Primary chronic cold agglutinin disease: An update on pathogenesis, clinical features and therapy

Chronic cold agglutinin disease (CAD) is a subgroup of autoimmune hemolytic anemia. Primary CAD has traditionally been defined by the absence of any underlying or associated disease. The results of therapy with corticosteroids, alkylating agents and interferon-a have been poor. Cold reactive immunoglobulins against erythrocyte surface antigens are essential to pathogenesis of CAD. These cold agglutinins are monoclonal, usually IgMκ auto antibodies with heavy chain variable regions encoded by the VH4-34 gene segment. By flowcytometric and immunohistochemical assessments, a monoclonal CD20+κ+B-lymphocyte population has been demonstrated in the bone marrow of 90% of the patients, and lymphoplasmacytic lymphoma is a frequent finding. Novel attempts at treatment for primary CAD have mostly been directed against the clonal B-lymphocytes. Phase 2 studies have shown that therapy with the chimeric anti-CD20 antibody rituximab produced partial response rates of more than 50% and occasional complete responses. Median response duration, however, was only 11 months. In this review, we discuss the clinical and pathogenetic features of primary CAD, emphasizing the more recent data on its close association with clonal lymphoproliferative bone marrow disorders and implications for therapy. We also review the management and outline some perspectives on new therapy modalities.


Introduction
Autoimmune hemolytic anemia (AIHA) is classified into warma nd cold reactive antibody types. Several entitiesare recognized within the cold antibody group; chronic cold agglutinin disease (CAD), acute cold antibody mediated AIHA complicating Mycoplasma pneumoniae or virali nfections, and paroxysmal cold hemoglobinuria. Only CAD will be furtheraddressed in this review.C AD has traditionally been classified into ap rimary or idiopathic type which has been regarded unrelated to underlying conditions, and a secondaryt ypea ssociated with malignant disease, most often lymphoma [1 -3]. The term "cold" is primarily derived from the immune biology of CAD, not from the clinical features whichw ill be discussed in detail below [4,5].
Cold hemagglutination was first reported by Landsteiner in 1903 [6] andfound to occur in human beings in 1918 [7]. The association of cold hemagglutination with hemolysis was described in 1937 by Rosenthal and Corten [8]. During the 1960s, Dacie [9]a nd Schubothe [10] published systematic descriptions of 16 CAD patientseach. The autoantibodies responsible forh emagglutinationa tl ow temperatures, cold agglutinins (CA), maybefound in the sera of healthy subjects as well as in patientsw ith AIHA of the cold reactive types [5,9]. CA bind to erythrocytes urface antigens at at emperature optimumo f0-48 C [ 4,11]. In contrast to polyclonal CA in healthy individuals, monoclonal CA often haveahigh-thermal amplitude, which contributes to their pathogenicity at temperatures approaching37 8 C [4,11 -13].
Management was largely unsatisfactoryu ntil the last decade [3,19,20]. Recently,considerable progress has been madei nt he knowledge of clinical features, bone marrow pathology, humorala nd cellular immunology,c andidate targets fort herapy,a nd more efficient management. We will review relevant findingsb yo ur groupa nd others on clinical, immunological andp athogenetic features of primary CAD.B ased on these results, we will provide an overview of more recentt herapeutic measures and give somes uggestions for furthers tudies.

Epidemiologica nd clinical features
In single-center series, primaryC AD has been found to account for1 3-15% of thec ases of AIHA [1,21,22]. In ap opulation-basedc linical study of primaryCAD in Norway,the prevalence was found to be 16 per million inhabitants and the incidence rate 1 per million inhabitants per year [3]. Little is known aboutp ossible geographic variations. Median age of CAD patients was 76 yearsand median age at onset of symptomsw as approximately6 7y ears [ 3].T he male/female ratio has been reported to be 0.5 -0.6 whichisnot very different from amale/female ratio of 0.72 in an age-matched general population. The frequency of auto-immunedisordersotherthan CAD does probably not differ from what is to be expectedin an elderly population with some female predominance [3,4]. Median survival was about1 2.5 yearsf rom diagnosis and median age at deathwas 82 years, which implies al ife expectancyi nt hese patients similar to that of an age-matched general population [3].
