Stress-related gastric ulceration in critical illness

This chapter deals with protecting ICU patients from their own gastric secretions. This keeps appearing in the Second Part exam:

• Question 1 from the first paper of 2003
• Question 10 from the first paper of 2018
• Question 11 from the first paper of 2020
• Question 1 from the second paper of 2022

The issue of PPI prophylaxis was absent from the CICM Part II exam between 2003 and 2018. Because of this prolonged gap and because the content of assessment is the widely acknowledged driver for Part II revision, the modern-day candidates may not have been well-prepared for this topic. In order to attend to this shortcoming, this chapter has enjoyed an update to reflect contemporary data. It seems important, given that the college keeps asking about it, and some of the more influential examiners are also co-investigators of the Australian trials. 

As far as published evidence goes, the time poor exam candidate would be wise to only choose one recent article and rely on it for all of their revision. Buendgens et al (2016) is a reasonable choice, as  it is available for free. 

"Is this seriously a problem?"

The modern day day trainee, blinded by the limitations of their youth, may disregard this topic as something of trivial importance. The junior doctor in charge of recharting medications may view the prescription of "life-saving pantoprazole" as a mindless application of a poorly supported opinion-based department protocol. However, it is not so. Rather, the almost total eradication of significant stress ulceration from the ICUs of the modern era can mainly be attributed to the constant use of some sort of stress ulcer prophylaxis measures, even if these only consist of enteral nutrition.

Historically, this was a major issue. Lucas et al (1971) report a 5% rate of stress ulcer associated with such blood loss that mortality from these events was 80%; 100% of acute trauma patients had some sort of obvious ulceration on endoscopy. Of highly at-risk patients (respiratory failure multi-trauma etc) Hastings et al (1971) found a bleeding rate of around 25%. In short, without "life-saving pantoprazole" fully one quarter of your patients might end up having a serious GI haemorrhage.

Causes of stress ulceration

Question 10 from the first paper of 2018 asked about this specifically, but five risk factoors would have only earned you 20% of the 10-mark SAQ. This makes sense because it would have been relatively easy to score full marks, because any damn thing will cause stress ulceration in the ICU. The gastric mucosa is like the canary in the coal mine, swooning at the first sign of any sort of calamity. According to Cook et al (1994) and Steinberg et al (2002),  independent risk factors for stress ulceration are:

• Respiratory failure with mechanical ventilation for >48 hrs
• Coagulopathy or anticoagulant use
• Hypotension, shock states of any sort
• Liver failure 
• Renal failure and use of renal replacement therapy
• Fasting state (no enteral feeding)
• Steroid use (especially dexamethasone)
• Presence of more than three comorbidities

According to Buendgens et al (2016), a whole other range of lower-risk associations exist, which include:

• Severe head trauma
• Surgeries lasting over 4 hours
• Burns
• Advanced age
• Male sex

So, virtually everybody in the ICU would qualify, according to these criteria. The perception that everybody is at risk has resulted in the use of PPI in about 80% of ICU patients. This makes some sense, because after three days of ICU stay we still see gastric erosions forming in about 90% of ICU patients, even in the modern era (Krag et al, 2016)

Pathogenesis of stress ulceration

This is a topic of some debate.

It is thought that the gastric mucosa is a very energy-hungry organ. The process of maintaining a massive pH gradient and secreting all that mucus is in fact a very ATP-hungry process. The oxygen extraction ratio of the gastric mucosa must approach 100% at normal physiological conditions.

Thus any decrease in gastric blood flow would result in a diminished capacity to secrete this precious protective mucus. The result would be horrific. As the mucus dissipates, the gastric mucosa finds itself defenseless against the hideous environment of the gastric lumen, which by its hospitality resembles the atmosphere of Venus.

Increase of gastric pH as a therapeutic goal

This is what people usually mean when they say "stress ulcer prrophylaxis". In order to protect the delicate gastric mucosa, one would want a slightly less acidic stomach. Thus, one ought to aim for a pH of around 3.5, roughly the same acidity as the inside of a grapefruit.

Even though this sounds sensible, nobody is entirely sure whether it works. Certainly the risk of gastric bleeding is reduced by this strategy, but there are other issues.

