Cellular Oncology

, Volume 36, Issue 2, pp 95–112

Tumour-microenvironment interactions: role of tumour stroma and proteins produced by cancer-associated fibroblasts in chemotherapy response

  • Matthew David Hale
  • Jeremy David Hayden
  • Heike Irmgard Grabsch
Review

DOI: 10.1007/s13402-013-0127-7

Cite this article as:
Hale, M.D., Hayden, J.D. & Grabsch, H.I. Cell Oncol. (2013) 36: 95. doi:10.1007/s13402-013-0127-7

Abstract

Background

Cytotoxic chemotherapy improves survival for some, but not all, cancer patients. Non-responders may experience unnecessary toxicity and cancer progression, thus creating an urgent need for biomarkers that can predict the response to chemotherapy. So far, the search for such biomarkers has primarily been focused on the cancer cells and less on their surrounding stroma. This stroma is known to act as a key regulator of tumour progression and, in addition, has been associated with drug delivery and drug efficacy. Fibroblasts represent the major cell type in cancer-associated stroma and they secrete extracellular matrix proteins as well as growth factors. This Medline-based literature review summarises the results from studies on epithelial cancers and aimed at investigating relationships between the quantity and quality of the intra-tumoral stroma, the cancer-associated fibroblasts, the proteins they produce and the concomitant response to chemotherapy. Biomarkers were selected for review that are known to affect cancer-related characteristics and patient prognosis.

Results

The current literature supports the hypothesis that biomarkers derived from the tumour stroma may be useful to predict response to chemotherapy. This notion appears to be related to the overall quantity and cellularity of the intra-tumoural stroma and the predominant constituents of the extracellular matrix.

Conclusion

Increasing evidence is emerging showing that tumour-stroma interactions may not only affect tumour progression and patient prognosis, but also the response to chemotherapy. The tumour stroma-derived biomarkers that appear to be most appropriate to determine the patient’s response to chemotherapy vary by tumour origin and the availability of pre-treatment tissue. For patients scheduled for adjuvant chemotherapy, the most promising biomarker appears to be the PLAU: SERPINE complex, whereas for patients scheduled for neo-adjuvant chemotherapy the tumour stroma quantity appears to be most relevant.

Keywords

Tumour-microenvironment Stroma Fibroblasts Chemotherapy response Prognosis Extracellular matrix 

Abbreviations

CAF

cancer associated fibroblast

CMF

cyclophosphamide methotrexate and 5-fluorouracil

CTGF

connective tissue growth factor

ECM

extracellular matrix

EFEMP1

fibulin 3

ELISA

enzyme linked immunosorbent assay

FBLN1

fibulin 1

FN1

fibronectin

HA

hyaluronan

HGF

hepatocyte growth factor

MMP

matrix metalloproteinase

PTK

protein tyrosine kinase

SCLC

small cell lung cancer

SDC1

syndecan 1

SERPINE1

serine protease inhibitor type-1

SPARC

secreted protein, acidic, cysteine-rich

TIMP1

tissue inhibitor of metalloproteinase-1

PLAU

urokinase-type plasminogen activator

1 Introduction

Cancer-associated fibroblasts (CAFs) can be derived from different precursor cell types, and the development of CAF precursors into CAFs may result from various genetic and micro-environmental factors (depicted in Fig. 1). The proteins produced by CAFs may stimulate tumour cells to acquire various cancer-associated characteristics [1]. An overview of these proteins and their resulting characteristics is shown in Fig. 2. In addition, many of the structural and secreted proteins produced by CAFs have been related to patient prognosis (summarised in Table 1). The aim of the current review was to assess putative associations between tumour stroma quantity, the proteins produced by CAFs, and the patient’s response to cytotoxic chemotherapy. The Medline database (1946–2012) was searched using the terms listed in Table 2. Publications were only included when they referred to epithelial cancers and cytotoxic chemotherapy, irrespective whether this chemotherapy was given alone, or before or after surgery. Publications including patients exposed to both chemotherapy and radiotherapy were excluded. Biomarkers were selected for review from those known to be associated with cancer-related characteristics [1] and known to effect patient prognosis. These biomarkers were identified by reviewing publications dealing with CAFs and the references therein.
Fig. 1

Origins of cancer associated fibroblasts. The top left box denotes the factors of normal tissues and their secreted proteins which stimulate local fibroblasts, fibroblast precursors and myofibroblasts to become CAFs. The bottom left box denotes additional cell types that can also become CAFs and the mechanisms by which this occurs. The right side of this diagram (denoted in orange) shows the positive feedback contribution of the tumour cells and the already established CAFs in further promoting CAF formation. Note that although osteopontin is usually produced by fibroblasts, some tumour cells are also capable of producing osteopontin. Although it is commonly accepted that CAFs and reactive fibroblasts can further stimulate CAF production, the exact contributions of chemokines and growth factors to this process are currently unknown. Abbreviations used: C: complement factor; CAF: cancer associated fibroblasts; EGF: epidermal growth factor; EMT: epithelial mesenchymal transition; FGF: fibroblast growth factor; ICAM: intracellular-adhesion molecule; MCP: monocyte chemotactic protein; MMPs: matrix metalloproteinases; PDGF: platelet derived growth factor; ROS: reactive oxygen species; TGF-β: transforming growth factor β; VCAM: vascular cell adhesion molecule. Different arrows represent the extent of the different contributions: Open image in new window indicates a probable contribution, Open image in new window indicates a definite contribution and Open image in new window indicates the main contribution. This diagram is composed of information retrieved from references [22, 34, 37, 48, 52, 135] and references therein

