Machine learning in diagnostic support in medical emergency departments.

Diagnosing patients in the medical emergency department is complex and this is expected to increase in many countries due to an ageing population. In this study we investigate the feasibility of training machine learning algorithms to assist physicians handling the complex situation in the medical emergency departments. This is expected to reduce diagnostic errors and improve patient logistics and outcome. We included a total of 9,190 consecutive patient admissions diagnosed and treated in two hospitals in this cohort study. Patients had a biochemical workup including blood and urine analyses on clinical decision totaling 260 analyses. After adding nurse-registered data we trained 19 machine learning algorithms on a random 80% sample of the patients and validated the results on the remaining 20%. We trained algorithms for 19 different patient outcomes including the main outcomes death in 7 (Area under the Curve (AUC) 91.4%) and 30 days (AUC 91.3%) and safe-discharge(AUC 87.3%). The various algorithms obtained areas under the Receiver Operating Characteristics -curves in the range of 71.8-96.3% in the holdout cohort (68.3-98.2% in the training cohort). Performing this list of biochemical analyses at admission also reduced the number of subsequent venipunctures within 24 h from patient admittance by 22%. We have shown that it is possible to develop a list of machine-learning algorithms with high AUC for use in medical emergency departments. Moreover, the study showed that it is possible to reduce the number of venipunctures in this cohort.

POCT urine dipstick versus central laboratory analyses: Diagnostic performance and logistics in the medical emergency department.

OBJECTIVES: The aim of this study was to evaluate the logistics and diagnostic performances of dipstick analyses compared to their counterpart central laboratory analyses for detection of bacteriuria, proteinuria, hyperglycemia, ketosis and hematuria. DESIGN AND METHODS: Urine dipstick results, urine culture results, flow cytometric cell counts, U-albumin-to-creatinine ratio, P-glucose and P-beta-hydroxybutyrate were retrospectively reviewed in a cohort of consecutive patients admitted to the medical emergency departments of two Danish hospitals. Sensitivity, specificity and predictive values of traditional dipstick analysis were estimated and dipstick was compared to flow cytometry for detection of significant bacteriuria using logistic regression. Turn-around-time for central laboratory analyses were assessed. RESULTS: For each comparison, 1,997 patients or more were included. Traditional dipstick analyses for proteinuria, bacteriuria and ketosis reached sensitivities of up to 90%, while sensitivity for hyperglycemia was 59%. Flow cytometry outperformed traditional dipstick analysis for detection of bacteriuria with a difference in the area under the ROC-curve of 0.07. Turn-around-times for 95% delivery of central laboratory analysis results ranged from approximately 1½ to 2 h. CONCLUSIONS: For the detection of bacteriuria and albuminuria, central laboratory analyses reach better performance than dipstick analysis while achieving acceptable turn-around-times and are thus viable alternatives to dipstick analysis. For detection of ketosis and hyperglycemia, dipstick analysis does not perform adequately, but as very short turn-around-time is often required, these conditions may be best diagnosed by point-of-care blood test rather than dipstick or central laboratory analyses. Dipstick hemoglobin analysis, flow cytometry and microscopic evaluation may serve each their distinct purposes, and thus are relevant in the emergency department.

Increased serum levels of microfibrillar-associated protein 4 (MFAP4) are not associated with clinical synovitis in rheumatoid arthritis but may reflect underlying cardiovascular comorbidity.