For instance, the alkaline environment of the stomach will now support the growth of oral bacteria, and thus expose the patient to the risk of ventilator-associated pneumonia. Not only aspiration is a concern - one study had linked the use of PPIs with a two-fold increase in the risk of C.difficile infection

Also, PPIs are not exactly benign drugs. Pantoprazole, for instance, can cause all sort of liver function abnormality. In short, it is no wonder that large scale studies of stress ulcer prophylaxis have not revealed any improvement in mortality. Some episodes of bleeding are prevented, to be sure - but some episodes of VAP may be created as a result, and ultimately a balance is achieved.

A recent meta-analysis suggests that the quality and quantity of the evidence for the use of stress ulcer prophylaxis is actually quite poor. There is neither a mortality improvement nor any increase in the risk of nosocomial pneumonia.

Choice of stress ulcer prophylaxis agent

Generally speaking, you have choice of four major drug classes:

• PPIs
• Antacids
• Sufralfate
• H2-receptor antagonists

Additionally, enteral nutrition could be viewed as a non-pharmacological form of stress ulcer prophylaxis.

Each class has its advantages and disadvantages. Question 10 from the first paper of 2018 also wanted the candidates to "include in your answer the available evidence for these"

• Antacids - These were the mainstay of ulcer prophylaxis in the 1970s and 1980s (see Gonzalez et al, 1985). They were given hourly, and required hourly gastric pH measurement to titrate (the usual goal was to aim for a gastric pH of > 5.0). This strategy was certainly more effective than doing nothing. In 1971, Hastings et al performed an RCT of antacid titration (to gastric pH >3.5) versus no therapy, among highly at-risk patients: the bleeding rate was decreased from 24.5% to 3.9%, i.e. the risk was six times higher without the antacid.  Unfortunately, the available agents tend to have nasty metallic ions eg. magnesium, aluminium, sodium (eg. Mylanta is a mixture of aluminium hydroxide and magnesium hydroxide). Also, one might develop some metabolic alkalosis, or worse yet a bowel obstruction.
• Sucralfate - an aluminium salt of sucrose octasulphate - coats ulcers and increases mucus production, but does absolutely nothing for gastric pH. This may be an advantage, as the low pH can continue killing pathogenic organisms, while the mucosa is protected by a thick coat of nicely viscous mucus. Again, constipation is a major side-effect. Sucralfate can form a bezoar, clogging the gastric outlet, and it can prevent the absorption of nutrients by binding to the components of enteral feeds. It is also thought to have a tendency to absorb or adsorb useful medications. The popularity of sucralfate decreased in the late 1990s when a trial compared it unfavorably to ranitidine (Cook et al, 1998); in fact ranitidine was twice as good.
• H2-receptor antagonists like ranitidine became popular in the wake of trials which favourable compared them to antacids and sucralfate. Unfortunately these drugs have a tendency towards tachyphylaxis after the first day of therapy. Their use has been largely superceded by PPIs; however their one major remaining advantage is cost. Wikipedia lists the wholesale price of one tablet as 0.01$ (US), making it attractive in the developing world
• Proton pump inhibitors eg. pantoprazole and omeprazole became available in the nineties and have subsequently superceded the H2 receptor antagonists . Alshamsi et al (2016) in a meta-analysis of 19 trials (2117 patients) found that overall there was a reduction in the risk of clinically signficant bleeding when compared to H2 receptor antagonists, but with a NNT of 37. There was no mortality benefit. Also, thus far nobody can tell which patients are at sufficiently high risk to benefit from them
• Enteral nutrition buffers gastric pH and suppresses stress-induced vagal stimulation, while proving trophic and circulatory benefits to the gut mucosa, but is occasionally impractical (i.e. not always can you feed them). Furthermore, these benefits were demonstrated mainly in animal studies and retrospective audits (MacLaren, 2001). A more recent prospective RCT (El-Kersh, 2018) found no difference, but only 2 patients had GI bleeding in each group. Ultimately, the benefit of enteral nutrution as stress ulcer prophylaxis has been extrapolated from the finding that it is usually safe to withold PPIs from patients who are tolerating enteral nutrition, i.e. with no additional GI bleeding risk (Alhazzani et al, 2017)

There does not seem to be any agreement regarding which agent is best, if one were going to give stress ulcer prophylaxis. The Danes recommend PPIs, because they are more effective at keeping the gastric pH at over 4.0. Recent meta-analysis publications listed in Buendgens et al (2016) redemonstrated their superiority over antihistamines and the superiority of antihistamines over sucralfate.

