Is retention of zoledronic acid onto bone different in multiple myeloma and breast cancer patients with bone metastasis?


Address correspondence to: Kent Søe, MSc, PhD, Vejle Hospital, Department of Clinical Cell Biology, Kabbeltoft 25, 7100 Vejle, Denmark. E-mail:


Zoledronic acid (Zol) is used to treat bone disease in both multiple myeloma (MM) and breast cancer patients with bone metastasis (BC). However, bones of MM and BC patients show a difference in retention of the bisphosphonate used for bone scintigraphy. Therefore, we hypothesized that disease-specific factors may differently influence Zol retention in MM and BC patients. We tested this hypothesis in an investigator initiated phase II clinical trial in which we compared the whole-body retention (WBrt) of Zol in a cohort of 30 multiple myeloma (MM) and 30 breast cancer (BC) (20 Zol naive and 40 with six or more previous administrations). On average, 62% of the administered Zol was retained in the skeleton of both MM and BC patients and independently of the number of treatments. WBrt of Zol did not correlate with cross-linked C-telopeptide (CTX) levels, but linear regression analyses showed that WBrt of Zol correlated with bone-specific alkaline phosphatase (bALP) levels in BC (p = 0.001), and with CTX/bALP in Zol naive MM patients (p = 0.012). Especially in BC patients, WBrt correlated with age (p = 0.014) independently of kidney function. In MM patients WBrt was found to primarily correlate with the extent of bone disease (p = 0.028). Multivariate linear regression analyses of the entire cohort pointed out that WBrt of Zol was best predicted by age (p < 0.000), osseous lesions (p < 0.001), and the preceding Zol dosing (p < 0.005) (r2 = 0.97). Comparing bone scintigrams with CT/X-ray images showed a poor correlation between sites of active bone disease and binding of scintigraphy bisphosphonate in 36% of MM patients and in 13% of BC patients. We conclude that WBrt of Zol is primarily determined by two non-disease related factors and only one disease related, but that there may be differences in retention or drug delivery at individual sites of bone disease between MM and BC patients. In order to find the optimal dosing of Zol, these observations should be taken into account.


More than 40 years ago the first important discoveries were published on bisphosphonates (BPs), which paved the way for the first BP used in clinical practice, etidronate, in the 1970s. Now, at least 11 BPs are registered for various clinical applications. Today BPs are primarily used to treat bone loss in cancer as well as osteoporosis, but also for diagnostic use, in which radioactively labeled BP is used for bone scintigraphy.[1]

One of the most recent and also most potent BPs is zoledronic acid (Zol), also called zoledronate. For about a decade it has been used to treat bone disease in the context of cancer bone metastases and multiple myeloma (MM).[2] According to the guidelines, cancer patients should be given 4 mg Zol/month. However, BPs have been found to have long-lasting effects on bone turnover[3-6] and there have been concerns regarding BPs causing osteonecrosis of the jaw[7-9]; this has resulted in a debate about the optimal dosing and duration of treatment with Zol in cancer patients.[6, 10-14] In light of this important discussion it is important to know how much Zol is retained per infusion and which factors may influence how much Zol is retained. These aspects have not been extensively investigated in humans (as discussed in a recent review[11]), but two studies concluded from investigations of mixed cohorts of cancer patients that approximately 60% to 65% of the Zol dose was retained in the skeleton and that multiple treatments did not affect the level of retention.[15, 16] Cremers and colleagues[17] made similar observations for pamidronate in BC patients. Furthermore, it was found that the pretreatment levels of serum collagen type 1 cross-linked C-telopeptide (CTX) and bone-specific alkaline phosphatase (bALP) positively correlated with the whole-body retention (WBrt) of pamidronate. Similar findings were obtained in another study of patients with Paget's disease of bone.[18]

In our study we investigated the WBrt of Zol in MM compared to BC patients with bone metastasis. These two patient groups are both treated with 4 mg Zol,[19, 20] but a comparison of Zol retention between these two patient groups has not been performed, although it is well known that there are qualitative differences in the binding of BP used for scintigraphy between MM and BC patients. The so-called cold lesions in bone scintigrams have been known for many years and are more frequently found in MM compared to BC patients.[21, 22] These cold lesions are sites where the BP used for scintigraphy is excluded from foci (compared to surrounding signal intensities) of osteolytic activity, as demonstrated by CT scan or X-ray.[22-26] The known difference in binding of scintigraphy BP between MM and BC patients prompted us to study whether there was a more global difference in the retention of BP used for treatment of bone disease between MM and BC patients. A phase II trial was established and recruited a total of 60 patients, 30 MM and 30 BC patients with evidence of bone disease by CT scan or X-ray. Ten of each group had never been treated with BP before, whereas 20 from each group had previously received at least six treatments with Zol. We determined the WBrt of Zol and compared the results to outcomes of bone scintigraphy, bone markers, demographic data, and clinical data.

