Site-specific human breast cancer (MDA-MB-231) metastases in nude rats: Model characterisation and in vivo effects of ibandronate on tumour growth

Abstract

Animal models are important tools to study the development of bone metastases and to evaluate strategies for their prevention and treatment. We here describe a new model in which tumour inoculation is achieved by injection of cancer cells into the femoral artery. This approach results in the development of multiple osteolytic lesions in the distal femora and proximal tibiae within 18 days after inoculation, with a success rate of 95–100% and no additional comorbidity. In untreated animals, osteolyses expanded continuously at a growth rate of 4.7–8.2 mm²/4 days, causing extensive destruction of resident bone structures by the tumour, significant loss of tibial bone density and a transient rise in urinary bone resorption markers. Continuous daily treatment with ibandronate (10 μg/kg) inhibited further growth of fully established metastases and reduced the mean osteolytic growth rate to 0.03 mm²/4 days. In lesions <6 mm bisphosphonate treatment resulted in a negative growth rate (−0.33 to −0.81 mm²/4 days). When ibandronate was started 3 days prior to tumour cell inoculation, the development of osteolytic lesions was substantially reduced (take rate only 17%) and bone density and structure were mostly preserved. We conclude that the intra-arterial approach used in this new model of metastatic bone disease results in site-specific osteolytic lesions with high take rates, steady tumour growth and no additional morbidity. While serial bone marker assessments did not prove useful to monitor osteolytic growth, our studies provide in vivo evidence that ibandronate treatment induces tumour remission by reversal of tumour growth. © 2003 Wiley-Liss, Inc.

Bone metastases are frequent complications of many malignant tumours, particularly of breast, prostate and lung cancers.1, 2 Animal models are important tools to study the mechanisms involved in the development of metastatic bone disease and to evaluate in vivo strategies for its prevention and treatment.3 Most current animal models utilise suspensions from malignant cell lines that are brought to the skeleton either via direct4, 5, 6 or indirect inoculation. With intracardial injections,3, 7–11 cells are inoculated at a relatively high tumour dose and, in most models, malignant dissemination occurs into both skeletal and visceral peripheries, resulting in varying degrees of animal comorbidity (e.g., cachexia, paraplegia). Direct inoculation of tumour cells into the bone marrow leads to site-specific osteolyses, but these models lack the processes associated with extravasation of tumour cells, while the injection itself causes bone damage that might result in nonspecific wound responses.12, 13, 14, 15

Furthermore, all currently characterised animal models of metastatic bone disease use either syngeneic tumours or human xenografts in small animals (e.g., mice) or syngeneic tumours in larger animals (e.g., rats). While small-animal models are often restricted to end-point analyses or radiographic follow-up, studies in larger animals allow the longitudinal assessment of bone turnover by serum or urine markers. So far, however, no site-specific models of metastatic bone disease using human breast cancer xenografts in rats have been reported.

We here describe a new model in athymic rats, using a direct approach by intraluminal inoculation of cancer cells into the femoral artery. The model permits sequential assessment of the development of site-specific metastases in untreated animals as well as the effects of pharmacologic interventions. The present report characterises the model and describes the effects of ibandronate on bone metastases. From these latter studies, we provide in vivo evidence that bisphosphonates reduce tumour burden and possibly induce tumour remission by reversal of tumour growth.

a.i., analytic interval; BMD, bone mineral density; CBA, computer-based analysis; DPD, desoxypyridinoline; DXA, dual X-ray absorptiometry; GFR, glomerular filtration rate; NK, natural killer; p.i., postinoculation; PTHrP, parathyroid hormone–related protein; PYD, pyridinoline; ROI, region of interest; S-Ca, serum calcium; S-TAP, serum total alkaline phosphatase; U-Ca, urinary calcium. The authors certify that they have not entered into any agreement that could interfere with their access to the data on the research or their ability to analyse the data independently, to prepare manuscripts and to publish them. F.B. is an employee of Roche Diagnostics GmbH, makers of ibandronate, and has declared a financial interest in this company.

MATERIAL AND METHODS

Animals

Male Sprague-Dawley rats, 7–8 weeks old and kept under standard conditions, and congenitally athymic male nude rats (rnu/rnu Rowett nudes) of the same age, kept under specific pathogen-free conditions, were obtained from Harlan-Winkelmann (Borchen, Germany). All animals had free access to tap water and were fed ad libitum with a commercial standard chow containing 1.1 g calcium and 0.8 g phosphorus/100 g dry weight (Sniff Spezialdiäten, Soest, Germany). After adaptation to laboratory conditions for 6 days, animals entered the study. During experiments, they were kept in single metabolic cages for 24 hr in either 7- or 3-day intervals. Experiments were approved by the governmental Animal Ethics Committee (Regierungspräsidum, Karlsruhe, Germany).

Tumour cells

The human oestrogen-independent breast cancer cell line MDA-MB-231¹⁶ was kindly provided by Dr. T. Guise (University of Texas, San Antonio, TX). Cells were cultured in DMEM supplemented with 10% FCS and 0.2% penicillin–streptomycin solution (medium and supplements from Boehringer Mannheim, Mannheim GmbH, Germany) at standard conditions (humidified atmosphere 37°C, 5% CO₂).

