Pranzatellietal.JournalofNeuroinflammation2013,10:10 JOURNAL OF http://www.jneuroinflammation.com/content/10/1/10 NEUROINFLAMMATION RESEARCH Open Access BAFF/APRIL system in pediatric OMS: relation to severity, neuroinflammation, and immunotherapy Michael R Pranzatelli1*, Elizabeth D Tate1, Nathan R McGee1, Anna L Travelstead2, Jerry A Colliver3, Jayne M Ness4 and Richard M Ransohoff5 Abstract Background: B-cell dysregulation has been implicated but not fully characterized inpediatric opsoclonus-myoclonus syndrome (OMS), a neuroblastoma-associatedneuroinflammatory disorder. Objective: Toassess the role ofB-cell activatingfactor (BAFF) and a proliferation-inducing ligand (APRIL), two critical B cell-modulating cytokines, as potential biomarkers of disease activityand treatmentbiomarkers inOMS. Methods: SolubleBAFF and APRIL were measuredin cerebrospinal fluid (CSF) and serum by ELISA in433 children (296OMS, 109 controls, 28 other inflammatoryneurological disorders (OIND)). BAFF-R receptors on circulating CD19 +B cellswere measured by flow cytometry.A blinded scorer rated motor severity on theOMS Evaluation Scale. Immunotherapieswere evaluatedcross-sectionally and longitudinally. Results: The mean CSF BAFF concentration, which was elevated inuntreated OMS and OIND, correlated with OMS severity category (P = 0.006), and reduction by adrenocorticotropic hormone or corticotropin(ACTH)(−61%)or corticosteroids (−38%) was seen ateach level ofseverity.In contrast, CSF APRIL was normal in OMS and OIND and unaffected by immunotherapy. When the entire OMS dataset was dichotomized into ‘high’versus ‘normal’CSF BAFF concentration, the phenotype of thehighgroup included greater motor severity and number ofCSF oligoclonal bands, and a higher concentration of inflammatory chemokines CXCL13 and CXCL10 in CSF and CXCL9 and CCL21 inserum. Serum APRIL was 6.7-fold higher inthe intravenous immunoglobulins (IVIg) group, whereas serum BAFF was 2.6-fold higher in therituximabgroup. The frequency ofB cell BAFF-R expression was similar in untreated and treated OMS. Longitudinal studies of CSF BAFF revealed a significant decline in ACTH-treated patients (with or without rituximab)(P < 0.0001). Longitudinal studies ofserum APRIL showed a 2.9-fold increase after 1 to 2g/kg IVIg monotherapy (P =0.0003). Conclusions: Striking distinctionsin BAFF/APRIL signaling were found. OMS displayed heterogeneity in CSF BAFF expression, which met many but not allcriteria as a potentialbiomarker of disease activity.We speculate that CSF BAFF mayhave more utility in a biomarker panelthanas a stand-alone biomarker, and that theselective upregulation of both serum APRIL byIVIgand BAFF by rituximab, as well as downregulationof CSF BAFF by ACTH/ steroids, may have utility as treatment biomarkers. Keywords: BAFF-R, Cerebrospinalfluid cytokines,Childhood neuroinflammatory disorders, Corticosteroids, Corticotropin, IVIg,Neuroblastoma, Opsoclonus-myoclonus, Paraneoplastic syndrome, Rituximab *Correspondence:[email protected] 1DepartmentofNeurology,NationalPediatricMyoclonusCenterand NeuroimmunologyLaboratory,andSouthernIllinoisUniversitySchoolof Medicine,POBox19643,Springfield,IL62794-9643,USA Fulllistofauthorinformationisavailableattheendofthearticle ©2013Pranzatellietal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and reproductioninanymedium,providedtheoriginalworkisproperlycited. Pranzatellietal.JournalofNeuroinflammation2013,10:10 Page2of9 http://www.jneuroinflammation.com/content/10/1/10 Introduction or sedation/anesthesia, drug treatments outside the Opsoclonus-myoclonus syndrome (OMS) is a neuroin- study scope, orconcurrent autoimmunedisorders. flammatorydisorderofchildrenandadults,whichisdem- The cross-sectional part of the study involved 276 chil- onstrably paraneoplastic in about 50% of the cases [1]. drenwithOMS,meanage(SD)3.8±3.0years(range0.89 Our research on CSF in pediatric OMS has shown that B to 17 years; boys n = 120; girls n = 156), who were en- cellsareexpanded[2],BcellchemoattractantCXCL13[3] rolled after consent was signed by parents and assent by and B/T cell chemoattractant CXCL10 [4] are likewise older children.OMS groupsincludeduntreated,currently overexpressed,and 35% of the patients harbor oligoclonal treated, and previously treated patients. Videotapes were bands.[5]Responsestocurrentimmunotherapeuticstrat- made using a standardized procedure and later scored egies, such as intravenous immunoglobulins (IVIg), corti- blindedontheOMSEvaluationScaletoyieldatotalscore costeroids, and anti-CD20, also are consistent with B-cell [2]. Further details of treatment and scoring are provided involvement[2]. in the figure legends. The children underwent a morning Two critical cytokines for B-cell activation, prolifera- lumbarpuncturefollowingourspecifiedprotocol[2]. tion, homeostasis, and survival are a proliferating- Controlscomprisedchildrenwithvariousnon-inflammatory inducing ligand (APRIL) and B-cell activating factor neurological disorders (NIND), including ataxia, devel- (BAFF) [6]. They are members 13 and 13B, respectively, opmentaldisorders,headache,movementdisorders,sei- of the tumor necrosis factor