Clinical Cancer Research Vol. 6, 2381-2392, June 2000
© 2000 American Association for Cancer Research
Molecular Oncology, Markers, Clinical Correlates |
Noninvasive Diagnosis of Bladder Carcinoma by Enzyme-linked Immunosorbent Assay Detection of CD44 Isoforms in Exfoliated Urothelia
Anthony C. Woodman,
Steve Goodison,
Marcus Drake,
Jeremy Noble and
David Tarin1
Cranfield Biomedical Centre, Institute of Bioscience and Technology, Cranfield University, MK43 0AL Bedfordshire, United Kingdom [A. C. W.]; University of California, San Diego Cancer Center and Department of Pathology, La Jolla, California 92093-0658 [S. G., D. T.]; and Department of Urology, Oxford Churchill Hospital, OX3 7LJ Oxford, United Kingdom [M. D., J. N.]
 |
ABSTRACT
|
|---|
The expression of
variant isoforms of the adhesion molecule CD44 is correlated with the
onset of neoplasia in many carcinomas. We have previously shown that
noninvasive detection of bladder carcinoma is possible by analysis of
anomalous CD44 expression in exfoliated urothelia. Although the
sensitivity and specificity values obtained for the detection of
bladder tumors using RT-PCR and Western blotting methods were superior
to those obtained using urine cytology, the application of such
techniques is inconvenient for routine diagnostic use. We now report
the design and development of a sandwich-ELISA system for the reliable
detection of CD44 protein extracted from sedimented urothelial cells in
voided urine. Naturally micturated urine samples were obtained from 53
patients with newly diagnosed bladder cancer and from 65 subjects with
no evidence of disease; patients with gross hematuria were excluded
because of interference with the assay. To demonstrate the diagnostic
potential of the system, a "gate" was imposed at N (max),
i.e., the highest absorbance value obtained from a
sample known to be tumor free. All values above this value were assumed
to be indicative of the presence of a tumor. Using this parameter, 42
of 53 (81.1%) patients with histologically confirmed bladder tumors
were correctly diagnosed. Correspondingly, under these conditions, the
assay is 100% specific for tumor detection, with a sensitivity of
81.1%, which equates to a positive predictive value of 100% and a
negative predictive value of 81.1%. A further 54 patients who had
previously received treatment for bladder cancer but were currently
clinically disease-free were also investigated. Of these, 47 of 54
(87%) were correctly diagnosed to be tumor-free, which in this group
equates to a positive predictive value of 87% and a negative
predictive value of 100%. The data presented demonstrate that the
rapid and accurate detection of elevated levels of CD44 protein
isoforms in exfoliated urothelial cells is applicable to the
identification and monitoring of primary and recurrent bladder cancer.
|
INTRODUCTION
|
|---|
Bladder carcinoma has a high prevalence in many industrialized
countries, which is in part a consequence of the disease having a
strong association with cigarette smoking and a number of occupations,
e.g., the rubber industry and the use of organic solvents.
Bladder cancer rates are stable at around 17 per 100,000, and it is
estimated that there will be >50,000 new cases in 1998 in the United
States.2
Approximately 60% of superficial bladder tumors recur after initial
resection, and 20% of these will progress to invasive malignancy
(1)
. Treatments using chemotherapy and/or surgery are
generally successful, particularly when the tumors are detected before
invasion through the basement membrane of the bladder wall. However,
because of the inaccessibility of the bladder to unaided visual
examination, internal investigation is only considered when macroscopic
hematuria or other symptoms occur. Although cystoscopy, contrast
urography, and ultrasound are to date the most powerful methods for the
diagnosis and monitoring of bladder tumors, they are uncomfortable and
labor-intensive procedures. Even with such invasive procedures, it is
sometimes difficult to reach a definitive diagnosis, especially after
the resection of an earlier neoplasm, either because the lesion is too
small to find or is inaccessible. Thus the development of simpler,
preferably noninvasive methods for the detection of urothelial
malignancy are urgently required. Furthermore, because bladder lesions
have a strong tendency to recur, the monitoring of asymptomatic
patients for recurrence after treatment is particularly important.
Urine cytology, although simple and highly specific, has insufficient
sensitivity to be routinely useful, particularly in the diagnosis of
well-differentiated, early-stage neoplasms, which are the most amenable
to successful treatment (2)
. Recently identified molecular
abnormalities that occur in neoplasia offer new opportunities for the
early diagnosis of bladder cancer, and the analysis of the expression
of the transmembrane glycoprotein CD44 may be useful in this context.
Occupying 60-kb of the short-arm of chromosome 11, the CD44
gene (Fig. 1)
contains at least 20 exons,
of which 10 are constantly expressed in almost all tissues (CD44s),
with the inclusion of the remaining exons (CD44v) being subject to
alternative splicing. This processing of CD44 transcripts can
theoretically lead to the production of several hundred protein
isoforms, all of which are subject to extensive posttranslational
modifications, particularly glycosylation, yielding isoforms with
apparent molecular weights of between
Mr 85,000 and
Mr 250,000. Having been initially
characterized as the lymphocyte homing receptor (3
, 4)
,
the CD44 family of proteins is now believed to play a role in many
cellular activities, including T-lymphocyte activation, cellular
adhesion, embryonic development, and hyaluron metabolism
(5, 6, 7)
. However, much interest has recently been directed
to the role of CD44 in malignancy. Animal studies with transfected cell
lines have implicated a role for one of the splice variants in
experimental tumor metastasis (8, 9, 10)
, whereas
RT-PCR3
based
studies of human tissues have demonstrated aberrant CD44
gene expression in many human tumors (11, 12, 13, 14, 15)
, indicating
that it could be a useful diagnostic marker. Subsequently there have
been many publications describing abnormalities of CD44 transcription
and translation in a variety of types of malignant disease (for recent
reviews, see Refs. 6
, 16
, 17
,
and 18
).
