Bioequivalence and Bioavailability Forum

Mahesh M ★ India, 2015-10-14 14:03 (3903 d 03:01 ago) Posting: # 15556 Views: 7,509	sparse sample design [Design Issues] Post reply
	Dear All, What is the sparse sample design? How it use in bioequivalence studies? Is it acceptable by regulatory body like FDA and EMA? I am very thankful to you please share some basics of that. Regards Mahesh

Helmut
★★★

Vienna, Austria,
2015-10-14 14:51
(3903 d 02:13 ago)

@ Mahesh M
Posting: # 15558
Views: 6,685

sparse sampling ⇒ population PK

Post reply

Hi Mahesh,

❝ What is the sparse sample design?

⇐ click here
In such a design only a few samples are collected from subjects. Hence, the name. Since C_max, AUC, etc. are not directly accessible, it is always used together with a population pharmacokinetics (PopPK) model. This designs is mainly employed in the target population (patients). It is possible to collect as few as 2–3 samples / subject:

Routine sampling in Phase II/III.
Special populations (children, cancer/AIDS, critical care patients, elderly…).

Unlike in “rich” data sets missing data and imbalance are not problematic:

Different doses / subject.
Different number of samples / subject.
Different sampling times / subject.

❝ How it use in bioequivalence studies?

No way. BE requires “rich” data sets, where NCA / SHAM is used to calculate PK-metrics – as opposed to PK modeling, where PK-parameters are estimated.

EMA BE-GL Section 4.1.4
A sufficient number of samples to adequately describe the plasma concentration-time profile should be collected. The sampling schedule should include frequent sampling around predicted t_max to provide a reliable estimate of peak exposure. […] The sampling schedule should also cover the plasma concentration time curve long enough to provide a reliable estimate of the extent of exposure which is achieved if AUC_(0-t) covers at least 80% of AUC_(0-∞).
FDA ANDA-guidance (draft) Attachment
We recommend drawing blood samples at appropriate times to describe the absorption, distribution, and elimination phases of the drug. For most drugs, we recommend collecting 12 to 18 samples, including a predose sample, per subject, per dose. This sampling should continue for at least three or more terminal elimination half-lives of the drug. […] The sample collection can be spaced in such a way that the maximum concentration of drug in the blood (C_max) and terminal elimination rate constant (K_el) can be estimated accurately.

❝ Is it acceptable by regulatory body like FDA and EMA?

No.

EMA BE-GL Section 4.1.5
The use of compartmental methods for the estimation of parameters is not acceptable.
FDA
Modeling not mentioned; only NCA throughout the text.

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

Mahesh M ★ India, 2015-10-14 15:55 (3903 d 01:09 ago) @ Helmut Posting: # 15560 Views: 6,200	sparse sampling ⇒ population PK Post reply
	Thank you so much for your response. Please share your thougts on below artical Thomas Jaki, Philip Pallmann, and Martin J Wolfsegger Estimation in AB/BA crossover trials with application to bioequivalence studies with incomplete and complete data designs Stat Med. 2013;32(30):5469–83 doi 10.1002/sim.5886 Regards Edit: Journal and DOI added. [Helmut]

Helmut
★★★

Vienna, Austria,
2015-10-14 17:14
(3902 d 23:50 ago)

@ Mahesh M
Posting: # 15562
Views: 6,210

incomplete ≠ sparse

Post reply

Hi Mahesh,

the paper should be read in conjunction with:

Yan Z. Comments on ‘Estimation in AB/BA crossover trials with application to bioequivalence studies with incomplete and complete data designs’. Stat Med. 2013;32(30):5484–6. doi 10.1002/sim.5995
Jaki T, Pallmann P, Wolfsegger MJ. Authors' reply to Comments on ‘Estimation in AB/BA crossover trials with application to bioequivalence studies with incomplete and complete data designs’. Stat Med. 2013;32(30);5487–8. doi 10.1002/sim.6000

The paper deals with AUC_0-t where only a few sampling time points are missing from a complete (rich) data set. In their example only one (prior to t_max) was missing. The paper does not deal with C_max. IMHO, the method is not applicable to a “sparse sample design”.

However, Thomas and Martin are members of the forum. They are by far more qualified than I am. ;-)

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

sparse sample design [Design Issues]

sparse sampling ⇒ population PK

sparse sampling ⇒ population PK

incomplete ≠ sparse