Tissue Biomarkers in Renal Cell Carcinoma: Issues and Solutions

Abstract

Renal cell carcinoma (RCC) is an aggressive malignancy associated with a high rate of metastasis. Although several promising therapeutic strategies are now available for the treatment of patients with metastatic kidney cancer, the prognosis of these patients remains poor. Research is ongoing to identify RCC-specific biomarkers that can improve early diagnosis, surveillance of tumor progression, and prediction of patient prognosis. The identification of biomarkers that may predict response to specific therapies will also be useful in stratifying RCC patients for treatment selection. Unfortunately, biomarker detection and measurement in kidney tumor tissues can be significantly biased by the lack of standardization in tissue sample acquisition, storage, and analysis. As a consequence, the establishment of standardized operating procedures is necessary to maximize the accuracy of tissue-based biomarker assays. Herein, we discuss current issues in tissue-based translational research aimed at identifying clinically useful biomarkers for kidney cancer.

Renal cell carcinoma (RCC) represents approximately 4% of all adult neoplasms, with an increasing incidence over the years. It has been estimated that in the United States 54,390 new cases of RCC and 13,010 deaths attributable to the disease will be reported in 2008¹. This high rate of mortality is due to lack of effective treatments for patients with advanced disease. Because RCC is resistant to conventional chemotherapy and has a low rate of response to immunotherapy, the prognosis remains poor, with 2-year survival reached by only 18% of patients with metastatic disease^{2, 3}. Although novel targeted therapies have proven to be effective, responses are unfortunately partial and not durable^4-10. Therefore, alternative molecular therapeutic targets need to be discovered..

The identification of tissue-based RCC biomarkers that can improve early tumor detection and predict patient prognosis and response to therapies is warranted.Despite long-standing efforts toward biomarker discovery and validation, no tissue biomarker is currently used in the clinical management of patients with kidney cancer. Many variables are involved in the various phases of RCC tissue biomarker analysis (Table 1) and the lack of standardization in many of the procedures involved in the biomarker development process represents a major limitation for the advancement of the field. Herein, we review current issues in tissue-based biomarker research in kidney cancer, with particular emphasis on immunohistochemical assays.

Table 1.

Variables that may exert a significant impact on data accuracy are identified for each phase of the biomarker evaluation process

Pre-tissue acquisition	Post-tissue acquisition	Tissue analysis by immunoassays
Drugs administration	Size of tissue aliquots	Antigen retrieval (FFPE tissue)
Type of anesthesia	Sample handing conditions	Tissue sections fixation (frozen tissue)
Duration of anesthesia	Rate of freezing	Antibody validation
Blood pressure variations	Type of fixative	Antibody incubation parameters
Intra-operative blood loss	Time in fixative	Type of detection system
Renal artery clamping time	Tissue embedding protocol	Use of control tissues
Pre-nephrectomy renal artery embolization	Storage temperature	Scoring by pathologist
Type of surgical procedure	Storage duration	Image analysis platform

Tissue Acquisition, Processing, and Storage

Kidney cancer is a heterogeneous entity that not only comprises histologically and genetically different subtypes but also shows a significant heterogeneity within the same lesion. Detection and quantification of tissue biomarkers within different areas from the same lesion may be extremely important to clarify the molecular mechanisms underlying this phenotypic heterogeneity. To achieve this goal it is crucial to use high-quality tissues, in which both morphology and biomarker integrity are guaranteed. Unfortunately, tissue acquisition, processing, and storage procedures have been shown to affect the quality of the specimen collected.

A change in the molecular profile of tumor cells can occur early in the tissue acquisition process because of surgical variables. Anoxia and changes in local pH due to anesthesia, renal artery clumping, prenephrectomy renal artery embolization, systemic blood pressure variations, and intraoperative blood loss represent stress events that can induce activation or deactivation of given molecular pathways, changing, for example, the phosphorylation status of key signaling molecules^{11, 12}. Investigators at the National Cancer Institute (NCI) are currently leading an important initiative aimed at systematically recording all the parameters used in each surgical procedure performed on patients with kidney cancer¹³. Future investigations on the tumor tissues obtained from this patient cohort will lead to a better understanding of the impact of each surgical variable on the molecular characteristics of the samples .