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 1. Schematic map of CD44 gene
structure. Arrows, gene sequences encoding for antibody
binding domains.
|
|
Previously, we have presented data describing the noninvasive detection
of bladder carcinoma by detection of anomalous CD44 mRNA and proteins
by RT-PCR/Southern hybridization and Western blotting, respectively
(13
, 19)
, in exfoliated urothelial cells collected from
naturally micturated urine. Although the sensitivity and specificity
values obtained for the detection of bladder tumors were superior in
those studies to those obtained using microscopic urine cytology, the
application of the techniques used is presently not feasible for
routine diagnostic use.
Immunohistochemistry and ELISA are techniques more suitable for the
high-throughput, routine assessment of diagnostic molecular markers.
The immunohistochemical assessment of CD44 protein isoforms in solid
bladder tissue is informative (19, 20, 21)
and has revealed
useful information concerning the expression of the CD44
gene during tumor progression in the bladder (22)
.
Furthermore, immunocytochemical analysis of CD44 proteins on exfoliated
urothelial cells has been shown to be a useful adjunct to cytology
(19)
. However, the application of immunohistochemistry to
the rapid, large volume analysis of exfoliated cells is inconvenient
and labor-intensive in interpretation. The assay of choice for such
purposes is an ELISA-based system, where high throughput is coupled
with a nonsubjective digital result. ELISA systems have been described
for the detection of circulating soluble CD44 in the serum of patients
with colon (23)
, breast (24
, 25)
, and ovarian
cancers (26
, 27)
, but the technology has not been applied
to the measurement of CD44 proteins in exfoliated cells. We report the
design and development of a sandwich-ELISA system for the detection of
CD44 protein isoforms in exfoliated urothelial cells and discuss its
application for the noninvasive detection of bladder cancer.
|
MATERIALS AND METHODS
|
|---|
CD44 Gene Nomenclature
The following nomenclature, as described previously (6
, 17
, 21) , is used for description of CD44 gene exon
arrangement. Alternatively spliced variant exons 6 to 15 (v1–10) can
be included in the transcript by alternative splicing between standard
exons 5 (s5) and 16 (s6).
Preparation of Protein Lysates
Cell Lines.
RT112 (human bladder carcinoma cell line) and HT29 (human colon
carcinoma cell line) were grown to 75% confluence in RPMI 1640 medium
(Sigma, Poole, United Kingdom) containing 10% fetal bovine serum
(Sigma) and 20 mM HEPES at 37°C. The cells were harvested
using a flexible cell scraper, pelleted by centrifugation (600 x
g, 4°C, 5 min) and washed with ice-cold wash buffer [PBS
containing a protease inhibitor cocktail of aminoethylbenzenesulfonyl
fluoride, 0.4 mg/ml EDTA-Na (0.5 mg/ml), leupeptin (0.5 µg/ml), and
pepstatin (0.5 µg/ml) supplied by ICN/Flow (Thame, United Kingdom)].
The cell pellet obtained by centrifugation (600 x g,
4°C, 5 min) was resuspended 1:1 (w/v) in lysis buffer [20
mM tris (pH 8.0), 150 mM
NaCl, 20 mM CHAPS, and protease inhibitor
cocktail], snap-frozen in liquid nitrogen, and stored at -80°C
until required.
Exfoliated Urothelial Cells.
Naturally voided urine specimens were collected from urology
out-patient clinics at the Oxford Churchill Hospital and processed as
described previously (19)
. Briefly, specimens of 
50 ml
were collected from 53 patients with newly diagnosed bladder cancer, 54
patients who had previously received treatment for bladder cancer but
were presently disease free as assessed by cystoscopy, and from 65
persons with no evidence of disease. (Details of age, sex and clinical
status for each of these three groups is summarized within the results
presented in Table 2
). Additional specimens were also investigated from
eight patients with benign prostatic hyperplasia and from six with
confirmed carcinoma of the prostate. To assess the effect of hematuria
on the assay, 10 patients with macroscopic bleeding and 7 patients with
cystic stents in situ, prone to microscopic hematuria, were
also assessed. The urine was collected into vessels containing protease
inhibitor cocktail at a final concentration as described above and kept
on ice during transport to the laboratory. By assaying the inability of
urine cell lysate preparations to cleave resofurin-labeled casein, we
have previously documented that complete inhibition of protease
activity in the urine is obtained using this protocol
(28)
. The exfoliated urothelial cells were subsequently
pelleted, washed, and lysed as described above for cell lines.
View this table:
[in this window]
[in a new window]
|
Table 2 Results of CD44 ELISA analysis of urine samples
from bladder cancer patients either currently presenting with a tumor
or in remission compared with control volunteers
|
|
Before the assay, all lysates were thawed and kept on ice for 1 h
to ensure complete lysis, and insoluble debris was removed by
centrifugation (15,000 x g, 4°C, 30 min). The
concentration of total protein was determined (Bio-Rad protein assay
kit) and adjusted to 500 µg/ml by dilution with wash buffer for
analysis by ELISA. For Western blot analysis, aliquots were adjusted to
1 mg/ml and mixed 1:1 (v/v) with SDS-PAGE gel loading buffer and boiled
for 5 min.
|
Labeling of Capture and Detection Antibodies
|
|---|
mAbs F-10–44-2 [recognizing protein domain encoded by exon 1
(s1)] and VFF-18 [recognizing protein domain encoded by exon 11 (v6);
both at 1 mg/ml PBS] were biotinylated with Biotin-NHS (Boehringer
Mannheim) according to the manufacturer’s protocol. The labeled
antibodies were column-purified (Sephadex G50), and concentration was
determined by absorbance at 280 nm. Stock aliquots were diluted to 20
µg/ml PBS and stored at -20°C.
