SEARCH

SEARCH BY CITATION

Keywords:

  • Biology;
  • cancer of unknown primary;
  • metastases;
  • tumour regression

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

Background

Cancer of unknown primary (CUP) ranks among the ten most common malignancies worldwide. Cancer of unknown primary presents as disseminated disease, has a dismal prognosis and remains a diagnosis of exclusion. The natural history and biology of the disease is poorly understood, and efforts are focused on identifying the specific ‘CUP signature’.

Materials and methods

We collected and analysed all published research in the biology of CUP from 1974 till present (Medline, Embase, ASCO and ESMO Congresses).

Results

Current scientific evidence suggests that aneuploidy and karyotype changes are frequent, while more subtle molecular aberrations, such as epidermal growth factor receptor family proteins, cKit/PDGFR are frequently overexpressed, although without prognostic significance. Loss of function of tumour suppressor genes, active angiogenesis, a hypoxic genetic programme and a mesenchymal transitory phenotype have been reported in CUP and may be indicative of unfavourable prognosis. Molecular pathway analyses have identified various biomolecules impacting on survival (pAKT, pMAPK, c-Met, p21 and pPRS6). Finally, circulating tumour cells have recently been reported as a frequent phenomenon in CUP.

Conclusions

Overall, advances in understanding CUP biology have been weak and the application of gene expression profiling failed to identify an as yet elusive ‘CUP molecular signature’. MicroRNA, epigenetic and proteomic studies are warranted to better characterize the biological profile of CUP and unravel its mystery.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

Cancer of unknown primary (CUP) accounts for 3–5% of cancer in both sexes and is the fourth most common cause of cancer death in both sexes [1]. Cancer of unknown primary is a diagnosis of exclusion, essentially every histologically confirmed metastatic malignancy for which the primary tumour site cannot be identified after a standardized diagnostic approach (physical examination, blood tests, endoscopies and high-quality imaging methods).

There is a long-lasting debate about the clinical importance of identifying the primary site in metastatic tumours of uncertain origin. On the one hand, there is the belief that CUP is merely disseminated disease from a ‘hidden’ primary tumour that could not be identified because it was too small or in a difficult anatomical location to be visualized by imaging techniques. In this case, the epidemiology, presentation, treatment and the course of the disease would not differ from the corresponding features of typical metastatic tumours of known primary. On the other hand, there is the more ‘heretic’ belief that CUP is a completely distinct clinical entity in which the origin does not really matter as there is a unique molecular/biological signature that differentiates it in every respect from typical metastatic cancers of known primary. This is the so called true or genuine CUP. In this setting, the disease is driven by a pro-metastatic, aggressive «CUP signature», patients have a short symptom history of less than 3 months and present with high-volume metastatic disease, often in uncommon sites (kidney, distant lymph nodes, scalp, soft tissues, skin and heart). They usually have short-lived responses to any kind of treatment and have a dismal prognosis with median survival rarely exceeding 12 months.

Currently, two main CUP groups are recognized based on of risk of death and clinicopathologic characteristics [2] (Table 1). The favourable prognostic group accounts for 20% of the cases and consists of clinicopathologic subsets with biological and clinical profiles corresponding to those of equivalent primary tumours. Unfortunately, the majority of CUP cases (80%) belongs to the poor-risk group, which consists of visceral metastatic disease and have an aggressive behaviour with limited response to treatment.

Table 1. The clinicopathological prognostic subgroups of cancer of unknown primary (CUP)
Favourable subsetUnfavourable subset

Women with papillary peritoneal adenocarcinoma

Women with axillary adenocarcinoma

Men with blastic bone metastases and elevated prostate-specific antigen (adenocarcinoma)

Poorly differentiated carcinoma with midline nodal distribution

Poorly differentiated neuroendocrine carcinoma

Squamous-cell carcinoma involving cervical lymph nodes

Adenocarcinoma with a colon-cancer profile (CK20+, CK7−, CDX2+)

Isolated inguinal adenopathy (squamous carcinoma)

Single potentially resectable tumour

Merkel cell adenopathy of unknown origin

Adenocarcinoma metastatic to the liver or other organs

Multiple cerebral metastases

Multiple lung or pleural metastases)

Multiple metastatic lytic bone disease (non-PSA)

Squamous-cell carcinoma of the adomino/pelvic cavity

Squamous abdominopelvic CUP

Recently, there has been active effort to disentangle the molecular profile of CUP to allow clinicians to understand the biology and potentially apply targeted agents for the treatment of this clinical entity. Herein, we review the up-to-date published data on the molecular biology of CUP and their prognostic/predictive relevance.

Genetic instability

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

Numerical and structural chromosomal abnormalities are a hallmark of cancer. Aneuploidy is a frequent event in solid tumours (70–90% of cases) and correlates with a dismal prognosis and chemoresistance [3]. The incidence of aneuploidy in CUP has been reported up to 70%, in a series of 153 patients [4]. Although there was no difference in median survival between patients with diploid and aneuploid tumours, 18% of the patients with diploid tumours survived for more than 2 years as compared to 9% of those with aneuploid tumours. Moreover, abnormalities of the short arm of chromosome 1 (1p) in 12 of 13 patients with CUP studied [5] and alterations in chromosome 12 i(12p) in 12/40 poor-prognosis patients with CUP have also been reported [6]. The presence of chromosomal alterations of chromosome 12 could predict response to cisplatin (75% vs. 18% response to cisplatin therapy in patients with tumours showing or not chromosome abnormalities, respectively; P = 0·002). [6] The predictive role of i(12p) for benefit from cisplatin-based chemotherapy was confirmed in other studies as well [7, 8], while gain of 12p can be used for the identification of extragonadal germ cell tumours in undifferentiated malignancies [7].

