Ketoacidosis at childhood type 1 diabetes onset negatively affects residual beta-cell functions during the first year after diagnosis

Abstract

This study investigates whether pediatric patients presenting with diabetic ketoacidosis (DKA) at the time of type 1 diabetes (T1D) diagnosis experience increased insulin requirements, reduced rates of partial remission, and diminished C-peptide secretion over the first year post-diagnosis, relative to children without DKA. The study included 101 children with newly diagnosed T1D. DKA was classified based on the International Society of Pediatric Diabetes (ISPAD) guidelines, and participants were categorized by DKA occurrence and followed for one year post-diagnosis. HbA1c, insulin requirements, and stimulated C-peptide in a mixed meal tolerance test were assessed within 14 days of diagnosis and 6 and 12 months after diagnosis. Our results showed that DKA at T1D negatively affects residual beta-cell function during the year after diagnosis. The DKA cohort had higher insulin requirements across all study visits; however, the occurrence of partial remission did not differ significantly between study groups. Children with DKA at T1D diagnosis exhibit a significant reduction in stimulated C-peptide by 6 months post-diagnosis (AUC: DKA: 0.54 nmol/L vs. non-DKA: 0.68 nmol/L, p < 0.001; Peak: DKA: 0.66 nmol/L vs. non-DKA: 0.88 nmol/L, p < 0.001), a difference that persists at 12 months compared to non-DKA participants (AUC: DKA: 0.42 nmol/L vs. non-DKA: 0.52 nmol/L, p < 0.003; Peak: DKA: 0.51 nmol/L vs. non-DKA: 0.71 nmol/L, p < 0.001).

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-38533-4.

Introduction

Type 1 diabetes (T1D) is a prevalent chronic condition in children, and its incidence continues to rise globally^1,2. The disease is characterized by autoimmune-mediated destruction of pancreatic beta-cells, resulting in progressive insulin deficiency³. While beta-cell function declines significantly at clinical onset, earlier stages, particularly following seroconversion to islet antibody positivity, often show a slower rate of deterioration. Accelerated beta-cell loss typically occurs within 6 to 18 months preceding diagnosis^4,5. Primarily, the reduction of pancreatic beta-cell mass is driven by autoimmune destruction processes⁶. However, this impairment may not solely result from the disease process; other factors may affect residual beta-cell function before and after clinical diagnosis. The rate of progressive destruction of beta-cells differs among individuals, with some maintaining prolonged residual C-peptide secretion. There is an ongoing debate about the factors associated with this prolonged residual function, including genetics, disease duration, age at diagnosis, and others^7–9. Variations in beta-cell loss after diagnosis may also be partially attributable to ethnic differences. For instance, Hispanics exhibited higher fasting C-peptide levels at the beginning of the Trial Net New Onset Intervention Trials compared to non-Hispanic whites⁹.

Diabetic ketoacidosis (DKA) is a common presentation at the onset of T1D and has been linked to increased risk of mortality and prolonged hospital stays, as reported in some clinical studies^10,11. Recent findings suggest that newly diagnosed T1D pediatric patients presenting with DKA may show reduced beta-cell function¹², require more insulin¹⁰, persistently high HbA1c levels^13,14, increased body mass index¹⁴, and face neurocognitive risks¹⁵. These individuals may also be at greater risk for subsequent DKA episodes¹⁶, although findings remain inconclusive. A population-based study in Sweden reported no significant association between initial DKA presentation and recurrent episodes during the course of T1D. But, it was pointed out that parental education level and low household equivalized disposable income were recognized as socioeconomic background factors that potentially influence the risk of developing DKA, both at the time of T1D diagnosis and subsequent DKA episodes. Yet, this finding should be interpreted in the context of the overall low prevalence of DKA at the time of diagnosis, as well as during established T1D in Sweden¹⁷.

The prevalence of DKA at diagnosis persists at high levels in many countries^18,19. Although the general recognition of T1D symptoms, awareness among healthcare providers, and parents or legal guardians remains suboptimal, with notable gaps in knowledge regarding risk factors, clinical presentation, and management strategies for T1D²⁰. Preserving endogenous insulin secretion, as reflected by C-peptide levels, plays a critical role in long-term disease outcomes. Findings from the Diabetes Control and Complications Trial (DCCT) demonstrated that individuals with stimulated C-peptide concentrations exceeding 0.2 nmol/L experienced better glycemic control, fewer episodes of hypoglycemia, and reduced risk of chronic complications such as retinopathy and nephropathy^8,9.

