Biopsy material should be examined by an anatomical pathologist with expertise in this area, and may require referral for a second opinion. The use of specific IHC panels is promising in assigning a likely primary site of origin (Park et al. 2007). The choice of which IHC stains or ancillary pathology investigations to perform is highly dependent on the clinical setting, the morphology of the malignant cells and the available material in biopsy specimens. Therefore, no one set of IHC stains or ancillary tests is appropriate for all patients. The 20 cytokeratin (CK) subtypes are typically expressed in carcinomas. A CK7 plus CK20 staining pattern may lead to additional IHC staining and specific clinical tests, so these should be considered for all CUPs (Losa et al. 2018).

Important principles to consider in selecting the appropriate ancillary testing to perform on tissue include the following:

• Tissue biopsy is usually better than FNA because it provides more tissue for ancillary testing, including IHC.
• It is important to identify highly treatable non-epithelial malignancies such as lymphoma, melanoma and germ cell tumour.
• The need to optimally classify CUP with extensive ancillary testing that may exhaust tissue needs to be balanced against the limited tissue available; therefore, some testing may need to be triaged.

Examples where ancillary molecular testing may be appropriate include in patients where there is a strong clinical suspicion of a specific tumour type, where standard molecular profiling to look for actionable genomic abnormalities for that tumour type should occur. For example, patients with TTF-1 adenocarcinoma thought most likely to be lung cancer (even in the absence of an identifiable lung primary) should have testing for epidermal growth factor receptor mutations and, if this is negative, anaplastic lymphoma kinase re-arrangements.

The optimal role of molecular profiling tests to determine the tissue of origin is the subject of ongoing research. Existing guidelines do not recommend the use of gene-expression-based profiling to identify primary tumours. However, where this is available via CUP research projects, it may be helpful.

The use of gene panel DNA sequencing to identify actionable mutations is still under investigation. It appears that, in a minority of patients, potentially actionable mutations may be found (Gatalica et al. 2014, Ross et al. 2015, Tothill et al. 2013, Varghese et al. 2017). In addition, other findings such as mutational signatures indicating tobacco or UV exposure may assist in diagnosing the primary site (Tothill et al. 2013, Varghese et al. 2017). The use of immunotherapy may be effective in some patients, especially those with high tumour mutation burden (Gröschel et al. 2016). Panel testing may also identify germline risk alleles. There is currently a lack of evidence as to whether patients receiving this panel testing have an outcome benefit. However, where this is available via CUP research projects, it may be helpful.

Timeframe for obtaining a diagnosis

Timeframes for diagnosis should be informed by evidence-based guidelines (where they exist) while recognising that shorter timelines for appropriate consultations and treatment can reduce patient distress.

Investigations should be completed within two weeks.