Systematic review for the serological testing for cold agglutinins: The BEST collaborative study

The investigation of suspected cold agglutinins (CA) can be challenging and highly resource-intensive. While progress has been made in the clinical understanding and management of patients with CA with new therapeutic options,1-5 there has been less systematic evaluation of the diagnostic approaches for CA leading to inconsistent practices. CA are IgM autoantibodies with specificity for red blood cell (RBC) surface antigens and are classically described as RBC autoantibodies with an optimal binding temperature of approximately 4°C.1, 2, 6 Clinically active CA can initiate the activation of the classical complement pathway leading to intra- or extravascular hemolysis, also called cold antibody-mediated autoimmune hemolytic anemia (cAIHA).1, 2, 6 cAIHA can be divided into cold agglutinin disease (primary CAD) or cold agglutinin syndrome (CAS). CAD is recognized as a well-defined low-grade clonal lymphoproliferative disorder, classified as a separate entity in the recent international consensus classifications.7, 8 CAS on the other hand, is more heterogeneous and found secondary to an underlying disease, including B-cell lymphoma, autoimmune disorders, and specific infections.2, 6 It is important to distinguish clinically significant CA from insignificant CA, often incidentally detected during ABO blood grouping and antibody screening. Recommendations from the first international consensus meeting6 suggest defining cAIHA by hemolysis in combination with a monospecific direct antiglobulin test (DAT) strongly positive for C3d and a CA titer of 64 and greater at 4°C. Other tests, such as the thermal amplitude (TA) to determine the risk of hemolysis at body temperature, are also used in some clinical practices.6, 8 However, attempts to define significant CA with diagnostic studies may be limited by the reference studies for the optimal diagnosis of CA given they are scarce and historical, making applications to current practice challenging. Despite the fact that the DAT is often easily accessible, there is a lack of specificity for the diagnosis of significant CA. Both the monospecific C3d DAT and CA titers only indicate binding of IgM antibodies in vivo and the putative concentration of antibodies, which would not necessarily indicate the ability to activate complement for hemolysis. Testing methodology and sample collection is heterogeneous, leading to pre-analytical error which may confound the diagnosis. Markers of anemia in isolation are also not specific for the diagnosis of hemolysis. The aim of this review is to describe the serological techniques and cut-offs to diagnose CAD or CAS, which would be clinically significant for cAIHA based on the published literature. We sought to inform future research for an evidence-based approach to the diagnostic investigation and interpretation of CA supporting the clinical diagnosis CAD/CAS. MEDLINE, EMBASE, and CENTRAL databases were searched from inception to May 2023. The search strategy was adapted from a prior systematic review on warm autoantibodies, incorporating keywords representing three concepts: (1) cold reactive autoantibodies, (2) serological detection methods, and (3) clinical significance.9 These concepts were combined in the search strategy, outlined in Appendix A. Additional manual searching of reference lists were performed during the full-text review and of reference texts. All articles containing information regarding CA were included, even if CA was not the primary focus of the article. All publication types were eligible if deemed relevant by the investigators. Criteria for exclusion included: (1) case studies and series with fewer than 5 patients, (2) manuscripts only discussing management with no elements regarding testing and/or serological diagnosis, (3) manuscripts detailing only warm autoantibodies including IgM warm autoantibodies, (4) gray literature, and (5) articles not in the English language. After duplicates were removed, three investigators (A.W.S., M.T.S.Y., and M.J.) independently performed the title and abstract screen, where discrepancies were resolved by a third reviewer. All investigators contributed to full-text screening independently, where adjudications done in pairs with discrepancies were resolved by a different study investigator as a third reviewer. Both these steps were performed with a systematic review reference manager (Covidence). All screening steps were done with a calibration exercise to determine concordance between reviewers before screening using Covidence. A standardized data extraction template was created and pre-piloted by two investigators (A.W.S. and M.J.), with input provided from all authors and the final template determined by consensus. All investigators contributed to data abstraction independently, where adjudications done in pairs with discrepancies were resolved by a different study investigator as a third reviewer. Data abstracted included: study characteristics (study methods, aim, number of participants, and underlying clinical disease studied), definitions of CA, definitions of hemolysis, testing methodologies for CA, ancillary testing, the number of participants with CA and/or hemolysis, if tests were associated with clinically significant CAD/CAS (including titer threshold, TA, DAT strength, antigen specificity), and other testing results. Where possible, patients with mixed hemolytic anemias were excluded (DAT C3d and IgG) to only extract data on patients with CAs. Studies were assessed qualitatively as the heterogeneity of patient populations and study designs. The analysis was stratified according to study design, where the proportion of articles supporting and/or having recommendations for serological testing of clinically significant CAD/CAS was determined. Review articles were not included in the primary analyses given these articles did not represent primary evidence and therefore a risk of duplicating results would occur for review articles citing similar studies. The literature search yielded 3585 articles, of which 611 were duplicates. After removal of duplicates, 2974 articles underwent title and abstract screening, resulting in a further 2722 articles being excluded. 