The authors investigated whether deletion of chromosome 9p in clear cell renal cell carcinoma (ccRCC) predicted worse disease-specific survival (DSS) and recurrence-free survival (RFS) and whether it was associated with more aggressive behavior in small renal masses.
In total, 703 ccRCC tumors were analyzed using fluorescence in situ hybridization (316 tumors) and cytogenetics (388 tumors). Tumor grade, classification, and size; 9p status; Eastern Cooperative Oncology Group performance status (ECOG PS); lymph node involvement; and the presence of metastasis were recorded. Outcomes were stratified by 9p status, and a Cox proportional hazards models was constructed using TNM staging, ECOG PS, tumor size, tumor grade, and 9p status.
Deletions of 9p were detected in 97 tumors (13.8%). At presentation, 9p-deleted tumors were larger and were more likely to be high grade (grade 3 or 4), to have a high tumor (T) classification (T3-T4), and to have lymph node or distant metastases (P < .01). The median DSS for patients with and without 9p deletions was 37 months and 82 months, respectively (P < .01). In patients with localized disease, the median RFS in those who had 9p deletions was 53 months and was not reached in those without 9p deletions (P < .01). In patients who had localized lesions that measured ≤4 cm in greatest dimension, 9p-deleted tumors were more likely to recur (19% vs 2%; P = .01).
The well described stage migration in renal tumors that results from increased use of abdominal imaging has resulted in increasing numbers of incidentally discovered, localized lesions. The segment comprised of tumors <4 cm has grown faster than other categories,1 and the need to treat many of these lesions has been questioned because only a small percentage of them are perceived to pose an imminent threat to patients.2 However, a subset of small renal tumors is associated with a well described metastatic potential.3 Core biopsy can determine subtype and grade (high/low) with reasonable accuracy,4 but lesions with a similar histology, grade, and tumor size remain notoriously heterogeneous in their potential for growth and spread. Therefore, additional markers of aggressiveness that can be determined by biopsy would be useful in selecting high-risk but clinically low-stage lesions for more intense imaging or exclusion from surveillance strategies. Furthermore, many patients with higher stage localized tumors are at high risk for the development of metastatic disease after nephrectomy. Biomarkers of aggressiveness can help select patients most likely to benefit from an adjuvant agent, several of which are currently in phase 3 trials.
Clear cell renal cell carcinoma (ccRCC) classically is associated with deletions of chromosome 3p or mutations of the von Hippel-Lindau gene found at 3p26-p25; these deletions and mutations are present in over half of ccRCC tumors and are associated with less aggressive disease.5, 6 Recently published results from our institution regarding cytogenetic changes in 282 ccRCC tumors indicated an association between deletions of 4p, 9p, and 14q and more aggressive tumors, with 9p deletions maintaining an independent association when factoring in tumor grade and stage.7 Previous investigators also have reported an association with aggressive lesions and losses of chromosome 9.8-11 In addition, a small comparative genomic hybridization study at our institution suggested that chromosome 9 deletions were associated with a poor response to immunotherapy in patients with metastatic disease.12 For these reasons, we judged that it would be worthwhile to undertake a larger study.
Deletions of chromosome 9p can be detected by cytogenetic analysis of cultured tumor cells or by fluorescence in situ hybridization (FISH). Cytogenetics can be performed only on fresh tissue, whereas FISH may be performed on fresh or archived specimens. At our institution, cytogenetic analysis has been performed in a prospective fashion on most RCC specimens since 2001. In addition, an institutional RCC tissue microarray (TMA) contains sections from hundreds of nephrectomy specimens between 1989 and 2000. For the current study, we analyzed the impact of chromosome 9p deletions on the survival of 703 patients with ccRCC based on a cohort comprised of 316 patients who were evaluated by FISH and of 388 patients who were evaluated by cytogenetic analyses, including 282 patients who were the focus of a previous publication.7 To our knowledge, this constitutes the largest study of chromosome 9p deletions in RCC to date, and has allowed us to determine the prognostic significance of chromosome 9p deletions in ccRCC in relation to clinical and pathologic variables and within specific patient subsets. An additional 68 nonccRCC tumors also were analyzed as a basis of comparison.