ligand superfamily (TNFLS) zures, and miscellaneous disorders. Their mean age was [7,8]. Both are expressed by astrocytes [9,10], and also 8.4±5.6years,and96wereboysand91weregirls.Chil- dendritic cells, macrophages, and monocytes [11]. In drenwithOIND,whowereevaluatedlocallyandthrough human cerebrospinal fluid (CSF) and serum, they are theCenterforPediatric-OnsetDemyelinatingDiseasein found insoluble form and often studied together in neu- Alabama, served as a comparison group for specificity. roinflammatory diseases [12-14]. The BAFF receptor Theirdiagnosesincludedacutedisseminatedencephalo- (BAFF-R or BR3), one of three receptors to which BAFF myelitis, encephalitis, meningitis, multiple sclerosis and binds [11], is the principal BAFF receptor [15] expressed related disorders, and miscellaneous disorders. The bynearly allmatureB cells [10]. OINDmeanagewas7.7±5.8years(range0.1to18years; Our preliminary work on BAFF showed an increase in boysn=17;girlsn=23). OMS and decrease by corticotropin (ACTH) or corticos- Two longitudinal OMS studies were performed. In the teroids[16].ThepresentreportinanincreasedOMSsam- BAFF study, 42 children from the cross-sectional study ple size tests the hypothesis that BAFF is a biomarker of were treated using ACTH-based conventional therapy disease activity and treatment biomarker (not diagnostic with (n = 31) or without (n = 11) rituximab. There were biomarker). It includes additional immunotherapy cross- 16 boys and 26 girls, mean age 3.7 ± 2.8 years (range 1.8 sectional groups; analysis of the relation to OMS severity to 18 years). CSF was obtained before and at 8.0 ± 2.1 (videodocumentedandscored)andduration;comparison months of treatment. In the APRIL study, 20 new chil- withAPRIL,inflammatorychemokines,oligoclonalbands, dren with OMS were recruited for monotherapy with and lymphocytesubsets;longitudinal studies of BAFFand intravenous immunoglobulins (IVIg) under FDA BB- APRIL, and comparative studies in other inflammatory IND No. 5839. There were 11 boys and nine girls, mean neurologicaldisorders(OIND).Toidentifythe‘phenotype’ age4.6±1.4years(range1.8to6.4years).Theyreceived of the patients with high CSF values compared to those standard monthly clinical doses of 1 or 2 g/kg, and with normal (control-like) values, we are now able to serum was collected before and at 4.1 ± 3.9 months dichotomize the dataset. We also measured the BAFF-R oftreatment. receptoroncirculatingBcellsinthecurrentstudy. Cytokine/chemokinedetection Methods CSF and serum were collected on ice, aliquotted, and Patientsandprocedures stored at −80°C in biorepository freezers. Samples were This prospective, case–control, translational research thawedontheassaydayandBAFFandchemokineswere study is part of the ‘Study of Cytokines in Children with measured using Quantikine ELISA kits per instructions Opsoclonus-Myoclonus Syndrome’ (ClinicalTrials.gov by the manufacturer (R&D Systems, Inc., Minneapolis, NCT 00806182) [3]. Approximately 300 children were MN, USA). The comparison chemokine panel included recruited through the National Pediatric Myoclonus CXCL9, CXCL10, CXCL12, CXCL13, CCL17, CCL21, Center, regardless of severity or treatment status. Inclu- and CCL22 kits from the same vendor. Human APRIL sion criteria included clinically confirmed diagnosis of ELISA kits were purchased from eBioscience (formerly OMS and ages 0.6 to 18 years,without restriction due to Bender MedSystems, Vienna, Austria). Assays were gender, race, ethnicity, or socioeconomic status. Exclu- performed in duplicate on undiluted samples, each sion criteria were contraindications to lumbar puncture 96-well plate containing both control and OMS samples. Pranzatellietal.JournalofNeuroinflammation2013,10:10 Page3of9 http://www.jneuroinflammation.com/content/10/1/10 Samples with values above the highest standard were re- analysis utilized Pearson correlations. Although the mean run at a 1:2 or higher dilution. Outliers were verified by ageoftheOMSgroupwassignificantlylowerthanincon- repeat measurement.The kit userhad no patient contact. trols or the OIND group(which did not differ), no correl- BAFF sensitivity was 0.73 ± 6.7 pg/mL in CSF, and 1.5 to ation between age of controls and CSF BAFF 11.9 pg/mL in serum; APRIL, 0.4 pg/mL. The inter-assay concentration (P = 0.72) or CSF APRIL (P = 0.19) was coefficient of variance (CV) was 9.4% (n = 22) for CSF found,andtheagerangeofthegroupswasequivalent. BAFF, 8.5% (n = 20) for serum BAFF, 7.8% (n = 10) for Secondary analysis was performed on the OIND group CSFAPRIL,and4.8%(n=8)forserumAPRIL.Thecorre- compared to the other groups. Also, OMS immunother- spondingintra-assayCVwas4.8%(n=11),6.0%(n=10), apy groups were bundled based on common treatment 7.2% (n = 6), and 6.7% (n = 6). Freezer storage time effects for graphic presentation. Lastly, the dataset was and the concentration of CSF BAFF (P = 0.21) or APRIL dichotomized based on cytokine levels 2 SD above the (P=0.16),orserumconcentrations,werenotcorrelated. control mean to determine if two different phenotypes emerged, and family-wise Bonferroni corrections (α at BAFF-Randlymphocytesubsets 0.5/n) were performed