Since Marx (1) first reported bisphosphonate (BP)-related osteonecrosis of the jaw (BRONJ), several cases have been reported worldwide (2). Anti-receptor activator of nuclear factor κ-B ligand (RANKL) antibodies, including denosumab or antiangiogenic agents, are also known to cause ONJ (2). Accordingly, the American Association of Oral and Maxillofacial Surgeons (AAOMS) changed the defined term BRONJ to medication-related ONJ (MRONJ) in 2014 (2). AAOMS basically recommends conservative treatment for the majority of MRONJ cases, excluding those of stage 3 disease or those exhibiting a well-defined sequestrum. However, the optimal treatment strategy remains controversial. In recent years, previous studies described the effectiveness of extensive surgery in the early stages of MRONJ (3,4). Our previous study also observed good outcomes of extensive surgery for MRONJ (5).

The histopathological findings of BRONJ have been evaluated in several previous studies (6–8), which revealed that the viable osteoclasts exhibit the feature of multinucleated giant cells. These giant osteoclasts are detached from the smooth bone surface and have lost their resorptive function (6–8). Furthermore, these abnormal osteoclasts may persist on the site (9). Denosumab-related ONJ was first reported in 2010 (10), with only a few previous reports regarding this being published since then (11–18). However, reports of ONJ caused by single application of denosumab are scarce. In addition, none of the above mentioned reports have described the histopathological features of this condition. Even if histopathological analysis was performed, viable osteoclasts and other bone remodeling-related cells, including osteoblasts and osteocytes were not described, since only the sequestrum, which has no viable cells, was surgically resected, according to the AAOMS recommendations for MRONJ (18).

The present study described the clinical and histopathological features of ONJ caused by single application of denosumab in two patients who were subsequently treated by extensive surgery.

The resected surgical segment was subjected to histopathological analysis, which revealed nearly identical findings in each specimen. Hematoxylin and eosin (HE) staining revealed sequestrum without viable cells, granulation tissue and viable bone with inflammation (Fig. 3A–D). In the necrotic bone, granulation tissue, containing neutrophils, lymphocytes and plasma cells, was observed. A bacterial mass was attached to the sequestrum, which revealed no osteoclasts, osteoblasts and osteocytes, with completely necrotic bone and empty osteocytic lacunae as characteristic findings (Fig. 3A and B). By contrast, in the viable bone, osteocytic lacunae, including viable osteocytes were observed, indicating the viability of the bone in this region (Fig. 3C and D). Bone resorption cavities were observed on the surface. However, the surrounding osteoclasts exhibited specific features, including being few in number despite the presence of bone resorption cavities. In case 2 in particular, barely any osteoclasts were observed. Furthermore, the existing osteoclasts had very few nuclei, giving a morphologically immature appearance. It was occasionally difficult to identify osteoclasts using HE staining. Immunohistochemistry using cathepsin K, which is regarded as a marker for osteoclasts (19), confirmed the findings of the HE staining. The cathepsin K-positive cells with very few nuclei that existed along, or were detached from, the bone surface were observed, predominantly in the case 1 specimen (Fig. 3E and F).

Reportedly, the risk ratio for MRONJ in patients who receive anti-RANKL inhibitors for cancer treatment ranges between 0.7 and 1.9% (20,21), which is equivalent to that reported for patients who receive zoledronate treatment (22,23). Although certain previous case reports on denosumab-related ONJ have been published (10–18), only one described the histopathological features (18), which revealed complete osteonecrosis with empty osteocytic lacunae and no osteocytes, osteoblasts or osteoclasts. The authors concluded that the histological features of denosumab-associated ONJ were similar to those of BP-associated ONJ. However, the authors only evaluated the necrotic sequestrum, since only this portion was surgically resected, according to the AAOMS recommendations for MRONJ (2). By contrast, in the present report, viable bone with cells responsible for bone remodeling was observed since extensive surgery was performed, which involves resection of not only the sequestrum, but also viable, inflamed bone (3–5). Furthermore, osteocytic lacunae with osteocytes were clearly observed, permitting distinction between viable and necrotic bone. The osteoclasts in this viable region were few and revealed a decrease in the number of nuclei. It was hypothesized that the maturation of immature osteoclasts around the sequestrum in patients treated with denosumab is inhibited. Although a few immature osteoclasts were observed in the case 1 specimen, there were barely any in the case 2 specimen. This was probably a result of case 2 being older than case 1; therefore, the bone turnover rate was higher in case 1 compared with in case 2. Weinstein et al (6) performed a transiliac bone biopsy in patients who received BP treatment and suggested that this treatment is associated with an increase in the number of osteoclasts, which include distinct, giant, hypernucleated and detached osteoclasts that are undergoing protracted apoptosis. Additionally, Cho et al (8) observed a notable number of osteoclasts, which were detached from the bony trabeculae in patients with stage 3 BRONJ, treated by partial mandibulectomy. These results suggested that osteoclasts generally remain viable following BP treatment. Therefore, evaluation of viable bone containing cells, including osteoclasts, is important for assessing the effects of denosumab on bone cells.

Denosumab is a fully human monoclonal antibody that targets RANKL (24). Generally, RANKL activates osteoclast differentiation by binding to RANK, a single transmembrane receptor expressed in osteoclast lineage cells. RANKL inhibition prevents the fusion of monocytes and macrophages to form multinucleated osteoclasts. Denosumab prevents RANKL from binding to RANK and subsequently inhibits osteoclast formation, function and survival. By contrast, BPs bind to bone minerals and these are taken up by mature osteoclasts at sites of bone resorption. These osteoclasts subsequently lose their resorptive function and persist (9). Denosumab was found to result in nearly complete disappearance of osteoclasts in an ovariectomized human-RANKL mouse model (25). From this perspective, the histopathological findings of denosumab-related ONJ in the two cases reported in the present study are acceptable. Denosumab is considered to exhibit a faster offset of action compared with BP, and its effects on bone remodeling are mostly diminished within 6 months of treatment cessation (26). Denosumab must be discontinued prior to surgery in patients with MRONJ, if systemic conditions permit. However, the effects of discontinuation remain to be elucidated. The necrotic sequestrum and inflamed bone formed in patients with ONJ are different from normal bone, and it remains unclear how they exhibit the identical metabolism. In the present study, the serum calcium level was decreased or maintained low by denosumab treatment, indicating the effects of this drug on bone metabolism. Since it was impossible to discontinue denosumab for a long duration in these cases, the present study performed extensive surgery 1 month following discontinuation. The prognosis of each case was good during the postoperative follow-up.

In conclusion, the present study described the clinical and histopathological features of denosumab-related ONJ in two patients. More data should be collected to describe the bone metabolism at the site of denosumab-related ONJ and to elucidate the mechanisms underlying denosumab-related ONJ.