Changes in levels of serum sCD26/DPP4 titer pre/post YS110 administration, as documented by Box and Whisker plot
Several crucial parameters were included in this phase I trial such as 1) tumor histology: 19 MM, 6 RCC and 1 UTC; 2) YS110 dose: 0.1-6 mg/kg; 3) frequency of drug administration: once every 2 weeks (Q2W) for three doses in 18 cases, once every week (Q1W) for five doses in 8 cases. In addition, examination of background factors between SD and PD cases indicated that no bias was found in age, BMI, absolute value of tumor volume or serum sCD26/DPP4 titer before YS110 administration, except for gender (data not shown). In contrast to male patients (4 SD and 7 PD in MM, and 2 SD and 3 PD in RCC), YS110 appeared to be more effective in female patients (6 SD and 2 PD in MM, 1 SD in RCC, and 1PD in UTC), as shown in Additional file 1 (Tables S1) and file 2 (Table S2). Since the number of cases in each antibody dose cohort was not sufficient for statistical analysis, in the present study, a total of 26 cases were further categorized by 1) tumor histology and 2) frequency of drug administration, to examine whether serum sCD26 titer variation can be a prognostic biomarker for YS110 treatment. Detailed information and the pertinent demographics of the 26 cases forming the basis for analyses are shown in Additional file 1 (Tables S1).
We first examined the serum sCD26 titer variation during a course of YS110 treatment in each group by Box and Whisker plot analysis. Serum sCD26 titer was consistently reduced immediately following YS110 administration on day 1, 15, 29, and gradually recovered until the next YS110 infusion, although it never returned to its former pre-dosing level (Fig. 1a). This pattern was similarly observed in the 18 cases treated on the Q2W drug administration schedule (Fig. 1b). In contrast, a clear difference was observed in 8 cases treated on the Q1W schedule. As shown in Additional file 1 (Tables S1), relatively high antibody dose (2–6 mg/kg) was administered in the Q1W cases as compared with the Q2W cases (0.1-2 mg/kg). These differences in the antibody dose and administration frequency strongly affected the serum sCD26 titer on day 15pre and day 29pre (Fig. 1d). Recovery of serum sCD26 titer following YS110 administration was not clearly observed with the more frequent drug administration of the Q1W cases. 14 male cases and 4 female cases received Q2W administration, while 2 male cases and 6 female cases received Q1W administration (Additional file 2 (Table S2)). The distribution bias between the male cases with Q2W and Q1W and the female cases with Q2W and Q1W was significant (p = 0.026 by Fisher’s exact test). In addition, the number of cases in Q1W (8 cases) was not sufficient for additional statistical analysis. Therefore, we mainly focused on the Q2W cases and the male cases for additional analyses. The initial fall and subsequent recovery of serum sCD26 titer were similarly observed in both the 19 MM cases and 6 RCC cases (Fig. 1e and h), including when the groups were further stratified into such cohorts as the 14 male cases with Q2W administration, 12 MM cases with Q2W administration, and 9 male MM cases with Q2W administration (Fig. 1c, f and g). As shown in Additional file 3 (Figure S1), the absolute value or titer variation of serum sCD26 titer was strongly correlated with level of serum DPP4 enzyme activity (r = 0.908, p < 0.001 or r = 0.974, p < 0.001, respectively).
Differences in serum sCD26/DPP4 titer variation on day 29pre and tumor volume variation on day 43 for SD and PD cohorts by Scatter plot analysis
We next investigated a potential relationship between pre/post serum sCD26 titer variation on days 1, 15 and 29, and tumor volume variation on day 43 by Scatter plot analysis after the start of YS110 administration, with the total of 25 cases stratified by SD and PD cohorts. The tumor volume variation of the SD group would be expected to naturally be lower than that of the PD group. Serum sCD26 titers were markedly decreased in both SD and PD cohorts immediately post YS110 infusion on days 1, 15 and 29 (Fig. 2a, c and e). On the other hand, a noticeable difference between the SD and PD groups in the serum sCD26 titer variation was observed on day 29pre. The serum sCD26 titer variation on day 29pre of the SD cohort was at a lower level compared with the PD group (Fig. 2d). Moreover, this phenomenon was clearly observed in each stratified group such as the 17 cases with Q2W administration, 14 male cases with Q2W administration, 18 MM cases, 11 MM cases with Q2W administration, 9 male MM cases with Q2W administration, or 6 RCC cases (Fig. 2f-k, respectively). These Scatter plot analyses indicate that the serum sCD26 titer variation of the SD cohort was lower than that of the PD cases when measured prior to YS110 administration, and the difference was particularly evident on day 29pre with the Q2W administration.
