Leukopenia as a risk factor for osteonecrosis of the jaw in metastatic prostate cancer treated using zoledronic acid and docetaxel


Haruki Kume, Department of Urology, the University of Tokyo Hospital, Tokyo, Japan. e-mail: KUMEH-URO@h.u-tokyo.ac.jp


Study Type – Harm (case series)

Level of Evidence 4

What's known on the subject? and What does the study add?

The common clinical practice for advanced prostatic adenocarcinoma is combination therapy of zoledronic acid (ZA) and docetaxel (TAX). Little is known about the consequences of this combination therapy with regard to osteonecrosis of the jaw (ONJ).

This study shows that the combination therapy of ZA and TAX increases the risk of ONJ and that tooth extraction and leukopenia induced by TAX are the risk factors.


  • • To determine whether docetaxel (TAX) can increase the risk of osteonecrosis of the jaw (ONJ) in patients with prostatic adenocarcinoma (PC) receiving zoledronic acid (ZA), a bisphosphonate (BP) used in the treatment of patients with cancer.


  • • The medical records of 111 patients with PC who received ZA between September 2006 and March 2011 at our institutions were reviewed to assess the incidence and risk factors for ONJ.


  • • Nine patients (8.1%) developed ONJ during a median follow-up of 14.5 months.
  • • Using univariate analysis we found that TAX chemotherapy (P = 0.037, hazard ratio [HR] 6.611), tooth extraction during ZA therapy (P < 0.001, HR 11.254), and high prostate-specific antigen level (P = 0.019, HR 8.008) at the start of ZA were predictive factors.
  • • Using multivariate analysis we found that TAX chemotherapy (P = 0.011, HR 56.35), steroid use (P = 0.044, HR = 17.795), and tooth extraction (P = 0.039, HR 7.471) were independent predictors.
  • • Among those receiving TAX chemotherapy, multivariate analysis identified tooth extraction (P = 0.009) and nadir WBC counts <1000/µL during TAX chemotherapy (P = 0.030) as independent risk factors.


  • • Tooth extraction and nadir WBC counts <1000/µL were found to be risk factors for ONJ in metastatic prostate cancer treated with ZA and TAX combination therapy, showing that leukopenia is an important factor in the development of ONJ.



osteonecrosis of the jaw


prostatic adenocarcinoma


zoledronic acid




hazard ratio


white blood cell


American Association of Oral and Maxillofacial Surgeons


granulocyte colony-stimulating factor.


Prostatic adenocarcinoma (PC) is the second leading cause of male cancer deaths in the USA [1,2]. PC is also one of the most common cancers in Japan, accounting for 10.6% of estimated new cancer cases and 5.1% of estimated cancer deaths in men [3]. Bone is the most common metastatic site, representing >80% of advanced PC cases [4]. In addition, complications from bone metastases, including pain, pathological fracture and abnormal serum calcium level, are a major cause of morbidity in PC.

Zoledronic acid (ZA) is a bisphosphonate (BP), which has a strong affinity for hydroxyapatite and concentrates within bone and shift bone turnover by inhibiting osteoclasts and promoting osteoblastic activity [4–6]. ZA has greater potency than pamidronate or other BPs used in several preclinical models of bone resorption [7], and has been shown to decrease and/or delay the onset of skeletal-related events, reducing tumour-induced hypercalcaemia and bone pain [4–6]. However, osteonecrosis of the jaw (ONJ), which has rarely been observed for other BPs [9], has been known to be a significant adverse event associated with use of ZA since the first report in 2003 [8–9].

The common clinical practice for advanced PC is combination therapy of ZA and docetaxel (TAX), an antineoplastic agent [10,11], but little is known about the consequences of this combination therapy for ONJ. The aim of the present study was to determine whether co-treatment with TAX affects the risk of ZA-related ONJ in a retrospective study cohort.


