Introduction

A serious side effect of treatment using bisphosphonate (BP) therapy is medication-related osteonecrosis (MRONJ). Bisphosphonates prevent osteoclastic resorption and they are used to treat osteoporosis, but they also reduce the normal repair of bone. The suggested risk factors for MRONJ are many but include poor dental and periodontal health,1 smoking and steroid intake2 and mandibular tooth extraction.3

Diagnosis is based upon the following:

  1. 1

    Current or previous treatment with anti-resorptive or anti-angiogenic agents;

  2. 2

    Exposed bone or bone that can be probed through an intraoral or extraoral fistula(e) in the maxillofacial region that has persisted for more than eight weeks and

  3. 3

    No history of radiation therapy to the jaws or obvious metastatic disease to the jaws.

The consequences of MRONJ are pain and infection, with loss of bone in severe cases. Treatment is usually prolonged, with prevention the preferable option. The incidence of MRONJ in osteoporosis patients receiving oral BP therapy has been reported4 at 0.00038% increasing to 0.21% (21 cases per 10,000) in patients with longer than four years of oral BP exposure.5 A retrospective study analysing patients who had undergone intravenous bisphosphonate therapy (Vidal-Real et al.6) reported a high prevalence of MRONJ of 12.9% (13 cases per 100 patients). This is because intravenous BP is more potent than oral BP. Because BPs persist in the bone, past use can have clinical effects well after treatment has stopped.7 What this highlights is that dose, duration of treatment and potency of the particular bisphosphonate will have an impact on MRONJ risk.

For the dental practitioner, a patient with a history of BP therapy raises concerns about management, particularly when an extraction or other surgery is needed. Knowledge of future bisphosphonate treatment can help minimise MRONJ risk by ensuring that patients are dentally fit and ensuring that any dentoalveolar surgery is undertaken before commencing BP treatment. Although a key diagnostic sign of MRONJ is exposed bone, the American Association of Oral and Maxillofacial Surgeons (AAOMS) classified suspected osteonecrosis with no bone exposure but with nonspecific symptoms or clinical and radiographic abnormalities that may be due to exposure to an anti-resorptive agent as stage 0.8 The radiological findings in stage 0 were listed as:

  • Alveolar bone loss or resorption not attributable to chronic periodontal disease

  • Changes to trabecular pattern – dense woven bone and persistence of unremodelled bone in extraction sockets

  • Regions of osteosclerosis involving the alveolar bone and/or the surrounding basilar bone

  • Thickening/obscuring of periodontal ligament (thickening of the lamina dura and decreased size of the periodontal ligament space).

The evidence for these signs was, however, limited to a single citation related to the last of the four.9 These researchers had reported the pre-operative radiographs of 55 patients with a history of BP therapy and found that 24 of these patients (83%) with MRONJ exhibited periodontal ligament (PDL) widening associated with extracted teeth, whereas only three patients (11%) who healed normally demonstrated PDL widening. The radiological sign of periodontal ligament widening, presented by Fleisher et al.,9 is the opposite of the sign listed by Ruggiero et al.8 Thus, the radiological signs presented by Ruggiero et al.8 do not appear to have any apparent evidence to support them. Other studies have found MRONJ is associated with radiological sclerosis, but Bagan et al.10 found that sclerosis of the jaw was greater in later stage I and II MRONJ patients than in those with stage 0. However, sclerosis was significantly greater only in the later stage III patients (n = 6). Other studies have reported that osteosclerotic changes are not seen in the early stages of MRONJ.11 Periodontal disease has also been associated with MRONJ,12 but this is a nonspecific finding.

Others dispute the existence of stage 0, given the nonspecific nature of the radiological findings, and prefer to classify those with these findings as being 'at risk' of MRONJ.13 Their patients with no bone exposure or symptoms (stage 1 in this newly proposed classification) had osteonecrosis which was identified histologically. Given that only 3 of their 14 cases with non-exposed MRONJ progressed to develop bone exposure, they were unable to determine which characteristics in their patients would be predictive of future bone exposure. Histological diagnosis of osteonecrosis by the general dental practitioner is clearly not feasible.

The purpose of our research was to review the literature and explore whether radiographic imaging could be used to assess the risk of MRONJ, to determine the earliest radiological signs of its onset and to explore the implications for clinical practice. Identification of specific, early radiological features of mandibular MRONJ indicating imminent bone exposure in a patient would indicate a need for frequent review and targeted advice, as well as greater caution in undertaking surgical procedures. Using bone exposure as a criterion for diagnosis of MRONJ, leads to delayed diagnosis and a poor response to treatment.

