A relatively long-tailed distribution, such as a sustained release formulation, could have this property. It will happen if one treatment has a large Cmax and the distribution for the other treatment is "stretched out".
It looks like intrasubject variability for Cmax was larger than assumed. It resulted in very wide confidence interval, which exceeded both lower and upper limits. Increasing number of subjects could help.
I would look to reformulate personally, increasing the number of subjects could look like 'forced' BE and you may run into ethical concerns. Unless your innovator is also highly variable in which case you may need to look at RSABE. there are free workflow templates for Phoenix WinNonlin to help you with such analyses.https://support.certara.com/forums/topic/705-rsabe-templates-for-phoenix-14/
90% CI of Cmax falling above 1.2500or below 0.8000 will not be acceptable to the regulatory body unless the drug concerned can be shown to have inherently high variability in bioavailability and the 90% CI is very wide.
Normally, when the Cmax falls below 0.8000, it means that the dissolution is slower than that of the innovator and when above 1.2500, the dissolution is faster. You will have to reformulate the product and compare the dissolution at pH values of 1.2, 4.5 and 6.8. The last pH profile may not be that critical for immediate release products.
To increase the chance of achieving a narrow 90% CI and hence achieving bioequivalence, it is important to understand how the 90% CI is calculated. The residual variance obtained from the ANOVA of the log transformed raw data is used to calculate the 90% CI. This residual variance comprises variability among the different units of the product being tested (wide variation in disintegration/dissolution among the same batch of tablets for example), assay error of the analytical method (poor precision and accuracy) and intrasubject variability (nothing we can do about it). For the former 2, we should strive to have a good formulation with small variability among the units and for the assay, make sure we use a high end equipment with good precision and accuracy as well employing personnel with good laboratory skills.