1 Clinical Scenario
A 32-year old woman is referred to the radiology department, after coming to the emergency room at night. She is presenting with dyspnea, pain in the chest as well as upper back, especially upon inspiration and she has palpitations. Within the last half an hour she started coughing up blood. Her medical history includes an episode of DVT (deep vein thrombosis) 2 years ago, while she was pregnant. Her blood pressure is 140/90 mmHg and the heart rate is 120 bpm. She is currently taking oral contraceptive medication to prevent another pregnancy. She has no family history of thrombotic diseases. Although there is a wide variety of conditions of both the respiratory and the cardiovascular tract that could cause the aforementioned symptoms, based on the patient’s DVT during pregnancy as well as the fact that she in a hypercoagulable state due to procoagulant medication (contraceptives), it is likely that she has a PE. The patient’s risk of suffering from a PE is between 10 and 25 percent.
The current golden standard for patients suspected of having a PE is the CTPA (computer tomography pulmonary angiography). There are several advantages of CT, such as the fact that it is incorporated in the emergency setting of many hospitals, its capacity to identify many conditions pertaining to the differential diagnosis of PE, as well as the fast nature of the diagnostic procedure. However, CTPA is also associated with levels of radiation high enough to increase cancer risk, especially in cases of pregnancy or young age. The fact that CTPA is not indicated during pregnancy is a great impediment as the risk of PE increases six times in pregnant women. Moreover, patients suffering of renal failure are contraindicated CTPA because the contrast solution is nephrotoxic. Contrastingly, you know that MRI does not pose risk of radiation and in unenhanced version of MRI there are no risks regarding the contrast solution either. You are also aware that in case of untreated PE the mortality between 23 and 87%, and this condition is accounts for 15% of the total number of maternal deaths. You are questioning whether an MRI would be able to reliably rule out a PE in this woman.
2 Clinical question
By way of comparison with CTPA, is the NPV of MRI high enough to reliably rule out PE in a 32-year old woman who presenting with dyspnea, chest pain, hemoptysis and tachycardia, currently taking contraceptives and with a positive medical history of DVT?
3 Literature search
Query: (((((((((("pulmonary embolism/diagnostic imaging"[MeSH Terms] OR "pulmonary embolism/mortality"[MeSH Terms]) AND "computed tomography angiography"[MeSH Terms]) OR "computed tomography angiography/methods"[All Fields]) AND "magnetic resonance imaging"[MeSH Terms]) OR "magnetic resonance imaging"[All Fields]) AND "predictive value of tests"[MeSH Terms]) OR "negative predictive value"[All Fields]) AND "female"[MeSH Terms]) AND "young adult"[MeSH Terms]) AND "prospective studies"[MeSH Terms]) AND ("survival rate"[MeSH Terms] OR ("survival"[All Fields] AND "rate"[All Fields]) OR "survival rate"[All Fields])
Selected article: Pasin L, e. (2018). Magnetic Resonance Imaging of Pulmonary Embolism: Diagnostic Accuracy of Unenhanced MR and Influence in Mortality Rates. – PubMed – NCBI . Ncbi.nlm.nih.gov. Retrieved 20 October 2018, from https://www.ncbi.nlm.nih.gov/pubmed/28116500
Motivation: The query elicited 15 results. The selected article was the most suitable for the above clinical scenario as it is dealing with patients of ages between 22 and 89 years old, the majority of which are women and who, based on symptomatology are referred to radiology with a clinical suspicion of PE. Moreover, the chosen article has an acceptable study population of 91 subjects. The only other article that might have fitted with the clinical scenario (Vamsidhar et al) was focusing on risk stratification of PE rather than on establishing a diagnosis.
4 Evaluation of the article
++ The results of the index test, MRI, were compared to gold standard, the CTPA, which has a high quality.
– Although broadly applicable, the results are not characterizing for a population of patients with PE that have a particular symptomatology and the recognized risk factors the disease
++ The entire population was subjected to both CTPA and MRI.
++ The study was performed with consistent effort to avoid expectation bias. All evaluators were blinded to each other’s results. Moreover, the specialists were chosen in order to leave as little chance to error as possible.
++ With a high kappa value of 0.87 study shows a good reproducibility
+/- The 95% CI’s are acceptable for the negative predictive value (lower limit of 87.7%) but unacceptable for sensitivity (lower limit of 61.1%) which are the two most important parameters of the MRI for excluding PE.
++ Clear cut-off points were established for PE diagnosis (clear thrombus seen on the image, breakoff of one of the pulmonary vessels, abrupt changes signal intensity within a pulmonary vessel) which are unanimously recognized
++ The index place is conducted at the most logical point in time given the nature of the disease i.e. right after the golden standard test is conducted.
+ Conclusion: CTA has a good sensitivity of 85% and an excellent specificity of 98.6%.
4. Evidence table
Pre-test Probability
(%) Sensitivity (%)
[95% CI] Specificity
(%)
[95% CI] PV+
(%)
[95% CI] PV-
(%)
[95% CI]
Population from article 22
85
[61.1; 96.0]
98.6
[91.3; 100] 94.4
[70.6; 99.7] 95.9
[87.7; 98.9]
Upper-limit pre-test probability range patient 25
85 98.6
95.3 95.2
Lower-limit pre-test probability range patient 10 85 98.6
87.1 98.3
Table 1: Evidence table based on the test characteristics at the upper and lower limit of prior probability.
