Evidence-Based Imaging - part 8 doc

However, it is not clear whether anticoagulation was withheld in patients with low-probability scans in this study. One systematic review by van Beek et al. (13) reported negative and positive predictive values of 99.7% and 88%, respectively. C. Modality 3: Computed Tomography Pulmonary Angiography (Scanners with Fewer than Four Detectors) Summary of Evidence: Computed tomography pulmonary angiography (CTPA) is increasingly being used for the diagnosis of PE. Level I (strong evidence) studies using a clinical outcome reference standard find rates of PE recurrence to be 0% to 6%, with negative predictive values of 94% to 100%. Studies using a conventional pulmonary angiography reference standard find broad variations in sensitivities, specificities, and positive and negative predictive values, likely due to variations in detection of sub- segmental emboli. Despite these variations, there is strong evidence to show that it is safe to withhold anticoagulation in patients with negative CTPA. Supporting Evidence: A systematic review of all published literature from 1966 to 2003 (Eng et al., in press) identified eight primary prospective levels I to II (strong to moderate evidence) studies in which all subjects under- went both CTPA and conventional angiography, the latter being con- sidered the reference standard. Among the eight primary studies, the sensitivities ranged from 45% to 100%, and specificities ranged from 78% to 100%. Nine major studies were found in our search that evaluated the nega- tive predictive value of CTPA using clinical outcomes. One prospective level I (strong evidence) study with a total PE prevalence of 25% followed 378 patients with negative CTPA for 3 months (14). No patients were lost to follow-up, none were anticoagulated during the follow-up period, and no patients were excluded for other reasons. Four of the 378 patients devel- oped PE (recurrence rate = 1%, negative predictive value = 99%). In all studies, recurrence rates ranged from 0% to 6%, and negative predictive values ranged from 94% to 100%. The study with the highest recurrence rate and lowest negative predictive value (level II) followed 81 hospital- ized patients from cardiology and pulmonary wards with a PE prevalence of 38%, a majority of whom (82%) had underlying cardiorespiratory disease (15). D. Modality 4: Multidetector Computed Tomography Summary of Evidence: Multidetector computed tomography, with higher image acquisition rates than non-MDCT scanners, reduces the rate of res- piratory and motion artifacts, particularly in sections obtained during the end of the scan when patients may not be able to maintain apnea, and improves overall spatial resolution. Limited evidence in clinical outcome studies demonstrates that the recurrence rate in patients with MDCT findings negative for PE is 1%, with a negative predictive value of 99%. Although definitive evidence is still forthcoming, it is reasonable to assume the performance of MDCT is at least as good as that of non-MDCT. It is safe, therefore, to withhold anticoagulation in patients with negative MDCT findings. 404 K. Juluru and J. Eng Supporting Evidence: There have been no major direct comparisons of con- ventional CTPA with MDCT. While we expect MDCT to be more sensitive for clots, negative predictive values cannot be much improved beyond the 94% to 100% achievable by conventional CTPA with clinical outcome as a reference standard. It is possible that subsegmental clots missed by con- ventional CTPA may have no clinical significance. The benefit of MDCT over non-MDCT appears to be the reduction in the number of patients with inconclusive scan results. Two prospective level III (limited evidence) studies were identified in our search evaluating MDCT against a clinical outcome reference standard (16,17). The studies evaluated a total of 236 patients, with PE prevalence of 18% to 19%. Patients were referred for MDCT scanning by clinicians who also had the option to choose other imaging modalities (e.g., nuclear imaging), thus introducing potential selection bias. Both studies reported a PE recurrence rate of 1% and negative predictive value of 99%. In com- parison to non-MDCT scan, MDCT scans had fewer respiratory and cardiac motion artifacts, higher rates of interpretation down to subseg- mental