Noval classification of newly diagnosed DM

1. Noval classification of diabetes
mellitus
Professor Mohammed Ahmed Bamashmos
Professor of internal medicine and endocrinology
Faculty of medicine Sanaa University

2. Ahlqvist et al. performed a data-driven cluster analysis using six variables (glutamate
decarboxylase antibody [GADA] level, age at diagnosis, body mass index (BMI),
haemoglobin A1c [A1c], and homeostasis model assessment estimates of beta-cell
function (HOMA2-B) and insulin resistance (HOMA2-IR)) at the onset of diabetes
mellitus in a Sweden and Finland cohort. They identified five exclusive subgroups [13].
Remarkable differences were reported in the cumulative incidence of diabetic
complications among the five subgroups. The reproducibility of this classification of
adult-onset diabetes mellitus has been validated by reports on American and Chinese
[14], German [15], Denmark [16] and Japanese [17] populations

3. The sub-classification based on the cluster analysis of Ahlqvist et al. [13] may have an
advantage over the others because it uses simple but pathophysiologically feasible
factors and is proven to be applicable over ethnic and racial differences to estimate
diabetic complications [14], [15], [16], [17]. The sub-classification may provide a new
framework for personalised diabetes treatment of established and/or novel
complications [18]. However, the rationale and strategy of antidiabetic treatment for
clustering-based sub-classification have not yet been fully discussed [19]. This review
aimed to inform readers about the current knowledge and perspectives on potential
therapeutic strategies based on clustering-based sub-classifications of adult-onset
diabetes mellitus.

4. Current classification of adult-onset diabetes mellitus: T1DM and
T2DM
There is an international consensus on the sub-classification of T1DM into three
subtypes [1], [2]: acute-onset [20], slowly progressive [20], [21], and fulminant [22]
depending on the mode of onset and progression. In contrast to T1DM, there is no
international consensus on the sub-classification of T2DM. Obese and non-obese
subclasses of T2DM are often used in current clinical practice. However, the
pathophysiological differences between obese and non-obese T2DM patients remain
to be elucidated. We discussed several issues on current classification of T1DM and
T2DM.

5. T1DM
T1DM occurs due to autoimmune β-cell destruction, usually leading to absolute
insulin deficiency. Environmental factors such as viral infections and genetic factors
such as the human leukocyte antigen (HLA) allele are linked to β-cell destruction.
Autoantibodies against islet antigens are often detected in the early stages of the
disease [1], [2], [3].
Representative courses of the three types of T1DM are illustrated based on disease
onset and progression [1], [20], [21], [22] (Fig. 1A).

7. Acute-onset T1DM presents as hyperglycaemic symptoms and an abolished insulin
secretory capacity, typically over several weeks to several months (Fig. 1A) [23]. Latent
autoimmune diabetes in adults (LADA) or slowly progressive type 1 insulin-dependent
diabetes mellitus (SPIDDM) is characterised by late age at onset and progressive β-cell
failure, both of which are associated with an initial non-insulin-requiring state but an
ultimate insulin-dependent state after several years [3], [22]. In fulminant T1DM, β-
cell destruction, hyperglycaemia, and ketoacidosis progress extremely rapidly, showing
no detectable insulin levels within one week [22]. Numerous acute-onset and all
LADA/SPIDDM, but not fulminant, cases feature islet autoantibodies [1], [20], [21],
[22].

8. . 1. Representative changes in plasma glucose and insulin levels during onset in type 1
diabetes mellitus (T1DM) and obese and lean type 2 diabetes mellitus (T2DM): a
hypothetical model. A. T1DM. T1DM is classified as acute, slowly progressive and fulminant
based on the manner of onset and progression [2], [20], [21], [22]. Acute T1DM (AT1DM)
typically develops hyperglycaemic symptoms along with an abolishment of the insulin
secretory capacity over several weeks to several months [23]. Latent autoimmune diabetes
in adults (LADA) or slowly progressive type 1 insulin-dependent diabetes mellitus
(SPIDDM) is characterized by late age at onset, progressive β-cell failure which is associated
with an initial non-insulin-requiring state and an ultimate insulin-dependent state over
several years, and persistent islet cell autoantibodies [21]

9. Fulminant T1DM, in which the process of β-cell destruction and the progressions of
hyperglycaemia and ketoacidosis are extremely rapid shows no detectable insulin
levels and severe hyperglycaemia within the first week [22]. B. T2DM. Although
fasting and postprandial glucose levels are altered to similar levels before and after the
onset of diabetes in lean and obese T2DM (upper panel), insulin sensitivity (dotted
line) and postprandial insulin levels (solid line) are quite different for the two
phenotypes of T2DM. In obese T2DM (red lines), insulin resistance increases
decade(s) before the onset of diabetes and fasting (not shown), and postprandial
insulin levels rise to compensate for insulin sensitivity, causing hyperinsulinemia

