Fluorine-18 deoxyglucose positron emission tomography, magnetic resonance imaging and bone scintigraphy for the diagnosis of bone metastases in patients with lung cancer: which one is the best?--a meta-analysis

Aims: To carry out a meta-analysis to compare fluorine-18 deoxyglucose ((18)FDG) positron emission tomography (PET), magnetic resonance imaging (MRI) and bone scintigraphy imaging for the diagnosis of bone metastases in patients with lung cancer.

Materials and methods: MEDLINE, EMBASE, Scopus and other databases were searched for relevant original articles published between January 1995 and January 2010. Inclusion criteria were as follows: (18)FDG PET, MRI or (99m)Tc-MDP bone scintigraphy was carried out to detect bone metastases in patients with lung cancer; sufficient data were presented to construct a 2×2 contingency table; histopathological analysis and/or close clinical and imaging follow-up and/or radiographic confirmation by multiple imaging modalities were used as the reference standard. Two reviewers independently extracted data. META-DiSc was used to obtain pooled estimates of sensitivity, specificity, diagnostic odds ratio (DOR), summary receiver operating characteristic (SROC) curves and the *Q index.

Results: In total, 14 articles that consisted of 34 studies fulfilled all inclusion criteria. On a per-patient basis, the pooled sensitivity estimates for PET, MRI and bone scintigraphy were 91.9, 80.0 and 91.8%, respectively. The sensitivity for PET and bone scintigraphy were significantly higher than for MRI (P<0.05). There was no significant difference between PET and bone scintigraphy (P>0.05). The pooled specificity estimates for PET, MRI and bone scintigraphy were 96.8, 90.6 and 68.8%, respectively. The specificity for PET was significantly higher than for MRI and bone scintigraphy (P<0.05), and the specificity for MRI was significantly higher than for bone scintigraphy (P<0.05). The pooled DOR estimates for PET, MRI and bone scintigraphy were 365.5, 53.8 and 34.4, respectively. The DOR for PET was significantly higher than for MRI and bone scintigraphy (P<0.05). There was no significant difference between MRI and bone scintigraphy (P>0.05). The SROC curve for PET showed better diagnostic accuracy than for MRI and bone scintigraphy. The SROC curve for MRI was better than for bone scintigraphy. The *Q index estimates for PET, MRI and bone scintigraphy were 0.933, 0.903 and 0.857, respectively. The *Q index for PET and MRI were significantly higher than for bone scintigraphy (P<0.05). There was no significant difference between PET and MRI (P>0.05). On a per-lesion basis, the pooled sensitivity estimates for PET, MRI and bone scintigraphy were 95.0, 83.8 and 71.5%, respectively. The sensitivity for PET was significantly higher than for MRI and bone scintigraphy (P<0.05), and the sensitivity for MRI was significantly higher than for bone scintigraphy (P<0.05). The pooled specificity estimates for PET, MRI and bone scintigraphy were 94.6, 96.3 and 91.0%, respectively. The specificity for MRI was significantly higher than for PET and bone scintigraphy (P<0.05), and the specificity for PET was significantly higher than for bone scintigraphy (P<0.05). The pooled DOR estimates for PET, MRI and bone scintigraphy were 431.9, 158.1 and 9.0, respectively. The DOR for PET was significantly higher than for MRI and bone scintigraphy (P<0.05) and the DOR for MRI was significantly higher than for bone scintigraphy (P<0.05). The SROC curve for PET and MRI showed better diagnostic accuracy than for bone scintigraphy. There was no significant difference between PET and MRI. The *Q index estimates for PET, MRI and bone scintigraphy were 0.953, 0.962 and 0.778, respectively. The *Q index for PET and MRI were significantly higher than for bone scintigraphy (P<0.05). There was no significant difference between PET and MRI (P>0.05).

Conclusion: (18)FDG PET was found to be the best modality to detect bone metastasis in patients with lung cancer, both on a per-patient basis and a per-lesion basis; MRI had the highest specificity on a per-lesion basis. For the subgroup analysis of (18)FDG PET, PET/computed tomography was shown to be better than PET and there were no significant differences between using (68)Ge and computed tomography for attenuation correction on a per-patient basis.