· Childhood leukemia survivors are at a higher risk for various short and long-term complications, including insulin resistance, obesity, and hyperglycemia.

· Glycated albumin (GA), an alternative marker of hyperglycemia, was positively correlated to glycated hemoglobin in pediatric patients with acute leukemia as in patients diagnosed with type 2 diabetes mellitus, suggesting that GA may serve as a complementary biomarker of glycemic status in this specific group.

Acute leukemia is the most common malignancy diagnosed during childhood. Advancements in treatment have led to a marked increase in survival rate, with some studies reporting survival rates of up to 90% depending on the risk group [1]. Issues related to the quality of life of survivors, including long-term complications such as hyperglycemia, insulin resistance, dyslipidemia, hypertension, obesity, and cardiovascular disease, have emerged as major concerns. Treatment methods for acute leukemia, which include drugs such as corticosteroids and asparaginase as well as cranial radiotherapy, can induce hyperglycemia and potentially cause childhood obesity during and/or following treatment [2]. Although the exact mechanism underlying these effects is unclear, these treatments may affect carbohydrate and lipid metabolism, resulting in insulin resistance and obesity [3]. Moreover, lack of physical activity and unhealthy eating habits may further increase the risks of these complications. Considering the adverse effects of these complications on the morbidity and mortality of pediatric patients, methods to minimize these sequelae are essential.

Hyperglycemia is often observed in patients undergoing chemotherapy. Transient hyperglycemia is especially prominent during remission induction therapy with glucocorticoid and asparaginase treatment and has been reported in up to 34% of pediatric patients with acute lymphoblastic leukemia (ALL) [4]. Steroid treatment may damage pancreatic beta cells; reduce insulin production; and increase insulin resistance, gluconeogenesis, and lipolysis, resulting in impaired glucose tolerance [5]. L-Asparaginase can also reduce insulin synthesis and secretion by depleting asparagine [6]. This drug is often associated with pancreatitis; it causes beta cell dysfunction and reduces insulin secretion [4]. Some of these effects are permanent and can lead to drug-induced diabetes [6]. Treatment of drug-induced diabetes includes careful glucose monitoring, diet modification, and insulin therapy when conservative management fails. Although most hyperglycemic events are transient, childhood leukemia survivors have a higher burden of long-term comorbidities, including diabetes mellitus [7]. Thus, close monitoring and follow up are crucial in this specific population.

Glycated hemoglobin (HbA1c) is a well-known marker of glycemic control that reflects the blood glucose status over the previous 2–3 months. Unfortunately, HbA1c level can be affected by the red blood cell (RBC) cycle [8]. As HbA1c is a product of the non-enzymatic chemical reaction of hemoglobin (Hb) with blood glucose, changes in Hb can affect HbA1c levels. Thus, hemolytic anemia, major blood loss, and blood transfusions may lead to misleading HbA1c results. RBC transfusion has been shown to reduce HbA1c levels [9]. Diseases such as anemia, hemoglobinopathies, and cystic fibrosis may also be associated with decreased HbA1c levels despite the presence of a high blood glucose level. Though debatable, a study found that HbA1c had low sensitivity for the detection of diabetes and impaired glucose tolerance in childhood cancer survivors treated with hematopoietic stem cell transplantation and total body irradiation [10]. Moreover, measurement methods of this marker varied widely among institutions [11]. To overcome these limitations, new markers of hyperglycemia have been investigated. Glycated albumin (GA) and fructosamine are products of non-enzymatic serum protein glycation and are independent of the RBC cycle. Glycation of albumin follows a relatively shorter cycle than that of Hb, and GA reflects the serum glucose level in the previous 2–3 weeks. These markers show faster improvement than HbA1c during recovery from acute diabetes ketoacidosis or diabetes onset [12].

However, there is a lack of data on the usefulness of GA in the pediatric population, particularly in patients with acute leukemia, who are at increased risk of diabetes mellitus compared to healthy peers [7]. Therefore, the objective of this study was to examine the relationship among GA, fasting blood glucose (FBG), and HbA1c levels in a pediatric population, particularly pediatric patients diagnosed with acute leukemia.

To the best of our knowledge, this is the first study to investigate the potential relationship between GA and HbA1c levels in pediatric patients with acute leukemia. In our study, GA, FBG, and HbA1c levels showed strong correlations with T2DM in pediatric patients, as expected. Although the GA level was only slightly lower in the acute leukemia group, it was positively correlated with the HbA1c level in the acute leukemia group and the T2DM group but not in the normal control group. Notably, GA was positively associated with increases in HbA1c levels in patients with acute leukemia as in patients with T2DM, even after adjustment for age, sex, and BMI z-score. Given the elevated risk of T2DM in children treated for acute leukemia, vigilant surveillance for endocrine complications is important. Our findings suggest that GA measurement may have a role in patients with acute leukemia and can complement conventional methods for evaluating glycemic status in this specific population.

