Development of a deep learning model for the automated detection of green pixels indicative of gout on dual energy CT scan.

Background: Dual-energy CT (DECT) is a non-invasive way to determine the presence of monosodium urate (MSU) crystals in the workup of gout. Color-coding distinguishes MSU from calcium following material decomposition and post-processing. Most software labels MSU as green and calcium as blue. There are limitations in the current image processing methods of segmenting green-encoded pixels. Additionally, identifying green foci is tedious, and automated detection would improve workflow. This study aimed to determine the optimal deep learning (DL) algorithm for segmenting green-encoded pixels of MSU crystals on DECTs. Methods: DECT images of positive and negative gout cases were retrospectively collected. The dataset was split into train (N = 28) and held-out test (N = 30) sets. To perform cross-validation, the train set was split into seven folds. The images were presented to two musculoskeletal radiologists, who independently identified green-encoded voxels. Two 3D Unet-based DL models, Segresnet and SwinUNETR, were trained, and the Dice similarity coefficient (DSC), sensitivity, and specificity were reported as the segmentation metrics. Results: Segresnet showed superior performance, achieving a DSC of 0.9999 for the background pixels, 0.7868 for the green pixels, and an average DSC of 0.8934 for both types of pixels, respectively. According to the post-processed results, the Segresnet reached voxel-level sensitivity and specificity of 98.72 % and 99.98 %, respectively. Conclusion: In this study, we compared two DL-based segmentation approaches for detecting MSU deposits in a DECT dataset. The Segresnet resulted in superior performance metrics. The developed algorithm provides a potential fast, consistent, highly sensitive and specific computer-aided diagnosis tool. Ultimately, such an algorithm could be used by radiologists to streamline DECT workflow and improve accuracy in the detection of gout.

Accuracy in Facial Trustworthiness Impressions: Kernel of Truth or Modern Physiognomy? A Meta-Analysis.

Being able to identify trustworthy strangers is a critical social skill. However, whether such impressions are accurate is debatable. Critically, the field currently lacks a quantitative summary of the evidence. To address this gap, we conducted two meta-analyses. We tested whether there is a correlation between perceived and actual trustworthiness across faces, and whether perceivers show above-chance accuracy at assessing trustworthiness. Both meta-analyses revealed significant, modest accuracy (face level, r = .14; perceiver level, r = .27). Perceiver-level effects depended on domain, with aggressiveness and sexual unfaithfulness having stronger effects than agreeableness, criminality, financial reciprocity, and honesty. We also applied research weaving to map the literature, revealing potential biases, including a preponderance of Western studies, a lack of "cross-talk" between research groups, and clarity issues. Overall, this modest accuracy is unlikely to be of practical utility. Moreover, we strongly urge the field to improve reporting standards and generalizability of the results.

Estimation of inhibitory quotient using a comparative equilibrium dialysis assay for prediction of viral response to hepatitis C virus inhibitors.

The relationship of inhibitory quotient (IQ) with the virologic response to specific inhibitors of human hepatitis C virus (HCV) and the best method to correct for serum protein binding in calculating IQ have not been addressed. A common method is to determine a fold shift by comparing the EC(50) values determined in cell culture in the absence and presence of human serum (fold shift in EC(50) ), but this method has a number of disadvantages. In the present study, the fold shifts in drug concentrations between 100% human plasma (HP) and cell culture medium (CCM) were directly measured using a modified comparative equilibrium dialysis (CED) assay for three HCV protease inhibitors (PIs) and for a novel HCV inhibitor GS-9132. The fold shift values in drug concentration between the HP and CCM (CED ratio) were ∼1 for SCH-503034, VX-950 and GS-9132 and 13 for BILN-2061. These values were ∼3-10-fold lower than the fold shift values calculated from the EC(50) assay for all inhibitors except BILN-2061. Using the CED values, a consistent pharmacokinetic and pharmacodynamic relationship was observed for the four HCV inhibitors analysed. Specifically, an approximate 1 log(10) reduction in HCV RNA was achieved with an IQ close to 1, while 2-3 and greater log(10) reductions in HCV RNA were achieved with IQ values of 3-5 and greater, respectively. Thus, use of CED to define IQ provides a predictive and quantitative approach for the assessment of the in vivo potency of HCV PIs and GS-9132. This method provides a framework for the evaluation of other classes of drugs that are bound by serum proteins but require the presence of serum for in vitro evaluation.

