mipi blood 2008

doi:10.1182/blood-2007-06-095331 Prepublished online Oct 25, 2007;2008 111: 558-565

Network German Low Grade Lymphoma Study Group (GLSG) and the European Mantle Cell LymphomaEimermacher, Hannes Wandt, Joerg Hasford, Wolfgang Hiddemann, Michael Unterhalt and for the Peter, Wolfram Jung, Bernhard Wörmann, Wolf-Dieter Ludwig, Ulrich Dührsen, HartmutKluin-Nelemans, Michael Pfreundschuh, Marcel Reiser, Bernd Metzner, Hermann Einsele, Norma Eva Hoster, Martin Dreyling, Wolfram Klapper, Christian Gisselbrecht, Achiel van Hoof, Hanneke C.

cell lymphomaA new prognostic index (MIPI) for patients with advanced-stage mantle

http://bloodjournal.hematologylibrary.org/cgi/content/full/111/2/558Updated information and services can be found at:

(2962 articles)Clinical Trials and Observations � collections: BloodArticles on similar topics may be found in the following

http://bloodjournal.hematologylibrary.org/misc/rights.dtl#repub_requestsInformation about reproducing this article in parts or in its entirety may be found online at:

http://bloodjournal.hematologylibrary.org/misc/rights.dtl#reprintsInformation about ordering reprints may be found online at:

http://bloodjournal.hematologylibrary.org/subscriptions/index.dtlInformation about subscriptions and ASH membership may be found online at:

. Hematology; all rights reservedCopyright 2007 by The American Society of 200, Washington DC 20036.semimonthly by the American Society of Hematology, 1900 M St, NW, Suite Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published

For personal use only. by on October 23, 2010. www.bloodjournal.orgFrom

http://bloodjournal.hematologylibrary.org/cgi/content/full/111/2/558

http://bloodjournal.hematologylibrary.org/cgi/collection/clinical_trials_and_observations

http://bloodjournal.hematologylibrary.org/misc/rights.dtl#repub_requests

http://bloodjournal.hematologylibrary.org/misc/rights.dtl#reprints

http://bloodjournal.hematologylibrary.org/subscriptions/index.dtl

http://bloodjournal.hematologylibrary.org/subscriptions/ToS.dtl

http://bloodjournal.hematologylibrary.org


CLINICAL TRIALS AND OBSERVATIONS

A new prognostic index (MIPI) for patients with advanced-stage mantle celllymphomaEva Hoster,1,2 Martin Dreyling,1 Wolfram Klapper,3 Christian Gisselbrecht,4 Achiel van Hoof,5 Hanneke C. Kluin-Nelemans,6

Michael Pfreundschuh,7 Marcel Reiser,8 Bernd Metzner,9 Hermann Einsele,10 Norma Peter,11 Wolfram Jung,12

Bernhard Wormann,13 Wolf-Dieter Ludwig,14 Ulrich Duhrsen,15 Hartmut Eimermacher,16 Hannes Wandt,17 Joerg Hasford,2

Wolfgang Hiddemann,1 and Michael Unterhalt,1 for the German Low Grade Lymphoma Study Group (GLSG) and theEuropean Mantle Cell Lymphoma Network

1Department of Internal Medicine III, and 2Department of Medical Informatics, Biometry, and Epidemiology, University of Munich, Munich, Germany; 3Institute ofPathology, Lymph Node Registry, University of Kiel, Kiel, Germany; 4Department of Hematology, Hopital Saint Louis, Paris, France; 5Department of Hematology,Akademisch Ziekenhuis St-Jan, Bruges, Belgium; 6Department of Hematology, University Medical Center Groningen, University of Groningen, Groningen, theNetherlands; 7Department of Internal Medicine I, Saarland University Medical School, Homburg, Germany; 8Department of Internal Medicine I, University ofCologne, Cologne, Germany; 9Department of Oncology and Hematology, Klinikum Oldenburg, Oldenburg, Germany; 10Department of Internal Medicine II,University Hospital of Wurzburg, Wurzburg, Germany; 11Department of Hematology/Oncology, Carl-Thiem Hospital, Cottbus, Germany; 12Department ofHematology and Oncology, University of Gottingen, Gottingen, Germany; 13Stadtisches Klinikum Braunschweig, Braunschweig, Germany; 14Department ofHematology and Oncology, Helios Klinikum Berlin-Buch, Berlin, Germany; 15Department of Hematology, University Hospital Essen, Essen, Germany;16Department of Hematology and Oncology, Katholisches Krankenhaus Hagen, Hagen, Germany; and 17Medical Department 5, Klinikum Nurmberg Nord,Nuremberg, Germany

There is no generally established prog-nostic index for patients with mantle celllymphoma (MCL), because the Interna-tional Prognostic Index (IPI) and Follicu-lar Lymphoma International PrognosticIndex (FLIPI) have been developed fordiffuse large cell and follicular lymphomapatients, respectively. Using data of455 advanced stage MCL patients treatedwithin 3 clinical trials, we examined theprognostic relevance of IPI and FLIPI andderived a new prognostic index (MCLinternational prognostic index, MIPI) of

overall survival (OS). Statistical methodsincluded Kaplan-Meier estimates and thelog-rank test for evaluating IPI and FLIPIand multiple Cox regression for develop-ing the MIPI. IPI and FLIPI showed poorseparation of survival curves. Accordingto the MIPI, patients were classified intolow risk (44% of patients, median OS notreached), intermediate risk (35%,51 months), and high risk groups (21%,29 months), based on the 4 independentprognostic factors: age, performance sta-tus, lactate dehydrogenase (LDH), and

leukocyte count. Cell proliferation (Ki-67)was exploratively analyzed as an impor-tant biologic marker and showed strongadditional prognostic relevance. The MIPIis the first prognostic index particularlysuited for MCL patients and may serve asan important tool to facilitate risk-adaptedtreatment decisions in patients with ad-vanced stage MCL. (Blood. 2008;111:558-565)

© 2008 by The American Society of Hematology

Introduction

Mantle cell lymphoma (MCL) is a relatively rare lymphoma entityaccounting for approximately 3% to 6% of all non-Hodgkin lymphoma(NHL) cases.1-3 It has a poor prognosis with reported median overallsurvival (OS) of only 30 to 43 months.1,2,4 In the late 1980s, it wascontroversially debated whether centrocytic lymphoma, as defined bythe Kiel classification,5 represented a separate lymphoma entity andwhether it corresponded to the intermediate lymphocytic lymphoma orlymphocytic lymphoma of intermediate differentiation of the WorkingFormulation.6 After detection of the characteristic translocation t(11;14)associated with those lymphoma subtypes and based on furthermorphologic analyses, Banks et al7 proposed in 1992 the term “mantlecell lymphoma” for this now widely accepted entity of malignantlymphomas.

Since the international acceptance of MCL in the revised European-American lymphoma (REAL) classification8 in 1994 many reports on

clinicopathologic features, treatment, clinical course, and prognosticfactors for patients with MCL have been published.1,2,4,9-17 Treatmentresults have overall been unsatisfactory, although a substantial variationin outcome was noted among individual cases with a small fraction ofpatients even achieving long lasting remissions. Unlike the InternationalPrognostic Index18 (IPI) for diffuse large B-cell lymphomas or theFollicular Lymphoma International Prognostic Index19 (FLIPI) forfollicular lymphomas, no broadly accepted prognostic index has beendefined for MCL so far. Previous reports revealed serious limitationsapplying the IPI and FLIPI to MCL patients,1,2,4,9-14,16,17,20 and thedevelopment of a prognostic index specific to MCL patients has beenpostulated.12,20 In addition, controversial results have been publishedconcerning the prognostic value of several candidate prognostic factors,but results were mostly based on limited samples of fewer than130 patients.

Submitted June 16, 2007; accepted September 27, 2007. Prepublished onlineas Blood First Edition paper, October 25, 2007; DOI 10.1182/blood-2007-06-095331.

The online version of this article contains a data supplement.

Presented in part in oral form at the 48th annual meeting of the American

Society of Hematology, Orlando, FL, December 12, 2006.

The publication costs of this article were defrayed in part by page chargepayment. Therefore, and solely to indicate this fact, this article is herebymarked ‘‘advertisement’’ in accordance with 18 USC section 1734.

© 2008 by The American Society of Hematology

558 BLOOD, 15 JANUARY 2008 � VOLUME 111, NUMBER 2




We analyzed the pooled data of the most recent randomized trials ofthe German Low Grade Lymphoma Study Group (GLSG) and theEuropean MCL Network to validate the IPI, FLIPI, and other prognosticfactors, and to develop, if needed, a new prognostic index that isparticularly suited for patients with advanced stage MCL.

Methods

Patients

Data from 3 consecutive randomized trials, GLSG1996,21 GLSG2000,22

and European MCL Trial 1,23 which were performed between 1996 and2004, were included in this analysis. Local ethics committees of theparticipating centers approved the study protocol, and written informedconsent was obtained from all patients in accordance with the Declarationof Helsinki. Patient entry criteria of all trials were identical with theexception of age, which was limited up to 65 years in the European MCLtrial. Otherwise all 3 trials recruited treatment-naive patients with advancedAnn Arbor stage III or IV MCL. Central pathology review was mandatoryto ensure a high quality of histologic diagnosis. Exclusion criteriacomprised serious concomitant diseases, bad performance status, or signifi-cant impairment of organ function.21-23 The initial diagnostic work-upcomprised a bone marrow biopsy, ultrasound examination of the abdomen,and computed tomography (CT) scans of chest and abdomen. Normal organfunction was assured by the respective laboratory tests, as well as byechocardiograms and electrocardiograms.

Treatment protocols

The GLSG1996 trial21 compared the response rates of initial chemotherapywith CHOP (cyclophosphamide, hydroxy-daunorubicin, vincristine andprednisone) and MCP (mitoxantrone, chlorambucil, prednisone). In 2000,the GLSG2000 trial22 was initiated to assess the additional efficacy ofincorporating the monoclonal CD20-antibody rituximab to primary CHOPin terms of time-to-treatment failure. In both GLSG trials, patients olderthan 65 years and responding to initial chemotherapy received interferon-�(IFN�) maintenance therapy. Younger patients were randomized betweenconsolidating myeloablative radiochemotherapy followed by autologousstem cell transplantation (ASCT) and IFN� maintenance within theEuropean MCL Trial 1.23 Detailed study design and results of the respectivetrials have been published previously.21-23

Candidate prognostic factors and outcome parameters

All baseline variables were considered as candidate prognostic factors,namely age, sex, Eastern Cooperative Oncology Group (ECOG) perfor-mance status,24 Ann Arbor stage, presence of B-symptoms, bone marrowinvolvement, spleen involvement, number of involved nodal areas, numberof extranodal sites, size of largest involved lymph node, and laboratoryparameters such as white blood cell (WBC), lymphocyte, granulocyte,monocyte, and platelet counts, hemoglobin (Hb), lactate dehydrogenase(LDH), albumin, and serum �2-microglobulin levels. Because upper limitsof normal (ULN) differed considerably between laboratories for LDH,albumin, and �2-microglobulin, ratios to ULN were calculated. In addition,the known prognostic indices established in other lymphomas, IPI,18 andFLIPI19 were assessed. The outcome parameter was OS calculated from theday of recruitment to death from any cause or to the latest follow-up date.

In line with IPI and FLIPI, our aim was to establish a clinical prognosticindex (mantle-cell lymphoma international prognostic index, MIPI) easilyavailable in clinical practice. Besides the clinical parameters, the prolifera-tion markers Ki-67 and number of mitoses, possibly the most importantprognostic markers in MCL,1,2,4,11,14-17 have been assessed centrally and in astandardized way. Thus, we analyzed the additional prognostic relevance ofcell proliferation and exploratively developed a combined biologic index(MIPIb) incorporating the effect of cell proliferation.

Statistical methods

To validate IPI and FLIPI, Kaplan-Meier OS curves stratified for risk groupwere calculated and compared by the log rank test. In addition, theindividual risk factors constituting IPI and FLIPI, respectively, werechecked for independent prognostic relevance by multiple Cox regressionanalyses.

