Romosozumab is effective in patients with variants in LRP5, LRP6, or WNT1

Abstract

Summary

Romosozumab has proven efficacy in postmenopausal osteoporosis, but data in LRP5/6 or WNT1 variant carriers are scarce. In 33 patient-years of women carrying such variants, romosozumab improved BMD and bone formation more than antiresorptives or no treatment. These findings suggest romosozumab as a potential therapy for genetically defined bone loss.

Purpose

The sclerostin inhibitor romosozumab is an osteoanabolic drug that enhances the canonical Wnt-β-catenin pathway. While its efficacy in postmenopausal osteoporosis is known, clinical data on the use of romosozumab in patients with variants in the LRP5, LRP6, or WNT1 gene affecting this pathway remain limited.

Method

We present clinical routine data of 33 patient-years in 27 postmenopausal women with heterozygous deleterious or established risk alleles variants (ACMG class 2–5) in the LRP5/6 or WNT1 gene, using dual-energy X-ray absorptiometry (DXA), high-resolution peripheral quantitative computed tomography (HR-pQCT), and laboratory parameters. Among these, 12 were treated with vitamin D and calcium only (NoT), 11 with antiresorptive therapy (AR), and 10 with romosozumab (ROMO).

Results

Patients treated with romosozumab showed significantly greater gains in areal bone mineral density (aBMD) compared to others (lumbar spine—NoT = −0.12% ± 5.02%, AR = + 3.97% ± 4.71%, ROMO = + 15.74% ± 7.63%, p < 0.001 for ROMO vs. AR and vs. NoT). This was accompanied by an increase in trabecular bone density and mass in the tibia, as well as elevated bone formation markers such as bone alkaline phosphatase after 6 months (AR = −3.36 µg/L ± 8.044 µg/L vs. ROMO = + 3.08 µg/L ± 2.21 µg/L, p = 0.026). Our findings indicate that romosozumab is effective in patients with heterozygous variants in the LRP5/6 or WNT1 gene. Here, romosozumab was superior to established antiresorptive treatments.

Conclusion

Thus, romosozumab might be a specific treatment for these patients. However, further studies are needed with larger patient cohorts and in particular with regard to biallelic loss-of-function variants in the LRP 5/6 or WNT1 gene.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00198-026-07932-8.

Introduction

Osteoporotic fractures cause pain and immobility, thereby reducing quality of life for the affected patients, but also increased healthcare costs. The majority of patients with osteoporosis, including postmenopausal women, are older than 50 years, and there are specific therapeutic guidelines to counteract bone loss and to prevent osteoporotic fractures for this patient group. Importantly, however, the treatment of patients with early-onset osteoporosis remains challenging, especially since the causes of the respective pathologies are different between individual patients. In the last years it became evident that a large number of patients with early-onset osteoporosis exhibit a specific genetic cause explaining their skeletal symptoms. [1] Theses genetic variations may concern specific pathways of the bone metabolism displaying distinct phenotypes requiring specific treatment.

The canonical Wnt-β-catenin pathway represents one of the most important osteoanabolic pathways in bone metabolism. Key mediators of this pathway include the co-receptors LRP5 and LRP6, which transduce Wnt signals essential for bone formation. Inhibitors of the Wnt pathway, such as sclerostin, [2] are secreted by osteocytes [3] in response to reduced or absent mechanical loading [4], acting as natural suppressors of bone formation. Loss-of-function (LOF) variants in LRP5/6 or WNT1 reduce the efficacy of the Wnt signaling, leading to compromised bone formation and mineralization. Heterozygous variants in the LRP5/6 and WNT1 genes are linked to early-onset osteoporosis (EOOP) [5, 6], as they disrupt the normal signaling mechanisms that regulate bone formation, resulting in a state of low-turnover bone metabolism. Moreover, whereas homozygous or compound heterozygous (likely) pathogenic variants in WNT1 cause osteogenesis imperfecta type XV, biallelic (likely) pathogenic variants of LRP5 lead to the severe osteoporosis-pseudoglioma syndrome (OPPG).[5] Additionally, established risk alleles for LRP5 have been described, such as p.(Val667Met), which can be found in up to 15.4% of young patients with osteoporosis and is associated with reduced bone mineral density.[7] Furthermore, preclinical studies have shown that this variant not only leads to a significant reduction of Wnt signaling in cell culture 7 but also significantly reduces the bone mineral density and osteoblast activity in a respective mouse model.[8]

Today, romosozumab, a monoclonal antibody that inhibits sclerostin and thereby enhances Wnt signaling, is available and approved for female patients with severe postmenopausal osteoporosis and vertebral fractures. While romosozumab has been shown to be effective in treating postmenopausal osteoporosis within the ARCH and FRAME trial, [9, 10] its application in patients with variants in LRP5/6 or WNT1 is rarely reported in the literature. [11] Preclinical data of a mouse model for osteoporosis-pseudoglioma syndrome (OPPG, homozygous inactivation of Lrp5) suggests that activation of the Wnt pathway through sclerostin inhibition can yield anabolic benefits. [12] On the other hand, a mouse model of combined Lrp5 and Lrp6 inactivation showed no osteoanabolic effects following anti-sclerostin antibody treatment, [13] which raises the clinically relevant question if romosozumab can increase bone mass in patients carrying variants in LRP5/6 or WNT1.

