Incidence of Fractures in Patients With Type 2 ... - Diabetes Care

Diabetes Care

1

Incidence of Fractures in Patients With Type 2 Diabetes in the SAVOR-TIMI 53 Trial

Ofri Mosenzon,1 Cheryl Wei,2 Jaime Davidson,3 Benjamin M. Scirica,4 Ilan Yanuv,1 Aliza Rozenberg,1 Boaz Hirshberg,5 Avivit Cahn,1 Christina Stahre,6 Krzysztof Strojek,7 Deepak L. Bhatt,4 and Itamar Raz1

DOI: 10.2337/dc15-1068

OBJECTIVE

Patients with type 2 diabetes have an increased risk of bone fractures, the predisposing factors for which are unknown. Treatment with thiazolidinediones (TZDs) further increases the incidence of osteoporotic fractures. In the Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR-TIMI 53) trial, fractures were considered an adverse event of special interest, and information regarding fractures was collected. RESEARCH DESIGN AND METHODS

RESULTS

During a median follow-up of 2.1 years, 241 patients (2.9%) in the saxagliptin group and 240 (2.9%) in the placebo group experienced a fracture (hazard ratio [HR] 1.00 [95% CI 0.83–1.19]). Event rates for fractures were the same in both treatment arms: 14.7 per 1000 patient-years in the entire population and 14.0 in the on-treatment population (first event only). Fracture risk was similar in patients treated with saxagliptin or placebo across different subgroups defined by race, cardiovascular risk, and renal function. A multivariable Cox regression analysis showed that risk of fracture was associated with female gender (P < 0.0001), longer diabetes duration (P < 0.0001), older age (P = 0.002), major hypoglycemic events (P = 0.01), noncompliance with study drug (P = 0.01), and treatment with TZDs (P = 0.03). CONCLUSIONS

In a large population of older patients with type 2 diabetes, treatment with saxagliptin was not associated with an increased risk of fractures. The association between longer diabetes duration and increased risk of bone fracture is an intriguing finding. There is a growing body of evidence that type 1 and type 2 diabetes both predispose to bone fractures (1,2). A recent nested cohort study demonstrated a hazard ratio (HR) of 1.66 (95% CI 1.6–1.72) for the risk of fractures requiring surgery in patients with diabetes compared with age- and sex-matched control subjects (3). The same study also demonstrated increased risk for postfracture complications among the population with diabetes, including deep wound infection, septicemia, and mortality.

1

Diabetes Unit, Hadassah Hebrew University Hospital, Jerusalem, Israel 2 AstraZeneca, Gaithersburg, MD 3 Touchstone Diabetes Center, the University of Texas Southwestern Medical Center, Dallas, TX 4 Thrombolysis in Myocardial Infarction (TIMI) Study Group, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA 5 AstraZeneca Research and Development, Wilmington, DE 6 AstraZeneca, Mölndal, Sweden 7 Department of Internal Diseases Diabetology and Cardiometabolic Diseases, School of Medicine with the Division of Dentistry in Zabrze Medical University of Silesia, Katowice, Poland Corresponding author: Itamar Raz, e-mail: ntv5022netvision.net.il. Received 20 May 2015 and accepted 10 August 2015. Clinical trial reg. no. NCT01107886, clinicaltrials .gov. © 2015 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered.

Diabetes Care Publish Ahead of Print, published online September 10, 2015

PATHOPHYSIOLOGY/COMPLICATIONS

We compared the incidence of fractures among the 8,280 patients who were assigned to treatment with saxagliptin with that in the 8,212 patients who were assigned to placebo. We further analyzed the participants’ baseline characteristics and fracture risk.

2

Fractures Incidence in SAVOR Trial

Recent findings indicate that bone strength is closely linked to the regulation of energy metabolism and insulin sensitivity (4). Bone and glucose homeostasis are controlled by common regulatory factors, including insulin (5), peroxisome proliferator activated receptor-g (6,7), and gastrointestinal hormones such as glucose-dependent insulinotropic peptide (8) and glucagon-like peptide (GLP) (9). Owing to the frequent coexistence of type 2 diabetes and osteoporosis and the higher incidence of fractures among patients with type 2 diabetes, it is important to investigate any interaction between antidiabetic drugs and fracture risk. Thiazolidinediones (TZDs) are strongly associated with a decrease in bone mineral density (BMD) and a twofold increase in the risk for fractures in women (10–12). Although the evidence linking TZD use and fractures is strong, evidence supporting the effect of other, older, antidiabetic drugs on bone strength is much weaker (protective: metformin (12–14), or deleterious: insulin (13,14) and sulfonylurea (13– 15)). In a 4 year follow-up of the Look Action for Health in Diabetes (AHEAD) substudy (16), intensive lifestyle intervention led to weight loss that was associated with a modest increase in bone loss at the hip in men. Animal models indicate that incretins may have beneficial effects on bone mass and bone quality (8,9). A metaanalysis of phase 2 and 3 clinical trials enrolling 20,000 patients treated for 24 weeks or longer, in which dipeptidyl peptidase-4 (DPP-4) inhibitors were compared with placebo or active treatment, showed that treatment with DPP-4 inhibitors was associated with a reduced risk of fractures (17). In a recently published retrospective population-based cohort study in which DPP-4 inhibitor (sitagliptin) users were compared with patients treated with other antidiabetic drugs (excluding insulin) and control subjects without diabetes matched for age and sex, no association was found between DPP-4 inhibitor use and osteoporotic fracture risk (18).We report here the results of the predefined outcome of the adverse event of special interest (AEOSI) of fractures in the Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus (SAVOR-TIMI 53) trial.

