Some Thoughts on AI Stimulated by Michael Wooldridge's Book "The Road to Conscious Machines. The Story of AI".

Following a brief over view of the contents of Michael Wooldridge's book I give an account of my own background in computing and AI. I then cover a range of topics stimulated by reading the book including machine learning's relationship to AI, applications to medical areas, the need to consider probabilistic effects on decisions, the importance of self-reproduction and whether AI can be made moral. I finish with a discussion of the mind-brain relationship and what makes us human.

Myofibroblast activation in colorectal cancer lymph node metastases.

BACKGROUND: Myofibroblasts have an important role in regulating the normal colorectal stem cell niche. While the activation of myofibroblasts in primary colorectal cancers has been previously described, myofibroblast activation in lymph node metastases has not been described before. METHODS: Paraffin-embedded lymph node sections from patients with macrometastases, micrometastases and isolated tumour cells were stained to identify myofibroblasts and to characterise the distribution of different cell types in tumour-containing lymph nodes. The extent of myofibroblast presence was quantified and compared with the size of the metastasis and degree of proliferation and differentiation of the cancer cells. RESULTS: We show substantial activation of myofibroblasts in the presence of colorectal metastases in lymph nodes, which is intimately associated with glandular structures, both in micro- and macrometastases. The degree of activation is positively associated with the size of the metastases and the proportion of Ki67+ve cancer cells, and negatively associated with the degree of enterocyte differentiation as measured by CK20 expression. CONCLUSION: The substantial activation of myofibroblasts in tumour-containing lymph nodes strongly suggests that these metastatic cancer cells are still significantly dependent on their microenvironment. Further understanding of these epithelial-mesenchymal interactions could lead to the development of new therapies in metastatic disease.

5-Fluorouracil response in a large panel of colorectal cancer cell lines is associated with mismatch repair deficiency.

BACKGROUND: Colorectal cancer is (CRC) one of the commonest cancers and its therapy is still based on few drugs. Currently, no biological criteria are used to choose the most effective of the established drugs for treatment. METHODS: A panel of 77 CRC cell lines was tested for sensitivity to 5-fluorouracil (5FU) using the SRB assay. The responses were grouped into three categories and correlated with genetic changes in the cell lines. RESULTS: The strongest and most clearcut correlation was between 5-fluorouracil response and replication error status (mismatch repair deficiency). All the other significant correlations (loss of heterozygosity for DCC and mutations in TGFbIIR) are secondary to the association with replication error status. INTERPRETATION AND CONCLUSION: Our findings validate previous analyses based mainly on clinical data, and indicate that replication error status could be a useful guide to 5-fluorouracil-based CRC therapy. Essentially, all previously described correlations with 5FU response are secondary to the association with replication error status.

An update to HLA nomenclature, 2010.

The WHO Nomenclature Committee for Factors of the HLA System met during the 15th International Histocompatibility and Immunogenetics Workshop in Buzios, Brazil in September 2008. This update is an extract of the main report that documents the additions and revisions to the nomenclature of human leukocyte antigen (HLA) specificities following the principles established in previous reports.

Humanised IgG1 antibody variants targeting membrane-bound carcinoembryonic antigen by antibody-dependent cellular cytotoxicity and phagocytosis.

