Bioequivalence and Bioavailability Forum

Hi Mr. Rao,

you confirmed in registration that you have read and will comply with the forum’s policy. Please read the relevant section again. You managed it to full quote in all of your replies. That’s an amazing percentage.
Hence, this is the first warning.

You shouldn’t expect to learn the basics from an internet forum. Suggested reading:

1. Hauschke D, Steinijans V, Pigeot I. Bioequivalence Studies in Drug Development. Methods and Applications. Chichester; Wiley: 2007.
   Chapter 8: Analysis of pharmacokinetic interactions.
   
2. Chow S-C, Liu J-p. Design and Analysis of Bioavailability and Bioequivalence Studies. Boca Raton; Chapman & Hall / CRC Press: 3^rd ed. 2009.
   Chapter 18.2: Drug Interaction Studies.
   
3. Patterson S, Jones B. Bioequivalence and Statistics in Clinical Pharmacology. Boca Raton; Chapman & Hall / CRC Press: 2^nd ed. 2017.
   Chapter 8.4: Food Effect Assessment and Drug-Drug Interactions (DDIs).
   
4. International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. Drug Interaction Studies. M12. 08 June 2022. Online.

If a clinically relevant interaction is found, there’s no reason to panic. It only has to be stated as such in the label / prescribing information.

❝ […] concluded there is no clinically relevant drug–drug interaction between CHF 5993 and result of cimetidine co-administration, with total exposure increased by 16 %, and peak concentration by 26 %.

❝ I need to know how they concluded.

What do you fail to understand?

Methods
This two-period, open-label, crossover study examined the drug–drug interaction of CHF 5993 and cimetidine. In one period, subjects received cimetidine 800 mg twice daily for 6 days; on the fourth day they also received CHF 5993 (BDP/FF/GB 400/24/100 µg). In the other, they received CHF 5993 alone. Primary objective was to compare the area under the plasma concentration–time curve from time 0 to last quantifiable concentration (AUC_0–t) of GB, with and without cimetidine. Secondary endpoints included GB AUC_0–12h, maximum concentration (C_max) […]

Results
Twenty-six subjects were randomised; 25 completed. Co-administration of CHF 5993 and cimetidine resulted in small, statistically significant increases in GB AUC_0–t, AUC_0–12h and C_max vs CHF 5993 (ratios 1.16, 1.21 and 1.26, respectively) […]

Conclusions
Overall, this study indicates that there is no clinically relevant drug–drug interaction between GB and the other components of the triple combination of GB/FF/BDP that comprise extrafine CHF 5993, and cimetidine […]

Statistically significant = 90% confidence interval of the ratio of adjusted geometric means does not include 1. Unfortunately only three significant digits are given in Table 3: AUC_0–t 1.16 (1.07, 1.27), AUC_0–12h 1.21 (1.08, 1.36), C_max 1.26 (1.00, 1.58). Perhaps the lower confidence limit of C_max in full precision was >1.

Contrary to bioequivalence, where the clinically relevant difference \(\small{\Delta}\) is commonly set to 20%, leading to the limits \(\small{\{\theta_1,\theta_2\}=\{1-\Delta,\,(1-\Delta)^{-1}\}=\{0.8,\,1.25\}}\), in DDI-studies a different (while still pre-spe­ci­fied) \(\small{\Delta}\) can be chosen. The paper does not give \(\small{\Delta}\). From the sample size section (page 272) one can assume that \(\small{\Delta=0.15}\), which would be pretty strict. Since the authors claimed that the increase was not clinically relevant, one could recalculate \(\small{\widehat{\Delta}_\text{r}}\) from the upper confidence limit by \(\small{\widehat{\Delta}_\text{r}=\left|CL_\text{upper}^{-1}-1\right|}\). For the ‘worst’ PK metric C_max, we get \(\small{\widehat{\Delta}_\text{r}\approx\left|1.58^{-1}-1\right|\approx0.367}\). IMHO, that’s a lot. If we consider only the primary PK metric AUC_0–t, \(\small{\widehat{\Delta}_\text{r}\approx0.213}\). The study was extremely underpowered (≈55%) for this PK metric. Interesting that the within-subject CV of C_max was smaller than the one of AUC_0–t. Not impossible but rare.

-script below.

---

library(PowerTOST)
design   <- "2x2x2"
n        <- 25
metrics  <- c("AUC0-t", "AUC0-12", "Cmax")
ratio    <- c(1.16, 1.12, 1.26)
CL.lower <- c(1.07, 1.08, 1.36)
CL.upper <- c(1.27, 1.36, 1.58)
res      <- data.frame(metric = metrics, ratio = ratio,
                       CL.lower = CL.lower, CL.upper = CL.upper,
                       Delta.r = NA_real_, CV = NA_real_,
                       power = NA_real_)
for (j in seq_along(metrics)) {
  res$Delta.r[j] <- abs(CL.upper[j]^-1 - 1)
  res$CV[j]      <- suppressMessages(
                      CI2CV(lower = CL.lower[j],
                            upper = CL.upper[j],
                            design = design, n = n))
  res$power[j]   <- suppressMessages(
                      power.TOST(CV = res$CV[j], theta0 = ratio[j],
                                 theta1 = 1 - res$Delta.r[j],
                                 design = design, n = n))
}
res[, 5:7] <- signif(res[, 5:7], 4)
print(res, row.names = FALSE, right = FALSE)

Gives:

metric  ratio CL.lower CL.upper Delta.r  CV     power
AUC0-t  1.16  1.07     1.27     0.2126   0.1780 0.5453
AUC0-12 1.12  1.08     1.36     0.2647   0.2410 0.8761
Cmax    1.26  1.36     1.58     0.3671   0.1555 0.9996