179 Combination treatment with pioglitazone and fenofibrate attenuates pioglitazone-mediated acceleration of bone loss in ovariectomized rats

Rana Samadfam, Malaika Awori, Agnes Be´nardeau1, Frieder Bauss2, Elena Sebokova1, Matthew Wright1 and Susan Y Smith Charles River Laboratories, 22022 Transcanadienne, Senneville, Montre´al, Que´bec, Canada H9X 3R3 1F. Hoffmann-La Roche AG, Basel, CH-4070 Switzerland 2Roche Diagnostics GmbH, Penzberg, DE-82377 Germany (Correspondence should be addressed to S Y Smith; Email: [email protected])

Abstract Peroxisome proliferator-activated receptor (PPAR) g ago- mineral content (w45%) and bone mineral density (BMD; nists, such as pioglitazone (Pio), improve glycemia and lipid w60%) at the lumbar spine. Similar effects of treatments were profile but are associated with bone loss and fracture risk. Data observed at the femur, most notably at sites rich in trabecular regarding bone effects of PPARa agonists (including bone. At the proximal tibial metaphysis, concomitant fenofibrate (Feno)) are limited, although animal studies treatment with PioCFeno prevented Pio exacerbation of suggest that Feno may increase bone mass. This study ovariectomy-induced loss of trabecular bone, resulting in investigated the effects of a 13-week oral combination BMD values in the PioCFeno group comparable to OVX treatment with Pio (10 mg/kg per day)CFeno (25 mg/kg controls. Discontinuation of Pio or Feno treatment of per day) on body composition and bone mass parameters OVX rats was associated with partial reversal of effects on compared with Pio or Feno alone in adult ovariectomized bone loss or bone mass gain, respectively, while values in the (OVX) rats, with a 4-week bone depletion period, followed PioCFeno group remained comparable to OVX controls. by a 6-week treatment-free period. Treatment of OVX rats These data suggest that concurrent/dual agonism of PPARg with PioCFeno resulted in w50% lower fat mass gain and PPARa may reduce the negative effects of PPARg compared with Pio treatment alone. Combination treatment agonism on bone mass. with PioCFeno partially prevented Pio-induced loss of bone Journal of Endocrinology (2012) 212, 179–186

Introduction selectivity/agonistic effects on PPAR subtypes. Thiazolidine- diones (TZDs) work predominantly through the activation of Bone is a highly specialized and dynamic tissue that undergoes PPARg and have been shown to improve insulin sensitivity constant remodeling by balancing bone formation and and glucose homeostasis in type 2 diabetes mellitus resorption (Clarke 2008, Eriksen 2010), processes that are (Spiegelman 1998, Lalloyer & Staels 2010), while fibrates regulated by osteoblasts and osteoclasts respectively (Clarke activate PPARa, with the primary effect of improving plasma 2008, Eriksen 2010). Osteoblastogenesis, osteoclastogenesis, lipids (Sierra et al. 2007, Lalloyer & Staels 2010). Accumulating and activity of the resultant cell types are controlled by a evidence from animal studies as well as clinical trials indicates variety of hormonal and humoral factors such as estrogen that activation of PPARg results in loss of bone mass and/or (Manolagas et al. 2002), thyroid hormones (Eriksen 2010), strength (Kahn et al. 2006, Glintborg et al. 2008, Schwartz growth factors (Eriksen 2010), and cytokines (Clarke 2008, 2008) via effects on bone formation and resorption (Sottile Eriksen 2010), the actions of which are mediated via specific et al. 2004, Grey et al. 2007), underlying the increased risk of receptors, including nuclear receptors (Boyce et al. 2009, fractures observed clinically with these agents (Kahn et al. Eriksen 2010). 2006, Glintborg et al. 2008). The role of PPARa (and the third The peroxisome proliferator-activated receptors (PPARs) member of the PPAR family, PPARd) in regulating bone are a group of ligand-activated nuclear receptors that play a key health is much less understood although, recently,fibrates were role in the regulation of lipid and carbohydrate metabolism, as reported to increase bone mass and strength in normal rats well as inflammation, immunomodulation, and cellular (Syversen et al. 2009) and in an animal model of osteoporosis, differentiation (Karpe & Ehrenborg 2009, Lalloyer & Staels the ovariectomized (OVX) rat (Stunes et al. 2011). 2010). PPAR agonists are a diverse group of compounds Combined (dual) activation of PPARa and PPARg has and, although often grouped together, exhibit different been pursued for over a decade due to the expectation of

Journal of Endocrinology (2012) 212, 179–186 DOI: 10.1530/JOE-11-0356 0022–0795/12/0212–179 q 2012 Society for Endocrinology Printed in Great Britain Online version via http://www.endocrinology-journals.org

