Volume 37, Issue 9

Since the early nineties the number of scientific papers reporting on artificial neural network (ANN) applications in medicine has been quickly increasing. In the present paper, we describe in some detail the architecture of network types used most frequently in ANN applications in the broad field of laboratory medicine and clinical chemistry, present a technique-structured review about the recent ANN applications in the field, and give information about the improvements of available ANN software packages. ANN applications are divided into two main classes: supervised and unsupervised methods. Most of the described supervised applications belong to the fields of medical diagnosis (n=7) and outcome prediction (n=9). Laboratory and clinical data are presented to multilayer feed-forward ANNs which are trained by the back propagation algorithm. Results are often better than those of traditional techniques such as linear discriminant analysis, classification and regression trees (CART), Cox regression analysis, logistic regression, clinical judgement or expert systems. Unsupervised ANN applications provide the ability of reducing the dimensionality of a dataset. Low-dimensional plots can be generated and visually understood and compared. Results are very similar to that of cluster analysis and factor analysis. The ability of Kohonen's self-organizing maps to generate 2D maps of molecule surface properties was successfully applied in drug design.

Mitochondria play a pivotal role in cellular metabolism and in energy production in particular. Defects in structure or function of mitochondria, mainly involving the oxidative phosphorylation (OXPHOS), mitochondrial biogenesis and other metabolic pathways, have been shown to be associated with a wide spectrum of clinical phenotypes. The ubiquitous nature of mitochondria and their unique genetic features contribute to the clinical, biochemical and genetic heterogeneity of mitochondrial diseases. We will focus on the recent advances in the field of mitochondrial disorders and their consequences for an advanced clinical and genetic diagnostics. In addition, an overview on recently identified genetic defects and their pathogenic molecular mechanisms will be given.

In human carcinomas, mutations that alter tumour genes such as the K RAS , P53 , or APC genes, are mostly point mutations. The detection of mutant alleles of tumour genes in specimens such as urine, pancreatic juice, sputum, and stool holds great promise for an early diagnosis of cancer. In addition, the detection of mutant tumour genes in tissue samples, such as lymph nodes or resection margins, may allow a sensitive diagnosis of residual malignant disease. However, the reliable detection of mutant alleles in excess of wild type alleles remains an unresolved analytical problem when the mutations are not known a priori . In the present communication, a new approach is described which makes possible the detection of unknown point mutations in tumour genes at excess of wild type alleles. The method is based on the removal of wild type alleles by hybridisation to immobilised complementary oligonucleotides. Using this approach, an enrichment of mutant K RAS , P53 and APC alleles of one mutant in up to 10 3 normal alleles has been achieved. Parallel miniaturised separation units with oligonucleotides complementary to defined sequences of a wild type allele should allow the detection of unknown point mutations as well as small insertions or deletions which occur in the sequence range covered by the oligonucleotides.

About one-third of hereditary non-polyposis colorectal cancer-related mutations in the mismatch repair gene hMLH1 result in the loss of entire exons from the wild type transcripts. Here we describe quantitative differences of hMLH1 transcripts without exon 15, exon 16 or exon 17 in several members of a family with hereditary non-polyposis colorectal cancer. The transcript lacking exon 15 is caused by a G to A transition affecting the last nucleotide of the respective exon and results in a truncated protein. The transcripts lacking exon 16 or exon 17, which are in-frame deletions, were also found in all tested samples of a normal population and represent common isoforms. Reverse transcription-polymerase chain reaction-based relative quantification revealed about 50 % signal intensity for the mutation-based transcript, but less than 10 % for the common isoforms, if compared to the wild type. All aberrant transcripts were detected from blood-derived cDNAs but not from samples of normal colon epithelium. Although the biological significance of the common isoforms is unknown, they might lead to false risk assessment in hereditary non-polyposis colorectal cancer cases.

