Risedronate use to attenuate bone loss following sleeve gastrectomy: Results from a pilot randomized controlled trial

Summary

The purpose of this study was to explore the efficacy of 150 mg once monthly oral risedronate use in the prevention of sleeve gastrectomy (SG) associated bone loss. Twenty-four SG patients (56 ± 7 years, 83% female, 21% black) were randomized to risedronate or placebo for 6 months, with an optional 12-month assessment. Outcome measures included 6 (n = 21) and 12 (n = 14) month change in dual energy x-ray absorptiometry-acquired regional areal bone mineral density (aBMD). Six-month treatment effect estimates [mean (95% CI)] revealed significant between group aBMD differences at the femoral neck [risedronate: +0.013 g/cm² (−0.021, 0.046) vs. placebo: −0.041 g/cm² (−0.067, −0.015)] and lumbar spine [risedronate: +0.028 g/cm² (−0.006, 0.063) vs. placebo: −0.029 g/cm² (−0.054, −0.004)]; both p ≤ 0.02. When followed postoperatively to 12 months, differential aBMD treatment effects were observed at the total hip [risedronate: −0.035 g/cm² (−0.061, −0.009) vs. placebo: −0.072 g/cm² (−0.091, −0.052)] and lumbar spine [risedronate: +0.012 g/cm² (−0.038, 0.063) vs. placebo: −0.052 g/cm² (−0.087, −0.017)]; both p < 0.05. Preliminary treatment effect estimates signal 6 months of risedronate use may be efficacious in reducing aBMD loss at the axial skeleton post-SG, with benefit largely maintained throughout the 1-year postoperative period. Confirmatory data from an adequately powered trial are needed.

1 |. INTRODUCTION

Despite well-recognized improvements in obesity-related comorbidities, increasing evidence implicates surgical weight loss in the onset of adverse skeletal health outcomes. Sleeve gastrectomy (SG) is currently the most commonly performed bariatric surgery,¹ which is accompanied by an increase in bone turnover² and a 3%–7% loss of areal bone mineral density (aBMD) at the axial skeleton.³ Encouragingly, recent data suggest that fracture risk associated with SG is at least no greater and may potentially be less^4,5 than other bariatric procedures,^6,7 such as Roux-En Y gastric bypass (RYGB). Exploration of bone-sparing strategies for this population, however, still appears prudent and may be especially relevant for older patients who are increasingly undergoing bariatric surgery⁸ and for whom the absolute risk of fracture is elevated prior to surgery.

Bisphosphonate therapy reduces osteoporotic fracture risk⁹ and may be effective in minimizing bone loss associated with surgical weight loss. Once-monthly oral risedronate is a safe and efficacious bisphosphonate^10,11 with a favourable gastrointestinal (GI) profile,¹² which acts by inhibiting the activity of osteoclast cells,¹³ thereby decreasing the rate of bone resorption. Because surgical weight loss is associated with significantly increased bone resorption,^2,14 bisphosphonate use may mitigate concomitant bone loss in this patient population; yet, this hypothesis has not been formally tested.

To address this knowledge gap, the main objective of the pilot Weight Loss with Risedronate for Bone Health (WE RISE) randomized controlled trial (RCT) was to begin to examine the efficacy of orally administered 150 mg once monthly risedronate versus placebo in the prevention of surgical weight loss associated bone loss among 24 SG patients followed for up to 12 months. Study design and 6-month feasibility outcome data are published¹⁵; herein, we report preliminary treatment effects on regional dual-energy x-ray absorptiometry (DXA)-acquired aBMD and trabecular bone score (TBS), as well as biomarkers of bone turnover [procollagen type 1 N-terminal propeptide (P1NP) and C-terminal telopeptide of type 1 collagen (CTX)]. We hypothesize that SG patients randomized to risedronate will experience less bone loss via reduced bone resorption as compared with individuals randomized to placebo at the end of active treatment period (6 months; primary end point), and this effect will be sustained throughout the duration of the follow-up period (12 months; secondary end point).

