Combination Therapies With Insulin in Type 2 Diabetes

  1. Hannele Yki-Järvinen, MD, FRCP1
  1. 1Department of Medicine, Division of Diabetes, University of Helsinki, Helsinki, Finland

    The U.K. Prospective Diabetes Study (UKPDS) demonstrated that intensive glucose control with insulin or sulfonylureas markedly reduces the risk of microvascular complications (1). For myocardial infarction, the reduction in risk (16% for a 0.9% decrease in HbA1c) was of borderline significance but corresponded closely to epidemiological predictions (14% decrease for a 1% drop in HbA1c) (2). These data demonstrated that neither insulin nor sulfonylureas, despite causing hyperinsulinemia and weight gain, have adverse effects on cardiovascular outcome. Glycemic control deteriorated continuously, however, even in intensively treated patients in the UKPDS (1).

    In the UKPDS, the worsening of glycemic control has been attributed to the natural course of type 2 diabetes and lack of efficacy of current antihyperglycemic therapies (1). Insulin therapy consisted of a single injection of ultralente or isophane insulin. If the daily dose exceeded 14 U, regular insulin was added and home-glucose monitoring was encouraged (1). Combination therapy regimens with insulin and oral agents were not used. We now know that 14 U of long-acting insulin is insufficient to control fasting glycemia in most type 2 diabetic patients (3). Since 1977, when the UKPDS was started, several studies have tried to define the optimal insulin treatment regimen for type 2 diabetic patients. These studies are the focus of this review and include studies comparing insulin alone to combination therapy with insulin and sulfonylureas (subject to meta-analyses in 1991 and 1992) (4,5) and more recent trials using metformin, glitazones, or acarbose in insulin combination therapy regimens. They do not contain data on cardiovascular end points but only on surrogate markers of risk of micro- and macrovascular complications, mostly data on glycemia, body weight, insulin doses, lipids, and in a few studies, also accurate data on the frequency of hypoglycemias.

    According to a Medline search (1966–2000), insulin alone has been compared with insulin combination therapy in a total of 34 prospective studies that lasted at least 2 months and reported data on HbA1 or HbA1c in type 2 diabetic patients. Studies comparing glycemic control, weight gain, hypoglycemias, and insulin requirements between the two modes of treatment in insulin-naïve patients are listed in Table 1 and in previously insulin-treated patients are listed in Table 2. The studies have been ranked according to glycemic control at the end of the trial.


    Insulin-naïve and previously insulin-treated patients.

    In insulin-naïve patients in a total of 15 comparisons (10 studies), glycemic control was similar in most (11 of 15) comparisons and better with the insulin combination than the insulin-alone regimen in four comparisons (Table 1). In all studies, the daily insulin dose was lower with insulin combination therapy than with insulin alone. The weighted mean for the insulin-sparing effect of two drugs (sulfonylureas and metformin) in addition to insulin was 62%, i.e., 1.5–2.0 times that with regimens combining either metformin alone (−32%) or sulfonylureas alone (−42%) (Table 1) with insulin. These data imply that oral agents still have significant glucose-lowering effects even in patients who are poorly controlled on oral drugs. One may also predict from these data that if the insulin dose is lowered less than ∼30% when patients are transferred from insulin alone to insulin combined with sulfonylurea or metformin, glycemic control will be better during insulin combination therapy. This is documented by analysis of data from studies in previously insulin-treated patients (Table 2). In these comparisons, glycemic control was better in most (19 of 25) comparisons, but the insulin dose was decreased by only 19% in the combination regimens using metformin and by 21% in comparisons using insulin and sulfonylureas (Table 2). Thus, although in most comparisons (30 of 45) glycemic control has been better with insulin combination therapy regimens than with insulin alone, the difference may be at least partly explained by how the insulin dose has been decreased during insulin combination therapy. All glitazones have improved glycemic control when added to previous insulin treatment (Table 2). In a study directly comparing the insulin-sparing effects of troglitazone (600 mg/day) and metformin (1,700 mg/day), troglitazone had a greater (−53%) insulin-sparing effect than metformin (−31%). This was explained by insulin-sensitizing effects of troglitazone but not metformin (6).


    What determines weight gain during insulin therapy?

