Summary of evidence: implications for clinical practice
Overall, the above data suggest that the basal-plus regimen is efficacious for improving glycemic control in patients with T2DM, with a low incidence of hypoglycemia and generally minor increases in bodyweight. Based on a small body of available evidence, the timing of short-acting insulin administration and use of different monitoring/titration approaches appear to have minimal clinical impact on efficacy and safety. When compared with premixed insulin, the basal-plus approach appears to result in largely comparable outcomes, whereas compared with basal-bolus regimen, it may result in generally non-inferior glycemic improvements with less weight gain, less hypoglycemia and fewer daily injections. A basal insulin/lixisenatide combination may offer several advantages over the basal-plus regimen.
Choice of patients for basal-plus
In the investigated set of studies, ongoing use of a basal insulin±OADs accompanied by an elevated HbA1c level was a common prerequisite for inclusion.8 13 14 20–24 26–28 This explains the low baseline FBG levels and minimal changes in FBG seen in these particular studies, as apparently FBG concentrations already were under adequate control by the basal insulin in the BOT setting. Logically, patients in studies including insulin-naive patients had the highest baseline FBG levels and the greatest reduction in FBG over the course of the study.15 16 However, the majority of this improvement occurred during the basal insulin run-in phase, prior to prandial insulin initiation. Consequently, in line with current guidelines,1 basal-plus patients in clinical studies are typically those with well-controlled fasting glucose levels, but inadequately high postprandial excursions leading to unsatisfactorily high HbA1c.
Tight glycemic control is particularly important in hospitalized patients, given the associations between hyperglycemia/hypoglycemia and higher rates of infection, increased length of hospital stay and mortality.29–32 Therefore, insulin regimens are commonly employed in non-critical-care patients, regardless of prior diabetes management approaches. Indeed, subjects in basal-plus inpatient studies were previously managed by diet, OADs only or low-dose insulin.18 25 The greatest risk of abrupt implementation of insulin therapy and tight titration in inpatients represents hypoglycemia. Accordingly, a basal-plus approach may be more appropriate than a full basal-bolus regimen in this particular population, supported by the lower rates of hypoglycemia reported by Umiperrez et al (5% vs 16% for basal-plus and basal-bolus patients, respectively).18 Conversely, little differences were observed between basal-plus and premixed insulin regimens in this context.25 However, the former allows for a greater degree of flexibility, which may have particular utility in a volatile hospital setting. In summary, any non-ICU hospitalized patient with suboptimal glycemic control may be a potential basal-plus candidate, although supporting literature is sparse.
A further consideration for all patients requiring insulin intensification is their individual risk of hypoglycemia. Two meta-analyses33 34 combining data from 4 of the 15 clinical studies included in the present systematic review (OPAL, ELENOR, POC and 1-2-3- trials)8 9 15 17 21 identified female gender, a longer diabetes duration and higher IGlar doses to be predictive of symptomatic hypoglycemia on a basal-plus regimen.33 34 Finally, we recommend that caution and close monitoring should be employed in such patients, with alternative intensification methods such as addition of lixisenatide meriting consideration.
Basal and bolus timing and approaches to titration
Times of the day for basal insulin administration were not prespecified in most of the trials, except the studies of Raccah, Jin, Tinhones and Gross, where insulin was injected in the evening.14 20 22 26 In most of the trials, the basal insulin was not further titrated on addition of the short-acting insulin, but rather kept at stable doses. Moreover, in all studies the long-acting basal insulin analogs IGlar and IDet as opposed to neutral protamin hagedorn (NPH) insulin were used. Therefore, daytime differences in basal insulin applications may be negligible.
Several basal-plus trials have compared the effect of administering short-acting insulin at different mealtimes. For both IGlar/IGlu and IGlar/ILis basal-plus regimens, reduction in HbA1c was found to be unaffected by the timing of short-acting insulin administration.8 14 This suggests that preferential dosing at breakfast, lunch or dinnertime has little bearing on glycemic efficacy, and that a degree of flexibility may be acceptable. However, in the subset of patients with baseline HbA1c>7%, achievement of a value below this target was more common when their preprandial IGlu bolus was administered prior to their main meal of the day.8 Consequently, administration of short-acting insulin with the patient’s largest meal may be recommended. This begs the question of how to define the main meal. The STEPWISE trial assessed outcomes of the basal-plus regimen when the main meal was defined as the “largest meal from the patient’s perspective” (with titrations based on preprandial blood glucose) compared with the “meal with the largest postprandial glucose excursion,” and found no significant differences between these groups.28 This suggests that either method is valid, although larger-scale studies would be helpful to confirm this.
