Oncology–Original Article

Veterinary Pathology
2017, Vol. 54(2) 204-211
Feline Injection-Site Sarcoma: Matrix ª The Author(s) 2016
Reprints and permission:
sagepub.com/journalsPermissions.nav
Remodeling and Prognosis DOI: 10.1177/0300985816677148
journals.sagepub.com/home/vet

I. Porcellato1, L. Menchetti1, C. Brachelente1, M. Sforna1,

A. Reginato2, E. Lepri1, and L. Mechelli1

Abstract
Feline injection-site sarcoma (FISS) is an aggressive tumor believed to arise from the proliferation of fibroblasts and myofibroblasts
in areas of chronic inflammation, particularly at sites of injection. Local recurrence is frequent after surgical excision. Gelatinases
(MMP-2 and MMP-9) and their inhibitor (TIMP-2) are endopeptidases pivotal in extracellular matrix remodeling and therefore in
tumor invasiveness. The aim of this study was to investigate the immunohistochemical expression of MMP-2, MMP-9, and TIMP-2
in FISS to assess their usefulness as prognostic factors. Size, soft tissue sarcoma (STS) grading system, depth of infiltration, surgical
margins, and Ki-67 index were evaluated as additional prognostic markers. Twenty-four cases of primary FISS were classified
according to clinical follow-up as nonrecurrent (NR, n ¼ 14; 58.3%) and recurrent (R, n ¼ 10; 41.7%). MMP-2, MMP-9, and TIMP-2
were variably expressed in the FISS examined, confirming their role in tumor invasiveness, yet they did not show significant
differences between the R and NR groups. These results could be due to different tumor stages or to the multiple activities of
these enzymes, not limited to ECM remodeling. The immunohistochemical expression of these enzymes considered alone does
not seem to be useful as a prognostic marker. STS grading system, depth of infiltration, surgical margins, and Ki-67 index did not
relate to recurrence. Instead, the size of the tumor, measured after formalin fixation, with an optimal cutoff of 3.75 cm (accuracy
¼ 86%; P < .05), and the mitotic count, with an optimal cutoff of 20 mitoses/10 HPF (accuracy ¼ 80%; P < .05), could be evaluated
as useful prognostic markers.

Keywords
cats, immunohistochemistry, matrix metalloproteinases, mitotic count, neoplasm, postvaccinal sarcoma, skin, vaccine-associated
fibrosarcoma

Feline injection-site sarcoma (FISS) is a subcutaneous tumor
that rarely arises in the dermis.17 It commonly develops at sites
of vaccination; hence, in the past it was named ‘‘vaccine-
associated fibrosarcoma’’ or ‘‘postvaccinal sarcoma.’’9,18,22
Recent studies postulate that, besides vaccines, several differ-
ent stimuli, such as injection of foreign material, trauma, and
microchip implantation, induce chronic inflammation in the
subcutis or in the muscle and can be responsible for neoplastic
transformation and FISS occurrence.6,10,25,26 Feline injection-
site sarcomas are characterized by the proliferation of atypical
spindle cells and often by a variable amount of multinucleated
giant cells. Histologically, the majority of these tumors are
fibrosarcomas, but other histologic variants such as rhabdo-
myosarcoma, myxosarcoma, chondrosarcoma, osteosarcoma,
and histiocytic sarcoma have been recognized.9,17 Lymphocy-
tic aggregates can be seen mostly at the periphery of the tumor;
necrosis is often prominent.15 Feline injection-site sarcomas
show high local aggressiveness and tend to recur frequently
after surgical excision. Metastases are rare.25
The aggressiveness of a tumor is directly related to its ability
to invade and infiltrate the surrounding healthy tissue and to
metastasize to distant organs. Alterations in extracellular
proteolysis can lead to unregulated tumor growth, tissue remo-
deling, inflammation, tissue invasion, and metastasis.23 Extra-
cellular matrix (ECM) turnover is finely regulated in
physiological conditions, but when deregulation occurs, it
becomes a cause of many diseases, including invasive cancer.30
Matrix metalloproteinases (MMPs) are proteolytic enzymes
involved in ECM remodeling and tumor invasion.4 Since their
activity can allow cancer cell spread, they have been widely
investigated as prognostic factors and possible therapeutic tar-
gets in different types of tumors, including sarcomas, both in
human and veterinary oncology.1,3,12,14,34 Gelatinases, namely
1
Department of Veterinary Medicine, University of Perugia, Perugia, Italy
2
CDVet Diagnostic Laboratory, Via Ugo Guattari, Rome, Italy
Supplemental material for this article is available on the Veterinary Pathology
website at http://journals.sagepub.com/doi/suppl/10.1177/0300985816677148.
Corresponding Author:
I. Porcellato, Department of Veterinary Medicine, University of Perugia, Via San
Costanzo, 4, Perugia, 06126, Italy.
Email: ilariaporcellato@yahoo.it
Porcellato et al
MMP-2 and MMP-9, are a subgroup of MMPs that cleave
gelatin (denatured collagen), different types of collagen, and
other extracellular matrix components.7,24 These proteases are
inhibited by tissue inhibitors of matrix metalloproteinases
(TIMPs). TIMP-2 inhibits both MMP-2 and MMP-9 activity and
is important in regulating ECM remodeling, since it can inhibit
tumor cell invasion and migration.29 Altered expression of these
enzymes is critical in different physiological and pathological
processes and has been investigated in numerous feline condi-
tions.2,31,32,39,41 Sorensen et al36 evaluated a group of MMPs on
fresh FISS biopsies with reverse transcriptase-polymerase chain
reaction (RT-PCR), showing that MMP-2 and membrane type-
MMP-16 were correlated with survival time.
The soft tissue sarcoma (STS) grading system was derived
from human guidelines;8,42 adapted to dogs,11 it is based on 3
parameters: degree of differentiation, mitotic count, and tumor
necrosis. It has also been applied to FISS but showed variable
prognostic significance.9,15,33
Another interesting prognostic factor seems to be the status of
surgical margins.15 However, the possibility of recurrence
despite clean excisional margins suggests that an aggressive
surgical excision could not nullify the risk of local recurrence.33
The purpose of this retrospective study were to evaluate
the prognostic value of the immunohistochemical expression
of gelatinases (MMP-2, MMP-9) and their inhibitor (TIMP-
2) in FISS. Prognostic significance of size (major diameter),
tumor grade according to the STS grading system, depth of
infiltration, status of surgical margins, and Ki-67 expression
were evaluated.
Materials and Methods
Case Selection
Formalin-fixed and paraffin-embedded (FFPE) specimens
from 24 cats with primary FISS were selected from the archive
of the Department of Veterinary Medicine (University of
Perugia, Italy). Patients were diagnosed between 2008 and
2013 and followed through December 2015. Selection criteria
were as follows:
1. localization of the tumor (interscapular/dorsal);
2. type of treatment (curative excisional surgery);
3. the availability of follow-up information (telephone
interview with referring veterinarian);
4. a minimum 24-month follow-up period, considering
FISS disease-free intervals reported in literature;5,15,19,27
and
5. consensual diagnosis of FISS, based on typical histologic
features.9
Age, breed, sex, and spaying/neutering state of the animals
were collected from the department’s database. The size (major
diameter) of the tumors was measured after formalin fixation
(24 to 48 hours depending on specimen size) with a caliper and
recorded at the moment of trimming due to the lack of available
clinical information on all cases.
205
Table 1. Histologic Criteria Used for Soft Tissue Sarcoma Grading.a
Score Meaning
Mitotic Score: Mitoses in 10 HPF (FN22/40 objective)b
1 0–9
2 10–19
3 > 19
Tumor Necrosis Score
0 Absent
1  50%
2 > 50%
Tumor Cell Differentiation Score
1 Good
2 Moderate
3 Poor
Total Grading Score
Grade 1 3
Grade 2 4–5
Grade 3 6
Abbreviation: HPF, high power fields.
a
Adapted from Trojani et al.42
b
2.37 mm2 ¼ Standardized area for mitotic counts using a microscope.28
Histologic Evaluation
All specimens were cut into 5-mm sections, routinely stained
with hematoxylin and eosin, and examined by light micro-
scopy. Blind histologic evaluations were performed for each
sample by 3 operators (IP, CB, MS), with re-evaluation in case
of disagreement. The initial diagnosis of FISS was confirmed
and tumors were graded and classified based on the STS grad-
ing system40 (Table 1). Mitotic count (MC) was blindly eval-
uated, counting the number of mitoses on 10 contiguous high
power fields (HPF) (field number of the ocular of 22 and a 40
objective28), starting from hotspots and avoiding areas with
necrosis or severe inflammation. The total number of mitoses
was recorded and the mean value for the 3 operators was
assessed. Depth of infiltration was recorded (D ¼ dermal;
S ¼ subcutaneous; M ¼ muscular). Evaluation of margins was
possible on 19 of 24 cases (79.2%): the specimens were
trimmed in 3-mm slices, with parallel slicing at regular inter-
vals.21 The most representative sections, in particular the ones
where the tumor macroscopically approached the surgical mar-
gins, were selected. The number of histologic cassettes
obtained was related to the dimension of the tumor (from 2
to 10 cassettes). Histologically, margins were classified as sug-
gested in the literature11 into I (infiltrated ¼ specimens where
neoplastic cells were continuous with at least 1 surgical margin
in any plane) and NI (noninfiltrated ¼ tumor cells did not reach
surgical margins). Close margins (<3 mm) and clean margins
(distance of at least 3–5 mm) were considered as NI.11
Immunohistochemistry
Samples were cut into slices of 5 mm and mounted on positively
charged slides, dewaxed, and rehydrated. Immunohistochem-
istry was performed with standard protocols32 using antibodies
206
against MMP-2, MMP-9, TIMP-2, and Ki-67. Protocols for
each antibody are summarized in Supplementary Table 1.
Briefly, after incubation with primary antibodies, slides were
treated with an ABC ready-to-use kit (Abcam, Cambridge, UK)
following the manufacturer’s instructions. Positive reaction
was revealed with 3-amino-9-ethilcarbazole (Dako, Glostrup,
Denmark). Carazzi’s hematoxylin was used as a counterstain.
Feline healthy myocardium was used as a positive control for
MMPs and TIMP-2 as previously described.2 Vimentin immu-
nostaining was performed as a control of mesenchymal origin
of the tumors. Negative controls were incubated with TBS,
omitting the primary antibody. Immunostaining of MMP-2,
MMP-9, and TIMP-2 was blindly evaluated by 3 operators
(IP, AR, CB) and positive cells were given a percentage value
(0%–100%) on the total number of neoplastic cells. Nuclear
positivity stain for Ki-67 was evaluated in 10 randomly
selected HPF, avoiding areas of inflammation and necrosis.