Fig. 2

Influence of CAFs on cancer progression by stimulating tumour proliferation, inflammation, angiogenesis, metastasis, invasion and necrosis. This diagram displays the products of CAFs and their role in stimulating the development of the ‘hallmarks’ of cancer. Abbreviations used: CCL5 (RANTES): C-C family cytokine ligand 5; COX: cyclo-oxygenase; CTGF: connective tissue growth factor; CXCL: CXC chemokine ligand; CXCR: CXC chemokine receptor; ECM: extracellular matrix; EDF: epidermal growth factor; EMT: epithelial mesenchymal transition; FGF: fibroblast growth factor; FSP: fibroblast specific protein; HGF: hepatocyte growth factor; IGF: insulin-like growth factor; IL: interleukin; MCP: monocyte chemotactic protein; MMP: matrix metalloproteinase; PLAU urokinase-type plasminogen activator; SDC1: syndecan 1; SDF: stromal derived factor; SFRP: secreted frizzle-related proteins; SERPINE1: serine protease inhibitor type 1; TGF-β: transforming growth factor β; VEGF: vascular endothelial growth factor. This diagram is composed of information retrieved from references [22, 34, 37, 48, 49, 50, 51, 95, 123, 136, 137] and references therein

Table 1

Proteins produced by CAFs measured in tumour tissue with a known impact on patient prognosis

Factor

Effect of high expression of factor on prognosis

Analysis type

Reference

Proteins

α-SMA

Worse

Univariate

[33, 138, 139]

α-SMA : collagen deposition ratio

Worse

Multivariate

[140]

Collagen IV in the BM

Worse

Univariate

[42]

CTGF

Improved

Univariate

[65]

FAP

Worse

Univariate

[33]

Fibronectin

Worse

Univariate

[35]

HGF

Worse

Multivariate

[87]

HIF-1α

Worse

Univariate

[141, 142, 143]

Hyaluronan

Worse

Multivariate

[84, 85]

Laminin in the BM

Worse

Univariate

[42]

Lysosomal proteases

Improved

Univariate

[144]

MMP1

Worse

Multivariate

[96]

MMP7

Worse

Multivariate

[95]

Periostin

Worse

Multivariate

[145]

Podoplanin

Improved

Multivariate

[146]

SDC1

Worse

Multivariate

[123, 124, 147]

SPARC

Worse

Multivariate

[148]

TIMP1

Worse

Multivariate

[136, 149, 150]

PLAU

Worse

Multivariate

[137, 151, 152, 153, 154]

SERPINE1

Worse

Multivariate

[137, 151, 152, 153, 154]

Vimentin

Worse

Multivariate

[155, 156, 157, 158]

Structural components

Fibrotic foci

Worse

Multivariate

[54, 159]

Quantity of tumour stroma

Worse

Multivariate

[42, 46, 47, 158, 160, 161]

TSR

Improved

Multivariate

[31, 43, 44, 45, 162]

α-SMA alpha smooth muscle actin, BM basement membrane, CTGF connective tissue growth factor, FAP fibroblast activating protein, HGF hepatocyte growth factor, HIF hypoxia induced factor, MMP1 matrixmetalloproteinase-1, MMP7 matrixmetalloproteinase-7, SDC1 syndecan 1, SERPINE1 serine protease inhibitor type 1, SPARC secreted protein, acidic, rich in cysteine, TIMP1 tissue inhibitor of metalloproteinase-1, TSR tumour stroma ratio, PLAU urokinase-type plasminogen activator

Table 2

Terms and strategy used in Medline literature search. Outline of the search strategy used for this review and the terms used to search for each subject

 

Search Terms

Broad tumour stroma searches

Tumour stroma

“desmoplasia” OR “ECM” OR “extracellular matrix” OR “microenvironment” OR

“stroma” OR “stromal collagen” OR “stromal signature” OR “tumor cell density” OR

“tumour density”

Structural component searches

Tumour stroma ratio

“histology” OR “morphological” OR “morphology” OR “reactive stroma grade” OR

“relative stroma abundance” OR “relative stromal abundance” OR “TCD” OR

“TSR” OR “tumor cell density” OR “tumor density” OR “tumor stroma ratio” OR

“tumour cell density” OR “tumour density” OR “tumour stroma ratio”

Fibrotic foci

“fibrotic foci”

Fibroblast searches

Cancer associated fibroblasts

“carcinoma associated fibroblast” OR “carcinoma associated fibroblasts” OR “carcinoma-associated fibroblast” OR “carcinoma-associated fibroblasts” OR “activated fibroblast” OR “activated fibroblasts” OR “cancer associated fibroblast” OR “cancer associated fibroblasts” OR “fibroblast” OR “fibroblasts” OR “myofibroblast” OR “myofibroblasts”

ECM component Searches

Proteins produced by cancer associated fibroblasts

“aggrecan” OR “basic fibroblast growth factor” OR “bFGF” OR “caveolin” OR “collagen” OR “connective tissue growth factor” OR “fibrin” OR “fibronectin” OR “fibrous” OR “fibulin” OR “fibroblast activating protein” OR “hepatocyte growth factor” OR “HGF” OR “hyaluronic acid” OR “hyaluronan” OR “hypoxia induced factor” OR “integrin” OR “integrins” OR “laminin” OR “metalloproteinase” OR “MMP” OR “MMPs” OR “nidogen” OR “osteonectin” OR “osteopontin” OR “PDGF” OR “PDGF receptor kinase inhibitor” OR “PAI” OR “SERPINE1” OR “periostin” OR “podoplanin” OR “proteoglycan” OR “proteoglycans” OR “secreted protein acidic and rich in cysteine” OR “smooth muscle actin” OR “SPARC” OR “stromal derived factor” OR “syndecan” OR “tenascin” OR “TIMP” OR “tissue inhibitor of matrixmetalloproteinase” OR “urokinase type plasminogen activator” OR “PLAU” OR “versican” OR “vimentin”

bFGF basic fibroblast growth factor, ECM extracellular matrix, HGF hepatocyte growth factor, MMP matrix metalloprotease, PAI plasminogen activator inhibitor, SERPINE1 serine protease inhibitor type 1, PDGF platelet derived growth factor, SPARC secreted protein, acidic, rich in cysteine, TCD tumour cell density, TIMP tissue inhibitor of metalloproteinase-1, TSR tumour stroma ratio, PLAU urokinase-type plasminogen activator

2 Cytotoxic chemotherapy

The purpose of cytotoxic chemotherapy is to kill cancer cells, to ablate micro-metastases, to reduce the size of the primary tumour and to improve survival [2]. Cytotoxic chemotherapy is administered as adjuvant, neo-adjuvant, curative or palliative treatment for numerous cancer types [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]. However, not all patients respond equally well to cytotoxic chemotherapy, and several factors have been associated with chemotherapy resistance such as patient age, histological grade of the primary tumour, lymph node status and particular RNA or protein expression profiles [17, 18, 19, 20]. Patients with tumours resistant to cytotoxic chemotherapy may receive ineffective treatment, experience toxicity or suffer from disease progression. Furthermore, a potentially delayed surgical resection could result in incomplete resection (R1 or R2) and, consequently, a worse prognosis.