Pantou et al. used cytogenetic analyses to characterize 20 CUP cases [9]. They could detect multiple chromosomal rearrangements (1–51 changes) involving mainly the chromosomes 1, 6, 7, 11, while some of them were characteristic for specific histology subtypes (e.g. 4q31, 6q15, 10q25 and 13q22 were more frequently encountered in adenocarcinomas). The presence of complex karyotyping was prognostic for worse survival as compared with patients who had up to five alterations (median survival 3 vs. 19·4 months, respectively; P = 0·003) [9]. Similarly patients with CUP-adenocarcinoma and chromosomal instability (CIN, eight of the 14 cases) had a shorter survival than those with CIN-low tumours. The difference did not attain statistical significance, possibly due to the limited number of cases [10]. In the same study, it was demonstrated that nearly all the metastatic lesions were clonally related to the primary tumour identified at autopsy, highlighting the rapid tumour progression of this subtype.

Oncogenes

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

Oncogenes have a crucial role in the development of cancer through either overexpression or amplification. More interestingly, new targeted therapies have been developed for the protein products of these genes challenging clinicians to explore the biological profile of CUP and to find the genetic signature, which would render this clinical entity a potential field for the application of these agents. Overall, the incidence of protein overexpression or gene mutations of oncogenes in CUP is no different from rates reported in metastatic cancer of known primary, with heterogeneity present as anticipated.

Epidermal growth factor receptor (EGFR) family

EGFR and Her-2/C-Erb2 have been extensively studied in CUP (Table 2) [11-17]. EGFR overexpression varies between 12 and 61% of CUP cases while that of Her-2 between 4 and 35%. EGFR overexpression has been reported to be more frequent in squamous CUP (P < 0·001) [17]. In most of the cases, neither of them could be associated with a difference in survival. Exceptions include the coexpression of EGFR, HER2 and COX-2, which doubled the survival time as compared with patients bearing tumours without coexpression (24 vs. 12 months, P = 0·0079) [14]. EGFR protein expression was reported to predict for response to cisplatin-based chemotherapy (response rates were 50 and 22% in patients with EGFR-positive tumours and EGFR-negative tumours, respectively; P < 0·05) [16]. Despite the significant overexpression of these receptors in CUP, the clinical relevance is uncertain. Anti-EGFR and anti-Her2 agents need to be evaluated in patients with CUP overexpressing these receptors or harbouring mutated EGFRs, although it is suggested that these mutations maybe missing [15]. The two other members of the family (Her-3 and 4) have not been analysed yet in CUP.

Table 2. Epidermal growth family receptor family (EGFR/Her-2)
AuthorNrMethodResultPrognostic/Predictive Relevance
  1. IHC, immunohistochemistry; HER-2, human epidermal growth factor receptor type 2; COX-2, cyclooxygenase-2; EGFR, epidermal growth factor receptor; PCR, polymerase chain reaction; qPCR, quantitative polymerase chain reaction; p-EGFR, phosphor-EGFR.

Pavlidis et al.. [11]26IHCHER-2 expression, 65%; overexpression, 27%None
Hainsworth et al. [12]100IHCHER-2 overexpression, 11%None
Van de Wouw et al.[13]45IHCHER-2 overexpression, 35%None
Rashid et al.[14]76IHC

HER-2 expression, 68%; overexpression, 24%

EGFR expression, 75%; overexpression, 61%

Coexpression of EGFR, HER-2, and VEGF or COX-2 in 54%

Superior survival with coexpression of EGFR, HER2, and COX-2
Dova et al. [15]50

IHC

PCR

qPCR

EGFR expression, 74%;overexpression, 12%

Wild-type exon 18, 19, and 21 EGFR gene in 96%

One case with exon 21 SNP:R836R

One case with exon 19 intronic splicing variant: IVS19_24G_A

No exon 18, 19, or 21 EGFR gene amplification

None
Massard et al. [16]54IHCHER-2 overexpression, 4%; EGFR expression 66% overexpression 35%EGFR expression was significantly correlated with response to cisplatin-based chemotherapy
Koo et al. [17]69

IHC/FISH

IHC

HER-2 overexpression/amplification 10·1%

p-HER (Py877) 14·5%

EGFR overexpression 21·7%

p-EGFR (Tyr 1068) 4·3%

None

c-Kit, platelet-derived growth factor receptor (PDGFR)

The c-Kit and the PDGFR proteins are transmembrane receptors with tyrosine kinase activity. TKIs targeting c-Kit and PDGFR have shown remarkable activity in case of chronic myeloid leukaemia and gastrointestinal stromal tumours [18, 19]. c-Kit is variably expressed in CUP as this was assessed by immunohistochemistry (IHC), while no kit gene mutations of exon 11 could be demonstrated [20]. No c-Kit or PDGFR inhibitors have been used in CUP up to date, as the incidence or clinical significance of these oncogene aberrations in the disease has not been elucidated (Table 3).

Table 3. c-kit/PDGFR
AuthorNrMethodResultPrognostic/Predictive Relevance
  1. IHC, immunohistochemistry; PDGFR, platelet-derived growth factor receptor; PCR, polymerase chain reaction.

Rashid et al.[14]76IHCc-Kit expression, 12%; overexpression, 4%None
Dova et al. [20]50

IHC

PCR

c-Kit expression, 81%; overexpression, 13%

PDFGR expression in 50%

Overexpression 3%

No exon 11 c-kit gene mutations

No exon 12, 18 PDGFR mutations

None
Massard et al. [16]54IHCPDGFR expression, 10%None
Koo et al. [17]69IHCPDGFRβ expression, 24·6%None

Other proto-oncogenes that have been studied include C-Myc and Ras, which were almost universally expressed in the 26 CUP cases studied (96% and 92%, respectively) [11] and the antiapoptotic protein Bcl-2 which was overexpressed in 40% of cases [21]. In either case, no prognostic value could be demonstrated.

