Published online Aug 15, 2026. doi: 10.4239/wjd.122101
Revised: May 13, 2026
Accepted: July 3, 2026
Published online: August 15, 2026
Processing time: 117 Days and 20.6 Hours
Adults presenting with ketosis or diabetic ketoacidosis at diabetes onset pose a diagnostic challenge because ketosis-prone (KP) type 2 diabetes mellitus (T2DM) and latent autoimmune diabetes in adults (LADAs) share overlapping clinical features but require fundamentally different long-term management strategies. In many clinical settings, limited access to islet autoantibody testing further com
To characterize the clinical, metabolic, and endocrine features of ketosis-onset diabetes in adults and to identify potential non-antibody biomarkers to differentiate LADA from KP-T2DM and distinguish ketosis-onset from nonketotic diabetes.
A total of 294 newly diagnosed adult patients were classified into LADA, KP-T2DM, and nonketotic T2DM groups. Clinical and metabolic characteristics were compared. Multivariable logistic regression was performed to identify factors associated with ketosis onset and LADA classification, and receiver operating characteristic analysis was used to evaluate the discriminatory per
LADA accounted for 27.3% of ketosis-onset cases. Compared with patients with KP-T2DM, those with LADA exhibited significantly impaired β-cell function and lower insulin resistance. Multivariable analysis identified postprandial C-peptide as the strongest independent discriminator between LADA and KP-T2DM. Receiver operating characteristic analysis demonstrated that 2-hour C-peptide showed good discriminatory performance for differentiating LADA from KP-T2DM (area under the curve = 0.852). Other metabolic parameters, including high-density lipoprotein cholesterol, gamma-glutamyl transferase, alkaline phosphatase, free triiodothyronine, fasting plasma glucose, and homeostasis model assessment of insulin resistance, showed additional but more limited discriminatory value.
Among adults presenting with ketosis at diabetes onset, LADA represents a substantial proportion of cases and should be considered in the differential diagnosis. Readily available non-antibody biomarkers, particularly 2-hour postprandial C-peptide, demonstrated promising discriminatory performance for differentiating LADA from KP-T2DM, whereas other metabolic markers provided supplementary information. These findings may support the clinical evaluation of atypical diabetes in settings where autoantibody testing is unavailable; however, independent validation is required before routine clinical application.
Core Tip: More than one-quarter of adults presenting with ketosis at diabetes onset in our cohort were diagnosed with latent autoimmune diabetes in adults, highlighting the importance of considering autoimmune diabetes in this clinical setting. We identified a panel of readily available non-antibody biomarkers, including 2-hour C-peptide, high-density lipoprotein cholesterol, gamma-glutamyl transferase, homeostasis model assessment of insulin resistance, alkaline phosphatase, free triiodothyronine, and fasting plasma glucose, that demonstrated discriminatory value for differentiating latent autoimmune diabetes in adults from ketosis-prone type 2 diabetes mellitus and for distinguishing ketosis-onset from nonketotic diabetes within the study cohort. These findings may provide supportive information for the clinical evaluation of atypical diabetes when islet autoantibody testing is unavailable; however, independent validation is required before routine clinical application.
- Citation: Pei SQ, Ren JL, Zhang L, Gao YX, Yang X, Liu X, Li SN, Liu P, Xu XY. Non-antibody biomarkers for differentiating ketosis-onset diabetes: A retrospective study. World J Diabetes 2026; 17(8): 122101
- URL: https://www.wjgnet.com/1948-9358/full/v17/i8/122101.htm
- DOI: https://dx.doi.org/10.4239/wjd.122101
The landscape of diabetes mellitus is increasingly recognized not as a simple dichotomy of type 1 diabetes mellitus and type 2 diabetes mellitus (T2DM), but as a complex spectrum of metabolic disorders with overlapping clinical phenotypes[1,2]. This heterogeneity presents a substantial challenge in clinical practice, particularly when an adult patient presents with an acute metabolic crisis of ketosis or diabetic ketoacidosis (DKA)[3,4]. This diagnostic ambiguity primarily involves two distinct, yet often conflated entities: Latent autoimmune diabetes in adults (LADAs) and ketosis-prone (KP) T2DM[5,6].
LADA, often termed “type 1.5 diabetes”, is defined by adult onset[7], the presence of circulating islet autoantibodies, and a transient period of insulin independence. It is the most common form of adult-onset autoimmune diabetes in China, with a prevalence of about 6% among newly diagnosed T2DM patients, translating to more than 10 million affected individuals[8]. Misdiagnosis of LADA as T2DM, which occurs in 2%-12% of cases, can delay appropriate insulin therapy and thereby accelerate β-cell failure[9]. By contrast, KP-T2DM is an antibody-negative subtype of KP diabetes, typically presenting with severe hyperglycemia and ketosis at onset[10,11]. These patients often exhibit marked but reversible β-cell dysfunction during the acute phase, and a large proportion regain sufficient endogenous insulin secretion to allow insulin withdrawal after stabilization[12]. Epidemiological studies suggest that KP-T2DM accounts for up to one-third of newly diagnosed T2DM patients presenting with ketosis[13], whereas LADA constitutes roughly one-quarter of adults with ketosis at onset[14].
