Texture analysis of (125)I-A5B7 anti-CEA antibody SPECT differentiates metastatic colorectal cancer model phenotypes and anti-vascular therapy response.

BACKGROUND: We aimed to test the ability of texture analysis to differentiate the spatial heterogeneity of (125)I-A5B7 anti-carcinoembryonic antigen antibody distribution by nano-single photon emission computed tomography (SPECT) in well-differentiated (SW1222) and poorly differentiated (LS174T) hepatic metastatic colorectal cancer models before and after combretastatin A1 di-phosphate anti-vascular therapy. METHODS: Nano-SPECT imaging was performed following tail vein injection of 20 MBq (125)I-A5B7 in control CD1 nude mice (LS174T, n=3 and SW1222, n=4), and CA1P-treated mice (LS174T, n=3; SW1222, n=4) with liver metastases. Grey-level co-occurrence matrix textural features (uniformity, homogeneity, entropy and contrast) were calculated in up to three liver metastases in 14 mice from control and treatment groups. RESULTS: Before treatment, the LS174T metastases (n=7) were more heterogeneous than SW1222 metastases (n=12) (uniformity, P=0.028; homogeneity, P=0.01; contrast, P=0.045). Following CA1P, LS174T metastases (n=8) showed less heterogeneity than untreated LS174T controls (uniformity, P=0.021; entropy, P=0.006). Combretastatin A1 di-phosphate-treated SW1222 metastases (n=11) showed no difference in texture features compared with controls (all P>0.05). CONCLUSIONS: Supporting the potential for novel imaging biomarkers, texture analysis of (125)I-A5B7 SPECT shows differences in spatial heterogeneity of antibody distribution between well-differentiated (SW1222) and poorly differentiated (LS174T) liver metastases before treatment. Following anti-vascular treatment, LS174T metastases, but not SW1222 metastases, were less heterogeneous.

Immunohistochemical expression of VEGF predicts response to platinum based chemotherapy in patients with epithelial ovarian cancer.

PURPOSE: For patients with epithelial ovarian cancer (EOC) cytoreduction, with a combination of taxane and platinum, is the standard of care. Despite this, approximately 50% of patients with advanced disease will relapse and moreover 15-20% of cases of EOC are resistant to platinum based chemotherapy. Vascular Endothelial Growth Factor (VEGF), an angiogenic factor, is associated with poor prognosis. This study was undertaken to examine whether there is an association between VEGF-A expression in the tumour of EOC patients and their response to platinum based chemotherapy. METHODS: The study cohort consisted of 66 patients with advanced stage EOC (FIGO III-IV). Ovarian cancer tissue was analysed for VEGF-A expression immunohistochemically. Protein expression was measured and correlated, with platinum sensitivity and overall patient survival. RESULTS: Median age of patients was 53 years, 45 patients had platinum sensitive disease (68%), the remaining patients being platinum resistant (32%). Of the platinum resistant group, 18 (86%) patients had high VEGF score compared to only 1 (2%) with high VEGF score in the platinum sensitive group. Median survival was 11 months in the patient group with high VEGF score versus 32 months in that cohort with low VEGF score. VEGF expression was significantly inversely correlated with overall survival (P < 0.0001). CONCLUSION: We demonstrated that tumours of patients with platinum resistant EOC exhibit higher levels of VEGF expression compared to the platinum sensitive group. VEGF in EOC, may be of clinical and therapeutic relevance and suggests a role for first line anti-angiogenic therapy.

Characterisation and radioimmunotherapy of L19-SIP, an anti-angiogenic antibody against the extra domain B of fibronectin, in colorectal tumour models.

