Bone Morphogenetic Proteins (BMPs)

Where BMP sits. BMP is one input into a process, not the process. The biology it drives is bone healing and the cycle underneath it is bone remodeling, acting through the osteoblast lineage; the other molecules in the same conversation are on growth factors in bone healing. Clinically it is used where that process has failed or is being asked to do too much - nonunion, open tibial shaft fractures and lumbar fusion - and its signature complication, bone where bone should not be, is heterotopic ossification.

What they are. Bone morphogenetic proteins are secreted signalling molecules in the transforming growth factor-beta (TGF-beta) superfamily that induce bone and cartilage formation. More than 20 subtypes have been identified; only BMP-2 and BMP-7 have regulatory approval.

How they were found. Marshall Urist showed in 1965 that demineralised bone matrix implanted subcutaneously in rabbits induced ectopic bone formation, the founding demonstration of osteoinduction. Wozney and colleagues cloned the proteins responsible in 1988, and recombinant BMP-2 and BMP-7 followed from that.

What BMP adds. Autograft is osteogenic and osteoconductive, allograft osteoconductive only. BMP supplies the third property: the signal that recruits host mesenchymal stem cells and drives them into the osteoblast lineage. It can give that property to a synthetic scaffold which has none of its own.

Where it is used. rhBMP-2 (InFuse, Medtronic) holds FDA approval; rhBMP-7, also called OP-1 (Stryker), holds only a humanitarian device exemption for recalcitrant long bone nonunion. Off-label use in posterior spinal fusion, fracture nonunion and revision arthroplasty is common but controversial.

Placing BMP against the alternatives is a standard viva opening. The discriminators are the three biological properties - osteogenesis (living cells that form bone directly), osteoconduction (a scaffold) and osteoinduction (signals that recruit and differentiate host stem cells).

The molecule. BMPs are secreted polypeptide growth factors, and the active ligand is a homodimer of two identical chains joined by disulfide bonds through a cysteine-rich domain. BMP-2 is 114 amino acids per chain and 26 kDa as a dimer; BMP-7 is 139 amino acids per chain and 35 kDa. A monomer has no biological activity, because it takes the dimer to bind the type I and type II receptors on the cell surface.

Family resemblance. BMPs belong to the TGF-beta superfamily and share its cysteine knot motif and its conserved receptor-binding domains.

The receptor complex. Signalling needs a heterotetrameric complex - two type II and two type I receptors - which assembles when BMP binds. The type II receptor is constitutively active and phosphorylates the type I receptor; the activated type I receptor is the kinase that phosphorylates Smad1/5/8.

Where BMP is found. BMP is abundant in bone matrix, stored there and released during resorption, and is expressed during fracture healing with a peak at three weeks. It is also present in cartilage, tendon and periosteum, at low levels in most adult soft tissues, and at high levels in embryonic development.

The family. More than 20 BMPs have been identified, divided into subgroups on structure and function. One subgroup induces bone, one opposes it, and the rest belong to tendon, joint and embryonic development.

BMP-3 is the exception. Also called osteogenin, it is a negative regulator of bone formation rather than an inducer of it.

The two that reached the clinic. BMP-2 and BMP-7 were the two developed for clinical use, although neither is the most potent osteoinductor in the laboratory. Of the pair, BMP-2 has the greater osteoinductive potency and the higher complication rate.

Receptor preference. Each group of ligands binds a characteristic set of type I receptors.

The route in. A BMP dimer assembles its receptors at the cell surface, the type II receptor phosphorylates the type I receptor, and from there the signal divides into a canonical Smad arm and a non-canonical MAPK arm.

View larger

View larger

Radiographs follow the healing, taken at 6, 12 and 24 weeks, looking for bridging bone, for the fusion mass in spinal cases, and for ectopic bone appearing where it should not be.

CT is the standard for judging fusion. Fine-cut images with three-dimensional reconstruction show the trabecular bridging and cortical continuity that plain films cannot resolve, and it is the study that settles complex anatomy.

Clinically, healing shows as pain settling at the fracture or fusion site, stability on examination and the return of weight-bearing. Afterwards the watch list is soft-tissue swelling, a neurological examination for the effects of ectopic bone, and dysphagia screening in the cervical spine - where BMP should not have been used in the first place.

Laboratory markers are research tools rather than routine clinical tests.

Histology, equally a research tool, means bone biopsy with immunohistochemistry for Runx2 and osterix, assessment of osteoblast differentiation markers, quantification of new bone formation and evaluation of how far the carrier has degraded.

View larger

Before using BMP the workup is the ordinary one: routine preoperative bloods (FBC, UEC, coagulation), nutritional assessment with vitamin D and calcium where bone healing is a concern, and inflammatory markers (CRP, ESR, WCC) where infection is. No BMP-specific test is required.

The decision starts with whether the problem genuinely needs enhanced bone formation, and then with whether the indication is an approved one or an off-label one that has to be justified.

Algorithm

View larger

When BMP is reasonable.

• ALIF at L4-S1 - the approved spinal indication
• Severe open tibial fractures, Gustilo IIIA/B
• Recalcitrant nonunion after autograft has failed
• High-risk fusion: smokers, revision surgery

Patient factors that push the decision that way are inadequate autograft quantity, concern about donor-site morbidity, and metabolic bone disease.

When not to. BMP is not used at all in these situations:

• Anterior cervical spine - not approved, airway swelling
• Active malignancy
• Pregnancy or breastfeeding
• Known hypersensitivity to BMP or to bovine collagen

Active infection, a recent cancer history and immunocompromise are relative contraindications.

Dose follows indication, and the off-label doses are the higher ones.

The carrier is not packaging. The absorbable collagen sponge is a bovine type I collagen matrix that absorbs the BMP solution and delivers it slowly, degrading over 4-8 weeks and acting as an osteoconductive scaffold while it does. It is the only FDA-approved carrier for BMP-2.

Preparation. Improper preparation reduces osteoinductive efficacy, so the sequence is prescribed.

Combining BMP with graft is off-label but common. Allograft chips soaked in BMP solution, or a BMP sponge layered with allograft, add structural support and osteoconduction to the osteoinductive signal; demineralised bone matrix and synthetic ceramics (TCP, hydroxyapatite) are used the same way, and the combination may reduce the BMP dose needed. Local autograft from the decompression, mixed with the BMP sponge, brings osteogenic cells as well and spares an iliac crest harvest.

Supraphysiologic dosing and off-label use are what lie behind the recognised complications and the arguments that still surround them.

Retrograde ejaculation is the BMP-specific complication that the technique viva and the FDA MAUDE signal both raise and the complications table omits. Antegrade ejaculation is lost, semen passing retrograde into the bladder, through dysfunction of the superior hypogastric plexus - the sympathetic network draped over the L5-S1 disc and sacral promontory, exactly where the anterior approach dissects and where the BMP sponge sits.

The association. Both independent re-analyses in this topic's evidence base report higher retrograde ejaculation and urogenital problems with rhBMP-2 in ALIF than with autograft (Carragee 2011; Fu 2013, where the increase was present but not statistically significant). The proposed mechanism is a BMP-driven local inflammatory reaction injuring the adjacent autonomic plexus, additive to the dissection itself.

Reducing the risk means blunt rather than monopolar dissection over the promontory, a left-sided retroperitoneal approach, and containing the sponge within the cage away from the plexus - or a posterior approach, or autograft, in a young man who wishes to father children.

How much of this is off-label. Studies suggest over 50% of BMP use is off-label - posterior fusion, cervical spine, nonunion - and the absence of approval for those indications creates medico-legal exposure. Evidence quality across them is variable, and industry-sponsored trials dominate the literature.

The malignancy signal. The independent YODA / Annals individual-patient-data meta-analysis (Fu 2013) found a statistically significant increase in new cancers at 24 months (RR 3.45, 95% CI 1.98-6.00), and the concern was dose-dependent - most prominent in the high-dose rhBMP-2 ("Amplify", around 40 mg) posterolateral-fusion programme, which the FDA did not approve.

Why it is not accepted as a proven causal risk. The event numbers were small and the tumour types heterogeneous, no single cancer dominating; follow-up was short; and harm ascertainment in the source industry trials was poor. Larger and longer administrative-database and registry studies at standard clinical doses have not confirmed a consistent causal increase.

The practical position. There is no proven causal cancer risk at standard doses, but the concern is biologically plausible, BMPs being mitogenic, and the signal was dose-dependent. So rhBMP-2 is avoided in active malignancy and recent cancer, and high off-label doses are discouraged. That nuanced synthesis, rather than a flat "safe" or "carcinogenic", is the exam-ready answer.

Follow-up after BMP watches two things at once: the healing, and the complications that belong to the BMP rather than to the operation.

Returning to activity after spinal fusion means walking on day one, light activities at 6 weeks, desk work at 6-8 weeks, physical work at 3-6 months and contact sport at 6-12 months, with fusion confirmed on CT before unrestricted activity.

