Limitations of preoperative biopsy in patients with metastatic renal cell carcinoma: comparison to surgical pathology in 405 cases


  • The present study was presented at the Society of Urologic Oncology Annual Meeting 2010.

Christopher G. Wood, MD, FACS, Professor and Deputy Chairman, Department of Urology – Unit 1373, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030, USA. e-mail:


Study Type – Diagnostic (cohort)

Level of Evidence 2b

What's known on the subject? and What does the study add?

Although there have been many investigations of biopsy for small renal masses, there are scant data on the accuracy of biopsy in the setting of metastatic renal cell carcinoma (mRCC). We report a large series of biopsies and compare with nephrectomy pathology in patients with mRCC.

The present study highlights the inaccuracy of biopsy in the setting of metastatic disease, which is related to sampling error because of heterogeneity within the tumour and among metastases. These limitations are important to realize when designing trials that depend on pathological findings from biopsy and not nephrectomy. In addition, we found that biopsy of primary tumours were more likely than biopsy of metastatic sites to be diagnostic of RCC. Future studies with multiquadrant biopsies of primary tumours could yield the most accurate pathological results for future studies.


  • • To evaluate the ability of preoperative biopsy to identify high-risk pathological features by comparing pathology from preoperative metastatic site and primary tumour biopsies with nephrectomy pathology in patients with metastatic renal cell carcinoma (mRCC).


  • • We reviewed clinical and pathological data from patients who underwent biopsy before cytoreductive nephrectomy for mRCC at MD Anderson Cancer Center (MDACC) from 1991 to 2007.
  • • Percutaneous biopsy techniques included fine-needle aspiration, core needle biopsy or a combination of both techniques.


  • • The pathology of 405 preoperative biopsies (239 metastatic site, 166 primary tumour) from 378 patients was reviewed at MDACC before cytoreductive nephrectomy.
  • • The biopsy and nephrectomy specimens had the same histological subtype in 96.0% of clear-cell renal cell carcinomas (RCCs) and 72.7% of non-clear-cell RCCs.
  • • Of 76 nephrectomy specimens where sarcomatoid de-differentiation was identified, only seven (9.2%) were able to be identified from the preoperative biopsy.
  • • In 38.3% of patients, the same Fuhrman grade was identified in both the biopsy and nephrectomy specimens.
  • • A definitive diagnosis of RCC was more likely to be reported in primary tumour biopsies than in metastatic site biopsies. (P < 0.001).


  • • Preoperative biopsy has limited ability to identify non-clear-cell histological subtype, Fuhrman grade or sarcomatoid features.
  • • When surgical pathology is not available, a biopsy obtaining multiple samples from different sites within the primary tumour should be recommended rather than limited metastatic site biopsy to identify patients for clinical trials.

fine-needle aspiration


MD Anderson Cancer Center


metastatic RCC


Patients being evaluated for metastatic RCC (mRCC) often undergo a biopsy of the metastatic site or primary tumour to obtain a tissue diagnosis. Prognostic information from the biopsy can be used to help counsel patients before surgery or systemic therapy. Known poor prognostic factors in patients with mRCC can be identified from biopsy, including high Fuhrman grade, non-clear-cell histology and the presence of sarcomatoid features.

The accuracy of biopsy findings in small indeterminate renal masses has improved significantly [1]; however, the ability of percutaneous primary tumour biopsy to identify advanced pathological features in patients with large renal masses is limited [2]. Possible explanations for the inaccuracy seen in larger primary renal tumours include tumour heterogeneity or inadequate sampling. Biopsy of metastatic lesions can be obtained to confirm that a patient has mRCC, but it is unknown whether the pathological features of metastatic lesions correlate to the primary tumour. These questions are becoming ever more important as clinical trials enroll patients for pre-surgical or non-surgical therapy with targeted agents [3,4]. The purpose of the present study is to evaluate the ability of preoperative biopsy to identify high-risk pathological features by comparing preoperative metastatic site and primary tumour biopsy specimens with nephrectomy specimens in a series of patients with mRCC undergoing cytoreductive surgery.


After institutional review board approval, we reviewed the records of all patients who underwent biopsy before cytoreductive nephrectomy for mRCC at MD Anderson Cancer Center (MDACC) from 1991 to 2007. Patients were considered eligible for the study if their biopsy was either done at MDACC or done at an outside institution and reviewed by an expert in kidney cancer pathology at MDACC. Percutaneous biopsies were performed with fine-needle aspiration (FNA) and/or core needle techniques using ultrasonography, CT or fluoroscopy for guidance. The location of biopsy performed was at the discretion of the physician who treated the patient.

