Xu MR, Li MH, Wang JW, Zhang YW, Lu JX, Ke JF, Li LX. Prevalence and clinical characteristics of diabetic retinopathy in patients newly diagnosed with ketosis-onset diabetes: A real-world study. World J Diabetes 2026; 17(3): 115465 [DOI: 10.4239/wjd.v17.i3.115465]
Corresponding Author of This Article
Lian-Xi Li, MD, PhD, Department of Endocrinology and Metabolism, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai Clinical Center for Diabetes, Shanghai Diabetes Institute, Shanghai Key Laboratory of Diabetes Mellitus, Shanghai Key Clinical Center for Metabolic Disease, No. 600 Yishan Road, Shanghai 200233, China. lilx@sjtu.edu.cn
Research Domain of This Article
Endocrinology & Metabolism
Article-Type of This Article
Observational Study
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Mar 15, 2026 (publication date) through Mar 15, 2026
Times Cited of This Article
Times Cited (0)
Journal Information of This Article
Publication Name
World Journal of Diabetes
ISSN
1948-9358
Publisher of This Article
Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA
Share the Article
Xu MR, Li MH, Wang JW, Zhang YW, Lu JX, Ke JF, Li LX. Prevalence and clinical characteristics of diabetic retinopathy in patients newly diagnosed with ketosis-onset diabetes: A real-world study. World J Diabetes 2026; 17(3): 115465 [DOI: 10.4239/wjd.v17.i3.115465]
Man-Rong Xu, Meng-Han Li, Ya-Wen Zhang, Jun-Xi Lu, Lian-Xi Li, Department of Endocrinology and Metabolism, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai Clinical Center for Diabetes, Shanghai Diabetes Institute, Shanghai Key Laboratory of Diabetes Mellitus, Shanghai Key Clinical Center for Metabolic Disease, Shanghai 200233, China
Jun-Wei Wang, Shanxi Bethune Hospital, Shanxi Academy of Medical Sciences, Third Hospital of Shanxi Medical University, Tongji Shanxi Hospital, Taiyuan 030032, Shanxi Province, China
Jiang-Feng Ke, Department of Radiation Oncology, Clinical Oncology School of Fujian Medical University, Fuzhou 350014, Fujian Province, China
Co-corresponding authors: Jiang-Feng Ke and Lian-Xi Li.
Author contributions: Xu MR undertook data curation, formal analysis, software, visualization, and writing original draft; Xu MR and Li MH contributed equally to this article, they are the co-first authors of this manuscript; Lu JX, Ke JF, and Li LX handled funding acquisition; Li MH, Wang JW, Zhang YW, Lu JX, and Li LX managed methodology, validation, and writing review and editing; Ke JF and Li LX were responsible for conceptualization, resources, project administration, supervision, and investigation, they contributed equally to this article, they are the co-corresponding authors of this manuscript; and all authors contributed to the article and approved the submitted version.
Supported by the National Natural Science Foundation of China, No. 81770813 and No. 82070866; the Joint Funds for the Innovation of Science and Technology, Fujian Province, No. 2023Y9453; Shanxi Research Program of Application Foundation, No. 202403021212199; and China Postdoctoral Science Foundation, No. 2024M751910.
Institutional review board statement: This study was approved by the Medical Ethics Committee of Shanghai Sixth People’s Hospital, approval No. 2018-KY-018(K).
Informed consent statement: Each subject gave written consent prior to participating in this study.
Conflict-of-interest statement: All the authors report no relevant conflicts of interest for this article.
STROBE statement: The authors have read the STROBE Statement-checklist of items, and the manuscript was prepared and revised according to the STROBE Statement-checklist of items.
Data sharing statement: All data generated or analyzed during this study are included in this article. Further enquiries can be directed to the corresponding author.
Corresponding author: Lian-Xi Li, MD, PhD, Department of Endocrinology and Metabolism, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai Clinical Center for Diabetes, Shanghai Diabetes Institute, Shanghai Key Laboratory of Diabetes Mellitus, Shanghai Key Clinical Center for Metabolic Disease, No. 600 Yishan Road, Shanghai 200233, China. lilx@sjtu.edu.cn
Received: October 17, 2025 Revised: November 23, 2025 Accepted: January 6, 2026 Published online: March 15, 2026 Processing time: 146 Days and 8.2 Hours
Abstract
BACKGROUND
The classification of ketosis-onset diabetes is controversial, and the prevalence and clinical characteristics of diabetic retinopathy (DR) in ketosis-onset diabetes mellitus remain unclear.
AIM
To compare the prevalence and risk factors of DR among patients with type 1 diabetes mellitus (T1DM), ketosis-onset diabetes, and non-ketotic type 2 diabetes mellitus (T2DM).
