ORIGINAL ARTICLE

original article

Diabetes, Obesity and Metabolism 16: 1070–1077, 2014. © 2014 John Wiley & Sons Ltd

Safety, efficacy and weight effect of two 11𝜷-HSD1 inhibitors in metformin-treated patients with type 2 diabetes T. Heise1 , L. Morrow2 , M. Hompesch2 , H.-U. Häring3 , C. Kapitza1 , M. Abt4 , M. Ramsauer4 , M.-C. Magnone5 & S. Fuerst-Recktenwald5 1 Profil, Neuss, Germany 2 Profil Institute for Clinical Research, Inc., Chula Vista, CA, USA 3 Internal Medicine IV, University of Tübingen, Tübingen, Germany 4 F. Hoffmann-La Roche, Ltd., Basel, Switzerland 5 Roche Pharmaceutical Research and Early Development, Roche Innovation Center Basel, Basel, Switzerland

Aims: We assessed safety and efficacy of two selective 11𝛽-HSD1 inhibitors (RO5093151/RO-151 and RO5027383/RO-838) in this randomized, controlled study in metformin-treated patients with type 2 diabetes.

Methods: Patients either received placebo (N = 21), RO-151 BID 5 mg (N = 24) or 200 mg (N = 20) or RO-838 QD 50 mg (N = 21) or 200 mg (N = 24) for 28 days. Metabolic assessments comprising of nine-point plasma glucose profiles, oral glucose tolerance tests and determination of metabolic biomarkers including insulin, C-peptide, glucagon, HbA1c and lipids were done at baseline and end of treatment. Results: Despite the short treatment duration, both RO-151 and RO-838 showed trends for improved HbA1c and consistent reductions in body weight (–0.86 to –1.67 kg) exceeding those observed with placebo (–0.28 kg, p = 0.019 for 200 mg RO-151 vs. placebo). Insulin sensitivity parameters (e.g. HOMA-IR and Matsuda-Index) improved non-significantly with 200 mg RO-151. Lipid parameters did not consistently improve with either compound, but RO-838 led to non-significant increases in triglycerides and VLDL-cholesterol versus placebo. Both compounds were well tolerated and showed inhibitory effects on 11𝛽-HSD1 activity based on urinary corticosteroid excretion. As reported for other 11𝛽-HSD1-inhibitors increased concentrations of ACTH and adrenal androgen precursors were found with RO-151, but not with RO-838. Conclusions: Slight metabolic improvements were seen, in particular with RO-151 high dose, however, the observed changes often did not reach statistical significance and were not clearly dose dependent. Studies of longer duration are needed to further investigate potential benefits and risks of these compounds. Keywords: cortisone, antidiabetic drug, glycaemic control, phase I-II study, randomized trial, type 2 diabetes Date submitted 30 December 2013; date of first decision 20 February 2014; date of final acceptance 8 May 2014

Introduction With the growing epidemic of obesity and type 2 diabetes mellitus (T2D), often accompanied with concomitant medical conditions such as insulin resistance, dyslipidaemia and hypertension, there is an increasing need for treatment options that address health issues beyond hyperglycaemia and that ideally avoid the risk of hypoglycaemia [1–3]. Glucocorticoids are one of the most important human hormones opposing the action of insulin by promoting differentiation and lipolysis of adipose tissue, and by increasing triglyceride storage and glucose production in the liver. Elevated plasma glucocorticoid levels lead to visceral obesity, insulin resistance, dyslipidaemia, hypertension and ultimately to diabetes. However, elevated circulating cortisol levels are not generally seen in patients with obesity [4,5]. Glucocorticoid levels in target tissues (liver and adipose) depend on circulating steroid concentrations (tightly controlled by the Correspondence to : Tim Heise, MD, Profil Institut für Stoffwechselforschung GmbH, Hellersbergstraße 9, D-41460 Neuss, Germany. E-mail: [email protected]

hypothalamic–pituitary–adrenal axis [HPA]) as well as on 11𝛽-hydroxysteroid dehydrogenases (11𝛽-HSDs), which control the amount of cortisol at a cellular level. In metabolic tissues such as liver and adipose tissue, 11𝛽-HSD1 converts the inactive steroid cortisone into the active steroid cortisol whereas 11𝛽-HSD2, mainly expressed in the kidney, inactivates cortisol by transformation into cortisone [6]. 11𝛽-HSD1 knock-out mice show improved glucose tolerance and insulin sensitivity in both liver and adipose tissue, reduced plasma triglycerides and improved lipoprotein composition compared with age-matched wild type controls [7]. In contrast, mice that express 11𝛽-HSD1 at elevated levels in adipose tissue exhibit visceral obesity, insulin resistance and hypertension [8,9]. Thus, inhibition of 11𝛽-HSD1 may offer a novel approach to improve hyperglycaemia and also positively affect co-morbidities of T2D such as hypertension, hyperlipidaemia or obesity. Indeed, first results with the 11𝛽-HSD1-inhibitor INCB13739 were promising showing significant reductions in HbA1c (glycated haemoglobin), homeostatic model assessment-insulin resistance (HOMA-IR) and body weight after 12 weeks in metformin-treated patients with T2D [10]. In addition,

