What actually happens, and when
The honest headline is that most patients get through the transplant itself. The risk is real, and it is concentrated later than people expect.
In 3,354 adults followed through transplantation across several centers, 4.1% had a cardiovascular event within 100 days and 13.9% within five years (Vasbinder, JACC CardioOncol 2023). Atrial fibrillation was the commonest event at both points, at 2.6% and 6.8%, followed by heart failure at 1.1% and 5.4%. Recipients of donor cells fared worse over the long run than recipients of their own, 16.4% against 12.1% at five years.
So the short-term rate is lower than the reputation of the procedure suggests, and the long-term rate is roughly three times it. A cardiology assessment that concentrates entirely on surviving the first hundred days is looking at the smaller half of the problem.
The things that predict those events are not the ones a general cardiology clinic would rank first. Prior anthracycline exposure, existing coronary disease and the type of transplant outrank most conventional cardiovascular risk factors (Vasbinder, JACC CardioOncol 2023).
Cellular therapies delivered by the same service follow a different clock again. After CAR T-cell infusion the heart is injured within days, and in proportion to the cytokine storm rather than to any cumulative dose (Ganatra, Circulation 2020). Everything below concerns transplantation itself.
Risk that can be known before the transplant
Risk identified before treatment starts is risk that can be prepared for. That is a different exercise from deciding who is eligible.
The CARE-BMT score was derived in 2,435 transplant recipients and validated in a separate 919 (Vasbinder, J Am Heart Assoc 2024). It uses age, transplant type, race, coronary disease, heart failure, peripheral artery disease, creatinine, triglycerides and cumulative anthracycline dose, all of them available on the day the decision is made.
It separates five-year cardiovascular event rates of 4.0%, 10.3% and 22.4% across its low, intermediate and high risk groups, with areas under the curve of 0.65 at 100 days, 0.73 at five years and 0.76 at ten (Vasbinder, J Am Heart Assoc 2024). It performs the same in recipients of their own cells and of donor cells. The point of it is not to find a group to turn away. It is to know which patients should have a cardiologist alongside them from the start rather than summoned once something has gone wrong.
What a low ejection fraction should and should not decide
The number most often used to refuse a transplant is the one with the least evidence behind it for that purpose.
Our position is on the record twice. The 2024 American Heart Association statement, which this lab led, holds that refusing transplantation on cardiovascular grounds should be reserved for the occasional patient with severe untreatable disease, poor cardiopulmonary reserve, or a life expectancy under a year (Hayek, Circulation 2024). We had put it more bluntly three years earlier: patients with an ejection fraction under 35% are often denied transplantation, and the data do not support the practice (Rotz, J Thromb Thrombolysis 2021).
The data in question are not ours, and they are consistent. No guideline sets a threshold. The comorbidity index usually invoked was built to predict death unrelated to relapse rather than cardiovascular events, and scores any cardiac history, ejection fraction included, as a single low-weight item (Sorror, Blood 2005). Where centers have compared patients transplanted with a reduced ejection fraction against matched patients with a normal one, cardiac complication rates and early mortality came out much the same (Qazilbash, Biol Blood Marrow Transplant 2009; Hurley, Biol Blood Marrow Transplant 2015).
The case should not be pushed further than it goes. A recent cohort of 852 recipients did find a pre-transplant ejection fraction below 50% independently predicted cardiovascular events in the first hundred days (Aghel, JACC Adv 2025), and every reassuring study is retrospective and shaped by the fact that the sickest candidates were never referred in the first place. Nobody has randomized anyone.
What follows from that is not a different threshold but a different question. Our own score, built to predict cardiovascular events specifically rather than death from any cause, does not use ejection fraction at all: what carried the signal was age, transplant type, prior anthracycline dose and vascular history (Vasbinder, J Am Heart Assoc 2024). A patient whose heart function is reduced has a risk that can be quantified, optimized in the weeks before conditioning, and watched closely afterwards. The choice is rarely between transplanting and not. It is between transplanting someone who has been prepared and transplanting someone who has not.
Atrial fibrillation is a signal, not a nuisance
The commonest cardiovascular event after a transplant has generally been treated as a passing consequence of a stressful illness. It does not behave like one.
In our own registry it was the commonest cardiovascular event after transplantation at both 100 days and five years (Vasbinder, JACC CardioOncol 2023). Frequency alone would not earn it a section. What does is what it turns out to predict.
Among 801 patients with multiple myeloma undergoing transplantation with their own cells, 8.7% developed atrial fibrillation, at a median of 13 days, predicted by age over 65, obesity and any previous episode of the arrhythmia. It carried a five-fold higher risk of death from any cause and a 4.5-fold higher risk of death not caused by the cancer coming back (Bolaji, JACC CardioOncol 2026).
That last figure is the one that matters, because death unrelated to relapse is the measure of what the transplant itself cost. An arrhythmia lasting days should not still be visible in mortality years later unless it is reporting something about the patient rather than about the moment. We have argued it is best read as a sentinel of how little cardiovascular reserve the patient had to begin with (Hayek, JACC CardioOncol 2026), which changes what should follow it: a reason to look at the whole patient, not a reason to reach for rate control and move on.
Whether any of it can be prevented
Knowing who is at risk is worth little if nothing can be done with the knowledge. This is the part of the program that is still open.
The most promising signal so far comes from a drug developed for something else. Among 11,682 transplant recipients with diabetes, the 12% who were taking a sodium-glucose cotransporter 2 inhibitor had roughly half the five-year mortality of matched patients who were not, along with fewer hospital admissions and fewer episodes of worsening heart failure (Yeo, Int J Cardiol 2026).
An effect that large in an observational comparison should be treated with suspicion, because patients well enough to be started on a new drug are different from patients who are not, and no amount of matching fully repairs that. It is a reason to run the trial, not a reason to act as though it has been run. Setting out the pathway from pre-transplant assessment through to long-term survivorship, and marking honestly which parts of it rest on evidence and which on consensus, is what the 2024 American Heart Association statement was for (Hayek, Circulation 2024).
Read next: