In frequency-division duplexing (FDD) massive multi-input multi-output (MIMO) systems, the codebook-based beamforming has been widely considered to overcome the overhead of channel state information (CSI) feedback while there is still inevitable complexity. In order to further reduce that or help with the codebook scanning, it is valuable to exploit partial reciprocity to reduce the complexity. In this letter, we propose an angle-adaptive subspace codebook for FDD systems with massive MIMO. By leveraging the partial reciprocity of FDD systems, it is possible to estimate the angle information of the downlink channel based on the angle information of the uplink channel. This estimation enables the construction of a subspace codebook to efficiently help with the beamforming process. The size of the proposed codebook is proportional to the number of primary arrival paths, which is much smaller than the number of antennas in massive MIMO systems. Simulation results show that the proposed angle-adaptive subspace codebook can significantly reduce the complexity of codebook scanning while maintaining satisfactory performance compared to the simple codebook.