original article

DIABETES, OBESITY AND METABOLISM

hyperlipidaemic patients showed improvements in total and low-density lipoprotein (LDL) cholesterol as well as in triglycerides. The compound did, however, lead to an increase in adrenocorticotropic hormone (ACTH) (usually within the normal range) which was not accompanied by changes in cortisol, testosterone (in men) and free androgen index (in women). Similar findings were reported for other 11𝛽-HSD1-inhibitors [11,12]. In addition, more recently published results with another 11𝛽-HSD1-inhibitor showed only modest improvements in HbA1c, blood pressure (BP) and body weight whereas neither fasting plasma glucose (FPG) nor postprandial plasma glucose (PPG) values were significantly lowered compared with placebo [13]. Other development programmes of 11𝛽-HSD1-inhibitors have already been abandoned [14]. In view of these conflicting results, it is still difficult to predict the future value of 11𝛽-HSD1-inhibitors for the treatment of diabetes or the metabolic syndrome, and differences between individual compounds of this class cannot be ruled out. RO5093151 (RO-151) and RO5027383 (RO-838) are selective and potent competitive inhibitors of human 11𝛽-HSD1. Both compounds had shown beneficial metabolic effects in rodents including improvements in FPG and PPG, body weight and liver triglycerides and were well tolerated in healthy subjects (Roche data on file). In addition, topical treatment with RO-151 accelerated wound healing in aged 11𝛽-HSD1-knockout (11𝛽-HSD1-KO) mice [15]. This study was designed as an exploratory placebo-controlled proof-of-concept trial to evaluate safety and efficacy of RO-151 and RO-838 in patients with T2D on metformin monotherapy. The objective was to compare different doses (high and low dose) as well as two different 11𝛽-HSD1-inhibitors to identify the safest and most effective compound for potential further development.

Materials and Methods This was a randomized, double-blind, five-arm parallel group, placebo-controlled 4-week study conducted in five centres (two in Germany, one in Austria and two in USA). The study consisted of a screening period, a 4-week pre-randomization period (for washout of oral anti-hyperglycaemic and weight-lowering medications other than metformin), a 4-week double-blind treatment period (with the first day of study drug intake being labelled as day 1), and a follow-up visit. The study was conducted in accordance with the Declaration of Helsinki and was approved by the local Ethical Committees and Institutional Review Boards at the participating sites. All patients provided written informed consent before screening. Male and female patients (37–65 years) with T2D, body mass index (BMI) between 26 and 42 kg/m2 and HbA1c ≥7.0% and ≤10.0% treated with stable metformin doses for at least 3 months were eligible. Exclusion criteria included history of diabetic ketoacidosis, proliferative diabetic retinopathy, autoimmune disease or chronic inflammatory condition, glucose-6-phosphate dehydrogenase deficiency, previous insulin therapy, treatment with thiazolidinedione or a dual peroxisome proliferator-activated receptor (PPAR) 𝛼/𝛾 agonist within 6 months of screening, lipoprotein modifying

Volume 16 No. 11 November 2014

therapy (fibrates, niacin) within a month from screening or corticosteroid therapy for more than 2 weeks within 3 months prior to screening. FPG >240 mg/dl prior to randomization or uncontrolled hypertension with systolic blood pressure (SBP) ≥160 mmHg and/or diastolic blood pressure (DBP) ≥100 mmHg at the time of screening also led to exclusion. Patients were randomized to receive 5 or 200 mg RO-151 twice daily (total daily dose of 10 or 400 mg) or 50 or 200 mg RO-838 once-daily or matching placebo all in combination with metformin. Patients were hospitalized from days –2/–1 to 2 (randomization visit) and from days 27 to 29 (end of treatment visit). During these in-house periods standardized meals were provided and several metabolic assessments were performed, in particular nine-point plasma glucose profiles (days –1 [or day –2 for patients participating in the adipose tissue biopsy], 1, 27, 28), oral glucose tolerance tests (OGTT) over 240 min (days –2, 26), the determination of FPG and PPG, insulin, C-peptide and glucagon concentrations (days –2, 14, 26), determinations of adiponectin, high sensitivity C-reactive protein (hs-CRP) and intact pro-insulin (days –2, 14, 26), determination of HbA1c (screening, days –1, 28) and lipids (screening, days –1, 14, 28). Safety assessments comprised of safety laboratory (screening, days –1, 2, 7, 14, 28, follow-up), electrocardiogram (ECG, screening, days –1, 1, 2, 14, 28, follow-up), vital signs assessments and 24 h ambulatory blood pressure monitoring (ABPM, using SpaceLabs 90207 device, Spacelabs Healthcare, Atlanta, GA, USA) (Days –2, 26) and assessment of HPA function with 24-h cortisol profiles and an ACTH stimulation test (intravenous injection of 250 μg ACTH [Synacthen ]) and measurement of cortisol and androgen profiles up to 1.5 h after ACTH administration on days –2, 14, 26. Laboratory efficacy parameters were determined by a central and Good Laboratory Practise (GLP)-certified laboratory (Quotient Bioresearch, Cambridgeshire, UK) using validated assays. Batch acceptance criteria were defined prior to analyses and data were only used if these acceptance criteria were met. The only exception to this rule was the glucagon assay which showed a rather high inter-assay variance. However, the full eight-point glucagon profiles of one individual were usually analysed within one batch and reproducibility of the results was acceptable. All sites had a thorough documentation on sample collection and sample handling, and no relevant pre-analytical errors were detected.

®

Statistical Methodology The primary efficacy endpoint was the absolute change in mean daily plasma glucose (MDPG) from baseline (day –1) to day 27 of the treatment period. MDPG concentration was calculated as area-under-the-curve. Insulin sensitivity was assessed from established OGTT-derived parameters (e.g. Matsuda-Index) and HOMA-IR [16,17]. Efficacy analyses were based on the intent-to-treat (ITT) population, which included all randomized patients. The safety population consisted of all patients who received at least one dose of study medication. A mixed model analysis of variance (anova) was used to estimate the effects of the high and low doses of RO-151 and RO-838 on the primary endpoint. Fixed

doi:10.1111/dom.12317 1071

original article

DIABETES, OBESITY AND METABOLISM

effect terms included baseline MDPG, treatment, and the stratification factors, gender and use of statin therapy. Treatment with anti-hypertensives was added as a covariate. No corrections for multiple testing were performed in this exploratory trial. Formal comparisons of RO5093151 and RO5027838 to placebo were carried out for the higher doses. Comparisons at the lower doses were performed and interpreted in an exploratory way. Point estimates and 95% confidence intervals (CIs) were calculated using contrasts from the anova. Similar analyses were undertaken for the secondary efficacy parameters and safety markers although no formal hypothesis testing was performed. Unless otherwise mentioned, data are given as mean values ± standard deviation (s.d.). A sample size of 110 patients was chosen to ensure that at least 100 patients completed the study. The sample size was based on the publication by Ahrén et al. [18] that reported a standard error of the difference in mean plasma glucose compared to placebo of 0.1. Assuming an inflation of this figure of up to 50% due to rounding error, and no change of mean plasma glucose in the placebo arm, a sample size of 20 evaluable patients per treatment group would give at least 90% power to detect a change from baseline in mean plasma glucose at day 28 of 1.1 mmol/l compared to placebo.

