Data available on genetic susceptibility suggest that LADA shows a lower genetic component [10,11,33] than T1DM. In particular, the human leukocyte antigen (HLA)-DRB1^*04-DQB1^*0302 and HLA-DRB1^*0301-DQB1^*0201, which are very common in young-onset T1DM and decrease in frequency with the increasing age at disease onset, are less frequent in LADA than in adult-onset T1DM [25,34]. Similar findings have been observed for the Cyst1858Thr single-nucleotide polymorphism in the protein tyrosine phosphatase nonreceptor 22 (PTPN22) gene and the insulin variable number tandem repeat (INS VNTR) I/I genotype expression [34,35]. Otherwise, the frequency of the cytotoxic T-lymphocyte-associated antigen 4 (CTLA4) Ala49Gly polymorphism in exon 1 has been shown not to be associated with the age of onset of T1DM, suggesting a similar role in LADA susceptibility [36].

Another study carried out in Swedish and Finnish populations, showed that the frequency of T2DM associated CT/TT genotypes rs7903146 in the transcription factor 7 like 2 (TCF7L2) gene was increased in LADA subjects as in T2DM subjects [37], as well as genetic similarities with T1DM have been observed related to HLA, INS VNTR, and PTPN22 [38]. These results suggest that patients with LADA may share genetic features with both T1DM and T2DM which further supports the concept that LADA is an admixture of the two major types of diabetes [38].

However, it should be noted that most of these studies have investigated genes previously associated with young-onset T1DM or T2DM [10,39,40,41,42]. Further studies are needed to better characterize the genetic susceptibility of LADA in order to establish preventive strategies as well as allow safe and effective therapies [43].

In T1DM activated T-cells to islet-cell proteins characterize the lymphomononuclear cell infiltration in the endocrine pancreas (insulitis), causing the progressive loss of insulin-secreting β-cells. On the other hand, islet cell autoantibodies seem to be an epiphenomenon rather than pathogenic factors in β-cells destruction, though their detection in plasma helps to diagnose autoimmune diabetes.

As a form of autoimmune diabetes, LADA is characterized by islet-cell specific autoantibody positivity and similar cell-mediated immune response although impairment of β-cells is slower than in classical T1DM [6,7,8]. Specifically, a recent study observed presence of insulitis by pancreatic scintigraphy using interleukin 2 (IL-2) radiolabelled with technetium-99m (^99mTc) and contrast-enhanced magnetic resonance imaging, providing evidence for the presence of activated mononuclear cells in the pancreas of LADA subjects, particularly within 1 year of initiation of insulin therapy, similar to T1DM at diagnosis [44]. Moreover, increased pancreatic ^99mTC-IL-2 uptake has been found in a subgroup of autoantibody-negative diabetic subjects, suggesting that insulitis may occur even in absence of islet-cell autoantibodies [45].

C-peptide, a marker of residual β-cell function, declines slower in LADA than in T1DM. Thus, a positive correlation between age at diagnosis of autoimmune diabetes and fasting C-peptide levels has been reported [46]. Indeed, LADA subjects show higher stimulated C-peptide values al all time-points following a mixed-meal tolerance than subjects with classical T1DM [12].

Islet-cell autoantibodies detectable in subjects with LADA are the same used to identify T1DM. Glutamic acid decarboxylase autoantibody (GADAs) is the most sensitive marker in both adult-onset T1DM and LADA, whereas insulin autoantibodies, protein tyrosine phosphatase IA-2 (IA-2A) and islet-specific zinc transporter isoform 8 (ZnT8) autoantibodies, which are detectable in younger T1DM patients, are positive only in a small percentage of LADA patients [47]. In the Action LADA study, 68.6% of subjects screened for diabetes-associated autoantibodies were positive for GADA only, 5% positive for IA-2A only and 2.3% positive for ZnT8A only, whereas at least two autoantibody types were found in 24.1% patients [4].