2 |. MATERIALS AND METHODS

2.1 |. Study design and patient population

Details regarding the study design and patient population are previously published.¹⁵ Briefly, this pilot, double-blinded, placebo-controlled, RCT (WE RISE: NCT03411902) involved 24 SG patients who were randomized to take 150 mg once monthly oral risedronate or placebo capsules over a 6-month period. Official study assessments occurred at baseline and 6 months, with all participants invited to partake in an optional 12-month assessment visit (occurring within 1.5 months of the surgical anniversary date). All eligible and interested participants read and signed an IRB-approved (protocol #48310) informed consent document prior to enrolment.

Potential participants were recruited from the Wake Forest Baptist Health Weight Management Clinic (WMC) in Winston Salem, NC and had to meet the following eligibility criteria, including: 1) scheduled SG surgery; 2) age 40–79 years; 3) weight < 204 kg (DXA scanner limit); 4) no chronic anti-reflux treatment, history of medical disorders known to affect bone metabolism, including osteoporosis (either self-reported or detected at the baseline assessment visit), use of bone-active medications, or a known allergy to risedronate; and 5) medical review and clearance by the study physician (J.D.A.), including medical examinations and clinical lab work [e.g. normal serum calcium or absence of significant renal dysfunction; estimated glomerular filtration rate (eGFR) <30 ml/min/1.73 m²] to identify any health concerns that would preclude individuals from safe participation in the study. (Of note, normal vitamin D status was not a criterion for entry into the study because normal vitamin D status or supplementation was required for surgical clearance.) As previously described,¹⁵ all participants followed the American Society of Metabolic and Bariatric Surgery (ASMBS) recommendations for the perioperative nutrition, metabolic, and non-surgical support of bariatric patients,¹⁶ including recommended consumption of at least 3000 IU/day of vitamin D (if serum levels were below 30 ng/ml), 1200–1500 mg/day of calcium and 90–120 mg/day of vitamin K. Additionally, all participants were asked to establish an exercise routine pre-operatively that was done for at least 30 min, 3-to-5 days per week, with exercise recommendations post-surgery including daily walking and strength training beginning after their 30-day follow-up visit.

2.2 |. Intervention descriptions

Six over-encapsulated 150 mg risedronate tablets or identical placebo capsules (filled with bulk powder) were dispensed to each participant from the Wake Forest Investigational Drug Service Pharmacy post-randomization. Participants were instructed to take each capsule orally and once monthly for 6 months (six doses). The first dose was taken 3–7 days prior to surgery, with the remaining five doses taken at monthly intervals thereafter. Among oral bisphosphonates, risedronate was specifically selected because of its monthly dosing frequency, tolerability and demonstrated efficacy in reducing vertebral, non-vertebral and hip fracture incidence.^10–12 Participants were instructed to follow label instructions (i.e., take medication with 177–236 ml of plain water; remain upright for the next 30 min; avoid eating, taking vitamins/mineral supplements, or antacids at the same time) and were queried by intervention staff at monthly intervals as to protocol compliance and onset of potential adverse events (AEs) during the active treatment period.

Detailed feasibility and safety data are previously reported.¹⁵ Briefly, among completers (n = 21), the average number of pills taken (out of six, total) was 5.9 ± 0.4 and 6.0 ± 0.0 in the risedronate and placebo groups, respectively (p = 0.21). Out of 134 contacts, five AEs were reported (3.7% AE rate), with two occurring in the risedronate group [scalp rash (mild, not related) and nausea (mild, definitely related)] and three occurring in the placebo group [headache (mild, not related), heartburn (moderate, not related), and nausea (mild, not related)]; p = 0.84.

2.3 |. Outcome measures and covariates

2.3.1 |. DXA-acquired bone metrics

All DXA-acquired metrics were assessed at baseline (n = 24), six (n = 21), and 12 (n = 14) months. aBMD of the total hip, femoral neck, lumbar spine and ultradistal radius; and TBS of the lumbar spine were determined by DXA (iDXA, GE Medical Systems, Madison, WI). Participants were categorized as osteopenic if any hip, spine, or distal radius T-score was between −2.5 and −1, respectively.¹⁷ All scans were performed and analysed in accordance with national recommendations by an International Society for Clinical Densitometry (ISCD) trained DXA technologist, as done previously,¹⁸ with CVs from repeated measurements at our institution of less than 2.0% and 4% for all axial BMD and TBS outcomes, respectively.

2.3.2 |. Biomarkers of bone turnover

Blood samples were obtained at baseline (n = 23), 6 (n = 21), and 12 (n = 14) months via venipuncture after an overnight fast (of ≥10 h) and abstinence from physical activity for the previous 24 h. After centrifugation for 20 min at 4°C, aliquots of plasma/serum were stored at −70°C until batch analysis at study close. Representative biomarkers of bone turnover, procollagen type 1 N-terminal propeptide (P1NP; formation) and C-terminal telopeptide of type 1 collagen (CTX; resorption), were selected per international recommendation¹⁹ and measured using commercially available ELISAs [Novus Biologicals (P1NP) and Immunodiagnostic Systems (CTX)], as done previously.¹⁸ Inter- and intra-assay variability for P1NP and CTX are 5.1% and 5.1% and 7.7% and 2.2%, respectively.

2.3.3 |. Covariates

Self-reported demographic information (i.e. age, sex, race/ethnicity, education level) was assessed at baseline. Participants were queried at baseline, 6 and 12 months on medical information including medical history (including any prior fracture, excluding finger/toe fractures) and medication use to assess 10 year major osteoporotic and hip fracture risk using the FRAX^® tool (version 3.10).²⁰ Height was assessed without shoes to the nearest 0.25 cm using a stadiometer (Health O Meter^® Portrod) and body weight was measured to the nearest 0.05 kg using a calibrated and certified digital scale (Health O Meter^® Professional 349KLX), with excess weight defined as the amount of weight above the weight equivalent to a BMI of 25 kg/m² for each individual. Lastly, pertinent clinical blood work [including serum calcium, creatinine, eGFR and parathyroid hormone (PTH)] was completed at baseline following standard procedures at a clinical laboratory (Labcorp).

2.4 |. Statistical analyses

Baseline characteristics were summarized overall and by randomized treatment group as means and standard deviations (mean ± SD) for continuous variables or counts and percentages [n (%)] for discrete variables. Baseline characteristics of participants retained at 6 and 12 months were compared using t-tests for continuous measures and chi-square tests for discrete characteristics to determine whether characteristics of the sample changed due to attrition. Treatment effects of risedronate versus placebo were estimated using a mixed linear model fit with treatment assignment, visit, and treatment by visit interaction, adjusted for baseline characteristics of each outcome and assuming an unstructured covariance. Groups were compared using contrast statements at 6 and 12 months, with change over 6 months being the primary comparison. Biomarkers of bone turnover, P1NP and CTX, were analysed on the log scale, with model estimates exponentiated back to the original scale and reported as percentage change. Plots for weight and DXA variables were generated using model-adjusted outcomes as percent changes from baseline, and plots of biomarkers used changes in the log scale to maintain confidence interval symmetry. Two sensitivity analyses were conducted for aBMD outcomes: firstly, we added weight change as an additional covariate to the model to determine whether the degree of weight loss affected treatment effects; secondly, we analysed a sub-sample of only postmenopausal women (n = 15) to ensure estimates did not differ substantially compared to men and pre/peri-menopausal women. All statistical analyses were performed using SAS v9.4 (SAS Institute, Cary, NC) with significance based on a Type I error rate of 0.05. Because the study was designed as a pilot to determine the feasibility and gather evidence for future work, analyses were not adjusted for multiple comparisons and findings are not considered confirmatory.