separately on six clinical variables BAFF-R receptors were measured ex vivo by flow cyto- (P < 0.008), nine chemokines/cytokines and oligoclonal metry [17]. Peripheral venous blood was delivered to the bands (P <0.0055), and five lymphocyte subsets flow cytometrist within 1 h of collection. A 100 μL ali- (P < 0.01) to control for the multiple comparisons that quot was blocked with 0.2 mg/mL normal mouse IgG were made. for 15 min at room temperature. Directly conjugated monoclonal antibody probes (anti-BAFF-R, anti-CD19, Results anti-CD45),purchased from Becton Dickinson(San Jose, BAFFcross-sectionalstudy CA, USA), were added to the remaining cell suspension. DifferencesinCSF BAFFwerehighly significantbetween Then 2 mL FACS lysing solution was added to lyse ery- groups (Figure 1A). Mean CSF BAFF was 57% higher in throcytes, followed by a 10-min incubation at room untreated OMS than in controls. Twenty-three percent temperature in the dark. The assay suspension was cen- of untreated OMS had CSF BAFF concentrations >2 SD trifuged at 4°C for 7 min at 600 × g, supernatant was above the control mean. The three ACTH groups are removed, the pellet was washed twice with 2 mL cold shown bundled, as are the three steroid groups, for lack FACSbuffer,andrecentrifuged.Afterdecanting,100μLof of significant difference in CSF BAFF among them. CSF 1%paraformaldehydewasadded,andafter5minatroom BAFF was 56% lower in ‘All ACTH Groups’ and 45% temperatureinthedark,therewasanothercentrifugation, lowerin‘AllSteroid Groups.’ decanting,andresuspensionofcellsinFACSbuffer. IncomparisonsofindividualOMStreatmentgroupsver- Lymphocyte subsets, measured by flow cytometry as sus untreated OMS (data not shown), significantly lower described previously [2], included cells that were CD19 mean CSF BAFF concentrations (pg/mL) were found for +CD3- (B cells), CD4+CD3+ (T helper/inducer cells), ACTH (88 ± 59, -61%, P < 0.0001), steroid (140 ± 148, CD8+CD3+ (T cytotoxic/suppressor cells), TCRγδ+ -38%, P = 0.002), ACTH + IVIg (113 ± 92, -50%, (gamma/deltaTcells),orCD16/56+(NKcells).Datawere P < 0.0001), steroid + IVIg (114 ± 42, -49%, P = 0.0002), acquired ona FACSCalibur cytometer and analyzedusing ACTH + other (89 ± 76, -60%, P < 0.0001), and steroid + Cell Quest(Becton-Dickinson,SanJose,CA,USA).Qual- other(115±52,-49%,P=0.005).Ineachofthosegroups, ity control measures wererigorous, as reported elsewhere the mean BAFF concentration was below, though not [2].Theflowcytometristhadnopatientcontact. significantly different than, the control mean. Of actively treated patients, 22% had CSF BAFF concentrations Statisticalanalysis below the lowest controls. In previously treated patients, Because of inter-individual differences and the large range mean CSF BAFF concentration did not differ significantly of valuesinherentinhumancytokineandchemokinedata, fromcontrols. both means and medians were analyzed. Cross-sectional The OIND group also manifested an elevated mean groups were compared by analysis of variance (ANOVA) CSF BAFF concentration compared to untreated OMS withTukey post-hoc tests, and primary comparisons were (P = 0.04) and controls (P < 0.0001), though there was no between controls and untreated OMS, and between un- significant difference in medians. In four OIND, whose treatedandtreatedOMS.Medianswereanalyzedsecondar- diagnoses included encephalitis, neurolupus, and develop- ily by the Kruskal-Wallis test with Dunn’s post-hoc test. mental delay, the CSF BAFF concentration was above the Twogroupcomparisonsweremadebytwo-tailedttestsor highestcontrolinthe369to1,398pg/mLrange.TheCSF: Mann–Whitneytests,depending onvariance,but pre-and serumBAFFratiosdidnotdifferbetweenthesegroups. post-treatment comparisons utilized the paired t test. Per- ACTH and steroid monotherapy groups did not differ centages were compared by Fisher’s exact test. Correlation significantly in CSF BAFF concentrations (P = 0.06, Pranzatellietal.JournalofNeuroinflammation2013,10:10 Page4of9 http://www.jneuroinflammation.com/content/10/1/10 treatments (0.07 ± 0.05, -70%, P = 0.009), and steroid + othertreatments(0.08±0.05,-63%,P=0.02). The mean serum BAFF concentration did not reflect CSFconcentrations(Figure1B).Itwassignificantlyhigher only in rituximab-treated patients receiving combination immunotherapy. The BAFF elevation was two-fold in the ACTH + other group and 2.3-fold in the steroid + other group. When the ACTH + other and steroid + other groups were combined, the 50% treated with rituximab had a 2.6-fold higher serum BAFF (3,186 ± 1,774 pg/mL) thantheresttreatedwithchemotherapyorsteroidsparers (1,236±779pg/mL)(P<0.0001). APRILcross-sectionalstudy The CSF concentration of APRIL did not differ signifi- cantlybetweenthecontrolgroupandOMScross-sectional groups (Figure 2A). Neither ACTH nor steroids lowered the concentration of CSFAPRIL, and IVIg also had no ef- fect.CSFAPRILwasnotelevatedintheOINDgroup. Combined IVIg treatment groups displayed signifi- cantly higher serum APRIL concentrations (Figure 2B); Figure1Cross-sectionalBAFFconcentrations.