Skip to content

Biomarkers in thoracic aortic aneurysm and dissection

TL;DR — Exactly one circulating biomarker has earned a place in aortic practice, and it works only in the acute setting and only as a rule-out: D-dimer, pooled sensitivity 95–96% at a 500 ng/mL cutoff but specificity only 60–70%, and it must be paired with a pretest probability score to be safe (Nazerian 2017, PMID 29030346; Watanabe 2016, PMID 27230962). For the far larger clinical problem — who has an undiagnosed aneurysm, whose aneurysm will grow, and who will dissect below the surgical threshold — there is no validated blood test. Dozens of candidates (MMPs/TIMPs, TGF-β1, microRNAs, sST2, elastin cross-links) have been reported, mostly in single-centre case-control cohorts of 20–400 patients with AUCs of 0.7–0.9; the two that reached larger prospective testing — sST2 and plasma desmosine — did so in acute dissection or in the abdominal aorta, and none has been prospectively replicated against a thoracic growth or dissection endpoint. The 2022 ACC/AHA guideline therefore contains no biomarker-based recommendation. The most credible near-term candidates are not proteins at all but genetic (polygenic scores adding ~4–11 absolute percentage points of explained diameter variance) and imaging (18F-sodium fluoride microcalcification, 4D-flow-derived wall stress), and the biggest unknown is whether any systemic marker can be informative about a focal, mechanically driven, single-organ failure.

1. What a biomarker would have to do

Five distinct decision problems are conflated under "aortic biomarker". They have different populations, different prevalences, and different performance requirements.

# Decision Population Prevalence of target What a test must achieve
1 Rule out acute aortic syndrome in the ED Chest/back/abdominal pain, syncope 13–17% in enriched ED cohorts (PMID 29030346, 31226214) Failure rate <1% (miss rate), high efficiency to reduce CT use
2 Case-find undiagnosed aneurysm General adults Low — mean ascending diameter 3.04 cm (women) / 3.32 cm (men) in UK Biobank (PMID 36378208) Positive predictive yield good enough to justify imaging; the best clinical score needs 1.8–9.7 imaged per aorta ≥4.0 cm found
3 Predict growth Known small/moderate aneurysm Growth is 0.1–0.2 cm/yr; most aortas are stable Discriminate slow from fast growers over 1–2 yr
4 Predict dissection below threshold Known aneurysm <5.5 cm Events are rare per patient-year Reclassify individuals across an operative threshold
5 Monitor drug response Patients on medical therapy — Change faster and more sensitively than diameter

Only problem 1 has a partial solution. Problems 3 and 4 are the ones that would change surgical decision-making, and they are where the evidence is weakest — see risk stratification and size thresholds.

2. D-dimer: the one biomarker that made it into a diagnostic algorithm

2.1 ADvISED

The ADvISED study is the anchor. In a prospective, multicentre, 6-hospital, 4-country study (2014–2016) of 1,850 consecutive ED outpatients in whom acute aortic syndrome (AAS) was in the differential, 241 (13%) had AAS: 125 type A dissection, 53 type B dissection, 35 intramural haematoma, 18 aortic rupture, 10 penetrating aortic ulcer (Nazerian 2017, PMID 29030346). Performance of D-dimer alone and in combination with the aortic dissection detection risk score (ADD-RS, 0–3):

Strategy Sensitivity / specificity Failure rate (missed AAS) Efficiency (proportion ruled out)
D-dimer <500 ng/mL alone Sens 96.7% (93.6–98.6); Spec 64.0% (61.6–66.4) 8 of 241 AAS cases had D-dimer negative —
ADD-RS = 0 and D-dimer negative — 0.3% (0.1–1.9); 1 AAS in 294 patients 15.9% (14.3–17.6)
ADD-RS ≤1 and D-dimer negative — 0.3% (0.1–1.0); 3 AAS in 924 patients 49.9% (47.7–52.2)

The headline is that the ADD-RS ≤1 / D-dimer-negative rule safely dismisses half of the suspected population at a 0.3% miss rate — but that eight patients with genuine AAS had a normal D-dimer is the reason it is a rule-out adjunct and never a stand-alone test. The ADvISED authors themselves framed D-dimer as "inadequate as a stand-alone test" (PMID 29030346), a position they defended in published correspondence (Morello 2018, PMID 30571366).

