Basal Glucose Can Be Controlled, but the Prandial Problem Persists—It’s the Next Target!

Widespread self-testing of blood glucose and, more recently, continuous glucose monitoring (CGM) have drawn attention to the daily patterns of blood glucose in both type 1 and type 2 diabetes. Seeing these patterns has allowed separate consideration of fasting (basal) hyperglycemia and after-meal (postprandial) increments of glucose above basal levels. In entirely healthy individuals, fasting plasma glucose is rarely >100 mg/dL (5.5 mmol/L). Peak values after meals are <140 mg/dL (7.8 mmol/L) and return quickly to basal levels. In the early stages of dysglycemia, high basal glucose (described as impaired fasting glucose) is the main abnormality for some people, whereas for others hyperglycemia after meals or an oral glucose load (impaired glucose tolerance) is more evident. These differences reflect varying mechanisms, may predict specific responses to initial treatments, and deserve further study. However, both basal and postprandial hyperglycemia are routinely present later in the natural history of type 2 diabetes, in part because sustained hyperglycemia can impair both the secretion and the various actions of insulin. Both are always present in type 1 diabetes. In the last 25 years we have become very successful in managing basal hyperglycemia, but postprandial hyperglycemia and its associated abnormalities remain untamed.

In untreated type 2 diabetes, basal hyperglycemia is usually quantitatively greater than further elevation of glucose after meals, especially when A1C is >8.0% (64 mmol/mol) (1–4). This is fortunate because most treatments for type 2 diabetes are more effective in controlling basal hyperglycemia. Metformin, sulfonylureas, thiazolidinediones, and basal insulins (human NPH and insulins glargine, detemir, and degludec) all have relatively modest effects on postprandial hyperglycemia. The same is true for most of the newer agents, including dipeptidyl peptidase 4 inhibitors, sodium–glucose cotransporter blockers, and longer-acting glucagon-like peptide 1 (GLP-1) receptor agonists. Consequently, in a mixed population of type 2 diabetes, glycemic exposure from basal hyperglycemia is typically dominant when A1C is high, and when control is improved by treatment, the contribution from postprandial hyperglycemia becomes more prominent (5). Even when dietary or pharmacological treatment maintains A1C values below the 7.0% (53 mmol/mol) target, residual postprandial hyperglycemia commonly limits attainment of A1C <6.5%, the upper end of the normal range (4,6). Persistence of postprandial hyperglycemia is most evident when long-acting (basal) insulin is added to one or more oral agents and systematically titrated to a fasting glucose target. An analysis of self-measured glucose profiles in type 2 diabetes no longer well controlled with oral agents, pooled from six studies, found that 79% of the glucose elevation >100 mg/dL was due to basal hyperglycemia (7). Six months after initiation of basal insulin, 34% of the hyperglycemic burden was basal and 66% was postprandial. In studies of this kind, average fasting glucose after optimization of basal insulin generally ranges between 100 and 130 mg/dL (6.6 and 7.3 mmol/L) and mean A1C is close to 7.0% (53 mmol/mol). Even when basal insulin is expertly titrated, significant prandial hyperglycemia may persist. For example, an experienced research group showed that for 61 patients whose mean baseline A1C was 9.5% (80 mmol/mol) on one or two oral agents, 36 weeks of titration of insulin glargine with continuation of metformin led to excellent control of fasting glucose (mean 104 mg/dL [5.75 mmol/L]). However, the mean A1C was 7.14% (54 mmol/mol), and half the group still had A1C >7%, mainly because of daytime hyperglycemia (8).

