It is not always easy to separate data from patients with long-standing T2DM from that of T2DM in older patients, but the Diabetes in Aging Study showed that in long-standing T2DM, coronary artery disease and hypoglycemia were the most common nonfatal comorbidities (18.98 per 1000 person-years and 15.88 per 1000 person-years, respectively).⁷ Setting appropriate treatment goals that take prevalent comorbidities into consideration is the first step in the successful management of T2DM. These goals will likely change over time and with progression of the disease. 

Metformin is typically the first line of drug therapy for patients with T2DM. However, the choice of therapeutic agents after metformin becomes more complicated for older patients, and current treatment recommendations do not explicitly discuss this difficulty. Older patients with T2DM are a very heterogeneous group with multiple comorbidities, increased risk of hypoglycemia, and greater susceptibility to adverse effects of antihyperglycemic drugs, making their treatment particularly challenging. Ideally, preferred drugs in older patients should not be cleared by the kidneys and should have a low risk of adverse cardiovascular effects or hypoglycemia. Generally, metformin, incretin-based therapies (ie, glucagon-like pepetide-1 [GLP-1] agonists and dipeptidyl peptidase-4 [DPP-4] inhibitors), and long-acting insulin should be preferred over other treatment options (sulfonylureas, glinides, and glitazones). 

Type 2 diabetes mellitus is a progressive disease characterized by insulin resistance, β-cell dysfunction, and an impaired incretin effect. Chronic hyperglycemia and elevated lipid levels cause glucolipotoxicity, which in turn leads to further β-cell dysfunction and failure. As a result, exogenous insulin is eventually needed in patients with T2DM of long duration. 

The landmark United Kingdom Prospective Diabetes Study (UKPDS) showed the natural progression of T2DM and decline in β-cell function, which was already at 50% by time of diagnosis and was virtually nonexistent by 12 years of known T2DM.⁸ After 9 years, only 25% of patients in UKPDS had good glycemic control on monotherapy and lifestyle measures, and many of the patients in the study required insulin therapy.⁹ 

Insulin therapy, particularly basal insulin analog therapy, is relatively simple to start with gradual up-titration by the physician or the patient.¹⁰ Basal insulin analogs are associated with less nocturnal hypoglycemia compared with neutral protamine Hagedorn (NPH) insulin.¹¹ Longer-acting insulin analogs^12-14 and more concentrated insulin analogs¹⁵ currently in development are associated with even lower risk of nocturnal hypoglycemia. 

Figure 1 and Table 1¹⁰ provide insights into introducing patients to the concept of insulin therapy and having patients self-inject during the office visit while discussing initial dosing of basal insulin. The initial dose may be a standard 10 units or be weight-based, as suggested by the American Association of Clinical Endocrinologists (AACE) on prevailing glycated hemoglobin A_1c (HbA_1c) levels (for HbA_1c ⩽8%, a total daily dose of 0.1-0.2 U/kg; for HbA_1c ⩾8%, a total daily dose of 0.2-0.3 U/kg), and suggestions for self-titration algorithm and dose adjustment until fasting plasma glucose reaches the target set for each patient.^16-18 Multiple self-titration schedules are available in the published literature that can be considered for individual patients.^10-22 Although the self-titration algorithms differ, the principle remains the same, which is providing a tool to each patient for self-adjustment of the insulin dose that meets their individual needs. 

Sulfonylureas rely on endogenous β-cell function for their action as insulin secretagogues; thus, their usefulness in managing T2DM of long duration is limited. This is borne out by the A Diabetes Outcome Progression Trial, in which patients' HbA_1c levels rose more rapidly over the years (loss of control) with sulfonylureas than with metformin, thiazolidinediones, or insulin (Figure 2).^9,23,24 The other issue with sulfonylureas is the risk of hypoglycemia, which increases as patient β-cell reserve decreases. One should not choose a medication on the basis of acquisition costs alone, as the costs of managing hypoglycemia²⁵ are great and the adverse impact of hypoglycemia on patient's future adherence to medication^25,26 is difficult to quantify. 

