This isn't an IE issue at all but a browser issue in general. Just because somebody's exploit example doesn't call Opera is no indication that Opera or any other HTTP client couldn't be used in the place of IE. In fact a resourceful trojan writer might just read the system registry to determine the default browser and use it. Take a look at the Help menu of the various applications on your system. Often times you'll find a "Visit Website" menu entry. If Opera is the default browser does clicking the entry open Opera? It does indeed if the application is programmed to open the default browser and Opera is the default. If it can be done it response to a click it can also be done behind your back.
You will never be 100% secure but here are a few things you can do to minimize your risk.
Configure only your default browser to use Prox but configure your firewall rules so that the only way your other browsers can connect is through Prox. This way your less oft used browsers will be incapable of connecting out until you reconfigure them to use Prox to connect when you need them.
Configure your firewall rules such that access to the proxy on localhost is explicit rather than implicit. An example of implicit access is when you create a firewall rule to allow any application to connect to localhost via TCP/UDP by any port. This will allow any application to connect out through Prox. To prevent this you can create two localhost loopback rules.
The first localhost loopback rule should allow inbound/outbound TCP/UDP to/from localhost with a remote port range from 1 to 8079 (localhost:1-8079) for any application.
The second localhost loopback rule allows inbound/outbound TCP/UDP to/from localhost with a remote port range from 8081 to 65535 (localhost:8081-65535) for any application. You cannot just block localhost loopback because some of your apps and perhaps Windows itself will yield errors, hang or even crash.
Once you've done that create two rules for every browser on your system.
The first rule will allow your browser to connect out via TCP (and UDP in the case of IE) to any address. Restrict it to the local port range from 1024 to 5000 and remote port 8080.
The second is for SSL and will allow your browser to connect out via TCP(/UDP) to any address with a remote port of 443 (HTTPS). Again the local port range is 1024 to 5000. If you feel comfortable with it you can configure Prox to filter secure pages but it may result in these pages being cached to disk (I don't know how Prox handles temp files). I recommend disabling the SSL rules until you need them.
You could also create a third rule to allow FTP but I personally think it's safer and more practical to use a dedicated FTP client.
The firewall rule for Prox will allow it to connect out via TCP. If your firewall allows for it, configure a discrete set of remote ports, e.g. 80, 443, 3128, 7734, 8080 and so on as you require. The local port range is always the same, 1024-5000.
Now to minimize the risk of malware utilizing your browser to connect, don't configure Prox to start with Windows and don't even run Prox if you're not browsing the web. With the above firewall rules, if Prox is not running your browsers cannot connect out, period. I even shut Prox down while typing this.
Finally if you are practicing "Safe Hex" and performing regular security audits of your system then malware shouldn't be a problem for you but that's another topic.