MATERIALS AND METHODS
The University of California, Los Angeles (UCLA) RCC TMA was constructed using previously described methods. Briefly, formalin-fixed, paraffin-embedded kidney tumor specimens were obtained from the tissue archive of the UCLA Medical Center. Core biopsies that measured 0.6 mm in greatest dimension—3 from representative regions of each tumor and 1 from adjacent normal tissue—were arrayed using a custom-built instrument. Four micrometer-thick sections from the resulting TMA block were transferred to glass slides using the paraffin sectioning aid system (adhesive-coated slides PSA-CS4x, adhesive tape, ultraviolet lamp; Instrumedics Inc., Hackensack, NJ) to support the cohesion of 0.6-mm array elements.
FISH was performed using the locus-specific identifier (LSI) p16/centromeric enumeration probe 9 (CEP 9) Dual Color Probe (Vysis, Downers Grove, Ill). The LSI p16 SpectrumOrange probe spans approximately 190 kb and contains several genetic loci, including D9S1749, D9S1747, p16 (INK4A), p14 (alternate reading frame [ARF]), D9S1748, p15 (INK4B), and D9S1752. The CEP 9 SpectrumGreen probe hybridizes to alpha satellite sequences specific to chromosome 9. Slides were baked overnight at 56°C in a dry incubator then dewaxed, dehydrated, digested, and air-dried in an automated VP 2000 processor (Vysis). Ten microliters of the probe mixture were applied to the hybridization area and covered with a 22 × 30 mm glass coverslip (Fisher Scientific, Fremont, Calif). The edges were sealed with rubber cement, and the slides were incubated in a Hybridizer (HyBrite Assay; Vysis) at 73°C for 2 minutes then at 37°C for 20 hours to first codenaturate the probe and chromosomal DNA and then allow hybridization. Rubber cemented coverslips were then removed, and the slides were placed in a posthybridization wash solution (2 × standard saline citrate and 0.3% NP-40) at 72°C for 2 minutes. After rinsing the slides in 1 × phosphate-buiffered saline for 1 minute, they were air-dried in the dark for 30 to 60 minutes; then, 4′,6-diamindino-2-phenylindole dihydrochloride-1 counterstain (10 μL; Vysis) was applied to the hybridization area and coverslipped.
The slides were viewed under an Axioplan2 fluorescent microscope (Zeiss, Oberkochen, Germany) and were interpreted independently by 2 separate investigators. FISH analysis was performed by counting the number of signals in all tumor nuclei. To interpret the FISH probe counts and to define 9p deletion status, we counted the number of 9p21 (red) nuclear signals and centromeric CEP 9 (green) nuclear signals for the cells in each tumor spot. Tumor spots with a majority of 2 red and 2 green signals were considered normal, samples with a large majority of 1 red and 1 green signal were interpreted as monosomy (deletion of 1 copy) of chromosome 9, and those with 0 or 1 red signal and 2 green signals were interpreted as displaying specific deletion of 1 or both copies of chromosome 9p. Tumors that had fewer than expected red signals per cell were designated as 9p deleted. However, a tumor had to have at least 2 of the 3 microarray tissue spots display an abnormal pattern to be designated as having a 9p deletion. This approach can be used commonly in practical clinical assessment of gene copy number and generally reflects the average copy number of the cell population examined.
For cytogenetic analysis, viable tumor samples were collected immediately after surgery and aseptically minced into 2-mm to 3-mm pieces using a scalpel. After tissue dissociation with collagenase II (5 mg/mL/100 g tissue; Worthington Biochem, Freehold, NJ), the cells were washed and then cultured in RPMI 1640 supplemented with 20% fetal bovine serum and 1% penicillin/streptomycin at a final concentration of 1 or 2 × 105 cells/5 mL medium in a T-25 flask. The cultures were harvested when growth was subconfluent and the cells were actively dividing. Cells were subjected to hypotonic treatment (potassium chloride, 0.075 M) for 30 minutes at 37°C and were fixed in methanol and acetic acid (3:1). Chromosomes were banded using the G-bands by pancreatin using Giemsa technique. Twenty metaphases were investigated and analyzed in accordance with the International Standing Committee on Human Cytogenetic Nomenclature by 1 clinical cytogeneticist (N.R.).
