Cancer patients point to pain as their most important and one of the most distressing symptoms (1). The dissemination of cancer to bone can lead to significant pain, and up to 85% of terminally ill cancer patients suffer from cancer-induced bone pain (2–4).

Mouse models of cancer-induced bone pain have demonstrated substantial sprouting of both sensory and sympathetic fibers giving rise to an increased nerve fiber density as cancer cells invade and proliferate in the bone tissue (5, 6). The nerve sprouting is observed in periosteum, mineralized bone and bone marrow (7, 8). In healthy mouse bone, sensory and sympathetic nerve fibers display a distinct morphology. Sensory fibers present with a long and linear morphology, while sympathetic nerve fibers have a spiral-like appearance and wrap around associated blood vessels (6, 9). Upon the invasion of cancer cells, the morphology changes to a disorganized appearance where sensory and sympathetic nerve fibers are intermingled (6). It has been proposed that such pathological reorganization can render the nerve fibers highly sensitive to movement of the tumor-bearing limb, reflecting movement-evoked pain (10), which is a key symptom of metastatic bone disease (11, 12). Also, formation of neuroma-like structure has been described in animal models of cancer-induced bone pain (6), and it is speculated that these structures may lead to spontaneous discharge contributing to spontaneous breakthrough pain, which has a strong impact on the patients' quality of life (12, 13).

Although the first reports of cancer-induced sprouting in animal models of cancer-induced bone pain occurred more than 15 years ago (8, 14), it is yet unknown how these findings translate to humans with metastatic bone disease. The present study examines the histological appearance of nerve profiles in bone biopsies from breast cancer patients with metastatic bone disease compared to bone biopsies from healthy controls. The density of nerve profiles and their association with vascular structures and cancer cells are evaluated in the bone marrow.

Here we provide first evidence that neuronal sprouting and reorganization occur in human breast-cancer infiltrated bone tissue. Firstly, we observe a significant increase in the nerve profile density of cancer-infiltrated human bone marrow tissue, and secondly, we find that these nerve profiles are less associated with vascular structures.

Overall, we found a five-fold increase in the nerve profile density of cancer-infiltrated human bone marrow compared to the healthy control bone marrow. Of note, the level of nerve profile density in the control biopsies was comparable to what was previously described in the bone marrow of patients with primary hyperparathyroidism (15). A recent study, investigating innervation in axial and appendicular human cadaveric bone, demonstrated that the numbers of nerve fibers significantly declined per year of age in elderly individuals (age range 66–99). Extrapolating this finding to younger individuals, it is possible that the increase in nerve profile density in MBD is even higher in magnitude due to the lower median age of the healthy controls (16). The subgroup analysis revealed a significant increase in the nerve profile density in both groups of archived cancer-infiltrated bone marrow. Interestingly, anti-cancer treatment within two years of obtaining the bone marrow biopsy did not seem to affect the nerve density. No significant change in nerve profile density was identified in the cancer-infiltrated biopsies collected during joint replacement surgeries. These biopsies were isolated from various locations in femoral, humeral, and pelvic bones and thus represent a more heterogeneous composition than the biopsies isolated from the iliac crest. Additionally, these biopsies were decalcified in EDTA for 56–71 days, and although EDTA in general is believed to preserve epitopes, the specific effect on PGP9.5 is unknown. That said, in mouse models of cancer-induced bone pain, the bone marrow space has been described to become devoid of nerve fibers as it gets filled with tumor (8), and it is tempting to speculate that the non-significant increase in nerve profile density seen in bone biopsies from patients undergoing joint replacement surgery is due to a beginning denervation. In support, we do not observe any nerve profiles in necrotic areas, visually observed as areas without apparent cell nuclei. In addition, six of nine biopsies were isolated from the intramedullary space of the femoral or humeral shaft, which is the area where cancer cells are seeded in the mouse models of cancer-induced bone pain.

Pathological sprouting of nerve fibers is not only described in models of cancer-induced bone pain, but also in the periosteum of geriatric mice with painful arthritic joints and in painful non-healed bone fractures (17, 18). In the model of cancer-induced bone pain, nerve sprouting is in part believed to drive the pain, as blockade of the nerve sprouting can attenuate the development of late-stage pain related behaviors (6, 19). Nerve growth factor (NGF), released by cancer cells and associated stromal cells, is thought to be a key player in the nerve sprouting as therapies that block the NGF/tropomyosin receptor kinase A (TrkA) pathway in early stages block the sprouting and effectively attenuate the development of cancer-induced bone pain (6, 19). Interestingly, results from a placebo-controlled proof-of concept study and a non-controlled open-label study with tanezumab, a humanized monoclonal antibody that blocks the binding of NGF to TrkA, suggest that the NGF-driven pathology also applies to patients with metastatic bone disease (20). In this study, we did not have access to pain reports from patients; therefore it was not possible to investigate a potential relationship between bone pain and the increase in nerve profile density. However, the significant increase in nerve profile density found in our study combined with the effect of tanezumab in clinical studies suggest that the same mechanism may in part drive the pain in human metastatic bone disease.

Myelinated and unmyelinated fibers have been demonstrated to be mostly associated with vascular structures in human bones (15, 16, 21). In a recent study, examining the association of nerve profiles to vasculature in the bone marrow of iliac crest biopsies isolated from patients with primary hyperparathyroidism, we found that more than 90% of the nerve profiles were associated with vasculature (15). This is in accordance with the present finding in the control biopsies where the majority of nerve profiles were associated with vasculature. In the bone marrow of cancer-infiltrated bone tissue, we found a significantly lower percentage of nerve profiles associated with vasculature compared to healthy bone marrow tissue. This may indicate that structural changes occur in the neuronal network leading to a disorganization of the nerves. Of note, the association of nerve profiles to vascular structures was investigated from the innervation perspective only. Thus, the present data do not provide information on specific types of vasculature or any information on the total coverage of vascular structures. As a qualitative observation, the vasculature of the cancer-infiltrated bone marrow appeared denser and more disorganized as compared to the vasculature of the healthy bone tissue (not shown). Following inoculation of prostate cancer cells, the vasculature of mouse femurs has been reported irregular and disorganized (5). In human breast cancer, hyper-vascularization is generally observed and has been shown to play a role in cancer development, invasion, and metastasis (22). The increased and irregular vascular network visually observed in this study could indicate that cancer cells have the same effect on the vasculature in human bone.

We did not identify any correlation between tumor burden and nerve profile density suggesting that the sprouting is independent of close contact to cancer cells, but instead may be driven by humoral factors. A few cancer-infiltrated bone biopsies would present with no or very few cancer cells. Due to the method of area estimation, the latter would end up with a tumor burden of 0.0%. However, in these biopsies an increased density of neuronal profiles ranging from 19.5 to 120.2 profiles/mm² was still evident_. Of note, since the analysis is limited to a thin, 2-dimensional section, it is only possible to identify the 2-dimensional nerve profiles and their in-plane association with vasculature and cancer cells. In reality, nerves are elongated 3-dimensional structures that could be associated with vasculature and cancer cells above or below the section plane, meaning that the reported percentages underestimate the 3-dimensional associations with vasculature and cancer cells. In addition, the bone biopsies were isolated from different skeletal sites. Most biopsies were from the iliac crest, but the biopsies from the joint replacement surgeries came from femoral, humeral, and pelvic bones. It has recently been demonstrated that there are differences within the nerve profile density of different bone compartments, i.e. periosteum, cortical pores and bone marrow (15), but it is yet unknown if there are differences between individual skeletal sites, including the weight bearing long bones and less strained bones, such as the iliac crest. Lastly, it is not possible to determine how progressed the cancer is, as patients with metastatic disease per definition are in the most advanced stage of breast cancer (stage IV).

In conclusion, the increased nerve profile density and decreased association of nerve profiles to vasculature found in cancer-infiltrated biopsies strongly suggest that neuronal sprouting and reorganization occur in human cancer-infiltrated bone tissue.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by National Committee on Health Research Ethics, Denmark, and The Regional Committee, The Capital Region of Denmark. The patients/participants provided their written informed consent to participate in this study. The National Committee on Health Research Ethics granted permission to obtain archived biopsies without prior written consent.

MSø, TG-S, and MP enrolled patients and obtained surgical biopsies. Archived samples from Pathological Biobank, OUH were collected by EB and CA. Bone biopsies from healthy controls were collected by CE and TA. RH and CA designed the experiments. RH, AJ, and IK acquired and analyzed the data. EB and MSa acquired data. RH, A-MH, and CA wrote the manuscript. A-MH supervised the study. All authors contributed to the article and approved the final version.

The study was supported by the Danish Private Funding Body Advokat Bent Thorbergs Fond and by European Union's Horizon 2020 Research and Innovation Programme under the Marie Sklodowska-Curie grant agreement No. 642720.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.