German survey finds cardiologists and internists disagree on heart toxicity thresholds

When cancer drugs threaten the heart, who decides that the damage is real? A new multicentre survey of German hospital physicians suggests that the specialists who receive those referrals — cardiologists and internists — do not share a common answer. In the survey, published in the journal Cardio-Oncology, responses to a single, identically worded question about left-ventricular ejection fraction (LVEF) thresholds were scattered across every available option, and a substantial share of respondents in every specialty endorsed a rule that does not match the European Society of Cardiology’s definition of cancer therapy-related cardiovascular toxicity.

The researchers contacted 1,435 departmental physicians in cardiology, internal medicine, gynaecology and oncology across multiple German federal states and received 120 completed questionnaires, a response rate of 8.4%. Most respondents — 73.3% — worked at academic teaching or university hospitals. The sample comprised 44 cardiologists, 28 oncologists, 27 gynaecologists and 21 internists. Because participation was anonymous and responses were not verified against patient records, all findings are self-reported, exploratory and hypothesis-generating rather than evidence about German hospital practice as a whole.

The survey deliberately examined the receiving side of cardio-oncology. The 2022 ESC Guidelines on cardio-oncology assign baseline cardiovascular risk assessment to the treating oncology or haematology team, with cardiological involvement guided by the resulting risk category. Most previous surveys have asked the treating specialties — the demand side — how they manage cardiotoxicity. This analysis instead asked how cardiologists and internists actually practise: which LVEF declines they treat as relevant, whether they use strain imaging, how they schedule biomarker testing in a standardised high-risk scenario, and who they believe should start cardioprotective treatment.

The threshold question produced the study’s most striking heterogeneity. Cardiologists and internists were asked at which LVEF decline to a value below the reference limits they assume relevant cardiotoxicity and communicate it. Among 34 answering cardiologists, 55.9% chose a decline of more than 10 percentage points, 38.2% regarded any LVEF decline as cardiotoxicity, and a few selected thresholds above 30 or 40 points. Among 13 internists, the ordering reversed: 61.5% said any decline, and only 23.1% chose the more-than-10-point option. Gynaecologists and oncologists answered a non-equivalent version of the question without the below-reference-limits requirement, so their responses — 26.7% of gynaecologists and 39.1% of oncologists endorsing any decline — are reported separately and cannot be compared across versions.

The authors stress that the item did not reproduce the guideline definition, which grades cardiotoxicity by severity, specifies a drop of at least 10 percentage points to an LVEF of 40–49% for moderate dysfunction and a fall below 40% for severe dysfunction, and incorporates global longitudinal strain and cardiac biomarkers alongside LVEF. The authors note that the item also did not specify whether the decline was absolute or relative. What it can describe is the reporting threshold at which consultants flag and communicate cardiotoxicity — and that threshold matters clinically, because it is the point at which surveillance is intensified, cardioprotective drugs are considered and cancer therapy may be modified. A criterion set too low risks unnecessary interruption of effective oncological treatment; one set too high may delay recognition of dysfunction the guidelines regard as actionable. Both types of response rule appeared in the data, prompting the authors to identify education within cardiology itself as a target, three years after the guideline’s publication and in a predominantly academic sample.

Strain imaging showed a similar divide between the consulting disciplines. Routine echocardiographic strain use during follow-up after chemotherapy was reported by 76.5% of cardiologists (26 of 34) but only 36.4% of internists (4 of 11), an exploratory difference with a Fisher’s exact p-value of 0.026. Among the small number of respondents who did not use strain — eight cardiologists and seven internists — internists more often cited lack of expertise and insufficient acoustic windows, while cardiologists more often cited time and personnel constraints. The authors caution that these barrier counts are too small to establish distinct specialty-specific mechanisms.