3 |. RESULTS

3.1 |. Sample characteristics

Baseline descriptive and clinical characteristics of randomized participants (n = 11 risedronate; n = 13 placebo) were previously reported (see abbreviated Table 1¹⁵). Briefly, average age of the study sample was 56 ± 7 years, 83% of the study participants were female (63% were postmenopausal) and 21% were Black. Baseline BMI was 44.7 ± 6.3 kg/m², with the risedronate group having a significantly higher BMI compared with the placebo group (48.1 ± 7.2 kg/m² vs. 41.9 ± 3.8 kg/m², respectively). No difference in previous fracture was noted between groups (risedronate n = 5, placebo n = 6) and no participants presented with FRAX^® based major osteoporotic or hip fracture risk estimates above treatment thresholds; however, three individuals (12.5%, all in the risedronate group) were categorized with osteopenia at baseline. With the exception of weight (p = 0.03) and osteopenia prevalence (p = 0.04), all other covariates presented in Table 1 were balanced across groups at baseline.

TABLE 1.

Baseline characteristics of study sample, overall and by treatment group

Baseline variable	Overall	Risedronate	Placebo
	n = 24	n = 11	n = 13
Age (years)	55.7 ± 6.7	53.8 ± 7.7	57.3 ± 5.7
Female, n (%)	20.0 (83.0)	9.0 (81.8)	11.0 (84.6)
Postmenopausal status, n (%)	15.0 (62.5)	6.0 (54.5)	9.0 (69.2)
Black, n (%)	5.0 (20.8)	3.0 (27.3)	2.0 (15.4)
Education, n (%)
High school degree or less	4.0 (16.7)	3.0 (27.3)	1.0 (7.7)
Some college	12.0 (50.0)	4.0 (36.4)	8.0 (61.5)
College+	8.0 (33.3)	4.0 (36.4)	4.0 (30.8)
Weight (kg)	122.1 ± 22.6	132.9 ± 25.3	113.0 ± 15.7
Excess weight (kg)	54.0 ± 19.0	63.9 ± 21.5	45.7 ± 11.8
Body mass index (kg/m2)	44.7 ± 6.3	48.1 ± 7.2	41.9 ± 3.8
Clinical bone categorization, n (%)
Normal	21.0 (87.5)	8.0 (72.7)	13.0 (100.0)
Osteopenic	3.0 (12.5)	3 (27.3)	0 (0)
Calcium (mg/dl)	9.4 ± 0.43	9.5 ± 0.4	9.3 ± 0.3
Creatinine (mg/dl)	0.83 ± 0.17	0.85 ± 0.20	0.81 ± 0.15
eGFR (% >60 ml/min/1.73 m2)	22 (92)	10 (91)	12 (92)
Parathyroid Hormone (pg/ml)	50.3 ± 17.0	52.2 ± 18.5	48.8 ± 16.5

Note: Continuous data are presented as mean ± SD and categorical variables are presented as n (%). Abbreviations: eGFR, estimated glomerular filtration rate; dl, decilitre; kg, kilograms; m, meter; mg, milligram; min, minute; ml, millilitre; n, sample size; pg, picogram.

Three individuals, all in risedronate group, dropped out during the first 6 months, leaving 21 individuals with complete baseline and 6-month follow-up data. A study Consolidated Standards of Reporting Trials (CON-SORT) diagram showcasing participant flow through the first 6 months of the study is published,¹⁵ with a diagram including the 12-month time point included as a supplementary file (see Figure S1). Compared with those who dropped out, those with 6-month follow-up data were similar across all baseline characteristics, except starting weight (p = 0.02), which was greater among those who dropped out (150.2 ± 29.0 kg) versus those who were retained in the 6-month sample (118.1 ± 19.2 kg). A total of 14 individuals elected to participate in the optional 12-month follow-up assessment visit, with n = 5 and n = 9 in the risedronate and placebo groups, respectively. Participants with 12-month follow-up data were significantly older (58.2 ± 6.7 vs. 52.3 ± 5.4 years) and presented with a lower baseline BMI (42.4 ± 5.9 vs. 48.0 ± 5.6 kg/m²) than those who declined to participate in the 12-month follow-up visit.

