Autoimmune Hemolytic Anemia With Gel-Based Immunohematology Tests

  Am J Clin Pathol   2013;139:457-463 457457   DOI: 10.1309/AJCPTU9AEQZXVZD4   457 © American Society for Clinical Pathology Coagulation and Transfusion Medicine / D󰁩󰁡󰁧󰁮󰁯󰁳󰁩󰁳 󰁯󰁦 AIHA W󰁩󰁴󰁨 G󰁥󰁬-B󰁡󰁳󰁥󰁤 DAT  Autoimmune Hemolytic Anemia With Gel-Based Immunohematology Tests Marco Lai, MD, 1  Giuseppe Leone, 2  and Raffaele Landolfi 3 Key Words: AIHA; DAT; Gel technology DOI: 10.1309/AJCPTU9AEQZXVZD4 A b s t r a c t  We used gel centrifugation tests (GCTs) to analyze the relationship between the diagnosis and immunohematology tests used for autoimmune hemolytic anemia (AIHA). The study included 588  samples positive for the direct antiglobulin test (DAT). Of these, 52 were from patients diagnosed with  AIHA. Immunoglobulin (Ig) class, IgG1, IgG3, and complement were measured. DAT strength had the  strongest correlation with AIHA diagnosis (odds ratio [OR], 23), followed by anti-IgG titer 300 (OR, 8.4), anti-IgG titer 1,000 (OR, 10.5), and C3d agglutination  strength (OR, 1.7   ).  Decision tree analysis revealed that DAT strength and anti-IgG titer higher than 100 were the best predictors of AIHA. Multidimensional  scanning analysis found a high grade of similarity among DAT strength, anti-IgG titer, and IgG strength in the AIHA samples. This observation was not detected in DAT-positive samples from patients without AIHA.  DAT strength remained the best diagnostic indicator  for AIHA and had the strongest association with  AIHA compared with other commercially available immunohematology tests. The other tests, despite good correlation with AIHA diagnosis, did not add useful information. In the most common type of warm autoimmune hemolytic anemia (AIHA), a positive result on the direct antiglobulin test (DAT) is the cornerstone of diagnosis. 1-3  However, among hospital patients the DAT result is often positive in the absence of AIHA; it is present in approximately 7% to 8% 4  of hospital-ized patients and thus has a poor predictive value for AIHA. Previous studies investigated the use of additional immu-nohematologic tests to improve the diagnostic accuracy for AIHA; they analyzed immunoglobulin (Ig) and complement characteristics to establish an association with immunologic RBC destruction. 4-15  The IgG subclass test in DAT-positive samples provides further information about the serologic basis of AIHA. However, these reports show conflicting results 16-22  on the relationship between IgG subclasses and autoimmune hemolysis. This may occur because factors other than Ig and complement characteristics, such as activity of the reticuloen-dothelial system, affect the immune hemolysis. 23 The objective of the present study was to investigate whether commercially available immunohematology tests that use gel technology can be used to add useful information for AIHA diagnosis. We also analyzed our data to establish whether combining the tests improved our ability to diagnose AIHA and to better understand the serologic characteristics of AIHA in individual patients. Materials and Methods Patients The study included 588 blood samples from 588 patients; all of these samples yielded a positive DAT result. Blood spec-imens were collected in ethylenediaminetetraacetate tubes and  458   Am J Clin Pathol   2013;139:457-463 458   DOI: 10.1309/AJCPTU9AEQZXVZD4  © American Society for Clinical Pathology Lai et al / D󰁩󰁡󰁧󰁮󰁯󰁳󰁩󰁳 󰁯󰁦 AIHA W󰁩󰁴󰁨 G󰁥󰁬-B󰁡󰁳󰁥󰁤 DAT centrifuged at ×1,500  g   for 10 minutes at room temperature. All samples were processed the day they arrived, and none were stored below room temperature. All immunohematology tests were performed with commercially available gel cen-trifugation test (GCT) cards. A diagnosis of AIHA was made  based on the following: hemoglobin (Hb), less than 12 g/dL (120 g/L); corrected reticulocyte count, more than 2%; lactate dehydrogenase, more than 220 U/L (3.67 μkat/L); and total  bilirubin concentration, more than 1.5 mg/dL (25.7 μmol/L), with the majority of bilirubin unconjugated. Samples from AIHA patients were taken at diagnosis. Immunohematology All of the DAT cards and ID Diluent 2 solution were  procured from the manufacturer (Diamed, Cresier Sur Morat, Switzerland), and all tests were performed in accordance with the manufacturer’s instructions. Gel-based DAT was  performed with low ionic strength solution (LISS)/Coombs cards, which contain polyspecific antihuman globulin (AHG), and ID Diluent 2 (modified LISS for RBC suspension). The DAT with monospecific AHG reagents (IgG, IgA, IgM, C3c, C3d, and Ctrl) was performed with DC-screening I cards and ID Diluent 2 (Diamed). IgG subclass analysis (IgG1 and IgG3) was performed with the DAT IgG1/IgG3 ID card and ID Diluent 2. Each ID card contained 2 dilutions, 1/100 for IgG1 and 1/100 for IgG3. The DAT with AHG titration was  performed with the specific GCT cards DAT IgG-Dilution, which contained 5 different dilutions (10, 30, 100, 300, and 1,000) of anti-IgG, and ID Diluent 2. Statistical Analysis The relationship between the immunohematologic test results and a diagnosis of AIHA was evaluated by means of logistic regression analysis with the backward likelihood ratio method. Decision tree analysis was performed with the χ  2  automatic interaction detector (CHAID) method. Relation-ships between variables were detected with cluster analysis and multidimensional scaling (MDS) analysis. The stress value is used for judging the goodness of fit of an MDS solution. Stress and squared correlation (RSQ) in distances illustrates the percentage of the variance in the model that is explained by the 2 dimensions. Results A total of 588 DAT-positive samples from 588 patients were included. Fifty-two samples were from patients who had  been diagnosed as having AIHA. ❚ Table 1 ❚  lists the serologic DAT results of the patients with and without AIHA. ❚ Table 2 ❚  shows the cross-tabulation between C3d reaction strength, IgG subclasses (IgG1 and IgG3), and GCT anti-IgG titer in the AIHA cases. The DAT revealed the following frequencies of positive combinations in patients with AIHA: IgG alone (n = 23), 44.2 %; IgG/C3d (n = 16), 30.8%; IgG/IgM/C3d (n = 5), 9.6%; IgG/IgA (n = 7), 13.5%; and IgG/IgM/IgA/C3d (n = 1), 1.9%. When AIHA patient samples positive for IgG but not IgM or IgA were considered (n   = 39), the IgG1 frequency was 87.1%. IgG1 was the only subclass detected in 59% of these patients. IgG3 was detected in 35.8% of the 39 patients and was the only subclass detected in 7.7%. Two of the 39 samples positive for IgG but not IgM or IgA were negative for IgG1 and IgG3 (5.1%). Among all 52 AIHA- positive samples, IgG1 was detected in 46 (88.5%), IgG3 in 22 (42.3%), and IgG3 alone in 4 (7.7%); the IgG subclasses were negative in 2 samples (3.8%).In the DAT-positive samples from patients without AIHA (n = 536), IgG1 was detected in 25 cases (4.7%), IgG1/IgG3 in 5 cases (0.9%), and IgG3 in 1 case (0.2%). IgG was detected in 425 cases (79.3%), IgG/C3d was found in 50 cases (9.3%), IgG/IgA/C3d in 1 case (0.2%), IgG/IgM/C3d in 4 cases (0.7%), IgM/C3d in 1 case (0.2%), and C3d alone in 39 cases (7.3%). In 16 cases (3%), no DAT specific-ity was detected. Binary logistic regression analysis revealed that the reaction strength of the DAT odds ratio (OR) was 23 (95% confidence interval [95% CI], 5.3-98.8;  P   = .000), the 1:300 DAT dilution OR was 8.4 (95% CI, 2.2-31.2;  P   = .001), the 1:1,000 DAT dilution OR was 10.5 (95% CI, 2.3-48.4;  P   = .002), and the C3d reaction strength OR was 1.7 (95% CI, 