Measuring C-peptide remains a key method for evaluating endogenous insulin secretion and beta-cell function. Although fasting and random C-peptide measurements are available, the mixed-meal tolerance test (MMTT), which evaluates stimulated C-peptide levels, has been validated and it is considered the primary endpoint in clinical trials aimed at measuring insulin secretion in individuals with T1D^21,22.

This study investigated whether the presence of DKA at the onset of T1D in pediatric patients is associated with increased insulin requirements, reduced incidence of partial remission, and diminished C-peptide levels during the first year following diagnosis, in comparison to children without DKA at diagnosis.

Methods

The study was conducted prospectively, involving children with DKA at T1D onset and children without DKA at diagnosis. For the study purposes, we enrolled 101 participants with newly diagnosed T1D at the age of 7 to 18 years in the region of Greater Poland (Poland) in the years 2012 to 2014 who underwent MMTT within 14 days of diagnosis and then at 6 and 12 months after diagnosis.

The diagnosis of T1D was made based on the International Society for Pediatric and Adolescent Diabetes (ISPAD) guidelines²³. Hyperglycemia (blood glucose > 200 mg/dl), metabolic acidosis (pH < 7.3 and/or HCO3 < 18 mmol/L), and ketonemia (blood ß-hydroxybutyrate ≥ 3 mmol/L) or ketonuria were considered as DKA²³. According to the results, children were divided into the non-DKA and DKA groups.

Relevant eligibility criteria included: newly diagnosed T1D, age 7–18 years, and consent to participate in the study. Exclusion criteria included: non-T1D diagnosis, current treatment with drugs known to interfere with glucose metabolism, acute viral or bacterial infections at study visit, age less than 7 years or more than 18 years, or lack of consent to participate in the study.

Parameters obtained on enrolment and then 6 and 12 months after diagnosis included glycemia, HbA1c, insulin, and C-peptide plasma level. C-peptide concentration was analyzed using radioimmunoassay (C-PEP II-RIA-CT, DIA source Immunoassay, Belgium), with the normal range: 0.59–1.54 pmol/ml, while insulin level was assessed by Chemiluminescent Microparticle Immunoassay (Insulin ARCHITECT System, Abbott Laboratories, USA), with the fasting normal range < 15 µU/ml. HbA1c was measured using the NGSP-certified method (HbA1c ARCHI-TECT System, Abbott Laboratories, USA) with normal values < 5.7%.

Partial remission (PR) was defined as an insulin requirement of ≤ 0.5 units/kg/24h^24,25 and assessed at Visit 1, Visit 2, and Visit 3. This definition is most applicable under conditions where the treatment policy remains consistent. Then, to correct this issue, we also termed IDAA1c as insulin dose-adjusted HbA1c (IDAA1c) ≤ 9 ^25–27. IDAA1c was calculated as IDAA1c = HbA1c(%) + [4× insulin dose (U/kg/d)] at Visit 2 and Visit 1.

MMTT was administered after an overnight fast, with participants’ baseline glucose levels ranging from 70 to 200 mg/dl. Long-acting insulin and basal infusion rates for those using insulin pumps were continued without modification. Rapid-acting insulin boluses were permitted up to two hours prior to the test. Each participant consumed 6 mL/kg of a standardized nutritional solution (maximum 360 mL; Nestlé Boost) within a 10-minute window. Blood samples for C-peptide analysis were obtained at pre-meal baseline and subsequently at 30, 60, 90, and 120 min post-ingestion. The primary endpoint was the area under the curve (AUC) for serum C-peptide over the 2-hour testing period²¹.

Legal guardians have provided written informed consent, and the children’s assent when applicable. The study was approved by the Ethics Committee of the Poznan University of Medical Science (number of permission: 960/12), and all procedures were performed in accordance with the Declaration of Helsinki. Figure 1 shows a flow chart outlining the patient’s enrollment and study design.

Fig. 1.

Flow chart outlining the patient’s enrollment and study design. T1D, type 1 diabetes; MMTT, mixed meal tolerance test; IDAA1C, dose-adjusted HbA1c; HbA1c, glycated haemoglobin.

Statistical analysis

A two-sided p-value below 0.05 was interpreted as indicating statistical significance. Baseline characteristics and outcomes were summarized using means and standard deviations (SD) for continuous variables, and frequencies with percentages for categorical variables.