Results Demographics and Baseline Characteristics A total of 359 patients were screened of whom 110 were randomized. Four patients were withdrawn prematurely from

treatment; two patients in the low dose RO-838 group, one patient in the high dose RO-838 group, and one patient in the low dose RO-151 group. One of these withdrawn patients refused further treatment, another one was lost for follow-up, and two patients were withdrawn due to insufficient therapeutic response. The demographic and baseline characteristics in each of the five treatment groups are provided in Table S1. There were no meaningful differences across the treatment groups apart from a slightly higher weight in the placebo group (however, BMI was comparable across treatment groups).

Efficacy Main efficacy results are summarized in Table 1. There were no significant changes in the primary endpoint MDPG nor in FPG values. However, trends to improvements versus baseline values were seen for both RO-151 and RO-838 in several metabolic parameters including HbA1c (Figure 1). The difference in HbA1c-values versus placebo approached statistical significance for the lower dose of RO-151 with a mean difference of –0.37% (p = 0.053). In addition, small reductions in body weight were consistently observed in all treatment groups (Figure 2). Least squares (LS) mean estimates of the reduction in body weight at week 4 were –0.28 kg in the placebo group, –0.86 and –1.08 kg in the low and high dose RO-838 groups, respectively, and –1.11 and –1.67 kg in the low and high dose RO-151 groups, respectively. These changes reached statistical significance in all four active treatment groups at the end of treatment but not in the placebo group (Figure 2). After correcting for placebo, the change from baseline remained statistically

Table 1. Efficacy results, expressed as mean changes from baseline to 4 weeks with corresponding 95% confidence intervals.

Parameter

Placebo

RO-151 (5 mg bid)

RO-151 (200 mg bid)

RO-838 (50 mg qd)

RO-838 (200 mg qd)

Mean daily PG (mg/dl) HbA1c (%) Fasting Assessments Glucose (mg/dl) Insulin (% change) C-peptide (% change) Glucagon (% change) Adiponectin (% change) Hs-CRP (% change) Pro-insulin (% change) Postprandial Assessments Insulin (% change) C-peptide (% change) Glucagon (% change) OGTT Glucose-AUC0–240 (mg/dl) Glucose-Cmax (mg/dl) 2h-Glucose (mg/dl) Insulin-Cmax (pmol/l) Insulin-AUC0–240 (pmol/l) Glucagon-Cmax (ng/l) Matsuda-Index (L2 /[10 × mg × pmol]) HOMA-IR

4 [–9;18] 0.1 [–0.2;0.4]

6 [–7;19] –0.2 [–0.5;0.1]

6 [–9;20] 0.0 [–0.3;0.3]

–3 [–17;12] 0.1 [–0.3;0.4]

3 [–10;16] –0.1 [–0.4;0.2]

9 [–3;20] –12 [–29;9] 1 [–8.4;11.3] –18 [–35;3] –1 [–12;12] –3 [–25;26] 19 [–9;56]

1 [–10;12] 5 [–14;29] 10 [0.3;20.9]+ 8 [–11;31] –19 [–28;–9]+ * –10 [–30;15] 23 [–4;58]

2 [–10;14] –14 [–31;7] 2.3 [–7.5;13.1] –19 [–35;0] –9 [–20;3] –5 [–27;24] –20 [–42;10]

3 [–9;15] –8 [–26;15] –0.3 [–9.8;10.1] –7 [–30;24] –3 [–15;10] 50 [15;95]+ * 0 [–25;32]

–5 [–16;6] –27 [–40;–10]+ –1.6 [–10.3;8.0] 9 [–16;41] –12 [–21;–1]+ 19 [–7;53] –3 [–24;24]

–25 [–42;–3]+ 0 [–11;12] –23 [–39;–2]+

0 [–22;28] 7 [–4;20] 10 [–10;35]

–25 [–43;–3]+ –6 [–17;6] –18 [–37;7]

–20 [–38;4] 6 [–6;20] –13 [–41;28]

–29 [–44;–9]+ –1 [–12;11] –23 [–40;–3]+

9 [–10;28] 9 [–13;32] –0.9 [–23;25] –58 [–141;25] –43 [–87; 1] –7 [–17;3] 0.00 [–0.16;0.16] –0.76 [–2.00;0.49]

8 [–11;26] 7 [–15;28] 0.9 [–22;24] –9 [–88;70] –11 [–54;32] –3 [–12;6] –0.05 [–0.20;0.10] 0.06 [–1.13;1.25]

5 [–15;25] –2 [–25;22] –2.8 [–28;22] –119 [–206;–32]+ –56 [–102;–9]+ –12 [–23;–2]+ 0.11 [–0.05;0.28] –1.45 [–2.74;–0.15]+

9 [–11;29] 14 [–9;37] 8.8 [–16;33] –50 [–135;34] –39 [–85;8] –1 [–13;11] 0.02 [–0.14;0.18] –0.33 [–1.61;0.94]

–5 [–24;15] –10 [–32;11] –18 [–41;5] –25 [–105;54] –33 [–77;11] 2 [–8;12] 0.27 [0.12;0.42]+ –1.12 [–2.31;0.08]

Changes are absolute changes from baseline unless otherwise indicated. Statistically significant changes (p < 0.05) from baseline are marked with ‘+ ’, significant placebo-adjusted differences from baseline are marked with ‘*’.

