Odanacatib LOFT trial (McClung 2019)
McClung MR et al. (LOFT Investigators) · Lancet Diabetes Endocrinol 7(12):899–911 · 2019 · DOI: 10.1016/S2213-8587(19)30346-8 · PMID 31676222
NCT00529373 (ClinicalTrials.gov); EudraCT 2007-002693-66. Citation count: 161 (FWCI 20.6, citation_percentile 100 per OpenAlex 2026-06-02). Closed access; no PMC OA copy.
TL;DR
The LOFT trial (Long-term Odanacatib Fracture Trial) was a Phase 3 double-blind placebo-controlled RCT of odanacatib — a selective cathepsin K inhibitor — in 16,071 postmenopausal women with osteoporosis. Odanacatib produced large, statistically significant reductions in morphometric vertebral (HR 0.46), hip (HR 0.53), and nonvertebral (HR 0.77) fractures. However, a statistically significant 37% increase in stroke risk emerged in the combined LOFT+Extension analysis (HR 1.37, 95% CI 1.10–1.71, p=0.0051), without a corresponding atrial fibrillation signal — indicating the cerebrovascular risk was not AF-mediated. Merck discontinued the program in 2016 before regulatory submission. This trial is the canonical cautionary tale for cathepsin K inhibition as an aging-context bone intervention: strong fracture efficacy paired with an unexpected, mechanism-unexplained cardiovascular liability.
Background and rationale
osteoporosis affects an estimated 200 million women worldwide and is a major driver of fracture-related morbidity and mortality in older adults. Existing anti-resorptive agents (bisphosphonates, denosumab) reduce osteoclast number or survival. odanacatib (MK-0822) offered a mechanistically distinct approach: selective, reversible inhibition of cathepsin-k — the cysteine protease responsible for degrading Type I collagen in the resorption lacuna — without directly killing osteoclasts. The hypothesis was that preserving osteoclast viability would maintain osteoclast-derived anabolic coupling signals (including TGF-β release from demineralised matrix), potentially producing better bone quality than bisphosphonates. Phase 2 data showed dose-dependent BMD increases, motivating the Phase 3 programme.
Study design
| Feature | Detail |
|---|---|
| Design | Multicentre, randomised, double-blind, placebo-controlled |
| Sites | 388 clinical sites; 40 countries |
| Participants | Postmenopausal women aged ≥65 years with osteoporosis (BMD T-score criteria) |
| Randomisation | 1:1 odanacatib 50 mg once weekly vs. matched placebo |
| Blinding | Double-blind (participants, investigators, outcome assessors) |
| Registration | NCT00529373; EudraCT 2007-002693-66 |
| n (randomised) | 16,071 1 |
| LOFT follow-up | Median 36.5 months 1 |
| LOFT Extension | Eligible completers invited to continue; total combined median 47.6 months 1 |
| Primary endpoint | Morphometric (radiographic) vertebral fracture incidence |
| Key secondary endpoints | Hip fracture; nonvertebral fracture; cardiovascular safety (including stroke, AF) |
| Sponsor | Merck & Co. |
Efficacy results
All fracture outcomes below are confirmed from the PubMed structured abstract (PMID 31676222 efetch) 1.
LOFT (median 36.5 months)
| Endpoint | Odanacatib | Placebo | HR (95% CI) | p |
|---|---|---|---|---|
| Morphometric vertebral fracture | 3.7% | 7.8% | 0.46 (0.40–0.53) | <0.0001 |
| Hip fracture | 0.8% | 1.6% | 0.53 (0.39–0.71) | <0.0001 |
| Nonvertebral fracture | 5.1% | 6.7% | 0.77 (0.68–0.87) | <0.0001 |
LOFT + Extension (median 47.6 months combined)
| Endpoint | Odanacatib | Placebo | HR (95% CI) |
|---|---|---|---|
| Morphometric vertebral fracture | 4.9% | 9.6% | 0.48 (0.42–0.55) |
| Hip fracture | 1.1% | 2.0% | 0.52 (0.40–0.67) |
| Nonvertebral fracture | 6.4% | 8.4% | 0.74 (0.66–0.83) |
Fracture efficacy was consistent across skeletal sites and sustained across the extension period. The relative risk reduction for morphometric vertebral fracture (~54%) and hip fracture (~47%) is among the largest observed for any osteoporosis intervention at this time point.
| Dimension | Status |
|---|---|
| Pathway conserved in humans? | yes — human Phase 3 RCT; no extrapolation required |
| Phenotype conserved in humans? | yes — fracture is a hard clinical endpoint |
| Replicated in humans? | yes for efficacy; no independent Phase 3 replication (programme discontinued) |
Safety: the stroke signal
The cardiovascular safety findings are the primary reason this trial matters to the wiki.
Stroke
| Dataset | Odanacatib | Placebo | HR (95% CI) | p |
|---|---|---|---|---|
| LOFT alone | 1.7% (136/8,043) | 1.3% (104/8,028) | 1.32 (1.02–1.70) | 0.034 |
| LOFT + Extension | 2.3% (187/8,043) | 1.7% (137/8,028) | 1.37 (1.10–1.71) | 0.0051 |
Both the LOFT-alone stroke HR (1.32, 1.02–1.70, p=0.034) and the combined LOFT+Extension HR (1.37, 1.10–1.71, p=0.0051) are confirmed from the PubMed structured abstract (PMID 31676222 efetch) 1. The abstract provides event counts for the LOFT-alone period explicitly.
