# I don't understand this Zytiga chart - can anybody help me?

And then I looked at this chart VIEW IMAGE where the median survival is 14.8 months. If you eliminate the pasient numbers below the chart it makes sense. It seems like about 30% of the participants are still alive after 20.2 months. That corresponds to what I've read in other articles as well.

But what are these patient numbers below the chart? It sais that a total of 797 patients were on Zytiga and 398 on Placebo. But if you draw a vertical line from the graph at 14.8 months, under 100 people are alive. And at 20.2 months only 5 people are alive. Thes numbers doesn't match the graph at all?

At first I thought they were effectively changing the vertical scale by basing it on the number of patients surviving at that instant. But I ruled that out because the chart would have been more granular at the end. Then I thought maybe it was a derivative - a rate of change - but that did not work either.

I just don't get it. Something is being interpreted differently from what we see.

This will be an interesting statistics problem for someone here.

Jeff

I just don't get it, because I just read about someone who took part in the Zytiga clinical trial in 2011, and the next month the trial was cancelled due to early FDA approval. He can't be a part of this phase III chart, can he? In that case he will destroy all the statistics. Maybe he was intended to be in a part IV trial or something?

It is plausible that the chart shows the number of people under treatment for x months; However, there is another problem. If we interpret the bottom of the chart as the number of people on treatment for any particular number of months, the median would have to be somewhat greater than 9 months, roughly between 9-10 months as I eyeball the chart. This is not consistent with the 8 month figure given.

**proscapt said...**

Smurfine - not everyone starts a trial at the same time; there is a period where the trial is open for enrollment, and then when the trial fills up, enrollment stops. However, when a trial is stopped early either because the drug is too effective to withhold from the placebo arm -- or for other reasons -- everyone stops at once. This can cause the duration of treatment to vary widely. In addition, even if two people start at the same time, their duration of treatment can vary; for example if they fail to respond and move on to another treatment.

It is plausible that the chart shows the number of people under treatment for x months; However, there is another problem. If we interpret the bottom of the chart as the number of people on treatment for any particular number of months, the median would have to be somewhat greater than 9 months, roughly between 9-10 months as I eyeball the chart. This is not consistent with the 8 month figure given.

As mentioned in the article, this chart represents the results before anyone one placebo were allowed to switch to Zytiga. ("Though patients in the placebo arm were allowed to cross over into the abiraterone arm of the study after positive interim results, this study is a final analysis of data before crossover was allowed.")

I see what you mean with the chart, and median treatment time doesn't match with 8 months. As you say it should be a bit higher. But the numbers below the chart are definitely not representing the survivors left, so it's more likely to believe they have something to do with median treatment time, even if it doesn't match excactly.

**tarhoosier said...**

With Kaplan-Meyer charts it is confusing to correlate the graphic part of the chart with the (in this case) survivor numbers. Kaplan-Meyer is a way to graphically represent all the participants AS IF they started simultaneously, even if they did not. Duration of treatment median and time to death are not correlated in this specific chart. Median treatment time is added information in the chart and not represented by the chart itself.

It is not the numbers of survivors (since several hundred are still alive when this trial finished), and as we discussed earlier it might not be the number of people still on Zytiga after x months either. So what can these number be?

It is my assumption that patients who were no longer eligible for the trial due to progression on treatment were still followed for mortality from all causes. Those who had such mortality by the dates of the trial were included in this analysis, and chart.

** I am not a doctor. My opinions are not medical advice. **

DX age 54 9/2012 PSA 5.3, DRE-, high PSAV. Clinical T1c

Biopsy: Gleason 3+4 with PNI / PC in 6 of 14 cores / 10% of length

Tx: RRP 2/2010

Pathology: pT2cNxMx / G 3+4 / PNI+ / SM- / SV- / EPE- / Volume 3cc /

PSA in 2010: <0.01 | 0.01 | 0.01 PSA in 2011: .01 | .02 | .02 | .01

PSA in 2012: .039 | .039 | .043

http://cancerguide.org/scurve_km.html

**2cents said...**

The numbers below indicate the number patients who can be evaluated at that time point. As you get out to the right on the x-axis, some patients will have not been in the study long enough to contribute data to that time point. Patients can also be censored. The link below has a good explanation. Hope this helps.

http://cancerguide.org/scurve_km.html

But basically what you mean here is that after f.ex. 18 months only 2 persons have been on Zytiga for the whole period on 18 months? The rest of the people started later?

