Type 2 diabetes mellitus (T2DM) is considered a progressive metabolic disorder caused by two major pathophysiological defects: insulin resistance and β-cell dysfunction [1]. Large-scale, randomized controlled trials have proved that intensive glucose control decreases the risk of microvascular complications [2,3], and even the risks of cardiovascular events or mortality, as shown in long-term follow-up studies [4,5]. However, mainly because of progressive decline in β-cell function and clinical inertia, long-term durable glycemic control remains a difficult goal to attain in the management of T2DM [6,7,8].

During past decades, the development of novel glucose-lowering agents and strategies has demonstrated more positive impact on glycemic durability than previously observed. A report from the Swedish National Diabetes Register demonstrated that, in real clinical practice, metformin resulted in superior glycemic durability than sulfonylureas or meglitinides [9]. Similarly, in the A Diabetes Outcome Progression Trial (ADOPT) study, rosiglitazone proved to be a better choice in achieving glycemic durability than either sulfonylureas or metformin [6]. Among recently developed novel agents, some, but not all, sodium glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists have shown more tolerable glycemic control than metformin and sulfonylureas [10,11]. Results of several trials have suggested that an early intensive insulin therapy can even lead to a complete disease resolution in patients with new-onset T2DM and help attain glycemic durability [12,13]. However, these results are based on tightly regulated clinical trials comprising highly selective patients, which limit their applicability in real clinical practice. In fact, other studies have noted patient-related factors or characteristics other than the use of specific glucose-lowering agents, such as tolerability, patient acceptance, and costs, that can affect durable glycemic control [14,15]. On the other hand, current treatment guidelines and recommendations have reached a consensus that tight glycemic control is more beneficial to patients with short-duration diabetes or those who are free of related complications [16].

Nonetheless, despite improved understanding of inter-patient differences affecting responses to therapy, little information is available on factors that are strongly associated with durable glycemic control in real clinical settings. Therefore, we aimed to assess the major determinants of durable glycemic control in new-onset T2DM using 2 years of observational data.

Subjects were divided into the durability group (n=114) and the non-durability group (n=80). Fig. 1 and Supplementary Fig. 1 shows the changes in the mean HbA1c levels in both groups during the 2-year follow-up period. The baseline HbA1c level was significantly higher in the non-durability group than in the durability group (mean, 9.7% vs. 8.6%). Three months after diagnosis, the difference in HbA1c between groups was 0.6%, however, the gap increased during folllow-up, producing ~1.3% difference in the mean HbA1c level 2 years after diagnosis (7.5% vs. 6.2%, P<0.01).

The baseline patient characteristics are described in Table 1. Subjects in the durability group had a higher BMI (26.1 kg/m² vs. 24.9 kg/m², P=0.021), and waist circumference (90.1 cm vs. 85.3 cm, P<0.001) than those in the non-durability group. Although the lifestyle factors were comparable between the groups, a comparison of the level of physical activity indicated that subjects in the durability group were more physically active. The durability group also had higher HOMA-β levels than the non-durability group, although the HOMA-IR levels were comparable between the two groups.

Metformin was the most frequently prescribed initial glucose-lowering agent in both groups, followed by sulfonylureas. The difference in treatment regimen between the two groups was the use of insulin and sulfonylureas, which were used more frequently in the non-durability group than in the durability group (Table 1). The usage frequencies of each prescription glucose-lowering agent during the entire follow-up period are displayed in Supplementary Table 1.

This study showed that some clinical factors contribute to the maintenance of durable glycemic control in new-onset T2DM. HbA1c levels at the time of diagnosis, and at the first follow-up visit were important determinants of glycemic durability. In addition, higher levels of education, physical activity, and HOMA-β were associated with durable glycemic control during a 2-year follow-up period.

HbA1c is the most widely used marker of glycemic control, and reflects overall glycemic exposure for the previous 2 to 3 months [18]. A higher baseline HbA1c level in the non-durability group suggests long-term exposure to hyperglycemia before the diagnosis. Therefore, we assume that a higher glucotoxicity in the non-durability group may have impaired both insulin secretion and activity [19]; thereby, lowering the possibility of maintaining favorable glycemic control in these patients. A significantly lower HOMA-β level in the non-durability group further supported this explanation.