Downloaded from Bioscientifica.com at 09/25/2021 09:19:03AM via free access 180 R SAMADFAM and others . Effects of PPARa and g activation on bone mass

better control of cardiovascular risk factors vs solo activation was discontinued in the remaining ten animals from each of one of the PPARs, as both diabetic dyslipidemia and of these groups for 6 weeks (i.e. end of the study at week 19). glucose control/insulin sensitivity are targeted (Lalloyer & Body weight and food consumption were measured Staels 2010). Several combined (dual) PPARa/g agonists weekly, starting from the last week of the acclimation period have reached phase III trials – muraglitazar (Kendall et al. and extending through the treatment and treatment-free 2006), tesaglitazar (Bays et al. 2007), and aleglitazar (Henry periods. In addition, body weight was measured at et al. 2009) – and although the development of muraglitazar randomization and on the day of termination (as an overnight and tesaglitazar was terminated because of safety concerns, the fasted body weight). therapeutic potential of PPARa/g agonists remains of high clinical interest. To our knowledge, the effect of combined Laboratory analysis activation of PPARg and PPARa on bone biology has not been investigated previously. At the end of the bone depletion period, blood was collected The objective of this study was to investigate the effects of from overnight fasted animals at weeks 5 or 6 and 12 or 13 of combination treatment with an agonist of PPARa (fenofi- the treatment period and at week 19 (i.e. end of the brate (Feno)) and PPARg (pioglitazone (Pio)) on bone mass, treatment-free period). Plasma was analyzed for assessment of bone density, and markers of bone turnover compared with triglyceride (TG), insulin, and biomarkers of bone turnover the effect of Pio and Feno alone in OVX rats. In addition, the (serum osteocalcin (bone formation) and urinary effect of discontinuation of treatment was investigated to C-telopeptide of type I collagen (CTx; bone resorption)). evaluate the reversibility of observed effects. TG was measured by an enzymatic colorimetric assay (GPO-PAP, Roche Diagnostics GmbH), while insulin and osteocalcin were measured using rat RIA kits (LINCO Research, Billerica, MA, USA and Biomedical Technologies, Materials and Methods Inc., Stoughton, MA, USA respectively). CTx was assessed with a rat ELISA (Immunodiagnostic Systems, Tyneand Wear, Animals UK). Adiponectin was measured by a colorimetric ELISA The study was conducted in accordance with Standard assay (B-Bridge International, Inc., Cupertino, CA, USA). Operating Procedures of Charles River Laboratories, Montreal, and F. Hoffmann-La Roche AG and the protocol Bone densitometry measurements was approved by the Institutional Animal Care and Use Committee. Animals were appropriately housed and experi- Bone densitometry was evaluated by dual-energy X-ray mental procedures were performed in accordance with absorptiometry (DXA) and peripheral quantitative computed guidelines of the Association for the Assessment and tomography (pQCT). Animals were anesthetized using iso- Accreditation of Laboratory and Animal Care and appropriate flurane and were maintained under the effect during the scans. federal, state, or local guidelines. Scans were acquired once prior to surgery,at the end of the bone At the beginning of treatment, animals were w7 months of depletion period, during weeks 5/6 and 12/13 of the treatment age and ranged in weight from 318 to 454 g. Animals had free period, and at the end of the treatment-free period at week 19. access to purified water and standard laboratory diet (PMI Certified rodent 5002, PMI Nutrition International, Inc., Dual-energy X-ray absorptiometry Saint Paul, MN, USA) throughout the study, except where indicated. The overall average temperature and relative DXA was used to measure area (cm2), bone mineral content humidity during the study were 21.9 8C and 50% respectively. (BMC (g)), and bone mineral density (BMD (g/cm2)) using a After a minimum acclimation period of 3 weeks, animals Hologic Discovery A densitometer (Hologic, Inc., Bedford, underwent surgical procedure (sham operation or ovari- MA, USA) with small animal hi-res software version 12.3. ectomy). Prior to surgery, baseline bone densitometry and The percent coefficient of variation for rat BMD at the spine body weight measurements were obtained from all animals. and femur was 1.1–1.5%. In addition, lean mass and fat mass Using a randomization procedure stratified according to body were reported from whole-body DXA scans. Bone densito- weight, animals were assigned to one of the following metry measurements of the whole body, lumbar spine (L1– treatment groups: 1) sham vehicle control, 2) OVX vehicle L4), and whole and regionalized right femur were acquired control, 3) OVX Pio (10 mg/kg per day), 4) OVX Feno from all animals. The initial scans acquired from each animal (25 mg/kg per day), or 5) OVX Pio (10 mg/kg per day) were compared with follow-up scans to ensure appropriate CFeno (25 mg/kg per day). and consistent positioning of scan sites. Daily oral dosing by gavage of vehicle or test substance(s) commenced in treatment groups 1–5 at the end of the bone Peripheral quantitative computed tomography depletion period (4 weeks after surgery). Animals were treated at approximately the same time each day for 13 weeks. Ten of the In vivo pQCT scans were performed using an XCT Research animals in groups 1–5 were killed at week 13, while treatment SAC bone scanner with software version 5.50D (Stratec

Journal of Endocrinology (2012) 212, 179–186 www.endocrinology-journals.org

Downloaded from Bioscientifica.com at 09/25/2021 09:19:03AM via free access Effects of PPARa and g activation on bone mass . R SAMADFAM and others 181