I thought the point with this chart was to make it look that all people started at the same time?

But you do agree that after 20 months (when the trial was stopped) there are still several hundred people alive? But does it mean that some of the people that are still alive after 20 months started on Zytiga f.ex. 2 months earlier? If so, the chart doesn't mean anything in my opinion , because they've survived for only 2 months, but it looks like they've survived for 20! According to the chart it looks like they've lived for at least 20 months.

*Post Edited (smurfine78) : 2/27/2013 1:15:29 AM (GMT-7)*

Some of the confusion is due to the chart that you link has no context. It must be an analysis of a trial but there is no other information. The graph is separated from the much more detailed report which contains extensive information directly related to the graph. So where is this article?

The article to which you link in your original post is a news report and refers to some other analysis because the numbers mentioned are different than the numbers in the graph link. So please provide the article from which the graph was taken and it may (will) complete the information.

I agree with 2cents that the numbers under the chart likely refer to men still under observation in the trial. The chart itself refers to survival, not just those under observation. The information on men still under observation cannot be directly compared to the data in the chart because they measure different things.

Again, provide the context for the graph and all may be made clear.

Click on the results on "metastatic crpc following docetaxel"

The numbers in the other article I referred to seems to be the same as in the chart? At least the numbers of participants are the same

*Post Edited (smurfine78) : 2/27/2013 9:49:01 AM (GMT-7)*

about the graph (sorry - here comes more confusion):

For each time point they look at all the people on the abi or placebo, and who is alive or dead. Then the statistics guys calculate a 'rate of death' for the abi group and the placebo group based, on how the alive/deceased ratio changes from time point to time point. The shapes of the curves reflects the rates of death.

Then they compare the rate of death (shape of curve) for abi vs placebo. Indeed the numbers get very small and the end as the patients die or haven't been on study long enough. They stopped the study because the death rate (or inversely survival rate), was significantly better for the abi group. It seems like they made this judgement without having long term data, but statistically the trend was already established with what data they had and it was unethical to keep going.

"The final analysis found that after a median follow-up period of 20.2 months, the median OS in patients in the abiraterone arm of the study was 15.8 months (95% CI, 14.8–17.0) compared with 11.2 months in the placebo arm".

These are different numbers than the graph.

OK, I found the difference. There was an interim analysis as the trial was continuing. This decision to analyze is made by an independent group controlling the trial. At that point they found the data which is in your graph and which Zytiga copies at their commercial site. This was the basis of the submission to the FDA, as it appeared statistically significant that Abiraterone made a difference n survival. As the FDA considered this the trial data continued to accumulate. After approval a further, final analysis was conducted. This set of numbers is that which the news article refers in your link at the top.

Even more confusingly, Zytiga themselves use the interim data for the Kaplan-Meier graphat their site, while just above the graph printing the LARGE, bold, green numbers and survival from the final analysis, ensuring that anyone trying to reconcile the large numbers with the graph below will be unsuccessful. Whew.

I think a take-away here is that information from the drug makers may be incomplete, confusing, or both.

*Post Edited (tarhoosier) : 2/27/2013 11:41:55 AM (GMT-7)*

So is it not correct that 30% of the total participants at 1195 were still alive 20 months after starting to take Zytiga? That's how I interpret the chart. (But now I don't seem to understand anything anymore).

Is there any way at all to see how many people (in %) were alive after being on Zytiga (or Zytiga + other treatments) for 20 months?

Anyway, I read at several pages (f.ex this:abiraterone.blogspot.no/) that there are several hundreds 4-year survivors from the phase III trial. Have you also read that? That is indeed the most important thing for me to know.

But how do we know that these hundreds are from the post-chemo trials and not the pre-chemo trials?

*Post Edited (smurfine78) : 2/27/2013 2:38:35 PM (GMT-7)*

Median survival was better with it than without it.

Those who completed the post-docetaxel trial, which this discussion refers to, had more options after treatment.

Men with hormone refractory disease are surviving longer than ever.

New options, including trials, show more potential treatments soon.

That is information that keeps both of us going.

## Forum Information

There are a total of 3,036,072 posts in 331,883 threads.

View Active Topics