differing $request_time and $upstream_reponse_time in nginx

I'm running nginx with some backend hosts.

I've created custom log format with TR:$request_time US:$upstream_response_time` to watch delivery of the requests.

I've got log line (serving CSS file from backing) with lines TR:2.410 US:0.012 , which probably means "the request has been serverd by backend host in 0.012 second, but nginx itself took 2.410 seconds to serve the request".

Question: Does it mean that the receiving host received response more than 100times slower than the originating machine generated it? How this can happen? (other than the slow line)

The receiving machine of this log is my laptop located in different city (10ms TTL from server to the laptop). I'm quite surprised because for most request, the $request_time and $upstream_response_time are usually very close to each other.