The use of intravenous nitrogen-containing bisphosphonates (N-BPs) is occasionally associated with the appearance within 24 to 36 hours of fever and musculoskeletal pain.1–3 The fever was noted to be associated with a fall in circulating lymphocyte number1, 4 and increases in circulating IL-64 and tumor necrosis factor-alpha (TNF-α)4–6 but not interleukin (IL)-1.6 This is referred as the acute phase response (APR).
N-BPs inhibit osteoclastic bone resorption by blocking farnesyl-pyrophosphate-synthase, an enzyme in the mevalonate pathway. Recent work suggested that this action may underlie the development of the APR because intermediates in this pathway, isopentenyl diphosphate and dimethyl-allyl diphosphate, accumulate in monocytes when this enzyme is blocked and result in the activation of adjacent γδ T cells with the release of interferon-γ and TNF.7–11
Knowing which individuals are more at risk of APR is clinically relevant and might also give insight into the pathogenesis of the APR. Recently, by reviewing the patients participating in the HORIZON clinical trial for the registration of zoledronic acid (ZOL) for the treatment of postmenopausal osteoporosis it was reported that APR is more common in younger subjects, nonsteroidal anti-inflammatory drug users, and those having back pain and less common in smokers, diabetics, and calcitonin or previous bisphosphonate users.12 The APR is considerably more common after the first infusion, tends to disappear despite continuing therapy and both the incidence and the severity decrease substantially with subsequent treatments.1, 12
The aim of this study was to explore the correlation between the circulating lymphocytes or their subpopulations and the risk of APR after N-BPs iv administration.
Materials and Methods
Forty patients with postmenopausal or senile osteoporosis who required iv therapy with N-BPs were enrolled in this study. These were 5 men and 35 women, with a mean age of 72 years (range = 53–91 years). None of the patients had received parenteral N-BPs treatment previously. Patients with cancer, autoimmune diseases, immunodeficiency, severe liver or renal insufficiency (serum creatinine >1.0 mg/dL), or recent acute infections were excluded from this study. Patients were not eligible if they had been treated with nonsteroidal anti-inflammatory or analgesic/antipyretic drugs within 3 days before beginning the study or if they had been treated within the last 2 years with cytostatic drugs, corticosteroids, or immunotherapeutics. None of the patients presented clinical signs of infectious disease.
This study was approved by the local ethic committee, and the subjects' consent was obtained according to the Declaration of Helsinki.
Treatment and follow-up investigation
All study participants received a single 5-mg ZOL in 100 mL of 0.9% saline intravenous (iv) infusion over 15 minutes. Before the iv infusion all patients had been on vitamin D supplements for at least 2 months. Samples of peripheral blood were taken immediately before bisphosphonate treatment by using Vacutainer blood collection tubes coated with ethylen-ediamine-tetraacetic acid (EDTA). White blood cells (WBC) were counted with an automated hematology analyzer (ADVIA 2120i Siemens). Fifty microliters of blood were distributed into each tube by the automated BD FACS Sample Prep Assistant II (Becton Dickinson, San Jose, CA, USA), a mixture of monoclonal antibodies conjugated with different fluorochromes (FITC, PE, PerCP, PE-Cy7, APC, APC-Cy7, anti-γδ receptor; BD Biosciences, San Jose, CA, USA) were added, the red blood cells were lysed and finally the cells were fixed (BD FACS Lysing Solution). The lymphocytes were analyzed by flow cytometer (BD FACSCanto, Becton Dickinson) with BD FACSDiva software. The lymphocytes were isolated using CD45 versus SSC as gating strategy. Different subsets of T cells were counted using these monoclonal antibodies: APC-conjugated anti-CD3, FITC-conjugated anti-CD4, PE-Cy7-conjugated anti-CD8. Amounts of CD4−CD8− double-negative and CD4+CD8− double-positive cells were calculated within gated CD3+ T lymphocytes and γδ T cells were counted in the samples of CD4−CD8− double-negative T lymphocytes stained with anti-TCR γ/δ-FITC.
Nature killer (NK) cells were counted using APC-CY7-conjugated anti-CD16 and PE-conjugated anti-CD56. B cells were counted using APC-CY7-conjugated anti-CD19. The laboratory used UK NEQAS for leucocyte immunophenotyping as external quality.
Serum 25-hydroxyvitamin D (25OHD) levels were also evaluated by commercial kit (Diasorin, Saluggia, Italy).
Body temperature was determined with digital clinical thermometers immediately before the intravenous infusion and at 12-hour intervals for 3 days. Fever was defined as an increase in body temperature above 37 °C. Patients were instructed to register the temperature values on a diary together with any self-administer acetaminophen dose to treat fever or other symptoms of APR.
