Full Post

« Validation of San Fransisco Syncope Rule
tPA in CVA: Blood, Sweat and Tears »

CTA vs. CTA/CTV for Pulmonary Embolism

Annals was mostly about airway this month, and felt a little sparse (Levitan showed that BURP and cricoid pressure worsen the view compared to bimanual laryngoscopy– stop the presses! Also, a letter to the editor advocated for the mnemonic LEMONS over LEMON — the extra S is for O2 saturation, which of course you might otherwise fail to consider as you're prepping to intubate… sheesh).

So, instead, I thought I'd hit up that other noteworthy periodical, the New England Journal of Medicine. This week (June 1, 2006, Vol 354, No. 22) they've got an article (pdf) from the PIOPED II study about the diagnostic value of CT angio alone, vs. CT angio plus CT venography of leg veins.

I think the lead author is the same guy that showed S1Q3T3 is neither sensitive nor specific for PE, back in 1975. What a career.

Anyway, this group recruited patients over 18 that had suspected acute PE, inpatient or outpatient, who were being referred for diagnostic imaging, or whom some kind of PE "consult" was ordered, during daytime weekdays. Of the 7284 suspected PE patients, more than half had to be excluded (4022 excluded, for reasons like renal failure, unable to be tested within 36 hr, pregnancy, MI, on a vent already, on anticoagulants already, had a filter, were in shock, were in prison…) Another 2172 refused to enroll or couldn't finish the protocol.

Of the 1090 remaining, 61% were women, 31% were black, and 89% were in outpatient settings (including rehab and nursing homes). 52% were smokers, 52% had tachypnea, 36% had a swollen calf… Yet only 6% were classified as high probability (See Wells score, below).

Anyway, only 824 got a CT and a definitive diagnosis (by digital subtraction angiography)…

Big findings:

  • Positive results on CTA alone, in combination with a high or intermediate probability of PE (based on clinical assessment), had a PPV of 92-96%
  • Normal findings on CTA with a low clinical probability had a NPV of 96%. 
  • CTA alone had a sensitivity of 83%.
  • CTA with CTV boosted sensitivity to 90%. Specificity was 95%, with or without CTV.

Other notable findings: patients with low-probability who got a positive CTA or CTA-CTV — 42% to 43% of those are false positives. Yow.

And on the flipside, if a patient started with high pretest probability, 40% had false-negative CTA and 18% had false negative CTA-CTV. (But then again, we know that a negative CT in a patient with moderate or high probability requires further workup — see Kline, Annals Emerg Med 2000, 35:168).

So, I don't think this is blockbuster stuff. I mean, if we're not going to make clinical assessments anymore, then yes, we should do CTA/CTV to catch more PE's. But why not just apply Well's criteria before your CT-Angio? (I'm writing this out because I've always wanted a quickly accessible version online, and emcrit.org is just too thorough on this topic):

  • 1.0 points for Hemoptysis
  • 1.0 points Cancer
  • 1.5 points for Previous diagnosed PE , or DVT
  • 1.5 points for HR > 100
  • 1.5 points for immobilization more than three days(bedrest with bathroom mobility), or surgery within 4 wks
  • 3.0 points for "signs and symptoms" — leg swelling, pain to palpation over leg veins
  • 3.0 points for PE being as likely or more likely than alternatives, based on history, exam, EKG, CXR, bloodwork

Scores of less than 2 mean low probaility (meaning, only 3.6 % of these patients had a PE), between 2-6 is moderate prob, and more than 6.0 is high probablity (meaning, 66% of these pts had a PE).

The only real use of this paper, I think, is in cases where there's high pretest probability, where a CTA/CTV would cut down on false negatives that just irritate everyone and lead to more testing or delay therapy.

Supplementary material on these methods, scanners used, etc is available at nejm.org . Most of this data was with 4-slice CT, they don't have enough 8- or 16-slice data to say for sure if it's better. Does anyone have info on CTA/CTV availability at our sites, is it practiced?

For completion's sake, here's a link to the ACEP guidelines on PE (pdf) – they're comprehensive and can save your career… In 2003, they advised (level B+C recommendations) that spiral CT could replace V/Q scanning for PE evaluation. Now, with PIOPED II, it may become Level A.

Posted on Sunday, June 4th, 2006 at 10:21 pm by Nick. Filed under Pulmonary Embolism, Risk Stratification, Journal Club.
You may post a comment.

3 Responses to “CTA vs. CTA/CTV for Pulmonary Embolism”

  1. I didn't dwell on this in the post, but it's becoming important on the EMED-L discussion group: The authors' gold standard for CT wasn't always digital subtraction angiography, but more often, simply a high-probability V/Q scan…

    This perplexes some, who think the V/Q is less reliable than a CT. If the V/Q scan is likely to generate false positives, then calling it a gold standard means the CT will have more than its share of false negatives:

    Most importantly, 57% of the "true" positive tests were due to a high probability V/Q scan. I have a major problem with using a high probability V/Q scan as being the absolute reference standard. They claim that previous studies showed that this composite reference standard would only have a FP rate of
    Jeffrey Mann continues:

    I noted that Perrier wrote the accompanying editorial. I am amazed by his commentary. He laments the fact that sensivity of a CT scan was only 83% and that the FN rate was 17%. How does he account for the high FN rate of 17%? This is what he states-: "Therefore, the most likely explanation for the findings of the PIOPED II study is that although multidetector (mainly four-slice) CTA is more sensitive than singledetector imaging, it still misses small, peripheral subsegmental clots that are better detected by ventilation–perfusion scintigraphy or classic pulmonary angiography." Surely, this statement is crazy/invalid. If a CT scan is missing small peripheral subsegmental clots (which is indeed possible), how would we know that fact? Remember that 57% of the positive PE-diagnoses in the PIOPED II study were made on the basis of a high probability V/Q scan and I cannot understand how "missed" small peripheral subsegmental clots can produce a high probability V/Q scan.

    Wow, people are getting riled up. I thought false positives in V/Q scan were low (consistent with the


    Comment by Nick on June 6th, 2006 at 10:25 am

  2. I thought false positives in V/Q scan were low (consistent with the <10% finding above), but then again, I've never ordered one.

    Perrier's editorial, on page 2383 of this same journal, goes on to note that whatever false negatives CTA/CTV misses is disappointing, but probably these are small PE's that resolve spontaneously, ie, of questionable significance clinically. Perrier is confident of this, because he ran a study in last year's NEJM showing good outcomes in those NOT anticoagulated because they had normal CTA's.

    Perrier's final conclusion: "CTV does not appear to improve the diagnostic yield of the CTA enough to justify the additional radiation."


    Comment by Nick on June 6th, 2006 at 10:30 am

  3. False Positive rate of a High-Prob V/Q was 12% in PIOPED I. This comes pretty close to the 17% “false neg rate” in PIOPED II. PIOPED II’s methodology was a major disappointment to everyone who follows the PE literature. What could have been a definitive study nw just leaves the waters muddy.


    Comment by weings01 on June 23rd, 2006 at 7:43 pm



Leave a Reply