Tissue handling conditions have also been shown to significantly influence the expression of biomolecules. The amount of time that a tissue specimen is kept at room temperature between collection and processing should be minimized to prevent RNA and protein degradation. The effect of warm ischemia and room temperature handling on RNA expression levels of several genes was recently assessed on fresh mouse liver tissue incubated at 25°C or 37°C for up to 4 hours. Importantly, incubation at 37°C drastically decreased messenger RNA (mRNA) levels to one-tenth of those measured at 25°C¹⁴. Transporting the tissue in a sterile and RNAse-free container on wet ice and promptly freezing an aliquot appear to be the best way to preserve RNA and protein integrity. The tissue should be embedded in cryosection molds using cryopreservation media and frozen in liquid nitrogen, or isopentane cooled with dry ice, to better preserve tumor histologic features and avoid the freezer burn at the tissue periphery often observed in tissue frozen directly at −80°C in Cryovial tubes¹⁵ .

There is also evidence that short-term storage of tissue on ice ensures stability of gene transcript levels, whereas marked modulation of individual mRNAs is observed after storage at room temperature, in isotonic sodium chloride solution, or in a commercial RNA-stabilizing buffer (RNAlater)¹⁶. Because the impact of various tissue handling variables can be tissue specific, investigators at the NCI are currently assessing the impact of time and temperature storage on gene and protein expression in kidney tumors removed from patients with von Hippel-Lindau disease¹³.

For long-term storage, specimens can be stored either in freezers at −80°C or within nitrogen vapor freezers (−150°C), although lower temperatures seem to be better for long-term tissue preservation. Efficient systems that continuously monitor freezer temperature (power backup and alarm systems) are absolutely necessary to avoid unintentional thawing of tissue samples. Freeze-thaw cycles affect the integrity of biomolecules and should therefore be avoided.

Acquisition of tissue via cryobiopsy is a procedure that has the advantage to combine tissue sampling and cryofixation in vivo. This technique consists of a needle cooled at −10°C that freezes the tissue, which is then captured by a rotating cutter device. In breast carcinoma, it has been shown that the gene expression profile of the tissue obtained using this method was comparable to the one generated by the analysis of matched biopsy specimens immediately frozen after extraction from the tumor mass¹⁷. Importantly, it has also been demonstrated that acquisition of tissue via cryobiopsy allows optimal preservation of phosphoprotein (ie, p-AKT) levels compared with other tissue collection procedures, including surgical resection and fine-needle aspirate biopsy¹³.

Several fixation-related parameters, such as type of fixative, volume and concentration, temperature and length of fixation, and specimen size, can affect biomarker detection and quantification in tissue specimens. Archived formalin-fixed paraffin-embedded (FFPE) tissues represent major resources of DNA in research, especially for retrospective studies. However, compared with the DNA derived from frozen tissue, the DNA extracted from tissues fixed in 10% neutral buffered formalin (the most common fixative used in pathology practice) is significantly degraded and fragmented. Tissue fixation at 4°C and/or the use of DNA stabilizing solutions have been shown to result in the isolation of longer DNA fragments. Formalin fixation also seems to be responsible for recurrent mutation artifacts in sequencing studies, probably caused by the cross-linking between cytosines and formaldehyde¹².

Fixation not only alters nucleic acid integrity but also the immunoreactivity of the tissue. Fixation time constitutes one of the main factors that influences tissue antigenicity. It is recommended to fix tissues in neutral buffered formalin for 12 to 24 hours¹⁸. However, because antigens can significantly differ in formalin sensitivity and tissues specimens are usually fixed for different lengths in different institutions, the effect of formalin fixation should be carefully evaluated for any biomarker being considered for implementation into clinical practice.

