Vet. Pathol.

24509-5 14 ( 1 987)

Canine Ovarian Neoplasms: A Clinicopathologic Study of 71 Cases,

Including Histology of 12 Granulosa Cell Tumors
A. K. PATNAIK
AND P. G. GREENLEE

Department of Pathology, The Animal Medical Center, New York, NY

Abstract. In a retrospective study of 71 primary ovarian tumors in the dog, epithelial tumors (46%) were
more common than sex cord stromal (34%) and germ cell tumors (20%). There were more adenocarcinomas
(64%)than adenomas. Sex cord stromal tumors were equally divided into Sertoli-Leydig (12/24) and granulosa
cell tumors (12/24). There were equal numbers (7/14) of dysgerminomas and teratomas among the germ cell
tumors. Most teratomas (6/7) were malignant. Most granulosa cell tumors were solid; two were mostly cystic.
Patterns included sheets of round and ovoid to spindle-shaped cells separated by thin, fibrovascular stroma;
neoplastic cells formed rosettes or Call-Exner bodies. In some areas, neoplastic cells were in cords or columns
and formed cyst-like structures. Four granulosa cell tumors were macrofollicular, having cysts lined with gran-
ulosa cells. Median ages of dogs with different ovarian neoplasms were similar; all were more than 10 years
old, except the dogs with teratoma (mean age, 4 years). Most neoplasms were unilateral (84'?0), except the Sertoli-
Leydig cell tumors, many of which were bilateral (36%). Size of ovarian neoplasms varied (2 cm3 to 15,000
cm3).Twenty-nine percent of neoplasms metastasized; adenocarcinomas (48%) and malignant teratomas (50%)
had the highest rates, and distant metastasis was more common in malignant teratoma. Endometrial hyperplasia
was in 67% of the dogs; it was most common in dogs with sex cord stromal tumors (95%). Uterine malignancy
was not seen in dogs with granulosa cell tumors, although hyperplastic endometrium was in all dogs with this
tumor. Cysts in the contralateral ovaries were most common in dogs with sex cord stromal tumors.

Canine ovarian neoplasms can be broadly divided
into three major categories: (1) germ cell, (2) epithelial,
and (3) sex cord stromal. The rate of occurrence of
these three histologic types varies in different studies.
Epithelial neoplasms are most common in some stud-
ies,2J2and sex cord stromal tumors are most common
in o t h e r ~ . ~Adenocarcinomas
J~ were more common
than their benign counterparts in two studies that used
the same source m a t e ~ i a l . ~Germ
, ' ~ cell tumors were
less prevalent than the other two types of tumor in all
these studies, and dysgerminomas were more common
than teratomas in most of them.2s5*7J2J3 In all of these
~ t u d i e s ~except
, ~ , ~ Jone,I2
~ no distinction was made be-
tween granulosa cell tumor and Sertoli-Leydig cell tu-
mor. The World Health Organization (WHO) also in-
cludes both of these types among the granulosa cell
tumors in their classifi~ation.~ Norris, et al. categorized
Sertoli-Leydig tumors as a separate group, both in the
dog and in cattle.I0J2
In a recent study of feline ovarian neoplasms, the
largest percentage of tumors were sex cord stromal in
rigi in.^ Similarly, in cattle and horses, sex cord stromal
tumors are p r e d ~ m i n a n t . ~ JGranulosa
~ J ~ J ~ cell tumors
in cats and cattle are considered more malignant than
in the d ~ g . ~In, ~rats J ~and mice, benign epithelial tu-
mors, described as tubular adenomas, are the most
common of ovarian n e o ~ l a s m s . ~ J ~
In women, epithelial neoplasms constitute more than
509
two-thirds of ovarian neoplasms and 90% of malignant
t u m o r ~ . ~ Granulosa
~-I~ cell and Sertoli-Leydig cell tu-
mors compose 2.0% and 0.2% of all ovarian neoplasms
in women, re~pectively.'~J~ Benign teratomas (der-
moid cysts) are the most common among neoplasms
of germ cell ~ r i g i n . ~ ~ J ~
This is a retrospective study of the primary canine
ovarian tumors diagnosed at the Animal Medical Cen-
ter over a 10-year period. Clinical and pathologic find-
ings are described, including histologic classification
of 12 granulosa cell tumors based on the classification
of these tumors in man.
Materials and Methods
A 10-year retrospective survey of the medical records at
The Animal Medical Center yielded 7 1 dogs with adequate
tissue specimens to confirm a diagnosis of primary ovarian
neoplasm and to classify the tumors. Contralateral ovarian
and uterine specimens, whenever available, were also ex-
amined. Age, breed, clinical, and gross pathologic descrip-
tions were obtained from records. Tissues were fixed in 10%
buffered formalin, processed routinely, sliced at 6 pm, and
stained with hematoxylin and eosin (HE), mucicarmine, peri-
odic acid-Schiff (PAS), and modified Grimelius stains. His-
tologic classification was based on the major component in
each tumor. The World Health Organization (WHO) system
of classification for primary ovarian neoplasms in humans
was used for classifying these tumors in d o g ~ . ' ~ J ~
5 10 Patnaik and Greenlee