Attainment of A1C levels <7.0% is more likely when basal insulin is started before A1C is markedly elevated. Analysis of data from >2,000 participants in 12 studies showed that when baseline A1C was between 7 and 8% (mean 7.6% [60 mmol/mol]), a level <7.0% was reached in 6 months by 75% of patients (9). The Outcome Reduction With Initial Glargine Intervention (ORIGIN) trial further verified the safety and efficacy of timely initiation of basal therapies (10). More than 12,500 patients with impaired fasting glucose, impaired glucose tolerance, or type 2 diabetes treated with no more than one oral agent (and also selected for high cardiovascular risk) were randomized to either stepwise oral therapy (in most cases metformin and glimepiride) or insulin glargine added to any prior oral therapy and titrated to specific targets. For the 88% of patients in ORIGIN with overt diabetes at enrollment, the mean duration of diabetes was 4 years and median A1C was 6.6% (49 mmol/mol) (11). After 7 years of treatment A1C was 6.6% with the regimen based on oral therapy (with ∼10% needing to add insulin) and 6.3% (45 mmol/mol) with the basal insulin regimen. Safety findings were reassuring even in this high-risk population and similar between the regimens except for a threefold greater frequency of hypoglycemia in the basal insulin arm.

Hypoglycemia is associated with morbidity and mortality and is the leading limitation of basal insulin therapy. In ORIGIN, younger age, lower BMI, and lower attained dosage of glargine as basal insulin were independent predictors of nonsevere hypoglycemia (12). In contrast, and consistent with other studies, events requiring assistance by another person were associated with older age and evidence of renal or cognitive impairment (12,13). Although a causal relationship between hypoglycemia and poor medical outcomes has been difficult to verify in type 2 diabetes, recurrent or severe hypoglycemia calls for measures to mitigate risks (13,14). The simplest such measure is relaxation of goals for A1C, typically by increasing the target range from <7.0% to between 7 and 8% for higher-risk patients. Use of the new basal insulins (insulin degludec [Tresiba] and insulin glargine 300 units/mL [Toujeo]), both of which have longer and flatter glucose-lowering profiles, may reduce the risk of hypoglycemia while maintaining desired levels of glycemic control. Direct comparisons of degludec (15) and glargine 300 units/mL (16) with glargine 100 units/mL (Lantus) as ongoing basal therapy in type 2 diabetes have shown ∼30% lower risk of confirmed hypoglycemic events at night and smaller reductions of hypoglycemia at any time of day. Presumably, a subset of patients would be especially likely to benefit from using one of the new basal insulins. The analysis of predictors of hypoglycemia accompanying the use of glargine 100 units/mL described above suggests that individuals who have low BMI and low basal insulin requirements might be most likely to have more stable glycemic control with a longer-acting insulin (12). When the risk of hypoglycemia is reduced, further titration of dosage during long-term follow-up might favor more frequent attainment of A1C targets, but this expectation has been difficult to verify (17,18).

Long-acting GLP-1 receptor agonists are an alternative to basal insulin for patients with high risk of hypoglycemia and can be combined with basal insulins to provide further improvement of glycemic control beyond what is possible with either component alone (19). They have a favorable effect on weight but, like basal insulin, only a modest effect on increments of glucose after meals (20,21).

In summary, with an array of oral and injectable agents to control basal hyperglycemia, as well as the recent addition of basal insulins with improved profiles of action, type 2 diabetes can usually be managed well for the first decade after diagnosis. In this time frame, the leading barriers to maintaining A1C <7.0% are delay in diagnosis, clinical inertia in advancing therapy (22), and difficulty identifying patients at greatest risk for hypoglycemia to allow use of alternative treatment goals or methods.

After longer duration of type 2 diabetes (and for most patients with type 1 diabetes), control of postprandial hyperglycemia is the main unmet need, and it tends to become more difficult over time. Because type 2 diabetes is now often diagnosed before age 40, increasing numbers of patients no longer maintain adequate glycemic control with basal therapies alone at an age when they should have several decades of productive life ahead, provided microvascular complications can be prevented.

The usual recommendation for treatment intensification in this setting has been to start basal-bolus insulin therapy or at least twice-daily injection of premixed insulins. However, several lines of evidence suggest this may no longer be the best approach. First, population-based studies in the U.S. show that although the proportions of patients who are able to maintain desired A1C levels by using lifestyle alone or with oral agents have increased in the last decade, insulin-treated patients continue to have poor control (23). Second, in the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial and the Veterans Affairs Diabetes Trial (VADT), which enrolled patients with long-duration type 2 diabetes and high cardiovascular risk, seeking excellent glycemic control with extensive use of basal-bolus insulin had mixed results. Microvascular outcomes and some nonfatal cardiovascular events were reduced (24,25), but all-cause mortality increased 20% during intensive glycemic therapy in ACCORD (25) and significant hypoglycemia and weight gain occurred in both studies. Third, studies comparing ways of intensifying insulin therapy for type 2 diabetes have shown that basal-bolus treatment offers little further improvement of A1C when compared with simpler insulin regimens (26,27). For all these reasons, it is time to reconsider the prandial problem in type 2 diabetes, which includes interrelated difficulties with hyperglycemia, hypoglycemia, and weight control.

