It's worth remembering that the GISS dataset has 2005 as the warmest year.
The ENSO years will be warmer as they up the global atmospheric temperature by releasing oceanic-stored heat - heat absorbed from the atmosphere. They will add an additional anomally to non-ENSO years, but that doesn't invalidate their use in the temperature series - there are plenty of processes that affect the year on year temperature (and the use of calendar years is somewhat arbitrary - ENSO events tend to span the boundary). If global warming wasn't pushing the "base" temperature up, then it would be likely that a previous El Nino year would likely be the record.
However, for those who don't see this point, it wouldn't be too difficult to plot a series ignoring El Nino years and one just of El Nino years. In the former 1005 would be the hottest, in the latter, 1998 would be the hottest.