Application of machine learning methods to bridge the gap between non-interventional studies and randomized controlled trials in ophthalmic patients with neovascular age-related macular degeneration.

PURPOSE: The effectiveness of intravitreal anti-vascular endothelial growth factor agents is usually lower in real world settings compared with randomized clinical trials (RCTs), often limiting the use of real-world evidence (RWE) in regulatory and healthcare decisions. The current analysis aimed to develop and validate an algorithm to explain the difference in outcomes between RWE studies and RCTs in patients with neovascular age-related macular degeneration. METHODS: The algorithm was developed using ranibizumab real world data (RWD) from the US and validated on Australian and UK RWD. A decision model was developed using machine learning principles, in which the model learns how to partition the most influential factors (out of 59 variables) so that they maximally relate to the change in visual acuity (VA) over 12 months. RESULTS: The algorithm identified baseline VA <73 Early Treatment Diabetic Retinopathy Study letters, presence of baseline subretinal fluid, and administration of three loading doses by Day 90 from drug initiation as the characteristics with the greatest impact on VA at month 12. When applying the different criteria, RWE outcomes became similar to those obtained in known RCTs. CONCLUSION: Machine learning techniques can be used to classify real world cohorts and identify subsets of patients who benefit to the same extent as that reported in RCTs. This methodology may support the translation of clinical trial findings to treatment performance in the clinical practice setting.

Comparison of real-world treatment outcomes of femtosecond laser-assisted cataract surgery and phacoemulsification cataract surgery: A retrospective, observational study from an outpatient clinic in France.

PURPOSE: This study compared real-world safety and efficacy outcomes of cataract surgery performed with LenSx femtosecond laser-assisted cataract surgery or manual phacoemulsification cataract surgery procedures. METHODS: A retrospective observational study used data from anonymised electronic medical records to compare mean cumulative dissipated energy, the proportion of eyes reaching emmetropia, mean change in best-corrected distance visual acuity and the proportion of eyes with post-surgical complications, including corneal oedema and posterior capsule opacification. Results were adjusted for multiple comparisons for primary and secondary objectives. RESULTS: Data from 811 phacoemulsification cataract surgery and 496 femtosecond laser-assisted cataract surgery procedures were analysed. Mean cumulative dissipated energy was significantly lower for femtosecond laser-assisted cataract surgery (6.5 percent-seconds) than for phacoemulsification cataract surgery (14.3 percent-seconds; p < 0.0001) procedures. More femtosecond laser-assisted cataract surgery (81.2%) procedures achieved emmetropia (⩽ 0.5 dioptre) than did phacoemulsification cataract surgery (73.5%) procedures, although this difference was not statistically significant. Mean change in best-corrected distance visual acuity and the proportion of eyes with corneal oedema, posterior capsule opacification or other complications were not significantly different between cohorts when adjusted for multiple comparisons. CONCLUSIONS: In this single-centre, single-surgeon retrospective electronic medical record database study using divide and conquer technique, femtosecond laser-assisted cataract surgery was associated with significantly lower cumulative dissipated energy when compared to manual phacoemulsification cataract surgery. This supports the hypothesis that femtosecond laser-assisted cataract surgery involves less mechanical trauma, which might lead to more consistent refractive and safety outcomes than manual phacoemulsification cataract surgery, though such outcomes were found to be comparable in this study.

SIERRA-AMD: A Retrospective, Real-World Evidence Study of Patients with Neovascular Age-Related Macular Degeneration in the United States.