OBJECTIVES: To study circulating MFAP4 in rheumatoid arthritis (RA) and its associations with clinical phenotype. METHODS: Early RA (ERA): 47 patients with newly diagnosed, treatment naïve RA were included. Serum MFAP4, clinical and laboratory disease variables were recorded serially during 12 months of intensive synovitis suppressive treatment. Long-standing RA (LRA): 317 patients participated, all receiving DMARD treatment. Disease activity, autoantibody status, extra-articular manifestations and cardiovascular morbidity were recorded. Paired serum and synovial fluid samples were obtained from 13 untreated ERA patients. Healthy blood donors served as reference points. MFAP4 was quantified by AlphaLISA immunoassay. Univariate, multivariate and mixed effects regression models were applied in the statistical analysis. RESULTS: ERA: MFAP4 increased from baseline and was significantly elevated at the 12-month follow-up, 17.8 U/l [12.6;24.1] vs. healthy controls, 12.7 U/l [9.5;15.6], p<0.001. MFAP4 did not correlate with joint counts or C-reactive protein. LRA: MFAP4 was increased, 25.9 U/l [20.4;33.7] vs. healthy controls, 17.6 U/l [13.7;21.2], p<0.0001, but did not correlate with disease activity measures or presence of extra-articular manifestations. Notably, MFAP4 correlated inversely with smoking (p<0.0001) and presence of antibodies against cyclic citrullinated peptides (anti-CCP), p=0.005. There was a positive association with systolic blood pressure, p=0.001 and co-occurrence of three cardiovascular events and/or risk factors, p<0.0001. The serum:synovial fluid MFAP4 ratio was 2:1. CONCLUSIONS: MFAP4 increases from diagnostic baseline despite intensive treatment but does not associate with synovitis at early or late stages of RA. Correlation patterns indicate that increased MFAP4 may reflect enhanced RA-related vascular remodelling.

Site-specific absence of microfibrillar-associated protein 4 (MFAP4) from the internal elastic membrane of arterioles in the rheumatoid arthritis synovial membrane: an immunohistochemical study in patients with advanced rheumatoid arthritis versus osteoarthritis.

Microfibrillar-associated protein 4 (MFAP4) is a non-structural matrix protein with cell regulatory activities and a potential as seromarker for fibrosis. We aimed to study the occurrence of MFAP4 in the synovial membrane from patients with rheumatoid arthritis (RA) vs osteoarthritis (OA). Formaldehyde-fixed synovial tissue sections, from patients with RA (N = 6) and OA (N = 6) undergoing total hip arthroplasty, were deparaffinized and immunostained with monoclonal antibodies against MFAP4. Elastin was detected using ElastiKit. MFAP4 in serum (sMFAP4) and synovial fluid was measured by an immunoassay. MFAP4 was present in synovial biopsies from both RA and OA patients, particularly prominent in deep arterioles where it colocalized with elastin. Notably however, MFAP4 was absent from the internal elastic lamina in RA arterioles irrespective of disease duration and synovitis activity, while present although with irregular staining patterns in OA specimens. sMFAP4 was increased in RA and OA serum vs healthy controls: median (interquartile range) 29.8 (25.3-39.1) and 25.5 U/L (18.1-43.3) vs 17.7 U/L (13.7-21.2), p = 0.006 and p = 0.02, respectively The concentration of synovial fluid was lower than in serum in both RA and OA. These findings may suggest that MFAP4 is involved in adaptive vessel wall remodeling associated with chronic joint disease.

Localization of surfactant protein-D in the rheumatoid synovial membrane.

Surfactant protein-D (SP-D) is a collectin, which plays an important role in airway protection and inflammation. The molecule has both pro- and anti-inflammatory capacities depending on its molecular size. Its involvement in joint diseases is largely unknown and the aim of this investigation was to study SP-D occurrence and distribution in the synovial membrane of patients with long-standing rheumatoid arthritis (RA) and osteoarthritis (OA). Six RA patients and six OA patients, who underwent total hip arthroplasty, were included in the study. Synovial tissue biopsies were obtained during surgery and subsequently prepared for immunohistochemistry. In this first, small-scale comparative study on the occurrence of SP-D in the synovial membrane of RA and OA, we report that SP-D was only present in the microvascular endothelium in subsynovial and pannus tissue and that the immunostaining was much stronger than in OA. This distribution pattern suggests that SP-D modulates RA inflammatory activities.

Type I and III collagen turnover is increased in axial spondyloarthritis and psoriatic arthritis. Associations with disease activity and diagnostic capacity.