Diabetes Care

RESEARCH DESIGN AND METHODS Study Design and Patient Population

SAVOR-TIMI 53 was a multicenter, randomized, double-blind, placebo-controlled trial that randomized 16,492 patients at 788 sites in 26 countries. Patients had type 2 diabetes with HbA1c between 6.5% and ,12.0% within 6 months of randomization and a history of established cardiovascular (CV) disease (CVD) and age .40 years or multiple risk factors for CVD, including age .55 for men or .60 for women. Patients were randomized to receive saxagliptin, 5 mg daily (2.5 mg daily in patients with an estimated glomerular filtration rate [eGFR] of #50 mL/min), or matching placebo. Other background glucose-lowering and CV therapies were administered at the discretion of treating physicians. Open-label titration of glucose-lowering therapies was allowed throughout the trial in both arms, excluding the use of DPP-4 inhibitors or GLP-1 receptor agonists. During the trial, systemic steroid use at baseline and/or for more than 3 months was reported in 6 patients (1.2%) who had a fracture and in 272 patients (1.7%) who did not have a fracture (P = 0.59). The study protocol was approved by the relevant institutional review board at each participating site, and written informed consent was obtained from all patients. The primary and secondary outcomes of this trial were previously described (19). Fractures As Predefined AEOSI and Their Collection

In the protocol, 10 categories were defined as AEOSI, including bone fractures. Investigators were specifically trained to collect all data regarding AEOSI and record them on a specific page in the electronic case report form (eCRF). Fractures were identified by two criteria: events reported by the investigator as an AEOSI on the eCRF searching the full adverse event database for all patients with a diagnosis matching a prespecified Medical Dictionary for Regulatory Activities (MedDRA) version 15.1 list of preferred terms (excluding “tooth fracture”) Analysis was done on the entire population (“overall”) and on the population that was “on treatment.” Overall categories included all subjects who were randomized (intention-to-treat

[ITT] population), including subjects who never took the study medication, and included all AEs that occurred on or after the date of enrollment. “On treatment” refers to AEs occurring while the subject was receiving treatment. AEs were deemed on treatment if, for nonserious AEs, the event occurred on or before the first day after the last blinded drug dosing date, or for serious AEs (SAEs), the event occurred on or before the 30th day after the last blinded drug dosing date. For those who did not experience bone fracture, the on-treatment window ended at the last dose date + 30 days. Statistical Analysis

Baseline characteristics of subjects were examined, and subjects with any fracture event were compared with those without such an event. Analyses were conducted on an ITT basis among patients who underwent randomization and by ontreatment analysis among randomized patients who took at least one dose of the study medication. Categorical variables were compared using the x2 or Fisher exact test and continuous variables with the t test. A P value #0.05 was considered statistically significant. P values were not adjusted for multiple testing. KaplanMeier estimates of cumulative risk for the first occurrence of bone fracture were plotted. For the subgroup analysis, the risks of bone fracture by baseline characteristics were examined using the Cox proportional hazards regression model, with each baseline variable only as the model term. The 95% CIs for the hazard ratios (HR) are based on profile likelihood. Forest plots were used to show HR estimates and 95% CIs by baseline variables. The multivariate Cox regression model was used for analyses of strength of baseline characteristics association with bone fracture. The model included treatment and clinically meaningful baseline covariates as the model terms. Baseline covariates, either continuous or categorical factors, were evaluated for association with bone fracture as measured by the Wald x2 statistical test corresponding to their contribution to the model. Statistical software SAS 8.2 or higher (SAS Institute, Inc., Cary, NC) was used. RESULTS Incidence of Fractures

Fractures occurred in 241 of the 8,280 subjects (2.9%) randomized to the

care.diabetesjournals.org

saxagliptin arm and in 240 of the 8,212 subjects (2.9%) randomized to the placebo arm (HR 1.00; 95% CI 0.83–1.19). Both groups developed 14.7 fractures per 1,000 patient-years. Kaplan Meier analysis of the time to the first event of bone fracture in the ITT population, by different treatment arms, is shown in Fig. 1A. On-treatment analysis showed that there were 222 (2.7%) and 220 (2.7%) incident fractures in the saxagliptin and placebo arms, respectively. The exposure-adjusted rate of subjects with on-treatment AEs of bone fracture per 1,000 years of follow-up (based on the first event only) was 14.0 in both groups. Of the fracture AEOSI, some were also classified as SAEs, mainly due to hospitalization. In the saxagliptin arm overall, there were 95 (1.1%) SAE fractures (88 [1.1%] on treatment), and in the placebo arm overall, there were 84 (1.0%) SAE fractures (79 [1.0%] on treatment; P = 0.53). The proportions of subjects with an AE or SAE of bone fracture were similar between the saxagliptin and placebo groups across the following subgroups: sex, age, race, regions of residence, baseline CV risk, and baseline renal function categories (Fig. 2A). The occurrence of multiple fractures in the same subject was similar between the two treatment arms: In the saxagliptin group, 24 subjects (0.3%) had two to three events and 2 subjects (,0.1%) had four or more events. In the placebo group, 27 subjects (0.3%) had two to three events, and 1 subject (,0.1%) had four or more events. Kaplan-Meier accumulation curves according to diabetes duration in the combined two treatment arms (ITT population) showed a gradual increase in the risk of fractures with longer diabetes duration (Fig. 1B). Baseline Characteristics and Multivariable Analysis of Subjects Who Had Fractures

Baseline characteristics of subjects who had fractures during follow-up is summarized and compared with those without fractures (Table 1). Subjects with fractures were older, more likely to be female, Caucasian, and non-Hispanic or Latino, had longer diabetes duration, lower eGFR, and were more likely to have had a history of dyslipidemia or heart failure and less likely to have had a prior myocardial infarction. Patients