Table 1. Clinical and pathologic findings in 7 1 dogs with primary ovarian neoplasms.

1 "+-.."l:+..
MLGI aiiry
Median Size in cm3 Metastasis Ovary Endome-
trial
Histologic Classifi- No. of Median Age
cation Dogs (range) Total Unilat- Bilat- (range) (no. dogs) (contra- Hyper-
eral eral
lateral) plasia

Germ cell tumors 14

Dysgerminoma 7 12(10-13) 7 6 1 130(12-1,575) 1 5 3
Teratoma 7 4 (2-9) 7 7 0 504 (8-1,000) 3 1 1
Benign- 1
Malignant -6
Epithelial tumors 33
Adenoma 12 12(6-7) 7 7 0 18 (10-840) 0 5 8
Adenocarcinoma 21 lO(4-15) 18 13 5 125 (9-3,488) 10 2 9
Sex cord stromal tumors 24
Granulosa cell tumor 12 12(8-16) 11 11 0 250 (12-4,096) 2 11 12
Sertoli-Leydig 12 11 (9-15) 11 7 4 60 (2-15,000) 1 2 6
Total 71 61 51 10 17 26 39

Results
Clinical and pathologic findings show that neo-
plasms of epithelial origin were most common (33/7 1,
46%), and nearly two-thirds ofthem (21/33,64%) were
malignant (Table 1). One-third of the tumors (24/7 1,
34%) were of sex cord stromal origin, among which
were equal numbers of granulosa (12/24) and Sertoli-
Leydig (12/24) tumors. There also were equal numbers
(7/14) of dysgerminomas and teratomas among the
germ cell tumors, which was the smallest of the three
categories. Six of the seven teratomas were considered
malignant.
The median ages of the dogs in the three categories
of ovarian neoplasia were close; all median ages were
greater than 10 years (range, 10 to 14 years), except
that of the dogs with teratoma. The median age of these
dogs was 4 years (range, 2 to 9 years), and three-fourths
of the dogs with malignant teratoma were 4 years old
or less, which was the lowest median age of all the dogs
with ovarian tumors. The median age of the dogs with
dysgerminoma was 3 times that of the dogs with ter-
atoma.
Comparing the rate of occurrence of different breeds
of dog having ovarian neoplasia to the respective rate
of occurrence of the breeds in the hospital population
showed that Boston terriers (5.5% vs. O.lYo), mixed
breed dogs (30% vs. 26%), German shepherd dogs
(15.6% vs. 9%), and poodles (15.6% vs. 12%) had a
higher prevalence of ovarian neoplasia than other
breeds. However, because of the wide distribution of
breeds in each of the three histologic groups, preva-
lence of tumor according to breed was difficult to eval-
uate.
Most ovarian neoplasms in this study were unilateral
(84%); unilaterality prevailed in the three main cate-
gories (80% to 93%). Among individual histologic types,
bilaterality was higher among Sertoli-Leydig tumors
(36%) and adenocarcinomas (28%). None of the gran-
ulosa cell tumors was bilateral.
Tumors varied widely in size. The two largest tumors
were an anaplastic Sertoli-Leydig tumor (1 5,000 cm3)
and a granulosa cell tumor (4,093 cm3).Adenomas had
the smallest median size (18 cm3). The widest range
in size was seen among the Sertoli-Leydig cell tumors
(2 cm3 to 15,000 cm3).
Only 29% (17/58) of malignant ovarian neoplasms
had metastases. Malignant teratomas (316; 50%) and
adenocarcinomas (1 0/2 1; 48%) had the highest rate.
Peritoneal carcinomatosis was the most common form
of metastasis in all three groups, although distant vis-
ceral metastasis was more common in dogs with ma-
lignant teratoma.
In 58 dogs in which the uterus was examined, cystic
endometrial hyperplasia was seen in 39 (67%). Where-
as 18 of 19 dogs (95%) with sex cord stromal tumors
and all dogs with granulosa cell tumors had cystic en-
dometrial hyperplasia, only 17 of 31 dogs (55%) with
epithelial tumors, and none with malignant teratoma,
had this change. Endometritis and/or squamous meta-
plasia of the uterine mucosa was seen in a few dogs
with epithelial and Sertoli-Leydigcell tumors, and more
commonly among dogs with granulosa cell tumors
(4/12; 33%). Contralateral ovaries of dogs with uni-
lateral neoplasms were examined in 48 cases, and in
26 dogs (54%) there were cystic changes. All ovaries
in dogs with sex cord stromal tumors (1 3/ 13) had cystic
changes, but only seven of 25 dogs (28.1%) with epi-
thelial tumors, and two of 18 dogs (1 l0/o) with adeno-
carcinoma, had cystic changes.
 Canine Ovarian Neoplasms 511