Cold-induced circulatorys ymptoms, although often not emphasized by physicians, are considered typicalf or CAD [10,17]. We found that more than 90%o fp atients with primaryC AD hads uch symptoms, ranging from moderatea crocyanosis to severe Raynaud phenomena precipitated even by very slight cold exposure [3]. Although the importance of cold exposure for exacerbation of hemolysis has been questioned [18], characteristic seasonal variations are fairly well documented in the literature [9,10,17,23].
According to review articles, anemiai nC AD is variable andusually not severe [9,17]. However, this is definitively not always the case. Fiveo f1 6p atients described in an early reporth ad minimum hemoglobin (Hb) levelsb elow 7.0 g/dla nd oneb elow 5.0 g/dl [ 10]. In as erieso f8 6p atients, we found a median Hb level of 8.9 g/dl, and one-third of the patientsh ad Hb levels at presentation ranging from 4.5 to 8.0 g/dl. Approximately,5 0% of the patients were considered transfusion dependenta ts ome time during the course of the disease [3]. Paradoxically, hemolysis is enhanced during febrile illnesses in about two-thirds of the patients [ 3,4,24,25]. We found no overall changeo ver time in Hb levels and parameters of hemolysis. Hb levels decreased, however, by as mucha s7 .7 g/dl in individual patients during a median observation time of five yearsand increased by as much as 5.8 g/dlinothers [3]. Thus,CAD tends to be an on-progressive disease in most patients, althoughfl uctuations in the clinical manifestations are prevalent ( Figure 1) and it shouldb ee mphasized that there are considerable individual variations. The figures clearly document that CAD is nota n "indolent" disease in terms of major clinical symptoms and quality of life.

Immune biology
In the great majority of CAD patients, CA are specific for the Iantigen, an erythrocyte surface carbohydrate macromolecule [ 26,27]. Anti-Pra nd anti-P specificities havea lso been described [ 27,28]. The concept of CA should notb ec onfused with that of cryoglobulin, althougho bvious similarities do exist between primaryC AD and cryoglobulinemia type I and II [29]. Immunoglobulins have occasionally been described that possess both CA and cryoglobulin properties [30,31].
Them echanismso fr ed-cella gglutination and subsequent destruction havebeen elucidated in detail [13,15,16,32]. Cooling of bloodd uringp assage through acral parts of the bodya llows CA to bind to erythrocytes and precipitate agglutination ( Figure 2). The antigen -antibody complex induces Cb inding and activation via the classical pathway as shown in Figure 3. Thus, C1 esterase activates C4 andC 2, generating C3 convertase which leads to the formation of C3b.Upon subsequent warming to 378 Cwhenthe blood returns to the central parts of the body, CA detachesf romt he cell surface allowing the agglutinated erythrocytes to separate from eacho ther, while C3b remainsb ound. Some C3b-coated erythrocytes are sequesteredand destroyed by C3-receptor bearing reticulo-endothelial cells, mainly in the liver.O nt he surface of the surviving erythrocytes, C3b is cleaved into C3c and C3d, leaving large numbers of C3d macromolecules on the cell surface. Cactivation may proceed beyond this step, resultinginC5activation by C3b andf ormationo ft he membrane attackc omplex C5-9 with intravascular cell lysis. Most evidence suggest, however,t hatt he majorm echanism of hemolysis in stablep atientsi st he hepatic sequestration of C3b-coated erythrocytes [5,13,15,16,32].
The thermal amplitude, defined as the highest temperature at which the antibody bindst he antigen, appearst ob em ore important than the titer with respect to the pathogenicity of CA [12,13,33]. The CA found in someh ealthy individuals are usually presentinlow titers, and titersinexcess of 256 are very uncommon in this group [27,34]. Furthermore, the thermal amplitude of cold-reactive autoantibodies in healthy subjects does note xceed1 5-208 Ca nd, therefore, they are of no clinical significance [27].
