Comparison of three diagnostic strategies for suspicion of pulmonary embolism: planar ventilation-perfusion scan (V/Q), CT pulmonary angiography (CTPA) and single photon emission CT ventilation-perfusion scan (SPECT V/Q): a protocol of a randomised controlled trial

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • This management outcome study will be the first trial to provide strong data on the diagnostic performance of a standardised diagnostic algorithm based on the single photon emission CT (SPECT) V/Q for pulmonary embolism (PE) diagnosis.

  • If the study confirms the safety of a diagnostic strategy algorithm based on SPECT V/Q for PE diagnosis, it will support and validate current nuclear physicians’ practices according to evidence-based medicine principles.

  • It will resolve doubts about a diagnostic strategy algorithm based on SPECT V/Q for PE diagnosis among clinicians particularly among the specialists in venous thromboembolism and will ensure a safe patient care.

  • If we fail to demonstrate the safety of SPECT V/Q for PE diagnosis, it will be necessary to perform further investigations by possibly revising diagnostic interpretation criteria and/or the role of the technique in the diagnostic algorithm for PE diagnosis.

Introduction

Pulmonary embolism (PE) remains a diagnostic challenge. PE shares signs and symptoms with several life-threatening conditions (eg, acute coronary syndrome, pneumonia), and no clinical sign, symptom or simple diagnostic test is accurate enough to rule out or confirm the diagnosis as a stand-alone test.1 2 The diagnostic management of PE does not rely on a single test but on the use of diagnostic algorithms, including sequential use of the clinical probability, D-dimer testing and imaging tests. Although the indications for first-line imaging may differ from different guidelines,3 there are currently two validated imaging modalities for PE diagnosis: planar ventilation/perfusion (V/Q) scan and CT pulmonary angiography (CTPA) which have been validated in outcome studies using as primary outcome the 3-month risk of venous thromboembolism (VTE) in patients left untreated on the basis of an initial negative diagnostic workup.4 Reaching diagnostic certainty in every patient with clinically suspected PE is of utmost importance. Both missed diagnosis and false diagnosis have undesirable consequences. Untreated PE could be fatal in up to 25% of patients,5 while anticoagulant therapy exposes patients to a high risk of bleeding: approximately 2% during the first 3 months of treatment, of which 10% are fatal.6 Moreover, there is a trend towards indefinite duration of anticoagulant therapy after a first episode of PE that was not provoked by a major risk factor.7 A definitive and accurate diagnostic conclusion is therefore required for all patients with suspected PE. The goal has been to eliminate the need for pulmonary angiography, the historical gold-standard, which was invasive and carried a risk of major complications or death.

The planar V/Q scan constituted the first attempt to replace pulmonary angiography. The diagnosis of PE is made by comparing lung ventilation and perfusion images, after the inhalation and intravenous injection of a radiotracer, respectively. The test is interpreted according to the PIOPED (Prospective Investigation of Pulmonary Embolism Diagnosis) criteria. The PIOPED study is the V/Q landmark accuracy study versus pulmonary angiography.8 A limitation of the planar V/Q scan is that, except when it is normal or near-normal (which allows to safely rule out PE without further testing),9 its results need to be interpreted in combination with clinical pretest probability (PTP) and/or other tests such as compression ultrasonography (CUS) of the leg veins or D-dimer blood test. This is often perceived as too complex by busy clinicians. However, as stated by the 2019 European Society of Cardiology guidelines on the diagnosis of PE ‘The planar V/Q scan is an established diagnostic test for suspected PE’.10

Strategies using CTPA are simpler. CTPA can be used without any other imaging tests in most cases.11 12 It is more widely available (24/7 in most centres) and allows for alternative diagnoses (eg, aortic dissection). CTPA has become the main imaging test for patients with suspected PE. However, CTPA has drawbacks, including some risks and contraindications (eg, allergy to contrast, renal failure). The radiation dose to the chest is significantly higher than that of V/Q scan, which is an increasing concern, particularly in young women. Some studies suggest that a woman undergoing a CTPA at the age of 20 could have a 0.5% absolute increase in her lifetime risk of breast cancer.13 Additionally, there is an increasing concern in the literature about the risk of overdiagnosing PE when using CTPA.14 In a randomised clinical trial, Anderson et al did not show significant difference between CTPA-based and V/Q-based strategies in terms of diagnostic exclusion safety, however the CTPA-based strategy led to a 35% higher rate of PE diagnosis. This likely means that a significant proportion of PE diagnosed when using CTPA are false positive results or not clinically relevant.

