Effect of Insulin Degludec Versus Insulin Glargine U100 on Hypoglycemia in Hispanic Patients With Type 2 Diabetes: Results From the SWITCH 2 Trial

SWITCH 2 was a 2 × 32-week, double-blind, multicenter, treat-to-target, two-period crossover trial (ClinicalTrials.gov: NCT02030600). The design and primary results for the SWITCH 2 study have been reported previously (24). In brief, adults with type 2 diabetes for ≥26 weeks, A1C ≤9.5% (80 mmol/mol), BMI ≤45 kg/m², and treatment with a basal insulin with or without oral antidiabetic drugs for ≥26 weeks and at risk of developing hypoglycemia were included, reflecting the general type 2 diabetes population. When enrolling, patients were asked to self-identify as Hispanic/Latino, if applicable. Patients were randomized 1:1 to receive either degludec for 32 weeks followed by glargine U100 for a further 32 weeks or glargine U100 for 32 weeks followed by degludec for a further 32 weeks, all once daily. Each 32-week treatment period consisted of a 16-week titration period and a 16-week maintenance period (Supplementary Figure 1).

In the SWITCH 2 trial, overall symptomatic hypoglycemic episodes were defined according to the American Diabetes Association definition as those requiring the assistance of another person (severe [26]) and/or blood glucose–confirmed (<56 mg/dL [3.1 mmol/L]) episodes accompanied by typical symptoms of hypoglycemia. Symptomatic hypoglycemia with onset between 00:01 a.m. and 05:59 a.m. was classified as nocturnal. All reported episodes of severe hypoglycemia were adjudicated by an independent external committee (24).

In these post hoc analyses, the safety of degludec versus glargine U100 in Hispanic patients and non-Hispanic patients was assessed by comparing rates of overall symptomatic hypoglycemia, nocturnal symptomatic hypoglycemia, and severe hypoglycemia for degludec versus glargine U100 during the maintenance (weeks 17–32 and 49–64) and full (weeks 1–32 and 33–64) treatment periods. Rates of adverse events (AEs) were also analyzed for Hispanic and non-Hispanic patients. The efficacy of degludec versus glargine U100 was assessed by measuring the change from baseline in A1C, fasting plasma glucose (FPG), and prebreakfast self-measured plasma glucose (SMPG) levels for degludec versus glargine U100 in Hispanic and non-Hispanic patients. Baseline was defined as week 0 for treatment period 1 and week 32 for treatment period 2. Daily insulin dose with degludec versus glargine U100 was assessed at the end of treatment period 1 (week 32), the end of treatment period 2 (week 64), and over the total treatment period.

Post hoc analyses of safety (hypoglycemia), efficacy, and insulin dose were based on the full analysis set (all randomized patients [except for one patient excluded due to an unsigned casebook]). Descriptive summaries of safety (hypoglycemia, AEs) and insulin dose were prepared for the safety analysis set (patients receiving at least one dose of investigational product or comparator), and efficacy summaries were prepared for the full analysis set. Differences between degludec and glargine U100 were analyzed statistically within each subpopulation (Hispanic or non-Hispanic), and the results were compared descriptively between subpopulations. The number of hypoglycemic episodes and change from baseline in A1C were analyzed as per the prespecified primary models used in SWITCH 2 (24). Daily insulin dose was analyzed as per the post hoc analysis reported in Wysham et al. (24).

In SWITCH 2, 36.4% (262/720) of patients were Hispanic, and 63.6% (458/720) were non-Hispanic. The disposition of Hispanic and non-Hispanic patients is shown in Supplementary Figure 2. Baseline characteristics for Hispanic and non-Hispanic patients are summarized in Table 1. Sex, age, BMI, and A1C were generally comparable between Hispanic and non-Hispanic patients, whereas FPG tended to be higher for Hispanic patients at baseline.