    Although most patients with type 2 diabetes are overweight, weight loss precedes the diagnosis of type 2 diabetes (7). This weight loss is due to hyperglycemia-induced wasting of energy, as glucosuria and as energy used to overproduce glucose (8). When glucose control is improved with insulin and/or sulfonylureas, energy loss in the urine decreases or ceases, weight increases, and basal metabolic rate (kJ/min) (810) and dietary intake (11) remain unchanged. The increase in body weight increases basal metabolic rate, but this is counterbalanced by improved glycemic control, which decreases basal metabolic rate because less energy is needed for glucose overproduction. Because dietary intake remains unchanged (11), weight gain is proportional to reduction of glucosuria and can indeed be predicted based on fasting glucose concentrations (11). Because glucosuria appears when the fasting glucose concentration exceeds 10–12 mmol/l, weight gain is inevitable if insulin therapy is postponed until significant glucosuria occurs. In our experience, a 5-mmol/l (90-mg/dl) decrease in fasting glucose or a decrease in HbA1c by 2.5% from a baseline of 15 mmol/l (270 mg/dl) is associated with a 5-kg weight gain during 1 year (or 2 kg/1% decrease in HbA1c) (11). Thus, the main predictors of weight gain are initial glycemia and its response to treatment (11). The patient with poor glycemic control before initiation of insulin therapy but with a good treatment response is at greatest risk for weight gain.

    Choice of oral agent and weight gain in insulin-naïve patients.

    Only one trial has compared the combination of insulin with that of insulin and metformin alone in previously insulin-naïve patients (12). In this study, which lasted 12 months, the bedtime insulin-metformin regimen was superior to three other bedtime insulin regimens with respect to glycemic control, weight gain, and hypoglycemias (Table 1) (12). The ability of metformin to counteract weight gain and improve glycemia, when combined with insulin, has been confirmed in abstract reports (13,14). In these studies, weight gain was less despite comparable glycemia (13) or weight gain was similar despite better glycemic control (14) in patients using metformin and insulin compared with those using insulin and sulfonylureas or insulin alone. Data are heterogenous regarding the ability of metformin to influence weight gain when combined with both insulin and sulfonylureas, compared with regimens containing insulin alone (Table 1). In two comparisons, weight gain was less with the combination regimen than with insulin alone (15,16), whereas two other comparisons revealed no difference (12,15) (Table 1, Fig. 1). The ability of metformin to counteract weight gain during insulin combination therapy has been attributed to a decrease in dietary intake (11).

    Choice of oral agent and weight gain in previously insulin-treated patients.

    Switching patients from treatment with insulin alone to insulin combination therapy with metformin has been associated with less weight gain in two of three studies, whereas no difference was found in any of the 16 comparisons in which insulin plus sulfonylurea was compared with insulin alone, despite better control in 10 of 16 comparisons (Table 2). It is unclear whether this is because weight was not accurately recorded or because the larger dose of exogenous insulin or the greater number of insulin injections used in insulin alone as compared with the insulin combination regimen had independent weight-promoting effects. In studies comparing insulin-glitazone treatment with insulin alone, glycemic control was better in each study with the insulin-glitazone combination than with insulin alone. Better glycemic control was also associated with greater weight gain in each of the three studies. Although the amount of weight gain relative to the improvement in glycemic control seemed slightly greater than with sulfonylureas (Fig. 1), data on insulin-glitazone combination therapy are still sparse. The significance of weight gain with glitazones is also difficult to judge, because glitazones may be beneficial in redistributing fat from visceral to subcutaneous sites (1719). A small fraction of weight gain with glitazones could be due to peripheral edema (20,21).


    Frequency of hypoglycemias in type 1 versus type 2 diabetes.

    In both patients with type 1 (22) and type 2 (12) diabetes, the frequency of hypoglycemias is inversely proportional to glycemic control. In the Diabetes Control and Complications Trial, in patients with HbA1c between 7 and 8%, the frequency of severe hypoglycemias requiring assistance in the provision of treatment was 0.62 per patient per year (22). In contrast, in the FINFAT study (12), the Kumamoto study (23), or other studies of type 2 diabetes in which comparable glycemic control was achieved (24,25), there were no severe hypoglycemias. The frequency of biochemical hypoglycemias (blood glucose <3.5 mmol/l) was 1.9 per patient per year in patients treated with insulin plus metformin and approximately twice as high in the other groups in the FINFAT study (12). In the latter study, HbA1c averaged between 7 and 8% in all groups for 1 year. These data, although derived from separate studies, suggest that hypoglycemias are much less of a problem in type 2 diabetic patients than in type 1 diabetic patients.

