OBJECTIVE: To evaluate the dose-response relationship of nuclear abnor
malities in tumor cells collected by serial scrape smears from oral ca
ncer patients on fractionated radiotherapy. STUDY DESIGN: The study in
cluded 31 patients with squamous cell carcinoma of the oral cavity tre
ated by radiotherapy (60 Gy in 25 fractions; 2.4 Gy per fraction). Ser
ial scrape smears were taken from each tumor before treatment and afte
r delivery of various fractions, usually 2 (4.8 Gy), 5 (12.0 Gy), 8 (1
9.2 Gy) or 12 (28.8 Gy). The smears were stained by Giemsa stain and e
valuated by light microscopy, and the number of micronucleated, binucl
eated, nuclear budded and multinucleated cells were scored. Their rela
tion to cumulative dose teas analyzed by Kruskal-Wallis one-way analys
is of variance. The results were expressed in terms of 1,000 mononucle
ated cells. RESULTS: Even before treatment, most of the tumors showed
various abnormally nucleated cells, and, despite the high intertumoral
variation (ns indicated by the high variance), all of them showed sta
tistically significant dose-related increases. The mean values before
treatment and after irradiation with 28.8 Gy, respectively, were 2.8 a
nd 19.5 (P<,0001) for micronucleated cells, 1.5 and 8.5 (P<.000001) fo
r nuclear budded cells, 8.2 and 35.5 (P<.0001) for binucleated cells,
and 3.7 and 16.8 (P<.0001) for multinucleated cells. When the differen
t types of nuclear abnormalities were combined and analyzed as ''abnor
mally nucleated cells,'' the mean count before treatment and after 28.
8 Gy were 7.9 and 44.9 (P<.00001), respectively. CONCLUSION: The study
showed that radiation-induced micronucleation, multinucleation, binuc
leation and nuclear budding in ol nl cancer cells has statistically si
gnificant dose-related increases that become evident in the initial fe
w days of radiotherapy and that they can be differentiated well by cyt
ology. This dose-response relationship and the high intertumoral varia
tions suggest that serial assay of these changes has potential use for
radiosensitivity prediction.