Post by silentbob on Mar 31, 2014 17:38:58 GMT -5
The central point of concern for longs is probably Efficacy in Type1, with good reason.
The FDA raises several issues and points out several things on how the study might have failed given a change of certain factors. I include my comments below.
- FDA: Afrezza is statistically worse versus the comparator group (+0.22 A1c)
-> This difference is small enough for it to become a matter of patient preference.
-> Mannkind will also point to the fact that the comparator has a large basal component (the long tail) which influences the results.
- FDA: Afrezza just only just stayed within the pre-specified non-inferiority margin, and one worst case analysis would point to a potential study failure.
-> The non-inferiority margin is 0.4 for a reason. If 0.4 was not enough, the margin would be lower.
-> The worst case analysis does not seem relevant considering mean A1c of dropouts was nowhere near what the worst case analysis implies (see p173).
- FDA: The comparator group may not have been sufficiently titrated.
-> Being type1's, patients in the comparator group have been injecting basal and prandial insulin for many years. Probably decades.
They know where they will get too many hypos and would be unwilling to increase dosage further.
-> Hypoglycemia is the wall that injected insulin titration runs into, and patients are wise to be wary of it.
- FDA: Basal insulin use was higher in the Afrezza group, and the trial may have failed if basal was equal across groups.
-> Technically this is true, but only for a bureaucrat.
-> The reality is that only Afrezza allows increasing basal dosing without causing excessive hypos, as evidenced by the lower hypo rate.
This is one of the key benefits of Afrezza yet the reviewer tries to paint a negative view of it.
- FDA: Hypos are lower with Afrezza but this may be due to the higher A1c.
-> If you only consider the whole group you may agree with this.
-> However, for patients reaching targets <6.5 A1c and <7.0 A1c Afrezza still shows significantly lower hypos.
This directly contradicts the reviewers statement.
Clearly there is risk here.
However I believe Mannkind has good arguments against the reviewer negativity.
A statistician without any interest in the real world of diabetics may side with the reviewer.
However I think any logical human being would look beyond the strict A1c numbers and consider how the comparator is actually partly basal insulin, and partly slow prandial insulin, and that this combination does have a slightly bigger effect on A1c but really makes it a worse pransial insulin.
Even through all the negativity I cannot imagine panelists to vote against a revolutionary treatment like this simply because the FDA has found a way to massage the data just so that a negative conclusion could be drawn. Assuming they do not have pre-existing bias against Afrezza I expect a 6-4 or better vote in favor of Afrezza in Type1's.
I don't say this with 100% confidence though. As we have learned before, you never know what is going on behind the curtain.
In particular there are various unfairly negative remarks in the document, while positive effects are downplayed.
I hope this is not a sign of FDA bias.
I hope the panelists give such an overwhelmingly positive vote so that the choice is removed from the FDA's hands.
The FDA raises several issues and points out several things on how the study might have failed given a change of certain factors. I include my comments below.
- FDA: Afrezza is statistically worse versus the comparator group (+0.22 A1c)
-> This difference is small enough for it to become a matter of patient preference.
-> Mannkind will also point to the fact that the comparator has a large basal component (the long tail) which influences the results.
- FDA: Afrezza just only just stayed within the pre-specified non-inferiority margin, and one worst case analysis would point to a potential study failure.
-> The non-inferiority margin is 0.4 for a reason. If 0.4 was not enough, the margin would be lower.
-> The worst case analysis does not seem relevant considering mean A1c of dropouts was nowhere near what the worst case analysis implies (see p173).
- FDA: The comparator group may not have been sufficiently titrated.
-> Being type1's, patients in the comparator group have been injecting basal and prandial insulin for many years. Probably decades.
They know where they will get too many hypos and would be unwilling to increase dosage further.
-> Hypoglycemia is the wall that injected insulin titration runs into, and patients are wise to be wary of it.
- FDA: Basal insulin use was higher in the Afrezza group, and the trial may have failed if basal was equal across groups.
-> Technically this is true, but only for a bureaucrat.
-> The reality is that only Afrezza allows increasing basal dosing without causing excessive hypos, as evidenced by the lower hypo rate.
This is one of the key benefits of Afrezza yet the reviewer tries to paint a negative view of it.
- FDA: Hypos are lower with Afrezza but this may be due to the higher A1c.
-> If you only consider the whole group you may agree with this.
-> However, for patients reaching targets <6.5 A1c and <7.0 A1c Afrezza still shows significantly lower hypos.
This directly contradicts the reviewers statement.
Clearly there is risk here.
However I believe Mannkind has good arguments against the reviewer negativity.
A statistician without any interest in the real world of diabetics may side with the reviewer.
However I think any logical human being would look beyond the strict A1c numbers and consider how the comparator is actually partly basal insulin, and partly slow prandial insulin, and that this combination does have a slightly bigger effect on A1c but really makes it a worse pransial insulin.
Even through all the negativity I cannot imagine panelists to vote against a revolutionary treatment like this simply because the FDA has found a way to massage the data just so that a negative conclusion could be drawn. Assuming they do not have pre-existing bias against Afrezza I expect a 6-4 or better vote in favor of Afrezza in Type1's.
I don't say this with 100% confidence though. As we have learned before, you never know what is going on behind the curtain.
In particular there are various unfairly negative remarks in the document, while positive effects are downplayed.
I hope this is not a sign of FDA bias.
I hope the panelists give such an overwhelmingly positive vote so that the choice is removed from the FDA's hands.
