Basic Information

Stress Hyperglycemia Ratio and Atrial Thrombus After Stroke: Thrombotic Signal or Stress Response?

Ünsal Vural

University of Health Sciences, Istanbul, Türkiye

Correspondence: unsalvural@gmail.com

Critical narrative review

Abstract

The stress hyperglycemia ratio (SHR) expresses acute glucose relative to estimated background glycemia. Its association with atrial thrombus after ischemic stroke raises a clinically relevant question. Does it identify a thrombogenic atrial substrate, reflect the response to cerebral injury, or capture both? This critical narrative review compares human studies of glycemic disturbance, inflammatory biomarkers and thrombus-related outcomes. Anatomically detected thrombus, composite echocardiographic findings, coronary thrombus burden and laboratory thrombogenicity are evaluated separately. The identified atrial SHR evidence is observational. Coronary studies provide supporting context but examine a different vascular process. C-reactive protein (CRP) and related inflammatory indices also show associations with atrial thrombus, although discrimination and adjustment vary. Experimental hypoglycemia studies demonstrate changes in platelet activity, fibrin structure and ex vivo thrombus formation. They do not establish hypoglycemia-induced atrial thrombosis. Interpretation is further limited by measurement timing, selected imaging populations, possible cohort overlap and imperfect thrombus classification. Neither an adjusted association nor a high multivariable-model area under the curve establishes a causal pathway or a safe imaging rule. Future studies should test whether SHR adds clinically useful information beyond atrial structure, rhythm, inflammation and stroke severity. Current evidence supports mechanistic investigation and prospective validation rather than biomarker-directed anticoagulation or omission of indicated imaging.

Keywords: stress hyperglycemia ratio; hypoglycemia; atrial thrombus; C-reactive protein; inflammation; transesophageal echocardiography; reverse causation; narrative review

1 Introduction

A negative transthoracic echocardiogram does not exclude left atrial (LA) or left atrial appendage (LAA) thrombus. Transesophageal echocardiography (TEE) can identify a cardiac source missed by transthoracic echocardiography (TTE). Such findings can influence secondary stroke prevention. The clinical challenge is deciding who needs further imaging after an unrevealing initial assessment. [1,2]

SHR has been proposed as an accessible marker of occult intracardiac thrombus after ischemic stroke. [3] This proposal requires three separate judgments. The first concerns biological plausibility. The second concerns diagnostic information beyond existing clinical assessment. The third concerns whether using the marker improves decisions or outcomes. Evidence for one does not automatically establish the others.

This review examines those distinctions across hyperglycemia, hypoglycemia and inflammation. The central question is whether an acute metabolic measurement identifies a pre-existing atrial source or mainly reflects the consequences of an embolic event. The assessment also considers how investigators define thrombus and how directly each study addresses atrial thrombosis.

2 Search approach and appraisal of evidence

A targeted narrative literature search was completed on 25 September 2026. Web-based searches were used to locate PubMed-indexed records, primary journal reports and accessible author manuscripts. Search combinations linked stress hyperglycemia ratio, hyperglycemia, HbA1c or hypoglycemia with thrombus, thrombosis, left atrium, left atrial appendage, platelet activation and fibrinolysis. Additional searches combined C-reactive protein, CRP, CRP-to-albumin ratio, neutrophil-to-lymphocyte ratio or platelet-to-lymphocyte ratio with atrial thrombus. Reference lists and related correspondence were checked for relevant primary studies.

Priority was given to human studies with a defined exposure, sample size and ascertainable outcome. Anatomically detected atrial thrombus was considered the most directly relevant endpoint. Coronary thrombus studies were retained as evidence from another vascular setting. Controlled experiments were included when they measured hemostatic or inflammatory responses to glycemic change. Studies of stroke prognosis alone were used for context. They were not treated as evidence that an atrial clot was present.

Tables 1 and 2 present selected studies rather than an exhaustive systematic inventory. Data were checked against primary full texts where accessible and indexed primary abstracts otherwise. Study design, participant and event counts, exposure timing, outcome definition, adjustment and validation informed the appraisal. Causal confidence and directness were assessed separately. The evidence descriptions are qualitative judgments, not formal GRADE ratings. No pooled estimate was calculated because populations, exposures and endpoints were not sufficiently comparable. This approach remains vulnerable to incomplete retrieval and publication bias.

ChatGPT (OpenAI) assisted with literature-search planning, manuscript drafting, organization, and language revision.

3 The atrial substrate remains central

Atrial thrombus develops where blood stasis, atrial tissue abnormalities and prothrombotic conditions intersect. Atrial fibrillation (AF) is a major contributor, but a sinus-rhythm tracing does not exclude intermittent arrhythmia or atrial cardiomyopathy. Enlargement, fibrosis, impaired appendage emptying, mitral disease and ventricular dysfunction can alter the local thrombotic environment. [4,5,6]

These factors are related but are not interchangeable. Atrial fibrosis can coexist with rhythm abnormalities. Mitral stenosis can promote atrial enlargement and slow flow. Reduced appendage velocity and spontaneous echo contrast (SEC) provide information about stasis. However, these features are usually characterized during TEE. They cannot ordinarily select patients for the examination that first reveals them. [1,5]

Anatomy and rhythm should anchor interpretation before glucose. A circulating biomarker may reflect systemic stress without identifying the anatomical location of a clot. Conversely, a local thrombogenic substrate may remain present when a blood marker is normal. Evaluation of a metabolic marker therefore requires comparison with the structural and rhythm information already available.