The prognostic relevance of the candidate prognostic factors wasevaluated using univariate Cox regression for OS. Subsequently, multipleCox regression with backward variable selection was performed to identifyindependent prognostic factors. Size of largest involved lymph node, spleeninvolvement, lymphocyte, granulocyte, and monocyte counts, albumin, andserum �2-microglobulin levels were not included in multiple regressionbecause of a high number of multivariately missing values. Bone marrowinvolvement was excluded because of high congruence with stage IV.Continuous parameters were not categorized a priori because this wouldhave negatively affected the power of the analysis.25 ECOG performancestatus was classified into 3 groups: asymptomatic (ECOG 0), symptomaticbut ambulatory and able to carry out light work (ECOG 1), and unable towork or bedridden (ECOG 2-4). WBC count, LDH, and �2-microglobulinwere analyzed on the log-scale because of highly skewed distributions. Theproportional hazards assumption and the linearity assumption for the finalmodel were checked using scaled Schoenfeld residuals26,27 and martingaleresiduals,28 respectively.

Prognostic groups were defined by categorizing the prognostic score(linear predictor) of the final Cox regression model. 2 optimal cutpointswere found maximizing the log rank statistic according to the “minimal Pvalue approach,”29 and P values for the log rank statistic were adjusted formultiple testing by the Bonferroni method. For practical reasons referring toclinical application of the prognostic index, the goal was to define 3 riskgroups, none of which comprised more than 50% of the patients. Anexplorative simplification of the new classification rule was developedcategorizing the prognostic factors according to standard cutpoints. Thesimplification with the best concordance to the original index measured byCohen weighted kappa30 was chosen.

Internal validation was performed applying the refined bootstrapdescribed by Efron.31 Random data splitting in training and validationsample was not performed because this internal validation procedurereduces the power for both model development and validation and is knownto be inferior to bootstrap validation.31,32 Bootstrap validation used thelikelihood ratio or log rank test statistic, the calibration slope (regressioncoefficient of the prognostic score), Harrell concordance index c33 forcensored outcome variables, and the measure of separation SEP34 forsurvival curves.

Statistical analyses were performed using SAS version 9.1 (SASInstitute, Cary, NC) and R version 2.2.1 (www.r-project.org). Significancelevel was 5% for all analyses.

Results

Patient characteristics, treatment, and outcome

Between May 1996 and October 2004, 455 patients with advancedstage MCL entered the trials. The median age was 60 years(34-86 years) and 76% of the patients were male. Eighty-fourpercent of the patients were in Ann Arbor stage IV, 79% had bonemarrow involvement. Forty-three percent presented with B-symptoms and 9% were unable to work or bedridden (ECOC 2-4).Further patient characteristics are depicted in Table 1.

Initial cytoreductive chemotherapy comprised CHOP in 56% ofthe patients, 31% of the patients received R-CHOP, 11% MCP, and2% other chemotherapy regimens. Of the 438 patients evaluable fortreatment response, 351 (80%) achieved a complete or partialremission, and 80 (18%) a complete remission. Treatment inremission was ASCT in 80 patients, IFN� maintenance in199 patients, and 72 patients obtained no therapy in remission. Of

PROGNOSTIC INDEX FOR ADVANCED MANTLE CELL LYMPHOMA 559BLOOD, 15 JANUARY 2008 � VOLUME 111, NUMBER 2




the 455 patients, 159 had died and the median OS was 57 monthswith a median follow-up of surviving patients of 32 months.

IPI and FLIPI

According to the IPI, 99 patients (23%) were classified as low risk(LR), 173 patients (40%) as low intermediate risk (LIR),119 patients (28%) as high intermediate risk (HIR), and 41 patients(9%) as high risk (HR); in 23 patients the IPI could not be evaluateddue to missing data. The median OS was not yet reached in the LRgroup with a 5-year OS rate of 59%, and 61 months in the LIRgroup, 45 months in the HIR group and 20 months in the HR group(P � .001, Figure 1). LIR and HIR groups comprised togethermore than two-thirds of the patients and were not well separated. Inmultiple Cox regression age, ECOG performance status and LDHwere prognostic for OS, whereas the number of extranodal siteswas not (Table 2) .

The FLIPI could be determined in 418 patients of whom 26(6%) were classified as LR, 124 patients (30%) as intermediate risk(IR), and 268 patients (64%) as HR. The median OS was not yetreached for LR and IR patients with a 5-year OS rate of 61% and57%, respectively, and 48 months in the HR group (P � .001,Figure 2). LR and IR groups were not separated and the large highrisk group had relatively good outcome. In the multiple Coxregression, age and LDH were prognostic for OS, whereas neitherhemoglobin nor the number of involved nodal areas were ofprognostic relevance (Table 3).

Prognostic factors

Neither sex, Ann Arbor stage (III vs IV), bone marrow involve-ment, number of extranodal sites or number of nodal areas, norplatelet count or albumin showed prognostic relevance for OS inunivariate analyses. In contrast, age, ECOG performance status,B-symptoms, spleen involvement, maximal lymph node size,WBC, lymphocyte, granulocyte, and monocyte counts, LDH, Hb,and �2-microglobulin had a significant impact on OS (Table 4).

Sex, stage, B-symptoms, number of extranodal sites, number ofnodal areas, platelet count, and hemoglobin lost their prognosticrelevance in multiple Cox regression with backward variableselection on the dataset of 409 complete cases. Age, ECOGperformance status, LDH, and WBC counts were identified as the4 independent prognostic factors for OS. Relative risk (RR) ofincreased age was 1.04 per year (95% CI 1.02-1.06, P�.001), RRof patients with poor performance (ECOG � 1) was 2.01 (95% CI1.19-3.39, P � .009), and patients with a tenfold elevated LDH orWBC count had a RR of 3.92 (95% CI 1.48-10.37, P � .006) and2.56 (95% CI 1.66-3.95, P � .001), respectively (Table 5). Accord-ingly, the MIPI prognostic score was calculated as

MIPI score � [0.03535 � age (years)] � age (years)]� 0.6978 (if ECOG � 1)� [1.367 � log10(LDH/ULN)]� [0.9393 � log10(WBC count)]

Mantle cell lymphoma international prognostic index (MIPI)

The median value of the prognostic score in 409 patients was 5.78(4.31-9.18), 10% of the values were below 5.16 and 90% of the

