Primary cardiac lymphoma (PCL) represents a rare subset of non-Hodgkin lymphoma, characterized by poor outcomes. The authors aimed to construct a framework of known clinical presentations, diagnostic features, disease complications, treatment, and outcomes to improve prognostication.
Individual patient data were obtained from defined cases of PCL (1949-2009) and systematically analyzed.
The authors report results of a review of 197 cases of PCL, with half of all cases reported since 1995. Survival was affected by 4 factors: immune status, left ventricular involvement, presence of extra-cardiac disease, and arrhythmia. Median overall survival (OS) for immunocompromised and immunocompetent was 3.5 months (m) and not reached, respectively (HR 0.29, 95% CI, 0.13-0.68; P = .004). LV involvement was uncommon (26%) and associated with an OS of only 1m, whereas patients free of LV involvement had a median OS of 22m (HR 0.28, 95% CI, 0.12-0.64; P = .002). Patients with extracardiac disease had shorter median OS compared with those without (6m vs 22m, HR 0.49, 95% CI, 0.26-0.91; P = .02). Those patients with an arrhythmia of any type had a median OS that was not reached (n = 55), whereas those without rhythm disturbances (n = 41) had median OS of 6m (HR 0.51, 95% CI, 0.29-0.91; P = .024). Overall response rate to therapy was 84%, with long-term OS over 40%.
Cardiac involvement of disseminated lymphoma is common, with between 9% and 24% incidence having been reported on autopsy series,.1-3 Primary cardiac lymphoma (PCL), however, occurs with far lower frequency. Defined as extranodal lymphoma involving only the heart and/or pericardium, it was first assigned a distinct pathological categorization in 1978 on the basis of 4 cases that composed only 0.5% of a series of extranodal lymphomas.4 Applying the same definition, PCL was discovered in only 0.05% of a series of over 12,000 autopsies.5 In 1996, citing the well recognized difficulty in assessing the precise location of the lymphoid cell(s) at the time of incipient clonal expansion, the Armed Forces Institute of Pathology broadened the inclusion criteria to include those cases of lymphoma presenting with cardiac manifestations, particularly if the bulk of disease was found in the heart or pericardium.6-9
A review of primary malignant tumors of the heart published in 1949 attributes the first recognized case to a 1939 report in a German publication.10, 11 Since then, accumulated knowledge of PCL has been derived mainly from isolated case reports, both pre- and postmortem, although series with as many as 9 cases have been published.12
It has been recognized that the incidence of extranodal NHL increased during the latter part of the 20th century in both the United States and elsewhere.13, 14 This has been attributed to multiple possible factors, including improved diagnostic techniques, increased population of immunosuppressed patients, exposure to environmental toxins, and the advent of the human immunodeficiency virus (HIV).15 With the discovery of HIV and its role in NHL in the early 1980s, authors began to distinguish reports of PCL vis-a-vis immune status, observing that PCL was more common—and increasing in incidence—in the immunocompromised population.7, 16-18 Published reviews to date have encompassed up to 9 HIV-positive patients and 38-48 immunocompetent patients, where a more recent study reviewed 68 patients without regard to immune status.12, 19-21
We present here an analysis of 197 cases of primary cardiac lymphoma. We have reviewed the cases for multiple discreet variables in an effort to establish patterns to these disparate presentations, treatments, and outcomes to develop a rationale for prognostication in this disease.
MATERIALS AND METHODS
PubMed searches using the terms primary, extranodal, cardiac, and lymphoma were used to identify published, potentially eligible case reports and case series. Works cited by each of these reports were also evaluated to identify any cases that may have been missed. Articles were excluded if 1) the reported patients failed to meet the current definition of PCL, defined as cardiac involvement of non-Hodgkin lymphoma, present upon initial lymphoma diagnosis, without evidence of disease elsewhere as evidenced by a lack of tumor encountered after a) both computed tomography of the body and bone marrow biopsy, b) autopsy, or c) an extensive search as reported by the author(s); 2) if insufficient information was provided to make a judgment of the diagnosis; or 3) if the source article could not be located.
Analyses were performed using Excel (Microsoft Corp, Redmond, Wash), in which all patient data were tabulated. Variables sought included demographic information, immune status, clinical presentation, radiographic and other diagnostic characteristics, disease complications, therapy rendered, therapeutic response, survival, and cause of death. The year of estimated diagnosis for each patient was assessed according to when the respective report was submitted (or year from publication if submission date not provided), and subsequently adjusting for survival achieved at the time of report (eg, in the case of a patient with 5-year survival reported in 1995, the year of diagnosis was estimated to be 1990). Echocardiography results were assumed to be transthoracic, unless otherwise specified. Any response categorization provided (eg, complete response, refractory disease) was assumed to be accurate, even if supporting data were not provided. Where such categorization was not provided, either a conclusion was drawn from supporting data or, if these were lacking, it was considered unknown. Cases that met criteria for an unconfirmed complete response (CRu) were grouped with complete responses (CR). All statistical analyses and graphing were performed using Prism 5.0 (Graphpad Software, La Jolla, Calif).
