We reconstruct the principal prognostic factors in pedi- atric DCM on the basis of recent high-quality evidence and consequently we arrange them categorically. A.

Clinical and Demographic Prognostic Fac-tors

Age at diagnosis has always been an important prognos-tic factor for pediatric DCM patients from a clinical per-spective. ⁵ Later childhood/adolescence at presenta-tion usually exhibits worse transplant-free survival than younger children among several cohorts. ^{2, 3} Older age at diagnosis in the PCMR (for instance, above 6 years) was still a risk factor for death or heart transplantation independently of the other confounding factors. ² Sev-eral series have considered young age as a positive prog-nostic factor possibly due to a higher proportion of re-versible conditions at early ages (such as myocarditis) and an ability to recuperate for those who survive the initial very risky period. ^{5, 2} Nevertheless, infant-onset DCM ( < 1 year) is not always protective; some groups find that infants have a higher short-term mortality rate, which is probably due to the disease progressing so quickly that there is no time to get advanced support. ⁴ This is in line with the idea that age, risk relationship is non-linear, with the highest risk being at the extremes of age (very early infancy and older childhood) rather than a simple linear effect. ⁴ From a clinical perspective, adolescent-onset DCM is a case for a higher level of risk alert, whereas the age factor should be considered in combination with etiology and severity at presentation. ^{2, 4, 9}

In pediatric DCM, etiology serves as the main “risk con-text” since it determines how severity markers (EF, LV dila-tion, MR, and biomarkers) are to be interpreted. ² Data from the registry era reveal a large variation in the out-come depending on the cause, with significantly better scenarios in the categories of potentially reversible versus primary/genetic phenotypes. ² Different studies show that myocarditis-associated DCM is more likely to recover function and have fewer death/HTx events than idio-pathic DCM, which is in line with an acute inflammatory injury that may be facilitated by early support. ^{2, 10} By contrast, idiopathic DCM, which is largely accounted for genetic cardiomyopathy cases in modern cohorts, has been associated with lower transplant-free survival and a more deteriorating progression of the disease. ^{2, 11} Familial/genetic disease could be at a higher risk in some cases, and the family history can be a convenient bedside indicator for increased monitoring; for instance, a single-center cohort study found family history as a risk fac-tor in time-to-event analysis. ¹² Neuromuscular DCM (e.g., dystrophinopathies) is also typically linked to worse prognosis, which is influenced by the presence of sys-temic comorbidities and, in some cases, limited advanced-therapy candidacy. ² Importantly, new data have con-nected the words “etiology” and “genotype”: in a study involving children who had their myocarditis confirmed through biopsy and showed a DCM phenotype, it was found that pathogenic/likely pathogenic cardiomyopathy variants were highly present and that event-free survival was significantly lower in those that had such variants, thus indicating that myocarditis may reveal an inherited substrate. ^{10, 13} Therefore, the etiologic classification should be clear when talking prognostics, as it not only changes the baseline risk but also how genetic testing can further refine the risk even within presentations labeled as myocarditis. ^{2, 10, 13}

Severity of HF at diagnosis is strongly indicative of a near-term prognosis in pediatric DCM. Children manifesting advanced symptoms (Ross/NYHA III, IV) are found to be at a notably greater risk of death/HTx as compared to those with less severe presentations. ¹⁴ In a signifi-cant group of children, the researchers created a nomo-gram and a simple bedside score. They found that an advanced functional class not just predicted the outcome but was also the factor with the greatest weight, which is in agreement with the idea that severe heart failure is a highly indicative marker of immediate danger. ¹⁴ Markers of critical illness, such as admission to the ICU, being on inotropes for a longer period, and the need for MCS, are mainly seen in children very close to cardiogenic shock and thus, these markers are highly correlated with early death or very quick progression to HTx. ¹⁵ Besides baseline visual acuity being a starting point, prognosis is not static: analysis of registry data shows that the greatest risk of death or heart transplant is right after the diagno-sis, and those who stabilize continue to have an improv-ing conditional prognosis with recovery patterns varying according to the cause and treatment response. ^{2, 16} On the clinical side, when a patient is in a critical condi-tion (for example, low blood pressure, which keeps get-ting worse, needing increasing doses of vasoactive drugs, ventilatory support, or consideration of ECMO/MCS), it is advisable to seek the help of advanced HF/transplant services immediately. ^{15, 14} Severity scores are also useful; for instance, elevated NYU-PHFI scores have been linked with greater chances of death/HTx in pediatric DCM populations. ¹⁷