Medizintechnik, Pforzheim, Germany). Bone densitometry and comparable for the remainder of the study period measurements were performed on the right proximal tibia of (Fig. 1A). Pio-treated animals showed trends toward a greater all animals. One slice each was obtained in the metaphysis and increase in weight gain compared with OVX controls, while in the diaphysis, acquired at 14 and 50%, respectively, of the the Feno-treated group showed trends toward a decreased rate total bone length distal to the reference line set to the tibial of weight gain. These differences, however, did not attain proximal end. For follow-up scans, positioning and placement statistical significance. Mean body weights and body weight of CT scan lines were verified using the scout scan and gains in the PioCFeno-treated animals were comparable compared with the initial scout scan. The percent coefficient to OVX controls during the treatment (Fig. 1A) and of variation for pQCT parameters, including trabecular BMD treatment-free periods. and cortical BMD, was 0.4–0.8%. Plasma insulin and TG levels Statistical analyses OVX increased insulin (Fig. 1B), TG (Fig. 1C), and total All statistical procedures were pre-specified in the study cholesterol levels (data not shown). Compared with OVX protocol. Data are presented as group means and S.E.M., unless controls, there was a decrease in plasma insulin levels in Pio- otherwise stated. The homogeneity of group variances was treated animals and PioCFeno-treated animals at weeks 5 and evaluated using Levene’s test at the 0.05 significance level. If 12 (Fig. 1B). There were also decreases in plasma TG in Pio- differences between group variances were not found to be treated animals, in Feno-treated animals, and in PioCFeno- significant (PO0.05), then a parametric one-way ANOVAwas treated animals at weeks 5 and 12 (Fig. 1C). Adiponectin performed. When significant differences among the means % . levels were also significantly increased by treatment with Pio were indicated by the overall ANOVA F test (P 0 05), the C t-test on least squares means was used to perform all or Pio Feno during the treatment period (at week 12/13, combinations of two-by-two group mean comparisons, except for the sham vehicle control group, which was compared only A 550 with OVX vehicle control. If Levene’s test indicated % . heterogeneous group variances (P 0 05), then the non- 500 parametric Kruskal–Wallis test was used to compare all considered groups. When the Kruskal–Wallis test was 450 significant (P%0.05), the Wilcoxon rank-sum test was used to perform the pairwise group comparisons of interest as above. 400 Grams For densitometry data (DXA and pQCT), individual 350 treatment period results were adjusted to the related pre- surgery and end of bone depletion periods by computing the 300 relative difference (i.e. the percentage change from pre- Treatment free period surgery and baseline respectively). DXA lean mass, DXA fat 250 mass, pQCT muscle area, and pQCT fat area were also –5–4–3–2 –1 1 2 3 4 5 6 7 8 9 10111213141516171819 Randomization End bone Week adjusted to the most recent body weight relative to each depletion scanning occasion. These derived variables were submitted to B C the analysis described above. The original measurements 1·4 1·4 (unadjusted data) were submitted to the above statistical 1·2 1·2 1·0 † 1·0 analysis only for the pre-treatment period. A modified 0·8 0·8 † * Bonferroni correction was applied. 0·6 0·6 * * * * 0·4 * * ‡ 0·4 * Insulin (ng/ml) 0·2 0·2

0·0 (mmol/ml) Triglycerides 0·0 Results –1 5 12/13 –1 5 12/13 Week Week

For ease of visualization, data are shown from pre-surgery to Sham vehicle control the end-of-treatment period (i.e. treatment groups 1–5, OVX vehicle control Pioglitazone (OVX) 10 mg/kg per day nZ10 per group), while the effect of discontinuation of Fenofibrate (OVX) 25 mg/kg per day treatments is described within the text, except for body Pioglitazone + Fenofibrate (OVX) 10/25 mg/kg per day weight and BMD at the femur. Figure 1 (A) Group mean body weights, (B) insulin, and (C) triglyceride levels at pre-treatment and during the treatment period. Body weight chart also shows the effect of ovariectomy Body weight post-randomization and the impact of discontinuation of treatment. 0 Insulin and triglyceride data are meanGS.E.M.*P %0.05 vs OVX; Mean body weight gains were statistically significantly higher †P 0%0.05 vs pioglitazone; ‡P 0%0.05 vs fenofibrate; P 0Z adjusted for OVX controls compared with sham controls up to week 8 P value (modified from Bonferroni adjustment). www.endocrinology-journals.org Journal of Endocrinology (2012) 212, 179–186

Downloaded from Bioscientifica.com at 09/25/2021 09:19:03AM via free access 182 R SAMADFAM and others . Effects of PPARa and g activation on bone mass

AB Feno partially prevented OVX- and Pio-induced fat gains, 150 12 as the mean fat area was slightly lower for animals treated 10 C 125 with Pio Feno relative to Pio. In contrast, Feno (alone or 8 in combination with Pio) had no effect on muscle mass 100 6 * (Fig. 3C). During the treatment-free period, loss of fat 4 *

g/mmol creatinine) C 75 was noted for both Pio- and Pio Feno-treated animals, µ 2 * Osteocalcin (ng/ml) * * * with mean values comparable to OVX vehicle controls

50 CTx ( 0 End bone 5 12 End bone 5 12 (Supplementary Table 2, see section on supplementary data depletion Week depletion Week given at the end of this article). No effect was noted for