OBJECTIVES: To investigate the turnover of type I and III collagen by neo-epitope markers in patients with axial spondyloarthritis (axSpA) and psoriatic arthritis (PsA). METHODS: Patients with PsA (n=101) or axSpA (n=110) and healthy subjects (n=120) were included. Demographic and clinical data were recorded. Markers of type I and III collagen were quantified by RIA (ICTP) or ELISA (C1M and C3M). Non-parametric statistics were applied for intergroup comparisons and correlation studies. The diagnostic potential of these marker molecules was assessed by ROC analysis. RESULTS: C1M and C3M, which originate from soft connective tissues, were significantly higher in axSpA and PsA as compared with healthy control subjects. CIM and C3M correlated with ASDAS and DAS28. Overall, ICTP, which arises from bone degradation, did not differ between disease versus healthy. However, ICTP was lower in HLA-B27 positive than in HLA-B27 negative patients with axSpA. There was no association between bone and soft connective tissue collagen I markers (ICTP and C1M), while C1M and C3M were highly correlated (p<0.0001). C1M discriminated between healthy and diseased with AUCs of 0.83 for PsA and 0.79 for axSpA. C3M AUCs were 0.77 for PsA and 0.78 for axSpA. CONCLUSIONS: Type I and III collagen remodelling in soft connective tissue is increased in axSpA and PsA and associates with disease activity. Bone collagen degradation is lower in HLA-B27 positive compared with HLA-B27 negative axSpA, which may represent an aspect of enhanced enthesopathic bone proliferation in HLA-B27 carriers. C1M and C3M distinguish well between healthy and diseased individuals.

Chondrocyte activity is increased in psoriatic arthritis and axial spondyloarthritis.

BACKGROUND: Psoriatic arthritis (PsA) and axial spondyloarthritis (axSpA) are chronic inflammatory rheumatic diseases with complex origins. Both are characterized by altered extracellular matrix remodeling in joints and entheses that results in destructive and osteochondral proliferative lesions. There is a need for biomarkers reflecting core disease pathways for diagnosis and disease mapping. Pro-C2 reflects mature cartilage collagen type IIB formation, while C-Col10 represents turnover of type X collagen, which is exclusively expressed by hypertrophic chondrocytes. The objectives of this study were to study cartilage metabolism in axSpA and PsA by assessing Pro-C2 and C-Col10 and to evaluate their diagnostic utility against a healthy reference population. METHODS: Patients with PsA (n = 101) or axSpA (n = 110) were recruited consecutively from three rheumatology outpatient clinics. Demographic and clinical disease measures were recorded. Pro-C2 and C-Col10 were quantified in serum by using newly developed and specific competitive enzyme-linked immunosorbent assays based on monoclonal antibodies. One-way analysis of variance and Tukey's multiple comparison tests were performed on log-transformed data. ROC curve analysis was carried out to evaluate their discriminative power. RESULTS: Pro-C2 levels in serum were significantly increased in both axSpA (median concentration 1.11 ng/ml, 0.67-1.64) and PsA (median concentration 1.03 ng/ml, 0.53-1.47) compared with healthy controls (median concentration 0.30 ng/ml, 0.16-0.41) (p < 0.0001). Pro-C2 did not differ according to treatment. C-Col10 was slightly but equally elevated in the PsA and axSpA groups vs. the control group, but it was significantly lower in patients with axSpA undergoing tumor necrosis factor-α inhibitor (TNFi) treatment. ROC curve analysis revealed AUCs of 0.85 (95 % CI 0.79-0.89) for axSpA and 0.81 (95 % CI 0.75-0.86) for PsA. CONCLUSIONS: These findings indicate that cartilage collagen metabolism was enhanced in the axSpA and PsA groups compared with the healthy control group. The lower C-Col10 level in patients with axSpA undergoing TNFi treatment may reflect that hypertrophic chondrocytes in axSpA are targeted by TNFi. ROC curve analysis showed a diagnostic potential for Pro-C2 in axSpA and PsA.

Cartilage collagen type II seromarker patterns in axial spondyloarthritis and psoriatic arthritis: associations with disease activity, smoking and HLA-B27.

The aim of the study was to assess the possible association between type II collagen turnover seromarkers and disease profile in patients with axial spondyloarthritis (SpA) and psoriatic arthritis (PsA). Outpatients with axial SpA (n = 110) or PsA (n = 101) underwent clinical examination including disease activity measures and HLA-B27 typing. The procollagen IIA N-terminal peptide (PIIANP) and a matrix metalloproteinase-generated type II collagen fragment (C2M) were quantified in serum by ELISA. C2M was higher in SpA than in controls, 0.41 versus 0.36 ng/ml (p = 0.004), while PIIANP did not differ between patients and healthy subjects, 2252 versus 2142 ng/ml (p = 0.13). However, DMARD-naïve SpA patients had higher PIIANP, 2461 ng/ml (p = 0.01) and C2M, 0.44 ng/ml (p = 0.0007) levels than controls, and PIIANP correlated with CRP (ρ = 0.34). C2M was lower in SpA smokers, 0.36 ng/ml versus non-smokers, 0.43 ng/ml (p = 0.02), while PIIANP was higher in HLA-B27 positive, 2312 ng/ml versus negative patients, 2021 ng/ml (p = 0.03). In PsA, PIIANP and C2M did not differ between patients and controls, but PIIANP was elevated in patients not receiving DMARDs, 2726 ng/ml. In PsA, PIIANP and C2M did not differ according to smoking and HLA-B27. Cartilage degradation assessed by C2M is increased in SpA irrespective of treatment but not in PsA. Cartilage synthesis reflected by PIIANP is increased in untreated SpA and PsA. PIIANP correlates with CRP in SpA while not in PsA. In DMARD-naïve SpA but not in PsA, HLA-B27 positivity and smoking are associated with a chondro-proliferative metabolic pattern.

Tumour necrosis factor-α inhibitors are glucocorticoid-sparing in rheumatoid arthritis.

INTRODUCTION: Rheumatoid arthritis (RA) is a chronic disease with autoimmune traits of unknown aetiology which primarily affects synovial joints. Glucocorticoids (GCs) are still widely used in RA treatment despite the expanding use of targeted and very efficient agents. The objective of this study was to assess the impact of treatment with tumour necrosis factor-α inhibitors (TNFi) on GC utilization in real-life practice among Danish RA patients. MATERIAL AND METHODS: The DANBIO registry is a nationwide rheumatologic database which collects demographic and clinical data. This retrospective study included RA outpatients from a tertiary rheumatologic department recruited from the DANBIO registry who initiated their first TNFi treatment between January 2000 and February 2010 (n = 171). GC dosing during the year before and the year after TNFi initiation were compared. Patients acted as their own control. RESULTS: The median daily prednisolone dose was significantly decreased after initiation of TNFi treatment (p < 0.01). At TNFi initiation, 78 patients (46%) received prednisolone compared with 53 (31%) by the end of follow-up. After TNFi initiation, 30 patients (38%) discontinued prednisolone and in 34 (44%) prednisolone dose was reduced. Similarly, the number of GC injections decreased significantly at 13, 26 and 52 weeks following TNFi initiation (p < 0.01). CONCLUSION: GC utilization is significantly reduced after initiation of TNFi treatment. Among patients on prednisolone at TNFi onset, prednisolone was withdrawn in one third and the dose was reduced in nearly half. Furthermore, the need for GC injections decreased.

Uncoupling of collagen II metabolism in newly diagnosed, untreated rheumatoid arthritis is linked to inflammation and antibodies against cyclic citrullinated peptides.

OBJECTIVE: To investigate the relationship between markers of collagen II synthesis and degradation with disease activity measures, autoantibodies, and radiographic outcomes in a 4-year protocol on patients with early rheumatoid arthritis (RA) who are naïve to disease-modifying antirheumatic drugs. METHODS: One hundred sixty patients with newly diagnosed, untreated RA entered the Cyclosporine, Methotrexate, Steroid in RA (CIMESTRA) trial. Disease activity and radiograph status were measured at baseline and 4 years. The N-terminal propeptide of collagen IIA (PIIANP) and the cross-linked C-telopeptide of collagen II (CTX-II) were quantified at baseline by ELISA. PIIANP was also assayed at 2 and 4 years. Anticyclic citrullinated peptide (anti-CCP) was recorded at baseline. An uncoupling index for cartilage collagen metabolism was calculated from PIIANP and CTX-II measurements. RESULTS: PIIANP was low at diagnosis and 4 years on (p < 0.001), irrespective of treatment and disease activity. PIIANP was lowest in anti-CCP positive patients (p = 0.006), and there was a negative correlation between PIIANP and anti-CCP titers (rho = -0.25, p 0.002). CTX-II was increased (p < 0.001) and correlated positively with disease activity and radiographic progression, but not with anti-CCP (p = 0.93). The uncoupling index was not superior to CTX-II in predicting radiographic changes. CONCLUSION: These results suggest that cartilage collagen formation and degradation are unbalanced when RA is diagnosed. The different associations of collagen II anabolism (PIIANP) and collagen II degradation (CTX-II) with anti-CCP, synovitis, and radiographic progression indicate that at this early stage of RA, cartilage collagen degradation is mainly driven by synovitis, while anti-CCP antibodies may interfere with cartilage regeneration by inhibiting collagen IIA formation. Trial registration j.nr NCT00209859.