1072 Heise et al.

Volume 16 No. 11 November 2014

DIABETES, OBESITY AND METABOLISM

Figure 1. Absolute change in HbA1c (%) from baseline to Week 4. The error bars give 95% confidence intervals. There were no significant changes from baseline in either treatment group (i.e. the error bars did not extend beyond the red line of equality), however, the placebo-adjusted change for the low RO-151 dose group approached statistical significance (+ placebo-corrected LS mean – 0.37% [95% CI: –0.75, 0.01], p = 0.053)

Figure 2. Absolute change in body weight (kg) from baseline to week 4. The error bars give 95% confidence intervals. A continuous decrease in body weight was observed in particular in the high dose groups. Significant changes versus baseline are indicated by the error bars not extending beyond the red line of equality. As slight improvements in body weight were observed in the placebo group, too, the placebo-adjusted change in body weight was only significant for the high RO-151 dose group after 4 weeks of treatment (*placebo-corrected LS mean –1.39 kg, [95% CI: –2.54, –0.23], p = 0.019).

significant in the high dose RO-151 group (LS mean –1.39 kg [95% CI –2.54, –0.23 kg], p = 0.019). Other metabolic parameters such as fasting and postprandial insulin concentrations, postprandial glucagon concentrations and post-OGTT maximum insulin concentrations and area under the insulin concentration curve also improved versus baseline, in particular with the high dose of RO-151, but some of these parameters improved with placebo, too, so that no significant difference to placebo treatment was observed for any dose of the two compounds. Neither was there a consistent compound- or dose-related pattern (Table 1); significant changes versus baseline were sometimes restricted to the lower doses of RO-151 or RO-838. Those parameters that were determined repeatedly in the course of the study did not show a clear time-related trend to improvements. For instance, there were significant changes in PPG values at individual timepoints (day 7, 14 or 21) and for individual doses, but no compound showed

Volume 16 No. 11 November 2014

original article

Figure 3. Absolute change in postprandial plasma glucose (mg/dl) from baseline to week 4. The error bars give 95% confidence intervals. Significant changes versus baseline are indicated by the error bars not extending beyond the red line of equality. Postprandial plasma glucose values were obtained 2 h after standardized meals. The effects after 4 weeks were assessed from OGTTs, see Table 1.

a consistent trend over the whole treatment duration (Figure 3). Significant changes versus placebo were only observed for adiponectin levels for the low dose of RO-151 (end of treatment mean concentrations 4375 ± 2041 ng/ml vs. 5548 ± 2615 ng/ml [placebo], p = 0.009 for the difference in relative change from baseline levels) and hs-CRP for the low dose of RO-838 (increase by 50% to end of treatment mean concentrations of 6.3 mg/l, p = 0.017 for the relative change from baseline vs. placebo). Improvements in insulin sensitivity parameters were observed for the high dose groups with relative increases from baseline in the Matsuda-Index by 38% for RO-838 (p = 0.001 vs. baseline) and 26% for RO-151 (p = 0.027 vs. baseline) (Figure 4). However, the placebo-adjusted differences did only reach statistical significance for the high dose of RO-838 (p = 0.010), whereas the change in the high dose of RO-151 was not statistically significant (p = 0.308). Other insulin sensitivity parameters also showed significant improvements versus baseline, e.g. HOMA-IR for the high dose of

Figure 4. Relative (%) change in insulin sensitivity (Matsuda-Index, derived from OGTT insulin and glucose concentrations) from baseline to week 4. The error bars give 95% confidence intervals. Significant changes versus baseline are indicated by the error bars not extending beyond the red line of equality. Significant improvements in insulin sensitivity (i.e. significant increases in the Matsuda-Index from baseline) were observed for the high doses of both RO-151 and RO-838, however, the placebo-corrected changes did not reach statistical significance.

doi:10.1111/dom.12317 1073

original article

DIABETES, OBESITY AND METABOLISM

Table 2. Mean percent changes in lipid parameters from baseline with corresponding 95% confidence intervals. Parameter

Placebo

RO-151 (5 mg bid)

RO-151 (200 mg bid)

RO-838 (50 mg qd)

RO-838 (200 mg qd)

Total cholesterol HDL LDL Triglycerides VLDL Apolipoprotein A1 Apolipoprotein B Apolipoprotein C III Apo B/Apo A1 Free fatty acids

5 [–1;13] 4 [–3;13] 9 [0;19]+ 12 [–3;29] –13 [–27;3] –2 [–7;3] 16 [8;24]+ 15 [5;27]+ 23 [16;30]+ 27 [11;46]+

–3 [–9;4] 3 [–4;11] 0 [–8;8] 9 [–4;25] –13 [–26;2] –1 [–6;4] 7 [1;14]+ 11[3;21]+ 16 [9;22]+ 28 [12;46]+

–5 [–11;2]* –4 [–12;4] –6 [–14;3]* 14 [–2;32] 3 [–5;12] –6 [–11;–1]+ 3 [–4;11]* 10 [0;20] 15 [8;23]+ 25 [9;45]+

8 [0;16] 3 [–5;12] 9 [0;20] 17 [1;36]+ 2 [–16;24] 0 [–5;6] 17 [9;25]+ 16 [3;31]+ 19 [12;26]+ 20 [4;39]+

14 [7;22]+ 7 [–1;15] 17 [8;27]+ 40 [22;60]+, * 25 [5;48]+, * 2 [–3;7] 23 [16;31]+ 32 [19;45]+, * 22 [16;29]+ 25 [10;42]+

Statistically significant changes (p < 0.05) from baseline are marked with ‘+ ’, significant placebo-adjusted differences from baseline are marked with ‘*’.