Inoculation of tumour cells

After incubation, tumour cells were harvested, washed twice with PBS, suspended in 0.5 ml of PBS and injected via a small polyethylene catheter (approx. 0.1 mm in diameter) directly into the femoral artery of both legs. Following injection of tumour cells, arteries were completely ligated. This technique was chosen to ensure the development of bone metastases in the hind limbs without systemic spread. However, the effects of complete arterial ligation on hind limb blood supply, local bone density and composition as well as systemic bone metabolism are unknown. In a pilot study, these effects were therefore assessed in 2 groups of male Sprague-Dawley rats (n = 10/group). In group A, both femoral arteries were injected with 0.5 ml of isotonic saline, followed by complete vascular ligation. Animals of group B received a sham operation without any manipulation of the hind-limb arteries. All procedures were performed under general anaesthesia using 50 mg/kg body weight of ketamine hydrochloride s.c. (Parker-Davis, Berlin, Germany) and 15 mg/kg body weight of 2% xylazine hydrochloride s.c. (Bayer, Leverkusen, Germany). Starting on day 0, body weight, 24 hr urine, blood samples and whole-body X-rays were taken at intervals of 7 days after the operation. Urine and serum samples were stored at −80°C until further analyses. Tibial BMD was measured on days 0 and 42 by DXA. All animals were killed on day 42 and the femora and tibiae collected for physicochemical analyses.

Experimental protocols

To determine the optimal number of tumour cells for injection, 4 groups of nude rats (n = 7 or 8/group) were inoculated with 1 × 10⁴ (group I), 1 × 10⁵ (group II), 1 × 10⁶ (group III) tumour cells or PBS alone (group IV). Blood samples, 24 hr urine specimens, whole-body X-rays and measures of body weight were obtained on day 0 and every 6th day p.i. BMD was measured at both tibiae on days 0 and 42 by DXA. All animals were killed on day 42 p.i. and both tibiae collected for histologic analyses.

A total of 120 animals, divided into 10 groups of 12, were studied following 3 main protocols (Fig. 1). In each animal, both hind limbs were inoculated and studied separately. Animals in groups with uneven numbers served as tumour-free controls and were inoculated with PBS alone at baseline (day 0). Animals in groups with even numbers were inoculated on day 0 with MDA-MB-231 tumour cells at a dose of 1 × 10⁵ cells/leg.

Figure 1. Summary of experimental protocols. Each group consisted of 12 animals. Animals in groups with uneven numbers received PBS (open triangle), while animals in groups with even numbers received tumour at a dose of 1 × 10⁵ cells/leg (solid triangle). Ibandronate was given at a daily dose of 10 μg P/kg body weight. Grey horizontal bar indicates duration of ibandronate treatment. In subprotocols 1a, 2a, 5a and 6a, 4 animals each were killed (cross) on days 6, 12 and 18 to study early changes after tumour inoculation.

Protocol 1 was used to establish the model without any interference following the baseline intervention. Animals were injected with either PBS (group 1, control) or tumour cells (group 2). All animals remained untreated until death on day 30. A subprotocol of equal size and design was used for assessment of the early stages of the model. To this aim, 4 animals each of groups 1a (PBS) and 2a (tumour) were killed on days 6, 12 and 18 p.i. (Fig. 1).

Protocols 2 and 3 were designed to study the effect of systemic bisphosphonate administration on the development of bone metastases. Ibandronate [1-hydroxy-3-(methylpentylamino)propylidene] bisphosphonic acid (monosodium salt monohydrate)]17 (Roche Diagnostics GmbH, Mannheim, Germany) was dissolved in isotonic saline at a pH of 7.4. Doses are expressed as μg P/kg of ibandronate (1 μg P/kg corresponds to 0.016 μmol/kg or 5.14 μg P/kg of free acid equivalents). The volume of administration was 2 ml/kg body weight.

In protocol 2, animals were injected with either PBS (group 3, control) or tumour (group 4). From day 18 p.i. onwards (i.e., after radiographic appearance of lytic bone lesions in the tumour-bearing group), all animals received daily s.c. injections of ibandronate at 10 μg P/kg (Fig. 1).

In protocol 3, starting 3 days prior to the injection of tumour cells or carrier, animals received daily s.c. ibandronate injections (10 μg P/kg). On day 0, animals received either PBS (group 5, control) or tumour (group 6). Treatment with ibandronate was continued until day 30 p.i. A subprotocol of equal design and size was used to assess the early stages of the model. To this aim, 4 animals each of groups 5a (PBS) and 6a (tumour) were killed on days 6, 12 and 18 p.i. (Fig. 1).

Assessments

Timing of analyses.

Measures of body weight, whole-animal radiographs, 24 hr urine collections and blood samples were obtained in all groups on days −3, 0, 6, 12, 18, 22, 26 and 30 p.i. (subprotocols were terminated earlier, see above). Bone density measurements were performed in vivo on day 0 and immediately before death. All other analyses were performed after death at day 30.

Radiographic examinations.

Radiographs were taken with rats under general anaesthesia. Animals were placed in a.p. position on a high-resolution mammography film (Strukturix D4 DW ETE 18 × 24 cm; (AGFA-Gevaert, Köln, Germany) and exposed to an X-ray at 55 kV, 5 mA for 90 sec using a radiographic inspection unit for animals (Torrex 120D; EG&G Astrophysics Research, Long Beach, CA). Radiographs were scanned with a resolution of 480 dpi and analysed by 2 different investigators blinded to the protocol. Analyses were done at a magnification of ×300, using a CBA program (CBA 8000, Leica QWin Pro V2.0; Leica, Heidelberg, Germany). Metastatic foci >0.5 mm in diameter (detection limit) were recognised as radiolucent lesions and manually delineated to determine the number (n), the perimeter (mm) and the area (mm²) of osteolytic lesions. Since initially separated lesions showed a strong tendency to confluate during tumour growth, only the total osteolytic area per bone and animal was used for further analyses.

Changes in osteolytic area were assessed by comparing subsequent radiographs of each animal in 4-day intervals. The growth rate (in mm² per a.i.) was then calculated as follows: Growth rate = [osteolytic area at day x–osteolytic area at day (x–a.i.)]/a.i.

DXA.