A slightly smaller degree of weight gain appears to be associated with the use of ILis before a main evening meal compared with breakfast or lunch,14 suggesting that evening administration may be more attractive to patients for whom weight is of particular concern. However, this difference appears to be of minimal clinical significance, with the magnitude of the gains shown to be small and similar to those on BOT alone.21 Indeed, considering that more hypoglycemia has also been reported in patients administering their prandial insulin in the evening,8 14 patients may gain a greater safety benefit from consuming their main meal (and administering their short-acting insulin) at breakfast or lunchtime. These factors collectively need to be taken into account when establishing individual treatment strategies.
A number of factors have been used to determine the need for insulin dose titration or stepwise addition of additional prandial boluses in basal-plus patients. Basing treatment decisions on telecare monitoring or SMBG was found to result in comparable glycemic improvements, weight changes and hypoglycemia incidences,15 as was the use of preprandial or postprandial blood glucose readings.28 This suggests that any of the above may be appropriate for informing insulin requirements. Analysis of the economic value of each method alongside assessment of convenience and patient satisfaction would allow for more informed selection.
Basal-plus versus basal-bolus
The majority of studies found the basal-plus approach to be similar to the basal-bolus regimen in terms of change in HbA1c9 27 and daily blood glucose,18 with only the OSIRIS trial failing to report non-inferiority.20 However, there appears to be a consistent trend toward fewer basal-plus than basal-bolus patients meeting a <7% HbA1c target at around 6 months, despite not always reaching significance.9 20 27 Interestingly, the FullSTEP study found that by 32 weeks, this difference had disappeared, with comparable proportions attaining HbA1c <7%.27 This is in keeping with the idea that basal-plus is a more gradual approach to treatment intensification after BOT, likely requiring more time to achieve satisfactory glycemic control than the immediate intensive basal-bolus regimen. This should be taken into account when planning the duration of future basal-plus comparative studies, and also kept in mind by physicians, so that unnecessary delays do not hinder patient progress. Nevertheless, the basal-plus regimen has been shown by several studies to result in comparably less hypoglycemia (including severe hypoglycemia)9 18 27 and weight gain16 20 27 than basal-bolus, with fewer daily injections required. Only the OSIRIS trial failed to note a reduction in hypoglycemia, with only Davidson et al reporting no difference in weight gain.9 20 These are likely explained by differences in study design and study populations, such as the extent to prior insulin exposure, dosage and use of OADs. In summary, basal-plus regimens may be more acceptable to patients in need of, but resistant to, treatment intensification due to concerns over weight gain and hypoglycemia risk, as well as reluctance to multiple daily injections.
Basal-plus versus premixed insulin
Most studies comparing basal-plus and premixed insulin regimens show the former to be non-inferior for reduction in HbA1c,16 22 23 with an inpatient study reporting comparable improvements in daily blood glucose.25 Furthermore, at least comparable proportions of patients are reported to achieve HbA1c <7%,16 22 23 26 with hypoglycemia and weight gain also found to be similar by the majority of studies.16 22 23 25 26 On balance, it appears that premixed insulin and basal-bolus approaches result in similar improvements in glycemic control, with a comparable degree of safety. Furthermore, the number of necessary injections (two per day) is generally identical in each regimen, meaning that neither has a particular advantage in this respect. However, the addition of one prandial bolus to a previously established BOT regimen may be simpler to implement and allow a greater degree of flexibility in terms of dose titrations. Thus, convenience and individual patient preference are likely to be the most relevant factors when deciding between basal-plus and premixed insulin regimens.