A retrospective chart review was conducted for men with PC who had bone metastasis and received ZA at the authors' hospitals between September 2006 and March 2011. The present study was approved by the institutions' Ethics Committee (No. 3124). ZA was administered at 4.0 mg every 21 days. TAX was given at 75 mg/m2 by 1-h i.v. infusion every 21 days, along with a daily oral dose of 1.0–2.0 mg of dexamethasone [12]. Patients were followed at outpatient clinic every 3–4 weeks. We reviewed laboratory test data, including complete blood count, renal function, liver function and PSA levels. All patients with suspected ONJ were referred to a dentist for evaluation including panoramic x-rays. Biopsy of ONJ lesions was performed only when metastatic disease was suspected. ONJ diagnosis was made according to the following criteria stated in the American Association of Oral and Maxillofacial Surgeons (AAOMS) position paper: (i) current or previous treatment with a bisphosphonate; (ii) exposed, necrotic bone in the maxillofacial region that has persisted for more than 8 weeks; and (iii) no history of radiation therapy to the jaws [13]. Therapy for ONJ consisted of termination of ZA, patient education, antibacterial mouth rinse, antibiotic therapy, and/or surgical debridement [13]. ONJ-free survival was calculated from the first ZA administration. A log-rank test and Cox proportional hazards regression model were used for univariate and multivariate analyses of ONJ-free survival, respectively. All analyses were performed using the software, Stat Mate III ver. 3.07, ATMS Co., Ltd., Tokyo, Japan. A P value <0.05 was considered to indicate statistical significance.


A total of 111 patients with PC were included in the analysis. The median (range) patient age at the start of ZA treatment was 75 (56–91) years, and the median (range) initial serum level of PSA was 170.2 (0.48–9559) ng/mL. Radical prostatectomy (n = 20) and external radiotherapy (n = 1) had been performed as radical therapy. Of 94 men, 19 (20.2%, excluding 17 unknown) were smokers and 20 of 91 (22%, excluding 20 unknown) had diabetes. Dentures were used in 52 men (46.8%). Nine men had tooth extractions within 6 months. Extent of disease on bone scan was grade 1, grade 2, grade 3, grade 4, and unknown in 31, 36, 34, 4 and 6 men, respectively [14]. Owing to painful metastases, 37 (33.6%) needed analgesics.

Before the start of ZA, all patients had received LHRH agonists in combination with anti-androgens (bicalutamide or flutamide), 62 (56%) had started 1.0–2.0 mg oral dexamethasone, 92 (83%) had become castration-resistant, and 56 (50.4%) had received chemotherapy with TAX (75 mg/m2 every 3 weeks). The median (range) values for blood test results at starting ZA were: WBC, 6600 (2000–22 100)/µL; haemoglobin, 11.9 (4.3–15.2) g/dL; platelets, 21.7 (6.8–38.9) × 104/µL; alkaline phosphatase, 315 (77–6634) IU/L; C reactive protein, 0.33 (0.01–28.64) mg/dL; and PSA, 27.5 (0.02–2761) ng/mL. During ZA therapy, a WBC nadir <1000/µL occurred in eight men as an adverse event of TAX treatment.

During a median (range) follow-up of 14.5 (1.0–40.2) months, ONJ developed in nine (8.1%) cases. It occurred significantly more frequently in men treated with ZA and TAX combined chemotherapy (8/56, 14.3%) than men with ZA monotherapy (1/55, 1.8%, P = 0.040, chi-squared test with Yates' correction). Of the eight patients with ONJ receiving the combination treatment, three had a nadir WBC <1000/µL.

To assess the predictive factors for ONJ we performed the log-rank test using univariate analysis, which revealed the following predictive risk factors for ONJ: TAX chemotherapy (P = 0.037, hazard ratio [HR] 6.611), tooth extraction (P < 0.001, HR 11.254), and high PSA level at the start of ZA (P = 0.019, HR 8.008 [Table 1]). A Cox proportional hazards regression model identified TAX chemotherapy (P = 0.011, HR 56.346), steroid use (P = 0.044, HR = 17.795), and tooth extraction (P = 0.039, HR 7.471) as independent predictive factors (Table 1).

