Assessing agreement between multiple raters with missing rating information, applied to breast cancer tumour grading

Background

We consider the problem of assessing inter-rater agreement when there are missing data and a large number of raters. Previous studies have shown only ‘moderate’ agreement between pathologists in grading breast cancer tumour specimens. We analyse a large but incomplete data-set consisting of 24177 grades, on a discrete 1–3 scale, provided by 732 pathologists for 52 samples.

Methodology/Principal Findings

We review existing methods for analysing inter-rater agreement for multiple raters and demonstrate two further methods. Firstly, we examine a simple non-chance-corrected agreement score based on the observed proportion of agreements with the consensus for each sample, which makes no allowance for missing data. Secondly, treating grades as lying on a continuous scale representing tumour severity, we use a Bayesian latent trait method to model cumulative probabilities of assigning grade values as functions of the severity and clarity of the tumour and of rater-specific parameters representing boundaries between grades 1–2 and 2–3. We simulate from the fitted model to estimate, for each rater, the probability of agreement with the majority. Both methods suggest that there are differences between raters in terms of rating behaviour, most often caused by consistent over- or under-estimation of the grade boundaries, and also considerable variability in the distribution of grades assigned to many individual samples. The Bayesian model addresses the tendency of the agreement score to be biased upwards for raters who, by chance, see a relatively ‘easy’ set of samples.

Conclusions/Significance

Latent trait models can be adapted to provide novel information about the nature of inter-rater agreement when the number of raters is large and there are missing data. In this large study there is substantial variability between pathologists and uncertainty in the identity of the ‘true’ grade of many of the breast cancer tumours, a fact often ignored in clinical studies.     

Intramuscular beta2-agonist administration enhances early regeneration and functional repair in rat skeletal muscle after myotoxic injury.

Systemic administration of beta(2)-adrenoceptor agonists (beta(2)-agonists) can improve skeletal muscle regeneration after injury. However, therapeutic application of beta(2)-agonists for muscle injury has been limited by detrimental cardiovascular side effects. Intramuscular administration may obviate some of these side effects. To test this hypothesis, the right extensor digitorum longus (EDL) muscle from rats was injected with bupivacaine hydrochloride to cause complete muscle fiber degeneration. Five days after injury, half of the injured muscles received an intramuscular injection of formoterol (100 mug). Muscle function was assessed at 7, 10, and 14 days after injury. A single intramuscular injection of formoterol increased muscle mass and force-producing capacity at day 7 by 17 and 91%, respectively, but this effect was transient because these values were not different from control levels at day 10. A second intramuscular injection of formoterol at day 7 prolonged the increase in muscle mass and force-producing capacity. Importantly, single or multiple intramuscular injections of formoterol did not elicit cardiac hypertrophy. To characterize any potential cardiovascular effects of intramuscular formoterol administration, we instrumented a separate group of rats with indwelling radio telemeters. Following an intramuscular injection of formoterol, heart rate increased by 18%, whereas systolic and diastolic blood pressure decreased by 31 and 44%, respectively. These results indicate that intramuscular injection can enhance functional muscle recovery after injury without causing cardiac hypertrophy. Therefore, if the transient cardiovascular effects associated with intramuscular formoterol administration can be minimized, this form of treatment may have significant therapeutic potential for muscle-wasting conditions.     

CNTF genotype is associated with muscular strength and quality in humans across the adult age span.

The relationship between ciliary neurotrophic factor (CNTF) genotype and muscle strength was examined in 494 healthy men and women across the entire adult age span (20-90 yr). Concentric (Con) and eccentric (Ecc) peak torque were assessed using a Kin-Com isokinetic dynamometer for the knee extensors (KE) and knee flexors (KF) at slow (0.52 rad/s) and faster (3.14 rad/s) velocities. The results were covaried for age, gender, and body mass or fat-free mass (FFM). Individuals heterozygous for the CNTF null (A allele) mutation (G/A) exhibited significantly higher Con peak torque of the KE and KF at 3.14 rad/s than G/G homozygotes when age, gender, and body mass were covaried (P < 0.05). When the dominant leg FFM (estimated muscle mass) was used in place of body mass as a covariate, Con peak torque of the KE at 3.14 rad/s was also significantly greater in the G/A individuals (P < 0.05). In addition, muscle quality of the KE (peak torque at 3.14 rad x s(-1) x leg muscle mass(-1)) was significantly greater in the G/A heterozygotes (P < 0.05). Similar results were seen in a subanalysis of subjects 60 yr and older, as well as in Caucasian subjects. In contrast, A/A homozygotes demonstrated significantly lower Ecc peak torque at 0.52 rad/s for both KE and KF compared with G/G and G/A groups (P < 0.05). No significant relationships were observed at 0.52 rad/s between genotype and Con peak torque. These data indicate that individuals exhibiting the G/A genotype possess significantly greater muscular strength and muscle quality at relatively fast contraction speeds than do G/G individuals. Because of high positive correlations between fast-velocity peak torque and muscular power, these findings suggest that further investigations should address the relationship between CNTF genotype and muscular power.     

