Prediction of cardiovascular risk factors from retinal fundus photographs: Validation of a deep learning algorithm in a prospective non-interventional study in Kenya.

AIM: Hypertension and diabetes mellitus (DM) are major causes of morbidity and mortality, with growing burdens in low-income countries where they are underdiagnosed and undertreated. Advances in machine learning may provide opportunities to enhance diagnostics in settings with limited medical infrastructure. MATERIALS AND METHODS: A non-interventional study was conducted to develop and validate a machine learning algorithm to estimate cardiovascular clinical and laboratory parameters. At two sites in Kenya, digital retinal fundus photographs were collected alongside blood pressure (BP), laboratory measures and medical history. The performance of machine learning models, originally trained using data from the UK Biobank, were evaluated for their ability to estimate BP, glycated haemoglobin, estimated glomerular filtration rate and diagnoses from fundus images. RESULTS: In total, 301 participants were enrolled. Compared with the UK Biobank population used for algorithm development, participants from Kenya were younger and would probably report Black/African ethnicity, with a higher body mass index and prevalence of DM and hypertension. The mean absolute error was comparable or slightly greater for systolic BP, diastolic BP, glycated haemoglobin and estimated glomerular filtration rate. The model trained to identify DM had an area under the receiver operating curve of 0.762 (0.818 in the UK Biobank) and the hypertension model had an area under the receiver operating curve of 0.765 (0.738 in the UK Biobank). CONCLUSIONS: In a Kenyan population, machine learning models estimated cardiovascular parameters with comparable or slightly lower accuracy than in the population where they were trained, suggesting model recalibration may be appropriate. This study represents an incremental step toward leveraging machine learning to make early cardiovascular screening more accessible, particularly in resource-limited settings.

Renal Autotransplantation for Uncontrolled Hypertension in Nonatherosclerotic Renal Artery Stenosis-2 Case Reports and a Brief Review of the Literature.

Fibromuscular dysplasia is the most common cause of renovascular hypertension in young adults under 40 years old. It is potentially amenable to renal artery angioplasty, which frequently normalizes blood pressure. However, limited options exist if angioplasty is not technically possible, or restenosis occurs. Here, we describe 2 patients who presented with hypertension secondary to renal artery stenosis. In the first case, a young adult with hypertension secondary to renal artery stenosis (fibromuscular dysplasia), developed restenosis 11 weeks after an initially successful renal artery angioplasty. In the second case, a patient with neurofibromatosis type 1 was diagnosed with hypertension secondary to renal artery stenosis. Angioplasty was not possible due to multiple branch occlusions. Both individuals went on to have successful renal autotransplantations, which ultimately cured their hypertension. In this article, we review the background, indications, and blood pressure outcomes in relation to renal autotransplantation in nonatherosclerotic renal artery stenosis.

Clinical impact of tissue sodium storage.

In recent times, the traditional nephrocentric, two-compartment model of body sodium has been challenged by long-term sodium balance studies and experimental work on the dermal interstitium and endothelial surface layer. In the new paradigm, sodium can be stored without commensurate water retention in the interstitium and endothelial surface layer, forming a dynamic third compartment for sodium. This has important implications for sodium homeostasis, osmoregulation and the hemodynamic response to salt intake. Sodium storage in the skin and endothelial surface layer may function as a buffer during periods of dietary depletion and excess, representing an extra-renal mechanism regulating body sodium and water. Interstitial sodium storage may also serve as a biomarker for sodium sensitivity and cardiovascular risk, as well as a target for hypertension treatment. Furthermore, sodium storage may explain the limitations of traditional techniques used to quantify sodium intake and determine infusion strategies for dysnatraemias. This review is aimed at outlining these new insights into sodium homeostasis, exploring their implications for clinical practice and potential areas for further research for paediatric and adult populations.

Skin Sodium and Hypertension: a Paradigm Shift?

PURPOSE OF REVIEW: Dietary sodium is an important trigger for hypertension and humans show a heterogeneous blood pressure response to salt intake. The precise mechanisms for this have not been fully explained although renal sodium handling has traditionally been considered to play a central role. RECENT FINDINGS: Animal studies have shown that dietary salt loading results in non-osmotic sodium accumulation via glycosaminoglycans and lymphangiogenesis in skin mediated by vascular endothelial growth factor-C, both processes attenuating the rise in BP. Studies in humans have shown that skin could be a buffer for sodium and that skin sodium could be a marker of hypertension and salt sensitivity. Skin sodium storage could represent an additional system influencing the response to salt load and blood pressure in humans.

Novel Mechanism for Buffering Dietary Salt in Humans: Effects of Salt Loading on Skin Sodium, Vascular Endothelial Growth Factor C, and Blood Pressure.

