Osteoclastic expression of higher-level regulators NFATc1 and BCL6 in medication-related osteonecrosis of the jaw secondary to bisphosphonate therapy: a comparison with osteoradionecrosis and osteomyelitis

Patients and tissue collection

The present study analyzed retrospectively jaw bone samples (n = 70) from 70 patients (MRONJ (BP): n = 30, ORN: n = 15, OM: n = 15, control: n = 10). Patient cohorts and respective samples from this study are identical to those in our previous study [12]. All patients were treated in the Department of Oral and Maxillofacial Surgery of the University Hospital Erlangen between 2007 and 2015. The analyzed jaw bone samples were gathered intraoperatively as part of routine histopathological diagnostics. The control group consisted of patients with histopathologically inconspicuous jaw bone samples (n = 10) which were obtained during dental surgery procedures of teeth with no signs of local infection. The jaw bone probes were fixed in 4% formalin immediately after surgical sampling. Histopathological analysis was performed by the Department of Pathology of the University Hospital Erlangen. Specific clinical disease criteria were checked by the review of medical records and radiographs.

Inclusion criteria for MRONJ (BP) samples were: (1) histopathological confirmation of MRONJ (BP). (2) Evidence of more than 8 weeks of exposed jaw bone. (3) Documented bisphosphonate therapy. (4) No radiotherapy. (5) No therapy with denosumab, bevacizumab, pazopanib, sunitinib, mTOR inhibitors and sorafenib.

Inclusion criteria for OM samples were: (1) histopathological confirmation of OM with evidence of chronic inflammatory processes in the jaw bone. (2) No bisphosphonate therapy. (3) No radiotherapy. Patients with primary chronic OM (non-bacterial cause) were excluded.

Inclusion criteria for ORN samples were: (1) Evidence of devitalized and exposed jaw bone in a previously irradiated field in the absence of local neoplastic processes. (2) No bisphosphonate therapy.

Inclusion criteria for control samples were: (1) no histopathologic evidence of bone disease. (2) No bisphosphonate or local radiation therapy. (3) No medications significantly affecting jaw bone homeostasis. (4) No intraoral inflammation. (5) No relevant periodontitis. (6) No local malignancies. (7) No relevant systemic diseases (e.g., osteoporosis).

Immunohistochemical staining

All formalin-fixed samples underwent decalcification and were embedded in paraffin before being sliced in 3-μm sections using a microtome (RM2165, Leica, Nussloch, Germany). Special microscope slides with improved adhesion were used (SUPERFROST ULTRA PLUS, Gerhard Menzel GmbH, Braunschweig, Germany). The sections underwent dewaxing in xylene and rehydration in graded propanol and distilled water before staining.

Hematoxylin and eosin staining (H&E) was carried out according to standard protocols.

Immunohistochemistry was performed using an automated staining device (Autostainer plus, DakoCytomation, Dako Deutschland GmbH, Hamburg, Germany). Antigen retrieval consisted of section treatment with ethylenediaminetetraacetic acid (EDTA) (dilution 1:100, PMB4-125, Antigen Retrieval Buffer 4, Spring Bioscience, CA, USA) at 66.7 °C for 5 h. The reduction of background staining artifacts was achieved by performing peroxidase-blocking for 5 min (S2023, DAKO REAL, Peroxidase-Blocking Solution, Dako Deutschland GmbH, Hamburg, Germany).

EnVision Detection System Peroxidase/diaminobenzidine (DAB), Rabbit/Mouse (K5007 HRP/DAB+, Dako Deutschland GmbH, Hamburg, Germany) was used as staining kit. This kit provided a conjugated dextran and a DAB+ chromogen that were used for visualizing the antibody-marked proteins. Hematoxylin was used for nuclear counterstaining (CS700, Dako Deutschland GmbH, Hamburg, Germany). Positive and negative controls were included in each staining series.

