|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 25, No 27 (September 20), 2007: pp. 4324-4325 © 2007 American Society of Clinical Oncology. DOI: 10.1200/JCO.2007.12.3919
In ReplyDepartment of Neurology, Memorial Sloan-Kettering Cancer Center, New York, NY
Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, NY
Department of Radiation Oncology Memorial Sloan-Kettering Cancer Center, New York, NY
Department of Radiation Oncology, Thomas Jefferson University, Philadelphia, PA
Department of Radiation Oncology, Medical College of Wisconsin, Milwaukee, WI
Medical Director, Stanford Comprehensive Cancer Center, Palo Alto, CA
Division of Radiation Oncology, Mayo Clinic, Rochester, MN
Department of Radiation Oncology, Medical College of Wisconsin, Milwaukee, WI
Department of Neurology, Memorial Sloan-Kettering Cancer Center, New York, NY We would like to thank Drs Ferreri and colleagues for their interest in our prognostic model for primary CNS lymphoma (PCNSL). When we initially analyzed our patients we sought to confirm the International Extranodal Lymphoma Study Group (IELSG) PCNSL prognostic score, but our data did not support this model.1 The IELSG score was derived from retrospective data collected by questionnaire from 23 sites representing patients treated over a 19-year interval.2 Unlike their retrospective information, we could confirm the primary source data for all patients included in our analysis including review of films to ascertain the location of all PCNSL lesions. Furthermore, our follow-up extended for more than 15 years providing mature data in nearly all patients. Our analysis of this well-studied population could not confirm the three variables other than age and performance status suggested by the IELSG score. We did not seek to eliminate these variables; the data did not support their inclusion. Furthermore, only one of these three variables, deep brain location, has ever been confirmed as having prognostic significance by another outside group.3 This raises significant concern about the relative value and applicability of their other variables: serum lactate dehydrogenase and CSF total protein. In an effort to optimize the prognostic value provided by age and Karnofsky performance status, we used two different statistical analyses. The first was a cut point analysis to determine which age was associated with the greatest prognostic import. Most previous publications citing age as a prognostic value for PCNSL, including our own, used 60 as the critical age without any specific substantiation. To our knowledge, this was the first analysis to evaluate age critically as a continuous variable. Recursive partitioning analysis (RPA) was then used to derive three simple but statistically robust prognostic subgroups. The value of RPA analysis is the ability to handle complex interactions of continuous and discrete variables while retaining conceptual simplicity. Further recognizing that our patients had received heterogeneous treatment (as did the IELSG population) and were collected retrospectively, we then confirmed our results using an outside prospective data set from the Radiation Therapy Oncology Group. Since publication of our prognostic model, we have had a second group confirm our results in 175 homogeneously treated patients.4 Therefore, we would assert that the derivation of our model was adequately complex and that the strength of our model is the simplicity that allows widespread application for patients, practitioners, and interpretation/analysis of clinical data. We would like to address some of the specific concerns raised regarding the patient population used in our analysis. All patients included were studied from initial diagnosis of PCNSL. Furthermore, patients with evidence of systemic disease were excluded from the RPA analysis, and including or excluding these patients did not significantly change the results of the multivariate or cut point analysis. The majority of patients who did not have initial histologic confirmation of their PCNSL were confirmed at later biopsy or autopsy so we did not exclude these patients from analysis. It is possible that patients with primary intraocular lymphoma (PIOL) represent a unique subtype of lymphoma; however, to date PIOL has been considered a variant of PCNSL and patients with PIOL are often included in clinical trials and reported data. Furthermore, excluding these 22 PIOL patients did not alter our results (Fig 1).
We have several concerns regarding the data presented by Ferreri et al. It is not clear to us why they did not analyze all 378 patients in their data set using our score. It is not surprising that a retrospective group of 105 patients with short median follow-up (26 months) and heterogeneous treatment may not confirm our results. It is not valid to apply the Memorial Sloan-Kettering Cancer Center (New York, NY) prognostic score in a subset analysis of patients already stratified by the IELSG criteria as these authors show in Figures 1A to 1D. Figure 1B is not significant, Figure 1A has no P value included and Figures 1C and 1D have too few patients included to draw any statistical conclusions. The results presented from the Methotrexate, Cytarabine, Thiotepa, Idarubicin, Leucovorin, Dexamethasone (MATILDE) trial have incomplete data for one third of the patients and the sample size reported has insufficient power to support their conclusions.5 Their assertion that the five variables used in the IELSG model can be successfully collected in prospective clinical trials is not supported by their own trial where nearly one third of patients did not have a full collection of variables. The optimal prognostic score must contain adequate criteria to minimize variability within a subset without having so many parameters resulting in an inherent bias. Furthermore, the criteria included in any score should have independent confirmation of their disease-specific value. It is our hope that we can collaborate with the IELSG and other groups via the International PCNSL Collaborative Group in a productive way to derive an appropriate and useful model for PCNSL that will have widespread applicability and value in the management of our patients and conduct of future clinical trials. AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest.
REFERENCES
1. Abrey LE, Ben-Porat L, Panageas KS, et al: Primary central nervous system lymphoma: The Memorial Sloan-Kettering Cancer Center prognostic model. J Clin Oncol 24:5711-5715, 2006 2. Ferreri AJ, Blay JY, Reni M, et al: Prognostic scoring system for primary CNS lymphomas: The International Extranodal Lymphoma Study Group experience. J Clin Oncol 21:266-272, 2003 3. Jahnke K, Thiel E, Martus P, et al: Relapse of primary central nervous system lymphoma: Clinical features, outcome and prognostic factors. J Neurooncol 80:159-165, 2006[CrossRef][Medline] 4. Angelov L, Doolittle ND, Siega T, et al: Blood brain barrier disruption (BBBD) and intra-arterial methotrexate based therapy for newly diagnosed primary CNS lymphoma: The BBBD Consortium experience. J Clin Oncol 25:78s, 2007 (abstr 2014) 5. Ferreri AJ, Dell'Oro S, Foppoli M, et al: MATILDE regimen followed by radiotherapy is an active strategy against primary CNS lymphomas. Neurology 99:1435-1438, 2006
Related Correspondence
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2007 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
50, class 1; age older than 50 plus Karnofsky performance status (KPS) 
70, class 2; age > 50 plus KPS lower than 70, class 3.
