Lumbar plain radiograph is not reliable to identify lumbosacral transitional vertebra types according to Castellvi classification principle

BACKGROUND: The anteroposterior view of the lumbar plain radiograph (AP-LPR) was chosen as the original and first radiographic tool to determine and classify lumbosacral transitional vertebra with morphological abnormality (MA-LSTV) according to the Castellvi classification. However, recent studies found that AP-LPR might not be sufficient to detect or classify MA-LSTV correctly. The present study aims to verify the reliability of AP-LPR on detecting and classifying MA-LSTV types, taking coronal reconstructed CT images (CT-CRIs) as the gold criteria. METHODS: Patients with suspected MA-LSTVs determined by AP-LPR were initially enrolled. Among them, those who received CT-CRIs were formally enrolled to verify the sensitivity of AP-LPR on detecting and classifying MA-LSTV types according to the Castellvi classification principle. RESULTS: A total of 298 cases were initially enrolled as suspected MA-LSTV, among which 91 cases who received CT-CRIs were enrolled into the final study group. All suspected MA-LSTVs were verified to be real MA-LSTVs by CT-CRIs. However, 35.2% of the suspected MA-LSTV types judged by AP-LPR were not consistent with the final types judged by CT-CRIs. Two suspected type IIIa and 20 suspected type IIIb MA-LSTVs were verified to be true, while 9 of 39 suspected type IIa, 9 and 3 of 17 suspected type IIb, and 11 of 13 suspected type IV MA-LSTVs were verified to truly be type IIIa, IIIb, IV and IIIb MA-LSTVs by CT-CRIs, respectively. Incomplete joint-like structure (JLS) or bony union structure (BUS) and remnants of sclerotic band (RSB) between the transverse process (TP) and sacrum were considered to be the main reasons for misclassification. CONCLUSION: Although AP-LPR could correctly detect MA-LSTV, it could not give accurate type classification. CT-CRIs could provide detailed information between the TP and sacrum area and could be taken as the gold standard to detect and classify MA-LSTV.