Without additional information about the distribution, there is not much that we can assert **for certain**. However, if we are willing to make reasonable guesses, there is often a rough empirical relationship between mode, median, and mean, for fairly well-behaved unimodal distributions. The relationship is $$\text{mean}-\text{mode}\approx 3(\text{mean}-\text{median}).$$ We can use this approximate relationship, combined with a crossing of the fingers, to estimate the mean, though an answer to $2$ decimal places would certainly be inappropriate!. There are even proofs that under suitable restrictions the relationship roughly holds.
For a detailed discussion at a not difficult level, please look here.