Comparative Analysis of Proliferating Cell Nuclear
Antigen, Bromodeoxyuridine, and Mitotic Index in
Uveal Melanoma
Siavash Ghazvini,*^ Stewart Kroll* Devron H. Char,* and Hilary Frigillana*
Purpose. Recent production of a monoclonal antibody, PC10, against proliferating cell nuclear
antigen (PCNA) makes it possible to evaluate cell cycling in formalin-fixed tissues. In this
study, the authors quantitatively evaluated the relationship between PCNA expression and
two other measures of cell cycling, bromodeoxyuridine labeling index (BrdU LI) and mitotic
index (MI), in archival uveal melanomas. The authors also examined the relative prognostic
importance of each measure.
Methods. Serial sections from 35 formalin-fixed, paraffin-embedded uveal melanomas were
immunostained with PC10 and BrdU antibody using a standard avidin-biotin-peroxidase
method. A quantitative scoring system was used to evaluate the fraction of cells that were
positive for PCNA, BrdU, and mitotic figures in the regions of high cycling. The LIs of the
different markers were compared, and their prognostic importance was evaluated.
Results. The median PCNA LI was 3.05% compared to the median BrdU LI of 0.94% and the
median MI of 0.034%. The PCNA LI was more variable in replicate sections than either the
MI or the BrdU LI. The correlation between PCNA LI and BrdU LI was 0.58, between PCNA
LI and MI it was 0.46, and between BrdU LI and MI it was 0.81. The relative risk of tumorrelated mortality per doubling of BrdU LI was 2.35, and of MI it was 2.34. Although these
were significant, PCNA LI of 1.08 was not.
Conclusions. Proliferating cell nuclear antigen immunostaining did not demonstrate a strong
relationship with either BrdU LI or MI. Unlike MI and BrdU LI, PCNA LI was not correlated
with tumor-related mortality. Caution is warranted in the interpretation of PCNA immunostaining in uveal melanomas. Invest Ophthalmol Vis Sci. 1995;36:2762-2767.
A he cell proliferation rate of solid tumors is an important parameter for survival, choice of optimum
therapy, and treatment response. 1 " 4 In uveal melanoma, we previously have used both mitotic counts in
the area of highest tumor proliferation and detection
of in vivo incorporated BrdU with appropriate monoclonal antibody to measure tumor cell proliferation.1
We have noted that although both of these techniques
correlate well, the latter is a more sensitive method to
detect low levels of cycling.5 The rate of tumor cell
From the * Department of Ophthalmology, University of California, and the
^Department of Biology, San Francisco Slate. University, San Francisco, California.
Supported in part Iry Research to Prevent Blindness, That Man May See, and. The
jokn and. Norma Bowles Fellowship (SC).
Submitted, for publication April 25, 1995; revised August IS, 1995; accepted
August 18, 1995.
Proprietary interest category: N.
Reprint requests: Devron II. Char, Department of Ophthalmology, University of
California, San Francisco, 10 Kirkham Street, Box 0730, San Francisco, CA
94143.
2762
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
proliferation of uveal melanoma has been shown to
be predictive of survival and of local tumor response. 10
Several techniques, including immunohistologic
assessment of tumor cell proliferation, can be used to
measure tumor cell growth rates. Immunostaining for
intrinsic cellular proliferation markers represents a
straightforward, simple technique that can be performed routinely on formalin-fixed, paraffin-embedded specimens and allows simultaneous assessment of
standard morphology.
Proliferating cell nuclear antigen (PCNA), a 36kDa nuclear protein that functions as an auxiliary protein to DNA polymerase delta, is a marker of cell proliferation.7 Elevated nuclear levels of this protein are
maximum during late Gl and S-phase.8 Several antibodies recognizing PCNA, including PC10, are available commercially and are reactive in formalin-fixed,
paraffin-embedded tissues.9 In contrast to BrdU label-
Investigative Ophthalmology & Visual Science, December 1995, Vol. 36, No. 1H
Copyright © Association for Research in Vision and Ophthalmology
Anti-PCNA/Immunocytochemistry
ing, PCNA studies do not require in vivo administration of the thymidine analog, and they have been described as a sensitive technique for the assessment of
tumor cell proliferation in archival specimens.10
In this study, we assessed, in a masked manner, a
cohort of uveal melanomas with known outcome using
PCNA, BrdU, and mitosis counts to determine the
reproducibility of each proliferation measure, their
relationship with one another, and their association
with patient survival.