Despite their clinical importance, antibody testing (e.g., glutamic acid decarboxylase antibody, islet cell antibody, insulinoma-associated-2 antibody, zinc transporter 8 antibody), especially in primary care and resource-limited settings, is often unavailable or delayed, limiting its utility for early classification[15]. Therefore, there is an urgent need to identify pragmatic, non-antibody biomarkers to facilitate early classification of ketotic diabetes. Previous studies have suggested that C-peptide indices, lipid parameters, liver enzymes, and hormonal markers may provide diagnostic insights. Previous studies have suggested that liver enzymes such as gamma-glutamyl transferase (GGT) and alkaline phosphatase (ALP) may be associated with metabolic dysfunction.
However, most previous studies have focused on autoantibody status, C-peptide, or conventional metabolic pa
A retrospective study was conducted at a single tertiary care center (General Hospital of Ningxia Medical University), over 6 years (July 2017 to July 2023). Medical records of all newly diagnosed adult patients with diabetes who were hospitalized with ketosis or ketoacidosis at disease onset were reviewed, along with a comparative group of patients with new-onset T2DM without ketosis during the same timeframe. The study was approved by the Ethics Committee of General Hospital of Ningxia Medical University (Approval No. KYLL-2025-2436), with a waiver of informed consent granted due to its retrospective design. As this was a hospital-based study, potential selection bias may arise from the inclusion of a hospitalized cohort.
A total of 294 patients met the inclusion criteria and were stratified into three groups. The LADA group (n = 55) included patients with LADAs, defined as having an age of ≥ 18 years, diabetes duration of < 6 months, presentation with ketosis (or ketoacidosis) at onset, and at least one positive islet autoantibody (glutamic acid decarboxylase 65 kDa isoform, islet cell antibody, insulin autoantibody, or insulinoma-associated antigen-2). Notably, LADA patients were those not requiring permanent insulin therapy within the first 6 months after diagnosis, which distinguished them from classic acute-onset type 1 diabetes. KP-T2DM group (n = 146): Patients with KP-T2DM, defined as age ≥ 18 years, new-onset diabetes (< 6 months) with ketosis or ketoacidosis at presentation (positive blood ketones with blood glucose ≥ 13.9 mmol/L, with or without low arterial pH/bicarbonate), negative for islet autoantibodies, and no evidence of other specific types of diabetes. These patients maintained glycemic control without continuous insulin after the acute episode, confirming a type 2 phenotype. Conventional T2DM group (n = 93): Patients with new onset T2DM without ketosis, meeting standard diabetes diagnostic criteria but without ketosis or DKA at onset, and negative for autoantibodies. They were matched by diagnosis period for comparison. Patients were selected from the same time period with broadly comparable baseline characteristics. However, no strict matching for age, sex, or body mass index (BMI) was performed. The patient screening and selection process is summarized in Figure 1.
Inclusion criteria: Inclusion criteria were as follows: (1) Newly diagnosed diabetes (according to American Diabetes Association criteria) with < 6 months of symptoms; (2) Age ≥ 18 years; (3) No prior treatment for diabetes before admission; and (4) For ketosis-onset groups, documented ketonemia or ketonuria at presentation (positive plasma β-hydroxybutyrate; DKA defined by hyperglycemia with arterial pH < 7.3 or bicarbonate < 18 mmol/L, with mild cases of ketosis without acidosis also included). For LADA vs KP classification, the presence or absence of islet autoantibodies, as defined above, was required.
Exclusion criteria: Patients with secondary causes of diabetes (pancreatic disease, endocrinopathies, steroid-induced diabetes, etc.), monogenic diabetes, gestational diabetes, or any severe comorbid illness that could confound metabolic parameters (e.g., liver or kidney failure unrelated to diabetes) were excluded. Individuals with incomplete data (e.g., missing key laboratory results) were also excluded.
We collected comprehensive clinical data for each patient from hospital records using a standardized case report form.
Age at diagnosis, sex, body weight, BMI, and blood pressure on admission were obtained from hospital records at the time of first diagnosis and used for subsequent analyses.
Fasting plasma glucose (FPG) and routine blood glucose measurements during hospitalization (pre-meal, 2 hours postprandial, and bedtime capillary glucose) were recorded. Glycated hemoglobin was measured by high-performance liquid chromatography to assess chronic glycemic control.
Fasting C-peptide (FCP) and 2 hours postprandial C-peptide were measured by chemiluminescent immunoassay to evaluate endogenous insulin secretion. These values were used to calculate indices of β-cell function, such as homeostasis model assessment (HOMA)-β (HOMA-islet, based on C-peptide) and HOMA-insulin resistance (IR). Modified formulas for C-peptide-based HOMA indices were employed: HOMA-IR (C-peptide) = 1.5 + (FPG × FCP)/2800 and HOMA-β (islet, C-peptide) = 0.27 × FCP/(FPG - 3.5).
Total cholesterol (TC), triglycerides, high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cho
Alanine aminotransferase, aspartate aminotransferase, and ALP were assessed by a colorimetric rate method. Serum uric acid and creatinine were also recorded.
Autoantibody testing for glutamic acid decarboxylase antibody islet cell antibody, insulin autoantibody, and insulinoma-associated-2 antibody was performed in all patients with ketosis-onset diabetes to differentiate LADA from KP-T2DM. Zinc transporter 8 antibody was additionally measured in a subset of cases for research purposes. Thyroid function (thyroid-stimulating hormone, free thyroxine, free triiodothyronine) and adrenal indicators (morning cortisol and adrenocorticotropic hormone) were also assessed to evaluate endocrine differences between groups. Serum 25-hydroxy vitamin D levels were obtained because of the emerging interest in its role in diabetes and were measured by electrochemiluminescence immunoassay.