4 What glycemic measurements can establish

4.1 Relative hyperglycemia and measurement context

SHR compares acute glucose with estimated usual glycemia derived from glycated hemoglobin (HbA1c). A common formulation divides admission glucose by estimated average glucose, using matching units. In mg/dL, estimated average glucose is 28.7 × HbA1c (%) − 46.7. [7] SHR describes relative glycemic elevation. It does not directly measure catecholamines, endothelial injury or thrombogenicity.

Timing matters. Admission glucose, fasting glucose and later inpatient measurements are different exposures. Infection, nutrition, insulin and other treatments can alter the numerator. Hemolysis, blood loss, transfusion or altered erythrocyte turnover can distort HbA1c and therefore the denominator. A falsely low HbA1c can increase SHR without a corresponding rise in acute glucose. These limitations are particularly relevant in surgical and critically ill patients. [8,9]

An SHR below 1 does not itself diagnose hypoglycemia. It indicates glucose below the estimated average used in the denominator. Likewise, a higher mean SHR in one group does not establish absolute hyperglycemia in every participant. Both components should be reported. Thresholds derived from different sampling schedules or formulations should not be assumed equivalent.

4.2 Biological plausibility and temporal direction

Controlled experiments link acute hyperglycemia with endothelial and prothrombotic changes under specified insulin conditions. These findings provide biological plausibility for a relationship with thrombosis. [10] Biological plausibility, however, does not establish the direction of causation.

Metabolic disturbance might contribute to a prothrombotic environment before clot formation. Alternatively, an atrial thrombus might embolize first. Cerebral injury and its neuroendocrine and inflammatory responses could then raise glucose. Shared factors could also influence both measurements. A single blood sample after stroke cannot distinguish these pathways.

Stroke severity, infarct characteristics, sampling delay, infection and early treatment therefore deserve explicit consideration. Adjustment for diabetes alone does not capture these processes. Nor does adjustment for one inflammatory marker fully characterize the inflammatory response. Serial measurements would help separate persistent vulnerability from a transient consequence of illness.

5 Hyperglycemia and imaging findings

5.1 Direct atrial evidence

Cicek and colleagues studied 486 selected patients undergoing TEE after negative TTE. TEE identified atrial or appendage thrombus in 64 patients. The adjusted association with SHR persisted, but the study was retrospective and excluded previous anticoagulant use. Its area under the receiver-operating-characteristic curve (AUC) of 0.796 described the combined model, not SHR alone. The main quantitative results are summarized in Table 1. [3]

Subsequent correspondence discussed inflammation and the atrial substrate. [11] In their reply, the authors emphasized that SHR and CRP retained associations in the same model. They also acknowledged unavailable detailed measures of atrial cardiomyopathy and the need for external validation. [12] Mutual adjustment supports conditional statistical associations. It does not prove separate causal mechanisms or eliminate reverse causation.

A larger sample does not necessarily provide more direct evidence. Song and colleagues analyzed 1,217 patients with nonvalvular atrial fibrillation. Their 112 positive outcomes comprised 28 thrombi, four sludge findings and 80 SEC findings. HbA1c and CRP were important model features. The reported AUC of 0.97 belonged to a multivariable machine-learning model. A thrombus-only sensitivity analysis was described, but separate discrimination estimates were not given in that section. [13] The overall AUC cannot be attributed to either biomarker or assumed to apply to discrete thrombus alone.

Combining SEC, sludge and thrombus increases the number of positive outcomes but changes the research question. Such a model identifies an echocardiographic risk milieu. It does not necessarily identify an established clot. Feature importance also describes the behavior of a fitted model; it is not a causal effect estimate.

5.2 Coronary evidence and limits of transfer

Coronary studies provide a broader test of metabolic associations with thrombosis. Chu and colleagues examined 227 patients with diabetes and ST-elevation myocardial infarction. Algül and colleagues studied 1,222 patients with acute coronary syndrome. Both reported associations between SHR and angiographic thrombus burden. [14,15] A post hoc analysis of the prospective CorLipid cohort also linked stress-induced hyperglycemia with large coronary thrombus burden. [16]

These observations support a thrombotic context for acute dysglycemia. They do not establish a common mechanism across coronary and atrial disease. Coronary thrombosis involves plaque-related injury and arterial flow. Atrial thrombosis often develops in a setting of stasis and atrial dysfunction. In the acute coronary and stroke cohorts discussed here, glucose was measured after the presenting event. This leaves temporal ambiguity.

Cutoffs from acute coronary cohorts should therefore not be transferred to stroke-related TEE selection. Odds ratios based on categorical SHR, continuous SHR or absolute fasting glucose also represent different contrasts. Pooling them without a prespecified harmonization strategy would obscure their meaning.

Studies of recurrent stroke or functional outcome address another question. Roberts and colleagues evaluated glycemic measures in 300 stroke patients. An INSPIRES analysis studied 4,515 patients with mild stroke or high-risk transient ischemic attack. [17,18] These studies inform prognosis. They do not verify an atrial source or validate a rule for excluding thrombus.

6 Inflammation as a comparator and shared pathway

CRP provides an important comparator because it can accompany acute tissue injury, infection and chronic inflammatory burden. Its association with thrombus may represent a shared disease process, a response to injury or residual confounding. The same temporal questions raised for SHR therefore apply to CRP.