In this retrospective study, we compared the effects of different osteoporosis treatments (vitamin D and calcium only, antiresorptive as well as romosozumab therapy) on skeletal parameters in patients with variants affecting the canonical Wnt-β-catenin pathway, i.e., in LRP5, LRP6, or WNT1. Patients were characterized based on clinical features, BMD, and bone structure. Additionally, we evaluated bone metabolism through markers of bone formation and resorption to assess the impact of these therapeutic approaches. To account for differences between the different genes, we added a statistical evaluation of patients with LRP5 variants only.

Materials and methods

Study design

This study is a retrospective data analysis. The research was conducted at a single bone-specific outpatient center and followed local ethical guidelines, local laws, and the principles outlined in the Declaration of Helsinki. All patients gave written, informed consent to the evaluation of their patient files and data. Approval was obtained from the local ethics committee of the chamber of physicians (2022–100804-BO-ff).

Patients

Patient files from 01/2022 to 12/2024 were retrospectively screened for postmenopausal women with osteoporosis and reported variants in the LRP5/6 or WNT1 gene. Cases of secondary osteoporosis were excluded. Patients were then grouped based on their documented therapeutic approach into three categories: no treatment (NoT), antiresorptive treatment (AR), and romosozumab treatment (ROMO). Fracture history and prior bone-specific treatments were recorded. Patients were characterized through clinical interviews, biochemical analyses of bone metabolism, bone mineral density (BMD) measurements, and high-resolution peripheral quantitative computed tomography (HR-pQCT) prior to treatment. Follow-up examinations were conducted at 6 months for the treatment groups and at 12 months for all groups. All patients received vitamin D supplementation, while calcium was supplemented only in cases of suspected calcium deficiency, indicated by elevated or high-normal parathyroid hormone (PTH) levels. Six patients were included in two groups: three initially treated with antiresorptive treatment and subsequently with romosozumab (Table 2) and three initially without treatment and subsequent romosozumab treatment (Table 2).

Table 2.

Genetic variants of the included patients: all patients were heterozygous unless stated otherwise. Patient ID17 is listed twice due to two variants detected but only counted once. ERA, established risk allele; LRP5*, individual 17 (ID 17) carries two variants in the LRP5 gene; LRP5^§, homozygous, all other heterozygous; N, no treatment; A, antiresorptive treatment; R, romosozumab treatment; AA, amino acid

ID	Affected gene	Treatment			Base	AA Change	ACMG class	ACMG criteria	gnomAD frequency (total)
		N	A	R
1	LRP5	X			c.2377G>A	p.(Gly793Arg)	4	PM2_sup, PP3_str, PP4	Absent
2	WNT1		X		c.754G>C	p.(Gly252Arg)	3	PM2_sup, PP3_mod	5.1 × 10−4
3	LRP5	X			c.2377G>A	p.(Gly793Arg)	3	PM2_sup, PP3_str, PP4	Absent
4	LRP6	X		X	c.775C>T	p.(Arg259Cys)	3	PM2_sup, PP3	8.8 × 10−5
5	WNT1	X			c.703C>T	p.(Arg235Trp)	5	PS3, PS4, PM2_sup, PP1, PP3	Absent
6	WNT1		X		c.703C>T	p.(Arg235Trp)	5	PS3, PS4, PM2_sup, PP1, PP3	Absent
7	WNT1		X	X	c.264T>A	p.(Ser88Arg)	ERA	Association Study_str	0.003
8	LRP5	X			c.3107G>A	p.(Arg1036Gln)	ERA	Association Study_vstr, Functional Study	0.004
9	LRP5	X		X	c.2829G>A	p.(Pro943 =)	2–3	PM2_sup, BP4	4.4 × 10−4
10	LRP5			X	c.1141G>A	p.(Asp381Asn)	4	PS4, PM2_sup, PP3_mod, PP4	4 × 10−6
11	LRP5	X			c.3403C>T	p.(Arg1135Cys)	4	PS4, PM2_sup, PP3, PP4	2.6 × 10−4
12	LRP5			X	c.3403C>T	p.(Arg1135Cys)	4	PS4, PM2_sup, PP3, PP4	2.6 × 10−4
13	LRP5	X			c.1067C>T	p.(Ser356Leu)	5	PS1, PS3, PM2_sup, PP3_mod	6.7 × 10−5
14	LRP5	X			c.1999G>A	p.(Val667Met)	ERA	Association Study_vstr, Functional Study	0.038
15	LRP5		X		c.2134G>A	p.(Val712Met)	3	PM2_sup, PP3	2.4 × 10-5
16	LRP5	X		X	c.1999G>A	p.(Val667Met)	ERA	Association Study_vstr, Functional Study	0.038
17	LRP5*			X	c.1999G> A	p.(Val667Met)	ERA	Association Study_vstr, Functional Study	0.038
17	LRP5*			X	c.2989C>T	p.(Ala1330Val)	ERA	Association Study_vstr	0.13
18	LRP5	X			c.2234C>A	p.(Ala745Val)	4	PS4, PM2_sup, PP3_mod, PP4	Absent
19	LRP5		X		c.2234C>A	p.(Ala745Val)	4	PS4, PM2_sup, PP3_mod, PP4	Absent
20	LRP5		X	X	c.1999C>A	p.(Val667Met)	ERA	Association Study_vstr, Functional Study	0.038
21	LRP5§			X	c.1999C>A	p.(Val667Met)	ERA	Association Study_vstr, Functional Study	0.038
22	LRP6	X			c.1396G>A	p.(Gly466Arg)	3	PM2_sup, PP3_str	Absent
23	WNT1	X			c.481delG	p.(Gly162Alafs*37)	4	PVS1_str, PM2_sup, PP4	Absent
24	LRP5		X		c.1999G>A	p.(Val667Met)	ERA	Association Study_vstr, Functional Study	0.038
25	LRP5		X	X	c.1999G> A	p.(Val667Met)	ERA	Association Study_vstr, Functional Study	0.038
26	LRP5	X			c.3403C>T	p.(Arg1135Cys)	4	PS4, PM2_sup, PP3, PP4	2.6 × 10−4
27	LRP5		X		c.3245A>G	p.(Tyr1082Cys)	4	PS4_mod, PM2_sup, PP3_str	9.9 × 10−5