(A)CSFBAFF.Box so did individual IVIg groups (see figure legend). In con- andwhiskergraphsdepictthemeanasaplussign,themedianasa trast, serum APRIL in ACTH- or steroid-treated groups linewithinthebox,interquartilerangesastheupperandlowerbox did not differ from untreated OMS or controls, which lines,andtherangeasTukeyerrorbars.Primarystatistical didnot differfromeachother. TheAPRILconcentration comparisonswereofmeansforuntreatedOMSversuscontrols(*) andtreatedversusuntreatedOMS(†).Statisticalsignificancelevelis inserumdidnotreflectthatinCSF. indicatedbythenumberofsymbols:*0.05>P≥0.01,***P≤0.0001. As a result of increased serum APRIL, the CSF:serum Thepre-bundledtreatmentgroupswereACTH(n=37),steroids(n= APRIL ratio was lower by 53% in the IVIg monotherapy 22),IVIG(n=24),ACTH+IVIg(n=39),steroids+IVIg(n=27), group, 60% in the ACTH + IVIg group, and 81% in the ACTH+other(rituximab,chemotherapy,orsteroidsparers)(n=14), steroid + IVIg group compared to untreated OMS. In andsteroids+other(n=13).NoneofthethreeACTHgroups controls, the CSF:serum APRIL ratio was 5.8-fold higher differedfromeachotherinBAFFconcentration,sotheywere bundledinto‘AllACTHGroups’.Steroidgroupswerebundledthe than the CSF:serum BAFF ratio (P < 0.0001); in sameway.OINDwassignificantlydifferentthanallothergroups(‡). untreatedOMS,itwas2.8-fold higher (P < 0.0001). ThemeanACTHdoseinthecombinedACTHgroupswas28±22 IU/m2/day(alternatedaydosesaveragedtoprovideadailydose Relationtoclinicaldata equivalent).Themeansteroiddosewas1.3±1.3mg/kg/day.(B) Patients were separated according to OMS severity or SerumBAFF.TheACTHmonotherapygroupandtheACTH+IVIg group,whichwerenotsignificantlydifferentinBAFFconcentrations, duration. Mean CSF BAFF concentration increased sig- werebundledintogroup2;steroidswerehandledlikewiseingroup nificantly with OMS severity category in the combined 3.The‘Rituximab+Other’groupincludedrituximab,IVIg,andeither OMS dataset (n = 271) (P = 0.006, ANOVA); so did the steroidsorACTH. median (P = 0.02, Kruskal-Wallis test). For OMS dur- ation category, there was no significant relation to CSF t-test and Mann–Whitney test). When all ACTH groups BAFF (P = 0.11). The CSF:serum BAFF ratio correlated (n = 90) were compared to all steroid groups (n = 62), with OMS total score (r = 0.38, P = 0.018), OMS dur- there was no significant difference in CSF BAFFconcen- ation (r = −0.48, P = 0.0018), and CSF B cell frequency trations: 91 ± 78 vs 123 ± 95 pg/mL, respectively. Only (r = 0.59, P = 0.0087). When all groups receiving either the combined ACTH group was significantly lower than ACTH or steroids were analyzed separately from all theIVIggroup(184 ±136pg/mL). remaining OMS groups, they manifested the lowest CSF ThemeanCSF/serumBAFFratiowas64%higherinun- BAFF concentrations regardless of OMS severity treated OMS than in controls (P < 0.0001). Compared to (Figure 3A) or duration category (Figure 3B). Only in untreated OMS, there were significant ratio reductions in the remaining OMS groups did CSF BAFFconcentration treated groups. The lowest ratios were for ACTH (0.11 ± trend withOMSseverityandOMSduration. 0.06, -55%, P = 0.0007), steroids (0.12 ± 0.05, -49%, ForserumAPRIL,therewerenosignificant differences P=0.02),ACTH+IVIg(0.14±0.10,-38%,P=0.02),ster- between severity or duration categories when the entire oid + IVIg (0.15 ± 0.06, -49%, P = 0.008), ACTH + other OMS dataset was used (P = 0.27). However, when the Pranzatellietal.JournalofNeuroinflammation2013,10:10 Page5of9 http://www.jneuroinflammation.com/content/10/1/10 resulting ‘high’ BAFF group (n = 23), there was a signifi- cantlyhigherconcentrationofCSFCXCL13andCXCL10, CSF oligoclonal band number and frequency, and serum CXCL9andCCL21comparedtothe‘normal’BAFFgroup (n = 253). The clinical characteristics of this group were greatertotalscore(motorseverity)andahigherfrequency of untreated patients. CSF Tcell frequency was lower in the ‘high’ BAFF group (80.3 ± 8.1%) than the ‘normal’ BAFFgroup(85.2±7.0%)(P=0.004).Therewerenosig- nificant between-group differences after Bonferroni cor- rectionsforCSForserumAPRIL,plasmaCXCL12,serum CCL17, serum CCL22, or for CSF lymphocyte subset fre- quencies (CD19+ B cells, γδ Tcells, CD4+ Tcells, CD8+ Tcells,orNKcells)(datanotshown). Because certain treatments alter CSF BAFF concentra- tions,theanalysiswasthenperformedonlyforuntreated OMS, for which there were 10 ‘high’ and 34 ‘normal,’ with non-significant trends: CSF CXCL13 (P = 0.06), serum CXCL9 (P = 0.08). When the sample was increased by inclusion of all non-ACTH/steroid groups, there were 10 ‘high’ and 106 ‘normal’. The statistically Figure2Cross-sectionalAPRILconcentrations.(A)CSFAPRIL, significant results were for CSF CXCL13 (P = 0.01), CSF expressedasng/mL(1000×pg/mL).Thepre-bundledtreatment OCB frequency (P= 0.04) and number(P= 0.03), serum groupswereACTH(n=33),steroids(n=11),IVIg(n=14),ACTH+ CXCL9 (P = 0.02), and total score (P = 0.02); CXCL10 IVIg(n=29),steroids+IVIg(n=14).