BACKGROUND: The effect of glycoengineering a membrane specific anti-carcinoembryonic antigen (CEA) (this paper uses the original term CEA for the formally designated CEACAM5) antibody (PR1A3) on its ability to enhance killing of colorectal cancer (CRC) cell lines by human immune effector cells was assessed. In vivo efficacy of the antibody was also tested. METHODS: The antibody was modified using EBNA cells cotransfected with beta-1,4-N-acetylglucosaminyltransferase III and the humanised hPR1A3 antibody genes. RESULTS: The resulting alteration of the Fc segment glycosylation pattern enhances the antibody's binding affinity to the FcgammaRIIIa receptor on human immune effector cells but does not alter the antibody's binding capacity. Antibody-dependent cellular cytotoxicity (ADCC) is inhibited in the presence of anti-FcgammaRIII blocking antibodies. This glycovariant of hPR1A3 enhances ADCC 10-fold relative to the parent unmodified antibody using either unfractionated peripheral blood mononuclear or natural killer (NK) cells and CEA-positive CRC cells as targets. NK cells are far more potent in eliciting ADCC than either freshly isolated monocytes or granulocytes. Flow cytometry and automated fluorescent microscopy have been used to show that both versions of hPR1A3 can induce antibody-dependent cellular phagocytosis (ADCP) by monocyte-derived macrophages. However, the glycovariant antibody did not mediate enhanced ADCP. This may be explained by the relatively low expression of FcgammaRIIIa on cultured macrophages. In vivo studies show the efficacy of glycoengineered humanised IgG1 PR1A3 in significantly improving survival in a CRC metastatic murine model. CONCLUSION: The greatly enhanced in vitro ADCC activity of the glycoengineered version of hPR1A3 is likely to be clinically beneficial.

Detection of circulating tumour cells in peripheral blood with an automated scanning fluorescence microscope.

We have developed an automated, highly sensitive and specific method for identifying and enumerating circulating tumour cells (CTCs) in the blood. Blood samples from 10 prostate, 25 colorectal and 4 ovarian cancer patients were analysed. Eleven healthy donors and seven men with elevated serum prostate-specific antigen (PSA) levels but no evidence of malignancy served as controls. Spiking experiments with cancer cell lines were performed to estimate recovery yield. Isolation was performed either by density gradient centrifugation or by filtration, and the CTCs were labelled with monoclonal antibodies against cytokeratins 7/8 and either AUA1 (against EpCam) or anti-PSA. The slides were analysed with the Ikoniscope robotic fluorescence microscope imaging system. Spiking experiments showed that less than one epithelial cell per millilitre of blood could be detected, and that fluorescence in situ hybridisation (FISH) could identify chromosomal abnormalities in these cells. No positive cells were detected in the 11 healthy control samples. Circulating tumour cells were detected in 23 out of 25 colorectal, 10 out of 10 prostate and 4 out of 4 ovarian cancer patients. Five samples (three colorectal and two ovarian) were analysed by FISH for chromosomes 7 and 8 combined and all had significantly more than four dots per cell. We have demonstrated an Ikoniscope based relatively simple and rapid procedure for the clear-cut identification of CTCs. The method has considerable promise for screening, early detection of recurrence and evaluation of treatment response for a wide variety of carcinomas.

EpCAM and gpA33 are markers of Barrett's metaplasia.

AIMS: To characterise a specific and sensitive marker of Barrett's metaplasia (BM). METHODS: Cases of normal oesophageal squamous mucosa (11 fresh endoscopic biopsies and 10 formalin fixed, paraffin embedded tissue blocks), BM (11 biopsies and 11 tissue blocks), and normal gastric body mucosa (five biopsies and five tissue blocks) were analysed using reverse transcriptase PCR, Western blotting, and immunohistochemistry for EpCAM, and reverse transcriptase PCR for gpA33. RESULTS: Strong EpCAM mRNA expression was detected in all the BM cases, in contrast to weak expression in all the normal gastric mucosal samples and no expression in any of the normal oesophageal mucosal samples tested. Strong gpA33 mRNA expression was detected in all the BM cases, in contrast to weak expression in a quarter of the normal gastric mucosal samples and no expression in any of the normal oesophageal mucosal samples tested. Western blotting showed EpCAM protein expression in all the BM cases and in none of the normal gastric or oesophageal mucosal samples tested. Immunohistochemistry showed strong EpCAM protein expression in BM and complete absence of expression in normal oesophageal squamous epithelium. Scattered EpCAM expressing cells were found in the gland bases of normal gastric body mucosa. CONCLUSIONS: EpCAM protein and gpA33 mRNA expressions are specific and sensitive markers of BM.

A mutated HLA-A*0101 allele in the colorectal cell line HCA-7.