FITC-labeling of mAb Hermes-3 (1 mg/ml; recognizing protein domain
encoded by exon 5) was performed using a FluroTag FITC-labeling kit
(Sigma) according to the manufacturer’s instruction. The protein
concentration of the labeled antibody was determined by absorbance at
280 nm, and the FITC labeling ratio was determined by measuring the
absorbance at 490 nm. Stock aliquots of 1 mg/ml were kept at 4°C.
|
Western Blotting
|
|---|
Samples of exfoliated urothelial cell lysates and tumor cell
lines, prepared as described above, were subjected to 6% SDS-PAGE
under nonreducing conditions. The separated proteins were
electroblotted (0.8 mA/cm2; 1 h) onto an
Immobilon-P membrane (Millipore) using Tris-Glycine transfer buffer
[48 mM Tris, 39 mM glycine, 0.1% SDS, 20%
methanol (pH 9.2)]. Nonspecific reactions were blocked with TBS
containing 5% skimmed milk before the membrane was incubated with mAb
Hermes-3 at 4°C overnight and then with peroxidase-conjugated
antimouse IgG (Sigma; 1/1000 dilution) for 1 h at room
temperature. All antibodies were diluted in 5% skimmed milk in TBS,
and after each incubation, the membrane was washed with TBS containing
0.1% Tween 20. Signals were detected by enhanced chemiluminescence
using an enhanced chemiluminescence detection kit (Amersham).
|
Assay Optimization Using Cultured Human Tumor Cell Lysates
|
|---|
To optimize the assay conditions, CD44 protein isoforms were
analyzed in cell lysates from two human tumor cell lines. We have
previously described the presence of CD44 standard and variant protein
isoforms in human bladder (RT112) and colon carcinoma (HT-29) cell
lines (19, 20, 21)
using Western blotting and
immunohistochemistry. Analysis of these cell lines was used to define
the conditions for sample preparation, sample concentration, and
composition of sample diluent buffer. Negative controls used for all
ELISA analyses were: (a) no capture antibody; (b)
no detection antibody; (c) no antigen; and (d)
BSA as nonspecific protein in place of antigen at appropriate
concentrations. All data were corrected by subtraction of negative
control values.
Sample Preparation.
Incomplete solubilization of membrane proteins limits the availability
of CD44 protein isoforms in the resulting lysate, so incorporation of
detergents in sample buffers is advisable to maximize solubilization.
However, the choice of detergent is dependent on several factors,
including the intended assay system and the interference of detergents
on antibody epitope binding. To optimize the incorporation of
detergents in the sample lysis buffer, a comparative assay was
undertaken on lysates prepared by cell disruption by passage through a
21-gauge needle (physical shearing) or with hypotonic buffer,
containing either CHAPS, Tween 20 (1% v/v), or NP40 (1% v/v). HT29
cells were harvested as described and split into four aliquots
containing an equal number of cells. The differing efficiency of
protein solubilization was clearly observed on measuring total protein
within each of the clarified lysates. Lysis buffers containing either
CHAPS or NP40 released almost 25 mg/ml total protein, whereas with
Tween 20, this was reduced to 15 mg/ml. Extraction using detergent
buffers but without shearing resulted in protein yields of 3 mg/ml.
Analysis of HT29 (100 µg/ml total protein) cell lysates by ELISA
demonstrated that similar results could be obtained after extraction of
protein using physical disruption or the incorporation of CHAPS in the
lysis buffer. However, if NP40 or Tween 20 were used in the lysis
buffer, no CD44 could be detected by the ELISA (data not shown). Thus,
because CHAPS lysis buffer appeared to have no detrimental effects upon
the ELISA, yet was superior to physical disruption in releasing total
protein, this buffer was used in all further studies.
Concentration of Sample.
Before the assessment of CD44 in exfoliated urothelial cells, the
optimum sample concentration was determined by constructing titration
curves for the detection of CD44 in the RT112 and HT29 cell lines.
These studies demonstrated that the optimal sample concentration should
be between 100 and 1000 µg/ml total protein (Fig. 2)
.
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 2. Optimization of ELISA. Determination of optimal
sample concentration for the evaluation of CD44 protein isoforms.
Samples were extracted using CHAPS lysis buffer as described in
"Materials and Methods." Results are mean ± SE of
quadruplicate wells from three separate assays.
|
|
Concentration of Blocking Protein in ELISA Diluent Buffer.
Initial studies were conducted in ELISA blocking buffer (Boehringer
Mannheim) reconstituted according to the manufacturer’s instruction,
producing a buffer containing 1% (w/v) protein. When the data obtained
for both RT112 and HT29 was compared with the intensity of signal
observed in Western blotting (data not shown), the absorbance values in
the ELISA were lower than may have been anticipated, particularly with
HT29. A reduction in blocking protein to 0.5% (w/v) resulted in an
increase in assay sensitivity without an increase in background signal.
Specifically, when diluted with buffer (1% protein) a maximum
absorbance (± SE) of 0.68 (± 0.03) was obtained; however, with
a reduction to 0.5% protein, the maximum absorbance was 1.53 (±0.12).
The absorbance values for the negative controls were 0.187 (± 0.01)
and 0.199 (± 0.02), respectively.
|
Optimization of Multiple-Capture Antibody ELISA
|
|---|
Triplicate (10 µl) RT112 protein extracts (0–1000 µg/ml) were
assessed in an ELISA using either biotinylated F-10–44-2 (6 µg/ml)
as the sole capture antibody, or at a concentration of 3 µg/ml in
combination with 3 µg/ml biotinylated mAb 23.6.1 (epitope encoded by
exon 7; Ref. 19
), or with 3 µg/ml biotinylated VFF-18
(epitope encoded by exon 11; Bender Medsystem, Vienna, Austria). A
fourth ELISA was performed using all three antibodies at 2 µg/ml;
hence, in each ELISA, the total concentration of capture antibodies was
6 µg/ml.
|
Optimum ELISA Protocol for Detection of Urothelial Neoplasia
|
|---|
One hundred µl of ELISA blocking buffer (0.5% protein)
containing biotinylated F-10–44-2 (2 µg/ml) and biotinylated VFF-18
(2 µg/ml) were bound to streptavidin-coated microtiter plates (Pierce
and Warriner, Chester, United Kingdom) for 30 min at room temperature,
with shaking at 250 rpm. All exfoliated cell lysates (extracted with
CHAPS lysis buffer) were diluted to a total protein concentration of
500 µg/ml, and 10-µl samples were assayed in triplicate by
incubation with shaking (250rpm) for 1 h at room temperature. A
single preparation of cell line RT112 cell lysate (100 µg/ml) was
used as a positive control for all assays, allowing interassay
comparison. Detection of the captured proteins was achieved by
incubation with FITC-labeled Hermes-3 (1 µg/ml in blocking buffer)
for 1 h followed by peroxidase-labeled mouse anti-FITC antibody
(0.15 units in ELISA blocking buffer) for a further 30 min at room
temperature with shaking (250 rpm). Plates were washed between each
incubation step with four changes of wash buffer (PBS containing 0.05%
Tween 20).
|
RESULTS
|
|---|
ELISA Design.