Maehama and colleagues found 19 atrial thrombi among 190 patients with non-rheumatic atrial fibrillation. CRP was independently associated with thrombus. Nevertheless, the proposed cutoff had a positive predictive value of only 19%, despite a negative predictive value of 97%. [19] A high negative predictive value in a selected, relatively low-prevalence cohort is not sufficient to establish a safe replacement for imaging.

In rheumatic mitral stenosis, Belen and colleagues linked CRP and the platelet-to-lymphocyte ratio with atrial thrombus. Glucose did not differ significantly between groups. [20] This is a useful negative comparator, but it does not refute SHR in stroke. The exposure, anatomical substrate and clinical setting differ.

Inflammatory ratios also require careful interpretation. Cicek and colleagues evaluated CRP-to-albumin ratio (CAR) in 303 selected stroke/TIA patients. Thirty-four had atrial thrombus. Admission glucose did not differ significantly between groups (p = 0.888). [21] This finding does not test relative hyperglycemia. The CAR and SHR reports cover overlapping recruitment periods and cite the same ethics approval. Patient overlap is possible but unconfirmed. They should not be counted as independent replication without clarification. [3]

Zhou and colleagues studied 623 patients with nonvalvular atrial fibrillation, including 59 with atrial thrombus. The neutrophil-to-lymphocyte ratio (NLR) retained an adjusted association, but discrimination was modest. Other indices did not consistently retain significance after fuller adjustment. [22] These results illustrate why a collection of significant univariable markers is not equivalent to a useful diagnostic panel.

Ratios can also obscure the source of an association. A high CAR may reflect higher CRP, lower albumin, or both. The same principle applies to SHR. An informative analysis should compare the ratio with its individual components. This would clarify whether the ratio adds information or mainly repackages an existing signal.

A combined metabolic and inflammatory model is plausible. Its value must be demonstrated by direct comparison with a clinical model, followed by validation. Entering SHR and CRP into one regression does not establish complementary clinical usefulness. Correlation, measurement error and shared responses to acute illness must also be considered.

 

Table 1. Human imaging studies of glycemic and inflammatory markers in relation to thrombus

Study	Design and sample	Exposure and outcome	Principal finding	Evidence and main limitations
Cicek et al., 2026 [3]	Retrospective, single center. Selected stroke/TIA patients after negative TTE. n = 486; 64 atrial thrombi.	Admission SHR. TEE-defined LA/LAA thrombus.	aOR 2.393 (1.107–5.172). AUC 0.796 is for the combined model.	Direct atrial endpoint; limited causal evidence. Post-event sampling; prior anticoagulation excluded; incomplete atrial characterization; no external validation.
Song et al., 2026 [13]	Retrospective prediction study. NVAF, n = 1,217. Composite-positive n = 112.	HbA1c and CRP among model features. Composite: 28 thrombi, 4 sludge and 80 SEC findings.	Both markers ranked among leading SHAP features. AUC 0.97 is for the multivariable random-forest model.	Mixed anatomical/surrogate endpoint; exploratory prediction. Only 28 discrete thrombi. Internal testing; incomplete anticoagulation data. Feature importance is not a causal effect.
Cicek et al., 2024 [21]	Retrospective, single center. Stroke/TIA without known AF. n = 303; 34 LA thrombi.	Admission CAR and other inflammatory indices. TEE within 10 days.	Reported CAR aOR 2.70 (1.39–5.25); AUC 0.749. Admission glucose comparison: p = 0.888.	Direct atrial endpoint; limited causal evidence. Post-event sampling; possible overlap with SHR cohort. Inconsistent model terminology and ratio units; no external validation.
Maehama et al., 2010 [19]	Observational TEE-selected series. Non-rheumatic AF. n = 190; 19 LA thrombi.	CRP within one week before TEE. Discrete LA thrombus.	Adjusted association p = 0.03. At CRP 2.1 mg/L: sensitivity 84%, specificity 60%, PPV 19%, NPV 97%.	Direct atrial endpoint; small association/diagnostic series. Few events and low rule-in value. Predictive values depend on prevalence.
Belen et al., 2016 [20]	Prospective collection, cross-sectional analysis. Untreated rheumatic mitral stenosis. n = 351; 92 LA thrombi.	CRP and PLR; blood within 12 hours of TTE/TEE. Discrete LA/LAA thrombus.	CRP aOR 1.90 (1.40–2.60); PLR aOR 1.03 (1.00–1.06). Glucose comparison: p = 0.170.	Direct atrial endpoint; limited causal evidence. Distinct rheumatic substrate; highly selected untreated population. Concurrent measurements; no external validation.

