An MRI study of autism: The cerebellum revisited

To the Editor:

Piven et al.1 reported that in their autistic and control subjects, the midsagittal area of the cerebellar vermis lobules VI and VII was 0.31 mm² (that is, 310 μm², which is less than half the cross sectional area of the cell body of a single Purkinje neuron; for instance, Ito2 reports it to be 676 μm² in cats). The same report stated that the cerebellar vermis lobules I to V were 0.50 ± 0.0 mm² in autistic patients and 0.50 ± 0.06 mm² in control subjects, sizes just as impossibly small. Another cause for incredulity was the standard deviation of 0 mm².

When these impossibilities were brought to the attention of Piven by one of my colleagues, his laboratory staff replied that “mm²” was a typo and it should have been “cm².” Because 1 cm = 10 mm, and 1 cm² = 100 mm², this correction would bring the reported vermis area measures up to 31 mm² for lobules VI and VII and 50 mm² for lobules I to V; sizes in the rat cerebellum range. When the new concern was again voiced, Piven’s laboratory gave yet another set of numbers: 308.3 ± 91.5 mm² and 311.2 ± 98.0 mm² for lobules VI and VII and 449.5 ± 71.5 mm² and 457.3 ± 60.7 mm² for lobules I to V.

This third set of numbers still does not appear to be correct. Perhaps they reported their numbers rounded off. If so, their published figures of 0.5 mm² were inaccurate by 1,000 times; rescaling by this factor would bring their area size up to 500 mm², still about 11% in error of …