10. β-cell failure or β-cell loss lessens hyperinsulinemia and leads to the development of
diabetes mellitus, which gradually and progressively causes impaired insulin demand (a
relative deficit of insulin) throughout the process of diabetes mellitus spanning decades.
Obese T2DM can be classified into metabolically healthy obesity (MHO) and metabolically
unhealthy obesity (MUO) [35]. In lean T2DM (blue lines), insulin sensitivity is not severely
altered, but β-cell function is genetically low [5], [30]. As β-cell dysfunction progresses with
aging and daily glycaemic overload by consuming inappropriate foods, the intrinsic insulin
secretory capacity declines over decades, finally leading to the onset of T2DM. Of note, the
classification of T2DM into two types is convenient application for clinical use and there are
large differences in the degree of insulin resistance and diabetic complications within
obesity (MHO vs. MUO) and lean diabetes. *Lipodystrophy, a rare type of lean T2DM,
denotes a lean but insulin-resistant phenotype [24], [36],

11. T2DM
Obesity is a major risk factor for T2DM onset since it increases insulin resistance
through the accumulation of ectopic fat, such as in the liver and the skeletal muscle
[24], [25], [26], [27]. In Europe and the United States, T2DM onset and progression
have mainly been described as a model of obesity-related processes [28]. However, a
certain proportion of lean, non-obese individuals also develop T2DM [29], [30]. As
clarified in studies using cluster-derived classification, the proportion of individuals
with severe insulin-deficient and non-obese diabetes in Europe [13], the United States
[14], and Japan [17] was ∼ 18%. Therefore, T2DM onset and development in Asians
and non-Asians populations may be similarly divided into two major subtypes: obese
insulin-resistant and lean insulin-sensitive. We thus illustrated the natural courses of
hyperglycaemia, insulin sensitivity, and postprandial insulin levels in T2DM as two
typical models (lean vs obese) (Fig. 1B

12. In obese T2DM (Fig. 1B), obesity, primarily due to overeating and decreased physical
activity, lowers insulin sensitivity in proportion to visceral and ectopic fat excesses in
the liver and the skeletal muscles, increasing the insulin secretory capacity by
compensatory β-cell hypertrophy [31], [32]. However, when this hyperinsulinemic
compensation begins to fail, postprandial and fasting plasma glucose levels rise,
leading to the onset of T2DM. The insulin secretory capacity declines over decades,
whereas the low insulin sensitivity remains unchanged. Previous longitudinal studies
support this notion in obesity-related consequences for trajectories of glycaemia,
insulin sensitivity, and insulin secretion before the diagnosis of T2DM in Pima Indians
[33] and the Whitehall II study [34].

13. In contrast, there are few models of lean T2DM for the trajectories [5], [30]. In lean
T2DM (Fig. 1B), it is assumed that insulin sensitivity is generally not severely altered
but β-cell function is genetically low [5], [30]: as β-cell dysfunction progresses with
age and daily glycaemic overload by the consumption of inappropriate foods, the
intrinsic insulin secretory capacity declines over several decades, finally leading to the
onset of T2DM. This group is characterised by relative hypoinsulinemia at the onset of
T2DM, and a certain proportion of patients will require insulin treatment during the
long course of diabetes.

14. Prediabetes has heterogeneous pathogenesis and ability to predict progression to full-blown
T2DM [35]. The critical phenotype of prediabetes, non-obese or obese, enables the
stratification of cardiometabolic risk [35]. In the obesity phenotype, prediabetes may coexist
with either metabolically healthy obesity (MHO) or metabolically unhealthy obesity
(MUO) [36].
Compared to MUO, MHO has a higher capacity for storage in the subcutaneous fat. It may
explain less visceral obesity, less ectopic fat and subsequent lower insulin resistance in
MHO [36]. Such phenotypes of body fat distribution can be linked to differences in the risk
of atherosclerotic cardiovascular disease (ASCVD), dementia, and cancer between
prediabetes and diabetes with MUO and those with MHO [37

15. There are racial differences in the frequency of the obese and non-obese groups in the
development of T2DM: Asians have a higher burden of diabetes and prediabetes at lower
BMI than non-Asians [38]. There are two potential reasons for the burden of diabetes in
Asians. First, Asians have a low capacity for insulin secretion [39], [40]. In a Korean cohort
study, patients who progressed to diabetes had lower baseline insulin secretory capacity
without a compensatory increase in response to lower insulin sensitivity at diabetes onset
than non-diabetic controls [39], [40], [41]. Second, Asians are genetically predisposed to
insulin resistance because they have a greater predisposition to visceral adiposity at a lower
BMI presumably through a lower capacity for storage in the subcutaneous fat than non-
Asians [42], [43]. These may explain at least partly a much greater propensity for
cardiometabolic disorders in Asians [43].