Table 1. Baseline patient characteristics

Parameter Quantity n

Median age, y (range) 60 (34-86) 455

Sex, no. male (%) 344 (76) 455

ECOG 0, no. (%) 147 (33) 452

ECOG 1, no. (%) 263 (58) 452

ECOG 2-4, no. (%) 42 (9) 452

Stage IV, no. (%) 384 (84) 455

B-symptoms present, no. (%) 196 (43) 451

Bone marrow involvement present, no. (%) 360 (79) 453

Extranodal sites more than 1, no. (%) 143 (32) 445

Median no. involved nodal areas, (range) 8 (0-11) 432

Median max. lymph node size, cm (range) 4 (0.5-24) 387

Spleen involvement present, no. (%) 225 (54) 420

Median WBC count, 109/L (range) 7.9 (1.0-764) 451

Median lymphocyte count, 109/L (range) 2.1 (0.35-625) 422

Median granulocyte count, 109/L (range) 4.2 (0.19-26.4) 413

Median monocyte count, 109/L (range) 0.5 (0.014-10.9) 408

Median platelet count, 109/L (range) 188 (3-1346) 452

Median LDH/ULN (range) 0.86 (0.15-5.3) 443

Median hemoglobin (males),g /L, (range) 133 (55-175) 342

Median hemoglobin (females),g /L, (range) 124 (30-149) 109

Median albumin/ULN (range) 0.8 (0.36-1.26) 268

Median �2-microglobulin/ULN (range) 1.1 (0.06-8) 285

Data in column 3 (“n”) are number of data points for parameter.

Figure 1. Overall survival according to the International Prognostic Index riskgroups.18 LR indicates low risk; LIR, low intermediate risk; HIR, high intermediaterisk; and HR, high risk.

Table 2. Prognostic relevance of the IPI18 (n � 432) risk factors foroverall survival determined in a multiple Cox regression

Prognostic factor Comparison RR 95% LCL 95% UCL P

Age, y � 60 vs � 60 1.7 1.2 2.3 .002

ECOG 2-4 vs 0-1 2.8 1.8 4.5 �.001

LDH elevated vs normal 1.6 1.2 2.3 .006

No. of extranodal sites � 1 vs �1 1.2 0.8 1.6 .33

Ann Arbor stage was III or IV in all patients.RR indicates relative risk; LCL, lower confidence limit; UCL, upper confidence

limit; and ECOG, ECOG performance status.

Table 3. Prognostic relevance of the FLIPI19 (n � 418) risk factorsfor overall survival determined in a multiple Cox regression

Prognostic factor Comparison RR 95% LCL 95% UCL P

Age, y � 60 vs � 60 1.8 1.3 2.5 .001

LDH elevated vs normal 1.8 1.3 2.5 .001

Hemoglobin, g/L �120 vs �120 1.1 0.8 1.6 .50

No. involved nodal areas � 4 vs � 4 1.1 0.7 1.6 .80

Ann Arbor stage was III or IV in all patients.RR indicates relative risk; LCL, lower confidence limit; and UCL, upper confi-

dence limit.

560 HOSTER et al BLOOD, 15 JANUARY 2008 � VOLUME 111, NUMBER 2




values below 6.63. Potential cutpoints between 5.15 and 6.65 insteps of 0.05 were assessed. On this basis, the value 6.2 showed thebest discrimination (log rank statistic 53.06, 1 degree of freedom,df) and defined a high-risk group of 84 patients (21%, score � 6.2).For the definition of an intermediate-risk group, the value of 5.7was identified as best discriminator (log rank statistic 61.11, 2 df).In this way 145 patients were classified as intermediate risk (35%,5.7 � score � 6.2). The remaining 180 patients belonged to thelow-risk group (44%, score � 5.7). After Bonferroni correction forthe P values calculated from the log rank statistics, the separationof these risk groups remained statistically significant. Median OSwas not reached in the low risk group with a 5-year OS of 60%, andit was 51 months and 29 months in the intermediate-risk group andthe high-risk group, respectively (Figure 3).

The internal validation procedure correcting for overoptimismby bootstrap showed stability of statistical significance and prognos-tic separation (Table 6). In addition, the bootstrap correctedperformance measures for the new prognostic classification systemappeared to be better than the results of the external validation ofthe IPI and FLIPI.

Simplified MIPI

The classification according to the new prognostic index involvessome elementary mathematical operations best performed using anelectronic calculator. To make the index most practicable, weexploratively simplified it categorizing the individual prognosticfactors maintaining high concordance to the original index. Dichoto-mizing the prognostic factors, as is commonly done, at standardcutpoints (60 years for age, ULN for LDH, 10 � 109/L for WBCcount) yielded low concordance to the MIPI (weighted kappa0.56), and suboptimal separation of survival curves (Table 6, SEP).Instead, weighting age, ECOG performance status, LDH, andWBC with 0 to 3 points using 3 cutpoints each (Table 7) andclassifying patients with a total of at most 3 points as LR, patientswith 4 to 5 points as IR and patients with more than 5 points as HR,yielded high concordance (weighted kappa 0.79) and good separa-tion of OS curves (Table 6, SEP).

Prognostic relevance of cell proliferation

The percentage of Ki-67 positive cells was available in 236 lymphnode biopsies of the 455 MCL patients with a median of 14.5% andrange of 1.2% to 91.0%. The number of mitoses per squaremillimeter was counted in 162 cases with a median of 4, range 0 to72 and one extreme value of 1039, which was excluded fromfurther analysis. Both proliferation parameters showed strongunivariate prognostic relevance for OS with an RR of 1.29 (95% CI1.16-1.44, P � .001) for Ki-67 elevated by 10% and 1.27 (95% CI

Table 4. Prognostic relevance for overall survival according to univariate Cox regression

Candidate prognostic factor Comparison n RR 95% LCL 95% UCL P

Age 10 years older 455 1.38 1.15 1.65 �.001

Sex Male vs female 455 1.12 0.77 1.63 .55

ECOG performance status 1 vs 0 452 1.61 1.10 2.35 .014

ECOG performance status 2-4 vs 0 452 4.80 2.84 8.12 �.001

ECOG performance status 2-4 vs 0-1 452 3.48 2.22 5.43 �.001

Ann Arbor stage IV vs III 455 1.09 0.70 1.70 .70

B-symptoms present vs absent 451 1.60 1.17 2.19 .003

Bone marrow involved vs not 453 1.04 0.70 1.54 .85

No. of extranodal sites 1 vs 0 445 1.25 0.72 2.16 .42

No. of extranodal sites � 1 vs 0 445 1.57 0.89 2.77 .12

No. of extranodal sites � 1 vs 0-1 445 1.30 0.94 1.80 .11

No. nodal areas 1 more 432 1.03 0.97 1.08 .34

Maximum tumor size 1 cm larger 387 1.06 1.01 1.10 .016

Spleen involved vs not 420 1.40 1.01 1.95 .044

WBC count 10-fold* 451 2.25 1.59 3.18 �.001

Lymphocyte count 10-fold* 422 1.97 1.40 2.77 �.001

Granulocyte count 10-fold* 412 4.69 2.06 10.71 �.001

Monocyte count 10-fold* 408 2.61 1.65 4.12 �.001

Platelet count 100 �109/L more 452 1.01 0.88 1.16 .89

LDH 2-fold* 443 1.72 1.32 2.24 �.001

Hemoglobin 10 g/L more† 451 0.91 0.85 0.98 .009

Albumin �ULN‡ 268 0.23 0.05 1.08 .063

�2-microglobulin 2-fold* 285 1.53 1.22 1.92 �.001

LCL indicates lower confidence limit; UCL, upper confidence limit; and RR, relative risk.*Log scale†No significant interaction of hemoglobin and sex (P � .43)‡Quotient to ULN