3.2 |. Six-month treatment effects on DXA-acquired bone measures and biomarkers of bone turnover

Change in weight/BMI and treatment effects on DXA-acquired bone metrics are presented in Table 2. Over the first 6 months period, a marginal, albeit non-significant difference was observed for weight change, with the risedronate group presenting with less weight loss than the placebo group [mean (95% CI): −13.3% (−16.4, −10.2) vs. −17.1% (−19.3, −14.8]; p = 0.06. Regarding aBMD, significant between group differences in favour of risedronate were observed at the femoral neck [+0.013 g/cm² (−0.021, 0.046) vs. −0.041 g/cm² (−0.067, −0.015)] and lumbar spine [+0.028 g/cm² (−0.006, 0.063) vs. −0.029 g/cm² (−0.054, −0.004)]; both p ≤ 0.02. No differences in TBS were observed. Sensitivity analyses adjusting aBMD treatment effect estimates for weight change did not materially affect study findings (see Table S1). In contrast, previously significant aBMD treatment effects were attenuated when limiting the sample to postmenopausal women, only (n = 15); however, the direction of change was similar to full sample modelling results (Table S2). Finally, no difference in P1NP was observed by group or time [risedronate: −4% (−18, 13) vs. placebo: −4% (−15, 9)]; p = 0.98; however, elevation in CTX was significantly blunted in the risedronate group [+68% (38, 104)] compared with placebo [+175% (135, 222)]; p < 0.001.

TABLE 2.

Six- and 12-month treatment effect estimates on weight, body mass index and DXA-acquired bone metrics

Outcome Variable	Baseline	Risedronate			Placebo			Absolute Risedronate - Placebo	p
		Absolute Value	Change	% Δ	Absolute Value	Change	% Δ
Weight (kg)	122.1 ± 22.6
6 months		99.3 (95.6, 103.1)	−16.3 (−20.0, −12.5)	−13.3	94.7 (91.9, 97.5)	−20.9 (−23.7, −18.1)	−17.1	4.6 (−0.2, 9.4)	0.057
12 months		101.7 (95.4, 108.0)	−13.9 (−20.2, −7.6)	−11.4	95.6 (90.9, 100.3)	−20.0 (−24.6, −15.3)	−16.4	6.1 (−1.8, 14.0)	0.121
Body mass index (kg/m2)	44.7 ± 6.3
6 months		37.2 (35.8, 38.7)	−6.1 (−7.5, −4.6)	−13.6	35.7 (34.7, 36.7)	−7.6 (−8.6, −6.6)	−17.0	1.5 (−0.3, 3.4)	0.095
12 months		38.4 (35.9, 40.8)	−4.9 (−7.4, −2.5)	−11.0	36.1 (34.3, 37.9)	−7.2 (−9.0, −5.4)	−16.2	2.3 (−0.7, 5.4)	0.130
Total Hip aBMD (g/cm2)	1.063 ± 0.132
6 months		1.003 (0.981, 1.024)	−0.028 (−0.049, −0.006)	−2.6	0.984 (0.967, 1.001)	−0.047 (−0.063, −0.030)	−4.4	0.019 (−0.008, 0.046)	0.163
12 months		0.996 (0.970, 1.022)	−0.035 (−0.061, −0.009)	−3.3	0.959 (0.939, 0.979)	−0.072 (−0.091, −0.052)	−6.7	0.037 (0.004, 0.069)	0.030
Femoral neck aBMD (g/cm2)	0.976 ± 0.168
6 months		0.956 (0.923, 0.990)	0.013 (−0.021, 0.046)	1.3	0.903 (0.877, 0.929)	−0.041 (−0.067, −0.015)	−4.2	0.054 (0.011, 0.096)	0.016
12 months		0.916 (0.887, 0.945)	−0.028 (−0.057, 0.002)	−2.8	0.901 (0.879, 0.923)	−0.043 (−0.065, −0.021)	−4.4	0.015 (−0.022, 0.052)	0.401
Lumbar spine aBMD (g/cm2)	1.311 ± 0.184
6 months		1.333 (1.299, 1.367)	0.028 (−0.006, 0.063)	2.2	1.275 (1.251, 1.300)	−0.029 (−0.054, −0.004)	−2.2	0.058 (0.014, 0.102)	0.013
12 months		1.317 (1.266, 1.368)	0.012 (−0.038, 0.063)	0.9	1.253 (1.218, 1.287)	−0.052 (−0.087, −0.017)	−4.0	0.065 (0.002, 0.127)	0.045
Trabecular bone score	1.497 ± 0.119
6 months		1.478 (1.413, 1.543)	−0.001 (−0.067, 0.064)	−0.1	1.485 (1.436, 1.533)	0.006 (−0.043, 0.054)	0.4	−0.007 (−0.090, 0.076)	0.860
12 months		1.456 (1.393, 1.519)	−0.023 (−0.086, 0.039)	−1.6	1.436 (1.390, 1.482)	−0.043 (−0.089, 0.003)	−2.9	0.020 (−0.060, 0.099)	0.609
UD Radius aBMD (g/cm2)	0.493 ± 0.066
6 months		0.483 (0.466, 0.500)	0.000 (−0.017, 0.017)	0.1	0.471 (0.457, 0.484)	−0.013 (−0.026, 0.001)	−2.5	0.013 (−0.009, 0.035)	0.238
12 months		0.468 (0.440, 0.496)	−0.015 (−0.043, 0.013)	−3.1	0.467 (0.446, 0.488)	−0.016 (−0.037, 0.005)	−3.3	0.001 (−0.034, 0.036)	0.959