Correlation of pre/post serum sCD26/DPP4 titer variation on day 29 with tumor volume variation and/or PFS by PPMC/SRDC analyses
PPMC and SRDC analyses were conducted to examine the correlation between pre/post serum sCD26 titer variation on days 1, 15 and 29, and the tumor volume variation as determined by RECIST criteria at day 43 after YS110 administration or PFS. In the FIH phase I clinical trial, 13 cases were assessed as SD and 13 cases were assessed as PD by RECIST, and among the 13 SD cases, YS110 was particularly effective in 7 cases with PFS being longer than 180 days (Additional file 1 (Tables S1)). In a total of 25 cases, statistically significant correlation between day 29pre serum sCD26 titer variation and tumor volume variation on day 43 was observed (p = 0.006 or p = 0.009 by PPMC/SRDC, respectively (Table 1)). There was also statistically significant correlation between serum sCD26 titer variation and PFS (p = 0.011 by PPMC on day 29post for a total 26 cases (Table 1)). In addition, there was statistically significant correlation between variation in the serum titer of DPP4 enzymatic activity and tumor volume or PFS, similar to the case with serum sCD26 titer (Table 1). Statistically significant correlation was similarly observed in variation between day 29pre serum sCD26/DPP4 titer and tumor volume, and between pre and/or post day 29 serum sCD26/DPP4 titer and PFS in 18 cases with Q2W administration frequency and 14 male cases with Q2W administration frequency (Additional file 4 (Table S3) and file 5 (Table S4)). In 19 MM cases, statistically significant correlation between variation in pre/post day 29 serum DPP4 titer and tumor volume was observed by SRDC analysis, while the correlation between day 29pre serum sCD26 titer and tumor volume almost reached statistical significance (p = 0.065) by PPMC analysis. There was statistically significant correlation between day 29post serum sCD26 titer and PFS by PPMC analysis, while the correlation between day 29post serum DPP4 titer and PFS almost reached statistical significance (p = 0.056 by PPMC or p = 0.069 by SRDC analysis) (Additional file 6 (Table S5)). In 12 MM cases with Q2W administration frequency, no statistically significant correlation between variation of serum sCD26/DPP4 titer and tumor volume was observed. The correlation between day 29post serum sCD26/DPP4 titer and PFS did reach statistical significance (Additional file 7 (Table S6)). In the 9 male MM cases treated with Q2W administration, no significant difference was observed in variation between serum sCD26/DPP4 titer and tumor volume, although there was a trend for a correlation between pre/post day 29 serum sCD26/DPP4 titer and PFS (Additional file 8 (Table S7)). In the 6 RCC cases and 8 cases treated with Q2W and Q1W administration respectively, the number of cases were not enough for PPMC/SRDC statistical analysis. These results indicate that there was a correlation in variation between pre/post day 29 serum sCD26/DPP4 titer (before/after the third YS110 administration) and tumor volume or PFS. Importantly, statistical significance was reached although there was limited number of cases with each stratified cohort, particularly in the 18 cases and 14 male cases treated with Q2W administration.
Day29pre serum sCD26/DPP4 titer of SD cohort was significantly lower than that of PD cohort by Bar graph analysis
Based on scatter plot and PPMC/SRDC examination, Bar graph analysis was conducted of day 1pre, 15pre and 29pre serum sCD26 titer variation in SD and PD cases. Of the total 23 cases (12 SD and 11 PD), serum sCD26 titers of both SD and PD cohorts were reduced from day 1pre to day 29pre samples. Of note, the day 29pre serum sCD26 titer variation of SD cases was significantly lower than that of PD cases (p = 0.016) (Fig. 3a). Similar results were observed for each stratified group such as the 17 cases treated with Q2W administration (p = 0.007), 17 MM cases (p = 0.068), 11 MM cases treated with Q2W administration (p = 0.068), 9 male MM cases treated with Q2W administration (p = 0.020), or 6 RCC cases (p = 0.049) (Fig. 3b and e-h). Statistically significant difference between the SD and PD cohorts with the smallest p-value was observed in the 14 male cases treated with Q2W administration (p = 0.003) (Fig. 3c). In the 8 cases treated with Q1W administration, the serum sCD26 titer variation of SD cases was lower than that of PD cases, trending toward statistical significance (p = 0.053) on day15pre prior to the third YS110 administration, which represented the same timing for sample collection to evaluate day 29pre serum sCD26 titer in the Q2W treatment schedule (Fig. 3d).