Patients and Methods

Approval of clinical trial

The study was conducted as a prospective nonrandomized phase II clinical trial in accordance with the Helsinki declaration and “Good Clinical Practice” (GCP) guidelines and was monitored by the GCP unit at Odense University Hospital, Denmark. It was approved by the Danish Medicines Agency, the Regional Ethical Committee, and the Danish Data Protection Agency. All patients gave informed consent. The study received EudraCT number 2007-003777-13 and trial registration number NCT00760370 at ( Comparison of Zometa Retention and Effect in Multiple Myeloma and Breast Cancer).


A total of 60 patients were included in the trial: 30 MM and 30 BC patients. Ten patients from each disease group had never received any BP previously and 20 patients had received at least six previous treatments with Zol for treatment of bone lesions as a result of the current malignancy. Zol was given as an intravenous (iv) dose of 4 mg infused over 15 minutes.

Inclusion criteria were as follows: patients diagnosed with MM or BC with metastasis to bone, scheduled for Zol treatment, and women in menopause (10–12 months after last menstrual cycle) or men 50 years or older. Both patient categories should show clear signs of bone lesions by X-ray or CT-scan. Important exclusion criteria were as follows: patients under the age of 18 years, women receiving hormone therapy that induces menopause, women receiving estrogen substitution therapy, or patients that had received treatment with any BP prior to the current malignant condition. At baseline all patients had a bone scintigraphy done. For patient characteristics please refer to Table 1. Creatinine clearance was calculated according to Cockcroft and Gault.[27] Patients with MM were grouped as (1) active myeloma encompassing newly diagnosed patients plus patients with progressive disease (PD) or (2) myeloma in remission (partial response [PR] + complete response [CR]) (see Table 1) according to the International Myeloma Working Group (IMWG) criteria of response.[28] Patients with BC were evaluated according to RECIST 1.0 criteria[29]; all patients had PD (see Table 1).

Table 1. Patient Characteristics
 AllMM (treatments = 0)MM (treatments ≥6)BC (treatments = 0)BC (treatments ≥6)
  1. MM = multiple myeloma; BC = breast cancer with metastasis to bone; M = male; F = female; = calculated creatinine clearance; PD = progressive disease; PR = partial response; CR = complete response.
Patients (n)6010201020
Sex (M/F)M = 19; F = 41M = 7; F = 3M = 12; F = 8M = 0; F = 10M = 0; F = 20
Age (years)63.5 ± 8.665.5 ± 5.867.7 ± 6.360.6 ± 8.059.9 ± 10.1
Weight (kg)72.6 ± 14.687 ± 18.073.7 ± 13.967.2 ± 10.667.1 ± 10.1
Height (cm)167.2 ± 9.0173.3 ± 5.3168.5 ± 9.8163.8 ± 9.1164.5 ± 8.2 (mL/min)84.4 ± 27.0103.1 ± 31.475.4 ± 21.179.4 ± 22.783.0 ± 25.1
Disease statusPD = 39; PR/CR = 21PD = 7; PR/CR = 3PD = 2; PR/CR = 18PD = 10; PR/CR = 0PD = 20; PR/CR = 0

On the day of their protocol-related Zol infusion, the patients were fasting and had blood samples taken for the analysis of CTX and bALP. Starting from the time of infusion the total volume of urine was collected for 48 hours. This time period was chosen because the vast majority of unbound BP is eliminated through the urine within the first 48 hours after infusion.[17] During this period the urine was stored at room temperature. Swiftly thereafter the urine was brought to the hospital, immediately pooled, weighed, and a sample was frozen at –20°C until further processing. The weight of urine was converted into milliliters (mL) by assuming a density of 1 g/mL. Total amount of urine collected per patient varied from 1424 to 7104 mL. After 2 months of storage, the urine was sent on dry ice to an external contractor for determining the concentration of Zol in the urine (see “Zol concentration in urine”). Fourteen days after the Zol infusion an end-of-study blood sample was taken for CTX and bALP measurement.

