Veterinary Clinical Pathology ISSN 0275-6382
TECHNICAL REPORT
Reproducibility of nuclear morphometry parameters from
cytologic smears of canine cutaneous mast cell tumors—intra- and
interobserver variability
Daniella de F.L. Barbosa, Mariana D. Paraventi, Ricardo de F. Strefezzi
Department of Veterinary Medicine, Faculty of Animal Science and Food Engineering, University of S~ao Paulo, Pirassununga, Brazil
Key Words
Grading, mastocytoma, observer experience,
prognosis
Correspondence
Ricardo de Francisco Strefezzi, Departamento
de Medicina Veterinaria, FZEA-USP, Av. Duque
de Caxias Norte 225, Pirassununga, SP 13635-
900, Brazil
E-mail: rstrefezzi@usp.br
DOI:10.1111/vcp.12167
Introduction
Morphologic diagnosis is based on the recognition of
visual patterns for several tissues by a qualified obser-
ver. Although many of the criteria are well defined for
each disease, inter- and intraobserver variabilities have
been documented and reach up to 50–60% among
experienced pathologists. 1
Mast cell tumors (MCT) are the most frequent
canine cutaneous neoplasm, accounting for up to 27%
of malignancies.2,3 Some breeds are considered to be
predisposed like Boxer, Boston Terrier, Bull Terrier,
and Labrador Retriever2,4, and the tumor is more com-
mon in animals from 7.5 to 9 years of age.3,5,6 The
macroscopic diagnosis of this tumor is very difficult
due to the enormous variation of the lesions, but it can
be made by cyto- and/or histopathology.6–10 MCT are
considered potentially malignant neoplasms.7,8,10,11
Histopathologic grading is considered the most
accurate method to predict the course of the dis-
ease.9,11,12 The most widely used grading system is the
Vet Clin Pathol 43/3 (2014) 469–472 ©2014 American Society for Veterinary Clinical Pathology and European Society for Veterinary Clinical Pathology
Background: Qualitative and/or semiquantitative histopathologic grading
has been the method of choice to predict the biologic behavior of mast cell
tumors (MCT). Recently, computer-assisted nuclear morphometric analysis
of histologic or cytologic specimens has been shown to correlate with
histopathologic grading of MCT. However, it is still not clear what the repro-
ducibility and variability of these computer-assisted measurements are.
Objectives: The objective of this study was to estimate the intra- and inter-
observer variability in computer-assisted morphometric analysis of nuclear
parameters in fine-needle cytologic smears of canine cutaneous MCT.
Materials and Methods: Measurements were performed by 2 observers
with basic training on the recognition of neoplastic cells and operating the
image analysis software. The analyses were performed twice at 6 months
apart.
Results: No significant effects were detected for observer, time, and the
interaction observer–time.
Conclusions: Nuclear morphometric analysis of mast cell tumors using
cytologic smears is a reproducible method for grading of MCT.
one suggested by Patnaik et al.4, in which the tumors
were divided into 3 degrees of differentiation.5,6,9,12
However, it is based on several subjective characteris-
tics such as density of the cell population and cell
morphology, which lead to intra- and interobserver
variability.5,13,14 This fact motivates the search for
ancillary methods, which could make data collection
more objective and reproducible, and possibly improve
the prognostic accuracy of computer-assisted morpho-
metric methods.3
Computer-assisted image analysis is considered
one of the best solutions to eliminate observer variabil-
ity. Using a computer system, the measurement of
objective parameters such as the area, perimeter, and
diameter of cells and/or nuclei becomes easy and
precise, as well as recognition of their forms and also
staining intensities.1,15,16
A couple of studies have demonstrated the
correlation of computer-assisted nuclear morphomet-
ric parameters in cytopathology with the histopatho-
logic grading of MCT.5,17 These results were
469
Reproducibility of morphometry in mast cell tumors
corroborated in histopathology.18 The correlation of
nuclear morphometric parameters with survival of the
animals has also been shown, ie, mast cells presenting
larger nuclei were associated with shorter postsurgical
survival intervals.17 However, the occurrence of intra-
and interobserver variabilities in the morphometric
method have not been investigated in detail for MCT,
as the measurements in published works were per-
formed by a single observer. This analysis is especially
important for cytopathology, because the hematologic
staining methods can reveal the cytoplasmic granules,
which can impair the sharp visualization of nuclear
morphologic details.