Note: Raw baseline values are presented in aggregate and as mean ± SD, with 6- and 12-month values presented as model adjusted means (95% Confidence intervals). Treatment effects were estimated using a mixed linear model fit with treatment assignment, visit, and treatment by visit interaction, adjusted for baseline characteristics of each outcome.

Abbreviations: aBMD, bone mineral density; g/cm², grams per centimetres squared; kg, kilogram; m, meter; UD, ultradistal; %, percent; Δ, change.

3.3 |. Exploratory 12-month treatment effects on DXA-acquired bone measures and biomarkers of bone turnover

Plots showcasing the 12-month time course of percentage change in weight, total hip aBMD, femoral neck aBMD, lumbar spine aBMD, and log P1NP and CTX are presented by group in Figure 1. Weight change largely plateaued from 6 to 12 months, with a net change of −13.9 kg (−11.4%) and −20.0 kg (−16.4%) observed in the risedronate and placebo groups, respectively, over the 12 months period (between group p = 0.12). Although attenuated from the 6-month time point, differential treatment effects in favour of risedronate were retained for lumbar spine aBMD [+0.012 g/cm² (−0.038, 0.063) vs. −0.052 g/cm² (−0.087, −0.017)] and CTX [+65% (31, 109) vs. +143% (102, 194)]; both p < 0.05. By 12 months, the previously significant difference at the femoral neck was attenuated; however, the treatment effect for total hip aBMD was augmented so that by 12 months, the risedronate group presented with significantly less loss than placebo [−0.035 g/cm² (−0.061, −0.009) vs. −0.072 g/cm² (−0.091, −0.052)]; p = 0.03. As with 6-month data, no difference in P1NP was observed by group or over time.

FIGURE 1.

Twelve-month time course of percent change in weight (A), aBMD (B–D), and biomarkers of bone turnover (E,F) by group. Baseline values are presented in aggregate, with 6- and 12-month values presented as model-adjusted means (95% confidence intervals) and estimated using a mixed linear model fit with treatment assignment, visit, and treatment by visit interaction, adjusted for baseline characteristics of each outcome. *p < 0.05 for risedronate versus placebo at 6 months, ⁺p < 0.05 for risedronate versus placebo at 12 months; ^#p < 0.01 for risedronate versus placebo at 6 months

4 |. DISCUSSION

The main objective of the Weight Loss with Risedronate pilot RCT was to evaluate the feasibility of using a monthly, oral bisphosphonate to attenuate bone loss in SG patients. Herein, we report preliminary treatment effect estimates suggesting that 6 months of risedronate use may be efficacious in mitigating bone resorption and associated aBMD loss at the axial skeleton post-SG, with benefit largely maintained throughout the 1-year postoperative period. Findings extend previously reported feasibility outcome data¹⁵; and when taken together, support the future study of risedronate use for the prophylactic management of bone loss secondary to SG in a definitely powered RCT.

