Keywords

Osteonecrosis; Jaw; Bisphosphonates; Etiology; Propensity score matching

Introduction

Osteoporosis is a very common disease, affecting mainly older people, with fragility fractures its clinical complication [1,2]. Bisphosphonates are the first choice drug in most clinical guidelines for treating osteoporosis [3,4] but there are no data published about its long-term security. So, in recent years, some diseases such as atypical fractures and osteonecrosis of the jaw (ONJ) have been published as possible complications of long- term treatment with bisphosphonates [5-7]. Nevertheless, these complications have also been described with denosumab, which is a potent antiresorptive with no pharmacological relationship with bisphosphonates [8,9].

ONJ is a new clinical entity first described in 2003 by Marx et al. who reported exposed maxillar bone without healing to infection and necrosis. Although from the outset, its etiology was related to the use of bisphosphonates this relationship has not been completely stated, because most ONJ cases have been described in oncology patients receiving bisphosphonates in very high dose, not used in the treatment of osteoporosis, in addition to other drugs, chemotherapy and radiotherapy [10].

Although ONJ is a feared complication, its current incidence is very low and the studies performed to establish a direct relationship with bisphosphonates treatment sometimes have shown contradictory results. Although ONJ is a feared complication, its current incidence is very low and the studies performed to establish a direct relationship with bisphosphonates treatment sometimes have shown contradictory results. We have not find any of these studies performed with propensity score test, which is one of the interpretation of the concept of probability, in which there is a perfect matching between cases and controls.

Materials and Methods

This is a case-control study in which cases were the patients presented ONJ and controls patients suffering from osteoporosis. 24 patients were diagnosed of ONJ following the criteria of The International Task Force on Osteonecrosis of the Jaw [11] and were attended at the Maxillofacial Service at the Hospital University Insular. We included as controls 874 patients suffering from osteoporosis who were attended at the bone metabolic unit at the Hospital University Insular. Taking into account the clinical and biochemical data, propensity score matching was applied and only 20 cases (from 24 patients with ONJ) were perfectly matched to 20 controls from 874 patients of control group.

Statistical analysis

Univariate analysis: Categorical variables are expressed as frequencies and percentages and continuous as mean and standard deviation (SD) when data followed a normal distribution, or as median and interquartile range (IQR=25^th -75 ^th percentile) when distribution departed from normality. For independent data, the percentages were compared, as appropriate, using the Chi-square (χ²) test or the exact Fisher test, the means by the t-test and the medians by the Wilcoxon test for independent data. For dependent data, the percentages were compared using the McNemar test, the means by the t-test for paired data and the medians by the Wilcoxon test for dependent data.

Propensity score: After performing an initial comparison between both groups, patients and controls, we observed some statistically significant differences in some variables, as shown in Table 1 and Table 2. Because of this, we made a matching process with “propensity score”, selecting the variables by the multivariant logistic regression. The resulting model, presented in Table 3 included the following variables: age, TRAP, osteocalcin, rheumatoid arthritis and chemotherapy. To obtain a perfect pairing, we lost 4 patients and only 20 cases and 20 patients could be finally included.

Table 1: Characteristics of the populations before the matching: Clinical features.

Table 2: Comparison of biochemical markers of bone remodeling, hormones and densitometry between controls and patients with osteonecrosis of the jaws.

Table 3: Variables included in the model for calculation of propensity score.

To determine the association between the use of the bisphosphonates and the osteonecrosis of jaw, we selected for each case a similar control (matching). This process was based on a propensity score obtained by means of the logistic regression. More concretely, we consider as propensity score the probability:

which was defined by the logistic model:

logit

Age, bone metabolism markers that showed significant association with the ONJ in univariate analysis, cancer, chemotherapy and rheumatoid arthritis were entered into the multivariate analysis. Selection of variables based on complete enumeration algorithm and Bayes information criterion (BIC) was then performed. The model was summarized as coefficients (SE), p-values (likelihood ratio test) and odds-ratios, which were estimated by confidence intervals at 95%.

