A total of 28 studies (n=910) were included in the review. The mean overall % adherence to quality criteria for included studies was 68.44% (range 58.06% to 87.10%).
Fluorodeoxyglucose-PET (positron emission tomography): The pooled sensitivity of fluorodeoxyglucose-PET for the detection of unknown primary tumours was 0.78 (95% CI 0.72 to 0.84) and the specificity was 0.79 (95% CI 0.74 to 0.83), based on 21 studies with 480 participants. Significant between study heterogeneity was identified for specificity. The maximal joint sensitivity and specificity (Q*), derived from the summary receiver operating characteristic curve, was 0.78. Fluorodeoxyglucose-PET had a lower sensitivity for tumours at the base of the tongue (0.68) and tonsils (0.77).
Fluorodeoxyglucose-PET/CT (computed tomography): The pooled sensitivity of fluorodeoxyglucose-PET/CT for the detection of unknown primary tumours was 0.81 (95% CI 0.74 to 0.87) and specificity was 0.83 (95% CI 0.78 to 0.87), based on eight studies with 430 participants. Significant between study heterogeneity was identified for both sensitivity and specificity. The maximal joint sensitivity and specificity (Q*), derived from the summary receiver operating characteristic curve, was 0.83.
Pooled estimates of likelihood ratios and log diagnostic odds ratios were also reported.
No data were reported comparing the accuracy of fluorodeoxyglucose-PET with fluorodeoxyglucose-PET/CT.