Although frozen tissue banking efforts for research purposes are increasing, paraffin-embedded tumor blocks stored in pathology department archives still represent the major source of tissue currently used in research. The length and conditions of paraffin block storage may influence biomarker detection. The recently published NCI Best Practices for Biospecimen Resources addresses this issue and suggests that paraffin blocks be stored under stable conditions within an area with pest and humidity control and a temperature below 80°F (27°C)¹⁹. The use of multiple control tissue samples for which tissue handling variables have been carefully recorded could be useful to investigate variations in biomarker expression levels caused by different conditions and length of storage.

Immunohistochemical Assays

The expression of candidate markers identified by proteomic, genomic, and gene expression profiling studies needs to be validated on tumor tissue by molecular pathology techniques such as immunohistochemical analysis. Immunohistochemical analysis allows one to not only detect the presence or absence of the antigen but also to localize it within cellular and subcellular compartments and to provide potential quantitative data. This is particularly important in kidney cancer, where tumors are characterized by substantial heterogeneity, not only among the different tumor subtypes but also within the same lesion. Immunohistochemical evaluation of cancer-specific marker expression in tissue microarrays (TMAs) allows to perform high-throughput analysis or hundreds of tissues at the same time. Importantly, since the entire tumor series is contained in one or a few slides, standardized assay conditions are used for all cases, which is something very difficult to achieve when staining whole tissue sections. TMAs containing cores representative of morphologically distinct tumor areas could be helpful in clarifying whether different tumor characteristics, such as Fuhrman nuclear grade (FNG), tumor growth pattern, and cytological features, are associated with specific molecular profiles of tumor cells. TMAs constructed according to standardized criteria will be invaluable to address these kinds of questions and might contribute to a better understanding of tumor progression in RCC.

An example of a standardized RCC TMA construction protocol based on the FNG is given by The Tissue Acquisition Pathology and Clinical Data Core of the Dana-Farber/Harvard Cancer Center Kidney Cancer Specialized Program of Research Excellence¹³. In this protocol, for each tumor, 2 cores of the predominant and 2 cores of the highest FNG are sampled. Importantly, the highest FNG is a known independent prognostic factor in RCC^{20, 21}. Therefore, it might be important to assess the expression of RCC biomarkers specifically in the cell population that shows these nuclear characteristics. However, because of concerns regarding the representation of tissue heterogeneity in TMAs, it must be noted that validation studies comparing results in TMA and whole tissue sections should always be performed before specific TMA-based immunoassays are utilized for biomarker evaluation.

Immunohistochemical analysis of both standard and TMA tissue sections is a well-established technique, but unfortunately many variables are involved in this procedure, potentially affecting result concordance and biomarker validation across institutions. One example is provided by discrepant published results regarding the association between HIF1α protein expression and clinical outcome in clear cell RCC. By using Western Blot analysis, Lidgren et al. showed that HIF1α was an independent favorable prognostic marker in clear cell RCC²². In a subsequent immunohistochemical study from the same group, the survival difference between high and low HIF1α expressors did not reach statistical significance but patients with higher HIF1α levels in their tumor tissue tended to have a more favorable²³. In contrast with these results, Klatte et al. recently reported that RCC patients with high HIF1α tumor expression had significantly worse survival than patients with low expression, even in multivariate analysis²⁴. As suggested by Klatte and colleagues, this contradictory finding may be explained by the different immunohistochemical methodology for HIF1α used by Lidgren et al., with which cytoplasmic HIF1α staining, but not nuclear staining, was detected.