Angiogenesis is a characteristic feature of tumours and other disorders. The human monoclonal antibody L19- SIP targets the extra domain B of fibronectin, a marker of angiogenesis expressed in a range of tumours. The aim of this study was to investigate whole body distribution, tumour localisation and the potential of radioimmunotherapy with the L19-small immunoprotein (SIP) in colorectal tumours. Two colorectal tumour models with highly different morphologies, the SW1222 and LS174T xenografts, were used in this study. Localisation and retention of the L19-SIP antibody at tumour vessels was demonstrated using immunohistochemistry and Cy3-labelled L19-SIP. Whole body biodistribution studies in both tumour models were carried out with (125)I-labelled L19-SIP. Finally, (131)I-labelled antibody was used to investigate the potential of radioimmunotherapy in SW1222 tumours. Using immunohistochemistry, we confirmed extra domain B expression in the tumour vasculature. Immunofluorescence demonstrated localisation and retention of injected Cy3-labelled L19-SIP at the abluminal side of tumour vessels. Biodistribution studies using a (125)I-labelled antibody showed selective tumour uptake in both models. Higher recorded values for localisation were found in the SW1222 tumours than in the LS174T (7.9 vs 6.6 %ID g(-1)), with comparable blood clearance for both models. Based on these results, a radioimmunotherapy study was performed in the SW1222 xenograft using (131)I-Labelled L19-SIP (55.5 MBq), which showed selective tumour uptake, tumour growth inhibition and improved survival. Radio- and fluorescence-labelled L19-SIP showed selective localisation and retention at vessels of two colorectal xenografts. Furthermore, (131)I-L19-SIP shows potential as a novel treatment of colorectal tumours, and provides the foundation to investigate combined therapies in the same tumour models.

Immunohistochemical expression of vascular endothelial growth factor and microvessel counting as prognostic indicators in node-negative colorectal cancer.

This manuscript reports a carefully controlled study of patients with Dukes B colorectal cancer (Dukes stage A, n=12 and Dukes stage B, n=44). Immunohistochemistry has been used to demonstrate reactivity for vascular endothelial growth factor (VEGF), and to measure levels of microvessel density (MVD) in order to assess the relationship of tumor angiogenesis with clinical outcome. Immunohistochemistry was performed using antibodies to VEGF and CD34 (for intratumoral vessel identification) and counting was performed at the invasive margin of the tumor. Results showed that for Dukes stage A patients 4/12 died of their disease, none of whose tumor was VEGF positive. In contrast, 2 patients who survived were positive for VEGF cytoplasmically, but neither showed increased tumor MVD. In Dukes B patients 10/44 died, 5 of whose tumor demonstrated VEGF reactivity, both in malignant cells and in tumor vascular endothelium. MVD ranged from 11 to 53 (median 28) for Dukes A cases and from 9 to 69 (median 32.5) for the Dukes B group. Kaplan-Meier plots and log rank test statistics for Dukes B patients demonstrated that VEGF reactivity in cells, and in tumor vascular endothelium was correlated with survival (p=0.047 and p < or = 0.06, respectively). There was a significant relationship between the presence of VEGF reactivity on vascular endothelium and outcome by Fisher's exact test (p=0.018). Similarly, by the same test VEGF positivity was significantly correlated with patient mortality (p=0.032). The presence of endothelial VEGF reactivity correlated with VEGF in malignant cells (p=0.0001) by Mann-Whitney U test and a significant inverse relationship between vessel density and patient survival was demonstrated (p = 0.019). The finding that in Dukes B patients MVD was inversely correlated with mortality supports the hypothesis that a low microvascular count is predicted close to the invasive margin, where VEGF expression is upregulated in response to hypoxia, induced by a lack of a functional vasculature. These data will be used to identify cohorts of patients who have a high risk of relapse and can be selected for adjuvant therapies such as VEGF antibody or antitumor antibody-directed therapy.

Spatial accuracy of 3D reconstructed radioluminographs of serial tissue sections and resultant absorbed dose estimates.

Many agents using tumour-associated characteristics are deposited heterogeneously within tumour tissue. Consequently, tumour heterogeneity should be addressed when obtaining information on tumour biology or relating absorbed radiation dose to biological effect. We present a technique that enables radioluminographs of serial tumour sections to be reconstructed using automated computerized techniques, resulting in a three-dimensional map of the dose-rate distribution of a radiolabelled antibody. The purpose of this study is to assess the reconstruction accuracy. Furthermore, we estimate the potential error resulting from registration misalignment, for a range of beta-emitting radionuclides. We compare the actual dose-rate distribution with that obtained from the same activity distribution but with manually defined translational and rotational shifts. As expected, the error produced with the short-range 14C is much larger than that for the longer range 90Y; similarly values for the medium range 131I are between the two. Thus, the impact of registration inaccuracies is greater for short-range sources.

