Bone Mrrow Trnsplnttion, (1999) 24, 283 287 1999 Stockton Press All rights reserved 0268 3369/99 $12.00 http://www.stockton-press.co.uk/bmt Comprison of the clssic Glucksberg criteri nd the IBMTR Severity Index for grding cute grft-versus-host disese following HLA-identicl sibling stem cell trnsplnttion R Mrtino 1, P Romero 1, M Subirá 1, M Bellido 1, A Altés 1, A Sured 1, S Brunet 1, I Bdell 2, J Cubells 2 nd J Sierr 1 1 Division of Clinicl Hemtology nd 2 Deprtment of Peditrics, Hospitl de l Snt Creu i Snt Pu, Brcelon, Spin Summry: Acute grft-versus-host disese (AGVHD) severity is usully grded (grdes 0 IV) by the pttern of orgn involvement using the clssic Glucksberg Settle criteri (GSC). Recently, the Interntionl Bone Mrrow Trnsplnt Registry (IBMTR) developed new Severity Index by regrouping the ptterns of orgn involvement into five Indexes (0 D) tht ppered more predictive of trnsplnt-relted mortlity (TRM) nd trnsplnt filure (TF, relpse or TRM). We studied the predictive vlue of both grding systems of TRM, TF nd GVHDrelted mortlity (GTRM) in series of 114 consecutive ptients 12 yers old llogrfted from histocomptible sibling t our institution, 100 of whom were evluble for AGVHD. The IBMTR Severity Index showed better incrementl prediction of TRM (reltive risks (RR) of 1, 1.5, 1.4, 2 nd 2.5 for Indexes 0, A, B, C nd D), TF (RRs of 1, 1.6, 1.6, 2 nd 2.3, respectively) nd GTRM (RRs of 1, 2.2 nd 4.8 for Indexes B, C nd D) thn the GSC. With the GSC different outcomes for TRM nd TF were found only from grde 0 to I II nd 0 to IV or I III to IV, but not from I II to III. The GSC lso ppered less predictive of GTRM (RRs of 1, 0.4 nd 2.9 for grdes II, III nd IV). In our reltively smll ptient smple, the new IBMTR Severity Index ppered more predictive of trnsplnt outcome thn the GSC, especilly between no AGVHD, erly Indexes (A B) nd dvnced Indexes (C D). Keywords: stem cell trnsplnttion; grft-versus-host disese; grding Moderte-to-severe cute GVHD (AGVHD) develops in 20 50% of dult recipients of n HLA-identicl sibling llogeneic stem cell trnsplnt (SCT). Mortlity directly ttributble to AGVHD or its tretment occurs in 10 20% of ptients. 1 3 For clssifiction nd prognostic informtion, AGVHD hs been grded ccording to the clssic Glucksberg Settle criteri (GSC) for more thn 20 Correspondence: Dr R Mrtino, Servei d Hemtologi Clínic, Hospitl de l Snt Creu i Snt Pu, Av. Snt Antoni M Clret, 167, 08025 Brcelon, Spin Received 30 November 1998; ccepted 11 Mrch 1999 yers. 4,5 Recently, the Interntionl Bone Mrrow Trnsplnt Registry (IBMTR) designed new stging system from lrge dt set of dult ptients receiving n HLAidenticl sibling BMT. 6 This IBMTR Severity Index ppered more predictive of the risk of trnsplnt-relted mortlity (TRM) nd tretment filure (TF) thn the GSC. We report the results of comprtive nlysis of these two stging systems in ptients llogrfted t our institution from n HLA-identicl sibling over n 8-yer period. Mterils nd methods Ptients From Jnury 1990 to Mrch 1998, 114 consecutive ptients ged 12 yers received n llogeneic SCT t our institution. Ptient chrcteristics re shown in Tble 1. The medin ge ws 36 yers (rnge 12 58), nd 68% of ptients were mle. GVHD prophylxis consisted of CYA plus short-course MTX in 79%, CYA plus prednisone in 11% nd CYA lone in 10%. CYA ws given t 3 mg/kg/dy i.v. strted on dy 1, which ws lter switched to the orl route with the im of mintining trough whole blood concentrtion of 200 300 ng/ml. Short-course MTX consisted of 15 mg/m 2 on dy 1 nd 10 mg/m 2 on dys 3, 6 nd 11. The source of the stem cells ws BM in 72% nd PBSC in 28%. Prtil T cell depletion of the grft ws done in 20 cses (18%), with finl T cell content of 1 3 10 5 /kg CD3 cells. 7,8 One hundred ptients survived 21 dys with engrftment fter trnsplnt nd were evluble for this study. Definition nd grding of AGVHD AGVHD ws dignosed cliniclly nd confirmed by biopsy of ffected tissues nd/or postmortem histology. The nursing nd medicl records of ll ptients were reviewed by two BMT physicins (RM plus PR, MS or MB) to confirm the dignosis, retrieve relevnt dt for the study nd ssign the mximum grde of AGVHD reched by both the GSC nd the IBMTR Index, both bsed on the pek stge of skin, gut nd liver involvement reched. Whenever there were discrepncies between both physicins, the records were reviewed together nd consensus ws reched. Dt were collected prospectively from Jnury 1995 onwrd nd retrospectively before this dte.