Matching: We then carried out a 1-to-1 matched analysis without replacement on the basis of each patient’s estimated propensity score. After propensity score matching, baseline characteristics were compared with the McNemar tests for binary variables and the t-tests for Wilcoxon test, as appropriate, for continuous variables. In addition, we assessed the success of propensity score matching to balance covariates in the 2 groups using standardized differences. Standardized differences of less than 10% support the assumption of balance between the 2 groups.

Conditional logistic regression: The endpoint was the rate of subjects treated with bisphosphonates for five or more years. For each one of them, a logistic model having the binary variable presence/absence of ONJ as covariable was considered. These models were estimated by means of the conditional likelihood. From the models were obtained the corresponding odds-ratios, which were estimated by means of 95% confidence intervals. Statistical significance was set at p<0.05.

Physical examination

A complete physical examination was carried out of every patient included in the study. Height was measured without shoes, and weight with light clothes was estimated on a balance scale. Body mass index (BMI) was derived from the formula: BMI=weight (kg)/height (m)².

Dual-energy X-ray Absorptiometry (DXA)

Bone mineral density (BMD) was measured by a DXA Hologic QDR 4500 Discovery (Hologic, Spain). The area of interest at the lumbar spine measurement was L2-L4. At the femoral site, two regions were measured: femoral neck and total hip. The software provided by the manufacturer allowed anatomical separations [12]. The results were expressed in g/cm². Precision of the system (coefficient of variation) was 0.5% in vitro (standard bone phantom) and 0.9% in vivo (12 patients measured twice in the same day). All the determinations were measured by the same operator, so no inter-observer variation existed. T-scores were calculated from the reference values previously obtained from Canary Islands population [13].

Trabecular Bone Score (TBS)

All TBS measurements were performed using TBS iNsight Software, version 2.0.0.1 (Med-Imaps, Pessac, France). This software uses the raw DXA image of the anteroposterior spine for the same region of interest as the BMD measurement. The densitometer was calibrated using anthropomorphic phantoms.

Quantitative Ultrasound (QUS) measurements

All subjects underwent calcaneus measurement by QUS. This was carried out using the Sahara Clinical sonometer (Hologic Inc., Bedford, MA). The system consists of 2 unfocused transducers mounted co-axially on a monitor caliper. One transducer acts as the transmitter and the other as the receiver. The transducers are acoustically coupled to the heel using soft rubber pads and an oil- based coupling gel. The Sahara device measures both broadband ultrasound attenuation (BUA) and speed of sound (SOS) at a fixed region of interest in the midcalcaneus, and the BUA and SOS results are combined to provide an estimate of the quantitative ultrasound index (QUI) using the formula:

QUI=0.41 × (BUA+SOS)-571

For all QUS measurements, the corresponding T-scores and Z-scores were calculated according the normative data for the Spanish population, previously established by our working group [14].

Biochemical measurements

Serum specimens were obtained after an overnight fast. Blood was collected without additives between 8:00 and 9:00 a.m. After centrifugation at 1 500 g for 10 min, serum was aliquoted and frozen at -20°C within 1 h from phlebotomy until the biochemical analyses were performed. Serum parathyroid hormone (PTH) and 25-hydroxy-vitamin D (25-OHD) were measured by electrochemiluminescence with Elecsys 170 PP (Modular Analytics) of Roche Diagnostic^® (Basel, Switzerland). For PTH, total coefficients of variation (TCVs) ranged from 1.6% to 10.9%, and for 25 OHD TCVs was 4.9% using blinded quality control samples in our laboratory.

The measured remodeling bone markers for formation were Osteocalcin (OC) (electrochemiluminescence immunoassay, analyser Elecsys, Roche Diagnostics, IN), and N-terminal propeptide of type 1 collagen (PINP) (electrochemiluminescence immunoassay Roche Diagnostics). The markers for resorption were tartrate-resistant acid phosphatase 5β (TRAP5β, colourimetry, Hitachi 704 Boehringer Manheim GmbH) and carboxy-terminal telopeptide of type I collagen (CTX, enzymatic immunoassay, analyser Elecsys CrossLaps, Roche Diagnostics SL, Barcelona, Spain).