Quantitative immunohistochemical analysis

All stained histological sections were scanned and digitalized completely in cooperation with the Institute of Pathology of the University Hospital Erlangen using a Pannoramic 250 Flash III Scanner (3DHISTECH Kft., Budapest, Hungary). Before scanning, the sections were quality-checked under a bright-field microscope (Axioskop, Zeiss, Jena, Germany; at a magnification of 100×–400×). The analysis of the digitalized sections was done via virtual microscopy using CaseViewer version 2.2 (3DHISTECH Kft., Budapest, Hungary). Figure 1 illustrates the method of “whole slide imaging”, that was used in this study. Two visual fields per virtualized section were set within areas with a high probability for the presence of osteoclasts (bone trabeculae, subperiosteal bone, endosteal structures and connective tissue directly adjacent to the bone). If the visual field size exceeded the total section size, only one visual field was used. Areas of pure necrosis were omitted for analysis. Within the visual fields, non-bony medullary areas were marked (regions of interest = ROIs) (Fig. 1c). Any cell counting occurred only within ROIs. Cells were considered osteoclasts if they met the following morphological criteria: (1). Multinuclearity (at least two nuclei). (2). Large cell body (larger than two fused mononuclear cells). (3). Direct contact with bone or proximity to bone. (4). No proximity to granulomatous foci or foreign particles. Area determination within the visual fields was done with Pannoramic Viewer, whereas cell counting was performed with ImageJ (Rasband, W.S., ImageJ, US National Institutes of Health, Bethesda, Maryland, USA, http://imagej.nih.gov/ij/, 1997–2014). Section analysis was conducted by two medical students familiar with tissue morphology, IHC-methods and analysis. These students were blinded to the origin of the specimens. Regarding cell counting, inter-individual differences were checked and did not exceed 10%.

Statistical analysis

Statistical analysis was conducted after consultation with the Department of Medical Informatics, Biometry and Epidemiology (IMBE) of the Friedrich-Alexander University Erlangen-Nürnberg.

For quantitative analysis, not only the ratios of osteoclasts to ROI were determined, but also the respective labeling indices (positive osteoclasts of a ROI/all osteoclasts of a ROI) and the ratio of nuclear BCL6+ osteoclasts to cytoplasmic BCL6+ osteoclasts. Results are expressed as the minimum, maximum, average, median, interquartile range (IQR) and standard deviation (SD). Box plot diagrams visualize the respective values.

The Kolmogorov–Smirnov test was used for normal distribution testing. The Mann–Whitney U test was used for statistical hypothesis testing. P-values ≤ 0.05 were considered statistically significant. SPSS 22 (SPSS, IBM, New York, USA) was used for statistics.

Osteoclastic NFATc1 expression patterns

NFATc1-positive (NFATc1+) cells showed a brown nuclear staining (Fig. 2a). NFATc1-expression occurred not only in osteoclasts, but also in mononuclear cells (Fig. 2a, b). The expression of NFATc1 in osteoclasts was restricted to the cell nuclei. NFATc1+ osteoclasts were found in specimens from all groups. The amount of NFATc1+ osteoclasts per ROI were found to be significantly higher in MRONJ (BP) specimens (median: 25.1 osteoclasts/mm²) than in ORN (median: 1.8 osteoclasts/mm²; p < 0.022), OM (median: 0.0 osteoclasts/mm²; p < 0.004), and control specimens (median: 0.0 osteoclasts/mm²; p < 0.001) (Table 2; Fig. 2c). The analysis of the osteoclast NFATc1 labeling indices revealed that MRONJ (BP) specimens (median: 57.3%) featured significantly higher indices than OM (median: 0.0%; p < 0.019) and control specimens (median: 0.0%; p < 0.001) (Table 2; Fig. 2d). However, no significant differences between the osteoclast NFATc1 labeling indices of MRONJ (BP) (median: 57.3%) and ORN specimens (median: 24.8%; p < 0.240) were found. For detailed data, see Table 2.

	Group	Min	Max	Mean	Median	IQR	SD
Anti-NFATc1 staining
Nuclear expression
NFATc1 + osteoclasts per ROI (osteoclasts/mm2)	MRONJ (BP)	0.0	171.7	31.2	25.1	45.9	36.1
	OM	0.0	31.4	7.0	0.0	14.1	9.8
	ORN	0.0	36.5	11.4	1.8	22.3	13.8
	CONTROL	0.0	5.8	0.8	0.0	0.6	1.9
Labeling index (%/100)	MRONJ (BP)	0	1	0.569	0.573	0.790	0.381
	OM	0	1	0.304	0	0.650	0.396
	ORN	0	1	0.424	0.248	1	0.440
	CONTROL	0	0.420	0.065	0	0.060	0.144
Anti-BCL6 staining
Cytoplasmic expression
BCL6 + osteoclasts per ROI (osteoclasts/mm2)	MRONJ (BP)	0.0	113.1	33.2	29.1	29.4	25.9
	OM	0.0	43.4	11.4	9.2	16.5	12.4
	ORN	0.0	141.3	19.7	5.6	13.6	35.8
	CONTROL	0.0	9.2	1.7	0.0	3.5	3.1
Labeling index (%/100)	MRONJ (BP)	0	1	0.760	0.860	0.380	0.300
	OM	0	1	0.445	0.444	0.860	0.391
	ORN	0	1	0.714	0.839	0.560	0.349
	CONTROL	0	0.370	0.100	0	0.300	0.163
Nuclear expression
BCL6 + osteoclasts per ROI (osteoclasts/mm2)	MRONJ (BP)	0	73.4	17.6	13.1	21.4	16.9
	OM	0	11.4	2.9	2.1	4.1	3.7
	ORN	0	92.5	10.8	2.5	10.4	23.4
	CONTROL	0	9.3	1.9	0	4.0	3.3
Labeling index (%/100)	MRONJ (BP)	0	1	0.469	0.483	0.58	0.329
	OM	0	0.72	0.160	0.075	0.29	0.218
	ORN	0	1	0.416	0.282	0.46	0.342
	CONTROL	0	0.56	0.109	0	0.27	0.189
Nuclear BCL6 + osteoclasts to cytoplasmic BCL6 + osteoclasts (%/100)	MRONJ (BP)	0	1	0.513	0.503	0.433	0.308
	OM	0	0.450	0.164	0.127	0.355	0.177
	ORN	0	0.778	0.460	0.500	0.333	0.237
	CONTROL	0	1	0.243	0	0.550	0.424

