In this paper, we consider a symmetric pure jump Markov process X on a general metric measure space that satisfies volume doubling conditions. We study estimates of the transition density p(t, x, y) of X and their stabilities when the jumping kernel for X has general mixed polynomial growths. Unlike [22], in our setting, the rate function which gives growth of jumps of X may not be comparable to the scale function which provides the borderline for p(t, x, y) to have either near-diagonal estimates or off-diagonal estimates. Under the assumption that the lower scaling index of scale function is strictly bigger than 1, we establish stabilities of heat kernel estimates. If underlying metric measure space admits a conservative diffusion process which has a transition density satisfying a general sub-Gaussian bounds, we obtain heat kernel estimates which generalize [3, Theorems 1.2 and 1.4]. In this case, scale function is explicitly given by the rate function and the function F related to walk dimension of underlying space. As an application, we prove that the finite moment condition in terms of F on such symmetric Markov process is equivalent to a generalized version of Khintchine-type law of iterated logarithm at infinity.