252 studies were assessed for full-text eligibility, in which 196 studies were excluded. Fifty-six articles were included in the data abstraction. Details are listed in Figure 1. Included articles by publication type included 2 systematic reviews, 2 cohort studies, 5 case–control studies, 18 case series, and 29 review articles. Approximately half of articles did not contain a definition for CA or hemolysis. More than half of studies did not describe a CA titer for the determination of clinical significance, whereas studies that did describe a CA titer broadly defined the threshold a priori rather than based on the clinical findings of significance. A summary of the results of the case-control studies and case series is displayed in Table 1, with details of individual studies abstracted in Tables 2 and 3 for cohort/case-control studies and case series respectively (with additional detail in Appendix B). Findings of studies are described by methodological study type: systematic reviews, cohort studies, case control studies, case series, and review articles. Two systematic reviews were included. One was performed to assess the heterogeneity in AIHA terminology, including CAD.10 In this systematic review, 13/18 articles assessed provided a definition of CAD. Eight of these articles were from one group of investigators and defined CAD as including all of the following characteristics: chronic hemolysis, positive polyspecific DAT, monospecific DAT strongly positive for C3d (and negative or weakly positive with IgG in 3/8 articles; though strongly and weakly positive were not defined), and CA titers ≥64 at 4°C. The other systematic review included describes the correlation between SARS-CoV-2 infection and vaccination and cAIHA (as described in the review) in 18 and one patient, respectively.11 The classification of cAIHA was based on the individual author's reported diagnosis. DAT results were reported but were not part of the inclusion criteria. CA screening (defined as significant at ≥64, without temperature specifications) was performed in 14 patients in whom CA titers were determined in 9 cases. In 5 patients the TA was performed, and all reacted at 30°C or higher. Clinical relation between serologic test results and the severity of the disease was not described. A single-center retrospective cohort study of CA screening in 14,900 cardiac surgery patients undergoing cardiopulmonary bypass (CPB).12 CA screening was positive in 47 cases (0.3%), with one case previously diagnosed as CAD. Upon further investigation, 27 cases were deemed to be falsely positive, and 20 cases had antibody specificity found. In 44 patients with positive CA screening with available records, 15 received modified CPB to reduce hypothermia; there was no difference in outcomes reported compared to those patients who did not receive modified CPB. The authors also performed a literature review of all published case reports, suggesting that the practice of CA screening was beneficial in 2 reported cases of CA patients undergoing surgery, as the authors felt that CA screening would have changed management. The authors of the cohort study suggest that these 2 cases represent 1.2% of CA patients or <0.01% of the surgeries across their literature search and cohort study. In another cohort study, Juskewitch et al. studied the association between monoclonal antibody (M-protein), M-protein light chain (LC) glycosylation, CA titers, and CA mediated hemolysis in a reference laboratory study group (48 patients) and in a institutional study group (77 patients) undergoing CA titer evaluation.13 CA titer groups were defined as negative (<64), indeterminate (64–512), and positive (>512) at 2–6°C. In the institutional study group, the rates of patients with a clinical diagnosis of CAD were 2%, 75%, and 100% in the different titer groups respectively. Eleven patients were diagnosed with non-CAD hemolysis, of whom 91% had a negative CA titer. DAT results were available of only 50 patients (75%), of whom 26 (52%) had a positive polyspecific result with 92% of these positive for monospecific C3 and 46% positive for monospecific IgG. The rates of DAT positive for polyspecific and monospecific complement were significantly higher in the indeterminate and positive titer groups. There was a significant association between the finding of an M-protein and M-protein with LC glycosylation, with a higher CA titer and a clinical diagnosis of CAD. IgM and kappa were the most commonly found heavy and light chains. Of the five case-control studies, all noted ranges of CA titers and were from different investigator groups (Tables 1 and 2; with additional detail in Appendix B). Only two of the five studies suggested a serological definition of CA. Bendix et al. suggested a CA titer threshold that may be associated with CAD14 by attempting to determine CA ranges in a prospective cohort of blood donors, a retrospective cohort of historical and current patients, and blood donors. That study suggested a titer ≤64 for a lower 95% reference range for their current patient cohort. There was some association between DAT strength and the degree of hemolysis. Wang et al. found an association between CAD/CAS and lower serum complement (C3/C4) levels; with a negative correlation between serum C4 levels and CA/anti-C3d titers. This