Figure 2. Overall survival according to the Follicular Lymphoma InternationalPrognostic Index19 risk groups. LR, low risk; IR, intermediate risk; and HR, highrisk.





1.09-1.48, P � .003) for the number of mitoses elevated by10/mm2. In bivariable regression including Ki-67 and number ofmitoses, only Ki-67 remained independently significant, whereasthe number of mitoses lost its prognostic relevance (P � .68,n � 126), and was thus excluded from further analyses. Patientswith Ki-67 assessed showed significantly less spleen and bonemarrow involvement, better performance status, more involvednodal areas, and larger lymph nodes than patients without availableKi-67. However, there was no significant difference in OS ofpatients with or without data on Ki-67 (median OS 58 and57 months, P � .39).

Ki-67 remained independently significant from the MIPI clini-cal score (regression coefficient 0.02142, P � .001) and inclusionof Ki-67 did not substantially change the regression coefficient ofthe MIPI score (0.9554, P � .001). Thus, we calculated thecombined biologic score by addition of 0.02142 times Ki-67 (%) tothe clinical score. Optimal cutpoints were 5.7 and 6.5 for the logrank statistics with respect to OS, and Figure 4 shows OS accordingto the so defined combined biologic index (MIPIb).

Discussion

Based on data of 455 patients with advanced stage MCL from3 randomized trials of GLSG and European MCL Network, a newprognostic index (MIPI, mantle cell lymphoma international prog-nostic index) has been developed. The MIPI is the first prognosticindex specific to MCL patients and, more importantly, allows a

clear separation of 3 well-balanced groups of patients withsignificantly different prognoses. The low-risk group comprised44% of the patients with a median OS not yet reached after amedian follow-up of 32 months, and a 5-year OS rate of 60%. Theintermediate risk group consisted of 35% of the patients andrevealed a median OS of 51 months, while the high-risk group(21%) had a poor outcome with a median survival of only29 months. The MIPI is based on the 4 independent prognosticfactors age, ECOG performance status, LDH, and WBC counts,which are clinical variables easily and routinely available inclinical practice. A simplification of the MIPI was developedallowing for bedside application reproducing the original indexremarkably well.

The application of the IPI18 for diffuse large B-cell lymphoma toour MCL patient cohort revealed a lack of separation of the2 intermediate-risk groups, which comprised more than two-thirdsof the patients. The application of the FLIPI19 for follicularlymphoma to our MCL patients showed even worse results, with noseparation of low- and intermediate-risk groups and a largehigh-risk group with relatively good outcome. In addition, of theindividual IPI and FLIPI risk factors, the number of extranodalsites and the number of involved nodal areas showed no prognosticrelevance and hemoglobin no independent prognostic relevance inour MCL patient cohort. ECOG performance status, which showedhigh prognostic relevance in our analyses and many previousreports, was not included in the development of the FLIPI.

Previous attempts to apply the IPI in particular to patients withMCL proved rather unsatisfactory1,12-14,16,20 (Table 8). The IPIshowed either no prognostic relevance for OS,4,10,17 or the original4 risk groups were combined to 312,14 or 2 risk groups1,11,16 ordifferent risk groups had very similar outcome.2,9,13,20 In addition,of the 5 IPI risk factors, the number of extranodal sites showedmostly no, and Ann Arbor stage inconsistent, prognostic relevancein previous univariate analyses (Table 8).

In contrast to our results, a recent report by Møller et al20 statedsuperiority of the FLIPI over the IPI in terms of size and number ofrisk groups, separation of survival curves, and prognostic value ofindividual risk factors. However, the number of patients analyzed,93, was rather small, low- and intermediate-risk groups were notwell separated, and the high-risk group included more than half ofthe patients. Interpretation of multiple Cox regression results washampered by inclusion of the interdependent factors IPI, FLIPI andthe individual prognostic factors.

Of the independent prognostic factors of the MIPI, consistentprognostic relevance on OS had been shown for age, performancestatus, and LDH in previous univariate analyses (Table 8). Hemato-logic parameters were mostly not considered as candidate prognos-tic factors. Three studies,1,11,12 however, included the WBC countand described a prognostic relevance in univariate analysis. Inseveral studies peripheral blood involvement was included4,10-13

Figure 3. Overall survival according to the new prognostic index (MIPI). LRindicates low risk, prognostic score less than 5.7; IR, intermediate risk, score 5.7 ormore but less than 6.2; and HR, high risk, score 6.2 or more. The prognostic score iscalculated as [0.03535 � age (years)] � 0.6978 (if ECOG � 1) � [1.367 �log10(LDH/ULN)] � [0.9393 � log10(WBC count].

Table 5. Independent prognostic factors for overall survival identified by backward variable selection with multiple Cox regression onsignificance level 5% for the Wald statistic

Prognostic factor Comparison � SE RR 95% LCL 95% UCL P

Age 1 year older 0.03535 0.009604 1.04 1.02 1.06 �.001

ECOG 2-4 vs 0-1 0.6978 0.2663 2.01 1.19 3.39 .009

LDH 10-fold 1.367 0.4962 3.92 1.48 10.37 .006

WBC count 10-fold 0.9393 0.2220 2.56 1.66 3.95 �.001

The candidate prognostic factors included were age, sex, ECOG performance status, Ann Arbor stage, B-symptoms, number of extranodal sites, number of nodal areas,WBC count, platelet count, LDH, and hemoglobin. The model was developed using 409 complete cases for the candidate prognostic factors. The same model was identifiedusing forward variable selection or a significance level of 10%.