The number of positive nuclei was recorded by each operator
and a mean value was calculated for each case.
Statistical Analysis
Diagnostic graphics and Levene’s test were used to test
assumptions and outliers.13 At the end of the follow-up period,
cases were assigned to recurrent (R) or nonrecurrent (NR)
groups. Differences in age, tumor size, and number of mitoses
as well as in MMP-2, MMP-9, TIMP-2, and Ki-67 expression
between R and NR groups were preliminarily analyzed using
Mann–Whitney tests. Correlation between mitoses/HPF and
Ki-67 were examined using Spearman rank correlations (r).
Values are expressed as mean + standard deviation (SD), and
median (Mdn) with interquartile range (IQR). Data classified
into categories (breed, sex, spaying/neutering status, mitotic
score, differentiation score, necrosis score, STS grade, comple-
teness of margins, depth of infiltration) were preliminarily
compared between R and NR groups using Fisher exact test
and were expressed as number and percentage. We used the
Life Table method to determine survival probabilities and
Kaplan–Meier method to calculate disease-free and survival
times. The differences of survival rate between R and NR
groups were evaluated by the log-rank test. Cats lost to fol-
low-up and cats that died from causes unrelated to FISS were
censored and included for survival analysis. The prognostic
significance of each variable in recurrence and survival was
further evaluated by the univariate Cox proportional hazards
model (Cox regression) and expressed as hazard ratio (HR)
with corresponding 95% confidence intervals (CIs) and P values
(from Wald statistic). The receiver operating characteristic
(ROC) analysis methodology was used to assess the diagnostic
accuracy of the parameters to detect recurrence. On the basis of
the AUC statistic, the diagnostic test can be noninformative
(AUC ¼ 0.5), less accurate (0.5 < AUC  0.7), moderately
accurate (0.7 < AUC  0.9), highly accurate (0.9 < AUC < 1),
or perfect (AUC ¼ 1).16,40 Optimal cutoff values were deter-
mined as points on the curve closest to (0, 1) and by the Youden
index. ROC curves were compared by a critical ratio z.16
Veterinary Pathology 54(2)
Statistical analysis was performed using SPSS 20.0 (SPSS Inc.,
Chicago, IL) and P < .05 was considered statistically significant.
Results
Clinical Data
The age of the cats at the moment of the diagnosis ranged
between 6 and 16 years, with a mean value of 11 + 3 years
(Mdn ¼ 11, IQR ¼ 10–13). Twenty-two of 24 (91.7%) were
domestic shorthair and 2 of 24 (8.3%) were Siamese cats. Of
the 24 cats included in the study, 10 (41.7%) were males
whereas 13 (54.2%) were females. The sex of 1 cat was not
recorded. Particularly, 6 of 10 males (60.0%) were neutered
and 7 of 13 females (53.8%) were spayed. No statistically
significant differences between the R and NR groups were
observed regarding age, breed, sex, and spaying/neutering sta-
tus. Follow-up at the end of the study ranged from 739 to
2467 days, with a median follow-up of 1222 days. More than
70% of the study population had a follow-up at the end of the
study > 3 years, with 30% of cases being followed for > 4 years.
The 6-month, 1-year, and 2-year estimated probabilities of not
developing recurrence of cats that experienced primary FISS
(n ¼ 24) were 79% + 8%, 67% + 10%, and 62% + 10%,
respectively, whereas their estimated survival probabilities
were 96% + 4%, 75% + 9%, and 75% + 9%, respectively
(mean + SD). The follow-up recorded after a minimum period
of 24 months after surgical excision showed that FISS recurred
in 10 out of 24 cases (41.7%). No metastases were detected. For
cats that experienced recurrence (R group, n ¼ 10), the median
disease-free interval was 162 days (IQR ¼ 93–245 days),
whereas the median survival time was 327 days (IQR ¼ 273–
994 days). The 6-month, 1-year, and 2-year estimated survival
probabilities of R cats were 90% + 9%, 37% + 16%, and 37%
+ 16%, respectively. At the end of the study, 7 cats (70%) of
the R group died from causes related to FISS and 3 (30%) were
censored (2 cases lost to follow-up and 1 dead from causes
unrelated to FISS) (Fig. 1a). As the NR group had no deaths due
to tumor-related causes, survival time statistics were not calcu-
lated, whereas log-rank test showed significant differences in
survival time between R and NR groups (w2 ¼ 17.6, P < .001).
Tumor size (major diameter) of formalin fixed samples ranged
between 1.5 and 8.0 cm. R tumors were larger (mean + SD ¼ 5.3
+ 2.1 cm; Mdn ¼ 5.0 cm, IQR ¼ 4.5–7.0 cm) than NR tumors
(mean + SD ¼ 2.7 + 1.1 cm; Mdn ¼ 2.5 cm, IQR ¼ 2.0–3.0 cm;
P ¼ .005) and the difference was statistically significant (P < .05),
with the highest values of major diameter (þ50%) in the R group.
For each unit increase in tumor diameter, cats were 1.72 times
more likely to have recurrence (HR ¼ 1.72, 95% CI ¼ 1.20–2.47;
P ¼ .003) and 1.88 times more likely to have tumor-related mor-
tality (HR ¼ 1.88, 95% CI ¼ 1.10–3.20; P ¼ .020).