Circulating surfactant protein -D is low and correlates negatively with systemic inflammation in early, untreated rheumatoid arthritis.

INTRODUCTION: Surfactant protein D (SP-D) is a collectin with immuno-regulatory functions, which may depend on oligomerization. Anti-microbial and anti-inflammatory properties have been attributed to multimeric SP-D variants, while trimeric subunits per se have been suggested to enhance inflammation. Previously, we reported low circulating SP-D in early rheumatoid arthritis (RA), and the present investigation aims to extend these data by serial SP-D serum measurements, studies on synovial fluid, SP-D size distribution and genotyping in patients with early RA. METHODS: One-hundred-and-sixty disease-modifying antirheumatic drug (DMARD) naïve RA patients with disease duration less than six months were studied prospectively for four years (CIMESTRA (Ciclosporine, Methotrexate, Steroid in RA) trial) including disease activity measures (C-reactive protein, joint counts and Health Assessment Questionnaire (HAQ) score), autoantibodies, x-ray findings and SP-D. SP-D was quantified by enzyme-linked immunosorbent assay (ELISA) and molecular size distribution was assessed by gel filtration chromatography. Further, SP-D Met11Thr single nucleotide polymorphism (SNP) analysis was performed. RESULTS: Serum SP-D was significantly lower in RA patients at baseline compared with healthy controls (P < 0.001). SP-D increased slightly during follow-up (P < 0.001), but was still subnormal at four years after adjustment for confounders (P < 0.001). SP-D in synovial fluid was up to 2.5-fold lower than in serum. While multimeric variants were detected in serum, SP-D in synovial fluid comprised trimeric subunits only. There were no significant associations between genotype distribution and SP-D. Baseline SP-D was inversely associated to CRP and HAQ score. A similar relationship was observed regarding temporal changes in SP-D and CRP (zero to four years). SP-D was not associated to x-ray findings. CONCLUSIONS: This study confirms that circulating SP-D is persistently subnormal in early and untreated RA despite a favourable therapeutic response obtained during four years of follow-up. SP-D correlated negatively to disease activity measures, but was not correlated with x-ray progression or SP-D genotype. These observations suggest that SP-D is implicated in RA pathogenesis at the protein level. The exclusive presence of trimeric SP-D in affected joints may contribute to the maintenance of joint inflammation. TRIAL REGISTRATION: (j.nr NCT00209859).

Circadian pattern and the effect of standardized physical exercise on procollagen IIA N-peptide (PIIANP) in rheumatoid arthritis at different stages and in healthy individuals.

BACKGROUND: Variant collagen IIA is re-expressed in diseased cartilage. Low procollagen IIA N-peptide (PIIANP) levels in serum have recently been reported in rheumatoid arthritis (RA). We investigated circadian rhythmicity and effect of physical activity on PIIANP in early and longstanding RA and in healthy subjects. METHODS: Patients with early and longstanding RA and controls were included. Fasting and serial blood samples were collected during 24 h. PIIANP response to physical activity was studied before and serially after standardized exercise. RESULTS AND CONCLUSION: In RA at different stages and healthy individuals, PIIANP exhibited no circadian rhythmicity, and PIIANP in serum was not influenced by physical activity.

[Smoking--a risk factor for rheumatoid arthritis development]. / Rygning--en risikofaktor for udvikling af reumatoid artritis.

Rheumatoid arthritis (RA) is a chronic disabling inflammatory joint disease of unknown aetiology. Genetic and environmental factors are implicated in its pathogenesis. We performed a literature search on MEDLINE, the Cochrane database and EMBASE on the interaction between smoking and RA. It is concluded that previous and current smoking constitute a risk factor for RA development. Smoking acts synergistically with other RA risk factors, including IgM-rheumatoid factor, anti-CCP and shared epitopes. The effect of smoking on the course of RA has not been described in sufficient detail.