Christenson [35]a nd co-workersf ound in 1957 that CA may sometimes be seen as an abnormal peak in the g -region by electrophoretic separation of serum proteins on cellulosecolumns. Fudenberg and Kunkel showed that these antibodies usually haveahigh (19 S or 1000 kDa) molecularw eight [36]. Later, Harboe and co-authorsfurther characterizedt he CA in CAD as monoclonal IgMk [37 -39].Inarecentstudy of sera from 172 patientsw ith monoclonal IgM associated with avariety of clinical disorders, CA were identified in 10 sera (8.5%) [40].B oth, pentameric and significant levels of hexameric IgM have been detected in sampleso fp urified CA from CAD patients [41]. Absence of Jchains seems to enhance the formation of hexamericI gM andh as been interpreteda sa deleterious feature of IgM-mediated disorders, resulting in ah ighera bility to activate the Cc ascade and therebyi nahigher lytic efficiencyofI gM [41,42].
In ourp opulation-based descriptives tudy of primaryC AD,amonoclonal band was detected by electrophoresis and immunofixation in sera from 79 (94%) of 84 patients with available data [3]. The monoclonal immunoglobulin was of the IgM class in 71 patients( 90%), IgA and IgG in three patients (3.5%) each,w hile two patients (2.5%) had clonal bandso fb othI gG andI gM.T he light chain restriction was k in 74 patients( 94%), l in two (2.5%) and unknowni nt hree (3.5%). Since CA detach from the erythrocytes when the blood returns to the bodyc ore, specific direct antiglobulin test (DAT)i su sually negative when performed with anti-IgM. DATispositive for C3d by definition ( Figure 3) [3,13,15,16]. In ourr etrospective study [3], specific DATfor IgG was negative in 64 patients(79% of those with available data), while erythrocyte-bound IgG was detected in the remaining17(21%).Fivepatientswith monoclonal IgG or biclonalIgM and IgG in serum all displayed IgG on the erythrocyte surface. It has not been established whether this cell-bound IgG is a Figure 3. During passage through acral blood vessels, cooling allows IgM cold agglutinin to bindt oe rythrocytes, causing agglutination and binding of complement C1 complex. C1 esterase activates C4 and C2, generating C3 convertase which binds and splits C3, leading to deposition of C3b on the erythrocyte surface. Upon subsequent warming,I gM removes from the cell surface and the agglutinatedc ells are detached from each other, while C3b remains bound. C3b may in turnactivate C5, leading to the formation of the membrane attack complex and intravascular cell lysis. Most destruction of C3b-coated erythrocytes, however, is mediated by reticulo-endothelial cells in the liver [15,16,32]. Intrahepatic conversion of C3b is responsible for the deposition of C3d on the surviving erythrocytes which are released into the systemic circulation. polyclonal reactive antibodyormonoclonal CA of the IgG type.
During maturation, each B-lymphocyte constructs its specific immunoglobulin heavy chain by assembly of coding sequences from the variable (V H ), diversity (D), and joining (J H )g enes egments. The diversity created by this recombination process is further increased by enzymatic modification at the cut ends of the gene segments, followed by the event of somatic hypermutation, typicallyo ccurring in the hypervariable segments of V H genes. Pascual, Thorpe, Stevenson and others have shown that anti-I CA found in serum samples from patientsw ithp rimary CAD are preferentially encoded by the V H 4-34 gene segment, formerly termed V H 4.21 [43,44]. This gene segmenta ppearst ob eo verrepresenteda mong the coding unitrepertoire, although it accounts for avery small fraction of normal circulating immunoglobulins [43,45]. We assessed the frequencyo fV H 4-34 gene expression by testing sera from 11 CADpatientswith hemagglutinationi nhibitiona ssayu sing ther at monoclonal anti-idiotypic antibody 9G4, which is specific for V H 4-34 encodedp rotein. All patient sera were confirmed to be idiotope positive [4]. In contrast, "naturally" occurring CA in healthy individuals, as well as CA artificially induced by Rhesus (D) immunization, areo ften derivedf rom V H gene segments othert han V H 4-34 [45,46].

