Background. Several investigators have reported the ability to establi
sh xenografts in nude mice from children with neuroblastomas, but a co
rrelation of prognosis with this establishment and the growth patterns
of the neuroblastomas has not been reported. Methods. Tumor specimens
from 58 children with neuroblastomas were heterotransplanted into BAL
B/c nude mice. In 34 patients, heterotransplantation was done before t
herapy; in 24 patients, tumors were obtained after at least one course
of chemotherapy or radiation therapy. The histology, cytogenetics, an
d growth characteristics of serial passages of the xenografts were stu
died. Results. The engraftment rate was 34%. Neuroblastomas with diplo
id chromosome numbers did not engraft. Chromosomal abnormalities invol
ving 1p were seen in more than 50% of the xenografts. Cytogenetic feat
ures were retained between original tumors and resultant xenografts. X
enografts could be established only from tumors with unfavorable histo
logy, as defined by Shimada classification criteria. The histology of
each xenograft line was strikingly similar, and each was highly undiff
erentiated. Engraftment rates, doubling times, and lag times did not v
ary appreciably between xenografts established from treated tumors com
pared with xenografts established from untreated tumors. There was no
correlation between doubling or lag times and prognosis. Patients whos
e tumors engrafted had only a 5% 3-year survival rate. Conclusions. Fr
om these results, it appears that successful engraftment is the most i
mportant prognostic indicator for patients with neuroblastomas. Becaus
e of the commonality of the histologic features and the stability of t
he tumor clones from patients before and after heterotransplantation,
these xenografts may be useful as an in vivo model for studying drug r
esistance and for designing treatment regimens.