Growing knowledge of the mechanisms that normally regulate plasma glucose has clarified the problems posed by both basal and postprandial hyperglycemia (28–30). Fasting glucose is tightly controlled through regulation of hepatic glucose production by variable release of insulin into the portal vein, with modulation of insulin’s hepatic effects by glucagon and free fatty acids (FFA). In type 2 diabetes, basal insulin secretion is impaired and FFA and glucagon levels are high during fasting. Injection of long-acting insulin suppresses hepatic glucose production by acting directly on the liver and indirectly by reducing FFA release from adipose tissues. In type 1 diabetes, abnormalities of basal glucose regulation are more marked and delivery of basal insulin that closely matches physiological needs is mandatory for good control. Analogs of human insulin and improved delivery devices have greatly helped in meeting this need.

A more complex set of mechanisms comes into play after meals. Insulin secretion rapidly and markedly increases, accompanied by cosecretion of amylin (the other β-cell hormone), which suppresses glucagon, slows gastric emptying, and acutely signals satiety. Various other gastrointestinal peptides are modulated by meals, among them GLP-1, which also suppresses glucagon, slows gastric emptying, and provides a more sustained satiety signal. Neural factors may further regulate hepatic glucose production and peripheral insulin sensitivity. In type 2 diabetes, peak levels of insulin are lower than normal and occur 90–120 min after the meal begins rather than in the first 30 min. The rise of amylin is also delayed and reduced and thus may fail to suppress glucagon, regulate gastric emptying, and limit food intake during the usual duration of the meal. Secretion or action of GLP-1 may also be impaired. As for basal glucose, mechanisms for controlling postprandial glucose are more severely disturbed in type 1 diabetes. When both endogenous insulin and amylin are entirely lacking, prandial insulin treatment faces serious challenges.

In both type 1 diabetes and type 2 diabetes requiring prandial therapy, the size and timing of a prandial insulin dose should match the needs posed by each meal. However, meals differ widely in composition, size, and timing. Faulty dosing decisions lead to a mismatch between the meal and the profile of insulin delivered, and either marked postprandial hyperglycemia or later hypoglycemia can occur. Even when prandial insulin dosing is accurate, the resulting smaller increase of glucose may lead to further suppression of endogenous insulin and amylin, favoring reduced satiety and thus continued calorie intake. These difficulties contribute to the postprandial hyperglycemia, later hypoglycemia, and continuing weight gain that are commonly seen during basal-bolus insulin treatment. In addition, frequent testing of glucose (or CGM) to guide prandial dosing and the multiple injections needed are burdens that hinder regimen adherence.

For clinical care, CGM facilitates individualized decisions on prandial dosing, and for research, it offers new quantitative end points for testing treatment regimens. Various formulas have been proposed to describe glycemic variability, and there is active debate about how to use them (31–33). Among the most appealing is the “time in range.” An expert panel recently proposed the percentage of measurements in a day between 70 and 180 mg/dL (3.9 and 10.0 mmol/L) as a summary indicator of glycemic control (34). Practice guidelines from the American Diabetes Association also suggest <180 mg/dL as a target for control of glucose levels 1–2 h after a meal (35). Selection of 180 mg/dL (10 mmol/L) by both groups as the upper end of a nominally desirable range speaks to the difficulty of controlling postprandial hyperglycemia with current methods. An early report of CGM measurements for a group of 15 patients with type 1 and 15 with type 2 diabetes whose mean A1C was 7.5% (58 mmol/mol) found that glucose was above 180 mg/dL 33% of the time (36). In addition to limiting attainment of A1C targets, this much variability beyond the desired range is suspected of contributing to both the microvascular and cardiovascular complications of diabetes, independent of the mean level of control reflected by A1C (37).

