Abstract
Background: Renal cell carcinoma (RCC) follows a variable clinical course related to disease stage and metastatic spread (including to bone). Molecular and genetic factors bear prognostic significance in RCC, including proteins involved in extracellular matrix invasion. Among these, bone sialoprotein (BSP) and osteopontin (OPN) are physiologically implicated in bone metabolism, and have a prognostic role in several tumors. BSP expression was also predictive of bone spread propensity in lung and prostate carcinoma. In RCC, no data are available for BSP, while OPN has been correlated with tumor stage, grade and survival. We aimed to define the predictive (of bone spread) and prognostic role of BSP and OPN immunohistochemical expression in clear cell RCC. Materials and Methods: from a series of 305 renal tumors resected between 1993 and 2002, 75 surgically resected clear cell RCCs with tissue material, clinical data and follow-up information available, were selected for the preliminary series; a second group of 126 chemo-naïve, radically-resected, consecutive RCCs was collected as a validation series. Immunohistochemical expression of BSP and OPN on paraffinized samples was evaluated by H-score [=Σ (intensity × percentage of positively stained cells)]. Results: In the preliminary series, BSP and OPN reactivity was found in 85% and 77% of cases, respectively. No predictive role of bone spread propensity of RCC was identified. Conversely, both BSP and OPN were significantly associated with shorter survival considering median (p=0.002) and upper quartile (p=0.03) expression values, respectively. In the validation group, a prognostic role was confirmed for BSP only (p=0.008), while OPN showed a trend of association with poorer survival (borderline p-value of 0.058). Conclusion: BSP was shown for the first time to be an independent parameter associated with poor prognosis in RCC. Its coexpression with OPN identifies a subgroup of RCC having the worst outcome.
Renal cell carcinoma (RCC) accounts for approximately 2-3% (1, 2) of human malignancies and follows a variable clinical course, generally depending on the disease stage at diagnosis (3) and most (70%) cases (2) have clear cell appearance. Overall cancer-specific five-year survival is around 69% (4), fatal outcome being associated with tumor progression with distant metastatic spread at any site, particularly the lung, brain, liver and bone. Metastases at diagnosis (including bone) are described in up to one-third of cases, dramatically affecting prognosis, with a mean survival time of approximately 20 months (5, 6).
Several molecular factors and genetic findings were shown to have prognostic significance in RCC (7, 8), including the enhanced expression of proteins and enzymes involved in extracellular matrix interactions, disruption and invasion. The small, integrin-binding ligand, N-linked glycoprotein (SIBLING) family includes several proteins involved in bone metabolism and also in tumor cell invasion and metastasis (9). Among these, bone sialoprotein (BSP) and osteopontin (OPN) are constitutively expressed by human osteoclasts and also by macrophages, having a role in bone turnover and metabolism by binding metalloproteinase subtypes via integrins. In the kidney, BSP is expressed in both proximal and distal tubular cells (10), while OPN was found to be localized in the Henle loop and distal tubule cells (10, 11) and in the matrix of renal stones (12). These proteins were also detected in several human tumors (9). A prognostic role was reported for BSP in bone (13), breast (14), cervical (15), lung (16) and prostate carcinomas (17) and in multiple myeloma (18). Apart from these types of cancer (9), OPN expression was also prognostic in mesothelioma (19, 20). Finally, the selective expression of BSP or OPN, and of their targets or inhibitors, was associated with a propensity to bone metastases (16, 21, 22, 23).
In RCC, OPN expression was first detected in 1994 (24), and conflicting results were subsequently reported: OPN immunoreactivity was correlated with tumor stage, grade, proliferative index and survival in one series of RCC (25), but not in another study (26). Enhanced plasma levels of OPN were correlated with poor survival in metastatic RCC (27). No clear data are available on the role of BSP in RCC.
