Vol. 43 ı No. 5 ı 2007 www.biotechniques.com ı BioTechniques ı 639
Short Technical Reports
The prognosis associated with
renal cell carcinoma (RCC) can vary
widely. Small, incidentally diagnosed
renal tumors can be observed safely
for many years in patients who are
poor surgical candidates (1). Larger,
high-grade tumors carry risk of
recurrence, which may be as high as
50%, following re-section of a clini-
cally localized tumor (2). Accurate
determination of prognosis using
molecular markers may be useful
for patient counseling and selecting
treatment. However, there are no
molecular markers of prognosis in
routine clinical use for patients with
RCC. Although microarray-based
expression profiling studies have
identified a large number of poten-
tially prognostic markers, a barrier
to development of molecular test
for clinic use has been the lack of
readily available clinical samples for
validation of candidate markers.
Frozen tumor libraries are costly
to establish and maintain. However,
(FFPE) tissues are routinely archived
and stored by pathology departments.
Therefore, FFPE tumor samples are
widely available and can be annotated
with clinical information. We describe
steps for optimizing and validating
a quantitative reverse transcription
(RT-PCR) assay for expression
profiling of FFPE RCC. We evaluate
commonly used reference genes for
stability of expression in RCC and
determine the optimal number of
reference genes necessary to calculate
normalization factors when using
frozen or archival tissue. We validate
the assay by comparing expression
values measured in frozen and
archival tissue. Finally, we validate
individual prognostic markers for use
with archival tissue. These validated
markers can be used in large-scale
clinical studies, which are required
before a marker can be recommended
for clinical use. This approach can be
applied across tumor types to develop
assays for testing prognostic markers
using archive tissue.
MATERIALS AND METHODS
Patient Selection and Overall Study
To identify reference genes with
stable expression in renal tissue, 20 pairs
of clear cell RCC and adjacent normal
kidney were used. An equal number
of low-grade (Fuhrman grade 1 and 2)
and high-grade (Fuhrman grade 3 and
4) tumors were represented. Matching
pairs of frozen and FFPE, clear cell renal
tumors from 23 patients were utilized to
validate reference genes identified for
use with FFPE renal tumors. All tissue
samples were procured between 2002
and 2006 and were retrieved from the
Department of Pathology following
approval by the Institutional Review
Board (EDR 53605). Commonly used
reference genes (GUS, TFRC, TBP,
RPLP0, ACTB, HPRT1, SDHA, B2M,
and RPL13) (3,4) were evaluated for
stable expression in renal tumors.
For extraction of RNA from FFPE
tissue, four 10-μm sections were cut
from archival block. Excess paraffin
was trimmed using a scalpel cleaned
with RNaseZAP (Ambion, Austin, TX,
USA), and the sections were placed in
1.5-mL RNase-free Eppendorf tubes.
Sections were treated twice with 1
mL xylene for 30 min at 55°C while
rocking. The sections were washed
twice with 100% ethanol. RNA was
extracted from the paraffin samples
using the MasterPure kit (Epicentre
Biotechnologies, Madison, WI, USA).
For frozen tissue, 0.5 g snap-frozen
tissue was homogenized in TRIzol
and placed in a 1.5-mL RNase-free
Eppendorf tube. RNA was then
extracted using the TRIzol protocol
(Invitrogen, Carlsbad, CA, USA).
