Heart scans disagree in cancer patients: ejection fraction and strain often diverge

In a busy cardio-oncology clinic, two echocardiographic measures of the same pumping chamber can tell two different stories. A new retrospective study of 117 oncology patients undergoing routine cardiac surveillance reports that left ventricular ejection fraction and global longitudinal strain deteriorated together in only a minority of patients, while more than four in ten showed discordant patterns in which one measure flagged trouble and the other did not. The findings, published in Cardiovascular Ultrasound, underline a practical dilemma that clinicians monitoring patients on potentially cardiotoxic cancer therapy face regularly: which number to trust, and what to do when the two disagree.

The research team, led by Jamilah AlRahimi, reviewed echocardiographic records of adult oncology patients screened between 2021 and 2023. Of 254 patients assessed for eligibility, 117 had complete imaging at three time points: baseline, an early follow-up at 6 to 12 weeks, and a short-term follow-up at 3 to 6 months. The cohort was overwhelmingly female, at 95.7 percent, and breast cancer was the most common malignancy at 89.7 percent. Most patients had received anthracycline-based chemotherapy, HER2-targeted therapy such as trastuzumab with or without pertuzumab, or a combination of both, while the remainder received other systemic anticancer regimens. Because detailed regimen-level and cumulative-dose information was not uniformly retrievable from medical records, the authors could not reliably count patients per treatment category.

Ejection fraction, the long-standing cornerstone of cardiac monitoring during cancer therapy, was measured with the biplane Simpson’s method. Global longitudinal strain, a speckle-tracking measure of how much the heart muscle shortens longitudinally with each beat, was analyzed offline using vendor-specific Philips QLAB software from standard apical views, with frame rates typically between 50 and 80 per second. The researchers applied guideline-consistent thresholds: ejection fraction deterioration was defined as either an absolute reduction of at least 10 percent from baseline or a value below 53 percent at either follow-up, while strain deterioration required a relative reduction of at least 15 percent in the absolute strain magnitude.

At the level of group averages, the heart’s systolic function looked remarkably steady. Mean ejection fraction moved from 57.7 percent at baseline to 56.7 percent at first follow-up and 57.1 percent at second follow-up, with repeated-measures analysis of variance showing no significant change over time (p = 0.124). Mean absolute strain values were similarly flat, at 20.36, 20.54 and 20.44 across the three time points (p = 0.803). On the surface, then, the population appeared to tolerate its cancer therapy without measurable cardiac cost.

Individual patients told a different story. Clinically relevant ejection fraction deterioration occurred in 32 patients, or 27.4 percent of the cohort. Ten of these met the 10-percent-reduction criterion alone, fourteen fell below the 53 percent threshold alone, and eight met both criteria. Strain deterioration appeared in 25 patients, or 21.4 percent. The authors emphasize that this gap between stable group means and substantial individual deterioration is one of the study’s central lessons: relying on population averages can obscure meaningful changes in the individual patients whose management actually depends on them.

The discordance analysis produced the study’s most striking numbers. Only 8 patients, 6.8 percent, deteriorated on both measures simultaneously. Seventeen patients, 14.5 percent, showed isolated strain deterioration despite a preserved ejection fraction, while 24 patients, 20.5 percent, showed the reverse: ejection fraction decline without any concomitant strain drop. Taken together, discordant patterns appeared in 41 percent of patients. Formal agreement between the two classifications was poor, with a Cohen’s kappa of just 0.05, though McNemar’s test found no statistically significant difference in how often each measure flagged deterioration (p = 0.27), suggesting the two parameters identify overlapping-in-size but largely different subsets of patients.

When the researchers examined the relationship between the actual changes in each measure from baseline to first follow-up, the connection was weak. The correlation between change in strain and change in ejection fraction was positive but modest (r = 0.19, p = 0.036), and in linear regression the strain change was a statistically significant predictor of the ejection fraction change (β = 0.19, p = 0.036) yet explained only 3.8 percent of its variance. In other words, knowing how much a patient’s strain shifted told you almost nothing precise about how much their ejection fraction would move.

The authors interpret these patterns through the known physiology of the left ventricle. Ejection fraction reflects global volumetric pump performance and depends on loading conditions, so it can remain within the normal range until myocardial injury is fairly advanced, buffered by compensatory mechanisms. Strain, by contrast, captures myocardial fiber deformation, particularly in the subendocardial layers that are thought to be most vulnerable to early cardiotoxic injury. This framework explains why strain reduction has often been reported to precede overt ejection fraction decline in prior work, including studies by Thavendiranathan and colleagues and Negishi and colleagues. Yet the current findings also caution against over-reading isolated strain abnormalities: the authors note that such changes may not uniformly translate into later ejection fraction decline in every patient, and they point to the SUCCOUR randomized trial, which did not show a significant reduction in ejection fraction decline with strain-guided surveillance compared with standard monitoring, as evidence that the clinical impact of strain-guided management remains uncertain.

Several features of the study limit how far its conclusions travel. The retrospective design introduces potential selection bias and restricted control over treatment-related variables. The predominance of women with breast cancer limits generalizability to other malignancies and to male patients. The short follow-up window, covering only the first three to six months of surveillance, cannot capture late cardiotoxic effects that may emerge over longer timeframes, and the absence of clinical cardiovascular outcomes means the prognostic significance of the observed imaging changes could not be assessed. Observer variability was not formally assessed, and no formal sample size calculation was performed for this exploratory study. The authors also note that contemporary breast cancer treatment regimens are increasingly associated with improved cardiovascular safety, which may partly explain the overall preservation of systolic function at the group level.

The practical message the authors draw is not that one measure should replace the other, but that neither can stand alone. Current recommendations from the American Society of Echocardiography, the European Association of Cardiovascular Imaging and the 2022 ESC Cardio-Oncology Guidelines support strain as an adjunctive tool for early detection of myocardial dysfunction, and the present data reinforce that framing: the two parameters captured related but distinct aspects of ventricular performance, disagreed in two of every five patients, and each flagged a subset the other missed. The authors suggest that clinical decisions should incorporate the broader context, including patient risk profile and treatment exposure, rather than resting on a single imaging metric, and they call for prospective studies with longer follow-up and detailed treatment data to clarify what discordant findings actually mean for patients’ futures.

Subject of Research: Discordance between ejection fraction and global longitudinal strain in oncology patients undergoing routine echocardiographic surveillance during cancer therapy

Article Title: Discordance between left ventricular ejection fraction and global longitudinal strain during routine cardio-oncology surveillance

Article References: AlRahimi, J., Saemaldahar, A., Madi, R., AlJehani, R., & AlKathiri, N. (2026). Discordance between left ventricular ejection fraction and global longitudinal strain during routine cardio-oncology surveillance. Cardiovascular Ultrasound, 24(1), Article 25. https://doi.org/10.1186/s12947-026-00386-2

Image Credits: AI Generated

DOI: 10.1186/s12947-026-00386-2

Keywords: cardio-oncology, echocardiography, ejection fraction, global longitudinal strain, cardiotoxicity, breast cancer, speckle-tracking echocardiography, chemotherapy, trastuzumab, heart failure surveillance, cardiac dysfunction, retrospective study