RO-151 and post-OGTT insulin-AUC (area under the curve) value (Table 1). Numerical improvements were also seen for OGIS (oral glucose insulin sensitivity), pro-insulin and the insulinogenic indices ISIest (insulin sensitivity index) and MCRest (metabolic clearance rate of glucose) (data not shown), but none of the listed changes reached statistical significance when compared with placebo. No evidence of consistent treatment-related improvements in lipids was found for either compound (Table 2). While significant changes versus baseline were observed for some lipid parameters, these usually also occurred in the placebo group. However, some lipid parameters deteriorated with the high dose of RO-838 compared with placebo (Table 2). Compared to placebo, these changes reached statistical significance for triglycerides, very low density lipoprotein (VLDL) and apolipoprotein C III (apo C III). In contrast, the high dose of RO-151 led to small, but statistically significant placebo-corrected improvements from baseline in total cholesterol, LDL and apolipoprotein B (apo B).

11𝛽-HSD1-inhibition Based on the urinary concentrations of tetrahydrocortisol (THF) and tetrahydrocortisone (THE) the 11𝛽-HSD1 activity index (ratio of [𝛼THF + THF]/THE) showed clear and dose-dependent inhibitory effects on 11𝛽-HSD1 activity for both compounds, in particular with RO-151. The maximum inhibition observed on day 27 was 88% and 86% in females and males, respectively, in the low dose group, and 92% in either gender in the high dose group. Inhibition was sustained for the whole day at both doses; ≥79% at all timepoints in the low dose group and ≥87% in the high dose group. RO-838 showed less pronounced inhibition of 11𝛽-HSD1 activity, maximum inhibition was 54% and 55% in females and males, respectively, in the low dose group, and 69% and 62%, respectively, in the high dose group.

Safety RO-838 and RO-151 were well tolerated in this study. The incidence and the character of adverse events (AEs) were similar between the active treatment groups and the placebo group (incidence 33–45% across active treatment groups vs. 38% in

1074 Heise et al.

placebo). The majority of AEs were of mild intensity, transient in nature and resolved without sequelae. There were no serious AEs, no deaths and no withdrawals due to an AE. No clinically relevant effects on laboratory safety parameters were identified for RO-151 or RO-838 and no clinically relevant effects of RO-151 or RO-838 were observed on semi-supine heart rate, orthostatic changes in vital signs, ECG parameters or neurological assessments. A trend for a slight increase in BP values (semi-supine SBP and, to a lesser extent, for semi-supine DBP) from baseline was observed in the high dose RO-838 group, but only for the in-clinic measurements. Increases from baseline were observed at all time-points from day 2 onwards (mean change from baseline ranged from 1.4 to 5.1 mmHg for SBP and from 1.5 to 4.5 mmHg for DBP). This trend was not observed in the 24 h ABPM (night-time systolic values even trended to decrease in these measurements). In contrast to the high dose RO-838 treatment group, the low dose RO-151 group showed a significant decrease from baseline in ABPM SBP (LS mean –3.53 mmHg, p = 0.022), in particular at daytime (–4.56 mmHg, p = 0.010) and to a lesser extent during night (–1.80 mmHg, p > 0.05). These changes were still evident after correcting for the change in the placebo group. No corresponding trend was identified for semi-supine SBP or DBP nor for 24 h-ABPM values in the high dose RO-151 group. Compared with both baseline and placebo, plasma ACTH-levels were significantly higher with both doses of RO-151 on day 29 (relative increase in AUC0–24h vs. baseline was 44 ± 33% with the low dose and 69 ± 36% with the high doses compared with a decrease of 3 ± 23% with placebo). At end of treatment the number of patients with at least one ACTH concentration above the upper limit of normal was 16 (out of 24) with the low dose and 14 (out of 20) with the high dose of RO-151. The mean increase in ACTH concentrations was accompanied with an increase in plasma cortisone, but not plasma cortisol, levels in both RO-151 treatment groups which was statistically significant, but small (mean absolute change in baseline was 0.02 ± 0.01 μmol/l in both RO-151 treatment groups compared with 0.01 ± 0.01 with placebo). In contrast, no consistent change in ACTH or cortisone concentrations was observed with either dose group of RO-838. There was no evidence of downregulation of adrenal cortisol production following ACTH stimulation in any treatment group. Basal

Volume 16 No. 11 November 2014

DIABETES, OBESITY AND METABOLISM

cortisol production before ACTH stimulation and cortisol increments after ACTH stimulation were adequate in all five treatment groups at both baseline (day –1/1) and week 4 (day 28). However, mean increases from baseline in peak plasma cortisol concentration were higher with RO-151 than with placebo (79.4 and 165.8 nmol/l in the low and high dose RO-151 groups, respectively, compared with 46.4 nmol/l in the placebo group). Likewise, androgen precursors (dehydroepiandrosterone [DHEA], androstenedione, sex hormone binding globulin [SHBG]) significantly increased in the RO-151 groups, but not in the RO-838 groups (data not shown). The placebo-corrected changes from baseline reached statistical significance for DHEA only in females in the high RO-151 dose group whereas androstenedione concentrations were significantly higher in females in both dose groups and in males in the low dose group. The increase in SHBG was only statistically significant in males in the high RO-151 dose group. Testosterone concentrations did not differ between treatment groups and placebo.