284 Tble 1 Ptient chrcteristics (n = 114) Comprison of GSC nd IBMTR Severity Index for AGVHD R Mrtino et l Number % Ptient ge (medin, rnge), yers 36 (12 58) 12 15 11 10 16 34 48 42 35 55 48 Sex Mle 77 68 Femle 37 32 Sex mismtch Femle donor to mle recipient 39 34 Other combintions 75 66 Disese Acute leukemi/mds 66 56 CML 17 15 Lymphom 17 15 AA 8 7 CLL/MM 6 6 Do nd/or Re CMV serology Both negtive 18 16 One or both positive 96 84 Disese sttus Erly 52 46 Other 62 54 Conditioning therpy TBI-bsed 97 85 Chemotherpy only 17 15 GVHD prophylxis CYA/short-course MTX 90 79 CYA/PDN 13 11 CYA lone 11 10 Source of SCT BMT 82 72 PBSCT 32 28 Yer of SCT 1990 1993 62 54 1994 1998 52 46 Prtil T cell depletion ( 3 10 5 /kg CD3 + cells) 20 18 Erly TRM ( 21 dys w/o engrftment) 14 12 Evluble for cute GVHD (engrfted nd live 21 dys) 100 88 MDS = myelodysplstic syndrome; AA = plstic nemi; CML = chronic myelogenous leukemi; CLL = chronic lymphocytic leukemi; MM = multiple myelom; CMV = cytomeglovirus; TBI = totl body irrdition; CYA = cyclosporine; PDN = prednisone; MTX = methotrexte; SCT = stem cell trnsplnttion; TRM = trnsplnt-relted mortlity. Includes chronic myelogenous leukemi in first chronic phse, cute leukemi in first remission nd plstic nemi. Definitions Endpoints were time to TRM, TF nd AGVHD-relted deth. TRM ws defined s deth in ptients without relpse, with censoring t relpse or lst follow-up. TF ws defined s deth or relpse with ptients live without disese censored t time of lst follow-up. GVHD-relted mortlity ws defined s deth directly ttributble to progressive refrctory AGVHD or deth from n invsive infection during intensive immunosuppressive therpy for AGVHD or deth from chronic extensive GVHD or n infection during its tretment. Sttisticl methods Reltive risks (RR) of trnsplnt outcomes were determined using Cox proportionl hzrds regression. Ptients without AGVHD were the reference group for clculting the RR of TRM nd TF with both stging systems. Acute GVHD entered the model s time-dependent vrible so tht prior to onset the ptient ws considered in the reference group, nd ptients were censored t deth or relpse. For clculting the RR of GVHD-relted mortlity, the rtes observed in grdes II nd Index B were the reference vlues, since there were no AGVHD-relted deths in ptients with erly stges (0 I or 0 A, respectively) nd only two deths due to chronic GVHD (dt not shown). RRs were clculted seprtely for the different grdes/severity Indexes, nd differences in risk between ctegories were clculted using Chi-squre test. 9 Results Incidence of AGVHD ccording to GSC nd IBMTR Index Tble 2 shows the observed incidence of AGVHD ccording to both stging systems. Thirty ptients did not develop AGVHD. According to the GSC, the number of ptients (nd %) with grdes I, II, III nd IV AGVHD were 21, 25, 9 nd 15, nd with the IBMTR Index ptients (nd %) grded s A, B, C nd D Severity Index were 11, 21, 22 nd 16, respectively. As seen in Tble 2, the IBMTR Index tended to ssign higher overll grde for AGVHD severity thn the GSC. Reclssifiction depended lrgely on the mximum extent of skin involvement. Thus, ptients with GSC grde I were ctegorized s Index A or B depending on the extent of skin involvement ( 25% or 25 50% of the body surfce re (BSA), respectively), nd ptients with GSC grde II were clssified s Index B or C if the rsh ffected 25 50% or 50% of the BSA. Ptients with grde III AGVHD were ctegorized s Index C or D depending on the presence of stge 3 or 4 involvement in t lest one orgn, respectively, nd 3/9 ptients with GSC grde III were ssigned n Index B becuse the mximum extent of rsh ws 50% of the BSA, with no stge 3 or Tble 2 Glucksberg grde nd IBMTR index GSC Grde % Corresponding IBMTR IBMTR % Index for ech GSC grde Index Index % 0 30 0 30 0 30 I 21 A 11 A 11 B 10 II 25 B 8 B 21 C 17 III 9 B 3 C 22 C 5 D 1 IV 15 D 15 D 16