 

 

Table 1. Human imaging studies of glycemic and inflammatory markers in relation to thrombus (continued)

Study	Design and sample	Exposure and outcome	Principal finding	Evidence and main limitations
Zhou et al., 2026 [22]	Retrospective cross-sectional study. Pre-ablation NVAF. n = 623; 59 LA thrombi.	CBC-derived inflammatory indices. Definite thrombus on TEE.	NLR aOR 2.113 (1.087–4.108) per natural-log unit; AUC 0.601. Continuous SII was nonsignificant after full adjustment.	Direct atrial endpoint; limited causal evidence. Weak stand-alone discrimination; correlated markers; treatment and referral selection; no external validation.
Chu et al., 2020 [14]	Retrospective analysis of prospectively enrolled patients. Diabetic STEMI, n = 227; 77 large coronary thrombi.	SHR ≥1.19 versus <1.19. Reclassified angiographic TIMI thrombus grade 4 or 5.	Model 2 aOR 4.857 (2.304–10.236). SHR AUC 0.669.	Direct coronary endpoint; indirect for atrial thrombosis. Data-derived cutoff; selected diabetes cohort; temporal ambiguity; no external validation.
Algül et al., 2024 [15]	Cross-sectional ACS study. n = 1,222; 451 with high coronary thrombus burden.	SHR. High versus low coronary thrombus burden.	Reported aOR 1.328 (1.082–1.752).	Coronary association; indirect for atrial thrombosis. Abstract-level verification only: exposure scaling and adjustment set could not be independently inspected. Reverse causation remains possible.
Stalikas et al., 2022 [16]	Post hoc analysis of prospective CorLipid cohort. STEMI, n = 309; 135 large coronary thrombi.	Admission glucose >140 mg/dL after ≥8 hours fasting. Large angiographic coronary thrombus.	aOR 2.171 (1.270–3.709). Events: 68/121 with hyperglycemia versus 67/188 without.	Direct coronary endpoint; indirect for atrial thrombosis. Selected fasting early presenters; post hoc analysis; no randomized glucose intervention.

Values in parentheses after odds ratios are 95% confidence intervals. Effects retain each study’s exposure scale; their magnitudes are not directly comparable. Counts are participants with the stated outcome, not necessarily new incident thrombi. No pooled sample size or effect is presented.

Evidence descriptions are qualitative appraisals of design and endpoint directness, not formal GRADE ratings. All studies in this table are observational. Prospective collection does not by itself establish temporal causation.

CAR2024 uses Cox/HR terminology in its methods but logistic/OR terminology in its results. Ratio units are not consistently specified. The reported OR is retained without endorsing a transferable cutoff. Its recruitment period overlaps that of SHR2026; shared participants have not been confirmed.

ACS, acute coronary syndrome; AF, atrial fibrillation; aOR, adjusted odds ratio; AUC, area under the receiver-operating-characteristic curve; CAR, CRP-to-albumin ratio; CBC, complete blood count; CRP, C-reactive protein; HbA1c, glycated hemoglobin; LA/LAA, left atrium/left atrial appendage; NLR, neutrophil-to-lymphocyte ratio; NPV/PPV, negative/positive predictive value; NVAF, nonvalvular AF; PLR, platelet-to-lymphocyte ratio; SEC, spontaneous echo contrast; SHAP, Shapley additive explanations; SHR, stress hyperglycemia ratio; SII, systemic immune-inflammation index; STEMI, ST-elevation myocardial infarction; TEE/TTE, transesophageal/transthoracic echocardiography; TIA, transient ischemic attack; TIMI, Thrombolysis in Myocardial Infarction.

7 Hypoglycemia and experimental thrombogenicity

Hypoglycemia is relevant because prothrombotic responses are not confined to high glucose. Human clamp experiments can establish the timing of a physiological response more clearly than a post-event observational sample. However, their outcomes usually concern platelets, coagulation proteins or clot properties. They do not directly demonstrate an atrial thrombus.

Gogitidze Joy and colleagues compared euglycemic and hypoglycemic conditions in healthy volunteers and people with type 1 diabetes. Hypoglycemia increased several inflammatory and prothrombotic markers. [23] A later experiment compared glycemic and insulin conditions in healthy adults. It showed why insulin exposure and glucose concentration must be interpreted together. [10] These studies support mechanisms, not estimates of clinical atrial thrombus risk.

Wright and colleagues used a randomized, counterbalanced design in 32 adults. They observed inflammatory and platelet–monocyte responses, with a rise in high-sensitivity CRP in the nondiabetic group. [24] This links hypoglycemia to inflammatory signaling. It does not show that CRP mediates subsequent atrial thrombosis.

Chow and colleagues studied 12 participants with type 2 diabetes and 11 controls. Platelet responses occurred acutely, while adverse fibrin properties in diabetes persisted after recovery. Their paired clamp studies used a fixed sequence, with euglycemia preceding hypoglycemia. [25] The design strengthens temporal interpretation of laboratory changes but remains small and susceptible to order effects.

Yamamoto and colleagues assessed thrombus formation using a microchip flow chamber. One component examined ten patients with diabetes before and after comprehensive care. Another examined ten patients undergoing an insulin tolerance test without a concurrent euglycemic control. [26] The terminology requires care: a flow-chamber thrombogenic response is an ex vivo measurement, not a clot detected inside the left atrium.

The response is not uniform across settings. Hagelqvist and colleagues randomized the order of exercise-related and resting hypoglycemia in 15 men with type 1 diabetes. Exercise-related hypoglycemia increased clot strength and reduced fibrinolysis relative to baseline. Resting hypoglycemia increased fibrinolysis. [27] Exercise, insulin exposure and diabetes phenotype therefore matter when interpreting the net hemostatic response.

Table 2 separates these experiments from the clinical imaging studies. Their findings argue against treating glucose as a simple one-directional surrogate for thrombosis. They do not establish a U-shaped relationship between glucose and atrial thrombus. A controlled human study demonstrating hypoglycemia-induced, imaging-confirmed LA/LAA thrombosis was not identified in this targeted search.