Histologic Examination
Tumors were graded according to the STS grading system.
Results are shown in Table 2. Two of 24 cases were classified
Porcellato et al 207

Figure 1. Kaplan–Meier overall survival curves in relation to recurrence (a), diameter (b), and mitotic count (c). Cats that experienced
recurrence (n ¼ 10; continuous line) showed reduced overall survival compared with cats that did not experience recurrence (n ¼ 14; dashed
line; P < .001). With regard to histologic evaluations, the median survival time was lower in cats with tumors  3.75 cm (n ¼ 9; continuous line)
than in those with tumors < 3.75 cm (n ¼ 12; dashed line; P < .05) or in cats with mitotic count  20 (n ¼ 9; continuous line) compared with
those with mitotic count < 20 (n ¼ 15; dashed line; P < .05). Ticks represent censored observations.

Table 2. Soft Tissue Sarcoma Parameters and Grading.a mitoses/HPF; Mdn ¼ 29 mitoses/HPF, IQR ¼ 19–33
mitoses/HPF) compared to the NR group (mean + SD ¼ 13
Recurrence
+ 6 mitoses/HPF; Mdn ¼ 12 mitoses/HPF, IQR ¼ 9–19
Parameter No (n ¼ 14) Yes (n ¼ 10) P Value mitoses/HPF; P ¼ .001). Cox models confirmed that MC was
a good predictor of both recurrence (HR ¼ 1.09, 95%
Mitotic count, No. (%) CI ¼ 1.04–1.14; P ¼ .001) and mortality (HR ¼ 1.07, 95%
1 6a (42.9) 0a (0.0)
.008 CI ¼ 1.01–1.14; P ¼ .017).
2 6a (42.9) 3a (30.0)
3 2a (14.3) 7b (70.0) Depth of infiltration was assessed in 23 out of 24 cases
Necrosis, No. (%) (95.8%). None of the tumors was confined to the dermis; 15
0 0 (0.0) 2 (20.0) of 23 FISSs infiltrated the subcutis (65.2%: 9 NR and 6 R)
.165
1 7 (50.0) 6 (60.0) and 8 of 23 extended to the muscle layer (34.8%: 5 NR and
2 7 (50.0) 2 (20.0) 3 R). No associations between depth of infiltration and
Differentiation, No. (%) recurrence risk (P ¼ .714) or survival (P ¼ .283) were
1 3 (21.4) 5 (50.0)
.389 evidenced.
2 6 (42.9) 3 (30.0)
3 5 (35.7) 2 (20.0) Margins were evaluated in 19 out of 24 tumors and were
Grade, No. (%) infiltrated in 6 of 19 cases (31.6%; 3 R and 3 NR) and not
1 1 (7.1) 1 (10.0) infiltrated in 13 of 19 cases (68.4%; 5 R and 8 NR). No asso-
.830
2 5 (35.7) 5 (50.0) ciation between completeness of surgical margins and recur-
3 8 (57.1) 4 (40.0) rence risk was evidenced (P ¼ 1.000).
a
The only parameter statistically associated with recurrence was the mitotic
count. Value in bold shows a statistically significant association (Fisher exact
test). Each subscript letter denotes a subset of recurrence categories whose
Immunohistochemistry
column proportions did not differ significantly from each other at the .05 level. MMP-2. MMP-2 showed a highly variable percentage of cellu-
lar expression in both the R and NR groups. It was expressed in
multifocal to coalescent areas in 22 of 24 cases (91.6%). The 2
as grade 1 sarcoma (1 R and 1 NR), 10 of 24 cases as grade 2 negative tumors were in the R group (8.4%). R tumors showed
sarcoma (5 R and 5 NR), and 12 of 24 cases as grade 3 sarcoma a percentage of positive neoplastic cells between 0% and
(4 R and 8 NR). The STS grade did not differ statistically 83.3% (mean + SD ¼ 44.8% + 33.9%; Mdn ¼ 46.7%, IQR
between R and NR groups (P ¼ .830) and was not predictive ¼ 18.3%–80.0%). In the NR group, the percentage of positive
of the risk of local recurrence (P ¼ .885) or survival (P ¼ .762). cells ranged from 1.7% to 96.7% (mean + SD ¼ 37.6% +
Only the mitotic score showed a statistically significant differ- 32.6%; Mdn ¼ 25.8%, IQR ¼ 11.7%–66.7%). The expression
ence between the R and NR groups, with 7 of 10 animals in the of this gelatinase was seen predominantly in the cytoplasm of
R group (70%) having a mitotic score of 3 (P < .05). Hence, the mononucleated spindle cells (Fig. 2), with rare positivity in
number of mitoses was further investigated using MC values. multinucleated giant cells. No difference in the percentage of
MC was higher for the R group (mean + SD ¼ 27 + 10 MMP-2 expression between the R and NR groups was observed
208 Veterinary Pathology 54(2)

Figures 2–10. Feline injection-site sarcoma (FISS), cats. Immunohistochemistry, AEC chromogen, Carazzi’s counterstain. Figure 2. Case No.
17. Cytoplasmic MMP-2 expression in spindle cells of a recurrent tumor. Figure 3. Case No. 14. MMP-2 expression in peritumoral small vessels.