Min minimum, Max maximum, IQR interquartile range, SD standard deviation, ROI region of interest, MRONJ (BP) medication-related osteonecrosis of the jaw secondary to bisphosphonate therapy, OM osteomyelitis, ORN osteoradionecrosis, NFATc1 nuclear factor of activated T-cells, cytoplasmic 1, BCL6 B-cell lymphoma 6

Osteoclastic BCL6 expression patterns

BCL6-positive (BCL6+) cells showed a brown nuclear and cytoplasmic staining (Fig. 3a). BCL6-expression occurred not only in osteoclasts, but also in mononuclear cells (Fig. 3a). BCL6+ osteoclasts were found in specimens from all groups. The quantitative analysis of the nuclear expression revealed a significantly higher ratio of BCL6+ osteoclasts per ROI in MRONJ (BP) specimens (median: 13.1 osteoclasts/mm²) than in ORN (median: 2.5 osteoclasts/mm²; p < 0.014), OM (median: 2.1 osteoclasts/mm²; p < 0.001) and control specimens (median: 0.0 osteoclasts/mm²; p < 0.001) (Table 2; Fig. 3c). The quantitative analysis of the cytoplasmic expression revealed the same patterns as for nuclear expression with a significantly higher ratio of BCL6 + osteoclasts per ROI in MRONJ (BP) specimens (median: 29.1 osteoclasts/mm²) than in ORN (median: 5.6 osteoclasts/mm²; p < 0.010), OM (median: 9.2 osteoclasts/mm²; p < 0.003) and control specimens (median: 0.0 osteoclasts/mm²; p < 0.001) (Tbl. 2; Fig. 3d).

The analysis of the osteoclast BCL6 labeling indices both for nuclear and cytoplasmic expression revealed that MRONJ (BP) specimens (nuclear median: 48.3%; cytoplasmic median: 86%) featured significantly higher indices than OM (nuclear median: 7.5%; cytoplasmic median: 44.4%; nuclear: p < 0.002; cytoplasmic: p < 0.014) and control specimens (nuclear median: 0.0%; cytoplasmic median: 0.0%; nuclear: p < 0.001; cytoplasmic: p < 0.001) (Table 2; Fig. 4a, b). However, no significant differences between the osteoclast BCL6 labeling indices of MRONJ (BP) and ORN (nuclear median: 28.2%; cytoplasmic median: 83.9%; nuclear: p < 0.572; cytoplasmic: p < 0.912) specimens were found (Table 2; Fig. 4a, b).

The calculation of the ratio of nuclear BCL6+ osteoclasts to cytoplasmic BCL6+ osteoclasts revealed significantly higher values for MRONJ (BP) specimens (median: 50.3%) than for OM (median: 12.7%; p < 0.001) and control specimens (median: 0.0%; p < 0.012) (Table 2; Fig. 4c). For detailed data, see Table 2. For p-values, see Figs. 3 and 4.

Why does BCL6 also appear in the cytoplasm?