3 Tumour stroma

A clinically identifiable ‘tumour mass’ is composed of both tumour cells and tumour stroma [21]. The tumour stroma, also referred to as ‘tumour micro-environment’, is the tissue that directly surrounds the tumour cells. Tumour stroma is thought to be essential for tumour cell survival and is increasingly recognised as a key element of the tumour, affecting most of the so-called ‘hallmarks’ of cancer [1, 21, 22]. Tumour stroma is a complex tissue composed of extra-cellular matrix (ECM) components, cellular components (fibroblasts, immune cells and endothelial cells), nerves and vessels. Compared to the normal lamina propria, the tumour stroma contains a larger number of fibroblasts, the so called cancer-associated fibroblasts (CAFs), which represent the major cellular component of the stroma [23, 24, 25, 26]. These fibroblasts have been shown to promote tumour progression [27]. In a number of studies, the quantity of tumour stroma was found to exhibit substantial variation within the same tumour type and between tumours from different origins, as well as variation in its relation to patient prognosis [28, 29, 30, 31]. The quantity of tumour stroma can be measured by morphometry, estimated by visual inspection [32], or measured indirectly by quantifying the proteins expressed by CAFs [33]. For an overview of the different biomarkers used see [34].

It has been suggested that an increase in the quantity of the intra-tumour stroma may be associated with a reduced response to chemotherapy [35, 36] via hampering drug delivery, reducing drug efficacy, or providing a micro-environment that stimulates tumour growth [35, 37, 38, 39]. Additionally, a recent study on breast cancer has suggested that cytotoxic chemotherapy reduces CAF proliferation and increases CAF apoptosis in vitro and in vivo which, in turn, may affect the chemosensitivity of the entire tumour [40]. The effect of chemotherapy on CAF proliferation may depend on the p53 mutation status of the CAFs and may be tumour type specific, as it was not seen in some tumours including those of the lung [40, 41].

3.1 Quantity of intra-tumour stroma

A relative increase in intra-tumour stroma has been associated with a worse prognosis in patients with cancers of the breast, colon and rectum, lung, oesophagus, prostate, and stomach [30, 31, 42, 43, 44, 45, 46, 47]. As of yet, no studies have been published dealing with a putative association between the quantity of intra-tumoural stroma and chemotherapy response in patients undergoing neo-adjuvant chemotherapy. One study has been published investigating tumour stroma in an adjuvant setting in a series of 574 early breast cancer patients who either underwent only a mastectomy or a breast conserving surgery, or a surgery combined with either adjuvant radiotherapy, adjuvant chemotherapy or adjuvant endocrine therapy [31]. The presence of ≥50% tumour stroma in the resection specimens turned out to be an independent poor prognostic marker for relapse-free survival and overall survival when the whole patient cohort was analysed. In the subset of patients who received chemotherapy, the presence of ≥50% tumour stroma in the pre-treatment resection specimens was found to be associated with a shorter relapse-free survival period in a multivariate analysis, but not with overall survival. As there was no statistically significant association between the stroma content and systemic therapy response, the authors suggested that the prognostic effect of the stroma was independent of systemic therapy, i.e., that a high stroma content acts as a poor prognostic factor but does not predict the response to adjuvant therapy in breast cancer patients [31]. Although the prognostic value of the stroma content in this study was in concordance with that of other studies (for reviews see: [22, 34, 37, 48, 49, 50, 51, 52]), the proposed relationship between stroma content and response to adjuvant chemotherapy remains to be confirmed in an additional, independent study.

3.2 Quality of intra-tumour stroma

3.2.1 Fibrotic foci

Fibrotic foci are composed of CAFs and collagen fibres [53]. The presence of fibrotic foci is a strong predictor for a poor prognosis in breast cancer patients treated with surgery only [54, 55, 56, 57]. Fibrotic foci have been related to poor cancer-specific survival in univariate and multivariate analysis when assessed in pre-chemotherapy treatment biopsies from patients with invasive breast carcinoma [3]. Similarly, the presence of fibrotic foci in resection specimens has been significantly associated with lymph node metastases, distant organ metastases and reduced overall survival in breast cancer patients exposed to chemotherapy after surgical resection [53]. However, no relationship between patient outcome and the presence of fibrotic foci in the resection specimens after chemotherapy was noted, which may be related to difficulties in distinguishing fibrotic foci from other post-chemotherapy tissue alterations [3]. Both studies failed to include a surgery-alone group in their analyses for comparison. There is good evidence, however, that the presence of fibrotic foci serves as a strong prognostic factor in breast cancer patients who did not receive chemotherapy [54, 55, 56, 57]. It remains to be established whether the presence of fibrotic foci may also predict the response to chemotherapy.

3.2.2 Cellularity and proliferative status of fibroblasts in fibrotic foci

Interestingly, not only the presence of fibrotic foci appears to be related to patient survival, but also the biological characteristics of the fibrotic foci (i.e., cellularity and proliferative index of fibroblasts) in the surgical resection specimens appear to be associated with time to disease recurrence and survival in breast cancer patients treated with adjuvant chemotherapy [53, 58]. The authors hypothesised that this phenomenon might be related to an increased production of growth promoting factors by proliferating CAFs [53] (see also Fig. 2). Whereas in these studies a patient cohort was used in which some patients were treated with adjuvant chemotherapy and others by surgery alone, the authors did not assess whether there were any differences between these two sub-groups. The cellularity and proliferative status of fibrotic foci have, however, been found to be of prognostic significance in patients receiving adjuvant chemotherapy and they may, therefore, allow the identification of patients who should and those who should not receive adjuvant chemotherapy treatment.