Laparoscopic SG is a relatively new (originating in 1999) and increasingly popular,¹ bariatric procedure in which the stomach is reduced to about 20%–25% of its original size in order to combat severe obesity. While generating less bone loss than more malabsorptive procedures [i.e. RYGB or biliopancreatic diversion with duodenal switch (BPD-DS)], longitudinal studies conducted over the past decade uniformly show that SG yields 3%–7% bone loss at the axial skeleton in the 6–24 months post-surgery.³ Bone loss experienced by participants in the placebo arm of the present study align with these observations, with risedronate use effectively halving 12 month aBMD loss at the hip (−3.3% vs. −6.7%) and maintaining bone mass at the lumbar spine (+0.9% vs. −4.4%) regions. While there is presently insufficient data to confidently characterize fracture incidence secondary to SG-associated bone loss, literature focused on metabolic bariatric procedures (of which, SG is one) report a modest association (i.e. 1.3- to 2.3-fold increase) with incident fracture in the 2–5 years post-surgery.^6,7 Based on estimates from a recent meta-regression analysis of individual patient data from multiple RCTs of osteoporosis medication use where incident fracture was assessed,²¹ we postulate that the degree of aBMD preservation observed here has clinical relevance. However, we implore future research endeavours to prioritize longitudinal studies with fracture endpoints in this specific population – particularly among older SG patients who are already at elevated risk of fracture⁸ – to better understand the clinical utility of our preliminary findings.

The clinical effects of bisphosphonates in general, and risedronate in particular,^10,11 on bone are well described. While this study is the first RCT, to our knowledge, to utilize a pharmacotherapeutic intervention to mitigate bariatric surgery associated bone loss, several lines of reasoning suggest that bisphosphonates should be effective. Firstly, surgery-induced weight loss is inherently catabolic to the skeleton,¹⁴ with high-turnover bone loss occurring in both cortical and trabecular regions.^22–24 Secondly, clinical studies in postmenopausal osteoporotic women have repeatedly confirmed the efficacy of bisphosphonates in attenuating bone resorption,¹³ preserving trabecular and cortical structure²⁵ and ultimately reducing the risk of osteoporotic fracture.²⁶ Risedronate specifically has demonstrated efficacy in reducing vertebral, non-vertebral, and hip fracture incidence.^10,11 Finally, bisphosphonates are already used to manage low bone mass secondary to analogous BMD-reducing conditions, such as prolonged glucocorticoid use,²⁷ spinal cord injury,²⁸ and cancer treatment²⁹ – including gastric cancer patients post-gastrectomy.³⁰ Thus, we are encouraged, though not overly surprised, by the beneficial signal attributed to risedronate use on bone outcomes in our pilot trial. Interestingly, the blunted rise in CTX in the risedronate group may indicate that the surgical weight loss associated bone loss is resulting – at least in part – from osteocytic osteolysis,³¹ as bisphosphonates are very effective in diminishing osteoclast activity.¹³ We implore future mechanistic studies to expand on this finding and help clarify the biology of surgical weight loss associated bone loss, particularly among older SG patients.

The potential for ulceration and hypocalcemia is a legitimate concern regarding oral bisphosphonate use in bariatric surgery patients, albeit less for risedronate use in SG patients than with more potent antiresorptive agents and malabsorptive procedures. Encouragingly, we did not encounter these serious adverse events our study¹⁵; however, as these are rare events, pilot safety data cannot be considered definitive. Provided these concerns are obviated, however, current clinical practice guidelines do support the consideration of antiresorptive agents in bariatric surgery patients with osteoporosis,¹⁶ with data presented herein providing preliminary evidence to consider extending this recommendation to the prophylactic management of surgical bone loss among non-osteoporotic, older SG patients – though certainly more data are needed to confidently inform clinical practice. In addition, it is worth noting that our exploratory 12-month data suggest that the bone-sparing effect of risedronate appears to persist after the cessation of 6 months of treatment; thus, transient bisphosphonate use during the most active weight loss period (i.e. initial 6 months) may yield long term bone health protection, while minimizing exposure and potential for adverse events in this clinical scenario.³²