(B)SerumAPRIL,expressedas and CCL21 were not significant. Thus, the ‘high’ group ng/mL.ThethreeIVIggroups(IVIg,ACTH+IVIg,steroids+IVIg)did notdiffersignificantlyfromeachotherinserumAPRIL appeared statistically underpowered when the dataset concentration,sotheywerecombinedinto‘AllIVIggroups’. wasdelimited. Comparedtocontrols,theserumAPRILconcentrationwashigherby 6.7-foldintheIVIgmonotherapygroup(32.6±34ng/mL),by8.2- BAFF-R foldintheACTH+IVIggroup(39.9±45ng/mL),andby6.9-foldin BAFF-R wasexpressed onalmostall circulatingCD19+ B thesteroid+othergroup(33.3±26ng/mL).IVIgwasinfused monthly,andmostevaluationswerescheduledjustbeforethenext cells, whether from controls or OMS. Mean BAFF-R IVIgwasdue. frequency was 94.5 ± 9.3% for untreated OMS (n = 6), 97.1±5.4%forconventionally-treatedOMS(n=9),90.1± OMS dataset was split into no IVIg groups versus all 12.2%formultimodalimmunotherapy(n=13),and96.9± IVIg groups, significant differences were found for IVIg 2.2% in controls, with no significant differences between at each level of OMS severity (Figure 3C) and for two groups (P = 0.45). Median frequencies also did not differ categories of OMS duration (Figure 3D). There was no significantly(P=0.62,Kruskal-Wallistest). significant difference acrossseveritycategories. The tumor OMS group (1.14 ± 95 pg/mL, n = 137) and LongitudinalCSFBAFFstudy the non-tumor OMS group (1.11 ± 0.68 pg/mL, n = 130) The mean CSF BAFF concentration in pre-treatment did not differ significantly in mean serum BAFFconcentra- OMSwashigherthanincontrols(P<0.0001).Itdeclined tion (P = 0.79), or in medians (P = 0.42). Mean serum significantly in children receiving ACTH-based immuno- APRILconcentrationinthetumorgroup(30.3±70ng/mL) therapy (Figure 4A), whether or not they also received and non-tumor group (19.7 ± 30 ng/mL) also was not sig- rituximab (Figure 4B,C). Of the 42 patients, BAFF nificantlydifferent(P=0.22). decreased in 36, did not change in two, and increased in TheOMScombineddatasetwasanalyzedforvariouscor- four. The only distinguishing characteristic of patients relations. CSF BAFF, not APRIL, correlated with CSF showing a BAFF decrease was lower CSF CXCL13 (2.6 leukocyte count (P = 0.0003, Pearson correlation), though pg/mL) than the others (4.1 pg/mL) (P = 0.02). Post- thervaluewaslow(r=0.24).Therewerenosignificantcor- treatment CSF BAFF concentration was in the normal relationswithCSFIgGconcentrationorthealbuminratio. range. Although there was an overall 71% reduction in motor severity (total score) (P < 0.0001), the change in Secondaryanalysis BAFF did not correlate with the change in total score, The whole OMS dataset was divided into two groups as even when patients without a decrease in BAFF were described under statistical analysis (Figure 4). In the excluded(r=0.19,P=0.29). Pranzatellietal.JournalofNeuroinflammation2013,10:10 Page6of9 http://www.jneuroinflammation.com/content/10/1/10 A C B D Figure3CSFBAFFandserumAPRILconcentrationversusOMSseverityanddurationcategory.(A)DotplotcomparisonofBAFFversus OMSseverityinallACTH/steroidgroupsversusallotherOMSgroups.AsteriskdenotesP<0.05bytwo-tailedttestswithincategories.Severity categorieswerepre-definedbasedontotalscoreasmild(0–12),moderate(13–24),orsevere(25–36).Between-categorycomparisonsweremade byANOVAandKruskal-Wallistests.(B)ComparisonofBAFFversusOMSdurationinallACTH/steroidgroupsversusallotherOMSgroups.OMS durationwaspre-definedasacute(0to3months),subacute(>3monthsto1year),orchronic(>1year).(C)SerumAPRILconcentrationsversus OMSseverityinallIVIggroupsversusallnon-IVIggroups.(D)SerumAPRILversusOMSdurationinallIVIggroupsversusallnon-IVIggroups. LongitudinalserumAPRILstudy changevariedgreatlybetweenpatients.Insixpatients,the The post-IVIg serum APRIL concentration (Figure 5D) post-IVIgAPRILconcentrationwas>1SDabovethecon- wassignificantlyhigherthanthepre-treatmentconcentra- trol mean; in four, >2 SD. The increase was significant in tion(meanby2.8-fold,medianby2.6-fold).SerumAPRIL children who received 2 g/kg (Figure 5F), but the sample was increased in all patients, though the amount of sizewassmallinthe1g/kggroup(Figure5E). Figure4DichotomizationofCSFBAFFconcentrationinto‘high’and‘normal’groupsforphenotyping.Percentageswerecomparedby Fisher’sexacttestsandmeansbyttests.Thehighandnormalgroupsdidnotdiffersignificantlyingenderratio,patientage,orOMSduration (datanotshown). Pranzatellietal.JournalofNeuroinflammation2013,10:10 Page7of9 http://www.jneuroinflammation.com/content/10/1/10 Figure5LongitudinaleffectsofimmunotherapyonCSFBAFFandserumAPRIL.(A)CSFBAFFpre-andpost-treatmentwithACTH-based combinationimmunotherapy.GroupmeanswithSEMandpercentreductionareshowntoeithersideofthelineplot.(B)ACTH-based immunotherapywithoutrituximab.Comparisonsweremadebypairedttests.(C)ACTH-basedimmunotherapywithrituximab.Pre-treatment meansinFiguresBandCdidnotdiffersignificantly;neitherdidpost-treatmentmeans.(D)Pre-andpost-IVIgtreatmentserumAPRIL.(E)Data from1g/kgIVIgdose.