3.3 Cancer-associated fibroblast (CAF) products

Although the expression of the CAF proteins alpha smooth muscle actin, fibroblast activating protein, hypoxia inducible factor, periostin and podoplanin have been associated with patient prognosis (for summary see Table 1), their potential value in predicting patient response to chemotherapy has not been investigated. The potential predictive associations of other CAF proteins with patient responses to chemotherapy are discussed below and listed in Table 3. The known effects of these proteins on the ‘hallmarks’ of cancer are listed in Table 4. Many of the CAF proteins are also produced by cancer cells. The proteins discussed in the current review are, however, predominantly produced by CAFs.
Table 3

Summary of factors with in vivo evidence for predicting cancer response to chemotherapy

Cancer

Variable studied

Analysis

Variable measurement indicating good response and reported effect

P value

Ref.

Breast

Tumour : Stroma Ratio

M

Low = Longer relapse free period

0.038

[31]

IDC

Fibrotic Foci

M

Absence = Less tumour related death

0.002

[3]

IDC

Fibrotic Foci

M

Absence = Less lymph node metastasis

0.011

[53]

IDC

Fibrotic Foci

M

Absence = Less distant organ metastasis

0.014

[53]

IDC

Fibrotic Foci

M

Absence = Longer overall survival

0.045

[53]

IDC

MIB-1 LI (%) of fibroblasts forming FFs

M

Low = Less lymph node metastasis

<0.001

[53]

IDC

MIB-1 LI (%) of fibroblasts forming FFs

M

Low = Less distant organ metastasis

0.009

[53]

IDC

Fibroblast : collage fibre ratio of FF

M

High = Less frequent lymph node recurrence

0.011

[53]

IDC

MIB-1 LI of fibroblasts in IDC w/o FFs

U

Low = Less lymph node metastasis

0.007

[53]

IDC

MIB-1 LI of fibroblasts in IDC w/o FFs

U

Low = Less distant organ metastasis

0.029

[53]

IDC

MIB-1 LI of fibroblasts in IDC w/o FFs

U

Low = Longer overall survival

0.012

[53]

IDC

MIB-1 LI (%) of fibroblasts in IDCs w/FFs

M

Low = Less distant organ metastasis

<0.001

[58]

IDC

MIB-1 LI (%) of fibroblasts in IDCs w/FFs

M

Low = Less distant organ metastasis

0.007

[58]

SCLC

Fibronectin matrix

U

Low = Longer overall survival

<0.05

[35]

Breast

CTGF expression

U

Low = Greater chemotherapy response

<0.05

[66]

Breast

FBLN3

M

Low = Shorter overall survival

0.037

[76]

Breast

HA

M

Low = Longer overall survival

0.006

[84]

NSCLC

MMP7

U

Low = Less chemotherapy resistance

0.036

[95]

NSCLC

MMP7

M

Low = Longer overall survival

0.001

[95]

Gastric

SPARC expression

M

High = Less early progressive disease

0.042

[115]

Gastric

SPARC expression

M

High = Longer overall survival

0.01

[115]

Breast

Syndecan 1

M

Low = Longer overall survival

0.03

[123]

Breast

Log TIMP1

M

Low = Greater chemotherapy response

0.03

[103]

Breast

PLAU & SERPINE1

M

Low both = Shorter disease free survival

0.003

[129]

Breast

PLAU: SERPINE1 complex

M

Low both = Reduced tumour recurrence

<0.001

[132]

CTGF connective tissue growth factor, FBLN3 fibulin 3, IDC invasive ductal carcinoma, M multivariate analysis, MIB LI (%) MIB labelling index, MMP matrix metalloproteinase, NSCLC non-small cell lung cancer, SDC1 syndecan 1, SERPINE1 serine protease inhibitor type 1, SCLC small cell lung cancer, SPARC secreted protein, acidic, rich in cysteine, TIMP1 tissue inhibitor of metalloproteinase-1, U univariate analysis, PLAU urokinase-type plasminogen activator

MIB LI (%) is a labelling index calculated as the percentage of fibroblasts which stain positive for Ki-67 antigen after immunohistochemistry and serves, therefore, as a measure of cell proliferation

Table 4

Influence of CAF products on the ‘hallmarks’ of cancer in patients treated with chemotherapy

CAF product

Effect of high expression of factor

Ref.

CTGF

Reduced apoptosis

[66]

FN1

Reduced apoptosis

[35, 71, 72, 163, 164]

LAMB1

Reduced apoptosis

[35, 71, 72, 163, 164]

Collagen

Reduced apoptosis

[35, 71, 72, 163, 164]

FBLN1

Reduced apoptosis

[75]

HA

Increased metastasis

[84, 85]

Reduced apoptosis

[83]

HGF

Increased apoptosis

[91]

Reduced apoptosis

[165]

MMP7

Increased apoptosis

[99]

Reduced apoptosis

[98]

SPARC

Increased apoptosis

[110, 111, 112]

SDC1

De-differentiation

[122]

Metastasis

[122]

TIMP1

Reduced apoptosis

[103]

No change in patient prognosis

[106]

PLAU: SERPINE1 ratio

Increased cell migration

[127]

Increased cell adhesion

[127]

CTGF connective tissue growth factor, FBLN1 fibulin 1, FN1 fibronectin 1, HA hyaluronan, HGF hepatocyte growth factor, LAMB1 laminin 1, MMP1 matrix metalloprotein-1, MMP7 matrix metalloproteinase-7, PLAU urokinase-type plasminogen activator, SDC1 syndecan 1, SERPINE1 serine protease inhibitor type 1, SPARC secreted protein, acidic, rich in cysteine, TIMP1 tissue inhibitor of metalloproteinase-1

3.3.1 Connective tissue growth factor

Connective tissue growth factor (CTGF) is expressed by several different stromal cell types and stimulates cell proliferation, migration, adhesion, chemotaxis, differentiation and angiogenesis [59, 60, 61]. Studies in different tumour tissues have shown that CAFs are the predominant cell type producing CTGF [59, 61, 62, 63, 64, 65]. Wang et al. found in a small series of breast cancers that low CTGF expression, measured in the post-chemotherapy breast cancer resection specimens, was associated with a better response to chemotherapy which, in turn, appeared to be related to a down-regulation of the apoptosis regulators Bcl-xL and cIAP1 [66]. The authors confirmed the putative relationship between CTGF, Bcl-xL and cIAP1 and response to therapy in several breast cancer cell lines. As this study only dealt with CTGF expression after chemotherapy, the effect of physiologically increased CTGF levels due to cell damage and wound healing on the overall level in the tumour remains to be established [67]. No information was provided in this study on the expression level of CTGF in the intra-tumoural stromal cells. Preliminary functional studies, carried out in cancer cell lines, indicate that CTGF expression by the tumour cells may be related to chemotherapy resistance, thus warranting further studies in other primary tumour samples.

3.3.2 Fibronectin, laminin and collagen

Fibronectin (FN1), laminins and collagens are the main structural proteins in the ECM (for review see [34]). These proteins are predominantly produced by fibroblasts, although there is some evidence that epithelial cancer cells may produce them as well [35, 68]. Sethi et al. [35] showed that an extensive deposition of FN1, collagen IV and tenascin in small cell lung cancers (SCLC), and an increased adherence of cancer cell lines to ECM proteins, protects the tumour cells against chemotherapy-induced apoptosis due to an increased protein tyrosine kinase (PTK) activation and a subsequent inhibition of caspase activation [35]. The observed adhesion of tumour cells to the ECM appeared to be mediated by β1 integrin in SCLC cell lines. Such functional information may have relevance in vivo since a small study of 16 patients with SCLC, who all received chemotherapy, showed that patients with a FN1-rich ECM within the resection specimens exhibited a significantly shorter survival time than patients with a FN1-depleted ECM [35] (see Table 3). It is unknown whether the chemotherapy was given as neo-adjuvant, adjuvant or definitive treatment in this latter study. Additional in vitro studies using bladder, prostate, lung and pancreatic cancer cell lines have confirmed that β1 integrin-mediated adhesion to ECM proteins confers chemoresistance to most of the cell lines investigated [69, 70, 71, 72]. These results strongly suggest that increased expression of β1 integrin may serve as a marker for chemoresistance, which warrants its further investigation in pre-treatment material. Targeting β1 integrin-mediated ECM adhesion may also be a way to reverse chemoresistance.

3.3.3 Fibulins

Fibulins encompass a family of secreted glycoproteins that are associated with basement membranes and elastic ECM fibres [73]. Fibulins are incorporated into fibronectin-containing ECM fibres, thereby contributing to the supra-molecular organisation of the ECM architecture (for review see [73]). All members of the fibulin family have been associated with both pro- and anti-tumour effects in a number of different cancer types (for review see [74]). The relationship between fibulins and response to chemotherapy has only been investigated for fibulin 1 (FBLN1) in a single breast cancer cell line study [75], and for fibulin 3 (EFEMP1) in a single breast cancer patient study [76]. In view of the interaction of FBLN1 with fibronectin, laminin 1, fibrinogen, nidogen and the proteoglycans aggrecan and versican [77], it was not surprising to find that breast cancer cell lines grown on Matrigel, which contains among others the ECM protein FBLN1, exhibited resistance to doxorubicin-induced apoptosis [75]. The authors confirmed that this effect was related to FBLN1 by blocking FBLN1 with monoclonal antibodies or suppressing FBLN1 expression using siRNA, both of which resulted in an increased chemosensitivity [75]. The role of EFEMP1 is currently less well understood than that of FBLN1, but it is thought to antagonise tumour angiogenesis [78]. In an in vivo study in 203 primary breast carcinomas, high EFEMP1 expression was found to be correlated with a worse disease-free and shorter overall survival than low EFEMP1 expression in patients receiving adjuvant anthracycline therapy, but not in those receiving combined chemotherapy of cyclophosphamide, methotrexate and 5-flourouracil (CMF) [76].

3.3.4 Hyaluronan

Hyaluronan (HA) is a glycosaminoglycan present in connective, epithelial and neural tissues and is essential for prenatal development (for review see [79]). Interestingly, cells from normal, single layered epithelia, such as those of the gastrointestinal tract, do not produce HA, and only minute amounts have been found in its associated stroma. In contrast, stratified epithelia, such as squamous epithelia, normally produce HA and have a HA-rich stroma (for review see [80]). HA expression in intra-tumour stroma, but also in the tumour cells themselves, has been associated with invasion, angiogenesis and local and distant metastases in a number of different tumour types (for review see [79]). Due to its physicochemical properties, HA can bind large amounts of water and form viscous gels. As such, it can act as a filter at the molecular level. Based on these properties, HA has been used as a chemotherapy drug carrier in colorectal and lung cancer patients [81, 82].

HA expression may also be used to predict the response to adjuvant therapy. In vitro experiments in breast cancer cell lines have suggested that a constitutive interaction between HA and CD44 may confer resistance to doxorubicin, which can be reversed using HA antagonists [83]. Furthermore, HA appears to be involved in drug resistance by regulating the PI3K/AKT pathway and, subsequently, increasing the expression of ABCB1 [83]. These results indicate (i) that HA may serve as a potential target in multidrug-resistant carcinomas, and (ii) that HA expression may serve as a useful biomarker to predict the response to chemotherapy. So far, two studies on clinical material from breast cancer and lung adenocarcinoma patients, who were treated by primary resection followed by adjuvant therapy in some of the cases, showed that a high HA expression level in the stroma, but not in the tumour cells, may serve as a poor prognostic factor [84, 85].