Autoantibodies in infectious mononucleosis have specificity for the glycine-alanine repeating region of the Epstein-Barr virus nuclear antigen.

Viruses have been postulated to be involved in the induction of autoantibodies by: autoimmunization with tissue proteins released by virally induced tissue damage; immunization with virally encoded antigens bearing molecular similarities to normal tissue proteins; or nonspecific (polyclonal) B cell stimulation by the infection. Infectious mononucleosis (IM) is an experiment of nature that provides the opportunity for examining these possibilities. We show here that IgM antibodies produced in this disease react with at least nine normal tissue proteins, in addition to the virally encoded Epstein-Barr nuclear antigen (EBNA-1). The antibodies are generated to configurations in the glycine-alanine repeat region of EBNA-1 and are crossreactive with the normal tissue proteins through similar configurations, as demonstrated by the effectiveness of a synthetic glycine-alanine peptide in inhibiting the reactions. The antibodies are absent in preillness sera and gradually disappear over a period of months after illness, being replaced by IgG anti-EBNA-1 antibodies that do not crossreact with the normal tissue proteins but that are still inhibited by the glycine-alanine peptide. These findings are most easily explained by either a molecular mimicry model of IgM autoantibody production or by the polyclonal activation of a germline gene for a crossreactive antibody. It also indicates a selection of highly specific, non-crossreactive anti-EBNA-1 antibodies during IgM to IgG isotype switching.

Infection of human thymocytes by Epstein-Barr virus.

The Epstein-Barr Virus (EBV) causes infectious mononucleosis, and has been strongly associated with certain human cancers. The virus is thought to exclusively bind to B lymphocytes and epithelial cells via receptors (CR2/CD21) that also interact with fragments of the third component of complement (C3). Recent evidence, however, has challenged this belief. We have used two-color immunofluorescence analysis using biotin-conjugated EBV and streptavidin-phycoerythrin along with fluorescein-conjugated anti-T cell antibodies and demonstrated that CD1-positive, CD3-dull (immature) human thymocytes express functional EBV receptors. In four replicate experiments, the binding of EBV to thymocytes ranged between 8 and 18%. This interaction is specific as evidenced by inhibition with nonconjugated virus, anti-CR2 antibodies, aggregated C3, and an antibody to the gp350 viral glycoprotein that the virus uses to bind to CR2. EBV can infect the thymocytes as evaluated by the presence of episomal EBV-DNA in thymocytes that had been incubated with the virus as short as 12 days or as long as 6 weeks. Episomal DNA analysis was performed by Southern blotting with a EBV-DNA probe that hybridizes to the first internal reiteration of the viral DNA. The presence of the EBV genome is also supported by the detection of EBV nuclear antigen 1 in infected thymocytes, assessed by Western blotting with EBV-immune sera. The EBV infection is specific as determined by blocking experiments using anti-CR2 and anti-gp350 antibodies. Finally, virus infection of thymocytes can act synergistically along with interleukin 2 and induce a lymphokine-dependent cellular proliferation. In view of previously reported cases of EBV-positive human T cell lymphomas, the possibility is raised that EBV may be involved in cancers of T lymphocytes that have not been previously appreciated.

Dissolved organic nutrient uptake by riverine phytoplankton varies along a gradient of nutrient enrichment.

The concentration of dissolved organic matter (DOM) in freshwaters is increasing in large areas of the world. In addition to carbon, DOM contains nitrogen and phosphorus and there is growing concern that these organic nutrients may be bioavailable and contribute to eutrophication. However, relatively few studies have assessed the potential for dissolved organic nitrogen (DON) or dissolved organic phosphorus (DOP) compounds to be bioavailable to natural river phytoplankton communities at different locations or times. Temporal and spatial variations in uptake, relative to environmental characteristics were examined at six riverine sites in two contrasting catchments in the UK. This study also examined how the uptake by riverine phytoplankton of four DON and four DOP compounds commonly found in rivers, varied with concentration. Total nitrogen (TN) and phosphorus (TP) concentrations, the proportion of inorganic nutrient species, and nutrient limitation varied temporally and spatially, as did the potential for DON and DOP uptake. All eight of the DOM compounds tested were bioavailable, but to different extents. Organic nutrient use depended on the concentration of the organic compound supplied, with simple compounds (urea and glucose-6-phosphate) supporting algal growth even at very low concentrations. DON use was negatively correlated with the TN and ammonia concentration and DOP use was negatively correlated with soluble reactive phosphorus (SRP) and dissolved organic carbon (DOC) concentration. The evidence indicates that DOM in rivers has been overlooked as a potential source of nutrients to phytoplankton and therefore as an agent of eutrophication.