A critical evaluation of routine stress ulcer prophylaxis

For the purpose of answering an (unlikely) future question on this topic, the savvy candidate will have prepared a certain small volume of clever-sounding bullshit to discharge onto the exam paper. Such a volume is offered below, as response to a request to critically evaluate the strategies for prevention of gastrointestinal bleeding in the critically ill.

Introduction

• Gastrointestinal bleeding in the critically ill patient may be due to a variety of causes; these include bleeding from stress ulceration, oesophageal varices, and colonic polyps.
• ICU patients are also more susceptible to GI bleeding because they often have concomitant coagulopathy, are under the effect of antiplatelet agents, or have poor splanchnic perfusion due to the effects of shock 
• Given that in the ICU GI bleeding is combined with a series of other major organ dysfunction syndromes, it tends to have a catastrophic mortality rate
• It is therefore important to be able to protect at-risk patients from this complication.

Rationale and counter-rationale

Advantages

• Decreased risk of gastrointestinal bleeding
• Decreased exposure to blood products, and the attendent risks thereof

Disadvantages

• Decreased gastric acidity, thus increased risk of non-sterile aspiration
• Increased risk of gastrointestinal bacterial overgrowth and translocation
• Increased risk of Clostridium difficile infections
• Economic disadvantage (cost)
• Side effects of the anti-ulcer agents (including drug interactions)

Evidence for the routine use of ulcer prophylaxis

• A recent meta-analysis suggests that the quality and quantity of the evidence is still poor, but on the weight of the available evidence there is neither a mortality improvement nor any increase in the risk of nosocomial pneumonia.
• There appears to be no benefit for stress ulcer prophylaxis in patients who are tolerating enteral feeding, and in these patients stress ulcer prophylaxis is not needed.
• There is insufficent evidence to recommend the mandatory use of stress ulcer prophylaxis in any specific patient group 
• Obviously, correcting coagulopathy (and not triggering any new coagulopathy, nor disabling the platelets with NSAIDs) is a good way to prevent catastropic bleeding in the ICU.
• SUP-ICU (Krag et al, 2018) did not find any mortality benefit, but there was a reduction in clinically significant UGI bleeding events. Having said this, even in the control group, the risk of this bleeding was only 4.4%. The take home message for many people was that PPIs were probably less essential than everybody had once thought, as many ulcer-producing disease states are being handled much better these days. For one, we are no longer ventilating people with tidal volumes of 1000ml.

Evidence to support one drug class over another

• At the time of the 2003 exam paper having been written, there was insufficient evidence to recommend any specific medication (e.e H₂As vs PPIs); however contemporray meta-analsysis suggests some benefit from the use of PPIs.  Pro-PPI studies include a big 2016 meta-analysis by Alshamsi et al, which revealed them to be more effective at preventing clinically significant episodes of bleeding. However, it must be pointed out that many of the studies which met inclusion criteria didn't even specify what they meant by "bleeding". When other meta-analysis authors selected studies  limited to ones with a low risk of bias, the results they arrived at were not significant (Barletta et al, 2016).
• PEPTIC, a 2020 ANZICS CTG megatrial, needs to be mentioned here because it was a direct comparison of H2 receptor antagonists and PPIs. Without going into too much detail (because The Bottom Line have already done it better), its results can be summarised as "there is not much difference". The 90-day mortality difference seemed to favour ranitidine, but did not reach statistical significance (0.8%), except in the group of post-cardiac-surgery patients.  Clinically significant bleeding did reach statistical significance, but the difference was 0.5% (1.3% vs 1.8%).  The Bottom Line authors felt that the statistically insignificant 0.8% mortality benefit in the H2 antagonist group was still a big deal, enough to say that "this study supports H2RB as the default agent for stress ulcer prophylaxis in ICU". Fortunately, the decision between PPIs and H2RBs in 2020 was rendered somewhat easier, as most of the world's supply of ranitidine ended up being contaminated with a hepatotoxic carcinogen.

Summary

PPIs are indicated in at-risk patient in ICU who are intolerant of enteral feeding, and who are otherwise at risk of gastrointestinal bleeding. H2 agonists may have a statistically insignificant significant mortality advantage over PPIs. Further research is required to identify the at-risk population.