Discussion This study investigated safety, tolerability and efficacy of two novel 11𝛽-HSD1 inhibitors (each given in a low and high dose) in comparison with placebo in metformin-treated patients with T2D over a treatment duration of 4 weeks. While MDPG, FPG or PPG did not significantly improve with either drug, we did see some indications of a treatment effect: despite the rather short treatment duration both drugs trended to improve HbA1c values and decreased body weight (which reached statistical significance for the high dose RO-151 group). In addition, the high dose RO-151 (non-significantly) improved insulin sensitivity parameters whereas the low dose of this compound showed slight decreases in SBP. On the other hand, most of these changes were modest and not always dose-dependent or statistically significant. Both drugs were well tolerated, but RO-151, in contrast to RO-838, increased concentrations of ACTH and adrenal androgen precursors (particularly in females) as reported for other 11𝛽-HSD1-inhibitors [13,10,19]. Published results for other 11𝛽-HSD1-inhibitors (available for treatment durations of up to 12 weeks) also report positive trends or significant reductions on HbA1c in a magnitude (0.3%) similar to the results of this study for some treatment groups [10]. However, the HbA1c findings in our study are difficult to understand as the treatment duration was only 4 weeks and as there was no clear correlation to dose or blood glucose values. On the other hand, in line with our results, the 11𝛽-HSD1-inhibitor MK-0916 did not improve FPG or 2-h PPG despite a significant reduction in HbA1c. Similarly, the small reductions in body weight observed in our study with either compound are comparable to those reported for MK-0916 [13,19] and INC13739 [10] approaching or slightly exceeding a placebo-adjusted difference of 1 kg. The trend to improvements insulin sensitivity parameters as observed in particular with the high dose of RO-151 group are consistent with previously reported results by Hawkins et al. who reported significant improvements in insulin sensitivity (as measured by the euglycaemic hyperinsulinaemic clamp technique) in people with T2D after 28 days of treatment with INC13739 [20]. In this

Volume 16 No. 11 November 2014

original article proof-of-concept study we based insulin sensitivity assessments on OGTT-outcomes and did not use the very precise, but also cumbersome clamp technique which might explain why our results did not reach statistical significance. The effects on lipids were inconsistent in our study. Mild deteriorations were seen with RO-838 (particularly with the high dose) which, however, were only significant for VLDL cholesterol, triglycerides and apo CIII. In contrast, the high dose RO-151 led to small, but significant improvements in cholesterol, LDL and apo B, although these results might have been influenced by increases of these parameters in the placebo group. The few published results with other 11𝛽-HSD1 inhibitors do not indicate a pronounced effect on lipids, either. While slight improvements were observed with INCB13739 [10] these only reached statistical significance for cholesterol in the highest dose group and for triglycerides in a subgroup analysis of patients with pre-existing hyperlipidaemia at baseline. In general, it seems that the treatment effects reported for selective 11𝛽-HSD1-inhibitors so far have been rather small, and the current study is no exception. The reason for the small magnitude of effect is not clear. The reduction in the 11𝛽-HSD1 activity index (ratio of urinary [𝛼THF + THF]/THE) clearly indicates that both compounds indeed did inhibit 11𝛽-HSD1-activity. However, one has to keep in mind that this index primarily reflects 11𝛽-HSD1 inhibition in the liver [21] and is not necessarily indicative of 11𝛽-HSD1 activity in the adipose tissue [22]. The local inhibition of 11𝛽-HSD1 activity in the adipose tissue unfortunately is difficult to assess requiring adipose tissue biopsies. While results of adipose tissue biopsies are not available for this study, data have been reported for AZD-4017, another 11𝛽-HSD1-inhibitor [12] that showed quite pronounced inhibition of 11𝛽-HSD1 (>90%) after single doses. Thus, from the limited data available so far on biomarkers of 11𝛽-HSD1-activity (urinary THF/THE-ratio and 11𝛽-HSD1-activity in adipose tissue) it seems that 11𝛽-HSD1-inhibitors do indeed inhibit 11𝛽-HSD1, although it is still not entirely clear if the primary inhibition occurs in the liver or in the adipose tissue (and this could also differ between compounds). Another potential reason for the rather small effects of 11𝛽-HSD1-inhibitors observed in this study may be the relatively short treatment duration of only 4 weeks. Considering that 11𝛽-HSD1-inhibitors act primarily in the adipose tissues influencing hormone concentrations at an intracellular (rather than a systemic) level, it might take some time for 11𝛽-HSD-inhibitors to elicit their full effect. It is not entirely clear to what extent and how long cortisol concentrations in the adipose tissues would have to be suppressed to elicit systemic metabolic effects, e.g. on glucose and lipid concentrations. One should keep in mind that other anti-diabetic agents such as 𝛾-PPAR-agonists have also shown increasing effectiveness in their blood-glucose lowering effect over the first 3–4 months of treatment [23]. While 11𝛽-HSD1-inhibitors have a different mode of action than 𝛾-PPAR-agonists they have also been shown to improve insulin sensitivity both in a euglycaemic glucose clamp setting [23] and using HOMA [10]. Our study also indicates an improvement in insulin sensitivity with RO-151 which reached statistical significance for

doi:10.1111/dom.12317 1075

original article OGTT-based parameters such as the Matsuda-Index and also for HOMA. However, it may take longer than 4 weeks before these changes translate into significant improvements in HbA1c or other metabolic parameters such as lipids. Noteworthy, the full effect of 𝛾-PPAR-agonists on lipids is reached only after 16 weeks [23]. On the other hand, a tachyphylaxis phenomenon cannot be completely ruled out for 11𝛽-HSD1-inhibitors as AZD-4017 showed pronounced inhibition of 11𝛽-HSD1 (based on both urinary corticosteroid measurements and assessments in adipose tissues) after single doses, but these effects were markedly attenuated after 9 days of treatment [12]. Nevertheless, the unchanged effect of both RO-151 and RO-838 on the 11𝛽-HSD1 activity index and also the continuous trend to a reduction in body weight (Figure 2) over the 4 weeks of treatment suggests that the metabolic activity of both compounds is sustained over 4 weeks. Finally, at this early stage of development it cannot be ruled out that even the high doses of RO-151 and RO-838 were insufficient to elicit sufficient and sustained metabolic effects. Based on the data for INCB13739 Hollis and Huber pointed out that substantial enzyme inhibition (close to 90% maximal inhibition) might be required to produce a sufficient decrease in intracellular cortisol so as to attenuate glucocorticoid receptor-mediated transcription [24]. On the basis of the 11𝛽-HSD1 activity index these high levels of inhibition were not achieved with RO-838 (maximum inhibition was below 60% in the low dose and still below 70% in the high dose). While enzyme inhibition was higher with RO-151 (approaching 90% with the low dose and exceeding 90% with the high dose) the level of enzyme inhibition in the target organ (adipose tissue) could not be assessed in our study. Another potential reason for the small degree of metabolic changes that sometimes even point into opposite directions could be analytical errors. However, with only one exception (glucagon) all assays met pre-specified validation criteria and were run by an experienced and GLP-certified laboratory. Therefore, it seems more likely that the lack of profound changes is due to biological rather than to analytical variation. One has to keep in mind that while the study was well powered to detect clinically meaningful changes in the primary endpoint based on available literature data, it was (as usual for phase 2a studies) not powered for secondary endpoints. In addition, no correction for multiple testing was done in this exploratory trial (indeed, powering for multiple testing in all the secondary endpoints studied would result in an unrealistically high sample size and render the study unfeasible). Thus, all analyses of secondary endpoints should be understood as exploratory in nature keeping in mind that with the many secondary endpoints the statistical probability to have a p-value below 0.05 just by chance is greater than 75%. Overall, however, only very few comparisons were statistically significant between active treatments and placebo, and the observed changes from baseline were usually rather small and of doubtful clinical relevance supporting the view that in this rather short-term trial the two 11𝛽-HSD1 inhibitors studied did not induce unambiguous metabolic changes. In line with published results on other 11𝛽-HSD1-inhibitors the two compounds investigated in our study showed a good