DXA was performed under general anaesthesia using a Hologic QDR 1000/W and the high-resolution software for small animals provided by the manufacturer (Hologic, Waltham, MA). BMD was determined in lumbar vertebrae 1–4 and in the proximal quartile of both tibiae.

Femoral and tibial X-ray densitometry.

Femoral and tibial X-ray densitometry was performed using computerised X-ray analyses as described previously.18, 19 In brief, bones were X-rayed with an aluminium wedge as a reference object. The film was scanned and the electronic image then analysed by a real-colour image analysis system (CBA 8000; Leitz, Wetzlar, Germany). Density is expressed as millimeters of aluminium equivalents (coefficient of variation ≤1%). The ROI was the metaphyseal area of the distal femur and the proximal tibia between the 2 cortices.

Femoral and tibial tissue volume, dry weight and ash analysis.

Rat femora and tibiae were submerged in distilled water and hydrated for 24 hr under low vacuum in a desiccator. After blotting with moistened blotting paper, the tissue volume of bone (bone + bone marrow volume) was measured by volume displacement of distilled water using a pressure transducer (TSE 2000; TSE, Bad Homburg, Germany). To obtain femoral and tibial dry weight, bones were dried to constant weight at +80°C in an incubator and weighed after adaptation to room temperature. Bones were then ashed in a muffle furnace (M 110; Heraeus, Hanau, Germany) at 600°C for 24 hr, and ash weight/bone was determined. Each specimen was then dissolved in 5 ml of 5N HCl for atomic absorption spectrophotometry.

S-Ca and U-Ca.

S-Ca and U-Ca were determined by atomic absorption spectrophotometry (model 2100; Perkin-Elmer, Ueberlingen, Germany) after dilution of the sample with lanthanum nitrate (1% in 0.1%N HCl).

Serum and urinary creatinine and GFR.

Serum creatinine was measured in a multichannel analyser, while urinary creatinine was determined by an automated modification of the Jaffe technique.20 The GFR was calculated as follows: C_u × V_u = C_p × V_p = C_u × V_u/C_p, where C is the concentration and V is volume (u, urine; p, plasma). Concentrations of urinary parameters were corrected for GFR and the results expressed as pmol/(ml · min^–1).

S-TAP.

S-TAP was measured by an automated colorimetric assay using a BM/Hitachi (Tokyo, Japan) System 704 analyser, following the standard method of the International Federation of Clinical Chemistry.21 The intra- and interassay coefficients of variation were 5 ≤%.

Urinary cross-links.

Urinary concentrations of total PYD and DPD were measured by an automated, reverse-phase ion-pair HPLC, as previously described.22 In brief, urine aliquots were hydrolysed with HCl to convert all urinary cross-links to the peptide-free form. Following partition chromatography on CF1 cellulose, cross-links were separated by ion-pair HPLC and concentrations determined by fluorometry of the eluted peaks. The overall reproducibility of the assay, including the cellulose formation step, was between 4% and 8%. Urinary concentrations (mmol/24 hr) were calculated on 24 hr excretion.

Histology.

Histologic assessment of tumour burden was done in both tibiae. Bones were fixed in 10% neutral buffered formalin and subsequently decalcified in a 14% EDTA solution. Following conventional processing of paraffin embedding, sections were cut in the anterior, median and posterior thirds at a distance of 200 nm. Sections showing maximum tumour extent were used for estimating tumour burden by counting the number of metastases and measuring their respective areas by manual delineation at a final magnification of ×400. Results were then processed by CBA using specific software (CBA 8000, Leica QWin Pro V2.0).

Statistical analyses

Statistical analyses were performed using the SAS (Cary, NC) software package. Descriptive data are expressed as means ± SD. Between-group comparisons were performed using Student's t-test after testing for normal distribution of values. A p value of <0.05 was considered significant. Correlations between parameters were calculated as simple Spearman correlation coefficients.

RESULTS

Local and systemic effects of vascular manipulation

After inoculation of the tumour cells, both femoral arteries were fully ligated to prevent bleeding. When the effects of these vascular manipulations on local and systemic bone balance were evaluated, no difference between ligated and sham-operated animals was found with respect to tibial BMD, markers of bone turnover (S-Ca, S-TAP, U-Ca, U-DPD, U-PYD), tissue volume, dry weight and ash analysis of femoral and tibial bones analysed at day 42 p.i. Previous investigations from the same laboratory (F. Bauss, unpublished data) found a transient rise of serum creatinine kinase levels shortly after femoral artery ligation. However, no macroscopic or microscopic changes were found in muscle tissues after death at a later point in time. Complete ligation of the femoral arteries therefore appeared to have no detectable effect on bone macrostructure, metabolism and chemical composition.

Dose-finding study

Following the intervention on day 0, all animals showed a reduction in body weight. In all groups and animals, baseline body weight was regained by day 12–18. At study termination, body weight was not different between groups and no cachexia was observed. Four animals died prematurely due to reasons not related to the study and were excluded from all analyses.

The dose-dependent effects of increasing tumour cell numbers on radiographic take rate (i.e., the number of legs developing osteolytic lesions compared to the number of successfully inoculated legs), time to occurrence and remission of skeletal metastases and on tibial BMD are summarised in Table I. Compared to controls (group IV), tibial BMD on day 42 was significantly reduced in tumour-bearing animals as a function of cell numbers inoculated (p < 0.05). In contrast, concentrations of S-Ca, U-Ca and S-TAP remained similar between tumour-bearing and control animals throughout the course of the study. Compared to controls, a significant increase in U-PYD and U-DPD was seen in group II (1 × 10⁵ cells) on day 12 p.i.: PYD (group mean ± SD) 3.77 ±0.41 vs. 2.99 ± 0.25 nmol/mmol creatinine; DPD 3.49 ± 0.41 vs. 2.69 ± 0.21 (p < 0.05 for both parameters). No significant differences in U-PYD and U-DPD were noted in the other groups of the dose-finding study.