Differences in baseline characteristics between DKA and non-DKA groups were assessed using Welch’s two-sample t-test for continuous variables to account for unequal variances, and Pearson’s Chi-squared test or Fisher’s exact test for categorical variables, depending on expected cell counts.

Longitudinal outcomes were analyzed using generalized estimating equation (GEE) models with an exchangeable correlation structure to account for repeated measures within individuals over time. C-peptide levels were log-transformed before modeling, with results back-transformed to the original scale (nmol/L) for interpretability. The AUC for C-peptide concentrations during the MMTT was estimated using the trapezoidal rule, with the AUC divided by 120 min to report the mean C-peptide level over the 2-hour period. GEE models were adjusted for age at diagnosis and BMI as potential confounders, selected based on clinical relevance and the statistical significance in univariate analysis. Estimated marginal means (EMMs) were assessed, with 95% confidence intervals (CIs) calculated on the back-transformed scale.

Pairwise contrasts between DKA and non-DKA groups at each time point (for C-peptide levels) and changes from baseline (Time 0) were estimated with Sidak adjustment for multiple comparisons to control the family-wise error rate. Differences in C-peptide levels and changes from baseline were reported as absolute differences (nmol/L), with 95% CIs approximated on the back-transformed scale.

PR was analyzed as a binary outcome using Pearson’s Chi-squared test at each time point. Sensitivity analyses were performed by re-running GEE models without adjustments for age and BMI. Analyses were conducted using the R Statistical language (version 4.3.3; R Core Team, 2024).

Results

The study included 101 children (51 females) at T1D onset with a mean age at baseline of 10.1 years. DKA was present in 50.5% of study participants. Insulin and C-peptide levels at admission were markedly lower in participants with DKA (1.6 µU/mL and 0.3 nmol/L, respectively; p < 0.001) compared to those without DKA (2.9 µU/mL and 0.5 nmol/L, respectively; p < 0.001). In addition, children with DKA had significantly elevated blood ketone levels (5.1 mmol/L) compared to those from the non-DKA group (2.9 mmol/L, p < 0.001). A detailed characteristic of the studied population is shown in Table 1.

Table 1.

Clinical and laboratory characteristics at diagnosis of type 1 diabetes in the study cohort, stratified by diabetic ketoacidosis.

Characteristic	All Patients (n = 101)	DKA (n = 51)	Non-DKA (n = 50)	p-value
Age (yrs)	10.1 (2.3)	10.2 (2.2)	10.0 (2.4)	0.673
Female, n (%)	51 (50.5%)	25 (49.0%)	26 (52.0%)	0.765
BMI (kg/m²)	16.3 (3.2)	15.6 (2.8)	17.1 (3.4)	0.024
BMI z-score	-0.62 (1.8)	-0.96 (1.59)	-0.28 (0.73)	< 0.001
Glycemia (mg/dl)	399 (126)	402 (96)	396 (151)	0.843
NGSP HbA1c (mmol/mol); [%]	99 (7.0) [11.4 (2.1)]	108 (6.0) [11.7 (2.0)]	97 (7.0) [11.0 (2.1)]	0.135
Insulin (µU/mL)	2.3 (1.5)	1.6 (0.5)	2.9 (1.8)	< 0.001
C-peptide (nmol/L)	0.4 (0.2)	0.3 (0.1)	0.5 (0.2)	< 0.001
Blood ketones (mmol/L)	4.0 (1.7)	5.1 (1.0)	2.9 (1.6)	< 0.001
DKA status, n (%)				< 0.001
Mild	14 (13.8%)	14 (27.4%)	0 (0.0%)
Moderate	27 (26.7%)	27 (52.9%)	0 (0.0%)
Severe	10 (9.9%)	10 (19.6%)	0 (0.0%)

Notes: Data are presented as mean (SD) for continuous variables or n (%) for categorical variables. p-values were calculated using Welch’s Two Sample t-test for continuous variables, Pearson’s Chi-squared test for sex, and Fisher’s exact test for DKA status.Abbreviations: BMI, body mass index; HbA1c, glycated hemoglobin; DKA, diabetic ketoacidosis.

