Variations in Clinical PET/CT Operations: Results of an International Survey of Active PET/CT Users

DISCUSSION

The aim of this survey was to gather data on PET/CT practice and operational heterogeneity in clinical routine. The total response rate of 14% is within accepted response ranges for Web-based surveys ( 9, 10). Most responses were collected from users in the United States. Today, about 45% and 15% of PET/CT systems are installed in the United States and Europe, respectively, which corresponds roughly to the regional input collected in our survey. PET/CT operations in Asia-Pacific are reflected less adequately, based on the number of responses collected from that region. The overall fraction of responses from public (60%) and private settings (30%) corresponds roughly to the public and private operations of PET/CT in Europe (80% and 20%, respectively) and Germany (80% and 20%, respectively).

In general, total numbers of PET/CT installations and operations are unknown, aside from estimates provided in recent publications on the status of PET in Germany ( 15), PET/CT in India ( 16), and trends in PET ownership in the United States ( 17).

According to our survey, only 24% of all PET/CT systems are operated at multidisciplinary centers, such as dedicated PET/CT centers, cancer centers, and diagnostic imaging centers, thus illustrating the close collaboration that is required among clinical and imaging specialists for meaningful ¹⁸F-FDG PET/CT image interpretation despite an ongoing “ownership” debate ( 7, 18). Our survey found that the adoption of PET/CT appeared unrelated to experience with stand-alone PET systems: 59% of sites had prior PET experience and 41% of sites did not. Most sites had 4, 5, or over 10 y of PET-only experience before the installation of a PET/CT system.

The answers to questions about the average number of staff members involved in clinical PET/CT operations raise several points. First, it appears that many more technologists than physicists or medical doctors are involved. A possible reason may be the fact that some sites, governed by their local law, need to staff their PET/CT systems with both a radiology-trained technologist and a nuclear medicine–trained technologist. Second, systems with high throughput may require the availability of a larger number of technologists in order to limit the total radiation burden of each. Minimum physics support per PET/CT operation appears to have decreased since the onset of this technology, illustrating its more routine availability and greater use today. There appear to be more nuclear medicine experts involved than radiologists, as is reflected also in the low frequency of contrast-enhanced PET/CT examinations ( Table 6). Only 2 sites reported the engagement of dual–board-certified physicians. This fraction may change as more residents welcome a combined radiology and nuclear medicine training program. A recent survey by the European Society of Radiology and EANM showed that 77%–84% of all members of those 2 societies favor an interdisciplinary training program ( 19).

Based on the survey results, the 3 main diseases studied with PET/CT are lymphoma, lung cancer, and breast cancer, followed by colon cancer and head/neck cancer. This response corresponds to most of the PET/CT indications identified in a recent study from Germany ( 15), although the ranking of lymphoma studies and lung studies is reversed. The 2 main indications for cardiac PET/CT were myocardial viability studies and rest–stress perfusion studies. PET/CT in neurology was used mostly for dementia workups, brain tumors, and epilepsy ( Table 3).

Overall, PET/CT was used for oncology, radiation therapy planning, cardiology, and neurology in, respectively, 87%, 4%, 4% and 5% of patients. This response corresponds to the uses discussed by Kotzerke et al. ( 15).

An important factor in judging the availability of PET/CT technology is the average time patients have to wait for an appointment. In some countries, national health boards have put forward recommendations on cancer patient workups, including recommendations on the choice of modality and reasonable waiting times ( 20). Our survey indicated that 9%–15% of patients had to wait more than 1 wk for a PET/CT appointment.

Optimal and appropriate use of PET/CT requires attention to the quality of CT and PET images. Adherence to the technical and clinical guidelines put forward by multidisciplinary expert panels ( 4– 6) ensures a minimum set of quality standards, simplifies cross-center and cross-system data comparison, and is a basis for the standardization of PET/CT ( 21) in multicenter studies. Our survey included several additional questions about the PET/CT protocol ( Table 6).

Although current guidelines suggest a minimum fasting period of 4–6 h before a ¹⁸F-FDG PET examination, our survey indicated a striking variability in minimum fasting time: 72% of sites adhered to 4–6 h, whereas 21% of sites extended the fasting period to up to 20 h. One third of the sites established additional, frequently nonsensical, dietary requirements for oncology studies. Most sites (99%) measure blood glucose levels before the administration of ¹⁸F-FDG. However, cutoff values varied from 120 mg/dL ( 6) to 150–200 mg/dL ( 4, 15). Interestingly, 3% of sites set their cutoff at 250 mg/dL, whereas 7% use no cutoff (Supplemental Fig. 2).

Across all sites, the maximum ¹⁸F-FDG activities injected were 465 MBq and 524 MBq for 3D PET and 2D PET, respectively and, thus, were within the recommended activity range of 360–740 MBq ( 4). However, with novel PET detector materials and the availability of 3D and time-of-flight PET acquisitions, these activity ranges could be lowered without significant degradation of PET image quality ( 6). Accordingly, the EANM suggests using 380 MBq and 190 MBq for 2D PET and 3D PET, respectively, for a standard adult patient (75 kg). In obese patients, the maximum injected activity should be below 530 MBq—far lower than the maximum activities found in our survey, which were 740 MBq and 670 MBq for 3D PET and 2D PET, respectively. In many patients, such activities lead to acquisitions beyond the peak noise-equivalent counts. Therefore, particularly for heavy patients, extended emission acquisitions may yield image quality superior to that after increased activity levels, as shown by Masuda et al. ( 22) and as recommended by the Dutch protocol for standardized whole-body ¹⁸F-FDG PET/CT ( 21) and by the EANM ( 6). Patient weight–adapted emission imaging, first suggested by Halpern et al. ( 23), was used by 44% of the surveyed sites.

Tracer uptake time varied widely in our survey, nor is any consensus apparent from the 3 major guidelines. All surveyed sites adhere to an ¹⁸F-FDG uptake time of at least 45–90 min. Boellaard et al. limit the uptake time to 55–65 min ( 6), with which only 47% of the surveyed sites complied. Strict compliance with predefined uptake times is essential for high reproducibility within and among institutions and for comparability between serial studies. Likewise, a wide range of field-of-view limits was evident for torso PET/CT examinations. Unfortunately, existing guidelines are noncommittal. For example, Delbeke et al. defined a torso acquisition as one extending from “skull base to midthigh” ( 4), whereas Krause et al. defined the same examination range (“skull base to proximal thighs”) as being “whole body” ( 5). Similarly, Boellaard et al. refer to the “base of the skull base to mid thigh” range as being “whole body” ( 6). The results of our survey uncovered this confusion because sites reported various anatomic landmarks and simplified descriptors of coaxial imaging limits ( Fig. 1).

Most sites (73%) use a dedicated low-dose CT scan for attenuation correction, consistent with guidelines aimed at minimizing patient radiation exposure. The use of contrast-enhanced CT images for attenuation correction is still a matter of debate ( 24, 25). A recent poll among radiologists and nuclear medicine physicians indicated that 60% of experts anticipated a growing use of contrast-enhanced CT as part of integrated PET/CT ( 19).

Of the sites we surveyed, 31% used no intravenous contrast material, whereas 20% used intravenous contrast material in most patients. Oral CT contrast material was used in most patients by 46% of sites. This preference for oral over intravenous contrast material is striking because oral contrast material may yield artifactual uptake patterns and bias in attenuation-corrected PET, much like focal concentrations of intravenous contrast material ( 26), unless a water-based oral agent ( 27, 28) is used. However, only 26% of the respondents who use oral contrast agents use a water-based agent. Clearly, further training seems appropriate to help PET/CT users understand the potential bias introduced by CT contrast material and to leverage the full potential of diagnostic PET/CT.

