False-negative MRI biomarkers of tumour response to targeted cancer therapeutics.

BACKGROUND: Non-invasive quantitative imaging biomarkers are essential for the evaluation of novel targeted therapeutics. Before deployment in clinical trials, such imaging biomarkers require qualification, typically through pre-clinical identification of imaging-pathology correlates. METHODS: First, in investigating imaging biomarkers of invasion, the response of orthotopic murine PC3 prostate xenografts to the Src inhibitor saracatinib was assessed using susceptibility contrast MRI. Second, the longitudinal response of chemically induced rat mammary adenocarcinomas to the VEGFR2 inhibitor vandetanib was monitored by intrinsic susceptibility MRI, to identify the time window of transient vascular normalisation. RESULTS: No significant differences in fractional blood volume (%), vessel calibre (µm), native T(1) (ms) or apparent water diffusion coefficient were determined, despite reduced expression of activated Fak and paxillin in the saracatinib cohort. Treatment with vandetanib elicited a 60% antitumour response (P<0.01), 80% inhibition in vessel density (P<0.05) and reduction in hypoxia (P<0.05). There was, however, no significant change in tumour baseline R(2)* (s(-1)) or carbogen-induced ΔR(2)* with treatment. CONCLUSION: Reporting negative imaging biomarker responses is important, to avoid the risk of clinical trials using the same biomarkers being undertaken with a false expectation of success, and the abandonment of promising new therapeutics based on a false-negative imaging biomarker response being mistaken for a true-negative.

Src inhibitors in early breast cancer: a methodology, feasibility and variability study.

Early clinical trials of anticancer agents may be enriched by robust biomarkers of activity. Surrogate measures used in trials of cytotoxic agents, such as tumor size regression, may not be informative when investigating targeted agents that act principally to inhibit invasion or proliferation. This study aimed to determine the validity of invasion-related biomarkers of activity for AZD0530, a potent Src inhibitor currently in clinical development. Focal adhesion kinase (FAK) and paxillin are downstream phosphorylation substrates of Src and mediate tumor cell adhesion and invasiveness. These were therefore selected as biologically relevant markers of Src inhibition. Early breast cancer was chosen as a model as multiple samples can be collected during standard treatment and there is an intervening period in which experimental intervention can be applied. Tumor tissue was collected from diagnostic core biopsies and subsequent surgical tumor excision samples in 29 women with early breast cancer attending a single center. Protein levels were assessed quantitatively by Luminex and qualitatively by immunohistochemistry. AZD0530 inhibited tumor growth in a manner independent of dose and inhibited phosphorylation of FAK and paxillin in a dose-dependent manner in a Calu-6 xenograft model. In the clinical study, agreement of within-visit and also of between-visit measurements was high and the estimated number of patients required to detect a drug effect would be low enough to allow use of these markers as endpoints in future dose selection studies.

Amplification and sequencing of genomic breakpoints located within the M-bcr region by Vectorette-mediated polymerase chain reaction.

The polymerase chain reaction (PCR) cannot be used to amplify the breakpoint in the chimaeric BCR-ABL gene in CML and acute leukaemias due to the large variation in the sites of breakpoint in the BCR gene (within a 5.8 kb region) and in the ABL gene (within a 150 kb region). The disease state is usually monitored using RNA-PCR to monitor abnormal transcripts. We have used a new modification of the PCR to amplify breakpoints within zone 3 of the M-bcr. A synthetic oligonucleotide linker, the Vectorette, is ligated to restriction digested DNA, and amplification is carried out between primers for a known target sequence and the Vectorette linker. Three Philadelphia chromosome Ph1-positive CML patients with breakpoints within the ALU region of zone 3 have been amplified and the sequence immediately around the breakpoint determined. The breaks occurred within 70 bp and two were only 14 bp apart. The Vectorette-PCR technique has the potential to rapidly identify and sequence breakpoints, and will enable the design of patient-specific primers to monitor disease progression, particularly following bone marrow transplantation.

Adaptation of the Boyden chamber to flow conditions: rheological effects on chemotaxis.

Although standard Boyden chambers assess chemotaxis under static fluid conditions they are routinely used as an experimental system to model the dynamic events associated with leukocyte extravasation in vivo. We have adapted the Boyden chamber system by incorporating it within the confines of a cone and plate viscometer which can then be employed to generate known shear conditions as the chemotactically responding cells migrate. Our results show that random locomotion of rat peritoneal exudate cells is stimulated under shear conditions within the range of 11.25-90/s relative to that under static conditions. Furthermore, chemotaxis to the synthetic tripeptide F-Met-Leu-Phe (FMLP) is also stimulated under shear conditions with the peak effect occurring near 22.5/s. This adaptation of the standard chamber system to allow the study of chemotaxis under flow conditions may provide further insight on the migratory properties of leukocytes in vivo.