Genetic methods

Patients with unclear (primary/idiopathic) osteoporosis, defined by history of BMD < T-score −2.5 and/or osteoporotic fractures before menopause, were offered genetic testing after exclusion of secondary causes for osteoporosis. DNA was isolated from blood samples and fragmented by an ultrasound device (Covaris, Woburn, MA, USA). Genetic testing was done by next-generation sequencing after enrichment by a custom designed Sure Select XT gene panel (skeletal disease-associated genome (sDAG) panel) (Agilent, Santa Clara, CA, USA) containing all genes described at the time in the nosology and classification of genetic skeletal disorders [14]. Sequencing data were generated on MiSeq, NextSeq, and HiSeq machines (Illumina, San Diego, CA, USA) with a coverage of > 100× leading to > 20× coverage of > 98% of the target region. Rare genetic variants were classified according to ACMG criteria [15]. For the classification of common (> 1% allele frequency) variants, we followed the recommendations for risk allele curation, classification, and reporting by ClinGen [16].

Bone density measurements

Dual-energy X-ray absorptiometry (DXA) was performed on the lumbar spine and both hips in accordance with international guidelines [17] using a Lunar iDXA device (GE Healthcare, Madison, WI, USA). In the romosozumab (ROMO) group, paired tests were used to assess absolute aBMD changes at the lumbar spine and hips over 1 year. For intergroup comparisons, changes were analyzed separately for the total lumbar spine (L1–L4, or the two lowest adjacent vertebrae if a fracture at L1, L2, L3, or L4 was present) and for the average total femur BMD across the three treatment groups except for lowest T-score choosing the lowest of the two available total femur T-scores and the lowest of at least two adjacent vertebral bodies for the lumbar spine. Vertebral fracture analysis (VFA) was performed at the first clinical presentation and repeated after 1 year to assess spinal fractures, also using the Lunar iDXA device.

HR-pQCT

High-resolution peripheral quantitative computed tomography (HR-pQCT) measurements were performed at the initial presentation and after 1 year of therapy within the clinical routine setup. Scans were acquired from the distal, non-dominant radius and the contralateral tibia, following international guidelines. Imaging was conducted using either a first-generation HR-pQCT device with an isometric voxel size of 82 µm (XtremeCT I, Scanco Medical AG, Brüttisellen, Switzerland) or a second-generation HR-pQCT device with an isometric voxel size of 60.7 µm (XtremeCT II, Scanco Medical AG, Brüttisellen, Switzerland). The HR-pQCT images were analyzed using the manufacturer’s established scripts, followingguidelines18  and their manuals.

Prior to evaluation, scans were screened for motion artifacts and scans with a higher than 3 grade were excluded. [18, 19] To match both generations of HR-pQCT results, all second-generation results were recalculated in accordance to Manske et al. matching first generation for comparability reasons. [20] Total volumetric bone mineral density (Tt.vBMD), trabecular volumetric bone mineral density (Tb.vBMD), and bone volume to tissue volume (BV/TV) were extracted for further analysis as well as trabecular thickness (Tb.Th), trabecular number (Tb.N), trabecular separation (Tb.Sp), cortical thickness (Ct.Th), and cortical volumetric bone mineral density Ct.vBMD.