Standardization of immunohistochemical analysis has been a matter of debate for a few decades, and many guidelines and regulations have been proposed over the years^{18, 25, 26}. For FFPE tissue, the type and duration of fixation can influence antigen preservation and recognition by a specific antibody. It has been suggested that aldehyde fixation induces formation of protein cross-linkages, causing protein folding that obscures the antibody-binding site²⁷. Antigen retrieval (AR) is capable of reversing at least in part this formalin-fixation effect and restoration of immunoreactivity to levels observed in fresh frozen tissue is achieved for some antigens^{28, 29}. However, in spite of the use of AR, significant differences in the expression levels of certain proteins are observed when frozen and FFPE tissues are immunostained utilizing the same primary antibody. For instance, the percentage of patients with clear cell RCC expressing the prognostic biomarker B7-H1 was reported to be 37% when the analysis was performed in fresh-frozen tissue but only 24 % when assessed in paraffin-embedded tissue^{30, 31}. Thus, the study of FFPE tissue appears to underestimate the presence of B7-H1 in clear cell RCC and possibly affect the evaluation of its prognostic value.

The AR techniques based on heating methods are the most common, but not necessarily the best, for unmasking all the antigens; for selected antigens enzymatic treatment may give better results. In general, to achieve the optimal retrieval level, the different variables in the retrieval process (type of enzyme, heating method, temperature, time, and buffer solutions) must be tested and optimized for each antigen. The optimized AR protocol should then be implemented by laboratories across institutions to facilitate data comparison.

Validation of antibody specificity and sensitivity is a critical step in the development of immunohistochemical assays. Various methods, including Western blot, flow cytometry, and immunoprecipitation, should first be used to select antibodies that specifically recognize the protein(s) of interest. Validation of such antibodies in immunohistochemical analysis of FFPE samples should then be carefully conducted using appropriate positive and negative control tissues and/or cell pellets. TMAs can be useful in validating novel antibodies by simultaneously testing multiple tumors or tissue types in a time- and cost-saving manner.

The assessment of the activation status of kinases through the study of phosphoprotein expression by immunohistochemical analysis is especially challenging in FFPE tissues. Since phosphoepitopes can be significantly affected by fixation and processing procedures, the specificity of phospho-specific antibodies should be evaluated extremely carefully.Controls for validation of such antibodies should include cell lines transfected with proteins in which the phosphorylation site has been eliminated by site-directed mutagenesis, treatment of tissue with phosphatases, or blocking the binding of the antibody with specific or unrelated phosphopeptides. The quality of the tissue used for both control and test samples is extremely important. In both tissue specimens, the level of preservation of the phosphoprotein should be optimal³².

The use of antibodies that are not available commercially in biomarker studies may limit their validation process and thus their use in the clinical setting. In clear cell RCC, carbonic anhydrase IX (CAIX) protein expression levels have been shown to correlate with patient prognosis and response to interleukin 2 immunotherapy, when assessed by immunohistochemical analysis, with a proprietary antibody (clone M75)^{33, 34}. A recent IHC study performed with the same antibody did not show CAIX to have independent prognostic significance³⁵. However, only 6% of the patients included in this study received high-dose IL-2 after nephrectomy, suggesting that CAIX might be a predictive rather than a prognostic biomarker for RCC. Recently, Hikmat et al have shown that a commercially available antibody (Novus Biological, clone NB100-417) can be used as a valid alternative to M75 in IHC detection of CAIX protein in clear cell RCC tissues³⁶. Additional studies are needed to further validate the use of this new antibody in the assessment of CAIX as a prognostic and predictive biomarker of clear cell RCC.

To be comparable, immunohistochemical studies should use not only the same antibody but also standardized detection systems, AR methods, reagents dilutions, and incubation times. The use of ready-to-use reagents in kit formats and automated immunostainers can reduce staining variation among samples due to differences in technical protocol implementation and batch effects. However, because preexisting protocols can be adjusted for specific needs, it is critical that immunohistochemical staining steps are described in detail in every publication.

The use of stored cut FFPE tissue sections for immunohistochemical analysis can affect staining results. It has been shown that tissue antigenicity decreases after sectioning most likely because of tissue oxidation and drying. This represents an important issue especially for TMA slides, which are frequently prepared in batches to minimize tissue loss by repeated sectioning of the paraffin block³⁷. The temperature and length of storage are crucial, although the extent and rapidity of immunoreactivity loss also seems to be antigen dependent, with phosphoprotein being more labile than other proteins^{32, 38}. Storage of paraffin recoated slides in a nitrogen chamber appears to be the best way to minimize tissue antigenicity loss³⁹.