To sum up and as previously mentioned, LADA shows halfway pathological features between T1DM and T2DM. In a recent Italian work, it has been suggested that different pathophysiological could explain the heterogeneous phenotypes of LADA [13]. Based on this model, in patients with moderate genetic susceptibility to T1DM, specific immunological factors can trigger an autoimmune process against islet cell antigens marked by the appearance of GADAs leading to β-cell apoptosis and insulin deficiency. On the other hand, in obese subjects with genetic susceptibility to T2DM, the low-grade inflammation, typical of visceral adiposity, might trigger a low-grade autoimmune process marked by IA-2 autoantibodies positivity, causing loss of β-cell function and an impairment of insulin secretion (Fig. 2).

LADA shares clinical and metabolic features with T2DM and T1DM as part of a continuum of variable severity of immune and metabolic dysfunction [48], described as “end of the rainbow” (Tables 2, 3) [8].

Metabolic syndrome and its components (blood pressure, lipid profile, waist-hip ration) are less prevalent in LADA patients than in those with T2DM, both in Caucasians and non-Caucasians subjects [5,16]. Nevertheless, this assumption changes when LADA patients are compared with patients with T1DM. The BOTNIA study [22] demonstrated that components of metabolic syndrome have a higher prevalence in LADA than in classical T1DM. Similar results emerge from the Action LADA project [49]. Another study, carried out in Spain, showed that adiposity parameters, blood pressure and triglyceride levels are more elevated in LADA than in T1DM but lower than in T2DM [50]. Similarly, data collected among a cohort of subjects with LADA in the United Arab Emirates showed higher body mass index (BMI), waist circumference, systolic blood pressure and glycated hemoglobin (HbA1c) values [5]. On the other hand, Swedish and Norwegian registers report that overweight and obesity are associated with increased risk of LADA [37], particularly in combination with family history for diabetes. However, as previously discussed, LADA is characterized by a great phenotypic heterogeneity, witnessed by variable degrees of insulin resistance and insulin insufficiency.

Recent evidence shows that the titre of antibody positivity, especially GADA, can influence clinical characteristics of subjects with LADA [14,25,28]. The NIRAD study [25] demonstrated the presence of a “bimodal distribution” of GADA titre in LADA sub-classifying two distinct forms of the disease. Patients with high GADA levels, typed as LADA 1, show more similar phenotypic features with T1DM (lower C-peptide levels, lower BMI, more ketosis prone) than with T2DM, whereas patients with low GADA titre, typed as LADA 2, are less ketosis and dyslipidemia prone as compared with LADA 1, but possess higher frequency of obesity, hypertension, dyslipidemia, and cardiovascular diseases. Similar results have been found in a Korean population where GADA levels showed a negative correlation with age of onset, total cholesterol, triglycerides, BMI, fasting and stimulated C-peptide, whereas a positive correlation was found with HbA1c and high density lipoprotein cholesterol [15]. The existence of two distinct subgroups of patients affected by LADA has been substantiated by genetic studies. PTPN22 risk genotype has been associated with a high GADA titre [51], whereas some but not all studies, have found an association between a lower GADA titre and the TCF7L2 risk allele for T2DM [52]. In addition, a higher frequency of other organ-specific autoantibodies (against thyroid peroxidase, steroid 21-hydroxylase, tissue, tissue transglutaminase, and parietal cell) was detected in patients with high GADA levels, highlighting a more extensive autoimmune process in this subpopulation than in those with a lower GADA titre [53]. Accordingly, several studies have testified an inverse relationship between GADA titre and C-peptide levels [54], as well as more pronounced traits of insulin deficiency were described in patients with higher GADA titre [25]. Another interesting finding concerns the relationship between clinical phenotypes of LADA and specific GADA binding patterns. Action LADA 12 [55] demonstrated an association between presence of N-terminally truncated GAD65 autoantibodies and the clinical phenotype of T1DM in patients with LADA. This may have important practical implications given the relatively large number of adult patients who are screened for GADA, both for diabetes classification and risk prediction for insulin therapy.