Table I. Dose Finding Study: Radiographic and Densitometric Results in Groups I–III Compared to Sham-Operated Controls (Group IV)

	Group I	Group II	Group III	Group IV
Take rate, number of legs developing osteolytic lesions compared to number of successfully inoculated legs. n.a., not applicable. * p < 0.05 ** p < 0.01 vs. controls
Number of animals	6	6	5	8
Number of inoculated cells	10^4	10^5	10^6	0
Day of first visible osteolytic lesions on X-ray	18	18	12	n.a.
Take rate (%)
Day 12	0/11 (0.0)	0/10 (0.0)	6/10 (60.0)	n.a.
Day 18	7/11 (63.4)	9/10 (90.0)	8/10 (80.0)	n.a.
Day 24	9/11 (81.2)	10/10 (100)	9/10 (90.0)	n.a.
Day 30	9/11 (81.2)	10/10 (100)	8/10 (80.0)	n.a.
Day 36	10/11 (90.9)	10/10 (100)	7/10 (70.0)	n.a.
Day 42	10/11 (90.9)	10/10 (100)	7/10 (70.0)	n.a.
Start of any spontaneous remission (day)	36	36	24	n.a.
Number of legs in remission on day 42	2/11 (18.2)	5/10 (50.0)	5/10 (50.0)	n.a.
Tibial BMD on day 42	0.216 ± 0.02*	0.213 ± 0.02**	0.210 ± 0.03*	0.228 ± 0.01

Histologic examination on day 42 p.i. showed large variations in the size of metastatic lesions. All tumours were highly vital by histologic judgement and typical for metastatic lesions induced by the MDA-MB-231 cell line.8 No further quantitative analyses were performed.

Since animals in group II showed homogenous occurrence of osteolytic lesions on day 18 p.i. and a late onset of spontaneous remissions, a dose of 1 × 10⁵ tumour cells was selected for the main experimental protocols.

Main experimental protocols

Following the intervention on day 0, all animals showed a reduction in body weight. In all groups, baseline body weight was recovered by day 18. At study termination, body weight was not different between groups and no cachexia was observed. Also, lumbar spine BMD was similar in tumour and control animals at death. Two animals died prematurely due to reasons not related to the study and were excluded from later analyses.

Protocol 1: model characterisation.

In animals injected with tumour cells (group 2), osteolytic lesions became radiographically visible on day 18 p.i. with a primary take rate of 95%. The number of legs showing osteolyses increased to 100% on day 22 p.i. and remained unchanged until the end of the study (Table II). When analysed by osteolytic area, lesions in all animals constantly increased in size (p < 0.001) (Table II, Figs. 2a,3a). Mean osteolytic growth rates varied between 4.72 ± 2.39 and 8.28 ± 11.67 mm²/a.i. and were not statistically different from each other (Table II). No specific radiographic changes were noted in control animals (group 1).

Table II. Development of Osteolytic Lesions in Tumour-bearing Animals

	Group 2 PBS (n = 11)	Group 4 IBN, days 18–30 (n = 11)	Group 6 IBN, days −3 to 30 (n = 12)
a.i., 4 days; OA, osteolytic area; IBN, ibandronate; n/a, not applicable/analyzed. The measurement <6 mm^2 and >6 mm^2 refers to the size of osteolytic lesions on day 18 p.i. * p < 0.05 ** p < 0.01 *** p < 0.001 vs. group 2 + p < 0.001 vs. day 18. ## p < 0.001 vs. lesions OA <6 mm^2 on day 18.
Take rate (n, % in parentheses)
Day 18	21/22 (95)	20/22 (91)	4/24 (17)
Day 22	22/22 (100)	20/22 (91)	4/24 (17)
Day 26	22/22 (100)	20/22 (91)	4/24 (17)
Day 30	22/22 (100)	18/22 (82)	4/24 (17)
Osteolytic area (mean in mm^2 ± SD)
Day 18	4.72 ± 2.39	6.47 ± 4.00	2.88 ± 3.89
Day 22	10.77 ± 4.57	11.01 ± 7.22	5.44 ± 3.99
Day 26	15.78 ± 7.41+	11.05 ± 7.76	10.02 ± 8.09
Day 30	21.69 ± 10.57+	11.52 ± 9.20*	12.94 ± 7.71
Mean osteolytic growth rate (mm^2 a.i. ± SD)
Day 18	4.72 ± 2.39	6.47 ± 4.00	2.88 ± 3.89
Day 22
Total group	6.05 ± 2.63	4.54 ± 3.76	2.55 ± 1.23
<6 mm^2	5.85 ± 3.12	2.25 ± 2.74*	n/a
>6 mm^2	6.58 ± 0.29	7.30 ± 2.92##	n/a
Day 26
Total group	5.01 ± 3.62	0.03 ± 3.45**+	4.59 ± 4.11
<6 mm^2	5.11 ± 4.17	−0.85 ± 1.51***	n/a
>6 mm^2	4.76 ± 2.13	1.10 ± 4.94##	n/a
Day 30
Total group	5.91 ± 3.95	0.47 ± 2.28***+	2.91 ± 1.26
<6 mm^2	5.03 ± 3.22	−0.33 ± 2.08**	n/a
>6 mm^2	8.28 ± 11.75	1.44 ± 2.32*	n/a

Figure 2. Development of metastatic bone disease in untreated animals (protocol 1 and subprotocol 1a). (a) Total osteolytic area in individual animals of group 2. (b) BMD in tumour-free and tumour-bearing animals before and 30 days after tumour inoculation. Values are means ± SD. *p < 0.05 vs. group 1. (c) Urinary excretion of total PYD in 24 hr urine collections corrected for GFR. Values are means ± SD (n = 11). *p < 0.05, **p < 0.01 vs. group 1. (d) BMD in tumour-free and tumour-bearing animals at 6, 12 and 18 days after tumour inoculation. Values are means ± SD (n = 4). *p < 0.05 vs. group 1.