Tumour suppressor and DNA repair genes

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

Tumour suppressor genes are frequently lost or inactivated during carcinogenesis promoting the initiation and progression of cancer through regulation of various biological processes such as cell proliferation, cell death and cell migration/invasion. The TP53 gene, the so called gatekeeper of the genome is the first tumour suppressor gene to be identified and the best one studied [22]. Mutations in TP53 gene are one of the most common genetic alterations in human cancers, and up to 50% of all human cancers contain mutations in both alleles of the TP53 gene [22].

In case of CUP, TP53 protein expression is observed up to 70% [21] and overexpression in 50% of cases. The expression of TP53 protein by itself had no prognostic value (Table 4) though coexpression of TP53 and Bcl-2 (20% of tumours) predicted for a higher response rate to cisplatin-based combination chemotherapy [21]. Bar-Eli et al. [23] investigated the frequency of p53 exon 5–9 mutations in a series of 15 CUP cases, and eight cell lines derived from CUP by means of PCR. A functional mutation was observed in six cases (26%) distributed in eight codons. In another cohort, no mutations or polymorphisms could be detected in 23 CUP cases confined to the cervical lymph nodes [24]. The unexpectedly low frequency of mutations is in discordance with IHC results, attributed partly to the differing specificities of antibodies for wild-type and mutated p53 gene products, the occurrence of metastases outside the exon 5–9 regions, and the differential impact of p53-regulating factors such as murine double minute (MDM)-2, p14 alternate reading frame (p14ARF) and p21 [25].

Table 4. Tumor suppressor and DNA repair genes
AuthorNrMethodResultPrognostic/Predictive Relevance
  1. IHC, immunohistochemistry; PCR, polymerase chain reaction; qRT- PCR, quantitative real-time–polymerase chain reaction; BRCA1, Breast Cancer 1; ERCC1, excision repair cross-complementation group 1; RMM1, ribonucleotide reductase M1.

Briasoulis et al. [21]40IHCp53 expression, 70%; overexpression, 53%Coexpression of Bcl-2 and p53 predictive for response to platinum
Van de Wouw et al. [13]48IHCp53 expression, 48%None
Bar-Eli et al. [23]23PCRExon 5–9 p53 gene mutations in 26%Not reported
Gottschlich et al. [24]23PCRNo exon 4–9 mutation in p53Not reported
Dova et al. [27]50PCROne case with KiSS-1 exon 4a 242C_G mutation (P81R)Not reported
Koo et al. [17]69IHC

P53 expression18·8%

ERCC1 expression 27·5%

RMM1 expression 88·4%

None
Souglakos et al. [29]56qRT- PCRHigh amplification rate of BRCA1, ERCC1, RRM1High ERCC1 mRNA expression prognostic for survival

Cancer of unknown primary is characterized by early dissemination, indicating that inactivation of metastasis-suppressor genes may be an early event in the tumorigenesis. KiSS-1 is the most extensively studied metastasis-suppressor gene. Loss-of-function of KiSS-1 was seen in several human malignancies and was correlated with advanced stage, high tumour burden and poor prognosis [26]. Dova et al. screened 50 cases of CUP for KiSS-1 gene mutations by PCR and direct sequencing and found only one mutated sample [27], suggesting that KiSS1 has not a central role in the metastasis cascade of CUP.

Defects in DNA repair genes are also highly encountered in human malignancies and are often predictive for chemotherapy [28]. Souglakos et al. studied the gene expression of BRCA1 (55/56 cases), ERCC1 (all), RRM1 (49/56) in a cohort of 56 cases by means of RT-PCR [29]. Patients with high ERCC1 mRNA levels had increased median overall survival (mOS: 21·8 months) in comparison with those with low ERCC1 expression (6·3 months; P = 0·027), while a statistical trend for increased survival was observed in patients with high RRM1 mRNA expression (P = 0·092). On the other hand, low expression of ERCC1 by IHC with no prognostic significance was reported by Koo et al. [17].

As a recurring theme, the incidence of tumour suppressor and DNA repair gene inactivation in CUP is similar to that reported for the variety of metastatic solid tumours of known primary. Accordingly, the peculiar CUP biology cannot be attributed to single-gene aberrations.

Angiogenesis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

Angiogenesis is the process through which new capillaries are formed from pre-existing vessels. Tumours are highly dependent on this process for their growth, survival and invasion. Hypoxia is a common condition found in a wide range of solid tumours and is often associated with poor prognosis. Tumour angiogenesis has been extensively investigated in cancer, while the relevance of angiogenesis in CUP progression has been highlighted lately (Table 5). Angiogenic deficiency of the primary tumour, inducing dormancy, with active angiogenesis in metastatic deposits may serve as a model for explaining CUP biology.

