Sclerostin Levels and Changes in Bone Metabolism After Bariatric Surgery

Context:

The role of sclerostin as a key regulator of bone formation remains unknown after Roux-en-Y gastric bypass (RYGB) or laparoscopic sleeve gastrectomy (SG).

Objectives:

The study objectives were evaluation of sclerostin and Dickkopf-1 (DKK-1) serum levels after surgery and correlations with bone turnover markers (P1NP, CTX), parathyroid hormone (iPTH) and areal bone mineral density (BMD), changes at total body, lumbar spine and total hip.

Design and Setting:

This was a prospective observational single-center two-arm study in premenopausal women with acute adipositas over 24 months.

Participants:

Participants were 52 premenopausal women (40 ± 8 years, BMI 43.4) after RYGB and 38 premenopausal women (41 ± 7 years, BMI 45.7) after SG.

Main Outcome Measures:

Prior to surgery and 1, 3, 6, 9, 12, 18, and 24 months after surgery sclerostin, DKK-1, CTX, P1NP levels and BMD were measured.

Results:

Sclerostin, CTX and (to a lesser extent) P1NP increased after surgery and remained elevated during the entire study period (P < 0.001). DKK-1 declined during months 3–9 (P < 0.005) and then remained unchanged, serum phosphate continuously increased (P < 0.001), iPTH remained within the upper normal limit. Sclerostin increases were significantly positively correlated with CTX and P1NP increases and negatively correlated with BMD loss. BMD independently declined regardless of RYGB and SG. Elevations of sclerostin, CTX, P1NP, and phosphate, but not DKK-1 and iPTH, were significant discriminating factors for BMD loss (AUC 0.920).

Conclusion:

Rapid and sustained increases of sclerostin, CTX, and to a lesser extent, P1NP cause an increase in bone metabolism and result in BMD loss at all skeletal sites.

Obesity, which has nearly doubled since 1980, is both a global health concern and an economic burden (1). Bariatric surgery is considered an effective method for immediate and ongoing weight reduction (2). Despite the cardiovascular and endocrinological benefits, there is cause for concern regarding the influence on bone metabolism (3, 4). Recent studies suggest that bone loss after bariatric surgery is not directly caused by weight loss, but rather, seems multifactorial (5–8). The exact pathological mechanism is currently unknown. Laparoscopic Roux-en Y gastric bypass (RYGB) is the most common surgical technique in bariatric surgery (9). A decline in bone mineral density (BMD) at the lumbar spine and at the hip and deteriorations of trabecular and cortical bone compartments, assessed by high-resolution peripheral quantitative computed tomography (CT) (HR-pQCT), were repeatedly reported after RYGB (10–12). The second most commonly performed bariatric technique is laparoscopic sleeve gastrectomy (SG) (9) demonstrating similar concerns with respect to bone loss (13, 14).

Furthermore, bariatric surgery is associated with immediate increases in bone turnover markers (BTM), but information about the longitudinal course of BTM is rare (11, 15). Moreover, the role of sclerostin, as a key regulator of bone formation after bariatric surgery is currently unknown. Sclerostin, which is chiefly produced by osteocytes, negatively inhibits the Wnt-pathway and thus osteoblast differentiation. Serum sclerostin levels differ according to sex (16), age, and physical activity (17) and may contribute to bone loss after bariatric surgery.

Hypothesis: We tested the hypothesis that RYGB and SG in obese young women strongly influence bone metabolism, BMD, and body composition, which is reflected by changes of osteocyte, osteoblast, and osteoclast activity.

Objectives: The primary objective of this study was to investigate any changes in the bone markers sclerostin, DKK-1, P1NP, and CTX during an observational period of 24 months.

Secondary objectives included the evaluation of differences between RYGB and SG with regard to changes in areal lumbar spine, total hip and total body BMD, as well as changes in body composition, medication and quality of life (QOL).

Materials and Methods

Study design

This was a prospective observational single-center two-arm study in premenopausal women with morbid adipositas. The study was performed at the St. Vincent Hospital, Medical Department II, in Vienna, Austria.

Patients were recruited at the Department of Visceral Surgery (>200 bariatric surgeries/y). The decision regarding the respective surgical method, RYGB or SG, was based on the determination of the department of surgery together with the patient́s own judgment.

The study was approved by the local ethics committee. All patients signed a written informed consent document prior to any study-related procedures. The study has been registered in Clinical Trials: NCT01739855. In this manuscript, data from RYGB and SG patients without any pre- or postoperative bone-specific supplementation/medication or physical exercise are presented.