Materials and methods

Information sources and search strategy: Data sources

A MEDLINE search was performed over the period 1946 to May 2017. In addition, the following databases were also included: Embase via OVID database (searched 1980–2017), Dentistry and Oral Sciences Source (1972–2017), and Proquest Dissertations and Theses (1861–2017). A hand search of the reference lists of the selected papers was undertaken to identify additional papers identifying early bone changes in MRONJ.

Eligibility criteria

The widest possible search was undertaken to include case series, literature reviews, clinical guidelines and clinical trials.

Inclusion criteria:

  • Studies on humans receiving or having received bisphosphonate therapy who had, or subsequently developed, MRONJ

  • No study design restrictions (case series, literature reviews, clinical guidelines and clinical trials all eligible)

  • Studies including descriptions of radiological findings and radiographic images.

Exclusion criteria:

  • Conference proceedings

  • Papers describing the bone changes resulting from tooth extraction and other later stages

  • Late stage MRONJ, ie patients presenting with bone exposure or MRONJ preceded or initiated by extractions

  • Animal studies.

Data selection

The searches involved search terms associated with the themes of 'osteonecrosis', 'bisphosphonates' and 'radiographs'. We identified 786 papers that contained references to radiographs or described imaging findings in MRONJ. A second more focused screening was performed where the full papers were examined by three appraisers (HD, JG-C and JB) applying the inclusion and exclusion criteria. Seventy-four articles were included. Of these, only 19 described early radiographic bone changes in MRONJ and gave a description or image of the earliest jaw bone changes. The quality assessment was performed on these 19 studies.

Quality appraisal of the published evidence: Data synthesis

In order to assess the original data on which recommendations had been made, the 19 selected studies were analysed using a modified 'Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies' (see https://www.nhlbi.nih.gov/health-pro/guidelines/in-develop/cardiovascular-risk-reduction/tools/cohort).

The questions referred to those factors which would be important in determining a diagnosis of MRONJ. They were:

  1. 1

    Was the research question or objective in this paper clearly stated?

  2. 2

    Did they mention MRONJ classification using American College of Oral and Maxillofacial Surgeons, 2007?

  3. 3

    Were inclusion and exclusion criteria for being in the study pre-specified and applied uniformly to all participants?

  4. 4

    Did it mention the average duration of bisphosphonate exposure for MRONJ patients?

  5. 5

    Were radiographic assessments performed with the radiologist blinded to bisphosphonate exposure?

  6. 6

    Steroid use/comorbidities recorded.

A score of 1 or 0 was given for each domain. The maximum score of 6 indicated that the paper satisfied all the above criteria. However, it must be acknowledged that each domain may not have same importance as the others and this is obscured when only the overall score is considered.

The methodology for this systematic review was registered with PROSPERO: International prospective register of systematic reviews in 2017. (See: https://www.crd.york.ac.uk/PROSPERO/index.php – 'Early radiological signs of MRONJ as predictors of ongoing disease', ID registration number CRD42017056028).

Quality assessment

From the 19 papers describing early radiographic changes, the radiological findings found in early MRONJ were tabulated and summarised (Table 1). Using the 'Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies', relevant publications were independently assessed for quality by two observers with a third acting as the arbitrator when disagreements were present. In order to assess the quality of original data on which published recommendations had been made, the 19 studies were analysed further in terms of their quality.

Table 1 Summary table of early radiographic signs of MRONJ

Results

The search produced 1,228 articles from the study search but 786 papers were included from the radiographic search (see Fig. 1). The abstracts were examined to exclude conference proceedings (as these were rarely available) and duplications. Seven hundred and eighty-six published papers were found which described an imaging modality in the diagnosis of MRONJ; this usually included a radiographic description. The full paper was obtained for these references using the inclusion and exclusion criteria and further analysed for early radiographic findings in MRONJ. Nineteen papers were identified.

Figure 1
figure 1

PRISMA flow diagram

Detecting the earliest radiographic signs of MRONJ (Table 1)

Of the 19 papers identified in our study describing the earliest radiographic changes in MRONJ, 15 cited focal osteosclerosis or osteolysis, nine mentioned a markedly thickened lamina dura, five papers a widened periodontal ligament space and one a thickened inferior mandibular cortical width.

Whereas nine of the authors mentioned thickening of the lamina dura as a diagnostic feature, Park et al.14 describe loss of lamina dura in one of their five cases reviewed. Hinchy et al.15 described a dense, sclerotic lamina dura but not all teeth were affected by the continuing osteoblastic activity. The appearance can be localised to only a few affected teeth.

The later stages of MRONJ, involving bone exposure, fracture and osteosclerosis, are easily recognised. However, clearer criteria are necessary to enable a better recognition of the earlier stages. Krishnan et al.16 described a case series of six patients with different stages of MRONJ, the paper by Park et al.14 described five patients and that of Rinchuse et al.17 described two patients. No comparison with a control group was used in these studies. The publications by Hawarth and Webb18 and Arce et al.19 were review articles and therefore contained no original primary data. Koth20 also reviewed the non-exposed stage of MRONJ, but concluded that the radiographic parameters determining early diagnosis were missing.

