Sirs: Relapsing neuromyelitis optica (RNMO) is a rare demyelinating disease in which the optic nerves and spinal cord are selectively and repeatedly targeted [8, 12, 15]. The immuno-pathogenesis of RNMO, distinct from classical multiple sclerosis (MS), remains unknown although recently a serum autoantibody highly specific to NMO (NMO-IgG) was discovered [4] and its diagnostic and pathogenetic roles are being investigated. Since there is evidence that MS is a helper T cell type 1 (Th1)-dominant disease [9, 14], it is of interest to compare Th1/ Th2 balance between RNMO and MS, but such studies are few. CD26 is a dipeptidyl peptidase-IV highly expressed on Th1 cells [2], and CD30 is a member of the tumor necrosis factor/nerve growth factor receptor superfamily preferentially expressed on Th2 cells [1]. Upregulation of CD26 during relapse and CD30 during remission have been reported in MS [3, 7, 13]. We quantified soluble CD26 (sCD26) and sCD30 in the cerebrospinal fluid (CSF) and sera of 13 patients (all women) with RNMO and 17 patients (14 women and 3 men) with MS by sandwich ELISA. The diagnosis was made based on the proposed criteria [6, 15]. The CSF and sera were obtained during relapse of acute myelitis and before methylprednisolone therapy was started. Fourteen patients with non-inflammatory neurological diseases (10 women and 4 men) were also analyzed as controls. RNMO and control patients were older than MS patients. The present study conformed to the guidelines of the institutional ethics committee, and patients gave spoken consent prior to the study. We used a sCD26 ELISA kit (Chemicon International, Temecula, CA) and a sCD30 ELISA kit (Bender Medsystem, Vienna) according to the manufacturers’ instructions. The minimum detectable levels of sCD26 and sCD30 were 11 ng/ml and 0.5 U/ml, respectively. Samples were analyzed undiluted in duplicate. The data in the three patient groups were compared with the KruskalWallis test and post hoc Scheffe test. Correlations were tested by Spearman rank correlation test. P < 0.05 was considered statistically significant. Data in RNMO, MS, and control patients and the statistics are shown in Table 1. The CSF-sCD26 levels were significantly elevated in RNMO and MS compared with control, but those in RNMO and MS were not different (Fig. 1A). Meanwhile, the serum-sCD26, CSFsCD30 and serum-sCD30 were not different between the 3 groups (Fig. 1B–D). There was no difference in CSFor serum-sCD26/ sCD30 ratios between the groups, either (Fig. 2A, B). The CSF-sCD26 levels were significantly correlated with the CSF-total protein levels in RNMO and MS. Our study suggests that a significant intrathecal Th1 response associated with CD26 expression develops during relapse in RNMO and MS but that overall Th1/Th2 balances in these diseases are not very different. Alternatively, it is also possible that differences in blood-CSF integrity have influenced the results as sCD26 and sCD30 levels were higher in serum than in CSF, and sCD26 and sCD30 indices were higher in RNMO and MS than in control but the differences did not reach statistical significance. We recently reported that CXCL10/IP-10, a Th1 chemokine, was significantly elevated in CSF of both RNMO and MS as compared with controls [11]. As for Th2 chemokines, CCL17/TARC was slightly higher in CSF in both demyelinating diseases, but the levels were much lower than the serum levels [11]. CCL2/MCP-1 capable of inducing Th2 polarization was significantly lower in CSF in MS but not in RNMO as compared with the level in controls [11]. These chemokine data are consistent with the present study that revealed a significant Th1 response in both RNMO and MS, though their Th2 responses may differ to some extent. We reported that RNMO lacked an IgG1 response seen in MS and other Th1-dominant diseases [10], and another group has suggested a pathogenic role of humoral immunity, mainly controlled by Th2, in their neuropathological study of NMO [5], suggesting less Th1-dominance in RNMO than in MS. NMO-IgG [4] may also imply the importance of humoral immunity or Th2 response in NMO as compared with MS. The present analysis did not detect such a difference, but further comparative studies on Th1 and Th2 components are needed to clarify unique pathomechanisms in RNMO. LETTER TO THE EDITORS