Strengths of the current study include the double-blind RCT design and excellent 6-month participant retention and adherence,¹⁵ necessary for robust assessment of treatment efficacy. It should be noted that loss to follow-up was exclusive to the risedronate group (drop out n = 3 at 6 months; did not elect to participate in optional assessment n = 5 at 12 months), which could bias our findings. However, 88% retention at 6 months is quite good³³; and, we are careful to weight 6 month findings more heavily than 12 months findings. Selection of the patient population (SG is currently the most popular bariatric procedure¹), active treatment (oral risedronate has a lower dosing frequency and incidence of gastric ulcers than alendronate¹²), and outcome measures (DXA-acquired aBMD is the primary metric to assess osteoporosis and is considered a surrogate outcome for fracture in RCTs²¹) were carefully and intentionally selected due to their high clinical relevance. That said, bone loss is more pronounced with malabsorptive bariatric procedures and other antiresorptive medications (such as ibandronate, denosumab, or intravenously administered zoledronic acid) or bone-active agents could be considered. Furthermore, it is well recognized that DXA-acquired aBMD does not fully explain fracture risk prediction,³⁴ is susceptible to undue influence of adipose tissue on aBMD accuracy and precision³⁵ and cannot evaluate elements of bone quality. TBS data from DXA does provide information regarding trabecular microarchitecture; however, it is currently validated for use in patients with a BMI between 15 and 37 kg/m². As the average baseline BMI of our study sample exceeded this upper bound, TBS results presented herein should be interpreted cautiously. Volumetric BMD (vBMD) obtained with quantitative computed tomography (QCT) is increasingly used to complement DXA-acquired aBMD estimates as less concern exists about obesity and weight loss-induced measurement error with three-dimensional imaging. QCT contributes additional information on bone strength and structure, thereby adding to fracture risk predictive power.³⁶ QCT bone metrics collected as a part of the WE RISE trial will be the subject of forthcoming work. As a pilot trial, this study also lacks objective metrics of diet and physical activity, which could influence bone health and we implore future studies to measure these covariates. Additionally, it is unfortunate that despite randomization participant weight and osteopenic status were imbalanced at baseline; however, primary model estimates adjust for each baseline outcome measure (which, in the case of aBMD acts as a surrogate for weight and osteopenia prevalence), and sensitivity analyses further adjusting for weight change did not materially alter primary results. Finally, we acknowledge that our sample size is small and may not be well suited for a parametric model; however, this approach yields estimates of the magnitude of difference between groups and the degree of change in key outcomes – data that are instrumental in powering future studies.

5 |. CONCLUSION

In conclusion, our pilot data signal that 6 months of risedronate use may be efficacious in mitigating aBMD loss following SG via blunted bone resorption. Benefit appears to be most strongly observed at the axial skeleton and largely maintained throughout the 1-year postoperative period. Findings support the future study of risedronate use for the prophylactic management of bone loss secondary to SG-induced weight loss, with definitive data from an adequately powered trial needed to confirm these pilot results. Finally, as risedronate appears efficacious in blunting, yet not fully preventing, increased bone resorption and aBMD loss secondary to SG, future studies in older adults could consider coupling bisphosphonate use with other bone-sparing strategies – such as dietary and/or exercise countermeasures^37–39 – in attempt to fully preserve bone during surgical and non-surgical⁴⁰ weight loss interventions.

What is already known about this subject?

• Mounting evidence implicates surgical weight loss as a cause of increased skeletal fragility and fracture risk.

• Bisphosphonate therapy reduces osteoporotic fracture risk and may be effective in minimizing bone loss associated with bariatric surgery.

What does your study add?

• Preliminary treatment effect estimates from this pilot randomized controlled trial suggest 6 months of risedronate use may be efficacious in reducing bone resorption and associated bone mineral density loss at the axial skeleton post-sleeve gastrectomy, with benefit largely maintained throughout the 1-year postoperative period.

• Definitive data from an adequately powered randomized controlled trial are necessary to confirm provocative pilot study findings.

DATA AVAILABILITY STATEMENT

Data that support the findings of this study are available from the corresponding author upon reasonable request.