arterial levels, and fewer inconclusive results (17). In our search, there were no major systematic reviews of MDCT or arti- cles that evaluated MDCT against an imaging reference standard. A mul- ticenter clinical trial, the PIOPED II sponsored by the National Heart, Lung, and Blood Institute, is currently obtaining data to assess the efficacy of mul- tidetector CT (among other tests) in patients suspected of having acute PE (18). E. Modality 5: Electron Beam Computed Tomography Summary of Evidence: Electron beam computed tomography has under- gone limited evaluation in the detection of PE, probably because this tech- nology is not widely available. One major level I (strong evidence) study using clinical outcome as a reference standard has shown that it is safe to withhold anticoagulation in patients with negative EBCT findings. When using conventional pulmonary angiography as a reference standard, EBCT has sensitivities and specificities similar to those of CTPA. Supporting Evidence: A level I (strong evidence) study by Swensen et al. (19) evaluated 993 patients with a PE prevalence of 34% who had negative EBCT findings and were not anticoagulated. At 3-month follow-up, seven patients developed PE or died from PE. No history was available in 19 patients who were known to have lived by the 3-month follow-up period. Recurrence of PE therefore ranged from 0.7% (7/993) to 2.6% [(7 + 19)/993]. One major level III (limited evidence) study evaluated EBCT against a pulmonary angiography reference standard (20). Sixty consecutive patients who had already been referred for conventional pulmonary angiography were imaged with EBCT. In this population with a PE prevalence of 38%, the sensitivity, specificity, and positive predictive and negative predictive values were 65%, 97%, 93%, and 82%, respectively. There have been no major systematic reviews evaluating EBCT. F. Modality 6: Magnetic Resonance Angiography Summary of Evidence: Magnetic resonance angiography has undergone limited evaluation, predominantly in populations referred for conventional Chapter 22 Imaging in the Evaluation of Pulmonary Embolism 405 pulmonary angiography. There is incomplete evidence to suggest that MRA can be used as the primary imaging modality in the evaluation of PE. Supporting Evidence: In four level III (limited evidence) studies, patients were selected from a population referred for conventional pulmonary angiography. The oldest study in 1994 (21), with a PE prevalence of 52%, reported problems with identification of pulmonary emboli at the seg- mental levels. Sensitivity and specificity of MRA in this study were 83% and 100%, respectively. In the remaining three studies performed between 1997 and 2002 (22–24) with PE prevalences ranging from 25% to 36%, prob- lems with identification of PE occurred mostly at the subsegmental levels. Sensitivities and specificities ranged from 77% to 100% and 95% to 98%, respectively. All three studies had at least two readers with interobserver agreement ranging from 57% to 91%, with lower values again noted mostly at subsegmental levels. In our search, there were no major systematic reviews of MRA, and there were no major studies that evaluated MRA against a clinical outcome reference standard. G. Modality 7: Ultrasound of Lung and Pleura Summary of Evidence: Evidence on the use of transthoracic ultrasound imaging of the lung and pleura to diagnose PE is limited. The available data show that this method does not have adequate sensitivity or speci- ficity for the detection sof PE. Supporting Evidence: One major level II (moderate evidence) study was identified in our search that used ultrasound imaging of the lung and pleura for evaluation of suspected PE against an imaging reference stan- dard (25). Ultrasound diagnosis of PE was made by the identification of (1) wedge-shaped hypoechoic homogeneous pleural-based lesions, or (2) sharply outlined pleural-based lesions with central hyperechoic reflection. Final diagnosis was established by a combination of nuclear lung imaging, clinical probability, CT, lower extremity Doppler ultrasound, and conven- tional angiography. In this study with a PE prevalence of 42%, the sensi- tivity, specificity, positive predictive, and negative predictive values were 71%, 77%, 69%, and 79%, respectively. There were no major systematic reviews of