The mechanistic evidence also limits therapeutic inference. If elevated SHR accompanies thrombus, it does not follow that aggressively lowering glucose will remove the clot or improve stroke recovery. SHINE randomized 1,151 patients with acute ischemic stroke and hyperglycemia. Intensive control did not improve the primary functional outcome and caused severe hypoglycemia in the intensive group. [28] The trial did not test SHR-guided treatment or atrial thrombus resolution.

 

Table 2. Human experimental studies of hypoglycemia, inflammation and thrombogenicity

Study	Design and participants	Glycemic exposure	Measured response	Evidence and main limitations
Gogitidze Joy et al., 2010 [23]	Controlled, partly paired clamp study. 59 unique participants: 35 healthy and 24 with T1D. Seventeen completed both arms.	Two-hour hypoglycemia at 2.9 mmol/L versus euglycemia at 5.2 mmol/L, with matched insulin.	Higher PAI-1, P-selectin and inflammatory/endothelial markers during hypoglycemia.	Controlled mechanistic evidence; indirect for atrial thrombosis. Only the crossover subset had randomized order. Partly unmatched groups; biomarkers rather than clinical clots.
Wright et al., 2010 [24]	Randomized, counterbalanced crossover. n = 32: 16 with T1D and 16 nondiabetic controls.	Sixty minutes at 2.5 versus 4.5 mmol/L. Follow-up through 24 hours.	Platelet–monocyte aggregation and CD40-related responses increased. hsCRP rose in nondiabetic participants; vWF/tPA increases were nonsignificant.	Randomized mechanistic evidence; indirect for atrial thrombosis. Small selected groups; multiple surrogate outcomes. Some contrasts reflected reductions during euglycemic insulin infusion.
Joy et al., 2016 [10]	Comparative glucose-clamp protocols. n = 45 healthy adults.	Hyperglycemia 11.1 mmol/L under basal/elevated insulin; hyperinsulinemic euglycemia 5.1 and hypoglycemia 2.9 mmol/L.	Euinsulinemic hyperglycemia and hypoglycemia increased endothelial, platelet and inflammatory-marker responses relative to hyperinsulinemic comparison conditions.	Controlled mechanistic evidence; indirect for atrial thrombosis. Insulin is an important co-exposure. No cardiac imaging or clinical thrombosis endpoint.

 

 

Table 2. Human experimental studies of hypoglycemia, inflammation and thrombogenicity (continued)

Study	Design and participants	Glycemic exposure	Measured response	Evidence and main limitations
Chow et al., 2018 [25]	Paired fixed-sequence crossover. n = 23: 12 with T2D and 11 matched controls. Euglycemia first; hypoglycemia 4–8 weeks later.	Two 60-minute periods at 2.5 versus 6 mmol/L, with matched insulin. Follow-up through day 7.	Acute platelet activation. In T2D, denser fibrin and prolonged clot lysis persisted through day 7.	Controlled mechanistic evidence; indirect for atrial thrombosis. Nonrandomized order; small sample without known cardiovascular disease; ex vivo clot properties.
Yamamoto et al., 2019 [26]	Two uncontrolled before–after studies: 10 T2D patients receiving comprehensive care and 10 nondiabetic patients undergoing pituitary testing.	Insulin-tolerance component: glucose fell from 5.2 to 1.7 mmol/L at 45 minutes.	Platelet-dependent thrombus formation accelerated in the microchip assay during hypoglycemia. PL-T10 fell from 156.4 to 109.7 seconds.	Exploratory ex vivo evidence; indirect for atrial thrombosis. No concurrent euglycemic control. Endocrine comorbidity and catecholamine/hematocrit changes; no in vivo clot.
Hagelqvist et al., 2024 [27]	Randomized crossover in 15 men with T1D. Another 15 healthy men contributed baseline comparisons only.	Hypoglycemia during exercise versus at rest. Measurements through 24-hour recovery.	Exercise-related hypoglycemia: clot strength +2.77 mm and LY-30 −0.45 percentage point versus baseline. Resting hypoglycemia increased fibrinolysis.	Randomized mechanistic comparison; indirect for atrial thrombosis. Exercise is a co-exposure; no euglycemic-exercise control. Young men only; no imaged clot.

None of these experiments used imaging-confirmed atrial thrombus as its outcome. Randomization strengthens interpretation of the measured physiological response; it does not convert a surrogate endpoint into direct evidence of clinical thrombosis. Participant totals are not pooled because overlap across laboratory reports has not been excluded.

For Hagelqvist et al., the 95% confidence intervals were 2.04–3.51 mm for the change in clot strength and −0.60 to −0.29 percentage point for LY-30. The randomized experimental sample was 15, not 30.

hsCRP, high-sensitivity C-reactive protein; LY-30, clot lysis 30 minutes after maximum amplitude; PAI-1, plasminogen activator inhibitor-1; PL-T10, time to a 10-kPa pressure rise in the platelet microchip assay; T1D/T2D, type 1/type 2 diabetes; tPA, tissue plasminogen activator; vWF, von Willebrand factor.