� indicates the regression coefficient; SE, standard error; RR, relative risk; LCL, lower confidence limit; and UCL, upper confidence limit.





and showed significance in 34,12,13 of 5 univariate analyses and in212,13 multiple regression analyses. The methods for the detectionof peripheral blood involvement, however, were mostly not ad-equately described and sensitivity of different methods may varysubstantially. Although peripheral blood involvement may occurwithout leukocytosis, a high WBC count probably reflects periph-eral blood involvement, and our results thus may confirm previousobservations.12,13

Despite these encouraging results, it must be emphasized thatthe current analyses did not include patients with limited stage Ior II MCL because for these patients no comparable clinical trialis available. However, the prognostic relevance of stage is notconsistently seen in the literature, the proportion of patients withstages I or II is rather low in MCL (Table 8) and they also requirea different therapeutic approach. Thus, as recommended byAltman,35 to avoid any treatment bias, we limited our investiga-tion to the advanced-stage MCL patients with standardizedtreatments in randomized trials. The present data are alsolimited to patients who can tolerate moderately intensivechemotherapy. This limitation, however, also applies to the IPI18

and FLIPI19 and does not hamper the relevance of the MIPIwithin these borders.

Spleen involvement, maximal tumor size, serum �2-microglobulin levels, and lymphocyte, granulocyte, and mono-cyte counts were excluded from multiple Cox regression due tomissing data, but had shown univariate prognostic relevance onOS. Nevertheless, further exploratory analyses revealed noadditional prognostic relevance of these variables to the 4 MIPI

prognostic factors. For �2-microglobulin this stands in contrastto the results of Khouri et al.36 Two MIPI adverse prognosticfactors, higher age and worse ECOG performance status, arealso contraindications against more aggressive experimentaltherapies, for which the MIPI could be applied to selectcandidate patients. However, excluding age and ECOG perfor-mance status as candidate prognostic factors, only LDH andWBC count remained independent prognostic factors stillallowing for patient stratification. In addition, in the cohort ofpatients younger than 65 years, the same 4 independent prognos-tic factors were identified, thus, there was no “age-adjusted”prognostic index as developed for the IPI. Finally, our findingsare independent of the choice of initial cytoreductive therapy,because inclusion of treatment variables did not change therelevance of the 4 independent prognostic factors. In particular,the inclusion of primary rituximab treatment as covariate did notchange the final prognostic model. In addition, a significantimprovement in OS has not been shown for ASCT so far forMCL patients.

Table 6. Bootstrap validation according to Efron et al31

Model Method Log rank-�2 df c SEP

Cox model optimistic 55.84* 4 0.6867 1.00†

Cox model bootstrap 46.62* 4 0.6791 0.91†

MIPI optimistic 61.11 2 0.6668 1.59

MIPI bootstrap 47.64 2 0.6536 1.49

MIPIb optimistic 39.78 2 0.7096 1.74

MIPIb bootstrap 31.78 2 0.6962 1.58

Risk groups using standard cutpoints optimistic 34.77 2 0.6393 1.51

Simplified MIPI optimistic 63.15 2 0.6716 1.60

IPI external 38.83 3 0.6359 1.41

IPI LR/LIR vs HI vs HR external 35.36 2 0.6149 1.40

IPI LR vs LI/HI vs HR external 35.10 2 0.6170 1.31

IPI LR vs LI vs HI/HR external 24.22 2 0.6178 1.41

FLIPI external 14.14 2 0.5773 1.36

df indicates degrees of freedom; c, Harrell measure of concordance34 between expected and observed survival times; SEP, measure of separation35 of survival curves;standard cutpoints: 60 years of age, upper limit of normal for LDH, 10 � 109 leukocytes/L; optimistic, estimation using the data set of model development; bootstrap, estimationcorrected for overoptimism; and external, estimation using an independent data set.

*Likelihood ratio statistic.†Calibration slope: regression coefficient of the prognostic score.

Table 7. Simplified prognostic index

Points Age, y ECOG LDHULN WBC, 109/L

0 �50 0-1 �0.67 � 6.700

1 50-59 — 0.67-0.99 6.700-9.999

2 60-69 2-4 1.000 -1.49 1.000-14.999

3 �70 — �1.5000 �15000

For each prognostic factor, 0 to 3 points were given to each patient and pointswere summed up to a maximum of 11. Patients with 0 to 3 points in summary wereclassified as low risk, patients with 4 to 5 points as intermediate risk, and patients with6 to 11 points as high risk. ECOG performance status was weighted with 2 points ifpatients were unable to work or bedridden (ECOG 2-4). LDH was weighted accordingto the ratio to the ULN. Thus, for an ULN of 240 U/L, the cutpoints were 180 U/L,240 U/L, and 360 U/L, for example.

— indicates not applicable.

Figure 4. Overall survival according to the combined biologic index (MIPIb) in220 patients with Ki-67 available. LR indicates low risk, combined biologic score(CBS) less than 5.7; IR, intermediate risk, CBS 5.7 or more but less than 6.5; and HR,high risk, CBS 6.5 or more. The combined biologic score is calculated as 0.03535times age (years) plus 0.6978 (if ECOG � 1) plus 1.367 times log10(LDH/ULN) plus0.9393 times log10(WBC count) plus 0.02142 times Ki-67 (%).





In an attempt to develop a combined biologic index aspostulated by Raty et al15 and Tiemann et al,17 we included theproliferation marker Ki-67 in our analyses and showed highprognostic relevance independent from the MIPI prognosticscore. Including Ki-67 we exploratively defined a combinedbiologic index (MIPIb) which revealed a low-risk group withrelatively good outcome. However, proliferation data wereavailable in only approximately half of the patients, and theselection of patients with Ki-67 appeared to be nonrandom withless spleen and bone marrow involvement and more and largerlymph node involvement. As it is assessed on lymph nodebiopsies, proliferation is not available in all MCL patients. Inaddition, standardization and reproducibility of proliferationassessment still requires improvement as essential prerequisiteto be used as molecular prognostic marker.37 Hence, a combinedbiologic prognostic index including Ki-67 is currently notapplicable except for research studies.

We performed internal validation by bootstrap, the methodthat achieves the highest possible power from the availabledata,31,32 and could confirm a high stability of the developedprognostic model. Nevertheless, an external validation on anindependent data set is still warranted to allow a broadapplication of this prognostic tool.