Antibody and radionuclide characteristics and the enhancement of the effectiveness of radioimmunotherapy by selective dose delivery to radiosensitive areas of tumour.

PURPOSE: Estimating the absorbed dose to tumour relative to normal tissues has often been used in the assessment of the therapeutic efficacy of radiolabelled antibodies for radioimmunotherapy. Typically, the calculations assume a uniform dose deposition and response throughout the tumour. However, the heterogeneity of the dose delivery and response within tumours can lead to a radiobiological effect inconsistent with dose estimates. The aim was to assess the influence of antibody and radionuclide characteristics on the heterogeneity of dose deposition. MATERIALS AND METHODS: Quantitative images of the temporal and spatial heterogeneity of a range of antibodies in tumour were acquired using radioluminography. Subsequent registration with images of tumour morphology then allowed the delineation of viable and necrotic areas of tumour and the measurement of the antibody concentration in each area. A tumour dosimetry model then estimated the absorbed dose from 131I and 90Y in each area. RESULTS: Tumour-specific antibodies initially localized in the viable radiosensitive areas of tumour and then penetrated further into tumour with continued tumour accretion. Multivalent antibodies were retained longer and at higher concentrations in viable areas, while monovalent antibodies had greater mobility. In contrast, non-specific antibodies penetrated into necrotic regions regardless of their size. As a result, multivalent, specific antibodies delivered a significantly larger dose to viable cells compared with monovalent antibodies, while non-specific antibodies deposited most of the dose in necrotic areas. There was a significant difference in dose estimates when assuming a uniform dose deposition and accounting for heterogeneity. The dose to the viable and necrotic areas also depended on the properties of the radionuclide where antibodies labelled with 131I generally delivered a higher dose throughout the tumour even though the instantaneous dose-rate distribution for 90Y was more uniform. CONCLUSIONS: The extent of heterogeneity of dose deposition in tumour is highly dependent on the antibody characteristics and radionuclide properties, and can enhance therapeutic efficacy through the selective dose delivery to the radiosensitive areas of tumour.

In-vitro uptake of radioactive lipiodol I-131 and I-125 by hepatoblastoma: implications for targeted radiotherapy.

Lipiodol, a stable iodine addition product of fatty ethyl esters derived from poppyseed oil, has been used as a vehicle for targeted cytotoxic or radiotherapeutic treatment in adults with unresectable hepatocellular carcinoma and in a few children with advanced hepatoblastoma (HB). Prolonged retention of lipiodol by the cancer cells might enable more effective targeted therapy to advanced, multifocal, or metastatic HB. To investigate the uptake and cytotoxic efficacy of beta and Auger electron-emitting radioconjugates on HB cells, monolayers of HB (C3a, Hep Tow1) and normal human hepatocyte cell lines were exposed to lipiodol131 (L131), lipiodol125 (L125), or a cocktail of both isotopes. Uptake of radioactivity was assessed autoradiographically using phosphoimages and the cytotoxicity assessed by trypan blue exclusion and clonogenic assay. The uptake of the different radioconjugates was identical in both HB cell lines. The "cocktail" of both radioactive lipiodols caused the greatest cytotoxicity to HB C3a, and HepTow1. L125 alone had a similar level of uptake and cytotoxicity as L131. Radioactive iodine alone did not show any cytotoxicity on any of the liver cell lines in culture for up to 72 h. These experimental results provide support for a clinical strategy of combinations ("cocktails") of radioconjugates in targeted radiotherapy in patients with HB.

Relationship between tumour morphology, antigen and antibody distribution measured by fusion of digital phosphor and photographic images.

