During early menopause, steady-state bone remodeling is perturbed; the number of basic multicellular units (BMUs) excavating cavities upon the endosteal surface exceeds the number (generated before menopause) concurrently refilling. Later in menopause, steady-state is restored; the many BMUs generated in early menopause refill as similarly large numbers of BMUs concurrently excavate new cavities. We hypothesized that risedronate reduces the number of cavities excavated. However, in younger postmenopausal women, the fewer cavities excavated will still exceed the fewer BMUs now refilling, so net porosity increases, but less than in controls. In older postmenopausal women, the fewer cavities excavated during treatment will be less than the many (generated during early menopause) now refilling, so net porosity decreases and trabecular volumetric bone mineral density (vBMD) increases. We recruited 324 postmenopausal women in two similarly designed double-blind placebo-controlled studies that included 161 younger (Group 1, ≤ 55 years) and 163 older (Group 2, ≥ 55 years) women randomized 2:1 to risedronate 35 mg/week or placebo. High-resolution peripheral computed tomography was used to image the distal radius and tibia. Cortical porosity was quantified using the StrAx1.0 software. Risedronate reduced serum carboxyterminal cross-linking telopeptide of type 1 bone collagen (CTX-1) and serum amino-terminal propeptide of type 1 procollagen (P1NP) by ∼50%. In the younger group, distal radius compact-appearing cortex porosity increased by 4.2% ± 1.6% (p = 0.01) in controls. This was prevented by risedronate. Trabecular vBMD decreased by 3.6% ± 1.4% (p = 0.02) in controls and decreased by 1.6% ± 0.6% (p = 0.005) in the risedronate-treated group. In the older group, changes did not achieve significance apart from a reduction in compact-appearing cortex porosity in the risedronate-treated group (0.9% ± 0.4%, p = 0.047). No between-group differences reached significance. Results were comparable at the distal tibia. Between-group differences were significant for compact-appearing cortex porosity (p = 0.005). Risedronate slows microstructural deterioration in younger and partly reverses it in older postmenopausal women, features likely to contribute to antifracture efficacy. © 2014 American Society for Bone and Mineral Research.
Bone's material composition and microstructure are maintained by bone remodeling; the focal replacement of a volume of damaged or older bone by new bone formed by osteoblasts of the basic multicellular unit (BMU).[1, 2] During young adulthood, remodeling is slow and in steady-state; the number of BMUs excavating cavities upon the intracortical, endocortical, and trabecular components of bones' inner (endosteal) surface equals the number of BMUs concurrently refilling cavities excavated earlier at other locations. There is little or no net bone loss or structural decay because remodeling is balanced; each BMU removes and deposits almost equal volumes of bone (Fig. 1A).
A BMU takes ∼3 weeks to excavate a cavity but ∼3 months to refill it, so there is a transitory deficit in mineralized bone matrix volume created by cavities just excavated, incompletely refilled cavities excavated weeks earlier, and incomplete secondary matrix mineralization, a process that takes many months.[3-6] This reversible deficit in mineralized matrix volume is ever present but not apparent during steady-state remodeling because sites being excavated are matched by equal numbers of sites excavated earlier that are concurrently refilling at different locations.
However, this reversible deficit in mineralized matrix volume becomes apparent and contributes to changes in morphology when steady-state remodeling is perturbed. For example, during early menopause, there is a rapid increase in the number of BMUs upon the endosteal surface excavating cavities and this number exceeds the few cavities excavated before menopause (when remodeling was slow) that are now refilling.[7, 8] This produces a net increase in cortical porosity and decrease in trabecular bone volume responsible for the rapid decrease in total volumetric bone mineral density (vBMD) (Fig. 1B).[9-11] The increase in birth rate of BMUs in early menopause is sustained into later menopause and advanced age, but now the large numbers of cavities excavated are matched by refilling of the equally large numbers excavated in early menopause; steady-state remodeling is restored, but at a higher remodeling rate than steady-state before menopause (Fig. 1C).[5, 6]
If the rapidity of remodeling was the only abnormality produced by menopause, bone loss would cease because equal numbers of sites are excavated and filled, as before menopause; but after menopause, there are more sites. Bone loss occurs because estrogen deficiency also increases the volume of bone resorbed and reduces the volume of bone formed by each BMU, producing a negative bone balance and structural decay.[12-14]
Antiresorptive therapies perturb remodeling by rapidly decreasing the number of cavities excavated by BMUs.[5, 15] We hypothesized that the net effect of risedronate on bone morphology will depend on whether remodeling was perturbed or in steady-state at the time of administration.[16, 17] We proposed that in younger postmenopausal women, the reduced numbers of cavities excavated during treatment will still exceed the fewer cavities (generated before menopause) now refilling so that net porosity will still increase, albeit less than in controls (Fig. 1B). Trabecular vBMD will also continue to decrease during treatment, but less so than in controls. In older postmenopausal women, when remodeling is likely to have returned to steady-state, the cavities excavated during treatment will be fewer than the large numbers (excavated in early menopause) now refilling, producing a net decrease in porosity and a net increase in trabecular bone volume fraction (Fig. 1C).