A prespecified ADvISED subanalysis added transthoracic focused cardiac ultrasound (FoCUS) in 839 patients (17.4% AAS): direct FoCUS signs had sensitivity 45.2% but specificity 97.4%; any FoCUS sign, sensitivity 89.0% / specificity 74.5%. ADD-RS ≤1 plus negative FoCUS gave a 1.9% failure rate; adding negative D-dimer drove the failure rate to 0% (95% CI 0–1.2%) (Nazerian 2019, PMID 31226214). This is the strongest available evidence that D-dimer's value is as one term in a multi-modal bundle, not as a test.

2.2 Meta-analytic performance and its limits

Meta-analysis Studies / subjects Pooled sensitivity Pooled specificity AUC
Watanabe 2016 (PMID 27230962) 22 studies, 1,140 AAS / 3,860 non-AAS 0.952 (0.901–0.978) at 500 ng/mL 0.604 (0.485–0.712) 0.946 (0.903–0.994)
Yao 2021 (PMID 34838062) 16 studies, 1,135 patients 0.96 (0.91–0.98) 0.70 (0.57–0.81) 0.94 (0.91–0.95)
Cui 2015 (PMID 25634194) 5 studies, 274 AAS / 469 controls 0.945 (0.781–0.988) 0.691 (0.437–0.865), p=0.136 0.916 (0.863–0.970)

Negative likelihood ratio is 0.06–0.08; positive likelihood ratio only 2.4–3.3 (PMID 27230962, 34838062). Translated: a negative D-dimer meaningfully lowers post-test probability; a positive one barely raises it. Watanabe's own conclusion is that D-dimer >500 ng/mL "moderately increases" the possibility of AAS (PMID 27230962) — it is not a positive diagnostic.

Documented failure modes of D-dimer in AAS:

  • Thrombosed false lumen / intramural haematoma and penetrating ulcer generate less fibrin turnover; these subtypes are over-represented among false negatives. ADvISED included 35 IMH and 10 PAU among 241 AAS cases (PMID 29030346).
  • Early presentation — D-dimer rises and falls with time from onset.
  • Poor specificity in the population where it is used: pulmonary embolism, sepsis, malignancy, and advanced age all raise it, so applying D-dimer alone would increase, not decrease, CT angiography volume.
  • No role outside the acute setting. D-dimer is not a marker of aneurysm presence or growth.

2.3 Other acute-phase markers

Smooth muscle myosin heavy chain, creatine kinase BB isoenzyme, calponin (smooth muscle troponin), soluble elastin fragments, and C-reactive protein have all been proposed for biochemical detection of acute dissection; none has an accepted strategy or a commercially deployed assay (Suzuki 2010, PMID 20717014). Fifteen years after that review, that statement is unchanged — this is one of the field's most durable failures. See aortic dissection.

Soluble ST2 (sST2) is the most developed alternative and the clearest cautionary tale. In 1,360 patients (1,027 in a retrospective discovery set, 333 in a prospective validation cohort), sST2 measured within 24 h of ED presentation gave an AUC of 0.97 (0.95–0.98) against 0.91 (0.88–0.94) for D-dimer and 0.50 (0.44–0.56) for cardiac troponin I; at a 34.6 ng/mL cutoff, sensitivity was 99.1%, specificity 84.9%, negative predictive value 99.7% (Wang 2017, PMID 29146682). An independent prospective European ED study of 297 patients (88 with AAS) did not replicate this: AUC 0.63 for sST2 versus 0.82 for D-dimer (p<0.001), and the only cutoff reaching 95.5% sensitivity (≥12 ng/mL) had a specificity of 10.8% (Morello 2020, PMID 32080259). A systematic review of 13 cohort studies of non-D-dimer biomarkers judged the two sST2 estimates irreconcilable (99.1%/84.9% versus 58%/70.8%) and concluded that alternatives to D-dimer are not ready for clinical use (Wren 2024, PMID 39107052). This is the field's cleanest example of a discovery-set AUC of 0.97 collapsing under external prospective validation.

