Forty years experience of treating cancer of unknown primary

Abstract

Cancer of unknown primary site (CUP) is not a rare tumour. It accounts as the seventh to eighth most frequently diagnosed cancer in a general medical oncology service. Since CUP is not a homogeneous disease and it consists of different favourable and unfavourable sub-sets, treatment of each clinicopathological entity requires a unique approach. The spectrum of therapeutic management includes both locoregional and systemic therapy and should intend to offer optimal benefit to favourable CUP patients and palliative care to unfavourable cases. This review article provides both a historical outline of CUP treatment as well as a helpful therapeutic guide to every oncologist who treats CUP patients.

Despite the recent technological advances, the diagnosis and treatment of patients with carcinoma of unknown primary (CUP) remains a real dilemma in daily practice. CUP is characterized by early dissemination, clinical absence of the primary tumour at diagnosis, unpredictable metastatic pattern and an aggressive behaviour. Unpredictable metastatic pattern is part of the different natural history of CUP patients from those patients with known primary tumours. Specifically, it refers to the differences in the incidence of splanchnic and non-splanchnic metastatic sites [1].

More than 50% of patients present with multiple sites of involvement, whereas the primary tumour can be identified antemortem in less than 20% of the cases. The most frequently detected primary sites are those of lung and pancreas [2].

CUP consists of a heterogenous group of malignancies with distinct clinicopathological entities. Certain clinicopathological CUP entities are considered as favourable sub-sets responding to platinum-based chemotherapy and exhibiting prolongation of survival [1–3].

In this review paper we refer mainly to the therapeutic management of patients with CUP during the last 40 years (1964–2006), by focusing separately on the treatment of each clinicopathological entity.

General aspects

Epidemiology

Cancer of Unknown Primary (CUP) accounts for 3–5% of all human malignancies. It represents one of the ten most frequent cancers and the fourth commonest cause of cancer death. CUP consists of a heterogenous group of malignant tumours that share distinct clinicopathological sub-sets. The annual age-adjusted incidence per 100 000 population in USA is 7–12 cases and in Europe (Netherlands) is 5.3–6.7 cases, while the median age is around 60 years being more common in males [1], [4–7].

Definition

An updated clinical definition of CUP should refer to patients who present with histologically-confirmed metastatic cancer in whom medical history, physical examination, full blood count, basic biochemistry battery, urinalysis, stool occult blood testing, immunohistochemistry with specific markers as well as imaging technology with chest x-ray, computed tomography of the chest abdomen and pelvis or mammography and MRI in certain cases, fail to detect the primary tumour [1], [8], [9].

Pathology

CUP is classified into four major histopathological sub-types: (i) adenocarcinomas well to moderately differentiated (50%), (ii) undifferentiated or poorly differentiated adenocarcinoma (30%), (iii) squamous cell carcinomas (15%) and (iv) undifferentiated neoplasms (5%) including poorly differentiated carcinomas, neuroendocrine tumours, lymphomas, germ cell tumours, melanomas, sarcomas and embryonal malignancies [1], [5].

Biology

Chromosomal abnormalities

Aneuploidy is seen in 70% of patients as well as, abnormalities of the short arm of chromosome 1(1p) and of chromosome 12 were reported. The presence of isochromosome i (12p) or a deletion in 12p, characterises a germ cell origin tumour and reflect patients presenting with midline nodal metastases [10–12].

Oncogenes, tumour suppressor genes and oncoproteins

Overexpression by immunohistochemistry of several oncogenes, tumour suppressor genes and proteins has been found. C-myc overexpression has been observed in 23% of patients, Ras in 23%, HER 2 in 4–24%, EGFR in 4–72%, C-Kit in 4–13%, Bcl-2 in 40%, and p53 in 48–53%. Whether this expression profiling may lead to better understanding of molecular pathophysiology of CUP or may help identifying patients suitable for targeted therapies, is still warranted. Search for gene mutations of various exons showed negative results [13–19].

Angiogenesis and proteolysis

VEGF, thrombospondin-1 and CD 34 (microvessel density) immunohistochemical expression has been correlated with adverse prognostic significance for survival [20], [21].

Matrix metalloproteinase −2 (MMP-2), MMP-9 and tissue inhibitor of metalloproteinase-1 (TIMP-1) are overexpressed in CUP patients, whereas TIMP-1 is associated with poor prognosis [22].

Diagnostic investigation of the primary site

Diagnostic pathology

Appropriate immunohistochemical investigation is mandatory in order to search for possible tumour primaries or to rule out potentially treatable cancers. Several monoclonal or polyclonal immunoperoxidase antibodies can identify enzymes, structural tissue components, hormonal receptors, hormones, oncofetal antigens or can be directed against various cytokeratin polypeptides [5], [23].

Electron microscopy, although not widely available, can be useful in distinguishing certain clinical situations, i.e. neuroendocrine tumours, melanomas or poorly differentiated sarcomas [5].

Molecular diagnostics could contribute in detecting tumour-specific chromosomal abnormalities such as the isochromosome of the short arm of chromosome 12 i (12p) in midline CUP nodal disease, or the translocation t (11,22) [q 24; q12] in PNET/Ewing's sarcoma or t(8;14) [q24; q32] in non-Hodgkin's lymphomas [24–26].

Imaging diagnosis and endoscopies

Computed tomography of the abdomen and pelvis can result in primary tumour detection in 30–50% of patients [27], [28]. Mammography or magnetic resonance imaging (MRI) of the breast are recommended only in female patients presented with isolated metastatic adenocarcinoma of the axillary nodes [29], [30]. Also, FDG-PET scanning may provide valuable information particularly in patients with squamous cancer metastatic to cervical lymph nodes [31].

In addition, it is important to point out that endoscopies are a useful diagnostic tool only in the presence of specific symptoms and signs [32].

Serum tumour markers

Routine evaluation of commonly used epithelial serum tumour markers (CEA, CA 15–9, CA 19–9, CA 125) has no proven prognostic or diagnostic value. Specific serum tumour markers should be requested only in certain CUP subsets i.e. CA 15–3 in patients with isolated axillary metastases, or CA 125 in peritoneal papillary adenocarcinomatosis, β-HCG or AFP in patients with midline nodal disease or PSA in men with bone osteoblastic metastatic deposits [33–35].

Clinicopathological entities

CUP is considered as a heterogenous disease. A number of clinicopathological entities has been recognized which are divided into the favourable and unfavourable subsets based on response to treatment and outcome (Table I). Unfortunately, patients diagnosed with unfavourable sub-sets are more common and carrying most of the times a dismal prognosis with short survival.

Table I.  Favourable and unfavourable clinicopatholical entities of CUP.

Favourable sub-sets		Unfavourable sub-sets
1.	Women with papillary adenocarcinoma of the peritoneal cavity	1.	Adenocarcinoma metastatic to the liver or other organs.
2.	Poorly differentiated carcinoma with midline distribution (extragonal germ cell syndrome).	2.	Non-papillary malignant ascites (adenocarcinoma).
3.	Poorly differentiated neuroendocrine carcinomas.	3.	Multiple cerebral metastases (adeno or squamous carcinoma).
4.	Women with adenocarcinoma involving only axillary lymph nodes.	4.	Multiple lung/pleural metastases (adenocarcinoma).
5.	Squamous cell carcinoma involving cervical lymph nodes.	5.	Multiple metastatic bone disease (adenocarcinoma).
6.	Men with blastic bone metastases and elevated PSA (adenocarcinoma)
7.	Isolated inguinal adenopathy (squamous carcinoma).
8.	Patients with a single, small, potentially resectable tumour.

Therapeutic management of CUP

Historical overview of systemic treatment

First-line chemotherapy

In 1960's and 1970's chemotherapy of CUP patients was based mainly on fluorouracil or adriamycin-containing regimens. Other drugs in use were cyclophosphamide, mitomycin-C and vinca alkaloids. Response rates were around 20% with a very few complete responses, while median survival ranged between 3 to 7 months (Table II) [36–47].

Table II.  Fluorouracil/Anthracycline-based chemotherapy in CUP patients.

Regimen	Number of Patients	Response rate (%)	Median Survival (months)	Reference
FU vs. various	103	6.2 vs. 2.6	9	Johnson et al., 1964 [36]
FU vs. combinations vs. various	213	16 vs. 9.5 vs. 8.5	3.5	Moertel et al., 1972 [37]
CAF	100	21	7^+	Valentine et al., 1979 [38]
FAM	28	21.4	7.6	Mckeen et al., 1980 [39]
CAV	20	50	8	Anderson et al., 1983 [40]
AV	38	13.2	7	Fiore et al., 1985 [41]
FAM	43	30.2	11	Goldberg et al., 1986 [42]
FU, DTIC, VCR, BCNU	61	20	3.6	Alberts et al., 1989 [43]
FAM	29	10.3	3.5	Al-Idrissi, 1990 [44]
AVM	57	30	–	Kambhu et al., 1990 [45]
PALA, MTX, FU, FA	21	5	–	Kelsen et al., 1992 [46]
Etoposide	25	8	–	van der Gaast et al., 1993 [47]

FU: fluorouracil, CAF: cyclophosphamide, adriamycin, fluorouracil, FAM: fluorouracil, adriamycin. mitomycin, CAV:cyclophosphamide, adriamycin, vincristine, AV:adriamycin, vindesine, VCR: vincristine, AVM: adriamycin, vincristine, mitomycin.

The use of cisplatin and carboplatin-based chemotherapy started in late 1980's. Combined cytostatics in use were etoposide, anthracyclines, vinca alkaloids, cyclophosphamide, bleomycin and more recently other novel drugs such as gemcitabine, vinorelbine or irinotecan. During this period several chemosensitive sub-groups to platinum-based chemotherapy have been identified i.e. papillary peritoneal adenocarcinomatosis, poorly differentiated carcinomas of midline nodal distribution or neuroendocrine tumours. Overall response rates were increased up to 40–50% and some more months were added to median survival. However, definite longer survivals have been documented for the above recognized chemosensitive sub-sets (Table III) [48–73].

Table III.  Platinum-based chemotherapy in CUP patients.

Regimen	Number of Patients	Response rate (%)	Median Survival (months)	Reference
FACP	23	17	6	Jadeja et al., 1983 [48]
PVeB	56	57	16	Greco et al., 1986 [49]
FAPH	85	21	6	Becouarn et al., 1989 [50]
PEB	34	79	8^+	van der Gaast et al., 1990 [51]
FEP	36	22	11	Raber et al., 1991 [52]
PFL	25	32	–	Lenzi et al., 1991 [53]
PE	16	19	8	Gill et al., 1991 [54]
PVB±Doxo	68	56	–	Hainsworth et al., 1991 [55]
P	21	19	5	Wagener et al., 1991 [56]
P-containing vs. Cb-containing	48	29	4.3	Pavlidis et al., 1992 [57]
PF	15	53	–	Khansur et al., 1995[58]
P-based	92	37.2	6.4	Farrugia et al., 1996 [59]
CbFL	40	25	7.8	Rigg et al., 1997 [60]
CbEpE	62	37	10	Briasoulis et al., 1998 [61]
PMiEp	40	50	9.4	Falkson et al., 1998 [62]
CbE	26	23	5.6	Warner et al., 1998 [63]
PF	44	27	–	Lofts et al., 1999 [64]
ECF	34	19	8.25	Parnis et al., 2000 [65]
PE	23	32	8	Voog et al., 2000 [66]
PAC	22	50	10.7	Guardiola et al., 2001 [67]
ECF	36	22	9	Karapetis et al., 2001 [68]
CA alternating with PE/GCSF	82	39	10	Culine et al., 2002 [69]
PMiF	31	27	7.7	Macdonald et al., 2002 [70]
PEG	30	36.6	7.2	Balana et al., 2003 [71]
CbEA	102	26.5	9	Piga et al., 2004 [72]
CbG	51	30.5	8.5	Pittman et al., 2006 [73]