Molecular analysis of environmental plant DNA in house dust across the United States

Despite the prevalence and costs of allergic diseases caused by pollen, we know little about the distributions of allergenic and non-allergenic pollen inside and outside homes at the continental scale. To better understand patterns in potential pollen diversity across the United States, we used DNA sequencing of a chloroplast marker gene to identify the plant DNA found in settled dust collected on indoor and outdoor surfaces across 459 homes. House location was the best predictor of the relative abundance of plant taxa found in outdoor dust samples. Urban, southern houses in hotter climates that were further from the coast were more likely to have more DNA from grass and moss species, while rural houses in northern, cooler climates closer to the coast were more likely to have higher relative abundances of DNA from Pinus and Cedrus species. In general, those plant taxa that were more abundant outdoors were also more abundant indoors, but indoor dust had uniquely high abundances of DNA from food plants and plants associated with lawns. Approximately 14 % of the plant DNA sequences found outside were from plant taxa that are known to have allergenic pollen compared to just 8 % inside. There was little geographic pattern in the total relative abundance of these allergens highlighting the difficulties associated with trying to predict allergen exposures based on geographic location alone. Together, this work demonstrates the utility of using environmental DNA sequencing to reconstruct the distributions of plant DNA inside and outside buildings, an approach that could prove useful for better understanding and predicting plant allergen exposures.     

Transgene flow in Mexican maize revisited: Socio‐biological analysis across two contrasting farmer communities and seed management systems

The flow of transgenes into landraces and wild relatives is an important biosafety concern. The case of transgene flow into local maize varieties in Mexico (the center of origin of maize) has been intensively debated over the past 15 years, including legal, political, and environmental disputes fanned by the existence of a significant scientific controversy over the methods used for the detection of transgenes. The use of diverse approaches and a lack of harmonized methods specific to the detection and monitoring of transgenes in landraces have generated both positive and negative results regarding contamination of Mexican maize with genetically modified material over the years. In this paper, we revisit the case of transgene contamination in Mexican maize and present a novel research approach based on socio‐biological analysis of contrasting communities and seed management systems. Two communities were used to investigate how different social and biological factors can affect transgene flow and impact transgene spread in Mexico. Our results show the presence of transgenes in one community and thus support the position that transgenes are highly likely to be present in Mexican maize landraces. However, our work also demonstrates that the extent and frequency with which transgenes can be found will significantly depend on the societal characteristics and seed management systems of the local communities. Therefore, we argue that future analysis of transgene presence should include social research on the seed management practices in the sampling area so that more robust and comprehensive understandings and conclusions can be drawn.     

Nitrogen dynamics during O3-induced accelerated senescence in hybrid poplar

Experiments were conducted to determine the fate of nitrogen (N) remobilized as a result of ozone (O₃)‐induced accelerated senescence in hybrid poplar subjected to declining N availability concurrent with O₃ stress. Cuttings were grown in sand culture where the supply of N to the plant could be controlled on a daily basis and reduced in half of the plants when desired. Plants all initially received 3·57 mm N daily until approximately the 20 leaf stage after which daily supply of N was reduced to 0·71 mm . Plants were grown in open‐top chambers in the field (Rock Springs, PA, USA) and received charcoal‐filtered air, half also received supplemental O₃ to a level of 0·08 µL L^?1. Allocation of newly acquired N was determined with ¹⁵N. The specific allocation (mg labelled N mg^?1 total N) of labelled N to upper, expanding leaf N was not affected by O₃, but was strongly affected by N treatment. However, O₃ increased the relative partitioning of labelled N to the expanding leaves and the roots. The balance between partitioning of newly acquired N to the upper leaves and roots was not affected by O₃, but was reduced by N withdrawal. Calculated net N flux was strongly negative in the lower leaves of O₃‐exposed, N withdrawal plants. Nitrogen uptake was not reduced by O₃. The allometric relationships between the roots and shoots were not affected by O₃ or N availability. The relative contribution of newly acquired versus remobilized N to new growth appears to be determined by N supply. Ozone exposure alters the allocation of newly acquired N via alterations in plant size, whereas N availability exerts a strong effect upon both plant size and N allocation.     

