In 2012, a team at the Scripps Translational Science Institute, Eric Topol among them, reported a striking finding in Science Translational Medicine. They used an automated rare-cell platform originally built to detect tumour cells. With it, they found that patients with acute myocardial infarction carried markedly elevated numbers of circulating endothelial cells. Many of these cells looked abnormal, with enlarged nuclei and multicellular clusters rarely seen in healthy controls. The vessel wall, it seemed, was shedding evidence of its own injury into the bloodstream.

That observation sits within a larger idea. The endothelium is a single layer of cells lining every blood vessel, spread across thousands of square metres. It is metabolically active and constantly signalling. Conventional risk tools, such as Framingham, the pooled cohort equations and even coronary calcium scoring, mostly describe accumulated burden or statistical likelihood. They say little about whether the endothelium is being damaged right now, or whether the body can still repair it.

Circulating endothelial cells (CECs) and endothelial progenitor populations offer a potential readout of both sides of that equation: injury and regeneration. This article covers three things. First, the biology that separates these populations. Second, the technical difficulty of counting cells present at a few per millilitre. Third, how the balance between them may eventually help stratify risk and set the intensity of prevention for individual patients.

CEC and EPC Biology: Two Cells, Two Stories

Circulating endothelial cells (CECs) are mature, terminally differentiated cells that have detached from the vascular intima and entered the bloodstream. They express endothelial markers such as CD146, CD31, CD105 and von Willebrand factor. They lack the leukocyte marker CD45 and the progenitor marker CD133. In healthy individuals they are very rare, typically single digits to low tens per millilitre depending on the method. Finding them in quantity is, almost by definition, a sign that something has gone wrong.

Several forces normally anchor endothelial cells to the vessel wall: integrin binding to the basement membrane, junctions between neighbouring cells, and survival signalling. Inflammatory cytokines, proteases, disturbed shear stress, oxidative injury and immune attack can all loosen that grip. Elevated counts were described in unstable angina and myocardial infarction as early as 1999. They have since been documented in vasculitis, sickle cell crises, sepsis, transplant rejection and pulmonary hypertension. CECs do not point to a specific disease. They point to injury.

Endothelial progenitors tell the opposite story. In 1997, Takayuki Asahara and Jeffrey Isner reported in Science that CD34-positive cells from adult peripheral blood could take on endothelial characteristics. These cells also incorporated into sites of new vessel growth. The finding challenged the assumption that vessel growth after birth depended only on local angiogenesis, and it launched two decades of research into vascular repair driven by the bone marrow. The working definition became CD34+/KDR+ (VEGFR2) cells, often with CD133 as a marker of immaturity.

The field later had to face an uncomfortable problem: the term EPC was being used for several different cells. Early-outgrowth cultures, including the colony-forming units used in many landmark studies, are mostly myeloid cells (CD14+, CD45+). They support repair by secreting VEGF, HGF and IL-8 rather than by becoming endothelium themselves. Late-outgrowth endothelial colony-forming cells (ECFCs) are true progenitors that can expand from a single cell and form functional vessels in vivo. A 2017 consensus statement recommended replacing the umbrella term with ECFCs and myeloid angiogenic cells.

Clinically, this gives a useful pairing. CECs are the debris of active damage. Progenitor populations are the reserve of repair capacity, released from bone marrow through VEGF, SDF-1/CXCR4 signalling and eNOS-dependent MMP-9 activation. A high CEC count alongside strong progenitor mobilisation suggests that injury is being answered. The same injury signal alongside depleted progenitors suggests a vascular system losing its ability to heal itself. Reading either number alone discards half the information.

Takeaway

A rising CEC count tells you the vessel wall is being damaged, while a falling progenitor count tells you the body is losing its ability to respond. Each number needs the other to mean much.

Detection Methodologies: Counting the Vanishingly Rare

Counting CECs is an unusually difficult rare-event problem. A millilitre of blood holds roughly five billion red cells and several million white cells, while the target population may number fewer than twenty. How the sample is handled before analysis matters as much as the assay itself. Anticoagulant choice, time to processing and even venipuncture technique affect results. The needle can dislodge endothelial cells from the vein wall, which is why protocols commonly discard the first tube drawn.

The earliest standardised approach, set out in a 2006 consensus, uses immunomagnetic isolation. CD146-coated beads capture the cells, which are then stained with Ulex europaeus lectin and counted under fluorescence microscopy. Checking each cell's size, nuclear features and lectin binding makes the method highly specific. Its weakness is throughput: it is labour-intensive, depends on the operator, and is hard to scale beyond specialised laboratories. Even so, it established reference ranges and remains the benchmark for judging newer methods.