Cardiac troponin I assays for Axsym (Abbott Diagnostics, Abbott Park, IL, USA) and Immuno 1 (Bayer Corporation, Tarrytown, NY, USA) analysers were evaluated. Heparin plasma or serum could be used for both assays. Samples were stable for 24 h at ambient temperature, 3 days at 4–8 ℃ and 3 months at −20 °C. After 10 months' storage at −80 °C, the recoveries were well above 100 % by both assays. Total coefficients of variation for Axsym assay were 9.0 %, 5.8 % and 5.3 % at concentrations of 2.6 μg/l, 9.83 μg/l and 34.3 μg/l respectively; for Immuno 1 these were 4.4 %, 1.6 % and 1.8 % at 2.3 μg/l,6.27 μg/l and 44.35 μg/l respectively. It was ≥ 20 % at concentration of ≤ 0.5 μg/l for Axysm assay and ≤ 0.15 μg/l for Immuno 1 assay. Recoveries were ≤ 90 % at ≤ 0.22 μg/l on Axsym and at ≤ 1.47 μg/l on Immuno 1. Neither method showed significant interference with haemoglobin, bilirubin, triglycerides or rheumatoid factor. Correlation between the two methods was excellent (r = 0.997, Y (Axsym) = 4.2X (Immuno 1) + 3.2). The highest concentrations detected in 50 healthy subjects were 0.3 μg/l and 0.1 μg/l by Axsym and Immuno 1 methods, respectively. Twelve out of 43 renal failure patients had troponin I 0.13–0.9 μg/l using Axsym method and 4 had levels of 0.07–0.13 μg/l using Immuno 1. In muscle trauma patients, troponin I was undetectable.

We have developed a new procedure for turbidimetric measurement of ferritin concentration in human serum, based on latex microparticle agglutination technology. The procedure has been automated using the Falcor 300 analyzer. Carboxilated latex particles (336 nm in diameter) were covalently coupled with immunopurified F(ab′) 2 fragments of anti-ferritin IgG antibodies. Coated microparticles were automatically mixed with undiluted sample and the resulting absorbance due to agglutination was measured at 550 nm. The procedure generated a calibration curve with a measuring range of 0 to 558 μg/l, showing a day-today imprecision lower than 5.7 %. The detection limit was 4 μg/l. There were no interferences from bilirubin, hemoglobin or rheumatoid factors. Turbid and lipemic samples caused an important interference which could be avoided by pretreating those samples prior to measurement. A prozone effect was provisionally obtained with ferritin concentrations over 1800 μg/l. The results suggested a hook-like effect due to a rapid microparticle precipitation in the reaction media, that could be avoided by increasing the reaction medium density by adding sucrose to the buffer, up to 150 g/l concentration. This sucrose addition resulted in a displacement of the Heidelberger curve with a prozone phenomenon occuring at concentration higher than 3000 μg/l of ferritin. Results obtained with the present procedure correlated well with those obtained by a nephelometric procedure and with those obtained by an RIA. We conclude that this latex turbidimetric immunochemical procedure is simple, rapid, has a good analytical and operational performance on the Falcor 300 analyzer and is well suited for the measurement of ferritin concentration in human serum.

Revised recommendations for diagnosis of diabetes introduce the intermediary risk group of impaired fasting glucose (IFG), defined as individuals with a fasting blood-glucose concentration between 5.6 and 6.0 mmol/l. We apply the concept of uncertainty to identifiable steps of sampling and measuring blood-glucose. Since many instruments in primary health care measure plasma-glucose and report results as blood-glucose and vice versa , factors affecting the transformation are also considered. The study identifies the measurement procedure as the major source of uncertainty, closely followed by preanalytical sources. The estimated uncertainties indicate that the presently available procedures do not allow identification of IFG by a single investigation. The approach to establish an uncertainty budget can be used to evaluate the clinical usefulness of measurements.

The study presents a comparison of data concerning lipid metabolism and lipid oxidation (oxidative stress) in children at the time of their birth and 3 months later, as well as of their mothers at the time of delivery, compared to a control group of non-pregnant women of the same age. The data confirm that labour represents an oxidative stress for both mother and child; it is expressed as a significant increase of malondialdehyde concentration in mothers immediately after delivery in comparison with non-pregnant women (p < 0.001). Its concentration in newborns was even higher than in their mothers (p < 0.005). Concentration of antibodies against oxidized LDL (oxLDLAb) was comparable in mothers and newborns due to their transplacental transport. During the first three months of life these autoantibodies increased almost two-fold. The importance of this unique observation is discussed with respect to possible early atherogenesis.

The following recommendations were drawn up by the Working Group “Point-of-Care Testing” of the DGKC and DGLM which was set up in 1997. This first document from the Working Group sets out the principles which should be observed when introducing point-of-care testing. These general recommendations are to be followed by further specific recommendations on individual procedures or groups of procedures, e.g. blood glucose, electrolytes in whole blood, blood gases, coagulation, toxicology, quality control and others, which will be drawn up by experts at the suggestion of the Working Group.