Antibody-directed cancer therapy has achieved encouraging responses despite poor localisation in tumour. This discrepancy may be attributed to heterogeneity of antibody delivery within tumours: preferential localisation in the better perfused and more radio- and chemosensitive areas provides a therapeutic advantage. Antibody distribution depends upon the interactions of many complex mechanisms. We have started to investigate this by studying the single and combined influence of two tumour-associated parameters, morphology and antigen, on antibody distribution. Tumours were taken from mice at 24 and 48 h after 125I-labeled anti-CEA antibody injection. Images of antibody distribution, antigen distribution and tumour morphology were acquired by radioluminography, radioimmunoluminography and digitisation of morphology, respectively. Image registration allowed correlation of pixel values of antibody distribution with corresponding values of antigen distribution and morphology. At 24 h there was little correlation between antibody and antigen distribution, but strong positive correlation between antibody distribution and morphology, with preferential localisation in viable tumour areas. Correlation between antibody distribution and morphology fell significantly between 24 and 48 h, while that between antibody and antigen distribution remained low. However, the combination of morphology and antigen distribution showed the largest influence on antibody distribution. This novel technique demonstrates potential for combining multi-factor information in order to provide a greater understanding of antibody distribution in tumours, facilitating the optimisation of clinical treatments.

A mouse model for calculating the absorbed beta-particle dose from (131)I- and (90)Y-labeled immunoconjugates, including a method for dealing with heterogeneity in kidney and tumor.

Flynn, A. A., Green, A. J., Pedley, R. B., Boxer, G. M., Boden, R. and Begent, R. H. J. A Mouse Model for Calculating the Absorbed Beta-Particle Dose from (131)I- and (90)Y-Labeled Immunoconjugates, Including a Method for Dealing with Heterogeneity in Kidney and Tumor. Radiat. Res. 156, 28-35 (2001). Conventional internal radiation dosimetry methods assume that the beta-particle energy is absorbed uniformly and completely in the source organ and that the radioactivity is distributed uniformly in the source. However, in mice, a considerable proportion of the beta-particle energy can escape the source organ, resulting in large cross-organ doses. Furthermore, the distribution of radioactivity is generally heterogeneous in kidney and tumor. Therefore, a model was developed to account for cross-organ doses and for the effects of heterogeneity in kidney and tumor in mice for two of the most important radionuclides used in therapy, (131)I and (90)Y. Most mouse organs were modeled as single-compartment ellipsoids or cylinders, while heterogeneity in kidney and in tumor was addressed by using two compartments to represent the cortex and the medulla and viable and necrotic cells, respectively. The dimensions of these models were taken from previous studies, with the exception of kidney and tumor, which were defined using radioluminography and mosaics of high-power microscopy images. The absorbed fractions in each compartment were calculated using beta-particle point dose kernels. The self-organ dose was significantly higher for (131)I compared to (90)Y in all compartments, but a considerable amount of beta-particle energy was shown to escape the source organ for both radionuclides, with as much as 85% and 36% escaping the marrow for (90)Y and (131)I, respectively. The cortex was found to occupy a greater proportion of the total kidney volume than the medulla, and consequently the self-dose was higher in the cortex. In addition, the thickness of the viable shell in the tumor increased with tumor size, as did the self-dose fractions in both necrotic and viable areas. This dosimetry model improves dose estimates in mice and gives a conceptual basis for considering dosimetry in humans.

Eradication of colorectal xenografts by combined radioimmunotherapy and combretastatin a-4 3-O-phosphate.

Solid tumors have a heterogeneous pathophysiology, which has a major impact on therapy. Using SW1222 colorectal xenografts grown in nude mice, we have shown that antibody-targeted radioimmunotherapy (RIT) effectively treated the well-perfused tumor rim, producing regressions for approximately 35 days, but was less effective at the more hypoxic center. By 72 h after RIT, the number of apoptotic cells rose from an overall value of 1% in untreated tumors to 35% at the tumor periphery and 10% at the center. The antivascular agent disodium combretastatin A-4 3-O-phosphate (CA4-P) rapidly reduced tumor blood flow to 62% of control values by 1 h, 23% by 3 h, and between 32-36% from 6 to 24 h after administration. This created central hemorrhagic necrosis, but a peripheral rim of cells continued to grow, and survival was unaffected. Changes in the pattern of perfusion across the tumor over time were zonal. Untreated mice showed perfusion throughout the tumor, with greatest activity at the rim. There was an overall reduction at 1 h, and total cessation of central perfusion from 3 h onward. A narrow peripheral rim of perfusion was always present, which increased in intensity and extent between 6 and 24 h, either through reperfusion or new vessel growth. Combining these two complementary therapies (7.4 MBq (131)I-labeled anti-carcinoembryonic antigen IgG i.v. plus a single 200 mg/kg dose of CA4-P i.p.) produced complete cures in five of six mice for >9 months. Allowing maximal tumor localization of antibody (48 h) before blood flow inhibition by CA4-P increased tumor retention by two to three times control levels by 96 h without altering normal tissue levels, as confirmed by gamma counting and phosphor image analysis. The success of this combined, synergistic therapy was probably the result of several factors: (a) the killing of tumor cells in the outer, radiosensitive region by targeted radiotherapy; (b) enhancement of RIT by entrapment of additional radioantibody after combretastatin-induced vessel collapse; and (c) destruction of the central, more hypoxic and radioresistant region by CA4-P. This work demonstrates the need to consider cancer treatment in a biologically heterogeneous setting, if results are to be effectively translated to the clinic.