METHODS
This real-world observational study enrolled 1777 patients with newly diagnosed diabetes between January 2003 and December 2012, including 211 with T1DM, 673 with ketosis-onset diabetes, and 893 with non-ketotic T2DM. DR was assessed using digital nonmydriatic fundus photography, and its severity was graded based on the ETDRS classification. Clinical characteristics and risk factors of DR were compared across the three groups.
RESULTS
After controlling for age and sex, DR prevalence was significantly higher in ketosis-onset diabetes (9.5%) than in T1DM (5.7%, P = 0.034), but not significantly different from non-ketotic T2DM (12.3%, P = 0.105). Risk factors for DR in ketosis-onset diabetes and non-ketotic T2DM included increased estimated glomerular filtration rate (eGFR) and urinary albumin excretion. In contrast, elevated eGFR and 2-hour postprandial C-peptide were independent risk factors for DR in T1DM. The prevalence and risk factors for DR in ketosis-onset diabetes were similar to those in non-ketotic T2DM, but differed from T1DM.
CONCLUSION
These findings further support classifying ketosis-onset diabetes as a subtype of T2DM rather than idiopathic T1DM. However, the distinctive clinical features of ketosis-onset diabetes should not be ignored.
Core Tip: This study demonstrates that the clinical features and risk profile of diabetic retinopathy in ketosis-onset diabetes resemble those of non-ketotic type 2 diabetes rather than type 1 diabetes. The similarity in diabetic retinopathy patterns supports the classification of ketosis-onset diabetes as a subtype of type 2 diabetes. Nonetheless, ketosis-onset diabetes presents with distinct clinical characteristics that should be carefully considered in clinical decision-making and individualized management.
Citation: Xu MR, Li MH, Wang JW, Zhang YW, Lu JX, Ke JF, Li LX. Prevalence and clinical characteristics of diabetic retinopathy in patients newly diagnosed with ketosis-onset diabetes: A real-world study. World J Diabetes 2026; 17(3): 115465
Diabetic retinopathy (DR) is a common microvascular complication of diabetes and leading cause of visual impairment and blindness worldwide[1,2]. In proliferative DR (PDR), the risk of severe vision loss within 2 years can reach 26% if left untreated[3]. A previous study also reported markedly higher prevalence of any DR and sight-threatening DR in individuals with type 1 diabetes mellitus (T1DM; 47.16% and 18.03%, respectively) compared with those with type 2 diabetes mellitus (T2DM; 26.49% and 7.59%, respectively)[4]. Therefore, the heightened risk of severe DR across different types of diabetes underscores the central role of glycemic control in DR prevention and treatment, and warrants the need for further in-depth investigation. Although the clinical characteristics of DR in T1DM and T2DM have been widely examined, major gaps remain in understanding ketosis-onset diabetes. To date, no study has specifically examined the prevalence or clinical features of DR in this population.
According to the American Diabetes Association Standards of Medical Care in Diabetes 2025, Section 2: Classification and Diagnosis of Diabetes, ketosis-onset diabetes is classified as idiopathic T1DM[5]. It is characterized by marked insulin deficiency and spontaneous ketoacidosis at diagnosis, despite the absence of islet-associated autoantibodies, features that resemble T1DM[5]. However, unlike T1DM, which is typically characterized by absolute loss of islet secretory function and lifelong insulin dependence, most patients with ketosis-onset have significantly improved pancreatic β-cell secretory function after glycemic amelioration[6]. Consistently, previous studies have found that patients with ketosis-onset diabetes exhibit higher peak C-peptide levels than in patients with T1DM after an oral glucose load[7]. Moreover, a meta-analysis reported that over one-third of individuals presenting with ketosis or ketoacidosis meet the 2019 World Health Organization criteria for ketosis-prone type 2 diabetes[8]. Therefore, the classification of ketosis-onset diabetes has remained controversial in recent years.
Our previous studies have provided accumulating evidence that patients with ketosis-onset diabetes exhibit a complication profile closely resembling that of individuals with non-ketotic T2DM[9-13]. Specifically, we demonstrated that the prevalence and risk of nonalcoholic fatty liver disease, hypertension, metabolic syndrome, as well as carotid and femoral atherosclerosis in ketosis-onset diabetes were largely comparable to those observed in non-ketotic T2DM[9-12]. Extending these observations to microvascular complications, our most recent study revealed that the prevalence and risk of chronic kidney disease (CKD) in ketosis-onset diabetes paralleled those in non-ketotic T2DM but differed substantially from those in T1DM[13]. Collectively, ketosis-onset diabetes shares considerable clinical overlap with non-ketotic T2DM across both macrovascular and microvascular complications. Ketosis-onset diabetes is characterized by transient insulin deficiency at presentation and marked glycemic variability[14], which may contribute to DR development through oxidative stress, inflammation, and endothelial dysfunction[15]. However, the classification of ketosis-onset diabetes using DR, a representative and clinically meaningful microvascular complication, remains unexplored.