Histological subtypes reported were based on the 2004 World Health Organization consensus classification, and the pathology was reviewed to conform to this classification when applicable. After initial pathological review by an expert genitourinary pathologist, individual immunohistochemical stains were used when deemed necessary and diagnosis was made according to morphology and staining characteristics. In cases where the attending pathologist was unable to report a histological subtype using the available tissue, specimens were designated as RCC, subtype unspecified.

Chi-squared or Fisher's exact test was used to analyse differences in accuracy among groups undergoing biopsy. A (two-sided) P < 0.05 was considered to indicated statistical significance. Stata v.10.1 (Statacorp, College Station, TX, USA) was used for all analyses.


An institutional database identified 566 patients undergoing nephrectomy with measurable metastatic disease from 1991 to 2007. The pathology of 405 preoperative biopsies in 378 patients was reviewed at MDACC. Patient characteristics are shown in Table 1. A biopsy was performed at a metastatic site in 239 patients and in the primary kidney tumour in 166 patients (Table 2).

Table 1.  Patient characteristics
 Median (range) or n (%) (unless noted otherwise)
No. of patients378
Age at surgery, years57.1 (15–84)
Diameter of primary tumour, cm9.0 (1.2–32)
Mean (range) BMI, kg/m227.7 (16–49)
 Male266 (70.4)
 Female112 (29.6)
 Caucasian316 (83.6)
 Hispanic38 (10.0)
 African-American17 (4.5)
 Other17 (4.5)
Table 2.  Pre-surgery biopsy characteristics
  n (%)
Total biopsies405
Metastatic site biopsies200 (100)
 Bone85 (42.5)
 Lung54 (27.0)
 Soft tissue41 (20.5)
 Brain10 (5.0)
 Liver4 (2.0)
 Adrenal4 (2.0)
Primary kidney tumour biopsies166 (100)
Lymph node biopsies39 (100)
 Retroperitoneal17 (43.6)
 Thoracic11 (28.2)
 Supraclavicular11 (28.2)
Biopsy type 
 FNA203 (50.1)
 Core30 (7.4)
 FNA + core32 (7.9)
 Excisional/curettage34 (8.4)
 Missing data107 (26.4)
Biopsy institution 
 MDACC163 (40.2)
 Elsewhere242 (58.8)
Imaging used for biopsy 
 CT85 (21.0)
 Ultrasonography46 (11.4)
 Fluoroscopy29 (7.2)
 None (open biopsy)34 (8.4)
 Missing data211 (52.1)
Patients receiving preoperative biopsy/cytoreductive nephrectomy
 1991–199679/94 (84.0)
 1997–2001132/247 (53.4)
 2002–2007168/219 (76.7)

Histological subtypes are shown in Table 3. In 171 cases, a histological subtype was assigned to both the biopsy and the nephrectomy specimens. The biopsy and nephrectomy had the same histological subtype assigned in 143/149 (96.0%) clear-cell RCCs and 16/22 (72.7%) cases with non-clear-cell subtypes of RCC.

Table 3.  Histological subtypes from biopsy and nephrectomy pathology
HistologyBiopsy [n (%)]Nephrectomy [n (%)]
 Clear-cell type156 (38.5)299 (73.8)
 Papillary16 (4.0)22 (5.4)
 Chromophobe03 (0.7)
 Collecting duct2 (0.5)6 (1.5)
 RCC, unclassified7 (1.7)39 (9.6)
 RCC, subtype unspecified125 (30.9)36 (8.9)
Neoplasm unspecified75 (18.5)0
No tumour identified24 (5.9)0

Of 76 nephrectomy specimens where sarcomatoid de-differentiation was identified, only seven (9.2%) were able to be identified in the preoperative biopsy. In three patients, sarcomatoid features were present in the biopsy (one primary tumour, two metastatic sites) but not identified in the nephrectomy specimen.

A numerical Fuhrman grade was assigned to 120 biopsy specimens (104 primary tumours, 16 metastatic sites). In 46 (38.3%) patients the same Fuhrman grade was identified in both the biopsy and nephrectomy specimens. When categorizing as high or low Fuhrman grade, 89 (74.2%) cases were able to be identified accurately from preoperative biopsy. There was no correlation between ability to identify the grade correctly and the year when the biopsy was performed (P= 0.40).

