Cancer of unknown primary (CUP) is a heterogeneous syndrome of metastatic cancers in which the primary site cannot be determined after a standard and comprehensive search, and accounts for 1–5% of all newly diagnosed malignancies (1,2).

CUP falls into either the clinicopathologically ‘favorable’ (~15% of cases) or ‘unfavorable’ (~85% of cases) subgroups (3–5). The former includes adenocarcinoma of the axillary lymph nodes, papillary serous carcinoma of the peritoneum (redefined as peritoneal cancer), squamous cell carcinoma of the cervical lymph nodes and extragonadal germ cell tumors in young men, which are derived from the middle of the body (4). For these patients, specific therapeutic management can be provided, and satisfactory local control and prolonged survival can be expected, with a median survival time of >24 months (6). On the other hand, the unfavorable subset (majority of patients) includes patients with adenocarcinoma metastasis to the liver or other organs, poorly differentiated carcinoma and squamous cell carcinoma of the abdominal cavity (1). These patients are usually treated with empiric chemotherapies based on platinum and taxanes (7) but are generally not chemosensitive and have a poor prognosis, with a median survival time of 5–8 months (4,8).

Delayed initiation of treatment and a tendency to choose empiric chemotherapy are associated with a poor prognosis (4). Therefore, a thorough workup at the time of diagnosis is important in CUP. The distribution and histology of the cancer are important for estimating the primary site. In addition to image-based and immunohistochemical examinations, the cancer genomic profiling (CGP) test has been used to estimate primary sites and identify potential targets for personalized therapies (9,10). Several studies using next-generation sequencing have demonstrated that molecular profiling for CUP is useful for predicting the tissue of origin (9,11–14). However, to the best of our knowledge, it has not been determined whether genomic profiling in CUP can help in providing tissue-specific therapies, including targeted therapies, to improve survival. A prospective randomized phase II trial (CUPISCO trial; NCT03498521) is currently underway to investigate the efficacy of CGP in CUPs; however, there are issues, such as insufficient sample volume and cases of misdiagnosis (15,16).

In the present study, a case of CUP is reported in which genomic profiling and lesion distribution were used to infer the primary organs of cancer and to determine a suitable therapeutic strategy.

A 68-year-old woman, gravida 1 para 1, who had a history of allergic dermatitis for 2 years and had been taking prednisone irregularly, presented to their primary care physician (Yokohama, Japan) in August 2020 with dyspnea on exertion. Right pleural effusion was observed, and 1,600 ml pleural fluid was aspirated by puncture. Papanicolaou staining was performed on the pleural fluid as previously described (17), and the cytological examination revealed Class V adenocarcinoma (Fig. 1A) according to the Papanicolaou classification (18). Upper and lower gastrointestinal endoscopy and gynecological examination were performed but the primary tumor was not identified. In October 2020, the patient was referred to the Department of Respiratory Medicine, Graduate School of Medicine, The University of Tokyo (Tokyo, Japan). By then, the pleural effusion had resolved spontaneously. Positron emission tomography-computed tomography (PET-CT) revealed mild fluorodeoxyglucose accumulation in the left predominant para-aortic and left internal iliac lymph nodes; however, this was considered to be a reactive change. At 11 months after the initial visit, the Krebs von den Lungen-6 level measured using chemiluminescent enzyme immunoassay (CLEIA) was elevated (1,243 U/ml; normal range, 0–464 U/ml), and the patient was referred back to the hospital for a close examination of the primary site. PET-CT revealed abnormal accumulation in the right axillary lymph node with a maximum standardized uptake value (SUV) of 3.8, and further accumulation in the peripancreatic, para-aortic, mesenteric, sacral and bilateral external iliac lymph nodes with a maximum SUV of 7.2. All lymph node accumulations were considered to indicate metastatic lymph nodes (Fig. 1B).