ultrasound in our search, and there were no major studies that evaluated ultrasound against a clinical outcome reference. H. Method 8: Echocardiography Summary of Evidence: Studies on the use of transthoracic echocardiogra- phy (TTE) have employed various criteria in the evaluation of pulmonary embolism. These include tricuspid regurgitation, right ventricular dilata- tion, right ventricular dyskinesis, right-sided cardiac thrombus, and flat- tening of the interventricular septum. Combinations of these criteria have yielded inadequate sensitivities and variable specificities. Data on the effectiveness of transesophageal echocardiography (TEE) for direct pulmonary thrombus visualization is limited, and this modality also suffers from poor sensitivity and specificity. The limited data on both TTE and TEE show that both modalities are inadequate as a primary imaging modality in the evaluation of PE. 406 K. Juluru and J. Eng Supporting Evidence: Two level II studies (moderate evidence) and one level III (limited evidence) study were identified in our search that utilized TTE for the diagnosis of PE against an imaging reference standard (26–28). Miniati et al. (28) studied a group of 110 patients with a PE prevalence of 39%. All patients had TTE followed by VQ scan. Conventional angiogra- phy was performed when the VQ scan was not normal. Echocardiographic criteria for diagnosis of PE included enlarged right ventricle, tricuspid regurgitation, or right ventricular hypokinesis. Sensitivity, specificity, and positive and negative predictive values were 56%, 90%, 77%, and 76%, respectively. Sensitivities in the other studies ranged from 19% to 52%, and specificities from 87% to 100%. A level III (limited evidence) study by Steiner et al. (29) utilized TEE and TTE for diagnosis of PE in 35 patients with a PE prevalence of 63%, using helical CT as a reference standard. Pulmonary embolism was diagnosed by visualization of thrombus in the main pulmonary artery, dilatation of the right ventricle or pulmonary artery, tricuspid regurgitation, or ab- normal motion of the interventricular septum. Sensitivity, specificity, and positive and negative predictive values were 59%, 77%, 81%, and 53%, respectively. There were no major systematic reviews of echocardiography in our search, and there were no major studies that evaluated echocardiography against a clinical outcome reference standard. I. Modality 9: Chest Radiography Summary of Evidence: There is limited evidence on the use of chest radiog- raphy in the evaluation of PE. Various chest radiographic findings are asso- ciated with poor sensitivity and only modest specificity. Chest radiography should not be the primary modality in PE evaluation. Supporting Evidence: In one major level II (moderate evidence) study by Worsley et al. (30), 1063 patients from the PIOPED group who underwent both diagnostic angiography and chest radiography were retrospectively evaluated. Radiographic signs evaluated included prominent central artery (Fleischner sign), enlarged hilum, enlarged mediastinum, pul- monary edema, chronic obstructive pulmonary disease (COPD), oligemia (Westermark sign), vascular redistribution, pleural-based areas of increased opacity (Hampton hump), pleural effusion, and elevated diaphragm. The highest sensitivity obtained was 36% for pleural effusion, and the highest specificity obtained was 96% for COPD. Combinations of the signs were not assessed. There were no major systematic reviews of chest radiography in our search, and there were no major studies that evaluated chest radiography against a clinical outcome reference standard. II. How Can Imaging Modalities Be Combined in the Diagnosis of Pulmonary Embolism? Summary of Evidence: Various proposed strategies have employed combi- nations of clinical exams, serum D-dimer measurement, lower extremity ultrasound, CTPA, VQ imaging, venography, impedance plethysmogra- phy, and conventional pulmonary angiography in the diagnosis of PE. Despite the heterogeneity in test utilization, recurrence rates for venous Chapter 22 Imaging in the Evaluation of Pulmonary Embolism 407 thromboembolism (VTE) in patients determined to be negative for PE were less than 2% in all major strategies identified. Safety, cost-effectiveness, and availability of resources may help to further differentiate these algorithms, and