3.3.5 Hepatocyte growth factor

Hepatocyte growth factor (HGF) is a protein produced by CAFs that has been implicated in regulating growth, motility and morphogenesis in multiple cell types [86]. Increased HGF expression measured in breast cancer extracts has been shown to predict a reduced relapse-free period and a reduced overall survival, independent of post-operative systemic therapy [87]. Concordantly, HGF protects cells from apoptosis in most physiological and pathological conditions, including cancer [88, 89, 90]. HGF has also been shown to increase the sensitivity of ovarian cancer cell lines to apoptosis induced by chemotherapy by phosphorylation of its receptor HGFR/MET, and by activating a caspase-dependent apoptotic pathway [91]. This latter observation suggests that a combination therapy of HGF and a chemotherapeutic drug, such as cisplatin, may reverse chemotherapy resistance and, simultaneously, protect normal cells from death. It is currently unknown why HGF increases sensitivity to cisplatin in ovarian cancer cell lines but not in other cell lines such as those derived from gastric and colorectal cancers [91].

3.3.6 Matrix metalloproteinase-7

Matrix metalloproteinases (MMPs) comprise a family of 23 zinc-dependent endopeptidases, which have been associated with tumour invasion and progression due to their ability to degrade the ECM and the basement membrane (for review see [92]). MMPs have been shown to be expressed in both tumour and stroma cells in lung and colorectal cancers [93, 94]. The role of MMPs in the tumour micro-environment is very complex and subject of intensive research, despite a lack of efficacy of MMP inhibitors in clinical settings [92]. High expression of MMP1 and MMP7 in cancer tissues has been associated with a poor prognosis [95, 96] (see Table 1). An association with response to chemotherapy has so far only been suggested for MMP7, and in vitro experiments have shown that MMP7 over-expressing cells are less sensitive to chemically induced apoptosis [97]. Similarly, MMP7 over-expressing colon cancer cell lines were found to be resistant to doxorubicin [98], and oxaliplatin resistance in colon cancer cell lines could be reversed by MMP7 inhibition [99]. Liu et al. studied MMP7 expression in pre-treatment tumour tissues from 159 patients with non small cell lung cancer who were subsequently treated with platinum-based neo-adjuvant chemotherapy [95]. In this study, low MMP7 expression was found to be related to an improved response to chemotherapy and an improved overall survival, thus confirming the in vitro experiments (see above). Together, these observations suggest that MMP7 may serve as a useful marker for both prognosis and response to chemotherapy, and as a potential therapeutic target.

3.3.7 Tissue Inhibitor of Metalloproteinase-1

Tissue inhibitor of metalloproteinase-1 (TIMP1) is one of the four endogenous protease inhibitors with a complex mode of action. TIMPs inhibit MMP-mediated degradation of the ECM, stimulate cell growth and inhibit apoptosis (for review see [100, 101]). In colorectal carcinomas, TIMP1 expression has almost exclusively been found in CAFs located at the invasive front of the tumour. In contrast, TIMP1 expression was rarely found in the stroma of normal colorectal mucosa, and was found to be absent in benign or malignant epithelial cells [102]. High TIMP1 expression in pre-treatment resection specimens and TIMP1 plasma levels were related to a poor response to chemotherapy, as well as poor disease-free and poor overall survival in patients with metastatic breast cancer [103, 104] and colorectal cancer, respectively [105], but not in patients with ovarian cancer [106]. These findings suggest that the predictive value of TIMP1 expression may be tumour type specific and depend on the type of chemotherapy applied, the method used to evaluate the response to chemotherapy, and how TIMP1 expression was determined. These factors varied in each of the studies reviewed [103, 104, 105, 106]. As these studies indicate that TIMP1 expression is highly specific for CAFs and can be measured in blood samples, thus obviating the need to obtain tissue biopsies, further studies into the predictive value of TIMP1 in different tumour types are warranted.

3.3.8 Secreted protein, acidic, cysteine rich

Secreted protein acidic and rich in cysteine (SPARC, also known as osteonectin or BM-40) is a protein that binds to fibrillar collagen and collagen in the basement membrane, thereby ensuring appropriate ECM assembly without participating in the structural ECM scaffold. In addition, SPARC has a number of other functions, including regulating the activity of MMPs, modulating growth factor signalling by cell surface receptors, preventing apoptosis in some but not all cells, inducing counter-adhesive properties, influencing cell proliferation, and exhibiting a potential tumour suppressor activity (for recent reviews see [107] and [108], and references therein). An observed relationship between SPARC expression and prognosis in patients with gastric cancer treated with chemotherapy is depicted in Table 1.

The expression pattern of SPARC, i.e., whether it is expressed by CAFs or tumour cells, appears to vary between different cancer types [108]. It has recently been shown that high levels of SPARC expression may be related to a better response to chemotherapy in tumours of the colon, breast, liver and pancreas through activating cellular senescence and enhancing apoptosis [109, 110, 111, 112], thereby supporting data from a previous study which indicated that therapy resistant colorectal cancers frequently exhibit low levels of SPARC expression [113]. Phelps et al. [114] showed, using platinum-resistant ovarian tumour cell xenografts, that (i) tumours were larger and mice lived shorter in SPARC null mice confirming that SPARC acts as a tumour suppressor and (ii) cisplatin therapy was more effective in tumours of SPARC null mice, the mechanisms of which remains to be elucidated. In contrast, high SPARC expression in 80 primary gastric cancers treated with docetaxel-based chemotherapy was found to be related to a worse patient survival in a multivariate analysis [115]. Whilst this study on gastric cancer suggests that the SPARC expression level could represent a useful prognostic marker, no clear conclusions can be drawn regarding the predictive value of SPARC expression as no control group was included. Also, information regarding the cell type which expresses SPARC was not provided in this study.

The mechanism by which SPARC expression may influence chemosensitivity is currently not fully understood, which is mainly due to the diverse biological functions of SPARC (for review see [107]). High SPARC expression has been shown to inhibit the proliferation of endothelial, fibroblastic, mesenchymal and smooth muscle cells in vitro [116, 117, 118]. Based on this information, one could speculate that SPARC may modulate the response to chemotherapy via the regulation of the proliferation of stromal cells, thereby reducing proteins produced by these cells, which are known to influence the response to chemotherapy.