16. Of note, the classification of T2DM into two types is convenient for clinical use (Fig.
1B). It should be emphasised that there are significant differences in the degree of
insulin resistance and diabetic complications within obese (MHO vs. MUO) and lean
diabetes. In the lipodystrophic phenotype, the risk of developing diabetes is high, even
in cases of extreme leanness [24], [36], [44]. An accumulation of ectopic fat in the
liver and the skeletal muscle caused by a low lipogenic capacity of the subcutaneous
fat results in severe insulin resistance, resembling the risk of complications in severely
obese patients with T2DM (Fig. 1B) [24], [36], [44].

17. Sub-classification of adult-onset diabetes mellitus based on
cluster analysis
discussed above, Ahlqvist et al. reported five exclusive subgroups for diabetes: cluster 1
(severe autoimmune diabetes [SAID]), cluster 2 (severe insulin-deficient diabetes [SIDD]),
cluster 3 (severe insulin-resistant diabetes [SIRD]), cluster 4 (mild obesity-related diabetes
[MOD]), and cluster 5 (mild age-related diabetes [MARD]) [13]. This classification is more
comprehensive than the classical classification, i.e., T1DM and T2DM [1], [2], because it
appraises obesity, aging, autoimmunity, insulin secretory capacity, and insulin resistance
[18]. T2DM can be classified into two subtypes in clinical practice: obese insulin resistance
and lean insulin deficiency. However, the former does not consider the concept of MHO
and MUO, while the latter inadequately predicts the risk of non-obese insulin resistance,
especially in Asians. Meanwhile, the newly identified Ahlqvist classification, which
subdivides obesity into metabolically healthy and unhealthy, allows for a better assessment
of complication risk.

18. One could claim that diabetic patients migrate between clusters with long duration of
diabetes [45]. However, it is considered that individuals are phenotyped before the
onset of diabetes and do not migrate between clusters, even with the progression of
disease duration [15], [46]. Cluster was a risk factor for diabetic complications
independent of the duration of diabetes [17]. Collectively, the cluster classification
may predict diabetic complications from the early to late phase of the diabetes time
course, independent of the duration of diabetes treatment. However, this notion
should be evaluated also in future studies.
Here we briefly discuss the clinical phenotypes of the five subgroups.

19. Cluster 1: SAID
The SAID cluster, defined by the presence of GADA, includes individuals with multiple autoantibodies.
Therefore, SAID is acute-onset or LADA [18]/SPIDDM [21], but not fulminant, which tests negative for
autoantibodies [22]. There is debate as to whether slowly progressive autoimmune diabetes with adult-onset
should be termed T1DM. However, the use of the term LADA [18]/SPIDDM [21] is common and acceptable in
clinical practice. It has the practical impact of increasing awareness of a population of adults likely to develop
overt autoimmune β-cell destruction [1]. Three distinct stages of autoimmune T1DM have been proposed by
considering methods of autoantibody detection: stage 1, autoantibodies plus normoglycaemia; stage 2,
autoantibodies plus asymptomatic dysglycaemia; and stage 3, autoantibodies plus symptomatic hyperglycaemia
[3]. The SAID nomenclature is in line with this staging to optimise the diagnosis and therapeutic strategy of
autoimmune diabetes under the rubric of T1DM. Regardless of age groups at onset and rate of progression
before symptomatic hyperglycaemia, acute-onset vs. LADA/SPIDDM, SAID is associated with low or depleted
insulin secretory capacity, low BMI, and poor glycaemic control (high A1c level) [18]; thus, there is a risk of
ketoacidosis [47]. SAID features the highest prevalence and incidence of diabetic retinopathy [17], [18] and a
risk of fracture [48].