The colorectal cell line HCA-7 expresses surface human leucocyte antigen-A*0201 (HLA-A*0201), but lacks expression of HLA-A*0101 whilst the normal B-cell line (EVA-1224), derived from the same individual, expresses both surface HLA-A1 and HLA-A2. Amplification refractory mutation system-polymerase chain reaction analysis, using sequence-specific primers, suggested that HCA-7 has a mutation in a 7 base pair (bp) cytosine repeat sequence located at the beginning of Exon 4 (bp 621-627). Cloning and sequencing revealed HCA-7 to have eight cytosine residues in this repeat sequence. In contrast, EVA-1224 contained only 7 cytosines. Analysis of the mRNA for HLA-A*010 using reverse trancriptase-polymerase chain reaction (RT-PCR), with an allele-specific 5' primer in exon 2 (bp 253-271) and a series of 3' primers in exons 3, 4 and 7 and in the 3'untranslated region, revealed that HCA-7 contained a shortened message terminating in the region of the exon 3/4 boundary. The insertion of an extra cytosine in this region, which is only two bases from the exon 3/4 splice site, is presumed to lead to a splicing defect between exons 3 and 4 resulting in the lack of expression of a functional HLA-A*0101 product. HCA-7 is mismatch repair (MMR) defective due to lack of expression of hMLH1 resulting from hypermethylation of the promoter region. The consequential increase in errors in single-nucleotide repeat stretches of DNA can account for the HLA-A*0101 mutation. This has probably then been selected for in the tumour to enable escape from immune attack against an HLA-A*0101-restricted tumour-specific determinant that has also arisen as a result of the tumour being MMR defective.

CDX1 is an important molecular mediator of Barrett's metaplasia.

The molecular pathogenesis of Barrett's metaplasia (BM) of the esophagus is poorly understood. The change to an intestinal phenotype occurs on a background of esophagitis due to refluxing acid and bile. CDX1, an important regulator of normal intestinal development, was studied as a potential key molecule in the pathogenesis of BM. CDX1 mRNA and protein were universally expressed in all samples of BM tested but not in normal esophageal squamous or gastric body epithelia. This tissue-specific expression was attributable to the methylation status of the CDX1 promoter. Conjugated bile salts and the inflammatory cytokines TNF-alpha and IL-1beta were all found to increase CDX1 mRNA expression in vitro. These effects were primarily mediated by NF-kappaB signaling but only occurred when the CDX1 promoter was unmethylated or partially methylated. The data suggest that CDX1 is a key molecule linking etiological agents of BM to the development of an intestinal phenotype. Although the initial trigger for CDX1 promoter demethylation is not yet identified, it seems likely that demethylation of its promoter may be the key to the induction and maintenance of CDX1 expression and so of the BM phenotype.

Immune responses in advanced colorectal cancer following repeated intradermal vaccination with the anti-CEA murine monoclonal antibody, PR1A3: results of a phase I study.

BACKGROUND AND AIMS: The aim was to determine the toxicity, clinical and immune responses to the murine monoclonal anti-carcinoembryonic antigen (CEA) antibody, PR1A3, in patients with advanced colorectal cancer. MATERIALS AND METHODS: Fifteen patients with advanced colorectal cancer received either 0.5-, 1.0- or 5.0-mg doses of PR1A3 mixed with 10% w/v Alum adjuvant (Superfos Biosector, Denmark) intradermally at 4-week intervals for 3 months. Patient serum was assessed for anti-idiotypic (Ab2), anti-anti-idiotypic (Ab3) and human anti-mouse antibody (HAMA) reactivity. Peripheral blood mononuclear cell (PBMC) proliferation with phytohaemagglutinin (PHA), CEA and PR1A3, stimulated IL-2, IL-4 and IFN-gamma levels and PR1A3-stimulated IL-2 receptor expression during immunotherapy were determined. Comparisons were made with 16 age-matched controls without malignant disease. RESULTS: Hyperimmune sera from 12 of the 15 patients showed Ab2 reactivity with no detectable Ab3 responses. Strong HAMA reactivity was recorded in 7 of the 15 cases with no adverse clinical effect. Delayed-type hypersensitivity (DTH) responses developed in 12 of the 15 patients. Pre-treatment PBMC proliferation with PHA was subnormal in each patient compared with controls, becoming normal (or supranormal) in all patients during immunisation (P<0.001). PBMC proliferation with CEA and PR1A3 increased during immunotherapy (P<0.001) along with stimulated production of IL-2, IFN-gamma and IL-2 receptor expression. Progressive disease was observed in 14 of the 15 patients with minimal toxicity. CONCLUSION: PR1A3 generated limited idiotypic responses but robust DTH reactivity in most patients. In vitro PBMC proliferation with mitogens and recall antigens is greatly increased during the course of immunisation, with a shift in stimulated cytokine profile.

