Abstract
Background: To better understand the significance of local insulin-like growth factor-binding protein 3 (IGFBP-3) in the development of osteosarcoma, IGFBP-3 levels and subcellular localization were compared in biopsies from osteosarcomas and unaffected normal bone tissues. Patients and Methods: IGFBP-3 levels were analyzed by immunohistochemistry in 21 osteosarcomas and 5 unaffected bone tissues. IGFBP-3 levels were compared for patient outcome. Results: Mature osteocytes of normal bone tissues contained high levels of predominantly nuclear IGFBP-3, whereas only 50% of osteosarcomas contained IGFBP-3-positive tumor cells, with predominantly cytoplasmic IGFBP-3 staining. Comparison of IGFBP-3 levels for patient outcome resulted in two groups. Patients with a low level of IGFBP-3 in osteosarcoma experienced a trend for a shorter survival time than did patients with a high level of IGFBP-3. Conclusion: Our results suggest that the levels and subcellular localization of local IGFBP-3 play a role in osteosarcoma development. Further prospective evaluation with higher clinical sample numbers might reveal a prognostic role for IGFBP-3 level in local tumors in patients with osteosarcoma.
Osteosarcoma is the most common primary malignant bone tumor with a propensity for pulmonary metastasis and a predilection for pediatric and geriatric age groups (1, 2). It is more common in males than in females and the distal femur and proximal tibia account for 80% of these tumors, followed in order by the proximal humerus, pelvis, jaw, fibula and ribs. The World Health Organization classification separates osteosarcomas into a number of subtypes but 95% of cases are central high-grade osteosarcomas of conventional type (3). The histology of osteosarcoma shows atypical cells (osteoblasts, mesenchymal cells, fibroblasts and chondrocytes) entangled in a mostly disorganized bony matrix (immature osteoid). The direct production of osteoid by malignant mesenchymal cells is the main feature of osteosarcoma (4).
The insulin-like growth factor (IGF)/IGF-binding protein (IGFBP) axis of growth-regulatory proteins controls cell proliferation, survival and differentiation in embryonic and various adult tissues (5-7). IGFs are the most abundant growth factors stored in bone and produced by osteoblasts (7). The IGF system plays an important role in the development of osteosarcoma (8) and several other major types of cancer (9). IGF-1 stimulates proliferation and survival mediated by a specific cell-membrane receptor, IGF-1R, which is involved in cell transformation (10-13). A family of high-affinity IGFBPs modulates the activity of IGF-1. IGFBP-3 is the major serum IGF-1 carrier protein controlling IGF-1 availability for its receptors in the extracellular milieu, thereby modulating the mitogenic and antiapoptotic actions of IGF-1 (9, 14). IGF-1/IGF-1R-independent activities of IGFBP-3 also play an important role in connection with its antiproliferative and proapoptotic functions (9, 15). IGFBP-3 can inhibit proliferation of immortalized murine fibroblasts with a targeted disruption of the IGF-1R (16) and induce programmed cell death via IGF-1-independent pathways in mouse fibroblasts (17) and in human prostate cancer (18) and breast cancer cells (19). Furthermore, cellular re-internalization of IGFBP-3 and subsequent nuclear localization was demonstrated (20). IGFBP-3 contains a nuclear localization sequence in its COOH-terminal domain, and importin-β-dependent import from the cytosol into the nucleus has been shown (21). Moreover, after addition to cell culture supernatants, IGFBP-3 has been detected in the nuclei of human lung, breast and prostate cancer cells (20-23), and human keratinocytes (24), suggesting that nuclear actions of IGFBP-3 may be important for its IGF-independent functions. Both nuclear and cytoplasmic IGFBP-3 can induce apoptotic cell death in U-2OS osteosarcoma cells (25). The expression of IGFBP-3 was shown to be tightly regulated at the transcriptional and post-transcriptional level. The expression of the IGFBP-3 gene is activated by the tumor suppressors p53 (26) and phosphatase and tensin homolog (PTEN) (27), and several proapoptotic and growth inhibitory factors, such as transforming growth factor β (TGF-β) and retinoic acid (28). Moreover, the turnover of extracellular IGFBP-3 is regulated by extracellular proteases (29), and intracellular proteases such as the endosomal and lysosomal enzymes cathepsin L and cathepsin D (30, 31) and the ubiquitin/proteasome system regulating the IGFBP-3 levels in the cytoplasm and nucleus, respectively (25).