Flow cytometry measures many markers per cell at once and is the main method for counting progenitors. A rigorous panel excludes CD45-positive cells and includes CD146, CD31, CD34 and CD133. It also adds a viability dye and a nucleic acid stain to exclude platelets, microparticles and debris. This matters because CD31 is abundant on platelets and CD146 appears on a subset of activated T cells. Reliable statistics require acquiring hundreds of thousands to millions of events. Without shared gating standards, published values have historically varied by orders of magnitude.

The CellSearch platform is FDA-cleared for counting circulating tumour cells. It has been adapted for CECs using CD146 magnetic enrichment, CD105 staining, DAPI nuclear staining and CD45 exclusion, followed by image-based review. It offers reproducibility and preserves cell morphology, the feature that enabled the Scripps myocardial infarction findings. However, its CEC application remains research use only, the marker set is fixed, and costs are substantial. Assessing ECFC function, meanwhile, still requires culture over several weeks.

Emerging technologies aim to combine scale with specificity. Imaging flow cytometry captures both markers and morphology at high throughput. Mass cytometry measures dozens of markers at once and has revealed variation within cell groups once assumed to be uniform. Microfluidic capture devices promise counting at the bedside, and machine learning classifiers are being trained to recognise injured endothelial cells automatically. Perhaps most intriguing, methylation patterns in cell-free DNA can identify fragments released by endothelial cells, giving a cell-free proxy for endothelial turnover. Each approach trades one limitation for another, and none has yet been validated prospectively for cardiovascular outcomes.

Takeaway

In rare-cell biology, the assay is part of the biomarker. A CEC count reported without its method, gating strategy and sample handling cannot be meaningfully compared with any other.

Clinical Applications: The Injury-to-Repair Ratio

The strongest prognostic evidence concerns progenitors. In 2003, Hill and colleagues reported in the New England Journal of Medicine that colony-forming unit counts fell as Framingham risk score rose. These counts also predicted brachial flow-mediated dilation better than conventional risk factors did. Two years later, Werner and colleagues followed 519 patients with coronary artery disease. Higher levels of CD34+/KDR+ cells were associated with a lower risk of cardiovascular death and first major cardiovascular events over twelve months.

CEC data come mostly from acute and inflammatory settings. Counts rise in acute coronary syndromes and track with markers of heart muscle and endothelial injury. In several cohorts, higher counts have been associated with later adverse events. Elevations are also documented in diabetes, chronic kidney disease and systemic autoimmune disease. In all of these, accelerated atherosclerosis is well established but poorly captured by traditional risk scores. In these settings, the CEC count works as a direct readout of ongoing vascular injury.

The CEC/EPC ratio tries to combine both signals into an injury-to-repair index. The logic is intuitive. Two patients with identical lipid profiles might differ fundamentally if one sheds endothelium faster than she can replace it. Studies in diabetic, kidney and rheumatological populations suggest that this imbalance tracks with endothelial dysfunction and vascular events. However, the ratio inherits the measurement variability of both components, and validated cut-off values from prospective studies do not yet exist.

Its most compelling potential use is guiding how intensive prevention should be. Progenitor counts can be changed. Statins increase circulating EPCs within weeks, and structured exercise training mobilises them. Smoking cessation, renin-angiotensin blockade and better glycaemic control have also shown favourable effects. CEC counts fall with effective treatment in vasculitis. A patient at intermediate conventional risk with an unfavourable ratio might justify earlier or more intensive therapy. Repeat measurements could then confirm whether the vascular biology is actually responding.

Candour is essential here. No guideline endorses CEC or EPC measurement for cardiovascular risk, and no outcome trial has compared ratio-guided therapy with standard care. The nearer-term roles lie in research cohorts, transplant vasculopathy and specialised vascular medicine. They also include monitoring endothelial damage from anticancer agents such as VEGF inhibitors. In the longer term, the value may come from combining cellular measures with polygenic risk, proteomics, imaging and continuous physiological monitoring into individualised risk models.

Takeaway

Most cardiovascular risk scores are snapshots of damage that has already accumulated. An injury-to-repair ratio aims to show what is happening now, but only once the methods behind it are reliably standardised.

Counting circulating endothelial cells and progenitors shifts cardiovascular assessment away from accumulated burden and toward biology as it is happening. It asks whether the vessel wall is being injured now, and whether the body can still repair it.

For centres pursuing precision vascular care, a defensible protocol pairs two measurements. The first is a standardised CEC count, using immunomagnetic isolation or CellSearch. The second is a flow cytometry count of CD34+/KDR+ progenitors. Both need consistent sample handling and gating. Measurements are most informative at baseline and after an intervention, read alongside lipids, inflammatory markers, imaging and genetic risk. The ratio is a tool for calibration, not a verdict.

The endothelium has always been signalling its condition into the circulation, one detached cell at a time. Those signals exist. The open question is whether we can measure them reliably enough to act on them, and that is what the next decade of vascular precision medicine will decide.