Fractures assessment

Prevalent vertebral fractures were assessed on standard lateral spine radiographs in all subjects. Vertebral fractures were defined following the radiological semiquantitative criteria of Genant [15]. The presence of nonvertebral fractures was documented firstly by a self-reported history with later confirmation in medical hospital records or X-ray films.

Results

Table 1 shows the characteristics of the studied populations before the performed matching by the propensity score method. Patients with ONJ had a higher age than controls and also had a higher prevalence of cancer, chemotherapy treatment, rheumatoid arthritis, oral steroids therapy and a higher use of bisphosphonates for more than 5 years, while controls showed a higher prevalence of fragility fractures and maternal hip fractures.

Table 2 shows biochemical, densitometric and ultrasonographic values of the populations studied before the matching. Patients with ONJ had higher values of TRAP, and lower values of beta- crosslaps, P1NP and osteocalcin than controls. Bone mineral density was higher in cases than controls, but only significantly at L2-L4 (p<0.001). There was a lower prevalence of osteoporosis densitometric values in patients with ONJ (T-score<-2.5 at any of the measured sites: lumbar spine, femoral neck or total hip). There were no statistical differences in trabecular bone score (TBS) and quantitative ultrasound parameters measured at the calcaneus between both groups of patients.

Table 3 shows the propensity-score obtained by means of the multivariate logistic regression for ONJ. Chosen values were age, serum values of TRAP and osteocalcin, the presence of rheumatoid arthritis, and having received chemotherapy. The propensity score deduced from this model is:

Figure 1 shows the paired propensity scores. This scatter shows that propensity scores are practically identical within the paired subjects.

Figure 1: The major BAT complex in humans.

Table 4 shows obtained data when cases and controls were compared after the matching performed by the paired propensity scores. There are no statistically significant differences between both groups with the exception of serum TSH levels, which were into the normal range. Nevertheless, we performed the next logistic regression using this hormone as co-variable.

Table 4: Comparison of the variables studied after matching by propensity score.

Table 5 shows the results of the conditional logistic regression for the exposure to bisphosphonates for 5 years. In both cases, there was no statistical association between the use of bisphosphonates and the presence of ONJ, either alone or analyzing them using TSH value as a co-variable.

Table 5: Conditional logistic regression for the exposure to bisphosphonates for 5 years.

Conclusion

In conclusion, our results confirm that there is not a direct casual association between the use of bisphosphonates after 5 years and the development of ONJ. This is probably due to the fact that in the pathogenesis of this disease can exist a number of clinical factors in addition to the use of bisphosphonates, such as the presence of a cancer and the chemotherapy used in its treatment, poor oral and dental health, corticoids, diabetes and teeth extractions.