3. Circulating markers of aneurysm presence and growth

3.1 The evidence, with cohort sizes

Marker Design and n Reported result Replication status
MMP-1/2/3/8/9, TIMP-1/2/3/4 (plasma + tissue) 21 BAV + 21 TAV ascending TAA vs 10 non-aneurysmal, surgical specimens (Ikonomidis 2013, PMID 23312977) Distinct plasma "cassettes" by valve morphology; plasma–tissue correlation reached significance only for MMP-8, TIMP-1, TIMP-3, TIMP-4 Not prospectively replicated against growth
Serum MMP-9 79 TAA, 105 AAA, 112 controls, case-control ELISA (Li 2018, PMID 30373522) AUC 0.83 for TAA (sens 70%, spec 91%); AUC 0.69 for AAA. MMP-9 higher in TAA than AAA Single centre; confounded by age and hypertension in the same paper
Total TGF-β1 53 MFS vs 74 controls, plus Fbn1^C1039G/+ mice (Matt 2009, PMID 19635970) MFS 15 ± 1.7 ng/mL vs controls 2.5 ± 0.4 ng/mL (p<0.0001); fell with losartan or beta-blocker; correlated with aortic root diameter in mice (p=0.002) Contested — see below
Circulating TGF-β in acute dissection Correspondence (Suzuki 2011, PMID 21816317) Title reports circulating TGF-β levels in acute dissection; PubMed carries no abstract, so direction and effect size are not verifiable from the record Letter-format; not a definitive cohort
miR-574-5p Tissue microarray 19 TAA vs 19 controls; serum 28 TAA vs 20 controls (Boileau 2019, PMID 31409059) Down-regulated in tissue, up-regulated in serum; serum AUC 0.87 Discovery-scale only (n=48 serum); direction of change inverts between compartments
miR-21-5p, -29b-5p, -126-5p/-3p, -181b-5p, -92a-3p Degenerative TAA vs stable CAD vs controls (Ekedi 2023, PMID 37511774) Panel discriminated TAA from CAD and controls with "very high predictive ability" on ROC; miR-92a-3p reported as novel Not externally validated; sample-size reporting internally inconsistent in the published abstract
miR-1, -21, -29a, -133a, -143, -145 Same BAV/TAV surgical cohort as above (PMID 23312977) miR-1 and miR-21 differed significantly between BAV and TAV aortic tissue Tissue only

3.2 Three often-cited candidates, and what their primary data actually cover

Desmosine, sST2 and lumican appear on every list of TAA biomarker candidates. Primary human data exist for all three, but almost none of it comes from the population the list implies — patients under surveillance for a stable thoracic aneurysm (searched 2026-08-28).

  • Desmosine / isodesmosine (elastin cross-link turnover). The only thoracic cohort is small: 30 thoracoabdominal aneurysm patients versus 30 age- and sex-matched controls, plasma desmosine 0.40 ± 0.31 vs 0.22 ± 0.15 ng/mL (p<0.001), AUC 0.82, and at a 0.27 ng/mL threshold sensitivity 78.6% / specificity 76.7%; plasma levels correlated with intramural MMP-2 (ρ=0.68) and TIMP-1 (ρ=0.72) (Doukas 2026, PMID 41683665). The prognostic evidence is abdominal: across 507 AAA patients and 162 controls, plasma desmosine had the strongest correlation with diameter of any serum marker (r=0.39) and predicted AAA events after adjustment for baseline diameter (HR 2.03 per SD, 1.02–4.02) (Mordi 2019, PMID 31595818). A faster LC-MS assay found roughly 3-fold higher desmosine in acute dissection than in controls (Kuzmanova 2025, PMID 39891463). No study has tested desmosine against thoracic growth or incident dissection prospectively (Gombert 2026, PMID 41976840).
  • Soluble ST2 has the largest primary dataset of the three — and all of it is acute-dissection diagnosis, where it failed external prospective validation (§2.3). No cohort has tested sST2 in stable thoracic aneurysm.
  • Lumican. Serum lumican was higher in 70 acute dissections (2.32 ± 4.29 ng/mL) than in 12 intramural haematomas (0.72 ± 0.32) or 30 healthy volunteers (0.85 ± 0.53; p=0.003), correlating weakly with radiological severity (ρ=0.37–0.47) (Gu 2016, PMID 26998013); serum levels are higher in acute than chronic dissection, and Lum-null mice have higher dissection mortality and rupture rates (Chen 2021, PMID 34310653); post-operative levels track prolonged ventilation and length of stay in 58 aortic surgery patients (Hsu 2021, PMID 33661915). The only aneurysm data are a 15-patient SWATH-MS discovery pilot in bicuspid-valve patients, in which lumican was one of five proteins showing lower abundance variability in the aneurysmal group (Harrison 2018, PMID 29929532).