A: adriamycin, B: bleomycin, C:cyclophosphomide, Cb: carboplatin, E: etoposide, Ep: epirubicin, F: fluorouracil, G: gemcitabine, H: hexamethymelamine, L:leucovorin, Mi: mitomycin-C, I:ifosfamide, I_f:interferon-α, I_R: irinotecan, P:cisplatin, Ve: vinblastine.

After the year 2000, taxanes have been integrated mainly to platinums, but both response rates and survival did not reveal any significant difference. Paclitaxel was the most commonly taxane used, accounting for 67% of the combinations with carboplatin, cisplatin, etoposide or gemcitabine (Table IV) [74–85].

Table IV.  Taxane/Platinum-based chemotherapy in CUP patients.

Regimen	Number of Patients	Response Rate (%)	Median Survival (months)	Reference
PLCbE	55	47	13.4	Haisnworth et al., 1997 [74]
PLCb	77	38.7	13	Briasoulis et al., 2000 [75]
PLCbE	71	48	11	Greco et al., 2000 [76]
PDX	22	33	8	Bouleuc et al., 2001 [77]
DX	29	7	6	Darby et al., 2001 [78]
PLPG	29	50	–	Gothelf et al., 2002 [79]
PLCbG	120	25	9	Greco et al., 2002 [80]
PLCbE g GIR	132	30	9.1	Greco et al., 2004 [81]
PLP	37	42	11	Park et al., 2004 [82]
DXG	35	40	10	Pouessel et al., 2004 [83]
PLCb	22	23	6.5	El-Rayes et al., 2005 [84]
PLCbE	78*	53	14.5	Hainsworth et al., 2006 [85]

P_L: paclitaxel, Cb: carboplatin, E:etoposide, P:cisplatin, G: gemcitabine, I_R:irinotecan, D_x: docetaxel.

*neuroendocrine CUP.

Small prospective phase II randomized studies are also available in the literature. Initially, non-platinum combinations have been compared while in the most recent ones comparisons were done between platinum and/or taxane combinations. Unfortunately, these randomized studies were unable to show any considerable difference in response rates or survival (Table V) [86–96].

Table V.  Prospective randomized-studies in CUP patients.

Regimen	Number of Patients	Response Rate (%)	Median Survival (months)	Reference
MiA vs. CMF	47	36 vs. 4.5	4.2 vs. 3	Woods et al., 1980 [86]
F vs.FAC	36	0 vs. 0	3.5 vs. 3.0	Shild et al., 1983 [87]
MA vs. CVB	95	42 vs. 32	4.1 vs. 5.8	Miliken et al., 1987 [88]
MA vs. MAP	55	7.1 vs. 18.5	5.5 vs. 4.6	Eagan et al., 1987 [89]
EP/AC vs. high dose	60	42 vs. 39	11 vs. 8	Culine et al., 1999 [90]
PDX vs. CbDX	73	26 vs. 22	8 vs. 8	Greco et al., 2000 [91]
CbE vs. PLFL	34	19 vs. 19	8.3 vs. 6.4	Dowell et al., 2001 [92]
PG vs. PIR	80	42 vs. 25	22% vs. 23% (1y-survival)	Lortholary et al., 2002 [93]
F vs. F + Mi	88	11.6 vs. 20	6.6 vs. 4.7	Assersohn et al., 2003 [94]
PG vs. PIR	80	55 vs. 38	8 vs. 6	Culine et al., 2003 [95]
PLCb vs. GVL	90	21.5 vs. 21.5	10.7 vs. 6.9	Huebner et al., 2005 [96]

F: fluorouracil, A:adriamycin, C: cyclophosphamide, Mi: mitomycin, M:methotrexate, Cb: carboplatin, E:etoposide:, L: leukovorin, V:vincristine, B: bleomycin, I_f:interferon, G: gemcitabine, I_R: irinotecan, D_X: docetaxel, P_L: paclitaxel, V_L: vinorelbine.