The tumor-suppressive functions of IGFBP-3 suggest that high systemic IGFBP-3 levels reduce the risk of developing cancer. In fact, many prospective studies investigating the association between serum or plasma levels of IGFBP-3 and relative cancer risk showed an inverse relationship between the levels of IGFBP-3 and death from major cancer types (9, 14, 32). However, other studies, for example on breast cancer risk, challenge the predictive value of serum IGFBP-3 levels (9, 14). This suggests that changes in serum IGFBP levels may not always reflect changes at the level of specific tissues. Hence there is a need for studies analyzing the local IGFBP-3 protein levels in cancer tissues. IGFBP-3 plays an important role as a proapoptotic factor in osteosarcoma cells (25), and although the IGFBP-3 gene expression (33) as well as protein turnover is tightly regulated in bone cells (25), the expression and subcellular localization of IGFBP-3 have not yet been thoroughly explored in osteosarcoma cells in clinical samples. In the present study, we compared the levels and subcellular localization of IGFBP-3 in biopsies from osteosarcomas and unaffected normal human bone tissues. Moreover, we analyzed how the IGFBP-3 protein expression levels in osteosarcoma tissues are associated with the survival of patients harboring an osteosarcoma.
Patients and Methods
Patients. Specimens of osteosarcoma were obtained from archives of surgically excised tumors of 21 patients treated at Innsbruck Medical University, Austria; McMaster University, Hamilton, Canada; Leipzig University Hospital, Germany; and South West Area Pathology Service of Liverpool, Australia. All patients were treated by surgery and received a mean of 780.3 days of follow-up care after surgical treatment. A pathological evaluation established the classification according to World Health Organization's guidelines and staging in all of the patients. The histology of the primary tumors was centrally reviewed by a bone pathology center in all cases. The minimum length of follow-up care was 30 days. All of the clinical and pathological information and follow-up data were based on reports from our tumor registry services. All patients provided written informed consent according to the local Investigational Review Board requirements. The study was reviewed and approved by the institution's Surveillance Committee to allow us to obtain tissue blocks and other pertinent information from the patients' files according to the different regulations of the four countries. The general clinical characteristics of the patients are shown in Table I.
Relation of IGFBP-3 level in osteosarcoma to patient data.
Immunohistochemical staining of IGFBP-3. Osteosarcoma tissues were gently decalcified using EDTA. Hematoxylin and eosin-stained sections were used to select homogenously cell dense areas in all samples and the region of the tumors with the highest cell rate was identified. Four-μm-thick formalin-fixed and paraffin-embedded tissue slices were placed on poly-L-lysine-coated slides (DAKO Corp, Carpinteria, CA, USA). After incubating for 1 hour at 60°C, the samples were kept at room temperature. After deparaffinization with xylene (2×10 minutes) the slides were gradually rehydrated through graded alcohols and water. Antigen retrieval was conducted for 30 minutes in citrate buffer (0.01 mol/l sodium citrate buffer, pH 6) in a steamer at 700-800 W. The slides were cooled to room temperature, rinsed in water and treated with 3% H2O2/methanol for 15 minutes to quench endogenous peroxidase activity. The tissue slices were washed in water, rinsed in phosphate-buffered saline (PBS) for 10 minutes and incubated for 15 minutes in blocking buffer (DAKO rabbit serum diluted 1:10 in 1% BSA/1 X Tris, pH 7.5). The tissue slides were incubated with monoclonal α-IGFBP-3 antibodies (α-IGFBP-3 MAB 305 Clone 84728; R&D, Vienna, Austria; dilution 1:50 in blocking buffer; see also (34, 35)) at room temperature for 1 hour. After incubation with biotin-conjugated secondary antibody for 45 minutes at room temperature, avidin-biotin complex was added for 30 minutes at room temperature. The samples were stained for 5 minutes using a DAB substrate kit (Sigma, Vienna, Austria). After washing in water then counterstained with hematoxylin, the slides were treated with variable density alcohol and sealed with balsam solution. A negative control slide was prepared using secondary antibody only.
The slides were assessed using a randomization of the histological fields and cell counting was carried out automatically using a Jenoptik digital camera system mounted on a BX50 Olympus microscope linked to Pentium IV-powered personal computer running ImageJ® (http://www.info.nih.gov/ij; this software and its Java source code are freely available and no license is required for its use) on Windows® XP platform and Java environment. The intensities, percentages, and patterns of immunohistochemical staining of each section were recorded uniformly using a standardized threshold. Six magnification fields were randomly selected and immunoreactive cells, total area and fraction areas were calculated at ×200 magnification. The staining results were first divided into four categories based on the estimated percentage of nuclear-stained tumor cells: negative (0-10%), weak (11-25%), moderate (26-50%) and strong (>50%), but a statistical cluster analysis performed on Kyplot software (Japan) (KyensLab Incorporated, Tokyo, Japan) identified two groups of osteosarcoma, one with diffuse and strongly stained cells (‘high level’ group) and one with few stained cells (‘low level’ group). Adjacent normal-appearing tissue within the tissue sections served as a positive internal control. As negative controls, a different section of each biopsy was used without addition of the primary antibody. IGFBP-3 labeling index (LI) was defined as the percentage of tumor cells displaying membranous and/or cytoplasmic immunoreactivity and was calculated by counting the number of IGFBP-3-stained tumor cells among more than 1,000 tumor cells from representative areas of each tissue section (36). All of the slides were evaluated and scored blindly to the clinical information pertaining to the patients.
Image analysis. The images obtained under light microscopy in the RGB system were converted into grayscale images and the intensity of each pixel was determined using ImageJ®. Data containing arrays of the type (x, y, intensity), where x and y are the coordinates of the pixel positioning, were collected using ImageJ®.
Statistical analysis. Two-sided unpaired t-test was used to analyze the association between two variables between the two groups (low and up-regulation). The Kaplan-Meier estimator was used to compute survival probability as a function of time. The log-rank test was used to compare patients' survival time between groups. Cox's regression analysis was used to evaluate the prognostic value of protein expression, and several clinical variables and histological subtypes. Overall survival, disease-specific survival (from the date of diagnosis to date of death specifically from cancer-related causes) and disease-free survival time (from the date of completion of surgery to the date of relapse or death of cancer-related causes) were analyzed. All of the statistical tests were two-sided. Statistical analysis was carried out using either KyPlot or Graph Pad (GraphPad Software, Inc., La Jolla, CA, USA) software for Windows XP®. An alpha-level of 5% of statistical significance was used for all our investigations.
Results
IGFBP-3 levels in normal unaffected bone tissues and in osteosarcomas. The levels and subcellular distribution of IGFBP-3 were investigated in immunohistochemistry experiments using monoclonal anti-IGFBP-3 antibodies (34, 35). A strong focal positive immune reaction was observed in mature osteocytes analysed in unaffected bone tissue biopsies derived from five different adults without bony lesion of degenerative or neoplastic disease. The anti-IGFBP-3 antibodies predominantly stained the nucleus of the mature human osteocytes, while the cytoplasm of these cells was stained to a minor extend. A representative example is shown in Figure 1A. The data indicate that IGFBP-3 was predominantly localized within the nucleus or around the nuclear envelope of mature human osteocytes in non-cancerous bone tissues.