Interestingly, we observed that a lower HbA1c level at the first follow-up visit was a strong indicator of glycemic durability. The significance of this association was maintained even after adjusting for baseline HbA1c levels and the type of glucose-lowering agents. Two possible explanations, with divergent mechanisms, should be considered for this observation. First, a rapid correction of hyperglycemia may be responsible for the glycemic durability in patients with new-onset T2DM. In this context, several studies have reported that early and intensive therapy produced favorable outcomes, i.e., recovery and maintenance of β-cell function, in drug-naïve patients at the onset of T2DM. In a meta-analysis to assess the efficacy of intensive insulin therapy, approximately 42.1% of patients experienced drug-free remission of diabetes during a 2-year period with only 2 to 3 weeks of intensive insulin therapy at the onset of diabetes [12]. Intensive insulin therapy provides rest to the β-cells by decreasing the hepatic glucose production, and by reducing glucotoxicity and lipotoxicity [13]. Our results, although retrospectively analyzed, expanded on the concept that a rapid correction of hyperglycemia, with or without insulin, may aid in the long-term maintenance of hyperglycemia within the target range. Second, patients in the durability group may be “good responders” to the glucose-lowering treatments. When the analysis was done among patients who attained HbA1c <7.0% at 6 months, we also observed similar findings (Supplementary Tables 2 and 3). This may partly be due to relatively short-term exposure to hyperglycemia or due to their having genetically healthier β-cells than patients in the non-durability group. However, individual differences that may affect therapeutic responses remain unclear. More targeted studies will be required to recognize the degree to which alterations in specific aspects of glucose homeostasis will differ between individuals, and how an individual will respond to a specific medication in a real clinical setting [20].

Our results further demonstrated that T2DM patients' educational and physical activity levels were associated with long-term glycemic durability. Several studies have identified patient-specific factors that may influence the durability of glycemic responses, emphasizing the need to personalize therapies based on patient characteristics. In a Japanese study, the avoidance of weight gain contributed to the maintenance of better glycemic control [21]. In another study by Mamza et al. [7], female gender, smoking, longer duration of diabetes, and a higher baseline HbA1c level were associated with a poor durability of therapeutic efficacy.

Patients' education level and physical activity are factors that need to be considered as seriously as socioeconomic status and patient attitude [22,23]. Patients with low socioeconomic status are more exposed to unhealthy lifestyles and adverse environmental factors such as obesity, physical inactivity, and smoking, as they are less likely to avail themselves of routine health checkups and health education compared to those with high socioeconomic status [24]. The European Diabetes (EURODIAB) Prospective Complications study reported that a healthy lifestyle was more prevalent among better-educated men and women with diabetes [22]. A previous study noted that exercise training, which promotes a higher physical activity level, could directly reduce HbA1c level in addition to reducing the risk of developing diabetic complications [25]. Another report identified patient adherence as an important determinant of response to treatment for T2DM [26]. Therefore, patients' attitude to treatment and their educational level should also be considered as predictive indicators of response to treatment and long-term glycemic control.

There are several limitations to this study. First, some factors varied because they were controlled by clinicians. Physicians in this study generally followed the current ADA/EASD guideline that recommends the following: HbA1c <7.0% as the glycemic target, metformin as the initial glucose-lowering agent, and lifestyle modifications. However, the glycemic targets were set individually, and the choice of glucose-lowering agent was solely dependent on the clinician's discretion. This is an inevitable limitation of a retrospective study design. Second, because we included only those patients who attended follow-up visits for at least 2 years, all our study subjects demonstrated good adherence to treatment, which may limit the applicability of our results to a broader patient spectrum.

In conclusion, our findings will help illuminate inter-individual differences in responses to therapy by providing evidence for various factors that can affect the durability of glycemic control. Above all, an early treatment response in terms of glycemic control was an important predictor of continuing durable glycemic control. Further well-controlled trials will be needed to confirm this hypothesis.