The effects of the B16 malignant melanoma on induced inflammatory responses in host mice: inhibition of macrophage exudation.

C57BL6 mice were injected with B16F1 or B16F10 cells and granulocyte- or macrophage-rich inflammatory responses were subsequently induced. Analysis of cell yields showed that the percentages of macrophages in the inflammatory exudates of tumour-bearing animals were significantly depressed. No significant changes were noted, however, in the levels of granulocytes in peritoneal exudates from tumour-bearing mice when compared to appropriate controls. In parallel studies, the levels of resident macrophages and granulocytes following lavage of non-stimulated peritoneal cavities of mice with and without tumours were monitored as were the levels of blood monocytes and leukocytes. No consistently significant changes were noted in the levels of blood leukocytes or monocytes in normal or tumour-bearing animals. Furthermore, no consistently significant changes were found in the percentages of granulocytes normally resident in the peritoneal cavities of mice with or without the B16 melanoma. It was found, however, that the percentages of resident macrophages in mice bearing the B16 malignant melanoma were significantly depressed relative to those found in normal animals.

Cell adhesion and experimental metastasis: a study using the B16 malignant melanoma model system.

The adhesive properties of B16F1 and B16F10 cells have been studied following their rates of attachment to various substrates and by analysis of their rates of aggregation within the defined environment provided by a cone and plate viscometer. Contrary to previous reports, we found that the B16F10 cells (with significant lung-colonizing potential following tail vein injection) were less homotypically adhesive than B16F1 cells (which colonize lungs poorly). The adhesiveness of B16F10 cells approached that of B16F1 cells only under conditions (low shear, low cell number) where cell collisions were thought to be so few that quantitative differences in aggregation rate could not be determined. B16F1 cells also adhered more to lung cells than did B16F10 cells when assessed by aggregation rate. However, analysis of aggregate composition in which one cell type had been fluorescently labelled showed that B16F10 cells actually formed more mixed aggregates with lung cells than did B16F1 cells. There was no significant difference in the adhesiveness of B16F1 or B16F10 cells to liver cells as assessed by aggregation rate. Analysis of aggregate composition under these circumstances, however, showed that B16F10 cells formed fewer mixed aggregates with liver cells than did B16F1 cells. These results are consistent with the possibility that metastatic cells need to display poor homotypic adhesiveness in order to detach from the primary but enhanced heterotypic adhesiveness in order to colonize specific organs.

The adhesiveness of normal and SV40-transformed BALB/c 3T3 cells: effects of culture density and shear rate.

The adhesiveness of BALB/c 3T3 cells and their SV40 virally transformed counterparts as assessed by aggregation kinetics was found to vary as a function of cell culture density and aggregation conditions. In culture, SV3T3 cells were found to have a faster growth rate and a smaller cell volume than 3T3 cells. Although both cell-types displayed increasing adhesiveness with increasing culture density, the adhesiveness of SV3T3 cells was consistently lower than that of 3T3 cells of comparable culture density when the cells were aggregated under shear rate conditions less than or equal to 45/sec. When the shear rate was increased from 90 to 450/sec, however, the aggregation profile inverted, with the 3T3 cells becoming less adherent than the SV3T3 cells. The ability of the transformed SV3T3 to remain adherent under conditions of relatively high shear may facilitate extravasation during the process of tumour spread.

Cell adhesiveness and the cell cycle: correlation in synchronized Balb/c 3T3 cells.

The adhesiveness of Balb/c 3T3 cells partially synchronized either by drug treatment or by the mitotic cell shake-off procedure was assessed by following the aggregation kinetics of aliquots of the cells suspended under defined conditions within the confines of a cone and plate viscometer. M phase cells synchronized by either method were significantly more adherent than appropriate control cells. Cells arrested in S phase following drug treatment were found to be poorly adhesive whereas G1 phase cells prepared by allowing drug treated M phase cells to proceed through mitosis or by treating them with L-histidinol did not differ in their adhesiveness from appropriate control cells. The adhesiveness of control cell populations was found to vary with the extent of re-feeding indicating a possible metabolic dependency of cell adhesion. It is suggested that although M phase cells are poorly adherent to the substrate they are nevertheless considerably adherent in an aggregation system. There was no apparent correlation of the observed increase in intercellular adhesion with any particular stage of M phase (prophase, metaphase, anaphase, telophase). We propose that Balb/c 3T3 cells show an increase in cell-cell adhesiveness as they proceed into M phase and that this is maintained until after division is complete. Our suggestion that certain cell types may interact differently with each other and with the substrate in different stages of the cell cycle could have in vivo relevance in those processes such as tumour spread where considerable cell division is involved.