In this study, individuals who presented with diabetic ketosis at diabetes diagnosis and tested negative for islet-associated autoantibodies were defined as having ketosis-onset diabetes. This study investigated potential clues for classifying ketosis-onset diabetes by examining the clinical characteristics and risk factors of DR. Our findings may provide insights to guide future treatment and management strategies for ketosis-onset diabetes.
MATERIALS AND METHODS
Study population and design
The data of this real-world study were obtained from our registry study (macrovascular complications in patients with diabetes study: A real-world study of macrovascular complications in diabetic populations; clinical trial registration number: ChiCTR1800015893). With approval from the Ethics Committee of Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, approval No. 2018-KY-018(K), this cross-sectional study consecutively recruited 2166 newly diagnosed patients with diabetes between January 2003 and December 2012. According to our previous studies[11,12], newly diagnosed diabetes was defined as fasting plasma glucose (FPG) ≥ 7.0 mmol/L and/or 2-hour postprandial plasma glucose (2h PPG) ≥ 11.1 mmol/L in individuals without a prior history of diabetes. Diabetic ketosis was defined as the presence of hyperglycemia and moderate to severe urinary ketone bodies (15-150 mg/dL). Ketosis-onset diabetes was defined as the presence of diabetic ketosis at diabetes diagnosis, with the absence of glutamic acid decarboxylase (GAD) and/or tyrosine phosphatase-like islet antigen 2 (IA-2) autoantibodies. T1DM was defined as the presence of ≥ 1 islet autoantibody (GAD65 and/or IA-2A) measured at diagnosis using validated assays. Patients with newly diagnosed diabetes who were negative for islet autoantibodies and had no diabetic ketosis or ketoacidosis at onset were classified as non-ketotic T2DM. GAD and IA-2 autoantibodies were measured using enzyme-linked immunosorbent assay (EUROIMMUN Medizinische Labordiagnostika AG, Germany), and urinary ketone bodies were assessed using Legal’s test. Positivity thresholds for both autoantibodies were determined according to the manufacturer’s instructions[12].
Patients were excluded if they met any of the following criteria: (1) Age < 17 years; (2) Gestational diabetes mellitus; (3) Incomplete clinical data; (4) Absence of retinal photographs due to technical limitations, such as cataracts; or (5) Severe systemic or infectious diseases, or conditions, including cortisol therapy and renal insufficiency that could cause positive urinary ketones. Based on the fourth criterion, 97 patients (58 males and 39 females) were excluded. After applying all exclusion criteria, 211 T1DM, 673 ketosis-onset diabetes, and 893 patients with non-ketotic T2DM were included in the final analysis. All participants provided handwritten informed consent.
Examination and laboratory measurements
Data on smoking, alcohol consumption, and use of lipid-lowering drugs (LLDs) and antiplatelet agents (APA) were collected. Similarly, medical conditions including hypertension and CKD were recorded. Physical examination data included height, weight, waist and hip circumference, systolic blood pressure (SBP), and diastolic blood pressure (DBP). Waist-to-hip ratio (WHR) was calculated as waist circumstance divided by hip circumstance, and body mass index (BMI) was calculated as weight divided by the square of height. The definitions of CKD, hypertension, smoking, and alcohol consumption were consistent with our previous studies[9,16]. Specifically, CKD was defined as an estimated glomerular filtration rate (eGFR) < 60 mL/minute/1.73 m2 and/or 24-hour urinary albumin excretion (UAE) ≥ 300 mg/24 hours.
Blood samples were collected from patients following an overnight fasting and 2 hours postprandially. Glycolipid metabolism-related parameters included FPG, 2h PPG, glycosylated hemoglobin A1c (HbA1c), fasting C-peptide (FCP), 2-hour postprandial C-peptide (2h PCP), total triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C). Additional laboratory tests included C-reactive protein (CRP) and markers of liver and kidney function, such as alanine aminotransferase (ALT), aspartate aminotransferase, γ-glutamyl transpeptidase, total bilirubin, conjugated bilirubin, creatinine, and serum uric acid. UAE was calculated as the mean of three morning urine samples collected during hospitalization. The homeostasis model assessment index of insulin resistance (HOMA2-IR), homeostasis model assessment index of β-cell function (HOMA2-β), and eGFR were elucidated in detail in our previous method[13].