At initial visit, HbA1c levels were comparable between groups (DKA: 108 mmol/mol [11.7%], non-DKA: 97 mmol/mol [11.0%], p = 0.135). However, insulin doses were significantly higher in the DKA group (0.51 U/kg/day) versus non-DKA (0.40 U/kg/day, p = 0.005). PR (≤ 0.5 U/kg/day) was less frequent in children with DKA at disease onset (54.9%) compared to those without DKA (74.0%, p = 0.045). At 6 months post-diagnosis, HbA1c improved in both groups with no significant difference (DKA: 43 mmol/mol [6.1%]; non-DKA: 44 mmol/mol [6.1%], p = 0.777), but insulin doses remained higher in DKA patients (0.47 vs. 0.37 U/kg/day, p = 0.009). PR was again less frequent in the DKA group (62.75%) than in the non-DKA group (88%, p < 0.003), though based on IDAA1c, the difference was not significant (p = 0.238). At 12 months, HbA1c slightly increased, with DKA patients showing a lower mean (48 mmol/mol [6.5%] vs. 52 mmol/mol [6.9%], p = 0.050), and insulin doses tended to be higher in the DKA group (0.54 vs. 0.45 U/kg/day, p = 0.052). PR remained less common in DKA participants (47.1%) but was not statistically significant (p = 0.132). At the same time, IDAA1c showed no significant difference between groups (p = 0.160). Table 2 summarizes metabolic and clinical outcomes across visits.

Table 2.

Metabolic and clinical outcomes at visit 1, visit 2, and visit 3 following T1D diagnosis, stratified by diabetic ketoacidosis.

Visit number	Characteristic	DKA (n = 51)	Non-DKA (n = 50)	p-value²
Visit 1	NGSP HbA1c (mmol/mol); [%]*	108 (6.0) [11.7 (2.0)]	97 (7.0) [11.0 (2.1)]	0.135
(14 days post-diagnosis)	TDDI (U/day)	17.4 (9.4)	15.0 (9.1)	0.191
	Insulin dose (U/kg/day)	0.51 (0.2)	0.40 (0.2)	0.005
	PR, n (%)	28 (54.9%)	37 (74.0%)	0.045
Visit 2	NGSP HbA1c (mmol/mol); [%]	43 (2.3) [6.1 (0.6)]	44 (2.5) [6.2 (0.7)]	0.777
(6 months post-diagnosis)	TDDI (U/day)	17.3 (10.6)	13.9 (6.9)	0.061
	Insulin dose (U/kg/day)	0.47 (0.24)	0.37 (0.15)	0.009
	PR, n (%)	32 (62.7%)	44 (88.0%)	0.003
	PR-IDAA1c, n (%)	38 (74%)	42 (84%)	0.238
	IDAA1c	8.05 (1.35)	7.67 (1.08)	0.119
Visit 3	NGSP HbA1c (mmol/mol); [%]	48 (2.5) 6.5 (0.8)	52 (3.0) 6.9 (1.1)	0.050
(12 months post-diagnosis)	TDDI (U/day)	20.9 (10.6)	18.3 (10.5)	0.215
	Insulin dose (U/kg/day)	0.54 (0.22)	0.45 (0.20)	0.052
	PR, n (%)	24 (47.1%)	31 (62.0%)	0.132
	PR-IDAA1c, n (%)	31 (69.7%)	37 (74.0%)	0.160
	IDAA1c	8.61 (1.37)	8.66 (1.64)	0.862

Notes: Data are presented as mean (SD) for continuous variables or n (%) for categorical variables. P-values were calculated using Welch’s Two Sample t-test for continuous variables and Pearson’s Chi-squared test for categorical variables (PR). Abbreviations: HbA1c, glycated hemoglobin; TDDI, total daily dose of insulin; ID, insulin dose; PR, partial remission; IDAA1c,insulin dose-adjusted HbA1c; PR-IDAA1c, partial remission calculated based on IDAA1c; DKA, diabetic ketoacidosis.* Measurement taken at the time of diagnosis.

At 2 weeks post-diagnosis, no significant difference in stimulated C-peptide levels was observed between groups; DKA participants had slightly lower AUC C-peptide (0.60 vs. 0.64 nmol/L, p = 0.257) and peak C-peptide (0.77 vs. 0.82 nmol/L, p = 0.344). By 6 months, significant reductions appeared in the DKA group, with AUC C-peptide at 0.54 nmol/L vs. 0.68 nmol/L (p < 0.001) and peak C-peptide at 0.66 vs. 0.88 nmol/L (p < 0.001). At 12 months, residual beta-cell function remained lower in the DKA group, with AUC C-peptide at 0.42 vs. 0.52 nmol/L (p = 0.003) and peak C-peptide at 0.51 vs. 0.71 nmol/L (p = 0.001). C-peptide levels during the first year after T1D diagnosis between children with DKA and those without are shown in Table 3.