The pattern is identical in all three cases: the marker is real in acute dissection or in the abdominal aorta, and is then imported into thoracic aneurysm surveillance, where it has never been tested.

3.3 The non-replication problem, diagnosed

The candidate-marker literature has a consistent structural defect, visible in the table above:

  1. Case-control at the operating table. Cases are patients presenting for surgical repair — i.e. at 5.0–5.5 cm or larger — and controls are transplant donors, CABG patients, or healthy volunteers. This is maximal spectrum bias: the design asks "can we tell a 5.5 cm aorta from a 3.0 cm one?", which is a question CT already answers for free.
  2. Cross-sectional, not longitudinal. Almost no study uses growth rate or incident dissection as the endpoint. Discriminating prevalent disease is not the clinical need.
  3. Sample sizes of 20–100 per arm with AUCs reported to three decimal places and no held-out validation set.
  4. Compartment inconsistency. miR-574-5p is lower in aneurysm tissue and higher in serum (PMID 31409059); this may be genuine extracellular-vesicle biology, but it means tissue findings cannot be used to predict serum direction.
  5. Confounding by the disease's own risk factors. Serum MMP-9 was itself influenced by age and hypertension in the paper reporting it (PMID 30373522), and correlated with CRP (r=0.33) and homocysteine (r=0.199) — i.e. it is partly a general inflammation readout.
  6. The TGF-β cautionary tale. Circulating TGF-β1 is the best-documented case: a large reported effect in Marfan syndrome that fell with treatment (PMID 19635970), which would be an ideal pharmacodynamic marker — but TGF-β measurement is notoriously sensitive to platelet activation during sample handling, and a dedicated interventional study designed to measure circulating TGF-β in Marfan patients (NCT01361087) was withdrawn with zero enrolment. The mechanistic ambiguity is deeper still: the TGF-β story in TAA is a paradox rather than a gradient (see pathophysiology), and single-cell data show TGF-β receptor genes down-regulated in Marfan aortic wall even where ligand is up (PMID 35052435). A circulating ligand level cannot resolve that.

4. Genetic markers used as biomarkers

Genotype is the one "biomarker" already changing management, and it does so by a different route: it identifies who needs earlier surveillance and a lower operative threshold, not by measuring current disease activity.

  • Single-gene diagnoses carry quantitative natural-history differences. In ACTA2 Arg179 smooth muscle dysfunction syndrome, thoracic aortic aneurysm repair or dissection had occurred in 12/33 (36%) at a median age of 14 years, and aortic disease was fully penetrant by age 25 (Regalado 2018, PMID 29300374). In PRKG1 p.Arg177Gln families, 12/29 carriers had type B dissection at a median descending aortic diameter of only 4.1 cm (range 3.8–5.0), median age 31 (Shalhub 2019, PMID 30871887) — i.e. genotype predicts dissection at diameters that no size threshold would flag. See genetics of TAA and syndromic aortopathies.
  • Polygenic scores are the emerging quantitative version. A 1.1-million-variant score added to clinical factors raised explained variance in ascending diameter from 29.2% to 39.9% in UK Biobank and improved AUROC for diameter ≥4 cm from 0.765 to 0.834 (p=7.3×10⁻¹⁰), replicating in Mass General Brigham, Framingham, and All of Us (Pirruccello 2023, PMID 37662232). An ascending-aorta polygenic score was associated with incident thoracic aortic aneurysm at HR 1.43 per SD (95% CI 1.32–1.54) in 385,621 participants (Pirruccello 2022, PMID 34837083). Details in omics and emerging science.
  • The ceiling this exposes: a purely clinical 11-variable score explained only 28–33% of diameter variance and, at a fixed operating threshold, had sensitivity of 8.9–18.8% with specificity 96–99% (Pirruccello 2022, PMID 36378208). Adding genetics helps but does not change the regime — most of the variance in aortic size, and therefore most of the risk, remains unexplained by anything measurable in blood.