Sham vehicle control Feno-treated animals (Supplementary Table 2). OVX vehicle control Pioglitazone (OVX) 10 mg/kg per day Fenofibrate (OVX) 25 mg/kg per day Bone densitometry Pioglitazone + Fenofibrate (OVX) 10/25 mg/kg per day Dual-energy X-ray absorptiometry Figure 2 Mean levels of (A) osteocalcin (bone formation marker) Whole body. DXA analysis revealed that OVX prevented and (B) CTx (bone resorption marker) at the end of the bone BMD gains compared with sham controls over the study depletion period and during treatment. Data are meanGS.E.M. *P 0%0.05 vs OVX; P 0Z adjusted P value (modified from period (Fig. 4A). Treatment of OVX rats with Pio resulted in Bonferroni adjustment). further reduction of whole-body bone mass (BMC and BMD; BMC data are shown in Supplementary Table 3, see OVX controls: 8.6G0.8 mg/ml, Pio: 22.0G2.2 mg/ml, and section on supplementary data given at the end of this article). PioCFeno 17.7G1.8 mg/ml; P!0.01 for both vs OVX In contrast, treatment of OVX rats with Feno partially controls). The significant reductions in insulin and TG (w50%) restored OVX-induced reduction in bone mass, as observed with Pio were maintained after discontinuation evidenced by generally higher BMC and BMD values; of treatment but were lost in other treatment groups however, values remained lower compared with sham (Supplementary Table 1, see section on supplementary data controls at the end of the treatment period (Fig. 4A, BMC given at the end of this article). data are shown in Supplementary Table 3). At the whole- body level, combination of PioCFeno did not have a notable Biochemical markers of bone turnover effect on Pio-induced bone loss in OVX rats. Over the treatment-free period, a slight loss and a slight gain of bone Compared with the sham group, OVX increased plasma levels

of osteocalcin by 60% (Fig. 2A) and CTx by 151% (Fig. 2B). A AB90 45 ‡ characteristic age-related decline in both osteocalcin and CTx 85 40 80 * ‡ was also noted, but there was no effect of any treatment on 35 ‡ * 75 * 30 these markers of bone turnover in OVX rats (Fig. 2A and B). 70 † 25 * 65 ‡ 60 20 * ‡ Fat mass change (%) Fat 15 Lean mass change (%) 55 Body composition * 50 10 Pre- End bone 5/6 12/13 Pre- End bone 5/6 12/13 At the whole-body level, DXA values (relative to body surgery depletion Week surgery depletion Week weight) showed a loss of muscle mass (26.5%) and gain of CD‡ fat mass (64.5%) in OVX rats compared with the sham group 39 8 * 36 7 ‡ (Fig. 3A and B). Pio resulted in further gains in fat mass and * 33 6 * ‡ ‡ a slightly greater loss of lean mass, while Feno partially 30 * * 5 prevented the OVX-induced fat gain, with no notable effect 27 4 * * 24 ‡ 3 on muscle mass. Feno partially prevented (by w50%) Pio- * 21 area change (%) Fat 2 Muscle area change (%) induced fat gains, with no effect on muscle mass. During the 18 1 Pre- End bone 5/6 12/13 Pre- End bone 5/6 12/13 treatment-free period, the rate of loss of lean mass and gain in surgery depletion Week surgery depletion Week fat mass was slower for Pio-treated animals compared with Sham vehicle control OVX, suggesting reversibility of the Pio effect. OVX vehicle control Pioglitazone (OVX) 10 mg/kg per day These results were generally corroborated by pQCT Fenofibrate (OVX) 25 mg/kg per day analysis of soft tissue at the tibial diaphysis site, where Pioglitazone + Fenofibrate (OVX) 10/25 mg/kg per day OVX induced an accelerated loss of muscle area (14%) while Figure 3 Analysis of (A) lean mass and (B) fat mass by DXA at increasing the fat area (67%) (Fig. 3C and D). Using values whole-body level and (C) muscle area and (D) fat area by pQCT at the proximal tibia. Data are meanGS.E.M. and show percentage adjusted to body weight, treatment of OVX rats with Pio 0 change values adjusted to body weight. *P %0.05 vs OVX; significantly increased fat area (40%), with a slight loss of †P 0%0.05 vs pioglitazone; ‡P 0%0.05 vs fenofibrate; P 0Z adjusted muscle area (8%), compared with OVX vehicle controls. P value (modified from Bonferroni adjustment).

Journal of Endocrinology (2012) 212, 179–186 www.endocrinology-journals.org

Downloaded from Bioscientifica.com at 09/25/2021 09:19:03AM via free access Effects of PPARa and g activation on bone mass . R SAMADFAM and others 183

A B femur were generally similar to the lumbar spine, with reversal 6 9 * * * * 6 of bone loss in Pio-treated animals and a slight reduction in * 4 3 * BMD in Feno-treated animals (Fig. 4D). * 0 * –3 2 –6 ‡ ‡ ‡ ‡ –9 *‡ Peripheral quantitative computed tomography BMD change (%) BMD change (%) †‡ 0 ‡ –12 –15 *‡ In vivo pQCT at the proximal tibial metaphysis showed –2 –18 Pre- End bone 5/6 12/13 Pre- End bone 5/6 12/13 significant reduction of bone mass in OVX control rats surgery depletion surgery depletion (Fig. 5B and C), most of which was attributed to the Week Week trabecular bone compartment (62% BMC and 61% BMD). C D Treatment with Pio further increased bone loss (9% for 12 12 * 10 10 * both trabecular BMC and BMD), whereas Feno prevented 8 8 * 6 6 or slightly restored OVX-induced bone loss (Fig. 5A). 4 * 4 2 * 2 Concomitant treatment with Feno prevented Pio-induced 0 0 w