The histological patterns of osteosarcomas analyzed in this study included thirteen osteoblastic osteosarcomas, three chondroblastic osteosarcomas, two telangiectatic osteosarcomas, one pleomorphic osteosarcoma, one fibroblastic osteosarcomas, and four osteosarcomas not otherwise specified. The majority of these tumors showed a heterogeneous and variable IGFBP-3 staining pattern. Nearly 50% of the osteosarcomas showed areas with strong IGFBP-3 staining. In these biopsies, IGFBP-3 was predominantly localized in the cytoplasm of the tumor cells and occasionally in the membrane in association with cytoplasmic staining (Figure 1 B-D). Strong nuclear IGFBP-3 staining was only found in approximately 2% of the tumor cells in focal fields of osteosarcomas with high IGFBP-3 expression (Figure 1E). Thereby the nuclear association of IGFBP-3 was almost exclusively detected in osteoid-forming (highly differentiated) areas of osteosarcoma. Low-level IGFBP-3 staining or absence thereof was found in half of the osteosarcoma series; a biopsy without detectable IGFBP-3 is shown in Figure 1F. Osteosarcomas with low levels of IGFBP-3 showed fewer malignant cells but more osteoid or extracellular matrix and these cells were slightly smaller than malignant cells with high IGFBP-3 staining. A low or high level of IGFBP-3 was not a characteristic pattern of any particular subtype of osteosarcoma, although all three chondroblastic osteosarcomas had low IGFBP-3 staining. The staining signals were quantified and analyzed statistically (Figure 2A-D). Cell populations with low and high IGFBP-3 were distinctly identified in our osteosarcoma series with a clear-cut difference with respect to total area (p<0.0001), average cell size (p=0.0054) and area fraction (p<0.0001) of IGFBP-3 stained cells.
Clinicopathological parameters associated with loss of IGFBP-3 expression. On the basis of cluster analysis, 11 out of the 21 osteosarcoma specimens showed a loss of IGFBP-3 expression. Table I presents the associations between the IGFBP-3 expression status and the clinicopathological parameters. Four samples were teaching samples and anonymous. The IGFBP-3 levels in osteosarcomas did not display significant correlation with age, gender, histological grade of differentiation, disease stage, or presence of nodal involvement; however, a certain trend could be observed. Osteosarcomas with low levels of IGFBP-3 occurred most frequently in the femur of female patients whose mean age was 31.09±16.26 years, whilst high level IGFBP-3 expressing osteosarcomas occurred most frequently in the humerus of male patients whose mean age was 51.02±27.39 years.
Immunohistochemical detection of IGFBP-3 in normal unaffected bone tissues and osteosarcoma. A-F, Immunoperoxidase staining of sections from osteosarcoma and unaffected bone tissue with monoclonal anti-IGFBP-3 antibodies are shown. A, Unaffected bone tissue showing focal nuclear staining of IGFBP-3 in mature osteocytes. In the inset, a cell with prominent nuclear staining is shown. B, Osteosarcoma with cells containing high levels of predominantly cytoplasmic IGFBP-3. C, The same specimen as shown in B in a binarized grey-scale picture as was used for quantification with ImageJ®. D, Another osteosarcoma with cells containing high levels of predominantly cytoplasmic IGFBP-3. E, Tumor cells with predominantly nuclear IGFBP-3 staining in a highly osteoidal differentiated area of an osteosarcoma. In the insets, cells with prominent nuclear staining are shown. F, Osteosarcoma with absence of IGFBP-3 staining. Magnifications: Main images ×200, insets ×600.
Morphometric analysis. The morphometric analysis was conducted as described in materials and methods. A, Cell counting; B, total area; C, average cell size; D, area fraction.
Impact of IGFBP-3 level in osteosarcomas on patient outcome. Of the 21 patients with osteosarcoma, five died after a mean follow-up time of 857.8 days, and 16 patients were still alive at the time of the last follow-up report. The median follow-up duration among the patients who remained alive was 746.3 days. Of the patients who died, two were patients whose tumors showed loss of IGFBP-3 expression (2/11), whereas the other three were patients whose tumors showed IGFBP-3 expression (3/10). Patients with low IGFBP-3-expressing tumors did not have significantly shorter disease-specific survival (p=0.9458 by the log-rank test; Figure 3). Neither patient gender, histological grade nor disease stage were statistically significant predictors of disease-specific or disease-free survival.
Discussion
Research on osteosarcoma is well justified because this bone tumor is not only the most common primary bone malignancy with a high propensity for pulmonary metastasis and a predilection for pediatric and geriatric age groups, but also it is the cornerstone of current clinical management involving high-dose neoadjuvant chemotherapy, to which 40-60% of cases are non-responsive (37). Thus, the prognostication, stratification and individualization of treatment could be crucial for optimal and cost effective management in osteosarcoma.