Arrhythmias in pediatric DCM patients can be complica-tions as well as signs of the existence of a more malignant myocardial substrate. Although atrial arrhythmias and conduction disease usually are signs of an advanced or arrhythmogenic phenotype, sustained VT and advanced AV block are directly relevant to the risk of sudden death. ⁵ Previously, reviews showed inconsistent relation-ships between heart rhythm disorders and results, prob-ably because of different definitions and small numbers of events in cohorts. ⁵ Current evidence from the reg-istry era confirms that physicians should classify as “high-concern” cardiac arrest or life-threatening ventricular ar-rhythmias and severe conduction abnormalities on ECG, notably when there is also severe LV dysfunction or the presence of adverse remodeling. ¹⁸ Sudden cardiac death (SCD) is a smaller, but still significant, part of the total mortality in DCM (3% at 5 years). Hence, prag-matic stratification tools have been suggested that can be used to identify children who might be monitored more closely for their heart rhythm or even considered for de-vice therapy. ¹⁸ For example, in one model, it was demonstrated that the majority of SCD events occurred in children who at one point fulfilled the triad criteria (severe LV dilation, age below a certain threshold, and LV wall thinning), which is in line with a thin-walled, heavily remodeled ventricle that is vulnerable to malig-nant arrhythmias. ¹⁸ ECG phenotypes might provide additional prognostic information that is independent of other factors: lowvoltage QRS was independently asso-ciated with death/HTx in a contemporary pediatric risk score, ¹⁴ probably indicating diffuse myocardial disease or fibrosis burden. ¹⁴ In addition, conduction disease may indicate the presence of high-risk genotypes; LMNA-related DCM is typically accompanied by conduction ab-normalities and malignant ventricular arrhythmias, which justifies rhythm surveillance and escalation planning from a very early stage. ¹⁹ A similar example would be SCN5A-related phenotypes, which may present a mixture of conduction disorder/arrhythmias, together with struc-tural DCM, thus emphasizing the idea that “electrical red flags” might be an indicator of genotype-related risk. ²⁰ In a real-world scenario, a patient with sustained VT, a high number of ectopy/NSVT together with adverse re-modeling, or major conduction disease should be consid-ered for a more intensive follow-up and the involvement of an electrophysiologist, while decisions about the device should be individualized. ^{18, 19, 20}

The knowledge of cardiomyopathy or sudden cardiac death in a family is medically very significant as it can be used as a proxy for the presence of hereditary dis-eases. Therefore, it is common that this evidence leads to early genetic testing and cascade screening. ²¹ On the other hand, the majority of studies show that family his-tory alone has no clear effect on the outcomes of pediatric DCM. The variations in definition, referral patterns, and genetic testing intensity among cohorts might have con-tributed to the different evidence obtained. ²² A single-center cohort found a positive family history of DCM to be significantly associated with mortality in Cox regression. ¹² On the other hand, a registry-based comparison of familial DCM versus idiopathic DCM in children revealed a broadly similar transplant-free survival, and the family history of sudden death was not an independent predic-tor of death/HTx. ²² Similarly, in the context of SCD-centric modeling, family history has not always been ob-served as a standalone predictor. ¹⁸ Thus, family his-tory should ideally be handled less as a statistical risk-enhancing factor and more as a very effective prompt for obtaining a more comprehensive etiologic understanding, especially genetic testing and rhythm monitoring, since the actual risk is frequently determined by the genotype-associated malignant phenotypes rather than by family history itself. ^{21, 22}