5. Imaging biomarkers beyond diameter

Diameter is a poor surrogate for wall integrity; the imaging field is trying to measure the wall instead.

Modality What it measures Best evidence Status
18F-sodium fluoride PET Medial microcalcification ("granular medial calcinosis") 101 thoracic aortic specimens from 57 aortopathy patients + 18 controls: microcalcification higher in mild (6.17 [2.71–10.39]) and moderate (3.74 [0.87–11.80]) degeneration vs controls (0.79 [0.36–1.90]); 18F-NaF autoradiography correlated with histological microcalcification r=0.76, p<0.001 (Fletcher 2022, PMID 35770666) Ex vivo validated; non-monotonic — severe degeneration shows less microcalcification (ρ = −0.51, p<0.0001), tracking elastin loss
18F-sodium fluoride PET (abdominal proof-of-concept) Aortic microcalcification activity (AMA) as a disease-activity index 10 patients pre/post EVAR: AMA fell in suprarenal (ΔAMA 0.62, p=0.03), neck (0.72, p=0.02) and aneurysm body (0.69, p=0.02) but not thoracic aorta (0.11, p=0.41) (Debono 2023, PMID 37164479) AAA, n=10 — hypothesis-generating; shows the signal is modifiable
18F-FDG PET Metabolic/inflammatory activity Grade III aortic-wall FDG uptake is a standard imaging criterion for aortitis — giant-cell arteritis, Takayasu, clinically isolated aortitis (Espitia 2024, PMID 39034261). No validated FDG threshold for degenerative TAA disease activity exists (searched 2026-08-28); the thoracic FDG literature is dominated by graft-infection work, where uptake separates infected from non-infected grafts (mean SUV ratio 2.19 vs 1.63, p<0.001) but says nothing about wall degeneration (Hasse 2024, PMID 38679286) Not established for degenerative TAA
4D-flow MRI Wall shear stress, flow eccentricity, turbulent kinetic energy In 112 non-dysfunctional bicuspid valves, leaflet fusion length (mean 7.8 ± 3.2 mm, range 2.3–15.4) was independently associated with sinus (p=0.002) and ascending (p=0.028) diameter and positively related to flow asymmetry, vortices and circumferential WSS (Guala 2021, PMID 33977309). Fluid–structure-interaction models of 7 ATAA vs 6 healthy aortas: max TKE 155.0 ± 188.4 Pa vs 0.6 ± 0.5 Pa; max TAWSS 8.6 ± 6.5 vs 2.8 ± 0.7 Pa; peak wall stress 414 ± 108 vs 215 ± 31 kPa (Zhu 2025, PMID 40067580) Mechanistically rich, prognostically unproven; see hemodynamics and biomechanics
Arterial tortuosity Vessel geometry as a surrogate for connective-tissue severity In 65 vascular EDS subjects, height-adjusted vertebral tortuosity index ≥15.5 discriminated cardiovascular events by age 30 (sens 70%, spec 76%) and was associated with events in high-risk-variant carriers <40 yr (IRR 4.14, 1.13–15.10) but not overall (IRR 1.79, 0.76–4.24) (Stephens 2023, PMID 37776192) Genotype-conditional; not generalised to non-syndromic TAA

The 18F-NaF finding deserves emphasis because it inverts a naive expectation: microcalcification marks early aortopathy and then disappears as the wall degenerates and elastin is lost (PMID 35770666). Any PET-based staging system therefore cannot use "more signal = worse disease" — a lesson that generalises to every activity-based marker of a disease whose end state is tissue absence rather than tissue accumulation.

6. Why nothing is in the guidelines

No circulating biomarker carries a class of recommendation for aneurysm detection, surveillance, or operative timing in the 2022 ACC/AHA aortic disease guideline (see guidelines). The mechanistic reasons are structural, not accidental:

  1. Dilution. The ascending aorta is roughly 30–60 g of tissue perfused by 5 L/min of cardiac output. Any molecule released focally is diluted into a systemic pool dominated by other sources.
  2. The disease is mechanical and focal. Rupture and dissection are local failures of a heterogeneous wall. A well-mixed venous sample averages away exactly the spatial information that determines the event.
  3. No reference standard for "activity". Growth is the only accessible longitudinal endpoint, and it is measured with an inter-study error of the same magnitude as one year of growth (see imaging and surveillance). A biomarker cannot be validated against a noisier gold standard than itself.
  4. Events are rare. Validating a dissection-prediction marker requires accruing events at rates of order 1%/yr, which means cohorts in the thousands followed for years — the same barrier that blocks trials (see clinical trials landscape).
  5. Aetiological heterogeneity. BAV, tricuspid degenerative, Marfan, Loeys-Dietz, and ACTA2 aortopathies have different cellular pathology (see bicuspid aortopathy, pathophysiology). A marker calibrated in a mixed surgical cohort is calibrated to a mixture that no individual patient represents. Ikonomidis 2013 (PMID 23312977) is the clearest demonstration: the MMP/TIMP plasma signature differed between BAV and TAV aneurysms of the same size.

7. Performance targets a useful biomarker would have to hit

Working backwards from each decision problem:

Use case Required operating point Why
Improve on ADvISED for AAS rule-out Failure rate ≤0.3% at efficiency >50% Must beat ADD-RS ≤1 + D-dimer, which already achieves 0.3% / 49.9% (PMID 29030346)
Population case-finding Better than 1.8–9.7 people imaged per ≥4.0 cm aorta found Current clinical score benchmark across UK Biobank, Framingham and Mass General Brigham (PMID 36378208)
Growth prediction Discriminate ≥0.3 cm/yr from <0.1 cm/yr at 12 months Would change surveillance interval, the one decision made every year
Dissection prediction below threshold Reclassify across the 5.0–5.5 cm decision boundary, calibrated, with a decision-curve net benefit This is the endpoint TITAN:SvS (NCT03536312) is randomising against; a biomarker that reproduced its answer per-patient would be practice-changing
Pharmacodynamic monitoring Change detectably within 3–6 months Aortic root Z-score changes ~0.07–0.13/yr on/off ARB (Pitcher 2022, PMID 36049495) — too slow to run efficient dose-finding trials on

The last row is the underrated one. The absence of a fast pharmacodynamic marker is a cause of the empty drug pipeline, not merely a parallel failure: without one, every candidate therapy must be tested against multi-year imaging endpoints, which is why so few reach phase 2.

Open questions

  • Can any circulating marker discriminate fast from slow growth prospectively, rather than discriminating operated aneurysms from normal aortas cross-sectionally? Every candidate in §3 was validated on the latter design (PMID 23312977, 30373522, 31409059, 37511774), and none on the former.
  • Does 18F-sodium fluoride uptake predict incident growth or dissection in the thoracic aorta, and how should the non-monotonic relationship between microcalcification and histological severity (ρ = −0.51, PMID 35770666) be handled in a staging system?
  • Is the failure of circulating TGF-β1 (PMID 19635970; withdrawn study NCT01361087) an assay problem, a biology problem, or both — and is there any blood-accessible readout of aortic TGF-β pathway state given that receptor genes are down-regulated in Marfan aortic wall (PMID 35052435)?
  • Can 4D-flow-derived wall stress (peak wall stress 414 ± 108 kPa in ATAA vs 215 ± 31 kPa in controls, PMID 40067580) be reduced to a reproducible, scanner-independent scalar with a threshold, or does its value require patient-specific FSI modelling that will not scale?
  • Does a polygenic score, which adds 4.0–10.7 absolute percentage points of explained diameter variance across four cohorts (PMID 37662232), add anything on top of the measured diameter itself for predicting events — the only comparison that matters clinically, and one not yet reported.
  • Do BAV and tricuspid degenerative aneurysms require entirely separate biomarker panels, as the divergent MMP/TIMP signatures suggest (PMID 23312977)?