References

1. Black DM, Rosen CJ (2016) Postmenopausal osteoporosis. N Engl J Med 374(3): 254-262.
2. Cummings SR, Melton LJ (2002) Epidemiology and outcomes of osteoporotic fractures. Lancet 359(9319): 1761-1767.
3. González-Macías J, Del Pino-Montes J, Olmos JM, Nogués X (2015) Clinical practice guidelines in postmenopausal, glucocorticoid and male osteoporosis: Spanish society for research in bone and mineral metabolism (updated 3rd version 2014). Rev Clin Esp 215 (9): 515-526.
4. Tarantino U, Iolascon G, Cianferotti L, Masi L, Marcucci G, et al. (2017) Clinical guidelines for the prevention and treatment of osteoporosis: Summary statements and recommendations from the Italian society for orthopaedics and traumatology. J Orthop Traumatol 18(Suppl 1): 3-36.
5. Eriksen EF, Díez-Pérez A, Boonen S (2014) Update on long-term treatment with bisphosphonates for postmenopausal osteoporosis: A systematic review. Bone 58: 126-135.
6. Watts NB, Diab DL (2010) Long-term use of bisphosphonates in osteoporosis. J Clin Endocrinol Metab 95(4): 1555-1565.
7. Rizzoli R, Burlet N, Cahall D, Delmas PD, Eriksen EF, et al. (2008) Osteonecrosis of the jaw and bisphosphonate treatment for osteoporosis. Bone 42(5):841-847.
8. Selvi Sabater P, Rizo Cerdá A, Titos ARcos J, Espuny Miró A (2014) Posible denosumab-induced jaw osteonecrosis in the treatment of osteoporosis: A case report. Farm Hosp 38(3): 248-256.
9. Diz P, López-Cedrún JL, Arenaz J, Scully C (2012) Denosumab-related osteonecrosis of the jaw. J Am Dent Assoc 143(9): 981-984.
10. Marx RE (2003) Pamidronate (Aredia) and zoledronate (Zometa) induced avascular necrosis necrosis of the jaws: A growing epidemic. J Oral Maxillofac Surg 61:1115-1118.
11. Khan AA, Morrison A, Hanley DA, Felsenberg D, McCauley LK, et al. (2015) Diagnosis and management of osteonecrosis of the jaw: A systematic review and international consensus. J Bone Miner Res 30(1): 3-23.
12. Duboeuf F, Braillon P, Chapuy MC, Haond P, Hardouin C, et al. (1991) Bone mineral density of the hip measured with dual-energy X-ray absorptiometry in normal elderly women and in patients with hip fracture. Osteoporos Int 1(4): 242-249.
13. Sosa M, Hernández D, Estévez S, Rodríguez M, Limiñana JM, et al. (1998) The range of bone mineral density in healthy Canarian women by dual X-ray absorptiometry radiography and quantitative computer tomography. J Clin Densitom 1(4): 385-393.
14. Sosa M, Saavedra P, Muñoz-Torres M, Alegre J, Gómez C, et al. (2002) Quantitative ultrasound calcaneus measurements: Normative data and precision in the Spanish population. Osteoporos Int 13(6): 487-492.
15. Genant HK, Wu CY, van Kuijk C, Nevitt MC (1993) Vertebral fracture assessment using a semiquantitative technique. J Bone Miner Res 8(9): 1137-1148.
16. Review C, Communication S, Principles G (2004) World Medical Association. World Medical Association Declaration of Helsinki Ethical Principles for Medical Research Involving Human Subjects. J Int Bioéthique 15(1): 124.
17. Orriss IR, Key ML, Colston KW, Arnett TR (2009) Inhibition of osteoblast function in vitro by aminobis-phosphonates. J Cell Biochem 106(1): 109-118.
18. Allen MR, Burr DB (2009) The pathogenesis of bisphosphonate-related osteonecrosis of the jaw: So many hypotheses, so few data. J Oral Maxillofac Surg 67(5 suppl.): 61-70.
19. Rasmusson L, Abtahi J (2014) Bisphosphonate associated osteonecrosis of the jaw: An update on pathophysiology, risk factors, and treatment. Int J Dent 2014: 471035.
20. Huja SS, Fernandez SA, Hill KJ, Li Y (2006) Remodeling dynamics in the alveolar process in skeletally mature dogs. Anat Rec A Discov Mol Cell Evol Biol 288(12): 1243-1249.
21. Furuya T, Maeda S, Momohara S, Taniguchi A, Yamanaka H (2016) Dental treatments, tooth extractions, and osteonecrosis of the jaw in Japanese patients with rheumatoid arthritis: results from the IORRA cohort study. J Bone Miner Metab 35(3): 344-350.