these issues require further investigation. Supporting Evidence: Nearly all of the pathway articles identified in our search employed clinical pretest probability in the diagnostic algorithm. We excluded those articles in which clinical pretest probability was not explicitly defined. All six of the major studies identified included a 3- month follow-up on patients who were determined to be negative for PE by the algorithm (31–36), and all had recurrence rates of VTE of less that 2%, although in one study (33), the high percentage of patients who were lost to follow-up may make the reported recurrence rate unreliable. One algorithm by Kruip et al. (34) employed only clinical exam, D-dimer, and lower extremity ultrasound, but with a notably high conventional angiog- raphy rate of 63%. Another study by Hull et al. (36) published in 1994 is also less than ideal because it relied on impedance plethysmography, a diagnostic modality that is no longer widely available. A level I (strong evidence) study by Wells et al. (31) deserves special mention because it most effectively limited the number of patients receiv- ing intravenous contrast, thereby reducing the overall risk of contrast- induced renal insufficiency. The study included 1252 patients with a PE prevalence of 15% who presented with symptoms of PE, had no con- traindications to contrast media, and had an expected survival of greater than 3 months. Following clinical assessment, all patients received VQ scans, followed by single or serial lower extremity ultrasound exams (Fig. 22.1). Lower extremity venography or conventional pulmonary angiogra- phy was performed in only 2% of patients. Although this algorithm limited the number of contrast examinations, it did so at the expense of a high number of lower extremity ultrasound examinations. An estimated 3093 lower extremity ultrasounds were performed on 1252 patients in this study (2.5 ultrasounds per patient). At most 19 patients (including 13 who were lost to follow-up) out of 1070 who were not anticoagulated by the algo- rithm developed VTE, equal to a recurrence rate of 1.8%. A more recent level I (strong evidence) study by Perrier et al. (32) placed greater emphasis on D-dimer measurement and CTPA (D-dimer measure- ments were not performed in the Wells algorithm). This study involved 965 patients (PE prevalence of 24%) with suspected PE who had no con- traindications to CT and who could be followed for 3 months. Following clinical probability assessment (Table 22.1), serum D-dimer was obtained in all patients, and a value less than 500mg/L excluded PE. None of these patients had recurrent VTE on 3-month follow-up. The remaining patients received combinations of venous ultrasound, CTPA, and conventional angiography. Sixty-two percent of patients obtained a contrast study (com- pared to 2% in the Wells algorithm), and complications from contrast administration were not discussed. However, ultrasound examinations were performed in only 71% of patients (compared to 2.5 ultrasounds per patient in the Wells algorithm). At most 10 patients (including three who were lost to follow-up) out of 685 who were not anticoagulated developed VTE, equivalent to a recurrence rate of 1.5%. 408 K. Juluru and J. Eng Chapter 22 Imaging in the Evaluation of Pulmonary Embolism 409 Figure 22.1. Clinical pathway proposed by Wells et al. [Source: Wells et al. (31), with permission from the Annals of Internal Medicine.] Table 22.1. Criteria for evaluating the clinical probability of pulmonary embolism (PE) accord- ing to Perrier et al. (32) Variable Score Previous PE or deep vein thrombosis +2 Heart rate >100 beats per minute +1 Recent surgery +3 Age (years) 60–79 +1 ≥80 +2 PaCO 2 <4.8 kPa (36 mm Hg) +2 4.8–5.19 kPa (36–38.9 mm Hg) +1 PaO 2 <6.5 kPa (48.7 mm Hg) +4 6.5–7.99 kPa (48.7–59.9 mm Hg) +3 8–9.49 kPa (60–71.2 mm Hg) +2 9.5–10.99 kPa (71.3–82.4 mm Hg) +1 Chest radiograph Platelike atelectasis +1 Elevated hemidiaphragm +1 Clinical probability according to total score: low, 0 to 4 points; intermediate, 5 to 8 points; high, 9 or more points. 