Previous studies on bladder cancer (19
, 21
, 22)
have
indicated which epitopes are most useful for the detection of CD44
protein isoforms in neoplastic urothelia. On this basis, combinations
of mAbs recognizing epitopes encoded by exons 1 and 5 of the standard
region and exons 7 and 11 of the variant region were tested in a
sandwich-ELISA design for accurate CD44 detection by comparison with
Western blot profiles from the same samples.
Comparison of ELISA with Western Blot Profiles.
To validate the reliability of ELISA results, data obtained by this
technique were compared with levels of CD44 overexpression visualized
by Western blotting. Naturally voided exfoliated urothelial cells were
obtained from 16 patients with no malignant disease and from 17
patients with histologically confirmed bladder cancer, and cellular
protein was extracted as described above. Western blots were performed
with the Hermes-3 (exon 5 epitope) antibody as described previously
(19)
. Sandwich ELISA was conducted using the F10.44.2
(exon 1 epitope) as capture antibody and Hermes-3 as the detection
antibody. The protein concentration in each assay was 500 µg/ml.
The Western blot profile of these samples was as described
previously (19)
. Both nonmalignant and malignant samples
expressed standard form CD44, as demonstrated by the presence of a
Mr 85,000 isoform. Malignant
samples also exhibited multiple protein isoforms in the range of
Mr 150,000–200,000, a pattern
indicating the overproduction of variant CD44 isoforms. Samples from
normal counterparts did not exhibit this pattern.
Statistical analysis of the data obtained using this
sandwich-ELISA protocol did reveal significant difference
(P < 0.005) between the patient groups (data not
shown). However, comparison with the Western blot profiles revealed
that several samples exhibiting a characteristic range of high
molecular weight CD44 isoforms (19)
presented low
absorbance values in the ELISA. It appeared that this ELISA design was
unable to detect a significant proportion of the samples, which by
clinical examination/histology and Western blot analysis, displayed
malignant features. A possible explanation for this discrepancy was the
presence of large variant isoforms in tumor samples not being
effectively captured onto the surface of the microtiter plate. To
examine this possibility, we investigated the efficacy of using a
multiple-capture antibody modification of the ELISA.
Multiple-Capture Antibody Protocol.
We reasoned that the binding of large CD44 isoforms by multiple-capture
antibodies should reduce the "shearing" of the molecule from the
plate and provide a more reliable measure of the analyte. Accordingly,
additional capture antibodies were selected for testing on the basis of
our previous studies showing that elevated expression of CD44 exon 7
(v2) and exon 11 (v6) occurs in many bladder tumors (19
, 21 , 22)
. The epitope recognized by F-10–44-2 is at the
NH2 terminal end of CD44: therefore, the
incorporation of capture antibodies recognizing these variant epitopes
would give additional anchorage by binding at a carboxy-proximal
position (Fig. 1)
.
To investigate whether the use of a multiple antibody capture protocol
could increase reliability of identification of tumor cells, ELISAs
containing one to three capture antibodies were performed, as described
in "Materials and Methods." All other components of the ELISA were
unchanged.
Fig. 3
presents the titration curve for
the detection of CD44 isoforms under various assay conditions using
total protein extracted from the RT112 cell line. The sensitivity of
the assay was modulated by altering the combination of capture
antibodies used. Maximum sensitivity was achieved using F-10–44-2 (3
µg/ml) + VFF-18 (2 µg/ml) as capture antibodies, manifest by a
shift of the titration curve to the right compared with that obtained
using F-10–44-2 alone. The combination of using F-10–44-2 with mAb
23.6.1 was less sensitive than with VFF-18. When a sample total protein
concentration of 200 µg/ml was assayed, an absorbance of 1.00 was
measured, whereas the F10–44-2/VFF-18 capture antibody combination
gave an absorbance of 2.4 (Fig. 3)
.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 3. Optimization of ELISA. Evaluation of the
efficacy of various combinations of capture antibodies in detecting
CD44 protein isoforms extracted from the RT112 bladder tumor cell line.
Samples were extracted using CHAPS lysis buffer as described in
"Materials and Methods." Results are mean ± SE of
quadruplicate wells from three separate assays.
|
|
When the malignant and nonmalignant exfoliated urothelial cell samples
were reassessed using the F-10–44-2/VFF-18 capture-Hermes-3 detection
ELISA, the malignant samples gave higher ELISA values in the new assay
and compared more closely with the Western blot profile of the same
sample. There was no corresponding increase in the ELISA values of the
nonmalignant samples with the improved multiple capture antibody assay,
and background values remained comparably low.
Intra- and Interassay Variation of the Multiple-Capture ELISA.
Before a detailed assessment of the ability of the multiple-capture
ELISA to discriminate between malignant and nonmalignant bladder
disease, the intra- and interassay variability was determined using the
RT112 cell line.
Ten (10 µl) RT112 protein extracts (100 µg/ml) were assessed in
five separate assays. The results are summarized in Table 1
. These results show that the assay is
reliable and reproducible with a mean intra-assay variability of 2.36%
and a mean interassay variability of 4.97%.
Detection of CD44 Standard and Variant Isoforms in Exfoliated
Urothelial Cells.
Using the F-10–44-2/VFF-18/Hermes-3 multiple-capture ELISA, the
presence of CD44s and CD44v isoforms was determined in 53 malignant and
65 nonmalignant samples of exfoliated urothelial cells. Additionally,
samples were assayed from 54 patients who had previously had a bladder
tumor, but were presently clinically believed to be clear of any
malignant disease. Tables 2
and 3
along with Fig. 4
summarize the ELISA values obtained for
each sample (mean of triplicates after subtraction of negative control
values). The mean absorbance (±SE) value for samples from patients
with confirmed tumors was 0.566 ± 0.035 compared with 0.270 ± 0.01 in the group with no malignant disease. In the group that had
previously presented with bladder tumors but was presently
disease-free, the mean absorbance was 0.390 ± 0.018.