Relationships between defects in tumor suppressor mechanisms enabling malignant transformation and progression and the resistance to cancer chemotherapy have been suggested (38). This is relevant for IGFBP-3 because it has tumor suppressor-like activity as an inhibitor of cell proliferation and stimulator of apoptotic cell death by IGF-1-dependent as well as IGF-1-independent mechanisms (9, 15). Accordingly, IGFBP-3 in plasma/serum has been found to have a prognostic value in patients with tumors. IGFBP-3 may protect against the development of gastric adenocarcinoma by preventing the formation of intestinal metaplasia (39). High levels of IGFBP-3 are associated with a reduced risk for prostate cancer (40) and reduced expression of IGFBP-3 was shown as an early event in prostatic carcinogenesis (41). Moreover, an inverse relationship was demonstrated between serum/plasma levels of IGFBP-3 and death or metastases for lung, colon and bladder cancer and childhood leukemia (42-46). Furthermore, there was a trend toward increased survival in patients with Ewing sarcoma family of tumors who had a high IGFBP-3/IGF-1 ratio (47). Low serum IGFBP-3 may also be associated with increased risk of colorectal cancer (48), although another prospective study of colorectal cancer showed that high IGFBP-3 did not modify the risk related to high IGF-1, and patients in the highest quintile for IGFBP-3 actually had an increased risk (49). Other studies of breast (50) and lung (51) cancer risk also challenge the predictive value of serum IGFBP-3 levels.
Kaplan-Meier estimator of two distinct groups of osteosarcoma showing low and high levels of IGFBP-3. Patients affected with osteosarcoma and a low level of IGFBP-3 (n=9) experienced a slightly lower survival time of 760.3±339.8 days than patients affected with osteosarcoma and a high level of IGFBP-3 (n=8), who had a survival time of 800.4±358.5 days (p=0.9107). However, there was no significant difference between the two survival curves using the Kaplan-Meier estimator (p=0.9458). Four samples were teaching samples and anonymous.
While these studies may mainly reflect the role of IGFBP-3 as IGF-1-binding protein in the circulation, it is not clear how changes in serum IGFBP-3 levels reflect changes at the level of local cancer tissues. The IGFBP-3 levels in local tumor cells and in their extracellular environment are less well studied (9). For these reasons, in the present study, we assessed differences in the IGFBP-3 protein expression as well as subcellular localization in osteosarcoma relative to normal bone tissues and analysed whether these parameters could play a role as prognosticator for potential resistance/responsiveness to chemotherapy. While the IGFBP-3 levels were high in the vast majority of mature osteocytes in all unaffected bone tissues analysed, we found two groups of IGFBP-3 expressing osteosarcomas. High IGFBP-3 levels were associated with a longer survival relative to patients with low IGFBP-3 levels, although this trend was not significant. Osteosarcomas with low levels of IGFBP-3 were mostly found in young female patients with femoral localization, whilst high level IGFBP-3 osteosarcomas involved mainly older male patients with humerus localization. The gender data, which need to be confirmed in larger studies, could be of clinical and therapeutic relevance. According to current knowledge, low intracellular levels of IGFBP-3 could be explained by down-regulation of IGFBP-3 gene expression or by inhibition of one of the intracellular IGFBP-3 proteases in the given osteosarcoma cells (9, 25, 29). Similarly, the expression of the IGFBP-3 gene in tumor cells has been shown to be repressed by polymorphism or hypermethylation of CpG islands in the promoter region of the gene (52-54). Moreover, in keeping with a role in tumor suppression, it has been shown that the expression of IGFBP-3 is induced by apoptosis-inducing or growth-inhibitory factors, such as TGF-β1, retinoic acid, antiestrogens, antiandrogens, tumor necrosis factor α (TNF-α), and trichostatin A (55-57), suggesting that these agents may mediate their cytostatic effects through IGFBP-3. Furthermore, IGFBP-3 was identified as one of the genes induced by p53 (26), a tumor suppressor which regulates the transcription of many cellular genes that are involved in cell cycle arrest and apoptosis (58). The turnover of intracellular IGFBP-3 has been shown to be regulated by proteolysis. Cathepsin L and cathepsin D are described as cytoplasmic IGFBP-3 proteases (30, 31) and we have previously shown that the half-life of nuclear IGFBP-3 is directly regulated by ubiquitin/proteasome-dependent proteolysis (25). However, the precise mechanisms leading to low IGFBP-3 levels in osteosarcoma warrants further studies.