410 K. Juluru and J. Eng Table 22.2. Summary of representative performance of various imaging modalities in detection of pulmonary embolism Clinical outcome reference Imaging reference studies studies Modality Sn (%) Sp (%) PPV (%) NPV (%) NPV (%) Angiography — — — — 95 1 Nuclear ventilation- 98 2 97 3 86–88 3 96–100 2 99.8 perfusion imaging Non–multidetector 45–100 78–100 60–100 60–100 94–100 CT pulmonary angiogram Multidetector CT — — — — 99 pulmonary angiogram Electron beam CT — — — — 97 MR angiography — — — — — Ultrasound of lung 71 77 69 79 — and pleura Echocardiography 56 90 77 76 — Plain film 36 4 92 5 38 6 76 7 — 1 Excludes three of the oldest studies, performed between 1978 and 1988. 2 For a normal scan. 3 For high-probability scan. 4 Highest sensitivity obtained using pleural effusion. 5 Highest specificity obtained using oligemia. 6 Highest positive predictive value obtained, using oligemia. 7 Highest negative predictive value obtained, using oligemia, pleural-based areas of increased opacity, pleural effusion, or elevated diaphragm. Sn, sensitivity; Sp, specificity; PPV, positive predictive value; NPV, negative predictive value. Take-Home Points The findings of this review are summarized in Table 22.2. Note that the sensitivities, specificities, and positive and negative predictive values shown in the table are derived from studies that range from level I (strong evidence) to level III (limited evidence). Therefore, comparison of these values between imaging modalities must be done with caution because of the heterogeneity in evidence strength. All of the diagnostic algorithms for suspected PE were associated with similar performances. The algorithm developed by Wells et al. (31) (Fig. 22.1) most effectively limited the use of intravenous contrast. However, the high number of ultrasound examinations and the use of serial compres- sion ultrasound up to 2 weeks following initial presentation challenge the practicality of this approach. Furthermore, although the algorithm may be effective in diagnosing pulmonary embolism, alternative etiologies of the presenting symptoms are more often discovered with CT. In Figure 22.2, we suggest an algorithm that is a modification of that proposed by Perrier et al. (32). The Perrier et al. algorithm makes use of enzyme-linked immunosorbent assay (ELISA) D-dimer as an initial screen, followed by lower extremity venous ultrasound in all patients with posi- tive D-dimer values. Studies have shown that DVT is unlikely in the absence of the clinical features noted in Table 22.3 (37). In our algorithm, we propose that only patients with clinical features of DVT undergo venous ultrasound, followed by CTPA when venous ultrasound is nega- tive. In the absence of clinical features of DVT, we propose that patients immediately undergo CTPA. The remainder of the investigation matches the Perrier et al. algorithm. We feel this approach provides rapid diagno- sis of PE and offers the opportunity to identify alternative etiologies for the patients’ symptoms through use of chest CT. Overall cost-effectiveness and safety need further study. Chapter 22 Imaging in the Evaluation of Pulmonary Embolism 411 Figure 22.2. Suggested algorithm for evaluation of pulmonary embolism. Refer to Table 22.1 for method of determining clinical probability and Table 23.3 for clinical features of DVT. Table 22.3. Clinical features of deep venous thrombosis according to Wells et al. (37) Active cancer (treatment ongoing or within previous 6 months or palliative) Paralysis, paresis, or recent plaster immobilization of the lower extremities Recently bedridden for more than 3 days or major surgery within 4 weeks Localized tenderness along the distribution of the deep venous system Entire leg swollen Calf swelling by more than 3 cm when compared with the asymptomatic leg (measured 10 cm below tibial tuberosity) Pitting edema (greater in the symptomatic leg) Collateral superficial veins (nonvaricose) Imaging Case Studies These cases highlight the advantages and limitations of the different imaging modalities. Case 1 History A 20-year-old woman with sickle cell trait was diagnosed with right popliteal vein thrombosis. She presents with shortness of breath, fever, and bilateral leg pain. Imaging Multiple planar perfusion images demonstrate decreased perfusion to the left lung (Fig. 22.3A). Ventilation images demonstrate normal ventilation to both lungs (Fig. 22.3B). These findings suggest central left-sided pul- monary emboli or a mass compressing the left main pulmonary artery. The CTPA demonstrates bilateral pulmonary emboli (Fig. 