While the occurrence rates differ between sexes, sex, however, has not ultimately been identified as a stable independent predictor of death/HTx when the pheno-type and severity are taken into account. ^{2, 22} Un-der competing-risk models, male sex has failed to demon-strate a strong correlation with the combined outcome of death or transplantation, thereby implying that the dif-ference seen is usually a result of the severity, etiologic mix, or care pathways factors. ²² For example, sex-related disparities in advanced-therapy treatments rather than in the biology of intrinsic DCM could be more pro-nounced; transplant-era analyses, for instance, have doc-umented differences in pediatric waitlist survival by sex while the transplantation rates were the same. ²³ In comparison, growth and nutritional status are indicative of the overall whole-child severity. Stunting (low height-for-age z-score) has been linked in PCMR studies to a higher risk of death (even more than the risk of trans-plantation), which is in line with the understanding that growth failure reflects the chronic burden of heart fail-ure and the child’s susceptibility to decompensation. ⁵ From a clinical standpoint, failure to thrive or malnutri-tion can be considered effective warning signs that ne-cessitate increased monitoring and supportive treatment when combined with adverse cardiac indicators. ^{2, 5}

Resource setting has a major impact on the outcomes ob-served in pediatric DCM, mainly through access to ad-vanced HF therapies such as HTx and durable VAD sup-port. ²⁴ In LMICs, these options may be limited, and outcomes often reflect what can be achieved with med-ical therapy and supportive care alone; a prospective study conducted in Khartoum state (Sudan) has shown very high mortality rates, thus emphasizing the suffer-ing of the very ill patients without the availability of scal-ing up treatments. ²⁵ Other LMIC cohorts also report similar high event rates and persistent severe dysfunc-tion, which is in line with delayed presentation and lim-ited longitudinal HF infrastructures that exacerbate early risk. ²⁶ LMIC studies also uncover modifiable factors. For instance, an Ethiopian decade-experience series first linked malnutrition to bad outcomes and then enalapril use to better outcomes, implying that basic HF optimiza-tion in these patients can still bring significant improve-ments even when HTx is not available. ²⁷ On the other hand, in transplant-capable systems, HTx and durable support have the potential to change the paths of the pa-tients from death to survival with advanced intervention, thereby making the use of the composite endpoint “death or HTx” complicated for comparisons. ²⁴ Experience in the United States shows that using durable VADs as a bridge to transplant for a longer period can result in greater success of bridging to transplantation when com-pared with the use of extracorporeal support. This is an example of how the healthcare system’s ability to man-age a case can change what was initially a fatal course into a survival after transplantation. ²⁸ Through the years, registry studies have also revealed enhanced sur-vival without the need for a transplant, indicating that HF treatment improvements have been made in general and not just in the areas of transplant volume. ⁷ Hence, prognostic understanding must deliberately recognize re-source context; a given clinical severity, therefore, may result in different actual risks based on whether and how quickly advanced therapies are available. ^{24, 25} The major clinical and demographic predictors, together with representative supporting evidence, are summarized in Table 1.

Currently, among pediatric DCM cohorts, NT-proBNP is the biomarker with the most reliable prognostic power and it is often the main determinant even after the clinical and imaging covariates have been taken into consideration. ³¹ Through a prospective multicenter repeated-measures design, the study has revealed that el-evated NT-proBNP levels were a very strong predictor of death/HTx and, most importantly, the longitudinal pat-tern contributed substantial prognostic information: the children that eventually reached the endpoints had the NT-proBNP levels gradually increasing (and staying high) compared to the patients without events, thus, encour-aging “dynamic risk” evaluation over merely relying on one baseline snapshot. ³¹ Further support for the non-linearity of the relationship between NT-proBNP and risk, as well as the variation of this association with the time horizon and cohort characteristics, comes from registry-linked analyses which thus fortify the notion that the in-terpretation should be local and not connected to a sin-gle universal cutoff ( Table 3). ³⁷ BNP levels measured in pediatric outpatient clinics for chronic LV systolic dys-function patient groups (including cardiomyopathy phe-notypes) have shown excellent risk discrimination of chil-dren at high risk of adverse events in the near short-term ( Table 3). ³⁸ Essentially, the BNP/NT-proBNP are most effectively employed as (i) baseline severity markers and (ii) serial response-to-therapy markers, in which case a failure to decline or a rebound in elevations should raise a concern still higher and lead to a re-examination of the trajectory and escalation timing. ^{31, 37} 2)

