Background
Ketosis-onset diabetes or ketosis-prone diabetes, once termed atypical diabetes, has been categorized as idiopathic type 1 diabetes based on the American Diabetes Association [
1‐
3]. However, several studies suggested that ketosis-onset diabetes was characterized as higher age of onset, overweight, the family history of diabetes, discontinuation of insulin treatment after euglycemic remission, and a variable response to diet and oral hypoglycemic agents, which supported that ketosis-onset diabetes was more likely to be a sub-group of type 2 diabetes [
2,
4].
The immunologic and metabolic characteristics of ketosis-onset diabetes have been well established, such as lack of islet-related antibodies and human leukocyte antigen (HLA) genetic association, unprovoked ketoacidosis, a remarkable but transient defect in insulin secretion, near-normoglycemic remission, substantial beta-cell reserve and displayed muscle, adipose tissue and hepar tissue insulin resistance [
2,
5‐
7]. Similar to type 2 diabetics, patients with ketosis-onset diabetes were observed with low frequencies of type 1 diabetic susceptibility or resistance alleles [
7‐
9]. Mauvais-Jarvis et al. [
1] reported that insulin secretory capacity in ketosis-onset diabetes was lost slowly, approaching that in type 2 diabetes, whereas differing from that in type 1 diabetes. However, the clinical manifestations of ketosis-onset diabetes were mixed with those of both type 1 and non-ketotic type 2 diabetes, which reflected the difficulty of classifying this heterogeneous group. Our previous studies investigated the carotid and femoral atherosclerosis in ketosis-onset diabetes and found that ketosis-onset diabetes resembled non-ketotic type 2 diabetes in features of atherosclerotic lesions [
10,
11]. We also found that the frequency and risk factors of non-alcoholic fatty liver disease in ketosis-onset diabetes were close to those in non-ketotic type 2 diabetes rather than those in type 1 diabetes [
12].
HTN constituted one of the most common complications in patients with type 2 diabetes [
13], which doubled the risk of cardiovascular events and deaths compared with subjects with normal blood pressure [
14]. HTN was presented in over 50% of patients with type 2 diabetes [
15]. For example, Coats et al. [
13] reported that up to 80% of type 2 diabetes was observed to meet the HTN criteria. Furthermore, Sowers et al. [
14] indicated that the prevalence of HTN in patients with type 2 diabetes was about three times higher than in patients without diabetes.
In contrast, a lower prevalence of HTN was presented in type 1 diabetes versus type 2 diabetes. The American Diabetes Association and the European Association for the Study of Diabetes reported less than one-third of type 1 diabetic patients complicated with HTN [
16]. In EURODIAB study, the percentage of HTN was 24% in type 1 diabetes [
17]. However, the frequency and clinical characteristics of HTN in ketosis-onset diabetes remained unclear. Tan et al. [
18] evinced difference of both systolic and diastolic blood pressure among non-ketotic type 2, type 1, and ketosis-onset diabetes. However, the levels of blood pressure were observed to be comparable between ketosis-onset and non-ketotic type 2 diabetes in other studies [
19‐
21].
Furthermore, metabolic syndrome (MetS), comprised of impaired glucose regulation, HTN, dyslipidemia, and central obesity, existed in approximately two-thirds of type 2 diabetic patients or more [
22]. Our previous reports also found more MetS in type 2 diabetes [
23]. Both the PROSPER study and BRHS study indicated that MetS and its components were related to type 2 diabetes [
24]. O’Neil et al. [
25] mentioned that MetS increased the risk of type 2 diabetes fivefold. Also, type 2 diabetes was in relation to incremental levels of serum proinflammatory cytokines and non-esterified fatty acids, and reduced glucose disposal, which contributed to the development of insulin resistance, sympathetic activation, and MetS, but such inflammatory changes were mild or nonexistent in type 1 diabetes [
22,
26]. MetS was not originally introduced to identify the traits of type 1 diabetes. A previous review summarized that 8–40% type 1 diabetes were combined with MetS, lower than that in type 2 diabetes [
16]. The emergence of MetS in type 1 diabetes might be attributed to advanced age, duration of diabetes, renal function, and insulin therapy, all of which were independently relevant to MetS [
27]. However, Lontchi-Yimagou et al. [
7] mentioned that inflammatory responses between ketosis-onset diabetes and type 2 diabetes remained similar. Furthermore, the prevalence of MetS in ketosis-onset diabetes was seldom investigated, only with the description of some components of MetS.