Dissecting the Molecular Origins of Specific Protein-Nucleic Acid Recognition: Hydrostatic Pressure and Molecular Dynamics

The fundamental processes by which proteins recognize and bind to nucleic acids are critical to understanding cellular function. To explore the factors involved in protein-DNA recognition, we used hydrostatic pressure to perturb the binding of the BamHI endonuclease to cognate DNA, both in experiment and in molecular dynamic simulations. A new technique of high-pressure gel mobility shift analysis was used to test the effects of elevated hydrostatic pressure on the binding of BamHI to its cognate recognition sequence. Upon application of a pressure of 500 bar, the equilibrium dissociation constant of BamHI binding to the cognate site was found to increase nearly 10-fold. A challenge has been to link this type of pure thermodynamic measurement to functional events occurring at the molecular level. Thus, we used molecular dynamic simulations at both ambient and elevated pressures to reveal details of the direct and water-mediated interactions between BamHI and cognate DNA, which allow explanation of the effects of pressure on site-specific protein-DNA binding and complex stability.     

Seasonal changes in the distribution and abundance of benthic invertebrates in six headwater streams in central Florida

Seasonal variations in invertebrate assemblages at two sites (upstream and downstream) on six central Florida headwater streams were compared by sampling at quarterly intervals with core and dip net samplers. Two of the streams were reclaimed following phosphate mining (～6 yr prior to this study), two received runoff from mined lands, and two were disturbed by agriculture and/or residential developments. Physical and chemical characteristics of the reclaimed streams differed markedly from those of the non-reclaimed streams; principal differences between the streams were in current velocity, percent organic matter (POM), Mn, conductivity and alkalinity. Annual mean densities of meiofauna and smaller macrofauna for the 12 stream sites ranged from 20?896 to 175?212 m^?2 and the mean for all sites was 56?492 m^?2. The reclaimed streams and one of the streams influenced by agriculture had annual means of less than 40?000 m^?2, 3- to 5-fold lower than the other streams. Fall and winter core densities were ～2.4-fold greater than those for spring or summer when drought and low dissolved oxygen prevailed. Meiofauna comprised 68–91% of the core sample invertebrates in reclaimed streams but only 43–62% in the non-reclaimed streams; principal functional groups were: gathering collectors – 61.5%, predators – 19.3% and filtering collectors – 15%. The taxonomic composition of the reclaimed streams was predominated by crustaceans (60–71%) while chironomids and annelids were more abundant (71–92%) in the non-reclaimed streams. Dip net sampling added 21 larger macrofauna species (Odonata, Hemiptera and Coleoptera) to our list of taxa, producing a total of 209 species. Species richness and diversity (H′ and N ₂) indices were lower in the reclaimed streams, but evenness was more variable. The Czekanowski–Dice–Sørensen similarity index showed that the reclaimed stream sites were quite similar to each other, but differed markedly from the other stream types; there was large variation both within and between seasons. For central Florida headwater streams, drought appears to have a larger influence on invertebrates than the type of land use, however this relationship should be confirmed using streams of similar hydrology.     

Root Gravitropism and Below-ground Competition among Neighbouring Plants: A Modelling Approach

Competition for nutrients among neighbouring roots occurs whentheir individual depletion volumes overlap, causing a reductionin nutrient uptake. By exploring different spatial niches, plantswith contrasting root architecture may reduce the extent ofcompetition among neighbouring root systems. The main objectivesof this study were: (1) to evaluate the impact of root architectureon competition for phosphorus among neighbouring plants; and(2) to compare the magnitude of competition among roots of thesame plant vs. roots of neighbouring plants. SimRoot, a dynamicgeometric model, was used to simulate common bean root growthand to compare the overlap of depletion volumes. By varyingthe gravitropism of basal roots, we simulated three distinctroot architectures: shallow, intermediate and deep, correspondingto observed genetic variation for root architecture in thisspecies. Combinations of roots having the same architectureresulted in more intense inter-plant competition. Among them,the deep-deep combination had the most intense competition.Competition between deep root systems and shallow root systemswas only half that of deep root systems competing with otherdeep root systems. Inter-plant root competition increased assoil diffusivity increased and the distance among plants decreased.In heterogeneous soils, co-localization of soil resources androots was more important in determining resource uptake thaninter-plant root competition. Competition among roots of thesame plant was three- to five-times greater than competitionamong roots of neighbouring plants. Genetic variation for rootarchitecture in common bean may be related to adaptation todiverse competitive environments. Copyright 2001 Annals of BotanyCompany Root architecture, phosphorus, competition, common bean, Phaseolus vulgaris L. nutrient uptake, gravitropism