What Cancer of Unknown Primary Means in Practice
This diagnosis describes situations where metastatic disease is confirmed but standard investigation cannot identify the organ the tumour arose from. It is a diagnosis of exclusion, made only after an adequate work-up.
Its frequency has fallen over time. With better imaging, immunohistochemistry and molecular methods, a group that made up 3 to 5 per cent of all cancers in the early 1990s is now 1 to 2 per cent, some sources reporting under 1 per cent.
A consequence is that the diagnosis today describes a more selected group than it once did. Outcome figures in older publications come from an era of more limited investigation and do not carry across to a patient seen now.
Important: That fall carries a clear message: before saying the primary cannot be found, the work-up has to have been genuinely completed. An incomplete one creates uncertainty that need not exist.
Why the Primary Site Cannot Be Found
Three explanations are offered. The first is that the primary stayed too small to be seen on imaging, having shed cells early without growing. The second is that it regressed spontaneously. The third is that the metastasis has lost its organ-specific features and carries no markers pointing to an origin.
How often each applies is not known. What matters is that failing to find the source does not always mean the work-up was incomplete. In a proportion of patients the origin stays uncertain after adequate investigation, and the plan is built with that uncertainty.
This also explains why a wholly negative panel is not a dead end. It does not show that no origin exists, only that the markers looked for are not expressed here, and the search moves to broader methods.
How the Diagnosis Is Made
The process starts with a detailed examination of the biopsy material. The pathologist applies a series of stains to establish which tissue type the tumour came from, showing whether it carries markers specific to the lung, breast, bowel or prostate.
- The biopsy material is assessed for adequacy.
- The histological type is determined by light microscopy.
- Immunohistochemical stains are applied stepwise.
- Targeted imaging follows the organ the stains point to.
- Complementary tests follow: breast imaging in women, PSA in men.
- Where needed, PET-CT scans the whole body.
- If the picture is still unclear, molecular testing is considered.
Immunohistochemistry is the workhorse here. A first panel separates the broad families, carcinoma, lymphoma, melanoma and sarcoma, and a narrower second points towards an organ. Markers are read in combination, because none is wholly specific. An adequate biopsy matters: twelve stains cannot be run on a few cells, and the route then is set out in when the tissue biopsy is not enough.
PET-CT has been measured. In one review the primary site was identified in 22 per cent of patients, additional metastases in 14 per cent, and management changed in 35 per cent. In the Willemse analysis of individual data from 1,865 patients the pooled rate was 54 per cent, varying with the pattern of metastasis; the gap reflects patient selection and reference standards.
Send Your Reports for Assessment
Share your pathology, imaging and blood results on WhatsApp, or send them by email. Documents are reviewed before the consultation.
Send Your Reports for Assessment
Share your pathology, imaging and blood results on WhatsApp, or by email. Documents are reviewed before the consultation. Which markers are tested for in lung cancer is covered in a separate article.
The Favourable Subsets
This is not a single disease. Roughly a fifth of patients fall into subsets treatable as a specific cancer type, with a better outlook.
- Squamous carcinoma confined to neck nodes: treated as head and neck cancer.
- Adenocarcinoma in axillary nodes: treated as breast cancer.
- Serous adenocarcinoma with peritoneal spread: treated as ovarian cancer.
- Poorly differentiated midline carcinoma: treated as germ cell tumour.
- Blastic bone metastases with raised PSA: treated as prostate cancer.
- Single-site disease: local treatment becomes an option.
- Neuroendocrine tumour: has its own algorithm.
Each is recognised because a pattern of spread reliably points back to one organ. A woman with disease confined to axillary nodes is treated for breast cancer even with no lesion visible in the breast. The logic is not that the origin is proved but that treating it as known gives the best expected result.
Recognising these subsets is the most valuable step in the diagnosis. In the remaining group, described as unfavourable, median survival has been reported at around 10 to 12 months. Identifying a subset changes the treatment and what can be expected, so the first question is whether the picture fits a defined subset.
