Qini curve and uplift

Question

I am building an uplift model from clinical data and I would like to evaluate it and compare it with the performance of different models. For this, I plotted the Qini curve and obtained the Qini coefficients. Also, in parallel, I plotted the uplift for groups of patients split based on the estimated conditional average treatment effect. I was expecting the two measures to give similar indications, but I find that in several cases they do not match. Here is one example: I used the Qini function from here https://humboldt-wi.github.io/blog/research/theses/uplift_modeling_blogpost/ and extracted the cumincrgain and uplift values for the two plots. How can this be explained? Thank you