Bone scintigraphy

Whole-body images of the entire skeleton were acquired 2 to 3 hours after intravenous injection of 700 MBq 99mTc-HDP (Technescan, Petten, Netherlands) using a dual-head camera (Philips Sky Light/Philips Precedence, Best, Netherlands) with a low-energy high-resolution (LEHR) collimator, matrix size 1024 × 512, scan speed 15 cm/min. Digital images were visually interpreted using Hermes Medical and evaluated on the following three parameters. (1) Overall skeletal binding of the tracer relative to the activity of the renal parenchyma, scored as high, normal, or low. (2) Focal processes, scored as no abnormal foci; no foci of cancer suspicious character; foci of questionable cancer suspicious character; a single focus of cancer suspicious character; multiple foci of moderate cancer suspicious character; or multiple foci of very cancer suspicious character. For graphic representations, these were converted into an arbitrary scale: 0, 1, 2, 3, 4, or 5, respectively. (3) If abnormal foci were identified, then the intensity of focal activity was relative to the surrounding bone. Scored as cold lesion; weak, strong; or very strong. For graphic representations these were converted into an arbitrary scale: 0, 1, 2, or 3, respectively.

Evaluations of bone disease

Bone disease of MM patients was evaluated based on analyses of recent X-ray images obtained prior to the protocol-related treatment. The images were analyzed to indentify the extent of bone disease by estimating the number of osteolytic lesions. The following categories were used: 1 = no visible osteolytic lesions (despite of other signs of active bone disease); 2 = 1 osteolytic lesion; 3 = 2 to 4 osteolytic lesions; and 4 = > 4 osteolytic lesions. To evaluate the extent of bone disease in BC patients, we focused on the type of lesion, because all BC patients had extensive bone disease. The analyses were done based on recent CT images obtained prior to the protocol related treatment by an experienced MD in nuclear medicine. Because a single CT scan cannot detect bone formation but rather detects the presence or absence of mineral, we have chosen not to use the term “sclerotic,” because this by definition is coupled to new bone formation. But because the CT scan can detect “mineral-dense areas,” which most likely represent sclerotic areas, this is the term we have chosen for such areas. We have used the following categories to score the CT images for bone disease in BC patients: 1 = purely lytic; 2 = primarily lytic but with some minor pathological mineral dense areas; 3 = equal number of lytic areas and areas with pathological mineral dense areas; 4 = primarily pathological mineral dense areas with minor lytic areas; and 5 = all lesions were mineral dense areas.

Zol concentration in urine

Analysis of Zol concentration in urine samples was performed by the external contractor, SGS Cephac (Saint-Benoît, France) through the use of a specific radioimmunoassay as described.[30] In brief, polyclonal rabbit anti-zoledronic acid antibody was coated on microtiter plates by using sheep anti-rabbit antibody. A competitive binding assay was performed using 125I-labeled Zol and unlabeled Zol (as a standard) or urine samples were allowed to compete for the binding sites of the antibody. After washing, the extent of antibody-bound, 125I-labeled Zol was determined by gamma counting. Assay performance was performed along with the specific measurement and showed an assay accuracy of 88% to 109% and intraassay and interassay coefficients of variation <20%. Linearity of dilution was established for concentrations exceeding the upper range of the assay. The limit of quantification was 10 ng/mL.

Bone markers

Blood samples from fasting patients were taken on day 0 just prior to their protocol-related Zol-infusion and 14 days after the infusion. Blood samples for bALP measurement were collected in a noncoated tube, whereas the samples for CTX measurement were coated with EDTA. The samples were frozen at –80°C until further processing. Concentrations were measured by full-automatic ELISA. bALP was measured using Alisei (Seac Radim Group, Calenzano, Italy) and CTX on a Modular E (Roche, Hvidovre, Denmark). ELISA analyses were done according to the instructions by the supplier: bALP ELA kit (Quidel Corp., San Diego, CA, USA); CTX-1 kit (Roche).


Statistical tests for normal distribution, t tests, chi-square, linear regression, and correlation analysis (Spearman correlation analysis) were done using the GraphPad Prism 4.01 software (GraphPad Software Inc., La Jolla, CA, USA). A p value ≤0.05 was considered as significant.

For multiple linear regression analyses and likelihood ratio tests, StataIC 12.1 (Stata Corp., College Station, TX, USA) was used. To ensure that no severe multi-colinearity existed between predictor variables, variance inflation factors (VIFs) were computed for each predictor variable within the data set. Because no VIF was >5, we did not exclude any predictor variable. Variance homogeneity of the standardized residuals was confirmed. To find the best predicting variables for WBrt of Zol, the least significant variables from the dataset were removed step by step. Likelihood ratio tests were used to ensure that the predictive value of the dataset was not lost in this process. This was done until the further removal of variables in a likelihood ratio test resulted in a p value ≤0.05. In this case, the new model was rejected and the previous model was accepted as the best model to predict WBrt of Zol.