To test the hypothesis that morphometry is a tool
to minimize intra- and interobserver variability, the
nuclear morphometric analysis of cytologic specimens
of canine cutaneous mast cell tumors was performed
by 2 independent observers, and during 2 distinct
periods.
Materials and Methods
A total of 21 samples of cytologic biopsies, 11
stained with Diff-Quick (Hematocor, Biol ogica
Comercial Ltda., S~ao Paulo, SP, Brazil) and 10 with
H&E, obtained by fine-needle aspiration, were used.
The Diff-Quick-stained smears were previously air-
dried, and the H&E slides were air-dried, followed
by immersion in absolute alcohol for 24 h. The
slides were analyzed under a trinocular microscope
(Axio Imager, Zeiss, G€ ottingen, Germany) with a 5
Megapixel digital camera (MRc AxioCam, Zeiss,
G€ottingen, Germany), coupled to a computerized
image analysis system (Axiovision, version 4.8,
Zeiss, G€
ottingen, Germany).
An experienced pathologist had previously diag-
nosed the samples as MCT by cytology and
histopathology. The measurements were performed
by 2 independent observers with skills for the rec-
ognition of malignant mast cells and use of the
software. Random fields were saved as “.jpg” files
using the 9 20 objective. A total of 100 nuclei of
neoplastic mast cells per sample were semiautomati-
cally outlined, with the aid of a computer mouse.
The observers were guided to consider all the cells
with sharp margins on each picture to avoid active
selection of larger or smaller cells. The slides were
numbered and the analyses were performed with-
out knowledge of the clinical history of the animals
and histologic classification of the MCT, during 2
distinct periods with an interval of 6 months. The
slides for each case were the same at both time
470 Vet Clin Pathol 43/3 (2014) 469–472 ©2014 American Society for Veterinary Clinical Pathology and European Society for Veterinary Clinical Pathology
Barbosa et al
points. Based on results of previous studies5,17, the
evaluated parameters were the nuclear area and
perimeter.
The results were statistically analyzed using the
software GraphPad Prism (version 4.02; GraphPad
Software Inc., CA, USA) and comparisons between
different observers and the same observer at different
time periods were performed by 2-way ANOVA. The
level of significance was set at P < .05.
Results
The MCT were obtained from 6 female (42.8%) and 8
male (57.2%) dogs. Three of them were mongrel dogs
and 3 were Boxers (21.4% each), 2 Brazilian Filas and
2 Schnauzers (14.2% each). The remaining dogs were
a Pinscher, a Poodle, a Dachshund, and a Doberman
Pinscher. Age was determined for 13 dogs, with a mean
of 8.6 years. Four of the MCT were histologic grade I
(28.6%), 7 grade II (50%), and 3 grade III (21.4%).
The mean nuclear area measured in
Diff-Quick-stained smears was 61.53  7.38 during
the first time period, and 61.25  6.97 during the
second time period, and 40.78  4.26 and
42.79  3.27 for the smears stained with HE, respec-
tively. The mean nuclear perimeter values were
26.73  3.97 and 28.87  2.11 for Diff-Quick, and
23.84  1.08 and 23.71  0.90 for HE. The measure-
ments for each case and observer were compared by
a 2-way repeated measures ANOVA and no signifi-
cant differences were detected for the effects obser-
ver, time, and interaction between observer and
time. The P-values are listed in Table 1.
Discussion
Our results indicate that the morphometric analysis in
cytopathology of MCT is a reproducible method. There
were no statistically significant differences between
Table 1. P-values for the effects observer, time, and interaction
between observer and time in the morphometric evaluation of different
cytologic specimens of canine mast cell tumors.
P-Values
H&E Diff-Quick
Area Perimeter Area Perimeter
Observer .1189 .0895 .1663 .1384
Time .0965 .7396 .8774 .1079
Interaction .7494 .9029 .7606 .5532
Barbosa et al
observers, time, or interaction time-observer for MCT
stained with 2 different stains used in cytology (P > .05).
