you'll have to use for biomarkers for confirmation, morphology will give indications depending on how well differentiated the cells are. another option is using FDG-PET whole body scan of the subject to look for glycophylic regions/tissues which should guide further tests
You have also to take into account that sometimes it is imposible to determine the location of the primary tumors. This is a situation that clinicians have to face, they call it 'cancer of unknown origin'.
This is a popular topic, metastasis tumor for unknown origin. A carefully designed approach could help in solving some of the questions. Of cause for a pathologist you need to recognize the characteristic features of the tumor. Then work on the differentiate the carcinomas from sarcoma. Immunohistochemistry may aid in this. Next working on the tumors with specific marker of tissue origin, such as PSA, PAP from prostate, and AFP from hepatocellular carcinomas. Now a days, modern technologies could help a lot. For example, shared X-chromosome inactivation pattern in female patient, or same LOH pattern across multiple microsatellite loci, or even the tumors share the similar genomic mutations or chromosome alterations. See the chapter in the Book by Cheng, Liang MD. as attached.
A metastasis usually resembles the (normal) tissue of origin. A microscope is very helpful. If you find breast tissue in the liver, you know something is wrong. Histopathology is old but still powerful ;-)
I agree with the above discussions that routine histopathology, in other words microscopic identification of cell morphology is still best solution to identification of metastatic cell. It is also assisted by special stains, and Immunohistochemistry (IHC) to identify the type of cell with markers specific for epithelial cells, fibroblasts, stem cell markers, or even the CD markers for immune cells (B cells, T cells) are needed to characterize the cells of the tumours of unknown morphology. Also, the tissue context is important, as an example if there is metastasis in a lymph node, its own architecture is destroyed and is replaced by the specific pattern formed by the metastatic tumour, further staining with immunomarkers using IHC helps identify the tumour cells and thus identify the original/primary tumour. Sometimes the task may not be very easy, when the metastatic cell morphology is very undifferentiated depending on nature of the malignancy and stage of the tumour, if it is very advanced or aggressive in nature. CT and PET scans are done to search or locate the tumour of origin. Some marker antibody refs. below may be useful.
First check for the morphology features. Many tumors retain the original tissue characteristic, such as transition cell carcinoma from urinary tract. Second stain for tissue specific markers, such as PSA for prostate cancer, GFAP for gliomas and AFP for liver cancer or tumors with germ origin. If there was record on the primary tumor, the molecular features such as similar mutation, identical DNA methylation patterns, or concordant LOH across multiple loci.
If there was no clue of primary tumors, it appears of metastatic tumors with unknown origin. For such tumors a very first work is to determine if the tumor is epithelial, strmal or other origin. Then morphology features, tissue specific markers, or molecular work. Statistical data sometimes also provide a significant information as the knowledge gained about most frequent metastasis to the site could aid us to evaluate the chance.
I agree with all the previous answers and I wish this article help you.
Several hypotheses have been advanced to explain the origin of cancer metastasis. These involve an epithelial mesenchymal transition, an accumulation of mutations in stem cells, a macrophage facilitation process, and a macrophage origin involving either transformation or fusion hybridization with neoplastic cells.