I have similar issues from Sydney to us-east-1 and us-west-2. For example: I have a Rails app hosted in us-west-2 that responds in about 450ms but the Lambda function via the API GW is taking 750ms despite the Lambda function itself running in less than 50ms. us-east-1 was even worse at 950ms and you needed to double those times from a user perspective if it triggers a CORS pre-flight check.
In the end I’ve decided to no worry about it. I can find ways to keep the response times within acceptable limits and being able to easily scale wins.