In 99/405 (24.4%) patients, RCC was not able to be identified from the biopsy specimen. No tumour identified or unspecified neoplasm was reported in seven (4.2%) and nine (5.4%) of patients with primary tumour biopsy and in 17 (7.1%) and 64 (26.8%) patients with metastatic site biopsies. Of 75 biopsy specimens reported as unspecified neoplasms, the nephrectomy specimen was identified as clear-cell RCC in 42 (56%), RCC subtype unspecified in 14(18.7%), unclassified RCC in 11(14.7%), papillary RCC in four (5.3%), and collecting duct RCC in four (5.3%). In 24 specimens with fibrosis or benign histology in the biopsy, the nephrectomy pathology was clear-cell RCC in 19 (76%) patients and unclassified RCC in five (24%) patients. Primary tumour biopsy was able to identify RCC in 150/166 (90.4%), as compared with metastatic site biopsy which identified RCC in 156/239 (65.3%) patients (P < 0.001).

For distinguishing between clear-cell and non-clear cell subtypes, the sensitivity and specificity of metastatic site biopsy were 0.952 (95%CI: 0.875–0.985) and 0.625 (95% CI: 0.259–0.897), while those of primary tumour biopsy were 0.954 (95%CI: 0.864–0.988) and 0.769 (95%CI: 0.460–0.938). When considering the ability to identify sarcomatoid features, the sensitivity and specificity of metastatic site biopsy were 0.071 (95%CI: 0.019–0.205) and 0.990 (95%CI: 0.960–0.998), and those of primary tumour biopsy were 0.117 (95%CI: 0.038–0.283) and 0.992 (95%CI: 0.952–1.000). There was no statistically significant difference between metastatic site and primary tumour biopsy in the ability to identify clear-cell vs non-clear-cell subtypes (P= 1.00) or sarcomatoid de-differentiation (P= 0.69).


In the present series, the concordance between biopsy and nephrectomy specimens was highest when clear-cell RCC was identified from biopsy. Biopsy was limited in the ability to predict non-clear-cell histological subtype and Fuhrman grade or to identify the presence of sarcomatoid features. The recognition of these limitations is important to consider when using biopsy data to guide treatment decisions or enrolment in clinical trials.

In patients with a characteristic mass in the kidney and evidence of metastatic disease typical for RCC (e.g. lung or retroperitoneal lymph node) who are surgical candidates, a biopsy may not be obtained because evaluation of nephrectomy specimen will provide pathologic information. However, in patients with atypical sites of metastasis or non-characteristic renal masses, or in those considering clinical trials with pre-surgical therapy, obtaining a biopsy might be indicated to make a diagnosis of RCC. The number of patients receiving biopsy at our institution could be increased because of the large number of clinical trials available, which is apparent in the increased numbers of patients receiving biopsy in the targeted therapy era. When considering biopsy of metastatic sites, size >1 cm, lack of necrosis and a location that is amenable to biopsy are the primary considerations for selection. Often, the primary tumour fits these criteria best, but the decision regarding the site of biopsy is made on an individual basis given the pattern of metastasis and differential diagnosis.

With the recent emergence of targeted therapy as first-line systemic therapy for mRCC [5], physicians have more options available for treatment, and some researchers have advocated expanding the role of biopsy in gaining prognostic information on patients with mRCC [6]. Biopsy pathology could be especially important for those patients unable to undergo cytoreductive nephrectomy, or those enrolled in systemic therapy trials with their primary tumour in situ. Unfortunately, the accuracy of biopsy in metastatic kidney cancer is not well described as there are few studies that compare the primary tumour surgical pathology with biopsy of metastatic sites or the primary tumour in mRCC. Most evidence showing the accuracy of biopsy for RCC has been obtained in patients with small renal masses, with less chance for sampling error [1,7].

We have previously reported the inaccuracy of preoperative percutaneous biopsies of mRCC primary tumours in identifying sarcomatoid features, non-clear-cell histology and Fuhrman grade [2]. The findings of the present study in relation to biopsy specimens from RCC metastatic sites are similar. There was no difference between biopsies from metastatic sites and those from primary tumours in the ability to predict advanced pathological features. Possible explanations of our findings include sampling bias, observer variability, tumour heterogeneity and necrosis.