RNA from both FFPE and frozen
tissue was then treated with DNase
I for 30 min. The samples were
checked for residual genomic DNA
by TaqMan RT-PCR for ACTB. If
there was measurable DNA after 35
PCR cycles, the samples were treated
with DNase I for an additional 30
min, and the assay for residual DNA
Expression profiling of archival renal tumors by
quantitative PCR to validate prognostic markers
Sean T. Glenn, Craig A. Jones, Ping Liang, Dharam Kaushik,
Kenneth W. Gross, and Hyung L. Kim
Roswell Park Cancer Institute, Buffalo, NY, USA
BioTechniques 43:639-647 (November 2007)
Formalin-fixed paraffin-embedded (FFPE) tissues are routinely stored by most pathology
departments and are a widely available resource for discovery of clinically useful biomark-
ers. We describe our method for optimizing quantitative reverse transcription PCR (RT-PCR)
for expression analysis using frozen and archival tissue. Commonly used reference genes
were evaluated for stability of expression in normal kidney and clear cell renal cell carcino-
ma (RCC). Optimal reference genes for calculating normalization factors for RT-PCR were
ACTB, RPL13A, GUS, RPLP0, HPRT1, and SDHA when using FFPE RCC. The optimal
reference genes when using frozen RCC were ACTB, RPL13A, and GUS, confirming that
use of multiple reference genes improves accuracy when intact RNA from frozen renal tumors
are used. Expression of 16 markers previously reported to have prognostic significance in
RCC was determined in 23 matching frozen and FFPE renal tumors, representing a range
of tumor grades and stages; correlation coefficient for expression measured in frozen and
FFPE tumors was 0.921 (P < 0.001). All markers predicted survival when frozen tumors were
used and 14 of the 16 markers predicted survival when FFPE tumors were used as the source
of RNA. An optimized RT-PCR assay can accurately measure expression of most prognostic
640 ı BioTechniques ı www.biotechniques.com
Vol. 43 ı No. 5 ı 2007
Short Technical Reports
The final RNA concentration
(A260:0.025) and purity (A260:A280
ratio) was measured using a NanoDrop
ND-1000 spectrophotometer (NanoDrop
Technologies, Wilmington, DE, USA).
RNA quality was assessed for 28S
and 18S ribosomal degradation using
a 2100 Bioanalyzer and an RNA 6000
Nano LabChip kit (both from Agilent
Technologies, Santa Clara, CA, USA).
Reverse Transcription and
performed in an Applied Biosystems
2720 thermal cycler using the cDNA
Archive kit (Applied Biosystems,
Foster City, CA, USA). The reaction
was performed according to manufac-
turer’s recommendation, using random
primers and 1 μg total RNA. When
utilized, gene-specific primers were
included at 100 nmol/L for each gene.
Gene-specific primers were designed
using PerlPrimer v1.1.6 (perlprimer.
sourceforge.net). All TaqMan primers
and probes were custom designed
using Primer Express, version 2.0
(Applied Biosystems), and appropriate
mRNA consensus sequence from the
NCBI Entrez nucleotide database.
PCR product sizes were limited to 100
bases in length. TaqMan probes were
labeled with 5′-FAM as a reporter
and 3′-TAM as a quencher. All oligo-
nucleotides and probes were supplied
by Integrated DNA Technologies
(Coralville, IA, USA).
Expression of each gene was
measured in 25-μL reactions using
TaqMan Universal PCR Master Mix
(Applied Biosystems), 5 ng cDNA,
900 nM each primer, and 250 nM
probe. The PCR was carried out using
a MyiQ Real-Time PCR Detection
Hercules, CA, USA), and cycling
conditions were as follows: 50°C for
2 min and 95°C for 10 min for 1 cycle;
95°C for 15 s and 60°C for 1 min for
45 cycles. To compare gene expres-
sions between samples, the threshold
cycle (CT) was normalized using the
mean CT for reference genes. The
normalized mRNA level (expression as
log2 with arbitrary unit) was defined as
∆CT = CT (test gene) - CT (mean for the reference genes).
All reactions were performed in
triplicate, and results were averaged.
Each 96-well plate included a control
reaction using a previously charac-
terized cDNA and marker.
CT values were recorded in
Microsoft Excel 2003 for calculation
of expression. Reference genes were
evaluated using the freely available
geNorm software, version 3.4
is a Visual Basic applet for Microsoft
Excel that calculates a gene expression
stability measure M for a reference
gene as the average pairwise variation
for that gene with all other reference
genes and allows ranking of genes
according to stability of expression
(5). This method produces similar
results when compared with a strategy
that ranks reference genes based on the
standard deviation of the log-scaled
expression levels (6).
To determine the optimal number of
reference genes, a pairwise variation
V is calculated when a sequentially
increasing number of reference genes
are used. A large V indicates that the
added gene should be included for
calculation of the normalization factor.
