the assessment of children with hypoglycemia in combination with relevant information on personal and dietary history and appropriate biochemical investigations.

For patients with ketotic fasting intolerance (e.g., ketotic GSDs, idiopathic ketotic hypoglycemia, or any situation in which euglycemic ketonemia may occur, such as patients on ketogenic diets), parallel documentation of point-of-care capillary beta-hydroxybutyrate (BHB) concentrations by a portable ketone meter may provide important additional information. To the best of our knowledge, reference values for capillary BHB during the day are not available in the pediatric population, but information on BHB normal values after overnight fasting can be extrapolated from historical publications ( Table 4 ). Most studies provide BHB concentrations after prolonged fasting, but recently, Parmar et al. demonstrated that serum point-of-care (POCT) BHB measurement using a precision Xtra meter are comparable to serum BHB concentrations in healthy children after an overnight fast (51). In 94 children (age range 6 months to 18.7 years) scheduled to undergo elective surgery, POCT BHB concentrations ranged from 0 to 1.1 mmol/l, but unfortunately, detailed information on the correlation between age and BHB concentrations was not provided in their study.

Currently, only the BHB meters can be used for minimally invasive home monitoring to guide dietary management after a working diagnosis has been established. To our opinion, the combination of real-time CGM with alarm function and a portable ketone meter may be used in children undergoing parallel diagnostic and therapeutic metabolic profiling for hypoglycemic disorders [(52) and Table 5 ]. A hybrid protocol combining the exploration of fasting tolerance (by CGM) with so-called in vivo metabolic profiling (BHB and possibly additional biochemical parameters) may replace the historical approach in patients suspected with hepatic GSDs. Such approach will likely further decrease the use of diagnostic in vivo fasting tests. Indeed, the combination of information collected through different testing approaches will likely represent the future strategy to characterize childhood hypoglycemia, including both the diagnostic and monitoring phase. Additional factors (e.g., experience of the medical team, patients’ needs, available resources, insurance policy) may result in variations in local practice, even between specialized centers.

The flow of metabolites through a (disrupted) metabolic pathway is determined by both the rate of supply of substrates and the residual activities of the enzymes in the pathway. In patients with IMDs, changes in the distribution of fluxes occur secondary to enzyme defect(s) and/or altered substrate availability. By assessing the flow of metabolites through specific metabolic pathways in vivo, fluxomics can provide an overall description of tissue phenotypes in a relatively simple manner (53). (Radioactive or stable) isotope tracers have been used since 1930s to characterize such changes in cellular and whole-body metabolism (54). Stable isotopes techniques are well established methods in fundamental/translational research to assess metabolic fluxes/changes (55, 56). Proof-of-concept research has also shown the potential clinical (research and care) application of stable isotopes to characterize the biochemical and clinical phenotype of IMDs associated with hypoglycemia (57, 58). An overview on current evidence paving the path for future use of stable isotope techniques in diagnosis and monitoring of IMDs characterized by hypoglycemia is provided, with specific focus on GSD I.

Although the past years have witnessed major advances in understanding the pathophysiology of disorders associated with hypoglycemia, relevant limitations in assessment and monitoring strategies remain. The current diagnostic approach to children presenting with hypoglycemia includes, besides medical/dietary history and clinical examination, biochemical tests (performed on “critical samples”) and eventually genetic confirmation (21). However, critical samples or reliable biochemical data are not always available or may not provide conclusive information. Also, traditional biochemical markers are static parameters and do not always appear sufficient to explain the phenotypic variability in patients with IMDs associated with hypoglycemia (94). Furthermore, those biomarkers may not adequately reflect the dynamic pathophysiological processes taking place in response to external factors such as metabolic stressors or treatment. Thus, there remain a group of patients with unsolved diagnostic and/or monitoring questions, that may be addressed by dynamic methods, such as in vivo functional tests, CGM and stable isotope tracing.

Functional in vivo tests have the advantage of characterizing metabolic pathways, such as glycogenolysis, gluconeogenesis, mitochondrial fatty acid oxidation, ketogenesis, and ketolysis, in a dynamic fashion. Yet, their use is progressively decreasing. Therefore, we do not recommend performing such tests in unexperienced centers. Several functional in vivo tests (e.g., controlled fasting challenge) are currently designed to monitor management after novel, innovative treatments (NCT03517085, NCT03665636, NCT05095727, NCT NCT04538989, NCT04720859, NCT03761693) in centers of expertise. Fasting test may still play a role in the diagnostic process of congenital hyperinsulinism. Yet, the development of safer and simpler diagnostic and monitoring tools for clinical research/care remains a compelling need (95).

