Pulmonary Scintigraphy for the Diagnosis of Acute Pulmonary Embolism: A Survey of Current Practices in Australia, Canada, and France

DISCUSSION

Several conclusions can be reached based on responses from 331 nuclear medicine departments in 3 different countries regarding V/Q scintigraphy for the diagnosis of acute PE. First, in these countries, SPECT has largely replaced planar imaging, with almost half of the SPECT studies also incorporating a low-dose CT. Second, criteria used for interpretation are inconsistent, especially for V/Q planar imaging.

The observation that a large majority of nuclear medicine centers (72%) have switched from the traditional planar imaging to SPECT suggests that the nuclear medicine community considers V/Q SPECT to be a more appropriate test with higher diagnostic accuracy than planar scanning for diagnosing PE. By using 3-dimensional imaging, V/Q SPECT has an undoubted technical advantage over conventional planar imaging through its ability to eliminate overlap of perfusion defects; its visualization of the medial-basal segment; and its ability to better assess the size, shape, and location of defects (26). Many studies have demonstrated the higher diagnostic performances of V/Q SPECT when compared with planar imaging for the diagnosis of PE (11–14). In particular, V/Q SPECT has been reported to dramatically reduce the proportion of nondiagnostic tests (15,16,27). It is for these reasons that the majority of nuclear medicine physicians have adopted V/Q SPECT as routine for PE diagnosis.

However, 33% of institutions still perform planar V/Q scintigraphy, with 5% of departments acquiring both planar and SPECT images. The reason why these institutions have not adopted V/Q SPECT is unclear and was not specifically elucidated in the study. Possible barriers preventing the transition from planar to SPECT imaging include resistance to change, greater familiarity and experience with planar imaging, concerns relating to increased image acquisition times, and the lack of a suitable ventilation agent (26). Some physicians may consider that there is no need to change because planar V/Q is a validated and well-established examination. Other physicians, especially those performing both SPECT and planar acquisition, reported a willingness to switch from planar to SPECT V/Q but were constrained by a reluctance of the clinicians to adopt this new imaging method.

Although the proportion of nuclear medicine physicians performing SPECT acquisitions for diagnosis of PE is high in the countries surveyed, references to V/Q scanning in the medical literature by nonnuclear medicine specialists frequently refer to planar imaging, often with reference to its high nondiagnostic rates (21,22). In the few publications that reference V/Q SPECT, it is often portrayed as a test with future potential applicability, rather than an examination in routine use today (10). Although current clinical guidelines make reference to planar V/Q imaging (10), and presumably many referring clinicians assume that this is how V/Q is still performed, our survey shows that in at least the 3 countries surveyed, SPECT has largely replaced planar imaging in the investigation of PE. This raises concern that there is a lack of communication between nuclear medicine physicians and referring physicians about how the test is performed. This discrepancy may be related to several factors. First, studies describing V/Q SPECT have been mostly published in nuclear medicine journals rather than in high-impact general internal medicine journals. In contrast, methods such as CTPA, V/Q planar, or D-dimer testing are more widely described in the nonimaging literature. Furthermore, V/Q SPECT has not achieved the same standard of validation as others diagnostic tests including planar V/Q imaging (1,6–9), CTPA (28–30), or D-dimer testing (31,32), which may impede its widespread acceptance by clinicians. Indeed, although there is an extensive literature reporting the superior diagnostic performances of V/Q SPECT as compared with V/Q planar, the exact accuracy of V/Q SPECT for PE diagnosis is still unknown (20,23). Most importantly, although all diagnostic tests and strategies (including CTPA, planar V/Q, or D-dimer) for PE diagnosis have been validated in large management outcome studies, such a study with V/Q SPECT is still lacking. It is likely that achieving the same standard of validation will be an essential step in convincing clinicians, and especially venous thromboembolism specialists, of the usefulness of V/Q SPECT for diagnosis of PE.