Measurement of nuclear morphometric parame-
ters in cytologic specimens could be easier to perform
than in histology because the cells are spread out over
the surface of a glass slide. In contrast, measurement of
nuclear morphometric parameters in histologic speci-
mens must contend with the irregular nuclear outlines
that may be present in a histologic specimen.1 The
staining method can also contribute to this result, as
the hematologic dyes facilitate the identification of
granulated mast cells, thus decreasing the likelihood of
misdiagnosis of other cells instead of mast cells. The
higher values for area and perimeter in the Diff-Quick-
stained smears are probably due to the fixing process of
the samples. The Diff-Quick-stained smears were air-
dried, while smears for H&E staining were air-dried
and subsequently immersed in absolute alcohol for
24 h, which causes cell shrinking.
The convenience, speed, and low cost of such tests
for the diagnosis of neoplastic processes, especially
with regard to skin and subcutaneous round cell
tumors, can be combined with computer-assisted mor-
phometry, adding numerical values, which can
increase objectivity of observations.16,19–23 Small vari-
ations in size, shape, and texture in most cases are
imperceptible to the human eye even with the aid of
light and electron microscopy, and can be detected by
computer-assisted image analysis.16,21
In veterinary oncology, scientific studies on mor-
phometry as a prognostic indicator of neoplastic dis-
eases have evolved. Its potential has been observed in
mammary neoplasms24–28, melanoma29, squamous
cell carcinomas of the skin18,30, and cutaneous
MCT.5,17,18
Despite the known advantages of these methods,
histopathologic analysis is still considered the gold
standard for the formulation of a prognosis in cases of
canine cutaneous MCT, even with the knowledge of
the risk for inter- and intraobserver variability.1,5,13,14
However, our results have shown that the morpho-
metric analysis on cytopathology slides of MCT is a
reproducible method, and can be used as an ancillary
method in canine cutaneous MCT grading.17
Acknowledgments
Our appreciation goes to the Conselho Nacional de Desen-
volvimento Cientıfico e Tecnol
ogico (CNPq), for the under-
graduate scholarship (Grant no. 124850/2010-3).
Disclosure: The authors have indicated that they
have no affiliations or financial involvement with any
organization or entity with a financial interest in, or in
Vet Clin Pathol 43/3 (2014) 469–472 ©2014 American Society for Veterinary Clinical Pathology and European Society for Veterinary Clinical Pathology
Reproducibility of morphometry in mast cell tumors
financial competition with, the subject matter or mate-
rials discussed in this article.
References
1. Marchevsky AM, Erler BS. Morphometry in pathology.
In: Marchevsky AM, Bartels PH, eds. Image Analysis: A Pri-
mer for Pathologists. New York: Raven Press; 1994:360 p.
2. Vail DM. Mast cell tumors. In: Withrow SJ, MacEwen
EG. Small Animal Clinical Oncology. 4th ed. St. Louis,
MO: Elsevier - Health Sciences Division; 1996:864p.
3. Welle MM, Bley CR, Howard J, R€ ufenacht S. Canine
mast cell tumours: a review of pathogenesis, clinical
features, pathology and treatment. Vet Dermatol.
2008;19:321–339.
4. Patnaik AK, Ehler WJ, MacEwen EG. Canine cutaneous
mast cell tumor: morphologic grading and survival time
in 83 dogs. Vet Pathol. 1984;21:469–474.
5. Strefezzi RF, Xavier JG, Cat~
ao-Dias JL. Morphometry of
canine cutaneous mast cell tumors. Vet Pathol.
2003;40:268–275.
6. Strefezzi RF, Kleeb SR, Xavier JG, Cat~ao-Dias JL. Prog-
nostic indicators for mast cell tumors. Brazilian J Vet
Pathol. 2009;2:110–121.
7. Macy DW. Canine mast cell tumors. Vet Clin North Am
Small Anim Pract. 1985;15:783–803.
8. O’Keefe DA. Canine mast cell tumors. Vet Clin North Am
Small Anim Pract. 1990;20:1105–1115.
9. London CA, S eguin B. Mast cell tumors in the dog. Vet
Clin North Am Small Anim Pract. 2003;33:473–489.
10. Ginn PE, Mansell JEKL, Rakich PM. Skin and Append-
ages. In: Maxie , M.G. , ed. Jubb, Kennedy and Palmer’s
Pathology of Domestic Animals. 5th ed. California: Else-
vier; 2007:553–781.