Patient characteristics

The characteristics of the 60 patients included in the protocol are shown in Table 1. The difference in sex composition explains the differences recorded in weight, height, and creatinine clearance between the groups. Furthermore, the MM cohort was significantly older than the BC cohort. The differences reflect what may be expected for these two disease groups. There was no statistical difference between the two groups of patients apart from weight and creatinine clearance (the latter will be addressed more thoroughly in Søe and colleagues, unpublished work). Weight and creatinine clearance does not influence the retention of Zol (data not shown and Fig. 6F and Fig. 7). Thus, the differences do not disturb the interpretation of our results.

Bone scintigrams show clear differences in the binding pattern of scintigraphy BP between MM and BC patients

Figure 1A shows three bone scintigrams of MM patients that illustrate the different types of scintigrams obtained from 30 MM patients. Patient 2 in general shows a weak binding, but at two sites in particular the scintigrams show a distinct and local absence of binding (marked by arrowheads): a cold lesion. A cold lesion (arrowhead) is also seen in patient 6. Apart from this, patient 6 shows weak accumulation in spots at the ribs (arrows) that may represent pathological fractures. But, in general, both patients 2 and 6 show no distinct accumulation of scintigraphy BP despite the identification of several osteolytic lesions in both patients (as observed on recent X-ray images). In contrast, patient 13 shows a strong accumulation of scintigraphy BP at isolated spots (arrows) that are clearly pathological and correlate with the number of osteolytic lesions. Thus, in this patient the scintigraphy BP seems to bind specifically to affected bone sites. In BC patients 67 and 65, the scintigrams show strong specific binding of scintigraphy BP to sites of pathological bone turnover (marked by arrows). In contrast, in BC patient 60 no binding of scintigraphy BP can be demonstrated although several bone metastases with osseous lesions showing extensive mineral-dense areas (which may indicate sclerotic changes) were found on CT images (data not shown). However, this patient showed no cold lesion, but rather just lack of focal binding in general. Overall no cold lesions were observed in BC; however, they were observed in 6 MM patients (20%) (Fig. 1C, statistically significant chi-square test p < 0.05). Overall we found that there was a poor overlap between the scintigrams and the extent of bone disease as determined by X-ray/CT scan in 10 of 28 MM patients (6 with cold lesions and 4 with general weak binding; 36%) and 4 of 30 BC patients (no cold lesions but all with general weak binding as observed for patient 60; Fig. 1B; 13%), emphasizing that binding of scintigraphy BP is less site-specific in MM than BC (statistically significant, chi-square test p < 0.05). In general, the scintigrams were less intense in MM compared to BC (Fig. 1D) and the foci were less numerous (Fig. 1E). These results raise the possibility that there may be qualitative and quantitative differences in the binding of Zol between MM and BC.

Figure 1.

Qualitative differences in binding of bone scintigraphy bisphosphonate. Representative bone scintigrams of (A) 3 MM and (B) 3 BC patients. Arrowhead, points to “cold lesion”; arrow, points to a focal high affinity binding site. (C) Number of patients that fall in the different categories of scintigram intensities. CL, cold lesion; W, weak; S, strong; VS, very strong (statistics: chi-square p = 0.033) (MM, n = 30; BC, n = 30). Black bars, MM; gray bars, BC. (D) Intensity of scintigram signals (arbitrary scale; please refer to Patients and Methods for definitions) (MM, n = 30; BC, n = 30). (E) Number of foci on scintigram (arbitrary scale, please refer to Patients and Methods for definitions) (MM, n = 30; BC, n = 30).

WBrt of Zol is not different between BC and MM patients

There was no difference in WBrt of Zol between MM and BC (Fig. 2A). MM patients on average retained 2.5 mg Zol out of the 4 mg dose (63%) and BC patients 2.4 mg (60%). It is noted that there were big differences in the extent of retention between individuals ranging from 1.1 mg (28%) to 3.6 mg (90%). There is no statistical difference in WBrt between those that were treated for the first time and those that have received at least six previous treatments (Fig. 2B). If the total number of previous treatments with Zol for each individual were considered there was no statistical support for a saturating effect. A patient who has received up to 60 treatments with Zol still retains the same fraction as a patient treated with Zol for the first time (Fig. 2C). This gives the impression that there apparently is an infinite potential of bone to retain Zol. However, the fact that there is a slight but significant impact of the dosing regimen on WBrt of Zol (Fig. 2D) suggests that frequent treatments could result at least in partial saturation, especially in BC patients. This highlights that dosing is more important for the extent of retention than the accumulated number of treatments per se.

Figure 2.