    Does the oral agent influence the occurrence of hypoglycemias independent of glycemic control?

    The occurrence of hypoglycemia has been sparsely reported [eight comparisons in insulin-naïve patients (Table 1) and five comparisons in previously insulin-treated patients (Table 2)]. In insulin-naïve patients, use of insulin combination therapy with metformin has been associated with less hypoglycemias than with insulin alone, despite better glycemic control with the insulin-combination regimen (12). No difference was observed between insulin-alone and insulin-sulfonylurea regimens in five of seven studies; in two studies (25,26), there were more cases of hypoglycemia with insulin and sulfonylurea than with insulin alone (Tables 1 and 2). No difference in the incidence of hypoglycemia was observed between insulin alone compared with insulin plus sulfonylurea and metformin regimens (Tables 1 and 2). In the latter studies, there was also no difference in glycemic control. In all three studies comparing insulin-glitazone combination therapy to insulin alone, the frequency of hypoglycemia was higher and glycemic control was better with the insulin-combination regimen. These data suggest that with the possible exception of metformin, use of insulin combination therapy is accompanied by a similar frequency of hypoglycemia than is use of insulin alone.


    Insulin-naïve patients.

    Data on changes in serum triglycerides and glycemia in insulin-naïve patients are summarized in Table 3. As judged from the weighted means of insulin-alone regimens, a decrease in HbA1c from ∼10 to 8% (i.e., by 2%) is associated with a 0.7- to 0.8-mmol/l decrease in serum triglycerides from an initial concentration of 2.4–2.7 mmol/l. With insulin combination therapy regimens, a decrease of HbA1c by 2% decreases serum triglycerides by 0.4–0.6 mmol/l (Table 3). In all except one study, insulin alone lowered serum triglycerides slightly more than insulin combination therapy, although there was no significant difference in the lowering of serum triglycerides with the two modes of therapy in any of the studies (Table 3). LDL and HDL cholesterol concentrations remained unchanged in all studies, with no differences between regimens (12,15,16,25,2729).

    Previously insulin-treated patients.

    As summarized in Table 4, the greater improvement in glycemic control with insulin combination therapy than with insulin alone in 11 of 14 studies has not been consistently (4 of 11 studies) associated with a greater decrease in serum triglycerides. These data demonstrate that factors other than average glucose concentrations determine the degree of lowering of serum triglycerides. Overall, the available comparisons of changes in serum lipid and lipoprotein concentrations in both insulin-naïve and previously insulin-treated patients do not allow definitive conclusions and do not support choice of one treatment regimen over another.


    Regarding blood pressure, in a follow-up study of the patients participating in the FINMIS study (15,30), blood pressure increased significantly in the entire group of 100 patients during 1 year. Weight gain correlated both with the increase in blood pressure and with an increase in the LDL cholesterol concentrations (30). Three shorter studies reported data on blood pressure (15,25,27) but found no changes in blood pressure or differences between regimens.


    Choice of insulin regimen during insulin combination therapy: NPH insulin or insulin glargine?

    Regarding basal insulinization, the commonly used intermediate-acting insulin (NPH) is not ideal for once-daily use. In the FINMIS study, in patients with type 2 diabetes with a mean BMI of 29 kg/m2, injection of NPH insulin at 9:00 p.m. resulted in maximal glucose lowering between 4:00 and 8:00 a.m., but the effect was gone by 3:00 p.m., i.e., 18 h after the injection, and dinnertime glucose concentrations were unnecessarily high. The recently approved long-acting insulin analog insulin glargine seems to overcome these problems. In a study comparing NPH plus oral agents to insulin glargine plus oral agents in 423 insulin-naïve type 2 diabetic patients for 1 year, all hypoglycemias were 35% lower and nocturnal hypoglycemias were 56% lower with insulin glargine than with NPH (Fig. 2). Dinnertime glucose levels were also significantly lower with insulin glargine than with NPH (Fig. 2).

    Regular insulin or short-acting insulin analogs compared with NPH during combination therapy.

    Regular insulin three times per day plus a sulfonylurea has been compared with a single injection of NPH taken at bedtime and a sulfonylurea. No difference in glycemic control was found, but weight gain was significantly greater with three injections of regular insulin than with a single injection of bedtime NPH insulin (31). Greater weight gain with no difference in glycemic control has also been reported with three injections of lispro plus sulfonylurea compared with NPH plus sulfonylurea (32) (Table 5).

