Talk:Ranibizumab

Edits of Oct 5 2014

@Jmh649:, nice job on the update. One question I had is that you summarized the AE info by saying that the side effect profiles for rani and bev were similar. My read was that the older Cochrane analysis found a fairly high RR of ocular AEs of 2.8 and a relatively modest RR of GI and infection SAEs of 1.3, which was just barely SS. The new Cochrane analysis found no significant difference in RR of death or serious systemic AEs, but they could not exclude a RR of serious systemic AEs of up to 1.3 and they do not appear to have looked at ocular AEs, which is where the larger RR was seen in the first analysis.

I may have misread this, but could you give it another look? The ocular result found in the first study seems to be unchallenged by the second unless I missed something. Thanks Formerly 98 (talk) 04:49, 6 October 2014 (UTC)

No excellent points. In these trial the GI effects were a secondary analysis. Have added. We need to add the eye related AEs. Which Cochrane review was that? Doc James (talk · contribs · email) (if I write on your page reply on mine) 14:45, 6 October 2014 (UTC)

It was reference 3 that earlier found the higher ocular AE rates with bev, and I thought you were updating with the newer review that includes more trials but which mysteriously did not seem to look at ocular AEs. Not sure the older one was a Cochrane review, but one of the authors is from the German Cochrane center.

Frankly it escapes me why these should have different AE rates, there is only a moderate difference in the serum half life that presents itself as a possible explanation for enhanced systemic AEs, and I have no explanation for why bev should give higher occular AEs as long as they don't contaminate the vial contents.

Another minor issue here (I know this is WP:OR, but I think we're allowed to do this on the Talk page while discussing how to use references) is that I don't think Cochrane's use of calculated absolute risk differences using the rates from 1-2 year trials is entirely appropriate here, as few will use these drugs for only 1-2 years. If there is a 0.3% increased risk of death and 1.8% increased risk of serious systemic AEs over a 1.5 year period, the average patient will probably experience increased absolute risks of ~1.8% and ~11% over 10 years of treatment. To me it seems like these calculations of absolute risk differences should take into account how long the drugs will be used in clinical practice, and not just the duration of the available trials.

Its hard for me to get a handle on what people consider a clinically meaningful increase in the risk of death. But back of the envelop, isn't 0.2% excess death per year on the order of what was seen with Vioxx (http://www.ncbi.nlm.nih.gov/pubmed/19637402) or did I miscalculate?

But perhaps I'm nitpicking. Whatever you decide to do is fine by me. Formerly 98 (talk) 16:04, 6 October 2014 (UTC)

With velecoxib the differences in death were statistical. Here they are not.

Good point about the eye issues. Have re added the PLOS paper. Doc James (talk · contribs · email) (if I write on your page reply on mine) 16:53, 6 October 2014 (UTC)

James, I don't want to get on your nerves here, but I know a little bit about clinical trial design, and I think the Cochrane statistical analysis is flawed in this case.

Their null hypothesis which they are trying to disprove is that bev causes more deaths and systemic side effects than rani. In order to prove that they have to set a non-inferiority limit (say, they are setting their criterion as being that bev is not more than 10% worse than rani), and show that the upper limit of the 95% C.I. for the risk of death and serious systemic side effects is less than 1.1 x the rate seen with rani.

What they have done here instead is to show that with the amount of data available to them, they cannot show a statistical difference between the two drugs. This is not the same as proving non-inferiority, as there is an alternative explanation for not getting a p<0.05, that explanation being that their study did not have adequate statistical power. Put another way, failure to prove that something is true is not the same as having proved that it is untrue.

Respectfully, I don't think we should use that reference to say the the two drugs have equal rates of systemic effects. I know I'm way out on a WP:OR ledge here, but I think the fact that they are calling the two equivalent when they cannot rule out a Vioxx-sized problem at the 95% confidence level illustrates what I am concerned about. Some stuff on non-inferiority statistics here; http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3019319/#!po=8.33333.

I can't quote any policies supporting my position, but it seems to me that if the authors of a meta analysis make an addition error, we should not follow their lead. From my POV, this looks like a similar case.

The language that I would like to see here would be something along the lines of the second study "did not find evidence supporting a clinically meaningful difference in systemic AEs or the risk of death with bev, but because only limited data was available, could not rigorously rule out such an increase".

Formerly 98 (talk) 20:09, 6 October 2014 (UTC)

Have added details similar to what you suggest but in a little simpler language I hope. Let me know what you think. Doc James (talk · contribs · email) (if I write on your page reply on mine) 21:06, 6 October 2014 (UTC)

I think that looks great. Sorry I don't always explain myself well, but you reduced several rambling paragraphs down to a sentence or two very nicely. I appreciate your openness to my concerns. Formerly 98 (talk) 22:55, 6 October 2014 (UTC)

I appreciate your comments. Sometimes I over simplify. And we really are not that much in disagreement. Doc James (talk · contribs · email) (if I write on your page reply on mine) 00:00, 7 October 2014 (UTC)

History of the article

It would be useful if a small history subsection could be added, in particular explaining when it was researched, released, filed for admission etc..

This would be useful to visitors to quickly get this information, after reading that Ranibizumab is based on another older monoclonal antibody. 2A02:8388:1641:8380:E1C:C3B5:304A:1703 (talk) 20:30, 30 January 2020 (UTC)