Even the number of islet autoantibodies in LADA appears to be associated with the intensity of autoimmune response. Indeed, the United Kingdom Prospective Diabetes Study (UKPDS) group and the NIRAD group, have noted that the number of islet cell autoantibodies was directly proportional to the intensity of autoimmune response, predicting more sever insulin insufficiency [21,47,56].

Not only the titre, but also different patterns of antibody positivity could influence clinical features of LADA patients. In a cohort of approximately 17,000 subjects affected by LADA, the positivity for IA-2A only was associated with a clinical phenotype more similar to T2DM, whereas subjects positive for both IA-2A and GADA were more similar to T1DM [5]. Other authors observed that IA-2A was the only autoantibody that showed increased frequency with increasing BMI in patients affected by T2DM [57]. However, conflicting data have been reported. In the Action LADA 7 patients positive for at least two autoantibodies compared with those positive for a single antibody did not show relevant differences in demographic or clinical parameters [4]. Moreover, mixed results about the correlation between GADA titre and length of insulin-free period have been reported [14,22,54,58].

There are only few studies related to the occurrence of microvascular complications (nephropathy, retinopathy, neuropathy) in LADA and controversial results have been reported, partly due to a substantial heterogeneity regarding disease duration of study's subjects. Limited to patients with a short disease duration, microvascular complications in LADA appear to be less frequent than in patients affected by T2DM. In particular, the Fremantle Diabetes Study [59], showed less frequency of microalbuminuria in patients with a recent diagnosis of LADA than in those with T2DM. In addition, in the same study an association between GADA positivity and a reduced risk of developing microalbuminuria during a follow-up period (<5 years) was found. In a study among Chinese subjects [60], it was observed that the onset of nephropathy and retinopathy is less common in patients with a recent diagnosis (<5 years) of LADA than in those affected by T2DM. However, these findings could be explained by the fact that subjects with T2DM are usually diagnosed later than subjects with LADA, being chronically expose to hyperglycaemia which may ultimately lead to microvascular complications. Thus, studies evaluating patients with a disease duration >5 years did not find differences between LADA and T2DM with respect to the prevalence of nephropathy or retinopathy, as well as no association was found between GADA positivity and rate of microvascular complications [61,62,63]. However, the BOTNIA study [64] observed that patients with a disease duration >5 years had an increased risk of retinopathy; in addition, the rate of neuropathy appears more prevalent in LADA than in T2DM [59,62,64,65].

A lower risk of macrovascular complications—including coronary heart disease, stroke, peripheral artery disease—could be postulated on the basis of the healthier metabolic profile of patients with LADA respect to those with T2DM. However, current data showed similar cardiovascular outcomes in LADA and T2DM. In the BOTNIA study [64] no statistically significant differences were found in respect to coronary heart disease, stroke, overall mortality, and cardiovascular mortality between LADA and T2DM. Similar results emerged from other independent studies. The Fremantle Diabetes Study [59] did not evidenced different rate of cardiovascular disease and mortality in LADA versus T2DM, while the HUNT2 study observed similar increase of cardiovascular risk in patients positive for GAD antibodies and in those who were negative, compared to non-diabetic patients [66]. Altogether, these results suggest that different pathogenetic mechanisms might modulate the onset of macrovascular complications in LADA, regardless of the metabolic profile.

Impaired bone metabolism and increased risk of fractures have been observed in both T1DM and T2DM [67], partly due to reduction of bone formation. A recent cross-sectional study [68], showed that bone resorption is reduced in both LADA and T2DM compared to non-diabetic subjects whereas circulating sclerostin—which is an antagonist of the osteoblastic bone formation—is increased in T2DM only, probably due to the association between sclerostin levels and metabolic syndrome which is more prevalent in T2DM than in LADA. However, these data suggest that pathways involved in bone metabolism differ between these two types of diabetes and further studies are needed to clarify the pathogenesis of bone impairment and explore the potential role of sclerostin on bone fragility associated with diabetes.

