In this observational prospective multicentric study, we investigated the diagnostic performance of clinical, laboratory, CT parameters and candidate biomarkers in 61 critically ill patients with NOMI suspicion. We finally diagnosed 33 intestinal ischemia including 27 intestinal necrosis, defined by stringent criteria. The results of routine work-up investigations showed poor sensitivity of clinical signs, and CT retrieved no sign of mesenteric ischemia in 30% of patients. Interestingly, plasma I-FABP at the time of NOMI suspicion presented high accuracy for necrosis prediction. Additionally, we focused on clinical, biological and management features and their association with outcomes. While no patient survived without necrosis resection, ICU survival increased up to 38.5% in patients undergoing resection of necrotic intestinal segments.
NOMI is known from long time as a challenging diagnosis and is frequently suspected in ICU in a context of a clinical worsening [1, 11]. Recently, monocentric retrospective studies highlighted several conditions affecting critically ill patients potentially promoting NOMI such as septic shock, successfully resuscitated cardiac arrest, post-cardiac surgery or cardiogenic shock [2, 4,5,6, 10]. These observations reinforce the hypothesis that NOMI represents the late stage of acute gastrointestinal failure process [22], preceded by still unclear pathophysiological mechanisms including impaired tissue perfusion responsible for gut barrier failure and endotoxin translocation, endothelial dysfunction and ischemia–reperfusion injury with increased local cytokine production [1, 23,24,25]. A working group of the ESICM recently underlined the need to validate new biomarkers and to increase the pathophysiological understanding of NOMI genesis [14]. To our knowledge, our study is the first prospective multicentric study focused on NOMI diagnosis and prognosis.
The diagnosis process should answer 2 highly important issues. First of all is to perform an early NOMI diagnosis to avoid progression to transmural necrosis [1, 16]. The second one is to dispose reliably information on the presence of intestinal necrosis to guide decision toward surgical treatment. In this study, we used stringent criteria to evaluate the diagnosis features accuracy according to the presence of necrosis or ischemia. Confirming previously published data [4], clinical digestive signs lacked sensitivity and specificity in performing NOMI diagnosis. In the same way, the increase of routine laboratory markers reflecting tissue ischemia lacked specificity, but should reinforce NOMI suspicion in patients at risk [17, 26]. In our cohort, CT signs of mesenteric ischemia could also not differentiate intestinal necrosis from ischemia without necrosis. In the literature, bowel dilation has been proposed as a marker of transmural intestinal necrosis, being more accurate when associated with multiorgan failure and increased arterial lactate [27]. However, most of mesenteric ischemia from the cohort of Nuzzo et al. has a vascular occlusion origin, a setting with increased CT diagnosis performance. On the contrary, we already reported disappointing CT performance in the specific NOMI setting and here we observed the same findings [4], CT scan concluding to no sign of intestinal ischemia in almost one-quarter of patients with intestinal necrosis.
In this cohort, we prospectively investigated plasma I-FABP and citrulline performance in NOMI, as they represent potential promising biomarkers. In particular, experimental studies demonstrated early increase of I-FABP after gut ischemia onset [28, 29]. Comparing plasma I-FABP concentration at time of suspicion in 27 patients with definite intestinal necrosis to 13 patients with intestinal necrosis ruled out, we found an AUC of 0.83 [0.70–0.96], and proposed a threshold of 3114 pg/mL with good positive predictive value (90% [67–96]) and moderate negative predictive value (58% [36–93]). Thuiyjls et al. studied I-FABP accuracy in 22 AMI patients compared to 24 other patients with initial AMI suspicion finally ruled out [30]. In their work, urinary and plasma I-FABP AUC reached 0.93 and 0.70. However, in critical illness and particularly in NOMI, we observed a high prevalence of acute renal failure, and urine might not be available. Another study of Matsumoto et al. found an AUC of 0.88 for AMI diagnosis including 15 NOMI and 9 arterial occlusions [31]. The authors highlighted I-FABP increase in various non-vascular intestinal ischemia etiologies. Although both studies of Thuiyjls and Matsumoto suffered methodological issues concerning classification of ruled out cases of mesenteric ischemia, altogether, these results suggest I-FABP could be a reliable and early biomarker of NOMI. Importantly, I-FABP threshold for mesenteric ischemia diagnosis is not consensual [25] and its accuracy may differ according to ELISA kits [32]. While promising, plasma I-FABP integration in order to monitor intestinal ischemia is probably too early at this point and should be further explored to refine plasma I-FABP accuracy in larger cohorts.
Plasma citrulline, proposed as a marker of acute intestinal failure in critically ill patients [19], had never been studied in a NOMI cohort before. We observed no relationship between plasma citrulline levels and presence of NOMI. Furthermore, plasma citrulline levels in presence of intestinal necrosis were not associated with outcome. In the literature, low plasma citrulline in critically ill patients has been reported, and was found to be associated with clinical signs of intestinal dysfunction, bacterial translocation, elevated I-FABP and worse outcomes, suggesting rational for its use as a NOMI biomarker [19, 23, 33, 34]. Our findings could be explained by a delayed time of measurement compared to previously cited studies investigating it early after admission, and a high prevalence of acute renal failure in our cohort, which may lead to high plasma citrulline concentrations despite reduction of the enterocyte mass [20]. Our data do not support an interest of citrulline in diagnosis of late stage NOMI.
NOMI therapeutic management is based on scarce evidence in the literature. Angiography, enabling the in situ administration of a continuous infusion of vasodilatory drugs, was considered an efficient treatment for NOMI [35,36,37]. However, the clinical benefit of this technique is unknown at the stage of intestinal necrosis [12]. The tolerance of vasodilatory drugs in hemodynamically unstable patients is unclear and treatment relies mainly on necrotic intestinal segment resections. To our knowledge, increased survival associated with resection of necrotic intestinal segments (38.5% versus no survivors without surgical resection) has never been reported in NOMI setting before. These findings highlight the importance to improve the screening of NOMI patients who may benefit of surgery given their high expected mortality in absence of necrosis resection [38]. Importantly, our results found potential interest of I-FABP in this way, allowing diagnosis of intestinal necrosis. However, the statistical association of I-FABP with ICU mortality in presence of intestinal necrosis was not confirmed in multivariate analysis. This result could be explained by the small size of the population, powered primarily to investigate diagnosis performance. Larger studies could help to clarify the interest of I-FABP in surgical treatment decision-making for NOMI patients. Lastly, bacteremia related to intestinal translocation was observed in 30% of NOMI patients, suggesting the benefit of antibiotic regimen targeting bacteria from digestive tract. Endotoxemia related to gut barrier rupture had been observed by Grimaldi and al after cardiac arrest [23], reinforcing this finding.
While common in the field of clinical research on NOMI in critical care, our study’s limitations are mainly methodological. The pathophysiology and the time-course of gastrointestinal failure are still unprecise, may vary greatly, and definitions are lacking [14]. We have chosen to focus on NOMI, thought to represent the worst stage of gastrointestinal injury, and raising unsolved diagnosis and therapeutic issues. The low incidence of NOMI requires an appropriate selection of the study population with consideration of the pre-test probability leading us to propose inclusion criteria. These criteria for NOMI’s suspicion, based on current knowledge in the field, could be interpreted as too late, resulting in the high severity of illness at the time of diagnosis. However, a lower threshold of NOMI suspicion would have led to unjustified invasive exams. We recognize that this high pre-test probability may have resulted in the high diagnosis performance observed for plasma I-FABP. Additionally, classification of patients in which NOMI can be ruled out is challenging. Consequently, NOMI was diagnosed using stringent criteria, mainly based on macroscopic findings, increasing the validity of patients’ classification. Also, whereas abdominal distension presented an interesting trend in diagnosis performance and in prognosis value in patients who had a diagnosis of intestinal necrosis, it has to be acknowledged that abdominal distension is a non-parametric parameter subject to variability assessment. Intra-abdominal pressure measurements and abdominal compartment syndrome as defined by the World Society of Abdominal Compartment Syndrome would have provide interesting information regarding NOMI diagnosis and prognosis in the study population but were not recorded by the study centers in usual care [39]. Finally, while observing increased survival of patients with intestinal necrosis resection, the observational design of the study does not allow to conclude a causal link.