Basal-plus versus GLP-1 RA
In recent years, the addition of a GLP-1 RA to optimized basal insulin has been explored for improving glycemic control in patients with T2DM. Although several studies have compared a basal/GLP-1 RA regimen to a basal-bolus regimen, only one trial (GetGoal Duo-2) appears to have included a basal-plus arm.24 In this particular study, reductions in HbA1c were non-inferior in patients on lixisenatide compared with those on a basal-plus regimen. In accordance with the published literature,35 however, a small amount of weight loss (−0.6 kg) was seen in patients on lixisenatide, while basal-plus patients experienced a small weight gain (+1 kg). In addition, lixisenatide was associated with lower rates of hypoglycemia, with a complete absence of severe symptomatic hypoglycemia. This suggests that the main aims of antidiabetic treatment (namely improved glycemic control without excessive weight gain or hypoglycemia) may be better met by a basal/GLP-1 RA than basal-plus regimen. However, GLP-1 RAs have been associated with gastrointestinal side effects, affecting 35.2% versus 8.6% of the lixisenatide and basal-plus patients in the GetGoal Duo-2 Trial, respectively.24 Increased heart rate36 and kidney damage37 have also been associated in some studies with GLP-1 RAs. These adverse effects may significantly impact quality of life, adherence and persistence, undermining the advantage gained from fewer insulin injections compared with the basal-bolus approach. Consequently, all the above factors should be carefully considered when individualizing patient care plans.
Basal-plus insulin combinations
In all but two of the studies, the basal insulin used was IGlar 100 U/mL rather than IDet. This may partly reflect an earlier approval, with IGlar having received European Medicines Agency/Food and Drug Administration approval several years before IDet (2000/2000 vs 2004/2005). A meta-analysis comparing these two long-acting insulins found them to be comparable for target HbA1c achievement without hypoglycemia when used as part of a BOT regimen (RR=1.07; 95% CI: 0.87 to 1.33), as well as risk of symptomatic hypoglycemia (RR=0.99; 95% CI: 0.90 to 1.08).38 Similarly, pharmacodynamic studies have demonstrated very little difference in duration of action.39 However, IGlar has been associated with a lower rate of adverse events leading to treatment discontinuation (RR=0.40; 95% CI: 0.24 to 0.69)38 and a lower daily insulin dose requirement (RR=0.29; 95% CI: 0.25 to 0.32).40–42 The latter points toward potential cost savings and may be an additional explanation for the greater use of IGlar in the present group of studies. IDet may result in less weight gain (0.6±2.5 vs 4.2±4.1 kg, p=0.004),43 and all these factors must be considered when selecting basal insulins. Head-to-head studies comparing IGlar and IDet basal-plus regimens would be informative. The use of the newer, long-acting insulins—degludec and IGlar-300—are outside the context of this review but warrant similar investigation.
When deciding on the best prandial insulin to add to a BOT regimen, several rapid-acting analogs (IGlu, IAsp and ILis) are available. IGlu has been shown to provide similar or superior improvements in glycemic control compared with regular human insulin at similar doses, with less severe postadministration excursions and reduced hypoglycemia.44 45 In the studies included in the present analysis, the most recently approved analog IGlu was most commonly used in the basal-plus arm followed by ILis and IAsp. This preference may also be a reflection of its slightly faster onset of action compared with the latter two analogs, as demonstrated by a number of comparative studies.46–48 Theoretically, this allows more flexible bolus administration at mealtime.49 It is difficult to draw evidence supporting the use of any one particular analog from the present studies due to their heterogeneity. Direct head-to-head comparisons of different basal/prandial insulin combinations in a larger population would be necessary to address this question.
Limitations
This review is based on studies identified through a PubMed search and the recommendations of experts. Other data sources, such as Embase, Cochrane and Medline, may have identified other studies that could have been included in this manuscript. There is a high degree of heterogeneity between the included studies in terms of design (including aims, sample sizes, durations, inpatient vs outpatient settings, definitions of hypoglycemia and insulin combinations) and baseline patient characteristics (age, glycemic levels, use of insulin and OADs). Furthermore, the definition of basal-plus is variable, with some studies referring to the addition of just one prandial bolus per day, others to a stepwise addition of up to three boluses based on target achievement over time and some to the administration of short-acting insulin only where prandial blood glucose readings indicate its necessity. This begs for standardization of terms used in the field of diabetes to avoid confusion in future. This heterogeneity contributed complexity to a qualitative evaluation and precluded the performance of a meta-analysis. Furthermore, most of the included studies were RCTs. While this suggests high-quality data and validity of interstudy comparisons, tight inclusion criteria mean that study samples represent selected populations, limiting their generalizability to the overall population in daily clinical practice. Of note, a lack of blinding in most of the trials also means that the effect of experimenter bias cannot be ruled out.