Due to the reported prognostic and predictive role of bone metastases in other types of cancers (9, 16, 21, 22), a series of completely resected RCC with known follow-up information was collected with the aim of defining the prognostic significance of BSP and OPN, including the ability of such molecules to predict bone spread of RCC. We show here that neither OPN nor BSP expression are predictive of bone metastasis propensity in resected RCC, but do have a prognostic role. In particular, as far as we know, BSP overexpression is shown for the first time to be an independent parameter associated with poor prognosis in RCC, and its coexpression with OPN at high levels identifies a subgroup having the worst outcome.
Materials and Methods
Case series. Preliminary cohort: From a series of 305 renal tumors resected at the Division of Urology of the University of Turin between 1993 and 2002, a search was made for cases treated by radical nephrectomy, diagnosed as clear cell variant RCC, for which representative paraffin blocks, clinical data and adequate follow-up information were available (including data on the development of metastases and their locations). Seventy-five cases met all these inclusion criteria.
All cases were reviewed and the histological type and Fuhrman grade re-assessed by two of us, with a specific expertise in uropathology (EB and GM). A representative tissue block was selected upon review of the original hematoxylin and eosin stained-slides. Serial sections were processed for immunohistochemistry after the tissue block had been de-identified and anonymously coded by a pathology staff member not involved in the current project.
External validation cohort: To confirm the prognostic and predictive impact of BSP and OPN in RCC, a consecutive series of 126 chemonaïve, radically resected RCC of the clear cell type was collected from the Pathology file of the University of Verona (Italy) and used for a validation analysis of results. The validation cohort was analyzed using tissue microarray (TMA) prepared assembling three core per case of RCC. The choice to use TMAs, thus improving the overall number of cases assessed, was made on the basis of previously reported technical analysis which validated the TMA-facilitated molecular test. Cells staining positive for BSP or OPN were recorded as occurring in less than a third of the core, 30-60% of the core or more than 60% of each core and score as score1, score 2 and score 3, respectively. Cases were considered positive when at least one of the three cores was scored 2 or 3, or when all three cores scored 1.
Ethic Statement: The study has been conducted according to the principles expressed in the Declaration of Helsinki.
Immunohistochemistry. Five micron-thick paraffin sections were collected onto charged slides, deparaffinized and rehydrated in water. Antigen retrieval (for BSP, only) was performed in citrate buffer solution, pH 6.0, in a microwave oven. Slides were washed and incubated with a blocking reagent (1% bovine serum albumin in tris buffered saline) in order to block non-specific antibody reactions and then incubated with primary antibodies against human OPN (polyclonal, 1/650; Sigma Aldrich, Italy) or human BSP (monoclonal, 1/1000; Chemicon International, Temecula, CA). Immunoreactions were revealed by a dextran-chain detection system (Envision; DakoCytomation, Glostrup, DK) using diaminobenzidine as chromogen. The specificity was validated in serial negative control sections by omitting the primary antibody for each immunohistochemical run. A section of lung parenchyma containing alveolar macrophages served as positive control. BSP and OPN immunostains were also evaluated on normal kidney tissues included in the samples of preliminary cohort and on 5 additional arrays.
Immunohistochemical data interpretation. Immunohistochemical findings were independently evaluated by two of the authors (EB, LR) and discrepant cases jointly re-evaluated at a multi-head microscope. All cases of the preliminary series were analyzed using a semiquantitative histological scoring (H-score) method. Briefly, immunostaining intensity of each case was scored as follows: 0, none; 1, weak; 2, moderate and 3 strong, this latter referring to the strong intensity of intra-tumoral macrophages. In addition, the percentage of positive neoplastic cells was evaluated. The H-score value was obtained multiplying the percentage of reactive cells by the corresponding observed intensity and the results were summed [H-score=Σ(I×PC), where I and PC represent the observed intensity and the corresponding percentage of positively stained cells, respectively], obtaining a final score ranging from 0 to 300.