Clinical data were gathered from the Institutional Review Board-approved UCLA kidney cancer database. After nephrectomy, patients with localized disease typically were followed according to a risk-stratified surveillance protocol using the UCLA Integrated Staging System. The protocol, which has been described elsewhere in detail,13 involves intermittent imaging of the chest, abdomen, and pelvis. Recurrence was defined as the appearance of suspicious, enlarging lesions at typical sites of RCC spread (lungs, mediastinal or retroperitoneal lymph nodes, liver, bone, brain). Lesions in atypical locations (eg pancreas, skin) underwent diagnostic biopsy.
Clinical and pathologic variables were recorded and compared between the patients with and without 9p deletions. Categorical variables were compared using the chi-square test and the Fisher exact test, as appropriate. Continuous variables were summarized as means with ranges and were compared using Student t tests. Disease-specific survival (DSS) was estimated using the Kaplan-Meier method, and group differences were compared with the log-rank test. A multivariate Cox proportional hazards model was used to analyze variables that were associated with time to death from RCC and the time to recurrence in patients who presented with localized disease. The predictive accuracy of the prognostic models was assessed by using the Concordance index (C index). Statistical analysis was performed using the SPSS 15.0 software package (SPSS Inc., Chicago, Ill).
A cohort comprised of 703 ccRCC tumors was analyzed by either FISH (316 tumors) or cytogenetics (388 tumors). Chromosome 9p deletions were detected in 97 tumors (13.8%). The analytic method did not affect the detection rate of 9p deletions (13.6% for cytogenetics vs 13.9% for FISH). In the FISH analysis, adjacent normal tissue, which served as a negative control for each tumor, exhibited no 9p deletions in any specimen. The majority of deletions (79%) were caused by monosomy of the entire chromosome 9. Pathologic variables for the patients who had ccRCC with and without 9p deletions are summarized in Table 1. Sixty-eight additional non-ccRCC tumors on the TMA also were analyzed with FISH. Only 2 of 39 papillary tumors (5%), 0 of 8 chromophobe tumors, and 0 of 4 collecting duct tumors had 9p deletions. The 2 papillary tumors that demonstrated 9p deletion were papillary type 2 tumors that were advanced at presentation and were associated with rapid disease progression and RCC-specific death in both patients. It is noteworthy that none of the benign tumors (8 angiomyolipomas and 9 oncocytomas) displayed 9p deletions.
Table 1. Association of Chromosome 9p Deletions With Pathologic Variables
No. of Patients (%)
No 9p Deletion, n=606
9p Deletion, n=97
Mean size, cm
Lymph node status
Lymph node or metastasis positive
The mean tumor size was significantly larger in tumors with 9p deletions (7.9 cm vs 6.7 cm; P < .01). When the tumors were divided into high and low tumor (T) classification categories (T1 or T2 vs T3 or T4), significantly more 9p-deleted tumors had a high T classification compared with tumors that did not have 9p deletion (60% vs 42%; P < .01). Categorizing Fuhrman nuclear grade in a similar fashion, 9p-deleted tumors were more likely to be high grade (grade 3/4 vs grade 1/2) than tumors without deletions (55% vs 39%; P < .01). Lymph node and/or distant metastases at presentation also were more likely in 9p-deleted tumors (57% vs 34%; P < .01).
