Cancer of unknown primary—state of the art

The cancer of unknown primary (CUP) syndrome does not represent a single clinical entity; instead, it encompasses a diverse group of malignant tumors. CUP is defined as a histologically confirmed metastatic malignant disease for which the primary tumor cannot be identified after completing initial diagnostics [1]. CUP syndrome is relatively common in clinical practice, with 3–5% of malignant tumors presenting as CUP [2]. Interestingly, recent literature suggests a decreasing incidence of CUPs (1–2% of malignant tumors) [3]. This decline may be attributed to advancements in molecular diagnostics, imaging, and the evolving definition of CUPs, which no longer includes mesenchymal tumors. Men and women are affected equally, with a median age peak at 65 years [4]. The clinical course of CUP syndrome is heterogeneous, marked by a poor prognosis, unpredictable metastatic patterns, and intrinsic resistance to therapy [3, 4]. The median overall survival (OS) for CUP patients in the subgroup with a poor prognosis is 3 months. OS for patients with CUPs of squamous differentiation is better than for those with undifferentiated carcinomas or adenocarcinomas. CUPs are categorized into four subgroups based on their histology: well or moderately differentiated adenocarcinomas (50%), poorly differentiated adenocarcinomas (30%), squamous cell carcinomas (15%), undifferentiated carcinomas (5%) [1]. It should be noted that, according to recent European Society of Medical Oncology (ESMO) guidelines, neuroendocrine carcinomas, sarcomas, germ cell tumors, melanomas, and hematological neoplasms are by definition no longer considered CUPs, even if no primary tumor is found [1].

The pathogenesis of CUPs remains elusive. Rassy et al. suggested several possible causes for CUP carcinogenesis [4]:

Metastatic spread occurs, when motile cells migrate prior to uncontrolled local proliferation of non-motile neoplastic cells at the primary site. Alternatively, migratory cells can disseminate from a primary tumor that initially formed but has subsequently been selectively eliminated by microenvironmental factors, thereby favoring the outgrowth of tumor cells at the metastatic site [4].

The primary goal of diagnosing CUPs is to rapidly identify the patient population that can benefit from specific therapy, potentially curative treatment (e.g., solitary inguinal or cervical lymph node metastases of squamous cell carcinoma) and has a favorable prognosis (20%).

In daily clinical practice, CUP patients often undergo a protracted diagnostic work-up in an attempt to locate the origin of the primary tumor. However, given the poor prognosis for most CUP patients (80%), it is advisable to start therapy promptly.

In addition to obtaining a medical history that includes an oncological history and questions about spontaneously regressed or resected lesions, a physical examination is recommended. Routine laboratory testing (including Lactate dehydrogenase (LDH) measurement) should also be performed. In men (especially with adenocarcinoma histology), Prostatic specific antigen (PSA) levels should be determined. For mediastinal tumors, beta human chorionic gonadotropin (β‑HCG) and alpha-fetoprotein (AFP) levels are relevant. Cancer antigen (CA) 15‑3 measurement may be useful in women with axillary or supraclavicular lymph nodes and adenocarcinoma histology, while CA-125 measurement should be performed in women with adenocarcinomas in the mediastinum, peritoneum, inguinal region, or a malignant pleural effusion. Carcinoembryonic antigen (CEA), CA 19‑9, and CA 72‑4 are relevant in cases where a gastrointestinal origin is suspected. Chromogranin testing should be considered in patients where a neuroendocrine tumor is suspected after pathological work-up. It should be noted that the determination of various tumor markers in serum has no independent diagnostic, prognostic, or predictive value and elevation (especially CA-125, CA 15‑3, CEA, and CA 19-9) can be nonspecific. However, these tumor markers are suitable for therapy monitoring.

Imaging typically includes a computed tomography scan of the neck, chest, abdomen, and pelvis. The role of Positron Emission Tomography and Computed Tomography (PET-CT) is a matter of debate among guidelines. While European Society for Medical Oncology (ESMO) and National Comprehensive Cancer Network (NCCN) guidelines recommend PET-CT mainly for patients with cervical CUPs and oligometastatic CUPs, international Swiss/Austrian/German Onkopedia guidelines recommend performing PET-CT early in the work-up [1, 5, 6]. A meta-analysis showed a detection rate of the primary tumor of 34% employing PET-CT (sensitivity and specificity both 84%) [7]. However, large prospective studies are lacking, so the role of PET-CT in the initial work-up has not been definitively established.

For female patients, mammography plus ultrasound is recommended, especially when axillary or supraclavicular lymph nodes with adenocarcinoma histology are present. A gynecological examination should be performed as well.

Depending on the results of these basic diagnostics, additional work-up employing endoscopies (gastroscopy, colonoscopy, bronchoscopy) may be indicated.

Table 1 provides an overview of the recommended clinical basic work-up according to ESMO guidelines.

An essential step for adequate work-up is histopathological diagnosis based on the immunohistochemical (IHC) marker profile, which plays a central role in CUP diagnosis. It is worth noting the ESMO and NCCN guidelines at this point, which recommend a detailed work-up based on IHC profiles. In addition, performing mismatch repair testing and, when available, tumor mutational burden testing appears meaningful.

Numerous assays that compare gene expression profiles, DNA methylation patterns, and mRNA expression of CUPs to tumors of known origin have been evaluated in clinical studies. The accuracy of these “CUP classifiers” ranges from 54 to 98% [4]. However, this initially promising result must be viewed critically. These assays are trained using a database of molecular profiles from tumors of known origin. Due to the pathogenesis of CUPs as described above, the validity of applying these assays to predict the tissue of origin (TOO) is unclear.