    Timing of the intermediate-acting insulin injection.

    The pros and cons of timing of the intermediate-acting insulin injection has been examined in three studies (15,33,34). In two studies, a bedtime injection was recommended because it resulted in less weight gain (15) or less hypoglycemias (34) than a morning injection. In the third study, no differences in glycemia or weight gain were found (33).


    Variation in hepatic insulin sensitivity seems to be much more important than insulin absorption in determining insulin requirements during combination therapy with NPH insulin (3). Hepatic insulin sensitivity cannot be routinely measured but correlates with various indexes of obesity (3). In type 2 diabetic patients with a mean BMI of 29 kg/m2, to achieve an average HbA1c of ∼7.5% from a baseline value of 10%, the mean bedtime NPH insulin dose for body weights of 70, 80, 90, and 100 kg has been 0.2, 0.3, 0.4, and 0.5 IU/kg (3,11). However, interindividual variation is large and has varied 20-fold between 8 and 168 IU per day (12), which implies that these average predictions are not accurate enough to be used on an individual level.

    Autoantibodies to glutamic acid decarboxylase (GADA) predict an increased likelihood of insulin requirement in both young and old adults with type 2 diabetes (35). In 3,672 newly diagnosed patients in the UKPDS, 34% of those aged 25–34 years and 7% of those aged 55–65 years had GADA (35). Among patients older than 55 years at diagnosis, 34% of those with GADA and 5% with autoantibodies to neither GADA nor islet cell cytoplasm required insulin therapy. In these older patients, only the presence of GADA but not phenotypic features such as BMI predicted insulin requirement. There are no studies comparing insulin combination regimens with insulin alone in these patients, who are often classified as having type 2 diabetes but actually have type 1 diabetes (36). In patients in whom signs of absolute insulin deficiency (rapid weight loss, ketonuria) ultimately develop, the presence of GADA may guide the choice of a basal-bolus–type full insulin-replacement regimen (37).


    In studies in which the insulin dose is titrated aggressively to reach glycemic targets, the decrease in HbA1c will be directly proportional to its initial level. Of other factors, obesity predicts a poor response to any type of insulin therapy, especially if insufficient doses of insulin are used (30,38). In addition, and as discussed above, GADA may predict poor response to combination therapy.


    Initiation of insulin therapy on an ambulatory basis in type 2 diabetic patients has been shown to be as safe and effective as an inpatient program (39). Regardless of the insulin treatment regimen chosen, the insulin dose should be adjusted to reach glycemic targets. Considering the large interindividual variation in insulin requirements, it is difficult to define the correct insulin dose by performing dose adjustments only at outpatient visits, unless these are very frequent. In our experience of treating insulin-naïve patients (12,15), the best method of defining the insulin dose is teaching the patient to self-adjust the dose based on results of home glucose monitoring. This is easiest to perform if the dose adjustment is based on measurement of fasting plasma glucose only. Fasting plasma glucose is not influenced by size, composition, or rate of absorption of meals as much as by postprandial glucose levels, and its measurement does not interfere with daily activities. The maximal action of NPH given at bedtime is exerted on fasting glucose, which, therefore, is a particularly suitable target for titration of the dose when insulin combination therapy with NPH or insulin glargine is used. However, because especially NPH insulin is unable to adequately control postdinner glycemia, fasting glucose must be in the normal range (4–6 mmol/l) for average glycemic control, measured using HbA1c to be <7.5%. A simple method of initiating insulin therapy, developed based on experience from the FINFAT study, is shown in Table 5 (12). The patient is assumed to be insulin-naïve and on maximal doses of sulfonylureas and metformin. The recommendation to discontinue the sulfonylurea (glyburide) but not metformin after insulin combination therapy is started is based on the inability of some patients to adequately titrate the dose of bedtime NPH because of hypoglycemia (12). An increase in mild hypoglycemias was also reported by Riddle and Schneider (25) with glimepiride combined with a single injection of 30/70 insulin at 6:00 p.m. compared with two injections of 30/70 insulin. Hypoglycemias may not be a problem with peakless insulins such as insulin glargine (40). Discontinuation of sulfonylurea when insulin therapy is started may retard achievement of good glycemic control unless the insulin dose is rapidly increased (12,25). Glitazones could be an additional or alternative component in the oral hypoglycemic agent regimen, but there are no studies in insulin-naïve patients.