(F)Datafrom2g/kgIVIgdose.Nosignificantdifferenceswerefoundbetweenpre-treatmentmeansorbetweenpost- treatmentmeansinFiguresEandF. Discussion reflected CSF levels in other disorders [9,13,19] or in This study provides new insights on the BAFF/APRIL the present study. system in pediatric OMS. Correlation of CSF BAFF with Differences in APRIL and BAFF responses also may clinical severity and co-segregation of high CSF BAFF reflect differences in their functions [11]. Both BAFF with CSF inflammatory chemokines and oligoclonal and APRIL bind to BCMA and TACI receptors, how- bands suggest a potential role of CSF BAFF as one of ever, only BAFF binds to BAFF-R, and APRIL also severalbiomarkers of disease activity inOMS.BAFFalso binds to surface proteoglycans [21]. The expression of showed promise as a treatment biomarker in its remark- these receptors differs on pre-immune B cells and able sensitivity to ACTH or corticosteroids in cross- antigen-experienced B cells (memory B cells and long- sectional as well as longitudinal studies. APRIL was not lived plasma cells) [22]. BAFF and APRIL may localize a biomarker of disease activity in OMS, but showed a to different anatomic niches, with distinctive local striking treatment effect of IVIg. BAFF-R expression on interactions. APRIL modulates certain aspects of B cell circulating B cells was not altered in OMS, nor is it in activation and isotype switching [22]. Also, BAFF plays multiple sclerosis[10]ormyastheniagravis[18]. a role in T cell activation and polarization to Th1 The fact that CSF BAFF and APRIL did not trend [11,15], and APRIL suppressesTh2 cytokine production in the same direction in OMS is not without prece- and antibody responses in vitro [23]. dent [12]. Although both are increased in neuropsychi- Extraordinary differential effects of immunotherapy atric lupus [12,14], only CSF BAFF is increased in were found for APRIL and BAFF. To our knowledge, untreated multiple sclerosis [19,20]. Another difference increased serum APRIL concentration as an effect of between BAFF and APRIL signaling in neuroinflamma- IVIg therapy has not been reported previously in a tory diseases is that both CSF and serum BAFF are neurological disorder. The only previous report we elevated in neuro-Behcet’s disease [13], whereas only encountered was of 11 children with Kawasaki disease, CSF BAFF is elevated in multiple sclerosis. The highest an acute vasculitis that responds to IVIg [24], in whom CSF BAFF concentrations in multiple sclerosis were an IVIg dose of 2 g/kg raised APRIL (6.7-fold) but low- found in patients with more than six oligoclonal bands ered BAFF (−41%) in plasma. In OMS, we found no [21], which is similar to the higher band counts we IVIg-lowering effect on serum BAFF. Serum BAFF found in OMS. Serum BAFF and APRIL have not increasedonly afterrituximabtherapy,whichwe showed Pranzatellietal.JournalofNeuroinflammation2013,10:10 Page8of9 http://www.jneuroinflammation.com/content/10/1/10 to be an early response to B cell depletion in OMS [25]. (CXCL10, CXCL13), and oligoclonal bands [2-5]. Such Only ACTH and corticosteroids lowered CSF BAFF, but molecules necessary for B-cell recruitment, activation, they did not affect CSF APRIL. Previously, a reduced and survival may be working together to promote neu- concentration of serum BAFF has been reported in dis- roinflammation [20]. Because of the exceptional sensi- orders with elevated BAFF, such as in Wegener’s granu- tivity of CSF BAFF to peripherally administered ACTH lomatosis [26]. These differences might be clinically or corticosteroid therapy, its potential utility as a re- exploited should they be found to serve as response pre- sponse predictive biomarker or response identification dictive biomarkers (identifying subpopulation according biomarker warrants validation. The immunomodulatory to response potential) or response identification (relating effect of IVIg on APRIL signaling merits evaluation in a biological and clinical responses to treatment) biomar- variety of neuroinflammatory disorders. kers inlongitudinalstudies. The clinicalimpactof IVIg-inducedelevation in serum Abbreviations ACTH:Adrenocorticotropichormoneorcorticotropin;APRIL:Aproliferating- APRIL is difficult to predict from observational data. In inducingligand;ANOVA:Analysisofvariance;BAFF:Bcellactivatingfactor; diseases associated with increased APRIL in serum, BAFF-R:BAFFreceptor;CD:Clusterofdifferentiation;ELISA:Enzyme-linked APRIL is thought to have a pathologic role. However, immunoassay;IVIg:Intravenousimmunoglobulins;NIND:Non-inflammatory neurologicaldisorders;OIND:Otherinflammatoryneurologicaldisorders; IVIg induces clinical improvement in OMS [1], and OMS:Opsoclonus-myoclonussyndrome. APRIL is not elevated in untreated OMS, raising the possibility that boosting serum APRIL could be thera- Competinginterests RichardM.RansohoffisamemberoftheScientificAdvisoryBoardof peutic in OMS. Interestingly, APRIL has been suggested Chemocentryxandholdsstockinthecompany.Heisamemberofthe to be involved in downregulation of serological and clin- ScientificAdvisoryBoardofVertex.Therestoftheauthorshavenothingto ical activity in patients with systemic lupus erythemato- disclose. sus [27]. Thisview is consistent with the proposed mode Authors’contributions of IVIg action, which is thought to be immunomodula- MRPwastheprincipalinvestigator,andparticipatedinthedrafting/revising tion of the cytokine network [28]. However, modulation ofthemanuscript,thestudyconceptanddesign,acquisitionofdata,analysis andinterpretationofdata,andstatisticalanalysis.EDTwastheco- ofAPRILmaynotbethemechanisminvolved. investigator,andparticipatedinpatientrecruitmentandevaluation,study