The single photon emission CT ventilation-perfusion scan (SPECT V/Q) is a different method of image acquisition.15 The only difference between a planar V/Q and a SPECT V/Q is the acquisition mode of images. Similar equipments (gamma cameras) and radiotracers are used. Compared with the planar V/Q scan, SPECT V/Q offers the advantage of tomographic sections that improve contrast resolution and limit the overlapping of perfusion defects by other structures.16 Many centres around the world have already replaced planar V/Q with SPECT V/Q in clinical practice.17 Several diagnostic accuracy studies have been performed on SPECT V/Q, and most reported improved diagnostic performances as compared with V/Q.18 However, as highlighted in a recent systematic review, methodological issues impede any firm conclusion to be drawn from these studies: in many of them, the SPECT V/Q result participated in the final diagnostic conclusion (incorporation bias).19 Most importantly, to date, there has been no formal diagnostic management outcome study or randomised trial evaluating a standardised diagnostic algorithm based on the SPECT V/Q for PE diagnosis.

This final step is now essential to confirm the application of this diagnostic method in the management of a patient with suspected acute PE. We present the protocol of a randomised multicentre, international management study comparing SPECT V/Q with validated strategies.

Methods and analysis

This protocol follows the recommendations from the EQUATOR network statement on Standard Protocol Items: Recommendations for Interventional Trials (online supplemental appendix A).20

Supplemental material

Our aim is to assess the performance of SPECT V/Q in a diagnostic strategy for the management of patients with a clinical suspicion of PE. Our hypothesis is that a strategy based on SPECT V/Q could be an alternative to previously validated strategies using either CTPA or planar V/Q scan for the diagnosis of PE. SPECT V/Q could combine the advantages of CTPA (safe diagnostic exclusion, simple algorithm), and of planar V/Q scan (no overdiagnosis/overtreatment of clinically non-relevant PE, lower radiation exposure no contraindication).

Two criteria need to be met prior to considering V/Q SPECT-based strategies validated for clinical use. First, its safety must be verified, that is, the risk of missing a PE. In VTE) studies, by convention, the risk of thromboembolic event during the 3 months following a negative workup has to be <3%, which is based on the risk observed after a negative pulmonary angiography, the historical gold standard for PE diagnosis. Second, we need to ensure that patients with positive testing actually require anticoagulant therapy.

Choice of study design

We have elected to design a randomised multicentre, international clinical trial rather than a management outcome study. A management outcome study in patients with clinically suspected PE, using a strategy based on the SPECT V/Q, and assessing the 3-month risk of VTE in patients left untreated on the basis of a negative workup could address the safety of diagnostic exclusion (should the upper limit of the 3-month risk of VTE be <3.0%).14 However, this study design would not allow us to compare the 3-month risk of VTE with that of a control group of patients sharing similar characteristics and investigated in the same centres during the same period using a previously validated diagnostic strategy. It is important to ensure that SPECT V/Q will be non-inferior to current diagnostic strategies.

We have decided to randomise into three different groups each using a different diagnostic strategy based on SPECT V/Q arm (investigational), CTPA arm (comparator) and planar V/Q arm (comparator) (see online supplemental appendix B for global design). We could have chosen a single control group, using either a CTPA or a planar V/Q-based strategy, given that these two strategies have demonstrated similar diagnostic exclusion safety. However, we want to be able to compare the proportion of patients with confirmed PE when using the SPECT V/Q, with that obtained when using a CTPA or planar V/Q-based strategy, in order to assess for a potential overdiagnosis, and to assess for the extent of overdiagnosis.

Supplemental material

Objectives and outcomes

Assessing the safety is a critical step. Missing a PE could result in a catastrophic outcome.12 Effective treatment exists in patient with confirmed PE and several others diagnostic strategies proved safe in excluding PE. Therefore, our primary objective will be to determine whether a diagnostic strategy based on SPECT V/Q is non-inferior to previously validated strategies (based on CTPA or planar V/Q scan) in terms of diagnostic exclusion safety, that is, at not missing the detection of clinically important PE, as assessed by the 3-month risk of VTE in patients with a negative diagnostic workup. The primary outcome is the rate of adjudicated VTE during the 90±4 days follow-up period (3-month VTE risk) in patients left untreated after a negative diagnostic workup. This outcome is commonly used in PE diagnostic studies. All suspected VTE events and death during follow-up will be adjudicated by an independent committee. The criteria used for adjudication of outcomes (VTE, bleeding, death) during follow-up are detailed in online supplemental appendix C. Briefly, VTE will be defined as an objectively confirmed proximal (ie, involving a veinous segment before the trifurcation of popliteal vein) lower limb deep vein thrombosis and/or PE using any validated diagnostic algorithm. Beyond verifying safety, our objectives will also be to determine whether a diagnostic strategy based on SPECT V/Q could lead to a potential overdiagnosis, as assessed by the proportion of patients diagnosed with PE in each of the three groups (SPECT V/Q, planar V/Q and CTPA-based strategies), and to evaluate the risk of bleeding, death, complications and need for additional diagnostic tests in each trial arm. The secondary outcomes will be the proportion of patients deemed to have PE according to the strategy in each arm (ie, a strategy with a significantly higher rate of PE diagnosis might correspond to a possible higher rate of false positive results and therefore undue treatment), the proportion of patients for whom additional tests are requested in each arm (lower limb venous CUS, CTPA, V/Q, other), the incidence of major bleeding, clinically relevant non-major bleeding episodes in each arm and the incidence and cause of death in each arm. We will report the proportion of patient who did not complete the anticoagulation for 3 months, and will describe recurrent events in patients with a confirmed PE at inclusion according to whether they occurred despite anticoagulation or not.

Supplemental material

Study population

Our study population consists of adult patients presenting to the emergency room of participating centres with a clinically suspected acute PE requiring chest imaging. PTP will be assessed by using a PTP score (ie, Wells or Geneva), with a dichotomous approach (ie, patients with a non-high or unlikely PTP but a positive D-dimer, and those with a high or likely PTP will be eligible).10 Patients with a non-high (intermediate or low) or unlikely PTP and a negative D-dimer test will not be enrolled, since PE is excluded without the need for imaging tests. Regarding the D-dimer threshold, either conventional or age-adjusted threshold, where implemented, will be used.

We will exclude patients with an already confirmed diagnosis of PE, those with clinically suspected massive PE (different diagnostic workup), on therapeutic dose of anticoagulant for ≥48 hours (decreased testing sensitivity), with another indication for long-term use of anticoagulation (eg, atrial fibrillation) (preventing primary end point assessment), with a contraindication to CTPA (insufficient renal clearance, allergy to contrast dyes), pregnant or breastfeeding women (different diagnostic workup), patients with a life expectancy <3 months (preventing primary end point assessment) and patients unable or unwilling to consent to participate.

Diagnostic algorithm

Reference PE strategies based on planar V/Q scan and CTPA will be consistent with previously validated diagnostic criteria and algorithm and described in figure 1 and figure 2.11 21 For the intervention arm, we had to decide on which criteria to use for SPECT V/Q interpretation and how to integrate SPECT V/Q in a standardised algorithm (figure 3). The European Association of Nuclear Medicine proposed criteria based on expert consensus.22 They chose to report the results of the V/Q SPECT as a binary response (‘PE’ or ‘no PE’), and to abandon the probabilistic approach used for planar V/Q (normal, low, intermediate or high probability). We previously performed receiver operating characteristic curve analyses according to the size and number of mismatches on V/Q SPECT to determine the best cut-off for PE diagnosis. The most discriminant cut-off was obtained using ‘≥1 segmental or 2 subsegmental mismatches’ as positive for PE diagnosis, corresponding to a sensitivity, specificity, positive and negative likelihood ratios of 92% (95% CI 84% to 100%), 91% (95% CI 87% to 95%), 10.2 (95% CI 6.5 to 16) and 0.09 (95% CI 0.04 to 0.23), respectively.18 Using these criteria, V/Q SPECT may be integrated in a diagnostic algorithm in combination with PTP. In a systematic review of PTP scores, we found the proportion of confirmed PE to be 8% and 34% among patients classified as having an ‘unlikely’ and ‘likely’ PE by the Wells score.21 However, after undergoing D-dimer test, the proportion of patients with confirmed PE increased among the subgroup of patients with an ‘unlikely’ PTP but a positive D-dimer test (usually around 25%). Applying our likelihood ratios to these PTP, the post-test probability of PE in case of a negative V/Q SPECT in a patient with an ‘unlikely’ PTP but a positive D-dimer should not exceed 3%. It will likely be slightly higher than 3% in patients with ‘likely’ PTP, which could require a bilateral leg CUS, similar to what has been used in previous algorithms using the planar V/Q scan or first-generation single-detector helical CTPA. In case of a positive V/Q SPECT, the post-test probability of PE should be ≥80% in both groups of patients with a ‘likely’ PTP or an ‘unlikely’ PTP but positive D-dimer.