Why the Source of a Metastasis Cannot Be Found
Why the organ a tumour arose from sometimes cannot be identified even when metastatic disease is confirmed, and what is done then.
Tissue-of-Origin Testing
Tests using gene expression profiles predict which organ a tumour came from. They propose an origin in most cases; the question is whether that improves treatment.
In a Japanese randomised trial, site-specific treatment guided by gene expression profiling was compared with standard chemotherapy. One-year survival was 44.0 per cent against 54.9 per cent, not a significant difference, and median survival 9.8 against 12.5 months. Predicting the origin did not improve treatment.
In the Chinese Fudan CUP-001 trial randomising 182 patients the result went the other way: with site-specific treatment guided by a 90-gene assay, median progression-free survival was 9.6 months against 6.6 months. The difference in overall survival was not significant.
What the test measures should be understood correctly. It compares the tumour’s gene expression pattern with known organ patterns and reports the closest match. The result is a probability estimate, not a direct demonstration as pathology is, so where it conflicts with the stains and imaging, clinical findings take precedence.
Important: Because two randomised trials reached different conclusions, tissue-of-origin testing is not recommended for everyone. It is considered in selected patients and through a board decision.
Genomic Profiling and Targeted Treatment
In recent years the approach has shifted from finding the origin to finding a targetable alteration. In the Ross study of 200 specimens, at least one genomic alteration was found in 96 per cent of tumours, and one matchable to an approved drug or trial in 85 per cent.
The trial that best shows what that means is CUPISCO. Of 1,505 patients screened, 636 entered and 436 who achieved disease control on platinum-based chemotherapy were randomised. Median progression-free survival was 6.1 months with molecularly guided therapy against 4.4 months with continued chemotherapy. Overall survival showed no significant difference and the data were immature.
This shows the distance between a finding and a treatment. A report listing an alteration changes nothing by itself: a drug has to exist for it, the patient has to be well enough to take it, and it has to be obtainable. That 436 of 1,505 screened could be randomised gives the real-world proportion.
Looking at the Target Rather Than the Origin
How the genetic alterations a tumour carries can guide treatment selection when the organ it arose from cannot be found.
Points to Watch When Reading These Trials
In the large trial of molecularly guided treatment, randomisation required disease control on platinum-based chemotherapy. The comparison was therefore made in a selected group, and patients who deteriorated at first assessment are not inside the result. The gap between the numbers screened and randomised shows that selection.
In the same trial the difference in progression-free survival did not carry through to overall survival, and the data were immature. That caveat stands: a progression-free survival gain does not by itself mean a gain in length of life.
For tissue-of-origin testing, two randomised trials reached different conclusions, and in one, treatment guided by the predicted origin was not shown to be superior to standard chemotherapy. The right position is then not to recommend it routinely and to decide at board level. The immunotherapy data come from a small single-arm study, and its response rate carries no comparison.
Where Immunotherapy Fits
In the single-arm NivoCUP study of 56 patients, nivolumab produced a response rate of 22.2 per cent among the 45 treated before, with median overall survival 15.9 months. Responses were more evident in tumours with higher PD-L1 expression, higher tumour mutational burden and microsatellite instability.
This also sums up how the approach has changed: even when origin cannot be established, markers describing the tumour’s relationship with the immune system can open a treatment option. Microsatellite instability and tumour mutational burden are therefore routinely assessed.
Assessing these markers opens a decision route independent of origin. Microsatellite instability or a high mutational burden allows a treatment option to be discussed even when the organ is unknown.
The study was small and single-arm; even so, it showed immunotherapy can be an option here, the decision resting on the biomarkers.
How the Process Runs
The sentence most often heard with this diagnosis is that nobody can find where it started. Yet the aim is not only to find the origin but to gather the information that will direct treatment.
- The pathology report states whether the material suffices for molecular testing.
- It also states which organ groups the stains covered.
- PET-CT comes in when the stains do not point to one organ group.
- Breast imaging in women and PSA in men are standard.
- Genomic profiling looks for an alteration that could direct treatment.
- Board discussion is the rule here, not the exception.
- Whether a suitable trial exists is considered with the plan.