8 Reliability of the thrombus endpoint

TEE is highly useful, but an echodensity interpreted as thrombus is not identical to surgical confirmation. In Manning and colleagues’ prospective intraoperative study, surgery confirmed 12 of 14 TEE-positive findings among 231 patients. Two positive findings were not confirmed. Both had been reported by only one observer. [29]

The denominator matters. The unconfirmed proportion among positive TEE findings was 2/14, or 14.3%. The conventional false-positive rate among surgically thrombus-negative patients was 2/219, or 0.9%. These are different quantities. Neither is a universal estimate for contemporary stroke investigations.

Kaymaz and colleagues evaluated 474 surgical patients with rheumatic mitral disease. TEE sensitivity and specificity for appendage thrombi were both 98%; performance differed for thrombi in the main atrial cavity. [30] A recent case report also documents an apparent appendage thrombus that was not found during surgery. [31] A case report establishes possibility, not frequency.

Pectinate muscles, anatomical ridges, reverberation and dense SEC can complicate interpretation. [1,31] When imaging and surgery are separated in time, interval dissolution or embolization can also explain disagreement. Non-confirmation should not automatically be equated with an initial diagnostic error.

For biomarker research, blinded image adjudication and explicit definitions are therefore essential. Definite thrombus, sludge, SEC and an equivocal mass should be recorded separately. Historical surgical percentages should not be applied as correction factors to a different cohort. The concern is outcome reliability, not an assumption that misclassification occurred in the SHR study.

9 Integrated critical appraisal and clinical usefulness

9.1 Causal interpretation and the missing temporal sequence

The evidence contains a consistent gap. Imaging cohorts identify thrombus but usually cannot establish whether dysglycemia preceded it. Clamp studies establish exposure timing but do not demonstrate clinical atrial thrombosis. Combining these designs improves biological interpretation. It does not close the causal gap between a transient metabolic disturbance and formation of an atrial clot.

An atrial thrombus detected after stroke may be residual evidence of the embolic source. A glucose measurement obtained later may instead reflect injury severity. This is a plausible competing explanation, not proof that the reported association is entirely noncausal. Future studies need measurements before the event or repeated measurements linked to serial imaging to distinguish these possibilities.

Inflammation should not be treated as a competing explanation that automatically displaces glucose. It could precede both dysglycemia and thrombosis, accompany the same stress response, or lie on a causal pathway. Each possibility changes the meaning of statistical adjustment. A prespecified causal model is therefore preferable to selecting adjustment variables solely because their univariable p values are significant.

9.2 Prediction requires more than statistical association

The appropriate clinical question is incremental value. A baseline model should use information available when the imaging decision is made. This includes rhythm history and monitoring, mitral disease or prosthesis, available TTE findings, ventricular function and stroke characteristics. Variables first obtained from the proposed TEE cannot legitimately justify selecting patients for that same test.

SHR should be compared with admission glucose, HbA1c and inflammatory markers as alternative additions to the same clinical model. Model discrimination, calibration and uncertainty should be reported. A combined-model AUC cannot be assigned to one component. Internal validation also does not establish performance in a new institution or a different stroke population. [32]

The number of events matters as much as the total sample. Sixty-four thrombi provide less modeling information than 486 participants might suggest. Composite outcomes may increase statistical precision while reducing clinical specificity. Models with many candidate predictors require protection against overfitting. Any imputation, feature selection or resampling should remain within the training process.

9.3 Selection, treatment and transportability

Selection for TEE is another important limitation. Clinicians may refer patients because of rhythm abnormalities, embolic patterns or other risk indicators. Analyses restricted to those patients may not generalize to all stroke presentations. Exclusion of anticoagulated patients creates a further limitation when applying results to routine clinical practice.

Anticoagulant exposure requires more detail than a yes/no variable. Timing, adherence and treatment adequacy can affect the probability of detecting residual thrombus. Short rhythm monitoring can also miss intermittent AF. Residual confounding by atrial disease may persist even after adjustment for recorded AF and atrial diameter.

9.4 Clinical decisions and the limits of intervention

A proposed imaging rule should state how many examinations it avoids and how many actionable thrombi it misses. Decision-curve analysis can assess expected net benefit across clinically plausible thresholds. [33] Ultimately, a prospective impact study should test whether using the rule improves decisions or outcomes. An association is not yet a TEE-selection policy.

A low SHR should not override an established indication for imaging. A high SHR or CRP alone does not establish an atrial source or justify anticoagulation. Equivocal imaging requires assessment of anatomy and clinical context. These biomarkers remain candidates for evaluation within a broader diagnostic strategy.

10 Research priorities and conclusions

Future cohorts should recruit patients at the point when TEE is being considered. Sampling times, stroke severity, infection, treatment and anticoagulant exposure should be documented prospectively. Glucose trajectories and HbA1c reliability should be assessed. Rhythm burden and atrial functional measures would help characterize the underlying substrate.

The primary imaging endpoint should be definite LA/LAA thrombus. SEC, sludge and uncertain findings should remain distinct secondary outcomes. Imaging adjudication should be blinded to biomarker results. Analyses should compare the same baseline model with and without SHR, glucose and inflammatory markers. External validation should precede implementation.

The current evidence supports an association between metabolic or inflammatory disturbance and thrombus-related findings in selected populations. It does not establish that an elevated SHR caused an atrial clot. Hypoglycemia experiments strengthen the biological link between glycemic disturbance and hemostasis, but remain indirect for atrial thrombosis. The decisive question is whether a marker adds reliable information that improves an individual patient’s imaging decision and subsequent care.