Predictive power of serum sCD26/DPP4 titer variation on outcomes of SD or PFS by ROC analysis in the stratified groups
ROC analysis was employed to determine the cut-off titer (the Index) of serum sCD26/DPP4 titer variation on day 29pre/post YS110 administration for the Outcomes of SD and PFS ≥ 90 or ≥ 180 days. The probability was evaluated by Fisher’s exact test (Table 2). A total of 23 cases was examined to determine the Index (46.4% or 18.2%) for the Outcomes SD, PFS ≥ 90 or ≥ 180 days, with statistically significant results (p = 0.003 for SD, and 0.005 or 0.003 for PFS, respectively, and with Area Under the Curve (AUC) 0.795, 0.697 or 0.759, respectively) (Table 2; Column Total). On Column Q2W (17 or 18 cases), Q2W, males (14 cases), MM (17 or 18 cases), MM, Q2W (11 or 12 cases), and MM, Q2W, males (9 cases), the Indexes of each column for the Outcomes were determined to have statistical significance or tendency toward significance. Particularly on Column Q2W, males (14 cases), the Index 37.7% on day29pre YS110 administration for the Outcome SD was statistically significant (p < 0.001 with AUC 1.000). Also, the Index 37.7% for the Outcome PFS ≥ 90 or ≥ 180 days was statistically significant (p < 0.001 or p = 0.027, and with AUC 0.950 or 0.879, respectively). Taken together, our analyses of serum sCD26/DPP4 titer variation during a course of YS110 treatment demonstrate that serum sCD26/DPP4 titer variation, particularly at the time point immediately prior to/following the third YS110 infusion on Day29 in the Q2W administration schedule, was a potential prognostic biomarker for YS110 anti-tumor therapy.
Addition of humanized anti-CD26 mAb reduced sCD26 levels in culture supernatants of CD26-expressing MM cell lines and non-tumor cells
Since sCD26 serum levels were markedly decreased in patients with CD26-expressing tumors following YS110 treatment in the phase I study (Fig. 1), we investigated the in vitro effect of YS110 on sCD26 production from MM cell lines. For this purpose, we selected various human CD26-positive or negative MM cell lines. MSTO parent was an endogenous human CD26-deficit cell line, while MSTO-CD26 stably expressed a full-length human CD26 [6]. Stable shRNA knockdown of CD26 in JMN, an endogenous human CD26-positive cell line, markedly reduced CD26 expression as compared with JMN ctrl-shRNA cells [19]. Cell surface expression of CD26 on MM cell lines was shown in Additional file 9 (Figure S2a). We first measured the amount of sCD26 contained in the culture supernatants from a 3-day culture of CD26-positive or negative cells. sCD26 could be quantified in the culture supernatants of CD26-positive MSTO-CD26, JMN ctrl-shRNA and H226 cells, whereas sCD26 could not be detected in the culture supernatants of CD26-negative MSTO parent and JMN CD26-shRNA cells, regardless of YS110 treatment (Fig. 4a). Treatment with YS110 clearly reduced the amount of sCD26 in the culture supernatants of MSTO-CD26, JMN ctrl-shRNA and H226 cells, as compared with those cells incubated with vehicle or control human IgG (Fig. 4a). We next examined the production of sCD26 from non-tumor (normal) cells. CD26 was clearly expressed on the cell surface of HDMVEC and TIG-1, while CD26 was hardly expressed on HUVEC and MCF10A, and partially expressed on MeT-5A (Additional file 9 (Figure S2b)). sCD26 could be quantified in the culture supernatants of CD26-positive TIG-1 and HDMVEC cells, whereas sCD26 could not be detected in the culture supernatants of CD26-negative or low MCF10A, HUVEC and MeT-5A cells (Fig. 4b). Similar with the results shown in Fig. 4A, YS110 treatment clearly reduced the amount of sCD26 in the culture supernatants of TIG-1 and HDMVEC cells, as compared with those cells incubated with vehicle or control human IgG (Fig. 4b). Treatment with YS110 resulted in decreased production of sCD26 from both MSTO-CD26 and TIG-1 cells in a dose-dependent manner (Fig. 4c). Subsequent time course analysis showed that sCD26 level in the supernatant of a 3-day culture of MSTO-CD26 cells was slightly enhanced compared to 1-day culture of MSTO-CD26 cells, and increased sCD26 level was observed in the supernatant of a 7-day culture of MSTO-CD26 cells (Fig. 4d). Reduction of sCD26 level following YS110 treatment was consistently observed at any culture period. Taken together, these data indicate that sCD26 was produced from both CD26-positive tumor cells and non-tumor cells, and the addition of YS110 reduced sCD26 production from those cells in an antibody dose-dependent manner. It is our hypothesis that these in vitro effects are reflected in the marked reduction of sCD26 level in the serum of patients with CD26-expressing tumors following YS110 administration.