Atrial fibrillation / flutter
LOFT: HR 1.18 (95% CI 0.90–1.55), p=0.24 (not statistically significant) 1. The absence of an AF signal is mechanistically significant: stroke risk was not driven by odanacatib-induced new-onset atrial fibrillation (a known concern for some anti-resorptive agents). The mechanism of the cerebrovascular signal therefore remains unexplained and unresolved. no-mechanism
Mechanistic hypothesis for stroke
Cathepsin K is expressed in vascular smooth muscle cells and macrophages in atherosclerotic plaques; preclinical work suggested cathepsin K degrades extracellular matrix components that stabilize plaque integrity. Inhibiting cathepsin K at vascular sites might impair fibrous cap stabilisation, potentially increasing plaque vulnerability or thrombus propagation. This hypothesis is speculative and was not tested mechanistically in LOFT. no-mechanism
Discontinuation decision
Merck voluntarily halted the odanacatib development programme in September 2016 — approximately three years before this primary report was published — based on an interim unblinded cardiovascular safety review. The regulatory submission was never filed. The published 2019 paper presents the full dataset, confirming the interim signal. This sequence (discontinuation before publication) is unusual and means the formal trial report post-dates the public health decision by three years.
The discontinuation was based on overall benefit-risk balance: the stroke hazard was judged sufficient to outweigh the fracture efficacy, particularly given the availability of approved alternatives (bisphosphonates, denosumab) and the lack of a mechanistic explanation that would allow the cardiovascular liability to be engineered away.
Interpretation and wiki role
Why this trial matters for aging research
- Cathepsin K inhibition is a validated resorption mechanism — the fracture efficacy data are among the strongest ever generated for postmenopausal osteoporosis, confirming the target hypothesis.
- The stroke signal is a cautionary tale for selective enzyme inhibition in aged tissues — cathepsin K is expressed widely (bone, vasculature, lung, kidney, brain), and blocking it with high selectivity may produce on-target effects at non-skeletal sites. This is a canonical example of the problem of pleiotropic enzyme targets in aging-adjacent drug development.
- AF-independence of stroke risk — rules out the mechanism seen with early-generation bisphosphonates (which were associated with AF); the cerebrovascular signal from odanacatib is genuinely unexplained and biologically interesting.
- Next-generation cathepsin K inhibitors — the LOFT data motivate tissue-selective or bone-targeted delivery strategies. The underlying target (collagen degradation in resorption lacunae) remains valid.
Cross-references
| Page | Relationship |
|---|---|
| osteoclasts | Primary citing page; odanacatib section with LOFT stroke citation at [^mcclung2019loft] |
| osteoporosis | Clinical phenotype being treated; LOFT is the pivotal efficacy dataset |
| odanacatib | Implicit stub — compound page for MK-0822 (cathepsin K inhibitor); this study is its primary Phase 3 anchor |
| cathepsin-k | Implicit stub — protein page; molecular target of odanacatib; expressed in osteoclasts and vasculature |
| osteoblasts | Osteoblast coupling (TGF-β from demineralised matrix) is part of the mechanistic rationale for odanacatib vs bisphosphonates |
| chronic-inflammation | Inflammaging as a driver of osteoclast activation and osteoporosis background risk |
Limitations and gaps
Study limitations (internal):
- All participants were postmenopausal women ≥65 years; findings do not extend to men or younger premenopausal populations.
- The LOFT Extension was an observational continuation of the blinded trial — enrolling completers introduces survivor bias; Extension results should be interpreted with that caveat.
- Cardiovascular adjudication methodology and blinding integrity for safety endpoints are not confirmed from the structured abstract; would require full-text methods section. no-fulltext-access
Wiki gaps:
no-fulltext-access — Lancet Diabetes Endocrinol is closed-access; local PDF not downloaded (archive status: not_oa). All load-bearing quantitative claims (fracture HRs, stroke HRs with event counts, AF HR with CI) have been confirmed from the PubMed structured abstract (efetch). The following items remain unverifiable without full text: cardiovascular adjudication methodology and committee blinding; Extension survivor-bias handling (whether eligible completers were pre-specified or convenience-enrolled); subgroup breakdowns for stroke by geography/baseline CV risk.
no-mechanism — The biological mechanism linking cathepsin K inhibition to increased stroke risk is unresolved. Vascular cathepsin K plaque-stabilisation hypothesis is speculative.
needs-replication — No independent Phase 3 replication exists because the programme was discontinued. The fracture efficacy data stand; the stroke signal also stands but its mechanism and generalisability to other cathepsin K inhibitor structures are unknown.
Footnotes
Footnotes
-
mcclung-2019-loft (this page) · doi:10.1016/S2213-8587(19)30346-8 · PMID 31676222 · McClung MR et al. (LOFT Investigators) · rct · Lancet Diabetes Endocrinol 2019;7(12):899–911 · n=16,071 postmenopausal women ≥65 yr; odanacatib 50 mg/week vs placebo; median follow-up 36.5 months (LOFT), 47.6 months combined; fracture efficacy HRs, both stroke HRs (LOFT-alone 1.32 [1.02–1.70] and combined 1.37 [1.10–1.71]), and AF HR (1.18 [0.90–1.55]) all confirmed from PubMed structured abstract (efetch, PMID 31676222) · closed-access full text; load-bearing quantitative claims verified against structured abstract ↩ ↩2 ↩3 ↩4 ↩5 ↩6