In conclusion, our new prognostic classification tool, MIPI,might be helpful to allow individualized, risk-adapted treatmentdecisions in patients with advanced stage MCL, to hopefully

optimize treatment and to improve outcomes of this aggressivedisease. In addition, our results will allow stratification in clinicaltrials, interstudy-comparisons of clinical trial results according topatient risk profiles, and provide a basis for establishing futurenovel biologic prognostic markers.

Acknowledgments

We thank the German Low Grade Lymphoma Study Groupand the European MCL Network for participation in thisstudy. Many thanks to Olaf Determann from the Lymph NodeRegistry Kiel, who also worked intensively in providing theproliferation data.

This work was supported in part by grants from the DeutscheKrebshilfe (T14/96/Hi 1, project No. 70-2208-Hi 2), the Bundesmin-isterium fur Bildung und Forschung, Kompetenznetz MaligneLymphome (no. 01 GI 9994), the European Commission (no.LSHC-CT-2004-503351), and the Lymphoma Research Founda-tion (no. MCLI-04-016).

Authorship

E.H. analyzed and interpreted data, performed statistical analysis,and drafted the manuscript; MD, W.H., and M.U. designed the

Table 8. Previous reports on clinical characteristics, outcome, and prognostic factors of patients with MCL

First author Tiemann Raty Samaha Andersen Schrader Oinonen Møller Argatoff Weisenburger Zucca Bosch Decaudin Velders

Reference 17 15 13 1 16 12 20 4 14 9 11 10 2

Number 304 127 121 105* 95 94 93* 80* 68 65 59 45 41*

Median age, y 64 65 63 66 62% (� 60) 66 68 65 64 64 63 59 68

Male, % 76 64 67 75 77 59 71 70 75 67 74 78 61

Stage I/II, % 8 17 13 16 8 24 17 14 25 22 5 13 20

Median OS, mo �36 34 37 30 �30 41 37 43 38 42 49 �56 31.5

Older than 60 y � na �† �‡ � � na na �‡ na na

Sex na na na na na na na

ECOG 2-4 vs 0-1 � na � � �§ � � � � na �

Stage III/IV vs I/II � na � � � na � ¶ �

B-symptoms � na – � na na � � na na

Spleen inv. na na � � na na na na � – na

Bone marrow inv. � na na � na na � na � �

Peripheral blood inv. na na � na na � na � na na � na

Extranodal sites, more than 1 �� na � � na na �

Bulk, larger than 10 cm na na na na na na na na na na na

Elevated LDH � na � � � � na na � �

WBC count na na na � na � na na na na � na na

Hemoglobin na na � � na � � na na na na na

Growth pattern � � na na na na � na na na na

Cytology � na na na na � � na � � na

Mitotic index � � na na na na na � na � na na

Ki-67 � � na � na na na na na na na �

IPI na �** � � �†† �‡‡ � �†† � �‡‡

IPI risk group, no. 2 na 4 2 2 3 4 4 3 4 2 2 4

Prognostic relevance reported from univariate analyses on significance level .05 for overall survival.� indicates prognostic; , not prognostic; inv, involvement; and na, not analyzed.*Population based data.†Age cutpoint 70 years.‡Age cutpoint 65 years.§ECOG performance status 0 vs � 0.¶Stage I-III vs IV.�P � .1.**LR, LIR, HIR not separated.††LIR, HIR not separated.‡‡LIR, HIR, HR not separated.�Extra nodal involvement, yes vs no.





trials and revised the manuscript; W.K. provided proliferation data;J.H. provided statistical expertise and revised the manuscript; C.G.,A.v.H., H.C.K.-N., M.P., M.R., B.M., H. Einsele, N.P., W.J., B.W.,W.-D.L., U.D., H. Eimermacher, and H.W. recruited, treated, anddocumented patients.

E.H. is a candidate at the Faculty of Medicine of the Ludwig-Maximilians-University Munich for her doctoral degree and part ofthis work has been developed in partial fulfillment of the require-ments for her degree.

A complete list of the members of the German Low GradeLymphoma Study Group and the European MCL Network whocontributed to this analysis appears in Document S1, available on

the Blood website; see the Supplemental Materials link at the top ofthe online article.

Conflict-of-interest disclosure: M.P. is a member of the Advi-sory Board of Roche, Genentech, and Lilly. M.D. received supportfor clinical studies and speaker honoraria from Roche. W.H. ismember of the Advisory Board of Roche and received support forclinical studies and speaker honoraria from Roche. The remainingauthors declare no competing financial interests.

Correspondence: Dr. Michael Unterhalt, MedizinischeKlinik III, Klinikum Gro�hadern, Marchioninistr. 15, D-81377Munchen, Germany; e-mail: [email protected].

References

1. Andersen NS, Jensen MK, de Nully Brown P, Gei-sler CH. A Danish population-based analysis of105 mantle cell lymphoma patients: incidences,clinical features, response, survival and prognos-tic factors. Eur J Cancer. 2002;38:401-408.

2. Velders GA, Kluin-Nelemans JC, De Boer CJ, etal. Mantle-cell lymphoma: a population-basedclinical study. J Clin Oncol. 1996;14:1269-1274.

3. A clinical evaluation of the International Lym-phoma Study Group classification of non-Hodgkin’s lymphoma. The Non-Hodgkin’s Lym-phoma Classification Project. Blood. 1997;89:3909-3918.

4. Argatoff LH, Connors JM, Klasa RJ, HorsmanDE, Gascoyne RD. Mantle cell lymphoma: a clini-copathologic study of 80 cases. Blood. 1997;89:2067-2078.

5. Lennert K, Stein H, Kaiserling E. Cytological andfunctional criteria for the classification of malig-nant lymphomata. Br J Cancer. 1975;31 (suppl)2:29-43.

6. National Cancer Institute sponsored study ofclassifications of non-Hodgkin’s lymphomas:summary and description of a working formula-tion for clinical usage. The Non-Hodgkin’s Lym-phoma Pathologic Classification Project. Cancer.1982;49:2112-2135.

7. Banks PM, Chan J, Cleary ML, et al. Mantle celllymphoma. A proposal for unification of morpho-logic, immunologic, and molecular data. Am JSurg Pathol. 1992;16:637-640.

8. Harris NL, Jaffe ES, Stein H, et al. A revised Eu-ropean-American classification of lymphoid neo-plasms: a proposal from the International Lym-phoma Study Group. Blood. 1994;84:1361-1392.