Inflammatory Markers (Neutrophil-to-Lymphocyte Ratio (NLR) and Systemic In-flammatory Index (SII)):

CBC-derived indices could reflect systemic inflammatory activation and physiologic stress in pediatric DCM. In a re-cent study of the pediatric DCM cohort, the non-survivors had higher NLR and SII. Besides that, such indices show quite a good discrimination for mortality by the ROC anal-ysis, and more importantly, even after adjusting for other variables, these two parameters are still significantly cor-related with mortality within that dataset ( Table 3). ¹² The parameters are appealing as they both inexpensive and widely accessible, and thus, they could have poten-tial in environments that lack advanced biomarker assays. However, due to the fact that the current evidence base is still limited by the number of cohort and single-center de-sign, one should consider them as complementary signals that need to be validated multicenter. ¹² 3)

Other Biomarkers and Routine Laboratory In-dicators:

Markers of myocardial injury (troponins) and general in-flammation (e.g., CRP) have not been reliable prognostic tools in chronic pediatric DCM except for those cases with specific etiologies; in the same pediatric cohort, there was no significant difference between troponins and CRP in terms of outcome and they were not significant predictors in the univariate time-to-event analyses ( Table 3). ¹² Routine chemical imbalances in the body may still indi-cate more advanced conditions. For instance, a reduced level of sodium in the blood (serum sodium) was linked with death and demonstrated a signal for prognosis in the univariate analysis of a pediatric dilated cardiomyopathy (DCM) cohort, which is in line with hyponatremia being a sign of severe heart failure (HF) neurohormonal activa-tion and patient vulnerability ( Table 3). ¹² Care should be taken when interpreting observations of other routine measures (magnesium for example) as the sample size is small and there could be scale effects. ¹² Besides that, the severity of the biomarkers together with the physio-logical correlates can actually reveal quite a lot: In a tiny pediatric cohort, it was also found that non-survivors had lower transmitral E-wave velocity, thus, indirectly con-firming the idea that very low forward-flow surrogates might be typical of advanced dysfunction. Nevertheless, the discoveries have to be replicated in bigger datasets before they can be used clinically. ¹² D.

Genetic Factors and Genotype-Phenotype Correlations

Genetic testing has fundamentally changed our understanding of the causes and outcomes of pediatric DCM. Today, an increasingly standard part of care is compre-hensive genomic testing, mainly because it can (i) iden-tify the cause of the disease, (ii) facilitate family mem-bers’ testing through cascade screening, and (iii) pro-vide risk assessment if the genotype, outcome associations are known in pediatric populations. ^{11, 39} Diagnostic yield varies among cohorts because of different pheno-type definitions (isolated vs. syndromic DCM), age com-position, and sequencing/interpretation pipelines. How-ever, various studies have demonstrated that a signifi-cant portion of children initially diagnosed as “idiopathic” cases, are eventually discovered to have a genetic basis for their condition when thoroughly examined. ^{11, 39} Studies of contemporary pediatric DCM cohorts show that the yield varies considerably between identifying genetic causes, and pediatric outcome data now reveal that the “gene-positive” status can have prognostic implications, i.e., it is not solely diagnostic. Genome sequencing of a nationwide Dutch pediatric DCM cohort (2010, 2017) un-covered LP/P variants in more than one-third of the chil-dren tested, and the carriers of the variants had a signif-icantly worse transplant-free survival, with a greater risk (hazard) of death/HTx during the follow-up period. ¹¹ One of the things that most clearly reveals the usefulness of clinical utility is a lot of sequencing work done in stud-ies of pediatric cardiomyopathy. For instance, the exome-based work in a large pediatric cohort found a molecular diagnosis in about a third of patients overall. The study also showed that the yield depended on the age and was lower in infants. This therefore supports the notion that the yield from a cohort is determined by its make-up and the risk interpretation given afterwards. ^{40, 41}

