force slope in fitlm

I'd like to use a linear regression, but to force the slope to take a fixed (=predefined) value. How is this possible using fitlm?

Backround:

I use a qqplot to check for normality of the residuals. In addition, I'd like to include a 95% confidence interval. One way to obtain the confidence interval is to fit a linear rgeression line and let matlab to the work:

mdl         = fitlm(X,Y);            
[ypred,yci] = predict(mdl,X);

Unfortunately, the linear regression line uses all the data to obtain an estimate for the slope. If outliers are present, the slope of the linear regression is not robust. Therefore, I'd like to determine a robust estimate of the slope in advance (e.g. using the first and third sample quantile -- which is also used in matlab's qqplot) and restrict the slope. Then, matlabs predict() provides a simple way to obtain a confidence interval. How is this done?