All laboratory assessments were performed within three days prior to surgery (baseline), after one month, after three months and quarterly in the ongoing first year. During the second year of the observation period, all patients had two additional visits. The timeframe for each visit was ±30 days (with the exception of the first visit one month after surgery: ±5 days). DXA was performed at baseline and 6, 12, 18, and 24 months after surgery (Figure 1).

Figure 1.

Study flowchart.

Lab, fasting serum laboratory examinations; MH, medical history; QoL, QOL; BMD, BMD (BMD total body; BMD total hip, BMD L1–L4).

Inclusion and exclusion criteria

Premenopausal obese women with a body mass index (BMI) ≥ 40 kg/m² and a total body weight ≤ 160 kg were included. A cortisol stress test was performed prior to surgery. Patients were excluded if they had any prior oral calcium and/or vitamin D supplementation, antiresorptive or anabolic therapy. Further exclusion criteria were any ongoing therapy with insulin, oral antidiabetic drugs, elevation of liver enzymes (ASAT > 45 IU/L, ALAT > 45 IU/L, GGT > 60 IU/L), eGFR < 90 mL/min/1.73 m², elevation of alkaline phosphatase, systemic or inhalative glucocorticoid use, 25-hydroxyvitamin D deficiency <12 ng/mL or alcohol use >3 U/d.

Laboratory analyses

Blood sampling was performed between 8 and 10 a.m. after an overnight fast. Samples were immediately centrifuged, cooled, and stored at −70°C for later analysis.

The sclerostin and Dickkopf-1 (DKK-1) levels from serum were quantitatively determined using an established enzyme immunoassay (EIA) kit [intra-assay coefficient of variation (CV) is 5–6% for sclerostin and 4–7% for DKK-1; Biomedica]. Crosslaps (CTX), Procollagen type 1 Amino-terminal Propeptide (P1NP), intact parathyroid hormone (PTH), and 25-hydroxyvitamin D (25-OH vitamin D) were measured via chemoluminescence on the IDS-iSYS microparticle immunoassay system (Immunodiagnostics Systems Ltd.). The intra- interassay coefficients of variation were as follows: CTX 2.1–4.9%; P1NP 2.6–3.0%; PTH 1.1–3.7% and 25(OH) Vitamin D 5.5–7.1%. Total serum calcium levels were photometrically determined on the Architect ci8200 platform (Abbott Laboratories).

DXA measurements

Areal lumbar spine, total hip areal, and femoral neck BMD was measured after daily cross-calibrations with standardized control phantom using DXA (GE LUNAR iDXA scanner, software version Encore 13, 50 040). The CV for the spine was 0.41% and 0.53% for total hip. Body composition including total skeletal BMD, total body (kg), lean body mass (kg), were also measured with this DXA scanner - CV for fat distribution was 2.3%; DXA measurements were taken by two well-trained and IOF-ISCD certified technicians. According to the manufacturer's recommendation, the upper weight limit for all DXA measurements was set to 160 kg with the “fat scan mode” > BMI 30.

Quality of life

To evaluate changes in QOL, the Short-Form Health Survey (SF-36) questionnaire was used. It includes the following eight domains: physical functioning, physical role functioning, bodily pain, general health perceptions, vitality, social role functioning, emotional role functioning, and mental health.

All patients were invited to complete a questionnaire prior to surgery and at months 6, 12, 18, and 24 after surgery.

Statistical analysis

The study was designed to enroll approximately 80 patients. Thirty finishers in each group would have at least 80% power to detect a mean within a group difference of 0.0225 g/cm² in areal lumbar spine BMD, assuming a standard deviation (SD) of 0.043 g/cm². Continuous outcome variables are described by the median [interquartile range]. Changes in variable values from baseline measurements to 24 months are tested for statistical significance using the unpaired t test or the nonparametric Wilcoxon signed rank test in case of non-normally distributed differences. Analyzing the complete time course of biochemical markers, repeated measurements ANOVA models were performed. Multiple comparisons were done using the Dunnet test, comparing the values at each visit to the baseline measurement. The Pearson correlation coefficients were used to describe correlations between the relevant outcome variables and to evaluate the association between the changes in these variables. The partial Pearson correlation coefficient was calculated to describe early changes (from baseline to month 1 testing). Log-transformed values were used for statistical analyses of variables with skewed distributions.

Univariate and multiple logistic regression models with baseline and study endpoint data were calculated to evaluate the changes (the lowest quartile vs the higher quartiles) of sclerostin, DKK-1, BTMs, and other contributing factors in BMD depletion. The impact of the factors considered in the logistic regression models are described by odds ratios/SD (OR) and 95% confidence intervals.