study defined a positive CA titer as ≥64 a priori, wherein patients with positive CA titers and/or anti-C3d DAT, a negative correlation between serum C3/C4 levels and indirect bilirubin was also found.15 No case-control studies evaluated the clinical significance of TA. Case series also did not offer a uniform serological definition of CA (Tables 1 and 3; with additional detail in Appendix B). A positive C3d + DAT was the most common element used to identify patients with CA, followed by a CA titer threshold. Only 3 studies defined a clinically significant CA titer of ≥64, though this was a threshold determined a priori to include patients in the case series. Six studies reported titers in patients with clinically significant hemolysis, though did not define a specific CA titer threshold that would be associated with clinical significance. Antibody specificity was determined in 13 studies, with 6 suggesting an association with clinically significant CA, notably with anti-I. Ten case series described the use of TA testing, with 3 suggesting an association with clinically significant CA. Five case series found an association with increased DAT strength to increased hemolysis. Of the 29 review articles included, in which some included references to the articles described above, 11 had a definition of CA that included a C3d + DAT,16-26 5 included a CA titer threshold,6, 16, 18, 21, 27 1 included TA,20 and 9 included both a CA titer threshold and TA.17, 19, 23-25, 28-31 Four articles only defined CA as a cold reactive antibody with specificity in keeping with CA,2, 32-34 1 article defined CA as a cold reactive antibody that can agglutinate RBCs where TA is pathogenic,35 and 6 articles did not have a clear serological definition.36-41 One article included three definitions of clinically significant CA observed in the literature, hemolytic anemia with: (1) strongly positive C3d DAT with a CA titer ≥64, (2) C3d DAT positivity only with a CA titer ≥64 and anti-I specificity, and (3) C3d DAT positivity only with cold autoagglutination up to at least 30°C in saline or albumin with a CA titer of ≥256.23, 32, 42, 43 Only 13 articles had a definition of hemolysis, which included anemia and laboratory markers of hemolysis.2, 6, 18, 21, 23-26, 28, 29, 31, 35, 42 Testing strategies supported by specific literature in the included articles are listed in Table 4. Based on individual studies, most studies used elevated CA titers to determine clinical significance, followed by a C3d + DAT, presence of TA, and then DAT strength. Clear cut-offs for those tests were not identified. In 13 studies (excluding review articles), antibody specificity was determined and in 6 studies a clinical relationship between the presence of anti-I was either correlated with increased severity of disease or present in clinically significant disease. Only 2 case series used bacterial and viral testing as adjunct strategies for determining CAD/CAS. Other ancillary tests in included studies that suggested CA included the presence of agglutinating RBCs on blood smears, positivity on the monocyte monolayer assay, a positive complement binding assay, positive enzyme-linked DATs, M-protein with or without LC glycosylation, low serum complement levels, and cold in vitro hemoysis assays. No defined cutoff:53 ≤64 not indicative of CAD:14 We found the literature was not consistent in defining the characteristics of CA in CAD/CAS, in accordance with findings from a recently published systematic review which also highlighted the heterogeneity in the definition of CAD used in literature.10 We also did not find evidence-based guidelines for how to conduct the investigation and interpretation of CA supporting the clinical diagnosis CAD/CAS. The most used laboratory techniques, the DAT, the determination of the titer, TA, antibody specificity, or combination of these, were used in varying degrees to support the diagnosis of CAD/CAS. Systematic efforts to determine the serological testing that is most appropriate for clinical diagnosis and trending treatment efficacy are needed. The most widely used laboratory test in the diagnosis of CAD/CAS causing AIHA is the DAT. If complement deposition is detected, it is usually interpreted as the presence of CA IgM antibodies, which can bind to RBC surface antigens to cause RBC agglutination as well as complement activation. Because of low avidity of the CA at temperatures above 4°C, the DAT in CAD/CAS is often positive for complement deposition only.44 Wilen et al. reported that a positive DAT with complement deposition can be used as an for further CA testing, as a negative DAT using has a negative of for CA titers that may be associated with clinically significant The DAT strength the concentration and avidity of CA, but only two case series reported on the association of agglutination strength with presence and severity of 47 and 13 case series did not find a clinical Other to the DAT (including and are not well to CAD/CAS broadly The CA titer is one of the other widely used tests to determine CA which may the concentration and avidity of the CA. A CA titer of ≥64 at is often defined as being indicative of clinically significant We did not find clear evidence to this threshold and more than half of the articles in literature review did not determine a CA titer (Tables 1 and However, Juskewitch et al. did a significant correlation between their and positive CA titer definitions and the of patients with the clinical diagnosis of though with no described correlation to disease Bendix et al. reported that CA titers in blood and patients without had a reference range of not more than This that low titers of CA can be in the of that may not be indicative for the presence of CAD/CAS. Despite the of CA titer testing, review that the clinical of