Figure 4. Case No. 1. MMP-2-positive vessels surrounded by a lymphocytic aggregate. Figure 5. Case No. 2. Variable MMP-9 expression
in different tumor lobules. Figure 6. Case No. 3. Expression of MMP-9 in multinucleated giant cells. Figure 7. TIMP-2 expression in FISS.
Figure 8. Case No. 9. Low fraction of Ki-67-positive nuclei in a recurrent FISS with > 1 mitoses/HPF. Figure 9. Case No. 3. High fraction of Ki-
67-positive nuclei in a nonrecurrent FISS with < 1 mitoses/HPF. Figure 10. Case No. 21. Nuclear Ki-67 staining in a giant multinucleated cell.

(P ¼ .666; Fig. 11). Cox models confirmed that it was not
a prognostic factor for recurrence (HR ¼ 1.00, 95%
CI ¼ 0.99–1.03; P ¼ .653) or survival (HR ¼ 1.01, 95% CI
¼ 0.99–1.01; P ¼ .405). As an additional finding, positive
staining was seen in intraneoplastic and peritumoral
endothelial cells of small vessels (Fig. 3), particularly in the
capillaries within lymphocytic aggregates (Fig. 4).
MMP-9. MMP-9 expression was extremely variable and posi-
tive in multifocal to coalescent areas (Fig. 5). Five FISSs were
Porcellato et al
Figure 11. Box plots comparing MMP-2, MMP-9, and TIMP-2 expres-
sion in feline injection-site sarcomas according to clinical follow-up
(recurrent: n ¼ 10; nonrecurrent: n ¼ 14).
completely negative (2 R and 3 NR). The percentage of posi-
tive cells in the R group ranged from 0.0% to 55.0% (mean +
SD ¼ 14.9% + 22.9%; Mdn ¼ 1.7%, IQR ¼ 0.0%–40.0%).
Percentage values ranged from 0.0% to 50.0% in the NR group
(mean + SD ¼ 21.2% + 19.1%; Mdn ¼ 18.3%, IQR ¼ 3.3%–
43.3%). Positive immunostaining was predominantly seen in
the cytoplasm of multinucleated giant cells (Fig. 6) and seldom
in mononucleated spindle cells with a weak staining intensity.
No difference in the percentage of MMP-9 expression between
the R and NR groups was observed (P ¼ .235; Fig. 11). Cox
models confirmed that the percentage of MMP-9 expression
was not associated with the hazards of tumor recurrence
(HR ¼ 0.99, 95% CI ¼ 0.96–1.03; P ¼ .630) or death
(HR ¼ 0.99, 95% CI ¼ 0.95–1.04; P ¼ .750).
TIMP-2. TIMP-2 expression (Fig. 7) was seen in every tumor
with a percentage of positive neoplastic cells ranging from
5.0% to 93.3% in the R group and from 22.0% to 93.0% in the
NR group (R: mean + SD ¼ 59.1% + 31.9%; Mdn ¼ 65.0%,
IQR ¼ 33.0%–93.0%; NR: mean + SD ¼ 63.1% + 26.1%;
Mdn ¼ 63.5%, IQR ¼ 43.0%–87.0%). Positive immunostain-
ing of cells was seen in the cytoplasm of both spindle and
multinucleated giant cells. TIMP-2 expression did not differ
between R and NR groups (P ¼ .886; Fig. 11). Cox models
confirmed that TIMP-2 was not a prognostic factor for recur-
rence (HR ¼ 0.99, 95% CI ¼ 0.97–1.02; P ¼ .527) or survival
(HR ¼ 0.99, 95% CI ¼ 0.96–1.01; P ¼ .287).
Ki-67. Ki-67 immunostaining was predominantly seen in mono-
nucleated cells (Figs. 8, 9) and only rarely in multinucleated
giant cells (Fig. 10). The number of Ki-67 positive nuclei ran-
ged from 4 to 38 for HPF (R: mean + SD ¼ 21.4 + 14.4; Mdn
¼ 18.7, IQR ¼ 12.4–25.0; NR: mean + SD ¼ 23.7 + 17.3;
209
Figure 12. Receiver operating characteristic curves of mitotic count
(AUC ¼ 0.882) and major diameter (AUC ¼ 0.861) for predicting
recurrence.
Mdn ¼ 26.5, IQR ¼ 8.5–43.1) but did not correlate with
mitoses/HPF (r ¼ 0.173; P ¼ .418). There was no difference
in the number of Ki-67þ nuclei between R and NR groups
(P ¼ .752). Ki-67 was not predictive of recurrence
(HR ¼ 0.99, 95% CI ¼ 0.96–1.03; P ¼ .736) or survival
(HR ¼ 1.00, 95% CI ¼ 0.95–1.05; P ¼ .975).
ROC analysis. Variables with a P value of < .05 from the bivari-
ate analyses were considered candidate variables for detecting
recurrence and submitted to ROC analysis (ie, tumor size and
mitoses, expressed as both MC value and 1–3 score). Area
under ROC for major diameter was 0.861 (95% CI ¼ 0.653–
1.000; P ¼ .007; Fig. 12) and Youden index analysis revealed
that the optimal cutoff for predicting recurrence was 3.75 cm
(sensitivity ¼ 88%, specificity ¼ 85%, accuracy ¼ 86%). We
found a lower median survival time in cats with tumors  3.75
cm than in cats with tumors < 3.75 cm (medians not reached;
log rank test: w2 ¼ 5.7, P ¼ .017; Fig. 1b). The area under ROC
for MC was 0.882 (95% CI ¼ 0.747–1.000; P ¼ .002; Fig. 12).