Two-sided p-values <0.05 were considered as indicating statistical significance. The SAS software (version 9.4, SAS Institute Inc., 2002–2012) was used for data analyses.

Results

Demographic data, BMI, serum values

One hundred seventeen premenopausal obese women were invited to participate in the study, and of these, 100 signed a written informed consent document. Ten patients experienced immediate and serious postoperative complications (anastomotic leaks and/or sepsis) resulting in a prolonged stay at the intermediate care unit, and were excluded from all analyses. The remaining 90 patients participated in this study with a median age of 40 years and a median BMI of 44 kg/m² (53% body fat), 52 with RYGB, and 38 with SG surgery. Baseline values in both groups were comparable and are reported in Table 1.

Table 1.

Baseline Characteristics of the Study Population

	RYGB Median/IQR n = 52		SG Median/IQR n = 38		P Values
	Age (y)	40	32:47	41	34:48	.544
Body mass index (kg/m2)	43.4	40.4:48.0	45.7	42,2:50,8	.877
Height (cm)	165.5	160:170	160	156.5:166.0	.447
Weight (kg)	119.3	110.2:128.4	127.8	112.4:131.3	.433
Body fat (kg)	61.1	54.8:70.5	66.4	59.5:69.5	.855
Proportion fat (%)	53.1	44.8:58.8	52.2	46.7:56.3	.437
Body lean mass (kg)	54.3	52.4:60.1	56.6	50:57.8	.534
Glucose (mg/dL)	104	94:112	102	92:113	.768
HbA1c (%)	6.6	5.4:7.4	6.5	5.5:7.1	.588
Uric acid (mg(dL)	6.0	5.4:7.0	6.2	5.4:7.9	.633
Cholesterol (mg/dL)	197	175.5:218.3	223.5	190:236.8	.571
HDL cholesterol (mg/dL)	44.5	37.8:52.8	45.5	38.8:52	.754
LDL cholesterol (mg/dL)	129	112.5:150.5	147.5	113.5:176.3	.717
Triglycerides (mg/dL)	116.5	91.5:161.3	183.3	154:201.3	.681
TSH (μU/mL)	1.5	1.1:2.5	1.54	1.23:2.24	.219
Serum protein (g/dL)	6.2	5.9:6.7	6.4	6.1:6.7	.462
Alkaline Phosphatase (IU/L)	109.5	76.4:132.2	111.5	74.8:127.3	.591
Estradiol E2 (pg/mL)	112.2	78.6:134.5	104.7	70.5:129.5	.623
Cortisol (μg/dL)	18.4	12.3:23.1	17.8	11.9:23.2	.499
hs-CRP (mg/L)	10.0	6.9:14.2	10.2	7.6;14.3	.935
Calcium (mmol/L)	2.37	2.26:2.46	2.42	2.38:2.49	.138
Phosphate (mmol/L)	1.06	1.0:1.16	1.03	0.97:1.15	.523
iPTH (pg/mL)	79.6	49.4:94.5	81.9	67.8:115.3	.161
25-OH vitamin D [ng(mL)]	16.3	13.4:22.6	16.2	13.3:22.0	.467
CTX (ng/mL)	0.26	0.20:0.32	0.25	0.12:0.33	.834
P1NP (μg/L)	38.5	32.7:46.1	31.8	26.1:40.7	.089
Sclerostin (pmol/L)	33.9	25.4:46	37.5	21.9:50.3	.146
Dickkopf-1 (pmol/L)	27.7	15.1:43.2	25.9	17.4:41.1	.554
BMD L1-L4 (g/cm2)	1.264	1.190:1.388	1.243	1.054:1.359	.487
BMD total hip (g/cm2)	1.197	1.096:1.226	1.073	0.981:1.228	.612
BMD femoral neck (g/cm2)	1.110	0.952:1.183	1.001	0.892:1.101	.492
BMD total body (g/cm2)	1.304	1.210:1.364	1.298	1.203:1.356	.138

Values are presented as median and interquartile ranges.

During the 24 month observational period there was an expected change of Δ −45% in BMI, a Δ −60% reduction in body fat, but also a loss of Δ −25% in body lean mass (P < .001 for all values).

Sixty two percent in the RYGB and 67% in the SG group were noted as smokers, and no change in smoking behavior occurred during the observation. Adjustment for smoking status did not significantly alter the results of the primary and secondary objectives.