Heterologous protection against influenza by injection of DNA encoding a viral protein.

Cytotoxic T lymphocytes (CTLs) specific for conserved viral antigens can respond to different strains of virus, in contrast to antibodies, which are generally strain-specific. The generation of such CTLs in vivo usually requires endogenous expression of the antigen, as occurs in the case of virus infection. To generate a viral antigen for presentation to the immune system without the limitations of direct peptide delivery or viral vectors, plasmid DNA encoding influenza A nucleoprotein was injected into the quadriceps of BALB/c mice. This resulted in the generation of nucleoprotein-specific CTLs and protection from a subsequent challenge with a heterologous strain of influenza A virus, as measured by decreased viral lung titers, inhibition of mass loss, and increased survival.

Does attractiveness in men provide clues to semen quality?

The psychological mechanisms underlying attractiveness judgements in humans are thought to be evolved adaptations for finding a high quality mate. The phenotype-linked fertility hypothesis proposes that females obtain reliable information on male fertility from male expression of sexual traits. A previous study of Spanish men reported that facial attractiveness was positively associated with semen quality. We aimed to determine whether this effect was widespread by examining a large sample of Australian men. We also extended our study to determine whether cues to semen quality are provided by components of attractiveness: masculinity, averageness and symmetry. Each male participant was photographed and provided a semen sample that was analyzed for sperm morphology, motility and concentration. Two independent sets of women rated the male photographs for attractiveness, and three further sets of 12 women rated the photographs for masculinity, symmetry or averageness. We found no significant correlations between semen quality parameters and attractiveness or attractive traits. Although male physical attractiveness may signal aspects of mate quality, our results suggest that phenotype-linked cues to male fertility may not be general across human populations.

Rheumatoid arthritis synovial membrane contains a 62,000-molecular-weight protein that shares an antigenic epitope with the Epstein-Barr virus-encoded associated nuclear antigen.

A monoclonal antibody, selected for reactivity with the Epstein-Barr virus (EBV)-encoded antigen EBNA-1, exhibited strong reactivity with the synovial lining cells in joint biopsies from 10 of 12 patients with rheumatoid arthritis (RA) and adherent cells eluted from these tissues. No staining of RA synovial membrane frozen tissue sections or eluted synovial-lining cells was obtained with monoclonal antibodies directed against other EBV-encoded antigens (anti-p160, anti-gp200/350) or with monoclonal antibodies directed against antigens encoded by cytomegalovirus, herpes simplex viruses, or human T cell leukemia virus type I. Among 12 osteoarthritis and normal synovial biopsies only rare reactive cells were noted. Characterization of the antigen(s) in RA synovium by the Western immunoblotting technique revealed a 62,000-molecular-weight (mol-wt) protein, in contrast to the 70,000-85,000-mol-wt EBNA-1 antigen found in EBV-transformed cells. The structural basis for the cross-reactivity of the RA synovial membrane 62,000-mol-wt protein and the EBNA-1 antigen appears to reside in the glycine-alanine rich region of these molecules. A rabbit antibody directed against a synthetic peptide (IR3-VI-2) derived from the glycine-alanine-rich region of EBNA-1 reacted with the 70,000-85,000-mol-wt EBNA-1 antigen in EBV-infected cells and with the 62,000-mol-wt molecule in RA synovial membrane extracts. Since strong antibody responses to EBNA-1 are known to exist in RA patients, these results suggest that immune responses to a cross-reactive antigen may play a role in the pathogenesis of RA.

Epstein-Barr virus-induced autoimmune responses. II. Immunoglobulin G autoantibodies to mimicking and nonmimicking epitopes. Presence in autoimmune disease.

During infectious mononucleosis, IgM autoantibodies are generated to a protein, p542, which contains a glycine-rich 28-mer epitope cross-reactive with the Epstein-Barr nuclear antigen-1 through Epstein-Barr nuclear antigen-1's glycine/alanine repeat. In normal individuals it is uncommon to find IgG anti-p542, but among patients with progressive systemic sclerosis, systemic lupus erythematosus, and ulcerative colitis high IgG anti-p542 (> 3 SD above the mean of normal 20-50 yr controls) occurred frequently. Lesser elevations occurred in Sjögren's syndrome, rheumatoid arthritis, ankylosing spondylitis, and Crohn's disease, but none with chronic hepatitis B infection. The reactive epitopes on p542 were mapped with deletion mutants, which indicated that the glycine-rich 28-mer was the major antigenic determinant, with lesser antibody responses to other epitopes. We conclude that normally there is an inability to generate IgG autoantibodies to the cross-reactive (mimicking) epitope of the p542 host protein, but that this inability is overcome in a proportion of patients with autoimmune disease. We conclude also that non-cross-reactive autoepitopes exist on p542 protein, to which IgG autoantibodies can commonly be formed in autoimmune disorders. The mechanisms responsible for the latter must involve different mechanisms than those responsible for autoantibodies to the mimicking epitope.

Epstein-Barr virus-induced autoimmune responses. I. Immunoglobulin M autoantibodies to proteins mimicking and not mimicking Epstein-Barr virus nuclear antigen-1.

In previous studies of infectious mononucleosis, we found IgM autoantibodies which react with hematopoietic cell antigens. Many of these were inhibited by synthetic glycine/alanine peptides representing the glycine/alanine repeat of Epstein-Barr virus nuclear antigen-1. We have cloned and expressed fragments of genes encoding two of these autoantigens. One gene (p542) encodes a protein containing a glycine-rich 28-mer, which is its chief autoantigenic epitope and which represents a newly identified class of evolutionarily conserved autoepitopes. The other gene (p554) encodes a protein that is not demonstrably cross-reactive with Epstein-Barr virus nuclear antigen-1 or with any other EBV protein, but forms complexes with other proteins. Immunoaffinity-purified anti-p542 and anti-p554 have relatively high binding affinities, as evidenced by inhibition at 10(6)-10(8) M-1, and neither autoantibody showed polyreactivity with other common antigens. The data thus suggest that neither autoantibody is simply an expression of polyclonal B cell activation. We conclude that the two autoantigens stimulate autoantibody synthesis by different mechanisms. One autoantigen shares homology to a viral protein which generates cross-reacting antibodies to the autoantigenic epitope. The other has no recognizable cross-reaction with the infecting pathogen and may become immunogenic through complexing with other proteins.

Risk communication in the dental practice.

The communication of risk in dental settings is a routine task that most clinicians are familiar with in their clinical encounters. However, work from medical settings has suggested that using this process in order to support health behaviour change in people may well be undermined by difficulties in understanding risk information, in presenting the information in a way that is clearly understood by the recipient and in the effects that such information may have for supporting further health behaviours by patients. This paper synthesises literature in the area that addresses these issues and explores approaches dental care professionals might consider when communicating risks in the dental surgery.

Unscheduled apoptosis during acute inflammatory lung injury.

Apoptosis is a mode of cell death currently thought to occur in the absence of inflammation. In contrast, inflammation follows unscheduled events such as acute tissue injury which results in necrosis, not apoptosis. We examined the relevance of this paradigm in three distinct models of acute lung injury; hyperoxia, oleic acid, and bacterial pneumonia. In every case, it was found that apoptosis is actually a prominent component of the acute and inflammatory phase of injury. Moreover, using strains of mice that are differentially sensitive to hyperoxic lung injury we observed that the percent of apoptotic cells was well correlated with the severity of lung injury. These observations suggest that apoptosis may be one of the biological consequences during acute injury and the failure to remove these apoptotic cells may also contribute to the inflammatory response.

The immune response to Epstein-Barr nuclear antigen: conformational and structural features of antibody binding to synthetic peptides.

Naturally developing human antibodies to the Epstein-Barr nuclear antigen recognize synthetic peptides containing sequences from the unusual glycine-alanine region of this protein. We tested antibody binding to a series of peptides of from five to 20 amino acids in length. Peptides as small as seven amino acids could bind but optimal results required chain lengths of 15. Binding was extremely sensitive to small changes in the length and sequence of the peptide, and also to the temp of the reaction. The changes can be ascribed to two factors: (1) deletion of the site of antigen binding and (2) loss of peptide secondary structure.

Prostatic involution in rats induced by a novel 5 alpha-reductase inhibitor, SK&F 105657: role for testosterone in the androgenic response.