Hypoglycemia is common not only in type 1 diabetes mellitus but also in T2DM. It is underrecognized and undertreated. The fear of hypoglycemia contributes to provider clinical inertia and can lead to patient defensive eating and medication nonadherence. It has also been implicated as a cause in cardiovascular complications of T2DM. The risk of hypoglycemia is greater in T2DM of long duration and in elderly patients. Therefore, when contemplating insulin intensification strategies beyond basal insulin analogs, hypoglycemia is an important consideration.²⁷ 

Beyond insulin resistance and decreased insulin secretion, other factors are also involved in the natural history of T2DM.²⁸ Glucose homeostasis is largely regulated by the incretins, which are gut-derived hormones released in response to nutrient (primarily glucose and fat) ingestion. These hormones elicit several important effects, including stimulation of insulin, a glucose-dependent effect. During the development and progression of T2DM, signals from gut-derived factors are diminished because of defective release of incretin hormones or resistance to their action. Combined β-cell dysfunction and incretin deficiency aggravate hyperglycemia, which further impairs insulin secretion and action. The reduced incretin effect is a substantial contributor to insulin deficiency. It is estimated that up to 70% of overall postprandial insulin response to glucose is mediated by the incretin hormones.²⁹ 

Postprandial hyperglycemia has important implications for the treatment of patients with long-standing T2DM.⁶ At some point after the addition of basal insulin and its subsequent titration to achieve fasting blood glucose (FBG) targets, glucose control may need to be further intensified because of postprandial hyperglycemia, which is usually a significant contributor as HbA_1c levels become closer to 7% (the higher the proportion of postprandial glucose to fasting plasma glucose, the lower the HbA_1c level).³⁰ 

From a pathophysiologic perspective, the next action should logically be to add a medication that targets postprandial glucose levels. What are the choices? Until now, treatment algorithms have guided physicians toward the addition of prandial insulin to basal insulin regimens. This approach leads to better glycemic control, but at a cost of increased blood glucose monitoring, greater probability of more insulin-associated weight gain, and greater risk of hypoglycemia. Knowing that incretins are responsible for up to 70% of the overall postprandial insulin response to glucose forms a basis for the use of incretin-based therapies as an alternative to the addition of prandial insulin in the most recent version of the AACE Comprehensive Diabetes Treatment Algorithm.³ Thus, the use of basal insulin to target fasting blood glucose levels and an incretin-based therapy to target postprandial glucose levels makes pathophysiological sense. The fact that incretin-based therapies work only in the presence of hyperglycemia helps reduce the risks of hypoglycemia. In addition, when comparing the incretin agents, GLP-1 receptor agonists are associated with weight loss and DPP-4 inhibitors are considered weight neutral; therefore, neither exacerbate the weight gain typically associated with insulin therapy. This finding has been borne out by a meta-analysis of clinical trial data, which in some cases also showed the ability to reduce the doses of basal insulin when GLP-1 receptor agonists were used with insulin.³¹ 

Currently, 4 oral DPP-4 inhibitors are available in the United States, all of which have been studied in combination with insulin therapy (Figure 3).^32-34 In the study on sitagliptin, the mean duration of diabetes was 12 to 13 years.³³ In general, the addition of the DPP-4 inhibitor reduced mean fasting plasma glucose levels modestly, reduced postprandial glucose levels more dramatically, and did so with low risk of hypoglycemia and without a change in weight. 

Of the 3 GLP-1 receptor agonists available, 2 are approved for use with basal insulin. Specifically, exenatide twice daily is approved for use with insulin glargine, and liraglutide once daily is approved for use with basal insulin (either insulin glargine or insulin detemir). If using GLP-1 receptor agonists with insulin, consider lowering the dose of the insulin to reduce the risk of hypoglycemia. Insulin and GLP-1 receptor agonists should not be injected at the same site, and GLP-1 receptor agonists are not a substitute for insulin. 