Acknowledgments

The author used ChatGPT (OpenAI) for assistance with literature-search planning, manuscript drafting, organization, and language revision. The author reviewed and edited the content and takes full responsibility for the manuscript.

Institutional Review Board Statement

Not applicable. This article is a literature-based review and does not report a new study involving human participants or animals.

Informed Consent Statement

Not applicable. This review does not report new data from individual participants.

Data Availability Statement

No new datasets were generated or analyzed for this review.

Funding

No financial support was received for this work.

Author Contributions

Ünsal Vural: Conceptualization, critical interpretation of the literature, literature review, drafting, revision, and final approval of the manuscript.

Conflicts of Interest

The author declares no conflict of interest.

 

 

References

1. Saric M, Armour AC, Arnaout MS, Chaudhry FA, Grimm RA, Kronzon I, et al. Guidelines for the use of echocardiography in the evaluation of a cardiac source of embolism. J Am Soc Echocardiogr. 2016;29:1–42. doi:10.1016/j.echo.2015.09.011.
2. Kleindorfer DO, Towfighi A, Chaturvedi S, Cockroft KM, Gutierrez J, Lombardi-Hill D, et al. 2021 guideline for the prevention of stroke in patients with stroke and transient ischemic attack. Stroke. 2021;52:e364–e467. doi:10.1161/STR.0000000000000375.
3. Cicek V, Erdem A, Kozan Çıkırıkçı EH, Yılmaz İ, Günhan E, Uygun E, et al. Stress hyperglycemia ratio is associated with intracardiac thrombus in acute ischemic stroke. Turk Kardiyol Dern Ars. 2026;54(5):395–402. doi:10.5543/tkda.2026.50611.
4. Kamel H, Healey JS. Cardioembolic stroke. Circ Res. 2017;120:514–526. doi:10.1161/CIRCRESAHA.116.308407.
5. Akoum N, Fernandez G, Wilson B, McGann C, Kholmovski E, Marrouche N. Association of atrial fibrosis quantified using LGE-MRI with atrial appendage thrombus and spontaneous contrast on transesophageal echocardiography in patients with atrial fibrillation. J Cardiovasc Electrophysiol. 2013;24:1104–1109. doi:10.1111/jce.12199.
6. Wang X, Xu X, Wang W, Huang H, Liu F, Wan C, et al. Risk factors associated with left atrial appendage thrombosis in patients with non-valvular atrial fibrillation by transesophageal echocardiography. Int J Cardiovasc Imaging. 2023;39:1263–1273. doi:10.1007/s10554-023-02841-x.
7. Nathan DM, Kuenen J, Borg R, Zheng H, Schoenfeld D, Heine RJ, et al. Translating the A1C assay into estimated average glucose values. Diabetes Care. 2008;31:1473–1478. doi:10.2337/dc08-0545.
8. Dungan KM, Braithwaite SS, Preiser JC. Stress hyperglycaemia. Lancet. 2009;373:1798–1807. doi:10.1016/S0140-6736(09)60553-5.
9. NGSP. Factors that interfere with HbA1c test results. Updated June 23, 2026. Accessed September 25, 2026. https://ngsp.org/factors.asp
10. Joy NG, Perkins JM, Mikeladze M, Younk L, Tate DB, Davis SN. Comparative effects of acute hypoglycemia and hyperglycemia on pro-atherothrombotic biomarkers and endothelial function in non-diabetic humans. J Diabetes Complications. 2016;30(7):1275–1281. doi:10.1016/j.jdiacomp.2016.06.030.
11. Çetin T. Letter to the Editor: Stress Hyperglycemia Ratio Is Associated with Intracardiac Thrombus in Acute Ischemic Stroke. Turk Kardiyol Dern Ars. 2026;54(6):534–535. doi:10.5543/tkda.2026.79395.
12. Cicek V, Erdem A, Kozan Çıkırıkçı EH, Yılmaz İ, Günhan E, Uygun E, et al. Reply to the Letter to the Editor: Stress Hyperglycemia Ratio Is Associated with Intracardiac Thrombus in Acute Ischemic Stroke. Turk Kardiyol Dern Ars. 2026;54(6):536–537. doi:10.5543/tkda.2026.67374.
13. Song L, Niu X, Wang B, Xu X, Wan C, Liu F, et al. Research on left atrial appendage thrombogenic milieu prediction model in patients with nonvalvular atrial fibrillation based on machine learning algorithm. BMC Cardiovasc Disord. 2026;26:99. doi:10.1186/s12872-025-05396-y.
14. Chu J, Tang J, Lai Y, Gao Y, Ye Z, Guan C, et al. Association of stress hyperglycemia ratio with intracoronary thrombus burden in diabetic patients with ST-segment elevation myocardial infarction. J Thorac Dis. 2020;12(11):6598–6608. doi:10.21037/jtd-20-2111.
15. Algül E, Özbeyaz NB, Şahan HF, Aydınyılmaz F, Sunman H, Tulmaç M. Stress Hyperglycemia Ratio Is Associated With High Thrombus Burden in Patients With Acute Coronary Syndrome. Angiology. 2024;75(7):645–650. doi:10.1177/00033197231167054.