Second-line chemotherapy

From 2001 to 2005 four phase II studies including almost 120 patients who failed first-line treatment have been published. Combinations of fluorouracil, leucovorin, gemcitabine, irinotecan or docetaxel were applied. Most studies revealed no encouraging results, although the combination of docetaxel and gemcitabine in 15 patients showed a response rate of 28% and a median survival time from the start of second-line treatment of 8 months (Table VI) [97–100].

Table VI.  Second line chemotherapy in CUP patients.

Regimen	Number of Patients	Response Rate (%)	Median Survival (months)	Reference
G	39	8	–	Hainsworth et al., 2001 [97]
FL	25	0	9^*	Culine et al., 2001 [98]
DXG	15	28	8**	Pouessel et al., 2003 [99]
GIR	40	10	4.5	Hainsworth et al., 2005 [100]

G: gemcitabine, I_R: irinotecan, F: fluorouracil, L: leucovorin, D_x: docetaxel.

*from diagnosis **from start of 2^nd-line treatment.

Targeted treatment

Two anecdotal cases with CUP have been treated one with thalidomide and steroids and another one with the combination of trastuzumab and vinorelbine. Both patients showed good partial response lasted for 17 and 7 months, respectively [101], [102]. In a phase II study 51 patients were treated with bevacizumab and erlotinib as second line treatment. Four patients (8%) achieved partial response and 30 patients (59%) stable disease with a median overall survival of 8–9 months [103].

Treating the favourable CUP subsets

Women with papillary adenocarcinoma of peritoneal cavity

This sub-set of CUP patients should be managed as FIGO stage III ovarian cancer. A combination of optimal surgical cytoreduction followed by platinum-based chemotherapy is mandatory. Relatively high tumour responses including complete responses, as well as progression-free, overall and long-term survival benefit, have been observed. A retrospective analysis of 47 patients treated with platinum-based or platinum taxane-based chemotherapy, showed an objective tumour regression of 53% with 36% complete responders, as well as a median progression-free and overall survival of 10 and 15 months, respectively. The 5-year survival was 10% [104], [105]. No firm data on intraperitoneal chemotherapy are available.

Poorly differentiated carcinoma with nodal distribution

This entity resembles extragonadal germ-cell syndrome. It affects mainly men less than 50 year old and is usually localized in the midline nodal chain (supraclavicular, mediastinal or paraortic lymph nodes), but occasionally can be manifested with concomitant lung parenchymal metastases. Serum β-HCG or AFP levels can be increased in less that 20% of the cases. It is considered as a chemosensitive tumour to platinum-based chemotherapy and should be treated as poor prognosis germ cell malignancies. High responses up to 50% with 15–25% complete remissions and 10–15% long-term disease-free survivors have been reported [51], [106].

Poorly differentiated neuroendocrine carcinoma

Since 1988 this sub-set has been recognized as a highly responsive tumour to cisplatin-based chemotherapy. In this initial report, response rate was 72% (24% complete response) and 15% long-term disease-free survivors. In a very recent publication by the same research group using the combination of paclitaxel, carboplatin and etoposide, 53% of patients had major responses (15% complete remissions) with a median survival of 14.5 months and a 24% 3-year survival [85], [107], [108]

Women with isolated adenocarcinoma in the axillary lymph nodes

Despite the lack of randomized studies the management of these women should be similar to stage II or III breast cancer. Locoregional therapy with or without systemic treatment is recommended. Patients with N1 disease (mobile nodes) should undergo axillary clearance followed by either a simple mastectomy or breast radiation therapy. Premenopausal patients with positive oestrogen receptors should be treated on an adjuvant setting with chemotherapy and then hormonotherapy alone. Women carrying N2 disease (fixed nodes) neoadjuvant chemotherapy should be preceded followed by appropriate guidelines for stage III breast cancer. However, in non-responding patients or in elderly women, radical radiotherapy should be offered followed by endocrine treatment in oestrogen receptors-positive cases. Long-term survival is achievable. The 5- and 10-year overall survival rates are 75% and 60%, respectively [109–111].

Squamous cell carcinoma involving cervical lymph nodes

These patients should be treated as locally advanced head and neck cancer with locoregional management. The estimated 5-year survival is 35 to 50%, while some long-term survivors have been documented. Only patients with p N1 neck disease without extracapsular extension can be treated with surgery alone. In all other cases surgery alone is still inferior.

Radiotherapy should be provided to both sides of the neck and the mucosa in the entire pharyngeal axis and larynx. On the contrary, radiotherapy to the ispilateral cervical nodes alone does not prevent locoregional recurrences [112], [113].

Although the impact of systemic chemotherapy in these sub-sets of patients remains uncertain, it can be recommended particularly in patients with N₂ or N₃ nodal disease.

Males with blastic bone metastatic lesions from an adenocarcinoma with elevated serum PSA

This a rare sub-set of the favourable CUP group. It refers to male patients that should be considered as suffering from metastatic prostate cancer and should be managed with endocrine treatment [1], [5].

Isolated inguinal lymphadenopathy from squamous cell carcinoma

In these rare cases nodal dissection with or without local irradiation is the treatment of choice. Long-term disease-free survivors have also been reported. No chemotherapy data are available [1].

CUP patients with a single small metastasis

Many of these rare patients enjoy palliative benefit or even long disease-free survival with only local resection with or without local radiotherapy [1].

Treating the unfavourable CUP sub-sets

The majority of CUP patients belong to the unfavourable group. Most of these patients have adenocarcinoma and are presented with multisplachnic metastases in the liver, lungs, brain or bones.

In three large studies with almost 400 CUP patients with hepatic metastases, the response rate to various chemotherapies (mostly platinum-based) was less than 20% and median survival less than 7 months [114–116]. In general, chemotherapy offers poor results although several platinum or taxane/platinum regimens have reported to produce better responses as compared to old combinations. In addition, they claim longer median or 1- and 2-year survivals. It is important to notice though, that the median survival of patients enrolled in clinical trials and those of unselected CUP population is almost 3-fold higher (6 to 10 months versus 2–3 months). A number of selection factors i.e. median age, performance status or inclusion of some favourable sub-sets might explain this difference [1], [5].

In conclusion, unfavourable CUP patients of relatively young age and good performance status could offer a chance of platinum-based chemotherapy. Alternatively, best supportive care should be recommended.

Prognostic variables in CUP

Several studies were able to identify independent prognostic factors that are associated with survival in patients with CUP.

Among the various clinicopathological, serological and biological factors or other comorbid conditions the following variables were found to inversely correlate with survival: male sex, poor performance status, high number of metastatic sites, unfavourable histopathologic sub-sets, presence of liver metastases, elevated serum alkaline phosphatase, elevated serum lactate dehydrogenase, low serum albulin levels and lymphopenia [117–121].

Useful guideline web-sites

CUP guideline recommendations are available on ESMO, NCI/PDQ and StART electronic sites:

1. ESMO Clinical Recommendations (available in full text/PDF/Palm OS format) http://www.esmo.org/reference/reference.guidelines.htm

2. NCI/PDQ: http//www.cancer.gov/cancerinfo/pdg/treatment/unknownprimary/healthprofessional/

3. START: http://www.startoncology.net