MATERIALS AND METHODS
Sample Selection
All 35 study patients with uveal melanoma underwent
primary enucleation between December 1985 and
March 1992. Each patient received a 15-minute intravenous infusion of BrdU (200 mg/m2) 1 hour before
tumor removal. Informed consent was obtained after
an explanation of the nature and possible consequences of the study, which adhered the tenets of
the Declaration of Helsinki. Permission to administer
BrdU was obtained from the Human Experimentation
Committee at the University of California, San Francisco and from the National Cancer Institute. Followup was available on all 35 patients through August
1994. Tumors selected were either amelanotic or
lightly pigmented because bleaching tumors with
more melanin had a negative effect on specific staining. Heavily pigmented tumors, patients receiving any
form of radiotherapy, and tumors with inadequate
representative sections were excluded from the study.
One patient in whom mitotic and BrdU indices were
calculated was excluded because of insufficient specimen for PCNA labeling.
The median largest tumor diameter of these melanomas was 17 mm (range, 6 to 24 mm), and the median height determined from A-scan ultrasonography
was 11.5 mm (range, 3.3 to 20 mm). The median
patient age was 61 years (range, 23 to 83 years). Histologically, the melanoma cell type was spindle in six
patients, predominantly spindle in 12 patients, mixed
in 16 patients, and epithelioid in 1 patient. Tumors
in 12 of 35 patients involved the ciliary body, and in
5 patients there was extraocular extension. Metastases
were detected in 15 patients, with the median time to
detection 2.6 years (range, 0.1 to 6.3 years). The overall median follow-up was 3.4 years; 15 of 20 patients
without detectable metastases were followed for at
least 3 years.
Immunohistochemistry
After enucleation, specimens were placed in 10% buffered formalin and processed in a standard manner
in the eye pathology laboratory. Contiguous 10-/im
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
2763
serial sections were obtained. Sections were stained
with hematoxylin and eosin, BrdU, and PC10 in duplicate.
Immunohistochemical detection of PCNA and
BrdU was performed using a standard avidin-biotinperoxidase method." Briefly, sections were deparaffinized by passage through xylene and graded alcohols. After blocking for at least 8 hours in 10% horse
serum at 4°C, anti-PCNA monoclonal antibody (Boehringer Mannheim, Indianapolis, IN) was applied in
a dilution of 1:200. At this concentration, overnight
incubation at 4°C gave consistent nuclear immunostaining patterns across replicate sections. A secondary
antibody, biotinylated anti-mouse immunoglobulin G
(Vector Laboratories, Burlingame, CA) was applied
in a dilution of 1:200. Endogenous peroxidases were
quenched with 3% hydrogen peroxide for 30 minutes.
Proliferating cell nuclear antigen was detected using
an avidin-biotin-complex method (ABC Elite, Vector) and Vector Violet (Vector) as the chromogen.
Negative controls were established by staining normal
pineal gland and substituting mouse immunoglobulin
for primary antibody in the protocol.
Immunohistochemical detection of BrdU was performed as previously described.1 Briefly, slides were
treated with 2 N HC1 for 60 minutes at room temperature, followed by treatment with 0.1% protease
(Sigma, St. Louis, MO) before they were placed in
blocking serum. Anti-BrdU monoclonal antibody (Caltag, San Francisco, CA) was applied to sections in a
dilution of 1:2500 and incubated for 45 minutes at
room temperature. Detection of antigen was performed as described above.
Cell Counting
All slides were read in a masked manner. The staining
pattern of PCNA antigen and the BrdU labeled cells
showed considerable regional variation within tumors.
Diagrams of each tumor were drawn from histopathologic sections. The region of highest activity was determined by scanning full tumor sections for the intensity
BrdU staining and mitotic figure presence. The area
of highest cycling activity was noted on each sketch.
Counting was begun in one corner of the predetermined area of highest cycling, and 20 high-power
fields (X400) were counted. The number of positive
cells in each field was recorded for analysis. All cells
displaying nuclear PCNA staining (light or heavy)
were counted as positive. Similarly, all cells displaying
nuclear BrdU staining were counted as positive. The
mitotic index was determined by counting mitotic
figures according to criteria of absence of a nuclear
membrane and presence of hairy projections of nuclear material.