WBrt of Zol depends on dosing. WBrt of Zol in (A) MM and BC cohorts (statistics: t test, p = ns), (B) in MM and BC depending on the treatment subgroups, those that have never received Zol-treatment before and those that have received a minimum of 6 previous treatments (statistics: t test), (C) in each individual and their prior number of treatments (statistics: Spearman correlation analysis, p = ns), and (D) in each individual according to their prior dosing regimen (average dose of Zol per month) prior to inclusion in the protocol. An average of 4 mg/month represents a monthly treatment whereas 2 mg/months represents on average treatment every second month (statistics: linear regression analysis; MM, black line, r2 = 0.05, p = 0.2623; BC, gray line, r2 = 0.18, p = 0.0184). Black diamonds, MM; gray triangles, BC.

WBrt of Zol correlates with bone markers

Levels of bALP, a biomarker of osteoblastic activity, predicts the WBrt of Zol in BC patients both in general (Fig. 3A) and when split in subgroups of patients previously treated or not (Fig. 3B). The fact that statistical significance is borderline as shown in Fig. 3B most likely only reflects loss of power because of the lower number of individuals and because the two lines are very similar. In contrast, there is no correlation between bALP and WBrt of Zol in MM, neither for the cohort (Fig. 3C) nor as separate groups (Fig. 3D).

Figure 3.

Correlation of WBrt of Zol with bALP. WBrt of Zol in (A) all BC patients, bALP levels upon inclusion in the protocol compared to WBrt of Zol (statistics: linear regression analysis, r2 = 0.33, p = 0.0011), (B) BC patients stratified according to the treatment subgroups (statistics: linear regression analysis, untreated, gray line, r2 = 0.39, p = 0.0537; at least 6 treatments, black line, r2 = 0.16, p = 0.0845), (C) all MM patients, bALP levels upon inclusion in the protocol compared to WBrt of Zol (statistics: linear regression analysis, r2 = 0.01, p = 0.5889), and (D) MM patients according to the treatment subgroups (statistics: linear regression analysis, untreated, gray line, r2 = −0.06, p = 0.4791; at least 6 treatments, black line, r2 = 0.08, p = 0.2276). Gray dots, untreated patients; black boxes, at least 6 treatments.

The bone resorption marker, CTX, was not found to correlate with WBrt either for BC or for MM patients (data not shown). Previously it was found that the ratio between CTX and bALP was a useful prognostic marker for osteolysis in MM.[31] A linear correlation analysis showed that there was no correlation between CTX/bALP ratio and WBrt of Zol in BC patients (Fig. 4A, B); however, for MM patients it was borderline significant (Fig. 4C). When splitting the MM cohort into subgroups it became evident that it was only the previously untreated MM patients in whom the WBrt of Zol showed a correlation with CTX/bALP, whereas this was not the case for MM patients treated multiple times (Fig. 4D).

Figure 4.

Correlation of WBrt of Zol with CTX/bALP ratio. WBrt of Zol in (A) all BC patients, CTX/bALP ratio upon inclusion in the protocol compared to WBrt of Zol (statistics: linear regression analysis, r2 = 0.001, p = 0.8549), (B) BC patients stratified according to the treatment subgroups (statistics: linear regression analysis, untreated, gray line, r2 = 0.12, p = 0.3310; at least 6 treatments, black line, r2 = 0.024, p = 0.5332), (C) all MM patients, CTX/bALP ratio upon inclusion in the protocol compared to WBrt of Zol (statistics: linear regression analysis, r2 = 0.12, p = 0.0661), and (D) MM patients stratified according to the treatment subgroups (statistics: linear regression analysis, untreated, gray line, r2 = 0.57, p = 0.0119; at least 6 treatments, black line, r2 = 0.00, p = 0.8981). Gray dots, untreated patients; black boxes, at least 6 treatments.

WBrt of Zol correlates with the extent of bone disease

With respect to MM patients we found that the number of osteolytic lesions correlated significantly with WBrt of Zol in MM (Fig. 5A). Because BC patients displayed a gradient of osseous lesions ranging from purely lytic (category 1) to purely mineral-dense areas (category 5; possibly reflecting sclerosis; see Patients and Methods for definitions), we analyzed whether this may significantly affect WBrt of Zol (Fig. 5B). We did not find a significant correlation, but with a p value of 0.12 there is some extent of correlation, which becomes more evident in the multiple linear regression analysis shown in Fig. 7. We also scored the extent of osseous changes in BC patients, but including this parameter into the correlation did not noticeably change the correlation (data not shown). Thus it seems that it is primarily the type of osseous lesion that affects retention (although not significant in this single regression analysis), most likely because all BC patients in our cohort had extensive bone disease and therefore little variation.

Figure 5.

