Identification of ALDOA and FH as antigens recognized by serum of patients with TIA
We observed two independent clones in the serum of 19 TIA patients by expression cloning (Fig. 1). Specifically, we found a sequence homology with aldolase, fructose-bisphosphate A (ALDOA) (Accession number: NM_184041) and fumarate hydratase (FH) (Accession number: NM_000143), respectively. The region between amino acids 70 and 469 of ALDOA was obtained as a pBluescript II clone and then recombined into a pGEX-4T-2 expression vector. Similarly, the cloned region between amino acids 1 and 185 of FH was recombined into pGEX-4T-2 vectors. Recombinant ALDOA and FH proteins were expressed in E. coli as GST fusion proteins. They were subsequently purified by affinity chromatography using glutathione-Sepharose.
Presence of serum antibodies was confirmed by Western blotting
We aimed at confirming the presence of anti-ALDOA (ALDOA-Abs) and anti-FH (FH-Abs) antibodies in serum. To this end, Western blotting was performed using serum obtained from TIA and CI patients. Using an anti-GST antibody, GST- ALDOA, GST- FH, and GST proteins were recognized as reactions of 65-kDa, 67-kDa, and 28-kDa proteins, respectively (Fig. 2). On the contrary, GST-ALDOA and/or GST-FH, but not GST, reacted with serum antibodies of patients #350 and #692. Based on these findings, we suggest that most, if not all, of GST fusion antigen proteins’ reactivity with serum antibodies may be due to antigen proteins rather than the GST domain. In the present study, specific reactions against ALDOA or FH proteins were estimated by antibody levels toward GST-tagged antigen proteins subtracted the levels toward GST.
Levels of ALDOA-Abs and FH-Abs are increased in patients with TIA and CI
In order to quantitatively analyze ALDOA-Ab and FH-Ab levels in serum, AlphaLISA was used. To this end, we examined the serum of HDs and patients with TIA, aCI or oCI. The Alpha counts represent the luminescent photon counts corresponding to the antibody levels. Our results showed that the levels of ALDOA-Abs and FH-Abs in the three ischemic cerebrovascular diseases (i.e., TIA, aCI, and oCI) were significantly higher compared with HDs (P < 0.05) (Table 2). Furthermore, we observed no significant differences in the Alpha counts among patients with the three ischemic cerebrovascular diseases (Fig. 3). Therefore, ALDOA-Ab and FH-Ab levels may be closely related to ischemic cerebrovascular diseases, but not the disease type.
We performed the ROC analysis to evaluate the ability of these markers to detect TIA, aCI, and oCI. The areas under the curve (AUC) of ALDOA-Abs and FH-Abs for TIA were 0.63 [95% confidence interval (CI): 0.56 – 0.69] (Fig. 4a) and 0.63 (95% CI: 0.56 – 0.70) (Fig. 4d), respectively. While the AUC for aCI were 0.63 (95% CI: 0.60 – 0.67) (Fig. 4b) and 0.63 (95% CI: 0.59 – 0.67) (Fig. 4e), respectively. Additionally, the AUC for oCI were 0.62 (95% CI: 0.54 – 0.70) (Fig. 4c) and 0.67 (95% CI: 0.60 – 0.75) (Fig. 4f). At a cutoff value of ALDOA-Ab levels of 14,869, the antibody level’s sensitivity and specificity for TIA diagnosis were 69.57% and 54.74%, respectively (Fig. 4a). Such levels were similar to those for aCI diagnosis (69.40% and 51.58%, respectively) (Fig. 4b). Furthermore, sensitivity and specificity for FH-Abs are shown in Fig. 4d, 4e, and 4f.
