In Shyamalan’s Unbreakable, two men make a perfect pair: David Dunn (Bruce Willis) walks away from a train disaster alone, without a single scratch, while Elijah Price (Samuel L. Jackson) is born with bones that break from an awkward movement. Mr. Glass is a caricature of disease; Dunn, a caricature of health: one is one super-strong pole, the other super-fragile, with nothing in between.
The film came out in 2000, and two years later Yale University’s press office released a statement about a newly discovered mutation causing extraordinarily high bone density in a Connecticut family. Richard Lifton, chair of genetics, said of these people quite literally: if the character in Unbreakable has living prototypes, these are them. The line appeared in an official press release—the real world had edged right up against fantasy.
As soon as “unbreakability” acquired a molecular address, it turned out that people pay for it with their skulls and hearing, while the heart is a separate line item. And that super-dense bone in a living body means disabling one very specific brake.
Bone looks like a material akin to concrete, but it is alive and constantly rebuilding itself. Two types of cells work in tandem: osteoclasts dismantle old tissue, osteoblasts pour in new. This constant cycle is called remodeling, and it explains why the skeleton responds to life: where bone is put under pressure, it adds material; where there is no load, it recedes.
Who decides where to add it is not immediately obvious. Buried deep in the bone are osteocytes, star-shaped cells walled into their own construction site. They act as mechanical-stress sensors: they detect deformation and signal the body to build bone precisely where it is loaded. A mouse whose LRP5 receptor has been disabled only in osteocytes loses the ability to respond to load—its bone stops noticing that it is being loaded.
LRP5 itself is a transmembrane receptor in the Wnt signaling pathway, 1615 amino acids long. And this pathway is inhibited by the protein sclerostin: it binds to the receptor and suppresses bone formation. If you imagine the skeleton as a machine with two pedals, sclerostin is the brake pedal, pressed down by default. It was this brake that was eventually released to treat osteoporosis. But nature released it first—in people who paid for it with their health.
In 1997, a group of researchers described a large pedigree in which bones were unusually dense, with no other outward abnormalities at all. Among the twelve carriers of the trait, the average spinal bone-density Z-score was 5.54; among the other family members, 0.41. A Z-score is a deviation from the population average, and 5.5 sigma turns a person into a statistical anomaly. The trait appeared in people from 18 to 86 years old, meaning this was about peak skeletal mass, accumulated by age twenty, rather than its current state.
The genetic locus landed in a region of chromosome 11, and in 2002 New England Journal of Medicine published the paper with the exact letter: a glycine-to-valine substitution at position 171 (G171V) within the β-propeller domain of the LRP5 receptor. Of the twenty family members who gave blood samples, seven had off-the-charts bone density, nine had completely normal density. One gene, one amino acid, two different people.
When endocrinologist Karl Insogna told Lifton about the family, an almost anecdotal detail emerged: another doctor independently knew of a patient with unbelievably dense bones. Comparing the two cases, the researchers realized they were relatives—two scattered patients had assembled into one pedigree. And the family itself, by all accounts, complained not about fractures (there were none), but about the fact that it is impossible to stay afloat in a pool: their bones are so dense, they sink.
People with this mutation occur, by Lifton’s estimate, about once in a million. But even for them, the phenotype is far from “zero risk and zero side effects.” In 2011, a study compared 258 people with unexplained high bone density, 236 of their relatives, and 58 spouses as controls.
Alongside buoyancy problems came an enlarged lower jaw, extra bone at tendon attachment sites, a broad build, greater height and weight—and with them, jaw pain, bone pain, and headaches. Medicine describes this cluster of features as a mild form of skeletal dysplasia.
G171V delivered a beautiful scientific surprise. Logic suggested that the mutation boosts the signal, that the receptor becomes more aggressive, that bones grow. Reality turned out to be inside out.
G171V disrupts LRP5’s binding to the chaperone Mesd—a guide protein that escorts the receptor to the cell surface. The binding broke down, leaving fewer receptors on the membrane. And yet the bone grew. Less receptor, more bone: intuition trained on “more protein, stronger effect” trips here, while molecular biology shrugs and keeps working.