CGM constitutes a highly informative technique to (non-invasively) monitor glucose levels in a dynamic fashion in patients with hypoglycemia (39–41, 45). Although physicians working in centers of expertise have become experienced with the CGM technology, its use in less experienced centers remains limited. Therefore, knowledge dissemination on the CGM technology and data interpretation is paramount. In fact, given its relatively low organizational burden and invasiveness, CGM may constitute a valuable technique. Ideally, individual patients’ CGM results could be compared with (1) the patient’s historical CGM data before and after intervention, and CGM data from both (2) an age-matched disease cohort and (3) age-matched healthy controls. Shah et al. recently reported CGM profiles from 153 nonpregnant, healthy, nondiabetic children and adults (age ≥6 years) with nonobese body mass index, using a blinded Dexcom G6 CGM, with once-daily calibration, for up to 10 days (96). Advanced CGM data analysis represents a promising strategy for minimally invasive patient monitoring (48). However, several challenges constrain CGM use in patients with hypoglycemia. First, neither general agreement on relevant CGM-related outcome parameters nor reference values for such parameters exist for patients with IMDs. Ideally, individual patients’ CGM results would be comparable once these issues have been solved. Second, the lag time (i.e., the time delay between a change in plasma glucose and the reporting of this change by the CGM device) may affect CGM reliability in patients experiencing sudden drops in glucose concentrations, as in hepatic GSDs. Third, most CGM devices do not contain predictive alarms (i.e., predicting an upcoming low glucose event using trend information) and rely upon the user to identify hypoglycemia and make manual adjustments. Also, current predictive alarms are based on a linear function of glucose trends, which may not adequately reflect the exponential decrease observed in hypoglycemic GSD patients (97). Likely the progress in machine learning and deep learning approaches and artificial neural networks will develop reliable algorithms for hypoglycemia prediction in patients with IMDs (35). Fourth, some devices require fingerprick calibration, which currently prioritizes hyperglycemia over hypoglycemia, thus constituting a potential additional source of inaccuracy, particularly a in the low range of blood glucose (35). Notably, most recent devices such as Dexcom G6 and Freestyle Libre are factory-calibrated (98). Fifth, as no specific CGM device is formally licensed for patients with IMDs, several types of CGM devices are used ( Table 3 ) presenting with different functionalities, advantages, and limitations. Additionally, data management applications currently vary, depending on the type of CGM device installed and interconnectivity between CGM software packages, and digital health records are lacking. Harmonization of data management applications as well as improved interconnectivity between CGM software packages and data health records are strongly warranted to facilitate data sharing among healthcare professionals and beyond towards citizen science applications. Sixth, while a 14-day data collection is recommended in DM patients (99), the number of measurements required to generate reliable CGM profiles in patients with IMDs is still unknown.

Stable isotope techniques appear attractive as they give access to the in vivo dynamics of hypoglycemia from a pathophysiological perspective. In principle, this may prevent a late diagnosis in patients presenting with milder phenotypes (100, 101). Yet, this technique has been explored experimentally in a limited number of conditions (e.g., idiopathic ketotic hypoglycemia, GSD I, congenital hyperinsulinism), but it is currently not clinically implemented, neither applied in clinical trials. Potential clinical benefit of this approach includes the definition of an easy and safe method to guide the diagnosis of patients with hypoglycemia and to improve patient monitoring, possibly constituting an additional tool to assess the effect of novel treatments (e.g., gene therapy, mRNA therapy). Given the safety of this methodology, it appears well suited for clinical trials in humans. Traditionally, venous blood samples are collected before the intravenous tracer infusion (after a priming dose) and several times after achieving the steady state (60). Such a procedure may potentially hamper the translation of this approach into clinical research/care in its current form. Previous research has shown the reliability of dried blood spot (DBS) sampling (compared to traditional venous sampling) in mice by tail bleeding to assess glucose kinetics following infusion of D-[U-¹³C]-glucose, D-[1-²H]-galactose, and [2-¹³C]-glycerol (102) or after an intraperitoneal injection of D-[6,6-²H]-glucose (58). Although these findings pave the way for possible oral (non-steady-state) administration of stable isotope tracers to humans (who are metabolically in a steady state, i.e. upon fasting without any unknown (unmeasured) and/or non-steady additional exogenous glucose input), the effect of this administration route on the tracer bioavailability remains unknown. Future studies exploring this approach in humans are warranted. Additionally, the use of stable isotope tracers in animal models can guide further dissection of pathophysiological derangements occurring in IMDs.

In addition to functional in vivo tests, metabolic profiling by CGM and stable isotope techniques, the microdialysis technique, at least in theory, would enable continuous sampling of multiple metabolites from the extracellular space from various tissues, such as blood, adipose tissue, or brain in experimental models (103, 104). The combination of stable isotope tracing and microdialysis in animal models appears intriguing (105). As microdialysis allows not only glucose monitoring (106, 107) but also sampling of other metabolites (e.g., glycerol, acylcarnitines) (108), it has a great potential for experimental research in patients with IMDs. Still, there are currently no commercial systems available.

AR, MGSR, MHO and TGJD wrote the first version of the manuscript. THvD, BMB, D-JR and TGJD critically reviewed the manuscript. AR drafted and wrote the final version of manuscript. TGJD critically reviewed the final version of the manuscript. All authors substantially contributed to the work and were involved in (a) conception and design of the study and/or analysis and interpretation of data, and (b) revising the article critically for important intellectual content. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work. All authors confirm the absence of previous similar or simultaneous publications. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling editor declared a past co-collaboration with the authors.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.