BMD change (%) BMD change (%) bone loss in OVX rats ( 80% for trabecular BMD), resulting –2 †‡ –2 ‡‡ –4 ‡ –4 ‡ * *‡ in BMD levels comparable to OVX controls. At the end of –6 –6 Pre- End bone 5/6 12/13 Pre- End bone 5/6 12/13 19 the treatment-free period, in vivo pQCT values were similar surgery depletion surgery depletion Week Week to values at the end of treatment, showing no evidence of Sham vehicle control reversal of the effects at the tibial metaphysis. In contrast to OVX vehicle control Pioglitazone (OVX) 10 mg/kg per day the effects observed at the tibial metaphysis, no clear effects Fenofibrate (OVX) 25 mg/kg per day Pioglitazone + Fenofibrate (OVX) 10/25 mg/kg per day of treatment were observed on BMD at the tibial diaphysis (Fig. 5D) (cortical bone site). Figure 4 BMD analysis by DXA at (A) whole body, (B) lumbar spine, (C) global femur in the treatment groups, and (D) global femur during treatment and follow-up treatment-free period. Data depicted are meanGS.E.M. and show percentage change from pre- surgery levels. BMD, bone mineral density. *P 0%0.05 vs OVX; Discussion †P 0%0.05 vs pioglitazone; ‡P 0%0.05 vs fenofibrate; P 0Z adjusted P value (modified from Bonferroni adjustment). Clinical use of PPARg agonists such as Pio or rosiglitazone is associated with an increased risk of fractures (Kahn et al. 2006, mass were noted for Feno- and Pio-treated animals, Schwartz 2006, Glintborg et al. 2008). In vivo preclinical respectively, indicating partial reversibility of the effect of studies in models of osteoporosis, such as the OVX rat, these compounds on bone (Supplementary Table 3).

Lumbar spine. At the lumbar spine, DXA analysis showed a AB8 20 * * * marked loss of bone (24% BMC and 19% BMD; BMC data 4 * * 0 * 0 –20 –4 * are shown in Supplementary Table 3) in OVX rats from the –40 –8 * –60 end of the bone depletion period and throughout the study –12 †‡ ‡ ‡ –80 (Fig. 4B). Treatment with Pio resulted in further loss of bone BMD change (%) –16 *‡ BMD change (%) *‡ –20 *‡ –100 mass (6% BMC and 6% BMD), whereas Feno reduced Pre- End bone 5/6 12/13 Pre- End bone 5/6 12/13 OVX-induced bone loss (41% prevention of BMD loss at surgery depletion surgery depletion week 12), as demonstrated by the higher BMD values com- Week Week pared with OVX controls. At this site, Feno partially (w60%) CD9 4 6 3 prevented Pio-induced bone loss in PioCFeno-treated OVX * * * 3 * 2 rats. During the treatment-free period, trends for reversal of 0 †‡ 1 –3 effects of Pio and Feno were noted. *‡ 0

BMD change (%) –6 *‡ BMD change (%) –9 –1 Femur. At the femur, most notably at sites rich in trabecular Pre- End bone 5/6 12/13 Pre- End bone 5/6 12/13 bone (distal and proximal ends (data not shown)), a significant surgery depletion surgery depletion Week Week loss of bone mass (15% for both BMC and BMD; P%0.05) was Sham vehicle control noted in OVX vehicle controls compared with sham vehicle OVX vehicle control Pioglitazone (OVX) 10 mg/kg per day controls over the study period. Treatment of OVX rats with Fenofibrate (OVX) 25 mg/kg per day Pio resulted in further loss of bone mass (3% BMC and 3% Pioglitazone + Fenofibrate (OVX) 10/25 mg/kg per day BMD), whereas Feno attenuated OVX-induced bone loss Figure 5 (A) Total, (B) trabecular, (C) cortical/subcortical, and (36% prevention of BMD loss at week 12), as indicated by the (D) cortical BMD analysis by pQCT in the proximal tibial higher BMD values compared with OVX controls (Fig. 4C). metaphysis (A–C) or diaphysis (D). Data depicted are meanGS.E.M. C and show percentage change from pre-surgery levels. BMD, bone Combination treatment of OVX rats with Feno Pio mineral density. *P 0%0.05 vs OVX; †P 0%0.05 vs pioglitazone; indicated that Feno partially prevented Pio-induced bone ‡P 0%0.05 vs fenofibrate; P 0Z adjusted P value (modified from loss. The results obtained over the treatment-free period in the Bonferroni adjustment). www.endocrinology-journals.org Journal of Endocrinology (2012) 212, 179–186

Downloaded from Bioscientifica.com at 09/25/2021 09:19:03AM via free access 184 R SAMADFAM and others . Effects of PPARa and g activation on bone mass