22. Sosa Henríquez M (2009) Osteonecrosis of the jaws: SEIOMM consensus document. Rev Osteoporos and Metab Miner 1(1): 41-51.
23. Ruggiero S, Gralow J, Marx RE, Hoff AO, Schubert MM, et al. (2006) Practical guidelines for the prevention, diagnosis, and treatment of osteonecrosis of the jaw in patients with cancer. J Oncol Pract 2(1): 7-14.
24. Bocanegra-Pérez MS, Vicente-Barrero M, Sosa-Henríquez M, Rodríguez-Bocanegra E, Limiñana-Cañal JM, et al. (2012) Bone metabolism and clinical study of 44 patients with bisphosphonate-related osteonecrosis of the jaws. Med Oral Patol Oral Cir Bucal 17(6): e948-e955.
25. Sosa-Henríquez M, Vicente-Barrero M, Bocanegra-Pérez MS (2011) Osteonecrosis of the jaws: new evidence on its etiopathogenesis. Rev Osteoporos Metab Miner 3(1): 5-6.
26. Marx RE, Sawatari Y, Fortin M, Broumand V (2005) Bisphosphonate-induced exposed bone (osteonecrosis/osteopetrosis) of the jaws: risk factors, recognition, prevention, and treatment. J Oral Maxillofac Surg 63(11): 1567-1575.
27. Raje N, Woo SB, Hande K, Yap JT, Richardson PG, et al. (2008) Clinical, radiographic, and biochemical characterization of multiple myeloma patients with osteonecrosis of the jaw. Clin Cancer Res 14(8): 2387-2395.
28. Assael LA (2009) Oral bisphosphonates as a cause of bisphosphonate-related osteonecrosis of the jaws: clinical findings, assessment of risks, and preventive strategies. J Oral Maxillofac Surg 67(5 suppl.): 35-43.
29. Otto S, Hafner S, Mast G, Tischer T, Volkmer E, et al. (2010) Bisphosphonate-related osteonecrosis of the jaw: is ph the missing part in the pathogenesis puzzle? J Oral Maxillofac Surg 68(5): 1158-1161.
30. Gaudin E, Seidel L, Bacevic M, Rompen E, Lambert F (2015) Occurrence and risk indicators of medication-related osteonecrosis of the jaw after dental extraction: A systematic review and meta-analysis. J Clin Periodontol 42(10): 922-932.
31. Leizaola-Cardesa IO, Aguilar-Salvatierra A, Gonzalez-Jaranay M, Moreu G, Sala-Romero MJ, et al. (2016) Bisphosphonates, vitamin D, parathyroid hormone, and osteonecrosis of the jaw: Could there be a missing link? Med Oral Patol Oral y Cir Bucal 21(2): e236-e240.
32. Ristow O, Gerngroß C, Schwaiger M, Hohlweg-Majert B, Kehl V, et al. (2014) Effect of antiresorptive drugs on bony turnover in the jaw: Denosumab compared with bisphosphonates. Br J Oral Maxillofac Surg 52(4): 308-313.
33. Troeltzsch M, Woodlock T, Kriegelstein S, Steiner T, Messlinger K, et al. (2012) Physiology and pharmacology of nonbisphosphonate drugs implicated in osteonecrosis of the jaw. J Can Dent Assoc (Tor) 78(1): 85.
34. Ramírez L, López-Pintor RM, Casañas E, Arriba L de, Hernández G (2015) New non-bisphosphonate drugs that produce osteonecrosis of the jaws. Oral Health Prev Dent 13(5): 385-393.
35. Huang YF, Chang CT, Muo CH, Tsai CH, Shen YF, et al. (2015) Impact of bisphosphonate-related osteonecrosis of the jaw on osteoporotic patients after dental extraction: A population-based cohort study. PLoS One 10(4): e0120756.
36. Peer A, Khamaisi M (2015) Diabetes as a risk factor for medication-related osteonecrosis of the jaw. J Dent Res 94(2): 252-260.
37. Heim N, Warwas FB, Wilms CT, Reich RH, Martini M (2017) Vitamin D (25-OHD) deficiency may increase the prevalence of medication-related osteonecrosis of the jaw. J Craniomaxillofac Surg 45(12): 2068-2074.
38. Silva BC, Leslie WD, Resch H, Lamy O, Lesnyak O, et al. Trabecular bone score: A noninvasive analytical method based upon the DXA image. J Bone Miner Res 29(3): 518-530.
39. Glüer CC (2007) Quantitative Ultrasound-It is time to focus research efforts. Bone 40(1): 9-13.
40. Raum K, Grimal Q, Varga P, Barkmann R, Glüer CC, et al. Ultrasound to assess bone quality. Curr Osteoporos Rep 12(2): 154-162.