3.3.9 Syndecan 1

Syndecan 1 (SDC1), also known as CD138, is a cell surface proteoglycan that regulates cell migration, cell-cell and cell-matrix interactions, and stimulates protease activation [119, 120]. Studies in SDC1 knock-out mice have suggested that SDC1 expression by CAFs is required for the growth-promoting effect of CAFs on breast carcinoma cells [121]. SDC1 expression was found to be increased in tumour cells and the surrounding tumour stroma in breast cancers, and it has been suggested that SDC1 may contribute to the de-differentiation of tumour cells and the acquisition of a metastatic phenotype [122]. High stromal expression of SDC1 in breast cancer patients before resection was found to be significantly associated with a worse survival after subsequent treatment with adjuvant chemotherapy [123] (see Table 3). A conflicting study has, however, reported that patients with cancers positive for stromal SDC1 appeared to exhibit a better response than those without stromal SDC1 [124]. Any effect of SDC1 expression on patient response to chemotherapy may be due to an interaction between SDC1 and Ki67 and, in addition, expression of the progesterone receptor. High expression of SDC1 in intra-tumour stroma has been associated with a high tumour cell Ki67 index, a low tumour cell progesterone receptor expression and a low overall survival in breast cancer patients treated with adjuvant chemotherapy [125]. Since SDC1 may be uniquely related to predictive markers of both the tumour and its stroma, these findings may enhance the value of SDC1 as a predictive marker.

3.3.10 Urokinase-type plasminogen activator

Urokinase-type plasminogen activator (PLAU) is a serine protease that cleaves plasminogen to form the active enzyme plasmin, one of several enzymes required for normal physiological processes such as tissue remodelling [126]. PLAU and its inhibitor, serine protease inhibitor type 1 (SERPINE1) are expressed by both tumour cells and CAFs. It has been suggested that a combined expression of PLAU and SERPINE1 by CAFs has a higher impact on cancer prognosis than either factor independently [127]. Most studies dealing with PLAU and SERPINE1 expression in vivo have employed enzyme-linked immunosorbent assays (ELISAs) in fresh tissues [126, 128, 129, 130, 131]. By using this type of assay, however, it is impossible to discriminate PLAU and SERPINE1 expression from cancer cells or CAFs.

The putative association between PLAU and/or SERPINE1 expression levels, measured in pre-chemotherapy resection specimens, and responses to chemotherapy in primary breast cancers is still controversial. One study has indicated that high expression levels of PLAU and/or SERPINE1 in the pre-treatment resection specimens predict a longer disease-free survival or a reduced recurrence rate [129], whereas another study has indicated that patients with high expression levels of PLAU and/or SERPINE1 in the pre-treatment resection specimens exhibit a shorter overall survival and an increased disease recurrence [128]. A third study found no association between PLAU or SERPINE1 expression levels and cancer recurrence [132]. In contrast, it was found that a high PLAU:SERPINE1 complex expression level was associated with a significantly reduced recurrence rate as compared to a low expression level of this complex [132]. The effect of the PLAU: SERPINE1 complex expression was found to be comparable to that of PLAU and SERPINE1 combined [132].

Interestingly, patients with a high PLAU and/or SERPINE1 expression level in the pre-treatment resection specimens have been shown to exhibit a (non-significant) trend towards a longer survival when treated with adjuvant anthracycline chemotherapy, whereas no change in survival was observed after treatment with adjuvant CMF. This latter finding suggests that a high PLAU and/or SERPINE1 expression level in pre-treatment specimens may be used to determine which patients will benefit from an anthracycline chemotherapy regimen [131]. This notion may have important implications, as anthracycline treatment may indeed be more effective than CMF treatment, but also carries a higher risk of inducing cardiotoxicity and secondary cancers [131]. Selecting out poor responders will thus be essential. The association of PLAU and SERPINE1 expression with patients’ responses to 5-FU, epirubicin and cyclophosphamide and anthracycline-taxane adjuvant chemotherapy is currently being investigated in the NNBC1 trial [126].

4 Conclusions and perspectives

Cytotoxic chemotherapy has been used to treat cancer for more than two decades now. Considering the narrow therapeutic index and the danger of potentially life-threatening toxicity, there is an urgent clinical need to identify biomarkers that can be used to predict which patients will benefit most from cytotoxic chemotherapy. Studies in the past focusing on the characteristics of cancer cells have not led to the identification of a clinically useful marker to predict patient response to cytotoxic chemotherapy. More recently, it has been recognised that the intra-tumour stroma contributes significantly to most of the so-called ‘hallmarks’ of cancer [21] and a number of studies have uncovered the intra-tumour stroma, or components thereof, as prognostic factors in cancer patients (listed in Tables 1 and 3). It has been suggested that increases in intra-tumour stroma quantity may be related to the response to chemotherapy [133] by reducing drug delivery (for a recent review see [36]), reducing drug efficacy [35], or by providing a micro-environment that stimulates tumour growth [37].

The aim of this review was to evaluate current knowledge on the role of intra-tumour stroma and proteins produced by CAFs in chemotherapy response. The main findings are summarised in Table 5. So far, most studies confirmed an association between intra-tumour stroma and prognosis in patients treated by surgery alone or by surgery followed by adjuvant chemotherapy. This association appears to be independent of the biomarker used (see Table 5). This latter observation could be explained by the fact that an increase in any one factor produced by CAFs may result in an increase in other CAF factors, as has been shown in SCLC [35]. Some of the CAF factors have also been shown to predict chemotherapy response in in vitro experiments (see Table 5).
Table 5

Potential biomarkers for predicting patient response to cytotoxic chemotherapy

Factor

Originates from intra-tumour stroma?