Due to elevated cancer antigen 125 (CA125) levels (309 U/ml; normal range, 0–35 U/ml) measured using CLEIA and the distribution of the metastatic lymph nodes, gynecological cancer was suspected; however, pelvic magnetic resonance imaging showed no obvious primary tumor and multiple uterine fibroids (Fig. 1C). The cervical and endometrial cytologies were negative. Fine-needle aspiration cytology of the right axillary lymph node revealed metastatic adenocarcinoma (data not shown). For histological diagnosis, a laparoscopic para-aortic lymph node biopsy was performed. Hematoxylin-and-eosin (H&E)-stained slides were made from formalin-fixed paraffin-embedded (FFPE) blocks as previously described (17). Immunohistochemical staining was performed using the Ventana BenchMark XT automated staining system (Roche Diagnostics) according to the manufacturer's protocol, as previously described (19). The primary antibodies used were as follows: Cytokeratin 7 (cat. no. 790-4462; clone SP52; Roche Diagnostics), estrogen receptor (EP1, Envision FLEX-ER, Dako; Agilent Technologies, Inc.), paired box 8 (clone 10336-1-AP; Proteintech Group, Inc.), p53 (clone DO-7; Roche Diagnostics), Wilm's tumor 1 (clone 6F-H2; Dako; Agilent Technologies, Inc.), GATA binding protein 3 (clone L50-823; Biocare Medical, LLC), cytokeratin 20 (clone SP33; Roche Diagnostics) and D2-40 (clone D2-40; Dako; Agilent Technologies, Inc.). All the cytopathological, histological and immunohistochemical images were examined and captured using a light microscope (BX51; Olympus Corporation). The tumor was diffusely positive for cytokeratin 7, estrogen receptor, paired box 8, p53 and Wilms tumor 1, whereas it was negative for GATA binding protein 3 and cytokeratin 20 (Fig. 2A). The histological and immunohistochemical diagnosis was of a high-grade serous carcinoma (HGSC) possibly derived from the gynecological organs, based on the WHO Classification (20).

Based on the distribution of the tumors, high CA125 levels, histology and immunostaining results, HGSC of the ovary or fallopian tubes was suspected to be the primary carcinoma. For the purpose of diagnosis and tumor debulking, a total abdominal hysterectomy, bilateral salpingo-oophorectomy, pelvic and para-aortic lymphadenectomy, and partial omentectomy were performed (Fig. 2B). Pathological examination of the surgical specimen, including histological assessment of H&E-stained slides of FFPE tissues, revealed that there were no primary tumors in the parenchyma of the ovaries or fallopian tubes, and also no peritoneal dissemination, with negative peritoneal washing cytology, or serous tubal intraepithelial carcinoma; however, there were multiple metastatic adenocarcinomas in para-aortic and pelvic lymph nodes. Although some lymph node metastases were present on both sides, including the para-aortic, sacral and external iliac lymph node metastases, some were only on the left, including the common iliac, internal iliac and obturator lymph node metastases. Combining preoperative imaging findings and pathological diagnosis, the distribution of multiple lymph node metastases was recorded as predominantly left-sided (Fig. 2C). Although no apparent primary site was identified, lymphatic invasion of HGSC was microscopically detected (Fig. 2D) within the left ovarian parenchyma, and given the distribution of the lymph node metastases, it was suggested that the tumor was left ovary-derived (Fig. 2D). Reevaluation of the pleural fluid cytology image at this point revealed a high nucleus-to-cytoplasmic ratio, which was consistent with an HGSC-like appearance (21). As no primary tumor had been identified, the tumor was diagnosed as a CUP. A CGP test was conducted to diagnose CUP. FoundationOne® CDx (Foundation Medicine, Inc.) is a qualitative next-generation sequencing-based in vitro diagnostic test performed by Foundation Medicine, Inc. (22) that showed 14 somatic variants (Table I), including three likely pathogenic variants [TP53 (p.G266R), CIC (p.E1263Gfs*78) and PBRM1 (p.I223Yfs*36)], and nine gene amplifications of CCND2, CSF3R, FGF23, FGF6, KDM5A, MYC, PIK3C2G, RAD52 and RICTOR (Table II). The tumor was microsatellite-stable with a tumor mutational burden (TMB) of 10.1/Mb (high) and a loss of heterozygosity (LOH) score of 23%. Immune checkpoint inhibitors (ICIs) were recommended based on the tumor being TMB-high (TMB-H) and poly ADP-ribose polymerase (PARP) inhibitors were recommended based on the high LOH score. The tumor distribution, pathological and immunohistochemical examinations, and genomic examinations suggested gynecological organ-derived HGSC (left ovary-derived being suspected). Myriad MyChoice CDx® (Myriad Genetics, Inc.; protocol details not available), a next-generation sequencing-based in vitro diagnostic test performed by Myriad Genetics, Inc., revealed that the tumor was homologous recombination-proficient (HRP) with a homologous recombination-deficiency (HRD) score of 41 and no tumor BRCA1/2 mutation.