In Hispanic patients, the rate of overall symptomatic hypoglycemia was numerically lower with degludec compared with glargine U100, but differences were not statistically significant (Figure 1 and Table 2). The rate of nocturnal symptomatic hypoglycemia was significantly lower with degludec compared with glargine U100 in the maintenance (51.7 vs. 84.2 episodes/100 patient-year exposure [PYE]; estimated rate ratio [ERR] = 0.63 [95% CI 0.41–0.99]; P = 0.043) and total treatment period (56.0 vs. 71.3 episodes/100 PYE; ERR = 0.71 [95% CI 0.51–0.98]; P = 0.035) (Figure 1 and Table 2). ERRs for severe hypoglycemia during the maintenance period could not be calculated because of the small number of events reported. Rates of severe hypoglycemia were numerically lower with degludec compared with glargine U100 during the full treatment period, but differences were not statistically significant (Figure 1).

In non-Hispanic patients, there were statistically significant reductions in the rates of both overall and nocturnal symptomatic hypoglycemia with degludec versus glargine U100 during the maintenance period (overall symptomatic hypoglycemia 195.8 vs. 312.3 episodes/100 PYE; ERR = 0.63 [95% CI 0.53–0.74]; P <0 .001; nocturnal symptomatic hypoglycemia 57.2 vs. 98.9 episodes/100 PYE; ERR = 0.57 [95%.CI 0.42–0.77]; P <0.001). These results were consistent with the full treatment period (Figure 1 and Table 2). Rates of severe hypoglycemia were numerically lower with degludec versus glargine U100 during both treatment periods, but differences were not statistically significant (Figure 1). Rates of hypoglycemia appeared to be lower in Hispanic versus non-Hispanic patients, in both treatment groups, and across hypoglycemia definitions (Table 2).

During the trial, AEs were reported in 137 (53.1%) Hispanic patients and 364 (80.0%) non-Hispanic patients, at rates of 225.9 events/100 PYE in Hispanic patients and 414.1 events/100 PYE in non-Hispanics. A similar pattern was observed for serious AEs, with rates of 14.4 versus 27.5 events/100 PYE. The most common AEs in both populations were infections and infestations (Hispanic patients 71/258, 27.5%; non-Hispanic patients 234/455, 51.4%), whereas the most common serious AE in both populations was hypoglycemia (Hispanic patients 3/258, 1.2%; non-Hispanic patients: 8/455, 1.8%). Five treatment-emergent AEs had a fatal outcome in three Hispanic patients and two non-Hispanic patients, all during treatment with glargine U100. There were a further two deaths resulting from nontreatment-emergent AEs during follow-up. All but one of the fatal AEs were judged by the investigator as unlikely to be related to the trial product; the remaining fatal AE, sepsis in a 60-year-old Hispanic female, was judged possibly related to glargine U100. An overview of AEs is presented in Supplementary Table 1.

Reductions in A1C over time are shown for each patient population in Figure 2A and B. There were no significant differences between the change in A1C achieved with degludec or glargine U100 in either patient group or treatment period (Supplementary Table 2). At baseline and throughout the trial, FPG values tended to be higher in Hispanic patients compared with non-Hispanic patients, but reductions in FPG with degludec were similar to those with glargine U100 (Figure 2C and D). In Hispanic and non-Hispanic patients, mean prebreakfast SMPG level decreased in both degludec and glargine U100 groups during the first 16 weeks of the SWITCH 2 trial and remained stable for the remainder of the trial (Supplementary Figure 3A and B).

Over the total treatment period, insulin dose was lower with degludec versus glargine U100 across patient groups (estimated treatment ratio [ETR] = 0.97 [95% CI 0.94–1.00]; P = 0.046 and ETR = 0.95 [95% CI 0.94–0.98]; P <0.001 for Hispanic patients and non-Hispanic patients, respectively) (Supplementary Tables 3 and 4). A pattern of a higher insulin dose for both degludec and glargine U100 was observed in Hispanic patients compared with non-Hispanic patients as the trial progressed (Supplementary Tables 3 and 4).

In these post hoc analyses of the SWITCH 2 trial, the lower risk of hypoglycemia with degludec versus glargine U100 observed in Hispanic patients was generally consistent with that observed in non-Hispanic patients. All ERRs were numerically or significantly in favor of degludec versus glargine U100 during the maintenance and full treatment periods across different types of hypoglycemia in Hispanic and non-Hispanic patients. These new analyses support the overall findings of the SWITCH 2 trial, where treatment with degludec compared with glargine U100 resulted in significant reductions in the rates of overall symptomatic hypoglycemia and nocturnal symptomatic hypoglycemia over the 16-week maintenance period and the full treatment period (24). These new analyses provide more evidence of the favorable safety profile of degludec in various populations of patients with type 2 diabetes (27–29). In SWITCH 2, noninferiority of degludec compared with glargine U100 for A1C levels was confirmed for both treatment periods in the overall trial population (24). We report similar findings in these post hoc analyses of the Hispanic and non-Hispanic patient populations. Treatment with degludec versus glargine U100 resulted in similar improvements in glycemic control, but at a lower daily insulin dose over the total treatment period in favor of degludec, regardless of Hispanic or non-Hispanic ethnicity.

There have been numerous reports in the literature on the differences between Hispanics and non-Hispanics with respect to risk factors for diabetes (5,7,30), with guidelines including Hispanic/Latino ethnicity as a risk factor for prediabetes and type 2 diabetes (11). In addition, negative attitudes toward the use of diabetes therapy, in particular insulin, are common among Hispanics (20,22). In the current study, differences in the outcomes between Hispanic and non-Hispanic patients were compared descriptively, and some patterns were noted.

The mean rate of hypoglycemic episodes tended to be lower in Hispanics versus non-Hispanics across all hypoglycemia types, despite a trend toward a higher insulin dose/kg of body weight in Hispanic patients. Observed rates of AEs were consistently lower in Hispanic patients compared with non-Hispanic patients across system organ classes. A slightly higher proportion of patients in the Hispanic group had a history of hypoglycemia unawareness at baseline (20.2 vs. 16.6% in the Hispanic and non-Hispanic groups, respectively); this may have resulted in fewer reports of symptomatic hypoglycemia in this group. Alternatively, lower rates of hypoglycemia in Hispanic versus non-Hispanic patients could be a consequence of their slightly poorer glycemic control. Improvements in A1C tended to be smaller in Hispanic patients in comparison to their non-Hispanic counterparts after 32 weeks of treatment with degludec or glargine U100. Higher FPG values were also observed in Hispanics compared with non-Hispanics throughout the trial, although change from baseline was similar between the two groups. In contrast to a pattern of higher FPG values in Hispanic patients during the trial, self-reported prebreakfast SMPG values tended to be lower in Hispanic patients compared with non-Hispanic patients as the trial progressed, which would align well with intensive insulin titration but is puzzling.

Previous reports have indicated that Hispanic patients have poorer glycemic control than non-Hispanic patients (13,14) and that they are less likely to adhere to insulin therapy than their non-Hispanic white counterparts (20). Reports have suggested that lower adherence may be related to fear of hypoglycemia or to an inability to afford insulin (20). In the current analyses, however, observed mean doses/kg of both degludec and glargine U100 appeared to be higher in Hispanic versus non-Hispanic patients at the end of treatment; thus, it seems unlikely that a failure to intensively titrate insulin could have accounted for the lower reported rates of hypoglycemia. However, treatment adherence to diabetes medications is reported to be lower in Hispanics versus non-Hispanic white patients (31), highlighting the potential for differences in adherence rates between patient groups during the trial.

Taken together, the higher FPG values throughout the trial, smaller improvement in A1C after 32 weeks of treatment, and higher mean dose/kg of degludec and glargine U100 at the end of treatment observed in Hispanic versus non-Hispanic patients highlight the potential for greater insulin resistance in Hispanic patients. Ferrannini et al. (5) reported a 27% lower insulin sensitivity of glucose uptake in Mexican American versus Caucasian patients with normal glucose tolerance. Although speculative, there is the potential for increased insulin resistance in Hispanic patients with type 2 diabetes in comparison to their non-Hispanic white counterparts, which may have been reflected in the lower incidence of hypoglycemia observed in Hispanics during the trial. This contrasts with data on U.S. emergency department visits for hypoglycemia (1993–2005), where rates per 1,000 people with diabetes were almost twice as high in Hispanic versus non-Hispanic patients (32). In the United States, Hispanics face socioeconomic barriers to health care access, i.e., national data from 2011 to 2013 indicate that 41.5% of Hispanics were found to be lacking health insurance, compared with 15.1% of non-Hispanic whites (18). Hispanics are over two times more likely to live under the U.S. poverty line than non-Hispanic whites (18), with an increased incidence of severe hypoglycemia previously reported for patients of lower economic status (32).

Albeit from a large study popu-lation, these are post hoc analyses. This limits the interpretation of the findings, as the Hispanic and non-Hispanic patients were not two randomized groups. The small number of severe hypoglycemic events reported precluded the calculation of ERRs (degludec/glargine U100) for this definition during the maintenance period.

Hypoglycemia has negative effects on patient health and quality of life (33), while posing a significant economic burden through loss of productivity and increased health care costs (34,35). Fear of hypoglycemia has been reported as a significant barrier to the initiation of insulin therapy for Hispanic patients (21,22) and contributes to psychological resistance to initiating insulin (19). The evidence from the current analyses showing that lower rates of hypoglycemia with degludec versus glargine U100 are seen in both Hispanic and non-Hispanic patients could help to counteract this fear.

In conclusion, safety and efficacy results from the Hispanic subpopulation of SWITCH 2 were generally consistent with those from the non-Hispanic subpopulation. In Hispanic patients, there was a numerically lower risk of overall hypoglycemia and a significantly lower risk of nocturnal hypoglycemia with degludec versus glargine U100, and both degludec and glargine U100 led to similar improvements in glycemic outcomes. Evidence of lower rates of hypoglycemia with degludec versus glargine U100 may help to counteract psychological insulin resistance in Hispanics with type 2 diabetes, thereby removing a barrier to insulin therapy and improving care.

The authors would like to thank Kadriye Kaplan and Eskil Brandt Kreiner (Novo Nordisk) for their review of, and input to, the manuscript. Medical writing support was provided by Anna Campbell, PhD, and Tom Vizard, PhD, and technical editing support by Richard McDonald, of Watermeadow Medical, an Ashfield company, part of UDG Healthcare plc, funded by Novo Nordisk.

This study was funded by Novo Nordisk.

L.C. has served on an advisory panel for Intarcia Therapeutics and on a speaker’s bureau for Novo Nordisk. A.B. has served on advisory panels for Abbott, Janssen, and Sanofi and speaker’s bureau for Abbott, Sanofi, and AstraZeneca and received research support from Boehringer Ingelheim, Bristol-Myers Squibb, Novo Nordisk, Eli Lilly and Company, Dexcom, Medtronic, Sanofi, Mylan, Duke Clinical Research Institute, Janssen, Jaeb Center for Health Research, GlaxoSmithKline, Orexigen Thrapeutics, Hygieia Research, University of Oxford, and AbbVie. R.D.l.R. has received speaking honoraria from Novo Nordisk, Boehringer Ingelheim, and Sanofi and clinical trial support from Novo Nordisk, Janssen Pharmaceuticals, Sanofi, GlaxoSmithKline, Elcelyx Therapeutics, and Merck. C.H.W. has served as a speaker and/or an advisor for Abbott, AstraZeneca, Boehringer Ingelheim, Eli Lilly, Insulet, Janssen, Novo Nordisk, and Sanofi. L.N.T. and S.H.Ø. are employees of Novo Nordisk. A.P.-T. has served on advisory panels for AstraZeneca, Dexcom, Lilly, Merck, Novo Nordisk, and Sanofi and received research support from Dexcom, Janssen, Lilly, Mylan, and Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

L.C., A.B., R.D.l.R., C.H.W., and A.P.-T. contributed to the analysis and manuscript writing for this study. L.N.T. and S.H.Ø. contributed to the design, analysis, and manuscript writing for this study. L.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Some of the data included in the manuscript have been previously presented in poster form at the International Diabetes Federation Congress, 4–8 December 2017, Abu Dhabi, United Arab Emirates.

There are changes between this version of the manuscript and the version initially posted online on 26 September 2018. After this paper was accepted for publication and published online, it came to our attention that there was an error in Figure 2. In this updated version of the paper, the secondary y-axis labels have been corrected in Figure 2A and B to display values in mmol/mol rather than %. Data for Hispanics and non-Hispanics have also been clearly identified. The caption for Figure 2 has been updated to clearly detail ETDs.