Diagnosis criterion
Following the digital nonmydriatic fundus photography protocol described in our previous study, each participant underwent fundus photography including retinal discs, macula, and temporal vessels using a 45° digital prism-free camera (Canon CR6-45NM) after a 5-minute dark adaptation. A single experienced ophthalmologist, blinded to all participant characteristics, reviewed the bitmap images to identify DR. DR was diagnosed when any of the following lesions were present: Microaneurysms, hemorrhages, soft exudates, hard exudates, neovascularization, or laser photocoagulation scars. Non-PDR (NPDR) was defined based on the presence of microaneurysms, retinal hemorrhages, venous wall changes, and intraretinal microvascular abnormalities[17]. PDR was characterized by neovascular growth, pre-retinal or vitreous hemorrhages, and fibrous proliferations[18].
Statistical analysis
Statistical analyses were performed using SPSS version 15.0. A two-sided P value < 0.05 was considered statistically significant. Variables with a normal distribution were presented as mean ± SD, and differences among the three groups were assessed using independent samples t-tests or one-way analysis of variance with Least Significant Difference post hoc tests. Skewed variables and qualitative data were expressed as median interquartile range (25%-75%), and analyzed using the Mann-Whitney U or Kruskal-Wallis H tests. Additionally, skewed variables were transformed to approximate normality using a rank-based inverse normal transformation (normal scores in SPSS) prior to analysis. χ2 test was used to compare the differences in prevalence of categorical variables. To investigate risk factors associated with DR, binary logistic regression was applied to estimate odds ratios (OR). The models were adjusted for multiple potential confounders, including demographic characteristics (sex, age); medical history and medication use (hypertension, use of LLD and APA), behavioral factors (smoking, alcohol consumption); and clinical indicators from physical and laboratory evaluations, such as SBP, DBP, WHR, BMI, ALT, γGT, TC, HDL-C, LDL-C, TG, ApoA, eGFR, SUA, UAE, FPG, 2h PPG, CRP, HbA1C, FCP, and 2h PCP.
RESULTS
Characteristics of study participants
The clinical characteristics of the three diabetic groups are summarized in Table 1. Age and sex distributions differed significantly across groups, with the proportion of males being markedly higher in ketosis-onset diabetes (74.7%) than in typical T1DM and non-ketotic T2DM (P < 0.001), highlighting a distinct sex-specific pattern in ketosis-onset diabetes. After adjusting for age and sex, significant differences among the three groups remained for hypertension, CKD, use of LLD, SBP, DBP, WHR, BMI, FPG, 2h PPG, HbA1c, FCP, 2h FCP, TG, TC, HDL, LDL, ALT, γ-glutamyl transpeptidase, SUA, UAE, eGFR, CRP, HOMA2-β, and HOMA2-IR. Among these variables, the glucose metabolism-related indicators, FPG (9.71 mmol/L) and 2h PPG (16.26 mmol/L), exhibited the highest levels in patients with ketosis-onset diabetes, whereas HbA1c (11.48%) was highest in those with T1DM. In contrast, HOMA2-IR (1.80) reached its highest level among patients with non-ketotic T2DM. In addition, insulin secretion-related markers were the lowest in T1DM group (FCP 0.62 ng/mL; 2h FCP 1.04 ng/mL). Notably, patients with ketosis-onset diabetes exhibited higher FCP (1.36 ng/mL) and 2h FCP (2.54 ng/mL) levels compared to those with T1DM, but lower than those in the T2DM group, whose FCP and 2h FCP were 2.09 ng/mL and 4.84 ng/mL, respectively. These findings indicate that β-cell function in ketosis-onset diabetes lies between T1DM and T2DM.
Table 1 Characteristics of the subjects, n (%)/mean ± SD.
Risk factors associated with DR in each diabetic group are presented in Table 2. After adjustment for age, sex, hypertension, use of LLD and APA, smoking, alcohol consumption, SBP, DBP, WHR, BMI, ALT, TC, HDL-C, LDL-C, TG, eGFR, SUA, UAE, FPG, 2h PPG, CRP, HbA1c, FCP, and 2h PCP, multivariate binary logistic regression demonstrated that eGFR was strongly associated with DR risk in all three diabetic groups: OR = 3.638 [95% confidence intervals (CI): 1.057-12.524, P = 0.041] in T1DM; OR = 1.655 (95%CI: 1.075-2.527, P = 0.022) in ketosis-onset diabetes; and OR = 1.517 (95%CI: 1.115-2.066, P = 0.008) in non-ketotic T2DM. In T1DM, 2h PCP was also significantly associated with DR (OR = 3.511, 95%CI: 1.182-10.429, P = 0.024). Conversely, in ketosis-onset diabetes and non-ketotic T2DM, UAE emerged as an additional shared risk factor, with OR = 2.153 (95%CI: 1.279-3.625, P = 0.004) and OR = 1.653 (95%CI: 1.192-2.292, P = 0.003), respectively. However, the risk factors for DR in patients with ketosis-onset diabetes were not fully consistent with non-ketotic T2DM, as evidenced by the significantly negative correlation between BMI and DR, which was only observed in non-ketotic T2DM (OR = 0.842, 95%CI: 0.764-0.927, P < 0.001).
