Why does geodetic2aer give ranges differing extremely from euclidian distance?

DISCLAIMER: This is a duplicate of http://stackoverflow.com/questions/39129748/matlab-mapping-toolbox-function-geodetic2aer-gives-extremely-different-ranges-th and http://de.mathworks.com/matlabcentral/newsreader/view_thread/346036, but I don't seem to get a lot of responses there. So I thought I might just try MATLAB Answers for the first time.

Now to the problem: I am calculating slant range distances between different points, of which I have both the lat-lon-height as well as the ENU coordinates (UTM, same heights as for in lat-lon-height). Now, I am using the function geodetic2aer from MATLAB's Mapping Toolbox to determine the slant ranges between these points.

In addition, I calculate the euclidian norm in XYZ space as defined by the UTM ENU-coordinates. The strange thing is: I get differences in range of up to 35 meters! That can't be right, can it?

I try to provide a minimal working example here:

P1_utm = [19237270 4714162 246.197998046875];

P1_llh = [42.5351900557357 -72.1991953382484 246.197998046875];

P2_utm = [19239222.083941354 4716423.97292039 778.630006880859];

P2_llh = [42.5563144623905 -72.176008488885 772.845077514484];

range_utm = norm(P1_utm-P2_utm);

S = referenceEllipsoid('wgs84');

[~,~,range_llh] = geodetic2aer(P1_llh(1),P1_llh(2),P1_llh(3),P2_llh(1),P2_llh(2),P2_llh(3),S);

This way you will end up with

range_utm = 3034.90313695554

range_llh = 3067.98897478523

What's happening here?? The transformation between lat-lon-height and UTM coordinates has been carried out using Peter Wasmeier's Geodetic Transformations Toolbox from MATLAB file exchange, and transforming back and forth gives residual errors in the 1e-4 to 1e-6 range - so this shouldn't be the problem.