16. Stalikas N, Papazoglou AS, Karagiannidis E, Panteris E, Moysidis D, Daios S, et al. Association of stress induced hyperglycemia with angiographic findings and clinical outcomes in patients with ST-elevation myocardial infarction. Cardiovasc Diabetol. 2022;21:140. doi:10.1186/s12933-022-01578-6.
17. Roberts G, Sires J, Chen A, Thynne T, Sullivan C, Quinn S, et al. A comparison of the stress hyperglycemia ratio, glycemic gap, and glucose to assess the impact of stress-induced hyperglycemia on ischemic stroke outcome. J Diabetes. 2021;13:1034–1042. doi:10.1111/1753-0407.13223.
18. Li W, Yan H, Gao Y, Gao C, Li W, Pan Y, et al. Stress hyperglycemia ratio and adverse outcomes in acute mild ischemic stroke or high-risk transient ischemic attack: a secondary analysis of the INSPIRES trial. Stroke. 2026;57:349–361. doi:10.1161/STROKEAHA.125.052987.
19. Maehama T, Okura H, Imai K, Saito K, Yamada R, Koyama T, et al. Systemic inflammation and left atrial thrombus in patients with non-rheumatic atrial fibrillation. J Cardiol. 2010;56(1):118–124. doi:10.1016/j.jjcc.2010.03.006.
20. Belen E, Özal E, Püsüroğlu H. Relationship between the presence of left atrial thrombus in patients with mitral stenosis and platelet-to-lymphocyte ratio. Anatol J Cardiol. 2016;16:673–677. doi:10.5152/AnatolJCardiol.2015.6485.
21. Cicek V, Kilic S, Dogan S, Erdem A, Babaoglu M, Yilmaz I, et al. Predictive Value of Inflammatory Scores for Left Atrium Thrombosis in Ischemic Stroke Without Atrial Fibrillation. Medicina (Kaunas). 2024;60(12):2046. doi:10.3390/medicina60122046.
22. Zhou Y, Song X, Ma J, Ren Y, Liang E, Chen K, et al. Association between complete blood cell count-derived inflammatory indices and left atrial thrombus in patients with nonvalvular atrial fibrillation: a cross-sectional study. Front Med. 2026;13:1730240. doi:10.3389/fmed.2026.1730240.
23. Gogitidze Joy N, Hedrington MS, Briscoe VJ, Tate DB, Ertl AC, Davis SN. Effects of acute hypoglycemia on inflammatory and pro-atherothrombotic biomarkers in individuals with type 1 diabetes and healthy individuals. Diabetes Care. 2010;33(7):1529–1535. doi:10.2337/dc09-0354.
24. Wright RJ, Newby DE, Stirling D, Ludlam CA, Macdonald IA, Frier BM. Effects of acute insulin-induced hypoglycemia on indices of inflammation: putative mechanism for aggravating vascular disease in diabetes. Diabetes Care. 2010;33(7):1591–1597. doi:10.2337/dc10-0013.
25. Chow E, Iqbal A, Walkinshaw E, Phoenix F, Macdonald IA, Storey RF, et al. Prolonged Prothrombotic Effects of Antecedent Hypoglycemia in Individuals With Type 2 Diabetes. Diabetes Care. 2018;41(12):2625–2633. doi:10.2337/dc18-0050.
26. Yamamoto K, Ito T, Nagasato T, Shinnakasu A, Kurano M, Arimura A, et al. Effects of glycemic control and hypoglycemia on Thrombus formation assessed using automated microchip flow chamber system: an exploratory observational study. Thromb J. 2019;17:17. doi:10.1186/s12959-019-0206-8.
27. Hagelqvist PG, Andersen A, Maytham K, Andreasen CR, Engberg S, Pedersen-Bjergaard U, et al. Hemostatic Effects of Exercise-related Hypoglycemia in Male Persons With Type 1 Diabetes. J Clin Endocrinol Metab. 2024;109(11):2735–2744. doi:10.1210/clinem/dgae278.
28. Johnston KC, Bruno A, Pauls Q, Hall CE, Barrett KM, Barsan W, et al. Intensive vs standard treatment of hyperglycemia and functional outcome in patients with acute ischemic stroke: the SHINE randomized clinical trial. JAMA. 2019;322(4):326-335. doi:10.1001/jama.2019.9346.
29. Manning WJ, Weintraub RM, Waksmonski CA, Haering JM, Rooney PS, Maslow AD, et al. Accuracy of transesophageal echocardiography for identifying left atrial thrombi. A prospective, intraoperative study. Ann Intern Med. 1995;123(11):817-822. doi:10.7326/0003-4819-123-11-199512010-00001.
30. Kaymaz C, Ozdemir N, Kirma C, Sismanoglu M, Daglar B, Ozkan M. Location, size and morphological characteristics of left atrial thrombi as assessed by echocardiography in patients with rheumatic mitral valve disease. Eur J Echocardiogr. 2001;2(4):270-276. doi:10.1053/euje.2001.0093.
31. Libiran NS, Becher E, Kataoka M. A case report on the pitfalls in diagnosing left atrial appendage thrombus: lessons from a false-positive transesophageal echocardiography (TEE) in mitral valve surgery. Cureus. 2026;18(5):e109438. doi:10.7759/cureus.109438.
32. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis: the TRIPOD statement. Ann Intern Med. 2015;162:55–63. doi:10.7326/M14-0697.
33. Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Making. 2006;26:565–574. doi:10.1177/0272989X06295361.