F2 (To Infinity And Beyond...)

[ppscott]

I have a question about the use of the f2 equation.

My understanding is that the equation is written so that it implies that the numbers used should be from time point one to the last time point at standard condition. All time points are from point 1 to n. This would exclude the use of assuming zero at zero as a time point and would also exclude including an infinity time point in the calculation. I have seen many people do one or both when calculating f2. I don't think they are calculating f2 correctly.

Am I correct?

[ewething]

Pete,

Sorry I've been on vacation. Interesting point you have brought to the table. I will need to research further.

[ewething]

Pete,

I'm in the gray area on this one but I'll try. In the "Guidence for Industry" (Immediate Release) listed for examples of time points are 15,30,45, &60mins.

People may be trying to meet the other criteria of only using one measurement beyond 85%.

Hope that made sense! I would have to agree with you on this one.

[ppscott]

I think that some are trying to raise the f2 values by using results that would be very close together by their nature.
Zero and infinity actually tell you very little about the profile of a drug dosage form. Non-coated tablets should not even be assumed to read zero at zero time.

(BY the WAY as a side note...The largest differences that I have seen are right around the time of deaggregation. Often 5 minutes on an immediate release product. Often this time is not sampled. Too many want the later time points. I have seen a "profile" that looks like a straight line. 80%, 85%, 95%, 100%... Too few are trying for those early time points 5%, 10% 20% 30%...)

[johnt]

Pete,

I have to absolutely agree with your last statement. I was taught early in my dissolution career that you what to have a good profile. You want to be able to tell if there are variations in the manufacturing of the product. If a discriminating method has not been developed these variations can go undetected. I have seen many methods that give >80% in 10 minutes. People seem to want dissolution results that match assay results which is personally not the best way. In these instances were there is not a discriminating profile the f2 test will most likely always pass.

[rjoshi]

Firstly, I agree that the method developed should be discrimatory if the intent is to assess the effect of additives, or processing variables etc. of a dosage form. Selection of time points is indeed important. However, I wish to open up another topic of discussion related to f2 test.


Although present in the FDA guidance, f2 test is not a very good test in comparing dissolution profiles. Profile comparison is based on comparison of the means ONLY at each of those time points considered for the study. it DOESNOT take in to account the variances about the mean. I think this is important particularly for a test like dissolution. I beleive there are reaserchers working on this problem and I hope to see a better way of comparing profiles soon. Anybody have any thoughts on that ?

[ppscott]

Some research has been done showing the value of modeling the dissolution curve as a cumulative distribution weibull function. This is normally applied to product failure analysis. However, the accumulative decay feature matches nicely to dissolution properties. The function is %diss=100%(1-e exp[-(t/a) exp [b] ]). Where a is the scale parameter and b is the shape parameter.

This is a great way to compare to profiles looking at both the scale and shape of the curve.

[rjoshi]

Thanks. Would you suggest a reference for this ? I may wish to check this.

[ppscott]

Sure, The best paper that I have seen is:

Methods to Compare Dissolution Profiles and a Rationale for Wide Dissolution Specifications for Metoprolol Tatrate Tablets. James Polli, G. Singh Rekhi, Larry Augsburger, and Vinod Shah; Journal of Pharmaceutical Sciences, Vol 86, No. 6, June 1997.


It is a great articles that evaluates methods to compare profiles. He also looks into Multivariate ANOVA, Multiple univariate ANOVA, Level and shape approach,f1, f2, Index of Rescigno, Ratio of percent dissolved, ratio of area under the curve, ratio of mean dissolution time, Zero-order, First-order, Hixson-Crowell, Higuchi, Quadratic, Gompertz, and Logistic methods.

[rjoshi]

Wonderful! and thank you very much. I will get this article and take a look at it. Perhaps, that will trigger further discussion. Any thoughts on whether the multivariate ANOVA is being used currently by the industry and how FDA views it as opposed their suggested f2 test.

[ppscott]

Multivariate analysis is a very useful tool.
However, The FDA appreciates the simplicity of the f2 equation. It does not require software or knowledge of statistical procedures.
It can be hand calculated by the typical analyst. That's a tough battle to fight.

They do allow other methods, whether model dependant or model independant in the guidances. You have to show that it will accurately discriminate between non-equivalent batches.

[C MONT]

Pete

In european countries, we are using f2 equation for determining diss. profiles similaraiy, in accordance with the European Drug Agency Guidelines:

1)three time points are used.
2)only one time point reaches more than 85 % dissolved.
3)CV for all 3 points must remain below 10 %.

Therefore point 2 restricts the comparation of "stright lines" in the terminal portion of dissolution profiles, and point three takes into account diss results variability. I would appreciate your comments.

Carlos Montuenga
Madrid.Spain

[ppscott]

I think that the f2 equation can be very useful if itis properly applied. Many papers have been written in the past few years criticizing the f2 equation due to situation where it could be used badly. I agree that in stead of trashing a perfectly good idea. Just set some guidance on how to use it properly.

The FDA introduced the equation without much guidance (they suggested 15, 30, 45, and 60, as an example in the original guidance). The FDA suggested some guidance a few years after they endorsed the equation. However, I have found that the tradition of automatically sampling 15, 30, 45, and 60 is hard to change.

[ppscott]

DS posted this. I added it to this thread for Him/Her.

DS
Junior Member
Registered: Jan 2001
Location:
Posts: 1

To use f1 or f2 equation, is it required that we test 12 dosage units or can we use the equation if we had only 6 dosage units.
Thanks in advance.

RE:
If you are regulated by the FDA they recomend 12 units of each (test and Product).

[SPMayock]

Pete,

I know that it has been a while since this was posted, but I wanted to make sure that most of the issues raised by this thread were covered and your question answered.

Neither the zero time point nor the infinity dissolution time points should be included in the f2 calculation for similarity. Additionally, the guidance states that "only one measurement should be considered after 85% dissolution of both the products". This is important since one could add later time points until the f2 value shows similarity when in reality they were not (obviously).

If both dissolution profiles are greater that 80% at 15 minutes then the f2 calculations does not have to be performed.

Also, to address the data variability issue the CV for earlier time points (i.e. 15 minutes or earlier) should not be more than 15% and not more than 10% for later time points. (N=12)

You are correct that very early time points can show the greatest variability especially for aged coated tablets. I try to avoid these, but I do try to include the 15 minute time point.

There may be other use ways to compare profiles, but this is what we have to live with for the time being.

Happy dissoluting!

Steve