Japanese Journal of Clinical Oncology

INTRODUCTION

Cancer of unknown primary site (CUPS) constitutes up to 15% of cancer diagnoses at any large medical institution (1-5). Even though the primary site of origin is unknown, treatment schemes, treatment response and prognosis have been defined for various good response subgroups (6,7): poorly differentiated carcinoma and adenocarcinoma (8-15), poorly differentiated neuroendocrine carcinoma (16,17), female peritoneal carcinomatosis (18,19), squamous cell carcinoma of the head and neck and other regions (20,21), lymph node cancer of hilar or mediastinal regions (22) and isolated axillary lymph node metastases in female patients (1,3). However, in most CUPS patients, treatment response and survival are still poor.

The point at which a metastatic cancer is allocated to the diagnostic group of CUPS is influenced by the extent of the investigations performed, although the effectiveness of exhaustive examinations to find the primary tumor and the resultant cost incurred have been questioned (5,23). More importantly, limiting examinations after exclusion of treatable cancers has been justified, since finding the primary tumor did not influence treatment response or survival (5,24), but conflicting reports on improved survival when the primary tumor was identified have also been reported (25).

Pathology review is essential in the diagnosis of CUPS. Clarification through performance of immunohistochemistry (26-28), cytogenetic studies (29,30) and evaluation of biological markers is usually attempted (31), but still fails to identify the primary tumor in a significant number of patients. Cell lines from patients with CUPS have been established to study the biology of the disease (32), but the intriguing presentation of metastatic cancer without a known primary site remains elusive (33).

In Taiwan, where the incidence of viral associated tumors such as hepatoma (HBV and HCV related) (34,35), undifferentiated nasopharyngeal carcinoma (Epstein-Barr virus related) are very high (36), it was also of interest to ascertain the spectrum of viral serology positivity in local patients diagnosed with CUPS and also to assess the spectrum of tumor marker changes in this geographical region.

For ease of comparison, empirical chemotherapy for CUPS patients can be divided into adriamycin-based, platinum-based and a miscellaneous group (37). Randomized studies have not been performed to evaluate response and survival differences between the different regimens, but very high response rates have been attained with a cisplatin (CDDP) based regimen, with prolonged survival (14,38).

Cisplatin and etoposide are chemotherapeutic drugs that have been used successfully in the treatment of patients with CUPS (38), often in combination with other agents. Cisplatin is a highly protein bound drug, with a serum half life of 58-73 h, yet its cytotoxicity resides in the non-protein-bound fraction, having a half-life of only 40-45 min (39,40), making it an optimal drug for continuous infusion, a schedule which also ameliorates the gastrointestinal toxicity (41). Etoposide is a podophyllotoxin derivative, acting through interaction with topoisomerase II. Its activity is heavily schedule dependent, since resistance may be associated with decreased steady-state drug levels, with increased efficacy associated with repeated dosing (42,43). The treatment protocol was designed to administer cisplatin and etoposide as a continuous intravenous infusion over 72 h, since continuous infusion has the potential for higher efficacy in drug-resistant tumors (44,45).

The aim of this study was twofold, first to define response rate, response duration and toxicity of a continuous infusion etoposide- and cisplatin-based regimen in the treatment of Chinese patients with CUPS and second to evaluate the distribution of histological diagnoses, tumor marker characteristics and serology for endemic viral infection in Taiwan.

PATIENTS AND METHODS

Between March 1993 and February 1996, 20 consecutive chemotherapy-naive patients with measurable cancer of unknown primary site were enrolled for study. All had histological diagnoses, reviewed by a consultant pathologist after initial diagnosis and confirmed to be carcinoma. Appropriate immunohistochemical studies were performed (all were cytokeratin positive and leucocyte common antigen negative) and classification and differentiation ascertained. Written informed consent was obtained prior to treatment.

Pretreatment evaluation of all patients included a complete blood count, serum chemistry, tumor marker studies ([alpha]-fetoprotein, [beta]-human chorionic gonadotropin, carcinoembryonic antigen, CA125, CA153 and CA199 for all patients and prostate specific antigen for male patients only), viral serology studies, a chest X-ray and sonar of the abdomen. Further image studies were dictated by the site of disease. All patients received an upper gastrointestinal tract endoscopy and colonoscopy to exclude esophageal, gastric and colorectal neoplasia, ear, nose and throat consultation and check-up and gynecological consultation for female patients. Patients had to have normal cell counts, liver and renal function tests.

The general principle upheld for investigations in this study aimed at excluding all potentially curable cancers, such as lymphoma, germ cell tumor or cancers for which specific treatment is necessary, such as breast cancer.

The treatment regimen consisted of etoposide (80 mg/m²) and cisplatin (25 mg/m²) continuous intravenous infusion on days 1-3; chemotherapy was repeated every 21 days if tolerable, but could be delayed to 28 days for blood cell recovery. The etoposide dose would be decreased by 25 and 50% in case of grade 3 and 4 neutropenia and thrombocytopenia, respectively. The cisplatin dose was not modified in this study.

