Predictors of Survival of Patients with Cancer of Unknown Primary Site: A Retrospective Study from Two Institutions in Egypt

Background: Identification of prognostic factors in patients with cancer of unknown primary (CUP) is important to optimize their management. Aim: To study the clinicopathological characteristics of patients with CUP and to identify factors that influence their survival. Methods: A retrospective review of the medical records of 102 patients who presented with CUP in two Egyptian cancer care facilities during six years from 2012 to 2017 inclusive. Results: The median age of patients was 61 years (range : 40-96) and 63% were males. Well-/moderately-differentiated adenocarcinoma was the most common histopathological diagnosis (60%) followed by poorly-differentiated carcinoma (25%). The common sites of metastases were the liver (56%), lymph nodes (56%), lungs (44%), and bones (38%). The initial treatment plan was single modality treatment in 43% of patients, combined modality in 16%, and best supportive care in 41%. The 6-month time-to-progression (TTP) and overall survival (OS) rates were 52.7% and 56.1%, respectively. Eastern Cooperative Oncology Group (ECOG) performance status >1, bone metastasis, low serum albumin, elevated serum alkaline phosphatase, and single agent chemotherapy treatment (compared to combination chemotherapy) were associated with significantly shorter TTP. Age ≥65 years, ECOG performance status >1, comorbidities, >1 metastatic site, bone metastasis, low serum albumin, elevated serum alkaline phosphatase, best supportive care / single modality treatment plan and single agent chemotherapy treatment (compared to combination chemotherapy) were associated with significantly shorter OS. Conclusions: Many factors may affect the prognosis of CUP patients, e.g., old age, poor performance status, and low serum albumin. Further studies including a larger sample size are needed to develop predictive models based on these factors in patients with CUP.


Introduction
Cancer of unknown primary (CUP) site represents a challenging diagnostic dilemma for oncologists, in which the site of origin of metastatic tumors remains obscure, even after comprehensive investigations 1,2 . Cancer of unknown primary constitutes approximately 3%-5% of all cancers 1 . The biology of these tumors is unclear; however, current evidence suggests that metastatic dissemination can occur without primary tumor development due to cancer cells' inherent metastatic aggressiveness 3 . Chromosome instability has been proposed as part of the unusual clinical presentation and poor outcomes of CUP patients 4 . The biological mechanisms behind this unusual clinical behavior are unknown and no identifiable molecular markers have been connected to these malignancies 5 . As a result, individuals with CUP are diagnosed only after performing specific clinical and histopathological investigations 4 , including histologic examination and immunohistochemistry (IHC) staining 6 .
The treatment of CUP patients begins with identifying favorable subgroups (20%) of individuals with particular clinical and/or pathologic manifestations 6,7 . The current guidelines detailed that favorable subgroups include female gender, young patients, isolated adenopathy, welldifferentiated adenocarcinoma and squamous cell carcinoma, single metastatic site, and good performance status 8 . These individuals react rather well to specific treatments, and some have possibly curable malignancies 9 . Unfavorable subsets (80 %) of CUP patients typically receive platinum-based chemotherapy, an empiric chemotherapy regimen designed to effectively treat a wide range of cancer types. However, response and survival of these subsets of patients are generally poor 10 .
When a likely primary tumor is detected, and the proper therapy is performed, the prognosis usually improves 11 . Ten to forty percent of CUP patients have metastasis in their lymph nodes, whereas the rest of the patients have metastases in their internal organs 12 . Even though the underlying tumor is frequently undetected, several clinicopathologic characteristics of CUP indicate groups of individuals with a better prognosis 13 . The prognosis is particularly good in CUP restricted to lymph nodes and with histology other than adenocarcinoma. On the contrary, liver metastasis and several organs, including the brain, lung/pleura, and bone, indicate a poor prognosis 6 . No immediate critical function compromise accounts for the good prognosis in individuals with lymph node metastases. However, survival rates differ depending on whether lymph nodes are involved 14 .
This study aimed to identify the prognostic factors that influence the survival of Egyptian patients with CUP.

Study design, setting, and participants
This was a retrospective study, which was performed to collect data from the medical records of all patients presenting with CUP from January 2012 to December 2017 at two Egyptian cancer care centers, the Clinical Oncology Departments of Ain Shams and Helwan Universities in Cairo. Only patients aged ≥18 years with an established CUP diagnosis, as suggested by pathological, radiological, and IHC examination were included. We excluded patients who were hospitalized with life-threatening comorbidities, patients with brain metastasis, and patients whose records did not include survival data. Initially, records of 150 CUP patients were retrieved; the primary sites were identified in 48 (32%) patients and were excluded, while the remaining 102 patients fulfilled the inclusion criteria and were recruited retrospectively.