Table 2 Risk factors of diabetic retinopathy in each diabetic group.
DR prevalence stratified by age and sex in each diabetic group
As shown in Figure 1, the prevalence of DR did not differ significantly between men and women with T1DM (5.9% vs 5.3%; P = 0.690) or non-ketotic T2DM (12.0% vs 12.8%; P = 0.690; Figure 1A and C). In contrast, among patients with ketosis-onset diabetes, the prevalence of DR was significantly higher in women (15.9%) than in men (7.4%; P = 0.001), with women exhibiting more than twice the prevalence of men (Figure 1B, P = 0.205 for trend in T1DM, and Figure 1F, P = 0.970 for trend in non-ketotic T2DM). Similarly, stratification by age showed no significant difference in DR prevalence across age groups in patients with ketosis-onset diabetes (Figure 1E, P = 0.338). Unlike ketosis-onset diabetes and T2DM.
Figure 1 Diabetic retinopathy prevalence stratified by age and gender in each diabetic group.
A: Comparison of diabetic retinopathy (DR) prevalence between men and women patients with type 1 diabetes mellitus (T1DM) after adjusting for age (P = 0.690); B: Comparison of DR prevalence stratified by age after adjusting for gender in patients with T1DM (P = 0.205 for trend); C: Comparison of DR prevalence between men and women patients with ketosis-onset diabetes after adjusting for age (P = 0.001); D: Comparison of DR prevalence stratified by age after adjusting for gender in patients with ketosis-onset diabetes (P = 0.338 for trend); E: Comparison of DR prevalence between men and women patients with non-ketotic T2DM after adjusting for age (P = 0.645); F: Comparison of DR prevalence stratified by age after adjusting for gender in patients with non-ketotic T2DM (P = 0.970 for trend). DR: Diabetic retinopathy; T1DM: Type 1 diabetes mellitus; T2DM: Type 2 diabetes mellitus.
Comparison of DR across the three groups
Figure 2 compares the prevalence of DR, ORs for DR, and prevalence of DR stratified by severity across the three diabetic groups after adjustment for sex and age. The prevalence of DR was 9.5% in ketosis-onset diabetes and 12.3% in non-ketotic T2DM, both significantly higher than that in T1DM at 5.7% (Figure 2A; P = 0.034 for T1DM vs ketosis-onset diabetes and P = 0.007 for T1DM vs non-ketotic T2DM). However, the prevalence did not differ significantly between ketosis-onset diabetes and non-ketotic T2DM (P = 0.105). Using T1DM as the reference, patients with ketosis-onset diabetes had a 2.016-fold higher risk of DR (95%CI: 1.054-3.858, P = 0.049), while patients with non-ketotic T2DM had a 2.330-fold higher risk (95%CI: 1.258-4.313, P = 0.005; Figure 2B). Across the three groups, non-ketotic T2DM exhibited the highest prevalence of NPDR (11.3%) and PDR (1.0%; Figure 2C). Additionally, only 0.1% of patients with ketosis-onset diabetes had PDR, representing the lowest prevalence among the three groups.
Figure 2 Comparison of diabetic retinopathy across three groups.
A: Comparison of the prevalence of diabetic retinopathy (DR) across type 1 diabetes mellitus (T1DM), ketosis-onset diabetes and non-ketotic T2DM after adjusting for gender and age (T1DM vs ketosis-onset diabetes: P = 0.034; ketosis-onset diabetes vs non-ketotic T2DM: P = 0.105; T1DM vs non-ketotic T2DM: P = 0.007); B: Odds ratio with 95% confidence interval of DR for the ketosis-onset diabetes and the non-ketotic T2DM, with T1DM as reference (ketosis-onset diabetes vs T1DM P = 0.049; non-ketotic T2DM vs T1DM P = 0.005; ketosis-onset diabetes vs non-ketotic T2DM P = 0.126); C: Comparison the prevalence of normal, non-proliferative DR and proliferative DR patients across T1DM, ketosis-onset diabetes and non-ketotic T2DM subjects after adjusting for gender and age (P = 0.015 for trend). DR: Diabetic retinopathy; T1DM: Type 1 diabetes mellitus; NPDR: Non-proliferative diabetic retinopathy; PDR: Proliferative diabetic retinopathy.
Comparison of islet function-related parameters across the three groups
Figure 3 presents the islet function-related parameters after adjusting for sex and age. Serum FCP, 2h PCP, HOMA2-β, and HOMA2-IR increased progressively from T1DM to ketosis-onset diabetes then to non-ketotic T2DM (all P < 0.001 for trend). These results indicate that patients with newly diagnosed ketosis-onset diabetes exhibited an intermediate level of insulin secretion, lower than in non-ketotic T2DM but higher than in T1DM. A similar pattern was observed for insulin sensitivity.
Figure 3 Comparison of islet function-related parameters.
A: Comparison of serum fasting C-peptide levels among patients with type 1 diabetes mellitus (T1DM), ketosis-onset diabetes and non-ketotic T2DM after adjusting for gender and age (P < 0.001 for trend); B: Comparison of serum 2-hour postprandial C-peptide levels among patients with T1DM, ketosis-onset diabetes and non-ketotic T2DM after adjusting for gender and age (P < 0.001 for trend); C: Comparison of homeostasis model assessment index of β-cell function levels among patients with T1DM, ketosis-onset diabetes and non-ketotic T2DM after adjusting for gender and age (P < 0.001 for trend); D: Comparison of homeostasis model assessment index of insulin resistance levels among patients with T1DM, ketosis-onset diabetes and non-ketotic T2DM after adjusting for gender and age (P < 0.001 for trend). FCP: Fasting C-peptide; 2h PCP: 2-hour postprandial C-peptide; T1DM: Type 1 diabetes mellitus; T2DM: Type 2 diabetes mellitus; HOMA2-β: Homeostasis model assessment index of β-cell function; HOMA2-IR: Homeostasis model assessment index of insulin resistance.
DISCUSSION
The present study comprehensively compared the prevalence, clinical characteristics, severity, and risk factors of DR among three distinct diabetic populations, including T1DM, ketosis-onset diabetes, and non-ketotic T2DM. Our findings demonstrated that ketosis-onset diabetes shared more clinical characteristics with non-ketotic T2DM compared to that of T1DM, providing further evidence that ketosis-onset diabetes should be classified as a subgroup of T2DM.
First, ketosis-onset diabetes exhibited clinical characteristics similar to those of classic non-ketotic T2DM. Unlike T1DM, the SBP, DBP, WHR, and BMI of patients with ketosis-onset diabetes and non-ketotic T2DM were closely aligned, suggesting comparable blood pressure levels and anthropometric indices between the two groups. In line with our results, a recent meta-analysis found that patients presenting with DKA or ketosis at diabetes onset have a significantly higher BMI than those with T1DM[8].
Nevertheless, our findings also suggested that ketosis-onset diabetes exhibits distinct clinical characteristics, making its classification challenging in clinical practice. Contrary to the other two types of diabetes, ketosis-onset diabetes showed a male predominance of 74.7%. Similarly, Zhang et al[19] reported that eight of nine patients diagnosed with ketosis or ketoacidosis were male, suggesting that men may have a higher susceptibility to ketosis-onset diabetes. While the underlying mechanisms remain unclear, this sex disparity may be attributed to hormonal factors, body fat distribution, and higher insulin sensitivity in females compared to that of males[20,21].
Additionally, ketosis-onset diabetes exhibited intermediate characteristics between non-ketotic T2DM and T1DM in both islet function and insulin resistance. The levels of FCP, 2h PCP, and HOMA2-β in ketosis-onset diabetes were intermediate between those in non-ketotic T2DM and T1DM. This suggests that islet dysfunction in ketosis-onset diabetes is more severe than in T2DM but less severe than in T1DM. These findings were also observed in a study of 1072 patients with non-autoimmune new-onset diabetes[22]. Similarly, insulin resistance as indicated by HOMA2-IR in ketosis-onset diabetes was intermediate between the other two diabetic populations. Consistent with our findings, an investigation in a sub-Saharan population showed significantly higher HOMA2-IR in non-ketotic T2DM than in ketosis-prone diabetes group[23]. Therefore, ketosis-onset diabetes represented a distinct clinical entity with features of both T1DM and T2DM.
