Risk stratification of cardiac metastases using late gadolinium enhancement cardiovascular magnetic resonance: prognostic impact of hypo-enhancement evidenced tumor avascularity

Background Late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR) is widely used to identify cardiac neoplasms, for which diagnosis is predicated on enhancement stemming from lesion vascularity: Impact of contrast-enhancement pattern on clinical outcomes is unknown. The objective of this study was to determine whether cardiac metastasis (CMET) enhancement pattern on LGE-CMR impacts prognosis, with focus on heterogeneous lesion enhancement as a marker of tumor avascularity. Methods Advanced (stage IV) systemic cancer patients with and without CMET matched (1:1) by cancer etiology underwent a standardized CMR protocol. CMET was identified via established LGE-CMR criteria based on lesion enhancement; enhancement pattern was further classified as heterogeneous (enhancing and non-enhancing components) or diffuse and assessed via quantitative (contrast-to-noise ratio (CNR); signal-to-noise ratio (SNR)) analyses. Embolic events and mortality were tested in relation to lesion location and contrast-enhancement pattern. Results 224 patients were studied, including 112 patients with CMET and unaffected (CMET -) controls matched for systemic cancer etiology/stage. CMET enhancement pattern varied (53% heterogeneous, 47% diffuse). Quantitative analyses were consistent with lesion classification; CNR was higher and SNR lower in heterogeneously enhancing CMET (p < 0.001)—paralleled by larger size based on linear dimensions (p < 0.05). Contrast-enhancement pattern did not vary based on lesion location (p = NS). Embolic events were similar between patients with diffuse and heterogeneous lesions (p = NS) but varied by location: Patients with right-sided lesions had threefold more pulmonary emboli (20% vs. 6%, p = 0.02); those with left-sided lesions had lower rates equivalent to controls (4% vs. 5%, p = 1.00). Mortality was higher among patients with CMET (hazard ratio [HR] = 1.64 [CI 1.17–2.29], p = 0.004) compared to controls, but varied by contrast-enhancement pattern: Diffusely enhancing CMET had equivalent mortality to controls (p = 0.21) whereas prognosis was worse with heterogeneous CMET (p = 0.005) and more strongly predicted by heterogeneous enhancement (HR = 1.97 [CI 1.23–3.15], p = 0.005) than lesion size (HR = 1.11 per 10 cm [CI 0.53–2.33], p = 0.79). Conclusions Contrast-enhancement pattern and location of CMET on CMR impacts prognosis. Embolic events vary by CMET location, with likelihood of PE greatest with right-sided lesions. Heterogeneous enhancement—a marker of tumor avascularity on LGE-CMR—is a novel marker of increased mortality risk. Supplementary Information The online version contains supplementary material available at 10.1186/s12968-021-00727-2.


Background
Nearly 17 million Americans are living with cancer [1], among whom cardiac metastases (C MET )bear a major impact on therapeutic decision-making and prognosis. Survival has markedly improved for patients with advanced (stage IV) cancer, resulting in a growing population at risk for C MET and its serious consequences. Data from our group and others have shown C MET to be common with advanced cancer, occurring in up to 20% of patients. [2][3][4][5][6] Embolic events-which can occur when lesions dislodge from the heart-are a leading source of morbidity and mortality among patients with C MET . Given that a growing array of new therapies and anticoagulants are available to potentially reduce risk, improved strategies to guide therapy and refine prognostic risk stratification for patients at risk for C MET are of substantial importance.
Cardiovascular magnetic resonance (CMR) has been well-validated for tissue characterization of cardiac masses. [2,4,[7][8][9][10][11][12] Whereas neoplasms can vary in morphology, vascular supply is an intrinsic requirement for tumor growth and this property can be leveraged for diagnostic purposes. Using the technique of late gadolinium enhancement (LGE), CMR can identify neoplasms based on vascularity as manifested by presence of contrast-enhancement. [13] It is also known that neoplasms can vary in pattern of contrast enhancement on LGE-CMR, and that some lesions can include enhancing and non-enhancing components. [2][3][4] Consistent with this, pathology studies have shown that some neoplasms can have avascular foci ("tumor necrosis")-a finding linked to aggressive tumor growth and adverse outcomes. [14] Impact of tumor avascularity-as manifested by contrast hypo-enhancement on LGE-CMR-has yet to be tested as a prognostic marker among patients with C MET .
This study encompassed a broad cohort of systemic cancer patients with C MET as well as controls (without C MET ) matched for cancer etiology and stage. CMR was performed using a tailored protocol to assess presence and pattern of C MET enhancement-including standardized grading and quantitative analyses-as well as standardized assessment of lesion size and mobility. The goal was to test impact of C MET anatomic distribution and contrast enhancement pattern on embolic events and mortality.