Within the prostate, androgen stimulates glandular cell secretion and proliferation while inhibiting glandular cell death. Due to its predominant nuclear localization, higher affinity for the androgen receptor, and more than 10-fold higher intracellular concentration than testosterone, dihydrotestosterone (DHT), not testosterone, appears to be the active intracellular androgen within the prostate of intact male hosts. The issue has remained unanswered, however, whether testosterone itself, without irreversible conversion to DHT by the 5 alpha-reductase enzyme, is capable of androgenic effects in the prostate. To address this issue, a novel dead end (i.e. product) inhibitor of the 5 alpha-reductase enzyme, SK&F 105657, was administered to intact or castrated male rats treated with either exogeneous testosterone or DHT. When administered twice a day orally at 25 mg/kg.dose, SK&F 105657 reduced the prostatic DHT content of either intact or castrated rats maintained with exogeneous testosterone to the same low level as that produced by surgical castration. Unlike castration, however, such SK&F 105657 treatment increased the prostatic testosterone content by more than 5-fold. The decrease in prostatic DHT coupled with a raise in testosterone are specifically due to the in vivo inhibition of the 5 alpha-reductase activity, since they were not observed in castrated rats maintained with exogeneous DHT. Treatment of intact or castrated male rats with exogeneous testosterone and oral SK&F 105657 (25 mg/kg, twice daily) resulted in a substantial inhibition of prostatic secretion, an inhibition of prostatic glandular cell proliferation, and an increase in prostatic glandular cell death. The magnitude of the changes, however, was not as great as that observed after surgical castration. The results are, however, specific for 5 alpha-reductase inhibition, since they were not observed in castrated rats given exogeneous DHT. These results demonstrate that if the prostatic testosterone content is elevated to sufficient levels, androgenic effects are induced without a requirement for an elevation in prostatic DHT content. Thus, the conversion of testosterone to DHT appears to function as a means of amplifying androgenic stimulation in the prostate.

Clinical trial simulation of docetaxel in patients with cancer as a tool for dosage optimization.

BACKGROUND: Pharmacokinetic and pharmacodynamic analyses conducted during the development of docetaxel showed that patients with non-small-cell lung cancer with high baseline alpha1-acid glycoprotein levels had shorter time to progression and time to death. To assess whether such patients might benefit from dose intensification, we initiated a series of clinical trial simulations. METHODS: Pharmacokinetic and pharmacodynamic models for time to progression, death, and drop-out were developed and validated with the use of phase II data from 151 patients with non-small-cell lung cancer. The simulation process, in which these models were combined with a previously reported model for safety, was evaluated by comparison of the original phase II data with the predicted results. Simulations were undertaken for the evaluation of whether a trial (phase III) of 125 mg/m2 of docetaxel versus 100 mg/m2 of docetaxel in patients with high alpha1-acid glycoprotein levels would show improved survival. RESULTS: The hazard models showed that lower alpha1-acid glycoprotein levels, fewer number of organs involved, and higher docetaxel cumulative area under plasma concentration-time curve were significantly associated with enhanced time to progression and time to death. The simulation process produced data patterns similar to actual patterns. In the simulated phase III trial, although median survival was slightly longer in the 125 mg/m2 docetaxel group than in the 100 mg/m2 group (5.49 months versus 5.31 months, respectively), the difference was significant in only 6 of 100 trials. CONCLUSIONS: The low power to detect a difference due to dose intensification was the basis for the decision not to perform such a trial. The simulation exercise yielded valuable insight into how pharmacokinetic- and pharmacodynamic-based simulation of clinical trials may have an impact on decision making in drug development.

Population pharmacokinetics of riluzole in patients with amyotrophic lateral sclerosis.

OBJECTIVES: To characterize the population pharmacokinetic of riluzole in patients with amyotrophic lateral sclerosis (ALS). METHODS: One hundred patients with ALS who were participating in a multicenter phase III dose-ranging trial of riluzole were sampled on 179 visits. The sampling strategy (two samples per visit) was varied across patients to define the population kinetic profile (full screen). Riluzole plasma levels were determined by HPLC, and the data were analyzed by nonlinear mixed-effect modeling (NONMEM program) with use of a one-compartment structural model. The model incorporated interoccasion (visit-to visit) variability. RESULTS: In the basic one-compartment pharmacokinetic model, interindividual variability in plasma clearance (51.4%) was higher than intraindividual (visit-to-visit) variability (28.0%), indicating uniform pharmacokinetic behavior during long-term therapy. Riluzole clearance was independent of dosage (25 to 100 mg twice daily), treatment duration (up to 10 months), age, and renal function; gender and smoking were the most important patient covariates, with hepatic function having lesser influence. Typical value of clearance was 51.4 L/hr for a nonsmoking male patient. It was 32% lower in women than in men and 36% lower in nonsmokers than in smokers. Gender- and smoking-related variations in riluzole exposure at the recommended dosage (50 mg twice daily) were within the range of exposures achieved (with no untoward effect) in this dose-ranging study. CONCLUSION: The pharmacokinetics of riluzole has been characterized in patients during long-term therapy. Riluzole clearance is independent of dose and treatment duration. Within-patient variability is low. Gender and smoking status are the main covariates to explain interpatient variability.