Current evidence regarding DR prevalence in ketosis-onset diabetes remains limited. In this study, the overall prevalence of DR in ketosis-onset diabetes was 9.5%, closely aligning with the 12.3% observed in non-ketotic T2DM, but significantly higher than in the T1DM at 5.7%. Moreover, similar to non-ketotic T2DM, the OR for DR in ketosis-onset diabetes was more than twice as high as in T1DM. These results are consistent with a meta-analysis reporting a 13.1% prevalence of DR among newly diagnosed patients with T2DM[24]. Similarly, a large-scale study of over 60000 newly diagnosed patients with diabetic reported DR incidences of 3.8% in T2DM and 2.0% in T1DM after 1 year of follow-up, with T2DM showing a higher cumulative prevalence[25].
Although studies specifically examining DR prevalence in ketosis-onset diabetes are lacking, both our data and previous studies demonstrate that the overall comorbidity profile of ketosis-onset diabetes more closely resembles that of T2DM than T1DM, particularly regarding nonalcoholic fatty liver disease, hypertension, and metabolic syndrome[11,12,26]. Moreover, considering vascular complications of diabetes, including macrovascular complications, such as atherosclerosis, and microvascular complications, such as CKD, evidence consistently indicates that the clinical features and prevalence of these complications in ketosis-onset diabetes are more similar to those observed in T2DM rather than T1DM[9,10,13,27]. We hypothesize that this similarity may be attributable to fluctuating glucose levels and dysregulated lipid metabolism in ketosis-onset diabetes. Similarly, a study has shown that glucose fluctuations are more likely to drive retinal microvascular inflammation and neurodegeneration[28]. In addition, disordered lipid metabolism leads to oxidative stress in vascular endothelium[29], thereby jointly promoting DR development. These processes parallel the oxidative stress and inflammatory responses commonly observed in T2DM[30]. However, the molecular mechanisms underlying DR development in ketosis-onset diabetes remain to be explored in future studies.
However, we also found that patients with ketosis-onset diabetes exhibited distinct patterns of DR prevalence, distinguishing them from non-ketotic T2DM. The prevalence of DR in females with ketosis-onset diabetes was twice as high as that in males, revealing a remarkable sex difference. In contrast, no sex-based difference in DR prevalence was found in the T1DM and non-ketotic T2DM groups. This discrepancy may be attributed to women being more susceptible to ketosis-induced hyperglycemic stress, possibly due to higher systemic inflammation levels[31]. This may lead to the damage of the vascular endothelium and DR development. Additionally, fluctuations in female sex hormones and a higher visceral fat distribution may exacerbate impairment of insulin sensitivity during ketosis[32,33]. This may be directly damaging the microvascular endothelium[34] and potentially explaining the higher prevalence of DR among women with ketosis-onset diabetes. Therefore, while ketosis-onset diabetes shares certain DR-related characteristics with classic T2DM, it also presents unique features that highlight its distinct clinical profile. Prospective studies are warranted to further investigate sex-specific differences in DR risk among individuals with ketosis-onset diabetes.
The severity distribution of DR in patients with ketosis-onset diabetes differed from that observed in T1DM and non-ketotic T2DM. Only 0.1% of patients with ketosis-onset diabetes exhibited severe PDR, whereas the proportions in patients with T1DM and non-ketotic T2DM with severe PDR were five- and tenfold higher, respectively. In contrast, a global review reported a PDR prevalence of 3.3% (range 0%-13%) in T1DM and 0.6% (0.31%-1.5%) in T2DM[35]. This discrepancy likely reflects characteristics of our study cohort, which consisted exclusively of newly diagnosed patients with diabetes. As early-stage hyperglycemia is often asymptomatic, patients with non-ketotic and T2DM may have experienced prolonged undetected hyperglycemia prior to diagnosis. In contrast, T1DM and ketosis-onset diabetes tend to be diagnosed earlier owing to prominent clinical presentations. Therefore, patients with non-ketotic T2DM exhibited the highest prevalence of both NPDR and PDR in this study.
When evaluating risk factors for DR across the three diabetic populations, we found that renal lesion-related parameters, including eGFR and UAE, were common risk factors for ketosis-onset diabetes and non-ketotic T2DM. In contrast, WHR was significantly associated with DR development only in T1DM. Thus, ketosis-onset diabetes shares a comparable prevalence of DR with T2DM as well as exhibits similar risk factors. Our findings align with a large T2DM cohort study demonstrating that microalbuminuria confers nearly double the risk of developing DR compared with that of its absence, with the risk increasing almost sixfold in the presence of macroalbuminuria[36]. In addition, the positive association between eGFR levels and DR across all three groups may be attributable to the inclusion of newly diagnosed cases, in whom glomerular hyperfiltration is common and may contribute to this association[37].
