In this short editorial by Michael Thase, the argument is made that a diagnosis of bipolar II disorder is not necessarily a contraindication to antidepressant use for treating a depressive episode.
With these diagnostic categories, it is important to realize that there are relative risks of different management strategies, such as of antidepressants worsening rapid cycling. But not all individuals necessarily will experience such an adverse effect. We do not as yet have clear evidence ahead of time which can allow us to predict which patients with a particular diagnosis will experience benefit or adverse effect from a particular treatment.
Part of the reason for this is that single diagnostic categories such as "bipolar II" or "rapid cycling" may represent a wide variety of ailments, which current diagnostic schemes cannot resolve, and may also represent numerous subsets of individuals, who may benefit or have adverse effects from different treatment strategies.
The best we can do, I think, is to use a type of Bayesian reasoning, in which current broad evidence should be our starting point to estimate risk or benefit. In the case of rapid cycling, I think we must assume that there is a significant risk of adverse effects in rapid cycling bipolar patients.
But in an effort to treat a debilitating depressive state, there may be instances in which a riskier treatment could be warranted, as there is evidence that particular individuals can benefit.