Figure 1
Figure 1

CT pulmonary angiography (CTPA) arm—algorithm. CUS, compression ultrasonography; PE, pulmonary embolism.

Figure 2
Figure 2

Planar V/Q arm—algorithm. CTPA, CT pulmonary angiography; CUS, compression ultrasonography; PE, pulmonary embolism; PTP, pretest probability.

Figure 3
Figure 3

Single photon emission CT (SPECT) V/Q arm—algorithm. CTPA, CT pulmonary angiography; CUS, compression ultrasonography; PE, pulmonary embolism; PTP, pretest probability.

Given the study’s pragmatic design, interpretation will be performed by the attending nuclear medicine physician or radiologist at each centre. We will not analyse the inter-reader variability in this study which has already been reported.

Study follow-up period

All patients with a negative diagnostic strategy will be left untreated and followed clinically for 3 months. Follow-up will be conducted by a phone interview at 3 months by study personnel using a standardised script comprising questions about further diagnostic testing, anticoagulant therapy or hospital visits. In case of a suspected VTE, bleeding or death during follow-up (unscheduled visit or day 90±4 visit), the study personnel will collect related clinical data on the electronic case report form (online supplemental appendix D and E), and prepare an adjudication file, including symptoms, clinical notes, hospital discharge summary and results of all diagnostic tests. Individual participant study duration will be 3 months.

Supplemental material

Supplemental material

Patient and public involvement

Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research.

Statistical analysis

Analysis of baseline characteristics

Descriptive statistics will be used to examine the baseline characteristics of included patients. Proportions will be reported for categorical variables. Normality distribution of variables will be assessed. Mean values and SD will be reported for all characteristics expressed as continuous variables. Medians and ranges will be presented for discrete data. The incidence of VTE in patients left untreated during the follow-up period in each arm will be reported and 95% CIs will be provided.

Outcome analyses

The difference in the 3-month risk of VTE between the SPECT V/Q group and the combined planar V/Q and CTPA groups will be estimated, along with its 95% CI using the Wilson score without continuity correction. We will base the non-inferiority conclusion on a one-sided 95% CI on the difference between event rates entirely excluding 1%. The secondary outcomes, including PE prevalence, will be compared 2 by 2 between the three groups using χ2 test or Fisher’s exact tests as appropriate. The trial will be analysed based on the intention-to-diagnose principle. However, numbers and results of additional tests requested by attending physicians will be captured and reported. A per-protocol analysis including patients in whom diagnosis algorithm will be strictly adhered to will be performed.

Sample size

Previous studies assumed a 3-month risk of VTE after a negative standard-of-care diagnostic workup of 1.5%, and used a minimal clinically important difference (MCID) of 1.5%–2.5% to assess non-inferiority.14 23 However, most recent studies found lower 3-month risk of VTE in negative patients.23 If we assume a 3-month risk of VTE in the negative patients under standard-of-care (V/Q or CTPA) to be 1%,11 24 and an MCID of 1%, using an alpha of 0.05, this means that we require 1224 negative-strategy untreated patients per group to achieve 80% power to detect inferiority. A previous study showed no significant difference in the 3-month risk of VTE after a negative CTPA-based strategy and a planar V/Q-based strategy.14 Therefore, to spare the number of patients required in the study without compromising the analysis of our primary outcome, we decided to apply an unbalanced 2/1/1 randomisation scheme. We will require 1224 patients with negative workup and not receiving anticoagulants in the SPECT V/Q arm, 612 in the planar V/Q arm and 612 in the CTPA arm. The planar V/Q and the CTPA arms will be considered as a single group for the comparison of the 3-month VTE risk. We chose to split the standard-of-care arm evenly between planar V/Q scan and CTPA precisely so that we could retain some ability to investigate whether there is any difference between the two. The number of patients included in these two arms will allow statistical analyses of the secondary outcomes, in particular the comparison of the proportion of confirmed PE: we will test the hypothesis of a 5% difference (from 15% to 20%) in the rate of confirmed PE between the SPECT V/Q and each of the other arms. To account for patients who will be diagnosed with PE (up to 30% among patients with an ‘unlikely’ PTP but a positive D-dimer, or a ‘likely’ PTP), and for patients who will be diagnosed without PE but who will receive anticoagulant therapy during follow-up for another indication (mainly atrial fibrillation in previous studies, usually around 5%), our final sample size will be 3672: 1836 patients in the SPECT V/Q arm, 918 in the CTPA arm and 918 in the planar V/Q arm.