Behavioral interventions for postprandial hyperglycemia and weight control, although centrally important, are beyond the scope of this article. For the present discussion, several other approaches deserve comment.

Gastrointestinal procedures intended to alter food intake and absorption and otherwise improve prandial physiology are now entering the mainstream of treatment options (38). The mechanisms underlying their effects are not well understood (39), but a large body of information on short-term clinical outcomes is available. Diabetes cannot be said to be cured by this approach, but very impressive weight loss and improvement of prandial and overall glycemic control are common (40). A recent joint statement by international diabetes organizations has endorsed metabolic surgery as a treatment “recommended” for patients with type 2 diabetes and BMI 40 kg/m² or higher and “considered” under certain conditions for individuals with less severe obesity (41). However, long-term assessment of the risks versus benefits of this approach is not yet available.

α-Glucosidase inhibitors (AGIs) are not widely used in the U.S. but have greater acceptance in some other countries. Their leading drawbacks are gastrointestinal side effects, mainly flatulence and diarrhea. They can reduce postprandial hyperglycemia in type 2 diabetes by ∼50% and are effective in combination with therapies that target basal glucose control (42). Secondary analyses of data from the Study to Prevent Non-Insulin-Dependent-Diabetes (STOP-NIDDM) suggested that acarbose might reduce cardiovascular risk for patients with impaired glucose tolerance or type 2 diabetes (43). A large study testing this hypothesis is now underway (44). If cardiovascular benefit is confirmed, use of AGIs for patients with type 2 diabetes and adequate basal control but persisting postprandial hyperglycemia should be reconsidered.

Many patients with type 2 diabetes advised to use basal-bolus insulin therapy do not consistently follow the treatment plan. To address this problem, a simpler, stepwise approach to adding prandial insulin to basal therapy has been proposed (45). Mean A1C reductions of 0.3% to 0.5% after adding a single prandial injection before breakfast or another important meal are typical and sometimes sufficient (46). At the time of initiation of basal insulin for type 2 diabetes or soon after, adding a single injection of prandial insulin may be as effective as full basal-bolus treatment and also as effective as two injections of premixed insulin but with less risk of hypoglycemia (27,47). Of course, this approach is not appropriate for type 1 diabetes.

Normal secretion of endogenous insulin increases plasma levels sooner after a meal than injection of the current generation of “rapid-acting” insulin analogs. Hence, insufficiently rapid onset of prandial insulin has been proposed as the main limitation of basal-bolus therapy, whether by multiple injections or by continuous subcutaneous insulin infusion (CSII). Also, the subcutaneous depot of insulins aspart, lispro, or glulisine may continue to increase plasma levels longer than necessary to match needs after small meals containing mainly refined carbohydrate. Insulin formulations that are more rapidly absorbed after subcutaneous injection are under study (48), and an inhalable formulation with rapid early uptake (49) is now available as well. Whether these modifications will lead to sustained improvement of glycemic control in routine management of either type 1 or type 2 diabetes remains unknown.

Computerized algorithms linking CSII to CGM are under intense scrutiny for type 1 diabetes (50). The hardware and software for such systems continue to improve, and there are encouraging reports of their use for ambulatory patients (51,52). In addition to the pharmacokinetics of currently used insulins, limitations of these systems include the cost of the devices themselves and of clinical support for their use, the risk of system failure, and observations that, to date, substantial glycemic variability related to meals remains. Whether such systems will be widely acceptable for type 2 diabetes is unknown, as is whether weight gain associated with their use can be prevented. Systems supplementing insulin infusion with timely delivery of subcutaneous glucagon to protect against hypoglycemia have yielded promising early results (53).

Most studies of currently available GLP-1 agonists have focused on the longer-acting products. Liraglutide (Victoza), extended-release exenatide (Bydureon), dulaglutide (Trulicity), and albiglutide (Tanzeum) all have relatively long duration of action and greater effects on basal than on postprandial glucose (54–57). All are very effective in reducing A1C in patients who start with high levels, and they have desirable effects on weight. Their main side effects are nausea and other gastrointestinal complaints, and up to 20% of patients discontinue treatment in clinical studies. They are strongly marketed as alternatives to basal insulin.