Fig. 1. Microfollicular pattern in granulosa cell tumor; tubular and solid areas with thin stroma. HE.
Fig. 2. Granulosa cell tumor; round and ovoid cells formed as Call-Exner bodies (arrow). HE.
Fig. 3. Granulosa cell tumor; anaplastic, spindle cells formed as Call-Exner bodies (arrow). HE.
Fig. 4. Granulosa cell tumor; anaplastic cells with multiple mitotic figures to form Call-Exner bodies (arrow). HE.
512 Patnaik and Greenlee

thick, and hyalinized in areas, or cellular-having lip-

id-containing, spindle-shaped, or polygonal theca cells.
Neoplastic cells were uniform or moderately pleo-
morphic- small, round ovoid to spindle-shaped, or
polyhedral. Nuclei were round or ovoid with scattered
chromatin and not-so-prominent nucleoli. Cytoplasm
of these cells was scanty, granular, and slightly eosin-
ophilic. Mitotic cells were rare, except in two tumors,
in which neoplastic cells were more anaplastic than in
the other tumors, and mitotic cells ranged from two
to four per high power field (Fig. 4). Small cysts and
acinar-like structures lined with multiple layers of neo-
plastic cells, a few of which had pale eosinophilic se-
cretions, were occasionally seen. Large areas of necrosis
and hemorrhage were common in all eight tumors.
Moderate to focally-dense infiltrations of eosinophils
were seen among the neoplastic cells in all eight tu-
mors.
Four of the granulosa cell tumors were macrofollicu-
lar, characterizedby various-sized, cyst-like structures,
lined with multiple layers of granulosa cells, some
forming papillary structures (Fig. 5). Some cysts con-
tained eosinophilic fluid or blood and cellular debris;
others were empty. Neoplastic cells in and around the
cystic structures were arranged in a trabecular pattern
or in irregular cords (Fig. 6). Solid areas were seen
between cysts. Histologic patterns in these areas were
similar to those described for the above group of eight
Fig. 5. Macrofollicular granulosa cell tumor; various- tumors. Occasional tubular structures containing Ser-
sized, cyst-like structures containing secretions. HE. toli cells were seen in two of the four tumors.
Fig. 6. Granulosa cell tumor showing rows of cells sep-
arated by stroma (watered silk pattern). HE. Discussion
Uniform classification systems have not been fol-
lowed in previous epidemiologic studies of canine
ovarian n e o p l a s m ~ . ~Thus,
~ ~ ~ it
~ is
J ~difficult
J~ to com-
Granulosa cell tumors pare the prevalence of different histologic types of
Gross pathology. The I2 tumors were mostly solid, ovarian neoplasms in our dogs with the findings of
encapsulated, and grayish, some with small cysts and other investigators. In our study, epithelial neoplasms
areas of hemorrhage and necrosis. Two were predom- predominated among the three major categories, and
inantly cystic, consisting of various-sized cystic struc- germinal cell tumors were the least common. These
tures containing watery fluid, sometimes mixed with findings agree with some ~ t u d i e sand ~ , ~disagree with
clotted blood. Grayish solid areas were seen among the others.2J2In some studies, adenomas (68%)were more
cysts. common than adenocarcinomas,2J2whereas in ours
Histology. The general histologic features in eight and in another s t ~ d yadenocarcinomas
,~ predominat-
of the 12 granulosa cell tumors were similar, but dif- ed.
ferent histologic patterns predominated in different tu- The 12 granulosa cell tumors of this study composed
mors and in different areas of the same tumor. The 17% (12/72) of the 72 ovarian neoplasms and 50% of
most common histologic characteristic was irregular the 24 sex cord-stromal tumors. Prevalence of gran-
masses or columns of cells separated by stroma. In ulosa cell tumors was higher (15/84) than that of Ser-
some areas, the neoplastic cells formed irregular cords toli-Leydig cell tumors (6/84) in one previous study in
or rows and groups of rosette and tubule-like structures which these tumors were classified separately.I2 In
separated by stroma (Fig. 1). In larger groups and in another study, however, features of Sertoli-Leydig tu-
columns, peripheral cells were along the stroma, and mors were seen in 50% of stromal tumors, although
cells in the center often formed rosettes similar to Call- they were not classified as a separate group.2It is dif-
Exner bodies (Figs. 2,3). The stroma was fibrovascular, ficult to compare the prevalence of the Sertoli-Leydig
 Canine Ovarian Neoplasms 513