Study population
The population was comprised of adults (≥ 18 years) with advanced (stage IV) systemic cancer and C MET , and controls without C MET , matched (1:1) for cancer diagnosis: Presence or absence of C MET was established using the reference of LGE-CMR, on which it was defined via established criteria as a discrete tissue prominence with vascularity as evidenced by contrast-enhancement. [2][3][4] Subjects with CMR-evidenced intracardiac thrombi were excluded. Figure 1 provides a schematic of the research protocol. As shown, study participants were accrued at two tertiary care centers with dedicated cancer care programs (Memorial Sloan Kettering Cancer Center [MSKCC], Weill Cornell Medicine-New York Presbyterian Hospital, New York, New York, USA) that share an integrated CMR program. Clinical data were collected in a standardized manner, including cancer diagnosis and stage, anti-cancer and anticoagulant therapies, as well as clinically documented embolic events (pulmonary embolism (PE), cerebrovascular events (CVA), systemic [splenic, peripheral] emboli) within 6 months of CMR. Mortality was assessed to test prognosis in relation to anatomic and tissue characteristics of C MET .
This study entailed analysis of data acquired for clinical purposes between 2012 and 2020; no dedicated interventions were performed for research purposes. Ethics approval for this protocol was provided by the MSKCC and Weill Cornell Medicine institutional review boards, each of which approved a waiver of informed consent for analysis of pre-existing clinical data.

Image analysis
C MET was identified on LGE-CMR based on lesionassociated vascularity in accordance with established qualitative (visual) criteria. [2][3][4] To test modifying impact of C MET tissue properties on cancer associated outcomes, lesions were further classified into two distinct categories-diffuse enhancement (homogenous contrast enhancement throughout lesion) or heterogeneous enhancement (enhancing and non-enhancing components within a lesion).
C MET  Quantitative analyses were used to assess magnitude of contrast enhancement within C MET . Concordant with established methods used by our group, [2,4] aggregate signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) ratios were measured on the long TI LGE-CMR image on which the lesion was most prominent. Intracardiac lesion size (area, linear dimensions) was measured on cine-CMR datasets, which were co-localized with LGE-CMR for purpose of analyses. For patients with multiple C MET , the largest lesion was used for quantitative analysis and patient categorization (i.e. heterogeneous or diffusely enhancing). Ancillary analyses included quantification of cardiac chamber size and function, which were measured on cine-CMR using standard planimetry methods. [3]

Prognostic assessment
Electronic medical records were reviewed to assess embolic events (PE, CVA, systemic [i.e. splenic, peripheral] emboli) within 6 months of imaging, as well as allcause mortality after CMR so as to test clinical events in relation to presence and type of C MET . All-cause mortality and embolic events were ascertained blinded to CMR analyses.

Statistical methods
Comparisons between groups with or without C MET , and between C MET subtypes (heterogeneous, diffusely enhancing) were made using Student's t-tests (expressed as mean ± standard deviation) for continuous variables, and Chi-square or Fisher's exact tests for categorical variables: Paired testing (t-tests or McNemar's tests) was employed for matched case-control comparisons. Univariate logistic regression was used to test variables associated with mortality and embolic events; variables significantly associated with outcomes in univariate Overall schematic of multicenter enrollment as well as standardized cardiovascular magnetic resonance (CMR) acquisition and analysis. Note that both enrolling sites employed a tailored CMR protocol for assessment of C MET , inclusive dedicated long inversion time (TI) late gadolinium enhancement (LGE)-CMR for evaluation of contrast-enhancement pattern within lesions. Ancillary clinical data were collected in a uniform manner, including baseline cancer-related indices, embolic events, and mortality following CMR analysis were then then tested together in adjusted models. The Kaplan-Meier method was used to calculate survival; follow-up duration was reported as median with interquartile range (IQR). Cox proportional hazards models compared mortality risk between groups, including prognostic utility of C MET features. Calculations were performed using SPSS (Statistical Package for the Social Sciences, International Business Machines, Inc., Armonk, New York, USA). Two-sided p < 0.05 was deemed indicative of statistical significance.