The radiographic appearance of early stage MRONJ is nonspecific. The article by Bedgoni et al.21 stated that osteosclerosis characterises the early stages of MRONJ, quoting an article by Saia et al.22 to support their statement. However, this latter article described the occurrence of MRONJ in patients that had previously undergone tooth extraction.

In the paper by Matsuo et al.,23 five of their ten cases with the earliest features of MRONJ (stage 0) had both dental infection and radiographic abnormalities, but the superimposed infection may have complicated the radiographic signs. Similarly, in the patient group with the non-exposed variant of MRONJ described by Fedele et al.,24 half (51%) had sinus tracts and superimposed infection. However, the 2014 position paper by the American Association of Oral and Maxillofacial Surgeons8 does allow fistulae in the stage 0 or unexposed bone variant of MRONJ. In the paper by Cardosa et al.25 osteolysis was the only consistent finding in early stage 0 MRONJ, but only four patients were studied. Thickening of the lamina dura was found in only two of these patients. No control group was used in this study, so it is unclear whether the radiographic findings were normal anatomical variation. Bagan et al.26 compared the area of mandible affected by MRONJ with the contralateral area and with healthy controls and found an increased severity of sclerosis in the MRONJ-affected bone. There was no blinding of the examiners, so there was a high risk of bias in selecting regions of interest in the study. There were only nine patients with MRONJ stage 0 included in the study so it remains unclear whether the sclerosis was primarily related to other stages of MRONJ.

Quality assessment

All of the studies were at risk of bias, by failing to score positively on at least one criterion (see Table 2). Most studies examined radiographs from patients where the radiologists were diagnosing investigations of unknown swellings or indeterminate painful areas, and blinding would not be appropriate in these situations. The radiologists were not blinded in 16 of the 19 papers, which presented a strong risk of bias. Only 7 of the 19 papers described other co-morbidities/steroid use which might be important in modifying the radiographic bone appearance in MRONJ.

Table 2 Quality scores of papers for each domain

Quality assessment of the highest scoring publications (Table 2)

Hutchinson et al.27

The authors used a self-selected group of 2,159 patients taking oral bisphosphonates who reported complications after invasive dental procedures, such as tooth extractions. Thirty individuals with pain, delayed healing and infection underwent radiographic imaging, but only ten of them had radiographic evidence of osteosclerosis (and/or osteolysis). This would indicate that even in symptomatic patients, the prevalence of radiographic signs is low. Subtle signs of osteosclerosis were present in the control group. Conventional CT scanning techniques were used in most cases. This technique is not recommended for monitoring of patients in general dental practice taking oral bisphosphonates because of the cost, limited availability and high radiation dose to patients.

Krishnan et al.16

This study described the early magnetic resonance imaging and CT findings in a small group of patients (n = 6) with MRONJ. These imaging techniques would not be routinely available to the general dental practitioner. The experimental group contained patients at different stages of MRONJ and no control group was used. Most patients had superimposed infection.

Rocha et al.28

Although Rocha28 showed that there is a statistical difference in the mean number of radiographic abnormalities between those who received intravenous bisphosphonate (n = 190) and those who did not (n = 74), there still remains a high prevalence of 'abnormalities' in the normal population. Of the total number of radiographic abnormalities found, about 28% were present in the control group. No necrotic bone was exposed in 83% of the MRONJ patients. The radiographic abnormalities in the earliest stages of MRONJ are unspecific and not diagnostic.

Zaman et al.29

Zaman et al.29 examined pairs of radiographs taken an average of 19 months apart. The standardisation methodology they used for obtaining radiographs of the identical mandibular molar region was not explained or justified. They found that, even using digital subtraction software, sclerotic change in the alveolar bone was significantly distinguishable from controls in only two of the seven subjects receiving intravenous bisphosphonates. The mean sclerotic change in these two subjects amounted to only 14–15% of the pixels in their region of interest, and whether this would be detectable in clinical practice by a general dental practitioner remains unknown.

Iwata et al.30

Iwata et al.30 found that there was a significantly greater mandibular cortical width in MRONJ patients compared to controls or those symptomless patients taking anti-resorptive agents. However, there was a poor correlation in the measurements between observers with large standard deviations observed within the groups. Only two patients were examined with stage 0 MRONJ disease.

In summary, the evidence describing the earliest radiological signs of MRONJ are difficult to distinguish from normal radiological variation. Those radiological signs described in early MRONJ are often complicated by superimposed infection.

Discussion

The American Association of Oral and Maxillofacial Surgeons published a position paper in 2014, which described the 'at risk' patient and a further stage 0 category where there was no evidence of necrotic bone but where the patient complained of bone or sinus pain or radiographic changes were detected (such as osteosclerois and thickening of the lamina dura).8 They also classified periapical or periodontal fistulae in the stage 0 category, so papers describing these early bone changes may be complicated by the superimposed infection.

One can conclude that the earliest radiographic signs of osteonecrosis closely resemble abnormalities similar to early chronic osteomyelitis. Radiolucent and sclerotic lesions have been reported due to osteoclastic activity resulting in osteolysis and abnormal osteoblastic activity causing osteosclerosis. Given the similarity to osteomyelitis, infection may be an important causative factor in those patients who progress to MRONJ. In those with pre-existing dental infections, such as active periodontal disease or periapical disease, progression to MRONJ is more likely to occur.23 This has important implications for the general dental practitioner, as these patients require regular oral prophylaxis, oral hygiene instruction and recall. Careful surgical technique is also needed with surgery to avoid excessive soft tissue trauma. In addition, the elimination of denture trauma can reduce the likelihood that infection will ensue.

The general dental practitioner may be faced with a diagnostic dilemma. Identification of specific, early radiological features of mandibular MRONJ indicating imminent bone exposure in a patient would indicate a need for frequent review and referral, but is this possible? The radiographic bone changes that occur before oral infection, and resulting solely from using bisphosphonates, may result in a nonspecific mixture of bone erosion and sclerosis. Bone changes may take the appearance of thickening of the lamina dura, widening of the periodontal ligament, focal sclerosis, cortical erosion and radiolucent areas (Table. 1), but these are nonspecific and can occur because of 'everyday' dental disease and even normal anatomical variation in bone pattern. Unfortunately, most of the MRONJ studies describe the consequences of bone exposure that occur following the population of the wound by oral bacteria, inevitably resulting in superimposed bone infection and possibly sequestration. Matsuo et al.23 illustrate a radiograph from a patient with stage 0 MRONJ with a swelling and a discharge of pus.

Without doubt, practitioners with MRONJ patients exhibiting unexplained pain, mandibular osteosclerosis or radiolucency should suspect that the patient has an increased risk of impending bone exposure. However, the evidence is weak. But with the advent of computer software that can accurately measure mandibular cortical width31 and a detailed previously published knowledge of the normal widths in the population,32 it should be possible to analyse radiographs to automatically identify those patients at increased risk of MRONJ.

Most of the 19 reviewed papers used conventional radiography (panoramic or intra-oral) but 11 also used cross-sectional imaging either exclusively or in combination with conventional radiography. There is some evidence that cross-sectional imaging (CT or CBCT) improves the detection of early radiological change. The evidence is, however, insufficient to support a recommendation for routine cross-sectional imaging of patients who are at risk of MRONJ, particularly as this would incur additional financial costs and radiation exposure in an asymptomatic patient.

We assessed five papers as providing good quality evidence using our modified quality assessment tool.16,27,28,29,30 This is not to undervalue the other publications, where clinical observation is important, especially where formal experimentation has ethical limitations. Of these five papers, one by Hutchinson et al.27 reported on the radiographic findings of bisphosphonate-treated patients with stage 0 disease, but no healthy control group was used. In addition, only ten patients with stage 0 MRONJ were described. Of the other four papers, radiographic assessments were not performed with the radiologist blinded to the bisphosphonate exposure,16 or steroid use and other comorbidities were not recorded.28,29,30 The study by Krishnan et al.16 had a small sample size (n = 6).

Increased vigilance is required where the risk of infection is greater; depending on factors such as the bisphosphonate potency, duration of therapy (more than 2 years), increased age and local risk factors. Local risk factors include any dentoalveolar surgery, poor oral hygiene and even poorly fitting removable dental prostheses. MRONJ is often detected late only when clinical symptoms become obvious. The radiographic findings in the earliest stages include the differential diagnosis of chronic osteomyelitis and metastatic bone cancer. If metastatic bone cancer is suspected, then a bone biopsy may be indicated but this will increase the risk of MRONJ. Several studies have reported that up to 50% of patients with stage 0 MRONJ progress to bone exposure.24,33 At present, it is not possible, using radiographic investigations, to predict those patients who will develop increasing disease severity.

In conclusion, in those patients at risk of bone exposure with MRONJ, insufficient information is present in the literature to allow the general dental practitioner to identify the radiographic features indicating imminent bone exposure. However, automated computer vision methods are available to assist in diagnosis of other bone disease, such as osteoporosis.

Conclusion

Using bone exposure as a criterion for diagnosis of MRONJ, leads to delayed diagnosis and a poor response to treatment. A well-designed prospective study is needed to determine the earliest radiological features that can be reliably determined. We aim to stimulate interest, debate and discussion to enable earlier diagnosis and more effective treatment.