Figure 3. Radiographic development of osteolytic lesions without (a) and with (b) ibandronate treatment. (a) In untreated tumour-bearing animals (group 2), osteolytic lesions became visible on radiographs on day 18 p.i. (d18). Lesions were located in the distal femur and the proximal tibia, showing a constant progression of osteolytic area and coalescence over time. (b) When tumour-bearing animals were treated with ibandronate from day 18 p.i. onwards (group 4), osteolytic growth was reduced or halted in the majority of lesions at 4 days after initiation of therapy (d22). The following radiographs show a further and continuous decline in osteolytic size until study termination (d30).

Tibial BMD was similar in both groups at baseline. On day 30, tibial BMD in tumour-bearing animals was significantly lower (p < 0.05) than in controls (Fig. 2b). In both of these groups, tibial BMD was lower than in the ibandronate-treated animals of protocols 2 and 3 (p < 0.01) (Fig. 5b, Fig. 7b).

Figure 5. Development of metastatic bone disease in animals treated with ibandronate from day 18 p.i. onwards (protocol 2). (a) Total osteolytic area in individual animals of group 4. (b) BMD in tumour-free and tumour-bearing animals before and 30 days after tumour inoculation. Values are means ± SD (n = 11). *p < 0.05 vs. group 3. (c) Urinary excretion of total PYD in 24 hr urine collections corrected for GFR. Values are means ± SD (n= 11). *p < 0.05, **p < 0.01 vs. group 3. Arrow and horizontal line indicate start and duration of ibandronate treatment in group 4.

Figure 7. Development of metastatic bone disease in animals treated with ibandronate from day −3 onwards (protocol 3, subprotocol 3a). (a) Total osteolytic area in individual animals of group 6. Only 3 animals developed bone metastases. (b) BMD in tumour-free and tumour-bearing animals before and 30 days after tumour inoculation. Values are means ± SD. *p < 0.05 vs. group 5. (c) Urinary excretion of total PYD in 24 hr urine collections corrected for GFR. Values are means ± SD (n= 11). ⁺p < 0.001 vs. day −3. (d) BMD in tumour-free and tumour-bearing animals at 6, 12 and 18 days after tumour inoculation. Values are means ± SD (n = 4).

Urinary excretion of collagen cross-links was significantly increased in the tumour-bearing group on days 12 (p < 0.05) and 18 (p < 0.01) p.i. (Fig. 2c), while no significant differences between tumour and control groups were noted for S-Ca, U-Ca and S-TAP. There was no relationship between changes in biochemical markers of bone turnover and osteolytic area or osteolytic growth rate.

Histologic examinations showed the predilection area for osteolytic lesions to be the distal femora and proximal tibiae, close to the epiphysial growth plate. Metastases showed marked destruction of trabecular and cancellous bone structures and replacement of bone marrow by tumour cells. Activated osteoclasts were seen in abundance along the remaining bone surfaces (Fig. 4a). Sections showing maximum tumour extent were chosen to measure individual metastatic areas by quantitative CBA. The mean metastatic area was closely correlated with same-day radiographic osteolytic area (r = 0.815, p < 0.0001). No histologic changes were observed in control animals.

Figure 4. Representative sections of tibial bones from tumour-bearing animals obtained on day 30 p.i. (haematoxylin and eosin stain, ×60). (a) Untreated animal (group 2) showing massive metastatic growth and almost complete osteolytic destruction of resident skeletal structures. At higher magnifications, numerous osteoclasts could be seen. The growth plate is shown at the right upper margin. (b) Animal treated with ibandronate after establishment of osteolytic lesions (group 4). The tumour is still visible, but osteolytic destruction of resident skeletal structures is less pronounced. At higher magnifications, only few osteoclasts were seen. The growth plate is shown at the right upper margin. (c) Animal treated with ibandronate before tumour inoculation (group 6). At higher magnifications, very few osteoclasts were visible. The growth plate is shown at the right upper margin. In most animals, no residual tumour is seen and skeletal structures are intact. As indicated in Table II, viable tumour was found in only 3 of 12 animals (4 of 24 legs) and, in these animals, the mean osteolytic area was similar to that seen in group 4.

Subprotocol 1 looked at early changes during days 6–18 p.i. in untreated rats. On day 18, osteolyses became radiographically visible in tumour-bearing animals with a take rate of 100%. Tibial BMD was similar in both groups at baseline and subsequently increased in controls (group 1a). In contrast, BMD did not increase in tumour-bearing animals (group 2a); and from day 12 p.i. onwards, tibial BMD was significantly higher in the control than in the tumour-bearing group (p < 0.05) (Fig. 2d). Changes in bone markers were similar to those observed in main protocol 1, with significant differences between groups in U-DPD on days 12 and 18. Histologic analyses of the tibiae obtained from animals killed on days 6 and 12 p.i. showed small but distinct metastatic foci, which were not visible on radiographs. Radiographic osteolyses visible in animals killed on day 18 were histologically verified as tumour tissue. Due to the small number of animals with osteolytic lesions on radiographs, no attempt was made to correlate histologic and radiologic analyses.

Protocol 2: effect of systemic ibandronate treatment on established bone metastases.