A strategy for antitumor vascular therapy by targeting the vascular endothelial growth factor: receptor complex.

Vascular endothelial growth factor (VEGF) is produced by cancer cells in response to hypoxia and is the primary stimulant of vascularization in solid tumors. Endothelial cells lining the blood vessels of these tumors have a high concentration of receptor-bound VEGF on their surface, providing a target for antibody- directed cancer therapy. To obtain a cloned antibody to this target when bound to its receptor on tumor endothelium, we used phage display technology to create a single-chain Fv (sFv) antibody library from mice immunized with the 165-amino acid isoform of human VEGF-A. We selected, purified, and characterized LL4, an anti-VEGF sFv that was shown to react with receptor-bound VEGF. LL4 bound selectively to blood vessel endothelium, as shown by immunohistochemistry on tissue sections of human tumors. Furthermore, using autoradiography and grain counting of histological sections, systemically administered LL4 was shown to localize selectively to the endothelial lining of tumor blood vessels in human colorectal carcinoma xenografts in vivo. This study demonstrates the feasibility of targeting tumor vasculature using recombinant antibodies to the VEGF:receptor complex.

Effectiveness of radiolabelled antibodies for radio-immunotherapy in a colorectal xenograft model: a comparative study using the linear--quadratic formulation.

PURPOSE: To develop a model that relates the pattern of dose delivery during radio-immunotherapy to biological effect. This model was used to assess the efficacy of a range of antibodies labelled with 131I, 186Re and 90Y. MATERIALS AND METHODS: Pharmacokinetic data were obtained by injecting tumour-bearing nude mice with radiolabelled antibody. The dose-rate in bone marrow and tumour was then given by a two-compartment model description of the pharmacokinetics combined with the radionuclide properties. Response characteristics of tumour and marrow were defined in terms of radiosensitivity, repair capacity and proliferation, and the biological effect was assessed using the linear quadratic formulation. RESULTS: Tumour-specific antibodies with intermediate molecular weight and clearance from the circulation delivered the most effective doses to tumour due to their rapid uptake and prolonged retention in tumour coupled with efficient clearance from blood. Matching the radionuclide with antibody pharmacokinetics and tumour type further increased this effect. CONCLUSIONS: The model improves conceptual understanding of the relationship of parameters affecting therapy and makes it possible to optimize radio-immunotherapy by selecting the most effective antibody and radionuclide according to tumour biology.

Optimizing radioimmunotherapy by matching dose distribution with tumor structure using 3D reconstructions of serial images.

The biological effect of radioimmunotherapy (RIT) is most commonly assessed in terms of the absorbed radiation dose. In tumor, conventional dosimetry methods assume a uniform radionuclide and calculate a mean dose throughout the tumor. However, the vasculature of solid tumors tends to be highly irregular and the systemic delivery of antibodies is therefore heterogeneous. Tumor-specific antibodies preferentially localize in the viable, radiosensitive parts of the tumor whereas non-specific antibodies can penetrate into the necrosis where the dose is wasted. As a result, the observed biological effect can be very different to the predicted effect from conventional dose estimates. The purpose of this study is to assess the potential for optimizing the biological effect of RIT by matching the dose-distribution with tumor structure through the selection of appropriate antibodies and radionuclides. Storage phosphor plate technology was used to acquire images of the antibody distribution in serial tumor sections. Images of the distributions of a trivalent (TFM), bivalent (A5B7-IgG), monovalent (MFE-23) and a non-specific antibody (MOPC) were obtained. These images were registered with corresponding images showing tumor morphology. Serial images were reconstructed to form 3D maps of the antibody distribution and tumor structure. Convolution of the image of antibody distribution with beta dose point kernals generated dose-rate distributions for 14C, 131I and 90Y. These were statistically compared with the tumor structure. The highest correlation was obtained for the multivalent antibodies combined with 131I, due to specific retention in viable areas of tumor coupled with the fact that much of the dose was deposted locally. With decreasing avidity the correlation also decreased and with the non-specific antibody this correlation was negative, indicating higher concentrations in the necrotic regions. In conclusion, the dose distribution can be optimized in tumor by selecting the appropriate antibodies and radionuclides. This has the potential to lead to a considerable enhancement of the efficacy of RIT in the clinic.

Recombinant anti-carcinoembryonic antigen antibodies for targeting cancer.

Antibodies can be used to target cancer therapies to malignant tissue; the approach is attractive because conventional treatments such as chemo- and radiotherapy are dose limited due to toxicity in normal tissues. Effective targeting relies on appropriate pharmacokinetics of antibody-based therapeutics, ideally showing maximum uptake and retention in tumor and rapid clearance from normal tissue. We have studied the factors influencing these dynamics for antibodies against carcinoembryonic antigen (CEA). Protein engineering of anti-CEA antibodies, in vivo biodistribution models, and mathematical models have been employed to improve understanding of targeting parameters, define optimal characteristics for the antibody-based molecules employed, and develop new therapies for the clinic. Engineering antibodies to obtain the desired therapeutic characteristics is most readily achieved using recombinant antibody technology, and we have taken the approach of immunizing mice to provide high-affinity anti-CEA single-chain Fv antibodies (sFvs) from filamentous bacteriophage libraries. MFE-23, the most characterized of these sFvs, has been expressed in bacteria and purified in our laboratory for two clinical trials: a gamma camera imaging trial using 123I-MFE-23 and a radioimmunoguided surgery trial using 125I-MFE-23, where tumor deposits are detected by a hand-held probe during surgery. Both these trials showed that MFE-23 is safe and effective in localizing tumor deposits in patients with cancer. We are now developing fusion proteins that use the MFE-23 antibody to deliver a therapeutic moiety; MFE-23:: carboxypeptidase G2 (CPG2) targets the enzyme CPG2 for use in the antibody-directed enzyme prodrug therapy system and MFE::tumor necrosis factor alpha (TNFalpha) aims to reduce sequestration and increase tumor concentrations of systemically administered TNFalpha.

Radioimmunoguided surgery in colorectal cancer using a genetically engineered anti-CEA single-chain Fv antibody.

In radioimmunoguided surgery (RIGS), a radiolabeled antibody is given i.v. before surgery and a hand-held gamma-detecting probe is used to locate tumor in the operative field. The rapid blood clearance and good tumor penetration of single-chain Fv antibodies (scFv) offer potential advantages over larger antibody molecules used previously for RIGS. A Phase I clinical trial is reported on RIGS with scFv (MFE-23-his) to carcinoembryonic antigen (CEA). Thirty-four patients undergoing surgery for colorectal carcinoma (17 primary tumors, 16 liver metastases, and 1 anastomotic recurrence) and 1 patient with liver metastases of pancreatic carcinoma received 125I-labeled MFE-23-his scFv (125I-MFE-23-his) 24, 48, 72, or 96 h before operation. 125I-MFE-23-his showed biexponential blood clearance with alpha and beta half-lives of 0.32 and 10.95 h, respectively. The abdomen was scanned during surgery with a hand-held gamma detecting probe (Neoprobe Corp.). 125I-MFE-23-his showed good tumor localization; comparison with histology showed overall accuracy of 84%. Highest median ratios for tumor:normal tissue and tumor:blood were recorded 72 or 96 h after scFv injection for patients undergoing resection of liver metastases. High levels of radioactivity were found in the kidneys. Five patients had grade 1 fever, and three had a grade 1 rise in blood pressure according to the Common Toxicity Criteria. There was a significant correlation between these ratios and those measured in excised tissues using a laboratory gamma counter (P < 0.001). MFE-23-his scFv antibody localizes in CEA-producing carcinomas. The short interval between injection and operation, the lack of significant toxicity, and the relatively simple production in bacteria make MFE-23-his scFv suitable for RIGS.