Mosenzon and Associates

with fractures were also more likely to use diuretics, insulin, and TZDs at baseline and less likely to use metformin. The HR of the time to first event for bone fracture was calculated for the ITT population of the combined two treatment arms, according to the following baseline characteristics: CV risk, renal function strata, demographic subgroups (Fig. 2B), diabetes duration, HbA1c at baseline, the occurrence of hypoglycemia, and albumin-to-creatinine ratio at baseline (Fig. 2C), and by baseline diabetes medications (Fig. 2D). The HR for fracture was increased in the following subgroups: moderately reduced renal function versus normal renal function, older age ($65, $75) versus younger, female versus male, residents of North America versus Europe or Latin America, longer diabetes duration, patients with any and with major hypoglycemic events, insulin users, and those who did not use metformin. On multivariable Cox regression analysis, fractures were associated with the following characteristics: female versus male gender (HR 1.84 [95% CI 1.53– 2.21], Wald x2 = 41.96, P , 0.0001), longer diabetes duration (HR 1.02 per year [95% CI 1.01–1.03], Wald x2 = 19.24, P , 0.0001), older age (HR 1.02 per year [95% CI 1.01–1.03], Wald x2 = 9.93, P = 0.002), noncompliance with study medication (saxagliptin or placebo) (HR 1.50 [95% CI 1.10–2.03], Wald x2 = 6.64, P = 0.01), major hypoglycemic events during the study (HR 1.76 [95% CI 1.13–2.74], Wald x2 = 6.22, P = 0.01), and the use of TZDs (HR 1.44 [95% CI 1.03–2.01], Wald x2 4.63 P = 0.03). There was no statistically significant association between increased risk of fractures and the following parameters: eGFR (HR 1.00 [95% CI 0.99–1.00], Wald x2 = 3.56, P = 0.06), HbA 1c (HR 0.93 [95% CI 0.87–1.00], Wald x2 = 3.51, P = 0.06), or insulin use (HR 1.24 [95% CI 0.99–1.56], Wald x2 = 3.37, P = 0.07). Variables that were not associated with the risk for fractures were the use of sulfonylurea (P = 0.84) or metformin (P = 0.92), BMI (P = 0.63), and the randomized treatment arm (P = 0.95).

osteoporotic fracture was defined, according to the World Health Organization definition (20), as a fracture of the hip, vertebrae, radius/ulna, or humerus. The difference between the total number of fractures at these locations between saxagliptin (110 fractures) and placebo (102 fractures) was not significant (P = 0.62). CONCLUSIONS Saxagliptin Versus Placebo

The SAVOR-TIMI 53 trial was a large randomized controlled trial performed in relatively old (mean age 65.1 6 8.5) patients with type 2 diabetes, in whom fracture incidence was prespecified as AEOSI and actively collected. Over a median follow-up of 2.1 years, the incidence of fractures was equally distributed between saxagliptin- and placebo-treated arms, in the ITT population (14.7 fractures per 1,000 patient-years), and also in the on-treatment population (14.0 fractures per 1,000 patient-years). Further, no differences were found in the incidence of fractures between the different treatment arms in various predefined subgroups. This is especially important considering some of the high-risk groups that were included in the study: sizeable female population of 5,455, large elderly population of 8,561 aged $65 years old and 2,330 aged $75 years old, and a large population with renal dysfunction, with 2,240 with moderately decreased renal function (eGFR 30–50 mL/min) and 336 with severely decreased renal function (eGFR ,30 mL/min). Increased risks of fractures among patients with type 2 diabetes were previously reported in the hip (1,2,21,22), proximal humerus (1,2,21), vertebra (23), and all nonvertebral fractures (1,21,22,24). No significant differences were found between treatment arms in the distribution of the location of fractures in the SAVOR-TIMI 53 patients (Fig. 3). There was also no difference in major osteoporotic fracture as classified according to the World Health Organization definition (20). Therefore, in conclusion, saxagliptin therapy does not influence the risk of fractures among patients with type 2 diabetes.

Location of Fractures

The distribution of the location of the fractures at the different regions of the skeleton was generally similar between the two treatment arms (Fig. 3). A major

The Incretin System and Bone Metabolism

Despite the abundant evidence regarding the interaction between the incretin

3

4


Diabetes Care

Figure 1—Kaplan-Meier estimate of time to the first bone fracture (derived from CRF or Preferred Terms) (A) and by duration of diabetes mellitus (DM) (B) in the ITT population.

system and bone health, the results of the SAVOR-TIMI 53 trial do not support a correlation between the use of the DPP-4 inhibitor saxagliptin and the risk of fractures. In animal models, intermittent administration of glucose-dependent insulinotropic peptide had an anabolic effect and prevented bone loss, similar to parathyroid hormone (25). GLP-1 appears to have a double-inhibitory effect on bone resorption, one direct and another indirect, via the thyroid C cells (9,26). A beneficial effect on bone structure was observed in rats with impaired glucose metabolism treated with exendin-4 (27).

Considering data derived from incretinbased therapy in patients with type 2 diabetes, a 44-week treatment with exenatide versus insulin glargine did not affect BMD (28). After 1-year treatment with the DPP-4 inhibitor vildagliptin, levels of markers of bone resorption and calcium homeostasis were unaffected compared with baseline and placebo (29). A meta-analysis of 28 phase 2 and 3 clinical trials enrolling 20,000 patients treated for 24 weeks or longer showed that treatment with DPP-4 inhibitors was associated with a reduced risk of fractures compared with placebo or active treatment (17). Several factors weaken this meta-analysis compared