Uveal melanomas demonstrate heterogeneity of
cell size across tumors. To control for variations in the
2764
Investigative Ophthalmology & Visual Science, December 1995, Vol. 36, No. 13
6.4-
1.6"
*
*•
**
0.4"
•
•
0.1"
r = 0.81
0.005
0.02
0.08
0.32
Mitotic Index (%)
FIGURE 1. Bivariate plot of bromodeoxyuridine labeling index versus mitotic index. Crossed points correspond to patients whose tumors metastasized; periods correspond to patients without detectable metastases at the time of study.
One low outlier is omitted.
number of cells per high-power field, cell densities
were estimated from five high-power fields in the representative fields from the same area of highest activity
on the hematoxylin and eosin sections. This was used
to calculate the percent of tumor cells positive for
each of the proliferation markers as opposed to indices per millimeters.2 The PCNA and BrdU LIs and MI
were calculated as the percent of positive cells.
Statistical Techniques
Each of the proliferation markers was transformed to
a log scale. This transformation symmetrized the distribution of each marker across tumors. The univariate
distributions each appeared to be log normally distributed (see Figs. 1 to 3). Parametric and nonparametric
correlation measures were used. The effect of each
cell proliferation marker on tumor-related survival was
analyzed using univariate and multivariate Cox proportional hazards models.12 Multivariate models included patient age, tumor involvement of the ciliary
body, cell type, largest tumor diameter, and the respective cell proliferation measures.
standard deviation of counts divided by the mean of
the counts across tumor sections (Table 1). The variability in mitotic counts was caused by the low prevalence of mitosis. Similarly, variability in BrdU counts
was infrequent and usually was related to occasional
low counts.1 Although a similar pattern was evident in
the PCNA counts, this was not always the case. Triplicate PCNA observations with two sections stained on
the same day and one section stained on another day
were obtained in 13 tumors. The ratio of the relative
within-day variance to the relative across-day variance
was equal to 0.37. Thus, the variability across batches
studied on different days was greater than those on
the same day. However, some tumors with multiple
measurements on the same day also demonstrated excessive variability. To minimize the effects of outliers,
the median value of each cycling measure was used
for all analyses.
The median MI was 0.034% (interquartile range
[IQR], 0.018% to 0.071%). The median BrdU LI was
0.99% (IQR, 0.38% to 1.80%). The median PCNA
index was 3.05% (IQR, 1.45% to 4.80%). Bivariate
plots of each of the three paired cycling measures are
given in Figures 1 to 3. All plots exclude one tumor
that had both BrdU and PCNA indices less than
0.01%. The BrdU and mitosis measurements correlated better than either of the PCNA correlations. On
average, the BrdU LI was approximately 25 times
larger than the MI. Therefore, the length of S-phase
8.0-
£
2.0"
0.5"
0.1
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
1.6
6.4
BrdUrd Index (%)
RESULTS
Multiple observations from sections taken within 0.3
mm of each other were separately counted for mitoses,
BrdU uptake, and PCNA labeling. The PCNA index
demonstrated the greatest variability, defined as the
0.4
FIGURE 2.
Bivariate plot of proliferating cell nuclear antigen
index versus bromodeoxyuridine labeling index. Crossed
points correspond to patients whose tumors metastasized;
periods correspond to patients without detectable metastases at the time of study. One low outlier is omitted.
2765
Anti-PCNA/Immunocytochemistry
MI and the BrdU LI, as well as the moderate correlation between the PCNA index and each of the other
indices, we explored the failure for PCNA to have
prognostic importance. Patients whose tumors metastasized were highlighted in each of the bivariate plots.
Although the length of follow-up is ignored, high ratios of PCNA to either mitosis or BrdU were associated
with better outcomes. Similarly, low ratios of PCNA to
either BrdU or mitosis were associated with poorer
outcome. Inclusion of PCNA into a multivariate model
with either the BrdU LI or the MI resulted in better
prognosis with higher values of the PCNA index. The
latter prognostic results with PCNA were not statistically significant.
DISCUSSION
0.005
0.02
0.08
0.32
Mitotic Index (%)
FIGURE 3. Bivariate plot of proliferating cell nuclear antigen
index versus mitotic index. Crossed points correspond to
patients whose tumors metastasized; periods correspond to
patients without detectable metastases at the time of study.
One low outlier is omitted.
is approximately 25 times longer than mitosis. The
plot of PCNA with BrdU had slightly more scatter, and
the PCNA with mitosis plot had the poorest fit. High
PCNA indices in low mitotic tumors and low PCNA
indices in high mitotic tumors were present. In general, however, PCNA staining could be used to delineate the intratumor area of highest cell proliferation.
