Regression eqn in first differences. Converting fitted values back to levels

I have a regression eqn, where all variables(dependent and explanatory) enter in first differences except one explanatory variable, where I need to show the contribution of each explanatory variable to dependent variable in levels. There is a solution to interpret the coefficients back in levels like this: [login to view URL] here he assume y0=0 and x0=0. If they are not zero, how this solution would change. Also if there is another explanatory variable in levels, how would this change i.e assume my eqn is y(t)-y(t-1)=a+b(x(t)-x(t-1))+c*z(t)+e(t)