The upper limit and the lower limit of the prior range, as evaluated from the chosen clinical scenario can exclude a PE reliably, given the high negative predictive values
(95.2% and respectively 98.3%).
5 Comments:
Having taken into account the choice of index test, the concerted efforts to avoid bias and the reproducibility of the results, the methodological quality of the article can be regarded as satisfactory. Hence it can be concluded that the evidence presented is applicable to the patient in the clinical scenario. This patient fits with the study population, as the symptomatology (dyspnea, chest pain, hemoptysis, palpitations) and the medical history prompt a referral for CTPA due to suspicion of PE.
The index test is applied to a medical event that is fully treatable, if therapy is delivered in a timely manner. This test has certain advantages over the golden standard, such as absence of radiation and of nephrotoxic contrast substances. Moreover, MRI has a similar performance with CTPA. However, there is a particular scenario in which MRI underperforms: in the worst-case scenario in patient with high risk, this diagnosis tool cannot exclude with enough certainty the presence of a PE (displaying a negative predictive value below 90%). This is the only situation in which a CTPA would still need to be performed.
6 Bottom line:
The patient is in the clinical scenario who is a 32-year old female with a symptomatology indicative of PE (dyspnea, chest pain, hemoptysis and palpitations), under current treatment with contraceptives and medical history of DVT classifies as a high-risk patient. Given the insufficiently high negative predictive value (for high risk in worst-case scenario), MRI would not be sufficient for excluding PE in this patient, despite the good quality of the article.
APPENDIX
Critical appraisal of the article
Criterion 1:
In the chosen study, performance of the index test (MRI) is compared against the CTPA which although does not have a sensitivity of 100% represents the current gold standard. The literature offers various values for CTPA sensitivity for PE, varying from 80% to almost 100%. Out of the 93 patients that were initially included in the study 2 were excluded because they could not undergo MRI. The remaining 91 participants all undertook both the index and the reference test, with no delay between the two measurements so that they can be compared with each other. The advantage of the CTPA is that as the gold standard, treatment is provided immediately after, which is vital in the cases of acute PE.
Conclusion: The results of the index test, MRI, were compared to gold standard, the CTPA, which has a high quality. (++)
Criterion 2:
The study population comprised of 93 patients that were sent to the Radiology Department to have a CTPA performed. Out of them, 59 were women and 34 were men, with ages between 22 and 89 years old. The average age was 63 years. The patients were recruited solely based on sequence (consecutive patients). The reason for referral was suspicion of acute PE. The study did not consider the symptomatology that prompted the referral, possibly due to the non-specific manifestation of acute PE. The study also makes no reference to the risk factors of PE, such as previous DVT, pregnancy, malignancy or cardiovascular problems (hyperlipidemia, hypertension, atherosclerosis).
Conclusion: Although broadly applicable, the results are not characterizing for a population of patients with PE that have a particular symptomatology and the recognized risk factors the disease. (-)
Criterion 3:
Yes, the study methodology avoided work-up bias. The entire study population initially underwent a multidetector CTPA followed by an MRI. For patients with confirmed PE there was no delay in administering treatment after the CTPA. The exclusion criteria was referring to patients who were not able to undergo MRI from reasons such as claustrophobia or CTPA due to reactions to the contrast solutions.
Conclusion: The entire population was subjected to both CTPA and MRI. (++)
Criterion 4:
The results of the MRI tests were interpreted separately by two radiologists, with 9 and 10 years of working experience, who were not aware of the CTPA results. The recall bias was prevented by having a third independent reader with 25 years of working experience who assessed the CT findings. Lastly, there was a fourth radiologists who checked whether the embolisms found on the MRI correspond to those found on the CTPA.
Conclusion: The study was performed with consistent effort to avoid expectation bias. All evaluators were blinded to each other’s results. Moreover, the specialists were chosen in order to leave as little chance to error as possible. (++)
Criterion 5:
The MRI findings were examined by two different radiologists in order to evaluate inter-observer agreement. The parameters for the kappa values set before the analysis was poor agreement for kappa below 0.20, fair between 0.21 and 0.40, moderate between 0.41 and 0.60 and good between 0.81 and 1.00. These values are the generally accepted ones within the research community. The kappa obtained in the study was 0.87, which shows a strong agreement. Study can be rendered as reproducible
Conclusion: With a high kappa value of 0.87 study shows a good reproducibility. (++)
Criterion 6:
The confidence intervals for the values characterizing the validity refer to the analysis of MRI results per-patient, although the study includes also values calculated per-embolus. The reason for choosing to base the CAT on the per-patient analysis is that the results are more relevant for clinical practice, compared to the per-embolus one. Moreover, inter-reader agreement is calculated for the per-patient analysis alone.