Toxicity was graded according to the common toxicity criteria. Objective response was defined as at least a 50% decrease in the sum of all two-dimensional products of measurable lesions, lasting at least 4 weeks; stable disease was defined as a [le]50 to >25% increase in the lesion sizes; disease progression was defined as a [ge]25% increase in the sum of the products of all measurable lesions or appearance of any new lesion, regardless of size. Survival was counted from the day the patient received the first dose of chemotherapy until demise.

Results

Characteristics of the patient population are given in Table 1. The mean age of tumor occurrence was 63.3 years. There was a predominance of male patients, 15 male to five female patients, because the cancer center is located in a Veterans' hospital. Over half the patients had more than three metastatic sites. The only patient with lung-only metastases had multiple nodules. Many patients had extensive disease at diagnosis, with poor nutritional and performance status.

Tumor marker studies showed that more than 50% of the patients tested had raised CA125, CA199 and even CA153. One patient had an [alpha]-fetoprotein of 817 U/ml, multiple bone, lung and liver metastases and positive serology for hepatitis B surface antigen and histology was poorly differentiated adenocarcinoma (not hepatoma); the patient progressed rapidly after initiation of chemotherapy and died within 3 months. [beta]-Human chorionic gonadotropin was slightly raised (<50 mIU/ml) in two male patients, who also showed no response to chemotherapy. All male patients tested negative for prostate-specific antigen.

Of 13 patients with liver metastases, 10 had a raised lactate dehydrogenase (LDH), 11 had a raised alkaline phosphatase and only two had a raised [gamma]-glutamyl transferase level. Of 10 patients with bony metastases, nine had a raised ALP.

Table 1. Clinical and pathological characteristics of 20 patients with cancer of unknown primary site (CUPS)

	No. of patients	%	Response rate(%)
Mean age (yrs)	63.3
Range (yrs)	41-83
Performance status
ECOG 0,1	3	15	33.3(1/3)
ECOG 2	9	45	11(1/9)
ECOG 3	8	40	37.5(3/8)
Male/female	15/5
Number of metastatic sites
1	1	5	100(1/1)
2	8	40	37.5(3/8)
[ge]3	11	55	9(1/11)
Metastatic sites
Lymph nodes(cervical, axillary, inguinal)	13	65	15.4(2/13)
Lung	8	40	25(2/8)
Pleura/peritoneum	3	15	0
Liver	13	65	30.8(4/13)
Bone	10	50	30(3/10)
Other sites(skin, thyroid, soft tissue, adrenal)	6	30	0
Elevated tumor markers
[alpha]-Fetoprotein (>8.5 ng/ml)	1/20	5
[beta]-Human chorionic gonadotropin (>10 mIU/ml)	2/9	22
Lactate dehydrogenase (95-213 U/l)	15/20	75
Alkaline phosphatase (10-100 U/l)	15/20	75
Carcinoembryonic antigen (>6 ng/ml)	8/20	40
CA153 (>25 U/ml)	4/7	57
CA-199 (>35 U/ml)	10/17	59
CA125 (>35 U/ml)	10/18	55.5
Prostate specific antigen (>3.32 ng/ml)	0/15	0
Viral serology studies*
HBsAg	4/20	20
Ab hepatitis C virus	1/20	5
EBV early antigen IgA (>1/160)	0/8	0
EBV viral capsid antigen IgA (>1/160)	2/8	10
Pathology
Poorly differentiated adenocarcinoma/carcinoma	13/20	65	31(4/13)
Moderately and well differentiated adenocarcinoma[dagger]	4/20	20	25(1/4)
Miscellaneous (1 mucoepidermoid, 1 poorly differentiated    squamous, 1 poorly differentiated large cell[dagger]	3/20	15	0

*No patient had more than one positive titer in viral serology markers. ^[dagger]One patient had liver, para-aortic lymph node, peritoneal involvement, suspect cholangiocarcinoma. ^[Dagger]The patient had multiple lung and cervical lymph node metastases.

Baseline viral serology studies can be seen in Table 1. In Taiwan, 15-19% of the population are hepatitis B carriers (34), and 20% of our patients tested positive for HBsAg; 1-5% of the population have been infected with hepatitis C (35), 5% of our patients tested positive for antibody to hepatitis C virus. Of eight patients tested for Epstein-Barr virus (EBV) serology, two tested positive, EBV viral capsid antigen IgA was >1/160 and both patients had cervical lymph node involvement, but also disease below the clavicle. Histology was poorly differentiated cancer in one patient and poorly differentiated adenocarcinoma in the other.

Table 2. Toxicity profile

	Grade of toxicity*
	Grade 0	Grade 1	Grade 2	Grade 3	Grade 4
Leucopenia	20	5	20	35	20
Thrombocytopenia	60	0	0	20	20
Infection	70	0	0	25	5
Alopecia	15	30	55	0	0
Diarrhea	75	10	10	5	0
Stomatitis	75	10	10	5	0

*Grading according to common toxicity criteria; values are percentage of patients.

