Model Predicts Follicular
Lymphoma Survival
National Cancer Institute (NCI) researchers have developed a model to predict
survival of patients with follicular lymphoma based on the genetic "signatures"
of their tumors at diagnosis. According to the model, the activity of two sets
of genes - termed "survival-associated signatures" by lead researcher Dr. Louis
Staudt and colleagues - was associated with either more aggressive forms of the
cancer and shorter survival times, or slower moving forms of the cancer and
longer survival times.
The findings, published in the Nov. 18 New England
Journal of Medicine, could have implications for treatment of follicular
lymphoma. Survival among follicular lymphoma patients varies dramatically,
explains Dr. Staudt, a principal investigator in the NCI Center for Cancer
Research Metabolism Branch. "Understanding the molecular causes of such
differences in survival could provide a more accurate method to determine
patient risk," Dr. Staudt says, "that could be used to guide treatment and may
suggest new therapeutic approaches."
To perform gene expression profiles for
this study, researchers used DNA microarray analysis, a method for quickly
scanning thousands of genes for activity in a tumor sample. The researchers
used the Lymphochip - a glass chip with DNA "spots" on it from approximately
18,500 genes expressed in lymph tissue - created in Dr. Staudt's laboratory to
study lymphoid cancers.
Researchers analyzed follicular lymphoma biopsies of 191
patients before treatment; biopsies came from institutions participating in the
NCI-sponsored Lymphoma/Leukemia Molecular Profiling Project. After biopsy, all
patients received standard treatments; subsequent medical records were examined
to determine survival. The Lymphochip was used to determine which genes were
active in the first group of 95 tumor biopsies (the "training set") and at what
levels; researchers then determined which of these genes were statistically
associated with survival. Next, researchers identified subsets of good- and
bad-prognosis genes that tended to be expressed together; these subsets
constituted the survival- associated signatures. In the remaining 96 samples
(the "test set"), two signatures - indicating poor and good prognosis - had strong
synergy and together predicted survival better than any other model tested.
Unexpectedly, both came from nonmalignant immune cells that infiltrate the
tumors.
Based on the two-signature model, the NCI team divided patients into
four equal groups with average survival rates of 3.9, 10.8, 11.1, and 13.6
years. For the 75 percent of patients with survival rates of 10 years or
longer, "watchful waiting is appropriate," Dr. Staudt says. "On the other hand,
those patients in the group with the lowest survival rate should be considered
for newer treatments and clinical trials."
That the most predictive signatures
came from immune cells suggests an important interplay between the host immune
system and malignant cells in follicular lymphoma. "One possibility is that
immune cells with the good-prognosis signature are attacking the lymphoma and
keeping it in check," he suggests. "Alternately, these immune cells may provide
signals that encourage the cancer cells not to leave the lymph node, preventing
or delaying the spread of the cancer."
In 2002, Dr. Staudt's group published a
study on a similar model identifying a single 17-gene signature that predicted
patient survival for diffuse large B-cell lymphoma (DLBCL). This model will be
used in a phase III trial testing the current standard of care for untreated
DLBCL against a new regimen. Patient biopsies will undergo gene expression
profiling to determine what tumor features influence patient response to the
therapies.
|