On day 18 p.i., the radiographic take rate in group 4 was 91%. After ibandronate treatment was commenced on day 18 p.i., this number decreased to 82% on day 30 p.i., but this numerical change was largely attributable to a complete remission of osteolytic lesions in 2 animals (Table II). While mean osteolytic areas were similar in groups 2 and 4 on day 18, no further growth was seen in the ibandronate-treated group from day 22 p.i. onwards (Figs. 3b,5a). As a result, the final osteolytic area in ibandronate-treated tumour-bearing animals was only 53% of the osteolytic area observed in untreated tumour-bearing animals (group 2, p < 0.05; Table II). During ibandronate treatment, mean osteolytic growth rates decreased rapidly to a nadir of 0.03 mm²/a.i. (p < 0.01 vs. day 18 p.i.). On days 26 and 30 p.i., growth rates were significantly lower than in control animals of group 2 and ibandronate-treated animals of group 6 (p < 0.01).

Tibial BMD was similar in both groups at baseline. On day 30, tibial BMD in tumour-bearing animals was significantly lower (p < 0.05) than in controls (Fig. 5b). Compared to untreated animals of groups 1 and 2, both ibandronate-treated controls (group 3) and tumour-bearing animals (group 4) showed significantly higher tibial BMD at study termination (p < 0.01) (Figs. 2b,5b).

Tumour-bearing animals showed a significant increase in the urinary excretion of collagen cross-links on days 6 and 12 p.i compared to tumour-free rats (p < 0.05) (Fig. 5c). However, urinary DPD did not differ between any of the groups once ibandronate treatment was started. In contrast, S-Ca levels decreased 4 days after initiation of ibandronate and remained significantly lower in the bisphosphonate-treated groups than in untreated tumour-bearing controls (p < 0.001). No significant differences between tumour and control groups were observed for S-TAP and U-PYD. Changes in markers of bone turnover did not correlate with osteolytic area or osteolytic growth rate.

Histologic examination showed the same predilection area for metastases as in protocol 1 with partial replacement of cancellous and cortical bone by tumour cells. Animals in group 4 had a mean metastatic area similar to that seen in the untreated control group. Also, the extent of malignant tissue, the destruction of resident skeletal structures and the activation of osteoclasts were much less pronounced in treated animals than in untreated controls (Fig. 4b). Due to the small sample size, these changes did not reach statistical significance. As in group 2, a close correlation was noted between quantitative histologic metastatic area and radiographic osteolytic area on day 30 (r = 0.782, p < 0.001).

While in untreated animals, osteolytic lesions appear to grow at a constant rate of 5–8 mm²/a.i., the effects of ibandronate on osteolytic growth appear to depend on the size of the lesion early during osteolytic development. When assessed radiologically by quantitative CBA, lesions with an osteolytic area <6 mm² on day 18 p.i. had a negative growth rate on days 26 and 30. In contrast, animals with an osteolytic area >6 mm² on day 18 p.i. maintained a positive growth rate at subsequent time points (Table II, Fig. 6). Histologic CBA of final metastatic areas on day 30 (MA_d30) revealed a similar picture: in group 2, the mean MA_d30 (±SD) was similar independent of initial osteolytic size on day 18 p.i. (OA_d18 < 6 mm,² MA_d30 = 5.91 ± 6.86 mm;² OA_d18 > 6 mm² MA_d30 = 7.37 ± 6.28 mm²). Conversely, in ibandronate-treated animals (group 4), MA_d30 differed significantly with initial lytic size (OA_d18 < 6 mm² MA_d30 = 4.60 ± 9.69; OA_d18 > 6 mm² MA_d30 = 14.56 ± 10.83, p < 0.05).

Figure 6. Osteolytic growth rates in untreated and ibandronate-treated, tumour-bearing animals (protocol 2) stratified by initial osteolytic size. (a) Initial size of osteolytic lesion <6 mm². Note the negative mean growth rates in the ibandronate-treated group on days 26 and 30, indicating a decrease in the size of osteolytic lesions. (b) Initial size of osteolyses >6 mm². Note the positive mean growth rates in the ibandronate-treated group on days 26 and 30, indicating reduced but ongoing growth of osteolytic lesions. Untreated animals show no significant change in growth rate, indicating continuous growth of osteolytic lesions. Values are means ± SEM. Arrow and horizontal line indicate start and duration of ibandronate treatment in group 4. *p < 0.05, **p < 0.01, ***p < 0.001 vs. group 2; ##p < 0.001 vs. lesions OA < 6 mm² on day 18.

Protocol 3: effect of adjuvant systemic ibandronate on the development of bone metastases.

When systemic administration of ibandronate was started 3 days prior to tumour cell inoculation, the development of osteolytic lesions was inhibited, as demonstrated by a stable take rate of only 17% in group 6 on day 18 p.i. (Table II). In the 3 animals showing osteolytic lesions, the initial osteolytic area was smaller than in groups 2 and 4; but due to the small sample size, this difference did not reach statistical significance. There was a slow expansion in osteolytic area over time (Fig. 7a), and on day 30, the mean osteolytic area in group 6 was 60% of that seen in untreated tumour-bearing animals of group 2 and similar to that of animals in group 4. Osteolytic growth rates were stable at all times, except on day 26, and significantly lower than the growth rates recorded in group 2 (Table II).

In both groups, tibial BMD was similar at baseline. On day 30, BMD was significantly lower in tumor-bearing than in tumour-free animals treated with ibandronate (p < 0.05) (Fig. 7b). BMD at day 30 was significantly higher in animals receiving ibandronate from day −3 onwards than in untreated animals (protocol 1) or animals treated from day 18 onwards (protocol 2) (p < 0.001 each) (Figs. 2b,5b,7b).

Compared to untreated, tumour-bearing animals (group1), rats receiving ibandronate from day −3 onwards showed an early and rapid decrease (p < 0.001) in urinary excretion of collagen cross-links (Fig. 7c). In contrast, there was no difference between tumour and control groups within protocol 3 for any of the other biochemical measurements.

Histologic predilection areas for metastases were similar to those seen in the other protocols, but vital tumour was found only in the 3 animals with radiographically visible osteolytic lesions. The extension of malignant tissue in ibandronate-treated animals was less pronounced than in the untreated controls in 2 of these animals, and activated osteoclasts were almost absent (Fig. 4c). Due to the small number of animals with metastases, group 6 showed the greatest mean metastatic area, and a correlation between quantitative histologic metastatic area and radiographic osteolytic area was not attempted.

None of the animals following subprotocol 3 (groups 5a and 6a) developed osteolytic lesions visible on radiographs. In both groups, tibial BMD was similar at baseline and throughout the study and increased linearly (Fig. 7d). The rate of tibial BMD accrual was significantly greater in ibandronate-treated animals than in untreated animals of protocol 1. This was true for both the control and tumour-bearing groups (Figs. 2d,7d). Changes in bone markers were comparable to those observed in the main protocol, with no significant differences between groups. Histologic analyses of tibial bones obtained from animals killed on day 6 or 12 p.i. showed small metastatic foci, not visible on radiographs. Lesions visible on day 18 were histologically verified as metastatic tissue.

DISCUSSION

Our aim was to establish an animal model that generates vital osteolytic lesions, has high site specificity, enables serial assessment of bone and tumour-related parameters and does not cause potentially interfering comorbidity. To this aim, a technique was developed in which tumour cells were directly inoculated into the femoral artery. At a dose of 1 × 10⁵ cells/injection, this approach resulted in the development of multiple osteolytic lesions in the distal femora and proximal tibiae within 18 days p.i., with a success rate of 95–100%. In untreated animals, osteolyses expanded continuously, causing extensive destruction of resident bone structures, significant loss of tibial bone density and a transient rise in urinary bone resorption markers. Despite these significant changes, no comorbidity (i.e., cachexia, paraplegia) was noted. The model was then applied to longitudinally study the effects of pharmacologic interventions on metastatic bone disease. In keeping with other results,8, 11 treatment with ibandronate hindered further growth of established metastases. In smaller lesions, ibandronate not only arrested tumour growth but induced a reversal of tumour size, as demonstrated by a negative osteolytic growth rate 8 and 12 days after initiation of treatment. Although these observations need to be confirmed by further experimental work, they are supported by results from in vitro23, 24, 25 and in vivo11 experiments, showing that bisphosphonates exert antiproliferative and proapoptotic effects on human breast cancer cells.

The new animal model has several advantages. In contrast to systemic application of tumour cells,7–10, 26 malignant dissemination into visceral peripheries does not occur. Systemic tumour spread often leads to considerable morbidity (e.g., hind-leg paralysis26), which in itself may affect outcomes such as take rate, BMD or bone turnover. Animals in our model showed no significant morbidity, as reflected by the absence of cachexia, normal mobility and normal axial BMD. The latter finding is in keeping with the results of Guise et al.,27 who reported normal plasma concentrations of PTHrP in nude mice inoculated intracardially with MDA-MB-231 tumour cells. The undesired effects of systemic tumour spread may be circumvented by direct inoculation of tumour cells into the bone marrow, which also results in site-specific osteolyses. However, these models lack the processes associated with the migration of tumour cells across vascular and tissue boundaries that is relevant to the in vivo situation.12, 13, 14 In addition, local bone damage during the inoculation procedure may lead to spurious changes in bone metabolism due to wound responses.

Another important feature of this model is a primary take rate of nearly 100%. This observation may be of relevance when studying the effects of adjuvant treatment on metastatic spread. In models with lower spontaneous take rates, the effect of adjuvant treatment might not be distinguishable from the natural course. Finally, the use of larger rodents permits serial assessment of parameters that are hard to measure in mice, such as bone turnover markers.

The disadvantage of the present model is the need for nude rats, which are expensive and require special housing conditions. Also, we observed a dose-dependent effect, with higher doses of tumour cells resulting in an earlier occurrence of both osteolytic lesions and spontaneous remissions. Rejection of tumour tissue by increasing numbers of NK cells is a well-known phenomenon in athymic rats.28, 29, 30 The gain in immune competence depends on the degree of cell differentiation, the amount of implanted cells or tissue and the growth kinetics of the xenotransplant. In very young rats, low or absent titres of NK cells allow (tumour) implants to reach a size that makes them impervious to the subsequent development and attack of NK cells.26, 31, 32 This is consistent with our data as earliest remissions were seen in animals injected with the highest number of tumour cells (1 × 10⁶). Inoculation of 1 × 10⁵ tumour cells/leg yielded a consistent take rate of 90% at day 18, with remissions starting on day 36. This dose therefore provides a time window (days 18–30 p.i.) that is unlikely to be influenced by the immunologic response of the host.

Using the optimal dose of 1 × 10⁵ tumour cells/leg, a consistent take rate of 100% was achieved on day 22 p.i.; i.e., all legs injected without technical failure at baseline developed osteolyses visible on radiographs. However, histologic analyses of tumour-bearing animals from subprotocol 1 (group 2a) showed that small metastatic foci are already detectable as early as 6 days p.i. Once these lesions became visible on X-rays, a mean growth rate of approximately 1–1.5 mm²/day was observed. Although mean growth rates were rather stable between subsequent analytic intervals, there was considerable interindividual variation.