Association between TIA and clinical parameters including ALDOA-Ab and FH-Ab levels
Results of univariate and multivariate logistic regression analyses are shown in Table 3. Using the cutoff value of 14,869 and 2,849, respectively, univariate logistic regression analysis revealed that the elevated ALDOA-Ab (OR: 2.91, 95% CI: 1.76 –4.83, P < 0.0001) and FH-Ab (OR: 2.88, 95% CI: 1.78 – 4.67, P < 0.0001) levels were associated with the increased risk of TIA, respectively. We included the univariate P < 0.05 into the multivariate analysis factors. The multivariate logistic regression analysis revealed that elevated ALDOA-Ab (OR: 2.46, 95% CI: 1.31 –4.62, P = 0.0050) and FH-Ab (OR: 2.49, 95% CI: 1.35 – 4.63, P = 0.0037) levels were independent predictors of TIA, respectively. The predictive values of ALDOA-Abs and FH-Abs for TIA were similar to that of HT which is a typical risk factor of TIA. Specifically, these include the following: age (OR: 6.04, 95% CI: 3.15 – 11.58, P < 0.0001); HT (OR: 2.97, 95% CI: 1.61 – 5.45, P = 0.0005); and DM (OR: 5.31, 95% CI: 2.05 – 13.79, P = 0.0006).
Elevated positive prediction values by combination of ALDOA-Abs, FH-Abs, and clinical risk factors
Next, we calculated the positive rates of 92 patients with TIA and 285 HDs including the conventional risk factors, age, HT, and DM. We used the cutoff values of ALDOA-Abs and FH-Abs to detect TIA, 14,869 and 2,849, respectively, as mentioned above. Positive predictive values (PPVs) of age, HT, and DM alone were 48.0%, 51.3%, and 71.1%, respectively (Table 4). On the contrary, PPVs of ALDOA-Abs combined with age, HT, and DM increased to 63.1%, 63.5%, and 91.3%, respectively. Similarly, PPVs of FH-Abs combined with age, HT, and DM were 61.9%, 56.9%, and 94.1%, respectively. Furthermore, PPVs with the combination of HT, DM, and ALDOA-Abs or age, DM, and FH-Abs reached up to 100%.
Japan Public Health Center (JPHC) cohort analysis
The above logistic regression analysis proved that ALDOA-Abs and FH-Abs are independent early warning risk factors for TIA, which is one of prodromal stages of CI. To further validate their association with CI, we conducted a prospective case-control study nested within the JPHC-based prospective study (the interference of age, sex and area were excluded). ALDOA-Ab and FH-Ab levels were divided into four quartiles. We found that the ORs (95% CIs) were 2.38 (1.24 – 4.55) and 2.50 (1.26 – 4.96) for individuals with the second and highest quartiles of antibody level, respectively, compared with the lowest quartile (Table 5). Additionally, we observed that FH-Ab levels were also positively associated with the risk of aCI. Specifically, we observed that the ORs (95% CIs) were 2.17 (1.20 – 3.92) and 2.60 (1.41 – 4.80), for individuals with the third and highest quartiles of antibody level, respectively. These results indicate that ALDOA-Abs and FH-Abs are useful markers in predicting the onset of aCI.
Association between ALDOA-Abs and FH-Abs and clinical parameters
We then examined the correlation between serum ALDOA-Ab and FH-Ab levels and sex, other diseases, lifestyle, and obesity (Table 6). The ALDOA-Ab levels were well associated with blood pressure (P = 0.0022), CHD (P = 0.0400) and habitual smoking (P < 0.0001), but not with sex, DM, HL, alcohol intake, or obesity. FH antibody levels showed similar correlation as that of ALDOA except significant correlation with DM (P = 0.0183) and less correlation with smoking habit (P = 0.0566).
Next, correlation between antibody levels and other clinical parameters was examined with Spearman correlation analysis. The levels of both AODOA-Abs and FH-Abs were significantly correlated with age, blood pressure, maximum intima-media thickness (max IMT), and blood sugar (Table 7). Because max IMT is one of indices of atherosclerosis, both ALDOA-Abs and FH-Abs could reflect the degree of atherosclerosis. Consistently with the results of Table 6, DM-related blood sugar was more closely related to FH-Ab levels than ALDOA-Abs. On the other hand, the antibody levels were inversely correlated with albumin/globulin ratio, cholinesterase, total protein, albumin, total cholesterol, and red blood cell number.