The tumor had a high TMB, and ICIs should therefore have been considered; however, the HRD score was 41 (relatively high among HRP tumors) and the LOH score was also high, close to that of the HRD tumor (HRD score ≥42). In addition, HGSC generally has a high response rate to platinum agents and PARP inhibitors are effective only while tumors are platinum-sensitive (23–25). Therefore, it was decided to start a combination therapy of paclitaxel and carboplatin (TC therapy; paclitaxel 175 mg/m² and carboplatin AUC 6 mg/ml/min every 3–4 weeks) and use niraparib (200 mg/body) as maintenance therapy. After eight courses of TC therapy, the CA125 level was markedly reduced from 247 U/ml to 16 U/ml (Fig. 3), and the patient achieved complete remission. Based on the HRD score of 41, the patient was switched to niraparib maintenance therapy according to the treatment protocol for ovarian cancer (25). The patient has been followed up every month since August 2022, when niraparib was started, and will continue to be followed up at the same intervals. The patient has been relapse-free as of July 2023.

In the present case, the diagnosis was delayed by the spontaneous resolution of the cancerous pleural effusion. However, CUP was classified as cancer of ovarian origin based on the marked lymphatic invasion around the left ovary, the distribution of the lymph node metastases, the genomic information and the immunohistochemical data, and the patient was able to receive specific treatment for primary ovarian cancer.

The biological nature of CUP remains largely unknown. The most popular hypothesis is that it is a metastatic tumor arising from an undetectable primary tumor due to regression, dormancy and small size (26,27). Another hypothesis is that it is a single metastatic tumor without a primary tumor (3). The present case is unique in that there was no primary lesion in the uterus and adnexa, accompanied by only lymphatic invasion within the left ovary. These findings suggest that the primary tumor in the left ovary or fallopian tube might either be small or regressed spontaneously.

An important observation in this case was the disappearance of the malignant pleural effusion. To the best of our knowledge, the phenomenon of spontaneous resolution of malignant pleural effusions has not yet been reported, and its precise mechanism remains unresolved. A similar phenomenon, in which a pleural effusion appears with ovarian malignancy and disappears with the disappearance of the ovarian tumor, is known as pseudo-Meigs' syndrome (28,29). In the present case, if the malignant pleural effusion had disappeared with the spontaneous resolution of the primary ovarian tumor, it would be consistent with the pathogenesis of pseudo-Meigs' syndrome.

The reason for the disappearance of the primary tumor was subsequently considered. The patient was being administered systemic exogenous corticosteroids around the time the pleural effusion appeared. It is possible that the systemic administration of steroids suppressed cancer immunity (30). Furthermore, the tumor was TMB-H. HGSC is characterized by copy number alterations with low TMB (31–33). TMB-H tumors are characterized by high levels of neoantigens and immunogenicity (34). One hypothesis is that the patient developed highly immunogenic ovarian cancer, but that cancer immunity was suppressed during steroid administration, resulting in pseudo-Meigs' syndrome associated with ovarian cancer. Subsequent reactivation of immunity by steroid withdrawal may have triggered shrinkage of the primary tumor and disappearance of the pleural effusion.

In the present case, genomic data suggested that PARP inhibitors could be expected to be effective as maintenance therapy after TC therapy. In addition, the patient had a TMB-H tumor and was expected to benefit from ICIs. Although the patient was treated with TC therapy followed by maintenance therapy with a PARP inhibitor based on the treatment regimen for primary ovarian cancer, the efficacy of immunotherapy in CUP has been demonstrated in a phase II trial (35) and is expected to become more widespread in the future. Approximately one-third of patients with CUP have a tumor proportion score >1% for programmed cell death 1-ligand 1, and antitumor immunity-related gene expression in CUP have been reported to be comparable to those in ICI-responsive malignancies (36,37). The pleural effusion appeared at the time of steroid administration and the tumor was a TMB-H tumor, which is rare for ovarian HGSC (31,32). This may suggest that the pathophysiology of CUP in the present case report is highly immunogenic.

The present study had some limitations. First, the hypothesis of pseudo-Meigs' syndrome associated with ovarian cancer was proposed as a pathogenesis of CUP; however, it is not possible to prove this hypothesis since there was no evidence of a tumor in the left ovary. Second, based on the high LOH score, a PARP inhibitor was used after TC therapy; however, it was not possible to evaluate the response to the PARP inhibitor, as the TC therapy resulted in complete remission. Long-term observations are needed in the future to investigate the effects of PARP inhibitors and ICIs.

In the present case, lymphatic invasion, the distribution of the lymph node metastases, genomic analysis and immunohistological analysis suggested CUP of left ovarian origin, and specific therapy for ovarian cancer was provided. This unique course, which was characterized by the appearance and disappearance of the CUP, may have been associated with the immune response.