Prognostic for surgery alone

Prognostic for surgery + chemotherapy

Effects chemotherapy response in vitro

Method of testing

References

Quantity of intra-tumour stroma

Exclusively

Yes

Yes

Not investigated to date

H&E (FFPE)

[30, 31, 42, 43, 44, 45, 46, 47, 158, 160, 161]

Presence of fibrotic foci

Exclusively

Yes

Yes

Not investigated to date

H&E (FFPE)

[3, 53, 54, 55, 56, 57]

Cellularity and proliferative status of fibroblasts in fibrotic foci

Exclusively

Yes

Yes

Not investigated to date

IHC (FFPE)

[53, 58, 159]

CTGF

Predominantly

Yes

Yes

Yes

IHC (FFPE)

[59, 61, 62, 63, 64, 65, 66]

FN

Exclusively

Yes

Yes

Yes

IHC (FFPE)

[35, 68, 69, 70, 71, 72]

FBLN

Exclusively

Yes

Yes

Yes

IHC (FFPE)

[75, 76]

HA

Conflicting results

Yes

Yes

Yes

Biotinylated hyaluronan probe (FFPE)

[83, 84, 85]

HGF

Predominantlya

Yes

Not investigated to date

Conflicting results

ELISA (fresh tissue)

[87, 91]

MMP7

Predominantly

Yes

Yes

Yes

IHC (FFPE)

[86, 93, 94, 95, 96, 97, 98, 99]

TIMP1

Predominantly

Yes

Yes

Conflicting results

Plasma

[102, 103, 104, 105, 106, 136, 140, 150]

SPARC

Conflicting results

Yes

Yes

Yes

IHC (FFPE)

[108,–115,148]

SDC1

Predominantlya

Yes

Conflicting results

Yes

ELISA (fresh tissue)

[121, 122, 123, 124]

PLAU: SERPINE1

Predominantlya

Yes

Yes

Yes

ELISA (fresh tissue)

[129, 131, 132, 137, 151, 152, 153, 154]

aThe use of the ELISA test prevents differentiation between whether the factor measured is derived from the tumour or the intra-tumour stroma.

CTGF connective tissue growth factor, ELISA enzyme-linked immunosorbent assay, FBLN fibulin, FFPE formaldehyde fixed paraffin embedded tissue, FN fibronectin, H&E haematoxylin and eosin stained tissue, HA hyaluronan, HGF hepatocyte growth factor, IHC immunohistochemistry, MMP7 matrix metalloprotein-7, PLAU:SERPINE1 the quantity of urokinase plasminogen activator and serine protease inhibitor type 1 complex, SDC1 syndecan 1, SPARC secreted protein, acidic, rich in cysteine, TIMP1 tissue inhibitor of metalloproteinase-1

Upon review of the literature it appears that the effect of the quality of intra-tumour stroma on the response to chemotherapy is predominantly determined by a reduction in drug efficacy, i.e., by preventing apoptosis (see Table 4), rather than by providing a tumourigenic micro-environment or by reducing drug delivery as has previously been suggested [36, 37]. Despite the various studies reviewed, it is obvious that intra-tumour stroma and CAF proteins represent a still under-investigated resource for determining the prognosis of patients treated with chemotherapy. In light of the evidence supporting many of the factors included in this review, and the possibility that these factors may be surrogate markers of CAF quantity, we propose that the factors selected for introduction into clinical practice would be best determined based on the ease of the test protocol, the cost of testing and the material that is required for testing (listed in Table 5). The measurement of PLAU, the PLAU inhibitor SERPINE1, the PLAU: SERPINE1 complex, HGF and SDC1 levels appear to be valuable evidence-based tests to predict the success of treatment with adjuvant chemotherapy. Of these, the PLAU:SERPINE1 complex is most evidence-based. However, the ELISA-based assay for this complex requires substantial amounts of fresh/frozen tissue [134]. For patients treated with neo-adjuvant, definitive or palliative chemotherapy, where fresh/frozen tissue is usually not available, an alternative biomaker is required. For these patients one intriguing concept is that a simple quantitative assessment of the amount and cellularity of the intra-tumoural stroma, which can be performed using routine haematoxylin/eosin stained slides, may predict the response to chemotherapy. Studies in breast cancer have shown that a quantification of the stroma, or an assessment for the presence or absence of fibrotic foci and the cellularity within these foci, could both be related to the response to chemotherapy [3, 31, 53]. As this type of measurement can easily be implemented in a routine pathology setting using formalin fixed paraffin embedded tissue, it would be worth pursuing this option in a large prospective study in different cancer types in order to validate the preliminary findings from the breast cancer studies. The link between intra-tumour stroma content and therapy resistance appears to impinge on the adhesion of cells to the ECM via integrins and the subsequent prevention of apoptosis [35, 69, 70, 71, 72] (see Table 4). Future studies should ideally fulfil the key criteria outlined in Box 1. However, consideration needs to be given to the fact that cytotoxic chemotherapy is currently included in the standard care for many cancer types. This may hamper the assessment of potential prognostic factors in prospective randomised clinical trials with a surgery-alone control arm. Such studies will, however, be possible using archival material.

Box 1. Recommendations for future in vivo studies investigating factors predictive of patient response to chemotherapy:

1. Include a sample of adequate size to determine statistical significance

2. If possible, use matched pre-treatment and post-treatment tissue samples to determine the feature of interest

3. Ensure that a suitable control arm is included to enable determination of whether a change in the feature of interest has a prognostic or predictive value

4. Ensure that further independent studies are performed to validate the findings

In summary, the current lack of a clinical test to identify patients that will or will not respond to cytotoxic chemotherapy leads to unnecessary toxicity and treatment failure. The intra-tumour stroma has been shown to contain a number of highly specific biomarkers (listed in Table 5) with a clear clinical utility to determine the prognosis of patients treated with surgery alone or with chemotherapy.

Acknowledgements

This work was funded by the Jean Shanks Foundation.

Conflict of interest

The authors declare that they have no conflict of interest.

Copyright information

© International Society for Cellular Oncology 2013

Authors and Affiliations

  • Matthew David Hale
    • 1
  • Jeremy David Hayden
    • 2
  • Heike Irmgard Grabsch
    • 1
  1. 1.Section of Pathology & Tumour Biology, Leeds Institute of Molecular MedicineUniversity of LeedsLeedsUK
  2. 2.Department of Upper Gastrointestinal SurgeryLeeds Teaching Hospitals NHS TrustLeedsUK

Personalised recommendations