1076 Heise et al.

DIABETES, OBESITY AND METABOLISM

safety profile overall and were well tolerated. There was no increased frequency of AEs compared with placebo, also no clinically meaningful changes occurred in vital signs, ECG-parameters or safety laboratory. While a trend for an increase in semi-supine SBP (and to a lesser extent DBP) was observed over the course of the study with the high dose RO-838 group, these changes were small, did not reach statistical significance and were not confirmed by the more robust analysis of ABPM values. However, as for other 11𝛽-HSD1-inhibitors there was a significant increase in the concentrations of ACTH and adrenal androgen precursors with RO-151. Similar results have been reported previously for other 11𝛽-HSD1 inhibitors [13,10–12]. Based on the 12-week data with INCB13739 the ACTH-response seems to plateau after approximately 4 weeks of treatment and might therefore indicate an increase in the susceptibility of the adrenal glands to stimulation of the HPA axis as an adjustment to the inhibition of 11𝛽-HSD1 in order to maintain basal cortisol homeostasis [10]. Likewise, no signs or symptoms of androgen excess had been observed with INCB13739 [10]. While in our study RO-151 did also lead to slight increases in plasma cortisone levels, it is worth noting that higher cortisol responses were also observed in the ACTH-stimulation tests when compared to placebo. Again, this indicates that these changes are a compensatory response of the HPA axis to restore cortisol homeostasis in the presence of an 11𝛽-HSD1 inhibitor and might therefore not be a major concern. Nevertheless, for the future development of these compounds evaluations of the HPA axis over a prolonged treatment duration are certainly needed. RO-838 did not show an effect on ACTH, plasma cortisone or androgen precursors, but in view of the relatively low decrease in the 11𝛽-HSD1 activity index with this compound this might also indicate an insufficient inhibitory effect on 11𝛽-HSD1. Apart from the effects on HPA axis described above there were no relevant changes in safety laboratory parameters. Both study drugs were well tolerated. The observed slight increase in clinic-based semi-supine SBP/DBP in the high dose RO-838 group may not be clinically relevant in view of the general lack of an effect on BP in the more robust 24-h ABPM analysis. Furthermore, the low dose RO-151 group showed a rather favourable BP profile, particularly for SBP with a significant decrease from baseline in ABPM SBP, in particular at daytime. There were no other relevant changes in vital signs, ECG-parameters or neurological assessments and the incidence of AEs were similar between the active treatment groups and the placebo group. In summary, treatment with the two 11𝛽-HSD1 inhibitors RO-151 and RO-838 showed inhibitory effects on 11𝛽-HSD1 activity based on urinary corticosteroid excretion. This was associated with trends to improve some parameters of the metabolic syndrome, in particular body weight, HbA1c and insulin sensitivity. These trends, however, often did not reach statistical significance and were not clearly dose dependent. Both compounds were well tolerated, but, in contrast to RO-838, RO-151 (which also showed the higher effect on the 11𝛽-HSD1 activity index) increased ACTH, plasma cortisone and androgen precursors. As these efficacy and safety results are in general agreement with previously published data on

Volume 16 No. 11 November 2014

DIABETES, OBESITY AND METABOLISM

11𝛽-HSD1 inhibitors, the potential of this class of compounds for the treatment of diabetes and the metabolic syndrome needs to be further investigated, preferably in treatment studies with a longer treatment duration than 4 weeks.

Acknowledgements The authors thank all patients for their participation in this study. In addition to Profil, Neuss, Germany, and Profil Institute for Clinical Research, Inc., Chula Vista, CA. USA, the following centres participated in this study: Harrison Clinical Research Deutschland GmbH, Munich, Germany (principal investigator Dr Bärbel Huber), Clinical Pharmacology Of Miami, Miami, FL, USA (principal investigator Dr Kenneth C. Lasseter) and the Department of Internal Medicine, Institute of Endocrinology and Nuclear Medicine, Medical University Graz, Graz, Austria (principal investigator Prof Dr Thomas Pieber). The authors thank the study teams at all participating centres for their excellent work. The assistance of Dr Dina Speidel, Profil, Neuss, Germany, in manuscript preparation is gratefully acknowledged. This study was funded by F. Hoffmann-La Roche, Ltd., Basel, Switzerland.

Conflict of Interest L. M. and M. H. are employees and shareholders of Profil Institute for Clinical Research, Inc. T. H. and C. K. are partners of Profil Institut für Stoffwechselforschung GmbH. These institutions received research funds from several pharmaceutical companies including Hoffmann-La Roche, Ltd., the sponsor of this study. T. H. is an advisory board member of Novo Nordisk and received speaker honoraria from Eli Lilly and Novo Nordisk. M. A., M. R., M. C. M. and S. F. R. are employees for Hoffmann-La Roche, Ltd. H. U. H. has nothing to declare.