Table 3.

Estimated marginal means and contrasts for stimulated C-peptide (AUC and Peak, nmol/L) in children with type 1 diabetes by DKA status and time post-diagnosis.

Visit	Group	AUC C-peptide (nmol/L)					Peak C-peptide (nmol/L)
		EMM	95% CI	Difference (DKA – non-DKA)	95% CI for Difference	p-value	EMM	95% CI	Difference (DKA – non-DKA)	95% CI for Difference	p-value
2-Weeks	DKA	0.6	0.56, 0.65	-0.04	-0.11, 0.03	0.257	0.77	0.70, 0.85	-0.05	-0.15, 0.05	0.344
	non-DKA	0.64	0.59, 0.70				0.82	0.75, 0.91
6-Month	DKA	0.54	0.50, 0.59	-0.13	-0.20, -0.07	< 0.001	0.66	0.61, 0.72	-0.22	-0.30, -0.14	< 0.001
	non-DKA	0.68	0.63, 0.73				0.88	0.80, 0.96
12-Month	DKA	0.42	0.37, 0.47	-0.1	-0.17, -0.04	0.003	0.51	0.43, 0.59	-0.2	-0.31, -0.09	0.001
	non-DKA	0.52	0.48, 0.57				0.71	0.63, 0.81

Notes: AUC and Peak C-peptide values are estimated marginal means (EMMs) derived from separate Generalized Estimating Equation (GEE) models with log-transformed outcomes, adjusted for age at diagnosis and BMI. Differences represent the absolute difference (nmol/L) between DKA and non-DKA groups at each visit, back-transformed from the log scale. The 95% confidence intervals (CIs) for the differences are calculated on the back-transformed scale. P-values are derived from pairwise contrasts using the GEE models.

Children who experienced DKA at T1D onset had lower AUC C-peptide values than children without DKA throughout the first year post-diagnosis. Differences in C-peptide AUC during MMTT between both study groups during the first year after diagnosis of T1D are shown in Fig. 2. Across Visits 1 to 3, children with DKA exhibited lower C-peptide levels and smaller increments from baseline compared to the non-DKA group, with differences becoming more pronounced over time. By Visit 3, the non-DKA group maintained significantly higher C-peptide levels and greater responses during the MMTT (Fig. 3 and Supplemental Table S1-S3).

Fig. 2.

Serum C-peptide area under the curve (AUC) during MMTT in the DKA group and the non-DKA group during the first year follow-up after diagnosis of diabetes (Visit 1, 2, 3). MMTT, mixed-meal tolerance test; DKA, diabetic ketoacidosis.

Fig. 3.

C-peptide levels (A) and changes (B) in C-peptide levels compared to baseline (0 min) across groups at Visit 1 (a), Visit 2 (b), Visit 3 (c). MMTT. mixed-meal tolerance test; DKA, diabetic ketoacidosis. The asterisk refers to a significant difference (i.e., p < 0.05). Abbreviations: DKA, diabetes ketoacidosis.

Discussion

Although screening initiatives for T1D in pediatric populations have become more accessible in recent years²⁸, delayed diagnosis remains a pressing public health concern in many regions, contributing to persistently high rates of DKA^29,30. A more severe clinical presentation at onset is often associated with diminished pancreatic beta-cell function. In our cohort, children who presented with DKA at initial diagnosis exhibited markedly lower baseline serum insulin and C-peptide levels, as well as elevated blood ketone concentrations, compared to those without DKA. Interestingly, baseline glycemia and HbA1c levels showed no significant difference between groups, indicating comparable hyperglycemia severity at diagnosis. Nevertheless, prolonged hyperglycemia, commonly called glucotoxicity, may impair beta-cell function through metabolic stress, potentially accelerating functional decline in susceptible individuals. This phenomenon is known to compromise the insulin secretory capacity of pancreatic beta-cells through several mechanisms, including oxidative stress, endoplasmic reticulum stress, and impaired insulin gene expression^31–33. Nevertheless, not only does chronic hyperglycemia alone contribute to the depletion of functional pancreatic beta-cells, but lipotoxicity and systemic inflammation associated with DKA have also been shown to diminish residual beta-cell function further^32–34.