increased titers is with low CA titers at may have CA that is clinically significant and patients with elevated titers may not have clinically significant disease. et al. studied CA titers in patients with a positive DAT to complement In the patients with hemolytic titers from 8 to at 4°C, saline were the agglutination was performed with of all CA titers in these patients were ≥64 at 4°C, which the most commonly used of clinically significant CA. However, in the patients without hemolytic CA titers at also from to and 20 to with and with albumin respectively. The different for determining CA titers testing results but are often not described in literature. In a case series of patients defined as mixed AIHA (DAT positive for both IgG and a CA with a TA and warm IgG antibody in the all CA TA with CA titers 64 at However, these cases did not have anemia to cold and patients clinically the of most articles describe in the CA titer between individual patients with CAD/CAS, but an evaluation of a correlation between the titer and the hemolytic degree in vivo is et al. a titer range of in patients with CAD/CAS, in with a titer of in donors, without a reported correlation titers and the degree of The of determination of the TA of the CA in literature review was not the of studies did not a correlation between TA and clinical significance. In most of the case series, a thermal range of the CA is described only In the case series as above, et al. a correlation of the presence or of hemolytic anemia with the of CA titers determined at and These authors that CAD/CAS may be diagnosed if there is a CA titer ≥64 at in combination with CA up to 30°C if the agglutination was performed with of A case series by et al. in 26 patients with may also the of the determination of the TA to the CA In patients, the TA of the CA up to and in patients up to with CA titers This review a where laboratory techniques and patient to results. in et al. the findings of et al. where all CA in patients with cAIHA a TA of 30°C or higher if agglutination was using saline without In these authors published a case series results. In half of patients with CA with a TA of 30°C or higher using an with of CA was not to activate complement to hemolysis in and in less than of these patients, was indicative of in vivo hemolysis. et al. in vitro that even TA of a CA ranges from to complement binding was up to testing at This may suggest that an IgM CA may bind to RBCs and complement activation at higher temperatures than suggested by the TA, which is determined by an agglutination hemolysis even with a the IgM in CAD/CAS have specificity for the surface antigens on the and in of the the is of hemolytic anemia to CA are also described in CA specificity for the is more the antibody specificity severity of the disease, In only case series testing on CA specificity, the CA specificity was described as either being or used to the diagnosis in patients with In the case series by et with patients, the most common specificity was Antibody specificity was to be correlated with and hemolysis in et al. found that the specificity of CA was to the severity of the disease, with anti-I CA causing more hemolytic increased and positive also that with an anti-I specificity were more commonly found in patients with primary and with and other were more common in secondary The literature and with incidentally found CA during testing is to determine the of One cohort study by et al. in cardiac surgery a low diagnostic of CA the degree of for individual tests correlated with clinical significance, testing be done a of testing may be to a putative diagnosis of CAD/CAS, other of anemia or hemolysis, to a in and a search of the literature, may not have all relevant case series and case-control studies in gray literature, but systematic review not to include a priori given literature such as and often not contain detail on methodology results to inform A was not given the retrospective of studies they would be of lower methodological We included only criteria. We also that review only current of the literature, though the practice of CA for clinical significance has not changed significantly in the based on literature not the of in and CAD/CAS sample Data abstraction of of these studies significant which was by a number of and reviewers with and in CA testing, including common testing such as titer and TA, is highly heterogeneous in current to the literature which is to The association between CA test such as titer and TA, and clinical was often not or not The lack of consensus on the optimal diagnostic evaluation in patients with CA is for both clinically significant CA and also treatment without a defined by The current literature also information on the of for incidentally detected CA which is a common in there is important to there has been no systematic to determine the type of serological testing that is most appropriate for clinical management. the heterogeneity of case definitions of CA and clinical significance often only being defined by the methodology for testing CA and definitions for CA and hemolysis may be for study and the understanding of CAD/CAS. The findings of systematic review indicate a to and CA testing strategies to define the serological for clinically significant CA. test results without clinical may to of both patients and as well as of studies are to more information on CA testing in correlation with clinical with a of laboratory techniques We would to at of and at the for their in the of the authors have of to in relation to this CA screening positive in 47 cases (0.3%), where 20 had antibody specificity found No significant difference in outcomes modified CPB to reduce cold in 20 cases but not clear if clinically significant patients with thermal amplitude at patients at