1.1-2.6;  P   = .013); all these reaction variables were signifi-cantly associated with AIHA diagnosis. ❚ Table 3 ❚  reports the results of the logistic regression analysis. The CHAID tree method allowed us to further classify the immunohematol-ogy tests on the basis of their relationship with the AIHA diagnosis ❚ Figure 1 ❚ . The variables analyzed by CHAID ❚ Table 1 ❚ Serologic Characteristics of Patients With and Without Autoimmune Hemolytic Anemia (AIHA)  AIHA (n = 52) Without AIHA (n = 536)Ig Class No. (%) Ig Class No. (%) IgG 23 (44.2) IgG 425 (79.3)IgG/C3d 16 (30.8) IgG/C3d 50 (9.3)IgG/IgA 7 (13.5) IgG/IgA/C3d 1 (0.2)IgG/IgM/C3d 5 (9.6) IgG/IgM/C3d 4 (0.7)IgG/IgM/IgA/C3d 1 (1.9) IgM/C3d 1 (0.2) C3d alone 39 (7.3) None 16 (3)IgG subclass IgG subclass IgG1 28 (53.8) IgG1 25 (4.7) IgG1/IgG3 18 (34.6) IgG1/IgG3 5 (0.9) IgG3 alone 4 (7.7) IgG3 alone 1 (0.2) None 2 (3.8) None 505 (94.2) Ig, immunoglobulin.  Am J Clin Pathol   2013;139:457-463 459459   DOI: 10.1309/AJCPTU9AEQZXVZD4   459 © American Society for Clinical Pathology Coagulation and Transfusion Medicine / O󰁲󰁩󰁧󰁩󰁮󰁡󰁬 A󰁲󰁴󰁩󰁣󰁬󰁥 were GCT anti-IgG titer, IgG strength, IgM strength, C3d strength, GCT anti-IgG1 strength, GCT anti-IgG1 titer at 100 strength, and DAT strength. The first and most impor-tant splitting variable for AIHA diagnosis was DAT strength (  P   = .000, c 2  = 357.3), which divided the samples into 3 nodes. The first node contained cases with DAT strength less than or equal to 2+ (n = 459, 78.1% overall); none of these patients were diagnosed as having AIHA. The sec-ond node contained cases with DAT strength 3+ (n = 54, 9.2% overall), which included 2 cases of AIHA. The   third node contained cases with DAT strength 4+ (n = 75, 12.8% overall); in this node, there were 50 cases of AIHA (66.7% of the node). The   third node (DAT 4+) was split into nodes 4 and 5 on the basis of the AHG titration test result (  P   = .004, c 2  = 11.3). The fourth node contained 28 cases with AHG titration less than or equal to 100, of whom 12 were diagnosed as having AIHA (42.9% of the node). The fifth node contained 47 cases with AHG titration more than 100, of whom 38 were diagnosed as having AIHA (80.9% of the node) (Figure 1).We performed cluster analysis with Euclidean distances to find similarities and dissimilarities among the agglutination strengths. In the samples from patients with AIHA, smaller distances were found between the DAT strength, IgG strength, and anti-IgG titer ❚ Table 4 ❚ . In the non-AIHA cases, the analysis revealed larger distances among the variables com- pared with the AIHA cases (Table 4). We performed MDS analysis on the Euclidean distance matrix of both the AIHA and non-AIHA cases because this analysis can more precisely detect similarities between the different immunohematologic tests. In AIHA cases ❚ Figure 2 ❚ , MDS analysis showed that some of the variables were very close together in the com-mon space (stress = 0.02, RSQ = 0.99). These variables were DAT strength, IgG strength, and anti-IgG titer; we called this collection of variables “the DAT family.” MDS analysis per-formed for the DAT-positive samples from patients without an AIHA diagnosis (stress = 0.05, RSQ = 0.99) did not detect the similarities found in the AIHA cases ❚ Figure 3 ❚ . ❚ Table 2 ❚ Cross-Tabulation Between C3d Reaction Strength, IgG Subclasses (IgG1 and IgG3), and GCT anti-IgG Titer in the AIHA Cases (n = 52) a   GCT Anti-IgG Titer (count)C3d Strength Subclass 0 10 30 100 300 1,000 Total C3d– IgG1 0 0 0 4 11 9 24 IgG1 and IgG3 0 0 0 0 2 1 3 IgG3 0 0 1 1 1 0 3 Total 0 0 1   5  14 10 30C3d 1+ IgG1 and IgG3 0 0 0 0 0 1 1 Total 0 0 0 0 0 1 1C3d 2+ IgG1 0 0 0 0 0 1 1 IgG1 and IgG3 0 0 0 0 1 0 1 Total 0 0 0 0 1 1 2C3d 3+ No subclass 1 1 0 0 0 0 2 IgG1 0 