Table 5. Angiogenesis/Hypoxia
AuthorNrMethodResultPrognostic/Predictive Relevance
Hillen et al. [53]69IHCNo difference in CD34 microvessel density between 39 CUP liver metastases and 30 known primary liver metastasesCD34 microvessel density with adverse prognostic significance for survival
Van de Wouw et al. [13]48IHC

CD34 microvessel density, median 56/mm3

VEGF expression, 39%;

Overexpression, 26%

None
Karavasilis et al. [31]80IHC

CD34 microvessel density, median 59 microvessels/mm2

VEGF expression, 100%; overexpression, 83%

Stromal TSP-1 expression, 83%; overexpression, 20%

Increased microvessel density in unfavourable group CUP

Positive correlation of VEGF and inverse correlation of TSP-1 with microvessel density

Rashid et al. [14]75IHC

VEGF expression, 49%; overexpression, 29%

Coexpression of EGFR, HER-2, and VEGF or COX-2 in 54%

None
Koo et al. [17]69IHCHIF1a expression, 20·3%HIF1a expression is adverse prognostic factor in multivariate analysis
Souglakos et al. [29]56qRT- PCRPatients with high HIF1a and TXR1 mRNA expression presented significantly lower mOSPrognostic significance for HIF1a mRNA
Agarwal et al. [31]50

IHC, WB

RT-PCR

Decreased VEGF expression in CUP as compared with MKP

Decreased expression of the VEGF121 and VEGF165 isoforms

Not Reported

Therefore, studies suggest that high microvessel density (MVD) is related with adverse prognosis in both univariate and multivariate analysis [30], while it is more often encountered in the unfavourable CUP group [31]. There are no data directly correlating vascular endothelial growth factor (VEGF) expression with prognosis, except for the positive correlation of VEGF and MVD. [13, 31]. On the other hand, Agarwal et al. tested the hypothesis that CUP is of low angiogenic phenotype in 50 cases of squamous carcinomas metastatic to the cervical lymph nodes and compared them with 52 cases of metastases from a known primary (MKP) [31]. They did observe lower VEGF expression at protein level in CUP. They also described that certain isoforms (esp. VEGF121 and 165) were more frequently expressed in MKP, and they suggested a different pattern of metastatic spread in squamous CUP metastasizing to the cervical lymph nodes, independently of angiogenesis [31].

Whatever the prognostic significance of angiogenesis in CUP, the initial results for the role of antiangiogenic treatment show promising activity. The combination of bevacizumab and erlotinib has been studied in first line (in combination with chemotherapy) and second line phase II studies [32, 33]. In first line treatment, disease control could be achieved in 82% of patients with a median progression free survival and median overall survival of 8 and 12·6 months, respectively [32]. Moreover, in the second line therapy, the combination of the two targeted agents (without chemotherapy) yielded a high clinical benefit rate of 71% and a median OS of 7·4 months. Both of these regimens were well tolerated, while their evaluation in a phase III trial would be necessary before any conclusions could be drawn.

Hypoxia

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

Hypoxia is a common condition found in a wide range of malignancies and is often associated with poor prognosis. The expression of regulatory proteins, especially hypoxia-inducible factors (HIF), have a critical role in the adaptation of tumour cells to hypoxia by activating the transcription of genes, which are involved glycolysis, angiogenesis, migration and cell proliferation.[34] Koo et al. reported that a hypoxic phenotype defined by the IHC expression of HIF1a, GLUT1, COX2 in squamous cell carcinoma was associated with a poor prognosis compared with those without expression of these proteins (P = 0·048, P = 0·029, and P = 0·042, respectively) [17]. In addition, HIF1α expression was an independent poor-prognostic factor in multivariable analysis (P = 0·034, 95% CI 0·879–71·503). Similarly, Souglakos et al. highlighted that the patients with high HIF1a mRNA expression presented significantly lower median OS (6·9 months) in comparison with those with low HIF1a expression (19·8 months; P = 0·031). High TXR1 mRNA expression (an angiogenesis regulator) was significantly correlated with decreased survival (7·4 months vs. 18·3 months; P = 0·038) [29].

Tumour–stroma interaction and stem cell phenotype

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

The tumour–stroma interaction has been brought lately into attention concerning its role in tumour development, support and progression as well as tumour chemoresistance [35]. One of the initial events of tumour migration is the invasion to the surrounding tissues by disrupting the extracellular matrix through the excretion of proteolytic enzymes known as metalloproteinases (MMPs). The expression of MMP 2 and 9 alongside their inhibitor, the tissue inhibitor of metalloproteinases-1 (TIMP-1) were studied by Karavasilis et al. in 75 CUP cases [36]. MMP-2, MMP-9 and TIMP-1 were overexpressed in 49%, 36 and 44% of CUP tumour cells, respectively. Interestingly, MMP-2 and 9 were weakly expressed in the stroma. High TIMP-1 was associated with a shorter survival (7·5 vs. 12 months, P = 0·016), while neither of the two MMPs could predict for survival. Further on, our group has studied the epithelial–mesenchymal transition (EMT) phenotype in 100 CUP cases by IHC. An EMT phenotype was defined as low protein expression of E-cadherin, expression of N-cadherin or vimentin with concomitant expression of the transcription factor SNAIL [37]. EMT was infrequently seen in this series (8·1%) but was strongly associated with poor OS (median OS in EMT-neg =13 months vs. median OS in EMT-pos. = 8 months, P = 0·023). EMT was also correlated significantly with male gender, high grade and presence of visceral metastases (χ2 P < 0·05), and a potential causal role of NOTCH2 and 3 activation towards induction of the EMT was pointed out (Table 6). The rarity of the EMT phenotype in CUP may be due to its plasticity, restriction to the malignant invasive front and transient nature.

Table 6. Tumor/Stroma Interaction and molecular pathways
AuthorNrMethodResultPrognostic/Predictive Relevance
  1. IHC, immunohistochemistry; MMP, Metaloproteinase; TIMP-1, tissue inhibitor of metalloproteinases type 1; EMT, Epithelial Mesenchymal Transition; OS, Overall Survival. *Positive cases according to the cut-offs, #Cut-off value (% of positive cells).

