Pulmonary embolism (PE) is a complication of systemic disease in dogs. Antemortem diagnosis is challenging because of the lack of a confirmatory test.
Pulmonary embolism (PE) is a complication of systemic disease in dogs. Antemortem diagnosis is challenging because of the lack of a confirmatory test.
To retrospectively determine the diagnostic utility of D-dimer concentrations in dogs with necropsy-confirmed PE.
Ten dogs with PE confirmed at necropsy that had D-dimer concentrations measured and 10 control dogs with D-dimer concentrations available that lacked PE on necropsy.
The computerized medical record database was searched for dogs with necropsy-confirmed PE that had D-dimer concentrations measured at that visit. An age-, sex-, and breed-matched control group was identified. Signalment, location of PE, and coagulation profiles were collected. Sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) were calculated using a D-dimer concentration of 250 ng/mL.
Coagulation profiles were not different between dogs with and without PE. Using 250 ng/mL as a cut-off D-dimer concentration, the sensitivity and specificity were 80 and 30%, respectively, for the diagnosis of PE. The NPV and PPV were 60 and 53.0%, respectively. D-dimer concentration <103 ng/mL had 100% sensitivity for ruling out PE and no value was 100% specific.
D-dimer concentrations <250 ng/mL have a high sensitivity for the absence of PE, but PE still can occur in dogs with a normal D-dimer concentration. Increased D-dimer concentrations are not specific for PE.
negative predictive value
positive predictive value
activated partial thromboplastin time
disseminated intravascular coagulation
The incidence of pulmonary embolism (PE) in dogs is unknown. It has been reported in dogs with immune-mediated hemolytic anemia, sepsis, neoplasia, protein-losing nephropathies, hyperadrenocorticism, and cardiac diseases. Antemortem diagnosis is difficult because there is lack of a gold standard diagnostic test in veterinary medicine. Clinical signs are not specific for PE and mimic signs of other cardiopulmonary diseases. PE often is suspected in patients that develop respiratory difficulty in which no primary cardiac or respiratory disease can be confirmed, and when concurrent diseases predisposing to thromboembolic complications are present. Confirmation of PE rarely is obtained because of the need for advanced and invasive diagnostic tests such as ventilation:perfusion scanning or contrast computed tomography angiography.
D-dimer is a degradation product of cross-linked fibrin that can be increased with clot formation and fibrinolysis. Increases in D-dimer concentrations have been documented in dogs with DIC, thromboembolic disease, internal hemorrhage, neoplasia, renal disease, liver disease, and post-surgically. Therefore, increased D-dimer concentration alone is not useful for verifying clot formation in dogs with PE because of low specificity.[2, 3] In humans, a low D-dimer concentration has been used to exclude pulmonary embolism. In a group of 671 patients presenting to the emergency room, a D-dimer concentration of <500 ng/mL measured by ELISA had a diagnostic sensitivity of 99.5% and a negative predictive value (NPV) of 99% for PE using pulmonary angiography or lung scans as the gold standard. In another study evaluating human patients with suspected acute PE undergoing pulmonary angiography that had D-dimer concentrations <500 ng/mL measured by latex agglutination, sensitivity ranged from 97 to 100% and NPV from 94 to 100%, depending on the test kit used. Using a recently introduced 2-level clinical probability assessment scheme, a negative D-dimer result safely excluded PE in human patients that were clinically unlikely to have PE using a highly or moderately sensitive assay test. However, CT or pulmonary angiography still remain the gold standards for the exclusion of PE in human medicine.
In a group of 20 dogs with systemic and pulmonary thromboembolic disease, a D-dimer concentration of <500 ng/mL measured by latex agglutination had 100% sensitivity for the exclusion of thromboembolism. To the authors' knowledge, no studies have compared D-dimer concentrations in dogs with and without PE. The purpose of this study was to evaluate the diagnostic utility of D-dimer concentrations in dogs with necropsy-confirmed PE.
The electronic database of the UC Davis William R. Pritchard Veterinary Medical Teaching Hospital was searched from January 2004 to October 2012 for dogs with a necropsy diagnosis of pulmonary arterial thrombosis or embolism. Dogs were eligible for inclusion if they had necropsy-confirmed thrombosis or embolism in the pulmonary arterial circulation and a coagulation panel with D-dimer concentration measured during the visit in which they were euthanized or died. Pulmonary embolism was detected by either gross or microscopic examination of a hematoxylin and eosin-stained slide, or by histologic diagnosis only.
Medical records were evaluated for the presence of disseminated intravascular coagulation (DIC). A diagnosis of DIC was made on the following: thrombocytopenia (<150,000/μL) and fulfilling 2 of the following 3 criteria, a prothrombin time (PT) or activated partial thromboplastin time (aPTT) >50% of normal, a fibrinogen concentration below or above the reference interval, or an increase in D-dimer concentration. Any dog that fulfilled the criteria for DIC or had evidence of thrombosis in organs other than the lungs on necropsy was excluded.
A control group from the same time period was identified consisting of age, sex, and breed (if available) matched to the study group. The control group included dogs without pulmonary arterial thrombosis or infarction reported on necropsy, that did not have a diagnosis of DIC (see above), and with D-dimer concentration measured during the hospital visit at which the necropsy occurred within the same time period.
Medical records were reviewed for the collection of signalment, location, and distribution of PE, PT,1,2 aPTT,1,2 fibrinogen,1,2 and D-dimer concentrations.1,3 Two different assay methodologies were used during this time period for D-dimer measurements. One was latex agglutination2 which reports D-dimer as a range, and the midpoint of the range was used for statistical analysis. The other methodology1 is immunoturbidimetric and provides continuous results for D-dimer concentrations.
Data were assessed for normality using the Shapiro-Wilk normality test. Normally distributed data are expressed as mean and standard deviation; data not normally distributed are expressed as median and range. Descriptive statistics were performed using a commercially available software product.4 A 2-tailed Mann–Whitney U-test was used to compare continuous data. Fisher's exact test, sensitivity, specificity, NPV, and positive predictive value (PPV) were calculated for D-dimer concentration using the high end of the normal reference range as a cut-off value (250 ng/mL). P < .05 was considered statistically significant.
Ten dogs fit the inclusion criteria. Six dogs subsequently were excluded because of a concurrent diagnosis of DIC with 10 dogs remaining for analysis. The mean age was 7.9 ± 3.9 years and mean body weight was 20.8 ± 9.3 kg. There were 2 mixed breed dogs and other breeds were represented only once. There were 1 female intact, 4 female spayed, 2 male intact, and 3 male castrated dogs. The control population had a mean age 7.8 ± 3.7 years and mean body weight of 21.0 ± 15.7 kg. There were 2 mixed breed dogs and other breeds were represented only once. The sex and neuter status distribution was identical to the study group. Underlying diseases in the study group included neoplasia in 4 dogs, renal disease in 2, sepsis in 1, cardiac disease in 1, immune-mediated hemolytic anemia in 1, and an indeterminate disease process in 1 dog. In the 10 dogs of the control group, underlying diseases identified included neoplasia in 3 dogs, renal disease with protein-losing nephropathies in 2, seizures in 2, sepsis in 1, gall bladder rupture with bile peritonitis in 1, and pulmonary hypertension from chronic bronchitis in 1. Systemic thromboembolic disease was identified in 2 control dogs; 1 dog with seizures had evidence of cerebral cortex infarction with reperfusion and had an elevated D-dimer concentration, and 1 dog with sepsis had an arterial thrombus in an external ileac artery and had a normal D-dimer concentration.
Pulmonary embolization was reported as diffuse or involving multiple lung lobes in 9/10 (90%) dogs, and only 1 lung lobe was affected in 1/10 (10%). Embolism was histologically interpreted by a pathologist as both acute and chronic in 1 dog, acute in 1 dog, and chronic in 1 dog. The size of the pulmonary arterial segment embolized was reported in 6/10 dogs, with large arteries involved in 3/6 and small arteries in 3/6. The dog with only 1 lung lobe affected had small pulmonary artery emboli. Two dogs had capillary thrombosis only. Three dogs had either the main pulmonary artery or the first branch of the main pulmonary artery embolized.
Median PT, aPTT, fibrinogen, and D-dimer concentrations for dogs with and without PE are summarized in Table 1. There were no statistically significant differences in measured parameters between groups. For the PE group, 2/10 D-dimer concentrations were measured with the immunoturbidimetric method, whereas 3/10 in the control group were measured using this method.
|Test (Reference Interval)||Dogs with Pulmonary Embolism||Dogs without Pulmonary Embolism|
|Prothrombin time (7.5–10.5 seconds)||8.5 (6.6–11.7)||9.2 (7.3–16.4)|
|Partial thromboplastin time (9.0–12.0 seconds)||14.3 (9.4–20.4)||13.9 (11.2–26.4)|
|Fibrinogen (90–255 mg/dL)||427 (53–1640)||341 (127–830)|
|D-Dimers (0–250 ng/mL)||750 (142–>2000)||562 (55–>2000)|
Using the upper end of the reference interval (250 ng/mL) to define increased D-dimer concentration, the sensitivity and specificity (95% CI) of this test for PE were 80.0% (44.3–97.5%) and 30.0% (10.6–65.3%), respectively. The NPV and PPV (95% CI) were 60.0% (14.7–94.7%) and 53.3% (26.9–78.7%), respectively. There was no significant difference in the proportion of patients with increased D-dimer concentration >250 ng/mL between the group with a PE and the group without PE using Fisher's exact test (P = 1.0). A D-dimer value of <103 ng/mL was required to produce a 100% sensitivity for the diagnosis of PE.
Results of this study demonstrate that the D-dimer test is highly sensitive with a moderate NPV, but it is poorly specific for the presence of PE in dogs. In this study, the sensitivity and NPV of D-dimer concentration for excluding PE were lower than what has been found in human medicine (94–100%).[4, 5, 8] This difference may be explained by the fact that the studies in humans used pulmonary angiography instead of pathology as the gold standard. To directly compare this population to humans, pulmonary angiography ideally would have been performed.
A previous prospective study reported 100% sensitivity and 70% specificity for thromboembolic disease when D-dimer concentration exceeded 500 ng/mL. In contrast, using a cut-off of 250 ng/mL, we found only 80% sensitivity, whereas a cut-off of < 103 ng/mL was required to replicate 100% sensitivity. One potential reason for the difference between the 2 studies is that the previous study included dogs with both pulmonary and systemic embolism, whereas this study examined PE alone. In the earlier study, thromboembolism was confirmed at necropsy in only 15 of the 20 dogs, whereas the 5 remaining dogs had a clinical diagnosis of thromboembolism. Additionally, all patients in the thromboembolic group were clinically ill from thromboembolic disease, which might increase the likelihood that an acute increase in D-dimer concentration would be detected. Clinical presentation was not assessed in this study, and emboli were suspected to be chronic in some dogs, which could alter the concentrations of D-dimer detected. Diseases associated with PE in this study were consistent with what has been reported previously in dogs and humans.
Two dogs in this study with PE had a D-dimer concentration within the reference interval (<250 ng/mL). One dog had a small focal pulmonary artery thrombus, whereas the other had severe diffuse chronic thrombosis. The focal thrombus in this dog may not have been large enough to cause enough D-dimer release into the systemic circulation to increase the circulating concentration. Most of the research on D-dimer as a predictive test for PE has been performed in humans who are showing respiratory clinical signs. In this dog, PE may have been an incidental finding unrelated to death as the dog was not in respiratory distress at the time of euthanasia. The dog with severe chronic thrombosis and D-dimer concentration in the reference range likely did not have active fibrinolysis resulting in degradation of cross-linked fibrin and generation of D-dimer. In experimentally induced canine pulmonary thromboembolism, D-dimer concentrations fall to <250 ng/mL within 48 hours post-embolism. In humans with chronic thromboembolic pulmonary hypertension, D-dimer concentration is much less sensitive (37%) than in acute disease, illustrating the potential for chronic thrombosis to result in normal D-dimer concentrations.
Embolism location affects the sensitivity of the D-dimer assay for the diagnosis of PE in people, with subsegmental emboli resulting in lower D-dimer concentrations than larger artery emboli. Two of 3 dogs in this study with large branch PE had increased D-dimer concentrations, and 2 of 3 dogs with PE in a lobar or sublobar pulmonary artery also had increased D-dimer concentrations. Thus, in this small group of dogs, embolism location did not appear to affect the sensitivity of D-dimer concentrations for PE.
One of the strengths of this study was confirmation of PE by necropsy, but there were several limitations inherent to a retrospective study. Although the control group had necropsies performed, there is a chance that PE was not identified during the examination. Dissolution of 50% of the volume of PE within 3 hours of death has been reported in dogs because of high concentrations of plasminogen activator activity in canine platelets and pulmonary artery cells.[13, 14] Because of this and the delay from death or euthanasia until necropsy, antemortem PE in the control group could not be excluded. Additionally, in the control group systemic thromboembolic disease was present in 2 dogs, but only one had an increased D-dimer concentration. The combination of these factors could have affected the sensitivity and specificity reported here.
When D-dimer was first measured at our institution, a range was reported rather than an individual number because a latex agglutination assay was used. Later in the study period, an immunoturbidimetric assay was used providing a single concentration. The immunoturbidimetric assay has been reported to be a more sensitive assay, but the 2 methods have been shown to have good agreement in dogs. The use of the midpoint of the range and the 2 different methodologies used are potential confounding factors, but no result reported overlapped the normal and increased reference interval. Finally, a relatively small number of dogs fit the inclusion criteria. Large prospective studies using an antemortem diagnostic gold standard are needed to further investigate the role of D-dimer in PE. In these studies, positive and negative likelihood ratios for D-dimer could be identified in a group of dogs in which clinically suspected PE is confirmed antemortem, as has been performed in human medicine.
In this study and in human medicine, a low D-dimer concentration does not completely exclude the possibility of PE.[8, 16] Low concentrations should be interpreted in light of the index of suspicion for PE as well as in the context of clinical signs (acute or chronic), thoracic radiography results, and arterial blood gas analysis. Using probability assessment of the likelihood of PE in veterinary patients as is currently performed in human medicine likely would enhance recognition of this disorder and improve diagnostic ability.
This study was not supported by a grant. This study has not been presented at a meeting
Conflict of Interest Declaration: Authors disclose no conflict of interest.
STA-compact, Diagnostic Stago, Parsippany, NJ
Pacific Hemostasis, Thermo Scientific, Lafayette, CO
BBL Fibrometer, Becton, Dickinson and Company, Franklin Lakes, NJ
GraphPad Prism 6.0, Graph Pad Software, La Jolla, CA