Laboratory assessment

Analysis of the bone turnover was carried out within clinical routine, at the associated, local laboratory (Institute of Clinical Chemistry and Laboratory Medicine, University Medical Center Hamburg-Eppendorf, Germany). Serum levels of bone formation markers such as bAP (bone-specific alkaline phosphatase) and osteocalcin (OC) were measured by a Liaison XL from Diasorin. Urinary levels of deoxypyridinoline per creatinine (DPD) were measured using Immulite-XP by Siemens.

Statistical evaluation

Group comparisons were first assessed for normality. Statistical significance was defined as p < 0.05, with α set at 0.05. Parametric tests were applied for normally distributed data, while non-parametric tests were used for non-normally distributed data. For follow-up data within the same individuals, paired t-tests were employed for normally distributed data, and the Wilcoxon test was used for non-normally distributed data. Simple two-group comparisons were conducted using the t-test or the Mann-Whitney U test, depending on the data distribution. For comparisons involving more than two groups, ANCOVA was performed, with age, BMI, and pre-treatment included as covariates if significant for DXA and HR-pQCT. For comparison of frequency distributions, chi-squared tests were used. For comparison of the DXA results of the three treatment groups, baseline BMD was tested as a covariate and included if significant. To address differences between variants in LRP5/6 and Wnt1, we have additionally run the statistical analysis on patients with LRP5 variants only and present these results in the supplemental material. A threshold of < 1.6% aBMD gain at the hip over one year was used to define non-response using the %CV [21].

Results

Patients

In total, 27 women were included, carrying variants in the LRP5/6 or WNT1 gene (Table 1) adding up to 33 patient-years under specific treatment or no treatment (NoT). While most variants were heterozygous, ID17 carried two heterozygous variants and ID21 a homozygous variant. No significant differences were detected with respect to patient age comparing NoT and romosozumab (ROMO) group while the antiresorptive group (AR) was significantly older than the NoT group (NoT = 52.2 ± 9.3 years, AR = 63.6 ± 9.4, ROMO = 54.8 ± 7.5, p = 0.011 for NoT vs. AR). No significant differences were observed for BMI between the groups nor for frequency of LRP5, LRP6, and WNT1 genetic variants. Baseline aBMD did not differ between the three groups (Table 1). Patient-specific variants and treatments including bone-specific treatment history are listed in Table 2. When comparing the mean age of menopause onset, no differences were detected between the three groups (NoT = 49.9 ± 1.4 years, AR = 50.1 ± 3.7, ROMO = 50.0 ± 3.2, p > 0.9 for ANOVA). Comparing the frequency of previous, bone-specific treatments between the groups, a significantly lower fraction of patients in the NoT group did receive such treatments prior to the here reported treatment. Half of the patients of the ROMO group did receive an antiresorptive treatment before with an average duration of 22.6 ± 14.5 months. When stratifying for pre-treatment in the ROMO group, patients without pre-treatment did exhibit a significantly higher BMD gain at the spine than those with pre-treatment (Suppl. Figure 1—w/pre-treatment ROMO = 9.7 ± 5.3%, w/out pre-treatment ROMO = 21.7 ± 3.5%, p < 0.0001) with a preserved, significantly higher gain than in the NoT group for both, w/out and with pre-treatment (p < 0.0001, p < 0.005). Patients without pre-treatment in the ROMO group had a > fivefold higher BMD increase at the spine than those from the AR group (p < 0.0001) while patients with pre-treatment had a > twofold increase, yet not reaching significance (p = 0.15) given a small group size (Suppl. Figure 1).

Table 1.

Demographic characteristics; NoT, no treatment; AR, antiresorptive treatment; ROMO, treatment with romosozumab; PBST, prior bone-specific treatment; ANOVA, analysis of variance, ^aNoT vs. AR, lowest spine = lowest lumbar spine value of at least two adjacent vertebrae (L₁–L₄), avg. total femur = average value of the left and right femur, total spine = average BMD of L₁–L₄

	NoT	AR	ROMO	ANOVA
n = 33	n = 12	n = 11	n = 10
Age [years]	52.2 ± 9.3	63.6 ± 9.4	54.8 ± 7.5	p = 0.011a
Menopause [years]	49.9 ± 1.4	50.1 ± 3.7	50.0 ± 3.2	NS
BMI [kg/m2]	24.2 ± 6.2	23.5 ± 2.8	22.7 ± 2.5	NS
Baseline BMD
Total spine [gHA/cm2]	0.87 ± 0.09	0.87 ± 0.14	0.75 ± 0.08	NS
Avg. total femur [gHA/cm2]	0.79 ± 0.08	0.72 ± 0.12	0.72 ± 0.12	NS
Baseline T-score
Lowest spine	−3.0 ± 0.5	−2.8 ± 1.2	−3.9 ± 1.0	NS
Lowest total femur	−1.8 ± 0.6	−2.4 ± 1.0	−2.4 ± 1.0	NS
Genetic variant
LRP5	75%	63.6%	80%	NS
LRP6	8.3%	9.1%	10%	NS
WNT1	16.7%	27.3%	10%	NS
PBST	8.3%	63.4%	50%	p = 0.016a
Fractures
Vertebral	16.7%	72.3%	40%	p = 0.017a
Non-vertebral	75%	54.5%	50%	NS