20. Cluster 2: SIDD
The SIDD cluster shows clinical profiles similar to those of SAID except for GADA
positivity [18]. SIDD features a mildly to severely low insulin secretory capacity and low
BMI and develops at a relatively older age than SAID. There is also poor glycaemic control
(high A1c level) [18] and, thus, the risk of ketoacidosis [47] although its frequency is lower
than that in SAID. SIDD has a higher prevalence and incidence of diabetic retinopathy and
neuropathy [17], [18] and may confer a fracture risk [48]. There are conflicting reports on
whether the frequency of SIDD differs among Caucasians [13], [14], [15], South Asians
[49], and East Asians [17], [50]; this topic requires future evaluation. Ketosis-prone diabetes
is a unique form of diabetes mellitus, namely intermediate between T1DM and T2DM, and
may be related to SIDD [51

21. Cluster 3: SIRD
The SIRD cluster shows obesity and moderate to severe insulin resistance with
hyperinsulinemia. Glycaemic control is relatively mild (low A1c level). Under severe
metabolic deterioration based on insulin resistance through ectopic fat accumulation
[24], [25], [26], [27], individuals with SIRD are at high risk of developing ASCVD.
SIRD confers the highest risk of developing DKD, commonly in Caucasians [18] and
non-Caucasians [17]. Since insulin resistance and hyperinsulinemia are associated with
the onset and progression of cancer and dementia (Alzheimer's disease) [52], [53],
SIRD may also carry the risk of cancer and dementia. The frequency of this cluster is
not consistent within Asians, as it was slightly lower in Japan [17] but higher in India
[49] and China [50].

22. Cluster 4: MOD
The MOD cluster shows a high BMI and mild to moderate insulin resistance. Obesity
may be linked to the onset of MOD to a lesser extent than SIRD. Glycaemic control is
relatively good, and somewhat fewer micro- and macrovascular complications are
involved than with other clusters. MOD resembles MHO [54], [55] and has a lower
prevalence of obesity-related complications such as hypertension, dyslipidaemia, and
NAFLD [56] than SIRD.

23. Cluster 5: MARD
MARD has elderly-onset diabetes but no other clinical characteristics. This cluster is
the largest among Caucasians [13], [15] and Asians [17], [50]. Age-related decreases in
insulin secretory capacity and insulin sensitivity, which are generally caused by
changes in body composition with decreased muscle mass and relatively increased fat
mass [57], are the suspected main causes of MARD. Glycaemic control is good, and
there were few microangiopathies. In elderly MARD individuals, attention should be
paid to senile syndromes such as frailty syndrome, mild cognitive impairment,
osteoporosis and fracture, and ASCVD [45].

24. Therapeutic strategies for managing glycaemic control and
preventing diabetic complications
The basic strategies for glycaemic control in adult-onset diabetes mellitus depend on
the type, condition, age, and degree of complications of diabetes and metabolic
disorders [7], [8], [9], [10]. In the current guidelines, metformin and comprehensive
lifestyle managements are recommended as the first-line therapy for T2DM [7], [8],
[58]. The choice of antidiabetic drug classes is based on ASCVD, hypoglycaemic risk,
weight effects, cost, and patient preferences [7], [8], [9], [58]. This recommendation is
mainly based on evidence from clinical trials, but these statements do not fully
consider the patients’ pathophysiological background. In other words, currently
proposed therapeutic strategies remain limited as a regimen tailored to the pathology
of T2DM [59] in which responsiveness to the different classes of antidiabetic agents
varies from person to person.

25. In addition to the above evidence-based approach, a therapeutic strategy based on the
pathophysiological background is intriguing. Here we proposed a five-step approach of
non-pharmacological and pharmacological treatments based on the pathophysiology
of obesity-related insulin resistance or low insulin secretory capacity in adult-onset
diabetes mellitus (Fig. 3). To pursue a more integrated strategy, we performed two-
dimensional phenotypic mapping of diabetic complications and commodities for the
five clusters (Fig. 4). Here we discuss its possible therapeutic utilities.

27. Five-step approach for the non-pharmacological and pharmacological treatment of adult-
onset diabetes mellitus based on the pathophysiology of obesity-related insulin resistance or
low insulin secretory capacity. Adult-onset diabetes mellitus is classified into types with
obesity-related insulin resistance or with low insulin secretory capacity. Obesity-related
insulin resistance is strongly linked to visceral fat obesity and ectopic fat disposition [44],
[63]. Free fatty acids in visceral fat are delivered to remote organs such as the pancreas [32],
[44], liver, skeletal muscle, and kidney and are a crucial fuel for satisfying metabolic
demands, but they are accumulated when delivered more than required (ectopic fat
accumulation) [44], [63]. Ectopic fat accumulation is a main cause of insulin resistance, a
phenomenon known as lipotoxicity [

28. induces organ dysfunction and complications in the β-cells (by pancreas fat) [31],
[32], the liver (by liver fat) [90], the skeletal muscle (by muscle fat) [25], [26], [27],
the kidney (by kidney fat) [100] and the cardiovascular system (by cardiac fat:
epicardial and perivascular fat) [24]. The red colour indicates the accumulations of
ectopic fat and visceral fat. We introduce a simple method for assessing
pathophysiology in treating patients with diabetes based on the presence and absence
of accumulation of visceral fat and ectopic fat. Step 1: estimate the pathophysiological
conditions (yes or no visceral fat and ectopic fat); in patients with no ectopic and
visceral fat, consider insulin deficiency caused by β-cell failure (blue colour);

29. Step 2: determine optimal A1c targets according to the guidelines [7], [8], [9], [10];
Step 3: set optimal body weight targets. For the body weight target, it is crucial to
assess the degree of visceral and ectopic fat accumulation, not the total body weight
which includes lean body mass such as the skeletal muscle. When visceral and ectopic
fat accumulation is linked to insulin resistance and hyperglycaemia, it should be
reduced. When visceral and ectopic fat is not accumulated, current body weight should
not be reduced;

30. Step 4: prescribe diet and exercise to achieve optimal A1c and body weight targets;
Step 5: Select the optimal class of antidiabetic agents when optimal A1c targets are not
achieved in Step 3. While choosing an optimal class of antidiabetic agents, target
organs (β-cells vs. the liver, the skeletal muscle, and visceral and ectopic fat) should be
considered. The blue box represents the classes of drugs that act on β-cells and
stimulate insulin secretion or compensate for insulin demands. The red box represents
the classes of drugs linked to reduction or modulations in visceral and ectopic fat. FFA:
free fatty acid, SGLT2i: sodium glucose cotransporter 2 inhibitors, TZD:
thiazolidinedione, GLP1RA: GLP1 receptor agonist, DPP4i: dipeptidyl peptidase-4
inhibitor, α-Gi: α-glucosidase inhibitor, TG: triglycerides, HDL-C: high density
lipoprotein cholesterol, DKD: diabetic kidney disease, ASCVD: atherosclerotic
cardiovascular disease, AF: atrial fibrillation. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article

31. Step 4: prescribe diet and exercise to achieve optimal A1c and body weight targets;
Step 5: Select the optimal class of antidiabetic agents when optimal A1c targets are not
achieved in Step 3. While choosing an optimal class of antidiabetic agents, target
organs (β-cells vs. the liver, the skeletal muscle, and visceral and ectopic fat) should be
considered. The blue box represents the classes of drugs that act on β-cells and
stimulate insulin secretion or compensate for insulin demands. The red box represents
the classes of drugs linked to reduction or modulations in visceral and ectopic fat. FFA:
free fatty acid, SGLT2i: sodium glucose cotransporter 2 inhibitors, TZD:
thiazolidinedione, GLP1RA: GLP1 receptor agonist, DPP4i: dipeptidyl peptidase-4
inhibitor, α-Gi: α-glucosidase inhibitor, TG: triglycerides, HDL-C: high density
lipoprotein cholesterol, DKD: diabetic kidney disease, ASCVD: atherosclerotic
cardiovascular disease, AF: atrial fibrillation. (For interpretation of the references to
colour in this figure legend, the reader is referred to the web version of this article

33. Two-dimensional phenotypic mapping for diabetic complications and commodities and potential
therapeutic strategies among the five clusters of adult-onset diabetes mellitus. According to the cluster
analysis, adult-onset diabetes is categorized as severe autoimmune diabetes (SAID, cluster 1), severe
insulin-deficient diabetes (SIDD, cluster 2), severe insulin-resistant diabetes (SIRD, cluster 3), mild
obesity-related diabetes (MOD, cluster 4), and mild age-related diabetes (MARD, cluster 5). The five
clusters are then placed on the map with the two axes representing insulin secretory capacity and
insulin resistance. The characteristic phenotypes of diabetic complications including retinopathy,
diabetic kidney disease (DKD) and neuropathy and ketoacidosis, and the comorbidities including non-
alcoholic fatty liver disease (NAFLD)/non-alcoholic steatohepatitis (NASH), atherosclerotic
cardiovascular disease (ASCVD)/heart failure/atrial fibrillation (AF), frail, fracture, and mild cognitive
impairment (MCI) are shown. Therapeutic issues, including glycaemic control and other risk factors
and related treatment procedures are also placed. SGLT2i: sodium glucose cotransporter 2 inhibitors,
TZD: thiazolidinedione, GLP1RA: GL