11. Sim~oes JPC, Schoning P, Butine M. Prognosis of canine
mast cell tumors: a comparison of three methods. Vet
Pathol. 1994;31:637–647.
12. Misdorp W. Mast cells and canine mast cell tumours. A
review. Vet Q. 2004;26:156–169.
13. Kiupel M, Webster JD, Miller RA, Kaneene JB. Impact
of tumour depth, tumour location and multiple syn-
chronous masses on the prognosis of canine cutaneous
mast cell tumours. J Vet Med A Physiol Pathol Clin Med.
2005;52:280–286.
14. Northrup NC, Howerth EW, Harmon BG, et al. Varia-
tion among pathologists in the histologic grading of
canine cutaneous mast cell tumors with uniform use of
a single grading reference. J Vet Diagn Invest.
2005;17:561–564.
15. Dardick I, Caldwell DR, Mccaughey WTE, Al-Jabi
M. Nuclear morphologic and morphometric analyses
471
Reproducibility of morphometry in mast cell tumors
of large noncleaved cell and immunoblastic non-
Hodgkin’s lymphomas. Hum Pathol.
1984;15:965–972.
16. Eldar S, Sabo E, Cohen A, et al. The value of histomor-
phometric nuclear parameters in the diagnosis of well
differentiated follicular carcinomas and follicular ade-
nomas of the thyroid gland. Histopathology.
1999;34:453–461.
17. Strefezzi RF, Xavier JG, Kleeb SR, Cat~
ao-Dias JL.
Nuclear morphometry in cytopathology: a prognostic
indicator for canine cutaneous mast cell tumors. J Vet
Diagn Invest. 2009;21:821–825.
18. Maiolino P, Cataldi M, Paciello O, Restucci B, De VG.
Nucleomorphometric analysis of canine cutaneous
mast cell tumours. J Comp Pathol. 2005;133:209–211.
19. Duncan JR, Prasse KW. Cytology of canine cutaneous
round cell tumors. Vet Pathol. 1979;16:673–679.
20. Cowell RL. Cytology: part II. Vet Clin N Am Small.
2003;33:IX–X.
21. Dardick I, Payne CM. Rationale and potential for image
analysis in histopathology: an overview. Ultrastruct
Pathol. 1992;16:97–98.
22. Hamilton PW, Allen DC. Morphometry in histopathol-
ogy. J Pathol. 1995;175:369–379.
23. Ikegushi M, Oka S, Saito H, et al. Computerized nuclear
morphometry: a new morphologic assessment for
advanced gastric adenocarcinoma. Ann Surg.
1999;229:55–61.
472 Vet Clin Pathol 43/3 (2014) 469–472 ©2014 American Society for Veterinary Clinical Pathology and European Society for Veterinary Clinical Pathology
Barbosa et al
24. Ciurea D, Wilkins RJ, Shalev M, Liu Z, Barba J, Gil J.
Use of computerized interactive morphometry in the
diagnosis of mammary adenoma and adenocarcinoma
in dogs. Am J Vet Res. 1992;
53:300–303.
25. De VG, Maiolino P. Prognostic value of nuclear mor-
phometry in feline mammary carcinomas. J Comp
Pathol. 1997;117:99–105.
26. Simeonov R, Simeonova G. Computerized morphome-
try of mean nuclear diameter and nuclear roundness in
canine mammary gland tumors on cytologic smears. Vet
Clin Pathol. 2006;35:88–90.
27. Simeonov R, Simeonova G. Fractal dimension of canine
mammary gland epithelial tumors on cytologic smears.
Vet Clin Pathol. 2006;35:446–448.
28. Simeonov R, Simeonova G. Computerized cytomor-
phometric analysis of nuclear area, nuclear perimeter
and mean nuclear diameter in spontaneous canine
mammary gland tumours. Vet Res Commun.
2007;31:553–558.
29. Roels SL, Van DA, Van ME, Ducatelle RV. DNA ploidy
and nuclear morphometric variables for the evaluation
of melanocytic tumors in dogs and cats. Am J Vet Res.
2000;61:1074–1079.
30. Martano M, Damiano S, Restucci B, Paciello O, Russo
V, Maiolino P. Nuclear morphometry in canine
acanthomatous ameloblastomas and squamous cell
carcinomas. Eur J Histochem. 2006;
50:125–130.