Univariate and multivariate results of each of the
cycling measures are given in Table 2. The multivariate results include adjustment for tumor cell type, ciliary body involvement, patient age, and largest tumor
diameter. Each of these is a known prognostic factor
in uveal melanoma. Although both the MI and the
BrdU LI were significant prognostic factors of tumorrelated survival, the PCNA index was not prognostic
in either univariate or multivariate analyses.
In light of the prognostic importance of both the
TABLE
Although PC10 immunoreactivity has been shown to
correlate with BrdU labeling and with mitosis in several tumors, our results demonstrated distinct discordant features.M'14 In contrast to measurements of either BrdU or mitosis, PCNA did not correlate with
prognosis in uveal melanoma. We have noted previously that BrdU LI and MI in areas of highest cycling
activity within uveal melanoma correlated strongly
with outcome. We did not find PCNA LI to be a good
quantitative measure of cell cycling or outcome compared to either BrdU LI or MI. Evans and coworkers1'
found a similar paradoxic correlation between high
PCNA staining and good outcome in cutaneous melanomas. Woosley and colleagues"1 noted that PCNA LI
in cutaneous malignant melanomas had no prognostic
significance, whereas MI in the regions of high activity
did. Davidsson and coauthors17 also found that PCNA
was not of prognostic importance in malignant melanoma of the ear. Similar findings have been reported
in other tumors.18"20
The reasons for the disparity between these several cell cycling measures and prognosis is uncertain.
Both the MI and the BrdU LI are prevalence measures. Each represents the percent of cells in a specific
phase of the cell cycle (DNA synthesis phase or mitosis) at a specific point in time. The PCNA index is not
l. Relative Variability Comparisons
Relative Variability Across Sections
Median
Mitosis
BrdU
Proliferating cell
nuclear antigen
Interquartile
Range
0.00-0.25
25
0.16
0.13
0.06-0.30
0.05-0.29
27
0.24
0.11-0.48
19
Within-tumor relative variability of each cycling measure.
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
0.25-0.50
0.50-1.00
>1.00
Investigative Ophthalmology & Visual Science, December 1995, Vol. 36, No. 13
2766
TABLE
2. Summary of Cox Model Analyses of Cell Cycling Measures
Variable
Univariate Analyses
Mitotic index
BrdU labeling index
PCNA index
Multivariate Analyses*
Mitotic index
BrdU labeling index
PCNA index
Scale
Relative
Risk
95% Confidence
Interval
P
Log,
Log,
Log,
2.49
1.99
1.04
(1.36, 4.55)
(1.16, 3.44)
(0.79, 1.39)
0.003
0.013
0.765
Log2
Log,
Log,
2.34
2.35
1.08
(1.31,4.18)
(1.31, 4.22)
(0.65, 1.80)
0.004
0.004
0.762
* Mullivariate models each included largest tumor diameter, patient age, and indicator variables for
mixed-cell type and involvement of the ciliary body.
PCNA = proliferating cell nuclear antigen.
as clear; it represents the percentage of cells with
PCNA accumulation above a threshold as detected by
immunohistochemistry. The amount of PCNA present
in a cell is dependent mainly on its phase of the cell
cycle and the half-life of the protein. The biologic
half-life of PCNA has been estimated to be approximately 20 hours. 21 In uveal melanoma, the mean
length of DNA synthesis is approximately 14 hours. 22
In this study, we determined that the geometric mean
ratio of PCNA to BrdU was 2.5:1. This suggests that
the phase of the cell cycle in which PCNA may be
detectable is more than twice the length of DNA synthesis. Hence, a cell that had been in S-phase but
passed through G2 and mitosis may still have sufficient
PCNA protein to stain positively.
In some tumors, results were widely discrepant.
Four of 35 tumors had a PCNA index less than the
BrdU index, and 9 of 35 tumors had a PCNA index
more than five times greater than the BrdU index. In
tumors with much higher PCNA-to-BrdU ratios, there
were smaller paired stained nuclei in addition to
larger isolated stained cells. We postulate that the former were postmitotic daughter cells and the latter
were in S-phase at the time of fixation. Hence, in some
tumors, the PCNA threshold was higher than the level
of PCNA detected in postmitotic daughter cells. In
tumors with lower PCNA-to-BrdU ratios, it appeared
that not all cells in S-phase were PCNA labeled. These
tumor-dependent variations distort the relationship
between the PCNA index and the cell proliferation
rate.