A small randomized controlled trial, carried out in Japan, compared glibenclamide with insulin treatment in subjects affected by LADA. Results showed worsening metabolic control and progressive deterioration of β-cell function as measured by significant reduction of stimulated C-peptide ratio after 30 months follow-up in patients treated with sulphonylurea [71]. Similar results have been reported in the Tokyo study [72], a multicenter, randomized, nonblinded clinical study that evaluated 4,089 noninsulin-dependent LADA patients with a 5-year or shorter duration of diabetes. In this study Maruyama et al. [72] demonstrated that the progression rate to insulin-dependency in the sulphonylurea group was higher than that in the insulin group, and C-peptide during the oral glucose tolerance test was more preserved in patients undergoing insulin therapy. Altogether, these data suggest that sulphonylurea should not be used as first-line therapy in patients with LADA.

There insulin treatment is essential in all patients with complete β-cell loss and represents the most straightforward therapy for replacing missing endogenous insulin secretion [13]. However, subjects with a recent diagnosis of LADA are characterized by some degree of preservation of β-cell function as shown by higher C-peptide levels, and progress slower to absolute insulin dependency. Nevertheless, consistent data from randomized clinical trials highlight the importance of an early initiation of insulin therapy in LADA regardless of presence of some endogenous insulin secretion. The rationale behind this approach is to improve metabolic control while protecting β-cell function.

Even though the pathophysiological basis of the protective effect of insulin therapy is not well known, several preclinical studies showed that therapy with exogenous insulin would support β-cell function, reducing the hyperglycaemic stress [73] and decreasing severity of insulitis [74]. Other studies suggest the promotion of T helper 2 immunity and activation of insulin-specific regulatory T-cells mediated by the exposure to exogenous insulin [75,76]. Moreover, a suppression of autoreactive T-cells through local release of regulatory cytokines has been suggested in patients treated with insulin [77].

An intervention intended to preserve β-cell function should be pursued in patients with LADA. Recent immune-intervention trials have achieved promising results in term of preserving stimulated C-peptide levels and improving glycaemic control [78,79,80]. However, some drugs currently used for treatment of T2DM might be considered in LADA.

Dipeptidyl peptidase 4 (DPP-4) inhibitors represent a class of oral antidiabetic agents frequently used in T2DM which have been shown to preserve β-cell function and reduce insulitis in patients with T2DM as well as in mouse models of autoimmune diabetes [81,82,83], suggesting that they might be a valuable treatment option in LADA. A recent randomized-controlled study conducted in China [84] has observed that treatment with sitagliptin in addition to insulin preserved C-peptide concentration better than insulin alone in patients with LADA over a 1-year period. Similarly, sitagliptin improved glycaemic control in adults with T1DM [85]. Furthermore, Johansen et al. [86] have reported that another DPP-4 inhibitor, linagliptin, attenuated decline of C-peptide in LADA patients over a 2-year study period. In a post hoc analysis of data pooled from five randomized, placebo-controlled studies [87], saxagliptin was effective in lowering blood glucose levels and well tolerated in GADA-positive patients. Moreover, compared with placebo saxagliptin appeared to increase C-peptide levels at 24 weeks follow-up.

Other interesting findings come from our study of a post hoc analysis investigating treatment with dulaglutide, a glucagon-like peptide 1 receptor agonist (GLP-1RA) [88] in patients with T2DM among whom there were some GAD antibody positive patients. The effectiveness of dulaglutide in subjects with LADA was indicated by reduction of HbA1c and increase of β-cell function, without affecting the rate of hypoglycaemia over 1-year observation period. This study is the first to indicate that dulaglutide is an effective anti-hyperglycaemic treatment in T2DM patients positive for GAD antibodies. Taking together these data highlight the potential long-term effects of DPP-4 and GLP-1RA in patients with LADA. However, further studies with larger cohort and longer treatment duration are needed to assess whether these therapies translate into a reduced progression to insulin dependence and diabetic complications [87].