Catalytic activity of an in vivo tumor targeted anti-CEA scFv::carboxypeptidase G2 fusion protein.

Antibody-directed enzyme prodrug therapy (ADEPT) targets an enzyme selectively to a tumor where it converts a relatively non-toxic prodrug to a potent cytotoxic drug. Previous clinical work using antibody-enzyme chemical conjugates has been limited by the moderate efficiency of tumor targeting of these molecules. To address this a recombinant fusion protein composed of MFE-23, an anti-carcinoembryonic antigen (CEA) single chain Fv (scFv) antibody, fused to the amino-terminus of the enzyme carboxypeptidase G2 (CPG2) has been constructed to achieve ADEPT in CEA-producing tumors. MFE-23::CPG2 fusion protein was overexpressed in Escherichia coli and purified using CEA affinity chromatography. Efficacy of MFE-23::CPG2 delivery to tumors in vivo was assessed by measuring catalytic activity after intravenous injection of purified MFE-23::CPG2 into nude mice bearing CEA-positive LS174T human colon adenocarcinoma xenografts. Recombinant MFE-23::CPG2 cleared rapidly from circulation and catalytic activity in extracted tissues showed tumor to plasma ratios of 1.5:1 (6 hr), 10:1 (24 hr), 19:1 (48 hr) and 12:1 (72 hr). (125)I-MFE-23::CPG2 was retained in kidney, liver and spleen but MFE-23::CPG2 catalytic activity was not, resulting in excellent tumor to normal tissue enzyme ratios 48 hr after injection. These were 371:1 (tumor to liver), 450:1 (tumor to lung), 562:1 (tumor to kidney), 1,477:1 (tumor to colon) and 1,618:1 (tumor to spleen). Favorable tumor : normal tissue ratios occurred at early time points when there was still 21% (24 hr) and 9.5% (48 hr) of the injected activity present per gram of tumor tissue. The high tumor concentrations and selective tumor retention of active enzyme delivered by MFE-23::CPG2 establish that this recombinant fusion protein has potential to give improved clinical efficiency for ADEPT.

Clinical applications of phage-derived sFvs and sFv fusion proteins.

Single chain Fv antibodies (sFvs) have been produced from filamentous bacteriophage libraries obtained from immunised mice. MFE-23, the most characterised of these sFvs, is reactive with carcinoembryonic antigen (CEA), a glycoprotein that is highly expressed in colorectal adenocarcinomas. MFE-23 has been expressed in bacteria and purified in our laboratory for two clinical trials; a gamma camera imaging trial using 123I-MFE-23 and a radioimmunoguided surgery trial using 125I-MFE-23, where tumour deposits are detected by a hand-held probe during surgery. Both these trials show MFE-23 is safe and effective in localising tumour deposits in patients with cancer. We are now developing fusion proteins which use MFE-23 to deliver a therapeutic moiety; MFE-23::CPG2 targets the enzyme carboxypeptidase G2 (CPG2) for use in the ADEPT (antibody directed enzyme prodrug therapy) system and MFE::TNF alpha aims to reduce sequestration and increase tumor concentrations of systemically administered TNF alpha.

Enhancement of antibody-directed enzyme prodrug therapy in colorectal xenografts by an antivascular agent.

