Abstract
Aims
Among cancer patients presenting with acute dyspnoea, the prevalence of acute heart failure (AHF), resource use and diagnostic accuracy of natriuretic peptides remain unknown. This study aimed to address these knowledge gaps.
Methods and results
Patients presenting with acute dyspnoea to the emergency department (ED) were prospectively enrolled in a multicentre diagnostic study. AHF was centrally adjudicated by two independent cardiologists based on current guidelines. B‐type natriuretic peptide (BNP) and N‐terminal proBNP (NT‐proBNP) concentrations were measured at ED presentation. Cancer status, resource use, and long‐term outcomes were prospectively assessed. Among 2153 patients, 473 (22.0%) had an active or past cancer. AHF was the most common final diagnosis in both cancer and non‐cancer patients (44.4% vs. 51.0%, p = 0.01). Among the alternative diagnoses, pneumonia and cancer‐related dyspnoea were more frequent in patients with cancer, while anxiety disorder/hyperventilation was frequent in patients without cancer. Hospitalization rate and length of hospital stay were both higher in cancer patients (p < 0.01). Among AHF‐related signs, rales and pleural effusion showed a significant interaction with cancer status and had lower diagnostic accuracy in cancer patients. The area under the curve (AUC) of NT‐proBNP was lower in cancer than in non‐cancer patients (0.89 vs. 0.93, p = 0.01), while that of BNP was similar (0.93 vs. 0.95, p = ns). This difference was mainly due to active cancers.
Conclusions
Acute heart failure was the most common diagnosis in cancer patients presenting with acute dyspnoea. Rales, pleural effusion, and NT‐proBNP had lower diagnostic accuracy versus patients without cancer, while that of BNP remained robust.
Keywords: Acute heart failure, Biomarkers, Cancer survivorship, Diagnosis, Natriuretic peptides
Introduction
Cancer and heart failure (HF) are leading causes of death and morbidity in both high and low/middle income countries. 1 , 2 , 3 , 4 Both disorders are the most frequent alternative cause of death in patients suffering from one of them and they each represent a common shared comorbidity. 1 , 2 , 3 , 4 Beyond sharing common risk factors such as ageing, obesity, cigarette smoking and diabetes, increasing evidence shows that cancer and HF are deeply entangled and the hallmark of this interaction is the ability to modify cellular homeostasis, leading to oxidative stress and metabolic dysregulation. 5 , 6 Moreover, cancer therapy including anthracyclines, immune checkpoint inhibitors and tyrosine kinase inhibitors can cause HF, even years after it was administered. 7 As populations age, HF prevalence rises, and cancer survival improves, the number of patients with a history of active or past cancer presenting to the emergency department (ED) with symptoms suggestive of acute HF (AHF) such as acute dyspnoea is expected to increase.
With more than 10 million patients presenting with acute dyspnoea to the ED worldwide, acute dyspnoea is one of the most common presenting symptoms. 8 Prompt and definitive identification of the cause of acute dyspnoea such as AHF, pneumonia, pulmonary embolism, obstructive pulmonary disease, anaemia, and anxiety disorder is paramount for the rapid initiation of effective cause‐specific and evidence‐based medical treatment and management. Patient's signs and symptoms, chest X‐ray, chest ultrasound, and natriuretic peptides (NPs, i.e. B‐type natriuretic peptide [BNP] and N‐terminal proBNP [NT‐proBNP]) are the cornerstone for the differential diagnoses of acute dyspnoea in the ED. 9 The early diagnosis of AHF may be particularly challenging in patients with cancer as NPs may be influenced by the high burden of comorbidities including chronic kidney disease. 10 , 11
Moreover, cancer and cancer therapies themselves can lead to a range of potential causes of dyspnoea, such as pleural effusion, pneumonia, anaemia and pulmonary embolism. Also, uncertainties persist regarding the diagnostic performance of BNP and NT‐proBNP for AHF in patients with cancer. 7 , 12
To bridge these gaps, the aims of this study were to evaluate (i) the prevalence of AHF in patients with active or past cancer presenting with acute dyspnoea to the ED, (ii) the diagnostic performance of BNP and NT‐proBNP in the early diagnosis of AHF, and (iii) the performance of guideline‐recommended algorithms for BNP and NT‐proBNP.