Data collection and study's outcomes
The following data were collected: age, gender, Eastern Cooperative Oncology Group (ECOG) performance status, comorbidity, pathological findings, tumor grade, IHC findings, number of affected organs, sites, type of obtained biopsy, diagnostic tools performed as the type of performed endoscopy, type of performed imaging, laboratory findings including serum tumor markers, treatment modalities planned. Response to treatment (RTT) was assessed according to the revised RECIST 1.1 15 .
The primary outcome of the present study was to explore the frequencies and distribution of clinical characteristics and to define the correlations and dependencies.
The secondary outcomes were to investigate the factors significantly affecting time-to-progression (TTP) and overall survival (OS) among patients with CUP.

Reporting guidelines
We followed the STROBE statement recommendations during the preparation of this report 16 .

Results
Characteristics of the included patients and a summary of investigations done in the search for primary cancer are shown in Table 1. The median age of patients was 61 years (range: 40-96) and the majority had performance status >1 (65.6%), and more than half of the patients had comorbidities. The pathological examination revealed that well-/ moderately-differentiated adenocarcinoma is the most common pathology and nearly two-thirds of the patients had grade II tumors. CK20 and CK7 were studied in all patients. Thirty-three (32.4%) patients were CK20 -ve CK7 +ve, 36 (35.3%) CK20 +ve CK7 -ve and 30 (29.4%) CK20 +ve CK7 +ve. Carcinoembryonic antigen, CA15.3, and CA125 were elevated in 22.6%, 15.8%, and 28.9% of selected patients.  The treatment modalities and response to treatment are illustrated in Table 2. Twenty-one (20.6%) patients underwent surgery, mainly in the form of excisional biopsy. On the other hand, 44 (43.1%) patients received chemotherapy alone as a single agent in 10 (22.7%) and combined regimens in 34 (77.3%) patients.
The estimated mean TTP was 5 months (95%CI: 4.3-5.6) and the median was not reached. The 6month TTP rate for the whole group of patients was 52.7%. Eastern Cooperative Oncology Group performance status >1, bone metastasis, low serum albumin, elevated serum alkaline phosphatase as well as the administration of single agent chemotherapy were associated with significantly shorter TTP (Table 3).
The estimated mean OS was 8.6 months (95%CI: 7.7-9.6) and the median was 8 (95%CI: 6-10). The 6month OS rate for the whole group was 56.1%. Age ≥65 years, ECOG performance status >1, comorbidities, bone metastasis, low serum albumin,  Therefore, the application of well-designed models is recommended to avoid false indications. According to the Egyptian study published by El-Shebiney and Maria on 84 patients with CUP, there are many prognostic factors for CUP, including performance status, histopathological subtypes, liver metastasis, lung metastasis, brain metastasis, albumin level, and the number of metastasis locations. However, the authors developed a simple model based on the performance status and the number of metastasis locations. The utilization of these two factors was based on multivariate analysis. The model classified the patients into two groups: poor-risk and good-risk. They found a significant difference between both groups in terms of one-year survival (p<0.0001) 23 . Poor PS was also found to be an unfavorable prognostic factor in the studies of Culine et al. 24 and Seve et al. 25 . Several studies have found that when it comes to the number of organs involved by metastases, CUP patients with a single involved organ had a better survival time compared to individuals with two or more involved organs 21,26 . However, Abbruzzese et al. 12 and Grau et al. 27 found that CUP patients with three or more organs involved by the tumor did not have a worse prognosis.
In a series of 311 patients, Petrakis et al. demonstrated that the median of OS and PFS was 8 and 4 months, respectively 28 . They developed an algorithm that predicts the OS of CUP patients up to 36 months, using three parameters; performance status, white blood cell count, and the clinicopathologic subgroup. If the patient has a tumor within serous peritoneal or axillary nodal, they classified him as low risk with a median OS of 36 months, without any further investigations. However, if the patient has a visceral subtype with elevated white blood cell count (>10,000/mm 3 ) and worse performance status, he will be classified as high-risk, and the median of OS will be five months. We believe that this algorithm needs a larger sample to be validated.

Conclusions
In conclusion, the current study confirms the poor prognosis of CUP. Older age, poorer performance status (>1), >1 affected organ, presence of comorbidities, lower albumin level, elevated alkaline phosphatase level and bony metastasis are predictors of worse prognosis in CUP patients. On the other hand, the response to treatment is associated with favorable survival. Further studies with larger sample size are required to assess the role of these factors in predicting the prognosis of patients with CUP.