"Paradoxical" exacerbationd uring febrile illnesses
ReducedCfactor levels in CAD were describedearly by Jonsen [47] and others.In1998, Ulvestad reported on apatient who experienced that during the course of the disease, the initial cold-induced exacerbations were gradually substitutedf or "paradoxically" enhanced hemolytic anemia during febrile episodes [24]. The C4 levels decreased steadily andeventually becameu ndetectable, and the in vitro hemolytic activity of serum( CH50) declinedt oz ero.I na subsequent study,w ea ssessed Cp rotein levels in 15 CAD patientsand found reduced levels of C3 in nine and C4 in 11 patients, six of whom hadlow CH50 [4]. Basedo nt he records,e xacerbation of hemolysis during acute phase reaction had occurred in five patients. In our population-basedretrospective study, 64% of CAD patients(50 of 68 patientswith available data on such deteriorations) reported exacerbation of hemolytic anemiad uring febrile illnesses [3].
In order to further investigate these phenomena, we undertookalongitudinal, prospective, 12 month follow-up study of one single patient with "paradoxical" exacerbationso fh emolysis [25]. In the absence of any acutee vents, low C3 and undetectableC4levels were confirmed. We observed a non-functionalc lassical Cp athway and an ormal alternativep athway.E xacerbationo fh emolytic anemiao ccurred during pneumonia ando nce more following ah ip fracture with subsequent surgery, and was paralleled by increased CRP levels.D uring each acutee vent the serum IgM levels declined temporarily,and after the hip fracture we recorded increased C3 levels, detectable C4, significantly increased levels of thep ro-inflammatoryc ytokinesi nterleukin-6, tumor necrosis factor-a and interferon-g ,and slightly increased interleukin-1b [25]. The most plausible explanation for these observations is that amajority of CAD patientshavelow levels of C3 and especially C4 during steady state due to acontinuous consumption. Cf actor levels, in particular low C4 levels, are ratelimiting for hemolysis. During acutep hase reactions, C3 and C4 levels increase due to an enhanced production,r esultingi ne xacerbation of hemolysis.
The findings of Cc onsumptiona nd depletion may havec linical implications. First, administration of C-containing plasma products shouldp robably be avoided. Second, these data explain why amajority of patientsw ith CAD havee xacerbations during conditions associated with acutephase reaction. Third, a non-functionalc lassical Cp athway may affectt he therapeutic potential of monoclonal antibodies in CAD [48 -50].

Clonal B-lymphocytes in primaryCAD
Pathogenic B-lymphocyte clones in CAD haveb een suspected or postulatedf or decades, basedo nt he findings of monoclonal IgMk CA in most, if not all patients [ 3,10,31,[37][38][39].M orer ecently,i th as been possible to verify such cell clones directly. Flow cytometric investigations by Silberstein and co-workersd isclosed B-cell clones in at least some patients [ 51]. In 1995, we reportedt he findings of lymphoplasmacytic lymphoma in the bone marrow of threec onsecutive patientso therwisec lassified as having primary CAD [ 52]. In as ubsequent study by our group, patientsw ith no clinical or radiological evidence of an underlying lymphoma were examined by flow cytometric immunophenotyping of bone-marrowa spirates as well as morphological and immunohistochemical assessment of trephine biopsies [31]. Al ymphoproliferative bone-marrow disorder characterized by clonal CD19 þ CD20 þ k þ lymphocytes was detected in 10 of 11 patients.
In arecent retrospective study,the medical records of 86 patientso therwise classifieda sh avingp rimary CAD were re-examined with regard to the presence of ac lonal lymphoproliferative bone-marrowd isorder [3]. MonoclonalC D20 þ k þ lymphocytes were found in the bone marrow of most patients in whom afl ow cytometric assessment hadbeen performed. Based on previously published data [31,53], a k / l ratio . 3.5 by flow cytometryw as considered strongly indicative of ac lonall ymphoproliferativeB -celld isorder. The median k / l ratio was 7.8 (range 0.9 -186), and ar atio higher than 3.5 was found in 36 (90%) of 40 patientswith available data [3]. Data on bone-marrow histology are shown in Ta ble I. Morphologic and immunohistochemicals igns of non-Hodgkin'sB -cell lymphoma were found in 50 (76%) of 66 patients with available information (Figure 4). Applying the WHO classification [54], 33 patientshad lymphoplasmacytic lymphoma (50% of patients with available histology data and 66% of those with ad emonstrable clonal lymphoproliferative bone marrowd isorder).