have documented that TZD PPARg agonists exacerbate day for Pio and Feno respectively). These lower doses were OVX-induced bone loss (Sottile et al. 2004, Kumar 2009, chosen to minimize possible off-target effects of the Stunes et al. 2011). The underlying mechanisms of PPARg- compounds while providing adequate pharmacological induced bone effects are not fully understood although, activity, as evidenced by changes observed in several collectively, clinical and preclinical data suggest impacts on pharmacodynamic indicators, i.e. decreases in plasma TG in both bone formation and osteoblast activity, as well as bone all treated groups, decreases in insulin in Pio and PioCFeno resorption and osteoclastogenesis (Schwartz 2008). Whether groups, and increases in adiponectin in Pio and PioCFeno the deleterious bone effects of PPARg agonists can be groups. Fourthly, we studied several bone sites, including reversed by co-treatment strategies, or by discontinuation of the femur, lumbar spine, and tibia, and also included treatment, has not been investigated systematically, although a treatment discontinuation to assess the reversibility of preliminary non-clinical study showed that the bone treatment effects to increase the reliability of the conclusions. antiresorptive agent alendronate could reduce rosiglitazone- Finally, this manuscript reports the effects of combination induced loss of bone (Kumar et al. 2009). treatment with both a PPARa and a PPARg agonist, This study provides the first evidence that co-treatment providing data to support the predicted beneficial effects of with a PPARg and PPARa agonist reduces the deleterious dual agonism in comparison with the deleterious effect of effects of PPARg agonism on BMC and/or BMD in mature PPARg agonism alone. OVX rats. Treatment with Feno partially prevented Pio- As expected, ovariectomy resulted in increases in body induced bone loss, in particular at sites rich in trabecular bone, weight gain and fat mass accumulation, accompanied by a lack with BMC and BMD values remaining generally lower of gain in bone mass and/or loss of bone mass (particularly at compared with OVX controls, suggesting that administration sites rich in trabecular bone), consistent with marked increases of Feno counteracted some of the negative Pio bone effects. in bone turnover markers (CTx and osteocalcin), compared As Feno and Pio are selective PPARa and PPARg agonists, with sham controls. Treatment with Pio exacerbated the respectively, the data are consistent with the compounds OVX-induced loss of bone mass, as shown by DXA and acting as functional, rather than pharmacological, antagonists in vivo pQCT. In contrast, treatment with Feno reduced in bone. Furthermore, they suggest that dual agonists of OVX-induced bone loss. BMC and BMD values were PPARa/g may exhibit lower or no deleterious bone effects in generally higher in Feno-treated animals than OVX controls comparison with PPARg agonists. but remained lower compared with sham controls. In contrast Dual agonists of PPARa/g have long been pursued for to the clear effect of OVX on CTx and osteocalcin, none of their potential to provide better control of cardiovascular risk the treatments (Pio, Feno, or PioCFeno) exerted any effect factors vs solo activation of PPARa or PPARg, as both on these parameters in this study. Although both CTx and dyslipidemia and glucose control/insulin sensitivity are osteocalcin are used to monitor bone turnover (resorption targeted; however, their effect on bone has not been reported. and formation), the literature, particularly non-clinical The results of this study may thus have potential clinical safety reports, is not entirely consistent. For example, Feno implications for dual PPARa/g agonists, as they show that increased osteocalcin in intact female rats (Syversen et al. concurrent/dual agonism of PPARg and PPARa may 2009) but had no effect in OVX rats (Stunes et al. 2011), minimize the negative effect of PPARg agonism on bone despite clear bone protective effects in both studies. This may mass/density. be due to the elevated bone turnover levels in OVX rats, During the preparation of this manuscript, Stunes et al. which can mask additional effects of treatment with PPARa (2011) reported that Pio augmented OVX-induced loss of agonists. Furthermore, measurements of markers of bone BMD, while Feno (and another PPARa agonist, pirinixic turnover are made at single time points and therefore subtle acid) maintained BMD at sham levels, determined by DXA effects of treatments may not be reflected in levels of bone and micro-computed tomography at the femur. While the markers, despite the fact that changes are occurring in bone data from the Pio and Feno groups of our study are broadly in structure, as evidenced by BMC/BMD analysis. In addition, line with those reported by Stunes et al., a number of in vitro studies indicate the presence of both PPARa and important methodological differences should be highlighted. PPARg in preosteoblast cells (Jackson & Demer 2000, First, we used mature (7-month-old) OVX rats, which have Syversen et al. 2009) and a role for PPARa in osteoblast a lower bone turnover rate compared with young adult rats differentiation and osteoclastogenesis and bone resorption has (3-month-old), minimizing the effect of growth on bone. been suggested (Chan et al. 2007, Syversen et al. 2009): Feno Secondly, we selected a longer bone depletion period stimulated mRNA expression of several osteoblast differen- (4 weeks vs 1 week) to ensure that bone loss had been tiation markers (e.g. sialoprotein and collagen I) in MC3T3- established prior to initiation of treatment. Indeed, the peak E1 cells (Syversen et al. 2009) and inhibited osteoclast effect on bone mass, bone turnover markers, and body weight formation (Chan et al. 2007). In contrast, PPARg activation was noted 4 weeks following surgery, with only a slight by high doses of troglitazone and ciglitazone decreased progression of the response for the remaining study period. maturation of MC3T3-E1 cells (Jackson & Demer 2000) and Thirdly, the doses of Pio and Feno employed in our study rosiglitazone has been shown to decrease the expression were lower (10 and 25 mg/kg per day vs 35 and 90 mg/kg per of several osteoblast markers in vitro (Hasegawa et al. 2008).