Table 1. Predictive factors for ONJ in all patients
Variable n Univariate analysisMultivariate analysis
HR (95% CI) P HR (95% CI) P
  • *

    Significant (P < 0.05).

 ≤75 years62Reference0.310  
 ≥76 years491.393 (0.521–7.792)   
Karnofsky performance status     
 ≤70%16<0.001 (0.035–2.961)   
Diabetes mellitus     
 No71Reference Reference 
 Yes200.508 (0.111–3.003)0.5140.892 (0.054–14.695)0.936
 Yes19<0.001 (0.048–1.826)   
Tooth extraction within 6 months of treatment     
 No103Reference Reference 
 Yes811.254 (21.189– >100.0)<0.001*7.471 (1.103–50.588)0.039*
Denture use     
 No59Reference Reference 
 Yes522.651 (0.703–9.957)0.1501.008 (0.165–6.178)0.993
Extent of disease     
 Grades 0, 1 or 267Reference   
 Grades 3 or 4380.873 (0.186–4.083)0.862  
Visceral metastasis     
 Yes10<0.001 (0.013–9.250)0.525  
Castration-resistant prostate cancer     
 Yes92>100.0 (0.569–20.761)0.179  
PSA level     
 ≤27.5 ng/mL56Reference   
 >27.5 ng/mL558.008 (1.296–17.917)0.019*10.722 (0.637– >100.0)0.100
Elevation of alkaline phosphatase >360 IU/L     
 Yes500.308 (0.093–1.767)0.229  
Anaemia (haemoglobin <13.8 g/dL)     
 Yes952.227 (0.375– 24.272)0.299  
Analgesic use     
 No74Reference Reference 
 Yes370.400 (0.104–2.304)0.3661.941 (0.213–17.656)0.556
TAX therapy     
 No55Reference Reference 
 Yes566.611 (1.088–15.555)0.037*56.346 (2.550– >100.0)0.011*
Steroid use     
 No49Reference Reference 
 Yes622.172 (0.515– 7.740)0.31717.795 (1.082– >100.0)0.044*

To evaluate risk factors for ONJ among patients with TAX chemotherapy, we carried out univariate and multivariate analyses which identified tooth extraction (P = 0.009) and nadir WBC counts <1000/µL during combination therapy (P = 0.030) as independent risk factors (Table 2).

Table 2. Risk factors for ONJ among patients receiving TAX and ZA treatment
Variable n Univariate analysisMultivariate analysis
HR (95% CI) P HR (95% CI) P
  • *

    Significant (P < 0.05).

Age  0.247  
 ≤75 years39Reference 
 ≥76 years172.208 (0.540–11.007) 
Karnofsky performance status  0.291  
 ≤70%8<0.001 (0.035– 2.722) 
Diabetes mellitus  0.805 0.298
 Yes60.773 (0.111–5.524)0.102 (0.001–7.561)
Smoking  0.263  
 Yes11<0.001 (0.037–2.460) 
Tooth extraction within 6 months of treatment  <0.001* 0.009*
 Yes610.747(10.940–>100.0)15.974 (1.990–>100.0)
Denture use  0.425 0.916
 Yes271.753 (0.431–7.362)0.889 (0.099–7.983)
Extent of disease  0.311  
 Grades 1 or 234Reference 
 Grades 3 or 4200.365 (0.090–2.153) 
Visceral metastases  0.732  
 Yes3<0.001 (0.001–>100.0) 
PSA level  0.004* 0.165
 ≤27.5 ng/mL24ReferenceReference
 >27.5 ng/mL3210.976 (1.921–34.062)25.737 (0.263–>100.0)
Elevation of alkaline phosphatase 360 IU/L  0.322  
 Yes240.376 (0.088–2.225) 
Anaemia (haemoglobin <13.8 g/dL)  0.192  
 Yes51>100.0 (0.529–23.894) 
Analgesic use  0.526 0.790
 Yes220.533 (0.093–3.378)0.706 (0.054–9.163)
Steroid use  0.351 0.075
 Yes510.389 (0.010–5.229)35.710 (0.697–>100.0)
No. of ZA administrations  0.815  
 >5301.203 (0.248–5.878) 
Nadir WBC <1000/uL  0.083 0.030*
 Yes83.230 (0.800–39.386)12.450 (1.273–>100.0)


In the present retrospective study we found nine (8.1%) cases of ONJ in 111 men receiving ZA therapy for metastatic PC. The incidence was in the range of that previously reported at 2.9–18.6% [15,16].

