This Perspective summarises evidence of the presence of nerves within the skeleton, where they remained silent and invisible until recent technological advances in neurohistology broke the silence. An extensive sensory nerve supply is now visible. It must impact upon our concepts of skeletal physiology and pathology.

Beginning at the beginning, one might ask: do bones have a sensory nerve supply?

Clinically, the answer is an unequivocal yes, although objective proof has been difficult to obtain and the true picture has been slow to emerge over the past 160 years.

In clinical practice, the intense pain caused by trauma, neoplasm or inflammation in bone is proof that sensory nerves exist within bones and joints. Clinical neurologists in the past were prepared to deduce the presence of nerves in bone from clinical observations. For instance, the French neurologist Leriche endorsed Bohler’s observation that a single large injection of local anaesthetic agent between the ends of a fractured bone induces both relaxation of local spasm and anaesthesia, which facilitates relatively painless reduction of the fracture. The deep injection does not alter overlying skin sensitivity. Leriche concluded that this proved the presence of pain fibres within bone [1]. But to show nerves by anatomic dissection was an insuperable task.

Anatomic dissection

The history of the anatomic search for nerves in bone is littered with failures. The problem is a gross discrepancy in the texture of the two tissues: bone is the toughest tissue in the body whereas unmyelinated axons, mere threads of protein, are the most fragile.

Leriche recorded that in 1846, a French anatomist had shown a nerve accompanying the nutrient artery into the femur of a horse [1]. Small branches were given off into the periosteum. Attempts to trace the nerves into bone by gross dissection failed, and the interest of anatomists languished for over a century. Nerves were seen in bone marrow in 1945 [2], but their presence within Haversian canals was less convincing, and remained debatable.

“Many articular twigs are so delicate that they escape detection by gross dissection”, said Gardner in the 1940s [3–5]. Stilwell [6–8] showed “undissectable threads” of nerve fibres entering periosteum from adjacent muscles [6–8].

The scant treatment afforded the subject of the nerve supply of bone in standard texts on the skeletal and nervous systems has been remarked upon in the past [3]. Sherman in 1963 stated “there has been so little interest in the nerve supply of bone that most textbooks of anatomy, even books devoted solely to bone, scarcely mention the subject”, citing eight references [9]. Uncertainty led one author in 1971 [10] to declare that the nerve supply of bone was “unfortunately terra incognita”.

Histology

Originally, histology was as difficult as anatomic dissection. To show delicate unmyelinated axons within bone, especially within the hard cortex, challenged routine histological methods. We know from neurology that in adults, many nerve endings that subserve pain sensation are fine, bare, unencapsulated filaments that branch widely and overlap with their neighbours; they penetrate between other cells and are the most delicate of all sensory terminals. They are easily destroyed by routine formalin fixation and hot paraffin-embedding methods, even in soft tissues. But within bone cortex, the hardest of all tissues, axons are at even greater risk, as routine decalcification simultaneously damages axons. Preparation of one tissue destroys the other.

Routine laboratory methods to fix and stain bones for light microscopy have consistently failed to show axons. This has often been interpreted as absence of nerves in bones, rather than inability to demonstrate nerves in bones. Absence of evidence is not evidence of absence!

Early studies of bone marrow rather than cortex were more successful [2]. Hurrell’s review quotes Variot and Remy who treated marrow with gold chloride, osmic acid or picrocarmine in 1881, and found an extraordinarily rich supply of both myelinated and unmyelinated fibres [11]. Interest lapsed until the 1920s. Based upon the discovery that axoplasm has an affinity for heavy metals (Ag, Au, Pb, Os, Ur), De Castro [12] evolved a method of silver impregnation of nerves, showing them as black threads in decalcified tissues [11, 12]. He described the nerve supply to growing bones, and traced nerves entering bone (with a nutrient artery) to their endings, which were observed to contact the protoplasm of osteoblasts. He believed these nerves to be autonomic. De Castro described both myelinated and non-myelinated fibres within adult bone, terminating on the walls of blood vessels. Hurrell also cited Miskolczy’s study in 1926 of silver-stained periosteum showing nerve fibres entering the bone surface at Haversian canals and other points [11, 13].

Hurrell himself extended these observations in 1937 by tracing nerve fibres into and along the Haversian canals of adult bones into the bone matrix. He described their distribution and endings. Like De Castro, Hurrell noted that some fibres terminated in close relation to osteoblasts. He confirmed that nerve fibres entering bone with the main nutrient artery were destined for the marrow, whereas those to the bone matrix accompanied smaller vessels that entered the Haversian canals here and there over the surface of the bone. These unmyelinated nerve fibres ran parallel to one another within Haversian canals, and maintained this relationship even in the depths of the bone. They branched and overlapped in a free and irregular manner “covering well over half the distance to the next adjacent canal”. Occasionally, the nerves of adjacent canals appeared to communicate. In many places, the fibres were thickly intertwined and gave off terminal twigs. Nerve endings within bone matrix were similar to nerve terminals seen in the periosteum. Hurrell suggested that the nerve fibres he had shown “may be the two ends of a reflex arc governing bone growth and maintenance”. This hypothesis has not been pursued, but it remains a seminal scientific idea.