Several studies have shown that tumor-derived
growth factors may enhance the stabilization of the PCNA
mRNA and result in increased PCNA levels.23"25 Previous
studies from our laboratory have shown expression
of numerous cytokines in uveal melanoma. 26 These
factors may play a role in the accumulation of PCNA
without association to DNA synthesis. Variable fixation
times may also affect detection of the PCNA antigen.
Start and coworkers27 showed that PCNA immunoreactivity in nonneoplastic tissues is gready reduced after
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
48 hours of fixation in formalin and virtually is abolished after 72 hours. In our study, fixation times were
approximately 24 hours; it is unlikely that variable fixation intervals were an important factor in this study.
BrdU uptake shows distinct, reproducible, nuclear
immunolabeling across serial sections, whereas PCNA
immunostaining had a greater staining variability
across serial sections. This variability in PCNA intratumor staining intensity may have resulted in greater
subjective interpretation of positive cells. Because
PC10 antibody dilution was controlled, it is doubtful
that this parameter was responsible for the variable
staining, as suggested.28
Several other investigations29 have shown potential problems with PCNA data. Proliferating cell nuclear antigen staining was shown to overestimate the
growth fraction of LoVo cell line xenografts when
compared to mitotic counts. Coltera and Gown30 observed that PCNA staining defined different cell subpopulations compared to those that either were incorporating BrdU or were Ki-67 positive. Finally, Hall
and coauthors 8 noted that some neoplasias displayed
deregulated PCNA expression not correlated with cell
kinetics.
In conclusion, PCNA immunostaining did not appear closely related either to the S-phase fraction or
to the Mi, and it was not an important prognostic
factor in uveal melanoma. This is possibly the result
of the complex nature of PCNA and its half-life. In
view of the myriad factors known to influence PCNA
labeling, caution is warranted in its interpretation.
Proliferating cell nuclear antigen staining was useful,
however, for purposes of rapidly identifying regions
of high tumor cell proliferation. Further studies are
warranted to elucidate the paradoxic correlation between PCNA staining and outcome.
Key Words
bromodeoxyuridine (BrdU), mitotic index, prognosis, proliferating cell nuclear antigen, uveal melanoma
Anti-PCNA/Immunocytochemistry
References
1. Lattman J, Kroll S, Char DH, et al. Cell cycling and
prognosis in uveal melanoma. Clin Cancer Res.
1995; 1:41-47.
2. Delahunt B, Bethwaite PB, NaceyJN, Ribas JL. Proliferating cell nuclear antigen (PCNA) expression as a
prognostic indicator for renal cell carcinoma: Comparison with tumour grade, mitotic index, and silverstaining nucleolar organizer region numbers. JPathol.
1993; 170:471-477.
3. Pe'erJ, Gnessin H, Shargal Y, Livini N. PC10 immunostaining of proliferating cell nuclear antigen in posterior uveal melanoma: Enucleation versus enucleation
postirradiation groups. Ophthalmology. 1994; 101:5662.
4. Kindy-Degnan N, Char DH, Swift P, Kaleta S, Ljung
BM. Bromodeoxyuridine uptake in the assessment of
hyperthermic therapy for intraocular tumor. Arch Ophthalmol. 1989; 107:746-750.
5. Char DH, Crawford JB, Kaleta S, Howes EL, Lovato
AA, QuiveyJM. Analysis of radiation failure after uveal
melanoma brachytherapy. Am J Ophthalmol.
1989; 108:712-716.
6. Kroll S, Char DH, Butler P, Ghazvini S, O'Brien JM.
Uveal melanoma growth patterns and prognosis. Invest Ophthalmol Vis Sci. ARVO Abstracts. 1994; 35:1925.
7. Bravo R, Frank R, Blundell PA, MacDonald-Bravo H.
Cyclin/PCNA is the auxiliary protein of DNA polymerase delta. Nature. 1987;326:515-517.
8. Morris GF, Mathews MB. Regulation of proliferating
cell nuclear antigen during the cell cycle. J Biol Chem
MolBiol. 1989;264:13856-13864.
9. Hall PA, Levison DA, Woods AL, et al. Proliferating
cell nuclear antigen (PCNA) immunolocalization in
paraffin sections: An index of cell proliferation with
evidence of deregulated expression in some neoplasms. / Pathol. 1990; 162:285-294.