Population characteristics
The population comprised 224 adults with advanced (stage IV) systemic cancer undergoing CMR, including 112 patients with C MET as defined by LGE-CMR, and unaffected (C MET -) controls matched for primary cancer diagnosis and stage. Table 1 details population characteristics, together with comparisons between cancer patients with C MET and their respective controls. As shown, cancer diagnosis varied among patients with C MET : Sarcoma, lung, genitourinary, gastrointestinal cancers, and skin/melanoma comprised the leading primary cancer diagnoses, although the population also included patients with primary cancers not typically associated with C MET (e.g. endocrine, head/neck). Regarding anti-cancer regimen, patients with C MET were more likely to be treated with mediastinal radiation therapy and low molecular weight heparin (both p = 0.01) but were otherwise similar with respect to matched controls. Of note, nearly half (49%) of patients with CMR-evidenced C MET had a subsequent change in anti-cancer medication regimen following CMR; 16% received new mediastinal/chest radiation within 3 months after imaging.
Regarding cardiac indices, cancer-matched controls referred for CMR were more likely to have adverse left sided chamber remodeling-as evidenced by lower LV ejection fraction and larger chamber size (both p < 0.01), but groups had similar right sided structural and functional indices (p = NS).

Cardiac metastasis location in relation to contrast enhancement
Anatomic location of C MET varied (LV 32% │LA 22% │, RV 37%│ RA 30%│multi-chamber involvement 30%): Left and right sided chamber involvement were near equal in prevalence (50%, 58% respectively). Regarding C MET tissue properties, 53% of patients had heterogeneously enhancing lesions (enhancing and non-enhancing components on LGE-CMR), whereas 47% had diffusely enhancing lesions without non-enhancing components.   Table 2 compares heterogeneously enhancing and diffusely enhancing C MET . As shown, lesions were similar with respect to anatomic distribution, as evidenced by equivalent patterns of chamber involvement and rates of intra-cavitary location (all p = NS). Regarding lesion size, heterogeneously enhancing lesions were larger, based on linear dimensions (p < 0.05). Quantitative analyses were consistent with lesion classification, as evidenced by higher CNR (reflecting greater differences between enhancing and non-enhancing regions) and lower normalized SNR (reflecting impact of non-enhancing lesions on aggregate lesion signal intensity) in heterogeneously enhancing C MET (both p < 0.001 vs. diffusely enhancing C MET ).

Embolic events
Embolic events (within 6 months of CMR) were assessed to test if presence of C MET impacted likelihood of clinical events, and whether this was modified by lesion location or tissue characteristics. A total of 33 embolic events occurred in the study population; events occurred at a median interval of 2 weeks from CMR [IQR 0.5, 9.5 weeks]. As shown in Table 3A, embolic events were over twofold more common among patients with C MET as compared to cancer matched controls (21% vs. 8%, p = 0.006), including increased incidence of PE (13% vs. 5%, p = 0.08): Embolic events occurred at a median interval of 2 weeks from CMR [IQR 0.5, 9.4 weeks].
Data shown in Table 3A also demonstrates that C MET location modified likelihood of clinical events: Whereas patients with right sided lesions had a more than threefold increase in PE (20% vs. 6%, p = 0.02), those with left sided lesions had near identical rates of PE compared to cancer-matched controls (4% vs. 5%). Among patients with left sided lesions, CVA was more common compared to matched controls, although statistical differences between groups was not achieved in context of low clinical event rates (7% vs. 2%, p = 0.38). Sub-group analyses limited to intracavitary C MET demonstrated a stronger association between lesion location and embolic event rates: Among patients with right sided intracavitary C MET , PE occurred in over one fourth of cases-a rate more than fourfold higher than in matched controls (27% vs. 7%, p = 0.01). Two-thirds (8/12; 67%) of patients with right sided intracavitary C MET who developed PE had lesions graded as highly mobile on cine-CMR.
Notably, increased PE rates among patients with right sided C MET occurred despite frequent anticoagulation: Anticoagulant therapies (warfarin or direct oral anticoagulants) were more commonly utilized in patients with right sided C MET as compared to matched controls (33% vs. 15%, p = 0.02), but were equivalent when among C MET patients with left sided involvement and controls (16% vs. 21%, p = 0.63). Of note, 60% of patients with PE were on anticoagulation at the time of their clinical event (64% in C MET vs 50% in controls, p = 0.64).
Regarding impact of C MET tissue characteristics on embolic events, Table 3B demonstrates that rates of PE were similar between patients with heterogeneous and diffusely enhancing lesions (p = NS), as were rates of left sided embolic events. Figure 3 reports PE rates among patients stratified by both lesion location and tissue characteristics: As shown, PE rates were highest among patients with intracavitary right ventricular lesions (p < 0.001 vs. other groups), whereas partitioning based on lesion tissue characteristics did not stratify event risk   (p = NS). Table 4 demonstrates that increased PE risk among patients with right ventricular C MET was accompanied by impaired RV function, as evidenced by lower absolute RV ejection fraction (RVEF) and higher prevalence of RV dysfunction (both p < 0.05). Of note, among patients with RV C MET , RVEF was similar between those who had PE prior to, compared to those who had PE after CMR (51.0 ± 12.7% vs 53.5 ± 3.8%, p = 0.72)-consistent with the notion that event driven changes in RV systolic function were not responsible for observed associations between impaired RV function and PE. Data shown in Additional file 1: Table S1 tests both clinical and CMR parameters in relation to PE. As shown, both gastrointestinal cancer etiology and right sided C MET were each associated with increased likelihood of PE in univariate regression analysis, and each parameter remained associated with PE (p < 0.01) when the two parameters (gastrointestinal cancer etiology, right sided intracavitary C MET ) were tested together in an adjusted model.

Mortality
Follow-up was performed for a median duration of 0.8 years [IQR 0.3-1.67], during which a total of 145 deaths occured. Figure 4 provides Kaplan Meier survival curves for the overall cohort of C MET patients and cancer-matched controls, as well as for subgroups based on C MET tissue characteristics. As shown (a), mortality  Figure 4b demonstrates that mortality differed in relation to tissue characteristics of C MET : Whereas patients with diffusely enhancing C MET had near equivalent mortality to matched controls (p = 0.21), prognosis was worse among patients with heterogeneously enhancing C MET (p = 0.005)-including increased 6-month (44% vs. 26%) and 1 year (65% vs. 41%) mortality in respective casecontrol comparisons (both p < 0.05). As shown in Table 5

Discussion
To our knowledge, this is the first study to test LGE-CMR pattern of C MET as a prognostic marker in patients with systemic cancer, with focus on localized hypo-enhancement (a marker of tumor avascularity) as a novel marker of adverse prognosis. Results add to a growing body of literature by our group and others validating LGE-CMR in relation to histopathology and demonstrating clinical utility of this approach to guide diagnostic, prognostic, and therapeutic decision-making for patients with known or suspected cardiac masses. Key findings are as follows: First, among a broad cohort of advanced cancer patients, C MET contrast-enhancement pattern varied-prevalence of diffusely (47%) and heterogeneously enhancing (53%) lesions was near equivalent. Quantitative analyses demonstrated heterogeneously enhancing C MET to have Fig. 3 Pulmonary Embolism (PE) in Relation to C MET Location and Tissue Properties. Top: Clinically documented PE (within 6 months of CMR) among patients grouped based on presence and location of C MET . Note higher rate of PE among patients with C MET involving the right ventricle, with differences most marked in analysis limited to intra-cavitary lesions (p < 0.001). Bottom: Location-based comparisons of PE rates among patients with heterogeneous and diffusely enhancing C MET . Note equivalent rates of PE between patients grouped based on C MET contrast enhancement pattern more aggressive features, as evidenced by larger lesion size and lower SNR (as would be expected in context of tumor avascularity). Second, presence and distribution of C MET impacted likelihood of embolic events. Aggregate embolic events were higher among patients with C MET compared to cancer matched controls (21% vs. 8%, p = 0.006), C MET location modified likelihood of events: Whereas patients with right sided lesions had a threefold increase in PE (20% vs. 6%, p = 0.02), those with left sided lesions had near identical event rates to those of (C MET -) controls (4% vs. 5% p = 1.00). Embolic event rates did not vary in relation to C MET by tissue properties, as evidenced by equivalent rates of PE among patients with diffuse and heterogeneously enhancing right ventricular lesions. Third, mortality risk conferred by C MET varied in relation to contrast-enhancement pattern. During a median follow-up of 0.7 years [IQR 0.3-1.7], patients with and without diffusely enhancing C MET had equivalent mortality to controls (p = 0.21), whereas prognosis was worse among patients with heterogeneously enhancing C MET compared to controls matched for cancer etiology and stage (p = 0.005).
Our finding that heterogeneous lesion enhancement constitutes an adverse prognostic marker among patients with C MET is consistent with established concepts in tumor biology: Tumor necrosis-as would be expected to result in avascularity and thus impaired contrast uptake-is a known marker of aggressive phenotype: Uncontrolled oncogene driven proliferation of neoplastic cells exhausts oxygen supply from normal vasculature, resulting in localized hypoxia which upregulates production of angiogenic factors and triggers neovascularization. [15,16] However, vessels formed in response to hypoxia lack normal physiological angiogenesis-providing a nidus for chaotic tumor architecture and vascular leakiness. Hypoxia alters cancer metabolism to foster survival during stress and drive malignant progression, resulting in resistance to anti-cancer  [14,17,18]. Regarding embolic events, our data demonstrated C MET location to be strongly associated with outcomes-as evidenced by increased rates of PE among patients with RV intracavitary C MET . Our finding that embolic events were equivalent between patients with diffuse and heterogeneous C MET is not unexpected, given that avascular components were typically centrally located within lesions and would thus be unexpected to provide a nidus for embolization. Regarding mechanism, our data demonstrated patients with RV C MET to have lower RV systolic function than those with C MET in other locations (p < 0.05)possibly due to mechanical tumor effects or treatment related (i.e. radiation-induced) cardiac injury. Based on this, it is possible that localized stasis could contribute to development of super-imposed thrombosis on neoplastic lesions-providing a nidus for embolic events. Whereas heterogeneous enhancement (as would be expected in context of tumor with superimposed thrombus) was not identified as a risk factor for embolic events, it is possible that micro-thrombi developed prior to or following the time of CMR, or that spatial resolution of LGE-CMR was insufficient for diagnostic detection. It is also possible that emboli stemmed from tumor dislodgement rather than primary thrombotic processes or from insufficient anticoagulation-concepts supported by the fact that nearly two-thirds (63%) of C MET patients were on anticoagulation at the time of clinical events, as well as recent data showing high embolic event rates in non-cancer [10,19] and cancer populations [20] with cardiac thrombus treated with anticoagulants. Future research, including imaging using high resolution 3D LGE-CMR [21] and prospective trials testing relative efficacy of anticoagulant regimens or targeted resection in cancer patients with C MET , are necessary to further test these concepts.

Limitations
Several limitations should be noted. First, whereas our study encompassed a broad cohort of cancer patients undergoing CMR and clinical follow-up at two institutions, it should be recognized that embolic events were ascertained based on clinical documentation and/or diagnostic testing. In this context, it is likely that subtle clinical events were under-reported, or that clinical considerations may have influenced testing such that embolic Note that prognosis varied based on C MET tissue properties, as evidenced by equivalent mortality risk between diffusely enhancing C MET and controls (p = 0.21) but increased mortality for patients with heterogeneously enhancing lesions (p = 0.004) events such as stroke were under-diagnosed. It should be noted that our study was unable to reliably ascertain all potential cancer related clinical indices or derive aggregate classifications of disease chronicity and performance status. Thus, while our results demonstrate that location and contrast-enhancement pattern of LGE-CMR evidenced C MET impacts clinical outcomes, further research is warranted to test whether associations observed in the current study are modified by cancer etiology, treatment type, and/or overall health status. Second, whereas our study utilized LGE-CMR for tissue characterization of C MET , alternative approaches to assess vascularity such as quantitative LGE thresholds, perfusion, parametric mapping, or susceptibility weighted CMR were not tested. Whereas current findings demonstrate heterogeneous contrast uptake pattern within C MET to be a marker of increased mortality risk, knowledge gaps persist as to the relative utility of different imaging approaches to assess tumor necrosis or additional tissue properties such as hemorrhage and calcification. It is also important to recognize that whereas prior studies have validated LGE-CMR tissue characterization of masses in relation to pathology and other standards including metabolic imaging and outcomes, [3,4,10] lack of systematic biopsy sampling in the current cohort (for whom invasive cardiac tissue sampling was uncommon in context of advanced cancer) prohibited direct comparison of LGE-CMR enhancement pattern to pathology findings. An additional limitation relates to the relatively small number of embolic events (n = 33) in this study, which may explain the wide confidence intervals with respect to observed associations of C MET and GI cancer etiology with PE: In this context, current results should be considered more exploratory than definitive, thus highlighting the need to test them further in larger scale studies. Future research is also warranted to test whether alternative protocols using low dose contrast administration or non-contrast approaches (e.g. T1 mapping) provide equivalent diagnostic and prognostic utility in cancer patients with known or suspected C MET .

Conclusions
Location and contrast-enhancement pattern of C MET impact clinical outcomes, with RV lesion location associated with PE and heterogeneous enhancement conferring increased mortality. Given current findings, future research is warranted to test anticoagulant strategies in cancer populations, whether adverse prognosis conferred by heterogeneous lesion enhancement stems from accelerated tumor growth, and whether tailored therapiespaired to lesion tissue characteristics-improves clinical outcomes for cancer patients with C MET .