Immunohistochemical Quantification and Image Analysis Systems

The National Institutes of Health defines a biomarker as “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention”⁴⁰. In immunohistochemical studies, the quantification of tissue biomarkers tends not to be highly reproducible when staining results are evaluated by eye. This is because scoring procedures are not well standardized and the interpretation of immunohistochemical stains is subjective and relies on the experience and acuity of the interpreter. To improve reproducibility, different discrete scoring systems (H-score, quick score, Allred score) that take into account both the intensity and the extent of the staining in different combinations have been proposed⁴¹. However, the availability of various scoring methods and the use of different cut-off points to define positive staining can make studies difficult to compare¹⁸.

Quantifiable internal or external reference standard controls that can be used as calibration points for the test have the potential to significantly increase data accuracy. The use of cell or protein blocks with known concentration of the analyte, which are fixed and processed in the same way as the test block, can represent an alternative to internal controls. However, because protein degradation and changes in the activation status of some molecular pathways are known to start even before the tissue is removed from the patient¹¹, an internal control should be preferred to normalize the expression level of the biomarker of interest. Unfortunately, control and test proteins can differ substantially in stability, and their rate of degradation may not be comparable. For each biomarker under investigation, the appropriate quality control should therefore be carefully selected to obtain an accurate in situ protein measurement.

Pathologist-based scoring is labor intensive and time-consuming and is affected by high levels of interobserver and intraobserver variability. In contrast, automated image analysis systems have been shown to provide high-throughput, accurate, and sensitive measurement of protein expression within tissue specimens. Different instruments and software programs are commercially available, such as ACIS (Clarient), iVision system and GenoMx (BioGenex), ScanScope (Aperio Technologies), Ariol SL-50 (Applied Imaging Corporation), and AQUA (HistoRx). Some software programs can be used to quantify both chromogenic (ie, brown stain) and fluorescence-based staining, whereas others are restricted to just 1 of the 2 methods⁴². Of note, the use of fluorochromes provides a broader dynamic range of signal intensity and thus generates more accurate data compared with chromogens^{43, 44}.

In fluorescent-based technologies (eg, AQUA), the use of multiple antibodies tagged with different fluorochromes allows simultaneous measurement of multiple biomarkers, as well as detection within cellular, subcellular, and even virtual compartments (eg, quantification of the phosphorylated fraction over the total amount of a given protein). However, multiplex assays require the use of additional controls to detect possible interferences and cross-reactions among antibodies that can affect measurement accuracy⁴³.

Although automated image systems certainly have many advantages, several drawbacks can be identified. For instance, the high cost of the instruments makes them inaccessible to some institutions. Moreover, users need to be trained to develop a sufficient degree of expertise. It must also be noted that computer-based technologies still require a significant input from pathologists for quality control purposes.

The use of automated image analysis technology for tissue biomarker quantification can by itself introduce nonbiological variables that can potentially bias the analysis. Indeed, threshold and cut-off values for image acquisition and data extraction from the acquired images are critical for measuring staining intensity and should therefore remain constant for each particular assay. In addition, it has been suggested that the image capturing system should be carefully calibrated to ensure reproducible quantitation from day to day and within different laboratories³².

Conclusions

The identification and validation of clinically relevant tissue biomarkers in RCC are warranted. Besides the well-recognized difficulty in identifying candidate markers to evaluate, several sources of variability in sample acquisition, processing, storage, and analysis significantly affect accuracy and reproducibility of the data, further limiting RCC biomarker development. Automated quantitative imaging systems can be helpful in objectively measuring biomarkers in tissue samples. Nevertheless, these systems are sensitive to technical variables that can influence the intensity of the immunostaining. Thus, strict standardization of the overall assay, using standard operating procedures and appropriate quantifiable reference controls, must be a fundamental prerequisite to achieve accurate and reliable quantification of protein biomarkers in tissues.