4 orgn involvement. These were the only cses in which the grde ssigned by the IBMTR Index ws lower thn the GSC. Tble 3 shows the min clinicl chrcteristics of the cses of AGVHD observed. Most ptients (94%) hd skin involvement, with gstrointestinl nd/or liver involvement in 38%. The sequence of orgn involvement followed the usul ptterns, with skin followed by gstrointestinl in 28%, nd followed by liver involvement in 10%. Seventy percent received systemic tretment with steroids, usully methylprednisolone t 2 mg/kg/dy, t medin of 8 dys from the first sign of AGVHD, nd 15 ptients received secondry tretment with ntilymphocyte globulin t 15 mg/kg/dy for 7 dys. Sixteen ptients died from progressive AGVHD or n invsive infection during intensive immunosuppression, while six died from chronic GVHD or n invsive infection during its tretment. Assocition between GSC nd outcome RRs (compred to ptients without AGVHD) of TRM nd TF by GSC grde re shown in Tble 4. Both RRs incresed, lthough not significntly, from grde 0 to grdes I nd II, but the RR of both TRM nd TF decresed from grde II to grde III. Ptients with grde IV hd significntly higher risk of TRM nd TF thn those without AGVHD (P = 0.001 nd 0.002, respectively). Assocition between IBMTR Index nd outcome Tble 4 shows the RRs of TRM nd TF ccording to the IBMTR Index. As seen, ptients with Indexes A/B, C nd D AGVHD hd progressively higher risks of both TRM Tble 3 Chrcteristics of GVHD Number % Number of evluble ptients 100 Developed cute GVHD 70 70 Medin dy of onset, rnge 28 (15 65) Sequence of orgn involvement Skin only 39 56 Skin-lower GI (-liver) 20 28 Skin-liver (-lower GI) 7 10 Viscerl only 4 6 Systemic tretment 49 70 Dy strt 1st tretment, medin (rnge) b 8 (1 12) Tretment filures 54 54 AGVHD-relted deths c 16 16 Infection-relted deths 10 10 Relpses d 19 19 CGVHD-relted deths c 6 6 Other 3 3 GI = gstrointestinl; CGVHD = chronic GVHD. Forty-nine ptients received primry tretment nd 15 lso received secondry tretment (see text). b Dys from the dignosis of AGVHD to strt of primry tretment, which consisted of methylprednisolone or prednisone t 2 mg/kg/dy. c Defined in text. d Six ptients re live with disese or in remission following donor lymphocyte infusions. Comprison of GSC nd IBMTR Severity Index for AGVHD R Mrtino et l nd TF, which were sttisticlly significnt for Indexes C nd D with respect to Index 0. Prediction of GVHD-relted mortlity Tble 5 shows the RRs of GVHD-relted mortlity ccording to both stging systems. As my be seen bove, the RR for grde III ws lower thn for grde II, while it gin significntly incresed for grde IV (P = 0.001). However, with the IBMTR Index this RR progressively incresed between Index B, C nd D (P = 0.007 nd P 0.001 for B vs C nd B vs D, respectively). Discussion The concept of modifying the GSC grding system for AGVHD to mke it more objectively pplicble nd increse its ssocition with trnsplnt outcome ws ddressed t n interntionl consensus conference. 10 The conference concluded tht lthough the clssic GSC distinguished ptients with different risks of outcome, dditionl studies were needed to evlute possible diversity of outcome within GVHD grdes. The IBMTR performed such study using registry dt from 2129 dults receiving n HLA-identicl sibling BMT nd found tht within different grdes certin ptterns of orgn involvement hd risks of outcome similr to other ptterns from other grdes. 6 Ptterns with similr RRs were thus regrouped into four Severity Indexes, nd the results were confirmed in seprte set of 754 T cell-depleted HLAidenticl sibling BMT. Our overll incidence of grde II IV AGVHD ws 49%, which flls within the rnge of 20 50% reported in the literture. 1 3,11 The pttern of orgn involvement, nd the primry nd secondry tretments given were lso those usully reported. 