0 1 2 1 0 4 IgG1 and IgG3 0 0 0 0 3 4 7 IgG3 0 0 0 1 0 0 1 Total   1 1 1   3  4 4 14C3d 4+ IgG1 0 0 1 0 0 0 1 IgG1 and IgG3 0 0 0 0 3 1 4 Total   0   0 1  0 3 1 5 AIHA, autoimmune hemolytic anemia; GCT, gel centrifugation test; Ig, immunoglobulin. a  Numbers in bold indicate the serology of the cases with anti-IgG titer less than 300. ❚ Table 3 ❚ Logistic Regression Analysis Results for Variables not Associated and Significantly Associated With AIHA Diagnosis Logistic Regression OR (95% CI)  P   Value Variables not in the equation GCT anti-IgG titer 10 .61 GCT anti-IgG titer 30 .95 GCT anti-IgG titer 100 .26 IgG strength .14 IgM strength .94 IgG1 (Pos. or Neg.) .77 IgG3 (Pos. or Neg.) .43 IgG1 strength .89 IgG1 strength at titer 100 .21 IgG3 strength .31 IgG3 strength at titer 100 .81Variables in the equation DAT strength 23 (5.3-98.8) .000 GCT anti-IgG titer 1:300 8.4 (2.2-31.2) .001 GCT anti-IgG titer 1:1,000 10.5 (2.3-48.4) .002 C3d strength 1.7 (1.1-2.6) .01 AIHA, autoimmune hemolytic anemia; CI, confidence interval; DAT, direct antiglobulin test; GCT, gel centrifugation test; Ig, immunoglobulin; OR, odds ratio; Pos, positive; Neg, negative.  460   Am J Clin Pathol   2013;139:457-463 460   DOI: 10.1309/AJCPTU9AEQZXVZD4  © American Society for Clinical Pathology Lai et al / D󰁩󰁡󰁧󰁮󰁯󰁳󰁩󰁳 󰁯󰁦 AIHA W󰁩󰁴󰁨 G󰁥󰁬-B󰁡󰁳󰁥󰁤 DAT Node 0DAT strengthadjusted  P   value = .000, χ 2 = 357.3, df   = 2≤2+≤100>1002+, 3+>3+GCT anti-IgG titeradjusted  P   value = .004, χ 2 = 11.3, df   = 1Category  AIHA No AIHA Total%8.891.2100n52536588–Node 1Category  AIHA No AIHA Total%0.010078.1n0459459Node 2Category  AIHA No AIHA Total%3.796.39.2n25254Node 4Category  AIHA No AIHA Total%42.957.14.8n121628Node 5Category  AIHA No AIHA Total%80.919.18n38947Node 3Category  AIHA No AIHA Total%66.733.312.8n502575– ❚ Figure 1 ❚   c 2  Automatic interaction detector analysis, showing each node with the percentage of autoimmune hemolytic anemia (AIHA) and non-AIHA cases for the node and the percentage of cases included in the node compared with the total number of cases analyzed. DAT, direct antiglobulin test; GCT, gel centrifugation test; Ig, immunoglobulin. ❚ Table 4 ❚ Cross-Tabulation of Distance Matrix From Cluster Analysis of AIHA and Non-AIHA Cases Expressed in Euclidean Distances  Euclidean Distance DAT Strength Anti-IgG Titer IgG Strength IgG1 Strength C3d Strength IgG3 Strength Proximity matrix AIHA   DAT strength 7.810 3.742 13847 22.293 22.962 Anti-IgG titer 7.810 6.557 12.155 24.125 23.457 IgG strength 3.742 6.557 12.359 22.113 22.164 IgG1 strength 13.847 12.155 12.359 18.351 21.024 C3d strength 22.293 24.125 22.113 18.351 13.351 IgG3 strength 22.962 23.457 22.164 21.024 13.351 Proximity matrix non-AIHA DAT strength 30.116 14.248 37.756 37.510 41.388 Anti-IgG titer 30.116 26.038 16.568 26.230 22.045 IgG strength 14.248 26.038 33.904 37.855 37.603 IgG1 strength 37.756 16.568 33.904 22.814 15.476 C3d strength 37.510 26.230 37.855 22.814 17.664 IgG3 strength 41.388 22.045 37.603 15.476 17.664 AIHA, autoimmune hemolytic anemia; DAT, direct antiglobulin test; Ig, immunoglobulin.  Am J Clin Pathol   2013;139:457-463 461461   DOI: 10.1309/AJCPTU9AEQZXVZD4   461 © American Society for Clinical Pathology Coagulation and Transfusion Medicine / O󰁲󰁩󰁧󰁩󰁮󰁡󰁬 A󰁲󰁴󰁩󰁣󰁬󰁥 Discussion Previous studies have investigated the relationship  between DAT positivity and a diagnosis of AIHA. These studies investigated protein characteristics of the RBC sur-face, the amount of Ig (quantitative factor), 24  Ig class and subclass (qualitative factor), and complement activation. However, the importance of these parameters in identifying immune hemolysis is still under debate. In the current study, logistic regression analysis indicated that DAT strength was the variable most strongly related to AIHA diagnosis. This is in accordance with previous studies, 1,4,25,26  in which the authors found that most cases of AIHA had a strong DAT reaction. The other variables significantly related to a diag-nosis of AIHA were DAT with GCT anti-IgG titers of 300 or 1,000. The magnitude of the relationship between anti-IgG titer and AIHA diagnosis was less than for DAT strength and AIHA diagnosis. However, we think that the anti-IgG titer remains important for monitoring the IgG content on RBCs during AIHA treatment. 24  In a previous report on gel technology, of 14 patients with immune hemolysis, 12 had AIHA and all had an anti-IgG titer of 300 or more. In our study, 39 of the 52 cases with AIHA had an anti-IgG titer of 300 or more. However, we found that 5 of the samples from  patients with AIHA had an anti-IgG titer of 30 or less: 1 with IgG3 alone and 4 with IgG and a strong C3d reaction (3+ or 4+) (Table 2). Logistic regression analysis revealed that C3d strength was significantly related to the diagnosis of AIHA  but with a lower OR magnitude. In previous reports, 27-29  the authors found a relationship between immune hemolysis and the amount of C3d. CHAID analysis confirmed that DAT strength was the parameter most strongly related to AIHA diagnosis. In cases with a DAT reaction strength of 4+, approximately two-thirds of the patients were diagnosed with AIHA. Within the group of cases with a DAT strength of 4+, a GCT anti-IgG titer of 300 or more was most strongly related to a diagnosis of AIHA; 81% of these patients had AIHA (node 5, Figure 1). It should be emphasized that a greater likelihood of AIHA in cases with DAT 4+ and GCT anti-IgG titer of 300 or more results in fewer AIHA cases identified compared with DAT of 4+ alone (node 3, Figure 1). This means that the GCT anti-IgG titer should be considered with caution when this information is given to clinicians because, in our study, among the AIHA cases, 12 cases had a DAT reaction strength of 4+ and GCT anti-IgG titer of less than 300. Based on the results thus far, it seemed to us that quan-titative factors played the most important role in the cases of AIHA. Therefore, in a further step, we used MDS analysis to investigate the relationship between the agglutination strength of Ig and complement. MDS analysis of the cases with AIHA (Figure 2) revealed that the behavior of DAT strength, IgG strength, and the anti-IgG titer had a high level of similarity (DAT family); they were very close in the common space (bidimensional). In the same analysis, C3d strength, IgG3 strength, and IgG1 strength were far apart. Furthermore, they were far from the “DAT family” in the common bidimensional space. This indicates that the  behavior of IgG1 strength, IgG3 strength, and C3d strength were independent from the other variables included in the MDS analysis (ie, the DAT family). For IgG1 strength, this dissociation can be explained by the way in which the reagents were prepared; the starting dilution for IgG1 is for –1.0–0.50.0–2.0–1.00.0Dimension 1    D   i  m  e  n  s   i  o  n   2 1.02.0 Anti-IgGtiterIgG strengthIgG3 strengthIgG1 strengthC3d strengthDAT strength0.51.0 –1.0–0.50.0–2.0–1.00.0Dimension 1    D   i  m  e  n  s   i  o  n   2 1.02.0 Anti-IgGtiterIgG strengthIgG3 strengthIgG1 strengthC3d strengthDAT strength0.51.0 ❚ Figure 2 ❚  Multidimensional scaling analysis for autoimmune hemolytic anemia cases in which a Euclidean distance model is used to plot dimension 1 vs dimension 2. DAT, direct antiglobulin test; Ig, immunoglobulin. ❚ Figure 3 ❚  Multidimensional scaling analysis for non–autoimmune hemolytic anemia cases, in which a Euclidean distance model is used to plot dimension 1 vs dimension 2. DAT, direct antiglobulin test; Ig, immunoglobulin.