Cox regression model, hazard ratio , using categorical data.

Hello everyone!
I am studying the effect of 8 arm of treatments on a cohort of 80 mice. The variable I have are: time status and arm of treatment.
I generated KM curves and obtained by log rank test both a general p.value and pairwise comparisons between couples of curves.

Then, I needed to calculate the hazard ratio.

So I used cox regression model.
Considering ARM a categorical value ( scored from 0-7) I have some problems with my output.
If I don t consider ARM as a categorical value spss gives me just one value for expB. (Is it a general value ?)
If I consider it as a categorical ( as it is) it let me choose an indicator and it calculates the HR using as indicator or the final or the initial arm.

Is it correct? Is it possible to get an HR for each pairwise comparison?
Which p should I use for each pairwise comparison? the one from KM curves or the one generated with cox regression output for HR?

Thank you very much, please be kind with a MD who is just approaching to statistics