PIM inhibitors synergize with disruptors of ribosome biogenesis to kill AML cell lines in vitro and potentiate growth rate inhibition in vivo. (A) AML cell lines were treated with AZD1208 and BMH-21 (left) or AZD1208 and S6 kinase inhibitor LY2584702 or each drug alone at 5 concentrations (see supplemental Methods) in the ratios shown. Fa, fraction affected. (B) Tumor growth rates were calculated for EOL-1 tumors in NSG mice treated with vehicle (Control), 5 mg/kg AZD1208, 5 mg/kg BMH21, or 5 mg/kg AZD1208 + 5 mg/kg BMH21 . Mean tumor growth rates (mm³ per day) were as follows: control (vehicle treatment) = 0.1212; AZD1208 = 0.0967; BMH-21 = 0.0983; AZD1208 + BMH-21 = 0.0822. Asterisk (∗) indicates a statistically significant difference between the average of the mean tumor growth rates of AZD1208- and BMH-21-treated mice vs AZD1208 + BMH-21 (t test, P = .0366). For statistical analysis of tumor growth rates, we assumed that tumor volume follows an exponential growth model as described⁴⁹:

$log10V=a+b×days+ϵ(Eq1)$

where $V$ is the tumor volume, a and b are parameters of interest, and $ϵ$ is an error term, which has a normal distribution with equal variance for each observation. For 4 different groups, including the control group, we estimated a and b from Eq 1, then computed the mean of growth rates. We denote $μTA$⁠, $μTB$⁠, $μTAB$⁠, and $μC$ to be the mean growth rate for each of 3 treatment groups (A = AZD1208, B = BMH-21, AB = AZD1208 + BMH-21) and one for control group (C = DMSO), respectively. For instance, $μTA=(1/nA)∑i=1nbiˆI(i∈A)$⁠, where $nA$ is the number of mice in group A, $biˆ$ is the estimate of b for i-th mouse, and $I$ is an indicator function. $nC=9$⁠, $nA=12$⁠, $nB=9$⁠, and $nAB=9$⁠, which amounts to $n=nC+nA+nB+nAB=39$ samples in total. We interpret $biˆ$ as a growth rate of i-th observation, in other words, one expects that it may vary according to the group which i-th mouse belongs to. We then carried out a 2-sample t test to identify whether a significant difference exists among each of treatment groups and control group. We tested

$H0:μC=μj$

$HA:μC>μj,$

where j = A, B, AB. We concluded that the growth rate of the control group was significantly higher than the growth rate of any of the treatment groups on average with 95% confidence level (P value <.05). The mean growth rates were as follows: $μTA=0.0967$⁠, $μTB=0.0982$⁠, $μTAB=0.0822$⁠, and $μC=0.1212$⁠. $μTA$ and $μTB$ did not appear to be statistically significantly different from each other. We provide the mean of predicted tumor volumes $Vˆ$ according to time, $Vˆ=10aˆ+bˆ×days$ in Figure 5 which shows the similarity in predicted tumor volumes for groups A and B. The t test confirmed that there is no growth rate difference between A and B: the P value for testing $H0:μA=μB,HA:μA≠μB$ was >.05. To determine if the therapeutic effect of AZD1208 + BMH-21 was statistically significantly greater than AZD1208 or BMH-21 alone, we compared the tumor growth rate in groups AB, A, and B. We tested this by conducting a t test to determine whether the growth rate of AB is statistically significantly lower than those of groups A and B. Because it was shown that there is no difference between groups A and B, we performed the following t test

$H0:μA∪B=μAB$

$HA:μA∪B>μAB,$

where $μA∪B$ is the mean growth rate of both of A and B groups, that is, $μA∪B={1/(nA+nB)}∑i=1nbiˆI(i∈A∪B)$⁠. This test yielded a P value of <.05, hence, we concluded that AZD1208 + BMH-21 reduced the tumor growth rate significantly, compared with the effectiveness of AZD1208 or BMH-21 alone. (C) Kaplan-Meier graph for survival. AZD1208 + BMH-21 extends survival significantly compared with AZD1208 alone (P = .0032) or BMH-21 alone (P = .0039). Log-rank test (Mantel-Cox) P values are tabulated.