View this table:
[in this window]
[in a new window]
|
Table 3 Summary data and statistical assessment of the
evaluation of CD44 protein expression in exfoliated urothelial cells
using the multiple-capture ELISA protocol
|
|
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 4. Evaluation of CD44 protein expression in
exfoliated urothelial cells using a sandwich ELISA. Detection of CD44
protein isoforms in exfoliated urothelial cells from patients with
histologically confirmed bladder tumors or currently in remission and
clinically adjudged to have no urological malignancy and compared with
samples from healthy volunteers with no known malignancy. A limited
number of samples were also evaluated from patients with benign
prostatic hyperplasia, prostatic cancer, frank hematuria, and with a
urinary stent in situ. Capture antibodies, F-10–44-2 (3
µg/ml) + VFF-18 (3 µg/ml); detection antibody: Hermes-3 (1
µg/ml); sample, 100 µg/ml total protein. Results are mean ±
SE of triplicate samples measured on two separate occasions. *,
significantly elevated mean compared with normal group (one-tailed
t test on normalized data; P <
0.005).
|
|
Further analysis of these data using a one-tailed Student’s
t test (data normalized by Box-Cox root transformation)
showed that the ELISA values for the samples obtained from patients
with confirmed bladder cancer were significantly greater than the
values obtained with either nonmalignant samples (P <
0.005). Although the mean absorbance value obtained from patients who
were presently in remission was significantly lower than those with a
present tumor (P < 0.005), it was, however,
significantly higher than that of the normal control group
(P < 0.005).
To investigate the diagnostic potential of these data, a "gate" was
imposed at N (max), i.e., the highest absorbance value
obtained from a sample known to be tumor-free. All values above this
value were then assumed to be indicative of a tumor, whereas all those
below were taken to be tumor-free. Using this parameter, the assay is
100% specific for first-time tumor detection, with a sensitivity of
81.1%. Presented in a more conventional form, the positive predictive
value for this assay was 100%, and the negative predictive value was
81.1%.
The values for specificity and sensitivity obtained by ELISA are
similar to those we obtained previously using Western blot analysis
alone (19)
. To assess the correlation of diagnosis with
the two assays, all samples were analyzed by Western blotting and the
F-10–44-2/VFF-18/Hermes-3 ELISA. Fig. 5
is a representative sample of the comparative data for six normal and
six malignant samples. The diagnostic "end point" for the Western
blot analysis was the presence or absence of isoforms with an apparent
molecular weight of >Mr 150,000
(intensity of bands was not determined), whereas for the ELISA, the
"end point" was a quantitative measure as described above. When the
same urothelial cell samples were analyzed by both Western blotting and
the F-10–44-2/VFF-18/Hermes-3 ELISA, the larger the number, size, and
range of CD44 isoforms detected by Western blotting, the higher the
absorbance values obtained with the ELISA. Significantly, all samples
deemed to be positive by Western blot were also positive by ELISA,
whereas those that were negative in the one system were also negative
in the other.
View larger version (78K):
[in this window]
[in a new window]
|
Fig. 5. Comparative analysis of normal and malignant
exfoliated urothelial cells by Western blot and multiple-capture ELISA.
All samples were analyzed by Western blot using the Hermes-3 antibody
and by sandwich ELISA in triplicate on two separate occasions. The blot
and mean ELISA values obtained from six tumor and six normal samples
are presented. For both assays, the sample concentration = 100
µg/ml total protein. Arrows, the position of protein
size markers (kDa).
|
|
During the course of this study, a limited number of samples from
patients with benign prostatic hyperplasia or carcinoma of the prostate
were also assessed, but no striking evidence of elevated CD44 was
observed in the urine from patients with either condition (Table 3)
.
Effect of Hematuria on the CD44 ELISA.
Frank hematuria interferes with the detection of CD44 isoforms
(13
, 19) and Fig. 4
. This does not reduce the clinical
value of assays based upon this molecule because patients presenting
with hematuria would routinely be investigated by cystoscopy. However,
we considered it prudent to examine whether microscopic hematuria could
interfere with the assay. Samples of urine (50 ml) from a healthy
volunteer with no urological disease were "spiked" with 0–2000
µl of whole peripheral blood, and the presence of CD44s and CD44v was
determined using the ELISA described above. It was found that the
addition of up to 750 µl of whole blood to 50 ml of urine had no
significant effect on the absorbance values obtained (Table 4)
. After the addition of 750 µl of
blood, the resultant lysate has a light pink color indicative of macro-
as opposed to microscopic hematuria. With up to 1250 µl of blood, an
increase in absorbance was observed. Above 1250 µl, the absorbance
values declined to below control values obtained with normal exfoliated
urothelial cells alone. This phenomena may be related to total cell
numbers in the assayed sample. Spiking the 50-ml sample with 1–1.25 ml
of whole blood may increase the absorbance of the sample because of
CD44 release and/or absorbent molecules such as haem. Beyond that, the
increased cell number in the centrifuged pellet may either interfere
with efficient protein extraction, or high concentrations of released
factors may directly inhibit the binding of soluble ligand in the
assay.
View this table:
[in this window]
[in a new window]
|
Table 4 Effect of whole blood contamination on
performance of the multiple capture ELISA
Equal numbers of RT112 bladder tumor cells were "spiked" with
volume of whole blood as indicated before protein extraction. Samples
were then handled and assayed as indicated in the text. Sample, 100
µg/ml total protein. Results were analyzed using a one-sided
t test to determine whether "spiking" significantly
elevated the absorbance values as compared with nonspiked controls.
|
|
|
DISCUSSION
|
|---|
The data obtained in this investigation demonstrate that the
measurement of CD44 isoforms in protein lysates of naturally exfoliated
cells in urine by ELISA can reliably detect the presence of bladder
cancer. The presence of hematuria can interfere with the assay, but
this does not limit the potential clinical value of the test because
direct cystoscopic investigation of patients presenting with this
symptom is imperative. However, the feasibility of noninvasive
diagnosis of bladder cancer by examination of naturally micturated
urine is attractive because it eliminates investigative discomfort,
thereby increasing the likelihood of patient compliance with diagnostic
procedures. The development of this ELISA was driven by such needs, and
now that a proof of principle has been demonstrated, more comprehensive
patient-based studies will be undertaken. Urinary CD44 levels in
samples collected from larger groups of urology patients will be
correlated with further clinical parameters now that this ELISA enables
high throughput analyses to be handled. Repeat sampling regimes and
sampling before and after surgical or therapeutic intervention will
also be focused on.
