Radiographic indices for the diagnosis of atlantoaxial instability in toy breed dogs [corrected]

Toy breed dogs are overrepresented for atlantoaxial instability. Radiography is a standard diagnostic test, however published toy breed-specific radiographic measurements are lacking and diagnosis remains largely subjective. Aims of this retrospective, diagnostic accuracy study were to describe normal values and determine whether some quantitative radiographic criteria strongly support a diagnosis of atlantoaxial instability specific to toy breed dogs. Neutral lateral and ventrodorsal radiographs of 102 toy breed dogs (92 control, 10 affected) were reviewed. The median C1-C2 overlap (the distance of overlap between the C2 spinous process and the dorsal arch of C1) was +4.65 mm in control dogs and -5.00 mm in atlantoaxial instability cases. A C1-C2 overlap ≤ +1.55 mm was the most sensitive (100%) and specific (94.5%) radiographic measurement in the diagnosis of atlantoaxial instability. Three relative measurements were performed: the ventral atlantodental interval to dorsal atlantodental interval ratio, the dens/C2 ratio, and the C1-C2 angle. These three relative measurements had good specificity (94.5, 86.9, and 98.9%, respectively), lower sensitivity (80.0, 66.7, and 60.0%, respectively), and were not influenced by body weight (P > 0.05). Absolute measurements (including absolute dens length and atlantoaxial distance) were significantly correlated with body weight (P < 0.05) diminishing their utility in the diagnosis of atlantoaxial instability. Decreased C1-C2 overlap strongly supports atlantoaxial instability. The ventral atlantodental interval/dorsal atlantodental interval ratio, dens/C2 ratio, and C1-C2 angle may provide further support but may be normal in individual cases.