    Against the emerging epidemic of type 2 diabetes, studies comparing different insulin treatment regimens are sparse and include only a small number of patients treated for a maximum of 1 year (Tables 1 and 2). Data on effects of insulin-combination therapy versus insulin alone on diabetic microvascular and macrovascular complications are nonexistent. The main reason for the paucity of data may be the reluctance of private funding agencies to support studies using pharmacological agents and the reluctance of industry to support studies with established preparations. The development of new agents such as glitazones and insulin analogs have increased the number of patients included in various trials, but many company-initiated trials are designed to fulfill licensing requirements and must be performed in multiple centers to save time. Although some company-initiated trials are of superb quality, others suffer from inadequate glycemic control and may lack the comparisons the clinicians would be interested in. Despite these deficiencies, some conclusions regarding the role of insulin combination therapy in the treatment of type 2 diabetic patients seem justified.

    No study reported worse glycemic control with insulin combination therapy than with insulin alone. Glycemic control was better with insulin combination therapy than with insulin alone in most studies of previously insulin-treated patients, but this could be explained by a smaller difference (∼20% for metformin or sulfonylureas) (Table 2) in the insulin dose between the two modes of treatment than in studies performed in insulin-naïve patients (30–40%, Table 1). Combination regimens allow use of less insulin injections, which may ease titration of the insulin dose and compliance (12,15,41). These benefits must be balanced against the side effects of oral drugs and, in some countries, their cost. Abnormal renal or liver function also limits the use of many oral agents. Weight gain seems proportional to the number of insulin injections used (12,15,31,32) and can be counteracted by inclusion of metformin in the treatment regimen. Metformin also seems to reduce the incidence of hypoglycemias (12), as does the use of the peakless long-acting insulin analog insulin glargine compared with NPH (40). These considerations and the need to treat not only hyperglycemia but also other risk factors in type 2 diabetes support the use of simple insulin combination regimens such as insulin glargine and metformin and or a sulfonylurea (40). The prevailing view that patients who are poorly responsive to such a regimen benefit from adding additional insulin injections is not supported by existing data. Instead, special emphasis should be placed on increasing the dose of the single long-acting insulin to a dose that normalizes the fasting glucose concentration.

    Figure 1 —

    Weight gain in previously insulin-treated patients during treatment with insulin combination regimens containing metformin (MET) (24,42,43), various sulfonylureas (SU) (26,44–51), and glitazones (GLIT) (52–54) and in insulin-naïve patients treated with insulin and MET (12), SU+MET (12,15,16), and SU (12,25,27–29).

    Figure 2 —

    Upper panel: Diurnal glucose profiles after 52 weeks of treatment of 423 patients using oral hypoglycemic agents with either insulin glargine (•) or NPH (○). Lower panel: The percentage of patients experiencing any symptomatic hypoglycemia (ALL) or nocturnal hypoglycemia (NOCTURNAL) in the same study (40).

    Table 1 —

    Studies comparing combination treatment regimens with insulin to insulin alone in insulin-naive type 2 diabetic patients

    Table 2 —

    Studies comparing combination treatment regimens with insulin to insulin alone in previously insulin-treated type 2 diabetic patients

    Table 3 —

    Changes in glycosylated hemoglobin and serum triglycerides during treatment with insulin alone as compared with combination therapies with insulin and oral agents in insulin-naive patients

    Table 4 —

    Changes in glycosylated hemoglobin and serum triglycerides during treatment with insulin alone as compared with insulin combination therapy in previously insulin-treated patients

    Table 5 —

    Studies comparing insulin combination regimens with different insulin injection regimens in type 2 diabetic patients (oral agents similar in all regimens)

    Table 6 —

    Simple algorithm to start insulin combination therapy in an insulin-naive patient treated with oral combination therapy


    • Address correspondence and reprint requests to Hannele Yki-Järvinen, MD, FRCP, Department of Medicine, Division of Diabetes, P.O. Box 340, 00029 HUS, Helsinki, Finland. E-mail: ykijarvi{at}

      Received for publication 19 October 2000 and accepted in revised form 2 January 2001.

      A table elsewhere in this issue shows conventional and Système International (SI) units and conversion factors for many substances.


    | Table of Contents