Several new markers for analysis of voided urine have recently been
approved for clinical use, including bladder tumor antigen (29
, 30)
, nuclear matrix protein 22 (NMP22; Refs. 31
and
32
), and fibrin/fibrinogen degradation products
(33)
. Such markers can help to detect clinically occult
bladder cancer and can increase the interval of cystoscopic evaluation.
However, although these assays are more accurate than urine cytology,
they do have problems of low sensitivity and specificity, especially in
detection of low-grade tumors (30
, 32
, 34)
. In view of the
continuing need for a reliable, noninvasive test for bladder cancer, a
number of alternative molecular markers are under evaluation. These
include the Lewis-X antigen (35)
, cytokeratin 20
(36
, 37)
, microsatellite analysis of urine DNA
(38)
, and measurement of telomerase activity (28
, 39
, 40)
. The suitability of these assays for routine diagnostic
practice awaits further investigation. Widespread use of molecular
assays could lead to reduction in the burden of cystoscopic
investigations for primary and recurrent disease. The introduction of
cheap reliable assays in routine practice would also lead to
improvement of early detection of the disease even in asymptomatic
individuals.
Previously, using immunohistochemistry, Western blotting, and RT-PCR
(13
, 19)
, we reported that inappropriate CD44 protein is
detectable in exfoliated urothelial cells. These studies clearly
demonstrated that the detection of abnormal CD44 gene
expression in exfoliated cells in naturally voided urine could be used
to identify patients with bladder cancer with a high level of accuracy
(13
, 19) . RT-PCR/Southern blot hybridization analyses had
a specificity of 91% and a sensitivity of 83%. Western blot assays
recorded values of 100% specificity and 75% sensitivity for the
detection of bladder tumors. However, both are technically demanding
and time-consuming with many steps that are inappropriate for use in
routine laboratories, including electrophoresis, hybridization
protocols, and autoradiography. In contrast, the ELISA method produces
an unequivocal digital result that can define normal ranges and
diagnostic thresholds.
There are a number of parameters that must be optimized for the
deployment of an ELISA. Our initial studies using mAb F-10–44-2 for
capture and mAb Hermes-3 for detection clearly showed that the
diversity of CD44 isoforms generated by alternative splicing
complicates the design of a consistently reliable assay. For this
reason, we devised the ELISA described above using Western blot
analysis as a visual index of the overall level of CD44 expression in a
given sample for purposes of standardization. This led to the
recognition that introduction of a second capture antibody would prove
useful in tailoring the assay to detect a large and complex molecule.
This improvement was interpreted to be attributable to the availability
of additional "anchorage" sites and to the overexpression of CD44
isoforms containing epitopes encoded by exon 11 in malignant as opposed
to normal exfoliated urothelial cells (19)
. Such
refinement of the assay resulted in a 66% increase in sensitivity as
determined by the absorbance value obtained with 200 µg/ml protein
extracted from the RT112 bladder tumor cell line, while retaining a low
intra- and interassay variation.
The F-10–44-2/VFF-18/Hermes-3 ELISA was found to be an accurate yet
simple tool for the noninvasive detection of bladder tumors. On a
sample of 53 patients with newly diagnosed bladder cancer and 65
persons with no evidence of malignant disease, this ELISA resulted in a
specificity of 100%, a sensitivity of 80.2%, and a positive
predictive value of 100% for the presence of bladder cancer. These
values surpass those reported for almost all molecular markers in
bladder tumor detection.
Although CD44 expression was found to be significantly lower in the
"remission" group compared with those with confirmed tumors, the
mean ELISA values for those patients who were at that time tumor-free
are significantly greater than in "normal" subjects. This raised
"basal" level of CD44 expression in such patients may be
attributable to "field-change" effects, which are known to occur
throughout the transitional urothelium in cancer patients. Although a
significant difference was observed between the mean ELISA values for
the "remission" and tumor-bearing groups, the test still accurately
discriminated between individuals with and without bladder cancer.
In this study, we observed that macroscopic hematuria, in the absence
of a bladder tumor, elevates the mean ELISA absorbance values. We
therefore investigated the influence of trace amounts of blood in urine
samples by "spiking" the urine with increasing amounts of whole
blood. It was found that hematuria had no effect on the assay until the
blood became visible and by definition macroscopic. In such cases,
immediate cystoscopy would be indicated regardless of the status of the
molecular assay. The use of stents to aid urinary flow is common in
cases of urinary obstruction, and assessment of seven patients with
stents in situ revealed that this procedure can lead to an
elevation in ELISA values in patients without tumor. We hypothesize
that the presence of indwelling stents causes increased desquamation of
bladder epithelium, resulting in elevated CD44 levels caused by a
raised turnover of basal cells. It should be noted that irritation
and/or inflammation per se, such as is common after bacillus
Calmette-Guérin therapy, did not significantly interfere with the
assay.
One of the primary target groups for a noninvasive test such as this
ELISA would be those with early recurrence. Routine, affordable,
asymptomatic screening would be a major advance, even if restricted to
high-risk individuals or those in known high-risk occupations.
In conclusion, our results suggest that the detection of elevated
levels of CD44 protein isoforms in naturally voided urine is indicative
of urological neoplasia. The development of this ELISA may go some way
toward the goal of making available a reliable, routine, and
noninvasive method of early bladder cancer detection. The values of
sensitivity and specificity obtained in this study are promising and
could be further improved by analysis of repeat samples. Additionally,
we believe that the combination of this CD44 ELISA with other molecular
assays and/or with cytology could greatly facilitate the noninvasive,
early diagnosis of bladder cancer and improve the detection of
carcinoma recurrence after treatment, thereby reducing the reliance on
cystoscopy.
|
FOOTNOTES
|
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 To whom requests for reprints should be
addressed, at University of California, San Diego Cancer Center,
9500 Gilman Drive, 0658, La Jolla, CA 92093-0658. Phone: (619)
822-1222; Fax: (619) 822-0207. 