A V < 0.15 was used to determine the
Figure 1. Denaturing RNA microcapillary
electrophoresis. One microliter (approximately
220 ng) total RNA extracted from renal tumors
was analyzed using an Agilent 2100 Bioanalyzer
and RNA 6000 NanoChip. Lane 2 is a water
control. Lanes 3–5 contain RNA from three rep-
resentative frozen renal tumors, and lanes 6–8
contain RNA from three representative FFPE
renal tumors. The RNA from FFPE tissue is de-
graded and does not have detectable 28S and 18S
ribosomal RNA bands.
Figure 2. Determination of the optimal reference genes for normalization of gene expressions.
(A) Nine reference genes ranked based on stability of expression in renal tumors and normal kidneys.
TaqMan PCR analysis was performed for nine reference genes using total RNA extracted from 20 frozen
clear cell renal carcinomas (10 high grade and 10 low grade) and adjacent normal kidneys. The geNorm
software, version 3.4 calculates a gene expression stability measure M for a reference gene as the aver-
age pairwise variation for that gene with all other reference genes, and allows ranking of genes accord-
ing to stability of expression. RPL13A and ACTB were the most stable genes and TFRC was the least
stable gene. (B) Determination of the optimal number of reference genes for expression analysis using
formalin-fixed paraffin-embedded (FFPE) renal tumors. TaqMan PCR was performed for nine reference
genes using total RNA extracted from five FFPE renal tumors. The geNorm software calculates a pair-
wise variation V for each sequential increase in number of reference genes. A large V indicates that the
added gene should be included for calculation of the normalization factor. A cutoff of 0.15 was used to
determine the optimal number of genes. Therefore, adding a seventh reference gene is not necessary and
the optimal number of reference genes is six.
642 ı BioTechniques ı www.biotechniques.com
Vol. 43 ı No. 5 ı 2007
Short Technical Reports
optimal number of reference genes.
For survival analysis, disease-specific
survival was the end point, and
expression status was the covariate
for the logrank analysis. Graphs and
Pearson correlation coefficients were
generated using Stata 8 (College
Station, TX, USA), and a two-sided P
value of 0.05 was considered statisti-
RESULTS AND DISCUSSION
measurement of gene expression.
RNA is extracted from tumor tissue
and converted to cDNA. Hundreds
of reactions can be performed and
analyzed simultaneously on multi-
well plates, allowing large numbers of
cases to be rapidly assessed. The assay
has a wide dynamic range, making
it suitable for validation of disease
related markers identified by micro-
array-based genomic profiling studies.
Previously described expression
profiling assays for RCC have relied
on high-quality RNA extracted from
fresh, frozen tumors. Paik et al. first
described expression profiling for
use in clinical practice using TaqMan
RT-PCR and archival breast tumors
(7). Two key features of their assay
allow accurate quantification of gene
expression using degraded RNA. PCR
primers were designed to produce
short products, and multiple reference
genes were utilized to normalize
As shown in Figure 1, total RNA
extracted from frozen RCC had sharp
28S and 18S ribosomal RNA bands.
However, RNA from FFPE tumors was
degraded and had no ribosomal RNA
bands. Therefore, others have reported
using primers located immediately
upstream to the PCR product in order
to optimize the synthesis of cDNA (7).
TaqMan PCR results were compared
using cDNA synthesized with random
oligonucleotide primers with or
without gene-specific primers. The
use of gene-specific primers did not
affect the RT-PCR results when used
with RNA from FFPE tumors <5 years
old. Thus, the use of random primers
alone simplifies reverse transcription,
and allows cDNA to be prepared in
bulk without having to anticipate the
genes being profiled.
One strategy for quantifying
gene expression using RT-PCR
is to normalize the expression
value for a gene of interest using
reference genes known to be stably
expressed. It is standard practice to
use a single reference gene when
expression profiling with frozen
tissue (8); however, several studies
have documented that reference gene
expression can vary considerably
and that use of multiple reference
genes may improve accuracy (3,4,9).
Therefore, it is important to document
stability of expression and identify
appropriate combination of reference
genes for the tissue being studied,
especially when the biological
significance of small differences in
expression is being assessed.
