|
|
Primary Central Nervous System
Lymphoma: The Memorial Sloan-Kettering Cancer Center Prognostic Model
Lauren E. Abrey, Leah
Ben-Porat, Katherine S. Panageas, Joachim
Yahalom, Brian Berkey, Walter Curran,
Christopher Schultz, Steven Leibel, Diana
Nelson, Minesh Mehta, Lisa M. DeAngelis
From the Departments of Neurology,
Epidemiology and Biostatistics, Radiation Oncology, Memorial
Sloan-Kettering Cancer Center, New York, NY; Radiation Therapy
Oncology Group; Department of Radiation Oncology, Thomas Jefferson
University, Philadelphia, PA; Department of Radiation Oncology,
Medical College of Wisconsin, Milwaukee, WI; Stanford Comprehensive
Cancer Center, Stanford, CA; and the Division of Radiation Oncology,
Mayo Clinic, Rochester, MN -- Address reprint requests to Lauren E.
Abrey, MD, Department of Neurology, Memorial Sloan-Kettering Cancer
Center, 1275 York Ave, New York, NY 10021; e-mail: abreyl@mskcc.org
|
|
|
Purpose. The purpose of this
study was to analyze prognostic factors for patients with
newly diagnosed primary CNS lymphoma (PCNSL) in order to
establish a predictive model that could be applied to the
care of patients and the design of prospective clinical trials.
Patients and Methods.
Three hundred thirty-eight consecutive patients with newly diagnosed
PCNSL seen at Memorial Sloan-Kettering Cancer Center (MSKCC;
New York, NY) between 1983 and 2003 were analyzed.
Standard univariate and multivariate analyses were
performed.
In addition, a formal cut point analysis was used to
determine the most statistically significant cut point for
age.
Recursive partitioning analysis (RPA) was used to create
independent prognostic classes.
An external validation set obtained from three prospective
Radiation Therapy Oncology Group (RTOG) PCNSL clinical
trials was used to test the RPA classification.
Results. Age and
performance status were the only variables identified on
standard multivariate analysis.
Cut point analysis of age determined that patients age
≤ 50 years had significantly improved outcome
compared with older patients.
RPA of 282 patients identified three distinct prognostic
classes: class 1 (patients < 50 years), class 2
(patients ≥50; Karnofsky performance score [KPS] ≥
70) and class 3 (patients ≥ 50; KPS < 70).
These three classes significantly distinguished outcome
with regard to both overall and failure-free
survival.
Analysis of the RTOG data set confirmed the validity of
this classification.
Conclusion. The MSKCC
prognostic score is a simple, statistically powerful model
with universal applicability to patients with newly diagnosed PCNSL.
We recommend that it be adopted for the management of newly
diagnosed patients and incorporated into the design of prospective
clinical trials.
|