To date, several risk factors for ONJ have been suggested, including tooth extraction, use of dentures and diabetes [17,18]. Kyrgidis et al.[17], analysing 20 patients with breast cancer who had ONJ, found the risk of ONJ increased with tooth extraction during ZA treatment (P = 0.001, HR 16.4) and with use of dentures (P = 0.029, HR 4.9). Khamaisi et al.[18] reported a much higher incidence (58%) of diabetes in men with BP-related ONJ than men without (14%). Other possible risk factors, smoking and obesity, have not been substantiated [9]. Medications, such as steroids, thalidomide, and some chemotherapeutic agents, were also postulated to be risk factors, but the original AAOMS position paper concluded that there were no measurable associations with these potential risk factors [13].

A few studies have reported an increased risk of ONJ among patients exposed to chemotherapy agents. Jadu et al.[19] reported that cyclophosphamide and erythropoietin treatments were associated with an increased risk of ONJ. They also reported an increased risk with steroid use. By contrast, Wessel et al.[9] found no significant association between ONJ and use of cyclophosphamide, steroids, erythropoietin or thalidomide. As for TAX chemotherapy, Ortega et al.[20] detected a potential, but not significant, trend to increased risk of ONJ associated with TAX chemotherapy (HR 3.8, P = 0.24) in 52 men with PC, treated with ZA for bone metastasis. Similarly, in the present study, TAX chemotherapy was a significant predictive factor for ONJ in univariate and multivariate analysis. In addition, we identified tooth extraction and nadir WBC counts <1000/µL during TAX therapy as the significant risk factors. Thus, the present results indicate that TAX-induced leukopenia, rather than TAX use itself, is of more importance in ONJ development. Anti-angiogenic properties of both ZA and TAX may additively or synergistically impair jaw vascularization and cause a predisposition to bone necrosis [20].

The association between nadir WBC count and ONJ in patients with cancer treated with BPs has not been reported. Leukopenia induced by chemotherapy might serve as an aggravating factor for infection of the oral cavity such as mucositis. Although the pathogenesis of ONJ is not fully understood, colonization of the oral cavity and teeth by a complex microbial flora is considered an important initial step [21]. Once oral mucositis is induced, inflammation of the lesion might persist owing to microbes and other stresses, including mastication and dental procedures. Thus, it would be reasonable to postulate that oral infection resulting from TAX-induced leukopenia could lead to ONJ aggravation.

Oral mucositis is associated with decreased granulocyte counts and occurs at rates as high as 20% in patients treated with TAX [10]. Resolution of mucositis is related to neutrophil recovery [22–26].. Indeed, Gabrilove et al.[27] observed the reduced severity of mucositis in patients who received granulocyte colony-stimulating factor (G-CSF) after standard dose chemotherapy. Earlier administration of G-CSF could decrease the risk of ONJ by recovering WBCs more rapidly.

In conclusion, we found three independent significant predictive factors for ONJ in PC cases with bone metastasis: tooth extraction, steroid use and TAX chemotherapy. In addition, among cases with ZA and TAX combination therapy, we found that tooth extraction and nadir WBC counts <1000/µL were independent significant risk factors, suggesting that leukopenia induced by TAX rather than TAX use itself might be essential for ONJ development. Obviously, limitations of the present study are its retrospective nature and its limited number of cases. Further studies are needed to show the pathogenetic and therapeutic implications of leukopenia in developing ONJ among patients with cancer receiving BP treatment.


None declared.