Objectives: To establish an orthotopic murine floor-of-mouth cancer mo
del of the analysis of the role of proteases such as urokinase-type pl
asminogen activator (u-PA) and the matrix metalloprotease MMP-9 (MMP-9
) in in vivo invasion. Study Design: Randomized, prospective animal st
udy. Methods: Two human squamous cell carcinoma cell lines, UM-SCC-1 a
nd 022, were assayed via zymography for their in vitro secretion level
s of u-PA and MMP-9. Both cell lines (5 x 10(6) cells) were injected i
nto the cervical subcutaneous tissues of female athymic nude (nu/nu) m
ice superficial to the mylohyoid muscle. Mice were sacrificed after 30
days, and tumor invasion characteristics were histologically compared
. Additional mice were then inoculated with invasive UM-SCC-1 cells an
d sacrificed 10, 30, and 40 days after inoculation to identify distinc
t stages of invasion. Results: In vitro secretion levels of MMP-9 and
activity of u-PA were higher in UM-SCC-1 cells than in 022 cells. In t
he in vivo studies, tumors formed from 022 cells were found to be noni
nvasive, whereas tumors derived from UM-SCC-1 cells progressed through
distinct and readily identifiable histologic stages of invasion. Thes
e stages included invasion of adjacent muscle layers (mylohyoid, genio
hyoid, and genioglossus muscles) and of associated structures (blood v
essels, bone, nerve, and regional lymph nodes). A staging system was d
evised accordingly. Conclusion: We developed an in vivo quantitative c
ancer invasion model that allows determination of the effect of the ex
pression and activity levels of the proteases MMP-9 and u-PA. Tumor in
vasion occurred in an orderly and stepwise fashion involving muscles a
nd related vascular, nervous, and bony structures of the floor of the
mouth and tongue. This orderly invasion allowed the development of a s
taging system. We anticipate that this model will have wide applicabil
ity in the study of in vivo tumor response to a variety of novel thera
peutic approaches.