Figure 3. Analysis with frozen versus formalin-
fixed paraffin-embedded (FFPE) renal tumors.
Expression analysis was performed by TaqMan
PCR using matching frozen and FFPE renal tu-
mors. Each point represents mean expression for
one of 16 markers previously reported as a prog-
nostic marker in renal cell carcinoma. The corre-
lation coefficient is 0.902, and P value is <0.001.
Markers with perfect correlation between the two
tissue sources fall on the diagonal line.
RefSeq ID Sequence
TATA Box Binding ProteinReverse:5′-CGTGGCTCTCTTATCCTCATGA-3′
Ribosomal Protein L13aReverse:5′-TTGGTTTTGTGGGGCAGCAT-3′-3′
Succinate Dehydrogenase Reverse:5′-GTCGGAGCCCTTCACGGT-3′
Complex, Subunit AProbe:5′-AATGCCACCTCCAGTTGTCCTCCTCCA-3′
Phosphoribosyltransferase 1Probe: 5′-TGGCCTCCCATCTCCTTCATCACATCTC-3′
RPLP0 NM_053275 Forward: 5′-CCACGCTGCTGAACATGCT-3′
Ribosomal Protein, Large, P0Reverse:5′-TCGAACACCTGCTGGATGAC-3′
Table 1. Reference Genes Evaluated
Vol. 43 ı No. 5 ı 2007 www.biotechniques.com ı BioTechniques ı 643
Short Technical Reports
Nine commonly used reference
genes (3,4) were evaluated (Table 1).
For all reference genes and prognostic
tumor markers, TaqMan primers
and probes were custom designed to
produce products <100 base pairs in
size. To select the best reference gene,
total RNA samples from 10 frozen
renal tumors representing both high-
and low-grade cancers and 10 frozen
normal kidneys were evaluated to
determine the most stably expressed
genes (Figure 2A).
Multiple reference genes can be
used to minimize normalization bias
resulting from variation in expression
of reference genes between tissues.
The use of multiple reference genes is
critical when evaluating FFPE tissue
where RNA degradation produces
additional variations in the measured
expression of reference genes. The
number of reference genes required to
calculate a normalization factor was
determined (Figure 2B). The optimal
reference genes for expression
analysis using FFPE renal tumors
were ACTB, RPL13A, GUS, RPLP0,
HPRT1, and SDHA.
To determine the optimal number
of reference genes for expression
analysis using frozen tumors, RT-PCR
was performed for the nine reference
genes using total RNA extracted from
10 frozen renal tumors. The optimal
number of reference genes was three
when using a V of 0.15 as the cutoff.
This cutoff is arbitrary and has been
suggested by the authors of the
geNorm software. A V approaching
0.15 should also produce accurate
expression values. In this study, the
change in V in going from two to
three reference genes was 0.0169,
suggesting that use of two reference
genes may be adequate when using
To validate the reference genes
for use in FFPE tissue, expression
values for 16 prognostic genes, previ-
ously identified in a microarray-based
profiling study (10), were compared
when measured using FFPE and frozen
RCC (Figure 3). The correlation for
the mean gene expressions measured
in matching FFPE and frozen tumors
was 0.921. Cronin et al. performed a
similar experiment comparing gene
expressions from paired archival and
frozen breast tumors and reported a
statistically significant correlation
coefficient of 0.91 (11). This study
demonstrates that RT-PCR expression
profiling is feasible using archival
renal tumors and produces expression
values for genes of interest that
correlate closely with expression
measured using frozen tissue.
Each of the prognostic markers
was then validated for use with FFPE
RCC. Expression of the 16 markers
(10) was determined in 23 matching
frozen and FFPE renal tumors, repre-
senting a range of tumor grades and
stages (Table 2). All markers predicted
survival when frozen tumors were used
and 14:16 markers predicted survival
when FFPE tumors were used as the
source of RNA (Table 3). Therefore,
although all markers predicted clinical
outcome in our patients, 14 markers
have been validated for use in future
studies with archival renal tumors as
the source of RNA.