Subjects and Methods
Study design and subjects
The flowchart of the study design is summarized in Fig. 2. We recruited 324 postmenopausal women with osteopenia in two trials with similar designs. Both studies were phase IIIb, randomized, double-blind, placebo-controlled, and endpoints included the detection of a treatment difference expressed as percent change from baseline in microstructure for 12 months. In the first study, designated group 1, we recruited 161 women younger than 55 years (mean age 53 years; range, 44–55 years) with BMI between 18 and 28 kg/m2. In the second study, designated group 2, we recruited 163 women older than 55 years (mean age 60 years; range, 55–76 years) with similar range in BMI. The entry criteria in both studies required a T-score at the lumbar spine (LS, L1–L4) of –1 to –2.5 SD, and at the total hip T-score ≥ –2.5 SD, or a LS T-score ≥ –2.5 SD and at the total hip T-score between –1 and –2.5 SD. Exclusion criteria were: previous or current use of glucocorticoids, anabolic steroids, estrogens, fluorides, bisphosphonates, or calcium or vitamin D supplementation. Women were randomized 2:1 to either oral risedronate (35 mg/week) or an oral placebo for 12 months. All received supplementation of 500 mg of calcium and 400 IU of vitamin D daily. Measurements of fasting serum carboxyterminal cross-linking telopeptide of type 1 bone collagen (CTX-1), serum amino-terminal propeptide of type 1 procollagen (P1NP), of areal bone mineral density (aBMD) at LS and femoral neck and of distal radius and tibia microstructure by high-resolution peripheral quantitative computerized tomography (HR-pQCT) were made at baseline and at 12 months.
Assessment of bone microstructure
Microstructure at the distal radius and tibia was determined using HR-pQCT (isotropic voxel size of 82 µm; XtremeCT; Scanco Medical AG, Brüttisellen, Switzerland). Quality control was monitored by daily scans of phantoms (rods of hydroxyapatite [HA] of 0, 100, 200, 400, and 800 mg HA/cm3 in a soft tissue equivalent resin; QRM, Moehrendorf, Germany). Images were analyzed by StrAx1.0, a new algorithm that segments bone from background and bone into its compact-appearing cortex, outer and inner transitional zones, and trabecular compartment without thresholding (StrAxCorp, Melbourne, Australia). The 40 most proximal cross-sectional slices were analyzed using 3600 radial attenuation profile curves around each slice. Most voxels contain both mineralized bone matrix and void volumes; the proportions of each were quantified using an interpolation function derived from two referents. Empty voxels, with attenuation equivalent to background, were assigned 0%. Voxels with attenuation produced by fully mineralized bone (density 1200 mg HA/cm3) were assigned 100%. The void volume of a voxel = 100–mineralized bone volume fraction. Total porosity is the average of the void volume fractions of all voxels. The precision errors for segmentation of bone compartments and quantification of porosity expressed as root mean square coefficients of variation ranged from 0.54% to 3.98% and were <1.5% for vBMD.
Data are presented as mean ± SD unless otherwise stated. Normality was tested using the Shapiro-Wilk procedure. Bone turnover markers were not normally distributed and were transformed using a square root function. The percentage change in microstructural parameters between baseline and 12-month follow-up was calculated in patients with interpretable HR-pQCT images at least at one site at baseline and 12 months. The differences in percentage change from baseline between risedronate and control groups were tested using unpaired t test. Analyses were performed using SPSS 20.0 (IBM, Armonk, NY, USA). A significance level of p < 0.05 was chosen.
