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.

A pilot phase Ib/II study of whole-lung low dose radiation therapy (LDRT) for the treatment of severe COVID-19 pneumonia: First experience from Africa.

BACKGROUND: Low dose radiation therapy (LDRT) has been used for non-malignant conditions since early 1900s based on the ability of single fractions between 50-150 cGy to inhibit cellular proliferation. Given scarcity of resources, poor access to vaccines and medical therapies within low and middle income countries, there is an urgent need to identify other cost-effective alternatives in management of COVID-19 pneumonia. We conducted a pilot phase Ib/II investigator-initiated clinical trial to assess the safety, feasibility, and toxicity of LDRT in patients with severe COVID-19 pneumonia at the Aga Khan University Hospital in Nairobi, Kenya. Additionally, we also assessed clinical benefit in terms of improvement in oxygenation at day 3 following LDRT and the ability to avoid mechanical ventilation at day 7 post LDRT. METHODS: Patients with both polymerase chain reaction (PCR) and high-resolution computer tomogram (HRCT) confirmed severe COVID-19 pneumonia, not improving on conventional therapy including Dexamethasone and with increasing oxygen requirement were enrolled in the study. Patients on mechanical ventilation were excluded. Eligible patients received a single 100cGy fraction to the whole lung. In the absence of any dose limiting toxicity the study proposed to treat a total of 10 patients. The primary endpoints were to assess the safety/feasibility, and toxicity within the first 24 hours post LDRT. The secondary endpoints were to assess efficacy of LDRT at Day 3, 7, 14 and 28 post LDRT. RESULTS: Ten patients were treated with LDRT. All (100%) of patients were able to complete LDRT without treatment related SAE within the first 24 hours post treatment. None of the patients treated with LDRT experienced any acute toxicity as defined by change in clinical and respiratory status at 24hr following LDRT. Majority (90%) of patients avoided mechanical ventilation within 7 days of LDRT. Four patients (40%) demonstrated at least 25% improvement in oxygen requirements within 3 days. Six patients (60%) were discharged and remained off oxygen, whereas four progressed and died (1 due to sepsis and 3 in cytokine storm). Median time to discharge (n = 6) was 16.5 days and median time to death (n = 4) was 11.0 days. Patients who ultimately died showed elevated inflammatory markers including Ferritin, CRP and D-dimers as compared to those who were discharged alive. CONCLUSION: LDRT was feasible, safe and shows promise in the management of severe COVID-19 pneumonia including in patients progressing on conventional systemic treatment. Additional phase II trials are warranted to identify patients most likely to benefit from LDRT.

Supporting Kenyan women with advanced breast cancer through a network and assessing their needs and quality of life.

INTRODUCTION: Metastatic breast cancer (MBC) patients have several unmet needs. The needs and quality of life of MBC women living in sub-Saharan Africa (SSA) are understudied. Facilitating the interaction of various caregivers is beneficial in addressing the needs. Internet-based resources play an important role in reaching out to these patients. We aimed to bring the various stakeholders into a joint network force, create a web-based portal, understand the needs of MBC patients, and assess the utilization of web-based resources for women from Kenya. METHODS: A network of various stakeholders considered crucial in the care of Kenyan women with MBC was created. We conducted educational camps and assessed their needs, quality of life (QoL), and knowledge. We assessed the impact of utilizing web-based resources by MBC patients from here. RESULTS: We formed a network involving partners and launched the first dedicated website for MBC from Kenya. The website has received 13,944 visits and 310,379 hits in 2 years. One hundred fourteen women living with MBC were interviewed, and our findings show that psychological needs (63%), physical support needs (60%), and health care system needs (55%) are leading areas of needs that increase with rural residence (p = 0.001), less education (p = 0.003), and aggressive treatments (p = 0.008). Quality of life (QoL) confirmed better scores with urban residence (p = 0.002), internet access (p = 0.010), and stable disease (p = 0.042). CONCLUSIONS: Creating a network of caregivers provides opportunities for cohesive efforts in understanding the psychosocial and medical needs of patients with MBC. Internet-based resources are an effective way of reaching out to them. Kenyan patients show extremely good uptake of internet-based resources.

Mammographic Breast Density and Breast Cancer Molecular Subtypes: The Kenyan-African Aspect.

INTRODUCTION: Data examining mammographic breast density (MBD) among patients in Sub-Saharan Africa are sparse. We evaluated how MBD relates to breast cancer characteristics in Kenyan women undergoing diagnostic mammography. METHODS: This cross-sectional study included women with pathologically confirmed breast cancers (n = 123). Pretreatment mammograms of the unaffected breast were assessed to estimate absolute dense area (cm2), nondense area (cm2), and percent density (PD). Relationships between density measurements and clinical characteristics were evaluated using analysis of covariance. RESULTS: Median PD and dense area were 24.9% and 85.3 cm2. Higher PD and dense area were observed in younger women (P < 0.01). Higher dense and nondense areas were observed in obese women (P-trend < 0.01). Estrogen receptor (ER) positive patients (73%) had higher PD and dense area than ER-negative patients (P ≤ 0.02). Triple negative breast cancer (TNBC) patients (17%) had lower PD and dense area (P ≤ 0.01) compared with non-TNBCs. No associations were observed between MBD and tumor size and grade. CONCLUSIONS: Our findings show discordant relationships between MBD and molecular tumor subtypes to those previously observed in Western populations. The relatively low breast density observed at diagnosis may have important implications for cancer prevention initiatives in Kenya. Subsequent larger studies are needed to confirm these findings.

Ethnicity and breast cancer characteristics in Kenya.

PURPOSE: There are no published data from specific regions of sub-Saharan Africa describing the clinical and pathological characteristics and molecular subtypes of invasive breast cancer by ethnic group. The purpose of this study was to investigate these characteristics among the three major ethno-cultural groupings in Kenya. METHODS: The study included women with pathologically confirmed breast cancer diagnosed between March 2012 and May 2015 at 11 hospitals throughout Kenya. Sociodemographic, clinical, and reproductive data were collected by questionnaire, and pathology review and immunohistochemistry were performed centrally. RESULTS: The 846 cases included 661 Bantus (78.1%), 143 Nilotes (16.9%), 19 Cushites (2.3%), and 23 patients of mixed ethnicity (2.7%). In analyses comparing the two major ethnic groups, Bantus were more educated, more overweight, had an older age at first birth, and had a younger age at menopause than Nilotes (p < 0.05 for all comparisons). In analyses restricted to definitive surgery specimens, there were no statistically significant differences in tumor characteristics or molecular subtypes by ethnicity, although the Nilote tumors tended to be larger (OR for ≥ 5 cm vs. < 2 cm: 3.86, 95% CI 0.77, 19.30) and were somewhat more likely to be HER2 enriched (OR for HER2 enriched vs. Luminal A/B: 1.41, 95% CI 0.79, 2.49). CONCLUSION: This case series showed no significant differences in breast cancer tumor characteristics or molecular subtypes, but significant differences in sociodemographic characteristics and reproductive factors, among the three major ethnic groups in Kenya. We suggest further evaluation of ethnic differences in breast cancer throughout the genetically and culturally diverse populations of sub-Saharan Africa.