For fracture frequency, the AR group had a higher frequency of vertebral fractures than the NoT group (p = 0.017). Two patients had received denosumab, one an intravenous bisphosphonate, and two oral bisphosphonates. For the AR group, 6 patients received intravenous ibandronate (3 mg) every three months while 5 patients received subcutaneous denosumab (60 mg) every 6 months.

Additional information on the baseline characteristics excluding patients with variants in the LRP6 or WNT1 gene are available in the Supplemental Table 1. Using a threshold of ≤ 1.6% aBMD gain at the hip over 1 year to define non-response, we identified seventeen patients as non-responders. Only two of them had received romosozumab (ID4, ID20), who nonetheless showed a +15.83% and +5.40% BMD increase at the spine. Importantly, ID20 received an antiresorptive therapy prior to romosozumab. Among non-responders, four carried an ACMG class 3 variant and five an ACMG class 4 variant. The mean spinal aBMD change in this subgroup was +2.04% (range −3.7% to +15.83%). Three patients, including the romosozumab-treated individual, had prior bisphosphonate exposure. Neither the involved gene nor the specific variant or the ACMG class was the same in these patients. The affected variants included changes in WNT1, LRP5, and LRP6. In WNT1, we observed the variants c.354G>C (p.Gly252Arg) and c.264T>A (p.Ser88Arg). Variants in LRP5 included c.2377G>A (p.Gly793Arg), c.2829G>A (p.Pro943 =), c.3403C>T (p.Arg1135Cys), and c.2234C>A (p.Ala745Val). For LRP6, the variants c.775C>T (p.Arg259Cys), observed in two patients, and c.1396G>A (p.Gly466Arg) were identified.

Bone mineral density

Comparing the percentage change of the spinal areal bone mineral density (aBMD), patients treated with romosozumab had a significantly (p < 0.001) higher aBMD gain than NoT or AR patients = NoT = −0.12 ± 5.01, AR = 3.97 ± 4.57, ROMO = 15.74 ± 7.63; (Fig. 1A). Accordingly, the aBMD of the spine was significantly higher in the ROMO group after 1 year of treatment (p < 0.001, Fig. 1B). When comparing the spinal T-score prior to therapy and 1-year follow-up, significant gains in T-score were detected only in the ROMO group (T₀ = −3.90 ± 1.06 vs. T₁₂ = −3.06 ± 0.88; p = 0.003, Fig. 1C).

Fig. 1.

Bone mineral density comparisons: (A, D) Percent change in areal bone mineral density (ΔaBMD) from baseline to 12-month follow-up for the total lumbar spine (A) and average total femur (D). (B, E) Absolute change in aBMD from baseline to 12 months for the lumbar spine (B) and total average femur (E). (C, F) Lowest T-score measures at baseline and 12 months for the lumbar spine (C) and total hip (F) for the three groups of no, antiresorptive, and romosozumab treatment. NoT, no treatment; AR, antiresorptive treatment; ROMO, romosozumab treatment; ΔaBMD, percent change in areal bone mineral density from baseline; aBMD, areal bone mineral density

For hip measurements, a significantly higher BMD gain was seen in the ROMO group compared to the NoT group (NoT = 0.41 ± 2.35, AR = 1.76 ± 2.11, ROMO = 5.24 ± 4.52; p < 0.001 Fig. 1D). Accordingly, a significant increase in aBMD was detected in the ROMO treated group (Fig. 1B, E). While there was no difference between NoT and AR with respect to increase in aBMD, the T-score was significantly higher in AR after one year of therapy (T₀ = −2.39 ± 1.06 vs. T₁₂ = −2.28 ± 1.02, p = 0.045) and an even more pronounced aBMD increase in the ROMO group was seen (T₀ = −2.43 ± 1.02 vs. T₁₂ = −2.12 ± 1.04, p = 0.007, Fig. 1E, F).

Baseline BMD did not significantly predict the changes in BMD under treatment. Results for comparison of patients with LRP5 variants only are described in the Supplemental Figure 2–4. While differences changed as well as exact p-values, the general pattern remained the same after excluding patients with variants in LRP6 or Wnt1. When excluding patients with ERA or ACMG class 2 (including WNT1 with ACMG class ≥ 3), which are particularly present in the ROMO group, the same trends were observed with significant BMD gain in the spine (p < 0.001, Suppl. Figure 5 A) while, given the smaller sample size, hip gains were only at borderline (p = 0.054, Suppl. Figure 5D). The same pattern was observed for paired testing of total aBMD of the spine (p = 0.018, Suppl. Figure 5B) while hip aBMD gain did not reach significance.