During hospitalization, all patients received standardized diabetes management, including routine capillary glucose monitoring before and 2 hours after each meal and at bedtime. These data were used to evaluate overall glycemic control during the acute phase rather than for formal variability analysis.
We screened for microvascular complications despite the short duration of diabetes. Fundoscopic examinations were performed to detect diabetic retinopathy, which was graded according to the International Clinical Diabetic Retinopathy Scale. Peripheral neuropathy was assessed based on symptoms and, when available, nerve conduction studies. Diabetic nephropathy was evaluated using albumin-to-creatinine ratio or estimated glomerular filtration rate. We also assessed macrovascular disease by evaluating carotid artery plaques on ultrasound (plaque defined as intima-media thickness ≥ 1.5 mm) and by calculating each patient’s 10-year atherosclerotic cardiovascular disease risk score using risk factors according to guidelines. Presence of hypertension was defined according to Joint National Committee criteria (≥ 140/90 mmHg on three separate occasions), and non-alcoholic fatty liver disease was diagnosed using ultrasound criteria. These comorbidities were recorded to compare their prevalence between groups.
C-peptide levels were measured before initiation of insulin therapy whenever possible to minimize the impact of exogenous insulin. All laboratory assays were performed in a hospital-certified clinical laboratory with standard quality control procedures. For C-peptide levels in patients who received insulin during acute treatment, values were measured either before insulin initiation or calculated using corrected formulas to estimate β-cell function under hyperglycemic conditions, thereby mitigating the influence of exogenous insulin (acknowledging that acute glucose toxicity at admission could transiently suppress β-cell function). However, the potential influence of acute metabolic status at admission cannot be completely excluded. Patients with missing key variables required for diagnostic classification or primary analyses were excluded. No statistical imputation was performed. Therefore, all regression and receiver operating characteristic (ROC) analyses were based on complete-case data.
Descriptive statistics: Continuous variables were summarized as mean ± SD for approximately normally distributed data, or as median with interquartile range (25th-75th percentile) for skewed variables. Categorical variables were sum
Group comparisons: For comparisons between two groups (e.g., KP-T2DM vs T2DM, LADA vs KP-T2DM), we used the independent-samples t-test for normally distributed variables and the Mann-Whitney U test for non-normally distributed variables. Categorical variables were compared using the χ2 test or Fisher’s exact test, as appropriate. Key comparisons of interest included metabolic differences between ketosis-onset diabetes and non-ketosis diabetes, as well as between LADA and KP-T2DM.
Logistic regression models were constructed to identify independent factors associated with (a) Ketosis at diabetes onset among newly diagnosed T2DM patients; and (b) LADA diagnosis among patients with ketosis-onset diabetes.
For model (a), the dependent variable was the presence of ketosis/DKA in new-onset diabetes (yes/no, comparing KP-T2DM vs ordinary T2DM). We first performed univariate analyses, and variables that differed significantly between the KP-T2DM and T2DM groups were entered as candidates into a multivariable logistic regression with stepwise selection. Adjusted odds ratios with 95% confidence intervals were obtained from the final model. For model (b), the dependent variable was LADA vs KP-T2DM status among ketosis-onset cases. Similarly, variables that significantly differed between LADA and antibody-negative KP-T2DM were entered into a stepwise logistic regression to identify significant diagnostic predictors. Multicollinearity among independent variables was assessed using variance inflation factors, and variables with significant collinearity were excluded from the final model.
To evaluate the discriminatory performance of the identified biomarkers, ROC curves were constructed. For differentiating ketosis-onset from nonketotic T2DM (model a), ROC curves were generated for each significant variable (e.g., age, BMI, fasting C-peptide, and fasting plasma glucose). The area under the curve (AUC) with 95% confidence intervals was calculated. Similarly, to differentiate LADA from KP-T2DM (model b), ROC curves were generated for both the multivariable logistic regression model and individual biomarkers (e.g., 2-hour C-peptide, HOMA-IR, HDL-C, and GGT). An AUC greater than 0.70 was considered indicative of acceptable discriminatory performance within the study cohort. Optimal cutoff values were determined by maximizing Youden’s index, with corresponding sensitivity and specificity. Because all ROC analyses were performed using the same study cohort in which candidate biomarkers were identified, the resulting AUCs and cutoff values should be considered exploratory and require independent validation before routine clinical application.
We performed Spearman’s rank correlation analysis to explore relationships between biochemical markers and metabolic indices. In particular, we examined correlations between liver enzymes (ALP and GGT) and measures of insulin resistance and lipid metabolism because previous studies have suggested possible associations between these markers and metabolic dysfunction. This analysis was intended to explore associations between liver enzymes and metabolic characteristics within the study cohort rather than to infer mechanistic relationships.
We conducted subgroup analyses by sex and age group to determine whether the clinical differences remained consistent (data not shown; performed for internal validation). Two investigators cross-checked all results for accuracy. No formal internal validation (e.g., cross-validation or bootstrapping) was performed for the regression models.