Worse still, different mutations with the same end result—“dense bone”—build it in different places. Three mouse model lines were compared—with SOST knocked out, with G171V, and with A214V. The first two added bone along the outer, periosteal surface and proved almost indistinguishable. G171V added bone inward, endosteally, and differed from both in most respects. Behind the shared diagnosis of “strong bone” lie different scenarios, and the genetic letter determines which one a person gets.
There is no agreement on load experiments either. One paper reported that mice with G171V show an amplified response to mechanical force. A later comparison of the same mutation with A214V under standardized loading showed that G171V does not differ from normal in periosteal bone formation—only the endosteal response was amplified, meaning the threshold of sensitivity to deformation had fallen. Two studies of the same mutation diverge on their central conclusion. The story has not yet been reduced to a single number.
Alongside the “unbreakable” people, there is an opposite extreme: a disease in which bone never stops growing. This is sclerosteosis—a rare autosomal-recessive disorder in which it is the brake itself, the protein sclerostin, that fails, rather than the receptor.
The overwhelming majority of known patients are Afrikaners in South Africa, descendants of seventeenth-century Dutch settlers. In a closed population, the mutation spread through the founder effect; one study assembled 22 families with common ancestors. A person with two copies of the defective gene eventually acquires a skeleton that grows for life and begins to crowd the nerves inside it.
The auditory ossicles in the middle ear fuse and stop moving—conductive deafness follows. The facial nerve is compressed first in attacks, as in Bell’s palsy, then permanently. The cranial vault grows, intracranial pressure rises, and the optic and trigeminal nerves begin to suffer. People with this disease are exceptionally tall—men are often over 198 centimeters—and their bones do not have the slightest fragility: fractures are not their thing.
In one series of 25 people, there were two spoken of almost with pride: one became an amateur boxing champion; another earned a university degree in metallurgy.
And now the number that closes the subject of “happy unbreakability”: over 38 years of observation, 63 patients were followed, 34 of them died, and 24 deaths were linked to increased intracranial pressure caused by skull growth. The average age at death was 33 years. Several people died suddenly: autopsy revealed acute compression of the medulla oblongata at the foramen magnum. The bone grew until it squeezed the brainstem.
Van Buchem disease is a relative of sclerosteosis, involving the same chromosome and the same gene, but damaged beside it rather than inside it: in most Dutch patients, a stretch of 52 thousand base pairs downstream of the SOST gene is missing, where a regulatory element used to lie.
The protein is formally intact, but it does not work as it should—which means a milder course and fewer patients: around thirty people worldwide, versus more than sixty with sclerosteosis. Two populations, Afrikaners and the Dutch, are historically linked, and both paid for the same malfunction with their skulls.
By the early 2000s, pharma had an unusually rare picture: there was no need to search for a target, nature itself had pointed it out. In people without sclerostin, bone grows like crazy, so what was needed was a drug that would take sclerostin’s job away from it—an antibody that blocks the protein.
In July 2011, Amgen and Belgium’s UCB announced a partnership with NASA: an experiment involving an anti-sclerostin antibody flew on the final shuttle mission, STS-135. Thirty mice, half given the antibody and half a placebo, then after the flight their bone structure, composition, and strength were compared with ground controls.
The logic of the link is simple: in weightlessness, bone loses material precisely because the load vanishes, and sclerostin is the central brake in that process. UCB’s comment in that press release is worth quoting: the entire program grew out of discovering the genetic cause of a rare inherited high-bone-mass disease. The disease of Afrikaners became a blueprint for medicine in an aging world.
The drug was named romosozumab. In the FRAME study of 7,180 women with postmenopausal osteoporosis, it lowered the relative risk of a new vertebral fracture over one year by 73% (0.5% of cases versus 1.8% on placebo), and clinical fractures by 36%. For osteoporosis, these numbers are almost scandalously good.
Then came the second study, ARCH, which compared the drug directly with alendronate in 4,093 women. It beat fractures convincingly: vertebrae by 48%, hip fractures by 38%. At the same time, in the first year of treatment, serious cardiovascular events occurred in 2.5% versus 1.9% in the control group. In July 2017, the FDA sent the company a Complete Response Letter: not a rejection, but a demand to incorporate the ARCH data into the application and sort out the risk-benefit profile. The regulator held its pause for two years.