Archival FFPE renal tumors can be
utilized for expression profiling by RT-
PCR. ACTB, RPL13A, GUS, RPLP0,
HPRT1, and SDHA are identified
as the six most stably expressed
reference genes that should be used
for calculating the normalization
factor. Even when frozen renal tumors
are used for expression profiling,
the use of multiple reference genes
improves accuracy. The expression
of most prognostic markers can be
evaluated in FFPE tumors; however,
candidate markers of prognosis should
be validated for use with FFPE tissue
prior to embarking on large-scale
expression profiling studies.
This work has been supported in part
by the Cancer and Leukemia Group B
Foundation (CALGB/Novartis Clinical
Scholar Award) and the National Cancer
Institute (no. 1R21CA121212-01).
The authors declare no competing
1. Chawla, S.N., P.L. Crispen, A.L. Hanlon,
R.E. Greenberg, D.Y. Chen, and R.G. Uzzo.
2006. The natural history of observed enhanc-
ing renal masses: meta-analysis and review of
the world literature. J. Urol. 175:425-431.
2. Zisman, A., A.J. Pantuck, J. Wieder, D.H.
Chao, F. Dorey, J.W. Said, J.B. deKernion,
R.A. Figlin, et al. 2002. Risk group assess-
ment and clinical outcome algorithm to
predict the natural history of patients with
surgically resected renal cell carcinoma. J.
Clin. Oncol. 20:4559-4566.
3. Thellin, O., W. Zorzi, B. Lakaye, B. De
Borman, B. Coumans, G. Hennen, T.
Grisar, A. Igout, et al. 1999. Housekeeping
genes as internal standards: use and limits. J.
Mahadevappa, and M. Tsyganskaya. 2000.
Comparison of human adult and fetal expres-
P Value (Logrank Test)
Table 3. Markers Predicting
Survival in Matching Frozen
and Archival Renal Tumors
Table 2. Demographics
Vol. 43 ı No. 5 ı 2007 www.biotechniques.com ı BioTechniques ı 647
Short Technical Reports
sion and identification of 535 housekeep-
ing/maintenance genes. Physiol. Genomics
5. Vandesompele, J., K. De Preter, F. Pattyn,
B. Poppe, N. Van Roy, A. De Paepe and F.
Speleman. 2002. Accurate normalization of
real-time quantitative RT-PCR data by geo-
metric averaging of multiple internal control
genes. Genome Biol. 3:RESEARCH0034.
6. Szabo, A., C.M. Perou, M. Karaca, L.
Perreard, J.F. Quackenbush, and P.S.
Bernard. 2004. Statistical modeling for select-
ing housekeeper genes. Genome Biol. 5:R59.
7. Paik, S., S. Shak, G. Tang, C. Kim, J.
Baker, M. Cronin, F.L. Baehner, M.G.
Walker, et al. 2004. A multigene assay to
predict recurrence of tamoxifen-treated,
node-negative breast cancer. N. Engl. J. Med.
8. Suzuki, T., P.J. Higgins, and D.R.
Crawford. 2000. Control selection for RNA
quantitation. BioTechniques 29:332-337.
9. Spanakis, E. 1993. Problems related to the
interpretation of autoradiographic data on
gene expression using common constitutive
transcripts as controls. Nucleic Acids Res.
10. Zhao, H., B. Ljungberg, K. Grankvist, T.
Rasmuson, R. Tibshirani, and J.D. Brooks.
2006. Gene expression profiling predicts sur-
vival in conventional renal cell carcinoma.
PLoS Med. 3:e13.
11. Cronin, M., M. Pho, D. Dutta, J.C. Stephans,
S. Shak, M.C. Kiefer, J.M. Esteban, and J.B.
Baker. 2004. Measurement of gene expression
in archival paraffin-embedded tissues: devel-
opment and performance of a 92-gene reverse
transcriptase-polymerase chain reaction assay.
Am. J. Pathol. 164:35-42.
Received 6 June 2007; accepted
21 July 2007.
Address correspondence to Hyung Kim,
Roswell Park Cancer Institute, Elm and
Carlton Streets, Buffalo, NY 14263. e-mail:
To purchase reprints of this article, contact: