Hello, everyone. My name is Samir Khadri, and I'm a practicing intensivist at the NIH Clinical Center and head the Clinical Epidemiology section in the Clinical Care Medicine Department there. And I'm here to talk to you about biomarkers of infection in the ICU. My focus is going to be predominantly on procalcitonin as the biomarker that has entered into our guidelines that is being ordered very often and still remains highly controversial. I have no financial disclosures. I serve on the IDSA sepsis task force and CMS and ASPE technical expert panels for sepsis and antibiotic resistance, and I serve as an associate editor for the journal Critical Care Medicine. I'll start with some basics on biomarkers, sepsis, and antibiotic use, and then we'll take a deep dive into the properties and performance of procalcitonin, as well as its role in antibiotic initiation, prognostication, escalation, and antibiotic discontinuation, as well as its role in COVID-19 and future implications. Sepsis continues to be a huge problem, being the number one cause of hospital death today, with over a quarter million people each year in the US and over 5 million people around the globe dying of sepsis. Unfortunately, the diagnosis of sepsis remains subjective. Even among intensivists who diagnose sepsis every day, who were surveyed in the study, inter-rater reliability was extremely poor. And the problem is there is no good gold standard for sepsis even today. And this lack of a gold standard for sepsis diagnosis results in multiple layers of uncertainty in managing these patients. For instance, should I treat empirically? What's the bug? How long do I treat? Should I use broad-spectrum agents? And this uncertainty is what drives excessively broad therapy in sepsis. In this study of community-onset culture-positive sepsis, there was a large amount of broad-spectrum antibiotic being used and still a relatively smaller amount of resistant pathogens. And there's even greater uncertainty in the sizable proportion of cases across the sepsis spectrum that end up being culture-negative. And this time-to-appropriate antibiotics study has shown us that culture-negative sepsis is probably just as bad as culture-positive sepsis. As critical care providers, myself included, we often indulge in what is called hyperbolic discounting or ignoring the distant threat. And for the most part, it's for good reason. We live in this inner sphere, as you can see here, the provider, the patient, the pathogen, and the antibiotic. And we do everything we can to try to improve the outcome of the patient. And our decisions are focused on that. However, we cannot ignore that there is a environment around us, which is the hospital environment. And then there is an environment around that, which is the world we live in. And unfortunately, it's multiple decisions that we have taken in that inner sphere over time that has catapulted the antibiotic resistance crisis into what we know of it today. And I think it's really important for us to, while we're taking care of the patient, just remember in the back of our minds that our decisions can lead to future implications. And I think it's important to have both a patient as well as a population perspective in our decisions. And this slide is a good reminder of that. In 2007, in a point prevalence study, EPIC-II, 1265 ICU, 76 countries, the question was asked, what proportion of your ICU patients are on antibiotics? And it was 71%. Now, fast forward 10 years, and the EPIC-III study was performed similarly, a point prevalence study, 1150 ICUs, 88 countries. And the same question was asked again. Now, mind you, in this decade, a lot happened in increasing awareness, action, and implementation around antibiotic stewardship. Unfortunately, the proportion is still 70%. Can biomarkers help us do better? Biomarkers broadly defined are markers of a biological process or state. Before integrating biomarkers into diagnosis and management, we ought to ask ourselves the question, do they inform anything over conventional clinical judgment? Do they improve patient outcomes? And do they increase cross-effectiveness? There's a variety of biomarkers that we use today in managing patients. These can range from, for instance, markers of organ dysfunction, like the troponin, or microbiology, like pneumococcal urinary antigen. But what we're talking about today is those biomarkers that help us differentiate between bacterial infection versus not bacterial infection in these sort of nonspecific cases of clinical, clinical syndromes like sepsis and pneumonia, where we're not really sure. Over the years, we have learned a lot about the pathobiology, diagnosis, and management of sepsis. And part of that effort has been in understanding biomarkers. The idea of sepsis biomarkers came about in 1988, and a number of candidate molecules have been tried and tested. Some of these have been shown here. And really, the only one that has made primetime guidelines has been procalcitonin. Procalcitonin first received FDA approval in 2007 for risk assessment of sepsis on ICU admission. In 2016, it received approval for aiding in prognosticating severe sepsis using serial PCTs. And in 2017, to aid in the starting and stopping of antibiotics in loris protracted infection, as well as stopping of antibiotics in suspected sepsis. But before delving into the clinical evidence that underpins these recommendations, let's understand a little bit more about procalcitonin's properties and performance. Calcitonin is the pro-hormone for calcitonin. Calcitonin is secreted by the C cells of the thyroid. But procalcitonin is also secreted constitutively by multiple different cells in the body in response to bacterial infection. As seen here in this study by Asikot and colleagues, they looked at procalcitonin concentrations for a variety of scenarios. And as you can see here, that the levels in severe bacterial infection were much higher than local infections, viral infections, and no infections. And this finding was sort of a eureka moment that sort of resulted in the procalcitonin cascade of investigations that followed. There are some unique properties of procalcitonin kinetics that make it desirable for use as a biomarker. Like, for instance, if you compare it to CRP, here, procalcitonin has a rapid rise and a peak in 12 to 24 hours. In comparison, CRP has more of sort of a delayed upslope, making it less useful in the first couple hours of presentation. Procalcitonin outperforms many of the other biomarkers in discriminating bacterial infection. In this meta-analysis of 12 studies, PCT showed better accuracy than CRP in diagnosing bacterial versus no infection. However, that signal is not necessarily uniform across the board. In this prospective study of 279 ICU patients with SIRS, PCT performed less well than CRP. However, I think the take-home message from this slide for you all is that biomarkers, in general, have moderate to poor performance, at best, in discriminating sepsis from non-septic SIRS. In this meta-analysis of 30 studies spanning more than 3,000 critically ill patients, more than half of them were septic, the procalcitonin performance metrics were 77% for sensitivity, 79% for specificity, and a pretty decent AOC of 0.85. And so, you know, it's not a perfect marker, but that doesn't exist. As practicing clinicians interpreting procalcitonin results, it's important to know when false positive results can occur. These can happen in non-infectious conditions like trauma, surgery, hypothermia, renal failure, neoplasms, as well as even non-bacterial infections like malaria, dengue, and severe COVID-19. The one to definitely remember is acute renal failure. PCT levels are elevated in this condition that is seen so commonly in our ICU. In this study of patients who underwent elective aortic surgery, the patients were divided into those who had infection and didn't have infection. So even in the absence of infection, as seen on the left-hand side, AKI was associated with higher levels of procalcitonin. As providers that use continuous renal replacement therapy often, it's important for us to understand that even CRRT can decrease procalcitonin levels. As seen here, the afferent plasma concentrations of PCT are... were higher in this study compared to efferent plasma concentrations. Now let's see whether the variability of procalcitonin levels across pathogens is clinically helpful. PCT distributions differ across different types of pathogens. It's high for gram-negatives compared to gram-positives in fungal species. And one wonders about the clinical utility of this information because the... the sensitivity of 69 and a specificity of 77 percent, I wonder whether this tells us anything more than our clinical gestalt. Even in pneumonia, comparing typical versus atypical patients, there is clearly a difference as seen here in the distributions. But again, you know, is this really helpful in adequately ruling out one type versus the other? In this study led by Alex Lawandi, we showed that in patients with bloodstream infections, procalcitonin not only varies a lot by pathogen, but it also varies a lot by site, by presence or absence of shock, and by SOFA score. And I think this is important because as clinicians, we should realize that there are really multiple drivers here of the procalcitonin level. So, it's difficult to really pinpoint differences in procalcitonin levels by one of these factors. Even if you look at sensitivity, as seen here in that study, where patients have been divided with sepsis and bacteremia versus occult bacteremia, no sepsis, you can see why... that while the sensitivity is decent for sepsis, it's quite dismal for bacteremia without sepsis. And one might imagine that it is this group of the more occult cases that might have benefited from a biomarker that is able to capture cases that are often missed clinically. But that doesn't seem to be the case, at least looking at the sensitivity. Now let's see whether procalcitonin has a role in antibiotic initiation in loris protract infections and suspected sepsis. In this meta-analysis of 12 studies of patients with community-acquired pneumonia, the summary ROC curve shows a AUC of 0.73, and a sensitivity of 0.55, and a specificity of 0.76. So, I would say suboptimal for making an accurate diagnosis of community-acquired pneumonia. PCT discrimination seems to be suboptimal even for ventilator-associated pneumonia, as seen in this study, with an AUC of 0.51 for initial PCT and of 0.62 for a PCT increase in trying to discriminate ventilator-associated pneumonia adjudicated by clinicians. As such, both the ATS-IDSA guideline for community-acquired pneumonia, as well as for hospital and ventilator-associated pneumonia, do not recommend procalcitonin levels to aid in initiation of antibiotic therapy. Every hour gone without antibiotics increases the risk of progression from sepsis to septic shock by about 4%. The surviving sepsis campaign guideline update in 2021 actually has created some room for clinicians to use their gestalt in determining the sort of rapidity of antibiotic therapy that might be needed based on likelihood of infection. If the likelihood is high, they recommend 1 hour. If the likelihood is low, they recommend probably deferring and closely monitoring while rapidly investigating to see if infection concerns persist, and then to administer the antimicrobials within 3 hours. And so, you know, has this created room for perhaps a biomarker like procalcitonin to aid in this decision? Let's take a look. We tried to address this in the study of patients who presented with suspected bloodstream infection to the hospital. And what we find here is, at a cutoff of 0.5 for procalcitonin, we get a sensitivity of 68% and a specificity of 66% for diagnosis of bloodstream infection, and a negative predictive value of 93%. Now, if we alter the cutoff by increasing it, we get a higher specificity at the cost of sensitivity. When we reduce the cutoff to 0.25 or 0.1, the opposite occurs. And while across the board, the negative predictive value is in the 90s, which may give us some reassurance that perhaps we could use a negative procalcitonin to rule out bloodstream infection, but the predictive value of a test is contingent on the prevalence of the disease in the population in which it's being tested. In this study, the proportion of patients presenting with bloodstream infections ranged between 11% to almost 29%. And so, all in all, admission PCT is not reliable for diagnosing bloodstream infections in patients presenting to the hospital. And the moderate to severe discrimination of PCT on admission for diagnosing bacteremia was really seen across the spectrum of sepsis, as seen here. And so, really, the take-home for intensivists contemplating whether or not to use procalcitonin on admission for patients with suspected serious infection must take into account that they were likely to miss a good amount of bacteremia. And if they are likely to miss a good amount of serious infection like bacteremia, then how reliable, really, can we consider procalcitonin to be for admission for any serious infection, for that matter? And this problem is not limited to bloodstream infections. This other study showed a similar signal of an AUC of 0.69 for clinician-adjudicated infection in ICU patients, so not just bloodstream infections. And in this study, nearly a third of patients with no confirmed infection had a PCT of more than 1. And about 15% of cases with confirmed infection had a PCT of less than 0.5. And PCT did not improve concordance between ID and ICU physicians. As such, the 2021 Surviving Sepsis Campaign Guidelines tell us that for adults with sepsis and septic shock, they recommend against using PCT for deciding whether to initiate antibiotics. And granted, this is a weak recommendation with very low-quality evidence. Now, let's see whether procalcitonin can help with prognostication and treatment escalation in critically ill patients and those with sepsis. This study, called as the MOSES study, was a prospective cohort study of patients with sepsis that presented to US EDs and at least survived for 5 days. And those patients got daily PCT tests. And they... what they found was that the survival in the patients who had a greater than 80% decrease in their procalcitonin level was much higher than those that did not. In another study, and this time it was not necessarily patients with sepsis, it was all-comers to the ED. And at 3 hospitals, also, higher levels of procalcitonin were associated with lower levels of survival. Given the prognostic value of procalcitonin, this study, the PASS-RCT study, asked the question, in critically ill patients, if the PCT level stays high and doesn't go down, and we increase diagnostics, culturing, and mandated antibiotic escalation as compared to standard of care, does that result in a difference? Do we capture cases that we may have missed that are not doing well? And the bottom line is that they found no difference in survival, no difference in proportional antibiotic appropriateness. And instead, on important secondary outcomes, there was worse AKI, more dialysis and mechanical ventilation needs, and greater ICU length of stay in the PCT group. As such, procalcitonin-guided escalation in ICU management cannot be recommended at this time. Now, let's examine the role of procalcitonin in guiding antibiotic discontinuation. In this large meta-analysis, which was an individual patient meta-analysis of patients with acute respiratory tract infections, PCT-guided antibiotic therapy was associated with not only decreased antibiotic exposure and lower antibiotic side effects, but also improved survival. Now, there is something to be considered here. That is, there was both antibiotic discontinuation studies as well as antibiotic initiation studies included in there. And the survival benefit was driven by a single study that was an antibiotic discontinuation study. So these findings cannot be extrapolated to antibiotic initiation. And secondly, these results are difficult to generalize to community-acquired pneumonia per se, because this also included patients with diseases such as asthma exacerbations and COPD exacerbations. How about PCT-guided discontinuation in critical illness and sepsis? There's actually multiple meta-analyses that have been performed that have looked at this question that have uniformly determined that PCT-guided discontinuation in critically ill patients and those with sepsis is associated with decreased antibiotic usage. What's interesting is that while many of them did not show a survival benefit, which is okay considering if there is no difference in survival but decreased antibiotic use, that's still a win, two recent meta-analyses actually showed that procalcitonin-guided discontinuation in critical illness was associated with survival benefit. Now this is important because, you know, especially for critical care providers, survival benefit is really a game-changer. And this makes us wonder whether we should be using procalcitonin more for antibiotic discontinuation. But I think we have to understand where the signal is coming from and what it means, and whether we can really attribute the survival benefit to procalcitonin. To better understand this, we need to go back to the time where these trials were performed in these meta-analyses and understand what was usual care at the time. For pneumonia, the median was 10 days, as seen here in this Medicare study. For suspected sepsis in ICU patients, it was as high as 16 antibiotic days. But usual care is changing. Several trials have come out in the last decade that have shown comparable outcomes for short and long courses of therapy for a variety of bacterial infection syndromes. It's important to understand that usual care dictates today whether PCT might be useful or not. And to understand this, let's look at this comparison of two recent trials. On the left-hand side, a progress trial in Greece that was done in patients with sepsis. On the right-hand side is the Wang study, or the PROACT trial, that was done in patients with suspected lower respiratory tract infections who presented to US EDs. And the one on the left, in sepsis, it showed a benefit, a survival benefit, and reduced antibiotic duration. However, the one on the right-hand side showed no difference in outcome, both in terms of mortality and antibiotic duration. But look carefully at the duration of therapy. Now mind you, this is apples and oranges. It's sepsis and pneumonia. So, if you look at the duration of therapy, on the left-hand side, it's 10 days for sepsis, and on the right-hand side, it's only 4.3 days for pneumonia. And one wonders about whether the procalcitonin survival benefit and reduced antibiotic signal that we've seen over the years in these meta-analyses, are we still going to see that today in our practice with lower or shorter courses of antibiotics? The other concern is whether the evidence underpinning shorter is better is generalizable to critically ill patients or not. So, in this study by Hallie Prescott and colleagues, which was a meta-analysis of 19 randomized controlled trials of hospitalized patients with pneumonia, UTI, and intra-abdominal infections, shorter versus longer courses of antibiotics resulted in no differences in clinical efficacy, microbiological cure, as well as mortality. This may seem reassuring, however, all these studies predominantly occurred among ward patients. And so, in the absence of a representation of the patients that we treat in the ICU, can we really say that shorter is better in critically ill patients or not? This study showed higher initial severity and radiological burden in pneumonia was linked to treatment failures. And I'd be interested to see this sort of study performed in the era of shorter courses to see whether shorter courses for patients with a greater burden of disease is appropriate or not. As such, the surviving sepsis guideline update from 2021 does mention that one could use shorter courses over longer courses. However, it's a weak recommendation with a very low quality of evidence and without specific durations mentioned. Our group conducted yet another meta-analysis on the impact of procalcitonin-guided antibiotic discontinuation on survival and antibiotic durations in critically ill patients. And the objective was a little bit different. We wanted to understand sort of what types of patients and what types of studies showed benefit versus did not show benefit. And so, one thing that we realized is that all the