Approval arrived on April 9, 2019, and with it a bold boxed warning: risk of heart attack, stroke, and cardiovascular death. Treatment cannot be started within a year of a heart attack or stroke, and the drug is discontinued if a vascular catastrophe is suspected. The dosage is 210 milligrams under the skin once a month, for twelve doses total, and no more.
Twelve months is not caution for caution’s sake. Continuing to release the brake indefinitely would mean steering the patient toward sclerosteosis: the same bone that builds up on the spine as protection against fractures would happily build up on the skull and around the nerves. On top of that, the drug’s own lifespan is shortened by cardiac risk. A medicine devised from the skulls of reverse Mr. Glasses is itself limited by that same skull and by one more organ.
If sclerostin both restrains bone and protects the cardiovascular system, the obvious question is: can only one of its functions be touched? In 2022, Nature Communications published a paper that took the protein apart piece by piece. Sclerostin consists of three loops; experiments with shortened variants showed that loop3 is responsible for inhibiting bone formation, while the cardioprotective effect resides in other regions.
That leads to an almost perfect idea: target loop3 precisely with an antibody or aptamer—bone will grow, while sclerostin’s vascular effect will remain, and the heart-attack warning will become unnecessary. In 2025, aptamers targeting loop3 appeared that improved both bones and muscles in mice without vascular side effects. This is still the mouse stage, but the direction has been found with remarkable directness: the protein turned out to have, like a person, a divisible portfolio of duties, and some of them can be fired without touching the rest.
Right here comes fresh bad news. The same target, the same class of antibodies—but a different disease. In osteogenesis imperfecta (“brittle bone disease,” caused by a collagen defect), the antibody setrusumab made it to phase three, and on December 29, 2025, the developer announced that both pivotal phases, Orbit and Cosmic, had failed to reach their primary endpoint—a statistically significant reduction in the rate of clinical fractures.
Bone density rose, but fractures did not become less frequent. The developer announced spending cuts.
The same brake, the same approach, a different basis for the disease—and one failure against one approval. Bone density, it seems, is no longer a reliable surrogate at all: in FRAME it predicted success; in Orbit, it did not. That is a separate story about how a convenient measurable indicator lives its own life and sometimes has nothing to do with what is happening to the patient.
LRP5 also has a 180-degree reversal. In 2019, researchers described a 23-year-old runner who, until age eighteen, ran up to 80 miles a week without a single bone injury. Then stress fractures began, and each time they came with less and less strain: first at 40 miles a week, then 20, then 10. Testing found an A745V variant in his LRP5 gene—in the same Wnt-binding domain where the “unbreakability mutation” sits.
The difference is fundamental. In the Connecticut family, the receptor worked in a way that made bone build itself up; in the runner, the receptor stopped hearing load. The bone did not become diseased at rest—it became deaf to the one signal that makes it stronger. The man broke himself not because he ran too much, but because running stopped making him stronger.
His father, a carrier of the same mutation, did not run and had not the slightest problem with axial skeletal density. The same gene produces both super-strong bone and bone deaf to training; the outcome is decided not by diagnosis, but by lifestyle.
Back to the film. Dunn received invulnerability for nothing, as a gift, and spent two hours searching for his vulnerability; Price received fragility and spent his life searching for an explanation. Mr. Glass and David Dunn sit on the same regulator, just at different settings—and the real people with the G171V mutation turned out to be somewhere in the middle.
Not “half-invulnerable”: the skeleton you really would not want to break came bundled with a jaw that hurts, a head that hurts, and an inability to stay afloat in a swimming pool. The expected superhero bonus turns out, on inspection, to be an imbalance with its own cost: 19.4% neurological manifestations in a review of case histories, 33 years of average lifespan in the rarest extreme.
This cost cannot be calculated as a single number, and that is perhaps the honest conclusion. Medicine can express it in fractures, vascular events, and years of life, but it has no way to add all this up into one bill for a person who has neither a fracture nor a heart attack, only a useless jaw and the inability to swim.
By releasing the brake on bone, nature showed both the medicine and its price: the antibody lasts exactly twelve months, and beyond that begins the territory where strong bone becomes a diagnosis again. Guaranteed invulnerability does not exist in any version—not in Connecticut mutants, not in patients taking romosozumab.