Respiration artifacts are known to occur in PET/CT images if the CT images are acquired during a respiratory phase significantly different from mid expiration. Forty-four percent of our surveyed sites use breathing instructions ( 29) in most patients, whereas another 44% do not. Those sites that do not use breathing instructions may use PET/CT systems with multislice (n > 6) CT technology, in which quiet breathing would be acceptable ( 30). Alternatively, respiratory gating is available with state-of-the-art PET/CT technology and can also be used to improve CT-based attenuation correction ( 31). However, this approach is not popular and was used by only 20% of the sites. Similarly, the use of furosemide, a bladder catheter, or anesthesia is an option to further improve PET/CT image quality but, because of the invasiveness and associated patient discomfort, remains unpopular ( Table 6). Other efforts toward increasing diagnostic quality may include dedicated positioning devices that help immobilize the patient comfortably for the duration of the examination ( 32, 33); 41% of sites use such devices.

PET image reconstruction differed among the sites: 48% appeared to apply standard, predefined reconstruction protocols on 128 × 128 matrices only. However, higher-resolution PET images are feasible and beneficial when one is imaging smaller objects, such as the head, neck, or extremities. Therefore, indication-specific adjustments to reconstruction parameters have been recommended ( 6, 21). Such suggestions will help to standardize PET image acquisitions and analyses across institutions.

Further, an initial choice of optimum reconstruction parameters will help minimize repeated reconstructions, for which access to the raw emission sinogram data, at least, needs to be ensured. Only 22% of the sites store the emission sinograms. A considerable number of sites still report PET and CT separately; only 63% of the clinical reports are fully integrated, whereas for 17% of patients 2 separate reports are generated. Reports on combined PET/CT examinations should be integrated ( 4) not only to justify the use of a dual-modality technique but perhaps more so to inform the referring physicians adequately and not leave them with a decision to make based on potentially differing CT and PET reports.

Clearly, joint reading of PET/CT studies requires advanced training on the dual-modality technique, and various suggestions have been made to address this topic ( 19). Practical approaches to bring about dual-modality training need to be developed. Alternatively, dual-certified physicians could issue PET/CT reports. Our survey indicated that 37% of PET/CT reports are issued by dual-certified physicians whereas only 23% are issued by more than one physician jointly; all other reports are issued by either radiologists or nuclear medicine physicians.

Variations in the reported SUV parameter do exist; some of these variations result from the inability to automatically calculate total lesion glycolysis—or SUV corrected for lean body mass or body surface area—with existing image analysis platforms. Nonetheless, available SUV measurements should be included in the report, but only 90% of sites did so.

The time required for interpretation was as long as the time of actual image acquisition. With an average emission scan time of 3 min per bed position, a torso scan is completed in less than 25 min assuming a 7-bed-position examination. At our surveyed sites, 51% of PET/CT reports take 16–30 min, raising concern about the quality of reports in high-throughput scenarios with only limited staff available. Finally, it would seem appropriate to provide referring physicians with a copy of the PET/CT image data together with the report. However, only 30% of sites indicated that they do so for most patients.

This study had some limitations. First, we collected data on the heterogeneity of PET/CT operations in clinical practice. Because those who respond to a survey such as ours are typically the experts who are most interested and motivated ( 34), we assume that the actual operational and procedural heterogeneities are far larger than reported here. Second, several responders did not provide eligible answers to every question asked (so-called item nonresponse ( 35)). A problem inherent to survey research itself, regardless of format or mode of administration, is that maximizing the breadth of a survey increases the risk of a greater number of incomplete responses. Therefore, we could not categorize responses to questions on PET/CT protocols with respect to the origin of the response (e.g., public vs. private settings or United States vs. Asia-Pacific). Finally, our survey did not address in detail all aspects of PET/CT acquisition protocols. We composed a series of questions that—based on our experience—mirror the areas of greatest diversity among clinical PET/CT users. If adherence to guidelines were to be tested in clinical reality, a regional or global survey performed by a societal organization would be best. In addition, it seems mandatory that radiology and nuclear medicine associations need to work together on a global scale to clarify the existing ambiguities between guidelines.