It is very likely you will get differing results. Especially in response times. This is mainly due to latency caused by the Geographical distribution. This will become important if response times is one of the metrics you are using to understand the website's ability to cope/scale under load.
A simple reflection of this is;
go to http://amazon.co.uk/
and compare it to http://amazon.co.jp/
You can see the differences in "hops" it makes to reach the server and the latency caused by opening a CMD window (on windows) and entering the following command
tracert amazon.co.jp
followed by
tracert amazon.co.uk
The important thing to remember is that your load tests are comparisons of like for like. Running one load test from India and another from the USA means you are comparing differing setups which may lead to confusion in interpreting the results. If you run them all from India, conceptually that's fine, but (I'm guessing) the main users would be US based. I would therefore recommend running the tests from US servers if you can.
Services such as Blazemeter offer this functionality (as a service). You can also achieve similar if you run your existing JMeter suite from a remote location. For example, by running it from an AWS instance in that local (US East (Northern Virginia)).