GA level has been shown to be highly correlated with HbA1c level in previous studies of patients with diabetes. In adult studies, GA was a highly specific but less sensitive marker of hyperglycemia than HbA1c or FBG levels [15]. However, several previous studies have noted that HbA1c is a poor marker for glucose monitoring in gestational diabetes mellitus (GDM) as pregnancy occurs over the relatively short time period of 40 weeks and can involve iron deficiency [16]. Sugawara et al. suggested that measurement of the GA level and the GA/HbA1c ratio were more useful than the HbA1c level for predicting infant complications due to GDM [17]. The demand for new biomarkers has led to several studies of GA. Kim et al. [18] reported that a GA level > 16% was associated with poorer outcomes in stroke patients without diabetes and was a prognostic biomarker for short-term stroke outcomes. Another study recommended GA as a possible alternative biomarker of glycemic status in patients with end-stage renal disease who have altered glucose and insulin homeostasis due to decreased renal metabolism [19]. As GA reflects recent glycemic status and can identify glycemic changes before fluctuations in the HbA1c level, it may better assess glucose variability [20]. A prospective study by Desouza et al. [21] found GA to be a better short-term correlation factor for blood glucose changes than other glycemic indices. However, while studies in pediatric patients with diabetes also showed that GA better reflected the glycemic status than HbA1c as in adult studies, studies of GA in the general pediatric population are limited and controversial [11].

We found that GA and HbA1c levels were correlated in both the acute leukemia and T2DM groups. While the correlation between GA and HbA1c is well documented in adult populations, findings in pediatric populations are scarce and often contradictory. A Japanese study of a pediatric population without diabetes reported a surprising inverse correlation between GA and HbA1c (r=-0.139, P<0.001) [22]. More recently, a large study from the US found GA to be poorly correlated with traditional markers of hyperglycemia in healthy adolescents with normoglycemia or mild hyperglycemia [23]. In our study, while GA and HbA1c did not demonstrate any significant correlation in the normal control group, GA was significantly correlated with HbA1c in the acute leukemia group. Subgroup analysis also revealed significant correlation of HbA1c and GA in the prediabetes subgroup of the acute leukemia group but not in the prediabetes subgroup of the normal control group (P=0.005). Prior studies in adults have shown that GA measurements are weakly correlated with HbA1c and FBG measurements at lower levels but showed a strong correlation at higher levels [24]. This may explain the lack of correlation seen in the healthy pediatric group. Nevertheless, GA was significantly correlated with HbA1c in the acute leukemia group as in the T2DM group. Neither the normal control group nor acute leukemia group had been diagnosed with diabetes mellitus or any glucose metabolism problem. Previous studies have demonstrated that insulin resistance is affected by prolonged corticosteroid use [2]. Other chemotherapeutic drugs, such as L-asparaginase, can also alter glucose metabolism and lead to a spectrum of hyperglycemia, ranging from transient episodes to diabetic ketoacidosis or prolonged diabetes, which can impact survival outcomes [25]. The combination effect of various treatment-related chemotherapeutic agents and proinflammatory cytokines is known to contribute to various metabolic disturbances in these specific groups [26]. One study found no difference in obesity status among pediatric patients with leukemia with or without insulin resistance (obese, 45.2%; normal weight, 41.9%) [27]. This suggests that insulin resistance may occur in children with leukemia regardless of their obesity status. Such hyperglycemic events and insulin resistance may arise in children with acute leukemia as a result of the long-term synergistic effects of various medications. Suwa et al. [28] showed that GA levels and continuous glucose monitoring parameters were significantly correlated, suggesting the need for wider application of GA as a glucose monitoring marker, particularly in susceptible populations. Although most hyperglycemia events reported in acute leukemia treatment are usually transient, they may be associated with adverse long-term health outcomes [29]. Studies have found that not only do childhood cancer survivors suffer from a higher incidence of diabetes mellitus compared to the general population, but also that the age-specific cumulative prevalence of metabolic syndrome increases rapidly with age in this specific group [30,31]. Given that traditional glycemic markers may not adequately reflect glucose variability in specific circumstances, there is a pressing need for alternative glucose monitoring markers. While our study found that the GA level was not significantly correlated with FBG level, the samples may not have been from fasting serum and may not have reflected the average glucose levels of the subjects, as mentioned above. Furthermore, previous studies have shown that both HbA1c and GA correlate well with average glucose level on continuous glucose monitoring (CGM) rather than single-point glucose measurements [32]. Desouza et al. [33] reported GA to be a more sensitive marker of daily glycemic excursions than HbA1c in patients with diabetes. Unfortunately, due to the retrospective design of our study, we were unable to obtain data on average glucose levels or measure glucose variability via CGM. As FBG only represents a single-point serum glucose concentration, it may not reflect average glucose variability, which could explain the lack of correlation in the acute leukemia group. Nevertheless, we found a significant correlation between GA and HbA1c levels in both acute leukemia and T2DM groups, suggesting GA as a potential complementary glycemic marker in situations where short-term glucose monitoring and careful screening for hyperglycemia are required. Further studies with larger sample sizes and more diverse populations, such as pediatric patients with acute leukemia diagnosed with diabetes mellitus, are necessary to fully understand the role of GA in glucose monitoring and the discrepancies that may exist among patient groups.