Karavasilis et al. [36]76IHC

MMP-2 expression, 69%; overexpression; 49%

MMP-9 expression, 49%; overexpression, 36%

TIMP-1 expression, 79%; overexpression, 44%

 Adverse prognostic significance of TIMP-1 expression
Stoyianni et al. [37]100IHC

E-Cadherin, *

SNAIL, by H score

Vimentin,

N-Cadherin, OCT

EMT phenotype in IHC and 16% by H

78·8% (< 60%)#

61·9% (≥ 85%) #

23·2% (≥ 40%)#

13·8% (≥ 40%)#

0% (1%)

8·1% of cases by score

EMT is prognostic for adverse OS
Krikelis et al. [39]100IHC

c-Met *

pMAPK

Notch 1

Notch 2

Notch 3

Jagged 1

42% (> 20%)#

54% (> 40%)#

2% (> 5%)#

56% (> 20%)#

73% (> 80%)#

22% (≥ 5%)#

pMAPK and c-Met are prognostic for OS

pMAPK predictive for chemotherapy

Golfinopoulos et al. [43]100IHC

PTEN *

pAKT

pRPS6

p21

Cyclin D1

50% (60%)#

73% (85%)#

59% (6%)#

60% (10%)#

44% (20%)#

p21, pAKT and pPRS6 prognostic for survival

Co-expression of pAKT and pMAPK major adverse prognostic marker

Cancer stem cells are potentially responsible for tumour expansion, repopulation and metastases and are considered to be cells that have high intrinsic migration/dissemination potential [38]. Given the fact that CUP shows small or absent primaries and rapid dissemination, the presence of a stem cell phenotype has been sought for in CUP cases by our group. Among 100 CUP FFPE tumours, we defined stem cell phenotype as immunohistochemical expression of CD133 and OCT4 and failed to find any (data not published). However, CUP circulating tumour cells in peripheral blood did exhibit Aldehyde Dehydrogenase-1 (ALDH1) expression by immunofluorescence in seven of 14 cases (data not published). Accordingly, acquisition of stem cell phenotype by CUP may be an event which is rare, transient and/or dynamic.

Signalling pathways in CUP

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

It is obvious that the expression of single or a few genes is not prognostic neither can it interpret the peculiar biology of CUP. The interest has shifted to the characterization of various molecular pathways, which seem to depict better the complicated process of carcinogenesis.

Krikelis et al. studied the protein expression of key genes involved in cell proliferation, differentiation, survival and apoptosis [39]. The immunohistochemical (IHC) expression of Notch1,-2,-3, Jagged1, c-Met and pMAPK was evaluated in 100 CUP tumours using tissue microarrays. Notch3 and pMAPK were most frequently overexpressed (73% and 54%, respectively), while Notch1 and Jagged1 rarely so. Notch-2 and c-Met were also frequently overexpressed (Table 6). c-Met and Notch3 expression were found to be statistically more frequent in squamous carcinomas (positive in 90% of cases) and associated with a unique metastatic pattern (c-Met-high in soft tissue/lymph node metastases, P < 0·001, Notch3-high in visceral, peritoneal/pleural and soft tissue/lymph node metastases, P < 0·001). pMAPK emerged as the major adverse prognostic factor (median overall survival OS 9 vs. 17 months in pMAPK high versus low cases, P = 0·016), while c-Met expression had a favourable prognostic impact, both in univariate (median, OS 15 vs. 9 months, P = 0·05) and in multivariate analysis (relative risk (RR) for death 0·48, P = 0·025). Notch 3 overexpression was correlated to worse survival in the midline nodal CUP subset (12 mos vs. 31 mos – P = 0·05), while Notch 1 overexpression was linked to inferior PFS in the visceral group (3 mos vs. 7 mos P = 0·05). Elevated pMAPK expression intensity was also correlated with increased response to chemotherapy while c-Met and Notch-3 were weaker and borderline predictive biomarkers, respectively.

MET is a transmembrane receptor that is activated upon binding of the hepatocyte growth factor. MET overexpression and activation by somatic mutations at early stages of cancer onset might sustain aberrant execution of the invasive growth programme. The role of c-Met mutations in CUP biology has been studied by Stella et al. [40] In a cohort of 23 CUP cases, they demonstrated that c-Met mutations are encountered in 30%, that is, much higher incidence than the respective one in MKPs which is estimated to be around 4%. This exceptional high rate of mutation may indicate a potential prognostic marker for CUP, while the development of new agents targeting Met renders this therapeutic approach very attractive for CUP [41].

Another pathway of high translational interest is the PI3K/PTEN/AKT, which is involved in cancer initiation and progression, cell growth, proliferation, metabolism and survival [42]. It has extensive crosstalk with other key molecules such as RPS6 (40S ribosomal protein S6), p21 (cyclin-dependent kinase inhibitor 1A, CDKN1A) and CCND1 (cyclin D1). The role of the phosphorylated active forms of Akt and PTEN in CUP were studied in a cohort of 100 patients with CUP [43]. The rate of immunohistochemical PTEN and pAKT overexpression was quite high (60% and 85%, respectively), while that of pRPS6, p21 and Cyclin D1 was seen in less than 20% of cases (Table 6). In multivariate analysis, high p21IHC expression was associated with better survival (risk ratio [RR] = 0·34 [95% CI, 0·16–0·73], P = 0·005). At both univariate and multivariate analysis, high IHC expression of pAKT and pRPS6 were significantly associated with worse survival; pAKT (RR = 2·39 [95% CI, 1·23–4·66], P = 0·01) and pRPS6 (RR = 2·76 [95% confidence interval (CI) 1·31–5·84], P = 0·008). Prognosis was also worse upon concurrent high IHC expression of pMAPK and pAKT {median overall survival = 8 months [95% confidence interval (CI) 5·3–10·7] vs. 17 months [95% CI 13·1–20·9]}. The high expression of pAKT as well as the dismal prognostic impact of this biomolecule renders the implementation of new Pi3k/Akt inhibitors in CUP a future challenge as well as an opportunity. Dual inhibition of both AKT and MAPK seems also promising, although toxicity may represent an issue for clinical application [44].