Induction of apoptosis by anti-cancer drugs with disparate modes of action: kinetics of cell death and changes in c-myc expression.

Incubation of CCRF CEM C7A human lymphoblastic leukaemia cells with etoposide (VP16) or N-methylformamide (NMF) induced apoptotic cell death. The kinetics of onset of apoptosis was determined and compared with that for dexamethasone-treated cells. The drugs induced 50% apoptosis at different rates: etoposide by approximately 18 h, NMF by 40 h and dexamethasone (DEX) by 52 h. In each case, the onset of apoptosis above 10% was preceded by a delay period. This was 8 h for etoposide, between 8 and 12 h for NMF and 36 h for dexamethasone. When cells were incubated for 36 h with dexamethasone and the drug washed out, addition of NMF induced apoptosis without any delay, suggesting that certain common biochemical events are required to prime the cells for apoptosis. However, cells treated for 8 h with NMF did not undergo immediate apoptosis on the addition of DEX. Analysis of the cellular content of the c-myc protein showed this to be undetectable by 2, 6 and 12 h after treatment with etoposide, NMF and DEX respectively. The rapid onset of NMF-induced cell death after a 36 h DEX pretreatment occurred 24 h after the loss of expression of c-Myc protein, suggesting that the expression of c-myc is not required for drug-induced cell death. In contrast to DEX-induced apoptosis, concomitant incubation of cells with NMF or etoposide and 200 nM of the protein synthesis inhibitor cycloheximide did not inhibit apoptotic cell death. The idea that drugs with different modes of action initiate conserved responses which engage a programmed cell death is discussed.

A study of the adhesive, locomotory and invasive behaviour of Walker 256 carcinosarcoma cells.

We have succeeded in selecting two variant strains of the Walker 256 carcinosarcoma which display markedly different adhesive properties. Both the high (W256A) and the low (W256S) adhesive variants respond chemotactically towards 10(-8) M f-met-leu-phe (FMLP) although there is a significant difference in their locomotory ability. Nevertheless, the fact that the essentially non-adherent W256S cells can migrate in vitro argues against any simple relationship between adhesion and locomotion. We suggest that traction is important in locomotion but that it need not arise only from direct adhesive interaction. We have also tested the invasive behaviour of the W256 variants using an in vitro model system in which disruption of a cellular barrier by the invasive cells can be recorded electrophysiologically. Although leucocytes can penetrate such a barrier they do so only under chemotactic stimulation, whereas W256 tumour cells of either variant strain will do so spontaneously. The tumour variants induce cell retraction within the barrier and this may lead ultimately to cell detachment and death. The holes which arise may then be colonized by tumour cells, and in this way the invasive process could be promoted. The molecular mechanisms by which tumour cells achieve destruction of the cellular barrier are not clear, but it is likely that a number of enzymes are involved.

Isolation and preliminary characterisation of cDNA clones representing mRNAs associated with tumour progression and metastasis in colorectal cancer.

We have constructed cDNA libraries from the poly(A)+ RNA of normal colonic mucosa and a liver metastasis from a colonic adenocarcinoma. Differential screening of these libraries using 32P-labelled cDNAs transcribed from poly(A)+ RNAs isolated from specimens of four normal colonic mucosae, five adenocarcinomas, and three liver metastases by Grunstein-Hogness and dot-blot hybridization has identified a number of recombinant cDNA clones homologous to mRNAs that appear to differ significantly in abundance between normal and neoplastic colon and metastases. These cDNA clones, and others identified in the libraries, may be of considerable importance both as diagnostic tools and in defining the phenotypic changes associated with tumour progression and metastasis.

Detection of mRNA for ribosomal phosphoprotein P2 in normal colon and colonic tumours by in situ hybridization.