Baseline characteristics were similar in women receiving placebo or risedronate in each study (Table 1). In group 1, 161 postmenopausal women with osteopenia were randomized to risedronate (n = 112) or placebo (n = 49); 135 (84%) completed the 12-month follow-up (Table 1). In group 2, 152 (93%) completed the 12-month follow-up. The main reasons for discontinuation were voluntary withdrawals, adverse events and loss to follow-up (Fig. 2). Remodeling markers decreased by ∼40% in younger, and ∼50% in older, postmenopausal women receiving risedronate (Table 2).
|Baseline values||Group 1 (<55 years old)||Group 2 (<55 years old)|
|Age (years)||53 ± 2||53 ± 2||61 ± 4||62 ± 6|
|Height (cm)||163 ± 6||163 ± 7||159 ± 6||159 ± 5|
|Weight (kg)||63 ± 7||61 ± 7||59 ± 8||60 ± 8|
|BMI (kg/m2)||24 ± 3||23 ± 3||23 ± 3||24 ± 3|
|CTX (ng/mL)||0.53 ± 0.17||0.57 ± 0.21||0.53 ± 0.18||0.52 ± 0.18|
|P1NP (µg/mL)||68.02 ± 23.07||69.04 ± 22.46||61.00 ± 22.03||60.98 ± 20.45|
|Total proximal femur aBMD (g/cm2)||0.79 ± 0.08||0.81 ± 0.13||0.78 ± 0.06||0.79 ± 0.06|
|Total proximal femur T-score||−1.34 ± 0.69||−1.15 ± 1.03||−1.31 ± 0.47||−1.23 ± 0.49|
|Lumbar spine aBMD (g/cm2)||0.92 ± 0.07||0.92 ± 0.06||0.85 ± 0.05||0.85 ± 0.06|
|Lumbar spine T-score||−1.75 ± 0.56||−1.74 ± 0.48||−1.77 ± 0.46||−1.78 ± 0.57|
|Total vBMD (mg/cm3)||420.27 ± 107.80||438.09 ± 98.89||451.21 ± 88.18||481.34 ± 104.99|
|Cortical vBMD (mg/cm3)||1101.36 ± 86.96||1103.53 ± 110.49||1108.10 ± 83.62||1115.91 ± 95.46|
|Trabecular vBMD (mg/cm3)||144.78 ± 45.59||142.54 ± 41.01||136.71 ± 52.93||136.99 ± 43.30|
|Compact-appearing cortex porosity (%)||28.02 ± 4.61||27.41 ± 5.21||29.01 ± 5.45||28.73 ± 5.76|
|Outer transitional zone porosity (%)||35.35 ± 3.21||34.98 ± 3.49||37.10 ± 3.86||37.02 ± 4.15|
|Inner transitional zone porosity (%)||70.88 ± 2.97||70.62 ± 2.76||71.05 ± 2.90||71.37 ± 2.47|
|Total vBMD (mg/cm3)||273.86 ± 41.02||287.97 ± 45.28||278.74 ± 38.06||280.98 ± 37.55|
|Cortical vBMD (mg/cm3)||637.40 ± 67.36||659.81 ± 69.17||613.54 ± 72.49||611.55 ± 68.13|
|Trabecular vBMD (mg/cm3)||125.93 ± 25.88||128.48 ± 34.67||130.53 ± 29.41||133.10 ± 26.00|
|Compact-appearing cortex porosity (%)||38.99 ± 6.66||36.57 ± 6.74||42.48 ± 8.48||42.31 ± 7.95|
|Outer transitional zone porosity (%)||41.54 ± 3.73||40.82 ± 4.19||46.31 ± 5.33||46.55 ± 5.26|
|Inner transitional zone porosity (%)||77.26 ± 2.39||76.81 ± 2.22||78.21 ± 2.32||78.40 ± 2.33|
|Group 1 (<55 years)||Between group p||Group 2 (>55 years)||Between group p|
|Control (n = 40)||Risedronate (n = 95)||Control (n = 52)||Risedronate (n = 100)|
|CTX (ng/mL)||−1.21 ± 27.10||−41.25 ± 59.90*||<0.0001||0.10 ± 24.37||−51.08 ± 29.19*||<0.0001|
|P1NP (µg/mL)||−2.44 ± 33.66||−43.60 ± 27.87*||<0.0001||−4.92 ± 20.17||−52.44 ± 21.12*||<0.0001|
|Total proximal femur aBMD (g/cm2)||−0.93 ± 1.78*||0.60 ± 1.80*||<0.0001||−0.11 ± 1.41||1.51 ± 1.66*||<0.0001|
|Lumbar spine aBMD (g/cm2)||−1.94 ± 2.33*||1.39 ± 2.89*||<0.0001||0.20 ± 2.60||3.18 ± 2.34*||<0.0001|
|Distal radiusa||n = 34||n = 67||n = 37||n = 63|
|Total vBMD (mg/cm3)||−1.89 ± 3.72*||−0.97 ± 3.10*||0.191||−0.40 ± 1.82||−0.31 ± 2.35||0.839|
|Cortical vBMD (mg/cm3)||−0.85 ± 2.54||−0.16 ± 2.33||0.177||0.05 ± 1.45||0.13 ± 1.45||0.787|