After a long bone or nonunion, weight-bearing is protected for 6-12 weeks, progressive from 12-16 weeks, and unrestricted only after radiographic union, with serial radiographs and the callus guiding the progression.

The evidence pendulum. The early industry-sponsored randomised trials were favourable - Burkus in ALIF, Govender in open tibial fractures. The independent re-analyses that followed, Carragee in 2011 and the YODA individual-patient-data analysis by Fu in 2013, found markedly higher harms and no proven advantage over autograft. That tension is the heart of any BMP discussion.

Fusion rates. Reported ALIF fusion with rhBMP-2 runs 94-100%, against 85-90% for autograft. Off-label posterolateral fusion runs 85-95% and varies between series, with mixed evidence against autograft and higher complication rates.

What patients report. ODI improvement, VAS pain scores and return-to-work times are equivalent to autograft, and satisfaction is high wherever fusion is achieved. At 5-10 years the fusion is maintained, adjacent segment disease is no different, and the cancer concern has not been borne out as a causal risk at standard doses.

Reoperation overall is similar to autograft. It may be higher where ectopic bone has to be excised, and revision for pseudarthrosis is no different.

Cost is the argument against. BMP costs thousands of dollars per dose, the saving from avoiding a harvest is offset by that price and by the cost of managing complications, and QALY analyses show marginal benefit at best.

rhBMP-2 is the most widely used osteobiologic in orthopaedic surgery, controversy over safety and off-label use notwithstanding.

Spinal FusionFracture NonunionOther Applications

The approved indication is single-level anterior lumbar interbody fusion, L4-S1. In the pivotal randomised trial fusion was achieved in 94.5% of rhBMP-2 patients against 88.7% with iliac crest autograft at 24 months, with equivalent clinical outcomes on ODI and VAS, a shorter operation and no donor site to hurt afterwards.

Posterolateral fusion is off-label. It is not an approved approach, uses higher doses per level, and carries the complications of BMP laid in an open paraspinal gutter from which it can migrate towards neural elements: ectopic bone compressing nerve roots, osteolysis with screw loosening, and radiculitis. Evidence on fusion rates against autograft is mixed.

Cervical use is different in kind - the one off-label application with a documented risk to life.

BMP-2 Is Not for the Anterior Cervical Spine

The FDA has warned specifically against rhBMP-2 in anterior cervical fusion, which is not an approved indication. The enclosed prevertebral space amplifies inflammatory swelling into airway oedema, haematoma, dysphagia and dysphonia; the reported consequences include prolonged intubation, emergency tracheostomy and death.

The approved indication is the acute open tibial shaft fracture treated with an intramedullary nail - the fresh fracture, not the established nonunion.

The BESTT trial randomised 450 open tibial shaft fractures to standard care or rhBMP-2 on a collagen sponge. The 1.5 mg/mL group had a 44% reduction in the risk of secondary intervention for delayed or nonunion (RR 0.56, 95% CI 0.40-0.78), faster fracture and wound healing, and significantly fewer infections in Gustilo-Anderson type-III injuries.

Off-label in established nonunion, union rates of 75-85% are reported in recalcitrant tibial nonunion, comparable to autograft but without the donor site. The situations it is reached for:

• Atrophic nonunion, where the biological healing potential is what is missing
• Critical-size defects after debridement for infection or tumour
• Revision surgery after failed bone grafting

rhBMP-7 holds its humanitarian device exemption in exactly this territory, recalcitrant long bone nonunion after prior treatment has failed, but availability is limited and BMP-2 has largely taken over.

BMP in Nonunion

BMP is an adjunct, not a replacement for surgical principles. Adequate debridement, mechanical stability and vascularity are essential; an osteoinductive stimulus cannot compensate for instability, infection or inadequate fixation.

Revision arthroplasty with bone loss - acetabular defects in revision hip replacement, tibial defects in revision knee replacement - rests on case series only, with concerns about osteolysis and heterotopic ossification.

Osteotomy healing, in periacetabular and high tibial osteotomy, has a theoretical benefit and no randomised evidence.

Bone tumour reconstruction uses allograft-BMP composites for segmental defects and to accelerate distraction osteogenesis, on very limited clinical data.

Oral and maxillofacial use has an approval of its own: sinus floor augmentation and ridge preservation before implants, alongside mandibular reconstruction.

Off-Label Use Considerations

Off-label BMP use is common but controversial. Many indications have no randomised evidence, off-label dosing carries higher complication rates, and a complication carries medico-legal consequences. Informed consent is essential, and autograft remains the gold standard unless it is contraindicated.

View larger