Treatment-induced reduction of BAFF in the central conceptanddesign,acquisitionofdata,andrevisionofthemanuscript.NRM nervous system might decrease BAFF-dependent sur- wasthelaboratoryresearcher,andparticipatedintheacquisitionand analysisofdata,andstatisticalanalysis.ALTwastheflowcytometrist,and vival of plasma cells [9], which express BAFF-R [29]. In participatedintheacquisitionandanalysisofdata,andrevisingofthe pediatric OMS, elevated CSF BAFF (not APRIL) corre- manuscript.JACwasthestudystatistician,andparticipatedinthereviewof lated with CSF cerebellar autoantibodies [30], though themanuscript.JMNparticipatedinOINDpatientrecruitmentand evaluation,andvettingofthemanuscript.RMRwastheprojectconsultant, the effect of treatment was not studied. Increased serum andrevisedthemanuscript.Allauthorshavereadandapprovedthefinal BAFF after rituximab, which may be important to B-cell versionofthemanuscript. repopulation [31], could also increase survival of auto- reactive circulatingB cells [29]. Acknowledgements ThisresearchwasfundedbyfoundationgrantsfromtheThrasherResearch BAFF adds to the evolving picture of B-cell involve- Fund(02826–2),theSpasticParalysisResearchFoundation(Illinois-Eastern ment in OMS, which includes CSF B-cell subset expan- IowaDistrictofKiwanisInternational),TheChicagoInstituteofNeurosurgery sion [2], positive oligoclonal bands [5], intrathecal over- andNeuroresearch,RonaldMcDonaldHouseCharitiesofCentralIllinois,and investigator-sponsoredgrantsfromQuestcorPharmaceuticals(UnionCity, production of CXCL13 [3], selective over-expression of CA,USA)andGenentechInc./BiogenIDEC(SouthSanFrancisco/SanDiego, CXCR5 receptors on CSF memory B cells [3], and clin- CA,USA).Theauthorsthankthepatients,theirfamilies,andreferring ical response to adjunctive anti-CD20 monoclonal anti- physiciansfortheirparticipation.WeespeciallythankJamesR.andMichelle L.ShopeIII,SallyE.Durdan,ShannonBoone,DavidandSherryRitzenthaler, body rituximab [1]. One potential limitation of CSF JackL.andEllynJ.Pittsley,SteveandDonnaPittsley,andCharlesandKerri BAFF as a biomarker is its inter-individual variability. Drabekfortheirgenerousresearchdonations. Moreover, only 23% of patients with untreated OMS had Authordetails CSF BAFFconcentration >2 SD above the control mean. 1DepartmentofNeurology,NationalPediatricMyoclonusCenterand This shows significant overlap of CSF BAFF values be- NeuroimmunologyLaboratory,andSouthernIllinoisUniversitySchoolof tween patients and controls. Our current research Medicine,POBox19643,Springfield,IL62794-9643,USA.2FlowCytometry Center,SouthernIllinoisUniversitySchoolofMedicine,POBox19626, explores the hypothesis that inflammatory cytokines, Springfield,IL62794-9626,USA.3StatisticsandResearchConsulting,Southern none of which is elevated in all patients, may have IllinoisUniversitySchoolofMedicine,POBox19623,Springfield,IL greater predictive value in OMS as a biomarker cluster 62794-9623,USA.4DepartmentofPediatrics,UniversityofAlabamaat BirminghamSchoolofMedicine,Birmingham,AL35233,USA.5Cleveland than individually. ClinicFoundation,TheLernerResearchInstitute,MellenCenterforMS In conclusion, BAFF, not APRIL, joins a short list of TreatmentandResearch,MailCodeNC30,9500EuclidAvenue,Cleveland, other putative CSF biomarkers of disease activity in OH44195,USA. pediatric OMS that includes B cell frequency (total B Received:29October2012Accepted:8January2013 cells and B-cell subsets), B-cell chemoattractants Published:16January2013 Pranzatellietal.JournalofNeuroinflammation2013,10:10 Page9of9 http://www.jneuroinflammation.com/content/10/1/10 References Barrvirus-specificoligoclonalbandsinmultiplesclerosisandother 1. TateED,PranzatelliMR,VerhulstSJ,MarkwellSJ,FranzDN,GrafWD,Joseph inflammatorydemyelinatingneurologicaldiseases.JNeuroimmunol2011, SA,KhakooYN,LoWD,MitchellWG,SivaswamyL:Activecomparator- 230:160–163. controlled,rater-blindedstudyofcorticotropin-basedimmunotherapies 22. TremlJF,HaoY,StadanlickJE,CancroMP:TheBLySfamily:towarda foropsoclonus-myoclonussyndrome.JChildNeurol2012,27:875–884. molecularunderstandingofBcellhomeostasis.CellBiochemBiophys 2. PranzatelliMR,TravelsteadAL,TateED,AllisonTJ,VerhulstSJ:CSFB-cell 2009,53:1–16. expansioninopsoclonus-myoclonussyndrome:abiomarkerofdisease 23. XiaoY,MotomuraS,PodackER:APRIL(TNFSF13)regulatescollagen- activity.MovDisord2004,19:770–777. inducedarthritis,IL-17productionandTh2response.EurJImmunol2008, 3. PranzatelliMR,TateED,McGeeNR,TravelsteadAL,RansohoffRM,NessJM, 38:3450–3458. ColliverJA:KeyroleofCXCL13/CXCR5axisforcerebrospinalfluidBcell 24. DoiM,TakedaT,SakuraiY,KatoJ,HayashiT,FukudaK,TakaseT,ShimaM: recruitmentinpediatricOMS.JNeuroimmunol2012,243:81–88. AlteredimmunoglobulinAandMlevelsassociatedwithchangesin 4. PranzatelliMR,TateED,HoefgenER,VerhulstSJ:Interferon-inducibleprotein BAFFandAPRILafteradministrationofintravenousimmunoglobulinto 10(CXCL10)isincreasedinthecerebrospinalfluidofchildrenwith treatKawasakidisease.JInvestigAllergolClinImmunol2010,20:413–418. opsoclonus-myoclonussyndrome.AnnNeurol2007,Suppl11:S112–S113. 