Since insulin is a growth factor, due at least in part due to its agonist effect on the IGF-1 receptor, it's plausible that any insulin might promote tumor growth. However, the affinity of Lantus for the IGF-1 receptor is uniquely high, higher than that of human insulin and much higher than Levemir.
This plausible mechanism, along with an apparent dose-response relationship for the effect, suggests that there really may be something to the association.
In patients with low cancer risk the absolute risk associated with Lantus is probably very low, and the association needs additional study.
Pushing insulin to very high doses in obese insulin resistant type 2 diabetics likely compounds the problem. Such an approach is irrational and needs serious re-examination as I recently pointed out here.
Although insulin analogs are appealing and relatively easy to use hard data that they offer advantages over older insulins in terms of clinical outcomes are lacking.
Current evidence does not give clinicians clear guidance about what to do but for now, in patients needing an insulin analog for basal coverage I'm tempted to go with Levemir.