Correlation of WBrt of Zol with the extent of bone disease. WBrt of Zol in (A) MM patients and (B) BC patients depending on the extent or type of osseous lesions. (A) WBrt of Zol in MM patients depending on their extent of bone disease. The arbitrary scale for osteolytic lesions was defined as follows based on analyses of X-ray images: 1 = no osteolytic lesions visible; 2 = 1 osteolytic lesion; 3 = 2–4 osteolytic lesions; 4 = > 4 osteolytic lesions (statistics: linear regression analysis, n = 28, r2 = 0.17, p = 0.0279). (B) BC patients depending on their type of osseous lesion. The arbitrary scale was defined as follows based on analyses of CT scans: 1 = purely lytic; 2 = primarily lytic but with some minor pathological mineral dense areas; 3 = equal number of lytic areas and areas with pathological mineral dense areas; 4 = primarily pathological mineral dense areas and minor lytic areas; 5 = all lesions were mineral dense areas (statistics: linear regression analysis, n = 30, r2 = 0.08, p = 0.1233).

Age is a major determinant for WBrt of Zol

The age of the patients ranges from 45 to 83 years, and in Fig. 6A it is clearly seen that age is a significant determining factor of the WBrt of Zol. Based on the graph in Fig. 6A it can be estimated that an 80-year-old retains 41% more Zol than does a 40-year-old, making age a very powerful predictor of WB Zol retention. But this is only statistically significant for BC patients (Fig. 6B) and in particular for the previously untreated group with an r2 of 0.59 (Fig. 6C). MM patients show no significant correlation between age and WBrt of Zol, although for the previously treated group it is almost significant (Fig. 6D). Within the cohort we also found (as expected) a strong and significant negative correlation between age and creatinine clearance, both for BC and MM patients (Fig. 6E). However, it seems that WBrt of Zol correlates with age independently of kidney function because creatinine clearance does not correlate with Zol retention (Fig. 6F).

Figure 6.

WBrt of Zol increases with age. WBrt of Zol in (A) both BC and MM patients (n = 60) depending on their age (statistics: linear regression analysis, r2 = 0.11, p = 0.0092) or (B) BC (n = 30) and MM (n = 30) separately (statistics: linear regression analysis, MM, black line, r2 = 0.02, p = 0.4798; BC, gray line, r2 = 0.20, p = 0.0137). (C) Subgroups BC patients (statistics: linear regression analysis, untreated, gray line, r2 = 0.59, p = 0.009; at least 6 treatments, black line, r2 = 0.11, p = 0.15). (D) Subgroups MM patients (statistics: linear regression analysis, untreated, gray line, r2 = 0.18, p = 0.22; at least 6 treatments, black line, r2 = 0.15, p = 0.087). Gray dots, untreated patients; black boxes, at least 6 treatments. (E) Influence of age on the creatinine clearance (statistics: linear regression analysis, MM, n = 30, black line, r2 = 0.27, p = 0.0035; BC, n = 30, gray line, r2 = 0.29, p = 0.0020). (F) Influence of creatinine clearance on WBrt of Zol (statistics: linear regression analysis, MM, n = 30, black line, r2 = 0.04, p = 0.2647; BC, n = 30, gray line, r2 = 0.03, p = 0.5911).

Age, extent of bone disease, and dosing are the strongest predictors of WBrt of Zol in the cohort

In order to find those variables which best and independently predict WBrt of Zol for the entire cohort (except for 2 patients from each disease group due to missing information for a few of the parameters), we performed a multiple linear regression analysis coupled to a likelihood ratio test. From all the variables tested (Fig. 7, left column) we found “age” to be the most significant variable, followed by “osseous lesions” and “mg Zol/month before” (Fig. 7, right column).

Figure 7.

Three independent variables best predict the WBrt of Zol. An initial multiple linear regression model was generated including 15 different variables (in addition to the statistical variable “_cons”: constant or intercept) (table to the left) that could be considered to be of relevance for predicting WBrt of Zol in BC (n = 28) or MM (n = 28) patients. The table on the right-hand side shows the final optimized model containing those three variables that best predict WBrt of Zol together with the statistical results of the multiple linear regressions. In between the tables is shown the statistical parameters of the likelihood ratio test. The variable “osseous lesions” combines the information on the number of osteolytic lesions in MM patients and the type of osseous lesion observed in BC patients. cr. = creatinine; scinti. = scintigraphy.