The primary tumours of mRCC frequently have subpopulations of cells with different genetic alterations and appearance, causing intra-tumour heterogeneity [8]. In patients with metastatic tumours, specific subpopulations can also be identified in distant metastases [9,10], causing heterogeneity among metastatic sites within an individual patient. When carrying out a biopsy of a metastatic lesion or primary tumour, it is likely that only one subpopulation of cells is sampled, and prognostic information is therefore based on only one subpopulation of cells. Conversely, when the pathologist examines a nephrectomy specimen, multiple sections are available for interpretation and evaluation of the different histological patterns. Pathologists could use immunohistochemical techniques to help provide prognostic information [11], and in the future genetic analysis of primary tumours could lead to better identification of markers for metastatic potential in RCC [12].

The presence of sarcomatoid features in pathological specimens could be associated with a poor prognosis and a lack of benefit from cytoreductive nephrectomy [13,14]. The recommended systemic therapy is often different in patients with sarcomatoid features, and early identification could be beneficial for placement in clinical trials [15]. Unfortunately, in biopsy specimens from mRCC primary tumours or metastatic sites, sarcomatoid features were only identified in one of 10 patients before surgery. Histological and genetic heterogeneity has been demonstrated with sarcomatoid transformation [16,17] and further supports tumour heterogeneity as an explanation for pathological differences between biopsy and nephrectomy specimens.

Most clinical trials with targeted agents have excluded patients with non-clear-cell histological subtypes, and retrospective studies suggest that response rates to these agents is lower [18]. Patients with non-clear-cell histology might not benefit from cytoreductive nephrectomy [13], and with new investigational therapies being developed [19] it will become increasingly important to identify these patients before surgery. Given the limited ability to distinguish subtypes of non-clear-cell RCC, there is considerable opportunity for improvement with advances in immunohistochemistry and cytogenetic techniques [11,20].

There are multiple phase 2 and phase 3 clinical trials underway investigating systemic treatment of RCC using targeted therapy with the primary tumour in situ[10,12]. Biopsy of the primary tumour or metastatic site is used to establish a tissue diagnosis as part of entrance criteria. Patients with non-clear-cell histological subtypes or sarcomatoid elements are frequently excluded because of poor survival and decreased response to therapy. Both metastatic site and primary tumour biopsies are limited in their ability to identify these features reliably, because of sampling error and tumour heterogeneity. In the absence of primary tumour surgical pathology, a biopsy obtaining multiple samples from different sites within the primary tumour should be recommended, rather than limited metastatic site biopsy, to identify patients for enrolment in clinical trials.

The non-diagnostic rate in the current series was 5.9%, which is similar to rates seen in biopsy series of small renal masses [1]. The rate of unspecified neoplasm in preoperative biopsy (18.5%) could reflect the technical limitations with conventional histology of the limited tissue available from preoperative biopsy or the de-differentiated nature of high-grade mRCC. Routinely repeating the biopsy might not have been warranted when nephrectomy was planned for straightforward patients with a large renal mass and a common metastatic site. However, the biopsy rate of unspecified neoplasm would probably be decreased if repeat biopsy were performed on all patients. As immunohistochemical and cytogenetic techniques continue to evolve, our ability to classify neoplasms from small tissue samples is expected to continue to improve [11,20]

Most biopsies in the present study were obtained with FNA. When processing cytological specimens at our institution, cells are embedded in a paraffin block facilitating immunohistochemical analysis. However, we believe that a combined approach with both FNA and core biopsy is most informative [21], and it is our current practice to obtain tissue using both methods when carrying out a percutaneous biopsy [2].

We have reported the largest series comparing preoperative biopsy pathology with nephrectomy specimens in patients with mRCC. There are several limitations to the current study which are to do with its retrospective nature. All pathological specimens were reviewed by an expert genitourinary pathologist before surgery, but they were not reviewed independently for the current study. While modern techniques might improve the ability to make a diagnosis from biopsy tissue, it is unclear how this would affect the concordance between biopsy and nephrectomy pathology if this inaccuracy were the result of clonal heterogeneity within the primary tumour and metastases. Importantly, we evaluated concordance of biopsy pathology with nephrectomy pathology only when the pathologist was able to make a specific diagnosis from the biopsy specimen, and we did not include cases where diagnosis was in doubt from the biopsy specimen. We believe this is an accurate representation of pathological information from a centre of excellence and clinical decisions for nephrectomy were based on these data.

The data in the present study can be used to design prospective studies to evaluate the ability of biopsy to predict the clinical course of patients with mRCC. Multi-quadrant core biopsies of the primary tumour could help to improve the diagnostic potential of biopsies and decrease the sampling error in heterogeneous tumours. Physicians should be aware of the limits of biopsy in mRCC when counselling patients before pre-surgical therapy.


None declared.