Baseline comparisons between ketosis-onset and nonketotic patients indicated that the ketosis group was significantly younger, whereas sex distribution and body weight did not differ (Supplementary Table 1). When specifically comparing KP-T2DM with classic T2DM, KP-T2DM patients were younger (median 32 years vs 44 years, P < 0.001), more often male (78.8% vs 69.2%, P < 0.001), and had higher body weight and BMI (Table 1). Other biochemical and hormonal parameters did not differ significantly between the two groups (Supplementary Table 1).
| Parameter | KP-T2DM (n = 146) | T2DM (n = 93) | P value |
| Male | 115 (78.8) | 65 (69.2) | < 0.001 |
| Age (years) | 32 | 44 (34-51) | < 0.001 |
| BMI (kg/m2) | 25.10 (22.90-28.62) | 23.88 (22.49-26.28) | 0.037 |
| FPG (mmol/L) | 14.37 ± 3.75 | 12.73 ± 4.13 | 0.001 |
| HbA1c (%) | 12.10 ± 2.21 | 10.87 ± 2.12 | < 0.001 |
| FCP (ng/mL) | 0.70 (0.36-0.93) | 1.20 (0.61-1.51) | < 0.001 |
| 2hCP (ng/mL) | 1.67 (1.02-2.02) | 2.71 (1.46-3.13) | < 0.001 |
| TG (mmol/L) | 2.78 (1.63-5.34) | 1.95 (1.23-3.11) | < 0.001 |
| HDL-C (mmol/L) | 0.85 (0.75-0.88) | 0.91 (0.78-1.00) | 0.005 |
| ALP (U/L) | 114.5 (87.8-134.7) | 86.1 (68.0-97.5) | < 0.001 |
At diagnosis, the KP-T2DM group exhibited more severe metabolic decompensation, with significantly higher FPG and glycated hemoglobin levels. β-cell function was markedly impaired, as reflected by lower fasting and postprandial C-peptide levels. Biochemically, KP-T2DM patients showed higher triglycerides and ALP levels, and lower HDL-C compared with the T2DM group (Table 1). The prevalence of high-risk atherosclerotic cardiovascular disease and carotid artery plaques was significantly lower in the KP-T2DM group than in the T2DM group (Supplementary Table 2).
Multivariate logistic regression identified age, FCP, and free triiodothyronine as protective factors against a ketotic presentation, whereas higher body weight, FPG, and ALP were independent risk factors (Table 2). Multivariate logistic regression identified age, FCP, and free triiodothyronine as protective factors against a ketotic presentation, whereas higher body weight, FPG, and ALP were independent risk factors (Table 2). ROC curve analysis demonstrated the discriminatory performance of these markers within the study cohort, with free triiodothyronine (AUC = 0.819; cut-off = 2.70 pg/mL) and FPG (AUC = 0.812; cut-off = 11.12 mmol/L) showing the strongest performance. ALP also demonstrated discriminatory value (AUC = 0.746; cut-off = 89.4 U/L; sensitivity 74.7%; specificity 69.9%; Table 3). Correlation analysis further showed that ALP levels were positively associated with triglycerides and TC (Supplementary Table 3). These findings suggest that ALP is associated with the presence of ketosis at diagnosis; however, prospective studies are required to determine whether ALP has predictive value for future ketosis. As shown in Figure 2, key biomarkers demonstrated varying diagnostic performance, with free triiodothyronine and FPG showing relatively strong performance for predicting ketosis. The multivariate logistic regression results are presented in Figure 3.
| Variable | OR | 95%CI | P value |
| Age (per year) | 0.954 | 0.92-0.99 | 0.016 |
| Weight (per kg) | 1.074 | 1.03-1.12 | 0.002 |
| FPG (per mmol/L) | 1.420 | 1.24-1.63 | < 0.001 |
| FCP (per ng/mL) | 0.193 | 0.07-0.51 | < 0.001 |
| FT3 (per pg/mL) | 0.118 | 0.04-0.32 | < 0.001 |
| ALP (per U/L) | 1.016 | 1.00-1.03 | 0.024 |
| Variable | AUC | Sensitivity | Specificity | Optimal cut-off | P value |
| Age (years) | 0.71 | 0.76 | 0.61 | 40.50 | < 0.001 |
| FPG (mmol/L) | 0.81 | 0.82 | 0.70 | 11.12 | < 0.001 |
| FT3 (pg/mL) | 0.82 | 0.62 | 0.91 | 2.70 | < 0.001 |
| ALP (U/L) | 0.75 | 0.75 | 0.70 | 89.40 | < 0.001 |
When comparing the two ketosis-onset groups, the median age of onset was similar (32 years). However, LADA patients exhibited a markedly leaner phenotype, with significantly lower body weight, BMI, and both systolic and diastolic blood pressure (Table 4). LADA patients exhibited markedly reduced β-cell function and lower IR than KP-T2DM patients (Table 4). These findings indicate severe β-cell failure and impaired glycemic stability in LADA, in contrast to the preserved β-cell reserve but stronger IR observed in KP-T2DM. Biochemical indices further distinguished the two groups: LADA patients had lower triglycerides and higher HDL-C levels (Table 4).
| Parameter | LADA (n = 55) | KP-T2DM (n = 146) | P value |
| Male | 29 (52.7) | 115 (78.8) | < 0.001 |
| Age (years) | 32 | 32 | 0.772 |
| Weight (kg) | 55 | 75 | < 0.001 |
| BMI (kg/m2) | 20.20 (18.36-22.14) | 25.10 (22.90-28.62) | < 0.001 |
| Systolic BP (mmHg) | 114.07 ± 16.55 | 126.17 ± 16.35 | < 0.001 |
| Diastolic BP (mmHg) | 74.82 ± 11.39 | 81.60 ± 12.52 | < 0.001 |
| FCP (ng/mL) | 0.30 (0.15-0.46) | 0.70 (0.36-0.93) | < 0.001 |
| 2hCP (ng/mL) | 0.64 (0.37-0.80) | 1.67 (1.02-2.02) | < 0.001 |
| HOMA-β (C-peptide) | 8.10 (3.83-15.99) | 15.70 (7.47-28.32) | < 0.001 |
| HOMA-IR (C-peptide) | 2.85 (2.23-3.81) | 4.47 (3.14-6.24) | < 0.001 |
| TG (mmol/L) | 1.64 (1.13-2.62) | 2.78 (1.63-5.34) | 0.003 |
| HDL-C (mmol/L) | 1.04 (0.97-1.08) | 0.85 (0.75-0.88) | 0.018 |
| LDL-C (mmol/L) | 2.60 (2.54-2.60) | 2.97 (2.55-3.19) | 0.011 |
| ALP (U/L) | 100.40 (85.20-119.00) | 114.51 (87.75-134.65) | 0.047 |
| GGT (U/L) | 20.20 (14.80-25.80) | 44.65 (22.85-56.15) | < 0.001 |
Finally, comorbidities related to IR were markedly less frequent in the LADA group. The prevalence of fatty liver (18.2% vs 64.4%) and hypertension (1.8% vs 14.4%) was significantly lower in LADA than in KP-T2DM, and diabetic nephropathy was absent in LADA but present in 7.5% of KP-T2DM patients. Data on carotid plaque, neuropathy, and retinopathy were limited and did not indicate clear differences between the groups (Supplementary Table 2).
Multivariate logistic regression identified several independent predictors of LADA among ketosis-onset patients. Postprandial C-peptide was identified as the strongest independent discriminator (Table 5). ROC analysis confirmed that 2 hours C-peptide showed good discriminatory performance (AUC = 0.852). Other variables showed additional but less robust discriminatory value (Table 6).
| Variable | OR | 95%CI | P value |
| 2hCP (per ng/mL) | 17.89 | 4.96-64.45 | < 0.001 |
| HOMA-IR (C-peptide) | 0.68 | 0.51-0.91 | 0.010 |
| HDL-C (per mmol/L) | 0.00 | 0.00-0.02 | < 0.001 |
| LDL-C (per mmol/L) | 2.47 | 1.16-5.27 | 0.019 |
| GGT (per U/L) | 1.05 | 1.01-1.09 | 0.014 |
| Variable | AUC | Sensitivity | Specificity | Optimal cut-off | P value |
| 2hCP (ng/mL) | 0.85 | 0.86 | 0.77 | < 0.98 | < 0.001 |
| HDL-C (mmol/L) | 0.79 | 0.73 | 0.86 | > 1.03 | < 0.001 |
| GGT (U/L) | 0.79 | 0.95 | 0.56 | < 37.85 | < 0.001 |
| HOMA-IR (C-peptide) | 0.70 | 0.80 | 0.61 | < 4.01 | < 0.001 |
Correlation analysis provided further mechanistic insight. GGT was positively correlated with fasting and postprandial C-peptide, HOMA-β, HOMA-IR, triglycerides, and TC. These associations are consistent with previous studies linking GGT to metabolic dysfunction; however, hepatic insulin resistance and oxidative stress were not directly assessed in the present study, and therefore, no mechanistic conclusions can be drawn (Supplementary Table 3). Subgroup analysis stratifying LADA patients by glutamic acid decarboxylase 65 kDa isoform antibody titer revealed no significant differences in glycolipid metabolism or islet function between high-titer and low-titer groups, suggesting that auto
In this retrospective study of adults with newly diagnosed diabetes, more than one-quarter of patients presenting with ketosis were diagnosed with LADA, indicating that autoimmune diabetes is not uncommon in this clinical context. KP-T2DM showed a distinct phenotype compared with classic T2DM, including younger age at onset, more severe hyperglycemia, and impaired β-cell function. In addition, we identified non-antibody markers that may help differentiate LADA from KP-T2DM and identify patients at risk of ketosis at diagnosis. These findings should be interpreted with caution, given the retrospective design and potential sources of bias. Compared with previous studies, the present study specifically focused on adults presenting with ketosis at diabetes onset and compared LADA, KP-T2DM, and nonketotic T2DM within the same hospital-based cohort. This design allowed us to evaluate not only β-cell function but also lipid-related, liver enzyme, and thyroid function markers as accessible non-antibody indicators.
Our findings are consistent with previous reports indicating that KP-T2DM typically affects younger, more obese individuals, often male, and is characterized by severe hyperglycemia and ketosis in the absence of autoantibody positivity[16-18]. These patients may partially recover β-cell function after the acute episode, supporting the concept of reversible β-cell dysfunction and transient insulin dependence[19,20]. By contrast, LADA, despite its adult onset, is characterized by progressive autoimmune destruction of β-cells, a lean body habitus, and lower IR[21]. The present study contributes to the literature by demonstrating that a substantial proportion of ketosis-onset patients fall into the LADA category, highlighting the importance of considering autoimmune diabetes even in adults. This observation aligns with prior studies reporting LADA prevalence rates of 6%-12% in newly diagnosed adults with diabetes[9,22]. However, it emphasizes that the rate is substantially higher among ketosis-onset presentations. The lower prevalence of high-risk atherosclerotic cardiovascular disease and carotid artery plaques observed in the KP-T2DM group should be interpreted cautiously. Because patients with KP-T2DM were substantially younger and differed in other baseline characteristics, residual confounding cannot be excluded, and these findings may partially reflect demographic differences rather than disease phenotype alone.
