Bioequivalence and Bioavailability Forum

Hi Mahesh,

❝ For truncated design (AUC truncated to 72 hours) is it required to take additional samples 72.00 hrs onwards to identified intra subject variability of AUC0-∞ or regulatory agency accept PK data up to 72.00 hrs.?

Are the terms “truncated design” / “truncated study” common in India? Doesn’t make sense to me. Since you posted in the Regulatives / Guideline category I guess you are referring to the FDA?

You can use C_max and a suitably truncated AUC to characterize peak and total drug exposure, respectively. For drugs that demonstrate low intrasubject variability in distribution and clearance, you can use an AUC truncated at 72 hours (AUC_0-72 hr) in place of AUC_0–t or AUC_0–inf. For drugs demonstrating high intra­subject variability in distribution and clearance, AUC truncation should not be used.

It is beyond me how one should assess that. And, not or…

• Distribution
  For a drug described by a two-compartment model by a partial AUC from 2×t_max to the inflection point of the lin/log-plot?
  
• Elimination
  AUC_t–∞? t_½? In the latter case, which distribution does it follow? Lognormal? Doubt it. Use a nonparametric test of untransformed data? Duno.

The FDA received a few comments related to truncation:

1. Inflamax Research: What is the justification for this? El-Tahtawy et al. show in their Journal of BE & BA article (El-Tahtawy A, Harrison F, Zirkelbach JF, Jackson AJ (2011) Bioequivalence of Long Half-Life Drugs – Informative Sampling Determination – Parallel Designed Studies. J Bioequiv Availab 3: 056-061) that for highly variable drugs, truncation reduces the probability of falsely rejecting bioequivalence. Truncation does not increase the probability of obtaining false bioequivalence. Rather, truncated AUC shows lower RMSE% compared to AUC_inf, for example.
   
2. Sanofi suggested: For drugs that demonstrate low intra­subject (< X%) variability in distribution and clearance, you can use AUC truncated at 72 hours. For drugs demonstrating high intra­subject variability (≥ Y%) in distribution and clearance, AUC truncation should not be used.
   
3. Sandoz: Please remove the reference to intra-subject variability in distribution and clearance for use of AUC truncated. Use of an AUC truncated to 72 hours should be allowed for all long half-life drugs.
   
4. Manipal AcuNova: In case of truncated study when AUC is truncated at 72.00 hours and when there is a missing sample at 72.00 hours for one of the subject, then the handling of such data for AUC_0–t evaluation is not mentioned in this guideline.
   
5. Helmut Schütz: Kharidia et al. (Pharm Res 16(1):1999, 130–4) showed for amiodarone that truncation is applicable indeed. Furthermore, the guidance does not give a distinction between highly variable drugs (CVw ≥30% if administered as a solution in a replicate design) and highly variable drug products (CVw ≥30% if administered in a replicate design). In the light of global harmonization reconsider this section (EMA accepts truncated AUCs for all IR products independent from variability).

#2 is strange. Where would one get the X and Y from? Good point in #4. We had many discussions in the forum about it…

IMHO, to “demonstrate” low variability in a particular study is a half-baked idea. You would have to sample for a long time (3–5 × t_½). Hence, all the benefit (shorter hospitalization, missing samples in an ambulatory setting) would go with the wind. Let’s assume that you were able to show low variability, i.e., you have the AUC_0–t and AUC_0–∞ data and base BE on AUC_0–72. I bet the FDA will ask for BE of the conventional PK metrics as well.
Ways out? Literature data? Don’t fall into the trap of comparing the CV_w of AUC_0–∞ with the one of AUC_0–t. We would need the one of t_½ or of the extrapolated part. I have never ever seen these values.

Interested to learn how our experienced US-members deal with this stuff. John, Nathan, Linda, Angus, Louis?

---

PS: The EMA’s GL does not contain such a bizarre restriction for IR-products. Truncation is not acceptable for all MR-products. IMHO, makes sense for controlled release. Not accepting truncation for delayed release is debatable…