with the current study: its retrospective nature, post hoc analysis, the inclusion of only fractures that were reported as SAEs, comparison of various interventions, short duration of follow-up (median 35 weeks), and small number of fractures (11 of 5,877 in the DPP-4 inhibitors vs. 21 of 4,928 in the comparators). However, the discrepancy between the meta-analysis and this report cannot fully be explained by the limitations of the meta-analysis and remains puzzling. The discrepancy is further supported by the difference between the findings in this trial and the results of a post hoc pooled analysis of 20 phase 2 and 3 trials (n = 9,156) comparing saxagliptin with placebo or active comparator, which found an increased risk for fractures with saxagliptin treatment (HR 1.81 [95% CI 1.04–3.28]) (30). The patient population in the pooled analysis was younger, with a shorter duration of diabetes, better renal function, less concomitant medication, and generally healthier compared with the SAVOR TIMI-53 patient population. The difference in patient population may explain the discrepancy between the pooled analysis and the current study results. In a retrospective population-based cohort that compared 22,510 DPP-4 inhibitor (sitagliptin) users out of 216,816 patients treated with other antidiabetic drugs (excluding insulin) and 216,816 control subjects without diabetes matched for age and sex, no association was found between DPP4-I use and osteoporotic fracture risk (18). Despite the large number of patients, that study had a few inherent weaknesses compared with our study, including its retrospective unrandomized nature (differences in the distribution of age, sex, etc. between the groups), and shorter duration of DPP-4 inhibitor treatment (median 1.3 years). Patients With Type 2 Diabetes, Baseline Characteristics, and the Risk of Fractures

To help define factors that were associated with increased risk of fractures in our population, we conducted a Cox multivariable regression analysis. The following parameters were independently associated with a higher risk for fractures: female sex, longer diabetes duration, older age, noncompliance with study medication (saxagliptin or placebo), major hypoglycemic events



Figure 2—A: Forest plot of the HR of AE of bone fracture between the saxagliptin and placebo groups across subgroups by demographic and medical history parameters. Forest plot of the HR of the time to first event for bone fracture (derived from CRF or Preferred Terms) by baseline cardiovascular risk, renal function strata, and demographic subgroups (B) by demographic subgroups (C), and by baseline diabetes medications (DM) (D) in the ITT population. ACR, albuminto-creatinine ratio; CHF, congestive heart failure; MRF, multiple risk factors. The solid squares denote the mean difference and are proportional to the weights used, the horizontal lines represent the 95% CIs, and the vertical line indicates no effect. (A high-quality color representation of this figure is available in the online issue.)

during the study, and the use of TZDs. The following parameters were not significantly associated with the risk of fractures, although they were somewhat borderline: lower eGFR, lower HbA 1c , and insulin use. BMI and the use of metformin, sulfonylurea, or

saxagliptin were not associated with the risk of fractures in our population. Female sex, older age, and TZD use are well-known risk factors for fractures. Longer diabetes duration was previously shown to increase fracture risk in retrospective cohort studies (13,31,32) and in

prospective population-based studies: among women (21) and in previously known and treated patients with diabetes but not in newly diagnosed diabetes, in patients with impaired glucose tolerance, (22) or in the Chinese population (33). This association is strengthened by the similar findings in our randomized controlled trial. The recently published analysis of nonvertebral fractures from the Action To Control Cardiovascular Risk in Diabetes (ACCORD)-bone substudy (34) also demonstrated longer diabetes duration among participants with fractures compared with those without fractures; however, that trial was not controlled for age. As opposed to this finding, a case-controlled, nationwide study from Denmark did not find any association between diabetes duration and the risk of fractures (35); however, that study was done in a very different population with a mean age of 43 years. Longer diabetes duration might be associated with greater tendency to fall due to sight disturbances, peripheral and autonomic neuropathy, impaired proprioception and/or postural hypotension. Another possible explanation is that in patients with longer duration of diabetes, the bone is more fragile and more likely to break. A possible mechanism for this includes prolonged exposure to advanced glycation end products that damage bone collagen and affect bone metabolism (36) and increased visceral fat tissue, which is associated with ongoing inflammation that might also damage bone structure (37). We found no significant association between baseline eGFR and the risk of fractures in the Cox multivariable analysis. Our findings are strengthened by the ACCORD-bone analysis (34), which also did not demonstrate an association between the risk of fracture and baseline eGFR, but rather, found an association between the slope of deterioration of eGFR and the risk of fractures among women. We found a significantly increased risk for fractures among participants who had a major hypoglycemic event (HR 1.76 [95% CI 1.13–2.74]). This strengthens findings in two previously published studies, a case-controlled nationwide study from Denmark (35) that demonstrated an increased risk for hip fractures and a retrospective observational study that demonstrated an increased risk for fall-related fractures

5

6


Diabetes Care

Table 1—Baseline characteristics of subjects with or without fracture during follow-up by ITT analysis Fracture during follow-up Characteristic Demographic characteristics Age, median (IQR) years Female sex, n (%) Race Caucasian, n (%) All others, n (%) Hispanic/Latino ethnicity, n (%) Weight , median (SD) kg BMI, median (SD) kg/m2 Diabetes duration, median (IQR) years Current smoker, n (%) Hypertension, n (%) Dyslipidemia, n (%) Prior myocardial infarction, n (%) Prior heart failure, n (%) Prior coronary revascularization, n (%) Baseline measurements eGFR, median (SD) mL/min Albumin-to-creatinine ratio, median (SD) mg/mmol HbA1c, median (SD) % Fasting serum glucose, median (SD) mg/dL Baseline cardiovascular medications, n (%) Aspirin Statins b-Blockers Diuretics ACE inhibitors Baseline antihyperglycemic medications, n (%) Metformin Sulfonylurea TZD Insulin

Yes (n = 481)

No (n = 16,011)

P value

67 (13) 236 (49.1)

65 (11) 5,219 (32.6)

,0.0001 ,0.0001

407 (84.6) 74 (15.4) 74 (15.4) 85.2 (27.2) 30.86 (7.84) 12.66 (13.77) 60 (12.5) 408 (84.8) 368 (76.5) 150 (31.2) 78 (16.2) 219 (45.5)

12,000 (74.9) 4,011 (25.1) 3,467 (21.7) 86.0 (25.2) 30.45 (7.20) 10.29 (11.36) 2,159 (13.5) 13,084 (81.7) 11,371 (71.0) 6,087 (38.0) 2,027 (12.7) 6,904 (43.1)