For analyses of factors affecting survival times, patients were included only if they were diagnosed premortem and had reported survival data. Otherwise, unless specified, each percentage presented is of the entire 197-patient cohort.
Literature Search Results A total of 190 articles containing case reports were identified. Of these, 13 articles were excluded, as they were not available in English. Another 11 articles were excluded for the following reasons: a) reported patients failed to meet either definition of PCL (as provided above; 6 articles reporting on a total of 14 patients22-27); b) insufficient information was provided about a given patient to make such a judgment (2 articles reporting on 1 patient each2, 28); or c) the source article could not be located (3 articles with unknown number of case reports).29-31 This provided a total of 166 articles, reporting 197 unique cases, for inclusion in our analysis.
Of the 197 patients in our analysis, the male-to-female ratio was 1.94, with a median age of 63. Although patients ranged in age from 9 to 90 years old, only 5 were 17 years of age or younger. HIV status was not mentioned for 133 cases (68%), and of the remaining 64 cases, 26 were HIV-positive (41%). Global immune status was not mentioned in 119 cases (60%). Apart from HIV, 2 patients were otherwise reported as immunocompromised. Of these 2, 1 report does not provide the nature of the immunocompromise,32 whereas the other patient was on chronic corticosteroids and methotrexate for treatment of rheumatoid arthritis.33 Performance status was given in only 3 instances, making any assessment of this variable as a predictor of outcome (including as a component of the International Prognostic Index) impossible.20,34 Half of all cases included in our analysis were reported within the last 15 years.
Presenting complaints were provided for every patient included in our analysis (Table 1). Dyspnea was by far the most common presenting symptom, affecting 64%. Constitutional complaints and chest pain were the next 2 most common symptoms, at 26% and 24%, respectively. Ten patients presented with superior vena cava (SVC) syndrome. Of 178 patients for whom pertinent information was reported, 47% were in congestive heart failure.
Table 1. Presenting Symptoms and Clinical Findings in Primary Cardiac Lymphoma
In cases where the presenting heart rhythm was described, 56% (83 of 149) had an arrhythmia other than sinus tachycardia/bradycardia. The 2 most common abnormalities were atrial arrhythmias and atrioventricular (AV) block, at 23% and 22%, respectively. Interestingly, 20 of 33 reported cases of AV block (61%) were complete. In addition, 16 patients had either left or right bundle branch block, and 8 suffered ventricular arrhythmias.
In keeping with previous reports,35 right heart involvement was far more common than left heart in this population. A total of 179 patients had described chamber involvement (vs pericardial involvement only), of which 92% had either the right atrium (RA) or right ventricle (RV) involved. Only 7% of these patients had left-side involvement in the absence of right-side involvement. The order of rate of chamber involvement was RA > RV > left atrium (LA) > left ventricle (LV). Other commonly involved structures included the intra-atrial septum (41%) and the SVC (25%). Of 183 patients with appropriate information, 107 had pericardial effusions (58%), 53 had masses involving the pericardium (30%), and 38 patients had both (21%). Of those with a pericardial effusion, 34% demonstrated tamponade physiology.
The first cases using echocardiography36 and nuclear imaging37 were both reported for cardiac lymphoma cases in 1981. This may have dramatically improved the ability to detect PCL: 64% of cases (14 of 22) occurring before or during 1981 were diagnosed only at autopsy; since then, that figure has dropped to 15% (27 of 175). Only 28 cases (15%) reported neither transthoracic (TTE) nor transesophageal echocardiography (TEE) results, and 13 of these were reported before 1981. In our series, TTE had a sensitivity of 73% for right atrial involvement (n = 108) and 55% for right ventricular involvement (n = 86). Using TEE added to the findings of TTE in 19 of 36 cases (53%).
Tissue diagnoses were made in 43% with minimally invasive biopsy techniques 36% at surgery, leaving autopsy diagnoses accounting for the final 21%. Of those who underwent surgery, 48% had surgery performed on an urgent/emergent basis; 56% underwent partial or complete tumor resection at the time of surgery.
By far, the most common NHL subtype was diffuse large B-cell lymphoma (DLBCL), with 113 reports. Other reported subtypes constituted fewer than 10% each: 13 cases of Burkitt, 10 cases of T-cell lymphoma, 8 cases of small lymphocytic lymphoma, and 2 cases of plasmablastic lymphoma. The remainder (51 cases, 26%) remained unspecified by the authors. These data agree with the findings of previous authors, who noted DLBCL as the predominant histology since the early to mid-1980s,2, 38 irrespective of HIV status.
Tumor dimensions were reported in 81 instances, and, of these, 21 (26%) had disease greater than 7.5 cm and 7 (9%) greater than 10 cm. Immunohistochemistry (IHC) was consistent with the predominance of B-cell disease in those tested; only 3 cases (of 88 reported) were CD-20 negative, of which 2 were T-cell histology and 1 plasmablastic. Values for KI-67 were reported in 13 instances, with a median value of 80%. Just over 10% (10 of 98) of patients had bone marrow involvement, with over half of all reports not including this information.
It was observed that 112 of 163 patients (68%) met strict criteria for PCL (see Methods above), with the remainder of reports lacking sufficient information to make such a judgment. HIV status seemed to affect the likelihood of meeting these criteria: only 26% of HIV-positive patients (5 of 19) were free of extracardiac lymphoma (ie, met a “loose,” but not “strict” definition of PCL), whereas 76% of HIV-negative patients (28 of 37) and 74% of those with unknown HIV status (79 of 107) had disease limited to the heart and/or pericardium at the time of initial diagnosis.
These data were most remarkable for elevations in lactate dehydrogenase (LDH). Sixty-six reports included LDH values, of which 92% were above their respective reference ranges, with a mean of 770 IU/L (range 228-4820 for the elevated levels described). The exact LDH determinations of patients noted to have normal levels were not detailed.20, 39-42 Anemia affected 27 of 53 patients, with a mean hemoglobin of 11.4 g/dL. There were too few reports of levels of creatine phosphokinase (CPK), cardiac-specific enzymes, uric acid, or erythrocyte sedimentation rate (ESR) for analysis.
Of all patients included in the analysis, 57% had died by the time of reporting, and the cause of death was provided for 71% (80 of 113). The most common cause of death was heart failure, at 40%. This was followed by sepsis or other severe infection (26%) and progression of lymphoma (23%). Of the 21 cases in which infection was the cause of death, 12 patients (57%) had received chemotherapy, although the link between treatment and death from infection was generally not specified. As mentioned above, over half of all patients with PCL presented with an arrhythmia, which was fatal in 11% of such cases (9 of 83). Four patients presented with sudden cardiac death.23, 36, 43, 44
A total of 12 patients presented with embolic phenomena attributed to PCL, including 7 episodes of pulmonary embolism (PE) and 4 episodes of cerebrovascular accident (CVA). PE was fatal in 4 instances, CVA in 1. Although tamponade physiology was noted in 36 cases (see Presentation above), in only 1 of these cases was it described as fatal.45
A total of 142 patients were treated by any modality (72% of all patients, 91% of those diagnosed premortem). The most common modality was chemotherapy, which was administered to 89% of those treated. Surgical resection (either partial or complete) was undertaken in 28% and radiotherapy performed for 20%, with a median dose of 34.8 Gy (range = 15-59 Gy). A total of 23 patients received combined up-front chemoradiation, with another 2 patients having received both treatments asynchronously. Only 2 patients received radiation without any sort of systemic chemotherapy,46, 47 making a substantive comparison of combination therapy versus radiation alone impossible. High-dose chemotherapy, followed by autologous stem cell transplant or allogenic transplant, was given in 4 patients and 1 patient, respectively. Heart transplant was reported for another. Based upon the supposition that diagnostic rates improved with advances in imaging, a parallel analysis of treatment trends was undertaken. This analysis demonstrated that, before and during 1981 (when echocardiography was first reported as a diagnostic tool for a case of PCL), only 41% of patients (9 of 22) were treated by any modality. Since 1981, 79% of all patients (133 of 168) have received treatment (for 7 patients it was unclear whether treatment was given).
Anthracycline-containing regimens, primarily CHOP, were given to 89% (97 of 109) of those patients receiving chemotherapy. The use of rituximab first occurred in 2001 and, since that year, has been given to 38% of patients with non-T–cell histology (23 of 61). A median of 6 cycles of chemotherapy was administered to the 67 patients for whom such data were reported. One fatality from tumor lysis syndrome was reported.48 Taken together with the 12 patients who had received chemotherapy and developed sepsis, this is indicative of a treatment-related mortality (TRM) as high as 10% (13 of 126) for chemotherapy. Management of heart failure and arrhythmias arising as complications of PCL was mentioned in only isolated case reports; therefore, no such data were collected or analyzed.
Response and Survival Patterns
Of the 142 patients treated by any modality, 92 had a partial (PR) or complete response (CR) documented; 26 reports did not provide response data. This translates to an overall response rate (ORR) of 79% for cases with available data and a CR of 59%, which compares favorably to the 38% CR reported in a 2004 review of 40 cases.35 These proportions are lower than the reported 71% CR in a review of cases between 1989 and 2005, for which the total number of cases in that particular analysis was not provided.49 Refractory disease as a best response was documented in only 16 cases, and relapse following a CR in only 12. The median time to relapse was 7 months, with none reported beyond 12 months. Second-line therapy was rarely reported, with a 42% ORR (5 of 12) and 2 instances of CR.50, 51, Although surgical resection and radiation trended toward a positive impact upon survival, only chemotherapy made a statistically significant difference (30m vs 0.3m, P<.0001, Table 2). There was virtually no difference in survival patterns between those receiving chemotherapy alone and those receiving combined chemoradiation. This supports a previous assessment of lack of benefit associated with the addition of radiation therapy to chemotherapy in a smaller group of patients.20 No effect of histology on overall survival (OS) could be found; however, as noted, non-DLBCL histology was rare. Furthermore, subtypes of DLBCL (ie, germinal center vs nongerminal center) were not mentioned in any reports.
Table 2. Therapeutic Management and Overall Survival in Primary Cardiac Lymphoma
Type of Therapy
Median Overall Survival, Mo
N, number of patients; NR, not reached.
For those receiving combination chemotherapy and radiation, the comparator group is those receiving chemotherapy alone, as only 2 patients received radiation with no chemotherapy.
The median OS of those diagnosed antemortem was approximately 12 months (Fig. 1). In an effort to determine predictors of outcome, we analyzed survival patterns of the 128 patients diagnosed premortem, for whom survival data was reported. Among them, those who had extracardiac disease (n = 32) had a reported median OS of 6 months, whereas those without (ie, those who met strict criteria for PCL, n = 74) had a median OS of 22 months (Fig. 2, Panel A; HR 0.49, 95% CI, 0.26-0.91; log-rank P = .024), with all reports of survival >30 months (n = 13) coming in those without extracardiac disease. Median OS for those with immunocompromise (including HIV, n = 28) and immunocompetence (n = 50) were 3.5 months and not reached, respectively (Fig. 2, Panel B; HR 0.29, 95% CI: 0.13-0.68; log-rank P = .0042). The only chamber whose involvement was found to affect survival was the LV; the 19 patients with this chamber involved incurred a median OS of only 1 month, whereas the 103 patients free of LV involvement had a median OS of 22 months (Fig. 2, Panel C; HR 0.28, 95% CI: 0.12-0.64; log-rank P = .0024). Those patients with an arrhythmia of any type (as defined above in Methods) had a median OS of not reached (n = 55), whereas those without rhythm disturbances (n = 41) had median OS of 6 months (Fig. 2, Panel D; HR 0.51, 95% CI: 0.29-0.91; log-rank P = .024).
We have here presented the largest review of PCL to date. We present data that seem to indicate that PCL is now diagnosed with greater frequency, likely because of advances in imaging, although other factors—such as a larger population of HIV-positive or otherwise immunosuppressed patients with longer life expectancy52— may be contributing to this increase. In any case, we are unable to substantiate any claim that incidence is rising in HIV-positive or otherwise immunosuppressed populations. Our data demonstrate a fairly even distribution of 23 cases over the years 1981 to 1995 (between zero and 3 reports per year), followed by another 3 each in 2005, 2006, and 2007. Regardless of trends in overall incidence, it seems that PCL in these HIV patients, as compared with their immunocompetent counterparts, more often reflects only a portion of their lymphomatous involvement.
As noted, our analysis concurs with previous reports indicating that PCL favors the right side of the heart. The reason for this phenomenon remains unclear; however, a possible, novel, anatomic explanation might be the finding that the thoracic duct drains lymph into the SVC, and subsequently into the right atrium and ventricle, meaning the right heart would more readily and frequently be exposed to any pre-existing nodal lymphoma. This might also explain the finding that approximately a quarter of PCL patients have SVC involvement. However, if this hypothesis were correct, it would seem to indicate that PCL is more of an extranodal complication of what is presumed to be occult systemic (nodal) lymphoma, and supports the analysis above that chemotherapy offers optimal therapy for this lymphoma. This would further lead us to posit that PCL, rather than being primarily cardiac, is simply a manifestation of occult systemic lymphoma, thus stripping the “strict” definition of the disease of substantive meaning. It nevertheless seems warranted, at least currently, to maintain PCL as a diagnostic category. The cardiac manifestations would almost certainly affect aspects of care (eg, diagnosis, complications), compared with cases of systemic lymphoma without significant cardiac involvement. Moreover, survival rates would be measurably impacted (Fig. 2A).
In terms of predominance of presenting symptoms/findings, our results generally agree with previous reports (Table 1). An interesting exception is the much higher incidence of arrhythmia that we here report. The reason for this discrepancy is unclear, but may be the result of different inclusion criteria and/or sample size differences.20, 21, 35 We considered the possibility that the high rate of RA involvement might lead to frequent interference with the function of, or conduction from, the sinoatrial lymph node. However, the rates of arrhythmia were virtually identical: 58% in those with RA involvement (n = 101) and 59% in those without RA involvement (n = 39).
In our analysis of treatment outcomes, chemotherapy appeared to have the greatest effect upon survival, although it is understood there may have been a selection bias. Those diagnosed premortem and yet not receiving chemotherapy likely represented an extremely moribund subset of the population, with shorter survival times. Although a similar but weaker bias may have occurred with other treatment modalities, there was no significant difference in survival curves for radiation or surgical resection, compared with those not receiving these treatments. Our demonstration that radiation adds little to systemic chemotherapy in terms of OS is novel. The aggregate of these data regarding different modalities supports our hypothesis that PCL should be considered a systemic disease, for which treatment should always include chemotherapy.
We compared our treatment data and those of reported series of other extranodal lymphomas (Table 3), each of which, like our cohort of patients, was entirely or mainly composed of patients with DLBCL subtype. Chemotherapy was used more often in PCL than in any other subtype, whereas surgery and radiotherapy were used comparatively less; surgical rates were on par with those observed in primary CNS lymphoma, and rates of radiotherapy similar to those in gastric lymphoma. Though no formal statistical comparisons were undertaken, the overall and complete response rates in PCL seem to approach those seen in primary lymphomas of the breast, stomach, and testes, and exceed those in primary CNS lymphoma. Similarly, the OS rates are indicative of outcomes between primary CNS and the other 3 types of extranodal lymphoma. It also bears noting that multiple Kaplan-Meier survival curves for PCL (Figs. 1 and 2) point to subsets of PCL patients that are likely cured.
Table 3. Therapeutic Modalities for the Treatment of Extranodal Lymphomas
We have provided data in support of 4 adverse prognostic factors: immunocompromise, extracardiac disease, LV involvement, and the absence of arrhythmia. The first 2 come as no surprise; immunocompetent patients would be at an overall greater health risk, and subject to higher rates of extranodal lymphoma. The presence of LV involvement as a risk factor, although not necessarily expected, is not particularly surprising. It was subsequently noted that 77% of those patients with LV involvement also had right chamber involvement; those with LV disease may simply be a group with a higher overall burden of cardiac disease, and therefore at higher overall risk of mortality. It remains curious that those presenting with rhythm disturbances actually survived longer than those with normal rhythms, although there are at least 2 potential explanations: a) those with arrhythmias more often suffered symptoms that prompted early presentation; or b) the detection of arrhythmia upon presentation may have led to earlier imaging and/or treatment of the affected population.
We acknowledge that caution must be used in interpreting the presented data. Within the population of patients affected by PCL wide variations exist, such as the presence and nature of comorbidities, the sophistication of available diagnostic tools, and even treatment options. Moreover, when dealing with case reports as a source of data, there is a bias in favor of reporting those cases with unique features and a lack of standardization as to what features should or should not be reported.
With this in mind, we believe that our data raise some questions and point out what seem to be trends in PCL, but fail to definitively answer most issues. Our data indicate that arrhythmias are much more common in PCL than previously believed. This is particularly relevant in terms of further data we have provided indicating that the presence of arrhythmia may impact survival. We have shown that response rates approximate those seen in other types of extranodal lymphoma, but with inferior overall survival (with the exception of primary CNS lymphoma). Finally, we would like to note that PCL, though still rare, is more common than previously thought. By our analysis of existing reports, there have been as many as 14 cases in 1 year (in both 1990 and 2006), and an average of over 6 cases per year since 1996. There are certainly others that go unreported; we have been able to find an additional 3 cases of PCL treated at our own institution (not reported here) within the last 5 years. In light of the emerging role of gene expression profiling to sub-classify DLBCL and clarify prognosis,53, 54 comparison of PCL genotypic analyses might be informative. With these concerns in mind, it may be worthwhile to create a registry of PCL patients to better standardize available data and resolve questions regarding prognosis and appropriate management.