Failure Risk Factor For Non-Operative Treatment of Splenic Trauma

Background: Splenic trauma is a common pattern for admission in blunt abdominal trauma. The objective of this study is to identify risk factors for failure of non-operative management (NOM) in splenic trauma. Methods: This is a retrospective monocentric analysis of a prospectively collected database. All patients admitted in the university hospital of Nice [Centre Hospitalier Universitaire (CHU) de Nice, France] for a splenic trauma from January 1 st 2006 to January 6 th 2018 were included. Primary outcome was the need for delayed splenectomy as an indicator of NOM failure. Results: Two-hundred-eighty patients were included in this study. Most splenic lesions were severe grades (grade 3 or higher). In total, 83 splenectomies were performed urgently, i.e. 29% of patients; 88 angioembolizations, i.e., 31% of patients with a success rate greater than 80%; 14.7% of 136 patients who had no previous angio-embolization required secondary splenectomy; 19.7% of the 61 patients who had anterior angio-embolization required secondary splenectomy. Age was not found associated with a higher failure rate (44 years in successful embolization vs 37.5 years in NOM-failure group, p = 0.15). Higher drop in hemoglobin levels between admission and 6 hours after admission was detected in the embolization failure group (-1.44 g/dl) as compared with the successful group (-0.68 g/dl), which approached statistical signicance (p = 0.064). Conclusions: Hemoglobin monitoring in the hours following the admission of a patient with splenic trauma might be an important factor during the medical supervision of hemodynamically stable patients. Early identication of patients at high risk of NOM failure by hemoglobin monitoring may prevent late splenectomy.


Background
Abdominal trauma is present in 7 to 10% of severely injured patients. In the case of blunt abdominal trauma, 45% have a splenic lesion, making it the most frequently injured organ in blunt abdominal trauma (1).
CT-scan is the reference imaging exam for trauma patients whose hemodynamic status is considered stable or responds to vascular lling (2,3). CT-scan can identify splenic lesions, provide a morphological description according to the AAST classi cation (American Association for the Surgery of Trauma) (4), but the scanner is not the only parameter to consider when choosing the best treatment for trauma patients.
The gold-standard treatment for patients with splenic injury in blunt abdominal trauma with stable hemodynamic status is the non-operative management (NOM), which is performed in approximately 80% of patients (5,6). This medical management is effective in 62 to 92% of cases, likely because of the increasingly frequent use of splenic angioembolization (5,(7)(8)(9)(10). It is di cult to identify patients who will have NOM failure and require secondary splenectomy (5,6,(11)(12)(13)(14). Many risk factors for NOM failure have been proposed, but none can accurately predict the failure of NOM.
This study aims to nd parameters to predict NOM-failure in case of blunt abdominal trauma with a splenic lesion, which will help in selecting patients in whom splenectomy rst intention would be indicated.

Methods
This a retrospective monocentric study of a prospectively collected database. All patients admitted in the university hospital of Nice (Centre Hospitalier Universitaire (CHU) de Nice, France) for a splenic trauma from January 1st 2006 to January 6th 2018 were included. Primary outcome was the need for delayed splenectomy in patients treated with NOM as an indicator of NOM failure.
The following data were collected: epidemiological data, gender, circumstances of the accident, hemodynamic status at admission, injury associated with the calculation of ISS score (Injury Severity Score), patient management, length of stay, and possible morbidity and mortality [according to Clavien-Dindo classi cation (15)].
Abbreviated Injury Scale (AIS) classi cation was used to classify trauma severity from CT scans of hemodynamically stable patients or according to operative ndings.
Biological data were also analyzed, such as hemoglobin (g/dl), arterial pH, and arterial lactates at admission and every 6 hours during the rst 36 hours of management.
Hemodynamic stability was de ned by heart rate between 50 and 90 beats per minute and systolic blood pressure above 90 mmHg. Permissive hypotension was not a goal at the time of the study.
Patients with extra-abdominal trauma requiring therapeutic procedure other than splenic were included in the splenic NOM group.
Selective distal splenic embolization was proposed in hemodynamically stable patients for whom the CT scan revealed active contrast leakage or splenic pseudoaneurysm. Proximal prophylactic angioembolization was not proposed in this study.
In the case of NOM, abdominopelvic CT-scan with contrast injection at arterial and portal time was routinely performed on the 5th day after the trauma. In the event of discomfort, deglobulization, appearance of abdominal pain, or other clinical signs considered worrying, an emergency CT scan could be performed earlier.
The primary endpoint was NOM failure with secondary splenectomy.

Page 4/17
The failure of NOM was considered in the following cases: Multiple intrasplenic pseudoaneurysms (> 3) at the control CT scan performed at day 5 after the trauma; Illness with hypotension, hemorrhagic shock or massive deglobulization suggestive of a secondary rupture of the spleen and imposing an emergent splenectomy; Infection of the previously embolized splenic parenchyma areas: abdominal pain in the left hypochondrium, fever, biological in ammatory syndrome, and signs of infection of the embolized territory with CT injected (intra-parenchymal air bubbles). This latter situation represents the late failures of NOM, referred to as secondary splenectomy in our study.
The database is approved by the French committee for privacy statements (Commission Informatique et Liberté) and authorized by the University Hospital of Nice. The data is stored on a secure server at Nice University Hospital.
A statistical analysis was performed. Data extraction and descriptive statistics were performed using Stata→ 15.0 software (Stata Corp., College Station, USA).
Univariate analysis was performed looking for NOM failure factors and angioembolization using a t-test.

Results
A total of 280 patients were included in our study. Descriptive data are listed in Table 1. More than 70% of the patients were men, and most of the splenic lesions were severe (grades > 3). angioembolizations i.e., 31% of all patients, with a success rate greater than 80%.
The mortality rate after emergency splenectomy was high, reaching 10%. Table 2 shows the continuous variables. The median age was 39 years, and the Glasgow Coma Scale was 13 at admission. The ISS score was high, with a median of 28.9. There was no statistically signi cant difference with ISS score between successful NOM and failure of NOM. Hemoglobin levels were also monitored during the rst 36 hours, ranging from an average of 12 g/dl at entry to 10.8 g/dl at 36 hours.
Some data were missing and could not be included in our study. This mainly concerns biological data, such as pH changes, lactate levels, and changes in hemoglobin. This can be explained by the variability of care depending on the patient's condition and the blood test performed, which is considered repetitive for some. Figure 1 shows the time when angioembolization was performed in the case of NOM. In 57% of cases, angioembolization was performed on the day of admission, which corresponds to patients presenting with active bleeding or aneurysm and in a stable hemodynamic state. This rate then decreases until we observe a second peak on the 5th day (11% of patients), corresponding to the completion of a control CTscan.
It can be observed that 28% of splenectomies were performed on the day of the admission, followed by a decrease, with a subsequent increase at D5 and D8 (12.8% of splenectomies). Again, these peaks correspond to the control imaging with the presence of a signi cant lesion not accessible to angioembolization or deglobulization (Fig. 2).
Embolization was not found as a protective factor for performing secondary splenectomy. A total of 14.7% of the 136 patients (i.e. 20 patients) who did not have primary angioembolization required secondary splenectomy, and 19.7% of the 61 patients (i.e. 12 patients) who had primary angioembolization required secondary splenectomy (p = 0.382).
There is no statistically signi cant difference between age and failure of nonoperative management; the median age of the successful embolization group was 44 years vs 37.5 years in the angioembolization failure group (p = 0.15).
The last analysis compared the decrease in hemoglobin between admission (H0) and 6 hours after admission (H6) from management between the angioembolization group and the angioembolization failure group. The hemoglobin differential between H0 and H6 appears to increase in patients with failed angioembolization (Table 3). The group of patients who had an angioembolization failure had a decrease in hemoglobin level of -1.44 g/dl between H0 and H6 vs. -0.68 g/dl for those who had a successful angioembolization (p = 0.064).
Many risk factors for NOM failure have been proposed, but none can accurately predict the failure of NOM.
Arterial blush on CT in the absence of angio-embolization is described as a risk factor of conservative treatment failure (5,8,18), as well as diffuse hemoperitoneum (14,16).
A severe injury associated with a high ISS would be predictive of a higher rate of NOM failure. Different ISS values are reported in the literature as limiting to a NOM [e.g., 15 (9), 25, and others (14,16,17)]. Combining a splenic trauma with a brain injury can complicate the surveillance (19).
The importance of age in NOM has been debated. Age greater than 55 years would increase the risk of NOM failure, especially for high-grade lesions (5,14,20); age greater than 40 years could be a NOM failure factor, according to other authors (16). In our study, the median age was 39 years. We detected no statistically signi cant difference for age; nor did we detect a statistically signi cant difference according to AAST classi cation, ISS, or mechanism of injury.
Biological abnormalities, such as a decrease in hemoglobin, pH, or an increase in arterial lactates, might be factors for NOM failure. Alcohol consumption is common during trauma and might result in a falsely high value of lactate or prolong lactate clearance; it must, therefore, be taken into account when evaluating patients with a high blood alcohol level (21). The value of lactate and bicarbonates was identi ed as an important independent predictor of the polytrauma patient with acute alcohol and drug use (22,23).
In our study, there is a difference in hemoglobin values at H0 and H6 between the groups having failed angioembolization and the successful group of angioembolization. This result, however, remains nuanced since it is, by little, not statistically signi cant (p = 0.064). This result can probably be explained because of the small number of patients. This result is nevertheless interesting since it is understood that biological deglobulization may be the rst sign of true bleeding requiring secondary splenectomy.
In the literature, a decreased value of hemoglobin in the hours following the admission of trauma patients has never been described as NOM failure factors. On the other hand, the need for transfusion of red blood cells is a risk factor for NOM failure (16,18).
In our study, the lack of data can represent a bias. Thus, the study of lactate levels could not be performed; this is explained by the disparity of care and evolution of patients. Indeed, lactate often dosed at H0 was not necessarily dosed in the aftercare. A study addressing the variation of lactatemia as a failure factor of NOM in splenic trauma might be interesting.
In this study, we could not demonstrate that angioembolization prevented secondary splenectomy. However, this is not comparable to the lack of bene t and is likely due to the lack of power of the study.
A multivariate analysis of the different predictors of NOM failure would also be useful. Unfortunately, we could not achieve this because of the amount of missing data. A multicenter prospective study would address these limitations and should now be conducted to con rm the role of deglobulization in the management of patients with splenic trauma.

Conclusion
Surgical emergencies are an important part of the activity of our hospitals. The trauma patient is a key issue because of the acute care required and the choice of the most appropriate treatment. NOM has become the standard for a hemodynamically stable patient, even though it has a signi cant failure rate.
Deglobulization 6 hours after admission could represent a new risk factor for NOM failure. The need to highlight predictive factors of NOM failure in multicenter prospective trials is essential to achieve optimal patient care.