Kuntz and Richins [2] confirmed the presence of unmyelinated sympathetic fibres anatomically and functionally related to blood vessels, and afferent myelinated fibres partly related to blood vessels [2]. Some fibres were thought to be pain conductors, since the range of their axon calibre was similar to pain fibres elsewhere.

Ralston et al. [14] used amputation specimens to search for nerve endings in human fasciae, tendons, ligaments, periosteum and synovial membranes. They found three main varieties of nerve endings, similar to those already recognised as sensory receptors in the skin, and it seemed likely that these conveyed the same modalities of sensation. Periosteum was particularly richly innervated, especially at the sites of insertion of ligaments, tendons or capsules of joints.

The innervation of the vertebrae and associated structures was studied in man [5] by Gardner et al. [5], and in monkeys [6–8] by Stilwell [6], with similar findings. Somatic sensory fibres were shown to end in longitudinal ligaments, blood vessels, bone marrow, periosteum, joints and dura mater. Furthermore, a one-segment overlap in innervation of the skeletal structures was shown, corresponding to the typical overlap of cutaneous dermatomes.

Hilton’s early work on the innervation of joints (1863) was quoted by Barnett, Davies and MacConaill a century later. In 1961 they confirmed his observations that joint capsules and ligaments receive nerve fibres from neighbouring muscles and nerve trunks [15]. Meantime, Gardner (1944–1950) added much information about joint innervation [3–5]. His serial sections of fetal limbs showed that large joints have a much more extensive nerve supply than is indicated in most text books. As already mentioned, he found many of the articular twigs to be so small and delicate that they defy gross dissection. The nerve supply to fetal joints is from multiple sources, such as the local peripheral nerve trunks and neighbouring muscular branches. The territories of supply of particular nerves within the joints were shown to be more or less constant, with overlap between territories of adjacent nerves. Four types of sensory nerve ending were identified. Innervation of fibrous capsules and ligaments was rich, whereas that of synovial membranes was poor. Samuel in 1952 compared the density of innervation of joints from normal and sympathectomised animals, and concluded that a large proportion of myelinated fibres in the articular nerves are of somatic sensory origin.

Thus, by 1960, axons and nerve endings had been found in abundance in periosteum, cortex, marrow, and in joint capsules, ligaments, tendons and fascia, especially where these meet the bone. Fetal joints are particularly richly innervated, as are vertebrae and paravertebral structures. Sensory terminals are less dense in spongiosa and bone marrow than in cortical bone, and are relatively sparse in muscles.

Modern advances in neurohistopathology

Histology and neurohistopathology evolved rapidly after 1960. Old harsh preparations such as formalin/hot paraffin, which shrink and snap fine protein filaments like axons, have been replaced by new gentle fixatives (e.g. glutaraldehyde), and cold embedding techniques (araldite or Spurrs). Fixative perfusion where appropriate can provide undamaged in vivo quality neural tissue for examination. Araldite enables cutting of ultra-thin sections. Electron microscopy not only detects nerve axons and synaptic junctions, but reveals their internal structure. Electron microscopy (magnification, x > 50,000) reveals axons within undifferentiated embryonic limb buds [16] that were previously [17] said to be without nerves at light microscopy (magnification, x < 800).

Histochemistry evolved in the 1960s [18], immunocytochemistry in the 1980s, and DNA technology in the 1990s. By 1996, Carpenter’s textbook of neuroanatomy contained illustrations of nerves within human bones [19, 20]. Old anatomic barriers gave way before these powerful modern tools and allowed hormones, neurotransmitters and other chemicals to be identified in neural tissue. Labelled molecules can trace the axons in which they are located. Thus previously invisible nerve pathways are being revealed by histochemical mapping. The new science of “chemical neuroanatomy” confirmed the presence of nerves in the skeletal sites discussed above, and has also revealed nerves in other areas such as the epiphyseal plate [21], and, interestingly, within the early uncalcified callus of healing fractures [22].

What role or roles are nerves playing at these sites? To transmit sensations such as pain and proprioception is one obvious answer. But that is not all. Nerves in general, and sensory nerves in particular, have another (again silent!) function, that of neurotrophism, to stimulate growth. In amphibian regeneration, the neurotrophic role of sensory nerves is well established; it has been proved that sensory nerves stimulate cell division, thus promoting the initial phase of limb growth [23]. There is no logical barrier to the hypothesis that sensory nerves in human skeleton perform the same function—to trigger mitosis. Neurotrophism is an accepted concept today in neurology and neuropathology, but is not as well understood in other specialties of medicine.

The combination of nerves in bone plus their neurotrophic property raise some intriguing questions. Are nerves in the epiphyseal plate provoking cell division, hence growth? Do nerves in early fractures trigger mitosis and increase the mass of cells in soft callus to bridge the gap before final ossification? In longstanding sensory neuropathies, does failure of mitosis reduce skeletal repair? Does this explain the fragmentation, non-union and dislocation of neuropathic osteoarthropathy? Are sensory nerves responsible for directing repair and maintenance of the skeleton they supply, as suggested by Hurrell in 1937? If nerves fail with age, injury or disease, does a nerve-dependent stage of the repair process fail? Is this a factor in degeneration? If nerves provoke mitosis, can focal disorder or excitation of a neuron trigger undisciplined focal mitosis and thus initiate a neoplasm?

These are some of the questions around the nerve/bone interface that beg to be explored.