At the beginning of the study, we hypothesized that a history of multiple transfusions in pediatric patients with acute leukemia may have affected their HbA1c level. Sugimoto et al. found that autologous blood donation can reduce the HbA1c level in patients with T2DM [34]. The authors speculated that RBC turnover was accelerated following blood donation and caused a reduction in the HbA1c level [18]. Thus, HbA1c may not be a satisfactory indicator of actual glycemic control in cases with a history of blood transfusions. However, although levels of Hb were significantly lower in the acute leukemia group than in the other 2 groups, Hb and HbA1c did not show a significant correlation (P=0.44), in contrast to previous studies where the authors found that HbA1c was inversely correlated with iron deficiency anemia [35]. While all 63 patients with acute leukemia had a history of RBC transfusion, they were not undergoing "active" chemotherapy at the time of their visit to our clinic. Rather, patients had completed chemotherapy and/or transplantation at least several months to years prior. Nevertheless, the previous transfusion history as well as complex treatment history of these patients may result in life-long sequelae in these patients. The extent to which HbA1c and GA influence glucose metabolism in patients with acute leukemia remains unclear, as does the impact of their previous transfusion and treatment histories. GA may serve as an alternative glycemic measure in situations where HbA1c measurements are unreliable in children treated for cancer. Conducting further research involving patients on active chemotherapy, who frequently present with anemia and pancytopenia, may enable a better understanding of glucose metabolism in this population.

In a nationwide study conducted in the US, GA level was inversely associated with BMI and cardiometabolic risk factors [23]. A previous study in a non-diabetic adult population also reported that BMI was negatively correlated with GA [36]. Other studies have reported controversial results regarding the association between obesity and GA. In our study, BMI z-score showed an inverse relationship with GA level in normal children. Obesity and albumin turnover may be associated, and chronic low-grade systemic inflammation may increase the catabolic albumin turnover rate in obese patients [24,38,39]. Kim et al. [18] suggested that the association between GA level and low adiposity meant that GA was a less reliable marker at normal levels of GA. Increased inflammation in obese children may provide another possible explanation for the negative correlation between BMI z-score and GA. Koga et al. [36] suggested that plasma high-sensitivity C-reactive protein (hs-CRP), an obesity-related inflammatory marker, could play a role. The authors hypothesized that obesity-related cytokines promote hs-CRP synthesis in the liver, resulting in higher hs-CRP levels in obese patients compared to the non-obese group, The authors showed that BMI as well as hs-CRP were negatively associated with GA in non-diabetic patients. However, we did not observe a relationship between GA level and BMI status in pediatric patients with acute leukemia nor in patients with diabetes. Although we were unable to assess hs-CRP levels in our study, other non-glycemic factors may explain the association between GA and BMI z-scores, and further studies are warranted.

This study had several limitations. First, it relied on data from retrospective chart review, meaning that some participants had missing data, and there may have been differences in sampling times due to the reliance on participants to follow the instructions to have samples drawn early in the morning. Some patients' caregivers may not have followed the instructions and the patient may not have fasted. Second, we did not evaluate other parameters associated with obesity and metabolic syndrome. A more detailed evaluation of adipose tissue and additional parameters, including free fatty acids, hs-CRP, and nutritional status, may be required to analyze the relationships between GA and other factors. Third, the sample size was small and varied greatly among groups. Despite having a statistical power of 0.9, our acute leukemia group comprised only a small percentage of leukemia patients who visited the endocrinology center and underwent GA, FBG, and HbA1c evaluation from a single blood sample. Furthermore, due to the limitation of a retrospective design, we were unable to obtain a uniform sample size across groups. To further examine the relationship between GA and HbA1c, a prospective larger population group study is necessary. Another limitation of our study was that obese children may have been overrepresented in the normal control group. Recent studies in Korea have found a high prevalence obesity in children (between 17.2% and 31.4%, depending on the study) [39,40]. The percentage of obesity in the normal control group in our study was 32%. This may be because the study was conducted at the pediatric endocrinology department, where parents who are concerned with their children's weight visit compared to the actual obesity prevalence. Finally, all patients in the acute leukemia group were stable, having completed treatment months to years prior to when serum measurements and anthropometric data were obtained. Further studies of GA levels during the active treatment phase and long-term monitoring would contribute to a better understanding of glucose metabolism in this specific cohort. A strength of our study is that it is the first to demonstrate a positive relationship between GA and HbA1c levels in the pediatric acute leukemia population.

In conclusion, our findings suggests that GA can serve as a complementary biomarker of glycemic status in pediatric patients with acute leukemia, as well as in patients with diabetes. However, given the limitations of our study, further research is needed to determine the usefulness of GA as a reliable glycemic index in pediatric patients with acute leukemia. Such investigations could shed more light on the potential clinical applications of GA in the management of glycemic control in these patients.