10. Yu CCW, Woods AL, Levison DA. The assessment of
cellular proliferation by immunohistochemistry: A review of currently available methods and their applications. HistochemJ. 1992; 24:121-131.
11. Hsu SM, Raine L, Fanger H. The use of avidin-biotinperoxidase (ABC) complex in immunoperoxidase
technique: A comparison between ABC and unlabelled antibody (PAP) procedures. / Histochem Catacomb. 1981; 29:577-580.
12. Cox DA. Regression models and life-tables./R Stat Soc
(Series B). 1972;34:187-202.
13. Amin MB, Ma CK, Linden MD, Kubus JJ, Zarbo RJ.
Prognostic value of proliferating cell nuclear antigen
index in gastric stromal tumors: Correlation with mitotic count and clinical outcome. Am J Clin Pathol.
1993; 100:428-432.
14. Sasaki A, Naganuma H, Kimura R, et al. Proliferating
cell nuclear antigen (PCNA) immunostaining as an
alternative to bromodeoxyuridine (BrdU) immunostaining for brain tumors in paraffin embedded sections. ActaNeuro. 1992; 117:178-181.
Downloaded From: http://iovs.arvojournals.org/ on 06/17/2017
2767
15. Evans AT, Blessing K, Orrell JM, Grant A. Mitotic indices, anti-PCNA immunostaining, and AgNORs in thick
cutaneous melanomas displaying paradoxical behavior. J Pathol. 1992; 168:15-22.
16. Woosley JT, Dietrich DR. Prognostic significance of
PCNA grade in malignant melanoma. / Cutan Pathol.
1993;20:498-503.
17. Davidsson A, Hellquist HB, Villman K, Gunnar W.
Malignant melanoma of the ear. / Laiyngeal Otol.
1993; 107:798-802.
18. De Goeij A, Ader H, de Bruine A. Expression of PCNA
in paraffin sections from human colon carcinoma patients. HistochemJ. 1992; 24:613.
19. Thomas N, Noguchi M, Kitagawa H, Kinoshita K, Miyazaki I. Poor prognostic value of proliferating cell nuclear antigen labeling index in breast carcinoma. J
Clin Pathol. 1993;46:525-528.
20. Cummings MC, Furnival CM, Parsons PG, Townsend
E. PCNA immunostaining in breast cancer. Aust N Z
JSurg. 1993; 63:630-636.
21. Bravo R, MacDonald-Bravo H. Existence of two populations of cyclin/proliferating cell nuclear antigen
during the cell cycle: Association with DNA replication
sites. J Cell Biol 1987; 105:1549-1554.
22. Kroll S, Char D, Kaleta-Michaels S. A stochastic
model for dual label experiments: An analysis of the
heterogeneity of S-phase duration. Cell Prolif. 1995;in
press.
23. Bravo R, MacDonald-Bravo H. Induction of the nuclear protein cyclin in quiescent mouse 3T3 cells stimulated by serum and growth factors: Correlation with
DNA synthesis. EMBOJ. 1984;3:3177-3181.
24. Jaskulski D, Gatti G, Travalli S, Calabretta B, Baserga
R. Regulation of the proliferating cell nuclear antigen
cyclin and thymidine kinase mRNA levels by growth
factors. JBiol Chem. 1988;263:10175-10179.
25. Hall PA, Coates PJ, Goodlad RA, Hart IR, Lane DP.
Proliferating cell nuclear antigen expression in noncycling cells may be induced by growth factors in vivo.
Br] Cancer. 1994;70:244-247.
26. Flier RS, Song JC, Char DH, Kaleta-Michaels S. Uveal
melanoma cytokines. Klin Monastsbl AugenlwilkDa. 1993;
202:174-179.
27. Start RD, Cross SS, Clelland C, Silcocks PB, Rogers
K, Smith JHF. Delay in fixation does not affect the
immunoreactivity of proliferating cell nuclear antigen
(PCNA). J Pathol. 1992; 168:197-199.
28. McCormick D, Yu C, Hobbs C, Hall PA. The relevance
of antibody concentration to the immunohistological
quantification of cell proliferation-associated antigens. Histopathology. 1993; 22:543-547.
29. Scott RJ, Hall PA, Haldane JS, et al. A comparison
of immunohistochemical markers of cell proliferation
experimentally determined growth fraction. J Pathol.
1991;165:173-178.
30. Coltrera MD, Gown AM. PCNA/cyclin expression and
BrdU uptake define different subpopulations in different cell lines./Histochem Cytochem. 1991;39:23-30.