HR-pQCT

Measuring the three-dimensional, volumetric bone mineral density (Fig. 2A, B, D, E) as well as bone volume per tissue volume (Fig. 2C, F) in radius and tibia, no changes between the groups were observed in Tt.vBMD for tibia (Fig. 2A) or radius (Fig. 2D). However, a significant increase of vBMD was seen in the trabecular compartment in the load-bearing tibia (Fig. 2B) but not the radius (Fig. 2E) (Tibia = p = 0.046 for NoT vs. ROMO, p = 0.039 for AR vs. ROMO). The same pattern was detected for BV/TV with significant increases in the tibia (p = 0.042; Fig. 2E, F) comparing NoT to ROMO but no changes in the radius. No significant changes were detected for Ct.vBMD, Ct.Th, Tb.Th, Tb.N, or Tb.Sp. For HR-pQCT measurements, 8 patients were available for NoT, 8 for AR, and 7 for ROMO after excluding motion corrupt patient data. Results are shown in Suppl. Tabl. 2.

Fig. 2.

Three-dimensional bone structure and density evaluation: (A, D) Percent change in total volumetric bone mineral density (ΔTt.vBMD) at the tibia (A) and radius (D) from baseline to 12-month follow-up. (B, E) Percent change in trabecular volumetric bone mineral density (ΔTb.vBMD) at the tibia (B) and radius (E). (C, F) Percent change in bone volume to tissue volume (ΔBV/TV) at the tibia (C) and radius (F). NoT, no treatment; AR, antiresorptive treatment; ROMO, romosozumab treatment; ΔTt.vBMD, percent change in total volumetric bone mineral density; ΔTb.vBMD, percent change in trabecular volumetric bone mineral density; ΔBV/TV, percent change in bone volume to tissue volume

No differences between the patients were seen when excluding patients with variants in LRP6 or Wnt1 (Suppl. Figure 3) with 5 datasets available for NoT and AR and 7 for ROMO.

Biomarkers of bone metabolism

When comparing the metabolic status dependent on the time point including age, BMI, and pre-treatment as a covariable, differences were only seen between the groups comparing the intragroup changes by 6 months (Fig. 3C, D, G, H). No changes for bone formation (OC and bAP) or bone resorption (DPD) were seen comparing baseline and 1-year follow-up (Fig. 3A, E, I).

Fig. 3.

Bone metabolism examination: (A–C) Osteocalcin (OC) levels over time (baseline, 6 months, 12 months) for the NoT (A) (no 6 months data), AR (B), and ROMO (C) groups. (D) Absolute change of OC from baseline to 6 months in AR and ROMO treatment groups. (E–G) Bone-specific alkaline phosphatase (bAP) levels over time for NoT (E) (no 6 months data), AR (F), and ROMO (G). (H) Absolute change of bAP from baseline to 6 months for AR and ROMO treatment group. (I–K) Bone resorption marker levels (DPD/crea) over time for NoT (I) (no 6 months data), AR (J), and ROMO (K). (L) Absolute change of DPD/crea from baseline to 6 months in the AR and ROMO treatment group. NoT, no treatment; AR, antiresorptive treatment; ROMO, romosozumab treatment; OC, osteocalcin; bAP, bone-specific alkaline phosphatase; T0, baseline; T6, measurement at 6 months after baseline; T12, 1 year follow-up measurement; N/A, not assessed; Δ 6mo, absolute difference after 6 months of treatment

For the AR group, no changes in the course of therapy were seen with respect to the bone turnover markers; however, the mean variation decreased for bone formation towards a low-turnover state (Fig. 3B, F, J). When comparing the three time points of assessment for ROMO, a significant increase of serum bAP levels was seen after 6 months (T₀ = 12.10 µg/L ± 6.27 µg/L vs. T₆ = 15.18 µg/L ± 7.08 µg/L, p = 0.011; Fig. 3C), whereas the increase of OC levels did not reach the level of significance (p = 0.056; Fig. 3G). After 1 year of treatment, the bone formation markers (OC and bAP) dropped back to baseline values (OC: p = 0.018; bAP: p = 0.016; Fig. 3C, G). Bone resorption did not change significantly over the course of therapy at the assessed timepoints (Fig. 3K).

Comparing the change in bone formation and bone resorption between AR and ROMO for the time point at 6 months, significantly higher increases from baseline were seen for ROMO than AR for bone formation (OC—AR = −1.89 ± 5.20 vs. ROMO = 4.29 ± 5.20, p = 0.019; bAP—AR = −3.36 ± 8.04 vs. ROMO = 3.08 ± 2.21, p = 0.026; Fig. 3D, H) while no differences were seen for bone resorption (DPD/Crea) (Fig. 3L).