Journal of Endocrinology (2012) 212, 179–186 www.endocrinology-journals.org

Downloaded from Bioscientifica.com at 09/25/2021 09:19:03AM via free access Effects of PPARa and g activation on bone mass . R SAMADFAM and others 185

The latter is consistent with our finding that Pio exacerbated Declaration of interest OVX-induced loss of bone and is in accordance with several other studies showing that TZD PPARg agonists suppress A B, E S, and M Ware employees of F Hoffmann La-Roche, Switzerland. F B is employed by Roche Diagnostics GmbH, Germany. S Y S, R S, and M A are osteogenesis (Akune et al. 2004) and/or reduce alkaline employees of Charles River Laboratories, Canada. phosphatase activity (Hasegawa et al. 2008). Thus, PPARg may exert direct inhibitory effects on osteogenesis/osteoblast function in addition to its well-established effects through Funding stimulation of adipogenesis from marrow progenitor cells that can reciprocally give rise to adipocytes or osteoblasts (Akune This study was performed at Charles River Laboratories and was funded by F. Hoffmann-La Roche AG, Switzerland. et al. 2004). Related to this, we observed that treatment with Feno had a greater effect in counteracting Pio-induced bone loss compared with Pio-induced fat gain, i.e. the net effect of Author contribution statement combining PioCFeno was to suppress the Pio effect on both bone and fat. A B, E S, M W, F B, and S Y S were involved in the study design, data, and The partial protective effect of PioCFeno co-treatment on statistical analysis; interpretation of results and discussion; and writing and critical assessment of the manuscript. R S was involved in data acquisition, interpretation bone loss compared with Pio was apparent across a number of of results and discussion, and writing and critical assessment of the manuscript. bone sites, including lumbar spine, tibia, and femur. In M A was involved in study conduct and critical assessment of the manuscript. general, the effect of all treatments on bone mass at the tibial diaphysis was marginal. At the lumbar spine and tibial metaphysis, combined treatment with PioCFeno prevented Acknowledgements Pio-induced bone loss by up to 60 and 80%, respectively, w The authors thank A-E Salman, A Vandjour, W Riboulet, C Wohlgensinger, while an 36% prevention of bone loss was observed at V Griesser, and A Wallier for providing technical assistance for biochemical the femur. assays (F. Hoffmann-La Roche AG, Switzerland) and the imaging technical A further aspect of this study was an assessment of the effect team for scanning assessments (Charles River Laboratories, Canada). Editorial of a 6-week treatment-free period, which followed the support for the development of this manuscript was provided by Moh Tadayyon (MediTech Media, UK), funded by F. Hoffmann-La Roche Ltd. 13-week treatment. In vivo, the net effect of combining Pio and Feno was essential to cancel the opposing effects on fat and bone. During the treatment-free period, the loss of the negative effects of Pio and positive effects on Feno on bone References mass and fat mass resulted in all groups trending toward OVX vehicle controls. Akune T, Ohba S, Kamekura S, Yamaguchi M, Chung UI, Kubota N, Terauchi Y, Harada Y, Azuma Y, Nakamura K et al. 2004 PPARgamma insufficiency enhances osteogenesis through osteoblast formation from bone marrow progenitors. Journal of Clinical Investigation 113 846–855. (doi:10.1172/JCI200419900) Conclusions Bays H, McElhattan J & Bryzinski BS 2007 A double-blind, randomised trial of tesaglitazar versus pioglitazone in patients with type 2 diabetes mellitus. Data from this study show that Pio exacerbated bone loss in Diabetes & Vascular Disease Research 4 181–193. (doi:10.3132/dvdr.2007. 039) the mature OVX rat model, predominantly affecting sites rich Boyce BF, Yao Z & Xing L 2009 Osteoclasts have multiple roles in bone in in trabecular bone. These effects were generally reversible addition to bone resorption. Critical Reviews in Eukaryotic Gene Expression 19 after discontinuation of treatment. In contrast to Pio, Feno 171–180. increased bone mass and, importantly, co-treatment with Chan BY, Gartland A, Wilson PJ, Buckley KA, Dillon JP, Fraser WD & C Gallagher JA 2007 PPAR agonists modulate human osteoclast formation Pio Feno partially prevented Pio-induced loss of bone. and activity in vitro. Bone 40 149–159. (doi:10.1016/j.bone.2006.07.029) These data suggest that concurrent/dual agonism of PPARg Clarke B 2008 Normal bone anatomy and physiology. Clinical Journal of the and PPARa may minimize the negative effect of PPARg American Society of Nephrology 3 (Suppl 3) S131–S139. (doi:10.2215/CJN. agonism on bone mass in rats. While the translation of these 04151206) Eriksen EF 2010 Cellular mechanisms of bone remodeling. Reviews in data to humans remains to be established, they raise the Endocrine & Metabolic Disorders 11 219–227. (doi:10.1007/s11154-010- possibility that treatment with a dual PPARa/g agonist may 9153-1) be an effective therapeutic regimen to manage diabetes Glintborg D, Andersen M, Hagen C, Heickendorff L & Hermann AP 2008 cardiovascular risk and minimize negative effects on bone Association of pioglitazone treatment with decreased bone mineral density mass in patients at risk. in obese premenopausal patients with polycystic ovary syndrome: a randomized, placebo-controlled trial. Journal of Clinical Endocrinology and Metabolism 93 1696–1701. (doi:10.1210/jc.2007-2249) Grey A, Bolland M, Gamble G, Wattie D, Horne A, Davidson J & Reid IR Supplementary data 2007 The peroxisome proliferator-activated receptor-gamma agonist rosiglitazone decreases bone formation and bone mineral density in healthy This is linked to the online version of the paper at http://dx.doi.org/10.1530/ postmenopausal women: a randomized, controlled trial. Journal of Clinical JOE-11-0356. Endocrinology and Metabolism 92 1305–1310. (doi:10.1210/jc.2006-2646) www.endocrinology-journals.org Journal of Endocrinology (2012) 212, 179–186