2 Vital Statistics Of the United States. American
Cancer Society Facts and Figures,
http://www.cancer.org/statistics/cff98, 1997.
3 The abbreviations used are: RT-PCR,
reverse transcription-PCR; mAb, monoclonal antibody; TBS, Tris-buffered
saline; CHAPS,
3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate.
Received 10/27/99;
revised 3/ 6/00;
accepted 3/ 6/00.
|
REFERENCES
|
|---|
-
Torti F. M., Lum B. L., Aston D., MacKenzie N., Faysel M., Shortliffe L. D., Freiha F. Superficial bladder cancer: the primacy of grade in the development of invasive disease. J. Clin. Oncol., 5: 125-130, 1987.[Abstract]
-
Raab S. S., Slagel D. D., Jensen C. S., Teague M. W., Savell V. H., Ozkutlu D., Lenel J. C., Cohen M. B. Low-grade transitional cell carcinoma of the urinary bladder: application of select cytologic criteria to improve diagnostic accuracy. Mod. Pathol., 9: 225-232, 1996.[Medline]
-
Dalchau R., Kirkley J., Fabre J. W. Monoclonal antibody to a human brain-granulocyte-T lymphocyte antigen is probably homologous to the W3/13 antigen of the rat. Eur. J. Immunol., 10: 745-750, 1980.[Medline]
-
Jalkanen S., Jalkanen M., Bargatz R., Tammi M., Butcher E. C. Biochemical properties of glycoproteins involved in lymphocyte recognition of high venules in man. J. Immunol., 141: 1615-1623, 1987.[Abstract]
-
Lesley J., Hyman R., Kincade P. W. CD44, and its interaction with extracellular matrix. Adv. Immunol., 54: 271-335, 1993.[Medline]
-
Rudzki Z., Jothy S. CD44 and the adhesion of neoplastic cells. Mol. Pathol., 50: 57-71, 1997.[Abstract/Free Full Text]
-
Lesley J., Hyman R. CD44 structure, and Function. Front. Biosci., 3: D616-D630, 1998.
-
Gunthert U., Hofmann M., Rudy W., Reber S., Zoller U., Haussmann I., Matzku S., Wenzel A., Ponta H., Herrlich P. A new variant of glycoprotein CD44 confers metastatic potential to rat carcinoma cells. Cell, 65: 13-14, 1991.[CrossRef][Medline]
-
Rudy W., Hofmann M., Schwartz-Albiez R., Zoller M., Heider K. H., Ponta H., Herrlich P. The two major CD44 proteins expressed on a metastatic rat tumor cell line are derived from different splice variants: each one individually suffices to confer metastatic behavior. Cancer Res., 53: 1262-1268, 1993.[Abstract/Free Full Text]
-
Heider K-H., Hofmann M., Hors E., van der Berg F., Ponta H., Herrlich P., Pals S. T. A human homologue of the rat metastasis-associated variant of CD44 is expressed in colorectal carcinomas and adenomatous polyps. J. Cell Biol., 120: 227-233, 1993.[Abstract/Free Full Text]
-
Matsumura Y., Tarin D. Significance of CD44 gene products for cancer diagnosis and disease evaluation. Lancet, 340: 1053-1058, 1993.
-
Tanabe K. K., Ellis L. M., Saya H. Expression of Cd44R1 adhesion molecule in colon carcinomas and metastasis. Lancet, 341: 725-726, 1993.[CrossRef][Medline]
-
Matsumura Y., Hanbury D., Smith J., Tarin D. Non-invasive detection of malignancy by identification of unusual CD44 gene activity in exfoliated cancer cells. Br. Med. J., 308: 619-624, 1994.[Abstract/Free Full Text]
-
Dall P., Heider K-H., Hekele A., von Minckwitz G., Kaufman M., Ponta H., Herrlich P. Surface protein expression and messenger RNA-splicing analysis of CD44 in uterine cervical cancer and normal cervical epithelium. Cancer Res., 54: 3337-3341, 1994.[Abstract/Free Full Text]
-
Iida N., Bourguinon L. Y. W. New CD44 splice variants associated with human breast cancers. J. Cell. Physiol., 162: 127-133, 1995.[CrossRef][Medline]
-
Goodison S., Tarin D. Current status of CD44 variant isoforms as cancer diagnostic markers. Histopathology (Oxf.), 32: 1-6, 1998.[CrossRef][Medline]
-
Goodison, S., and Tarin, D. Clinical implications of anomalous CD44 gene expression in neoplasia. Front. Biosci., 3: E89–E109.
-
Naor D., Sionov R. V., Ish-Shalom D. CD44: structure, function and association with the malignant process. Adv. Cancer Res., 71: 241-319, 1997.[Medline]
-
Sugiyama M., Woodman A. C., Sugino T., Crowley S., Ho K., Smith H., Matsumura Y., Tarin D. Non-invasive detection of bladder cancer by identification of abnormal CD44 proteins in exfoliated cancer cells in urine. J. Clin. Pathol: Mol. Pathol., 48: M142-M147, 1995.
-
Borgya A., Woodman A. C., Sugiyama M., Donie F., Kopetski E., Matsumura Y., Tarin D. Isolation and characterisation of antibodies which specifically recognise the peptide encoded by exon 7 (v2) of the human CD44 gene. J. Clin. Pathol: Mol. Pathol., 48: M241-250, 1995.