Supporting Information Additional Supporting Information may be found in the online version of this article: Table S1. Patient demographics at screening values are given as mean values (range) or as mean values ± s.d.

References 1. Apovian CM. Management of diabetes across the course of disease: minimizing obesity-associated complications. Diabetes Metab Syndr Obes 2011; 4: 353–369. 2. Cefalu WT, Richards RJ, Melendez-Ramirez LY. Redefining treatment success in type 2 diabetes mellitus: comprehensive targeting of core defects. Cleve Clin J Med 2009; 76: S39–S47. 3. DeSouza C, Fonseca V. Therapeutic targets to reduce cardiovascular disease in type 2 diabetes. Nat Rev Drug Discov 2009; 8: 361–367. 4. Glass AR, Burman KD, Dahms WT, Boehm TM. Endocrine function in human obesity. Metabolism 1981; 30: 89–102. 5. Schäcke H, Döcke WD, Asadullah K. Mechanisms involved in the side effects of glucocorticoids. Pharmacol Ther 2002; 96: 23–43. 6. Tomlinson JW, Stewart PM. Cortisol metabolism and the role of 11beta-hydroxysteroid dehydrogenase. Best Pract Res Clin Endocrinol Metab 2001; 15: 61–78.

Volume 16 No. 11 November 2014

original article 7. Morton NM, Holmes MC, Fievet C et al. Improved lipid and lipoprotein, hepatic insulin sensitivity, and glucose tolerance in 11𝛽-hydroxysteroid dehydrogenase type 1 null mice. J Biol Chem 2001; 276: 41293–41300. 8. Masuzaki H, Paterson J, Shinyama H et al. A transgenic model of visceral obesity and the metabolic syndrome. Science 2001; 294: 2166–2169. 9. Masuzaki H, Yamamoto H, Kenyon CJ et al. Transgenic amplification of glucocorticpoid action in adipose tissue causes high blood pressure. J Clin Invest 2003; 112: 83–90. 10. Rosenstock J, Banarer S, Fonseca VA et al. The 11𝛽-hydroxysteroid dehydrogenase type 1 inhibitor INCB13739 improves hyperglycemia in patients with type 2 diabetes inadequately controlled by metformin monotherapy. Diabetes Care 2010; 33: 1516–1522. 11. Liu W, Katz DA, Locke C et al. Clinical safety, pharmacokinetics, and pharmacodynamics of the 11𝛽-hydroxysteroid dehydrogenase type 1 inhibitor ABT-384 in healthy volunteers and elderly adults. Clin Pharmacol Drug Dev 2013; 2: 133–151. 12. Sjöstrand M, Hansson GI, Hartford M et al. Pharmacodynamic effects of AZD4017, a selective 11beta-HSD1 inhibitor, in liver and adipose tissue. Diabetes 2011; 60: A319. 13. Feig PU, Shah S, Hermanowski-Vosatka A et al. Effects of an 11𝛽-hydroxysteroid dehydrogenase type 1 inhibitor, MK-0916, in patients with type 2 diabetes mellitus and metabolic syndrome. Diabetes Obes Metab 2011; 13: 498–504. 14. Amgen. Amgen Outlines Strategy, Growth Objectives and Capital Allocation Plans. Amgen Press Release Dated 21 April 2011. Available from URL: http://www.amgen.com/media/media_pr_detail.jsp?releaseID=15532 98. Accessed 1 December 2013. 15. Tiganescu A, Tahrani AA, Morgan SA et al. 11𝛽-Hydroxysteroid dehydrogenase blockade prevents age-induced skin structure and function defects. J Clin Invest 2013; 123: 3051–3060. 16. Matsuda M, DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care 1999; 22: 1462–1470. 17. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner R. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985; 28: 412–419. 18. Ahrén B, Simonsson E, Larsson H et al. Inhibition of dipeptidyl peptidase IV improves metabolic control over a 4-week study period in type 2 diabetes. Diabetes Care 2002; 25: 869–875. 19. Shah S, Hermanowski-Vosatka A, Gibson K et al. Efficacy and safety of the selective 11𝛽-HSD-1 inhibitors MK-0736 and MK-0916 in overweight and obese patients with hypertension. J Am Soc Hypertens 2011; 5: 166–176. 20. Hawkins M, Hunter D, Kishore P et al. INCB013739, a selective inhibitor of 11𝛽-hydroxysteroid dehydrogenase type 1(11𝛽HSD1), improves insulin sensitivity and lowers plasma cholesterol over 28 days in patients with type 2 diabetes mellitus. Diabetes 2008; 57: 344-OR. 21. Stewart PM, Boulton A, Kumar S, Clark PM, Shackleton CH. Cortisol metabolism in human obesity: impaired cortisone-->cortisol conversion in subjects with central adiposity. J Clin Endocrinol Metab 1999;84:1022-1027. 22. Stimson RH, Andrew R, McAvoy NC, Tripathi D, Hayes PC, Walker BR. Increased whole-body and sustained liver cortisol regeneration by 11beta-hydroxysteroid dehydrogenase type 1 in obese men with type 2 diabetes provides a target for enzyme inhibition. Diabetes 2011; 60: 720–725. 23. Czoski-Murray C, Warren E, Chilcott J, Beverley C, Psyllaki MA, Cowan J. Clinical effectiveness and cost-effectiveness of pioglitazone and rosiglitazone in the treatment of type 2 diabetes: a systematic review and economic evaluation. Health Technol Assess 2004; 8. Available from URL: http://www. hta.ac.uk/fullmono/mon813.pdf. Accessed 1 December 2013. 24. Hollis G, Huber R. 11𝛽-Hydroxysteroid dehydrogenase type 1 inhibition in type 2 diabetes mellitus. Diabetes Obes Metab 2011; 13: 1–6.

doi:10.1111/dom.12317 1077