Design, synthesis, and pharmacological characterization of (+)-2-aminobicyclo[3.1.0]hexane-2,6-dicarboxylic acid (LY354740): a potent, selective, and orally active group 2 metabotropic glutamate receptor agonist possessing anticonvulsant and anxiolytic properties.

2-Aminobicyclo[3.1.0]hexane-2,6-dicarboxylic acid (9) was designed as a conformationally constrained analog of glutamic acid. For 9, the key torsion angles (tau 1 and tau 2) which determine the relative positions of the alpha-amino acid and distal carboxyl functionalities are constrained where tau 1 = 166.9 degrees or 202 degrees and tau 2 = 156 degrees, respectively. We hypothesized that 9 would closely approximate the proposed bioactive conformation of glutamate when acting at group 2 metabotropic glutamate receptors (mGluRs). The racemic target molecule (+/-)-9, its C2-diastereomer (+/-)-16, and its enantiomers (+)-9 (LY354740) and (-)-9 (LY366563) were prepared by an efficient, stereocontrolled, and high-yielding synthesis from 2-cyclopentenone. Our hypothesis that 9 could interact with high affinity and specificity at group 2 mGluRs has been supported by the observation that (+/-)-9 (EC50 = 0.086 +/- 0.025 microM) and its enantiomer (+)-9 (EC50 = 0.055 +/- 0.017 microM) are highly potent agonists for group 2 mGluRs in the rat cerebral cortical slice preparation (suppression of forskolin-stimulated cAMP formation) possessing no activity at other glutamate receptor sites (iGluR or group 1 mGluR) at concentrations up to 100 microM. Importantly, the mGluR agonist effects of (+)-9 are evident following oral administration in mice in both the elevated plus maze model of anxiety (ED50 = 0.5 mg/kg) and in the ACPD-induced limbic seizure model (ED50 = 45.6 mg/kg). Thus, (+)-9 is the first orally active group 2 mGluR agonist described thus far and is an important tool for studying the effects of compounds of this class in humans.

An Epstein Barr virus-related cross reactive autoimmune response in multiple sclerosis in Norway.

In studies of patients in Norway with multiple sclerosis (MS), we have found cross reactive autoantibodies related to the Epstein Barr virus nuclear antigen-1 (EBNA-1). The MS patients had elevated IgG antibody to EBNA-1, as measured by reactivity with a synthetic glycine/alanine peptide, P62, which represents the glycine/alanine repeat in EBNA-1. The mean titer of anti-P62 in patients with acute relapse at the time of assay was significantly higher than in the remaining patients. Patients with remitting/relapsing MS also had elevated autoantibody to a lymphocyte protein, p542, cross reactive with EBNA-1 through a glycine/serine epitope. High titered anti-EBNA-1 antibodies from some MS, as well as from some SLE sera, were shown to cross react with 80-82 kDa and 60 kDa proteins in neuroglial cells.

What's lost in inverted faces?

Disproportionate inversion decrements for recognizing faces and other homogeneous stimuli are often interpreted as evidence that experts use relational features to recognize stimuli that share a configuration. However, it has never directly been shown that inversion disrupts the coding of relational features more than isolated features. Here we report three studies that compare inversion decrements for detecting changes that span the isolated-relational features continuum. Relatively large inversion decrements occurred for relational features (Thatcher illusion changes, internal feature spacing), with smaller decrements for isolated features (presence/absence of facial hair or glasses). The one discrepancy was a relatively large inversion decrement for detecting changes to the eyes and mouth, which we had classified as an isolated feature change. However, this decrement disappeared when the features were presented out of the face context (Experiments 2 and 3), suggesting that it occurs because subjects spontaneously code relations between the features and the rest of the face. Although the results support the interpretation of disproportionate inversion effects as evidence of relational coding, the difficulty of classifying changes as isolated or relational highlights an undesirable ambiguity in the isolated-relational feature distinction. We therefore consider alternative construals of the configural coding notion.