1 3,11 We found tht 30% of our ptients hd shift in the AGVHD clssifiction between both stging systems (Tble 2). These chnges were from grde I to Index B (10/21 grde I), from grde II to Index C (17/25 grde II) nd from grde III to Index B (3/9 grde III). Interestingly, very similr correltions between both systems hve been recently reported by the Settle tem in series of 838 dults nd children who received n llogeneic BMT from n unrelted or relted donor. 12 Thus, in this study the mjor chnges were from grde I to Index B nd grde II to Index C. The most relevnt finding in our study ws tht the predictbility of TRM nd TF ws better with the IBMTR Index thn with the GSC, minly due to the fct tht the RRs for both outcomes were similr for grdes II nd IV nd were lower for grde III. With the IBMTR Index, however, the RRs progressively incresed from Indexes A/B to C nd D, thus hving better incrementl correltion with trnsplnt outcome. We lso nlyzed the correltion of both stging systems with the probbility of dying from severe GVHD or its tretment (by n invsive infection), with similr results. Thus, the RR of GVHD-relted deth fell from grde II to III, while it showed progressive increse from Index B to D. Our study included ptients trnsplnted from bone mr- 285
Comprison of GSC nd IBMTR Severity Index for AGVHD R Mrtino et l 286 Tble 4 Reltive risk (RR) (95% CI) of trnsplnt-relted mortlity (TRM) nd tretment filure (TF) by the Glucksberg Settle grde nd the IBMTR Index Grde/Index n (%) of TRM RR of TRM P n (%) of TF RR of TF P of AGVHD 0 9 (30) 1 0.2 10 (33) 1 1 I 10 (48) 1.6 (0.8 3.2) 0.1 b 13 (62) 1.9 (1.04 3.4) 0.046 II 13 (52) 1.7 (0.9 3.4) 0.4 c 15 (60) 1.8 (1 3.3) 0.05 b III 4 (44) 1.5 (0.6 3.7) 0.001 d 4 (44) 1.3 (0.6 3.2) 0.5 c IV 12 (80) 2.7 (1.5 4.9) 12 (80) 2.4 (1.4 4.2) 0.002 d 0 9 (30) 1 0.3 e 10 (33) 1 0.2 e A 5 (45) 1.5 (0.7 3.5) 0.3 f 6 (54) 1.6 (0.8 3.4) 0.2 f B 9 (43) 1.4 (0.7 3) 0.04 g 11 (52) 1.6 (0.8 3) 0.02 g C 13 (59) 2 (1.04 3.8) 0.004 h 15 (68) 2 (1.1 3.7) 0.006 h D 12 (75) 2.5 (1.4 4.6) 12 (75) 2.3 (1.3 4) P vlue for comprison of ptients with no AGVHD vs grde I; b P vlue for comprison of ptients with no AGVHD vs grde II; c P vlue for comprison of ptients with no AGVHD vs grde III; d P vlue for comprison of ptients with no AGVHD vs grde IV; e P vlue for comprison of ptients with no AGVHD vs Index A; f P vlue for comprison of ptients with no AGVHD vs Index B; g P vlue for comprison of ptients with no AGVHD vs Index C; h P vlue for comprison of ptients with no AGVHD vs Index D. Tble 5 Reltive risk (RR) (95% CI) of GVHD-ssocited trnsplntrelted mortlity (GTRM) Grde/Index GTRM, n (%) RR of GTRM P of AGVHD II 8 (32) 1 0.04 III 1 (11) 0.4 (0.2 0.8) 0.001 b IV 11 (73) 2.9 (1.5 4.6) B 3 (14) 1 0.007 c C 7 (32) 2.2 (1.4 4.3) 0.001 d D 11 (69) 4.8 (2.5 6.9) P vlue for comprison of ptients with AGVHD grde II vs III. b P vlue for comprison of ptients with AGVHD grde II vs IV. c P vlue for comprison of ptients with AGVHD Index B vs C. d P vlue for comprison of ptients with AGVHD Index B vs D. row or peripherl blood stem cells nd severl pproches for GVHD prophylxis. Despite this reltive heterogeneity, the findings fvoring the IBMTR Index for prediction of trnsplnt outcome were significnt. Of note, in the originl report from the IBMTR, the predictive vlue of the Severity Index ws confirmed in n independent group of T celldepleted BMT, suggesting tht t lest for HLA-identicl sibling trnsplnts the Index is useful for different GVHD prophylctic strtegies. Possibly due to the few ptients included within ech IBMTR Index, we found tht Indexes A nd B hd similr RRs of TRM nd TF, finding tht ws not found in the originl report, in which both RRs significntly incresed from A to B. 6 It is probbly of greter clinicl relevnce, however, to identify different outcomes between moderteto-severe (grdes III/IV or Indexes C/D) thn between mild-to-moderte AGVHD (grdes I/II or Indexes A/B), since the risks of TRM nd TF increse minly from mildto-moderte to moderte-to-severe stges. 