The optimal cutoff of MC ¼ 20 was associated with a sensi-
tivity of 70%, a specificity of 86%, and an accuracy of 80%.
Median survival time was greater in cases with MC < 20 (med-
ian not reached) than in cases with MC  20 (Mdn ¼ 994 days,
75th percentile ¼ 280 days; log rank test: w2 ¼ 5.8, P ¼ .016;
Fig. 1c). The sensitivity and specificity of MC for the predic-
tion of recurrence were 70% and 86%, respectively (AUC ¼
0.800; 95% CI ¼ 0.610–0.990; P ¼ .014), whereas accuracy
was 83%. There was no significant difference in AUC between
diameter and MC (z ¼ 0.16, P > .05) or between MC and
mitotic score (z ¼ 0.69, P > .05).
Discussion
The aim of this retrospective study was to investigate the
expression of MMP-2, MMP-9, and one of their inhibitors,
210
TIMP-2, by neoplastic cells to assess their possible role in
predicting the biological behavior of the tumor. In addition,
the prognostic value of size, STS grade, depth of infiltration,
status of excisional margins, and Ki-67 index was investigated.
MMPs and TIMPs have been recognized in numerous stud-
ies as critical factors in tumor aggressiveness.1,3,12,14,34 MMP-2
is a gelatinase that can cleave different substrates, in particular
gelatine (denatured collagen). It was expressed by tumor cells
in almost all FISSs examined, except in 2 cases, both recurrent.
However, we did not find associations between MMP-2 and
recurrence or survival. The role of MMP-2 in angiogenesis is
controversial; in fact, it can both enhance cell survival and
promote angiogenesis or induce apoptosis and inhibit neovas-
cularization.35 In our study, the endothelial expression of
MMP-2 was seen in all cases, particularly at the periphery of
the tumor, supporting the hypothesis of its pro-angiogenetic
role in FISS. A strong endothelial immunostaining was also
seen in the middle of lymphocytic aggregates, confirming its
role in leukocyte migration.37 MMP-9 is capable of degrading
almost the same substrates of MMP-2, except collagen I. Its
expression can be induced in different cell types by inflamma-
tion and its increase has been demonstrated in malignant cell
lines.24 Mmp9-deficient mice had a reduced cell-cancer prolif-
eration compared to wild-type mice, indicating that MMP-9
produces growth-promoting signals.12 As with MMP-2, no sta-
tistical difference between R and NR groups was evident. The
variable results of MMP-2 and MMP-9 could have different
explanations. First of all, the expression of MMPs can show
high variability during different cancer stages. The disappoint-
ing results obtained in clinical trials with MMP inhibitors led to
the hypothesis that MMPs could be more important in the early
steps of cancer development than in late phases.12 Moreover, in
the past few years, it has been shown that MMP function is
more complex than initially thought and goes beyond degrada-
tion of physical barriers. Indeed, MMP-2 and MMP-9 can also
exhibit tumor-suppressing effects in several circumstances.23
Considering that there are more than 20 known MMPs, MMP-2
and MMP-9 could contribute to tumor invasiveness yet not be
critical factors. Our results suggest that the only immunohisto-
chemical evaluation of MMP-2 and MMP-9 expression is not
enough in defining FISS prognosis.
The MMP inhibitor TIMP-2 was expressed in every tumor
investigated with no statistical difference between the 2 groups.
This was an unexpected finding because TIMPs regulate MMP
activity and the extent of tissue remodeling. Accumulating
evidence demonstrates that TIMPs have physiologic function
independent of their ability to block MMP activity. Particu-
larly, TIMP-2 can promote differentiation of endothelial cells
and fibroblasts,20,38 perhaps contributing to tumor progression.
The lack of metastatic tumors in this study could be due to
the fact that not all cats were evaluated with radiographs or
computed tomography before death.
The size of the R FISSs was significantly higher than the NR
ones (P < .05), confirming that a prompt intervention could be
critical in determining the prognosis. The optimal cutoff was
set at 3.75 cm. This cutoff point provides a sensitivity of 88%
Veterinary Pathology 54(2)
and a specificity of 85% in predicting tumor recurrence. How-
ever, it must be noted that this threshold refers to formalin fixed
samples, due to the lack of clinical measurement for all cases,
and therefore cannot be extrapolated to clinical measurements.
Feline injection-site sarcomas are often considered as soft-
tissue sarcomas and several studies tried to evaluate the applic-
ability of the STS grading system in prognosis assessment.15,33
According to our results, the STS grading system was not pre-
dictive of lesion recurrence (P ¼ .627). Indeed, the only statisti-
cally significant histologic parameter between the R and the NR
tumors was the mitotic score (P < .05). Hence, the number of
mitoses was further investigated using values of MC. Cox models
showed that MC was a good predictor of both recurrence (P < .01)
and mortality (P < .05). ROC curve confirmed that MC could be
considered a useful prognostic parameter to predict FISS recur-
rence (P < .01) at the optimal cutoff of MC ¼ 20. Patients with
MC > 20 had a higher risk of recurrence and a lower median
survival time. Necrosis and degree of differentiation were not
significantly different between the 2 groups. Our data suggest that
they should not be considered useful morphologic data in prog-
nosis assessment, as already hypothesized in other studies.15
No significant difference was observed between FISSs with
I and NI margins. Cases where neoplastic cells were continuous
with at least 1 surgical margin in any plane were classified as
‘‘I.’’11 Instead, tumors with ‘‘close’’ margins and ‘‘complete’’
margins were grouped together as NI, considering the small
sample of this study. Our results seem to show that this 2-tier
classification is not useful as a prognostic marker and to con-
firm that a wide excisional margin is recommended.