According to recent criteria, Waldenström's macroglobulinemia (WM) is defined as lymphoplasmacytic lymphoma of theb onem arrowc ombinedw ith monoclonal IgM at any serumc oncentration [55]. Whent hese criteria were applied, 50% of CAD patientswith available immunoglobulin and histology data met the diagnostic criteria for both primary CAD and WM [3]. On the other hand, we haveobserved an occasional CAD patient with monoclonal IgMk for more than 21 yearsw ithout any demonstrable clonal B-cell population as repeatedly assessed by flow cytometrya nd immunohistochemistry. Transformation to diffuse large B-cell lymphoma appearstobea rare event, occurring in 3-4% of patientswith primary CAD after ad iseased uration of 10 years [ 3].
CAD patientsdiagnosed by us and others to have a low-grade lymphoproliferative bone marrow disorder undoubtedly represent the same majority that used to be classified as having primaryC AD [9,27,31,48]. Except in the uncommon event of transformation, these clonal lymphoproliferative disorderss eldom, if ever, show features of clinically overtlymphoma even after decades [ 3]. Furthermore, most of the rare patients traditionally classifieda sh aving secondary CAD suffer from ar eadily demonstrable lymphoma, often of an aggressive type, that may be associated with IgMl as well as IgMk CA [58,59]. Therefore,we still thinki ti sa ppropriate to apply the termp rimary CAD in patients not showing the classical features of the secondaryt ype.

Diagnosis
Basedonthe characteristics discussed in the preceding paragraphs and available literature [1,3,5,10,17,27], the criteria shown in Ta ble II should be used to define primaryC AD.T he demonstration of am onoclonal serum immunoglobulin and ac lonal,l ymphoproliferative bone-marrow disorder should notberegarded as an absolute prerequisite for diagnosis, since the cell clones mayb et oo small to manifest themselves by histopathologicfi ndings or be detected by flow cytometry, electrophoresis andi mmunofixation. Such verification of clonalitydepends to alarge extent on sensitivity and, in particular with respect to the electrophoretic findings, on optimalp reparation and examination of specimens.
Ta ble II lists the diagnostic examinationst hat shouldb ep erformed.P roblems in measurement of bloodcellcounts may sometimesbeencountered due to agglutination, but pre-warming of the EDTAbloods amples when necessaryw ill eliminate such difficulties. Fors erum immunoglobulin analyses, including cold agglutinin titration, electrophoresis, immunofixation andq uantification of immunoglobulin classes, it is essential to keep blood specimens at 378 Cf rom sampling until serum has been removed from the clot. Assessment of thermal amplitude may be informative, but is hardly needed for diagnostic or therapeutic decisions. Bone-marrowe xamination by flow cytometryofaspirateand careful assessment of a trephine biopsy sample should always be performed.

Management of primaryC AD
According to literature, counseling on cold avoidance shouldb et he mainstay in management of primary CAD [17,19,60]. In 63 (73%) of 86 patients reported by us, however, the physician and/or the patient had notp erceived suchm easuresa ss ufficient [3]. Corticosteroids anda lkylatinga gentsa re usually ineffective [3,19,20]. Improvement following interferon-a single agent therapy has been reported in a small retrospective series,but in anotherseries none of the patientsr esponded [ 61,62]. Furthermore, no response to cladribine monotherapy was observed in a small,p rospective study,b ut the doses of cladribine applied in this trial were low [63]. The potential of splenectomy hasn ot been studied systematically,b ut theoreticalc onsiderationsa nd clinical experience strongly discourage its use as atherapeutic procedure [3,15,16,19,32]. The recognition of primaryC AD as ac lonal lymphoproliferative CD20 þ B-cell disorder and the success of treatment with the monoclonal anti-CD20 antibody rituximabi nC D20 þ non-Hodgkin's lymphoma [64,65] made us ando theri nvestigators hypothesize that rituximab therapy might also be effective in CAD.The adverseeffects of rituximab are different from those of most cytotoxic drugs and less severe [64,65], andt he B-lymphocyte elimination is not cell cycle dependent [ 66]. One small and two somewhatl argerp hase 2t rials [48,67,68] haveb een published in addition to anumberofcase reports [69].