1,6,11,13 In this respect, the IBMTR Index ppers to be more useful thn the GSC. It remins to be seen whether or not the IBMTR Index will be more predictive of trnsplnt outcome thn the GSC in other SCT settings, such s volunteer unrelted or fmily mismtched BMT or in lrge series of sibling HLAidenticl PBSCT. This ltter study is currently underwy within the PBSCT subcommittee of the Spnish Group of Hemtopoietic Trnsplnttion (GETH). Both clssifiction systems re bsed on the pek severity of AGVHD involvement of skin, gut nd liver. Recently, the Settle tem hs proposed n lterntive pproch for AGVHD grding which plces specil emphsis on the overll clinicl course s reflected by disese progression nd response to tretment. 14 This novel pproch my prove more predictive of outcome thn the two sttic models since it includes cliniclly relevnt prmeters, nd thus it should lso be tested in future studies. In summry, our results suggest tht the outcome-bsed reclssifiction of ptterns of orgn involvement into new Severity Indexes my mke the IBMTR Index more dequte for studying AGVHD fter llogeneic HLAidenticl sibling SCT due to its better correltion with trnsplnt outcome thn the clssic GSC. References 1 Weisdorf D, Hke R, Blzr B et l. Tretment of moderte/severe cute GVHD fter llogeneic BMT: n nlysis of clinicl risk fetures nd outcome. Blood 1990; 75: 1024 1030. 2 Lzrus HM, Vogelsng GB, Rowe JM. Prevention nd tretment of cute GVHD: the old nd the new. A report from the ECOG. Bone Mrrow Trnsplnt 1997; 19: 577 600. 3 Mrtin PJ, Schoch G, Fisher L et l. A retrospective nlysis of therpy for cute GVHD: initil tretment. Blood 1990; 76: 1464 1472. 4 Glucksberg H, Storb R, Fefer A et l. Clinicl mnifesttions of grft-versus-host disese in humn recipients of mrrow from HLA mtched sibling donors. Trnsplnttion 1974; 18: 295 304. 5 Thoms ED, Storb R, Clift RA et l. Bone mrrow trnsplnttion. New Engl J Med 1975; 292: 895 902. 6 Rowlings PA, Przepiork D, Klein JP et l. IBMTR Severity Index for grding grft-versus-host disese: retrospective comprison with Glucksberg grde. Br J Hemtol 1997; 97: 855 864.
7 Gllrdo D, Grci-Lopez J, Sured A et l. Low-dose donor CD8 cells in the CD4-depleted grft prevent llogeneic mrrow grft rejection nd severe grft-versus-host disese for chronic myeloid leukemi ptients in first chronic phse. Bone Mrrow Trnsplnt 1997; 20: 945 952. 8 Urbno-Ispizu A, Solno C, Brunet S et l. Allogeneic trnsplnttion of selected CD34 cells from peripherl blood: experience of 62 cses using immunodsorption or immunomgnetic technique. Bone Mrrow Trnsplnt 1998; 22: 519 525. 9 Andersen PK, Borgn O, Gill RD, Keiding N. Sttisticl Models Bsed on Counting Processes. Springer: New York, 1993. 10 Przepiork D, Weisdorf D, Mrtin P et l. Consensus conference on cute GVHD grding. Bone Mrrow Trnsplnt 1995; 15: 825 828. Comprison of GSC nd IBMTR Severity Index for AGVHD R Mrtino et l 11 Grtwohl A, Hermns J, Apperley J et l. Acute grft-versushost disese: grde nd outcome in ptients with chronic myelogenous leukemi. Blood 1995; 86: 813 818. 12 Mrtin P, Schoch G, Gooley T et l. Methods of ssessment of grft-versus-host disese. Blood 1998; 92: 3479 3480. 13 Vn Lint MT, Uderzo C, Locsciulli A et l. Erly tretment of cute grft-versus-host disese with high- or low-dose 6- methylprednisolone: multicenter rndomized tril from the Itlin Group for Bone Mrrow Trnsplnttion. Blood 1998; 92: 2288 2293. 14 Mrtin P, Nsh R, Snders J et l. Reproducibility in retrospective grding of cute grft-versus-host disese fter llogeneic mrrow trnsplnttion. Bone Mrrow Trnsplnt 1998; 21: 273 279. 287