Circulating tumour cells

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

CUP is the prototype of early cancer dissemination through shedding of tumour cells from the primary site to the bloodstream, circulation and implantation in metastatic foci [45]. The identification and molecular characterization of circulating tumour cells (CTCs) is a vivid area of translational research in oncology. Our group, could detect the presence of CTCs in 15/24 (62·5%) patients with CUP, using immunofluorescent detection in peripheral blood mononuclear cells [46]. The presence and number of CTCs correlated significantly with the histological diagnosis of poorly differentiated carcinoma (P = 0·04), high grade (P = 0·02) and peripheral leukocytosis (WBC> 10·000, P = 0·022) but had no prognostic utility for OS or PFS. As stated previously, CUP circulating tumour cells in peripheral blood did exhibit Aldehyde Dehydrogenase-1 (ALDH1) expression by immunofluorescence in seven of 14 cases (data not published), suggesting acquisition of a stem cell-like phenotype.

Gene expression profiling

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

The critical question in CUP is if the biology is dictated by the primary tumour or if it is defined by the unique metastatic tropism irrespective of the primary. In the first instance, treatment with primary-tailored chemotherapy regimens would be optimal, while in the second case, effective targeting of biomolecules pivotal for metastatic dissemination would be needed. Multiple gene expression profiling platforms have been broadly used for the identification of the tissue of origin [47, 48]. The concept used, is to characterize the genes expressed in the various solid tumours (training set), validate the results in a second cohort of known tumours (validation set), identify gene sets differentially expressed between various tumour types and then apply them in CUP cases to «molecularly» assign them to a tissue of origin [49]. cDNA, mRNA or microRNA profiling have been used for the search of the primary site yielding more that 90% accuracy [50]. Gene expression assays are described in Table 7.

Table 7. Gene expression profile assays
AssayPlatformTissueNumber of tumor typesNumber of genesAccuracy in known tumors(%)
  1. FFPE, Formalin-Fixed, Paraffin-Embedded (tissue).

Veridex

RT-PCR

mRNA

FFPE61076

Pathwork Diagnostics

Tissue of Origin test

cDNA microarrayFrozen/FFPE15150089

Rosetta Genomics

MiReview met

RT-PCR

miRNA

FFPE2248 miRNAs86

bioTheranostics

Cancer Type ID

RT-PCR

mRNA

FFPE399286

Whole genome arrays have only been used for examining gene differential expression across several tumour types to select for tissue of origin classifiers, but not for examining gene expression differences between CUP and various solid tumours.

Hainsworth et al. treated patients with CUP prospectively, based on gene profiling assay-predicted tumour site [50]. Patients treated with primary-specific therapies seemed to have better survival as compared to historical controls treated with empiric chemotherapy [51]. In one of the few attempts at looking for CUP versus MKP biological differences, global microRNA profiling failed to demonstrate a significant microRNA expression difference between matched subgroups of CUP and MKP of favourable prognosis (axillary nodal CUP vs. breast cancer metastases, peritoneal CUP versus peritoneal deposits from ovarian cancer, squamous cervical nodal CUP vs. head neck cancer metastases) [52]. This result argues against the existence of a CUP-specific signature. However, we should keep in mind that the CUP signature is likely to be modulated by epigenetic changes and proteomics and likely resides in the group of the biologically genuine, poor-risk visceral CUP cases. Accordingly, it would be rational to be searched for in the latter subgroup.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

The identification of a CUP-specific signature remains elusive; therefore, the jury is still pending regarding the true nature of CUP: a disease with missing primary or missing biology? In the past decade, the explosion of gene profiling assays helped to biologically assign CUP cases to various tissues of origin, but generated no insights on the biological differences between CUP and typical metastatic solid tumours. Such a profiling research plan focusing on patients with poor-risk visceral CUP, coupled to microRNA, epigenetic and proteomic studies are needed to explore further the labyrinth of CUP.

Address

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References

Department of Medical Oncology, University Hospital of Larissa, Mesourlo, 41334 Larissa, Greece (Konstantinos Kamposioras); Department of Medical Oncology, Ioannina University Hospital, Niarxou Avenue, 45500 Ioannina, Greece (George Pentheroudakis, Nicholas Pavlidis) .