In contrast, the shorter-acting GLP-1 agonists, unmodified exenatide (Byetta) and lixisenatide (Adlixin), have received limited attention (58,59). When given just prior to a meal, both slow gastric emptying, prevent an inappropriate rise of glucagon, potentiate insulin secretion, and thereby markedly blunt postprandial hyperglycemia. They also favor weight loss. Dose-ranging studies have shown that these agents can reduce postprandial hyperglycemia at low doses (e.g., 5 μg before a meal) that rarely cause gastrointestinal side effects (60,61). In principle, their best use should be for patients with good control of basal glucose but persisting postprandial hyperglycemia.

Even less attention has been directed to pramlintide (Symlin), an amylin analog (62). This agent is approved in the U.S. as an adjunct to basal-bolus treatment of type 1 or type 2 diabetes. It is given as fixed doses by subcutaneous injection before meals, has a short duration of action, and markedly blunts postprandial hyperglycemia by suppressing glucagon and slowing gastric emptying. Like the GLP-1 agonists, it reduces food intake and favors weight loss. Pramlintide’s disadvantages for its current indications include the need for several daily injections in addition to multiple injections of insulin, frequent dosing decisions, and increased risk of nausea or hypoglycemia. Because of these difficulties and limited marketing efforts, it is not widely used.

Recent studies of GLP-1 and amylin agonists added to basal insulin for type 2 diabetes have been very encouraging. Key findings from these are summarized in Table 1.

In one study (63), 261 patients with type 2 diabetes with elevated A1C on oral therapies plus insulin glargine (Lantus) were randomized to add twice-daily exenatide (Byetta) or placebo injections before breakfast and dinner (Fig. 1). Glargine was titrated during 30 weeks of treatment. Mean baseline A1C was >8.0% in both groups, and the reduction from baseline was 0.69% greater with exenatide. After treatment, A1C ≤7.0% was attained by 60% of the group assigned to exenatide and 35% of the placebo group. Body weight increased with placebo but decreased with exenatide, with a between-treatment difference of 2.7 kg. Gastrointestinal complaints were more frequent with exenatide, but the frequency of hypoglycemia did not differ. Most of the between-group difference in A1C was due to blunting of the postprandial increments of glucose after the exenatide doses at breakfast and dinner.

Another study directly compared basal insulin plus twice-daily exenatide with conventional basal-bolus therapy of type 2 diabetes (64). All patients had previously used insulin glargine (Lantus) with oral agents, and after 12 weeks of titration of glargine dosage, 627 of them were randomized to add either exenatide (Byetta) or insulin lispro. From mean baseline levels of 8.2% and 8.3% (66 and 67 mmol/mol), both treatments reduced A1C to 7.2% (55 mmol/mol). Weight increased with lispro but decreased with exenatide, with a between-treatment difference of 4.6 kg. Hypoglycemia was less frequent but gastrointestinal symptoms more common with exenatide. Reduced postprandial increments accounted for most of the glycemic effect of both regimens.

A smaller but similarly designed study examined the effects of exenatide versus rapid-acting insulin on glucose profiles measured by CGM (65) (Fig. 2). After a run-in of 8 to 12 weeks on basal-bolus insulin, 102 patients with long-duration (median 15 years) type 2 diabetes and high cardiovascular risk were randomized to continue basal-bolus therapy or to switch to insulin glargine (Lantus) plus two or three mealtime doses of exenatide (Byetta). Adjustment of medications during 26 weeks of randomized treatment attained the intended goal of equivalent A1C between 6.7 and 7.3% (50 and 56 mmol/mol) with each regimen (final mean values 7.2% [55 mmol/mol] with basal-bolus insulin and 7.1% [54 mmol/mol] with basal insulin and exenatide). Glycemic variability, defined as the coefficient of variation of glucose levels measured by CGM, was the primary end point. Variability was reduced more with the exenatide-based regimen. Glucose values were in the 70–180 mg/dL range 75% of the time with exenatide and basal insulin and 71% of the time with basal-bolus insulin. Weight was unchanged with basal-bolus therapy but declined with exenatide and basal insulin, with a between-treatment difference of 5.5 kg.