PURPOSE: Characterize real-world baseline visual acuity (VA) and anti-vascular endothelial growth factor (VEGF) treatment patterns in neovascular age-related macular degeneration patients in 2012-2015. DESIGN: Retrospective, multicenter, noninterventional real-world evidence study. PARTICIPANTS: A total of 98 821 eyes from 79 885 patients receiving intravitreal anti-VEGF therapy. METHODS: Anonymized patient data routinely collected over 5 years were extracted from 58 United States centers to a central database using an electronic medical records system. MAIN OUTCOME MEASURES: Baseline VA, VA change from baseline, treatment frequencies, annual anti-VEGF injections, bilateral treatment frequencies, annual total clinic visits, and noninjection clinic visits. RESULTS: Baseline characteristics were comparable across years. Baseline VAs (Mean±standard deviation [SD] Early Treatment Diabetic Retinopathy Study [ETDRS] letters) were similar for 2012, 2013, and 2014 (53.6±23.3, 53.2±23.4, and 53.1±23.6, respectively), but was lower for 2015 (50.7±24.4). In eyes with 4-year follow-up, VA changes from baseline (ETDRS letters) were least squares means of +1.1 (95% confidence interval [CI], 1.0;1.3), -1.3 (95%CI, -1.5;-1.0), and -3.1 (95%CI, -3.5;-2.7), and -5.2 (95%CI, -6.0;-4.3) for years 1-4. Mean±SD number of injections was 7.5±1.9, 6.7±2.1, 6.6±2.3, and 6.4±2.3 for years 1-4. By year 4, 36.7% of eyes had ≤8-week dosing intervals (q8w) and 21.2% had ≥12-week dosing intervals. Eyes treated q8w increased 40% from Year 1 (32.4%) to Year 4 (45.3%). Baseline bilateral treatment frequency was 6.1%. Of the patients treated bilaterally, 32.0% received the first treatment in the better-seeing eye, and 68.0% received first treatment in an eye with vision the same as or worse than the fellow-eye. This trend was evident across all index years. CONCLUSIONS: This real-world study describes the treatment burden, initiation and monitoring patterns, and VA outcomes at a scale and timeframe that has not been previously reported. In this cohort, baseline VA was similar for the index years 2012-2014, but lower for 2015. In patients with 4-year follow-up, both VA and injection frequency declined, whereas the proportion of eyes treated more frequently than the recommended q8w interval increased. The reduction in dosing intervals may be a consequence of intensification of treatment due to year-on-year VA loss and disease progression.

Effectiveness of Continued Ranibizumab Therapy in Neovascular Age-Related Macular Degeneration versus Switch to Aflibercept: Real World Evidence.

PURPOSE: To assess the long-term comparative effectiveness of ranibizumab versus switching to aflibercept in neovascular age-related macular degeneration (nAMD). DESIGN: A 24-month, retrospective, comparative, nonrandomized, matched cohort study. PARTICIPANTS: Patients with nAMD initiated on ranibizumab who remained (nonswitchers) or who switched to aflibercept (switchers) captured from a United States electronic medical records database between July 1, 2011 and October 12, 2014. METHODS: Patient eyes were matched for baseline age, baseline visual acuity (VA), VA at month 3, and duration of follow-up. Matching ratio was 1:2 (switchers: nonswitchers) where possible and 1:1 otherwise. MAIN OUTCOME MEASURES: The primary outcome was VA change from baseline (first injection of ranibizumab) to month 24. Secondary end points were standardized area under the curve of VA change; patient eyes (%) gaining or losing ≥5, ≥10, or ≥15 letters, or with VA of >73 letters at month 24; number of injections and monitoring visits; and analysis of preswitch characteristics. RESULTS: A total of 454 switchers and 750 matched nonswitchers were included. The adjusted difference in mean VA change from baseline to month 24 for switchers to nonswitchers was 0.02 letters (95% confidence interval [CI], -1.63 to 1.68). The upper bound 95% CI (1.68) was below the predefined noninferiority margin of 5 letters. Switchers had a significantly higher annualized number of mean total visits compared with nonswitchers (10.0 vs. 9.0 for year 1; 8.7 vs. 7.4 for year 2), a higher number of injection visits (8.4 vs. 6.7 for year 1; 7.0 vs. 5.1 for year 2), but a lower number of monitoring-only visits (1.6 vs. 2.3 for year 1; 1.7 vs. 2.3 for year 2). During the preswitch period, switchers had a higher number of injection visits (7.6 vs. 6.5), fewer monitoring-only visits (1.5 vs. 2.2), and comparable total visits (9.1 vs. 8.7). Visual acuity change from baseline to switch was similar between switchers and nonswitchers (adjusted least squares mean difference, -1.36 letters; 95% CI, -2.76 to 0.05). CONCLUSIONS: Switching patients from ranibizumab to aflibercept resulted in no difference in VA change compared with those maintained on ranibizumab only. The lower retreatment rate in nonswitchers compared with switchers post switch does not support the view of a longer treatment efficacy.