The confidence intervals are as follows: 61.1 – 96% for sensitivity, 91.3 – 100 % for specificity, 70.6 – 99.7 % for positive predictive value and 87.7 – 98.9% for negative predictive value. In order to exclude PE with certainty following an MRI, the negative predictive value and the sensitivity are the most relevant characteristics. Whereas the negative predictive value has a good lower limit (of 87.7%), the lower limit for sensitivity (61.1%) is not satisfactory.
Conclusion: The 95% CI’s are acceptable for the negative predictive value (lower limit of 87.7%) but unacceptable for sensitivity (lower limit of 61.1%) which are the two most important parameters of the MRI for excluding PE. (+/-)
Criterion 7:
Yes, the study uses clear cutoff points for the establishing a PE diagnosis. There had to be an concordant result in two different planes together with one of the following findings: a clear thrombus seen on the image, breakoff of one of the pulmonary vessels, abrupt changes in how the intensity of the signal within a pulmonary vessel. These findings are the internationally accepted for PE imaging diagnosis.
Conclusion: Clear cut-off points were established for PE diagnosis (clear thrombus seen on the image, breakoff of one of the pulmonary vessels, abrupt changes signal intensity within a pulmonary vessel) which are unanimously recognized. (++)
Criterion 8:
As the purpose of the study is to evaluate whether the gold standard, namely MRI can be replaced by the index text, namely CTPA, the MRI procedure was not delayed after the moment when CTPA was conducted. It can be assumed that the reason for firstly using CTPA is that PE is an acute life-threatening and reliable diagnosis should be performed in order to administer treatment as quickly as possible.
Conclusion: The index place is conducted at the most logical point in time given the nature of the disease i.e. right after the golden standard test is conducted. (++)
Criteria 9:
PE+ PE- Total
MRI+ 17 1 18
MRI- 3 70 73
Total 20 71 91
Table 2: Table based on the pre-test probability and the number of patients in the selected article
Sensitivity: 85% (95% CI: 61.1; 96.0)
Specificity: 98.6% (95% CI: 91.3; 100)
PV+: 94.4% (95% CI: 70.6; 99.7)
PV-: 95.9% (95% CI: 87.7; 98.9)
Pre-test probability: 22% (95% CI: 14.2; 32.1)
Conclusion: CTA has a good sensitivity of 85% and an excellent specificity of 98.6%. (+)
Criteria for the evaluation of the applicability of a diagnostic test
Does the test identify a treatable disorder?
Yes, acute PE is a treatable disorder. The first line of treatment is thrombolytic medication. In cases in which patient has a massive PE which heavily obstructs pulmonary circulation leading to hemodynamic instability or in cases of ventricular dysfunction, a percutaneous pulmonary embolectomy is indicated.
Will the test be better than the test already being used?
There are same clear advantages that MRI possess above CTPA, the most obvious ones being the lack of irradiation and of nephrotoxic contrast. These features make MRI applicable for a broader range of patient, including pregnant women or renal insufficiency patients. Since the goal is to exclude PE, the most important parameter when deciding if MRI is a good alternative for CTPA is the negative predictive value. In both lower and upper limit prior probabilities, this value is satisfactory.
Is the test reasonably priced?
Although more expensive than CTPA, the difference of approximately 40 euros between the two modalities is still acceptable (160 vs 200 eur)
Will the test be accepted by the patient/Is the test burdensome for the patient?
Yes, unenhanced MRI, although taking longer is a non-invasive diagnosis tool. Nevertheless, for a patient who suffers from claustrophobia this test would be very burdensome (and so would the CTPA).
Will the test influence the treatment plan?
Taking first a negative result into account, both the upper and the lower limits of the prior probability are very good (above 90%) for the patient in the observed scenario. Most of the case scenarios also proved good values, which would lead exclusion of a PE without further diagnosis. However, there is an exception, when worst case scenario is applied to the upper limit of prior probability. In this case the result is 87.5% hence the probability that the patient has PE is still 12.5%. This value is too high and as a result patient still needs to undergo a CTPA, which disregards MRI as the superior alternative.
Taking a positive result into account, 85% is a reasonable value to start treatment. The reason for it is that treatment can cause bleeding and hence it should not be administered unless there is enough certainty that patient would benefit from administration. In the worst-case scenario, for both upper and lower limit of pre-test probability, the results are not sufficient to proceed with treatment and hence CTPA should follow.
Post-test probability of
PE (T+)
Observed case scenario (%) Worst case scenario (%) Best case scenario (%)
Prior probability 25% 95.3
(proceed
with treatment) 70
(CTPA scan)
100
(proceed
with treatment)
Prior probability 10% 87.1
(proceed
with treatment)
43.8
(CTPA scan) 100
(proceed
with treatment)
Post -test probability of
PE (T-) Observed case scenario (%) Worst case scenario (%) Best case scenario (%)
Prior probability 25% 95.2
(no further
diagnostic testing) 87.5
(CTPA scan) 98.9
(no further
diagnostic testing)
Prior probability 10% 98.3
(no further
diagnostic testing) 95.5
(no further
diagnostic testing) 99.6
(no further
diagnostic testing)
Table 3: Evidence table based on the observed, best- and worst-case scenario at the upper and lower limit of prior probability