For majority of cancers, primary lesions are usually identified first, followed by LNM or distant metastases. In patients with CUP, the metastatic lesions are identified first, and then attempts are made to track the primary lesion. In our study, as high as 95% of the pathology-confirmed primary sites in patients with clinical CUP were located at the drainage area of LNM, which first demonstrated that SLN theory is helpful in tracking the primary site of CUP. Based on this, a potential candidate primary lesion that could metastasize to the lymph nodes should be extensively investigated by thorough physical or radiological examination, or even biopsy. To our knowledge, there has been no previous report or published data focusing on this issue.
CUP was once viewed almost as a unique type of cancer. Now it is believed that most CUPs have primary sites and probably retain the gene signature of the putative primary origin5. Determining the primary site or the tissue-of-origin may have a substantial effect on therapeutic approaches as well as patient survival. However, most prior efforts have focused on IHC staining or gene expression profiling to determine the possible tissue of origin, and these approaches have problems such as high cost, the requirement for sufficient tumor tissue, and patients no longer being treatment candidates by the time results become available19. Moreover, they are unavailable in many districts, and can produce false-positive and false-negative data.
18F-FDG PET-CT is also a valuable diagnostic tool for patients with CUP. One meta-analysis showed that the overall primary tumor detection rate, pooled sensitivity, and specificity of 18F-FDG PET-CT were 37%, 84%, and 84% respectively20. Several studies found that 18F-FDG PET-CT could detect the occult primary tumor in as high as 49–57% of CUP cases21 − 23. The factors limiting the use of 18F-FDG PET-CT include its high cost and its limited value in small size tumor and tumors exhibiting a low FDG uptake24. One recent study by Cengiz et al. showed that 18F-FDG PET-CT does not represent a clear diagnostic advantage over conventional imaging methods regarding the ability to detect the primary tumor site25,26.
Therefore, radiological examination, pathological features, and molecular profiling are still not adequate for tracking of the primary site of CUP. Our study shows that as many as 95% of the confirmed primary sites in CUP cases are consistent with SLN theory. Clinical use of SLN theory includes cooperation between clinical oncologists and diagnostic experts to systemically review all available clinical information, to identify clues to indicate potential primary lesions, and then to undertake biopsy or operation on the prime suspect for pathological diagnosis. With this multidisciplinary approach, the tracking of primary lesions of CUP could be much improved.
Numerous studies have confirmed that SLNs are the first stop reached by metastatic cancer cells as they enter the regional lymphatic basin in the vast majority of cancer patients16. SLN theory has been widely used in surgery as SLN biopsy, resulting in fewer axillary lymph node dissections and fewer lymphedema cases, and thus improved quality of life27 − 30. Therefore, SLN theory has been of vast assistance in nodal staging and treatment options. However, to our knowledge, there has been no study to assess the role of SLNs in tracking the primary of CUP.
Early dissemination, aggressiveness, and unpredictable metastatic patterns are characteristic of CUP and it has been recommended that physicians should not rely on patterns of metastases to determine the primary site. However, a study by Hemminki et al. suggested that location of metastasis may predict site-specific cancer deaths and provide insights into the location of primary tumors31. SLN theory has also been used in some cases of CUP. For example, adenocarcinoma in axillary lymph nodes in women are often presumed to originate from breast cancer32, consistent with our findings.
To our knowledge, this is the largest sample size study showing that in clinical CUP patients, tumor grading affects the final identification of primary lesions, with a statistically significant difference of 94% vs. 77% for differentiated vs. un- or poorly-differentiated tumors, respectively. Previously, Peter et al. found that sinonasal undifferentiated carcinoma had higher rates of nodal involvement than sinonasal small-cell carcinoma33. Another study showed that patients with undifferentiated carcinoma had higher rates of advanced stage disease than patients with endometrioid adenocarcinoma34. These findings suggested that undifferentiated carcinoma was more likely to undergo early metastasis, making it harder to identify the primary site in CUP.
Besides the largest sample size, another strength is that 75% patients had 18F-FDG PET-CT scanning, providing a systemic review of cancer status both in and outside the lymph nodes for each patient. However, our study also has its own limitations. First, most cases did not receive second opinions from the CMUP multidisciplinary team. However, all our patients were diagnosed and treated following our institute’s guidelines or rules. Second, the diagnosis of LNM in 6% of patients was established by 18F-FDG PET-CT only. However, the accuracy of 18F-FDG PET-CT in predicting axillary LNM was as high as 78–95%35, and it detected abnormal lymph nodes more often than CT36. Therefore, 18F-FDG PET-CT is the best available method integrating the structure and function of lymph nodes. Third, the number of identified primary site of CUP cases with abdominal or pelvic LNM were fewer than those with cervical or axillary LNM. The possible reasons are: (1) LNM in abdominal or pelvic cavity are deep and hard to find, so there tend to be more than two directions of lymph drainage, making it less likely that potential primary lesions can be located. (2) CUP patients with cervical lymph node or axillary LNM are favorable types and have better prognosis, allowing more time for the primary lesion to be detected1.
As there was no statistically significant difference between group A and group B in terms of clinical and pathological features, it is reasonable to extrapolate the findings of group A to group B, and possibly to group C, or at least for differentiated CUP.