Therefore, our primary aims are to estimate the prevalence of HTN and MetS in ketosis-onset diabetes, and to comparatively study the clinical characteristics and risk factors of HTN and MetS across type 1, ketosis-onset and non-ketotic type 2 diabetes.
Discussion
According to our surveys, HTN and MetS were frequent findings in ketosis-onset diabetes. Ketosis-onset diabetes was comparable to non-ketotic type 2 diabetes but exceeded type 1 diabetes regarding the prevalence and clinical features of HTN and MetS, supporting the categorization of ketosis-onset diabetes.
Our findings showed that ketosis-onset diabetes presented a higher predominance in men, which was reported 1:5 to 6:1 male to female ratios in the previous observations, whereas type 1 and non-ketotic type 2 diabetes were more females [
30,
31]. Men accumulated more adipose tissue in the visceral area and higher levels of androgen related with insulin-resistance and were more susceptible to glucotoxicity or glucolipotoxicity triggering unprovoked ketosis [
30]. FPG levels ranged from 6.2 to 11.5 mmol/l and 2 h PPG levels ranged from 18.3 to 20.4 mmol/l in ketosis-onset diabetes reported in the past studies [
21,
31]. Likewise, our findings showed that patients with ketosis-onset diabetes had the highest FPG and 2 h PPG compared with type 1 or type 2 diabetic controls. Beyond other types of diabetes, high levels of plasma glucose in ketosis-onset diabetes exhibited the acute blunted insulin secretion and an initial reduction in glucose disposal at the episode of ketoacidosis [
2,
32]. Moreover, ketosis-onset diabetes was observed to be related with middle age at diagnosis and obesity, which were different from type 1 diabetes. HbA1c levels in ketosis-onset diabetes were higher than in type 2 diabetes, while remained similar in contrast to those in type 1 diabetes. According to the past surveys, HbA1c levels were from 5.7 to 11.5% in ketosis-onset diabetes, from 7.8 to 10.6% in type 1 diabetes, and from 11.8 to 12.5% in type 2 diabetes, which were found no statistically difference among three groups [
1,
6,
21,
33,
34].
HTN rate in the ketosis-onset diabetes was distinct from that in the type 1 diabetes, whereas showed no difference between ketosis-onset diabetes and non-ketotic type 2 diabetes in our current study. The proportion of non-ketotic type 2 diabetics with HTN was 42.7% in our present study, close to the 50% reported in the previous study [
15]. By comparison, our research reflected that 15.7% of type 1 diabetes was hypertensive, and approximately 11–59% of type 1 diabetic patients had HTN based on different diagnostic criteria by previous investigators [
17,
35].
The frequency of HTN in ketosis-onset diabetes reported herein was 34.4%, near to that of non-ketotic type 2 diabetes. There were only two previous studies referring to the prevalence of HTN in small samples of ketosis-onset diabetes without comparison with type 1 and non-ketotic type 2 diabetes [
33,
36]. Goodstein et al. [
36] reported 25 HTN patients in 33 ketosis-onset diabetes in the veteran population. Pinero-Pilona et al. [
33] also undertook a small observational study, in which the prevalence of HTN was 29.7% in 37 ketosis-onset diabetic patients at follow-up. In our study, 279 ketosis-onset diabetics were related to 2.8-fold risk of HTN compared with the type 1 diabetic subjects, whereas with undifferentiated OR compared with the type 2 non-ketotic diabetics. It assumed that ketosis-onset diabetes might correlate with insulin resistance and sympathetic activation, which always occurred in non-ketotic type 2 diabetes, progressing to develop HTN and cardiac abnormalities such as diastolic dysfunction, macroangiopathy and left ventricular hypertrophy [
26,
37].
It was described that high levels of BMI and SUA, which was the characteristics of HTN in type 2 diabetes, was also linked to HTN in ketosis-onset diabetes. In contrast, LDL-C and WHR were at risk for HTN independently in type 1 diabetes, which was different from those of ketosis-onset diabetes. Therefore, risk factors of HTN in the ketosis-onset diabetes were close to those of the non-ketotic type 2 diabetes instead of type 1 diabetes.