,0.0001

66.3 (31.3)

71.8 (29.3)

,0.0001

2.1 (9.7) 7.5 (1.6) 142 (60)

1.8 (6.9) 7.6 (1.8) 145 (64)

0.5697 0.1236 0.0652

349 (72.6) 392 (81.5) 284 (59.0) 232 (48.2) 252 (52.4)

12,055 (75.3) 12,525 (78.2) 9,878 (61.7) 6,833 (42.7) 8,688 (54.3)

0.1801 0.0919 0.2534 0.0170 0.4299

310 (64.4) 179 (37.2) 39 (8.1) 239 (49.7)

11,113 (69.4) 6,407 (40.0) 931 (5.8) 6,548 (40.9)

0.0211 0.2197 0.0389 0.0001

0.0007 0.8280 0.1572 ,0.0001 0.5873 0.0928 0.0091 0.0022 0.0262 0.293

Categorical variables were compared using the x2 or Fisher exact test and continuous variables with the t test. P , 0.05 was considered to be statistically significant. P values were not adjusted for multiple testing. IQR, interquartile range.

(38) in patients with major hypoglycemic events. The interaction between BMI and even BMD and the risk of fracture in patients with type 2 diabetes is complex and dependent also on fracture site and patient age (39,40). BMD seems to underestimate fracture risk in this population. Our study further indicated that higher BMI, and most likely higher BMD, should not be considered as completely protective factors against fractures or osteoporosis in the population with type 2 diabetes, and appropriate diagnostic measurements and treatment for this condition should therefore be considered. Antidiabetic Drugs and the Risk of Fractures

Our study found that the only antidiabetic drugs that had a statistically

significant effect on the risk of fractures were TZDs. Although only ;5% of our patient population was using TZDs at any time during the trial (19), the use of TZDs was associated with an increased risk for fractures, which further demonstrates the strong effect of this group of drugs on the risk of fractures. TZDs are strongly associated with a decrease in BMD and up to twofold increases in the risk for fractures among postmenopausal woman (10–12). TZDs increase the risk of fractures by multiple mechanisms, including an increase in lineage allocation of the stem cells toward adipocytes at the expense of osteoblasts in the marrow, resulting in reduced bone formation (6). In our study, metformin was used in ;70% of the entire study population (19). There was a significantly higher

percentage of patients who used metformin at baseline among patients who did not have a fracture compared with those that had a fracture (69.4% vs. 64.4%, P = 0.021) (Table 1) as well as decreased a HR of 0.77 (95% CI 0.64– 0.93) (Fig. 2D). However, multivariable analysis found that metformin use was not independently associated with a reduction in fracture risk (P = 0.92). This might partly be explained by the lower use of metformin among the large population of patients with renal dysfunction. Although the Rochester cohort suggests that metformin decreases fracture risk in patients with type 2 diabetes (with a HR of 0.7) (13), the A Diabetes Outcome Progression Trial (ADOPT) study did not demonstrate similar beneficial effects of metformin on fracture risk (12). It did show decreased levels of both bone resorption markers (C-telopeptide cross-linked collagen type 1) and, contrary to animal studies, decreased levels of bone formation markers (procollagen type 1 N propeptide) (15). No bone metabolic markers were measured in our study, and the follow-up might be too short to show any clinical difference with the use of metformin. More than 40% of the patients in our study were using insulin at baseline. The univariate Cox regression analysis showed that insulin was associated with an increased risk for fractures (HR 1.45 [95% CI 1.21–1.73]) (Fig. 2D). However, at the multivariate Cox regression analysis, insulin use was not associated with a significant change in the risk for fractures in our population, possibly due to the association between insulin use and diabetes duration. Contrary to our findings, in another study, insulin-treated women with diabetes had an almost doubled fall incidence (odds ratio 2.78 vs. 1.68) (14), which might lead to a higher risk for fractures. In the Rochester cohort, insulin use slightly but significantly increased the fracture rate (13). A prospective study demonstrated that insulin-treated older women with diabetes had more than double the risk of foot fractures (multivariate-adjusted risk ratio of 2.66) compared with patients without diabetes and noninsulinuser patients with diabetes (19). In our study, ;40% of the population was using sulfonylureas (19). The use of sulfonylureas was associated with increased risk for hypoglycemic events in



Figure 3—Distribution of the location of fractures at the ITT population for saxagliptin vs. placebo arms (Subjects with events in more than one category are counted in each category. Subjects with multiple events in the same category are counted only once in that category). (A high-quality color representation of this figure is available in the online issue.)

the saxagliptin arm compared with the placebo arm. Hypoglycemia and, specifically, major hypoglycemic events were associated with increased risk of fracture, and major hypoglycemia was an independent risk factor for fractures in the multivariate Cox regression analysis. However, sulfonylurea use was not associated with an increased risk of fractures at the single-variable analysis (Fig. 2D) or in the multivariate Cox regression analysis. Similar to our findings, the use of sulfonylurea (glyburide) did not have an effect on bone mass and fracture risk in the ADOPT clinical trial and in the Rochester cohort study (13). Study Strengths and Weaknesses

The strength of the study presented here is the large and diverse population of patients with type 2 diabetes monitored

prospectively in the SAVOR-TIMI 53 trial. This population includes a large prespecified subset of patients at a considerably high risk for fractures, including women, elderly patients, and patients with renal dysfunction. The diversity of the population regarding HbA1c at baseline, duration of diabetes, and antidiabetes treatment, improve the external validity of this research. Fractures were predefined AEOSI, and information was actively collected throughout the study, which enhances the confidence in the data presented here. A large percentage of fractures were also classified as SAE due to hospitalization, and more data were collected for these cases. The weakness of our study is that fractures were exploratory end points in this trial. Therefore, this study was not designed to collect the data

necessary to properly and thoroughly evaluate fractures in general and osteoporotic fractures in particular. All fractures, regardless of their mechanism, were classified together, and any evidence regarding BMD or bone turnover markers was missing. The lack in fracture ascertainment might attenuate the possible association between treatment arms and fracture risk. Another weakness is the relatively short duration of follow-up (median 2.1 years, with interquartile range of 1.8–2.3) (19), which was just barely longer than 1–1.5 years of use, which is the length of time needed for drugs used for the treatment of osteoporosis (such as bisphosphonates) to show a diminishing effect on the risk of fracture. Lastly, even though the SAVOR-TIMI 53 is one of the largest trials done in this population of patients