Peripheral leucocytes and lymphocytes subpopulations were compared in patients with and without APR, by Mann-Whitney U-test for nonparametric independent variables and then after correcting the values for any potential interfering factor by analysis of covariance (ANCOVA). A linear regression model with variance analysis was used to correlate the continuous variables.
The receiver operating characteristic (ROC) curves were constructed to investigate the cutoff points with the best sensitivity and specificity in predicting the occurrence of APR.
A two-tailed p value of 0.05 was considered significant. SPSS software (version 17.00, SPSS, Chicago, IL, USA) was used for statistical analysis.
Up to 42.5% of patients (17 of 40) receiving the infusion of ZOL experienced an APR defined by the occurrence of fever within 3 days after the infusion. The main baseline characteristics of the two groups of patients, with or without APR, are listed in Table 1. Mean age of APR patients was lower than in non APR patients (69 ± 7 years versus 74 ± 8 years, respectively; p = 0.054). Both proportion and absolute number of γδ T cells were significant higher in patients who experienced an APR (p = 0.02 and p = 0.013, respectively). Nonsignificant differences were found between the two groups for morning body temperature (below 37 °C in all subjects) white blood cells and for the other circulating lymphocyte subpopulations. Serum 25OHD was slightly higher in APR patients (Table 1); 25OHD levels lower than 50 nmol/L were found in eight non-APR patients (35%) and in six APR patients (33%). Serum C-reactive protein (CRP; Table 1) was slightly above normal (5 mg/dL) in five patients (three non-APR and two APR) but similarly elevated values could be retrieved in all of them several months earlier. Type 2 diabetes was present in three non-APR patients and in two APR patients.
Table 1. Main Characteristics of the Two Groups of Patients, With or Without APR
Age was inversely correlated with circulating γδ T cells (p = 0.003; Fig. 1) as well as with total lymphocytes, T cells, T8, and B (p ranging from 0.008–0.025; results not shown).
The difference in circulating γδ T cells between APR positive and negative patients persisted significant for values adjusted for age, both as absolute number and percent of T cells. By the ROC curves the best discriminating values (ie, yielding the best sensitivity and specificity) were 25 γδ T cells/microliter (p = 0.032) and 3.0% γδ T cells (p = 0.027), respectively. None of the patients with <3% γδ T cells experienced an APR (false negative), whereas 22% of patients with >3% γδ T cells APR was not observed (false positive).
An APR is the most frequent adverse effect of intravenous administration of N-BP drugs, independent of the treated disease.1, 12 Recently, it was proposed that N-BPs may act as nonpeptide phosphoantigens, capable of binding to γδT cells.9–11 This would lead to activation and proliferation of γδ T cells, with subsequent release of proinflammatory cytokines such IL-6, TNF-α, and IFN-γ.4–6
In this study we have shown that both the prevailing number and proportion of circulating γδ T cells represents a risk factor for the appearance of fever after the first infusion of ZOL. We identified a threshold value of 3% γδT cells below which APR was never observed, whereas 78% of patients with % values higher than 3% experienced an APR. This threshold might be used for implementing preventing strategies for APR, for example, based on the administration of statins in conjunction with aminobisphosphonates.13
It is known that both the risk of APR12 and the number and proportion of γδ T cells decreases with aging.14, 15 However, we found that also the age-adjusted γδ T cell number and proportion remain significant determinants of the occurrence of APR.
This observation gives additional support to the hypothesis that γδ T cells are the most relevant target cells for the occurrence of APR. The larger the number of circulating γδ T cell the greater the production of proinflammtory cytokines. On the other end the independent protective effect of aging seems to suggest that other age-associated factors (reduced cytokine secretion?) might be involved.
It is well established that APR is considerably both more common and more severe after the first infusion of N-BPs.1, 12 In keeping with our results, it is expected that N-BPs treatment is associated with persitent decrease of circulating γδ T cell. This information is not available while circulating γδ T cells were reported both to increase9 and to decrease16, 17 a few days after iv N-BPs.
In conclusion, the results of this study indicate that the proportion of circulating γδ T cells, together with age, are important determinant of the occurrence of APR after intravenous infusion of ZOL and possibly of any other N-BPs. The presence of more than 3% circulating γδ T cells might be used in the future for predicting the occurrence of APR.
All the authors state that they have no conflicts of interest.
Authors' roles: Study design: MR, SA, DG. Study conduct: MR, OV, RO, AV, EF. Data collection: MR, OV. Data analysis: MR, SA. Data interpretation: MR, SA, OV, EF, DGG. Drafting manuscript: SA, MR. Revising manuscript content: OV, RO, AV, EF, DG. Approving final version of manuscript: All. SA takes responsibility for the integrity of the data analysis.