The mean follow-up for the entire cohort was 40 months (range, 1-218 months). For patients who presented with localized disease, mean follow-up was 41 months (42 months for those with lesions ≤4 cm). The median Kaplan-Meier estimated DSS for all patients with 9p-deleted tumors was 37 months versus 82 months for patients with non-9p deleted tumors (hazard ratio [HR], 1.72; P < .001)(Fig. 1, top). At 5 years, 44% of patients without 9p deletions had died of RCC compared with 68% of patients with 9p deletions. For the 260 patients who presented with metastatic disease (lymph node or distant), there was no difference in DSS when they were stratified by 9p deletion status (HR, 1.02; P = .9)(Fig. 1, middle). Conversely, in the 443 patients who presented with localized disease, there were 64 recurrences and 48 deaths from RCC, and recurrence-free survival (RFS) and DSS in patients who had localized disease stratified by 9p deletion status differed significantly between groups. Thirteen recurrences developed in the 43 patients with 9p deletions, whereas 52 recurrences developed in the 400 patients without 9p deletions (30% vs 13%; P < .01). The median estimated RFS was 53 months for the group with 9p-deleted tumors and was not reached for the group without deletions, and their 5-year RFS rates were 49% and 77%, respectively (P < .001)(Fig. 2). Multivariate Cox regression analysis, including T classification, Fuhrman grade, tumor size, and 9p deletion status, was performed with respect to the time to RCC recurrence in the cohort with localized disease, and deletion of 9p demonstrated independence as a prognostic factor. The addition of 9p deletion to a pathologic model with T classification, Fuhrman grade, and tumor size resulted in a slight but significant increase in predictive accuracy (the C index increased from 79.9% to 80.4%; P = .01). Regarding cancer-specific survival, the 1-year, 2-year, and 5-year DSS rates were 96%, 86%, and 67%, respectively, for patients with 9p deletions versus 98%, 94%, and 87%, respectively, for patients without deletions (Fig. 1, bottom). Multivariate Cox regression analysis for this cohort with localized disease, including 9p status, Eastern Cooperative Oncology Group performance status, T classification, and Fuhrman grade demonstrated an independent effect for 9p deletion with respect to DSS (P = .015).
In a subset of 207 patients with small ccRCC tumors (tumors with a maximal dimension ≤4 cm), 9p deletions were present in 21 patients (10%). It is noteworthy that there was not a significant difference in the mean primary tumor size between patients with and without 9p deletions (3.0 cm vs 2.7 cm, respectively; P = .22). However, 9p deletions were associated significantly with the presence of lymph node or distant metastases in this group (23% vs 7.5%; P = .03).The 5-year DSS rate differed significantly in this cohort between patients with and without 9p deletion (56% vs 90%; P < .01)(Fig. 3). In 188 patients who had N0M0 tumors ≤4 cm, 3 of 16 patients (19%) with 9p deletions experienced recurrence versus 4 of 172 patients (2%) without 9p deletions (P = .01). Furthermore, RFS was significantly worse in the patients with 9p deletions (Fig. 4). The 1-year, 2-year, and 5-year RFS rates were 99%, 99%, and 97%, respectively, in patients without loss of 9p versus 93%, 93%, and 68%, respectively, in patients with 9p deletion (P < .01). In a Cox regression analysis that included T classification, tumor grade, tumor size, and 9p deletion status, 9p deletion (HR, 6.65; P = .021), but not tumor size (HR, 1.43; P = .54), had an independent effect on RFS (Table 2). The predictive accuracy of the model increased significantly when 9p deletion status was added (the C index increased from 86.3% to 89.0%; P = .03).
Table 2. Cox Regression Analysis for Predicting Recurrence-Free Survival in Patients With Localized Renal Cell Carcinoma (N0M0) <4 Centimeters
The main findings of the current study are that deletions of chromosome 9p are present in 14% of ccRCC, are associated with more aggressive lesions that are more likely to involve regional lymph nodes or distant sites at presentation, and are more likely to demonstrate metastatic disease progression after nephrectomy for clinically localized disease. Although 9p-deleted tumors are more likely to be present in larger tumors with a higher grade and stage, they also are present in smaller and lower grade tumors, and 9p status (but not tumor size) was associated independently with disease recurrence in localized, small renal tumors.
Knowledge of the 9p deletion status, therefore, provides additional relevant clinical information about the expected clinical course of these patients, particularly in small renal masses (SRMs) that appear to have a more lethal phenotype. Currently, there are no dependable preoperative prognostic factors that may be used to predict the malignant potential of these smaller tumors, which leaves little upon which to base a decision regarding the necessity of immediate treatment or the safety of pursuing a course of active surveillance. Within this subset, size does not appear to provide additional prognostic information, as demonstrated in this report and in a multi-institutional study.3 The growth rate of a lesion also may belie the nature of SRMs, with tumors that have no growth demonstrating a rate of malignancy similar to the rate observed in growing lesions.14 The 9p deletion HR for recurrence after resection of an SRM was 6.65 on multivariate analysis, which was higher than its HR for the overall cohort of patients with localized disease. In addition, our analysis on predictive accuracy indicated that 9p deletion status provided prognostic information in addition to tumor grade and markedly differentiated these small tumors from the tumors without 9p deletion. A preoperative biopsy can provide both approximate grade and the 9p deletion status (using FISH) and, thus, may provide a strong indication regarding the aggressive potential of SRMs. Furthermore, the finding that 9p deletions in SRMs are associated with lymph node or distant metastasis at presentation suggests that these lesions are not good candidates for observation.
