In the quest to improve the diagnostic accuracy for the detection of pulmonary embolus, it is important not to lose sight of the big picture. What really matters is the prevention of death by pulmonary embolus.

To achieve this end there are four principle goals of management

• Identify those at risk
• Confirm the clinical diagnosis
• Commence treatment
• Anticipate recurrence

The first goal is to identify those at risk. In order to do this it is not only important to identify clinical risk factors but also to estimate the prevalence of disease in the patient population.

The determination of prevalence of PE is difficult and can vary depending upon the test used to make the diagnosis.

Pulmonary angiography and helical CT imaging may detect a higher prevalence of PE compared to lung scintigraphy. To illustrate this point, in the following respiratory session I will present data from the Newcastle study where 717 patients were followed 1.5 to 2.5 years after lung scintigraphy for the diagnosis of pulmonary embolus and their outcome determined. Only 1% of these patients were investigated by pulmonary angiogram and the prevalence of disease was determined to be 23%. By comparison, the original PIOPED study and the recently published PISA-PED study investigated 70% and 46% respectively of their patients with pulmonary angiography. Their respective prevalence of PE were 33% and 39%.

Although prevalence of disease is important, it is even more important to determine the prevalence of disease that will kill and thereby facilitate risk stratification.

Unlike ischaemic heart disease, risk stratification for pulmonary embolus has not been well developed, largely due to a paucity of outcome data.

To demonstrate the limitations of the clinical diagnosis of PE, consider a pretest score using 3 categories: risk factors, investigations and clinical findings and a simple scoring system, with respective total scores of 15,6 and 9. The patients' scores can be expressed as a ratio of the total and summed in order to determine the pretest probability of PE.

Pre-test score

Risk Factors (score) Investigation (score) Clinical (score)

Post operative (2) Hypoxia (2) Dyspnoea (2)

Immobility (2) CXR-effusion (1) Chest Pain (2)

PHx DVT/PE (2) CXR-opacity (2) Haemoptysis (2)

Major Trauma (2) ECG-tachycardia (1) Raised JVP (1)

Cancer (2)

Travel (1)

Oestrogens (1)

Smoker (2)

Venous dis (1)

Obesity (1)

TOTAL 15 6 9

SCORE n/15 + n/6 + n/9

Arbitrarily, a summed fraction score of <0.69 can be regarded as a low pretest probability for PE.

When we applied this score system to our patients in the Newcastle study we found to our surprise there was no significant difference between the scores of patients with and without PE.

Pre-test Evaluation

Pulmonary Emboli unlikely if score

<(2/15+2/6+2/9) =0.69

Patients without PE

Score range: 0-1.39 (median=0.58)

Patients with PE

Score range: 0-1.19 (median=0.47)

Obviously this scoring system is not necessarily the most appropriate way of evaluating pretest probability, and the physician's gestalt may be more predictive of PE. This, however, illustrates the need for a test of much higher diagnostic accuracy, such as lung scintigraphy, pulmonary angiography or helical CT imaging.

The second principle of management to prevent death by PE is to confirm the diagnosis by investigation.

Let's just consider lung scintigraphy. Although the original PIOPED study has had an enormous impact on lung scintigraphy for the diagnosis of PE there remains lack of consensus with regard to technique and diagnosis criterion.

• Which technique is best?

1. V/Q v Perfusion alone
2. Inert gas v Aerosol v Technegas
3. SPECT v Planar

• Which diagnostic criterion is best?

1. Revised PIOPED
2. PISA-PED
3. McNeil
4. Biello

If we now consider the PISA-PED study and those authors' views:

The Argument For Perfusion Imaging Alone and a New Set of Diagnostic Criteria.

They note that:

• PIOPED study showed sensitivity for the detection of PE of only 41% (high prob.) Moderate probability =39% of total studies. Lung scintigraphy failed to detect 12% of PE in patients with low probability studies

Consequently, they devised a new set of diagnostic criteria consisting of normal, near normal, abnormal with a typical appearance of PE and abnormal without a typical appearance of PE. Using this criterion for perfusion imaging alone, they found a diagnostic sensitivity of 92% and negative predictive value of 97%. Using clinical assessment (which most probably was largely physician's gestalt) they found further improvement in the diagnostic accuracy.

In comparison to the PIOPED study, the PISA-PED study's conclusion, however, raise the question:

• If perfusion imaging is best, how does ventilation imaging reduce the diagnostic accuracy?
• The implications is that either the diagnostic criteria of PIOPED are inferior, or Xe-133 ventilation imaging is inappropriate for the diagnostic criteria used in PIOPED.

The Newcastle Study: (Technegas Ventilation/Perfusion)

• Revised PIOPED criteria used
• End points: Pulmonary embolus (PE), death by PE, death by hemorrhage, bleeding complications in treated patients
• 21/717 patients (3%) had the diagnosis missed and received no treatment (7/12 died)
• Only 10% of studies were "indeterminate"

If we compared the Newcastle study to the original PIOPED study, the difference in indeterminate studies becomes apparent. In addition, the lung scintigraphy result of the PISA-PED study is shown, highlighting what would appear to be a more decisive diagnostic criteria. But in their study 14% of cases were indeterminate.

Given PIOPED the PISA-PED studies' findings, one way of interpreting the Newcastle study's findings is that:

• Ventilation imaging with Technegas provides a means of better utilizing the previously established revised PIOPED diagnostic criteria by improved lesion matching in corresponding ventilation and perfusion images.

The third principle for the prevention of death by PE is commencement of appropriate treatment.

In the Newcastle study a total of 231 patients were anticoagulated, including those for reasons other than PE. However in those treated for PE there were 6 deaths by hemorrhage and 15 severe bleeding complications.

We need to be reminded that there is a mortality and morbidity associated with anticoagulation therapy in patients with PE and that it is equally important to discriminate between those who require treatment and those who do not.

A diagnostic test that is "too sensitive", may potentially result in the treating physician doing more harm than good.

The fourth principle for the prevention of death by PE is anticipating recurrence of pulmonary emboli. This can be achieved by:

Recurrence can be documented by a changed pattern of perfusion defects or perfusion mismatch on repeat ventilation and perfusion lung scintigraphy. However with patients with persisting risk factors hypercoagualable states, for example activated protein C resistance, clinical evaluation to identify such high risk states and clinical judgment in determining the risk: benefit ratio are paramount in deciding the duration of anti coagulation therapy.

So, to conclude. There remains a need for outcome studies among different patient populations (both treated and untreated). Similar outcome studies using angiography and spiral CT need to be done to determine the safest treatment thresholds.

There is a need for further testing and comparison of different diagnostic criteria.

There may be a shrinking role for pulmonary angiography.

If revised PIOPED criteria are used efforts should be made to further reduce the population of "indeterminate" studies.