We have localized the mRNA for the ribosomal phosphoprotein P2, a putative metastasis-related sequence, in normal colon and colonic carcinomas by in situ hybridization, using an oligonucleotide probe end-labelled with digoxigenin. The mRNA was identified in normal colonic epithelial cells, the intensity of the signal being greater in cells at the base of the crypts compared with those on the surface. A strong positive signal was also seen in plasma cells, in fibroblasts in granulation tissue, in ganglion cells, and in hepatocytes. A positive signal was identified in all 16 primary colonic tumours studied and in 7 hepatic metastases. In contrast to previous studies based on Northern blot analysis, we were unable to demonstrate increased expression in metastases as compared with primary tumours, nor could we demonstrate any increased expression in primary tumours which were associated with distant metastases.

Rectal cancer: evaluation of staging with endosonography.

PURPOSE: To investigate whether endosonography is reliable in making radiation therapy decisions in rectal cancer, with possible downstaging taken into consideration. MATERIALS AND METHODS: Ninety patients (52 men, 38 women; median age, 69 years) with rectal adenocarcinoma underwent endosonography within 2 weeks before surgery and radiation therapy (performed in 54 patients). The tumor invasive edge was used for radiation therapy decision making. RESULTS: The local stage was accurately assessed in 65 patients (39 with and 26 without irradiation). The tumor invasive edge was accurately assessed in 63 patients. Overstaging was present in 19 patients; the tumor had grown almost through the muscularis propria in six. The invasive edge (P = .1) and lymph node status were overstaged more often in the patients with than in the patients without irradiation. Tumor was understaged in eight patients: The invasive edge did not penetrate but there was budding beyond the muscularis propria in five; the invasive edge penetrated the muscularis propria in two. In seven of the eight patients, growth beyond the muscularis propria was smaller than the endosonographic resolution. Three patients with understaged, nonirradiated tumors developed pelvic recurrence. None of the patients with irradiation and none of the 16 patients without irradiation but with correct assessment developed pelvic recurrence. CONCLUSION: Preoperative irradiation decision making on the basis of endosonographic findings is uncertain. Downstaging after preoperative irradiation must be considered.

A sequence previously identified as metastasis-related encodes an acidic ribosomal phosphoprotein, P2.

We have used a metastasis-related human cDNA isolated from a liver metastasis from a colonic adenocarcinoma to screen a human breast carcinoma cDNA library for homologous sequences. Nucleotide sequence analysis of positive clones revealed that the cDNA represents a ribosomal phosphoprotein. P2. The expression of P2 mRNA was significantly higher (Student's t test, one tail; P less than or equal to 0.01) in seven fibroadenomas than in seven carcinomas, with an average five-fold difference. This enhanced expression level P2 mRNA in benign fibroadenomas compared with malignant carcinomas is contrary to that expected, based on earlier work with normal colonic mucosa, colorectal carcinoma and hepatic metastasis. The identification of gene transcripts which differ in abundance and correlate with the metastatic phenotype may be of considerable importance both as diagnostic aids and in defining the changes associated with tumour progression and metastasis at the molecular level. The possible role that ribosomal proteins may play in the progression of carcinoma of the breast is discussed.

Use of angiographic needles with or without stylets: pathologic assessment of vessel walls after puncture.

There is controversy as to whether angiographic needles without stylets produce more arterial damage than those with stylets. Iliac arteries from 15 fresh human cadavers were punctured 56 times with either an 18-gauge angiographic needle with a stylet or one without a stylet (28 punctures with each needle type). These puncture sites were serially sectioned and examined microscopically. Each needle tract was evaluated for margin irregularity, shape of puncture, and approximation of edges. No statistically significant differences in arterial wall changes were found. The authors' data suggest that the choice of beveled needle use in angiography can probably be made on a basis other than concern for differences in vessel wall damage secondary to the presence or absence of a stylet.

Vascular endothelial growth factor stimulates protein kinase C-dependent phospholipase D activity in endothelial cells.