Among institutions using V/Q SPECT, 46% simultaneously acquired CT images. The relatively high proportion of institutions that have adopted this approach would suggest that imaging specialists consider the CT to be useful for the interpretation of V/Q SPECT. However, there was a marked variability between the 3 countries. In France, where the proportion of centers using V/Q planar was the highest, 58% of institutions performing SPECT also acquired CT images. The proportion was 42% in Australia and only 11% in Canada. These results cannot be fully explained by differences in the availability of hybrid cameras. Indeed, although most Canadian institutions do have a hybrid SPECT/CT system, only 11% of centers performed V/Q SPECT CT acquisition. Dosimetric considerations may be a contributing factor, although the additional radiation dose is only on the order of 1 mSv whereas overall radiation dose of a CTPA is on the order of 8–20 mSv (11). Difference in cost and in reimbursement policies could also play a role in some jurisdictions. In addition, there are much less data validating the use of V/Q SPECT/CT as compared with V/Q SPECT alone. Studies have consistently reported a higher specificity of the test (24,33), which may be of relevance in the setting of the increasing concern about a possible overtreatment of PE and a trend toward the use of extended duration of anticoagulation therapy (34). An additional low-dose CT also allows the detection of alternative diagnoses, which has been reported as an advantage of CTPA (35). A limited number of institutions routinely use a low-dose CT in lieu of ventilation imaging (Q SPECT/CT). Recent data demonstrated similar sensitivity but lower specificity, with a high risk of false-positive results when compared with V/Q SPECT (24).

The second part of the survey was dedicated to the interpretation criteria. There is an extensive literature on the topic, with multiple criteria proposed for interpretation, likely as a result of the probabilistic reporting approach proposed in the PIOPED study. For V/Q SPECT, the most commonly used criteria (60%) were those proposed by the EANM, that is, a binary approach (PE or no PE) with a diagnostic cutoff of “1 segmental or 2 subsegmental perfusion mismatched defects” (18). This diagnostic cutoff has been reported to be optimal for PE diagnosis (36), and a recent observational study confirmed the diagnostic exclusion safety of a diagnostic management including a V/Q SPECT study interpreted with these criteria (37). Only 8% of institutions used the probabilistic approach for interpretation of SPECT, compared with 29% for planar. Our survey demonstrated that the shift from planar to SPECT imaging has also been associated with a replacement of probabilistic reporting with a more binary reporting approach.

Although there is some variability in the reporting criteria used for SPECT imaging, there is much greater inconsistency in the interpretation of planar imaging. In the survey, 29% of institutions use the probabilistic reporting approach, allowing the integration the V/Q scan results into validated diagnostic algorithms. However, most centers performing planar imaging use a binary reporting approach, with 35% of respondents using the EANM diagnostic cutoff and 20% of institutions using a 1 subsegmental cutoff. This is a relevant finding for referring specialists, as the reporting criteria used may not be those that have been validated in large outcome management studies (6–9,38). Overall, only 10% of centers currently use planar V/Q interpreted according to the probabilistic criteria, highlighting our observation that the approach to image acquisition and the reporting criteria used for V/Q scanning have changed in recent years, which may not be understood by referring clinicians.

Planarlike images can be derived from SPECT data using several approaches. In our survey, 63% of institutions performing V/Q SPECT also generate pseudoplanar images. In Australia, departments were further asked about the rationale of producing such images. For 36% of institutions, the aim is just to provide planar images to the referring clinicians, with the pseudoplanar images not being used for interpretation. Fifty-eight percent of institutions use them to confirm or correlate SPECT image interpretation. Although summed planar images have been shown to approximate true planar images, they are not perfectly concordant and may blur out small perfusion defects (39,40). Although the concordance between reprojected and true planar images may be better (39), how to manage discrepancies between SPECT and pseudoplanar images, and how they should be used by both reporting specialists and referring specialists, remains unresolved.

This study has some limitations. First, the questionnaire was composed of a small number of multiple-choice questions, which limited the possible responses. However, all questions included a response of other, which was chosen in a limited number of cases, suggesting that most physicians found an adequate response among the propositions. Second, only 1 answer per institution was consolidated to conform to the initial survey design performed in France. Accordingly, we did not assess the variability of practices among physicians from a same institution, which would be a subject of interest. Third, the participation rate was 30% in Canada and 44% in Australia, which could bias the results because nonresponders may hold divergent practices. Nonetheless, the participation rate was almost universal (209/210 departments) in France, where the survey demonstrated the same trend in the use of SPECT and similar inconsistency with regard to the interpretation criteria. Fourth, the conclusions drawn relate to the 3 countries that were surveyed and it may not be possible to extrapolate the results more broadly to other countries. Access to hybrid scanners varies around the world, and in particular, there is differing access to various scintigraphic ventilation agents. Of particular note is that technegas or Krypton are not available in the United States, which may be a contributing factor to the low adoption of V/Q SPECT imaging in that country, although some suggest that ^99mTc-DTPA and ¹³³Xe may be used.