In the first 16 case reports published, all patients improveda fter rituximab therapy, andahigh proportion of ther esponsesw erec lassified as complete [70,71]. The explanation for such ah igh response rate is probably that response rates estimated from case reports are likely to be strongly influenced by publication bias, lack of strict disease definitions, and heterogeneous or lacking response criteria.
We reported on 37 courses of rituximab single agent therapy administered to 27 patientsw ith primary CAD in ap rospective,u ncontrolled trial [48]. Each eligible patient received acourse of rituximab at adose of 375 mg/m 2 on day 1, 8, 15 and 22. Re-treatment wasp ermitted in patients whor espondeda nd subsequentlyr elapsed. Ther esponsec riteriaa re summarizedi nT able III. Fourteen of 27 patients responded to their first course of rituximab, andsix of ten relapsed patientsr espondedt or e-treatment. In both groups combined, responses were achieved after 20 of 37 courses, resultingi na no verall response rate of 54%. We observed one completea nd 19 partial responses. Respondersa chieved am edian increase in Hb levelso f4 .0 g/dl and am edian decrease in IgM levels by 54%. Clinical and laboratorydata indicated a benefit even in somep atients classified as nonresponders. Median time to response was 1.5 months (range,0 .5-4.0)a nd mediano bserved response duration was 11 months (range, . No serious adversee vents occurred. The results of as imilar trial in 20 patients by Schöllkopfand co-workersfitinvery well with our findings, althought hey reported a shorter response duration [ 68]. Some minor discrepancies between the results of the two studies maybe explained by slightly different inclusion and response criteria.

Possible directions for future research
Theb enefita chievedb yr ituximab single agent therapy in CAD is limited by a4 5-50% failurer ate and relatively shortresponse duration. Furtherstudies are warranted, therefore, in order to explain the variable effect of rituximab therapy,i dentify possible predictors,a nd improveo nr esponser ates and response duration. Even when aC D20 þ k þ lymphocyte clone can merely be detected and monoclonal IgM is present at low levels, patients may haveaclinicallysevere disease with ah igh CA titer or CA with high thermal amplitude [4,31]. Small B-cell clones that produce deleteriousp roteins arew ellk nown,a nd these conditions are veryo ften difficult to treat effectively [72,73]. Thus,a ne xplanation for the difficulties in achievingr emissions may be that in most cases, small cell clones produceb iologically highly active antibodies that must be nearly eradicated in order to achieve clinical improvement. On the otherh and, rituximab can induce good partial remissions even in patientsw ho achieve only am odestd ecrease in monoclonal IgM by about5 0% [48]. This may indicate that reduction of the lymphocyte clone and the concentrationofthe auto-antibody may not be the only pathway of therapeutic effect. In WM, the monoclonal B-cell population can induce expansion of circulating,p olyclonal B-lymphocytes [74]. To our knowledge, no studiesh aveb een done to explore the possible role of this phenomenon in CAD or any implications for therapy.
Rituximabhas been shown to killCD20 þ cellsbyat least threem echanisms; C-dependent cytotoxicity (CDC),a ntibody-directedc ellularc ytotoxicity (ADCC), andi nductiono fa poptosis by direct intracellular signaling [66]. Some in vitro and in vivo data indicate that CDC is an essential mechanism of actiona nd, therefore, the reduced availability of C proteins in many patientsw ith CAD mayt urno ut to be of clinical importance [49,50]. In our prospective trial, however, we found no association between C3 or C4 levelsand response to rituximab therapy [48]. The administration of interferon-a mayr aise serum C4 levels [75] and up-regulateC D20 expression on the surface of B-cells [76,77]. In ourrituximab study,we intended to evaluate whether combining rituximab and interferon-a could improveo ne fficacy [ 48]. Patient or physician preferences, however, resulted in only five patientsreceiving the combination,and it was impossible to put forward any firms tatementso nt he efficacyofc ombiningrituximabwith interferon-a .
