To be sure, this is a real point but it is not adverse selection. Adverse selection requires asymmetric information, namely that I know more about my brain tumor than does my potential insurance company. The more likely problem is that the tumor is common knowledge, or would be if I applied for insurance, and the company won't sell a policy for any price cheaper than the costs of treatment. There is no asymmetry of information, rather insurance simply is no longer possible. In the limiting case, imagine that a predictor-demon could forecast your lifetime medical expenditures with certainty, and then blog them by your social security number. Such a person, no matter how healthy, couldn't buy insurance either.
Scream all you want, but that is not inefficient per se (don't complain in the comments about the limits of the efficiency concept, and the cruelness of economists, I'm already on that one, scroll down to #7 under "microeconomics", alternatively you might make a complicated Rawlsian argument.) Covering these people, by the use of government policy, is a transfer, not an efficiency improvement, with an added caveat for imperfect capital markets.
Defenders of the adverse selection argument in reality believe the following: if someone is going to face death, or a very bad medical outcome, and can't buy their way out of it, government should put up the money, at least within limits.
Maybe yes, maybe no, but now we are comparing competing investments and which will bring the greatest utilitarian good and the greatest moral good. I'm far from convinced health care access wins that race or even comes in second.
Why does this matter? Because if it's adverse selection, that leads very quickly to a policy argument for a mandate on coverage and, in this market, a single payer system. Not so fast, say the bloggers at MR, and I think they are right.