Biomarkers and neurodevelopment in perinatally HIV-infected or exposed youth: a structural equation model analysis.

OBJECTIVE: To examine the relationship between markers of vascular dysfunction and neurodevelopmental outcomes in perinatally HIV-infected (PHIV+) and perinatally HIV-exposed but uninfected (PHEU) youth. DESIGN: Cross-sectional design within a prospective, 15-site US-based cohort study. METHODS: Neurodevelopmental outcomes were evaluated in relation to nine selected vascular biomarkers in 342 youth (212 PHIV+, 130 PHEU). Serum levels were assessed for adiponectin, C-reactive protein (CRP), fibrinogen, interleukin-6 (IL-6), soluble vascular cell adhesion molecule-1 (sVCAM-1), E-selectin (sE-selectin), monocyte chemoattractant protein (sMCP-1), intercellular adhesion molecule-1 (sICAM-1), and P-selectin (sP-selectin). The Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) was administered at entry, yielding a Full-Scale IQ score, and four index scores. Factor analysis was conducted to reduce the biomarkers to fewer factors with related biological roles. Structural equation models (SEMs) were used to measure associations between resulting factors and WISC-IV scores. RESULTS: Mean participant age was 11.4 years, 54% were female, 70% black. The nine biomarkers were clustered into three factor groups: F1 (fibrinogen, CRP, and IL-6); F2 (sICAM-1 and sVCAM-1); and F3 (MCP-1, sP-selectin, and sE-selectin). Adiponectin showed little correlation with any factor. SEMs revealed significant negative association of F1 with WISC-IV processing speed score in the total cohort. This effect remained significant after adjusting for HIV status and other potential confounders. A similar association was observed when restricted to PHIV+ participants in both unadjusted and adjusted SEMs. CONCLUSION: Aggregate measures of fibrinogen, CRP, and IL-6 may serve as a latent biomarker associated with relatively decreased processing speed in both PHIV+ and PHEU youth.

Point-of-care capillary blood lactate measurements in human immunodeficiency virus-uninfected children with in utero exposure to human immunodeficiency virus and antiretroviral medications.

OBJECTIVES: To estimate the prevalence of elevated point-of-care (POC) capillary blood lactate concentrations in human immunodeficiency virus (HIV)-exposed, uninfected children (HEU) and to determine if POC lactate varies with in utero antiretroviral (ARV) exposure. METHODS: The Surveillance Monitoring for Antiretroviral Therapy Toxicities protocol of the Pediatric HIV/AIDS Cohort Study enrolled 1934 children between 2007 and 2009, 0 to 12 years of age, born to HIV-infected mothers. POC lactate was measured annually on capillary blood using the Lactate Pro device. Associations of POC lactate with in utero ARV exposure and other characteristics were evaluated using logistic regression models, adjusting for maternal characteristics and other confounders. RESULTS: Of 1641 children with POC measurements (median age, 3.0 years), 3.4% had POC lactate >3 mmol/L. Median POC lactate level decreased with age (1.9 mmol/L, 1.7 mmol/L, and 1.6 mmol/L for children 0-<6 months [99% ≤6 weeks of life], 6-<24 months, and ≥24 months of age, respectively; P < 0.001). Prevalence of elevated POC lactate did not differ by in utero ARV exposure drug class, but was significantly higher in children exposed in utero to emtricitabine or efavirenz, cocaine or opiates, and those of white race. CONCLUSIONS: POC lactate testing is a useful rapid laboratory screening assay for HEU children with ARV exposure. ARV use during pregnancy has resulted in a dramatic decrease in mother-to-child transmission of HIV, and the risk of elevated lactate in HEU children is low. However, as new ARVs and more complex regimens are used during pregnancy by HIV-infected women, continued monitoring for infant toxicities is essential.