In patients with type 2 diabetes previously treated with basal insulin and oral agents, insulin glargine (Lantus) was titrated for 12 weeks, and patients with A1C >7.0% (n = 446) were randomized to add lixisenatide or placebo injected once-daily before breakfast (66). After 24 weeks, mean A1C had declined from 7.6 to 7.0% (60 to 53 mmol/mol) with lixisenatide and to 7.3% (56 mmol/mol) with placebo, with a difference of 0.32%. Values of A1C ≤7.0% were attained by 56% of patients taking lixisenatide versus 39% of those assigned to placebo. Gastrointestinal symptoms and hypoglycemia were more frequent with lixisenatide.

This head-to-head study examined 298 patients with a mean duration of type 2 diabetes of 12 years who were previously taking basal insulin and oral agents. After optimization of insulin glargine (Lantus) for 12 weeks, they were randomized to addition of lixisenatide once daily, rapid-acting insulin once daily, or rapid-acting insulin three times daily for 26 more weeks (67). From a mean A1C of 7.6% (60 mmol/mol) after glargine titration, levels declined to 7.2, 7.2, and 7.0% (55, 55, 53 mmol/mol) with lixisenatide and the two rapid-acting insulin regimens, respectively. The change of weight during randomized treatment favored the lixisenatide arm versus the two insulin arms by −1.7 and −2.0 kg.

A randomized, open-label comparison of pramlintide with rapid-acting insulin, with each of these prandial therapies added to titrated basal insulin, was performed in 133 patients with type 2 diabetes (68) (Fig. 3). After 24 weeks, A1C was similarly reduced from 8.2 and 8.3% (66 and 67 mmol/mol) to 7.2 and 7.0% (55 and 53 mmol/mol) with the pramlintide and rapid-acting insulin regimens, respectively. Reductions of fasting glucose and postprandial glucose increments were not significantly different between the regimens. Weight increased with rapid-acting insulin but not pramlintide, with a difference of 4.7 kg. Pramlintide caused more nausea but less hypoglycemia than rapid-acting insulin.

The preceding discussion suggests that basal hyperglycemia is no longer the main problem. Instead, the prandial problem deserves more attention, and there are promising but incompletely explored options for addressing it. From both the commercial and the public health points of view, the return on investment could be improved by shifting effort and resources from basal to prandial therapy.

Together with earlier diagnosis and timely initiation of basal therapy for type 2 diabetes, prandial treatment could be added earlier than in the past. The glycemic increment persisting after optimized basal therapy could be blunted by adding, with one or more meals, an AGI, a dose of rapid-acting or regular human insulin, or a short-acting GLP-1 agonist. Progressing stepwise in this fashion from a basal-only regimen to basal plus prandial therapy before attempting a full basal-prandial strategy is simpler and likely to be better tolerated. At present, continuation of basal therapies used alone, even with appropriate titration of dose, typically allows A1C levels to rise gradually to levels well above 7.0% after 10 or more years while additional treatment to restore A1C to the <7.0% “target” is delayed. A better approach might be to consider 7% A1C a “ceiling” and to seek values below this level by timely addition of prandial therapy. Overtitration of basal insulin leading to hypoglycemia would be less common if prandial therapy were regarded as easy and desirable rather than difficult and dangerous. Even without incorporating new methods, earlier prandial therapy might extend the time excellent glycemic control is maintained, leading to reduction of long-term microvascular and cardiovascular risks.