Randomisation

After consent, the patient will be randomised into one of the three study arms using the interactive web-based registration and randomisation system. Randomisation will be unbalanced, with twice as many patients in SPECT V/Q arm as in CTPA and planar V/Q arms. Randomisation will be stratified by site and performed in blocks of 6 or 9.

Discussion

Despite the lack of clinical validation, SPECT V/Q has already replaced planar V/Q scan for the diagnosis of PE in current practice in nuclear medicine centres.17 Indeed, most of nuclear medicine physicians consider SPECT V/Q as a reference whereas 2019 clinical guidelines made by European Society of Cardiology only says ‘SPECT V/Q may be considered for PE diagnosis’ but ‘large-scale prospective studies are needed to validate SPECT techniques’.10 It is still described as experimental by some clinicians while some others are not always aware of the technique used in daily practice and the possible related issues.

Thus, there is an urgent need for a clinical trial to assess the real-life performance of a diagnostic strategy based on SPECT V/Q. This management outcome study will be the first randomised clinical trial to provide strong data about a standardised diagnostic algorithm based on the SPECT V/Q for PE diagnosis.

Indeed, if the study confirms the safety of a diagnostic strategy algorithm based on SPECT V/Q for PE diagnosis, it will support and validate current nuclear physicians’ practices according to evidence-based medicine principles. In addition, it will resolve doubts among clinicians particularly among the specialists in VTE and will ensure a safe patient’s care.

Otherwise, if we fail to demonstrate the safety of SPECT V/Q for PE diagnosis, it will be necessary to understand what let to this result, by revising the interpretation criteria for SPECT V/Q, and/or to raise the question of the role of this technique in the diagnostic algorithm for PE diagnosis.

Ethics and dissemination

The study protocol was approved in France (by Biomedical Research Ethics Committee, n° CPP Ouest 6-937), in Switzerland (by Commission cantonale d’éthique de la recherche de Genève n° 2016-01195) and in Canada (by Ottawa Health Science Network Research Ethics Board 20220636-01H). Prior to enrolment in the trial, the investigator or delegate will inform each participant, by providing full and adequate verbal and written information regarding the objectives and procedures of the trial and the possible risks involved. The patients must be informed about their right to withdraw from the trial at any time. The investigator or delegate will obtain signed informed consent from all patients prior to inclusion in the trial in Canada (see online supplemental appendix F), and an oral consent in France and Switzerland (see online supplemental appendix G). A Data Safety and Monitoring Board (DSMB) will be created. The DSMB will be independent and composed of three members: an expert in thrombosis medicine and clinical trials (Chair), a biostatistician and an expert in thromboembolic diseases. All members of the DSMB will remain at arms-length from the study. All VTE will be reported to and reviewed by the DSMB. The DSMB will meet after every 400th participant completes follow-up and reports to the principal investigator.

Supplemental material

Supplemental material

The results of this study will be submitted for presentation at relevant national and international conferences, and for publication in a peer-reviewed journal.

The target audience for this study comprises all physicians involved in the diagnosis of PE (emergency medicine, thrombosis specialists, nuclear medicine physicians and radiologists). The results will be published in a general internal medicine journal, to reach a broad audience; they will be presented at scientific meetings of various medical specialties. Some members from our research team are on the writing panel for scientific society clinical practice guidelines and hold leadership positions in scientific societies or research networks, ensuring efficient diffusion of our results to the target audience.

Trial current status

The study was approved by the Ethics Committee in September 2016 and was registered on clincaltrials.gov in December 2016 (NCT02983760). The first patient was enrolled in Brest in April 2017. The study is now open in 10 centres in France and 1 in Switzerland. We are launching the study in Canada in 2023. We plan to complete recruitment at the end of 2024.

Data availability statement

No data are available.

Ethics statements

Patient consent for publication

This post was originally published on https://bmjopen.bmj.com