The patients received an average of 2.2 courses (range 1-6 courses) of chemotherapy; all patients received their first course at full dose. Of the 10 patients with progressive disease, seven received only one course and three received two courses of treatment, after which disease progression precluded further treatment. Thirteen patients went on to the second course of treatment, seven at 25% dose reduction, three at 50% dose reduction and only three received the second course at full dose. The overall response rate was only 25% (5/20), 31% in the patients with poorly differentiated carcinoma/adenocarcinoma and 25% in patients with moderate to well differentiated adenocarcinoma; there were no complete responses, 25% (5/20) had stable disease and 50% (10/20) had progressive disease. Survival was poor; median survival was 4 months (range 1.2-12.2 months), only slightly improved (4.9 months) in patients with poorly differentiated carcinoma/adenocarcinoma or in patients responding to chemotherapy (4.8 months, range 3-7.4 months). One patient died 9 days after initiation of chemotherapy from uncontrollable hypercalcemia, attributed to rapidly progressive disease; at the time of demise, his white cell count was 4 × 10³ cells/ml and platelet count 1 × 10⁵ cells/ml, with no signs of infection.

Toxicity data are given in Table 2; 55% (11/20) of patients experienced grade 3 or 4 neutropenia; 30% (6/20) had associated fevers, of whom only one was culture positive for Morganella morgagni, but there were no toxic deaths; 40% (8/20) also experienced grade 3 or 4 thrombocytopenia. Other less common toxicities included one episode each of grade 3 diarrhea and grade 3 mucositis. In spite of antiemetic use, grade 3-4 nausea and vomiting were noted in 35% (7/20) of patients.

Discussion

CUPS is difficult to diagnose. There is always the potential for detecting the primary tumor with more extensive and advanced investigations and techniques. The art lies in knowing when to terminate investigations and start treatment. In this study, critical pathology review excluded all potentially curable cancers or cancers that may have responded favorably to specific chemotherapy or hormone therapy and emphasis was placed on excluding tumors which are especially common in this geographical region, such as hepatoma, cervical cancer, nasopharyngeal cancer and esophageal and gastric cancers.

In an area where endemic viral-associated tumors constitute the most common tumors locally, it was of interest to ascertain viral serology markers in local patients with CUPS. In this study, 35% of patients had positive viral serology for HBsAg, antibody to HCV or raised IgA titer against EBV, with no overlap in viral markers; the prevalence of these markers is not remarkably different from that in the local general population. Further prospective epidemiological studies are necessary to evaluate the significance of positive viral serology in an area with a high incidence of virus-associated tumors.

The patients in this study had poor performance status (40% ECOG 3) and extensive disease (involvement of three or more sites in 55% of cases). The median age was 63.3 years, comparable to that in other studies. The spectrum of pathological diagnoses after critical review shows that 65% of patients had poorly differentiated cancers and only 20% were moderate/well differentiated adenocarcinomas.

Tumor marker studies showed elevated CEA, CA125, CA199 and CA153 in about half the patients and raised lactate dehydrogenase and alkaline phosphatase in 75% of patients; all were non-specific. No correlation between prognosis and level of alkaline phosphatase was found.

Etoposide and cisplatin in the combination and schedule administered in this study incurred moderate toxicity in this group of patients and the study was terminated early. Treatment response was 25% overall, with a trend of decreasing response with increasing number of metastatic sites; median survival was 4 months and negligibly better in patients with poorly differentiated cancers. Delay in seeking medical advice may have resulted in a time frame shift and thus shortened survival time after initiation of therapy. However, it should be noted from Table 1 that the group of patients with performance status 3 had the best response to chemotherapy compared with patients with performance status 0-1; most of these patients had poorly differentiated tumors, so that a poor performance status is not an absolute contraindication to cisplatin-based chemotherapy.

The etoposide and cisplatin dosages used in this study are between 60 and 75% of bolus dosages in previously reported studies (14,38), and 64% of the etoposide dose in an otherwise identical regimen of concurrent 72 h infusion of EP for refractory breast cancer trial (42). The decreased dose intensity in this trial could have resulted in a decreased response rate and possibly influenced survival. However, the toxicity incurred in this trial was high: grade 3 and 4 neutropenia occurred in 55%, grade 3 and 4 thrombocytopenia in 38% and grade 3 and 4 infection in 30% of patients. The severity of the toxic effects may be attributed to poor performance status. A similar regimen giving etoposide 80 mg/m² for 3 days and cisplatin 75 mg/m² for 1 day, both as daily 1 h infusions, to Chinese patients with non-small cell lung cancer resulted in only 30% grade 3 and 4 neutropenia (46). In conclusion, CUPS carries a grave prognosis and we are still a long way from understanding the etiology, pathogenesis, treatment and even best palliation of such patients.