CT angiography for acute GI haemorrhage is the dogs danglies: have we gone to far?

Created - 13.08.2013

My week of on call is around the corner and it comes with the joyous prospect of being awoken at night by a junior doctor asking for a mesenteric angiogram on a fit 90 year old with a do not resuscitate order.

This brings me onto the role of CT angiography in acute GI haemorrhage. I trained in the era where CT angiography is all the fashion. A quick Medline search would bring up numerous small studies all expounding its virtues. CT angiography is great, the feckless are happy that something is being done, it’ll probably show no bleeding, and be reported by the radiology registrar on call.

The inevitable is delayed until the morning, back to sleep, job done. I am being facetious. 

I am a believer in CT angiography. Ok, for upper GI haemorrhage where endoscopy has identified but failed to deal with the bleeding ulcer, CT has no role and in those cases nearly all of us (I think) would proceed to catheter angiography and consider embolising the GDA whether or not there is active bleeding. On a side note, empirical GDA embolisation is a win win situation, if the bleeding stops, it is because we coiled the artery, if they bleed again, we’ve done all we can, over to you maestro.

The dilemma lies with lower GI haemorrhage, where the initial CT angiogram is negative (and yes, it was performed when the patient was actively bleeding), or in upper GI haemorrhage in a patient with pancreatitis or a previous Whipples when the pseudoaneurysm is not identified amongst the upper abdominal mush. 

It will be almost three years since I have been let lose and allowed to practice autonomously (well kind of, I have a Scottish professor to answer to at times) and there have been numerous instances where I am glad I did a CT angiogram and a few where I wished I had done one. My dilemma is what to do when the initial CT is negative?

I am not sure I agree with the mantra of repeated CT angiography in the hope that eventually radiation exposure will stop the bleeding. There is a nice article by RoyChoudhury SH et al. (AJR. 2007 Nov; 189(5): W238-46) which uses an invitro model to assess the threshold of detection of active arterial bleeding. It states that CT can detect up bleeding as low as 0.35-0.5ml/min whilst catheter angiography detects only 1ml/min. But if you read on, the 1ml/min quoted for catheter angiography is for non-selective angiography (ie a pigtail in the aorta). If you perform more selective runs, this threshold drops to to 0.05ml/min with the catheter tip at the bleeding point (10 times better than CT).

But we all know this. Propping up a bar at many an IR meeting, I heard the more experienced amongst us voice their honest inebriated opinion. The vessel detail that catheter angiography gives you is not (yet) matched by CT. An abnormal vessel does not need to be actively bleeding to be identified on catheter angiography. Beaded, narrowed or truncated segments may often identify the perpertator.

I think that the treatment algorithim is still evolving. Although CT angiography plays an important role it would be wrong to dismiss the benefit of catheter angiography when CT is negative and the patient is still bleeding/unstable. As a side note, I generally get an anesthetist out of bed to put the patient to sleep if the patient is not compliant. If they refuse, then its pointless performing a suboptimal angiogram on a writhing elderly patient who cannot breathold. They can operate on the patient. Often at this point, the surgeon intervenes and I get my anaesthetist. 

I’ve never written for the newsletter before and it has been a liberating experience not to have to use endnote or back up my diatribe with evidence. However, as I write this I can sense my colleagues cycling away to distance themselves and I feel duty bound to say that any opinion expressed here is not my own but that of the unit in which I am incarcerated. Ha!

Vivek Shrivastava, Consultant interventional/ cardiovascular radiologist, Hull