The Ki-67 index did not show statistical differences between
the R and NR groups and the percentage of positive nuclei did
not correlate with MC. The independence of Ki-67 fraction and
MC has been recognized in different studies in both human and
veterinary medicine.20,43
In conclusion, MMP-2, MMP-9, and TIMP-2 seem to be
important in FISS behavior because of their high expression,
but none of them is sufficient as a unique prognostic marker.
On the other hand, our study suggests that size (major diameter)
and the MC should be taken into consideration as prognostic
markers in the assessment of a useful grading system for FISS.
Data from our study suggest 3.75 cm (major diameter of the
formalin fixed tumor) and MC of 20 as threshold values to
predict recurrence and define a prognosis in FISS. Since this
is a retrospective study on a relatively small caseload, further
investigations are needed to better assess additional useful
prognostic markers.
Authors’ Note
This article was prepared in the Uniform Requirements format. For
any additional inquiry on research materials, please contact the first
author via e-mail: ilariaporcellato@yahoo.it.
Acknowledgements
We would like to thank Dr Giovanni Angeli, Dr Emilia del Rossi, Dr
Marica Stazi, and Luca Stefanelli for their precious technical
assistance.
Porcellato et al
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to
the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, author-
ship, and/or publication of this article.
References
1. Akkoc A, Inan S, Sonmez G. Matrix metalloproteinase (MMP-2 and MMP-9)
and steroid receptor expressions in feline mammary tumors. Biotech Histochem.
2012;87(4):312–319.
2. Aupperle H, Baldauf K, März I. An immunohistochemical study of feline myo-
cardial fibrosis. J Comp Pathol. 2011;145(2–3):158–173.
3. Benassi MS, Gamberi G, Magagnoli G, et al. Metalloproteinase expression and
prognosis in soft tissue sarcomas. Ann Oncol. 2001;12(1):75–80.
4. Björklund M, Koivunen E. Gelatinase-mediated migration and invasion of can-
cer cells. Biochim Biophys Acta. 2005;1755(1):37–69.
5. Bray J, Polton G. Neoadjuvant and adjuvant chemotherapy combined with
anatomical resection of feline injection-site sarcoma: results in 21 cats. Vet
Comp Oncol. 2016;14(2):147–160.
6. Carminato A, Vascellari M, Marchioro W, Melchiotti E, Mutinelli F.
Microchip-associated fibrosarcoma in a cat. Vet Dermatol. 2011;22(6):
565–569.
7. Chakrabarti S, Patel KD. Matrix metalloproteinase-2 (MMP-2) and MMP-9 in
pulmonary pathology. Exp Lung Res. 2005;31(6):599–621.
8. Coindre JM. Grading of soft tissue sarcomas: review and update. Arch Pathol
Lab Med. 2006;130(10):1448–1453.
9. Couto SS, Griffey SM, Duarte PC, Madewell BR. Feline vaccine-associated
fibrosarcoma: morphologic distinctions. Vet Pathol. 2002;39(1):33–41.
10. Dean RS, Pfeiffer DU, Adams VJ. The incidence of feline injection site sarco-
mas in the United Kingdom. BMC Vet Res. 2013;9:17.
11. Dennis MM, McSporran KD, Bacon NJ, Schulman FY, Foster RA, Powers BE.
Prognostic factors for cutaneous and subcutaneous soft tissue sarcomas in dogs.
Vet Pathol. 2011;48(1):73–84.
12. Egeblad M, Werb Z. New functions for the matrix metalloproteinases in cancer
progression. Nat Rev Cancer. 2002;2(3):161–174.
13. Field A. Discovering Statistics Using SPSS. London, UK: SAGE; 2009.
14. Giantin M, Aresu L, Benali S, et al. Expression of matrix metalloproteinases,
tissue inhibitors of metalloproteinases and vascular endothelial growth factor in
canine mast cell tumours. J Comp Pathol. 2012;147(4):419–429.
15. Giudice C, Stefanello D, Sala M, et al. Feline injection-site sarcoma: recurrence,
tumour grading and surgical margin status evaluated using the three-
dimensional histological technique. Vet J. 2010;186(1):84–88.
16. Greiner M, Pfeiffer D, Smith RD. Principles and practical application of the
receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med.
2000;45(1–2):23–41.
17. Gross TL, Ihrke PJ, Walder EJ, et al. Skin Diseases of the Dog and Cat. 2nd ed.
Oxford, UK: Blackwell; 2005.
18. Hendrick MJ, Brooks JJ. Postvaccinal sarcomas in the cat: histology and immu-
nohistochemistry. Vet Pathol. 1994;31(1):126–129.
