Regularized Canonical Correlation Analysis

Hi, I have a question about sample size and regularized canonical correlation analysis.
Is a sample size of 15 unreasonable to use in a regularized canonical correlation analysis? If one has lots and lots of potential explanatory variables (multicollinearity expected) and multiple response variables, but only a sample size of 15 for each variable, is there any hope of learning something about the relationships between the response and explanatory variables? If not, what analyses might I want to employ to explore these relationships?