Detection of the primary tumor can change the prognosis of patients with CUP by enabling targeted treatment. Previous studies have indicated that PET and PET/CT are useful for the detection of primary sites [7, 8, 13–15]. In this study, neither the PET nor PET/CT imaging improved the detection of the putative primary sites in patients with CUP that already had thorough conventional diagnostic workups. Neither PET nor PET/CT detected the primary sites in any of the 20 patients with CUP.
Previous studies have reported that PET detects primary lesions in 24%-41% of patients with CUP . However, these studies differ in the definition of patients with CUP and conventional workups. Lassen et al.  identified primary cancers in nine out of 20 patients (45%) in their prospective study. Among them, eight patients had primary lung cancers and did not undergo chest CTs during the conventional workups. Bohuslavizki et al.  studied 53 patients with CUP, of whom primary tumor sites were detected in 20 (37.8%) using the PET. This study did not include CT or magnetic resonance imaging (MRI) in the conventional workup. Alberini et al.  investigated 41 patients with CUP. PET detected primary sites in 26 patients; however, 15 patients had the primary site revealed in the conventional workup. The results of previous studies differed from the findings of this study, which was limited to patients for whom a complete conventional workup, including CT or MRI of suspected lesions, failed to show a primary lesion.
According to the previous literature , 20-27% of primary sites are identified before the patients with CUP die. The primary site was not initially identified in all of our study patients and was detected in 2 of 20 cases by immunohistochemical staining of biopsied metastatic lesions during the follow-up period. We included only patients whose primary sites were not identified by initial complete diagnostic workups including CT of chest, abdomen, pelvis and endoscopic examinations. Therefore, the patients in our study might have less chance to detect primary sites.
The poor resolution of PET has been superseded by PET/CT, which identifies anatomical landmarks more accurately. The PET/CT detects the primary tumor in 22-73% of patients with CUP, according to a recent review article . However, in our study PET/CT did not improve the detection rate of primary sites in patients with CUP. The findings are consistent with those reported by Gutzeit et al. that the identification rate of primary cancers using PET/CT was 33%, but the diagnostic accuracy did not differ significantly from that of the other modalities even though it revealed more anatomical detail.
The PET and PET/CT have gained widespread acceptance as useful methods for the management of cancer . However, they do not appear to be effective in identifying a primary lesion after a thorough conventional workup fails to do so. This may be due to the biological characteristics of primary tumors. Primary tumors may disappear after seeding metastases because their angiogenetic incompetence leads to marked apoptosis and cell turnover . Primary tumors that have regressed would not be detected by PET or PET/CT. In this study, neither the PET nor PET/CT detected primary sites in six patients with cervical lymph node metastases, contrary to the findings of other studies [6, 12]. The 6 patients with cervical lymph node metastases in this study included 4 poorly differentiated carcinomas and 2 squamous cell carcinomas. The metastases of poorly differentiated carcinoma would have marked cell turnover and apoptosis and that leads to early regression of the primary site. A higher portion of poorly differentiated carcinoma would be one reason for a low detection rate of primary sites in cases with cervical lymph node metastases.
The PET or PET/CT revealed FDG uptake lesions other than the known metastases in seven patients. These additional uptake lesions were of no value for detecting the primary sites of tumors, and false positive FDG uptake lesions complicated the diagnosis. Despite no additional value of the PET or PET/CT in the detection of the primary site, primary lesions were identified in two cases by immunohistochemical staining of biopsied metastatic lesions during the follow-up period (Table 2). Various immunohistochemical markers were used to identify the primary site according to the pathology of the metastatic sites. In one patient (patient no. 6) with metastatic adenocarcinoma, immunohistochemical markers such as the estrogen-receptor, progesterone-receptor, C-erbB2, cytokeratin 7, cytokeratin 20, TTF-1(Thyroid Transcription Factor-1) and GCDFP-15(Gross Cystic Disease Fluid Protein 15) were used to identify the primary site. Immunohistochemical staining for the estrogen-receptor, progesterone-receptor and C-erbB2 were positive, but cytokeratin 7, cytokeratin 20, TTF-1 and GCDFP-15 were negative. Therefore, the primary cancer was presumed to be a breast cancer. Cytokeratin 7, cytokeratin 20 and TTF-1 were used in a patient with cervical lymph node metastasis (patient no. 11). Results of immunohistochemical staining showed positive cytokeratin 7, negative cytokeratin 20 and focal positive TTF-1 in this patient. The primary cancer was presumed to be a non-small cell lung cancer. A careful conventional workup that includes immunohistochemistry would be helpful for cases in which the primary site cannot be successfully identified using PET or PET/CT. PET or PET/CT scans are easy to perform because of their non-invasiveness ; however, subsequent invasive procedures and biopsies are inevitable for pathology confirmation of the results of the PET and PET/CT.
The limitations of this study included the following. First, the sample size was small and the study design was retrospective. Second, this study was performed in the early stages of PET and PET/CT, when the PET and PET/CT were not widely used. It is possible that the study results do not reflect current PET or PET/CT scanning.
In conclusion, neither PET nor PET/CT improved the detection of primary sites in patients with CUP in our study. Although it is inconclusive because of small sample size of the study, the additional value of PET or PET/CT for the detection of primary sites in patients with CUP might be less than expected; especially in patients that have already had extensive conventional diagnostic workups. Further study is needed to validate this finding.