The irregular nature of solid tumor vasculature produces a heterogeneous distribution of antibody-targeted therapies within the tumor mass, which frequently results in reduced therapeutic efficacy. We have, therefore, combined two complementary therapies: Antibody-directed Enzyme Prodrug Therapy (ADEPT), which targets tumor cells, and an agent that selectively destroys tumor vasculature. A single i.p. dose (27.5 mg/kg) of the drug 5,6-dimethylxanthenone-4-acetic acid (DMXAA), given to nude mice bearing the LS174T colorectal xenograft, destroyed all but a peripheral rim of tumor cells, without enhancing survival. The ADEPT system, in which a pretargeted enzyme activates a prodrug, consisted of the F(ab')2 fragment of anti-carcinoembryonic antigen antibody A5B7 conjugated to the bacterial enzyme carboxypeptidase G2 and the prodrug 4-[(2-chloroethyl)(2-mesyloxyethyl)amino]benzoyl-L-glutamic acid, which was given i.p. in three doses of 500 mg/kg at 72, 84, and 96 h post-conjugate administration (25 units of carboxypeptidase G2). The antibody-enzyme conjugate could be selectively retained at approximately twice the control levels by administration of the antivascular agent at the time of optimal conjugate localization within the tumor (20 h post-conjugate administration), as demonstrated by gamma counting, phosphor plate image analysis, and active enzyme measurement. This resulted in significantly enhanced tumor growth inhibition in groups of six mice, compared to conventional ADEPT therapy, with no concomitant increase in systemic toxicity. In a separate experiment, aimed at trapping the prodrug within the tumor, a 16-fold increase over control values was produced (means, 44.8 versus 2.8 microg/g tumor) when DMXAA was given 4 h prior to 4-[(2-chloroethyl)(2-mesyloxyethyl)amino]benzoyl-L-glutamic acid. The therapeutic window was small, with no significant enhancement of prodrug retention when DMXAA was given at either earlier or later time points. This correlated with the time of vascular shut-down induced by the antivascular agent. We are currently investigating whether it is more advantageous to trap increased levels of conjugate or prodrug within the tumor for maximal enhancement of conventional ADEPT. These studies demonstrate that combined use of antibody-directed and antivascular therapies can significantly benefit the therapeutic outcome of either strategy alone.

A comparison of image registration techniques for the correlation of radiolabelled antibody distribution with tumour morphology.

Image registration is a powerful tool for correlating functional images with images of anatomical structure. This facilitates more accurate quantitation of regional radiopharmaceutical uptake. Similarly, registration of images of radiolabelled antibody distribution, in tissue sections, with the equivalent histological images allows the comparison and measurement of radiopharmaceutical distribution with morphological structure. The images used were obtained by storage phosphor plate technology, for the radiopharmaceutical distribution, and by digitization of the stained histological sections. Here we compare four fully automatic registration techniques and one manual technique in terms of their spatial accuracy. We have found that there was no difference in accuracy between cross-correlation, minimization of variance and mutual information. These techniques were more accurate than principal axes and the manual technique. However, minimization of variance and mutual information were more time-consuming than the other methods. Consequently, cross-correlation is the method of choice for automatic registration of large numbers of these image pairs.

Radioimmunoluminography: a tool for relating tissue antigen concentration to clinical outcome.

Proteins control cell function and measurement of their concentration is crucial for understanding their roles in health and disease. However, current methods for their detection in tissue sections are not quantitative. Radioimmunoluminography (RILG) is a system for direct measurement of quantity and distribution of protein in histological sections. Histological carcinomas were reacted with antigen-saturating concentrations of 125I-antibody to carcinoembryonic antigen (CEA). Antibody distribution and concentration was mapped by phosphor imaging. Radioactivity in each pixel of the digital image was proportional to antigen concentration, calculated from a standard line generated from a nitrocellulose CEA dot-blot assay. RILG gave a linear correlation with standards of known CEA concentration (r = 0.999). Six tumour xenografts with differing CEA concentrations by radioimmunoassay (RIA) were studied by RILG and immunohistochemistry (IHC). RILG gave a linear correlation with CEA by RIA (r = 0.994) but IHC failed to do so (r = 0.42). CEA levels measured by RILG, in cryostat (n = 15) and paraffin (n = 19) sections from colorectal cancer patients showed a range of CEA concentration (38.9-594 ng g(-1) and 22.5-212.5 ng g(-1) respectively). Tumour CEA concentration by RILG was significantly correlated with dose of antibody (% injected radioactivity kg(-1)) localized in tumour (P = 0.04 and P < 0.02 respectively), in patients receiving radioimmunoguided surgery. Clinical relevance of RILG is illustrated by identifying patients with high tumour CEA, most likely to benefit from antibody targeted therapy. Knowledge of the pathophysiology of many diseases may be enhanced by quantitative estimation of antigen concentration related to tissue morphology.