7

8


with type 2 diabetes at high CV and fracture risk, the power of the trial was designed by the primary combined CV outcome and not by the fracture risk and, therefore, may not completely exclude possible change in fracture risk under saxagliptin treatment. Summary

Saxagliptin neither increased nor decreased the incidence of fractures in a large, diverse population at high risk for fractures. Because fractures are highly prevalent among patients with type 2 diabetes, this is an important safety issue to consider when choosing antidiabetic treatment. We found a strong association between longer diabetes duration and the risk of fractures. This is a significant factor for clinicians to consider, especially because BMD underestimates the risk for fractures among patients with type 2 diabetes. Further data collected through ongoing large outcome studies, as well as by observational studies, may further shed light on this important overlap between two of the largest epidemics of our time: type 2 diabetes and osteoporosis. APPENDIX

SAVOR-TIMI 53 Executive Committee: Eugene Braunwald (Study Chair), Deepak L. Bhatt (Co-Principal Investigator), Itamar Raz (Co-Principal Investigator), Jaime A. Davidson, Robert Frederich (nonvoting), Boaz Hirshberg (nonvoting), Gabriel Steg.

Funding. The SAVOR-TIMI 53 trial was funded by AstraZeneca and Bristol-Myers Squibb. Duality of Interest. O.M. discloses the following relationshipsdadvisory board: Novo Nordisk, Eli Lilly and Company, Sanofi, Merck Sharp & Dohme, Boehringer Ingelheim, Janssen Pharmaceuticals, Novartis Pharmaceuticals AG, AstraZeneca; grants paid to institution as study physician: AstraZeneca and Bristol-Myers Squibb; research grant support through Hadassah Hebrew University Hospital: Novo Nordisk; speaker’s bureau: AstraZeneca and BristolMyers Squibb, Novo Nordisk, Eli Lilly and Company, Sanofi, Novartis Pharmaceuticals AG, Merck Sharp & Dohme, and Boehringer Ingelheim. C.W. is an AstraZeneca employee. J.D. discloses the following relationshipsd advisory board: Amgen Inc., Aspire Bariatrics, AstraZeneca, Eli Lilly and Company, GlaxoSmithKline, Janssen Pharmaceuticals, LifeScan, Merck Sharp & Dohme, Novo Nordisk, Takeda Pharmaceutical Company Ltd, and Valeritas, Inc.; speaker’s bureau: AstraZeneca, Becton, Dickinson and Company, Janssen Pharmaceuticals, LifeScan, and Novo Nordisk; other: American Association of Clinical Endocrinologists board member, The

Diabetes Care

Journal of Diabetes associate editor; SAVOR principal investigator and executive committee, President WorldWIDE Diabetes, University of Texas Southwestern Medical Center Continuing Medical Education Executive Committee. B.M.S. discloses the following relationshipsd research grants: TIMI Study Group and Brigham and Women’s Hospital from AstraZeneca and Bristol-Myers Squibb, Daiichi-Sankyo Co., Ltd, GlaxoSmithKline, Johnson & Johnson, Bayer Healthcare, Gilead Sciences, Inc., Eisai Co., Ltd, and Merck Sharp & Dohme; consulting fees: Gilead Sciences, Inc., Lexicon, Arena, Eisai Co., Ltd, St. Jude Medical, Inc., Bristol-Myers Squibb, Forest Laboratories, Inc.,, Boston Clinical Research Institute, Decision Resources, University of Calgary, and Elsevier Practice Update Cardiology. B.H. is an AstraZeneca employee. A.C. discloses the following relationshipsd honoraria for advisory work and lectures: AstraZeneca, Boehringer Ingelheim, Elli Lilly and Company, Merck Sharpe & Dohme, Novartis Pharmaceuticals AG, Novo Nordisk, Sanofi, and Teva. C.S. is an AstraZeneca employee. K.S. discloses the following relationshipsdhonoraria for speaking engagements: Eli Lilly and Company, Novo Nordisk, Sanofi, Servier, and Janssen-Cilag; participation in clinical trials: Astra Zeneca. D.L.B. discloses the following relationshipsdadvisory board: Cardax, Inc., Elsevier Practice Update Cardiology, Medscape Cardiology, and Regado Biosciences; board of directors: Boston VA Research Institute, Society of Cardiovascular Patient Care; chair: American Heart Association Get With The Guidelines Steering Committee; data monitoring committees: Duke Clinical Research Institute, Harvard Clinical Research Institute, Mayo Clinic, and Population Health Research Institute; honoraria: American College of Cardiology (Clinical Trials and News senior associate editor, ACC.org), Belvoir Publications (Harvard Heart Letter editor in chief), Duke Clinical Research Institute (clinical trial steering committees), Harvard Clinical Research Institute (clinical trial steering committee), HMP Communications (Journal of Invasive Cardiology editor in chief), Journal of the American College of Cardiology (associate editor, section editor, pharmacology), Population Health Research Institute (clinical trial steering committee), Slack Publications (Cardiology Today’s Intervention chief medical editor), WebMD (continuing medical education steering committees), Clinical Cardiology (deputy editor); research funding: Amarin, AstraZeneca, Bristol-Myers Squibb, Eisai, Ethicon, Forest Laboratories, Ischemix, Medtronic, Pfizer, Roche, Sanofi, and The Medicines Company; unfunded research: FlowCo, PLx Pharma LLC, and Takeda Pharmaceutical Company Ltd. I.R. discloses the following relationshipsdadvisory panel: AstraZeneca/Bristol-Meyers Squibb, Eli Lilly and Company, Medscape, LLC, Merck Sharp & Dohme, Novo Nordisk, Sanofi, Orgenesis Inc., SmartZyme Innovation Ltd, and LabStyle Innovations Ltd; consultant: AstraZeneca/BristolMeyers Squibb, Insuline Medical Ltd, Gili Medical Ltd, Kamada Ltd, and FuturRx Ltd; speaker’s bureau: AstraZeneca/Bristol-Meyers Squibb, Eli Lilly and Company, Johnson & Johnson, Merck Sharp & Dohme, Novartis Pharmaceuticals AG, Novo Nordisk, Sanofi, and Teva; stock/ shareholder: Insuline Medical Ltd, LabStyle Innovations, SmartZyme Innovation Ltd, Orgenesis

