Hi Khai,
Unfortunately, there is no a constant ‘time to wait’ parameter in order to read the proper value from the buffer due to some factors such as I2C clock speed or SPI clock rate selection can be different from one application to another.
In this case, I would recommend that you make sure you are giving the proper time to the bridge to translate the data from I2C to SPI.
To be clear, please review the files attached to this response.
At the image01 (below), you can see a proper communication frame. After the I2C stop signal, the proper time is given to the bridge in order to translate the information (yellow line). So, the communication with the MCU + Bridge + SPI slave is correct.

On the other hand, the image02 (below) shows an incorrect communication due to the time is not enough for the bridge to translate the information (red line). In consequence, the data acquired will be incorrect, you can see the NAK signals later.

So, please make sure you are giving enough time to the bridge.
Regards,
Jose