Downloaded from Bioscientifica.com at 09/25/2021 09:19:03AM via free access 186 R SAMADFAM and others . Effects of PPARa and g activation on bone mass

Hasegawa T, Oizumi K, Yoshiko Y, Tanne K, Maeda N & Aubin JE 2008 The Manolagas SC, Kousteni S & Jilka RL 2002 Sex steroids and bone. Recent PPARgamma-selective ligand BRL-49653 differentially regulates the fate Progress in Hormone Research 57 385–409. (doi:10.1210/rp.57.1.385) choices of rat calvaria versus rat bone marrow stromal cell populations. Schwartz AV 2006 Diabetes, TZDs, and bone: a review of the clinical BMC Developmental Biology 8 71. (doi:10.1186/1471-213X-8-71) evidence. PPAR Research 2006 24502. (doi:10.1155/PPAR/2006/24502) Henry RR, Lincoff AM, Mudaliar S, Rabbia M, Chognot C & Herz M 2009 Schwartz AV 2008 TZDs and bone: a review of the recent clinical evidence. Effect of the dual peroxisome proliferator-activated receptor-alpha/gamma PPAR Research 2008 297893. (doi:10.1155/2008/297893) agonist aleglitazar on risk of cardiovascular disease in patients with type 2 Sierra ML, Beneton V, Boullay AB, Boyer T, Brewster AG, Donche F, diabetes (SYNCHRONY): a phase II, randomised, dose-ranging study. Forest MC, Fouchet MH, Gellibert FJ, Grillot DA et al. 2007 Substituted Lancet 374 126–135. (doi:10.1016/S0140-6736(09)60870-9) 2-[(4-aminomethyl)phenoxy]-2-methylpropionic acid PPARalpha agonists. Jackson SM & Demer LL 2000 Peroxisome proliferator-activated 1. Discovery of a novel series of potent HDLc raising agents. Journal of Medicinal receptor activators modulate the osteoblastic maturation of MC3T3-E1 Chemistry 50 685–695. (doi:10.1021/jm058056x) preosteoblasts. FEBS Letters 471 119–124. (doi:10.1016/S0014-5793(00) Sottile V, Seuwen K & Kneissel M 2004 Enhanced marrow adipogenesis and 01372-7) bone resorption in estrogen-deprived rats treated with the PPARgamma Kahn SE, Haffner SM, Heise MA, Herman WH, Holman RR, Jones NP, agonist BRL49653 (rosiglitazone). Calcified Tissue International 75 329–337. Kravitz BG, Lachin JM, O’Neill MC, Zinman B et al. 2006 Glycemic (doi:10.1007/s00223-004-0224-8) durability of rosiglitazone, metformin, or glyburide monotherapy. New Spiegelman BM 1998 PPAR-gamma: adipogenic regulator and thiazolidi- England Journal of Medicine 355 2427–2443. (doi:10.1056/NEJMoa066224) nedione receptor. Diabetes 47 507–514. (doi:10.2337/diabetes.47.4.507) Karpe F & Ehrenborg EE 2009 PPARdelta in humans: genetic and Stunes AK, Westbroek I, Gustafsson BI, Fossmark R, Waarsing JH, Eriksen EF, pharmacological evidence for a significant metabolic function. Current Petzold C, Reseland JE & Syversen U 2011 The peroxisome proliferator- Opinion in Lipidology 20 333–336. (doi:10.1097/MOL. activated receptor (PPAR) alpha agonist fenofibrate maintains bone mass, 0b013e32832dd4b1) while the PPAR gamma agonist pioglitazone exaggerates bone loss, in Kendall DM, Rubin CJ, Mohideen P, Ledeine JM, Belder R, Gross J, ovariectomized rats. BMC Endocrine Disorders 11 11. (doi:10.1186/1472- Norwood P, O’Mahony M, Sall K, Sloan G et al. 2006 Improvement of 6823-11-11) glycemic control, triglycerides, and HDL cholesterol levels with Syversen U, Stunes AK, Gustafsson BI, Obrant KJ, Nordsletten L, Berge R, muraglitazar, a dual (alpha/gamma) peroxisome proliferator-activated Thommesen L & Reseland JE 2009 Different skeletal effects of the receptor activator, in patients with type 2 diabetes inadequately controlled peroxisome proliferator activated receptor (PPAR)alpha agonist fenofibrate with metformin monotherapy: a double-blind, randomized, pioglitazone- and the PPARgamma agonist pioglitazone. BMC Endocrine Disorders 9 10. comparative study. Diabetes Care 29 1016–1023. (doi:10.2337/dc05-1146) (doi:10.1186/1472-6823-9-10) Kumar S, Homan S, Samadfam R, Mansell P, Guldberg R, Smith SY & Fitzpatrick L 2009 ASBMR 31st Annual Meeting MO0001-MO0445. Journal of Bone and Mineral Research 24 S370–S496. (doi:10.1002/jbmr. Received in final form 31 October 2011 5650241305) Lalloyer F & Staels B 2010 Fibrates, glitazones, and peroxisome proliferator- Accepted 7 November 2011 activated receptors. Arteriosclerosis, Thrombosis, and Vascular Biology 30 Made available online as an Accepted Preprint 894–899. (doi:10.1161/ATVBAHA.108.179689) 7 November 2011

Journal of Endocrinology (2012) 212, 179–186 www.endocrinology-journals.org

Downloaded from Bioscientifica.com at 09/25/2021 09:19:03AM via free access