Children with DKA at T1D diagnosis tended to have higher insulin requirements in the first few months after diagnosis. In our study, participants presenting with DKA consistently required higher insulin doses during follow-up visits. The differences were statistically significant at the first two visits, and a similar trend was observed at the third visit, approaching significance. The persistence of higher insulin needs in the DKA group aligns with their lower baseline and stimulated C-peptide levels, indicating ongoing beta-cell impairment. These findings are supported by an international study involving nine countries, which analyzed 9,269 children aged 0.5–15.9 years with newly diagnosed T1D followed for two years. That study demonstrated that children with severe DKA required more insulin than those without DKA at both 1 year (0.72 IU/kg/day vs. 0.62 IU/kg/day, P < 0.001) and 2 years post-diagnosis (0.76 IU/kg/day vs. 0.71 IU/kg/day, P = 0.003). No significant difference in insulin requirements was observed between the non-DKA and non-severe DKA groups¹⁴.

Notably, our study found that HbA1c levels remained similar between children with and without DKA across all study visits. This suggests that patients presenting with DKA at the time of T1D diagnosis may require higher insulin doses to achieve comparable glycemic control, likely reflecting more pronounced beta-cell dysfunction. Interestingly, at the 12-month follow-up, the non-DKA group exhibited a slightly higher mean HbA1c than the DKA group. This pattern could point to improved long-term glycemic management in the DKA group, potentially driven by more intensive treatment following their initial severe presentation. However, existing literature presents conflicting data regarding the impact of DKA at diagnosis on long-term HbA1c outcomes. Our results are consistent with a population-based cohort study from Western Australia, which explored the relationship between moderate-to-severe DKA at onset and glycemic control over a span of two decades. The study followed children under 16 years of age diagnosed with T1D, with follow-up extending up to 14 years. Initially, those with moderate-to-severe DKA had higher HbA1c levels. Between years 2 and 6 post-diagnosis, HbA1c values were similar across groups. From year 7 onward, however, the DKA group consistently showed higher HbA1c levels, with statistically significant differences emerging between 8 and 12 years (p < 0.05)³⁵.

However, some studies report that children with DKA at diagnosis have higher HbA1c levels in the first years after diagnosis^14,36,37. This discrepancy raises essential questions about the influence of additional factors on long-term glycemic control. Recent evidence suggests that automated insulin delivery (AID) systems may reduce long-term HbA1c disparities associated with DKA at diagnosis. A single-center study found that early AID use within a month of T1D diagnosis helped reduce the glycemic impact of DKA³⁸. Similarly, Dovc et al. showed that initiating therapy with AID mitigates the relation between DKA at diagnosis and higher HbA1c over time¹⁴.

Regarding PR occurrence in the two study groups, its rate depended on the definition implemented. We reported that PR, defined as an insulin dose ≤ 0.5 U/kg/day, was less frequent in the DKA group across visits compared to the non-DKA group. However, after integrating insulin requirements with HbA1c using IDAA1c, there was no difference in PR occurrence between study groups. Since IDAA1c is considered a more valuable measure for assessing PR in individuals with T1D, the present study concluded that DKA at disease onset does not adversely affect PR during the first year after diagnosis in children with newly diagnosed T1D. These findings are discordant with earlier studies, which reported that pediatric patients with a more severe initial presentation of T1D had a lower probability of attaining PR than those who did not present with DKA at diagnosis^10,39,40. However, it is essential to note that we did not observe a substantial intergroup difference in IDAA1c rates, in contrast to the significant differences observed in AUC C-peptide. This discrepancy can be explained by the fact that the formula for IDAA1c was derived using a C-peptide threshold of 300 pmol/L rather than the 200 pmol/L validated by the DCCT. Consequently, IDAA1c may underestimate PR occurrence in younger children who typically exhibit lower serum C-peptide concentrations due to a smaller β-cell mass^26,41.