Previous investigators observed the loss of regions of chromosome 9 in RCC, and these deletions have been associated with more aggressive tumors. In early study of 42 tumors, 33% had loss of heterozygosity (LOH) of at least portions of chromosome 9, and most had deletions at 9p21-22.11 A later study of 72 tumors reported LOH of 9p in 31% of a set locally advanced ccRCC tumors and an association with recurrence.8 More recently, a study of 73 ccRCC tumors with FISH identified deletions of 9p in 18% and an independent association of 9p deletions with DSS.10 These earlier studies were limited by their smaller numbers, which prevented subset analyses.
The gene(s) on chromosome 9 that, when deleted or otherwise inactivated, confers more aggressive behavior on ccRCC is uncertain. It appears likely that the region(s) of interest lies on 9p rather than elsewhere on chromosome 9 in light of previous results8-11 and based on the finding that 21% of our 9p-deleted lesions had 2 9q alleles. The cyclin D-dependent kinase 2NA (CDK2NA)/ARF tumor suppressor gene (INK4A) located at 9p21 that codes for the p16 and p14 regulatory proteins is a candidate whose loss or silencing potentially plays a role in more aggressive ccRCC. The protein p16 is a cyclin D kinase inhibitor that is active in the retinoblastoma/elongation factor 2 (Rb/EF2) pathway; whereas p14, a protein coded from exons that overlap the p16 coding region, is an inhibitor of the p53-inhibiting murine double-minute protein 2 (mdm2). The loss of function of 1 or both of these proteins may allow tumor cells to behave more aggressively. Carbonic anhydrase IX (CA9) also is located on 9p, and lower expression of CA9 has been associated with shorter survival and poorer response to immunotherapy in metastatic lesions.15, 16 In addition, the interferon α(IFN-A) and IFN β(IFN-B) genes are located on 9p distal to the p16/p14 locus, and exogenous IFN-A analogs are effective in the treatment of metastatic ccRCC. We used a FISH probe that was specific for the 9p21 locus in part because the probe is a commercially available, off-the-shelf product that is well validated and already is in clinical use as part of the UroVysion (Vysis) bladder tumor assay.
FISH and cytogenetics may underestimate the extent to which 9p alterations are present. After the loss of 1 copy, a solitary chromosome 9 can duplicate, leading to a normal copy number for chromosome 9 (uniparental disomy). However, abnormalities on the solitary chromosome will be present on the duplicated copy, which results in LOH because there is no counterbalancing allele. In previous studies using LOH analysis, higher rates of chromosome 9p alteration were observed,8, 9, 11 but it is unclear whether this was because of greater sensitivity or differences in the sets of tumors that were analyzed. Future immunohistochemical and gene expression studies will help further determine the loss of which gene(s) on chromosome 9 is responsible for the increased aggressiveness of 9p-deleted ccRCC lesions.
In conclusion, loss of chromosome 9p is present in 14% of ccRCCs and is associated with more aggressive lesions, which are more likely to involve regional lymph nodes or distant sites at presentation and are more likely to demonstrate disease progression after nephrectomy for clinically localized disease. Loss of 9p confers a more aggressive phenotype on these lesions and portends a worse prognosis in patients with localized disease. In the current study, 9p status, but not tumor size, was associated independently with disease recurrence in localized, small renal tumors, raising the possibility of using preoperative FISH analysis to characterize the malignant potential of SRMs and help determine which lesions should be managed aggressively. Further studies are needed to validate these findings and to elucidate the specific genes on 9p that are involved in conferring this lethal phenotype.
CONFLICT OF INTEREST DISCLOSURES
This research was supported in part by an unrestricted research grant from Novartis.