|Trabecular vBMD (mg/cm3)||−3.61 ± 8.21*||−1.60 ± 4.49*||0.190||−1.74 ± 5.65||−0.23 ± 5.73||0.207|
|Compact-appearing cortex porosity (%)||4.23 ± 9.19*||1.51 ± 7.33||0.111||−0.49 ± 3.74||−0.88 ± 3.44*||0.596|
|Outer transitional zone porosity (%)||1.67 ± 6.70||0.73 ± 5.38||0.443||−0.16 ± 2.96||−0.11 ± 2.69||0.941|
|Inner transitional zone porosity (%)||0.07 ± 1.47||−0.03 ± 1.27||0.727||−0.03 ± 1.15||0.00 ± 0.99||0.871|
|Distal tibiaa||n = 40||n = 89||n = 47||n = 91|
|Total vBMD (mg/cm3)||−0.63 ± 2.81||−0.20 ± 2.78||0.424||0.13 ± 1.44||0.76 ± 1.70***||0.033|
|Cortical vBMD (mg/cm3)||−1.09 ± 2.41*||−0.46 ± 2.39||0.171||−0.33 ± 1.75||0.50 ± 1.68*||0.007|
|Trabecular vBMD (mg/cm3)||−0.95 ± 3.69||0.24 ± 4.92||0.176||−0.09 ± 2.27||0.40 ± 1.51*||0.134|
|Compact-appearing cortex porosity (%)||3.44 ± 5.93**||1.36 ± 5.68*||0.060||0.76 ± 3.03||−0.75 ± 2.88*||0.005|
|Outer transitional zone porosity (%)||1.87 ± 4.81*||0.93 ± 4.74||0.302||0.16 ± 2.40||−0.66 ± 2.42*||0.060|
|Inner transitional zone porosity (%)||0.24 ± 0.87||0.10 ± 1.08||0.480||−0.04 ± 0.80||−0.30 ± 0.67***||0.050|
Effect of risedronate at the distal radius
Group 1 (younger postmenopausal women)
Distal radius total vBMD (mean ± SEM) decreased by 1.9% ± 0.6% (p = 0.005) in controls and by 1.0% ± 0.4% (p = 0.013) in risedronate-treated women. Compact-appearing cortical porosity increased by 4.2% ± 1.6% (p = 0.011) in controls and by 1.5% ± 0.9% (p = NS) in risedronate-treated women. Trabecular vBMD decreased by 3.6% ± 1.4% (p = 0.015) in controls and by 1.6% ± 0.6% (p = 0.005) in risedronate-treated women (Fig. 3A, B). Trends were similar for the porosity of the outer and inner transitional zones but none achieved statistical significance. No between-group difference reached statistical significance.
Group 2 (older postmenopausal women)
Trait changes in controls and risedronate-treated women did not achieve significance apart from a reduction in compact-appearing cortical porosity in the risedronate-treated women (0.9% ± 0.4%, p = 0.047) (Fig. 3C, D).
Effect of risedronate at the distal tibia
Group 1 (younger postmenopausal women)
In both control and risedronate groups, changes in total vBMD did not achieve statistical significance. Compact-appearing cortical porosity increased by 3.4% ± 0.9% (p = 0.001) in controls and increased by 1.4% ± 0.6% (p = 0.02) in risedronate-treated women. Trabecular vBMD remained unchanged in both groups (Fig. 4A, B).
Group 2 (older postmenopausal women)
Total vBMD remained unchanged in controls and increased by 0.8% ± 0.2% (p < 0.0001) in risedronate treated women (Fig. 4C). Compact-appearing cortical porosity remained unchanged in controls and decreased by 0.8% ± 0.3% (p = 0.016) in risedronate-treated women. Porosity of the outer and inner transitional zones remained unchanged in controls and decreased, by 0.7% ± 0.3% (p = 0.011) and 0.3% ± 0.1% (p < 0.0001) in risedronate-treated women, respectively. Trabecular vBMD remained unchanged in controls and increased by 0.4% ± 0.2% (p = 0.013) in risedronate-treated women (Fig. 4D). Between-group differences were significant for the changes in total vBMD (p = 0.033), compact-appearing cortex porosity (p = 0.005), and inner transitional zone porosity (p = 0.05) in this age group alone.