Biomarker surveillance in the standardised high-risk scenario — a patient with pre-existing heart failure receiving anthracycline chemotherapy — revealed a reactive pattern that persisted even among cardiologists. Cardiologists reported baseline NT-proBNP or troponin testing before chemotherapy in 64.7% of cases, and testing after each cycle or after completion in 38.2% each. But late testing in the first post-treatment year, during which much anthracycline-associated cardiotoxicity becomes clinically apparent, was endorsed by only 11.8% at six months and 20.6% at twelve months, and acute cardiac events remained a common trigger for 47.1%. Internists tested less often at every proactive time point and reported no testing at six or twelve months, with 18.2% determining no biomarkers at all. Serial troponin and natriuretic peptides are inexpensive, guideline-recommended tools that can detect cardiotoxicity before imaging changes, and the authors argue that the absence of a defined owner and schedule for post-treatment surveillance is a systemic problem rather than one confined to the treating specialties.

Ownership of baseline risk stratification itself was the least clearly assigned of any step in the pathway. Formal cardiotoxicity risk stratification showed the lowest reported cardiological involvement of any step, at 42.9% of cardiologists. Among the 51 cardiologists and internists asked directly, only about half performed the stratification — 41.2% within an interdisciplinary team and 9.8% independently — while 29.4% did not perform it at all and 19.6% regarded it as generally the treating oncologist’s or haematologist’s task, a position the authors note is actually consistent with the guideline pathway. The problem, they conclude, is not that the assessment is being skipped — 85.4% of gynaecologists and oncologists in the companion report from the same survey said they performed it — but that consultants do not agree on where the task sits. Where an assessment was made, the interdisciplinary process largely worked as intended: 88.5% of respondents reported that it was translated into a joint evaluation with the treating oncologist or gynaecologist, and only 7.7% issued a unilateral recommendation.

The work-up requested from cardiology reinforced the impression of an incomplete handoff. Oncologists requested a broader panel of investigations than gynaecologists, but in both groups echocardiography and electrocardiography predominated — requested by 95.8% of oncologists and 85.0% of gynaecologists — while troponin, natriuretic peptides and additional laboratory parameters were requested roughly half as often, and about a quarter of respondents did not request an ECG at all. This matters because a baseline ECG is recommended for all patients before potentially cardiotoxic therapy, and baseline biomarkers and cardiovascular risk factors contribute to the HFA-ICOS risk category that determines subsequent management. The referral, the authors suggest, is being framed as an imaging question rather than as part of a complete baseline risk assessment.

The authors frame their findings as candidate targets for prospective evaluation rather than as proven deficits: agreed LVEF and strain reporting conventions embedded in referral criteria, implementation of routine strain imaging, structured post-treatment biomarker and imaging surveillance within survivorship pathways, and explicit assignment of responsibility for prescribing and titrating cardioprotective therapy. They also emphasise the survey’s limitations. The 8.4% response rate, though comparable to other voluntary physician surveys, carries a risk of non-response bias whose direction cannot be determined; academic centres were over-represented, so the gaps described may be larger in community hospitals; respondents’ accounts were not verified against records; and several items rested on small subgroups. Because the survey captured physicians’ reports rather than measured care or outcomes, the authors write, these exploratory findings identify where the consulting side of cardio-oncology diverges — and where structured remedies, not general appeals to collaborate, might close the divide.

Subject of Research: Self-reported cardio-oncology practices, diagnostic thresholds and surveillance responsibilities among German hospital cardiologists and internists

Article Title: Cardio-oncology from the consulting side: diagnostic thresholds, surveillance and treatment responsibility among cardiologists and internists in a German multicentre survey

Article References: Poley, S., Payo-Anez, L., Heidrich, F. M., Ibrahim, K., Christoph, M., Haase, T., Hamann, J., Harst, L., & Quick, S. (2026). Cardio-oncology from the consulting side: diagnostic thresholds, surveillance and treatment responsibility among cardiologists and internists in a German multicentre survey. Cardio-Oncology, 12(1), Article 123. https://doi.org/10.1186/s40959-026-00588-7

Image Credits: AI Generated

DOI: 10.1186/s40959-026-00588-7

Keywords: cardio-oncology, cardiotoxicity, LVEF threshold, strain imaging, biomarker surveillance, ESC guidelines, anthracycline, risk stratification, Germany, physician survey, NT-proBNP, troponin