I can think of a reason expect this not to work very well. The impedance of twin-lead transmission line is dependent on the ratio between
- the diameter of the conductors, and
- the distance between their centers.
In twin-lead or any parallel-conductor transmission line, the insulation is designed to keep that distance stable. On the other hand, in speaker wire, the insulation is usually quite soft, and a small amount (<1 mm) of incidental squishing of the cable will quite significantly change the distance between the conductors, and thus the impedance.
But, that's a theoretical answer, on why we don't in general use cheap speaker wire or lamp-cord for transmission lines. What about in practice?
Well, I've got a piece of speaker wire handy (30 ft or so, 16 AWG). I unplugged it from my speaker and plugged it into a resistance decade box and my antenna analyzer, and made some measurements. Impedance, 0-30 MHz:

So, an adequate piece of transmission line around 110 Ω.
Does it have obvious impedance discontinuities measured in my analyzer's "TDR" mode?

Not very much! (The spike on the left end is the connection to the analyzer, and the wiggle on the right is the connection to the termination.)
I'd guess that the practical reasons not to use speaker wire are:
- It's not a standard impedance, so you'd need matching on both ends
- Any particular product might have a different impedance
- It's not coax, so you have to treat it like any twin-lead and keep it away from other things, but people generally prefer coax unless they're building a high-performance HF station, at which point why use the cheap stuff?