Risedronate suppressed bone remodeling markers by ∼50% in younger and older postmenopausal women. In younger postmenopausal women, risedronate blunted the increase in cortical porosity by ∼60% and blunted the decrease in trabecular vBMD by ∼70% seen at the distal radius in younger postmenopausal controls. In older postmenopausal women, risedronate modestly decreased porosity and tended to increase trabecular vBMD. Similar trends were observed at the distal tibia.
Bone loss during menopause and advancing age is the result of three abnormalities in bone remodeling. Two occur at the cellular level of the BMU; the volume of bone resorbed by osteoclasts of each BMU increases and the volume of bone deposited by osteoblasts of each BMU decreases.[12, 13, 21] This negative BMU balance is responsible for bone loss and structural decay, but the negative balance is only ∼5%; 95% of the bone removed by osteoclasts of a BMU is replaced by the osteoblasts of that BMU, so the third factor, the driving force responsible for bone loss and structural decay is the surface extent of bone remodeling, both its rate and whether or not it has been acutely perturbed.[7, 22-26]
Antiresorptive agents reduce bone loss mainly by reducing the number of remodeling sites eroding bone. However, the net effect depends on the number of remodeling sites excavated earlier that are now concurrently refilling. Risedronate administered in younger postmenopausal women reduced the number of excavated sites, but these probably still exceeded the fewer sites generated earlier now refilling, so net porosity increased, but ∼60% less than the increase in porosity in controls. This increase in porosity in treated women was not significant relative to baseline porosity at the distal radius. The deterioration in trabecular vBMD in controls was also reduced or prevented by risedronate. In older postmenopausal women, in whom remodeling was likely have returned to steady-state, risedronate reduced remodeling by ∼50% as in early menopause, but we suggest that now there are fewer sites excavated than the larger number (excavated in early menopause) now refilling, so there is a net reduction in porosity. This reduction was modest, making changes more difficult to detect in older women.
We suggest that the accelerated bone loss in early menopause is primarily due to perturbation of surface level remodeling because the rapid rise in the number of BMUs excavating cavities exceeds the fewer sites (excavated before menopause) now refilling. This perturbation from a slow steady state before menopause to a rapid steady-state is accompanied by the appearance of a negative BMU balance[5, 15]; each BMU removes more bone than each subsequently deposits. However, in early menopause, the rapid decline in vBMD is mainly due to the perturbation in remodeling as increased numbers of BMUs upon the endosteal surface erode bone. We suggest that this contributes relatively more to bone loss than the negative BMU balance produced by each BMU. We also suggest that this is why changes are larger in younger than older postmenopausal women.
The increase in the number of cavities excavated upon Haversian canal surfaces produces a rise in porosity, but this porosity is largely reversible and most of this void volume refills as the slower formation phase of a remodeling cycle reaches completion, refilling the excavated cavity to about 95% of the volume of bone removed. In later menopause, when steady-state remodeling is restored at a higher remodeling rate, bone loss slows because the contribution of the perturbation of remodeling at the surface level is gone. Now bone loss depends only on the negative BMU balance, which persists and may worsen as bone formation by each BMU continues to decline, and the rate of steady-state remodeling, which remains elevated but decreases upon trabecular surfaces (because they disappear as trabeculae are lost) and continues upon the intracortical and endocortical surfaces as they increase.
Bone loss occurs from both cortical and trabecular compartments. Cortical bone is 80% of the skeleton and has a low surface area/bone matrix volume configuration so there are fewer remodeling sites per unit cortical bone matrix volume. Trabecular bone loss is greater as a percentage of its peak mass, partly because its high surface area/bone matrix volume configuration facilitates remodeling; there are more remodeling sites per unit trabecular bone matrix volume (each with their negative bone balance) eroding a smaller peak volume of bone. Nevertheless, the slower loss of the larger volume of cortical bone and the more rapid loss of the smaller volume of trabecular bone (20% of the skeleton) contribute similar amounts of bone in absolute terms in the first decade after menopause, whereas after age 60 to 65 years bone loss becomes predominantly cortical.
The antifracture efficacy of a treatment is related, at least partly, to its ability to decrease cortical porosity. Resistance to bending of cortical bone decreases to the seventh power of its apparent density, the inverse of porosity. Moreover, resistance to bending is a fourth-power function of the distance of a unit volume of bone from the neutral axis. Therefore, even a slightly lower porosity in compact-appearing cortex is likely to decrease fracture risk. Prevention of the appearance of these cavities and reversal of porosity is therapeutic because it reduces stress concentrators.