cell tumors in our study or the study cited12with results Eighty-four percent of ovarian neoplasms in this
of other epidemiologic studies because Sertoli-Leydig study were unilateral. A similar finding has been seen
cell tumors are usually included with granulosa cell in the dog and other domestic animal^.^,^,^ This is also
tumor~.2,5,8,9,13 true in humans, but the frequency of bilaterality is
Sertoli-Leydig cell tumors are established as a sep- higher in dogs (16% vs. 5-10Y0).~~J~
arate group in human m e d i ~ i n e . ~They ~ J ~ have
J ~ com- In man, endometrial hyperplasia, endometrial car-
plex but distinct histologic characteristics that differ- cinoma, and breast cancer are associated with ovarian
entiate them from granulosa cell tumors. Granulosa neoplasms.1J6Even though endometrial hyperplasia,
cell tumors can easily be distinguished from Sertoli- sometimes associated with squamous metaplasia, has
Leydig cell tumors; the former are characterized by frequently been reported in dogs in this and other stud-
typical, microfollicular, solid, or cystic histologic pat- ies, no association with endometrial malignancy has
terns and Cell-Exner bodies and the absence of tubular been d o c ~ m e n t e d .In ~ .this
~ series, endometrial hyper-
structures, primitive stromal cells, and the male Sertoli plasia was most common among dogs with granulosa
cell elements-all three of which are required for a cell tumors (1 2/ 12), suggestingan estrogenic effect, and
diagnosis of Sertoli-Leydigcell tumor. In addition, Ser- high concentrations of sex hormones have been re-
toli-Leydig cell tumors often contain heterologous ele- corded in case reports of granulosa cell tumors.6 We
ments such as intestinal epithelium and neuroendo- did not have hormonal assays in our dogs to establish
crine cell components, which are not seen in granulosa this. In human patients, endometrial hyperplasia (up
cell tumors.’J6J8 to 62%) and endometrial carcinoma (5% to 25%) have
Histologic patterns of human granulosa cell tumors been associated with granulosa cell tumors.LJ6These
include microfollicular, macrofollicular,trabecular, in- endometrial changes are the result of a prolonged, un-
sular, watered silk, solid tubular, sarcomatoid, and leu- opposed estrogen effect on the endometrium.
tinized. Most commonly, a mixture of the different In our study, only 24% (17/71) of all the ovarian
patterns is seen within a single tumor. Similarly, neoplasms had metastasis, but 50% of teratomas ( 3 / 6 )
most of these histologic patterns were seen in the tu- and 48% of adenocarcinomas (10/2 1) had metastasis.
mors we studied, and there was a mixture of the pat- Very few teratomas have been described as malignant
terns in all 12 tumors. The rate of metastasis in our in previous studies, but a higher percentage of meta-
dogs, however (2/12, 17%), and in a previous study,12 static adenocarcinoma has previously been r e p ~ r t e d . ~ . ~
was lower than that observed in man (up to 30°/o).1,16 Most teratomas in human patients are benign.I”I6
However, most granulosa cell tumors in the cat and
cattle are considered malignant, with metastasis in up
to 50% ofthe ~ases.~JOJl These findings suggest a species Acknowledgements
variation in cases of granulosa cell tumor, which has The authors thank The Bodman Foundation for financial
been suggested previously.12 support and A. Christine MacMurray for editorial assistance.
Malignant teratomas (6/7), 50% with metastasis, were