When excluding patients with variants in LRP6 or Wnt1, the pattern of differences between the groups remained the same while exact means changed and the drop of osteocalcin in the ROMO groups was not significant after 12 months (Suppl. Figure 4).

Discussion

The canonical Wnt-β-catenin pathway is of major importance for bone metabolism [22] and defects in this pathway can be linked to decreased bone mineral density and fractures. [5, 23] Since sclerostin inhibition by romoszumab increases bone formation by activating this pathway, it is relevant to study if patients with genetically impaired Wnt signaling can respond to such treatment. [11, 12] In our study of 33 patient-years with genetic variants in LRP5/6 or WNT1 and low BMD, we observed significantly greater BMD gains in those treated with romosozumab compared to patients receiving no treatment or antiresorptive treatments, consistent with preclinical findings from mouse experiments indicating potential benefit for patients with LRP5/6 or WNT1 variants.

While BMI and the frequency of genetic variants were similar across groups, the AR group was significantly older. Since age affects bone mineral density and bone loss [24, 25], we included age as a covariate in further analyses to account for its influence. The AR group also had a higher frequency of vertebral fractures, likely, at least in part, due to their older age and thus longer disease duration. Consequently, the frequency of pre-treatment was also higher in this group, aligning with national treatment guidelines that adjust therapy thresholds based on age and fracture risk; thus, these factors were accounted for as covariates.

One year of romosozumab treatment was effective in significantly increasing the aBMD of hip and spine, compared to the NoT group. In the spine, romosozumab significantly increased aBMD compared to both control and antiresorptive treatment. Similarly, the significant increas in hip aBMD in the ROMO group compared to NoT aligns with this finding. Our results match the results from the literature, reporting more pronounced bone gain in the spine than the hip for antiresorptive treatments in general [26] but also romosozumab in particular [11, 26]. However, only the ROMO group showed significant hip bone gains over 12 months, while the AR group remained unchanged compared to NoT patients. Since the frequency of patients with ERA or ACMG class 2 variants was high in the ROMO group, we added statistical analysis of DXA results excluding such patients which led to the same pattern of significant aBMD gains only in the ROMO group. Taken together, romosozumab outperformed antiresorptive treatment in terms of bone mass gain in 12 months of treatment measured by DXA. Using romosozumab, Cosman et al. report a 13.3% and 6.8% increase in the lumbar spine and total hip [10], respectively, McClung et al. 11.3% and 4.1%  [27], and Ishibashi et al. 16.9% and 4.7% [28], all at the same range like we report on the use of romosozumab in patients with variants in LRP5/6 or WNT1 (15.7% at the lumbar spine and 5.2% at the total hip). Stratification of the ROMO group by prior antiresorptive treatment showed the well-known pattern of greater aBMD gains in treatment-naïve patients, consistent with previous studies [29]. Although the advantage of romosozumab over NoT and AR was preserved, the comparison with AR did not reach significance, likely due to the small subgroup size (n = 5) in the stratified comparison. These preliminary findings support the relevance of genetic testing in early-onset osteoporosis, as treatment-naïve ROMO patients showed markedly higher (about twice as high) aBMD gains than those in the ROMO subgroup with prior antiresorptive treatment (and approximately fivefold higher compared to the AR group).

To further quantify the main factors driving the aBMD change, we performed HR-pQCT measurements for three-dimensional analysis. The results indicate that romosozumab primarily affects the loaded trabecular compartment of the tibia, increasing mineralization and bone mass compared to AR and control (NoT), while no significant influences were seen at the radius. These results suggest a synergistic effect of romosozumab treatment and mechanical loading in LRP5/6- and WNT1-associated low BMD bone disorders. These findings align with literature reporting more pronounced bone gains in the weight-bearing tibia when treating osteoporosis with bisphosphonates [30] and with anabolic drugs like teriparatide [31]. Furthermore, the interaction of sclerostin and its reduction due to loading is already established for humans.[32]

While a case report on a patient with an LRP5 variant (c.1828G>A p.[Gly610Arg]), treated with romosozumab, states 5 new vertebral fractures under treatment [11], none of our patients exhibited a new fracture within the time period of 1-year follow-up. This may be caused, as discussed by the authors, by the severity of osteoporosis of the reported case. However, it may also be attributed to the low incidence of new vertebral fractures of 0.5% to 4.0% in the ARCH and FRAME study. [9, 10]