Our multivariate analysis identified several factors associated with ketotic onset in T2DM. Age and C-peptide were inversely associated with ketosis, whereas body weight and FPG were positively associated. These findings are generally consistent with previous studies linking younger age and impaired β-cell function to increased ketosis risk[23]. Previous studies on KP diabetes have mainly emphasized age, obesity-related features, glycemic burden, β-cell reserve, and metabolic indicators as predictors of ketosis[24,25]. Free triiodothyronine and ALP were also associated with ketotic presentation; however, their roles should be interpreted cautiously. Our study additionally evaluated ALP and free triiodothyronine, which may provide supplementary information but should be interpreted as exploratory associations[26].
In comparing LADA and KP-T2DM, postprandial C-peptide showed the best discriminatory performance, with a cut-off of < 0.975 ng/mL providing good sensitivity and specificity. Our finding that 2 hours C-peptide was the strongest discriminator is consistent with previous evidence emphasizing the importance of C-peptide for assessing residual β-cell function and supporting diabetes classification. Recent reviews also note that C-peptide, together with islet auto
These findings suggest potential diagnostic utility. In clinical practice, access to islet autoantibody testing may be limited, particularly in resource-constrained settings. Readily available non-antibody biomarkers may therefore provide supportive information for the clinical evaluation of patients presenting with ketosis at diabetes onset. Clinical and metabolic features were associated with the presence of ketosis at diagnosis[27], whereas higher glucose and elevated ALP levels were associated with ketotic presentation. Whether these markers predict future ketosis requires prospective validation[28]. Among patients with ketosis-onset diabetes, postprandial C-peptide, HDL-C, and GGT demonstrated discriminatory value for differentiating LADA from KP-T2DM within the present cohort. These biomarkers may provide supportive information for disease classification; however, their clinical applicability requires independent validation before routine use. From a clinical management perspective, differentiation between LADA and KP-T2DM remains important because their long-term clinical trajectories may differ. LADA is generally associated with progressive β-cell decline, whereas KP-T2DM may show partial recovery of β-cell function after the acute phase. Misclassification may therefore influence subsequent management strategies. In the present study, ALP and GGT were associated with metabolic differences between groups; however, their clinical significance remains uncertain and requires further investigation. Because of the retrospective cross-sectional design, the observed associations between ALP, GGT, free triiodothyronine, and ketosis-related phenotypes should not be interpreted as evidence of causal or mechanistic relationships.
This study has several strengths. We analyzed a well-defined inception cohort with comprehensive clinical and biochemical profiling. By evaluating multiple metabolic parameters, we explored a range of non-antibody markers relevant to ketosis-onset diabetes. To our knowledge, this is among the first studies in China to examine routinely available biochemical markers, including ALP and GGT, in this clinical context. These findings may help expand the scope of markers considered for differentiation, although their roles require further validation.
From a pathophysiological perspective, the observed differences between LADA and KP-T2DM may reflect two partially distinct but overlapping metabolic processes. LADA is primarily characterized by autoimmune-mediated β-cell dysfunction and progressive loss of insulin secretion, whereas KP-T2DM is more strongly associated with IR, obesity-related metabolic stress, and transient β-cell impairment during acute decompensation. In this context, C-peptide reflects residual β-cell function and therefore serves as the most direct discriminator between these phenotypes. In contrast, lipid-related markers and liver enzymes may reflect broader metabolic disturbances. Previous studies have linked GGT and ALP to metabolic syndrome, fatty liver, and oxidative stress; however, these enzymes should be interpreted as indirect indicators of metabolic dysfunction rather than specific markers of hepatic IR. Similarly, the association between free triiodothyronine and ketotic presentation may reflect systemic metabolic stress rather than a direct causal role. Reduced free triiodothyronine levels have been reported in acute illness and may represent a non-thyroidal illness response during severe hyperglycemia or ketosis. Therefore, these findings should be interpreted as associations rather than evidence of causality.
Several limitations of this study should be acknowledged. First, the retrospective, single-center design may limit the generalizability of the findings. Second, because the study population was derived from hospitalized patients, selection bias may exist, as these individuals may represent more severe metabolic phenotypes than those managed in outpatient settings. In addition, substantial baseline differences existed among the study groups, including age, sex, BMI, and obesity-related characteristics. Therefore, residual confounding cannot be excluded, and some observed metabolic and cardiovascular differences may partially reflect underlying demographic characteristics rather than disease phenotype alone. Future studies incorporating appropriate adjustment for baseline confounding, such as propensity score methods or multivariable adjusted analyses, are warranted. Third, the lack of longitudinal follow-up precludes assessment of long-term disease progression, including β-cell function decline and insulin dependence. In addition, C-peptide measurements obtained during acute presentation may have been influenced by glucose toxicity and metabolic stress, potentially affecting the assessment of β-cell function despite efforts to minimize this bias. Furthermore, because this was a cross-sectional observational study, the observed associations between biochemical markers and ketotic phenotypes should not be interpreted as evidence of causal or mechanistic relationships. Finally, the absence of an external validation cohort limits the robustness of the findings, and prospective multicenter studies with independent validation cohorts are required to confirm these findings before routine clinical application.