In other words, can one truly infer the origin of a tumor of unknown origin based on the molecular profile of a tumor of known origin? A retrospective postmortem analysis of CUP patients revealed discrepancies between the spectrum of primary tumors detected during autopsy and the molecular prediction of the TOO by the “CUP classifiers” [8].

A more clinically relevant question is whether molecular testing to determine the putative TOO can improve the patient’s prognosis. Surprisingly, the superiority of TOO-directed therapy compared to empirical chemotherapy has not been demonstrated to date. Results from the randomized GEFCAPI 04 phase III study, presented at the ESMO 2019 Congress, found that progression-free survival in the group receiving empirical chemotherapy with cisplatin/gemcitabine was comparable to the group receiving therapy based on molecular prediction of the TOO (HR = 0.95 [0.72–1.25]; p = 0.7) [9]. However, it should be noted that the specific therapies in this study were established in 2011, and thus, recent developments in oncology may not have been accounted for, which could have contributed to the negative outcome.

Nonetheless, for the reasons mentioned, there is no recommendation in the guidelines for the routine clinical application of these assays.

Next-generation sequencing (NGS) has identified clinically relevant mutations in 52 to 85% of CUP patients [2, 10]. Up to 30% of CUP patients harbor mutations such as BRAF V600E, ERB2, or FGFR, for which approved targeted therapies are available [2]. A study in 389 CUP patients demonstrated that 28% of these CUPs exhibit alterations such as microsatellite instability (1.8%), PD-L1 expression (22%), or high tumor mutational burden (11.8%) [11]. These biomarkers suggest that a subset of CUP patients could benefit from immune therapy using checkpoint blockade. The ESMO Translational Research and Precision Medicine Working Group introduced the ESMO Scale for Clinical Actionability of Molecular Targets (ESCAT), which provides guidance for classifying genetic alterations in tumors according to their relevance for targeted therapy [12].

However, a caveat should also be noted when using NGS in CUP patients. The sensitivity to targeted therapy also depends on the TOO (e.g., BRAF V600E inhibitors in different tumor entities), while, on the other hand, certain substances (such as NTRK inhibitors) have been approved for tumor-agnostic use. Similarly, Programmed death-ligand 1 (PD-L1) expression is not always associated with a response to immunotherapy. Currently, there is no evidence for the superiority of targeted therapy or immunotherapy over empirical chemotherapy in prospective CUP studies. However, the prospective randomized phase II CUPISCO study (NCT03498521), which is investigating the efficacy of NGS-based targeted and immunotherapy compared to empirical platinum-based chemotherapy, is addressing this data gap.

The treatment of CUP patients is guided by the results of the work-up described above. In addition, clinical knowledge of the metastatic pattern of tumors considered in the differential diagnosis and radiological features is crucial for classification (e.g., TTF-1-negative bronchial carcinoma vs. CUP or intrahepatic cholangiocellular carcinoma vs. CUP). When either clinical or molecular work-up suggests the origin of a primary tumor with a high probability, specific therapy according to the current guidelines of the respective tumor entity appears reasonable.

This subgroup is characterized by a favorable prognosis and includes:

For localized solitary metastases, as mentioned earlier, local therapy (resection or radiation plus/minus chemotherapy) should be considered.

For patients with unidentifiable CUPs, therapy selection should be based on performance status (PS) and LDH levels. Patients with an Eastern Cooperative Oncology Group (ECOG) PS ≤ 1 and normal LDH have a better prognosis (median overall survival of 12 months) compared to patients with elevated LDH and/or PS ≥ 2 (median overall survival of 4 months).

Platinum-based empirical chemotherapy (plus gemcitabine or taxanes) remains the first-line therapy for the majority of CUP patients, if no clinical trial is available. Doublet chemotherapy should be considered primarily in fit patients with normal LDH, while in all other cases, monotherapy or “best supportive care” is recommended.

In a meta-analysis that included 543 CUP patients with an unfavorable prognosis and compared 16 chemotherapy protocols, no advantage of a specific regimen was detected. It should be noted, however, that the confidence intervals were wide [13].

Evidence for the use of immunotherapy in CUP patients is still limited. In a phase II study 45 pretreated CUP patients were exposed to nivolumab. An overall survival of 15.9 months (95% CI 8.4–21.5 months) and a response rate of 22.9% were observed [14]. However, due to the heterogeneous and limited patient population, recommendations based on this study are limited.

Immunotherapy should be considered in CUP patients in the unfavorable prognostic group with mismatch repair deficiency/MSI‑H tumors in the second-line setting. First-line immunotherapy appears to be reasonable for the treatment of mismatch repair deficient/MSI‑H “colon-like” CUPs. Immunotherapy in the second-line setting is also an option for CUP patients in the unfavorable prognostic group with high PD-L1 expression or high tumor mutational burden (“TMB-high”).

Targeted therapy is recommended for NTRK fusion, ALK fusion, ROS‑1 fusion-positive, or EGFR-mutated CUPs. BRAF V600E-mutated CUPs should be treated with targeted therapy in the second-line setting (first-line in “NSCLC-like CUP”). After first-line therapy failure, targeted therapy based on the molecular profile (preferably discussed in a molecular tumor board) can also be considered.

CUP syndrome encompasses a heterogeneous group of tumors, with the majority having a poor prognosis. Identifying subgroups with a favorable risk profile that can benefit from specific therapy is crucial. Prolonged diagnostic investigations should be avoided to prevent treatment delays. The role of targeted therapy and immunotherapy for the treatment of CUP patients is still under investigation. Across all guidelines, there is a broad consensus strongly recommending the inclusion of CUP patients in clinical trials.