Inc., and GlucoMe Ltd. No other potential conflicts of interest relevant to this article were reported. Author Contributions. O.M. acquired, analyzed, and interpreted the data, drafted the manuscript and revised the manuscript for important intellectual content, corrected the manuscript according to reviewer comments, and approved the final version of the manuscript submitted. C.W. was part of the executive committee and helped to design the trial, acquire, analyze, and interpret the data, and approved the final version of the manuscript submitted. J.D. and K.S. reviewed the manuscript for important intellectual content and approved the final version of the manuscript submitted. B.M.S., B.H., and C.S. conceived and designed the study, helped to acquire, analyze, and interpret the data, reviewed the manuscript for important intellectual content, and approved the final version of the manuscript submitted. I.Y. and A.R. helped to acquire, analyze, and interpret the data, revised the manuscript for important intellectual content, and approved the final version of the manuscript submitted. A.C. helped to acquire, analyze, and interpret the data, reviewed and revised the manuscript for important intellectual content, and approved the final version of the manuscript submitted. D.L.B. conceived and designed the study, helped to acquire, analyze, and interpret the data, reviewed and revised the manuscript for important intellectual content, and approved the final version of the manuscript submitted. I.R. conceived and designed the study, helped to acquire, analyze, and interpret the data, and reviewed and revised the manuscript for important intellectual content. O.M., C.W., B.M.S., B.H., C.S., D.L.B., and I.R. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Prior Presentation. Parts of this study were presented at 74th Scientific Sessions of the American Diabetes Association, San Francisco, CA, 13–17 June 2014.

References 1. Nicodemus KK, Folsom AR; Iowa Women’s Health Study. Type 1 and type 2 diabetes and incident hip fractures in postmenopausal women. Diabetes Care 2001;24:1192–1197 2. Vestergaard P. Discrepancies in bone mineral density and fracture risk in patients with type 1 and type 2 diabetes–a meta-analysis. Osteoporos Int 2007;18:427–444 3. Liao CC, Lin CS, Shih CC, et al. Increased risk of fracture and postfracture adverse events in patients with diabetes: two nationwide population-based retrospective cohort studies. Diabetes Care 2014;37:2246–2252 4. Lecka-Czernik B. Safety of anti-diabetic therapies on bone. Clin Rev Bone Miner Metab 2013;11:49–58 5. Ferron M, Wei J, Yoshizawa T, et al. Insulin signaling in osteoblasts integrates bone remodeling and energy metabolism. Cell 2010;142: 296–308 6. Ali AA, Weinstein RS, Stewart SA, Parfitt AM, Manolagas SC, Jilka RL. Rosiglitazone causes bone loss in mice by suppressing osteoblast


differentiation and bone formation. Endocrinology 2005;146:1226–1235 7. Akune T, Ohba S, Kamekura S, et al. PPARgamma insufficiency enhances osteogenesis through osteoblast formation from bone marrow progenitors. J Clin Invest 2004;113:846–855 8. Xie D, Zhong Q, Ding KH, et al. Glucosedependent insulinotropic peptide-overexpressing transgenic mice have increased bone mass. Bone 2007;40:1352–1360 9. Yamada C, Yamada Y, Tsukiyama K, et al. The murine glucagon-like peptide-1 receptor is essential for control of bone resorption. Endocrinology 2008;149:574–579 10. Grey A, Bolland M, Gamble G, et al. The peroxisome proliferator-activated receptorgamma agonist rosiglitazone decreases bone formation and bone mineral density in healthy postmenopausal women: a randomized, controlled trial. J Clin Endocrinol Metab 2007;92: 1305–1310 11. Yaturu S, Bryant B, Jain SK. Thiazolidinedione treatment decreases bone mineral density in type 2 diabetic men. Diabetes Care 2007;30: 1574–1576 12. Kahn SE, Zinman B, Lachin JM, et al.; Diabetes Outcome Progression Trial (ADOPT) Study Group. Rosiglitazone-associated fractures in type 2 diabetes: an analysis from A Diabetes Outcome Progression Trial (ADOPT). Diabetes Care 2008;31:845–851 13. Melton LJ 3rd, Leibson CL, Achenbach SJ, Therneau TM, Khosla S. Fracture risk in type 2 diabetes: update of a population-based study. J Bone Miner Res 2008;23:1334–1342 14. Schwartz AV, Hillier TA, Sellmeyer DE, et al. Older women with diabetes have a higher risk of falls: a prospective study. Diabetes Care 2002; 25:1749–1754 15. Zinman B, Haffner SM, Herman WH, et al.; ADOPT Study Group. Effect of rosiglitazone, metformin, and glyburide on bone biomarkers in patients with type 2 diabetes. J Clin Endocrinol Metab 2010;95:134–142 16. Lipkin EW, Schwartz AV, Anderson AM, et al.; Look AHEAD Research Group. The Look AHEAD Trial: bone loss at 4-year follow-up in type 2 diabetes. Diabetes Care 2014;37:2822– 2829 17. Monami M, Dicembrini I, Antenore A, Mannucci E. Dipeptidyl peptidase-4 inhibitors and bone fractures: a meta-analysis of randomized clinical trials. Diabetes Care 2011; 34:2474–2476 [Erratum in Diabetes Care 2014;37:312]


