We consider limited feedback for zero-forcing-based coordinated beamforming over time-varying channels. Most previous work on limited feedback focuses on the feedback rate of the channel state information (CSI), while assuming that feedback occurs for every channel use. This feedback update policy, however, may not be optimal for time-varying channels. To take channel variations into account, we propose a scheme that adjusts both the feedback rate and the feedback update period, depending on channel statistics. We derive an upper bound of the expected rate loss caused by the limited feedback. Based on this bound, we formulate an optimization problem minimizing the residual interference power and derive closed-form solutions for the feedback rate and the feedback update period while fixing one of the parameters and optimizing the other. Using the closed-form expressions, we propose an algorithm that iteratively updates the parameters. Simulation results demonstrate that the proposed limited feedback scheme can outperform the conventional scheme, which feeds back the CSI at every channel use.