All subjects were initially vitamin D deficient (<20 ng/mL), and serum levels of vitamin D did not improve during the study period. PTH levels were close or above the upper normal range and only slightly declined. Serum calcium levels (total and adjusted for albumin), which were within the lower normal range, significantly declined after 6 and 9 months (Δ −8%, P < .05) and recovered to values close to baseline in both groups. Serum phosphate levels increased continuously, reaching moderate, but sustained significant values at month 3 until study endpoint for both groups (Δ +23% and Δ +25%, P < .001) (Figure 2 and Supplemental Figure 1).

Figure 2.

Median and interquartile serum values of bone parameters.

DKK-1, Dickkopf-1; CTX, C-terminal telopeptide of type I collagen (crosslaps); P1NP, procollagen type 1 amino-terminal propeptide; iPTH, intact parathyroid hormone; calcium, total serum calcium. #, P < .05 vs baseline for both groups (prior to surgery ≤ 3 days) for both groups; *, P < .001 vs baseline for both groups; ##, P < .005 vs baseline. Dotted horizontal lines indicate the age-adjusted reference range of sclerostin (17), DKK-1 (31), P1NP, and serum phosphate, respectively; single dotted horizontal lines indicate the lower limit of normal for vitamin D, CTX, and serum calcium, respectively, and the upper limit of normal for iPTH.

The levels of high sensitive (hs-) CRP were initially increased (median 10.6 mg/L) and slightly declined to median values of 6.1 ng/L (P < .05 vs baseline at months 18 and 24), but all median values remained elevated above the upper normal limit (Supplemental Figure 1 and Supplemental Table 2).

Bone mineral density

In both groups BMD at the lumbar spine and at the total hip persistently decreased. The most notable loss of BMD was observed in the total body BMD with an overall loss of −18% (P < .001 for all values). The decline of BMD in specific regions in the skeleton (arms, legs, trunk, and ribs) did not differ (Figure 3).

Figure 3.

Median and interquartile values of BMD and body composition parameters.

BMD, BMD; BMD total body; BMD lumbar spine L1–L4, BMD total hip; BMI, BMI; total body fat; body lean mass; #, P < .05 vs V1 for both groups; *, P < .001 vs baseline (prior to surgery ≤ 3 days) for both groups.

Fractures

There were two fragility fractures in the RYGB arm: One radius fracture after 14 months and one humerus fracture after 17 months.

Medications

Prior to surgery the number of patients receiving proton pump inhibitors was well balanced between both groups (24.6% and 26.9%). A marked increase was observed after three months (67% and 74% in RYGB and SG patients). At the end of the study, 72% and 70% of patients still received this medication. The increase in the use of selective serotonin reuptake inhibitors was distinct and mainly manifested itself in the second year (+24% and +19% in the two groups). The use of lipid-lowering agents, ACE inhibitors and angiotensin-II-receptor agonists was declined at study endpoint by 57%, 46%, and 57%, respectively.

Serum sclerostin, DKK-1, and BTMs

In both RYGB and SG patients, median serum sclerostin levels increased from 33.9 and 37.5 pmol/L at baseline to 47.3 and 46.9 pmol/L one month after surgery (P < .05) and remained elevated with a maximum of 83.4 and 84.7 pmol/L (P < .001) six months after surgery. These changes were independent of weight loss. DKK-1 levels (27.7 and 25.9 pmol/L at baseline) decreased to minimum values of 13.4 and 12.9 pmol/L at month 6 (P < .05 at month 3, P < .001 at month 6). After that, DKK-1 levels continuously increased to values comparable to baseline. Median CTX levels (0.26 and 0.25 ng/mL at baseline) continuously increased (P < .001 after one month) with a peak at month 12 (0.79 and 0.78 ng/mL) and remained elevated until the study end (0.65 and 0.71 ng/mL). Median P1NP levels (38.5 and 31.8 μg/L at baseline) increased less significantly (P < .05 after 3 months), also reaching peak values after 12 months (66.4 and 67.2 μg/L) and decreased steadily until month 24 (54.8 and 56.3 μg/L, P < .001 at all measured time points) (Figure 2).

Sclerostin positively correlated with CTX in both groups in the partial correlation model for immediate changes within the first month after surgery (r = 0.89 and 0.77, P < .001) and in the correlation model for the whole study period (r = 0.46 and 0.40, P < .01). A positive correlation was also found for CTX and P1NP for both groups (r = 0.57 and 0.59; r = 0.58 and 0.50, P < .005 for all values). Slight, but significant partial correlations were observed between DKK-1, sclerostin, and BTMs for the early changes, but not for the entire study period. No significant correlations were found for iPTH, total serum, and albumin-corrected calcium levels, or age and urinary calcium loss.

Sclerostin negatively correlated with the decreased BMD in both groups at lumbar spine (r = –0.45 and r = –0.49, P < .05), at total hip (r = –0.53 and r = –0.46, P < .05) and at total body BMD (r = −0.38 and −0.37, P < .05). Similar correlations were found for CTX and P1NP levels, but not for DKK-1 and serum phosphate (Table 2 and Supplemental Table 1).

Table 2.

Biochemical Variables of RYGB and SG Patients. Correlation Coefficients for the Changes During Study Period Were Adjusted for Age. Bold Numbers Indicate Statistically Significant Findings

Correlations RYGB	Sclerostin	CTX	P1NP	DKK-1	iPTH	Calcium	Phosphate	BMD L1–L4	BMD Total Hip	BMD Total Body
Sclerostin	1							−0.446	−0.528	−0.382
								P = .038	P = .041	P = .041
CTX	0.456	1						−0.522	−0.514	−0.477
	P < .0001							P < .005	P = .005	P = .002
P1NP	0.582	0.623	1					−0.733	−0.497	−0.501
	P = .005	P < .001						P = .006	P = .021	P = .038
DKK-1	0.438	0.529	0.629	1				0.432	0.398	0.297
	P = .634	P = .385	P = .631					P = .301	P = .823	P = .683
iPTH	0.387	0.183	0.318	0.409	1			0.137	0.482	0.328
	P = 375	P = .574	P = .098	P = .106				P = 675	P = .771	P = .485
Calcium	0.532	0.391	0.264	0.285	0.543	1		0.234	0.311	0.364
	P = .109	P = .632	P = .248	P = .395	P = .134			P = .699	P = .722	P = .688
Phosphate	0.256	0.483	0.842	0.230	0.275	0.285	1	0.147	0.242	0.139
	P = .328	P = .426	P = .175	P = .384	P = .363	P = .197		P = .439	P = .327	P = .414

Correlations SG	Sclerostin	CTX	P1NP	DKK-1	iPTH	Calcium	Phosphate	BMD L1–L4	BMD Total Hip	BMD Total Body
Sclerostin	1							−0.492	−0.463	−0.373
								P = .031	P = .025	P = .002
CTX	0.405	1						−0.632	−0.598	−0.469
	P < .001							P < .001	P = .004	P = .009
P1NP	0.503	0.543	1					−0.733	−0.532	−0.419
	P = .003	P = .035						P = .006	P = .032	P = .047
DKK-1	0.507	0.575	0.532	1				0.422	0.403	0.301
	P = .435	P = .458	P = .472					P = .354	P = .732	P = .582
iPTH	0.412	0.249	0.389	0.504	1			0.163	0.272	0.113
	P = .399	P = .479	P = .157	P = .295				P = 571	P = .552	P = .329
Calcium	0.255	0.533	0.376	0.433	0.479	1		0.141	0.398	0.025
	P = .244	P = .373	P = .523	P = .362	P = .107			P = .699	P = .399	P = .733
Phosphate	0.321	0.453	0.273	0.254	0.255	0.285	1	0.147	0.042	0.439
	P = .193	P = .243	P = .175	P = .321	P = .363	P = .197		P = .652	P = .754	P = .414

Abbreviations: CTX, C-terminal telopeptide of type I collagen (crosslaps); P1NP, procollagen type 1 amino-terminal propeptide; DKK-1, Dickkopf-1; iPTH, intact parathyroid hormone; calcium, total serum calcium; phosphate, serum phosphate; BMD, bone mineral density; L1–L4, lumbar vertebrae 1–4.

With the exception of baseline values, the CTX/P1NP ratio ranged from 37% to 89% (maximum at month 12; mean ratio months 1–24: 58%; P < .005 for all values). No significant ratios for DKK-1 with all other serum analytes were observed.

ROCC analyses

At the end of the study, changes in sclerostin, CTX, P1NP, and to a minor extent DKK-1, were statistically significant discriminators in the RYGB and SG groups. For the estimation of sensitivity and specificity of these markers at different skeletal sites, receiver operating characteristic (ROC) curves (ROCC) were calculated. ROCC analysis of BMD showed the best performance for sclerostin at the total body BMD [area under the curve (AUC) - AUC: 0.790 and 0.795], followed by CTX at the total hip (AUC: 0.765 and 0.759), and CTX at the lumbar spine (AUC: 0.761 and 746). P1NP showed a lower AUC: 0.596 and 0.536. The differences between sclerostin at the total body BMD and CTX at the total hip were significant (P = .04). The lowest AUC was observed for DKK-1 (lumbar spine BMD: 0.405, total body BMD: 0.425, and total hip BMD: 0.454).

Univariate and multiple logistic regression analyses

Associated potential risk factors for bone loss during the study period in the entire study population–due to comparable outcome measures between the two groups (Supplemental Table 2)–were analyzed by univariate and multivariate logistic regressions. Based on the number of the study population, a model was calculated to prove the impact of sclerostin, DKK-1, and BTM on BMD changes during the observational period as expressed by odds ratios (OR, standardized by standard deviations; 0.096 for sclerostin and 0.045 for total body BMD) (Table 3). With the exception of lean body mass (OR 1.6), no significant factors were observed prior to surgery. Changes in sclerostin, CTX, P1NP, serum phosphate, lean body mass, and BMI, but not DKK-1 were significant discriminating factors for BMD loss after bariatric surgery. The AUC for total body BMD was 0.920, 0.923 for lumbar spine BMD, and 0.927 for total hip BMD. Global AUC was 0.946, 0.944, and 0.932 resulting from the multiple logistic regression models when considering all values simultaneously (Figure 4). After exclusion of sclerostin the AUC was 0.845, 0.889, and 0.811, respectively.

Table 3.

Multiple Logistic Regression Analysis of Contributing Factors for BMD Loss per sd (as a Categorical Value, Adjusted for Age), Expressed as Odds Ratios (OR, 95% confidence intervals) in the Study Population.

Variables	BMD Total Body		BMD L1–L4		BMD Total Hip
	OR (95% CI)	P Value	OR (95% CI)	P Value	OR (95% CI)	P Value
Sclerostin (pmol/L)	3.045 (2.472:4.246)	.002	3.011 (2.438:4.375)	.003	2.505 (1.536:3.532)	.004
CTX (ng/mL)	6.523 (3.459:8.971)	.011	5.426 (3.402:7.004)	.032	5.802 (3.977:7.183)	.002
P1NP μg/L	1.543 (0.728:1.967)	.038	2.144 (1.634:3.043)	.042	2.543 (2.113:3.195)	.007
DKK-1 pmol/L	1.178 (0.673:1.304)	.105	1.063 (0.573:1.233)	.459	1.745 (1.124:2.402)	.093
iPTH (pg/mL)	1.454 (0.731:1.483)	.243	0.903 (0.463:1.944)	.436	1.274 (0.554:1.983)	.534
Serum phosphate (mmol/L)	1.797 (1.468:2.014)	.012	1.643 (1.132:2.054)	.003	1.863 (1.345:2.135)	.030
Body lean mass (kg)	1.634 (0.731:2.124)	.234	2.696 (1.553:4.543)	.036	2.923 (2.317:4.013)	.001
BMI (kg/m2)	3.752 (2.587:4.581)	.041	2.994 (2.002:4.112)	.004	2.476 (2.145:3.843)	.003

Abbreviations: CTX, C-terminal telopeptide of type I collagen (crosslaps); P1NP, procollagen type 1 amino-terminal propeptide; DKK-1, Dickkopf-1; iPTH, intact parathyroid hormone; BMI, body mass index.

Figure 4.

ROCC of the multiple logistic regression models with/without sclerostin and CTX, P1NP, DKK-1, serum phosphate, BMI, body lean mass for BMD loss (total body, lumbar spine L1–L4, total hip); AUC, area under the curve.

Quality of Life

During the entire study period no significant improvements with regard to social functioning, emotional role, and mental health were observed. Furthermore, physical summary and mental summary did not change significantly. The components of physical functioning, general health, and bodily pain improved from study month 9 and remained significant (P < .05) until the study end.

Discussion

This was a clinical trial in premenopausal untreated women with morbid obesity. We investigated the early and protracted effects of RYGB and SG on BTM and BMD over a period of 24 months and the possible role of sclerostin as a key regulator in this specific population.

Our data demonstrate for the first time increased levels of sclerostin, independent of the type of bariatric surgery, with ongoing increases of bone formation and resorption markers and a transient reduction of DKK-1 levels.

The findings on changes in CTX, phosphorus or vitamin D are only partly in line with previous published studies (10, 18, 19). In our population, baseline values of PTH levels were elevated, possibly related to the well-known increment of leptin due to the obesity of our patients or to low vitamin D levels. Secondary hyperparathyroidism is common in obese patients and surgery-induced weight loss does not improve vitamin D or PTH levels (7). Diverse animal models suggest secondary hyperparathyroidism with vitamin D malabsorption and increased CTX and low P1NP levels after RYGB responsible for cortical and trabecular bone loss (20–22). A recent RYGB rat model discovered metabolic acidosis as an independent contributing factor for bone loss. Upregulated 125 (OH)₂-vitamin D activation seems to be a compensation of the intestinal calcium malabsorption (6).

The role of sclerostin in our patients remains unknown. A study on postmenopausal women with primary hyperparathyroidism shows a trend towards lower sclerostin levels and higher CTX or bone alkaline phosphatase levels caused by an imbalance in Wnt function. Additionally, sclerostin was positively associated with age, body composition, BMD, physical activity, and negatively correlated with markers of bone formation and calcium. Weight loss in obese female and male adults (≥65 years, BMI ≥ 30 kg/m²) increases sclerostin levels, which can be stabilized by muscle exercise and/or protein diet (17, 23, 24). By way of contrast, in our subjects, sclerostin increased within 30 days, sustained with a peak after 6 months, steadily declined, but never attained comparable baseline values. This shift of bone formation towards resorption is represented by strong positive partial and global correlations of sclerostin and BTM. The lack of mechanical loading due to the rapid and excessive weight loss partly explains the increases of sclerostin serum levels via changes of SOST, RANKL, and OPG activity. This noncanonical Wnt-pathway promotes osteoclastogenesis via increased SOST and reduced OPG and causes undermineralization of bone matrix. Furthermore, we observed an increase of extracellular phosphate levels as a sign of enhanced liberation of resorbed bone (25–27). The phosphate homeostasis is linked to skeletal mineralization, and FGF 23 inactivation results in hyperphosphatemia. Animal models suggest that inactivation of FGF23 is associated with severe osteomalacia presuming that serum phosphate can act as a signaling molecule (28). None of these models focused on the activity of osteoblasts and possible mechanisms of endogenous compensation apart from canonical and noncanonical Wnt-pathways. LRP5 as a coreceptor of Wnt controls bone formation acts via osteoblasts by the inhibition of serotonin synthesis in the duodenum. RYGB and SG cause severe anatomical alteration within the intestine, P1NP increases can be partly interpreted as a compensation of bone loss (29). Furthermore, obesity is associated with chronic low grade inflammation and increased levels of CRP, IL-6, and TNFα. Bariatric surgery is associated with an ongoing decline, but the influence on bone metabolism is still unknown (30). We observed a decline of elevated hs-CRP levels, but median levels were still elevated at the study endpoint.

The Wnt pathway and its endogenous inhibitors sclerostin and DKK-1 play an important role in bone formation and regeneration, which appears to be associated with changes in bone mass. While sclerostin levels in our study population were initially within normal ranges, median DKK-1 levels were above normal ranges compared to healthy young females (31). Besides bone metabolism, DKK-1 is also involved in chronic inflammation, atherogenesis, endothelial dysfunction, and in the regulation of glucose metabolism (32). Initially, our patients had elevated levels of hs-CRP and fasting glucose levels within the upper limit of normal suggesting chronic inflammation and poor glycemic control. We observed a significant decrease in DKK-1 levels between months 3 and 9, subsequently increasing to baseline levels. The transient decline in DKK-1 could be hypothetically interpreted as a response within the Wnt pathway to the changes of sclerostin, but no correlation between DKK-1 and changes in sclerostin, BTMs, BMD, hs-CRP, or fasting glucose levels was found. This suggests that DKK-1 levels per se provide only limited information about the local microenvironment of bone tissue in patients after bariatric surgery compared to patients with diabetes or rheumatoid arthritis (RA) (33).

There are controversies regarding BMD and body composition imaging accuracy with DXA technology in obese patients as measurements as axial sites compared to appendicular sites have more overlying soft tissue. Despite this concern, bone loss after bariatric surgery is well documented and DXA is an easily obtainable noninvasive method (5, 34). We hypothesized changes in bone metabolism represented by severely and sustained elevated resorption and formation marker with negative effects on BMD. In this context, the role of sclerostin and its unexpected elevation remains crucial. Our findings on BMD, weight loss or BTM independent of the surgical method are in line with recently published data, only in super obese patients (BMI > 50) RYGB seems to have a benefit concerning weight loss (35, 36).

The multiple logistic regression models (baseline vs study endpoint) were calculated to evaluate the effect of changes of sclerostin and other variables in the observed bone loss. Sclerostin, in particular, provides further information in the interpretation of the observed decline of BMD (AUC for total body BMD 0.920 vs 0.845 without sclerostin).