seen more frequently in our dogs than in dogs of pre- References
vious ~ t u d i e s . ~Dysgerminomas
,~J~ were as prevalent
1 Anikwue C, Dawood MY, Kramer E Granulosa and
as teratomas (10% each); this does not agree with pre- theca cell tumors. Obstet Gynecol 51:214-220, 1978
vious studies in which dysgerminomas were more 2 Cotchin E: Canine ovarian neoplasms. Res Vet Sci 2:
common than the teratomas.2.12 133-142, 196 1
Median ages of our dogs were similar to those re- 3 Gelberg HB, McEntee K Feline ovarian neoplasms. Vet
ported in previous studies of ovarian neoplasms, with Pathol22572-576, 1985
the exception of dogs with teratoma (median age, 4 4 Gregson RL, Lewis DJ, Abbott DP: Spontaneousovar-
years). Age ranges, however, were wider than previ- ian neoplasms of the laboratory rat. Vet Pathol21:292-
ously r e c ~ r d e d . ~Wide
. ~ J ~age ranges were found in dogs 299, 1984
with adenocarcinoma (4 to 15 years) and teratoma (2 5 Hays HM Jr, Young JL Jr: Epidemiologic features of
to 9 years). Except for dogs with dysgerminoma and canine ovarian neoplasms. Gynecol Oncol 6:348-353,
Sertoli-Leydig cell tumor, our dogs were comparable 1978
6 McCandlish IAP, Munro CD, Breeze RG, Nash AS:
in age to human patients with ovarian tumors.I5J6In Hormone producing ovarian tumours in the dog. Vet
man, dysgerminomas and Sertoli-Leydig cell tumors Rec 105:9, 1979
develop mostly in the young, whereas in this and other 7 McEntee IS,Zepp CP Jr: Canine and bovine ovarian
reports of these two tumors, the dogs were o ~ ~ . ~ , ~ tumors. J ~ J ~Proc , ~1st~World Congress Fertility Sterility 2:
Certain breeds of dog allegedly have a predilection for 649-663, 1953
certain histologic types of ovarian tumor,5but we were 8 Moulton JE: Tumors of the female genital system. In:
not able to confirm this. Tumors in Domestic Animals, ed. Moulton JE, 2nd ed.,
514 Patnaik and Greenlee
pp. 330-336. University of California Press, Berkeley,
1978
9 Nielsen S W, Misdorp W, McEntee K: XV. Tumors of
the ovary. Bull WHO 53:203-215, 1976
10 Norris HJ, Taylor HB, Gamer FM: Comparative pa-
thology of ovarian neoplasms. 11. Gonadal stromal tu-
mors of bovine species. Vet Pathol6:45-58, 1969
11 Norris HJ, Gamer FM, Taylor HB: Pathology of feline
ovarian neoplasms. J Path 97138-143, 1969
12 Norris HJ, Gamer FM, Taylor HB: Comparative pa-
thology of ovarian neoplasms. IV. Gonadal stromal tu-
mours of canine species. J Comp Pathol 80:399-405,
1970
13 Priester WA, McKay FW: The occurrence of tumors in
domestic animals. National Cancer Institute Monograph
No. 54, US Department of Health and Human Services,
Bethesda, MD, 1980
Request reprints from Dr. A. K. Patnaik, The Department of Pathology, The Animal Medical Center, 5 10 East 62nd Street,
New York, NY 10021 (USA).
14 Rehm S , Dierksen D, Deerberg F: Spontaneous ovarian
tumors in Han :NMRI mice: histologic classification,in-
cidence, and influence of food restriction. J Natl Cancer
Inst 72:1383-1395, 1984
15 Scully RE: Ovarian tumors: a review. Am J Pathol87:
686-720, 1977
16 Scully RE: Tumors of the ovary and maldeveloped go-
nads. Atlas of Tumor Pathology, 2nd series, fasc. 16.
Armed Forces Institute of Pathology, Washington DC,
1979
17 Weiss NS, Homonchuk T, Young JL Jr: Incidence of
the histologic types of ovarian cancer: the US Third Na-
tional Cancer Survey, 1969-1 97 1. Gynecol Oncol5: 16 1-
167, 1977
18 Young RH, Scully RE: Ovarian Sertoli-Leydig cell tu-
mors: a clinicopathological analysis of 207 cases. Am J
Surg Pathol9: 543-569, 1985