Future studies should validate the diagnostic cutoffs identified in this study in larger, multicenter populations and evaluate their performance using independent validation cohorts as well as their association with long-term clinical outcomes. Mechanistic investigations are warranted to determine whether the observed associations of ALP and GGT with ketotic diabetes are related to hepatic insulin resistance, oxidative stress, β-cell dysfunction, or other metabolic pathways. If validated in prospective studies, these biomarkers could potentially be incorporated into predictive models or nomograms to support the early classification of atypical diabetes. Future mechanistic studies incorporating direct assessments of hepatic insulin resistance, oxidative stress, inflammatory pathways, and longitudinal follow-up are also needed to clarify the biological significance and clinical relevance of these observed associations before routine clinical implementation.
In adults with ketosis-onset diabetes, LADA is not uncommon and should be considered in the differential diagnosis. Postprandial C-peptide showed the strongest performance for distinguishing LADA from KP-T2DM, while other markers provided supplementary information. These findings may provide supportive information for differentiating atypical diabetes in settings where antibody testing is limited, although further validation is required.
| 1. | Krause M, De Vito G. Type 1 and Type 2 Diabetes Mellitus: Commonalities, Differences and the Importance of Exercise and Nutrition. Nutrients. 2023;15:4279. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 35] [Reference Citation Analysis (0)] |
| 2. | Tegegne BA, Adugna A, Yenet A, Yihunie Belay W, Yibeltal Y, Dagne A, Hibstu Teffera Z, Amare GA, Abebaw D, Tewabe H, Abebe RB, Zeleke TK. A critical review on diabetes mellitus type 1 and type 2 management approaches: from lifestyle modification to current and novel targets and therapeutic agents. Front Endocrinol (Lausanne). 2024;15:1440456. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 28] [Reference Citation Analysis (0)] |
| 3. | Eledrisi MS, Elzouki AN. Management of Diabetic Ketoacidosis in Adults: A Narrative Review. Saudi J Med Med Sci. 2020;8:165-173. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 9] [Cited by in RCA: 44] [Article Influence: 7.3] [Reference Citation Analysis (0)] |
| 4. | El-Remessy A. Diabetic Ketoacidosis Management: Updates and Challenges for Specific Patient Population. Endocrines. 2022;3:801-812. [DOI] [Full Text] |
| 5. | Imam SK, Hassan FM, Mohamed H. Latent Autoimmune Diabetes in an Adult Male Presenting With Diabetic Ketoacidosis (DKA). Cureus. 2023;15:e33847. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 2] [Reference Citation Analysis (0)] |
| 6. | Dubois-Laforgue D, Timsit J. The Etiological Diagnosis of Diabetes: Still a Challenge for the Clinician. Endocrines. 2023;4:437-456. [DOI] [Full Text] |
| 7. | Sacks DB, Arnold M, Bakris GL, Bruns DE, Horvath AR, Lernmark Å, Metzger BE, Nathan DM, Kirkman MS. Guidelines and Recommendations for Laboratory Analysis in the Diagnosis and Management of Diabetes Mellitus. Clin Chem. 2023;69:808-868. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 109] [Cited by in RCA: 89] [Article Influence: 29.7] [Reference Citation Analysis (2)] |
| 8. | Bavuma C, Sahabandu D, Musafiri S, Danquah I, McQuillan R, Wild S. Atypical forms of diabetes mellitus in Africans and other non-European ethnic populations in low- and middle-income countries: a systematic literature review. J Glob Health. 2019;9:020401. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 18] [Cited by in RCA: 29] [Article Influence: 4.1] [Reference Citation Analysis (0)] |
| 9. | Buzzetti R, Tuomi T, Mauricio D, Pietropaolo M, Zhou Z, Pozzilli P, Leslie RD. Management of Latent Autoimmune Diabetes in Adults: A Consensus Statement From an International Expert Panel. Diabetes. 2020;69:2037-2047. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 201] [Cited by in RCA: 177] [Article Influence: 29.5] [Reference Citation Analysis (4)] |
| 10. | Boike S, Mir M, Rauf I, Jama AB, Sunesara S, Mushtaq H, Khedr A, Nitesh J, Surani S, Khan SA. Ketosis-prone diabetes mellitus: A phenotype that hospitalists need to understand. World J Clin Cases. 2022;10:10867-10872. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in CrossRef: 3] [Cited by in RCA: 5] [Article Influence: 1.3] [Reference Citation Analysis (2)] |
| 11. | Ji B, Suresh S, Bally K, Naher K, Banerji MA. Ketosis-Prone Type 2 Diabetes (Flatbush Diabetes) in Remission: A Report of Two Cases. Cureus. 2022;14:e28514. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 1] [Reference Citation Analysis (0)] |
| 12. | Makahleh L, Othman A, Vedantam V, Vedantam N. Ketosis-Prone Type 2 Diabetes Mellitus: An Unusual Presentation. Cureus. 2022;14:e30031. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 5] [Reference Citation Analysis (0)] |