References

  1. Nazerian P, Mueller C, Soeiro AM, et al. Diagnostic Accuracy of the Aortic Dissection Detection Risk Score Plus D-Dimer for Acute Aortic Syndromes: The ADvISED Prospective Multicenter Study. Circulation. 2018;137:250-258. PMID 29030346
  2. Nazerian P, Mueller C, Vanni S, et al. Integration of transthoracic focused cardiac ultrasound in the diagnostic algorithm for suspected acute aortic syndromes. Eur Heart J. 2019;40:1952-1960. PMID 31226214
  3. Morello F, Mueller C, Soeiro AM, et al. Response by Morello et al to Letters Regarding Article, "Diagnostic Accuracy of the Aortic Dissection Detection Risk Score Plus D-Dimer for Acute Aortic Syndromes". Circulation. 2018;138:448-449. PMID 30571366
  4. Watanabe H, Horita N, Shibata Y, et al. Diagnostic test accuracy of D-dimer for acute aortic syndrome: systematic review and meta-analysis of 22 studies with 5000 subjects. Sci Rep. 2016;6:26893. PMID 27230962
  5. Yao J, Bai T, Yang B, Sun L. The diagnostic value of D-dimer in acute aortic dissection: a meta-analysis. J Cardiothorac Surg. 2021;16:343. PMID 34838062
  6. Cui JS, Jing ZP, Zhuang SJ, et al. D-dimer as a biomarker for acute aortic dissection: a systematic review and meta-analysis. Medicine (Baltimore). 2015;94:e471. PMID 25634194
  7. Suzuki T, Distante A, Eagle K. Biomarker-assisted diagnosis of acute aortic dissection: how far we have come and what to expect. Curr Opin Cardiol. 2010;25:541-545. PMID 20717014
  8. Wang Y, Tan X, Gao H, et al. Magnitude of Soluble ST2 as a Novel Biomarker for Acute Aortic Dissection. Circulation. 2018;137:259-269. PMID 29146682
  9. Morello F, Bartalucci A, Bironzo M, et al. Prospective diagnostic accuracy study of plasma soluble ST2 for diagnosis of acute aortic syndromes. Sci Rep. 2020;10:3103. PMID 32080259
  10. Wren J, Goodacre S, Pandor A, et al. Diagnostic accuracy of alternative biomarkers for acute aortic syndrome: a systematic review. Emerg Med J. 2024;41:678-685. PMID 39107052
  11. Ikonomidis JS, Ivey CR, Wheeler JB, et al. Plasma biomarkers for distinguishing etiologic subtypes of thoracic aortic aneurysm disease. J Thorac Cardiovasc Surg. 2013;145:1326-1333. PMID 23312977
  12. Li T, Jiang B, Li X, et al. Serum matrix metalloproteinase-9 is a valuable biomarker for identification of abdominal and thoracic aortic aneurysm: a case-control study. BMC Cardiovasc Disord. 2018;18:202. PMID 30373522
  13. Matt P, Schoenhoff F, Habashi J, et al. Circulating transforming growth factor-beta in Marfan syndrome. Circulation. 2009;120:526-532. PMID 19635970
  14. Suzuki T, Trimarchi S, Sawaki D, et al. Circulating transforming growth factor-beta levels in acute aortic dissection. J Am Coll Cardiol. 2011;58:775. PMID 21816317 (no abstract on PubMed)
  15. Boileau A, Lino Cardenas CL, Courtois A, et al. MiR-574-5p: A Circulating Marker of Thoracic Aortic Aneurysm. Int J Mol Sci. 2019;20:3924. PMID 31409059
  16. Ekedi AVNB, Rozhkov AN, Shchekochikhin DY, et al. Evaluation of microRNA Expression Features in Patients with Various Types of Arterial Damage: Thoracic Aortic Aneurysm and Coronary Atherosclerosis. J Pers Med. 2023;13:1161. PMID 37511774
  17. Doukas P, Bassett C, Hruschka B, et al. Plasma Desmosine Is Elevated in Thoracoabdominal Aortic Aneurysms and Is Associated with Intramural Proteolytic Activity. Int J Mol Sci. 2026;27:1236. PMID 41683665
  18. Mordi IR, Forsythe RO, Gellatly C, et al. Plasma Desmosine and Abdominal Aortic Aneurysm Disease. J Am Heart Assoc. 2019;8:e013743. PMID 31595818
  19. Kuzmanova E, McIntyre A, Syed M, et al. Accelerating isotope dilution LC-MS-based desmosine quantification for estimating elastin turnover. Bioanalysis. 2025;17:99-104. PMID 39891463