Importantly, our data revealed that children with DKA at T1D diagnosis have significantly reduced residual beta-cell functions, particularly 6 and 12 months post-diagnosis. This pattern might reflect DKA’s acute and sustained impact on beta-cell function, and is consistent with the lower baseline C-peptide levels observed in the DKA group. Throughout Visits 1 to 3, children diagnosed with DKA showed lower C-peptide levels and smaller increases from baseline compared to those without DKA, with differences becoming more pronounced over time. By Visit 3, the non-DKA group continued to have significantly higher C-peptide levels and greater responses during the MMTT. Our findings may suggest that DKA at childhood T1D onset negatively affects residual beta-cell function in the first year following diagnosis. The presented results align with other studies indicating that DKA at disease onset is associated with a higher risk of rapid C-peptide decline^42,43. Tables 1S to 3S in Supplementary Information present the EMMs, changes from baseline (Time 0), and contrasts for C-peptide levels in children with T1D, stratified by DKA status at diagnosis, across three visits.

Our study results suggest early clinical diagnosis of T1D may help preserve greater residual beta-cell function. This underscores the importance of increasing awareness of early symptoms so that caregivers and healthcare professionals can promptly recognize signs of hyperglycemia and facilitate timely intervention. Expanding access to T1D screening programs could significantly reduce the number of DKA cases in children and adolescents during initial disease presentation.

The Environmental Determinants of Diabetes in the Young (TEDDY) study found that children diagnosed with T1D during islet autoantibody screening had a greater residual beta-cell function than those diagnosed from the general population. TEDDY participants, identified at birth as high-risk based on human leukocyte antigen (HLA) profile, were followed closely until T1D onset, and after diagnosis, were compared to age-matched children diagnosed with T1D from a general population. The main evaluation, including HbA1c and MMTT, was conducted within the first month after the diabetes diagnosis. These assessments were then repeated at 3, 6, and 12 months, and continued every six months thereafter. This study showed 58% of TEDDY participants had no symptoms at diagnosis, and none experienced DKA, whereas 98% of controls had symptoms and 14% had DKA (P < 0.001 and P = 0.03). At diagnosis, TEDDY children had lower HbA1c (6.8% vs. 10.5%, P < 0.0001). They also exhibited higher C-peptide levels during the first year after diagnosis, indicating better preserved beta-cell function. In the TEDDY cohort, total insulin dose and IDAA1c were lower one year after diagnosis than in control children. This study showed that earlier diagnosis of T1D is strongly related to greater residual beta-cell function after one year following diagnosis⁴⁴.

The strength of the present study lies in using the gold standard MMTT to assess longitudinal beta-cell function in children with newly diagnosed T1D at multiple time points, including 14 days after diagnosis and then at 6 and 12 months post-diagnosis. The longitudinal follow-up of participants from diagnosis allowed for a comprehensive evaluation of the progression of beta-cell decline over time. This approach provides valuable insights into the early impact of DKA on pancreatic function in pediatric patients. However, some limitations of the current study need to be acknowledged. This study involved a relatively small sample size, primarily due to the frequent hospitalizations required to perform the MMTT. These frequent visits caused inconveniences for patients and their legal guardians. Consequently, the number of participants was limited to the total number of eligible or available patients in our Diabetes Clinic, making it a practical sample for comprehensive analysis. Furthermore, the IDAA1c definition was not applied within 14 days post-diagnosis because the brevity of this interval limited the feasibility of accurate assessment, and high HbA1c levels during this early period could disproportionately influence the IDAA1c values. Finally, there was no follow-up visit three months after the diagnosis. Yet, it might be a more suitable time for assessing PR according to the IDAA1c definition. Consequently, we also defined PR based on an insulin dose of ≤ 0.5 U/kg/day at all study visits.

In conclusion, the present study provides valuable data on the possible adverse effect of DKA at T1D diagnosis in children on residual beta-cell function throughout one year after diagnosis. Our work has highlighted that DKA at diagnosis was significantly associated with lower baseline insulin and C-peptide levels and higher blood ketone values. The DKA cohort required higher insulin doses; however, the occurrence of PR did not differ significantly between the study groups. Finally, this study demonstrated that children presenting with DKA at T1D diagnosis exhibit a significant reduction in stimulated C-peptide levels at 6 months post-diagnosis, a difference that persists at 12 months.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

EN contributed to study design and methodology, concept, data collection, investigation, analyses, and writing, reviewing, editing, and approving the manuscript. EN prepared all tables and figures. PW contributed to the design and methodology, interpretation of data, and editing and approving the manuscript. AK contributed to data acquisition, interpretation, reviewing, editing, and approving the manuscript. EN is the guarantor of this work and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

All data used in the analysis are available from the corresponding author upon request.

Declarations

Competing interests

The authors declare no competing interests.