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Genetic instability
  5. Oncogenes
  6. Tumour suppressor and DNA repair genes
  7. Angiogenesis
  8. Hypoxia
  9. Tumour–stroma interaction and stem cell phenotype
  10. Signalling pathways in CUP
  11. Circulating tumour cells
  12. Gene expression profiling
  13. Conclusion
  14. Address
  15. References
  • 1
    Pavlidis N, Fizazi K. Carcinoma of Unknown Primary. Crit Rev Oncol Hematol 2009;69:2718.
  • 2
    Pavlidis N, Pentheroudakis G. Cancer of unknown primary site. Lancet 2012;9824:142835.
  • 3
    Rajagopalan H, Lengauer C. Aneuploidy and cancer. Nature 2004;432:33841.
  • 4
    Hedley DW, Leary JA, Kirsten F. Metastatic adenocarcinoma of unknown primary site: abnormalities of cellular DNA content and survival. Eur J Cancer Clin Oncol 1985;21:1859.
  • 5
    Abbruzzese JL, Lenzi R, Raber MN, Pathak S, Frost P. The biology of unknown primary tumors. Semin Oncol 1993;20:23843.
  • 6
    Motzer RJ, Rodriguez E, Reuter VE, Bosl GJ, Mazumdar M, Chaganti RS. Molecular and cytogenetic studies in the diagnosis of patients with poorly differentiated carcinomas of unknown primary site. J Clin Oncol 1995;13:27482.
  • 7
    Summersgill B, Goker H, Osin P, Huddart R, Horwich A, Fisher C et al. Establishing germ cell origin of undifferentiated tumors by identifying gain of 12p material using comparative genomic hybridization analysis of paraffin-embedded samples. Diagn Mol Pathol 1998;7:2606.
  • 8
    Ilson DH, Motzer RJ, Rodriguez E, Chaganti RS, Bosl GJ. Genetic analysis in the diagnosis of neoplasms of unknown primary tumor site. Semin Oncol 1993;20:22937.
  • 9
    Pantou D, Tsarouha H, Papadopoulou A, Mahaira L, Kyriazoglou I, Apostolikas N et al. Cytogenetic profile of unknown primary tumors: clues for their pathogenesis and clinical management. Neoplasia 2003;5:2331.
  • 10
    Speel EJ, van de Wouw AJ, Claessen SM, Haesevoets A, Hopman AH, van der Wurff AA et al. Molecular evidence for a clonal relationship between multiple lesions in patients with unknown primary adenocarcinoma. Int J Cancer 2008;123:1292300.
  • 11
    Pavlidis N, Briassoulis E, Bai M, Fountzilas G, Agnantis N. Overexpression of C-myc, Ras and C-erbB-2 oncoproteins in carcinoma of unknown primary origin. Anticancer Res 1995;1:256367.
  • 12
    Hainsworth JD, Lennington WJ, Greco FA. Overexpression of Her-2 in patients with poorly differentiated carcinoma or poorly differentiated adenocarcinoma of unknown primary site. J Clin Oncol 2000;1:6325.
  • 13
    van de Wouw AJ, Jansen RLH, Griffioen AW, Hillen HFP. Clinical and immunohistochemical analysis of patients with unknown primary tumour. A search for prognostic factors in UPT. Anticancer Res 2004;24:297302.
  • 14
    Rashid A, Hess KR, Lenzi R, Raber MN, Abbruzzese JL, Varadhachary GR. Overexpression and prevalence of molecular markers in patients with cancer of unknown primary (CUP). Proc Am Soc Clin Oncol 2005;2:9683.
  • 15
    Dova L, Pentheroudakis G, Georgiou I, Malamou-Mitsi V, Vartholomatos G, Fountzilas G et al. Global profiling of EGFR gene mutation, amplification, regulation and tissue protein expression in unknown primary carcinomas: to target or not to target? Clin Exp Metastasis 2007;24:7986.
  • 16
    Massard C, Voigt JJ, Laplanche A, Culine S, Lortholary A, Bugat R et al. Carcinoma of an unknown primary: are EGF receptor, Her-2/neu, and c-Kit tyrosine kinases potential targets for therapy? Br J Cancer 2007;97:85761.
  • 17
    Koo JS, Kim H. Hypoxia-related protein expression and its clinicopathologic implication in carcinoma of unknown primary. Tumour Biol 2011;32:893904.
  • 18
    Joensuu H, Fletcher C, Dimitrijevic S, Silberman S, Roberts P, Demetri G. Management of malignant gastrointestinal stromal tumours. Lancet Oncol 2002;3:65564.
  • 19
    O'Hare T, Corbin AS, Druker BJ. Targeted CML therapy: controlling drug resistance, seeking cure. Curr Opin Genet Dev 2006;16:929.
  • 20
    Dova L, Pentheroudakis G, Golfinopoulos V, Malamou-Mitsi V, Georgiou I, Vartholomatos G et al. Targeting c-KIT, PDGFR in cancer of unknown primary: a screening study for molecular markers of benefit. J Cancer Res Clin Oncol 2008;134:697704.
  • 21
    Briasoulis E, Tsokos M, Fountzilas G, Bafaloukos D, Kosmidis P, Samantas E et al. Bcl2 and p53 protein expression in metastatic carcinoma of unknown primary origin: biological and clinical implications. A Hellenic Co-operative Oncology Group study. Anticancer Res 1998;18:190714.
  • 22
    Soussi T, Ishioka C, Claustres M, Béroud C. Locus-specific mutation databases: pitfalls and good practice based on the p53 experience. Nat Rev Cancer 2006;6:8390.
  • 23
    Bar-Eli M, Abbruzzese JL, Lee-Jackson D et al. p53 gene mutation spectrum in human unknown primary tumors. Anticancer Res 1993;13:161923.
  • 24
    Gottschlich S, Schuhmacher O, Gorogh T, Hoffmann M, Maune S. Analysis of the p53 gene status of lymph node metastasis in the head and neck region in occult primary cancer. Laryngorhinootologie 2000;79:4347.
  • 25
    Vousden KH, Woude GF. The ins and outs of p53. Nat Cell Biol 2000;2:E17880.
  • 26
    Makri A, Pissimissis N, Lembessis P, Polychronakos C, Koutsilieris M. The kisspeptin (KiSS-1)/GPR54 system in cancer biology. Cancer Treat Rev 2008;34:68292.
  • 27