Statistical analysis. The Mann Witney U-test and the Chi-square tests were used to analyze differential H-score values of BSP and OPN between groups. The Spearman and Chi-square tests were used to analyse the correlation between BSP and OPN expression levels assessed with the H-score (preliminary cohort) and four-tie score (validation cohort) method, respectively. Survival analyses were performed using Kaplan-Meier curves and log-rank tests. In the preliminary cohort, the median BSP and OPN expression level was used as cut-off and a further analysis was conducted by comparing the fourth versus the first, second and third quartile of OPN expression. In the validation cohort, survival comparisons were performed between positive and negative cases for both markers. Cox proportional hazards regression was used to estimate the overall survival hazard ratio (OS-HR) associated with BSP and OPN expression in uni and multivariate analyses. Statistical analysis was performed using Graphpad 4 (GraphPad Software, Inc., La Jolla, CA) software and results were considered statistically significant at p<0.05 levels.
Results
Clinicopathological data. The preliminary cohort included 55 male and 20 female patients, having a median age of 62 years. The tumors were Fuhrman grade 1 or 2 in 33 cases and grade 3 or 4 in 42; stages I or II were found in 25 cases, while the other 50 were stage III or IV. Metastases were found at diagnosis in 22 and at follow-up in 19 patients, for a total of 41 cases (17 to bone and 24 to other organs).
The validation cohort included 92 male and 34 female patients having a median age of 62 years. The tumors were grade 1 or 2 in 69 and grade 3 or 4 in 57 cases; stages I and II were found in 91 cases, while the other 35 were stage III or IV (14 patients had metastatic disease at the time of the diagnosis, with no information about the site).
Preliminary cohort.
BSP distribution and correlations. Normal tissues: BSP immunoreactivity was detected in macrophages, as expected, and in tubular cells with a finely granular expression pattern, generally diffuse and occasionally restricted to paranuclear areas, probably related to Golgi apparatus.
RCC: Most cases (64/75, 85%) had some degree of reactivity for BSP, with H-scores widely ranging from 1 to 260 (Figure 1). The median H-score was 20 and no correlation was found between BSP expression and patient age, while that for BSP was significantly higher in female patients (p=0.01). In addition, BSP expression was correlated to tumor stage (p=0.01) and grade (p=0.03), comparing low (I and II) versus high (III and IV) stages or Furhman grades. With regard to site of metastasis, BSP was not correlated to the presence of metastases in bone, nor at any other site, thus excluding a possible predictive role of BSP for RCC metastasis (Table I).
Conversely, BSP had a significant prognostic role: high (above the median value) BSP-expressing cases had a significantly shorter survival than low BSP-expressing cases (p=0.002, Figure 2A). At univariate analysis, high tumor stage, grade, presence of metastases and high BSP levels (>median value) were significantly associated with high risk of death; at multivariate analysis of survival other than presence of metastases (p=0.002, OS-HR=4.37), high BSP expression was associated with a significantly increased OS-HR (p=0.035, Table II).
OPN distribution and correlations. Normal tissues: As expected, OPN immunoreactivity was detected in macrophages, in tubular cells and some parietal cells of the glomerular Bowman capsule, with a finely granular cytoplasmic reactivity, generally diffuse and occasionally restricted to golgian paranuclear areas. The intensity of the staining was in most cases weaker than that of the reference normal macrophages.
Immunohistochemical cytoplasmatic expression of Bone sialoprotein (BSP, ×40) and Osteopontin (OPN, ×40) in renal cell carcinoma (H&E: Hematoxylin and eosin, ×40).
Correlation of Bone sialoprotein (BSP) and Osteopontin (OPN) expression with clinicopathological variables in the preliminary cohort of 75 clear cell renal cell carcinoma (Chi-square test).
RCC: Some degree of reactivity was found in 58/75 cases (77%), with H-scores ranging from 1 to 250 (Figure 1). The median H-score was 10 and no correlation between OPN expression and patient age, sex, tumor stage or grade was found. In addition, OPN expression was not associated to metastases at diagnosis nor was it able to predict a propensity for bone dissemination (Table I). At variance with BSP, high OPN expression (>median value) was associated with an increase in the OS-HR that did not reach statistical significance (p=0.55, Table II). Only when the population of patients was divided into quartiles of expression, and the first to third quartiles were compared to the fourth, higher OPN levels were significantly associated with poor survival (p=0.03; Figure 3A).