-
Woodman A. C., Sugiyama M., Yoshida K., Sugino T., Borgya A., Goodison S., Matsumura Y., Tarin D. Analysis of anomalous CD44 gene expression in human breast, bladder, and colon cancer and correlation of observed mRNA and protein isoforms. Am. J. Pathol., 149: 1519-1530, 1996.[Abstract]
-
Sugino T., Gorham H., Yoshida K., Bolodeoku J., Nargund V., Cranston D., Goodison S., Tarin D. Progressive loss of CD44 gene expression in invasive bladder cancer. Am. J. Pathol., 149: 873-882, 1996.[Abstract]
-
Guo Y. J., Liu G., Wang X., Jin D., Wu M., Ma J., Sy M. S. Potential use of soluble CD44 in serum as indicator of tumor burden and metastasis in patients with gastric or colon cancer. Cancer Res., 54: 422-426, 1994.[Abstract/Free Full Text]
-
Martin S., Jansen F., Bokelmann J., Kolb H. Soluble CD44 splice variants in metastasizing human breast cancer. Int. J. Cancer, 74: 443-445, 1997.[CrossRef][Medline]
-
Lackner C., Moser R., Bauernhofer T., Wilders-Truschnig M., Saminigg H., Berghold A., Zatloukal K. Soluble CD44 v5 and v6 in serum of patients with breast cancer. Correlation with expression of CD44 v5 and v6 variants in primary tumors and location of distant metastasis. Breast Cancer Res. Treat., 47: 29-40, 1998.[CrossRef][Medline]
-
Darai E., Leblanc M., Walker-Combrouze F., Brinuier A. F., Madelenet P., Scoazec J. Y. Expression of cadherins and CD44 isoforms in ovarian endometrial cysts. Hum. Reprod. (Oxf.), 13: 1346-1352, 1998.[Abstract/Free Full Text]
-
Darai E., Bringuier A. F., Walker-Combrouze F., Feldmann G., Madelenet P., Scoazec J. Y. Soluble adhesion molecules in serum and cyst fluid from patients with cystic tumours of the ovary. Hum. Reprod. (Oxf.), 13: 2831-2835, 1998.[Abstract/Free Full Text]
-
Yoshida K., Sugino T., Tahara H., Woodman A. C., Bolodeoku J., Nargund V., Fellows G., Goodison S., Tahara E., Tarin D. Telomerase activity in bladder carcinoma and its implication for non-invasive diagnosis by detection of exfoliated cancer cells in urine. Cancer (Phila.), 79: 362-369, 1997.[CrossRef][Medline]
-
D’Hallewin M. A., Baert L. Initial evaluation of the bladder tumor antigen test in superficial bladder cancer. J. Urol., 155: 475-476, 1996.[CrossRef][Medline]
-
Schamhart D. H., de Reijke T. M., van der Poel H. G., Witjes J. A., de Boer E. C., Kurth K., Schalken J. A. The Bard BTA test: its mode of action, sensitivity and specificity, compared to cytology of voided urine, in the diagnosis of superficial bladder cancer. Eur. Urol., 34: 99-106, 1998.[CrossRef][Medline]
-
Soloway M. S., Briggman V., Carpinito G. A., Chodak G. W., Church P. A., Lamm D. L., Lange P., Messing E., Pasciak R. M., Reservitz G. B., Rukstalis D. B., Sarosdy M. F., Stadler W. M., Thiel R. P., Hayden C. L. Use of a new tumor marker, urinary NMP22, in the detection of occult or rapidly recurring transitional cell carcinoma of the urinary tract following surgical treatment. J. Urol., 156: 363-367, 1996.[CrossRef][Medline]
-
Serretta V., Lo Presti D., Vasile P., Gange E., Esposito E., Menozzi I. Urinary NMP22 for the detection of recurrence after transurethral resection of transitional cell carcinoma of the bladder: experience on 137 patients. Urology, 52: 793-796, 1998.[CrossRef][Medline]
-
Schmetter B. S., Habicht K. K., Lamm D. L., Morales A., Bander N. H., Grossman H. B., Hanna M. G., Jr., Silberman S. R., Butman B. T. A multicenter trial evaluation of the fibrin/fibrinogen degradation products test for detection and monitoring of bladder cancer. J. Urol., 158: 801-805, 1997.[CrossRef][Medline]
-
Wiener H. G., Mian C., Haitel A., Pycha A., Schatzl G., Marberger M. Can urine bound diagnostic tests replace cystoscopy in the management of bladder cancer?. J. Urol., 159: 1876-1880, 1998.[CrossRef][Medline]
-
Pode D., Golijanin D., Sherman Y., Lebensart P., Shapiro A. Immunostaining of Lewis X in cells from voided urine, cytopathology and ultrasound for noninvasive detection of bladder tumors. J. Urol., 159: 389-392, 1998.[CrossRef][Medline]
-
Buchumensky V., Klein A., Zemer R., Kessler O. J., Zimlichman S., Nissenkorn I. Cytokeratin 20. a new marker for early detection of bladder cell carcinoma?. J. Urol., 160: 1971-1974, 1998.[CrossRef][Medline]
-
Klein A., Zemer R., Buchumensky V., Klaper R., Nissenkorn I. Expression of cytokeratin 20 in urinary cytology of patients with bladder carcinoma. Cancer (Phila.), 82: 349-354, 1998.[CrossRef][Medline]
-
Steiner G., Schoenberg M. P., Linn J. F., Mao L., Sidransky D. Detection of bladder cancer recurrence by microsatellite analysis of urine. Nat. Med., 3: 621-624, 1997.[CrossRef][Medline]
-
Kinoshita H., Ogawa O., Kakehi Y., Mishina M., Mitsumori K., Itoh N., Yamada H., Terachi T., Yoshida O. Detection of telomerase activity in exfoliated cells in urine from patients with bladder cancer. J. Natl. Cancer Inst., 89: 724-730, 1997.[Abstract/Free Full Text]
-
Kavaler E., Landman J., Chang Y., Droller M. J., Liu B. C. Detecting human bladder carcinoma cells in voided urine samples by assaying for the presence of telomerase activity. Cancer (Phila.), 82: 708-714, 1998.[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
V. Urquidi, D. Sloan, K. Kawai, D. Agarwal, A. C. Woodman, D. Tarin, and S. Goodison
Contrasting Expression of Thrombospondin-1 and Osteopontin Correlates with Absence or Presence of Metastatic Phenotype in an Isogenic Model of Spontaneous Human Breast Cancer Metastasis
Clin. Cancer Res.,
January 1, 2002;
8(1):
61 - 74.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
ABSTRACT
INTRODUCTION