Compared to tumour-free controls, urinary excretion of collagen cross-links significantly increased in tumour-bearing animals on days 6 and 12 p.i., when X-rays and histologic examinations were still normal. It is likely that these changes reflect increased bone destruction caused by the metastatic process. We have previously shown that the urinary excretion of DPD closely reflects the recruitment and activation of osteoclasts in rats.19 However, it remains unclear why in the present study, despite continuous osteolytic activity, the excretion of both markers returned to control values at later points in time. Since the increase in cross-link excretion occurs on days 6–18, it is unlikely that this change is attributable to postoperative33 or adaptive stress or to paracrine activities of the tumour cells (e.g., PTHrP). Possibly, the relatively discrete increase in collagen cross-links is masked by the high background level of bone turnover in growing rats. Consequently, in the present model, bone markers appear useless to assess and monitor the extent of osteolytic lesions in nude rats.

At study end, tumour burden was estimated by histologic, 2-dimensional CBA of the sections showing maximum tumour extent on day 30. Distinct metastases were counted and their respective areas measured. Although this technique allows for only an approximate estimate of tumour burden, it proved to be sufficient for the present study and its aims. Thus, metastatic areas as assessed by histologic CBA were closely correlated with the osteolytic area measured on plain radiographs (r = 0.815, p < 0.0001). It therefore appears that in the present model radiographic osteolytic area reflects metastatic area and, thus, approximate tumour burden in individual animals. Consequently, changes in osteolytic area calculated from the sequential radiographs taken during the course of the study are likely to reflect changes of similar size and direction in metastatic area. As a result of the randomisation procedure, which had led to a larger pretreatment tumour size in the ibandronate-treated group, the metastatic area on day 30 was larger in group 4 than in the untreated control group. However, when analysed for changes in osteolytic area during the treatment period, tumour growth in the placebo group was twice that of the ibandronate-treated group. This finding further substantiates potential antitumour effects of the bisphosphonate. We therefore have reason to assume that a negative osteolytic growth rate during treatment with ibandronate mirrors a reduction in tumour size, pointing to an antiproliferative effect of the bisphosphonate. It is well established that bisphosphonate treatment impairs tumour growth and affects tumour burden.8, 11 However, it remains unclear whether these antitumour properties are attributable to a direct cytoreductive effect of ibandronate on the tumour cells or whether they result from the drug's inhibitory action on bone cells. Furthermore, we did not study the muscle tissues surrounding affected bones; and although there was no macroscopic evidence of metastatic tumour in these soft tissues, the presence or absence of and potential changes in soft tissue metastases in response to bisphosphonate treatment were not analysed histologically.

In vitro and in vivo studies have shown that bisphosphonates cause apoptosis in malignant cells11, 25, 34–37 as well as in osteoclasts.11, 38 However, these effects appear to be restricted to bone and its microenvironment as ibandronate did not induce apoptosis in breast cancer cells implanted into mammary fat pads of mice.11 At least in vivo, it therefore appears likely that the antitumour effects of bisphosphonates are mediated through the inhibition of osteoclastic bone resorption. The bone matrix is rich in growth factors that are relevant to the survival, proliferation and propagation of cancer cells within the bone microenvironment.39, 40 These growth factors are usually released upon osteoclast-mediated bone resorption, and it is conceivable that inhibition of osteoclasts or induction of osteoclast apoptosis will lead to a reduction in the local concentration of bone-derived growth factors. In our model, the repressive effects of ibandronate on tumour take rate and growth appeared to be most pronounced immediately after initiation of treatment. Furthermore, tumour size at the time of treatment initiation may affect therapeutic efficacy as ibandronate influenced osteolytic growth rates preferentially in small metastases (Fig. 6). This may reflect the greater susceptibility of smaller tumours to microenvironmental changes such as a reduction in growth factor concentrations and/ or to direct antitumour effects.11, 25

Ibandronate, when given in an adjuvant fashion, may interfere with these processes by decreasing the local concentration of bone-derived growth factors through inhibition of osteoclast-mediated bone resorption. Since bisphosphonates are deposited on the surface of bone, they may also prevent MDA-MB-231 tumour cells from attaching to, spreading upon and invading bone tissues.23, 24, 41 This assumption is supported by our finding of a reduced take rate in pretreated animals on day 18 p.i. compared to untreated animals (Table II). However, we were unable to completely preclude tumour-induced osteolyses independent of whether ibandronate was administered before or after tumour inoculation. This is in accordance with previous results obtained in nude mice using the same tumour cell line but an intracardiac inoculation approach.8, 10 Early ibandronate treatment has been shown to prevent tumour-associated deterioration on bone density and bone strength following intraosseous inoculation of breast cancer cells.42 Taken together, these observations support the notion that in human cancers bisphosphonate treatment may be initiated early during the course of disease. Although in the present model, ibandronate prevented the development of osteolytic lesions in nearly 90% of animals, we do not know why in nonresponding animals tumour growth rates do not differ compared to untreated animals. This finding is somewhat unexpected and warrants further investigation into the differential effects of tumour therapy.

In conclusion, this animal model provides an additional tool to study the course of metastatic bone disease. Using a physiologically relevant approach through the circulation, the model effectively generates highly vital site-specific lesions, enables serial assessment of bone and tumour-related parameters and avoids the disadvantages of systemic tumour spread (comorbidity such as cachexia, paraplegia). When applied to the assessment of pharmacologic effects of bone tumour growth, the present model provides in vivo evidence that administration of ibandronate not only inhibits the development of new bone metastases but in certain instances may actually lead to a reversal in tumour growth.

Acknowledgements

We thank Mr S. Hoert, Ms M. Wagner and Mr M. Metz for excellent technical assistance during the animal experiments. We also thank Dr C. Kissling for his expertise in HPLC analyses and Mrs. J. Meixner for histologic preparations. Ibandronate was provided free of charge by Roche Diagnostics GmbH, Mannheim, Germany.