Validation cohort.
Prognostic impact of BSP and OPN. BSP-positive RCCs were 49/126 (39%) and no significant correlations with clinicopathological parameters were found (not shown). On the contrary, in the both univariate and multivariate analysis of survival, adjusted for the major clinicopathological parameters, BSP expression was an independent prognostic factor (p=0.04), as were tumor grade and stage (Table III). Patients with BSP-positive neoplasms had a significantly shorter survival (p=0.008, Figure 2B).
OPN-positive cases were 36/126 (29%), with no correlation to clinicopathological parameters (not shown). Patients with OPN expression had poorer survival, but the difference did not reach statistical significance (borderline p=0.058; Table III and Figure 3B).
Coexpression of BSP and OPN. In both series of RCC, BSP and OPN reactivities were correlated and significant co-expression was observed in both preliminary (R=0.35; p=0.002, Spearman test; Figure 4A) and validation series (p=0.01, Chi-square test; Figure 4B). Comparing the four possible combinations of BSP and OPN expression to survival, double-positive RCCs of the validation cohort (BSP+/OPN+) were those conferring the poorest outcome (p=0.003; Figure 4C).
Discussion
In this study, a significant independent prognostic role of BSP expression was demonstrated in clear cell RCC. Similarly, an increased mortality was associated with high OPN expression, although this figure did not reach statistical significance. In addition, we found that concurrent expression of BSP and OPN identified a subset of patients with RCC having the poorest outcome. Conversely, neither markers were predictive of bone metastasis propensity with RCC progression.
In the preliminary cohort, eligible cases were relatively few compared to the total number of resected renal tumors. The main reason for this was the difficulty of obtaining reliable information on disease progression, with special reference to the development of metastases in bone versus other sites. Unfortunately, this type of information was crucial for the purpose of this study and cases with equivocal or incomplete data were excluded. The results were therefore validated in an external series of 126 consecutive cases (validation cohort).
BSP was shown, as far as we know, for the first time to be an independent prognostic factor at multivariate analysis, high BSP levels being associated to a shorter survival of clear cell RCC-affected patients. The prognostic role of BSP expression was confirmed also in the validation cohort. These tumors were investigated on TMAs and scored with a different method, but despite the different percentage of reactive cases (29% versus 85% of the pilot series), a significant prognostic role was maintained. In our study, higher BSP expression correlated with tumor variables previously reported to be associated with patient outcome, as well as tumor stage and grade (3), most likely reflecting the effect of BSP on cell migration and apoptosis evasion (9). This correlation was not confirmed in the validation cohort, possibly due to the earlier tumor stage.
Conversely, in the present study on RCC, prognostic significance of OPN was obtained only by dividing the population into quartiles. Positive cases of the fourth quartile had a significantly higher risk of death. These results were partially confirmed in the validation cohort where survival analysis reached a borderline p-value of 0.058. In the literature, OPN was found to be prognostic in a relatively large series of 171 clear cell RCC (25). The percentage of OPN-expressing cases in these authors' series (37.5%, using a cut-off of ≥5% reactive cells) was lower than that observed for our pilot series (77%, with no cut-off threshold) and in the range of that recorded for the validation cohort (29%). These discrepancies need further investigation and are possibly related to different RCC selection criteria or treatment modalities, or to different antibodies used (monoclonal vs. polyclonal) or even to different evaluation criteria (H-score vs. percentage of cells). As a matter of fact, introducing a cut-off of 5% immunoreactive cells, a minimum H-score of 5 to 15 (depending on intensity values from 1 to 3) would be required to score a case as positive, this would reduce the percentage of OPN-reactive cases to 57% or 44%, respectively, almost in the range of Matusan et al.'s findings (25).
Bone sialoprotein (BSP) expression correlates with survival of patients with renal cell carcinoma using the median value as a cut-off in both the preliminary (A) and validation (B) series.
Osteopontin (OPN) expression weakly correlates with survival of patients with renal cell carcinoma in the preliminary (A) and the validation (B) series.