Attempts to formalize multivariate risk prediction for pe-diatric DCM are currently underway, but are still not as developed as adult heart failure scores. The reason be-ing that pediatric cohorts are smaller and endpoints are more often considered as time-to-event outcomes (e.g., death or heart transplantation [HTx]). ¹⁴ Clinicians in fact use essential predictors such as age at diagno-sis, severity of heart failure leading to hospital admis-sion, remodeling/ventricular dysfunction, and the etio-logical context to do a sort of informal multivariate risk formula even if there are no considered bedside scores. ^{14, 31} Recently, research has increasingly focused on cohort-derived point scoring tools. In a 2025 pedi-atric DCM study, Xu et al. created a nomogram and a streamlined score based on five predictors (age at onset, advanced functional class [III, IV], moderate-to-severe MR, low-voltage ECG, and requirement for vasoactive drugs) to forecast death/HTx, their results showed that the two methods had a discrimination of AUC 0.804 (nomogram) and AUC 0.794 (simplified score), respec-tively. A high-risk-cutoff ≥13.25 was proposed by the authors. At the same time, they stressed that external validation is a must before the model can be used rou-tinely. ¹⁴ Methodologically, a significant step forward in line with the principle that the trajectory matters is the move to time-updated models that incorporate re-peated measurements rather than using baseline snap-shots only. Serial evolution of biomarkers and imaging features can improve separation between responders and non-responders, and dynamic markers may outperform a single baseline value for predicting adverse outcomes over follow-up. ³¹ Similar methods focus on interme-diate trajectories that are closely connected to the down-stream risk, such as LV reverse remodeling after the ther-apy guided by the guidelines. As an illustration, Han et al. came up with a predictive model for reverse remod-eling in children with recent-onset DCM which can be used for more personalized counseling and determining the intensity of follow-up based on the expected remod-eling response. ⁴⁴ Lastly, machine-learning (ML) and data-driven phenotyping techniques are gaining ground in identifying subgroups with clinical significance that are not easily distinguishable by means of linear point scores.

Fu et al. characterized the phenotypes of patients with very different event risks in their electronic-record clus-tering study (n = 279) from 2025. The higher-risk phe-notype patients were at substantially elevated risk of ad-verse outcomes (reported HR 8.10, 95% CI 4.05, 16.19), “phenotype-driven” methods can be used to supplement standard regression, which, however, still has to be exten-sively externally validated and tested for transportability. ⁴⁵ Practical takeaway: the existing pediatric DCM risk look promising but the majority of them are lim-ited to single cohorts only and thus should be considered as decision-support prototypes until they get externally validated and calibrated across systems where’death vs HTx’ pathways differ. ^{14, 45} Representative prognostic modeling approaches in pediatric DCM, including score-based, time-updated, reverse-remodeling, and machine-learning approaches, are summarized in Table 5.

Most of the pediatric DCM evidence base is still largely composed of observational studies and also cohort het-erogeneity along with changing treatment eras. A lot of predictors are interrelated (collinearity), so it is not very clear how to separate the independent effects of each and the risk of “double counting” closely related mea-sures is also increased. There are several new markers (e.g., inflammatory indices and advanced CMR fibrosis measures) which are still based only on relatively small pediatric samples and thus require multicenter valida-tion. Moreover, the modelling challenge is made even more complicated due to the fact that both transplanta-tion and VAD, by resulting in patients bypassing their nat-ural progression, yield the observation of event rates that reflect referral patterns and resource availability. To sum up, although psychosocial and system factors (such as ac-cess barriers, adherence, and follow-up constraints) are scarcely recorded, they can significantly change the real-world risk.

Priorities set forth for the field are: (i) to carry out ex-ternal validation and transportability testing of already available pediatric risk tools in different settings; (ii) stan-dardizing definitions and handling of endpoints, includ-ing competing risks; (iii) prospectively harmonizing imag-ing and biomarker protocols in multicenter studies (this also includes strain and natriuretic peptide sampling); (iv) genotype-informed cohorts powered for gene-specific pediatric risk estimation and myocarditis, genotype inter-actions; and (v) implementation studies to assess whether structured, trajectory-based risk stratification improves referral timing and family-centered outcomes.