    Dova L, Golfinopoulos V, Pentheroudakis G, Georgiou I, Pavlidis N. Systemic dissemination in cancer of unknown primary is independent of mutational inactivation of the KiSS-1 metastasis-suppressor gene. Pathol Oncol Res 2008;14:23941.
  • 28
    Martin LP, Hamilton TC, Schilder RJ. Platinum resistance: the role of DNA repair pathways. Clin Cancer Res 2008;14:12915.
  • 29
    Souglakos J, Pentheroudakis G, Papadaki C, Cervantes A, Petrakis D, Roodriguez-Braun E et al. Prognostic significance of gene expression profile in patients with carcinomas of unknown primary. Ann Oncol 2010;128P:viii55.
  • 30
    Karavasilis V, Malamou-Mitsi V, Briasoulis E, Tsanou E, Kitsou E, Kalofonos H et al. Angiogenesis in cancer of unknown primary: clinicopathological study of CD34, VEGF and TSP-1. BMC Cancer 2005;5:2532.
  • 31
    Agarwal B, Das P, Naresh KN, Borges AM. Angiogenic ability of metastatic squamous carcinoma in the cervical lymph nodes from unknown primary tumours. J Clin Pathol 2011;64:76570.
  • 32
    Hainsworth JD, Spigel DR, Thompson DS, Murphy PB, Lane CM, Waterhouse DM et al. Paclitaxel/carboplatin plus bevacizumab/erlotinib in the first-line treatment of patients with carcinoma of unknown primary site. Oncologist 2009;14:118997.
  • 33
    Hainsworth JD, Spigel DR, Farley C, Thompson DS, Shipley DL, Greco FA. Phase II trial of bevacizumab and erlotinib in carcinomas of unknown primary site: the Minnie pearl cancer research network. J Clin Oncol 2007;25:174752.
  • 34
    Lu X, Kang Y. Hypoxia and hypoxia-inducible factors: master regulators of metastasis. Clin Cancer Res 2010;16:592835.
  • 35
    Iwatsuki M, Mimori K, Yokobori T, Ishi H, Beppu T, Nakamori S et al. Epithelial-mesenchymal transition in cancer development and its clinical significance. Cancer Sci 2010;101:2939.
  • 36
    Karavasilis V, Malamou-Mitsi V, Briasoulis E, Tsanou E, Kitsou E, Kalofonos H et al. Matrix metalloproteinases in carcinoma of unknown primary. Cancer 2005;104:22827.
  • 37
    Stoyianni A, Goussia A, Pentheroudakis G, Siozopoulou V, Ioachim E, Krikelis D et al. Immunohistochemical study of the epithelial-mesenchymal transition phenotype in cancer of unknown primary: incidence, correlations and prognostic utility. Anticancer Res 2012;32:127381.
  • 38
    Sampieri K, Fodde R. Cancer stem cells and metastasis. Semin Cancer Biol 2012;22(3):18793.
  • 39
    Krikelis D, Pentheroudakis G, Goussia A, Siozopoulou V, Bobos M, Petrakis D et al. Profiling immunohistochemical expression of NOTCH1-3, JAGGED1, cMET, and phospho-MAPK in 100 carcinomas of unknown primary. Clin Exp Metastasis 2012;29:60314.
  • 40
    Stella GM, Benvenuti S, Gramaglia D, Scarpa A, Tomezzoli A, Cassoni P et al. MET mutations in cancers of unknown primary origin (CUPs). Hum Mutat 2011;32:4450.
  • 41
    Peters S, Adjei AA. MET: a promising anticancer therapeutic target. Nat Rev Clin Oncol 2012;9:31426.
  • 42
    Vivanco I, Sawyers CL. The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nat Rev Cancer 2002;2:489501.
  • 43
    Golfinopoulos V, Pentheroudakis G, Goussia A, Siozopoulou V, Bobos M, Krikelis D et al. Intracellular signalling via the AKT axis and downstream effectors is active and prognostically significant in cancer of unknown primary (CUP): a study of 100 CUP cases. Ann Oncol 2012;23:272530.
  • 44
    Bartholomeusz C, Gonzalez-Angulo AM. Targeting the PI3K signaling pathway in cancer therapy. Expert Opin Ther Targets 2012;16:12130.
  • 45
    Alix-Panabières C, Pantel K. Circulating tumor cells: liquid biopsy of cancer. Clin Chem 2012. [Epub ahead of print].
  • 46
    Pentheroudakis G, Apostolaki S, Stoyianni A, Karaxalios D, Patramani S, Goussia A et al. Circulating tumour cells in cancer of unknown primary site:correlation with clinico pathologic characteristics and prognosis. Ann Oncol 2010; 21:Abstr 865: viii269.
  • 47
    Pentheroudakis G, Briasoulis E, Pavlidis N. Cancer of unknown primary site: missing primary or missing biology? Oncologist 2007;12:41825.
  • 48
    Pentheroudakis G, Golfinopoulos V, Pavlidis N. Switching benchmarks in cancer of unknown primary: from autopsy to microarray. Eur J Cancer 2007;43:202636.
  • 49
    Monzon FA, Koen TJ. Diagnosis of metastatic neoplasms: molecular approaches for identification of tissue of origin. Arch Pathol Lab Med 2010;134:21624.
  • 50
    Hainsworth JD, Rubin MS, Spigel DR, Boccia RV, Raby S, Quinn R et al. Molecular gene expression profiling to predict the tissue of origin and direct site-specific therapy in patients with carcinoma of unknown primary site: a prospective trial of the sarah cannon research institute. J Clin Oncol 2013;31(2):21723.
  • 51
    Pentheroudakis G, Greco FA, Pavlidis N. Molecular assignment of tissue of origin in cancer of unknown primary may not predict response to therapy or outcome: a systematic literature review. Cancer Treat Rev 2009;35:2217.
  • 52
    Pentheroudakis G, Spector Y, Krikelis D, Kotoula V, Meiri E, Malamou-Mitsi V et al. Global microRNA profiling in favorable prognosis subgroups of cancer of unknown primary (CUP) demonstrates no significant expression differences with metastases of matched known primary tumors. Clin Exp Metastasis 2012. [Epub ahead of print].
  • 53
    Hillen HF, Hak LE, Joosten-Achjanie SR, Arends JW. Microvessel density in unknown primary tumors. Int J Cancer 1997;74:815.