Many tumors produce vascular endothelial growth factor (VEGF), a paracrine factor acting selectively on endothelial cells. VEGF has many effects on cultured endothelial cells and mediates angiogenesis and enhanced vascular permeability in vivo. The endothelial signal transduction pathways of VEGF represent novel targets for cancer therapy because they are readily accessible to systemically administered drugs. We have examined VEGF-stimulated signals generated in HUVEC to identify potential targets for therapeutic intervention. The transphosphatidylation reaction has been used to monitor phospholipase D (PLD) activity; total inositol phosphates have been measured after prelabeling of cells with [3H]myoinositol; and intracellular free calcium has been measured using Fura-2 fluorescence. After HUVEC-stimulation with VEGF, there is an early influx of calcium (maximal by 100 seconds) followed by activation of PLD (half maximal by 100 seconds, EC50 70 pm). The PLD activity was inhibited by reducing extracellular calcium (150 nM, 50% inhibition), exposure to 12-O-tetradecanoylphorbol 13 acetate (200 nM, 24 hours, 100% inhibition), Roche 31,8220 (10 microM, 15 minutes, 72% inhibition), or genestein (100 microM, 30 minutes, 56% inhibition), which suggests a dependence on both protein kinase C and tyrosine phosphorylation. Activation of phospholipase C-catalyzed hydrolysis of phosphatidylinositol-4,5-bisphosphate was inferred from the production of inositol phosphates, although this response was slower (half maximal by 3 minutes). The phospholipase C activity was also dependent on influx of calcium and was partially inhibited by low (150 nM) extracellular calcium. PLD may be involved in mediating a number of endothelial responses to tumor-secreted VEGF, notably cytoskeleton-dependent effects such as the cell migration involved in angiogenesis. This signal transduction pathway could represent an accessible and vulnerable target for cancer therapeutic intervention and has the novelty of being located within normal cells rather than tumor cells.

Isolation of cDNA clones using yeast artificial chromosome probes.

The cloning of large DNA fragments of hundreds of kilobases in Yeast artificial chromosomes, has simplified the analysis of regions of the genome previously cloned by cosmid walking. The mapping of expressed sequences within cosmid contigs has relied on the association of genes with sequence motifs defined by rare-cutting endonucleases, and the identification of sequence conservation between species. We reasoned that if the contribution of repetitive sequences to filter hybridizations could be minimised, then the use of large cloned DNAs as hybridisation probes to screen cDNA libraries would greatly simplify the characterisation of hitherto unidentified genes. In this paper we demonstrate the use of this approach by using a YAC, containing 180 kb of human genomic DNA including the aldose reductase gene, as a probe to isolate an aldose reductase cDNA from a lambda gt11 human foetal liver cDNA library.

Walking, cloning, and mapping with yeast artificial chromosomes: a contig encompassing D21S13 and D21S16.

Chromosome 21 has often been used as a model system for the development of genome mapping and cloning strategies in humans. In this report methods for systematic chromosome walking, cloning, and mapping are exemplified in the construction of a 1.5-Mb yeast artificial chromosome (YAC) contig encompassing and extending 400 kb beyond each of the genetic loci D21S13 and D21S16. Isolation of insert-terminal sequences from YACs in this contig provides a set of closely spaced physical markers. These have been used to generate a long-range genomic restriction map.

Plasminogen activator inhibitor type 2 in breast cancer.

The serine protease urokinase plasminogen activator (uPA) is causally involved in cancer invasion and metastasis. Activity of this protease in vivo is controlled principally by two inhibitors, one of which is plasminogen activator inhibitor type 2 (PAI-2). In this study, we show that PAI-2 levels were significantly higher in primary breast carcinomas (n = 152) than benign breast tumours (n = 18). In the primary cancers, PAI-2 levels correlated weakly but significantly with those of uPA and PAI-1, but not with tissue type plasminogen activator (tPA) or uPA receptor (uPAR) levels. Using Northern blotting, mRNA for PAI-2 was found in 28.6% of 49 primary breast cancers. In contrast to findings at the protein level, PAI-2 mRNA levels failed to correlate with those for uPA or PAI-1. After immunocytochemistry with primary cancers, PAI-2 was detected predominantly in the malignant cells of primary carcinomas but was also present in stromal cells. Using the median value as a cut-off point, PAI-2 showed no significant relationship with either disease-free interval or overall survival. However, using an optimum cut-off value, patients with low levels of PAI-2 had a worse outcome than those with a high level. We conclude that, unlike PAI-1, high levels of PAI-2 may be a favourable prognostic marker in breast cancer.

Urokinase plasminogen activator as a predictor of aggressive disease in breast cancer.

Urokinase plasminogen activator (uPA) is a multifunctional protein involved in both extracellular proteolysis and signal transduction. uPA usually mediates its actions while attached to a membrane-bound receptor, termed uPAR. In this study, uPA and its receptor were measured at both protein and mRNA levels in breast cancer. At both levels, concentrations of uPA were significantly correlated with those for uPAR. uPA levels also correlated significantly with cathepsin B and cathepsin D but not with cathepsin L, MMP-8 or MMP-9 levels. Irrespective of the cut-off point used (e.g., median, tertile or quartile values), uPA was a significant prognostic marker for breast cancer.