One interesting result of this study is the difference in the subcellular localization of IGFBP-3 between mature osteocytes in normal bone tissues and osteosarcoma cells in malignant bone tissues. We demonstrate here that IGFBP-3 is predominantly associated with the nucleus/nuclear envelope of the vast majority of the mature osteocytes in normal bone tissues. In contrast, predominantly nuclear staining was observed only in a few tumor cells in all but two osteosarcomas showing high IGFBP-3 levels, whereas the vast majority of the tumor cells in this group of osteosarcomas had predominantly cytoplasmic IGFBP-3. Although the role of nuclear IGFBP-3 in mature osteocytes is not known, these findings certainly indicate that nuclear IGFBP-3 exists in arrested differentiated osteocytes in vivo. These data are in keeping with a recent study showing that IGFBP-3 is associated with the nucleus of fully differentiated chondrocytes in human articular cartilage (59). Moreover, evidence was presented suggesting that translocation of IGFBP-3 into the nucleus of porcine muscle satellite cells plays a role in mediating the proliferation-suppressing action of TGF-β1 (60); and, in fact, there is precedence that the IGF-1/IGFBP-3 axis plays an important role in the control of proliferation and differentiation of embryonic and adult bone tissues of zebra fish (6), mice (5) and humans (7). Nuclear IGFBP-3 has previously also been detected in tumor cell lines (20-23). All of the sporadically detected tumor cells with nuclear IGFBP-3 localization in the present study formed large amounts of osteoid, whereas the tumor cells with cytoplasmic IGFBP-3 did not. This suggests that nuclear localization of IGFBP-3 is consistent with a highly differentiated phenotype, while high cytoplasmic IGFBP-3 levels are consistent with de-differentiation. In keeping with these data, we previously showed that nuclear IGFBP-3 is degraded by the ubiquitin/proteasome system in de-differentiated human U-2 OS osteosarcoma cells (25), suggesting that nuclear IGFBP-3 is down-regulated by proteolysis in these cells. To summarize this paragraph, nuclear IGFBP-3 could play a role in the induction or maintenance of osteocyte differentiation while cytoplasmic IGFBP-3 is a characteristic of de-differentiated osteosarcoma cells.
In summary, this study has shown that two groups of IGFBP-3-expressing osteosarcomas can be delineated. High IGFBP-3 levels in the local tumors are associated with a trend toward longer survival. Further prospective evaluation with higher clinical sample numbers might reveal a prognostic role for IGFBP-3 level in local tumors in patients with osteosarcoma. Another interesting finding of this study is the change in IGFBP-3 subcellular localization from predominantly nuclear in normal mature osteocytes to predominantly cytoplasmic in osteosarcoma cells. The role of this shift in the subcellular localization of IGFBP-3 for the development of osteosarcoma warrants further studies.
Acknowledgements
This work was supported by the Austrian Cancer Society-Tyrol (Austria) to WZ, by the EuroBoNet to (Germany) to TA, and by the Austrian Cancer Society-Tyrol (Austria) and Peripro to CS. CS designed the experiments and was responsible for the performance, control, interpretation of data and for the coordination of the morphological study, and wrote the first draft of the manuscript. SR performed the histological staining and archived samples for analysis. WZ designed the experiments, analyzed and recorded morphological information, supervised the project, interpreted data, and wrote the manuscript. JR, TA and CL provided tissue samples and clinical data as well as patients' outcome data and reviewed morphological data.
- Received February 25, 2009.
- Revision received April 7, 2009.
- Accepted April 13, 2009.
- Copyright© 2009 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