19. Hershey AE, Sorenmo KU, Hendrick MJ, Shofer FS, Vail DM. Prognosis for
presumed feline vaccine-associated sarcoma after excision: 61 cases (1986-
1996). J Am Vet Med Assoc. 2000;216(1):58–61.
20. Jalava P, Kuopio T, Juntti-Patinen L, Kotkansalo T, Kronqvist P, Collan Y.
Ki67 immunohistochemistry: a valuable marker in prognostication but with a
risk of misclassification: proliferation subgroups formed based on Ki67 immu-
noreactivity and standardized mitotic index. Histopathology. 2006;48(6):
674–682.
211
21. Kamstock DA, Ehrhart EJ, Getzy DM, et al. Recommended guidelines for
submission, trimming, margin evaluation, and reporting of tumor biopsy speci-
mens in veterinary surgical pathology. Vet Pathol. 2011;48(1):19–31.
22. Kass PH, Barnes WG Jr, Spangler WL, Chomel BB, Culbertson MR. Epide-
miologic evidence for a causal relation between vaccination and fibrosarcoma
tumorigenesis in cats. J Am Vet Med Assoc. 1993;203(3):396–405.
23. Kessenbrock K, Plaks V, Werb Z. Matrix metalloproteinases: regulators of the
tumor microenvironment. Cell. 2010;141(1):52–67.
24. Klein T, Bischoff R. Physiology and pathophysiology of matrix metallopro-
teases. Amino Acids. 2011;41(2):271–290.
25. Martano M, Morello E, Buracco P. Feline injection-site sarcoma: past, present
and future perspectives. Vet J. 2011;188(2):136–141.
26. Martano M, Morello E, Iussich S, Buracco P. A case of feline injection-site
sarcoma at the site of cisplatin injections. J Feline Med Surg. 2012;14(10):
751–754.
27. McEntee MC, Page RL. Feline vaccine-associated sarcomas. J Vet Intern Med.
2001;15(3):176–182.
28. Meuten DJ, Moore FM, George JW. Mitotic count and the field of view area:
time to standardize. Vet Pathol. 2016;53(1):7–9.
29. Moore CS, Crocker SJ. An alternate perspective on the roles of TIMPs and
MMPs in pathology. Am J Pathol. 2012;180(1):12–16.
30. Nagase H, Visse R, Murphy G. Structure and function of matrix metalloprotei-
nases and TIMPs. Cardiovasc Res. 2006;69(3):562–573.
31. Petznick A, Madigan MC, Garrett Q, Sweeney DF, Evans MD. Contributions of
ocular surface components to matrix-metalloproteinases (MMP)-2 and MMP-9
in feline tears following corneal epithelial wounding. PLoS One. 2013;8(8):
e71948.
32. Porcellato I, Giontella A, Mechelli L, Del Rossi E, Brachelente C. Feline eosi-
nophilic dermatoses: a retrospective immunohistochemical and ultrastructural
study of extracellular matrix remodelling. Vet Dermatol. 2014;25(2):86–94.
33. Romanelli G, Marconato L, Olivero D, Massari F, Zini E. Analysis of prognos-
tic factors associated with injection-site sarcomas in cats: 57 cases (2001-2007).
J Am Vet Med Assoc. 2008;232(8):1193–1199.
34. Roomi MW, Kalinovsky T, Monterrey J, Rath M, Niedzwiecki A. In vitro
modulation of MMP-2 and MMP-9 in adult human sarcoma cell lines by cyto-
kines, inducers and inhibitors. Int J Oncol. 2013;43(6):1787–1798.
35. Shapiro S, Khodalev O, Bitterman H, Auslender R, Lahat N. Different activa-
tion forms of MMP-2 oppositely affect the fate of endothelial cells. Am J
Physiol Cell Physiol. 2010;298(4):C942–C951.
36. Sorensen KC, Kitchell BE, Schaeffer DJ, Mardis PE. Expression of matrix
metalloproteinases in feline vaccine site-associated sarcomas. Am J Vet Res.
2004;65(3):373–379.
37. Stefanidakis M, Koivunen E. Cell-surface association between matrix metallo-
proteinases and integrins: role of the complexes in leukocyte migration and
cancer progression. Blood. 2006;108(5):1441–1450.
38. Stetler-Stevenson WG. The tumor microenvironment: regulation by MMP-
independent effects of tissue inhibitor of metalloproteinases-2. Cancer Metas-
tasis Rev. 2008;27(1):57–66.
39. Stetler-Stevenson WG, Gavil NV. Normalization of the tumor microenviron-
ment: evidence for tissue inhibitor of metalloproteinase-2 as a cancer therapeu-
tic. Connect Tissue Res. 2014;55(1):13–19.
40. Swets JA. Measuring the accuracy of diagnostic systems. Science. 1988;
240(4857):1285–1293.
41. Tamamoto T, Ohno K, Takahashi M, et al. Plasma matrix metalloproteinase-9
activity in cats with lymphoma [published online June 5, 2014]. Vet Comp
Oncol.
42. Trojani M, Contesso G, Coindre JM, et al. Soft-tissue sarcomas of adults; study
of pathological prognostic variables and definition of a histopathological grad-
ing system. Int J Cancer. 1984;33(1):37–42.
43. van Lelyveld S, Warland J, Miller R, et al. Comparison between Ki-67 index
and mitotic index for predicting outcome in canine mast cell tumours. J Small
Anim Pract. 2015;56(5):312–319.