22.3C). Discussion This case demonstrates an instance in which both nuclear ventilation- perfusion imaging and CTPA detected evidence of pulmonary emboli necessitating treatment. However, it is notable that CTPA detected emboli in the right lung, where perfusion imaging was interpreted as normal. Case 2 History A 41-year-old woman has an extensive vascular history and DVT in both lower extremities. She presents with pleuritic chest pain and shortness of breath. Imaging Multiple planar perfusion images demonstrate heterogeneous activity throughout both lungs with large perfusion defects in the basal segments of both lower lobes (Fig. 22.4A). Additional small perfusion defects are seen in the right and left apices. The perfusion defects in the lower lobes do not correspond to any defects in ventilation imaging (not shown). These findings led to an interpretation of a high probability for pulmonary embolism. Findings on perfusion imaging also include abnormal activity in the liver, raising the suspicion for collateral circulation and vascular shunting. The CTPA demonstrates occlusion of the superior vena cava (Fig. 22.4B) and multiple collateral vessels around the liver (Fig. 22.4C). No pul- monary emboli were identified on CTPA. Discussion This case demonstrates an instance in which nuclear ventilation-perfusion imaging findings and CTPA findings are discordant. The CTPA detected occlusion of the superior vena cava with collateral vessels around the liver that were suggested by VQ scanning. The patient received anticoagulation therapy based on VQ findings without further imaging. 412 K. Juluru and J. Eng Chapter 22 Imaging in the Evaluation of Pulmonary Embolism 413 Figure 22.3. A: Anterior and posterior technetium 99m (Tc-99m) macroaggregated albumin (MAA) planar perfusion images demonstrate decreased perfusion to the left lung. B: Anterior single-breath and equilibrium-phase 133 Xe ventilation images in same patient demonstrate normal ventilation to both lungs. In combination with perfusion imaging, these findings led to an interpretation of high probability for pulmonary embolism. C: The CTPA demonstrates filling defects in right and left pulmonary arteries (arrows) consistent with pulmonary emboli. A B C [...]... Radiology 1 985 ;156:149–153 39 Zerhouni EA, Boukadoum M, Siddiky MA, et al Radiology 1 983 ;149:767–773 40 Siegelman SS, Khouri NF, Scott WW Jr, et al Radiology 1 986 ;160:313–317 41 Remy J, Remy-Jardin M, Wattinne L, Deffontaines C Radiology 1992; 182 : 80 9 81 6 42 Hillerdal G Chest 1 989 ;95(4):313–317 43 Westcott JH, Volpe JP Radiology 1991; 181 (P): 182 44 Rose RW, Ward BH Radiology 1973;106:179– 182 45 Eggli... 18FDG-PET with percutaneous needle aspiration biopsy for the evaluation of pulmonary nodules, 18FDG-PET was more sensitive for malignancy (100% vs 81 %), while transthoracic needle aspiration was more specific (100% vs 78% ) (61) H Cost-Effectiveness The cost-effectiveness of management strategies for SPNs was evaluated by Gould et al (86 ) in an analysis taken from a societal perspective The decision-analysis... lack of follow-up Results for 0. 8- to 1.0-cm lesions were significantly better than for 0. 5- to 0.7-cm lesions (sensitivity 88 % versus 50%; p = 013) The percentage of nondiagnostic percutaneous lung biopsies ranges from 4% to 40% in various series This wide variation reflects no differences in technique, study population, and prevalence of malignancy The most important complication of imaging- guided percutaneous... al., 19 98 (56) 89 0.7–4.0 66 Needle biopsy or 92 open lung biopsy 90 Prauer et al., 19 98 ( 68) 54 0.3–3.0 57 Surgery 90 83 Gupta et al., 19 98 (66) 19 1.0–3.5 63 Needle biopsy 10 Thoracotomy 8 Bronchoscopy 1 100 100 . embolism Clinical outcome reference Imaging reference studies studies Modality Sn (%) Sp (%) PPV (%) NPV (%) NPV (%) Angiography — — — — 95 1 Nuclear ventilation- 98 2 97 3 86 88 3 96–100 2 99 .8 perfusion imaging Non–multidetector. solid, (2) part- solid (mixed solid and ground-glass attenuation), or (3) non-solid (pure ground-glass attenuation). ᭿ The imaging strategy for the further evaluation of small solid pul- monary. evaluation with 18 FDG-PET, percutaneous needle biopsy, or video-assisted thoracoscopic surgery (VATS) is recommended. ᭿ Part- solid nodules (solid and ground-glass components) and non- solid nodules