I'm not sure that I completely understand your question, but I will attempt an answer. If I had three variables - A, B and C - that were associated with one another AND if I had clear evidence as to the time sequence for or experimental intervention between each variable (A precedes B, B precedes C), then there are a variety of ways that I might explore the associations with A as IV, B as IV and DV, and C as DV.
If you are dealing with cross-sectional data, then the differentiation between independent and dependent variable is always somewhat precarious.