Remodeling, as reflected in the suppression of bone remodeling markers, was reduced, but not abolished, by risedronate. We suggest that this continued remodeling (despite treatment) is likely to originate within the intracortical compartment because bisphosphonates are adsorbed into superficial subendosteal bone matrix and may not access deeper peri-Haversian cortical bone matrix, which continues to be remodeled when initiated upon Haversian canal surfaces. If most bisphosphonate is adsorbed upon superficial matrix and fails to penetrate the deeper matrix, osteoclasts will not encounter or engulf matrix-containing bisphosphonate, so remodeling and bone loss will continue in the intracortical compartment. Trabecular bone has a plate-like structure that is more readily infiltrated by adsorbed bisphosphonate, and so it is engulfed by osteoclasts, which then stop resorption.
Risedronate may also modify the lifespan of osteoclasts, reducing the depth of resorption pits. In beagle dogs, there is evidence suggesting that risedronate increases the volume of bone deposited by the osteoblasts of the BMU. Even if the negative BMU balance was lessened, if not fully corrected, slow continued remodeling is likely to produce slow structural deterioration despite compliance with therapy.
A strength of the study was the use of a state-of-the-art imaging method in assessing bone microstructure in vivo. In addition, we applied a new image-processing algorithm allowing automatic and accurate segmentation of bone compartments and accurate measurement of porosity. A limitation of this study was its brevity. Increases in cortical porosity or decreases in trabecular vBMD in controls (who received calcium and vitamin D, which may slow remodeling and bone loss) were small, limiting our ability to detect benefits of risedronate, particularly in older postmenopausal women in whom steady-state remodeling was restored. This may partly explain the few between-group differences that reached statistical significance. Thus, longer follow-up is needed to establish the effects of treatment on microstructure.[3, 4] Nevertheless, the trends were consistent throughout, with approximate halving of small reductions in total or trabecular vBMD and blunting of modest increases in porosity in controls during this brief observation period. Another limitation is the lack of dating of menopause, which prevented us from examining changes in microstructure as a function of years since menopause.
In conclusion, the morphological effects of antiresorptive therapy partly depend on when it is administered. Risedronate is likely to reduce fracture risk, in part, by reducing cortical porosity by allowing partial refilling of existing pores and by reducing the appearance of new cortical porosity. These benefits were observed within the first year of treatment and so support the rapid and sustained efficacy in the early reduction in fracture risk reported using this agent.[36, 37]
RC has received consulting fees, lecture fees, grant support from Amgen, Bioiberica, MSD, Novartis, Pfizer, Roche, and UCB. DF has received consulting fees, lecture fees, and grant support from Amgen, Chugai, GE, GlaxoSmithKline, Lilly, MSD, Novartis, Nycomed, Roche, Servier, TEVA, and Warner Chilcott. EM has received consulting fees, lecture fees, and grant support from Novo Nordisk and preglem. JR has received grant support from Warner Chilcott. TT has received consulting fees, lecture fees, and grant support from Abbvie, Amgen, Chugai, Gibaud, GlaxoSmithKline, Ipsen, Lilly, MSD, Novartis, and UCB. AC has received consulting fees, lecture fees, and grant support from Amgen, Lilly, MSD, Novartis, and Warner Chilcott. RZ has received consulting fees, lecture fees, and grant support from Amgen, MSD, and Servier. ES has received consulting fees, lecture fees, and grant support from Amgen, MSD, Novartis, Sanofi Aventis, and Servier. RR has received consulting fees, lecture fees, and grant support from Amgen, MSD, GSK, Servier, Danone, and Takeda.
The study was funded by Warner Chilcott (US) LLC and Sanofi. The funding sources participated in the design, conduct, and analysis of the study. We thank Miriam Annett (Warner Chilcott) and Dietrich Wenderoth (Warner Chilcott) for statistical and technical support. The authors are responsible for the content, editorial decisions, and opinions expressed in the article.
Authors' roles: Drafting manuscript: YB, ES, RC, and RR. Revising manuscript content: YB, RC, AC, DF, ML, EM, JR, TT, JZ, OB, AGZ, RZ, ES, and RR. Approving final version of manuscript: YB, RC, AC, DF, ML, EM JR, TT, JZ, OB, AGZ, RZ, ES, and RR. ES takes responsibility for the integrity of the data analysis.