On the other hand, MetS was more frequent in ketosis-onset diabetics and non-ketotic type 2 diabetics than in type 1 diabetics. The prevalence of MetS in type 2 diabetic population reached 65–75.6% [
24,
38]. In our previous observations, approximately 70% of Chinese type 2 diabetic outpatients had features of the MetS [
23]. Nonetheless, Hawa et al. [
22] found that MetS was not a characteristic of autoimmune type 1 diabetes, which had comparatively lower prevalence despite the study population and the diagnostic criteria. MetS was detected in 58.8% of the patients with ketosis-onset diabetes in our research, higher than in the type 1 diabetes (25.3%). Furthermore, Otiniano et al. [
39] reported that 74 ketosis-prone diabetic patients with MetS were inclined to have traits to type 2 diabetes.
Despite lacking direct research for MetS in ketosis-onset diabetes, there were a few studies observed that individual components of MetS were similar in patients with ketosis-onset diabetes and non-ketotic type 2 diabetes, but unlike those in type 1 diabetes [
1,
40]. As one of the key components of MetS, the average level of TG in ketosis-onset diabetes approached that of non-ketotic type 2 diabetes, in striking contrast with type 1 diabetes [
18,
41]. Moreover, TC, HDL-C, and LDL-C bore similarity between ketosis-onset diabetes and type 2 diabetes, in consistence with the past studies [
7,
20]. Additionally, the WHR values were found undifferentiated between the ketosis-onset and non-ketotic type 2 diabetes in the present research, which was in accordance with a previous study [
18,
41]. However, higher values of WHR were detected in ketosis-onset diabetics than in type 1 diabetics. Goodstein et al. [
36] showed that ketosis-onset diabetic patients accompanied by three or more metabolic-related risk variables of type 2 diabetes comprising obesity, dyslipidemia, and HTN, predicting beta-cell recovery. Bhalla et al. [
34] emphasized that ketosis-onset diabetes displayed a similar level of insulin resistance compared with type 2 diabetes. Lontchi-Yimagou et al. [
19] showed that HOMA-IR was higher in type 2 diabetes than in ketosis-onset diabetes. Of note, the insulin secretion evaluated by FCP and 2-h PCP and insulin resistance quantified by HOMA2-IR were intermediate between type 1 diabetics and non-ketotic type 2 diabetics in our research, which may occur due to remission duration after initial metabolic disturbance. As a result, ketosis-onset diabetic individuals had a higher frequency of MetS, approaching those of non-ketotic type 2 diabetes instead of type 1 diabetes subjects. Moreover, identically, risk factors of MetS in the ketosis-onset diabetes were almost in conformity with those in the non-ketotic type 2 diabetes, rather than those in the type 1 diabetes, which indirectly supported that ketosis-onset diabetes were proposed as a subtype of type 2 diabetes.
Our results had clinical and therapeutic inferences for ketosis-onset diabetes. Firstly, the prevalence of HTN and MetS were similar in participants with ketosis-onset diabetes and type 2 diabetes but was higher in both groups than in type 1 diabetic group, which elucidated that ketosis-onset diabetes tended to be classified into type 2 diabetes. Control of diet and exercise, and insulin sensitizers treatment to minimize insulin resistance and to mitigate the risk of relapse might be suitable for ketosis-onset diabetes after the initial hyperglycemia crisis. Secondly, we observed that ketosis-onset diabetes was associated with similar metabolic and hypertensive risk as type 2 diabetes. Senility, overweight, hyperuricemia, and hyperinsulinemia were the comorbidity with HTN or MetS in ketosis-onset diabetes. Therefore, early screening and stringent control of HTN, dyslipidemia, overweight, and other modifiable risk factors in MetS was necessary for estimation and reduction of cardiovascular risk in ketosis-onset diabetes.
Limitations should also be acknowledged. Firstly, potential confounding factors could affect the results, due to the single-center and observational study. However, we controlled these confounding factors as much as possible in the present analyses. Secondly, the study did not assess other diagnostic criteria of MetS. Although disparity was found in the diagnosis of criteria, the NCEP ATP III criteria were simple and appropriate to apply in Asian patients compared with the IDF criteria or the WHO criteria [
22].
Authors’ contributions
LXL and WPJ provided the hypothesis, revised the manuscript, and handled funding and supervision. JWW, APW, and MYC collected and analyzed the data. JWW and LXL drafted the manuscript. LXL, JWW, APW, JXL, and JFK participated in the revision of manuscript. For all authors, no competing or financial interests were declared. All authors read and approved the final manuscript.