Preclinical application of radioimmunoguided surgery using anti-carcinoembryonic antigen biparatopic antibody in the colon cancer.

Radioimmunoguided surgery (RIGS) has been known as a sophisticated tool to detect micrometastasis intraoperatively. A preclinical model of RIGS was designed to test the possible clinical applicability of the biparatopic antibody in detecting colorectal cancer. The biparatopic antibody was constructed using two anti-carcinoembryonic antigen (CEA)-specific antibodies, T84.66 and PR1A3, reacting against two different epitopes. (125)I-labeled biparatopic antibody was introduced via the principal colonic arteries at the end of operation in 10 operable patients with colon cancer. After 24 h, the radioactivities of the tumors and lymph nodes were counted using the gamma-detecting probe. The radioactivity count was performed ex vivo. The accurate detection in the primary tumors and metastatic lymph nodes were 100 and 88.7% respectively. False-positive detections occurred in 24 of 256 lymph nodes (9.4%), whereas false-negative detections occurred in 5 of them (2%). The most frequent cause of false-positive detection was dissociated radionuclides trapped in the lymphatic tissues. False-negative detections occurred mainly from weak targeting by radiolabeled antibody, probably due to weak expression of tumor CEA. Conclusively, as most detection errors appear to be reduced within 3 days in vivo, the biparatopic antibody can efficiently be applied to the clinical RIGS, thereby facilitating accurate detection and removal of occult cancer foci in colorectal cancer.

Loss of CDX1 expression in colorectal carcinoma: promoter methylation, mutation, and loss of heterozygosity analyses of 37 cell lines.

Expression of the homeobox protein CDX1 is lost or reduced in a significant proportion of colorectal carcinomas (CRCs) but the underlying mechanism for this is unclear. We have demonstrated absence of CDX1 mRNA expression in 7 of 37 CRC cell lines and shown that all 7 cell lines have a methylated CDX1 promoter. Twenty-five cell lines showed both CDX1 mRNA expression and an unmethylated CDX1 promoter. The five remaining cell lines had a partially methylated CDX1 promoter and all expressed CDX1 mRNA; when treated with the demethylating agent, 5-aza-2'-deoxycytidine, these five cell lines all showed increased CDX1 expression. No mutations were found in the promoter and coding regions of CDX1. One polymorphism was demonstrated in each of the promoter, 5' UTR, and coding region of exon 1 of CDX1, but there were no associations between CDX1 mRNA expression and different polymorphic genotypes. Similarly, there was no association between CDX1 mRNA expression and loss of heterozygosity at the CDX1 locus. In conclusion, absence or reduction of CDX1 expression in CRC seems to be primarily regulated by promoter methylation and is probably selected for because of its impact on the differentiation of colonocytes.

An ancestral Ashkenazi haplotype at the HMPS/CRAC1 locus on 15q13-q14 is associated with hereditary mixed polyposis syndrome.

The putative locus for hereditary mixed polyposis syndrome (HMPS) in a large family of Ashkenazi descent (SM96) was previously reported to map to chromosome sub-bands 6q16-q21. However, new clinical data, together with molecular data from additional family members, have shown 6q linkage to be incorrect. A high-density genomewide screen for the HMPS gene was therefore performed on SM96, using stringent criteria for assignment of affection status to minimize phenocopy rates. Significant evidence of linkage was found only on a region on chromosome 15q13-q14. Since this region encompassed CRAC1, a locus involved in inherited susceptibility to colorectal adenomas and carcinomas in another Ashkenazi family (SM1311), we determined whether HMPS and CRAC1 might be the same. We found that affected individuals from both families shared a haplotype between D15S1031 and D15S118; the haplotype was rare in the general Ashkenazi population. A third informative family, SM2952, showed linkage of disease to HMPS/CRAC1 and shared the putative ancestral haplotype, as did a further two families, SMU and RF. Although there are probably multiple causes of the multiple colorectal adenoma and cancer phenotype in Ashkenazim, an important one is the HMPS/CRAC1 locus on 15q13-q14.