Methods
Study design and population
This is a secondary analysis within the Basics in Acute Shortness of breath EvaLuation (BASEL V) study, a prospective, multicentre diagnostic cohort study aiming to advance the early management of patients with acute dyspnoea (ClinicalTrials.gov, NCT01831115). 8 , 13 Adult patients who presented with acute dyspnoea to the EDs of two University Hospitals in Switzerland (Basel and Zürich) were enrolled. Patients were enrolled irrespective of their renal function at ED presentation, but those with end‐stage kidney disease on chronic dialysis were excluded. Additionally, patients with an unclear final diagnosis, even after central adjudication, or those who were adjudicated as having cardiac dyspnoea due to acute coronary syndrome or arrhythmia without any other evidence of AHF, were also excluded. 8 , 13
The study was conducted in accordance with the Declaration of Helsinki and approved by the local ethics committees. All participants provided written informed consent. Reporting is in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology statement (online supplementary Table S1 ). 14
Cancer‐related variables
Cancer status (present/absent and active/inactive) was prospectively collected. The following additional cancer‐related variables were retrospectively retrieved from the medical charts: primary site of origin, stage as advanced/localized, and administered chemotherapeutic regimens. Non‐melanoma skin cancers as well as in‐situ carcinomas and monoclonal gammopathy of undetermined significance were not considered as malignant cancer. Cancer patients were considered as those who had active or past cancer. Cancer was defined ‘active’ if diagnosed or treated within 6 months from index visit, if recurrent, regionally advanced or metastatic, or not in complete remission (in case of haematological cancers). 15 Patients with ‘cured’ cancer or in complete remission were classified as ‘past cancer’. Active cancers included those detected or highly suspected during index ED visit (e.g. rib metastases or unilateral pleural effusion at chest X‐ray, full blood count compatible with haematologic malignancy). Solid cancers were defined ‘advanced’ if at any timepoint the TNM staging was ≥T3 or ≥ N1 or ≥ M1. 16
B‐type natriuretic peptide measurement and algorithms
At ED presentation, blood samples were collected in EDTA tubes to measure the concentrations of NT‐proBNP (Elecsys proBNP assay, Roche Diagnostics AG, Rotkreuz, Switzerland) and BNP (Biosite/Alere Diagnostics Inc. assay, San Diego, CA, USA). 8 , 13 BNP levels <100 pg/ml ruled out AHF and BNP ≥400 pg/ml ruled in AHF. 17 NT‐proBNP levels <300 pg/ml ruled out AHF, while NT‐proBNP levels ≥450 pg/ml if age <55 years, ≥900 pg/ml if age 55–75 years, and ≥1800 pg/ml if age >75 years ruled in AHF.
Routine clinical care and resource use
Patients underwent routine work‐up during ED and hospital stay, irrespective of their participation in the study, including clinical history, physical examination, 12‐lead electrocardiogram, routine blood work‐up including BNP or NT‐proBNP, chest X‐ray, echocardiography, lung function tests and computed tomography. Timestamps of their ED and hospital stay were also collected. Length of stay in the ED was defined as the time from ED presentation to home discharge or hospital admission or death. Duration of hospitalization was defined as time from hospital admission to home discharge or death.
Final diagnosis adjudication
The final underlying cause of acute dyspnoea including AHF was centrally adjudicated by two independent cardiologists/internists reviewing all available medical records obtained during routine clinical as well as extensive prospective study‐specific documentation including clinical history, physical examination, 12‐lead electrocardiogram, laboratory findings, chest X‐ray, echocardiography, lung ultrasound, lung function testing, computed tomography, the response to therapy, and also autopsy data for patients who died in hospital. Laboratory findings included one of the natriuretic peptides (BNP in Basel and NT‐proBNP in Zürich) with class I recommendation in current guidelines. 9 As recommended, AHF was diagnosed if in addition to acute dyspnoea, patients had symptoms typical of HF including fatigue, ankle swelling, and signs typical of HF: pulmonary rales, pleural effusion, elevated jugular venous pressure, peripheral oedema, and objective evidence of a structural or functional cardiac abnormality (e.g. increased left atrial size, left ventricular hypertrophy, increased E/e', increased pulmonary artery pressures in echocardiography, or impaired left ventricular or right ventricular systolic function, and raised BNP or NT‐proBNP indicating increased intracardiac pressures). In case of disagreement, a third independent cardiologist was involved.
Statistical analysis
Categorical variables are reported as count (percentage), and groups were compared using the chi‐square test or Fisher exact test, as appropriate. Continuous variables are reported as median (Q1–Q3) and were compared using the Mann–Whitney U test. Positive likelihood ratios (LR+) with 95% confidence intervals (CI) were calculated to assess the value of signs and symptoms for the diagnosis of AHF stratified by cancer status. The p‐value for interaction was calculated to assess statistical significance. Standardized prevalence ratios were calculated as the ratio of observed to expected cases of each of the most frequent cancer types. 18 The number of expected cases was extracted from previous studies and corrected for the male to female ratio for sex‐related malignancies. 19 , 20
The diagnostic accuracy of BNP and NT‐proBNP for AHF according to cancer status was assessed using receiver‐operating characteristic curves and their subtended area under the curve (AUC). Confidence intervals for AUCs and p‐values for comparison were calculated with the DeLong's test. 21 Additionally, to assess discrimination, a multivariable model was constructed including age, sex and estimated glomerular filtration rate (eGFR). Subgroup analyses were planned a priori for age, sex, cancer and cardiac‐specific variables and time since cancer diagnosis. Patients without a valid measurement of NPs were excluded from this analysis.
To evaluate the performance of the guideline‐recommended BNP and NT‐proBNP‐algorithms, safety was assessed as the sensitivity and negative predictive value for rule‐out and accuracy as the specificity and positive predictive value for rule‐in. The diagnostic performance measures for the algorithms were compared using binomial exact test and the Pearson chi‐square test (95% CIs were calculated using the Agresti–Coull method). A subgroup analysis of diagnostic accuracy and performance considering patients with active versus past cancer was carried out.
No specific sample size calculation was carried out, because this was a secondary analysis of the BASEL V study.
All hypothesis testing was two‐tailed, and p‐values <0.05 were considered statistically significant. All statistical analyses were performed using R version 4.1.3 (R Foundation for Statistical Computing, Vienna, Austria). The R packages used are listed in online supplementary Appendix S1 .
Results
Baseline characteristics
Among 2153 patients, 473 (22.0%) had active or past cancer (online supplementary Figure S1 ). Cancer patients were older (median 76 vs. 73 years), presented more often with productive cough and less frequently with paroxysmal nocturnal dyspnoea, were more likely to be smokers and to suffer from chronic obstructive pulmonary disease and previous pulmonary embolism versus patients without cancer (Table 1 ). The prevalence of most clinical signs and renal function was comparable in patients with versus without cancer, while haemoglobin and albumin levels were lower in cancer patients. Patients with active cancer presented more often with productive cough and symptoms of low cardiac output, were more likely to be smokers and had less prevalence of heart and kidney disease (coronary artery disease, previous episodes of AHF, atrial fibrillation, higher eGFR) (online supplementary Table S2 ). Duration of hospitalization was greater in patients with active cancer (online supplementary Table S2 ).
Table 1.
Baseline characteristics of patients according to cancer status
| Characteristic | Overall (n = 2153) | No cancer (n = 1680) | Cancer (n = 473) | p‐value |
|---|---|---|---|---|
| Demographics | ||||
| Age (years) | 75.0 (61.0–82.0) | 73.0 (60.0–82.0) | 76.0 (68.0–82.0) | <0.001 |
| Male sex | 1202 (55.8) | 920 (54.8) | 282 (59.6) | 0.060 |
| Presenting symptoms | ||||
| Fever | 389 (19.2) | 306 (19.4) | 83 (18.7) | 0.74 |
| Night sweats | 446 (22.2) | 349 (22.3) | 97 (21.6) | 0.72 |
| Cough | 1271 (60.6) | 961 (59.1) | 310 (66.0) | 0.007 |
| Sputum production | 893 (42.6) | 666 (40.9) | 227 (48.4) | 0.004 |
| Weight gain | 465 (23.2) | 372 (24.0) | 93 (20.5) | 0.11 |
| Nocturia | 654 (48.2) | 504 (47.9) | 150 (49.2) | 0.70 |
| Orthopnoea | 994 (48.3) | 759 (47.4) | 235 (51.4) | 0.13 |
| Duration of dyspnoea (days) | 5.0 (2.0–14.0) | 5.0 (2.0–14.0) | 5.0 (2.0–14.0) | 0.22 |
| Paroxysmal nocturnal dyspnoea | 801 (39.1) | 642 (40.3) | 159 (35.0) | 0.042 |
| NYHA class | 0.20 | |||
| II | 286 (13.3) | 234 (13.9) | 52 (11.0) | |
| III | 950 (44.1) | 730 (43.5) | 220 (46.5) | |
| IV | 917 (42.6) | 716 (42.6) | 201 (42.5) | |
| Chest pain | 808 (37.7) | 642 (38.4) | 166 (35.1) | 0.19 |
| Low‐output symptoms a | 116 (10.9) | 89 (10.4) | 27 (12.8) | 0.31 |
| Cardiovascular risk factors and previous cardiac disease | ||||
| Arterial hypertension | 1451 (67.5) | 1127 (67.2) | 324 (68.6) | 0.54 |
| Diabetes mellitus | 493 (22.9) | 391 (23.3) | 102 (21.6) | 0.43 |
| Dyslipidaemia | 903 (42.3) | 713 (42.7) | 190 (40.9) | 0.49 |
| Active or past smoker | 1407 (66.8) | 1073 (65.5) | 334 (71.5) | 0.015 |
| Coronary artery disease | 718 (33.4) | 566 (33.7) | 152 (32.2) | 0.53 |
| Previous myocardial infarction | 403 (18.7) | 322 (19.2) | 81 (17.2) | 0.32 |
| Previous PCI | 319 (14.8) | 249 (14.8) | 70 (14.8) | >0.99 |
| Previous CABG | 203 (9.4) | 170 (10.1) | 33 (7.0) | 0.039 |
| Previous valve replacement | 107 (5.0) | 87 (5.2) | 20 (4.2) | 0.40 |
| Previous AHF | 685 (31.9) | 552 (32.9) | 133 (28.2) | 0.054 |
| Comorbidities | ||||
| Obstructive lung disease | 720 (33.4) | 540 (32.1) | 180 (38.1) | 0.016 |
| Peripheral obstructive arteriopathy | 256 (11.9) | 194 (11.5) | 62 (13.1) | 0.35 |
| Previous stroke | 242 (11.2) | 192 (11.4) | 50 (10.6) | 0.60 |
| Chronic kidney disease | 611 (28.4) | 477 (28.4) | 134 (28.3) | 0.97 |
| Liver disease | 207 (9.6) | 168 (10.0) | 39 (8.2) | 0.25 |
| Previous pulmonary embolism | 189 (8.8) | 130 (7.8) | 59 (12.5) | 0.001 |
| Previous pneumonia | 410 (19.1) | 307 (18.3) | 103 (21.9) | 0.081 |
| Psychiatric disease | 475 (22.1) | 379 (22.6) | 96 (20.3) | 0.29 |
| Vital signs at presentation | ||||
| SBP (mmHg) | 137.0 (121.0–155.0) | 137.0 (121.0–155.0) | 136.0 (119.0–153.2) | 0.19 |
| DBP (mmHg) | 80.0 (68.0–91.0) | 80.0 (68.0–92.0) | 77.0 (67.0–90.0) | 0.010 |
| Peripheral O2 saturation (%) | 96.0 (93.0–98.0) | 96.0 (93.0–98.0) | 95.0 (92.0–98.0) | 0.008 |
| Respiratory rate (bpm) | 20.0 (16.0–27.0) | 20.0 (16.0–26.0) | 20.0 (16.0–28.0) | 0.33 |
| Heart rate (bpm) | 90.0 (75.0–106.0) | 90.0 (75.0–107.0) | 90.0 (77.0–104.0) | 0.94 |
| BMI (kg/m2) | 25.9 (22.5–30.0) | 26.1 (22.7–30.1) | 24.8 (21.6–28.7) | <0.001 |
| Clinical signs at presentation | ||||
| Lung crackles/rales | 930 (44.3) | 714 (43.6) | 216 (47.0) | 0.20 |
| Wheezing | 500 (24.0) | 397 (24.4) | 103 (22.6) | 0.43 |
| Peripheral pitting oedema | 870 (40.8) | 681 (41.0) | 189 (40.2) | 0.77 |
| Electrocardiogram | ||||
| Sinus rhythm | 1367 (69.2) | 1056 (68.3) | 311 (72.7) | 0.081 |
| Atrial fibrillation | 467 (23.6) | 374 (24.2) | 93 (21.7) | 0.29 |
| Right/left bundle branch block | 357 (18.1) | 278 (18.0) | 79 (18.5) | 0.82 |
| ST‐segment elevation | 36 (1.8) | 30 (1.9) | 6 (1.4) | 0.46 |
| ST‐segment depression | 106 (5.4) | 82 (5.3) | 24 (5.6) | 0.80 |
| Negative T wave | 171 (8.7) | 132 (8.5) | 39 (9.1) | 0.71 |
| Selected laboratory values at presentation | ||||
| Creatinine (μmol/L) | 87.0 (68.2–119.0) | 87.0 (69.0–119.0) | 87.0 (68.0–116.0) | 0.69 |
| eGFR (ml/min) | 69.0 (44.9–89.3) | 69.5 (44.8–90.0) | 66.0 (44.9–87.8) | 0.31 |
| K+ (mmol/L) | 4.1 (3.8–4.4) | 4.1 (3.8–4.4) | 4.1 (3.8–4.4) | 0.82 |
| Na+ (mmol/L) | 138.0 (136.0–141.0) | 138.0 (136.0–141.0) | 138.0 (135.0–140.2) | 0.10 |
| Haemoglobin (g/L) | 133.0 (118.0–146.0) | 135.0 (121.0–147.0) | 126.0 (110.0–140.8) | <0.001 |
| Albumin (g/L) | 36.0 (32.0–38.0) | 36.0 (33.0–39.0) | 34.0 (31.0–38.0) | <0.001 |
Note: Values are given as median (interquartile range), or n (%).
Abbreviations: AHF, acute heart failure; BMI, body mass index; CABG, coronary artery bypass grafting; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; NYHA, New York Heart Association; PCI, percutaneous coronary intervention; SBP, systolic blood pressure.
Reduced peripheral perfusion, oligo/anuria.
Among the 473 cancer patients, 403 had solid malignancies and 70 had haematologic malignancies. The most frequent primary sites for solid malignancies were lung (n = 91), breast (n = 73), prostate (n = 60) and colon‐rectum (n = 26), while the most frequent haematologic malignancies were non‐Hodgkin lymphomas (n = 20), acute myeloid leukaemia (n = 10), multiple myeloma (n = 9) and myelodysplastic syndromes (n = 8). Patients with active cancer were 252 (53.3%), of whom 95 (20.1%) were metastatic. Forty‐eight (10.1%) patients were on active chemotherapy (cytotoxic or biological or hormonal) at study inclusion.
Final adjudicated diagnosis
Acute HF was the adjudicated final diagnosis in 210 (44.4%) patients with cancer vs 857 (51.0%) in patients without cancer, respectively (p = 0.01), being the most common single diagnosis in both groups. Cancer characteristics stratified by a final diagnosis of AHF are presented in online supplementary Table S3 . Cancer patients adjudicated to have AHF were older (79 vs. 73 years), had a longer time since cancer diagnosis (1329 vs 815 days), and had less often received chemotherapy as compared to cancer patients with other causes of acute dyspnoea. AHF as the final diagnosis in patients with active cancer was adjudicated in 37%, while an alternative diagnosis was adjudicated in 63% of them (most commonly cancer‐related dyspnoea [29%], pneumonia [16%], bronchitis [11%]).
Among all 1067 patients with an adjudicated final diagnosis of AHF, 210 (19.7%) had an active or past cancer. AHF subtypes were comparable in patients with versus without cancer (online supplementary Table S4 ). Among the alternative diagnoses, pneumonia and cancer‐related dyspnoea were more frequent in patients with cancer, while anxiety disorder/hyperventilation and other non‐cardiac causes were more frequent in patients without cancer. Of relevance, cancer‐related dyspnoea was the final adjudicated diagnosis in 15.2% of cancer patients (Figure 1 ). Physical stress and pulmonary disease as AHF triggers were more prevalent in patients with active cancer (online supplementary Table S5 ).
Figure 1.
Barplot of final adjudicated diagnoses stratified by cancer status. AHF, acute heart failure; COPD, chronic obstructive pulmonary disease.
Length of stay in the emergency department, disposition decision and duration of hospitalization
The length of stay in the ED was comparable in cancer versus non‐cancer patients, and 6.1 (4.3–9.2) h in the whole cohort (online supplementaryFigure S2 ). Eighty‐one percent of patients were hospitalized, more commonly if they had active or past cancer (85.6% vs. 79.2%, p = 0.002). In a multivariable logistic regression model adjusted for age, sex, eGFR and need for oxygen therapy at arrival, cancer status was associated with a higher probability of hospitalization in younger patients, but not in older patients (overall odds ratio 1.00 [95% CI 0.69–1.45], p for difference between cancer and no cancer = 0.98) (Figure 2A ). Duration of hospitalization was longer in cancer versus non‐cancer patients (12 [6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18] vs. 10 [6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16] days, p < 0.001). This trend was maintained after adjusting for age, sex, eGFR and need for oxygen therapy at arrival (difference 1.59 [95% CI −0.03 to 3.24] days, p for difference = 0.05) (Figure 2B ).
Figure 2.
(A) Hospitalization rate per age according to cancer status. (B) Duration of hospitalization per age according to cancer status. Both models are adjusted for age, sex, estimated glomerular filtration rate, and need for oxygen therapy at arrival.
Standardized prevalence ratio of cancer subtypes
The prevalence of lung cancer, acute myeloid leukaemia and myelodysplastic syndromes, assessed with the standardized prevalence ratio, was higher than expected, while the prevalence of colorectal, ovarian and thyroid cancer and myeloproliferative syndromes was lower than expected (online supplementary Figure S3 ).
Transthoracic echocardiography
Transthoracic echocardiography (TTE) was performed in 182 (38.5%) cancer patients versus 748 (44.5%) patients without cancer (p = 0.02). When adjusting for the final diagnosis of AHF, cancer status was no longer associated with the performance of TTE (odds ratio 0.84 [95% CI 0.67–1.05], p = 0.12). TTE variables are shown in online supplementary Table S6 . Left ventricular ejection fraction was higher in cancer versus non‐cancer patients (55% [41–60] vs. 52% [35–60], p = 0.01).
BNP and NT‐proBNP
In patients with a final diagnosis of AHF, NP levels were similar in both cancer and non‐cancer patients at ED presentation. In contrast, NT‐proBNP at hospital discharge was higher in AHF in cancer patients versus those without cancer (online supplementary Table S7 ). In patients with an alternative diagnosis, BNP and NT‐proBNP concentrations were higher in cancer than in non‐cancer patients, both at ED presentation as well as hospital discharge (online supplementary Table S8 ). C‐reactive protein levels were higher in cancer patients regardless of the final diagnosis.
Diagnostic accuracy of signs, symptoms, and B‐type natriuretic peptides for acute heart failure
Overall, the diagnostic accuracy for AHF of symptoms reported by the patients, quantified by the LR+ were comparable in patients with versus without cancer. In contrast, the diagnostic accuracy of two signs obtained by physical examination and chest X‐ray, rales and pleural effusion respectively, showed a significant interaction with cancer status and a lower LR+ in cancer patients (Figure 3 , online supplementary Table S9 ). This effect was even more evident in patients with active cancer (online supplementary Figure S4 ).
Figure 3.
Positive likelihood ratios (LR), and their 95% confidence intervals, of predefined signs and symptoms for the diagnosis of acute heart failure according to cancer status. P‐value for interaction was statistically significant for pleural effusion and crackles.
The diagnostic accuracy of BNP was very high and comparable in cancer and non‐cancer patients (AUC 0.93 [0.91–0.96] vs. 0.95 [0.94–0.96], p = 0.30) (Figure 4 ). This was confirmed in a multivariate logistic regression analysis adjusted for age, sex and eGFR. Nonetheless, accuracy of BNP was reduced in active cancers (AUC 0.90 [9.85–0.95]), with borderline significance (p = 0.06; online supplementary Figure S5 ). In contrast, the diagnostic accuracy of NT‐proBNP was high, but lower in cancer than in non‐cancer patients (AUC 0.89 [0.86–0.92] vs. 0.93 [0.92–0.95], p = 0.01) (Figure 4 ). This was confirmed in a multivariate logistic regression analysis adjusted for age, sex and eGFR (p = 0.05). The reduction in accuracy was mainly driven by patients with active cancer (online supplementary Figure S5 ).
Figure 4.
Receiver‐operating characteristic curves of N‐terminal pro‐B‐type natriuretic peptide (NT‐proBNP) and B‐type natriuretic peptide (BNP) for the diagnosis of acute heart failure according to cancer status. p NT‐proBNP = 0.01, p BNP = 0.30. AUC, area under the curve.
Performance of BNP and NT‐proBNP algorithms
The diagnostic performance of guideline‐recommended BNP and NT‐proBNP algorithms according to cancer status are shown in Figure 5 . The BNP algorithm maintained very high sensitivity (>97%) and specificity (>92%) in cancer patients, albeit with a lower efficacy (70.1% vs. 77.1%, p = 0.009). This difference was mainly due to active cancer patients (online supplementary Figure S6 ). In contrast, the NT‐proBNP algorithm showed reduced rule‐out proportion (p = 0.004), reduced (but still high) sensitivity and negative predictive value for AHF rule‐out (p = 0.04 and 0.02, respectively), an increased grey zone proportion (p < 0.001), a decreased specificity and positive predictive value for AHF rule‐in (p < 0.001 for both) in cancer patients. This difference was mainly due to active cancer patients (online supplementary Figure S7 ).
Figure 5.
Performance of guideline‐recommended diagnostic algorithms using (A) B‐type natriuretic peptide (BNP) or (B) N‐terminal pro‐B‐type natriuretic peptide (NT‐proBNP) for the diagnosis of acute heart failure (AHF). NPV, negative predictive value; PPV, positive predictive value; Sens., sensitivity: Spec., specificity.
Discussion
To the best of our knowledge, this was the first multicentre prospective diagnostic study centrally adjudicating the prevalence of AHF and the diagnostic performance of clinical signs and BNP/NT‐proBNP in cancer patients presenting to the ED with acute dyspnoea (Graphical Abstract). We report five major findings. First, AHF was the most common cause of acute dyspnoea in both cancer (44%, 37% in active cancer) and non‐cancer patients (51%). This has straightforward clinical implications for emergency physicians and cardiologists. High awareness for AHF as a possibly and highly prevalent cause of acute dyspnoea also in patients with cancer is mandatory to ensure that this important and well‐treatable differential diagnosis is not missed. Second, in this study, prevalence of cancer among patients presenting with acute dyspnoea was high with a striking 22%, and prevalence of cancer in those adjudicated to having AHF as the main cause of acute dyspnoea was 19.0%. This extends and corroborates recent observations derived from linking a national AHF cohort with national registries of recent diagnosis of common solid cancers (breast, prostate, colorectal, and lung cancer) in the UK, where the prevalence of cancer in hospitalized AHF patients was 5.9%. Active or past cancer of any kind is therefore a very common comorbidity in patients with AHF. Of note, the cause of acute dyspnoea in one out of seven cancer patients was related to cancer itself. Again, this finding has important clinical repercussions, suggesting clinicians investigate cancer‐related causes with second‐level tests (e.g. computed tomography of the chest) once AHF has been excluded. Third, overall resource use as quantified by the duration of hospitalization was higher in cancer versus non‐cancer patients (increase of about 2 days), even after multivariate adjustments. This observation is consistent with clinical experience, as cancer patients often require additional diagnostic and therapeutic interventions, including cancer‐specific procedures. This finding aligns with similar studies in cancer patients presenting with acute chest pain to the ED, highlighting the added complexity of managing this population. 18 Fourth, among signs and symptoms related to AHF, pleural effusion, crackles and possibly lower limb oedema had a lower likelihood for AHF in cancer patients. This observation is well in line with the fact that pleural effusion can be an exudate caused by cancer or cancer metastasis, crackles can reflect fluid spillage in hypoalbuminaemia or carcinomatous lymphangitis, and lower limb oedema being due to pelvic lymphadenopathies or hypoalbuminaemia. 22 Fifth, cancer patients with other non‐AHF causes of acute dyspnoea had moderately higher BNP concentrations (+ ≈50%) and substantially higher (+ ≈150%) NT‐proBNP concentrations versus those without, a result in line with a small previous retrospective analysis. 10 This translated into a small non‐significant decline in AUC for AHF of BNP, and a moderate significant decline in AUC of NT‐proBNP. This difference seemed mainly driven by active cancers. Further research is needed to reveal the pathophysiological basis for the stronger confounding effect of cancer for NT‐proBNP versus BNP. Contributors possibly include the stronger confounding effect of inflammation, atrial fibrillation, age, and chronic kidney disease. 23 , 24 , 25 As in this cohort the difference remained significant also after adjusting for age and renal function, cancer‐related inflammation may be the key driver behind the lower diagnostic accuracy of NT‐proBNP versus BNP in cancer patients. This hypothesis is supported by a pilot study showing a positive correlation between NT‐proBNP and C‐reactive protein levels and no correlation for BNP. 26 Based on these observations, known or still undiagnosed cancer should be considered a possible cause of unexpectedly high NT‐proBNP concentrations. Sixth, the decrease in diagnostic accuracy of NT‐proBNP also affected the performance of its current guideline‐recommended diagnostic algorithm. In cancer patients, the age‐adjusted NT‐proBNP algorithm, but not the BNP algorithm, had reduced rule‐out efficacy, reduced (but still high) sensitivity and negative predictive value for AHF rule‐out, an increased grey zone, a decreased specificity and positive predictive value for AHF rule‐in. This difference was particularly pronounced in patients with active cancer. These observations underscore the need for a more nuanced approach when interpreting NT‐proBNP levels in cancer patients and support the potential preference for BNP in this clinical context.
Limitations
These findings are specific to patients presenting with acute dyspnoea to the ED and may not be generalized to NP screening in asymptomatic patients or patients presenting to an outpatient (cardio‐oncology clinic). Second, despite the strict central diagnostic adjudication protocol, a few patients may still have been misclassified and patients in whom the presence or absence of AHF could not be ascertained after adjudication were excluded. Third, there could have been an underreporting of very old cancers, causing possible misclassification of patients' cancer status. The effect of this is expected to be small. Fourth, due to its observational nature, this study cannot quantify the medical and economic relevance of the lower diagnostic accuracy observed for NT‐proBNP. Future studies are warranted to assess these important aspects. Fifth, the unblinding to BNP and NT‐proBNP concentrations in the final adjudication process may have introduced incorporation bias, leading to raised sensitivity and specificity. 27 Finally, as patients with end‐stage kidney disease on chronic dialysis and patients presenting with dyspnoea as an anginal equivalent were excluded, we cannot comment on these patients.
Conclusions
Among patients presenting with acute dyspnoea to the ED, AHF was the most common diagnosis in cancer patients and patients without cancer. In cancer patients, the presence of rales and pleural effusion had lower diagnostic accuracy for AHF. Interestingly, also NT‐proBNP, but not BNP, had lower diagnostic accuracy for AHF in cancer patients. Accordingly, the guideline‐recommended diagnostic algorithm for NT‐proBNP showed worse performance in cancer patients versus patients without cancer, while the one for BNP maintained excellent performance also in cancer patients. These differences were especially evident in patients with active cancer.
Supporting information
Appendix S1. Supporting Information.
Acknowledgments
We are profoundly grateful to all staff involved in the study. Open access publishing facilitated by Universitat Basel, as part of the Wiley ‐ Universitat Basel agreement via the Consortium Of Swiss Academic Libraries.
Funding
This work was supported by research grants from the University of Basel, the University Hospital Basel, the Swiss National Science Foundation, the Swiss Heart Foundation, Critical Diagnostics, Abbott, Alere, BRAHMS, Roche, and Singulex. The sponsors had no role in the design and conduct of the study, collection, management, analysis, and interpretation of the data, and preparation, review, or approval of the manuscript.
Conflict of interest: P.B. has received a grant from the Swiss Heart Foundation related to the present work (FF23062). D.W. reports research grants from the Swiss National Science Foundation (Grant Reference P500PM_225285), the Swiss Heart Foundation (Grant Reference FF22112), the University Hospital Basel and the German Heart Foundation (Grant Reference K22/13) as well as speaker honoraria from PHC, outside the submitted work. P.L.A. has received research grants from the Swiss Heart Foundation (FF20079 and FF21103) and speaker honoraria from Quidel, paid to the institution, outside the submitted work. G.K. has received research support from the Swiss National Science Foundation, the Swiss Heart Foundation, the University Hospital Basel and the University of Basel, as well as speaker honoraria/consulting fees from Janssen and consulting fees from PAGE Therapeutics, all unrelated to this work and paid to the institution. F.M. has been supported by Deutsche Gesellschaft für Kardiologie, Deutsche Forschungsgemeinschaft (SFB TRR219, Project‐ID 322900939), and Deutsche Herzstiftung. Saarland University has received scientific support from Ablative Solutions, Medtronic and ReCor Medical. Until May 2024, F.M. received speaker honoraria/consulting fees from Ablative Solutions, Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Inari, Medtronic, Merck, ReCor Medical, Servier, and Terumo. T.B. received research grants from the Swiss National Science Foundation (PASMP3‐134362); the University Hospital Basel, Department of Internal Medicine; University Hospital Basel; Abbott, and Roche, as well as speaker honoraria from Roche. C.M. has received research support from the Swiss National Science Foundation, the Swiss Heart Foundation, the KTI, the University Hospital Basel, the University of Basel, Beckman Coulter, Biomerieux, Brahms, Mitsubishi, Novartis, Ortho Clinical, QuAcuidel, Roche, Siemens, Singulex, and Sphingotec, as well as speaker honoraria/consulting honoraria from AstraZeneca, Bayer, Boehringer Ingelheim, BMS, Osler, Novartis, Roche, Siemens, SpinChip, and Singulex, outside the submitted work, and all paid to the institution. All other authors have nothing to disclose.
Contributor Information
Paolo Bima, Email: desireenadine.wussler@usb.ch.
Christian Mueller, Email: christian.mueller@usb.ch.
References
- 1. Conrad N, Judge A, Tran J, Mohseni H, Hedgecott D, Crespillo AP, et al. Temporal trends and patterns in heart failure incidence: A population‐based study of 4 million individuals. Lancet 2018;391:572–580. 10.1016/s0140-6736(17)32520-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Moliner P, Lupón J, de Antonio M, Domingo M, Santiago‐Vacas E, Zamora E, et al. Trends in modes of death in heart failure over the last two decades: Less sudden death but cancer deaths on the rise. Eur J Heart Fail 2019;21:1259–1266. 10.1002/ejhf.1569 [DOI] [PubMed] [Google Scholar]
- 3. Arnold M, Rutherford MJ, Bardot A, Ferlay J, Andersson TML, Myklebust TÅ, et al. Progress in cancer survival, mortality, and incidence in seven high‐income countries 1995‐2014 (ICBP SURVMARK‐2): A population‐based study. Lancet Oncol 2019;20:1493–1505. 10.1016/S1470-2045(19)30456-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Edwards BK, Noone AM, Mariotto AB, Simard EP, Boscoe FP, Henley SJ, et al. Annual report to the Nation on the status of cancer, 1975‐2010, featuring prevalence of comorbidity and impact on survival among persons with lung, colorectal, breast, or prostate cancer. Cancer 2014;120:1290–1314. 10.1002/cncr.28509 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. de Wit S, Glen C, de Boer RA, Lang NN. Mechanisms shared between cancer, heart failure, and targeted anti‐cancer therapies. Cardiovasc Res 2023;118:3451–3466. 10.1093/cvr/cvac132 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Wilcox NS, Amit U, Reibel JB, Berlin E, Howell K, Ky B. Cardiovascular disease and cancer: Shared risk factors and mechanisms. Nat Rev Cardiol 2024;21:617–631. 10.1038/s41569-024-01017-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Gevaert SA, Halvorsen S, Sinnaeve PR, Sambola A, Gulati G, Lancellotti P, et al. Evaluation and management of cancer patients presenting with acute cardiovascular disease: A clinical consensus statement of the Acute CardioVascular Care Association (ACVC) and the ESC Council Of Cardio‐Oncology – part 2: Acute heart failure, acute myocardial diseases, acute venous thromboembolic diseases, and acute arrhythmias. Eur Heart J Acute Cardiovasc Care 2022;11:865–874. 10.1093/ehjacc/zuac107 [DOI] [PubMed] [Google Scholar]
- 8. Mueller C, Scholer A, Laule‐Kilian K, Martina B, Schindler C, Buser P, et al. Use of B‐type natriuretic peptide in the evaluation and management of acute dyspnea. N Engl J Med 2004;350:647–654. 10.1056/NEJMoa031681 [DOI] [PubMed] [Google Scholar]
- 9. McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, et al. 2021 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: Developed by the Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC). With the special contribution of the Heart Failure Association (HFA) of the ESC. Eur J Heart Fail 2022;24:4–131. 10.1002/ejhf.2333 [DOI] [PubMed] [Google Scholar]
- 10. Bando S, Soeki T, Matsuura T, Tobiume T, Ise T, Kusunose K, et al. Plasma brain natriuretic peptide levels are elevated in patients with cancer. PLoS One 2017;12:e0178607. 10.1371/journal.pone.0178607 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Anker MS, Sanz AP, Zamorano JL, Mehra MR, Butler J, Riess H, et al. Advanced cancer is also a heart failure syndrome: A hypothesis. Eur J Heart Fail 2021;23:140–144. 10.1002/ejhf.2071 [DOI] [PubMed] [Google Scholar]
- 12. Lyon AR, López‐Fernández T, Couch LS, Asteggiano R, Aznar MC, Bergler‐Klein J, et al. 2022 ESC Guidelines on cardio‐oncology developed in collaboration with the European Hematology Association (EHA), the European Society for Therapeutic Radiology and Oncology (ESTRO) and the International Cardio‐Oncology Society (IC‐OS). Eur Heart J 2022;43:4229–4361. 10.1093/eurheartj/ehac244 [DOI] [PubMed] [Google Scholar]
- 13. Kozhuharov N, Martin J, Wussler D, Lopez‐Ayala P, Belkin M, Strebel I, et al.; BASEL V Investigators . Clinical effect of obesity on N‐terminal pro‐B‐type natriuretic peptide cut‐off concentrations for the diagnosis of acute heart failure. Eur J Heart Fail 2022;24:1545–1554. 10.1002/ejhf.2618 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. J Clin Epidemiol 2008;61:344–349. 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
- 15. Khorana AA, Noble S, Lee AYY, Soff G, Meyer G, O'Connell C, et al. Role of direct oral anticoagulants in the treatment of cancer‐associated venous thromboembolism: Guidance from the SSC of the ISTH. J Thromb Haemost 2018;16:1891–1894. 10.1111/jth.14219 [DOI] [PubMed] [Google Scholar]
- 16. Amin MB, Edge SE, Greene FL, Byrd DR, Brookland RK, Washington MK, et al. AJCC cancer staging manual. 8th ed. New York, NY: Springer; 2017. [Google Scholar]
- 17. Tsutsui H, Albert NM, Coats AJS, Anker SD, Bayes‐Genis A, Butler J, et al. Natriuretic peptides: Role in the diagnosis and management of heart failure: a scientific statement from the Heart Failure Association of the European Society of Cardiology, Heart Failure Society of America and Japanese Heart Failure Society. Eur J Heart Fail 2023;25:616–631. 10.1002/ejhf.2848 [DOI] [PubMed] [Google Scholar]
- 18. Bima P, Lopez‐Ayala P, Koechlin L, Boeddinghaus J, Nestelberger T, Okamura B, et al.; APACE and TRAPID‐AMI Investigators . Chest pain in cancer patients: Prevalence of myocardial infarction and performance of high‐sensitivity cardiac troponins. JACC CardioOncol 2023;5:591–609. 10.1016/j.jaccao.2023.08.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2021;71:209–249. 10.3322/caac.21660 [DOI] [PubMed] [Google Scholar]
- 20. Li J, Smith A, Crouch S, Oliver S, Roman E. Estimating the prevalence of hematological malignancies and precursor conditions using data from Haematological Malignancy Research Network (HMRN). Cancer Causes Control 2016;27:1019–1026. 10.1007/s10552-016-0780-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. DeLong ER, DeLong DM, Clarke‐Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics 1988;44:837–845. [PubMed] [Google Scholar]
- 22. Gradalski T. Edema of advanced cancer: Prevalence, etiology, and conservative management – a single hospice cross‐sectional study. J Pain Symptom Manage 2019;57:311–318. 10.1016/j.jpainsymman.2018.11.005 [DOI] [PubMed] [Google Scholar]
- 23. Hall C. NT‐proBNP: The mechanism behind the marker. J Card Fail 2005;11:S81–S83. 10.1016/j.cardfail.2005.04.019 [DOI] [PubMed] [Google Scholar]
- 24. Rørth R, Jhund PS, Yilmaz MB, Kristensen SL, Welsh P, Desai AS, et al. Comparison of BNP and NT‐proBNP in patients with heart failure and reduced ejection fraction. Circ Heart Fail 2020;13:e006541. 10.1161/circheartfailure.119.006541 [DOI] [PubMed] [Google Scholar]
- 25. Farnsworth CW, Bailey AL, Jaffe AS, Scott MG. Diagnostic concordance between NT‐proBNP and BNP for suspected heart failure. Clin Biochem 2018;59:50–55. 10.1016/j.clinbiochem.2018.07.002 [DOI] [PubMed] [Google Scholar]
- 26. Jensen J, Ma LP, Fu ML, Fu MLX, Svaninger D, Lundberg P‐A, et al. Inflammation increases NT‐proBNP and the NT‐proBNP/BNP ratio. Clin Res Cardiol 2010;99:445–452. 10.1007/s00392-010-0140-z [DOI] [PubMed] [Google Scholar]
- 27. Kohn MA, Carpenter CR, Newman TB. Understanding the direction of bias in studies of diagnostic test accuracy. Acad Emerg Med 2013;20:1194–1206. 10.1111/acem.12255 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix S1. Supporting Information.