Although BCL6, a member of the POZ/BTB-zinc finger protein family, fulfills its function as a transcriptional repressor at a nuclear site [27], the immunohistochemical analysis of the present study revealed the presence of BCL6 in the cell nucleus as well as in the osteoclast cytoplasm (s. Figure 3a). Therefore, the nuclear and cytoplasmic expression of BCL6 expression in osteoclasts was studied separately and the ratio of nuclear to cytoplasmic expression was calculated. The results of nuclear and cytoplasmic BCL6 expression revealed no appreciable differences (Figs. 3, 4). In contrast, NFATc1 was found predominantly in the cell nuclei (Fig. 2a). The occurrence of BCL6 in both cell compartments was also observed in microadenoma and colorectal cancer cells [32]. In general, protein translation takes place predominantly in the extranuclear cell compartment [33]. However, for active suppression of genes, BCL6 must be translocated back into the cell nucleus. Thus, the presence of BCL6 in both cell compartments might be due to a prolonged retention time in the osteoclast’s cytoplasm. The retention time of BCL6 in the osteoclast’s cytoplasm seems to be much longer than that of NFATc1, since NFATc1 could only be detected in nuclei (Fig. 2a).

Relevance of the findings for the understanding of MRONJ (BP) and its differential diseases

The analysis of the osteoclastic expression of NFATc1 and BCL6 allows deeper insights into the status of osteoclasts in the MRONJ (BP)-, ORN- and OM-affected jawbone.

We showed that the osteoclast profile of MRONJ (BP) is characterized by an increased osteoclastic expression of higher-level regulators, paradoxically both enhancer (NFATc1) and suppressor (BCL6) (Figs. 3, 4). In addition, the predecessor study elucidated that the osteoclast profile of MRONJ (BP) is further dominated by a high cell quantity, special cell morphology (giant hypernucleated osteoclasts), resorptive inactivity (TRAP expression low) and an increased cell–cell fusion rate (DC-STAMP expression high) [12]. In literature, the assumption that bisphosphonates inactivate and decimate osteoclasts is widespread [34–36]. This may be true for the bisphosphonate-exposed bone, but need to be put into perspective for MRONJ (BP), as osteoclasts in MRONJ (BP)-tissues can only be considered compromised in terms of their resorptive capacity (TRAP expression low) but not related to their overall cellular activity [12]. A general cellular inactivation of these osteoclasts cannot be assumed as “driving” regulatory mechanisms seem to be active (e.g. high expression of NFATc1).

The comparison between MRONJ (BP) and ORN osteoclast profiles revealed that these bone conditions are similar in terms of the expression patterns of NFATc1 and BCL6 (Figs. 2, 3, 4). However, as we could show previously, these entities differ significantly in terms of osteoclast quantity and morphology [12]. Osteoclasts in MRONJ (BP) tissue, although resorptive inactivate, seem to react to “driving” intracellular signals, such as an increased expression of NFATc1, by forming giant and hypernucleated cells. Contrary, radiation-affected osteoclasts in ORN tissues do not morphologically respond to an increased NFATc1 expression. This indicates that these pathologies, although clinically very similar, differ at the level of osteoclastic regulation.

After being first described in 2003, MRONJ (BP) was temporary not clearly distinguished from a type of OM [37]. However, a clear distinction between these two pathologies is mandatory as we could demonstrate that the osteoclast profiles of MRONJ (BP) and OM differ (Figs. 2, 3, 4, [12]). The increased expression of NFATc1 and the consecutive activation of DC-STAMP expression in MRONJ (BP) osteoclasts could occur as part of a compensatory reaction to the bisphosphonate-induced inhibition of the osteoclasts resorptive capability (TRAP expression low). In contrast, osteoclasts in OM tissue remain resorptive active (TRAP expression high) and thus may not require compensatory upregulation of NFATc1. This could explain the low osteoclast expression of NFATc1 in OM (Figs. 2, 3, 4). However, the observed differences in osteoclastic NFATc1 and BCL6 expression between MRONJ (BP) and OM tissues further strengthens the etiological delineation of these osteopathologies.

From bench-to-bedside?

Clinically but also histopathologically, the differentiation of MRONJ (BP), OM and ORN can be difficult. It requires reliable medical history and clinical information to make the correct diagnosis. The study of the osteoclastic expression of the higher-level regulators NFATc1 and BCL6 not only deepens the understanding of the activity of osteoclasts in these bone conditions, but also extends the profiling possibilities of these key cells. By examining and compiling the osteoclast profiles of the relevant diseases, histopathological examination could gain significance and importance in distinguishing these diseases and thus contribute better to a correct therapeutic decision. A better understanding of the pathophysiologic cellular signaling in jaw bone osteonecrosis is a prerequisite to identify targeted therapeutic approaches.

Limitations

The examined samples are from intraoperatively obtained routine biopsies. It follows that this study did not investigate standardized collectives and samples. Patient-dependent lifestyle factors affecting bone homeostasis, e.g. diet, exercise, and the age were not matched. The localization of the sampling within the jawbone also varied individually.