Better awareness of the differences between long-acting and short-acting drugs in this class is needed. Demonstration of a 22% reduction of cardiovascular mortality with liraglutide (in the Liraglutide Effect and Action in Diabetes: Evaluation of cardiovascular outcome Results [LEADER] trial [69]) has increased the attractiveness of this agent, but this GLP-1 agonist does not reliably control postprandial hyperglycemia. Whether long-term use of short-acting GLP-1 agonists, which have better prandial effects (20,21), can also reduce cardiovascular risk in a similar population is unknown, and no such benefit was seen with once-daily lixisenatide in the very high-risk population studied in the Evaluation of Lixisenatide in Acute Coronary Syndrome (ELIXA) trial (70). However, the short-acting GLP-1 agents might be an excellent option for patients whose postprandial hyperglycemia seriously impairs attainment of A1C goals to limit microvascular risk. Exenatide or lixisenatide should, for many patients with type 2 diabetes, be as effective as prandial insulin without causing weight gain and hypoglycemia. Gastrointestinal symptoms could be limited by minimizing doses. Delivery devices allowing selection of doses between 2.5 and 10 μg would permit the greatest flexibility, but they have not been made available. The best results are likely to derive from stepwise addition of up to three prandial doses (65). Pre- or postprandial glucose tests would not routinely be needed to guide dosing, allowing greater convenience and perhaps lower additional cost than prandial insulin. Further well-designed studies of this approach are needed.

Perhaps the most remarkable opportunity is for development of a fixed-ratio formulation of an amylin analog with insulin. Under normal conditions, endogenous insulin and amylin are cosecreted by β-cells in the same patterns during fasting and with meals. Use of pramlintide (off-label) as the only prandial therapy supplementing newly initiated basal insulin in the study described above (66) is a significant proof of concept, but not likely to be effective for longer-duration type 2 diabetes or type 1 diabetes. Especially in type 1 diabetes, where little or no β-cell function remains, coordinated replacement therapy for this bihormonal deficiency deserves further study. Preliminary reports show no obvious barriers to a combined formulation of pramlintide with regular human insulin for clinical use (71,72), and developing one could have both clinical and commercial appeal. Continuous delivery of a coformulation might have several advantages. Titration decisions would be simpler than with separate dosing. Postprandial hypoglycemia might be reduced, and weight control would surely be improved compared with insulin therapy alone. Nausea might be less frequent than with intermittent use of pramlintide. Combined with a closed-loop control system, this approach might attain near-normal glycemic control more reliably than an insulin-only or insulin-with-glucagon system. These are testable hypotheses.

This topic deserves a separate review, but several points are clear. There are many gastrointestinal peptide hormones, some with known structures and metabolic functions (73,74). Research on their roles in health and disease will be central to understanding the genesis of obesity and diabetes and perhaps the links between these and cardiovascular disease and cancer. Formulations of these peptides, or synthetic analogs, will become available for research and clinical use. Their dosing may be more flexible than that of insulin, allowing alternative (noninjection) routes of delivery.

Where can we find the resources for these new efforts? Product development and basic research are expensive. However, some of the resources now used for fine-tuning the treatment of basal hyperglycemia may now be better used for the prandial problem. We have multiple classes of oral and injected agents with powerful effects on basal glucose, as well as many choices in each class. Rather than more of these redundant therapies, better ways to deliver prandial insulins and short-acting GLP-1 agonists should be studied, along with coformulation of an amylin agonist with insulin. In addition, investment in closed-loop delivery systems and surgical modification of intestinal function should proceed but not be unlimited. They are what Lewis Thomas called “half-way technologies” (75). While amazing—we are lucky to have them today—they will become obsolete (like tuberculosis sanatoria and iron lungs) when research uncovers the root causes of the problems they address. The present challenge is to focus use of closed-loop CSII and metabolic surgery on subgroups of patients for whom they offer the best balance of benefit to cost and risk. Other resources should be reserved to explore new insights into islet, gut, and brain physiology (76) to produce the definitive therapies of the future. Regulatory and financial incentives toward this end are needed. The prandial problem—including postprandial hyperglycemia, weight gain, and hypoglycemia caused by overreliance on injected insulin—is an endocrine and neurologic puzzle that calls for further basic and clinical research.

Funding. Support for this work was from the Rose Hastings and Russell Standley Memorial Trusts.

Duality of Interest. M.C.R. has received research grant support through Oregon Health & Science University from AstraZeneca, Eli Lilly, and Novo Nordisk and honoraria for consulting or speaking from AstraZeneca, Biodel, Elcelyx, Eli Lilly, GlaxoSmithKline, Sanofi, Theracos, and Valeritas. These dualities of interest have been reviewed and managed by Oregon Health & Science University. No other potential conflicts of interest relevant to this article were reported.