9. Zucca E, Roggero E, Pinotti G, et al. Patterns ofsurvival in mantle cell lymphoma. Ann Oncol.1995;6:257-262.

10. Decaudin D, Bosq J, Munck JN, et al. Mantle celllymphomas: characteristics, natural history andprognostic factors of 45 cases. Leuk Lymphoma.1997;26:539-550.

11. Bosch F, Lopez-Guillermo A, Campo E, et al.Mantle cell lymphoma: presenting features, re-sponse to therapy, and prognostic factors. Can-cer. 1998;82:567-575.

12. Oinonen R, Franssila K, Teerenhovi L, Lappa-lainen K, Elonen E. Mantle cell lymphoma: clinicalfeatures, treatment and prognosis of 94 patients.Eur J Cancer. 1998;34:329-336.

13. Samaha H, Dumontet C, Ketterer N, et al. Mantlecell lymphoma: a retrospective study of 121cases. Leukemia. 1998;12:1281-1287.

14. Weisenburger DD, Vose JM, Greiner TC, et al.Mantle cell lymphoma. A clinicopathologic study

of 68 cases from the Nebraska Lymphoma StudyGroup. Am J Hematol. 2000;64:190-196.

15. Raty R, Franssila K, Joensuu H, Teerenhovi L,Elonen E. Ki-67 expression level, histologicalsubtype, and the International Prognostic Indexas outcome predictors in mantle cell lymphoma.Eur J Haematol. 2002;69:11-20.

16. Schrader C, Meusers P, Brittinger G, et al. Topo-isomerase IIalpha expression in mantle cell lym-phoma: a marker of cell proliferation and a prog-nostic factor for clinical outcome. Leukemia.2004;18:1200-1206.

17. Tiemann M, Schrader C, Klapper W, et al. Histo-pathology, cell proliferation indices and clinicaloutcome in 304 patients with mantle cell lym-phoma (MCL): a clinicopathological study fromthe European MCL Network. Br J Haematol.2005;131:29-38.

18. A predictive model for aggressive non-Hodgkin’slymphoma. The International Non-Hodgkin’s Lym-phoma Prognostic Factors Project. N Engl J Med.1993;329:987-994.

19. Solal-Celigny P, Roy P, Colombat P, et al. Follicu-lar lymphoma international prognostic index.Blood. 2004;104:1258-1265.

20. Møller MB, Pedersen NT, Christensen BE. Mantlecell lymphoma: prognostic capacity of the Follicu-lar Lymphoma International Prognostic Index.Br J Haematol. 2006;133:43-49.

21. Nickenig C, Dreyling M, Hoster E, et al. Com-bined cyclophosphamide, vincristine, doxorubi-cin, and prednisone (CHOP) improves responserates but not survival and has lower hematologictoxicity compared with combined mitoxantrone,chlorambucil, and prednisone (MCP) in follicularand mantle cell lymphomas: results of a prospec-tive randomized trial of the German Low-GradeLymphoma Study Group. Cancer. 2006;107:1014-1022.

22. Lenz G, Dreyling M, Hoster E, et al. Immuno-chemotherapy with rituximab and cyclophos-phamide, doxorubicin, vincristine, and pred-nisone significantly improves response andtime to treatment failure, but not long-term out-come in patients with previously untreatedmantle cell lymphoma: results of a prospectiverandomized trial of the German Low GradeLymphoma Study Group (GLSG). J Clin Oncol.2005;23:1984-1992.

23. Dreyling M, Lenz G, Hoster E, et al. Early con-solidation by myeloablative radiochemotherapyfollowed by autologous stem cell transplanta-tion in first remission significantly prolongs pro-gression-free survival in mantle-cell lymphoma:results of a prospective randomized trial of the

European MCL Network. Blood. 2005;105:2677-2684.

24. Oken MM, Creech RH, Tormey DC, et al. Toxicityand response criteria of the Eastern CooperativeOncology Group. Am J Clin Oncol. 1982;5:649-655.

25. Royston P, Altman DG, Sauerbrei W. Dichotomiz-ing continuous predictors in multiple regression: abad idea. Stat Med. 2006;25:127-141.

26. Schoenfeld D. Partial residuals for the propor-tional hazards regression model. Biometrika.1982;69:239-241.

27. Grambsch PM, Therneau TM. Proportional haz-ards tests and diagnostics based on weightedresiduals. Biometrika. 1994;81:515-526.

28. Therneau TM, Grambsch PM, Fleming TR. Mar-tingale-based residuals for survival models. Bi-ometrika. 1990;77:147-160.

29. Altman DG, Lausen B, Sauerbrei W, SchumacherM. Dangers of using “optimal” cutpoints in theevaluation of prognostic factors. J Natl CancerInst. 1994;86:829-835.

30. Cohen J. Weighted Kappa: nominal scaleagreement with provision for scaled disagree-ment or partial credit. Psychol Bull. 1968;70:213-220.

31. Efron B, Tibshirani RJ. An Introduction to theBootstrap. Vol. 57: Chapman & Hall/CRC; 1994.

32. Steyerberg EW, Harrell FE, Jr., Borsboom GJ,Eijkemans MJ, Vergouwe Y, Habbema JD. Inter-nal validation of predictive models: efficiency ofsome procedures for logistic regression analysis.J Clin Epidemiol. 2001;54:774-781.

33. Harrell FE Jr, Califf RM, Pryor DB, Lee KL, RosatiRA. Evaluating the yield of medical tests. JAMA.1982;247:2543-2546.

34. Sauerbrei W, Hubner K, Schmoor C, SchumacherM. Validation of existing and development of newprognostic classification schemes in node nega-tive breast cancer. German Breast Cancer StudyGroup. Breast Cancer Res Treat. 1997;42:149-163.

35. Altman DG. Systematic reviews of evaluations ofprognostic variables. BMJ. 2001;323:224-228.

36. Khouri IF, Saliba RM, Okoroji GJ, Acholonu SA,Champlin RE. Long-term follow-up of autolo-gous stem cell transplantation in patients withdiffuse mantle cell lymphoma in first diseaseremission: the prognostic value of beta2-micro-globulin and the tumor score. Cancer. 2003;98:2630-2635.

37. Sehn LH. Optimal use of prognostic factors innon-Hodgkin lymphoma. Hematology Am SocHematol Educ Program. 2006:295-302.





mipi blood 2008

Documents