  20. Gombert A, Mohapatra SR, Frankort JM, et al. Desmosine in Aortic Disease: Biology, Measurement, and Clinical Applications in Aortic Pathologies. J Clin Med. 2026;15:2540. PMID 41976840
  21. Gu G, Wan F, Xue Y, et al. Lumican as a novel potential clinical indicator for acute aortic dissection: A comparative study, based on multi-slice computed tomography angiography. Exp Ther Med. 2016;11:923-928. PMID 26998013
  22. Chen SW, Chou SH, Tung YC, et al. Expression and role of lumican in acute aortic dissection: A human and mouse study. PLoS One. 2021;16:e0255238. PMID 34310653
  23. Hsu ME, Cheng YT, Chang CH, et al. Level of serum soluble lumican and risks of perioperative complications in patients receiving aortic surgery. PLoS One. 2021;16:e0247340. PMID 33661915
  24. Harrison OJ, Cagampang F, Ohri SK, et al. Candidate plasma biomarkers for predicting ascending aortic aneurysm in bicuspid aortic valve disease. J Cardiothorac Surg. 2018;13:76. PMID 29929532
  25. Dawson A, Li Y, Li Y, et al. Single-Cell Analysis of Aneurysmal Aortic Tissue in Patients with Marfan Syndrome Reveals Dysfunctional TGF-β Signaling. Genes (Basel). 2022;13:95. PMID 35052435
  26. Regalado ES, Mellor-Crummey L, De Backer J, et al. Clinical history and management recommendations of the smooth muscle dysfunction syndrome due to ACTA2 arginine 179 alterations. Genet Med. 2018;20:1206-1215. PMID 29300374
  27. Shalhub S, Regalado ES, Guo DC, Milewicz DM. The natural history of type B aortic dissection in patients with PRKG1 mutation c.530G>A (p.Arg177Gln). J Vasc Surg. 2019;70:718-723. PMID 30871887
  28. Pirruccello JP, Khurshid S, Lin H, et al. AORTA Gene: Polygenic prediction improves detection of thoracic aortic aneurysm. medRxiv. 2023. PMID 37662232 (still preprint-only as of 2026-08-28)
  29. Pirruccello JP, Chaffin MD, Chou EL, et al. Deep learning enables genetic analysis of the human thoracic aorta. Nat Genet. 2022;54:40-51. PMID 34837083
  30. Pirruccello JP, Lin H, Khurshid S, et al. Development of a Prediction Model for Ascending Aortic Diameter Among Asymptomatic Individuals. JAMA. 2022;328:1935-1944. PMID 36378208
  31. Fletcher AJ, Nash J, Syed MBJ, et al. Microcalcification and Thoracic Aortopathy: A Window Into Disease Severity. Arterioscler Thromb Vasc Biol. 2022;42:1048-1059. PMID 35770666
  32. Debono S, Nash J, Fletcher AJ, et al. Aortic sodium [18F]fluoride uptake following endovascular aneurysm repair. Heart. 2023;109:1677-1682. PMID 37164479
  33. Espitia O, Toquet C, Jamet B, et al. [Aortitis]. Rev Med Interne. 2024;45:767-775. PMID 39034261
  34. Hasse B, Ledergerber B, Van Hemelrijck M, et al. 18F-fluorodeoxyglucose uptake patterns in positron emission tomography/computed tomography caused by inflammation and/or infection after graft surgery for thoracic aortic dissection. J Nucl Cardiol. 2024;36:101865. PMID 38679286
  35. Guala A, Evangelista A, Teixido-Tura G, et al. Leaflet fusion length is associated with aortic dilation and flow alterations in non-dysfunctional bicuspid aortic valve. Eur Radiol. 2021;31:9262-9272. PMID 33977309
  36. Zhu Y, Armour C, Li B, et al. A combined 4D flow MR imaging and fluid-structure interaction analysis of ascending thoracic aortic aneurysms. Biomech Model Mechanobiol. 2025;24:829-844. PMID 40067580
  37. Stephens SB, Shalhub S, Dodd N, et al. Vertebral Tortuosity Is Associated With Increased Rate of Cardiovascular Events in Vascular Ehlers-Danlos Syndrome. J Am Heart Assoc. 2023;12:e029518. PMID 37776192
  38. Pitcher A, Spata E, Emberson J, et al. Angiotensin receptor blockers and β blockers in Marfan syndrome: an individual patient data meta-analysis of randomised trials. Lancet. 2022;400:822-831. PMID 36049495