| 13. | Kovacs A, Bunduc S, Veres DS, Palinkas D, Gagyi EB, Hegyi PJ, Eross B, Mihaly E, Hegyi P, Hosszufalusi N. One third of cases of new-onset diabetic ketosis in adults are associated with ketosis-prone type 2 diabetes-A systematic review and meta-analysis. Diabetes Metab Res Rev. 2024;40:e3743. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 13] [Reference Citation Analysis (0)] |
| 14. | Jiang Y, Zhu J, Lai X. Development and Validation of a Risk Prediction Model for Ketosis-Prone Type 2 Diabetes Mellitus Among Patients Newly Diagnosed with Type 2 Diabetes Mellitus in China. Diabetes Metab Syndr Obes. 2023;16:2491-2502. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 5] [Reference Citation Analysis (0)] |
| 15. | Felton JL, Redondo MJ, Oram RA, Speake C, Long SA, Onengut-Gumuscu S, Rich SS, Monaco GSF, Harris-Kawano A, Perez D, Saeed Z, Hoag B, Jain R, Evans-Molina C, DiMeglio LA, Ismail HM, Dabelea D, Johnson RK, Urazbayeva M, Wentworth JM, Griffin KJ, Sims EK; ADA/EASD PMDI. Islet autoantibodies as precision diagnostic tools to characterize heterogeneity in type 1 diabetes: a systematic review. Commun Med (Lond). 2024;4:66. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 38] [Reference Citation Analysis (0)] |
| 16. | Wang X, Tan H. Male predominance in ketosis-prone diabetes mellitus. Biomed Rep. 2015;3:439-442. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 14] [Cited by in RCA: 20] [Article Influence: 1.8] [Reference Citation Analysis (0)] |
| 17. | Sjöholm Å. Ketosis-Prone Type 2 Diabetes: A Case Series. Front Endocrinol (Lausanne). 2019;10:684. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 16] [Cited by in RCA: 16] [Article Influence: 2.3] [Reference Citation Analysis (1)] |
| 18. | Arslanian S, Bacha F, Grey M, Marcus MD, White NH, Zeitler P. Evaluation and Management of Youth-Onset Type 2 Diabetes: A Position Statement by the American Diabetes Association. Diabetes Care. 2018;41:2648-2668. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 293] [Cited by in RCA: 262] [Article Influence: 32.8] [Reference Citation Analysis (4)] |
| 19. | Saisho Y. β-cell dysfunction: Its critical role in prevention and management of type 2 diabetes. World J Diabetes. 2015;6:109-124. [PubMed] [DOI] [Full Text] |
| 20. | Niu F, Liu W, Ren Y, Tian Y, Shi W, Li M, Li Y, Xiong Y, Qian L. β-cell neogenesis: A rising star to rescue diabetes mellitus. J Adv Res. 2024;62:71-89. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 54] [Cited by in RCA: 47] [Article Influence: 23.5] [Reference Citation Analysis (0)] |
| 21. | Yin W, Luo S, Xiao Z, Zhang Z, Liu B, Zhou Z. Latent autoimmune diabetes in adults: a focus on β-cell protection and therapy. Front Endocrinol (Lausanne). 2022;13:959011. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 17] [Reference Citation Analysis (0)] |
| 22. | Mohammadi M. Prevalence of latent autoimmune diabetes in adults and insulin resistance: a systematic review and meta-analysis. Eur J Transl Myol. 2024;34:12694. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 1] [Reference Citation Analysis (0)] |
| 23. | Hernandez A, Truckenbrod L, Federico Q, Campos K, Moon B, Ferekides N, Hoppe M, D'Agostino D, Burke S. Metabolic switching is impaired by aging and facilitated by ketosis independent of glycogen. Aging (Albany NY). 2020;12:7963-7984. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 14] [Cited by in RCA: 30] [Article Influence: 5.0] [Reference Citation Analysis (0)] |
| 24. | Mendez DA, Ortiz RM. Thyroid hormones and the potential for regulating glucose metabolism in cardiomyocytes during insulin resistance and T2DM. Physiol Rep. 2021;9:e14858. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 7] [Cited by in RCA: 16] [Article Influence: 3.2] [Reference Citation Analysis (0)] |
| 25. | Xu C, Wen S, Xu Z, Dong M, Yuan Y, Li Y, Zhou L. Low T3 Syndrome is Associated with Imbalance of Bone Turnover Biomarker in Patients with Type 2 Diabetes. Diabetes Metab Syndr Obes. 2024;17:3667-3682. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 1] [Cited by in RCA: 2] [Article Influence: 1.0] [Reference Citation Analysis (0)] |
| 26. | Iacovides S, Maloney SK, Bhana S, Angamia Z, Meiring RM. Could the ketogenic diet induce a shift in thyroid function and support a metabolic advantage in healthy participants? A pilot randomized-controlled-crossover trial. PLoS One. 2022;17:e0269440. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 7] [Cited by in RCA: 22] [Article Influence: 5.5] [Reference Citation Analysis (0)] |
| 27. | Xu C, Gong M, Wen S, Zhou M, Li Y, Zhou L. The Comparative Study on the Status of Bone Metabolism and Thyroid Function in Diabetic Patients with or without Ketosis or Ketoacidosis. Diabetes Metab Syndr Obes. 2022;15:779-797. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 12] [Cited by in RCA: 11] [Article Influence: 2.8] [Reference Citation Analysis (0)] |
| 28. | Masuda S, Ota K, Okazaki R, Ishii R, Cho K, Hiramatsu Y, Adachi Y, Koseki S, Ueda E, Minami I, Yamada T, Watanabe T. Clinical Characteristics Associated with the Development of Diabetic Ketoacidosis in Patients with Type 2 Diabetes. Intern Med. 2022;61:1125-1132. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 1] [Cited by in RCA: 6] [Article Influence: 1.5] [Reference Citation Analysis (0)] |