Although numerous studies have characterized the binding of BPs to bone or mineral, very little information is available on the actual retention and sites of retention in humans.[11] We therefore found it highly relevant to investigate these questions in humans. We found that the WBrt of Zol after 48 hours was around 2.5 mg in both MM and BC, corresponding to about 63% of the dose given. This is in good agreement with Chen and colleagues,[15] who found the WBrt of Zol to be 63% at 24 hours postinfusion, as well as with Skerjanec and colleagues,[16] who found a WBrt of around 65% at 24 hours postinfusion with 4 mg Zol. Both studies were performed in mixed cohorts of cancer patients. Our results are also in good accordance with the retention of other BPs.[17, 18, 32-35] Furthermore, we found that multiple treatments with Zol did not significantly affect the WBrt of Zol. Thus, at the first glance we did not find any differences in the WBrt of Zol between MM and BC patients. However, it is possible that the measurement of WBrt may miss differences in position-specific retention. This indeed seems to be the case when comparing the retention of scintigraphy BP of the patients prior to the protocol-related Zol treatment.

Clear differences were found between MM and BC patients in the retention of scintigraphy BP, both semiquantitatively (Fig. 1D, E) and qualitatively (Fig. 1AC). In general, scintigrams of 87% of BC and 64% of MM patients had clear cancer-suspicious foci that also correlated with the general pathologic bone status of the patient as identified by X-ray or CT scan. This could mean that BP may not bind specifically to sites of high bone turnover, where it is most acutely needed, in 13% of BC and 36% of MM patients. This coincides with the previously observed specificity of bone scans, in which up to 25% of bone scintigrams may show false-negative results in BC[25, 26, 36] and up to 60% in MM.[22, 23] Furthermore, we found that the scintigrams of 20% of all MM patients displayed cold lesions, which is in good accordance with previous findings,[21-23] but we did not find any cases in BC patients, which is also in agreement with the literature.[21] Because it has been documented that cold lesions coincide with sites of osteolytic lesions,[22-26] this suggests that BP is practically excluded from sites of osteolytic lesions in these MM patients. This is of important therapeutic value because it may imply that a large fraction of the MM patients and possibly a smaller fraction of BC patients may not have the full benefit of Zol treatment because Zol may not be bound at all pathological sites.

It is not surprising that we found a correlation between bALP and WBrt of Zol in BC patients (Fig. 3A, B) because this confirms previous results.[17, 18] However, no reports have been published on the parameters responsible for BP retention in MM. It is therefore more surprising that only CTX/bALP was found to correlate with WBrt of Zol, especially because neither CTX nor bALP alone had any predictive value. However, it is possible that the CTX/bALP ratio may be a surrogate marker for the extent of bone disease and it could therefore be that this is the true determinant of retention in MM as indicated in Figs. 5 and 7. But it is very interesting that the predictive value of the bone turnover markers is so different between BC and MM patients although the WBrt of Zol is the same. This suggests that there are other factors which determine binding of Zol than just bone formation or resorption.

In this context it is interesting that we found a correlation for the entire cohort between the WBrt of Zol and the age of patients. However, it was also evident in a linear regression analysis that this correlation was mostly caused by a good correlation for BC patients, especially for the previously untreated. Interaction analyses have ruled out that the differences between BC and MM with respect to age are caused by the presence of males in the MM cohort and not a result of time from disease onset (ie, the older the patients are the longer they have been sick), differences in bALP, CTX, or CTX/bALP levels (data not shown), or by reduced kidney function with age (Fig. 6E, F, and Fig. 7). We speculate that in the more elderly patients endocortical trabecularization may be progressed, thereby generating a larger bone surface for general nonpathologic binding of Zol. Apparently this effect is more pronounced in BC patients than in MM although the difference between the two disease groups is not statistically significant (interaction analysis), which may be due to smaller variation of age in the MM cohort compared to BC. Fogelman and Bessent[37] found that the WBrt of scintigraphy BP increased significantly with age in both sexes in a cohort of 250 healthy volunteers and thereby supports our interpretation.

Our data show that WBrt of Zol is independent of the accumulated number of infusions in accordance with previous findings for pamidronate[17] and Zol.[15, 16] Given the slow release of Zol from bone tissue, this implies that a reservoir of Zol is generated in both MM and BC patients. Based on the pharmacokinetics and calculations on pamidronate retention presented by Cremers and colleagues,[17] we could estimate that a year of treatment (12 infusions of 4 mg) with Zol results in an accumulation of approximately 24 mg Zol (50% of a yearly dose). This is in accordance with other observations.[15-17, 38, 39] However, although a large reservoir is built up, we can still see that dosing is a strong determinant for retention of Zol. This strongly suggests that although there is a global reservoir in the skeleton, there may be local patches where it is rapidly released and is therefore available for binding of more Zol. However, the difference is not dramatic and may therefore suggest that, overall, most Zol is bound nonspecifically in the skeleton and not preferentially at sites of high bone turnover, especially not in MM patients (based on scintigrams). This is very clearly seen from our multiple linear regression analyses in which a likelihood ratio test showed that two general parameters, age and dosing, explain most of the WBrt, whereas it is only the number or type of osseous lesions that indicate any dependency on specific binding to sites of pathological bone turnover. Together, these three variables predict about 97% of the variation in Zol retention. Thus, although a reservoir of Zol is built up, this may not accumulate at sites where it is most needed but rather in the skeleton overall. This has to be considered when discussing dosing regimens with BPs. Small local increases in bone turnover may not be detected as a result of the very powerful global, but not necessarily complete, suppression of bone resorption. Lund and colleagues[31] recently demonstrated in a MM cohort that even very small intrapatient-specific increases of approximately 0.04 µg/L (in the particular study this reflected an increase from 0.17 to 0.21 µg/L) were sufficient to predict onset of progressive disease. This study was done on a patient cohort in which 93% had received or were currently receiving BP treatment; thus, CTX values were already very low. We therefore speculate that if a local relapse or continuation of osteolytic processes occur, such local bone destruction during Zol treatment can only be detected by bone resorption markers if they are analyzed frequently (eg, monthly) and if the patient is used as his/her own reference. Thus, local sites of progressive bone disease may easily be missed in everyday practice. It is a well-known phenomenon that skeletal-related events can progress or arise despite effective global suppression of bone resorption markers by BPs.[20, 40-43]

We suggest that a possible explanation for continued progression of bone disease despite treatment with powerful BPs such as Zol may be the result of an ineffective retention of Zol at least in some patients. Why may there be such a local insufficient binding of Zol? We find it likely that this may be related to inappropriate vascularization at local sites. It has been demonstrated that human trabecular bone remodeling occurs within so-called bone remodeling compartments (BRC)[44] and that these are frequently broken or lacking in MM patients compared to controls[45] as a result of a breakdown of the canopy.[46] These compartments are intimately connected with vasculature.[45, 47] Local loss of blood flow near the bone surface may explain the lack of sufficient retention of Zol at local sites, such as cold lesions. A possible effect on the vasculature may be directly caused by tumor load, which has also previously been considered in the context of false-negative retention of scintigraphy BP.[22-24, 26] It is also possible that a local lack of bone formation, which may be more pronounced in MM than BC patients, may explain the local lack of retention, especially in MM patients. However, as shown in Fig. 1B (patient 60) a high degree of pathological mineralization is not always a guarantee for labeling with scintigraphy BP. Therefore, there may also be another explanation that may be related to vascularization and/or BRCs. We have previously shown that when BRCs are present bone resorption and formation is coupled, whereas if BRCs are broken they are uncoupled and only resorption takes place,[45, 46, 48] without any formation, which correlates with the appearance of osteolytic lesions in MM patients.[46] Thus it is possible that both differences in the local rates of bone formation and the breakdown of BRCs and the resulting inefficient vascularization of these sites can explain the local insufficient binding of BP. This only shows that binding of BP in humans may be more complex than commonly assumed and that more attention needs to be given to retention and targeted delivery of BP in humans if treatment of bone disease is to be improved.

Currently, a lot of attention is given to the optimal dosing and duration of treatment in both MM and BC patients.[6, 10, 12-14, 49, 50] Although our study is small, we find that our data suggest attention should be given to the fact that a substantial fraction of patients still develop skeletal-related events or fractures despite long-term treatment with BP. More investigations are needed to find the reason for this insufficient effect of BP at a few local sites, and more awareness should be created that such local insufficiencies may not be detected if bone resorption markers are not measured frequently. More knowledge and awareness in this area may make it possible to improve drug delivery and thereby the effectiveness of treating bone disease.


KS has received partial research funding for the present protocol by Novartis. TP has received research funding from Novartis. EHJ has been co-investigator on several Novartis trials but has never obtained any financial support.


We are deeply grateful for the professional, excellent collection and documentation of data according to the GCP guidelines by the project nurses Anne Tørsleff and Annette Rehmeier, and for the collecting and freezing down most of the urine samples by Vibeke Nielsen and Birgit MacDonald. We thank statistician René Holst for statistical advice. We thank Novartis for financial support and interest in the protocol. We also thank the GCP unit of Odense University Hospital for an uncomplicated cooperation. We thank the Department of Radiology at Lillebaelt Hospital for free access to X-ray and CT images.

Authors' roles: Study design: KS and JMD; Recruitment of patients: TP, EHJ, and CTH; Data conduct: KS, TP, EHJ, CTH, and HBJ; Data collection: KS, TP, EHJ, CTH, and HBJ; Data analysis: KS; Interpretation of data: KS assisted by JMD; Drafting manuscript: KS; Revision of manuscript: all authors; Approval of final version of manuscript: all authors; KS takes responsibility for the integrity of the data analysis.