Introduction

Antiresorptive agents, such as bisphosphonate (BP) and denosumab (Dmab), are widely used as the first-line therapies for patients with osteoporosis or metastatic bone tumors. Since Marx [1] first described patients with bisphosphonate-related osteonecrosis of the jaw (BRONJ) in 2003, case reports of BRONJ have increased. Recently, osteonecrosis among patients receiving Dmab or angiogenesis inhibitors has also been reported; therefore, the term BRONJ was replaced by medication-related osteonecrosis of the jaw (MRONJ), described in the American Association of Oral and Maxillofacial Surgeons (AAOMS) position paper 2014 [2]. Treatment strategies for MRONJ are controversial regarding a nonsurgical versus surgical approach for the first-line therapy. The AAOMS position paper 2014 [2] and Japanese position paper 2012 [3] recommended conservative therapies, such as antibiotic administration, oral rinse, and local washing, for stages 1–2 MRONJ. In contrast, several recent systematic reviews have shown that surgical therapy is more effective than conservative therapy [4,5,6]. In a multicenter retrospective study of 361 patients with MRONJ, we reported that the outcome of surgical therapy was significantly better than that of conservative therapy, minimizing background factor bias by propensity score method; moreover, extensive surgery, which removes necrotic bone with surrounding bone, was superior to conservative surgery, which removes only necrotic bone [7]. However, some patients did not achieve complete healing, even when extensive surgery was performed.

Recently, some investigators have reported that a periosteal reaction was frequently detected by the CT examinations among patients with MRONJ [8,9,10,11], although its clinical significance and the treatment outcome of patients exhibiting a periosteal reaction were not clearly described. We have experienced MRONJ cases with periosteal reactions; these often did not achieve a cure, despite extensive surgery. The purposes of this study were to investigate the relationships between various clinical factors, including CT findings, and treatment outcome of patients with MRONJ undergoing surgical therapy, and to clarify the clinical significance of the periosteal reaction.

Materials and methods

Patient

We enrolled 136 patients who underwent surgery for MRONJ at the Department of Dentistry and Oral Surgery, Kansai Medical University Hospital between 2014 and 2017, or at the Department of Oral and Maxillofacial Surgery, Nagasaki University Hospital between 2011 and 2017. Patients who were followed up for less than 3 months were excluded from the study. When surgery was performed on both the maxilla and the mandible, or when multiple surgeries were performed at different times, each surgery was counted separately. Finally, we performed the following examinations of 164 surgeries.

Variable

The following factors were examined from medical records or CT data: age, gender, site (upper or lower jaw), MRONJ stage [2], trigger of MRONJ (tooth extraction or others), primary disease (osteoporosis or malignant tumor), type of antiresorptive agent (BP or Dmab), duration of administration of antiresorptive agent, discontinuation of antiresorptive agent for more than 3 months, administration of corticosteroid, diabetes, number of leukocytes, serum albumin, creatinine, operation method (conservative or extensive surgery) [7], wound status (primary suture or open), CT findings, and treatment outcome. Regarding CT findings, separation of sequester (absent/present), osteosclerosis (mild/severe), and periosteal reaction (absent/present) were investigated (Fig. 1).

Fig. 1
figure 1

CT findings of MRONJ: a separation of sequester (arrow), and b severe osteosclerosis and periosteal reaction (arrowhead)

Full size image

Osteosclerosis from the alveolar bone to the lower edge of the mandible or maxillary sinus was defined as “severe” osteosclerosis, and partial or no sclerosis was defined as “mild”. Treatment outcome was divided into two types: healing or no healing. Healing meant the absence of all symptoms, including swelling, pain, redness, pus discharge, and bone exposure.

Statistical analysis

All statistical analyzes were performed using the SPSS software (version 24.0; Japan IBM Co., Tokyo, Japan). Correlations between each variable and treatment outcome were analyzed by univariate and multivariate Cox regression. Kaplan–Meier curves were drawn for some categorized data that were significantly correlated with treatment outcome. Then, factors related to periosteal reaction were analyzed by Fisher’s exact test or the Mann–Whitney U test, followed by multivariate logistic regression.

Ethics

This study conformed to the tenets of the Declaration of Helsinki. Ethical approval was obtained from the Institutional Review Boards (IRB) of Kansai University Hospital and Nagasaki University Hospital. This was a retrospective study, and therefore, we published research plan and guaranteed opt-out opportunity by the homepage of our hospital according to the instruction of IRB.

Results

Background factors of the patients are summarized in Table 1. 46 patients were males and 118 were females; the average age of all patients was 74.1 years. The site of MRONJ was upper jaw in 47 patients and lower jaw in 117. Type of antiresorptive agent was BP in 123 patients, and Dmab in 41. The primary disease was osteoporosis in 94 patients and malignant tumor in 70. Surgery was performed without discontinuing the antiresorptive agent for more than 90 days in 85 patients, while 79 patients discontinued medication for more than 90 days before surgery.

Table 1 Background factors of the patients. Values are expressed as median (25–75% tile)
Full size table

1- and 2-year cumulative cure rates of the 164 cases were 75.2% and 83.2%, respectively (Fig. 2). In univariate analysis, 7 variables were significantly correlated with poor treatment outcome: gender (male), separation of sequester (absent), osteosclerosis (severe), periosteal reaction (present), primary disease (malignant tumor), low albuminemia, and drug holiday (absent). In multivariate analysis, 2 variables were independent risk factors that reduced the cure rate: periosteal reaction (present) and primary disease (malignant tumor) (Table 2, Figs. 3, 4). The 1- and 2-year cumulative cure rates in patients with periosteal reaction were 37.9% and 53.4%, which were significantly lower than 85.1% and 90.8% in those without periosteal reaction. Similarly, the 1- and 2-year cumulative cure rates in patients with malignant tumor were 50.1% and 64.9%, which were significantly lower than 93.2% and 96.4% in those with osteoporosis.

Fig. 2
figure 2

Cumulative cure rates of all 164 patients. The X axis indicates the period (days) from the first visit to the day when the treatment outcome was judged

Full size image
Table 2 Factors related to treatment outcome
Full size table
Fig. 3
figure 3

Relationship between primary disease and treatment outcome

Full size image
Fig. 4
figure 4

Relationship between periosteal reaction and treatment outcome

Full size image

Since it was clear that the periosteal reaction is an important factor related to the cure rate, factors related to the periosteal response were investigated. In univariate analysis, three variables were significant factors related to the periosteal reaction: site (lower jaw), osteosclerosis (severe), and primary disease (malignant tumor). In multivariate analysis, four variables were significant factors associated with the periosteal reaction: sex (female), site (lower jaw), osteosclerosis (severe), and primary disease (malignant tumor) (Table 3).

Table 3 Factors related to periosteal reaction
Full size table

Discussion

Recently, it has been reported that surgical therapy significantly increases the cure rate for patients with MRONJ, compared with the conservative therapy [4,5,6]. We have performed extensive surgery for MRONJ, as reported previously [7]. However, some patients did not achieve full recovery, despite extensive surgery; many patients with periosteal reactions were among those who did not achieve full recovery. In this study, CT findings, such as periosteal reaction, separation of sequester, and osteosclerosis, were also included among independent variables; the relationships between various factors and the cure rate were examined. The results showed that the presence of a periosteal reaction, as well as primary disease of malignant tumor, were independent risk factors related to cure rate. In contrast, there was no significant difference in treatment outcome between BP and Dmab. Since all patients with malignant tumors were administered high dose antiresorptive agent and all patients with osteoporosis were administered low-dose antiresorptive agent, it is unknown which is more important as a risk factor related to poor outcome, primary disease, or dosage of antiresorptive agent. However, we think that the dosage of antiresorptive agent may influence the treatment outcome strongly.

Ida et al. [12] examined the CT images of the jawbone of 1142 patients, and revealed that a periosteal reaction was observed in 40% of patients with osteomyelitis; in 91% of those patients, periosteal responses were recognized in one or more layers parallel to the cortical bone. Fatterpekar et al. [8] reported 5 typical radiographic findings of MRONJ: (1) structural alteration of the trabecular bone, (2) cortical erosion, (3) sclerosis, (4) sequestrum, and (5) periosteal new bone. Indeed, the reported rates of periosteal reactions in MRONJ are relatively high: 11/75 (14.7%) [10], 15/34 (44.1%) [11], 3/6 (50%) [8], and 19/28 (67.9%) [9]. Conversely, a periosteal reaction was not found among patients with osteoradionecrosis (ORN) of the jaw, a condition that is partially similar to MRONJ [11, 13]. It may be that the antiresorptive agent mainly acts on osteoclasts and exerts a minimal effect on soft tissue; however, radiation therapy greatly influences both bone and the surrounding soft tissues, including the periosteum, reducing blood flow and promoting scar formation.

In this study, 35 of 164 MRONJ patients (21.3%) exhibited a periosteal reaction. Furthermore, we examined factors related to the occurrence of periosteal reactions and found that 4 variables were significantly correlated with periosteal reaction by multivariate analysis: gender (female), site (lower jaw), primary disease (malignant tumor), and osteosclerosis (severe). In contrast, there was no significant difference in frequency of periosteal reaction between BP and Dmab. It is unclear why females showed a higher percentage of periosteal reactions. In another clinical study, we found that female patients receiving a high dose of antiresorptive agent developed MRONJ significantly more frequently than males, as determined by the multivariate Cox regression analysis (unpublished data). We suspect that differences in male and female jaw structure may be related to the development of periosteal reactions or MRONJ. Cancer patients who were administrated a high dose of antiresorptive agent, or those with more severe osteosclerosis, were more susceptible to periosteal reactions. These findings indicate that a periosteal reaction is not a reactive phenomenon (e.g., similar to osteomyelitis); instead, this reaction represents a more destructive lesion formed because of obstacles to bone remodeling, associated with the apoptosis of osteoclasts. Further research is needed on periosteal reactions in MRONJ patients.

To the best of our knowledge, this is the first study to show that periosteal reactions in MRONJ are associated with a lower cure rate, despite extensive surgery. In the future, it may be necessary to reconsider the surgical method for patients exhibiting periosteal reactions. This study is limited, because it is a retrospective investigation with a small number of patients; therefore, generalization of the results may be difficult. Hence, a more detailed study with a larger number of patients is necessary. In summary, our results indicate that MRONJ with a periosteal reaction is associated with poor treatment outcome despite surgical treatment; thus, surgical methods for such patients should be reconsidered.