Analysis of chromosomal instability in human colorectal adenomas with two mutational hits at APC.

In vitro data show that the adenomatous polyposis coli (APC) protein associates with the mitotic spindle and that mouse embryonic stem cells with biallelic Apc mutations are karyotypically unstable. These findings led to suggestions that APC acts in chromosomal segregation and that APC inactivation leads to chromosomal instability (CIN). An alternative hypothesis based on allelic loss studies in colorectal adenomas proposes that CIN precedes and contributes to genetic changes at APC. We determined whether colorectal adenomas with two mutations at APC show features consistent with these models by studying 55 lesions (average size 5 mm; range 1-13 mm) from patients with familial adenomatous polyposis. A variety of methods was used depending on available material, including flow cytometry, comparative genomic hybridization, and loss of heterozygosity (LOH) analysis. Selected adenomas were assessed for proliferative activity by Ki-67 immunocytochemistry. Seventeen of 20 (85%) tumors were diploid, two were near-diploid, and one was hypotetraploid. Just one (near-diploid) tumor showed increased proliferative activity. LOH was found occasionally on chromosome 15q (2 of 49 tumors), but not on chromosome 18q (0 of 48). In 20 adenomas, LOH at APC was associated with loss at 5q but not 5p markers, with the former encompassing a minimum of 20 Mb. However, three of these lesions analyzed by comparative genomic hybridization displayed normal profiles, suggesting, together with other data, that the mechanism of LOH at APC is probably somatic recombination. Our results therefore do not support the hypothesis that CIN precedes APC mutations in tumorigenesis. Regarding the model in which APC mutations lead directly to CIN, if APC mutations do have this effect in vivo, it must be subtle. Alternatively, CIN associated with APC mutations might be essentially an in vitro phenomenon.

An insight into the genetic pathway of adenocarcinoma of the small intestine.

BACKGROUND: Although the adenoma to carcinoma pathway in colorectal cancer is well described, the mechanisms of carcinogenesis in the small intestine remain unclear. AIMS: The aim of this study was to investigate candidate genes in the genetic pathway of adenocarcinoma of the small intestine. SUBJECTS AND METHODS: A total of 21 non-familial, non-ampullary adenocarcinomas of the small intestine were analysed. DNA was extracted from formalin fixed paraffin wax embedded tissue using standard techniques. The replication error (RER) status was determined by amplification of BAT26. The mutation cluster region (MCR) of the adenomatous polyposis coli (APC) gene was screened using polymerase chain reaction single strand conformational polymorphism and direct sequencing. Immunohistochemistry was performed on formalin fixed paraffin wax embedded tissue using monoclonal antibodies for hMLH1, hMSH2, beta-catenin, E-cadherin, and p53. RESULTS: Fourteen male and seven female patients with a median age of 64 years (range 21-85) presented with adenocarcinoma of the duodenum (10), jejunum (7), and ileum (4). One cancer (5%) was found to be RER+, and all tumours stained positive for hMLH1 and hMSH2. No mutations were detected in the MCR of the APC gene. beta-Catenin showed increased nuclear expression with loss of membranous staining in 10 cancers (48%). Absent or decreased membrane expression of E-cadherin was found in eight cancers (38%). Strong staining of p53 was found in the nucleus of five cancers (24%). CONCLUSION: We did not detect mutations in the MCR of the APC gene, and this suggests that adenocarcinoma of the small intestine may follow a different genetic pathway to colorectal cancer. Abnormal expression of E-cadherin and beta-catenin was common and reflects an early alternative to APC in this pathway in which mutations may be found in adenocarcinoma of the small intestine.