Specifically, TLR4 was significantly associated with both DSS and OS in AJCC stages 2 and 4. Across all stages, we found that for two of the TLR4 probes (Short and Long2) a higher expression correlated with improved OS (exp(coef)short = 4.04, p = 0.019; exp(coef)long2 = 3.69, p = 0.06). By contrast, the remaining RGFP966 probes (Medium and Long1) showed decreased expression with improved survival (exp(coef)medium = 0.26, p = 0.019; exp(coef)long1 = 0.22, p = 0.034). Figure 3 CRC Survival and Relationship to TLR4 Expression. A) DSS, OS, DFS, and RFS are shown with their associated exponential

regression coefficients (exp (coef)) and significance levels. Note that the direction of the coefficients varied depending on probe. B and C) Cox Proportional

Hazard Curves (GSE14333) for DFS based on level of TLR4 expression. Three curves are generated for each probe based on quartiles of TLR4 expression. These graphs demonstrate that probability of DFS is significantly associated with TLR4 expression, and the direction of the association is probe-dependent. B) For TLR4 long probe 2,DFS is lowest in the group with the highest level of expression (75th ARN-509 nmr percentile). Cut-off values for TLR4 expression were as follows: 5.0 (25th percentile), 6.8 (50th percentile), and 8.0 (75th percentile). C) For the TLR4 Short Probe, higher levels of expression result in improved DFS. The same cut-off values were used as in Figure 3B. The association between survival and TLR4 expression was corroborated DNA Synthesis inhibitor by a strong correlation between TLR4 expression (Short and Long2) and DFS among 290 colon cancer patients ranging from Duke’s stages A through D (exp (coef) 0.78, p = 0.0008 and exp (coef) 1.47, p = 0.0006) (GSE14333) [17]. TLR4 expression

levels were divided into quartiles by probe. Survival curves were constructed per probe, meant to represent low, average, and high expression (Figure 3B, 3C). For Long2, higher expression of TLR4 was associated with lower probability of DFS (Figure 3B). The inverse relationship was demonstrated for Short (Figure 3C). This association between DFS and TLR4 expression was not supported by other GSE series examining the endpoints of OS, DFS (GSE12945) [26], relapse-free survival (HDAC inhibitor GSE8671) [18] and recurrence-free survival (GSE33113) [27]. In a separate series of 48 sporadic colon cancer samples, no association between TLR4 expression and survival was observed (exp (coef) = 1.13, p = 0.61) (GSE16125) [28]. When differentiating colon from rectal cancers, the tumor location was not significant in any models of survival, p > 0.80.