Since the number of bariatric surgeries is increasing worldwide, our findings might encourage prospective pharmacological interventional studies. The observed loss of BMD after bariatric surgery could be a risk factor for fragility fractures, in particular in this relatively young population close to the expected peak bone mass (37). Whether sclerostin (or DKK-1) antibody treatment or antiresorptives with/without anabolic combination treatment might be a suitable treatment option, it will be subject to preclinical and clinical studies. To date, these therapies were evaluated in postmenopausal women, but also in various animal models of bone disease (38, 39, 41).

The mean loss of −25% in body lean mass in our patients without a high-protein diet or structured muscle exercise training should be observed to prevent sarcopenia and can be partly explained by mechanical unloading of the skeleton owing to drastic weight loss (35).

Besides alterations in bone metabolism, there was an increase in both groups in the number of new patients taking proton pump inhibitors and, to a lesser extent, the current use of selective serotonin reuptake inhibitors. During their visits, many patients stated that they would not have undergone the surgery had they previously known the full extent of the personal difficulties they would encounter regarding food intake (vomiting, reflux, sweating during eating, psychosocial disorder). Another important psychological issue, mainly in the second year after the operation, was sagging skin at arms, legs, and abdomen, which caused significant discomfort and shame. While our findings on QOL are to some extent not in line with current studies which describe improved QOL on several levels after bariatric surgery, it is also known that patients with preoperative depression have inferior results in this regard (40).

Strengths and limitations

The strength of this study is its long observational period and the investigation of serum markers reflecting bone metabolism, as well as the changes in BMD at different skeletal sites and body composition in premenopausal patients. Additionally, we investigated changes in sclerostin and DKK-1 in this longitudinal setting.

One limitation is the lack of paired tetracycline labeled transiliac bone biopsies to evaluate the observed areal BMD changes in correlation with histomorphometric parameters. Volumetric BMD was not assessed by HR-pQCT measurements to analyze cortical and trabecular bone structure. Due to the obesity, we could not position the patient's arms or legs in the manufacturer's approved rack to perform a feasible examination without motion artifacts. A recently published HR-pQCT study in a study population with comparable BMI and age reported cortical bone deterioration after RYGB, especially at the tibia, with stable trabecular parameters. In contrast to our study, these patients had ongoing vitamin D and calcium supplementation, a possible protective factor (11). Based on the decision of the department of visceral surgery on the respective surgical method, this study also has a lack of a structured randomization (and therefore has differing number of patients in the two groups). Furthermore, our study was not designed to evaluate any potential clinical risks or benefits, such as fracture outcome of the investigated population.

Conclusion

Changes of sclerostin and BTM serum levels provide further important information on bone metabolism in patients after bariatric surgery. The observed increases in resorption, formation, and osteocyte markers result in a vast and ongoing loss of BMD and should be subject to further clinical or pharmacological investigation. Additionally, weight-bearing muscle exercise, a well-balanced diet, and vitamin D supplementation should also be taken into consideration to prevent iatrogenic sarcopenia.

Acknowledgments

The authors cordially thank our study nurse Dragana Simic and our secretary Monika Binder-Ziegler for the coordination of the participants. We also thank Sabine Klauss at Ulm/Germany for graphic design of the figures, Professor Tommy Vacca at Linz/Austria for proofreading, and we also acknowledge the work of Professor Stylianos Kapiotis and his staff of the central laboratory at St. Vincent Hospital Vienna.

Author's roles: Study design: C.M. and H.R. Study conduct: C.M. and R.K. Data collection: C.M., R.K., Ch.M., A.R.N. Data analysis and statistical calculations: C.M., R.K., G.K.M., and Ch.M. Data interpretation: C.M., R.K., G.K.M., H.R., and P.P. Drafting manuscript: C.M., R.K., G.K.M., and P.P. Revising manuscript content: C.M., R.K., Ch.M., A.R.N., G.K.M., H.R., and P.P. Approving final version of manuscript and revised manuscript: C.M., R.K., Ch.M., A.R.N., G.K.M., H.R., and P.P. C.M. takes responsibility for the integrity of the data analysis.

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• AUC

  area under the curve

 
• BMD

  bone mineral density

 
• BMI

  body mass index

 
• BTM

  bone turnover markers

 
• COV

  coefficient of variation

 
• CTX

  crosslaps

 
• EIA

  enzyme immunoassay

 
• HR-pQCT

  high-resolution peripheral quantitative computed tomography

 
• OR

  odds ratios

 
• P1NP

  Procollagen type 1 Amino-terminal Propeptide

 
• QOL

  quality of life

 
• RA

  rheumatoid arthritis

 
• ROC

  receiver operating characteristic

 
• ROCC

  receiver operating characteristic curves

 
• RTGB

  Roux-en Y gastric bypass

 
• SG

  sleeve gastrectomy.