18. Driessen JH, van Onzenoort HA, Henry RM, et al. Use of dipeptidyl peptidase-4 inhibitors for type 2 diabetes mellitus and risk of fracture. Bone 2014;68:124–130 19. Scirica BM, Bhatt DL, Braunwald E, et al.; SAVOR-TIMI 53 Steering Committee and Investigators. Saxagliptin and cardiovascular outcomes in patients with type 2 diabetes mellitus. N Engl J Med 2013;369:1317–1326 20. FRAX. WHO Fracture Risk Assessment Tool [article online], 2001. Available from http:// www.shef.ac.uk/FRAX/index.aspx. Accessed 20 May 2015 21. Schwartz AV, Sellmeyer DE, Ensrud KE, et al.; Study of Osteoporotic Features Research Group. Older women with diabetes have an increased risk of fracture: a prospective study. J Clin Endocrinol Metab 2001;86:32–38 22. de Liefde II, van der Klift M, de Laet CE, van Daele PL, Hofman A, Pols HA. Bone mineral density and fracture risk in type-2 diabetes mellitus: the Rotterdam Study. Osteoporos Int 2005;16: 1713–1720 23. Yamamoto M, Yamaguchi T, Yamauchi M, Kaji H, Sugimoto T. Diabetic patients have an increased risk of vertebral fractures independent of BMD or diabetic complications. J Bone Miner Res 2009;24:702–709 24. Strotmeyer ES, Cauley JA, Schwartz AV, et al. Nontraumatic fracture risk with diabetes mellitus and impaired fasting glucose in older white and black adults: the health, aging, and body composition study. Arch Intern Med 2005; 165:1612–1617 25. Bollag RJ, Zhong Q, Ding KH, et al. Glucosedependent insulinotropic peptide is an integrative hormone with osteotropic effects. Mol Cell Endocrinol 2001;177:35–41 26. Nuche-Berenguer B, Moreno P, Esbrit P, et al. Effect of GLP-1 treatment on bone turnover in normal, type 2 diabetic, and insulinresistant states. Calcif Tissue Int 2009;84:453–461 27. Nuche-Berenguer B, Moreno P, Portal-Nu~ nez S, Dap´ıa S, Esbrit P, Villanueva-Pe~ nacarrillo ML. Exendin-4 exerts osteogenic actions in insulinresistant and type 2 diabetic states. Regul Pept 2010;159:61–66 28. Bunck MC, Eliasson B, Corne´ r A, et al. Exenatide treatment did not affect bone mineral density despite body weight reduction in patients with type 2 diabetes. Diabetes Obes Metab 2011;13:374–377 29. Bunck MC, Poelma M, Eekhoff EM, et al. Effects of vildagliptin on postprandial markers of bone resorption and calcium homeostasis in

recently diagnosed, well-controlled type 2 diabetes patients. J Diabetes 2012;4:181–185 30. Hirshberg B, Parker A, Edelberg H, Donovan M, Iqbal N. Safety of saxagliptin: events of special interest in 9156 patients with type 2 diabetes mellitus. Diabetes Metab Res Rev 2014;30: 556–569 31. Leslie WD, Lix LM, Prior HJ, Derksen S, Metge C, O’Neil J. Biphasic fracture risk in diabetes: a population-based study. Bone 2007; 40:1595–1601 32. Lipscombe LL, Jamal SA, Booth GL, Hawker GA. The risk of hip fractures in older individuals with diabetes: a population-based study. Diabetes Care 2007;30:835–841 33. Koh WP, Wang R, Ang LW, Heng D, Yuan JM, Yu MC. Diabetes and risk of hip fracture in the Singapore Chinese Health Study. Diabetes Care 2010;33:1766–1770 34. Isakova T, Craven TE, Scialla JJ, et al.; Action to Control Cardiovascular Risk in Diabetes (ACCORD) Trial. Change in estimated glomerular filtration rate and fracture risk in the Action to Control Cardiovascular Risk in Diabetes Trial. Bone 2015;78:23–27 35. Vestergaard P, Rejnmark L, Mosekilde L. Relative fracture risk in patients with diabetes mellitus, and the impact of insulin and oral antidiabetic medication on relative fracture risk. Diabetologia 2005;48:1292–1299 36. Saito M, Marumo K. Collagen cross-links as a determinant of bone quality: a possible explanation for bone fragility in aging, osteoporosis, and diabetes mellitus. Osteoporos Int 2010; 21:195–214 37. Reid IR. Relationships between fat and bone. Osteoporos Int 2008;19:595–606 38. Johnston SS, Conner C, Aagren M, Ruiz K, Bouchard J. Association between hypoglycaemic events and fall-related fractures in Medicarecovered patients with type 2 diabetes. Diabetes Obes Metab 2012;14:634–643 39. Leslie WD, Morin SN, Lix LM, Majumdar SR. Does diabetes modify the effect of FRAX risk factors for predicting major osteoporotic and hip fracture? Osteoporos Int 2014;25:2817– 2824 40. Schwartz AV, Vittinghoff E, Bauer DC, et al.; Study of Osteoporotic Fractures (SOF) Research Group; Osteoporotic Fractures in Men (MrOS) Research Group; Health, Aging, and Body Composition (Health ABC) Research Group. Association of BMD and FRAX score with risk of fracture in older adults with type 2 diabetes. JAMA 2011; 305:2184–2192

9