25. PranzatelliMR,TateED,TravelsteadAL,VerhulstSJ:Chemokine/cytokine 5. PranzatelliMR,SlevPR,TateED,TravelsteadAD,ColliverJA,JosephSA: profilingafterrituximab:reciprocalexpressionofBCA-1/CXCL13and Cerebrospinalfluidoligoclonalbandsinchildhoodopsoclonus- BAFFinchildhoodOMS.Cytokine2011,53:384–389. myoclonus.PediatrNeurol2011,45:27–33. 26. KrumbholzM,SpecksU,WickM,KalledSL,JenneD,MeinlE:BAFFis 6. DaridonC,YouinouP,PersJO:BAFF,APRIL,TWE-PRIL:who’swho? elevatedinserumofpatientswithWegener’sgranulomatosis.J AutoimmunRev2008,7:267–271. Autoimmun2005,25:298–302. 7. SchneiderP,MacKayF,SteinerV,HofmannK,BodmerJL,HollerN,Ambrose 27. StohlW,MetyasS,TanSM,CheemaGS,OamarB,RoschkeV,WuY,Baker C,LawtonP,BixlerS,Acha-OrbeaH,ValmoriD,RomeroP,Werner-FavreC, KP,HilbertDM:InverseassociationbetweencirculatingAPRILlevelsand ZublerRH,BrowningJL,TschoppJ:BAFF,anovelligandofthetumor serologicalandclinicaldiseaseactivityinpatientswithsystemiclupus necrosisfactorfamily,stimulatesBcellgrowth.JExpMed1999, erythematosus.AnnRheumDis2004,63:1096–1103. 189:1747–1756. 28. LePottierL,SapirT,BendaoudB,YouinouP,ShoenfeldY,PersJO: 8. RickertRC,JellusovaJ,MileticAV:Signalingbythetumornecrosisfactor Intravenousimmunoglobulinandcytokines:focusontumornecrosis receptorsuperfamilyinB-cellbiologyanddisease.ImmunolRev2011, factorfamilymembersBAFFandAPRIL.AnnNYAcadSci2007, 244:115–133. 1110:426–432. 9. KrumbholzM,TheilD,DerfussT,RosenwaldA,SchraderF,MonoranuCM, 29. VallerskogT,HeimbürgerM,GunnarssonI,ZhouW,Wahren-HerleniusM, KalledSL,HessDM,SerafiniB,AloisiF,WekerleH,HohlfeldR,MeinlE:BAFFis TrollmoC,MalmströmV:DifferentialeffectsonBAFFandAPRILlevelsin producedbyastrocytesandup-regulatedinmultiplesclerosislegionsand rituximab-treatedpatientswithsystemiclupuserythematosusand primarycentralnervoussystemlymphoma.JExpMed2005,201:195–200. rheumatoidarthritis.ArthritisResTher2006,8:R167. 10. ThangarajhM,KisiswaL,PirskanenR,HillertJ:TheexpressionofBAFF- 30. FühlhuberV,BickS,KirstenA,HahnA,GerrietsT,TschernatschM,KapsM, bindingreceptorsisnotalteredinmultiplesclerosisormyasthenia PreissnerKT,BlaesF,AltenkämperS:ElevatedB-cellactivatingfactorBAFF, gravis.ScandJImmunol2007,65:461–466. butnotAPRIL,correlateswithCSFcerebellarautoantibodiesinpediatric opsoclonus-myoclonussyndrome.JNeuroimmunol2009,210:87–91. 11. MackayF,SilveiraPA,BrinkR:BcellsandtheBAFF/APRILaxis:fast-forward onautoimmunityandsignaling.CurrOpinImmunol2007,19:327–336. 31. KreuzalerM,RauchM,SalzerU,BirmelinJ,RizziM,GrimbacherB,PlebaniA, LougarisV,QuintiI,ThonV,LitzmanJ,SchlesierM,WarnatzK,ThielJ,Rolink 12. George-ChandyA,TrysbergE,ErikssonK:RaisedintrathecallevelsofAPRIL AG,EibelH:SolubleBAFFlevelsinverselycorrelatewithperipheralBcell andBAFFinpatientswithsystemiclupuserythematosus:relationshipto numbersandtheexpressionofBAFFreceptors.JImmunol2012, neuropsychiatricsymptoms.ArthritisResTher2008,10:R97. 188:497–503. 13. HamzaouiK,HoumanH,HentatiF,HamzaouiA:BAFFisup-regulatedin centralnervoussystemofneuro-Behcet’sdisease.JNeuroimmunol2008, 200:111–114. doi:10.1186/1742-2094-10-10 Citethisarticleas:Pranzatellietal.:BAFF/APRILsysteminpediatricOMS: 14. HopiaL,ThangarajhM,KhademiM,LaveskogA,WallströmE,Svenungsson relationtoseverity,neuroinflammation,andimmunotherapy.Journalof E,AnderssonM:Cerebrospinalfluidlevelsofaproliferation-inducing Neuroinflammation201310:10. ligand(APRIL)areincreasedinpatientswithneuropsychiatricsystemic lupuserythematosus.ScandJRheumatol2011,40:363–372. 15. NgLG,SutherlandAP,NewtonR,QianF,CacheroTG,ScottML,Thompson JS,WhewayJ,ChtanovaT,GroomJ,SuttonIJ,XinC,TangyeSG,KalledSL, MackayF,MackayCR:Bcell-activatingfactorbelongingtotheTNFfamily (BAFF)-RistheprincipalBAFFreceptorfacilitatingBAFFcostimulationof circulatingTandBcells.JImmunol2004,173:807–817. 16. PranzatelliMR,TateED,HoefgenER,SwanJA,ColliverJA:Therapeutic down-regulationofcentralandperipheralB-cell-activatingfactor(BAFF) productioninpediatricopsoclonus-myoclonussyndrome.Cytokine2008, 44:26–32. 17. KivisäkkP,LiuZ,TrebstC,TuckyB,WuL,StineJ,MackM,RudickRA,Campbell JJ,RansohoffRM:Flowcytometricanalysisofchemokinereceptor expressiononcerebrospinalfluidleukocytes.Methods2003,29:319–325. Submit your next manuscript to BioMed Central 18. SellamJ,RouanetS,Hendel-ChavezH,AbbedK,SibiliaJ,TebibJ,LeLoëtX, and take full advantage of: CombeB,DougadosM,MarietteX,TaoufikY:BloodmemoryBcellsare disturbedandpredicttheresponsetorituximabinpatientswith rheumatoidarthritis.ArthritisRheum2011,63:3692–3701. • Convenient online submission 19. PiazzaF,DiFrancescoJC,FuscoML,CortiD,PirovanoL,FrigeniB,Mattavelli • Thorough peer review L,AndreoniS,FrigoM,FerrareseC,TrediciG,CavalettiG:Cerebrospinal • No space constraints or color figure charges fluidlevelsofBAFFandAPRILinuntreatedmultiplesclerosis.J Neuroimmunol2010,220:104–107. • Immediate publication on acceptance 20. RaghebS,LiY,SimonK,VanHaerentsS,GalimbertiD,DeRizM,FenoglioC, • Inclusion in PubMed, CAS, Scopus and Google Scholar ScarpiniE,LisakR:Multiplesclerosis:BAFFandCXCL13incerebrospinal • Research which is freely available for redistribution fluid.MultScler2011,17:819–829. 21. FranciottaD,DiStefanoAL,JariusS,ZardiniE,TavazziE,BalleriniC, MarchioniE,BergamaschiR,CeroniM:CerebrospinalBAFFandEpstein- Submit your manuscript at www.biomedcentral.com/submit