Both exenatide twice daily and liraglutide once daily require dose titration to minimize gastrointestinal adverse effects. Exenatide twice daily should be given from 0 to 60 minutes prior to a meal. Liraglutide can be given without regard to mealtimes. The long-acting version of exenatide (exenatide extended-release) is given once weekly without regard to mealtimes. The extended-release formulation is a powder that must be reconstituted by the patient and then administered immediately after reconstitution. Exenatide once weekly is now available in a pen. 

Data from clinical studies support the therapeutic potential of the GLP-1 receptor agonist–insulin combination, typically showing beneficial effects on glycemic control and body weight and with a low incidence of hypoglycemia. In established insulin therapy, the addition of a GLP-1 receptor agonist to insulin may allow for reductions in insulin dose, which may mitigate weight gain or hypoglycemia concerns. Figure 4 shows data from Buse et al,³⁶ where the mean duration of T2DM was 12 years. Currently a fixed-dose combination of insulin degludec (a longer-acting basal insulin analog with a lower incidence of nocturnal hypoglycemia than insulin glargine) and liraglutide is in development.³⁷ 

Other issues to take into account when choosing medications in patients with T2DM of long duration include the effects of medications on cardiovascular risk. The DPP-4 inhibitors are, on the whole, generally well tolerated. Results from the Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR)-TIMI trial showed that DPP-4 inhibition with saxagliptin did not increase or decrease the rate of ischemic events, though the rate of hospitalization for heart failure was increased in a large group of patients with T2DM who had a history of, or were at risk for, cardiovascular events.³⁸ The Examination of Cardiovascular Outcomes: Alogliptin vs. Standard of Care in Patients with Type 2 Diabetes Mellitus and Acute Coronary Syndrome trial showed that, among patients with T2DM who had a recent acute coronary syndrome, the rates of major adverse cardiovascular events were not increased with alogliptin compared with placebo.³⁹ There are positive data from clinical studies of the GLP-1 receptor agonists exenatide and liraglutide on cardiovascular risk factors^40-44 in patients with T2DM, but so far there are no results from ongoing outcome trials. 

Renal impairment is another important issue that increases hypoglycemia risk and is more prevalent in older patients.⁶ Metformin is not recommended for patients with severe renal impairment, although data on its use in more than 50,000 patients with different levels of renal function showed no increased risk in cardiovascular disease, all-cause mortality, or acidosis or serious infection (even at a glomerular filtration rate of 30 to 45 mL/min/1.73 m²).⁴⁵ Insulin can be used in patients with renal impairment, as can thiazolidinediones; most sulfonylureas should not be used in these patients. Linagliptin is the only DPP-4 inhibitor not cleared primarily by the kidney and therefore requires no dose adjustment in renal impairment (Table 2)⁴⁶. 

Exenatide is eliminated by the kidney. Neither short-acting nor extended-release exenatide is recommended for patients with severe renal impairment (creatinine clearance, <30 mL/min). Both should be used with caution in patients with renal transplantation. No dosage adjustment of short-acting exenatide is required in patients with mild renal impairment (creatinine clearance, 50-80 mL/min). Caution should be applied when initiating or escalating doses of short-acting exenatide from 5 to 10 μg or when using extended-release exenatide in patients with moderate renal impairment (creatinine clearance, 30-50 mL/min). Liraglutide is not heavily metabolized by the kidney, and therefore no dosage adjustment is necessary in patients with renal impairment. Caution is recommended, however, in cases where drug-induced nausea and vomiting may result in dehydration and affect patients with renal impairment. 

In patients with long-standing T2DM, combination therapy that works in complementary fashion to address basal and prandial hyperglycemia is often required.⁶ Basal insulin is often necessary in a state of relative β-cell absence to meet basal insulin needs. Postprandial hyperglycemia can be addressed by prandial insulin, but it can now be conquered by agents that work in glucose-dependent means, with a low risk of hypoglycemia and a low risk of weight gain. Addition of oral DPP-4 inhibitors offers modest glucose lowering with weight neutrality and good tolerability. Glucagon-like peptide-1 receptor agonists, given by subcutaneous injection, offer more robust glucose lowering, a low risk of hypoglycemia, and the possibility of lower insulin needs and weight loss.