High dietary sodium intake triggers increased blood pressure (BP). Animal studies show that dietary salt loading results in dermal Na+ accumulation and lymphangiogenesis mediated by VEGF-C (vascular endothelial growth factor C), both attenuating the rise in BP. Our objective was to determine whether these mechanisms function in humans. We assessed skin electrolytes, BP, and plasma VEGF-C in 48 healthy participants randomized to placebo (70 mmol sodium/d) and slow sodium (200 mmol/d) for 7 days. Skin Na+ and K+ concentrations were measured in mg/g of wet tissue and expressed as the ratio Na+:K+ to correct for variability in sample hydration. Skin Na+:K+ increased between placebo and slow sodium phases (2.91±0.08 versus 3.12±0.09; P=0.01). In post hoc analysis, there was a suggestion of a sex-specific effect, with a significant increase in skin Na+:K+ in men (2.59±0.09 versus 2.88±0.12; P=0.008) but not women (3.23±0.10 versus 3.36±0.12; P=0.31). Women showed a significant increase in 24-hour mean BP with salt loading (93±1 versus 91±1 mm Hg; P<0.001) while men did not (96±2 versus 96±2 mm Hg; P=0.91). Skin Na+:K+ correlated with BP, stroke volume, and peripheral vascular resistance in men but not in women. No change was noted in plasma VEGF-C. These findings suggest that the skin may buffer dietary Na+, reducing the hemodynamic consequences of increased salt, and this may be influenced by sex.

Pre-dialysis systolic blood pressure-variability is independently associated with all-cause mortality in incident haemodialysis patients.

Systolic blood pressure variability is an independent risk factor for mortality and cardiovascular events. Standard measures of blood pressure predict outcome poorly in haemodialysis patients. We investigated whether systolic blood pressure variability was associated with mortality in incident haemodialysis patients. We performed a longitudinal observational study of patients commencing haemodialysis between 2005 and 2011 in East Anglia, UK, excluding patients with cardiovascular events within 6 months of starting haemodialysis. The main exposure was variability independent of the mean (VIM) of systolic blood pressure from short-gap, pre-dialysis blood pressure readings between 3 and 6 months after commencing haemodialysis, and the outcome was all-cause mortality. Of 203 patients, 37 (18.2%) patients died during a mean follow-up of 2.0 (SD 1.3) years. The age and sex-adjusted hazard ratio (HR) for mortality was 1.09 (95% confidence interval (CI) 1.02-1.17) for a one-unit increase of VIM. This was not altered by adjustment for diabetes, prior cardiovascular disease and mean systolic blood pressure (HR 1.09, 95% CI 1.02-1.16). Patients with VIM of systolic blood pressure above the median were 2.4 (95% CI 1.17-4.74) times more likely to die during follow-up than those below the median. Results were similar for all measures of blood pressure variability and further adjustment for type of dialysis access, use of antihypertensives and absolute or variability of fluid intake did not alter these findings. Diastolic blood pressure variability showed no association with all cause mortality. Our study shows that variability of systolic blood pressure is a strong and independent predictor of all-cause mortality in incident haemodialysis patients. Further research is needed to understand the mechanism as this may form a therapeutic target or focus for management.

Comparison of estimated protein output and urine protein: creatinine ratio in first and second voids with 24-hour urine protein.

BACKGROUND: Current UK guidelines for the identification, management and referral of chronic kidney disease advise an early-morning urine sample for the albumin:creatinine ratio or the protein:creatinine ratio (PCR) in order to quantify proteinuria. Estimated protein output (EPO) is an alternative and possibly better method of quantifying proteinuria which takes lean weight into consideration. METHODS: We carried out a single-centre study of 36 adult patients with proteinuric nephropathy over a period of 18 months. Urinary PCR and EPO estimates of 24-hour urine protein were compared with 24-hour urine collections by Bland-Altman analysis. RESULTS: Average 24-hour urine protein was 1.6 g (range 0.2-5.1 g). Best agreement with 24-hour protein was for first-void EPO (limits of agreement 0.33-1.59) followed by a second-void EPO (0.40-1.76), then second-void PCR (0.40-2.08) and lastly first-void PCR (0.28-2.03). None of the differences between estimates of urine protein excretion and 24-hour urine protein were statistically significant. All estimates of protein output had wide confidence intervals confirming that spot urine samples, while simple and convenient to do, are imprecise measures of 24-hour urine protein excretion. CONCLUSION: When estimating 24-hour urine protein from a spot urine sample, EPO may be marginally more accurate than PCR, and first-void urine samples slightly better than second-void urine samples, but a first- or second-void PCR will suffice in most instances.

Post-transplant lymphoproliferative disorder: case report and review of susceptibility to EBV in the Scottish adult renal transplant pool.

We report a case of high-grade non-Hodgkin's lymphoma following Epstein-Barr virus (EBV) infection in a 38-year-old renal transplant recipient who was successfully treated with rituximab and remains alive 6 years later with reasonable graft function. We subsequently undertook a survey showing that 1.8% of the Scottish adult transplant pool are susceptible to EBV infection. Though a vaccine for EBV is currently not yet available, routine screening of potential renal transplant recipients for EBV should help identify those at increased risk of post-transplant lymhoproliferative disorder (PTLD), while tailoring of immunosuppression and antiviral prophylaxis with Ganciclovir may help reduce the emergence of this potentially life-threatening disease.