I reference the recent New York Times article that cites the essential problem of holding trials for fatal conditions
1) The NY Times article was substantially biased when talking about the benign side effect - not even mentioning that 40% of treated patients get squamous cell skin cancers (easily removed when spotted - but nonetheless far from benign). And almost certainly getting other cancers as well, albeit not as easily spotted as squamous cell cancer.
2) You have mischaracterized the NY Times article - slanting it even further. Even the original (slanted) article did not go so far as to push for avoiding all trials in fatal conditions. In fact it only suggested that for treatments as dramatic as the PLX treatment, when used in short-term-fatal conditions, there should be a quicker path.
PS The p value for the CLDX trial is probably around 0.15. Nice for Ph ii - but nowhere near convincing. If, not-entirely-hypothetically, for every 10 worthwhile drugs entering their first randomized trial there are 100 that are no better than sugar water then for every 10 worthwhile drugs coming out of that trial with p<0.15 there will be 7.5 totally worthless drugs with p<0.15. There are things you could do to improve that - e.g. historical comparisons - but those are IMO nowhere near as robust as a randomized trial.
PPS Note that in the case of the PLX drug I agree that it would be desirable to somehow get it approved. But only in the terminally terminal ill. For people with a median life expectancy of, say, 15 months or longer, I believe a randomized trial is going to be required. Do other cancers pop up at high rate and cause their own mortality? Does it not improve survival - because the cancer 'catches up' after 6 or 7 months? ...