Uni/multivariate analysis of overall survival (OS) in the preliminary cohort of 75 clear cell renal cell carcinomas. Mets, Metastasis; M, male; F, female; HR, hazard ratio; NR, not reached; NA, not applicable.
Uni/multivariate analysis of overall survival (OS) in the validation cohort of 126 clear cell renal cell carcinomas.
An interesting observation is that although OPN expression alone did not reach statistical significance as an independent prognostic factor, its coexpression with BSP identified a subset of RCC having the worst prognosis compared to other tumor groups (p=0.003), thus indicating that the two markers may act synergistically during tumor progression of RCC.
In both series, we failed to demonstrate a predictive role of BSP and OPN expression with regard to bone metastasis propensity of RCC, as previously shown in other carcinoma types (16, 21, 22). This may be surprising since both proteins are known to be implicated in bone turnover and metabolism; in addition, it was also proposed that they cooperate in distant tumor cell implantation, in the process of metastatis, in particular to bone (9, 28). The observation in other osteotropic carcinomas, such as the currently analyzed RCCs, and follicular thyroid carcinoma (Papotti and Volante, unpublished observation), that BSP and OPN are not predictive of bone metastasis propensity needs to be further investigated.
Bone sialoprotein (BSP) and Osteopontin (OPN) expression correlated with each other in both preliminary (A) and validation (B) series. Double-positive (BSP+/OPN+) cases are those having the poorest outcome (C).
The mechanisms of action of BSP and OPN in RCC growth and progression are not known. Normal renal cells have constitutive expression of BSP at the tubular level, but probably the secreted amount is minimal and involved in other regulatory functions. In neoplastic conditions, high protein levels produced by tumor cells may cooperate to promote invasive growth and modulate distant metastatic spread, not necessarily restricted to specific bone colonization (9). These two molecules being members of the same family of SIBLING proteins, having several specific regulatory functions in tumor cell growth and invasion, it can be speculated that in RCC the pathway involving BSP and OPN is specifically activated or up-regulated in aggressive forms of clear cell carcinoma. In this respect, it may be of interest the fact that antibody-based anticancer therapy directed against OPN or BSP are under evaluation in both in vitro and in animal models (29, 30) with promising results. Additionally, both BSP and OPN are capable of binding enzymatic proteins involved in extracellular matrix degradation, such as the metalloproteinases (MMP), namely MMP types 2 and 3, respectively, via different integrin types [i.e. inhibitors of the above mediators have been developed (31, 32, 33)]. Selected cases of RCC overexpressing BSP and/or OPN might be targeted by such inhibitors, with the aim of controlling tumor cell growth and invasion by interfering specifically with the BSP/OPN pathway and its cascade. Preliminary data on this therapeutic approach already exist (30, 34). Since RCC therapy has limited tools in non-surgical patients, such targets may open new frontiers for possible combined treatments with conventional chemotherapy, as proposed for combined anti-angiogenetic treatments and chemotherapy (35).
The development of a method based on tissue arrays has created a powerful tool for evaluating hundreds of tumors simultaneously with histological and immunohistochemical analyses. Array analysis permit rapid molecular profiling, and tissue arrays enable the analysis of protein expression profiles on specimens to determine their potential clinical significance and role in RCC biology. For these reasons, we proposed to analize BSP/OPN immunophenotypical expression on TMA-facilitated test.
In conclusion, BSP was identified as an independent prognostic factor associated with poor prognosis of RCC and its coexpression with OPN was shown to be significantly related to RCC cases having the worst outcome. This may open interesting perspectives for therapies with inhibitors targeting the BSP/OPN and MMP pathway.
Acknowledgements
We are grateful to Professor Salvatore Rocca Rossetti (University of Turin) for his helpful suggestions. This work was supported by grants from the Italian Ministry of Research (ex-60% to MP) and the Regione Piemonte (Ricerca Sanitaria Finalizzata to FP, grant no. 38097 dated 27/11/2008).
- Received April 8, 2013.
- Revision received May 2, 2013.
- Accepted May 8, 2013.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.