trials that represent the evidence underpinning antibiotic discontinuation in critically ill patients based on procalcitonin were all open-label studies. And they all did... neither of the studies had a low risk of bias. More importantly, although we found the same survival benefit associated with antibiotic discontinuation with PCT in critically ill patients, this benefit was lost when we looked at patients with sepsis. This was lost when we looked at patients with more than 80% procalcitonin algorithm adherence, which basically means that when procalcitonin algorithms are being followed, we're not seeing that survival benefit. And also, that survival benefit was exclusively seen when PCT was used along with CRP, but not without CRP. Also, the benefit... the survival benefit that was seen was largely governed by this one trial called the SAPS trial. And in this study, which showed at least 2 days decrease in antibiotic use in the PCT algorithm and improved survival, there was low algorithm compliance. Less than half of the patients, or the providers, were following the algorithm. And also, the Modified Intention Treat Analysis Fragility Index was 9. So, basically, 9 patients moved to the other side could have flipped the results. And also, the mortality benefit was a bit of an unexpected finding. And so, I think the take-home here is that we need to take a step back and understand a little better what's driving the survival benefit before we sort of attribute it specifically to procalcitonin. In future studies, we need to understand whether procalcitonin is helping us rule out other conditions or rule in other conditions. And the treatment of which might be resulting in better outcomes. As such, according to the Surviving Sepsis Campaign 2021 update for adults with sepsis and septic shock, when the optimal duration is unclear, they suggest using procalcitonin and clinical evaluation to decide when to discontinue antimicrobials. But it is a weak recommendation with low-quality evidence. This talk would be incomplete without looking at the role of procalcitonin in COVID-19. As we saw during the pandemic, there was a very high rate of empiric antibiotic use, and yet a very low prevalence of co-infection. Multiple studies have shown that high procalcitonin levels predict severe COVID-19. This study compared procalcitonin and CRP for diagnosing bacterial co-infection in COVID-19 critical illness. And it found that in the first 48 hours of presentation, procalcitonin performed slightly better than CRP. However, we have to realize that we are in a different time now. We're actually seeing a lot of patients present with bacterial pneumonia that just happened to be tested for COVID-19 and have an incidental positive. And so I think the reliability of procalcitonin, even as a rule-out test, is a bit shaky today compared to previously. Now, let's look at implications for future practice. As we are seeing a growing equivalence between shorter and longer courses based on recent RCT data, it is imperative that we do a present-day reappraisal of real-world use and utility of procalcitonin in ICUs. We need to engage antibiotic stewardship programs more to help with the appropriate use and interpretation of procalcitonin. We also need to improve compliance with follow-up PCT tests. Most patients who are presenting to hospitals with suspected infection are getting PCT tested only once. And even if the original intent of the person who ordered the test was to follow up with multiple tests to allow for discontinuation, that is not being performed. It's well known that inappropriate empiric antibiotic therapy is associated with poor outcomes in bloodstream infections and in sepsis. In fact, in this study, 75% of all bacteremic deaths occurred in patients with Staph aureus and enterobacterials as the bloodstream pathogen. And as a result, if we can try to appropriately diagnose the most common pathogens early enough, we are likely to save many lives. And studies such as this have combined rapid diagnostics and antibiotic stewardship together to show faster detection, faster appropriate escalation, and faster and greater de-escalation, and fewer contaminants, with no differences in survival. And one wonders whether we would...we could have even better outcomes if we combine potential biomarkers along with rapid PCR testing and antibiotic stewardship. In summary, procalcitonin has limited utility in antibiotic initiation and escalation in the ICU. Procalcitonin-guided antibiotic discontinuation, on the other hand, in lower-respiratory tract infections, critical illness, and sepsis does not appear to be harmful and may be used. We are unable to recommend PCT for guiding antibiotic use in COVID-19 at the current time. Further research is needed on how providers respond to PCT results in the real world. We need to understand the mechanisms around any survival benefit attributable to PCT use. And we also need to understand the role of PCT in decreasing antibiotic exposure over and above the current-day shorter regimens. Thank you very much for your time.

biomarkers

infection

procalcitonin

sepsis

antibiotics

ICU

COVID-19