Overall, the shared risk factor patterns between ketosis-onset diabetes and T2DM, particularly regarding retinal impairment, reinforces the notion that ketosis-onset diabetes may represent a subtype of T2DM. However, unlike ketosis-onset diabetes, low BMI emerged as an additional risk factor for DR in T2DM. Consistent with our findings, a previous study reported a significant negative association between BMI and the 6-year risk of DR in two Asian cohorts (hazard ratio = 0.78, 95%CI: 0.66-0.92)[38]. This paradox may reflect the natural course of T2DM, wherein DR is linked more closely to declining β-cell responsiveness than to insulin resistance, potentially resulting in muscle and fat loss[39]. Therefore, the risk of DR in ketosis-onset diabetes has a unique feature. Collectively, these findings underscore the heterogeneity of diabetes, with each subtype exhibiting distinct clinical features.
Although this study provides valuable and reliable data, it has some limitations. First, as a cross-sectional study, it could not assess longitudinal effects associated with the progression of ketosis-onset diabetes. Second, due to the recruitment of patients newly diagnosed with diabetes, we could not obtain information on the duration of hyperglycemia before diagnosis. Third, the study was conducted at a single center and included only hospitalized patients, which may limit generalizability. In particular, the hospital-based recruitment approach may have introduced selection bias by enriching the study cohort with individuals experiencing more severe metabolic disturbances. Therefore, multicenter prospective studies are needed to further characterize DR in ketosis-onset diabetes and facilitate comparisons with other diabetes subtypes. Additionally, because non-mydriatic fundus photography may miss peripheral lesions, future studies should incorporate post-dilation imaging to reduce detection bias. Residual confounding may also remain due to the absence of formal variable selection strategies, such as directed acyclic graphs. Future research should include a more comprehensive set of DR-related covariates to strengthen model validity.
CONCLUSION
The similar prevalence and characteristics of DR observed in ketosis-onset diabetes and non-ketotic T2DM support classifying ketosis-onset diabetes as a subtype of T2DM rather than T1DM. Notably, ketosis-onset diabetes exhibits distinct clinical features that warrant attention in clinical practice. These findings support the implementation of T2DM-based personalized management for ketosis-onset diabetes and underscore the need for further research to refine diabetes classification.
ACKNOWLEDGEMENTS
We thank all participants, investigators, and other colleagues for their precious contributions to the design, execution, and analysis of the present study.
Yau JW, Rogers SL, Kawasaki R, Lamoureux EL, Kowalski JW, Bek T, Chen SJ, Dekker JM, Fletcher A, Grauslund J, Haffner S, Hamman RF, Ikram MK, Kayama T, Klein BE, Klein R, Krishnaiah S, Mayurasakorn K, O'Hare JP, Orchard TJ, Porta M, Rema M, Roy MS, Sharma T, Shaw J, Taylor H, Tielsch JM, Varma R, Wang JJ, Wang N, West S, Xu L, Yasuda M, Zhang X, Mitchell P, Wong TY; Meta-Analysis for Eye Disease (META-EYE) Study Group. Global prevalence and major risk factors of diabetic retinopathy.Diabetes Care. 2012;35:556-564.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 3524][Cited by in RCA: 3283][Article Influence: 234.5][Reference Citation Analysis (3)]
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: 9][Reference Citation Analysis (0)]
Lontchi-Yimagou E, Nguewa JL, Assah F, Noubiap JJ, Boudou P, Djahmeni E, Balti EV, Atogho-Tiedeu B, Gautier JF, Mbanya JC, Sobngwi E. Ketosis-prone atypical diabetes in Cameroonian people with hyperglycaemic crisis: frequency, clinical and metabolic phenotypes.Diabet Med. 2017;34:426-431.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 16][Cited by in RCA: 25][Article Influence: 2.8][Reference Citation Analysis (0)]
Porchia LM, Ochoa-Precoma R, Reyes-Albarracín Y, Gonzalez-Mejia ME, López-Bayghen E. Age-related differences in fluctuations in insulin resistance evaluated with HOMA-IR and triglyceride and glucose-based indices during the menstrual cycle, as determined using the NHANES cross-sectional dataset.Maturitas. 2024;188:108085.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 6][Reference Citation Analysis (0)]
Scientific quality: Grade B, Grade B, Grade B, Grade C
Novelty: Grade B, Grade B, Grade B
Creativity or innovation: Grade B, Grade B, Grade B
Scientific significance: Grade A, Grade B, Grade B
P-Reviewer: Huo WQ, PhD, Associate Professor, China; Tung TH, PhD, Associate Professor, Director, Statistician, Taiwan; Yap CG, PhD, Associate Professor, Malaysia S-Editor: Bai Y L-Editor: A P-Editor: Lei YY