When comparing the bone turnover data from baseline to 6 months, romosozumab was shown to significantly increase bone formation compared to antiresorptive treatments, as evidenced by elevated bAP levels. Accordingly, OC levels showed a numerical increase after 6 months, approaching significance (p = 0.056). Furthermore, OC levels significantly declined at 12 months compared to 6 months, consistent with the borderline increase observed at 6 months. At 1 year, bone formation returned to baseline, consistent with known romosozumab effects, aligning our findings with studies on postmenopausal osteoporosis. [9, 10] Antiresorptive treatment tended to suppress bone turnover further in this low-turnover cohort, which is potentially disadvantageous. In contrast, romosozumab significantly increased bone metabolism. While no significant changes in bone resorption were observed, the romosozumab group exhibited a trend towards reduced bone resorption at 6 months, consistent with findings from studies on postmenopausal osteoporosis. [9, 10] Taken together, our results suggest that romosozumab induces an anabolic effect associated with activation of the canonical Wnt-β-catenin pathway, even in patients with LRP5/6 or WNT1 variants. This positive effect may be attributed to the reduction of sclerostin’s inhibitory action on the Wnt-β-catenin pathway, highlighting romosozumab’s efficacy in promoting an anabolic response in this patient population.

These findings are promising for clinical applications, given the limited suitability of other bone-specific drugs in such cases. Antiresorptive treatments may prevent bone loss but do not address the underlying pathomechanism and the needed bone gain in particular for long-term treatment. In contrast, romosozumab offers an effective therapeutic option and, in addition to teriparatide, [33] may provide a targeted approach for managing patients with LRP5/6 and WNT1 variants in clinical practice.

These variants are linked to EOOP [5, 6] with variants of LRP5 found in up to 15.4% of young patients with osteoporosis [8]. This knowledge may put our results on the effectiveness of treatment in patients with variants in LRP5/6 or WNT1 into a new context regarding the use of anabolic therapies in EOOP. Of note, reporting of genetic variants differs between laboratories and countries. In particular, established risk alleles of ACMG class 2 variants are often not routinely reported. However, growing evidence for the deleterious effects of specific variants has led to their increasing reporting, although this is not yet standard practice everywhere.

This study has limitations. While all patients carried variants in LRP5/6 or WNT1 respectively, these variants are of different kinds and not homogeneously distributed between the groups. However, all variants (except for ID 9) were judged as relevant for the canonical Wnt-β-catenin pathway. To address possible differences in LRP5/6 or WNT1, we have conducted our statistical analysis additionally on an isolated group of patients with variants in the LRP5 gene. The pattern observed remained the same while some significant differences vanished, most likely also due to a drop in the group size. Additionally, the group specifications differed with respect to the frequency of previous treatment, with more older patients being more often treated with bone-active drugs in accordance with national treatment guidelines. However, we have taken care to minimize the effect of previous treatment and fractures by including them as covariates in statistical testing. Furthermore, half of the patients in the ROMO group received other bone-specific treatments, mostly bisphosphonates, prior to romosozumab. This is known to reduce the effectiveness of romosozumab with regard to BMD gain. [33, 34] Therefore, any influence on the results would more likely lead to an underestimation of the romosozumab effect. Furthermore, although no fractures occurred in any group during the 1-year observation period, the small cohort size limits any conclusions regarding fracture outcomes. Importantly, we included patients with ERA and ACMG class 2 variants, and a large portion of these were in the ROMO group, resulting in some imbalance between groups. Although our sub-analysis excluding ERA and ACMG2 variants still showed significant differences, including these variants (ERA and ACMG2) remains important because they are relatively common (matching the enrichment in our cohort) and known to reduce BMD. For example, the established risk allele LRP5 c.1999C>A, p.(Val667Met), has been reported to compromise BMD and bone matrix properties in osteoporosis.[8] The evaluation and inclusion of Wnt-attenuating variants is important as a proof-of-concept, as romosozumab effectiveness in such cases would directly support individualized therapy for genetically driven early-onset low bone mass.

In conclusion, we have been able to show for the first time that romosozumab is effective in treating low BMD in patients with variants in LRP5/6 or WNT1. This is of particular interest since romosozumab treatment activates bone formation by promoting Wnt signaling, a process being reduced by relevant variants in LRP5/6 or WNT1. Therefore, our results are of paramount importance to translate first results from preclinical studies to clinical application in treating patients with variants in LRP5/6 or WNT1 to effectively increase their BMD and thereby reduce their fracture risk.

Supplementary Information

Below is the link to the electronic supplementary material.

Funding

Open Access funding enabled and organized by Projekt DEAL. This study received funding from UCB within an institutional research grant to RO (ERoLO2020) and from DFG (German Research Foundation) within the Clinical Research Unit 5029 to MA, TS, and RO (project number 517063424).

Data Availability

The dataset analyzed in this study is available from the corresponding author upon reasonable request; however, restrictions may apply due to data privacy regulations.

Declarations

Conflicts of interest

FvB, ND, UK, TS, MA have no conflict of interest. FB has received speaker fees from Alexion, UCB, and Diasorin and has received institutional research grants from UCB and Alexion. RO has served as a speaker and advisory board member for Kyowa Kirin, Inozyme, Ipsen, Pharmacosmos, UCB, and Mereo and has received an institutional research grant from Kyowa Kirin, UCB, and Inozyme.