- A copy of the reports and slides stays with the patient.
Two things cost the most time here: an inadequate sample, and investigations not planned in parallel. Taking enough tissue at the first biopsy, and setting some aside for molecular testing, removes the need for a second procedure weeks later.
Board review matters particularly here, because the decision rests on more than one specialty’s interpretation. That process is described in how a tumour board works.
Frequently Asked Questions
What does the term cancer of unknown primary mean?
Cancer of unknown primary describes confirmed metastatic disease where standard investigation cannot identify the organ the tumour arose from. It is a diagnosis of exclusion, made after an adequate work-up.
How common is it?
The frequency has fallen with better imaging and pathology. It made up 3 to 5 per cent of all cancers in the early 1990s and is now 1 to 2 per cent, some sources reporting under 1 per cent.
Does the source have to be found?
Not always. A plan can be built on the biological features and targetable alterations of the tumour independently of its origin.
Should I have a tissue-of-origin test?
A tissue-of-origin test is considered in selected patients. In a Japanese randomised trial, treatment guided by predicted origin was not superior to standard chemotherapy; in a Chinese trial it lengthened progression-free survival.
Does genomic testing help?
At least one genomic alteration is found in the great majority of these tumours. In CUPISCO, molecularly guided therapy lengthened progression-free survival, with no significant difference in overall survival.
Is immunotherapy an option?
In the single-arm NivoCUP study, nivolumab produced a response rate of 22.2 per cent. Responses tracked with markers such as PD-L1 and microsatellite instability, and the decision rests on those markers.
Is the outlook the same for everyone?
No. Roughly a fifth of patients fall into subsets treatable as a specific cancer type, with a better outlook.
Related Articles
Related Pages
Comprehensive Genomic Profiling
Open the Page
Molecular Tumour Board
Open the Page
Treatment Pathway
Open the Page
Immunotherapy
Open the PageSources
- Krämer A, Bochtler T, Pauli C et al. Cancer of unknown primary: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol 2023;34(3):228-246. Link
- National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology: Occult Primary. Link
- Rassy E, Pavlidis N. Progress in refining the clinical management of cancer of unknown primary in the molecular era. Nat Rev Clin Oncol 2020;17(9):541-554. Link
- National Cancer Institute. Carcinoma of Unknown Primary Treatment (PDQ) — Health Professional Version. Link
- Hainsworth JD, Rubin MS, Spigel DR et al. Molecular gene expression profiling to predict the tissue of origin and direct site-specific therapy in patients with carcinoma of unknown primary site. J Clin Oncol 2013;31(2):217-223. Link
- Hayashi H, Kurata T, Takiguchi Y et al. Randomized phase II trial comparing site-specific treatment based on gene expression profiling with carboplatin and paclitaxel for patients with cancer of unknown primary site. J Clin Oncol 2019;37(7):570-579. Link
- Liu X, Zhang X, Jiang S et al. Site-specific therapy guided by a 90-gene expression assay versus empirical chemotherapy in patients with cancer of unknown primary (Fudan CUP-001). Lancet Oncol 2024;25(8):1092-1102. Link
- Krämer A, Bochtler T, Pauli C et al. Molecularly guided therapy versus chemotherapy after disease control in unfavourable cancer of unknown primary (CUPISCO). Lancet 2024;404(10452):527-539. Link
- Ross JS, Wang K, Gay L et al. Comprehensive genomic profiling of carcinoma of unknown primary site: new routes to targeted therapies. JAMA Oncol 2015;1(1):40-49. Link
- Tanizaki J, Yonemori K, Akiyoshi K et al. Open-label phase II study of the efficacy of nivolumab for cancer of unknown primary (NivoCUP). Ann Oncol 2022;33(2):216-226. Link
- Willemse JRJ, Lambregts DMJ, Balduzzi S et al. Identifying the primary tumour in patients with cancer of unknown primary using FDG PET/CT: a systematic review and individual patient data meta-analysis. Eur J Nucl Med Mol Imaging 2025;52(1):225-236. Link