Elimination of CD20 þ lymphocytes by anti-CD20 induced ADCCrequires binding of the Fc-domain of the CD20-bound antibody to the Fc-receptor of effector cells [66]. Polymorphism in the IgG Fcg receptor IIIa (Fcg -RIIIa) geneh as been proposed to influence the depletion of B-lymphocytes by rituximab [78,79]. Although the possible consequences of such genetic variations remain to be confirmed in CAD, clinical studies haves uggestedt hat Fcg -RIIIa polymorphism maye xplain the variability in the response to rituximab therapy in WM [80].
Purine analogueshaveshown aremarkable efficacy in low-grade lymphoproliferative diseases, including WM [81,82]. Although purine analogues do not seem promising in CAD when administered as monotherapy [63], remission has been reported in two single casesa fter thea dministrationo fc ladribine and fludarabine, respectively [3,83]. In as mall, prospective study,cladribinewas shown to reduce the number of clonal cells, althoughnot resultinginany significant clinical improvement [63]. As ynergistic effecto f fludarabine and rituximab haveb een shown in a follicularl ymphomaB -celll ine resistant to the cytotoxic activity of either drug alone, probably mediated through ad own-modulationo fm embrane CD55 [ 84]. In WM,p urine analogue and rituximab combination therapy has resulted in higher response Ta ble III. Response criteria used in therapeutic trials.

Complete response
Absence of anemia No signs of hemolysis Disappearance of clinical symptomso fCAD Undetectable monoclonal serum protein No signs of clonal lymphoproliferation as assessed by bone marrow histology, immunohistochemistryand flow cytometry Partial response Astable increase in hemoglobin levels by at least 2.0 g/dl or to the normal range Areduction of serum IgM concentrationsb yat least 50% of the initial levelo rt othe normal range Improvement of clinical symptoms Transfusion independence No response Failure to achievec omplete or partial response In order to qualify for any given response level,a ll criteria havetobefulfilled rates andm ore prolongedr emissions as comparedt o purine analogue single agent therapy [85]. Fludarabine may induce AIHA, but this adverseevent seems to occur mainly in patients with chronic lymphocytic leukemia, and recento bservations mayi ndicate that the additionofrituximabwill reduce the risk [86].
We are now running ap hase 2s tudy on the safety and efficacyofrituximab and fludarabine combination therapy in primary CAD [87], still using the response criteria listed in Ta ble III. By February2007, response evaluation was possible in the first nine patients, mediana ge 72 years( range, 59 -85).S ix had previously received rituximab single-agent therapy, resulting in one complete response and onep artial response, while four hadb een non-responders. Following combinationt herapy,f ourp atients achieved ac ompleter esponse,f oura chieveda partialr esponsea nd oned id notr espond. Hematologic toxicity was observedi nf our patients (grade 2, 3a nd 4, respectively)a nd infection grade 2, nausea and dermatitis in one each. Thus, rituximaba nd fludarabine combination therapy seems feasible even in elderly patients with CAD. Response rates are promising and suggestive of a higher efficacy, but superiority over rituximabs ingleagent therapy remainst ob ep roven in an extended study.
Since the hemolytic activity of CA is Cd ependent, one might consider directCmodifying agents as possible therapeutic options. Infusion of the humanized, monoclonal anti-C5a ntibody eculizumab has recently been documenteda sapowerfult herapeutic measure in paroxysmal nocturnal hemoglobinuria [88]. No reports haveb een published on its use in CAD.B ased on the mechanisms of CA mediated C activation andh emolysis discussed in the previous paragraphs, however, ones houldt heoretically not expectapronounced effect in stable CAD patients. Prospectivet rialsm ay still be justifiedi n refractoryp atientsw ith severe hemolysis or acute exacerbations.